TONAL DEVELOPMENT OF TAI LANGUAGES

TONAL DEVELOPMENT OF TAI LANGUAGES LIAO HANBO 廖漢波 Presented in Partial Fulfillment of the Requirements for the Degree of MASTER OF ARTS IN LINGUISTICS PAYAP UNIVERSITY May 2016 Title: Tonal Development of Tai Languages Researcher: Liao Hanbo Degree: Master of Arts in Linguistics Advisor: Assistant Professor Phinnarat Akharawatthanakun, Ph.D. Approval Date: 31 May 2016 Institution: Payap University, Chiang Mai, Thailand The members of the thesis examination committee: 1. _________________________________ Committee Chair (Professor Somsonge Burusphat, Ph.D.) 2. _________________________________ Committee Member (Assistant Professor Phinnarat Akharawatthanakun, Ph.D.) 3. _________________________________ Committee Member (Robert Wyn Owen, Ph.D.) Copyright © Liao Hanbo Payap University 2016 ACKNOWLEDGEMENTS First and foremost, I would like to extend my sincerest thanks to my thesis advisor, Assistant Professor Dr. Phinnarat Akharawatthanakun, for her continuous support and patience towards my research. She introduced the knowledge of literature to me during my M.A. studies. She further guided me to do the fieldwork, and provided critical feedback on my research skills, my views and my attitude towards the issue. Her insightful comments sharpened my mind. It is she who enlightened me while I pursued my academic career. I would like to acknowledge the financial assistance from the Ford Foundation International Fellowships Program for the postgraduate scholarship which supported the first two years of my study in Payap University, Chiang Mai, Thailand. Special thanks to Assistant Professor Dr. Tai Chung-pui for his steadfast support and encouragement during my studies, to Dr. Robert Wyn Owen for providing useful comments, to Professor Dr. Somsonge Burusphat for agreeing to be my thesis committee chair and for her time of reading and giving me suggestions, to Mr. Louis William Davis for helping with the proofreading of my thesis, to Researcher Alexis Michaud for his kind and patient directions on tonology, and to Ms. Chen Yen-ling for introducing me to the literature of my research area. I must thank Ms. Suthinee Promkandorn, my M.A. classmate, who introduced me to Thai language and culture when I was a new arrival to Thailand. Besides, she kindly helped me to translate my thesis abstract into Thai. I also want to express my gratitude to Mr. Suparak Techachareonrungrue who helped on the process of Thai translation of the abstract. I would also like to thank the scholars whom I met at various stages of my study. They are Assistant Professor Kao Ya-ning, Dr. Margaret Milliken, Professor Li Jinfang, Professor Stephen Matthews, Associate Professor Masayuki Yoshikawa, Dr. Shen Ruiqing, Professor Pan Qixu, Professor Fan Honggui, Mr. Sattanan i Saengsrichan, Associate Professor Lu Xiaoqin, Mr. Liang Xianning, Mr. Lu Yungao, Ms. Xu Xiaoming, and Mr. Lu Guangzhou. I am grateful to all the Language Resource Persons (LRPs) whom I interviewed during the fieldtrip. I was deeply moved by their enthusiastic help. I dedicate this thesis to my family in particular to my mother, who passed away during my first year of study in Payap. In the past few years, accompanied by my father, she must have been gazing at me graciously in heaven. My sister-in-law Ms. Yan Liyan has utilized all her personal connections to find the most suitable LRPs in Guangxi for me, for which I am grateful. I could never express my gratitude by words for their love. Liao Hanbo ii Title: Tonal Development of Tai Languages Researcher: Liao Hanbo Degree: Master of Arts in Linguistics Advisor: Assistant Professor Phinnarat Akharawatthanakun, Ph.D. Approval Date: 31 May 2016 Institution: Payap University, Chiang Mai, Thailand Number of Pages: 225 Keywords: Tai languages, Tonogenesis, Tonal split, Tone box, Tonology ABSTRACT This thesis aims to provide a full scheme of tonal development of Tai, from tonogenesis in proto-Tai to different diachronic hierarchies of tonal splits in different Tai groups and varieties, and further suggests a new viewpoint on Tai classification as well as the revised Tai tone box based on tonology. The analysis is based on data of tones from 42 Tai varieties, most of which were collected during my fieldworks in China and Thailand. Data collection and analysis has been accomplished by a revised version of Gedney’s tone box (Gedney 1989[1972], Liao & Shen 2012), which is supposed to contain the basic patterns of tonal split and merger in Tai languages. Any tone split which is not reflected in the tone box is emphasized to discuss. The discussion of the diachronic processes of tonal development of Tai languages is the main body of this thesis. Tonogenesis is treated as the very first step of the arising of tones in proto-Tai. After tone arose, two groups of proto-initials (voiceless-voiced) conditioned primary tonal splits whereas splits conditioned by other pharyngeal features, such as aspiration and pre-glottalization represent stages of secondary development. Some secondary tonal split patterns result in three-way register splits. However, the bearing capacity of the tone numbers in a language often cause the third register to merge into one of the other two to make two-way register splits always be the mainstream. On the basis of the full tonal development scheme explored, the other two hypotheses are attested as follows. First, based on the analysis of tonal behaviors with supplementary evidence from initial behaviors, vocalic behaviors, and exclusive lexical items, this thesis suggests two levels of Tai division as follows. Primary, it agrees on the two-term system iii suggested by Haudricourt (1956) to divide proto-Tai first into Dioi or ʔɤaɓ vs. Tai proper, by another two-term system, NT-YN (Northern Tai-Yongnan Zhuang) and Southern Tai. Li’s Central Tai and Southwestern Tai (Li 1977) can be divided from each other under the Southern Tai sub-group, in contrast Northern Tai and Yongnan Zhuang differ from one another under the NT-YN sub-group. Second, based on all the possible conditioning phonation types of initials at time of tonal splits, an integrated tone box is provided. However, a unified Tai tone box have been proved to be cumbersome if a Tai variety’s affiliation has been known. It is because not all Tai varieties obtain all types of secondary tonal splits. For instance, Northern Tai varieties lack of contrastive aspirated sounds, and unaspirated stops in NT never go with glottalized initials to condition secondary tonal splits. Therefore, a tone box designated with the divisions containing the rows of aspirated sounds and unaspirated stops appear to be unnecessary for Northern Tai. For this reason, Tai tone boxes designated with specific pertinence to different Tai periods and different Tai sub-groups are first suggested if the diachronic period of a Tai variety or the affiliation of a modern Tai variety are known. Nevertheless, the integrated Tai tone box is finally suggested to be able to capture the tonal distinctions in any Tai variety. iv ชืไอ฼รืไอง: ผูຌวิจัย: ชืไอปริญญา: อาจารย์ทีไปรึกษาวิทยานิพนธ์: วันทีไอนุมัติผลงาน: สถาบัน: จานวนหนຌา: คาสาคัญ: พัฒนาการของ฼สียงวรรณยุกต์฿นภาษาตระกูลเท ฼ลีไยว หຌาน฾ป ศิลปศาสตรมหาบัณฑิต ิภาษาศาสตร์ี ผูຌช຋วยศาสตราจารย์ ดร. พิณรัตน์ อัครวัฒนากุล 31 พฤษภาคม 2559 มหาวิทยาลัยพายัพ จังหวัด฼ชียง฿หม຋ ประ฼ทศเทย 225 ภาษาตระกูลเท กา฼นิดวรรณยุกต์ การ฽ยก฼สียงวรรณยุกต์ กล຋องวรรณยุกต์ การศึกษาวรรณยุกต์ บทคัดย่อ วิทยานิพนธ์ฉบับนีๅมุ຋ง฼สนอรูป฽บบพัฒนาการของวรรณยุกต์฿นกลุ຋มภาษาตระกูลเท ฾ดย฼ริไมจากการ กา฼นิดของวรรณยุกต์฿นภาษาเทดัๅง฼ดิมเปสู຋ลาดับขัๅนของการ฽ยก฼สียงวรรณยุกต์฽บบขຌามสมัย฿น ภาษาตระกูลเทกลุ຋มต຋าง โ ฽ละวิธภาษาเทต຋าง โ นอกจากนัๅนยัง฼สนอมุมมอง฿หม຋฿นการ฽บ຋งกลุ຋ม ภาษาตระกูลเท฽ละการปรับระบบกล຋องวรรณยุกต์ตาม฼กณฑ์การศึกษาวรรณยุกต์ การวิ฼คราะห์ ขຌอมูลอาศัยวรรณยุกต์จากวิธภาษาตระกูลเท ไโ ภาษา ซึไงส຋วน฿หญ຋เดຌจากการ฼กใบขຌอมูลภาคสนาม฿น ประ฼ทศจีน฽ละประ฼ทศเทย การ฼กใบขຌอมูล฽ละการวิ฼คราะห์ขຌอมูลเดຌยึด฽นวคิดกล຋องวรรณยุกต์ของ Gedney (แ้่้ [แ้็โ]ี ฽ละ Liao & Shen (โเแโี ทีไ฼สนอรูป฽บบพืๅนฐานของการ฽ยก฽ละรวม ฼สียงวรรณยุกต์฿นกลุ຋มภาษาตระกูลเท ฾ดยการวิจัยครัๅงนีๅเดຌ฿หຌความสาคัญกับการ฽ยก฼สียงวรรณยุกต์ ทีไเม຋สอดคลຌองกับ฽นวคิดกล຋องวรรณยุกต์฼พืไอนามาอภิปรายผลดຌวย การอภิ ป รายพั ฒ นาการของวรรณยุ ก ต์ ฿ นภาษาตระกูล เท฽บบขຌ า มสมั ย ถื อ ฼ป็ น ประ฼ดใ นหลั ก฿น วิทยานิพนธ์ฉบับนีๅ การศึกษากา฼นิดวรรณยุกต์ถือ฼ป็นขัๅน฽รกทีไนาเปสู຋การ฼กิดขึๅนของวรรณยุกต์ภาษา เทดัๅง฼ดิม ฼มืไอมีวรรณยุกต์฼กิดขึๅน ฼สียงพยัญชนะตຌนดัๅง฼ดิม โ กลุ຋ม ิพยัญชนะ฼สียงเม຋กຌอง – พยัญชนะ ฼สี ย งกຌ อ งี ถื อ ฼ป็ น ฼งืไ อ นเขปฐมภูมิ ทีไ มีผ ลต຋ อ การ฽ยก฼สี ย งวรรณยุก ต์ ฾ดยขัๅ น ต຋ อ มาการ฽ยก฼สียง วรรณยุกต์฼ป็นผลมาจากลักษณ์ของ฼สียง฿นช຋องคอ ฼ช຋น การพ຋นลม฽ละการนาดຌวยการกัก฼สຌน฼สียงซึไง ฼ป็น฼งืไอนเขขัๅนทุติยภูมิทีไมีอิทธิพลต຋อกระบวนการดังกล຋าว รูป฽บบการ฽ยก฼สียงวรรณยุกต์฿นขัๅน ทุติยภูมิบางรูป฽บบส຋งผล฿หຌ฼กิดการ฽ยก฼สียงวรรณยุกต์ออก฼ป็น ใ ทาง อย຋างเรกใดี ความสามารถ฿น การรองรับวรรณยุกต์฿นภาษาหนึไง โ มีผลทา฿หຌ฼กิดการรวม฼สียงวรรณยุกต์ทีไสามกับหนึไง฿นสองของ วรรณยุกต์ทีไ฼หลือจึงทา฿หຌการ฽ยก฼สียงวรรณยุกต์฽บบ โ ทางซึไงถือ฼ป็น฽บบหลัก หลังจากศึกษา รูป฽บบพัฒนาการของวรรณยุกต์อย຋างละ฼อียดดังกล຋าว จึงมีการทดสอบสมมติฐาน฼พิไม฼ติม โ ประการ ดังนีๅ v ประการ฽รก จากการวิ฼คราะห์พฤติกรรมของวรรณยุกต์ พรຌอมกับหลักฐาน฼พิไม฼ติมจากพฤติกรรมของ ฼สียงพยัญชนะตຌน พฤติกรรมของ฼สียงสระ ฽ละจากชุดคาศัพท์฼ฉพาะ วิทยานิพนธ์ฉบับนีๅเดຌนา฼สนอ การ฽บ຋งภาษาตระกูลเทออก฼ป็น โ ระดับ ฿นระดับ฽รก ผลการวิจัย฼หในพຌองกับระบบการ฽บ຋ง฼ป็น โ กลุ຋มทีไนา฼สนอ฿นงานของ Haudricourt (1956) ฿นการ฽บ຋งภาษาเทดัๅง฼ดิม฼ป็น Dioi หรือ ʔyai กับ ภาษาเททัไวเปตามระบบการ฽บ຋ง฼ป็น โ กลุ຋ม คือ NT-YN (ภาษาเท฼หนือ – จຌวง หยงหนานี ฽ละ ภาษาเท฿ตຌ ส຋วนภาษาเทกลาง฽ละเทตะวันตกจากงานของ Li (1977) สามารถ฽ยกออกจากกัน฾ดยจัด อยู຋ภาย฿ตຌกลุ຋มย຋อยของภาษาเท฿ตຌ ฿นทางตรงกันขຌาม ภาษาเท฼หนือ฽ละจຌวงหยงหนานกลับมีความ ฽ตกต຋างกันอย຋างสิๅน฼ชิง฾ดยจัดอยู຋ภาย฿ตຌกลุ຋ม NT-YN ประการทีไสอง จากลักษณะ฼สียงพูดของพยัญชนะตຌนทีไ฼ป็น฼งืไอนเขเดຌทัๅงหมด฿นกระบวนการ฽ยก฼สียง วรรณยุกต์ จึงมีการ฼สนอกล຋องวรรณยุกต์฽บบผสมผสาน ทัๅงนีๅกล຋องวรรณยุกต์ภาษาตระกูลเท฿นรูป ฽บบ฼ดิมนัๅน฽สดง฿หຌ฼หในถึงความยุ຋งยากซับซຌอน฼กินเป หากพิจารณา฿น฽ง຋ของความ฼กีไยวขຌองสัมพันธ์ กั น ของวรรณยุ ก ต์ ฿ นภาษาต຋ า ง โ ฼นืไ อ งจากเม຋ ฿ ช຋ ทุ ก ภาษาย຋ อ ย฿นตระกู ล เททีไ มี ก าร฽ยกตั ว ของ วรรณยุกต์฿นขัๅนทุติยภูมิ ตัวอย຋าง฼ช຋น วิธภาษา฿นกลุ຋มเท฼หนือเม຋มีการ฼ปรียบต຋างของ฼สียงพ຋นลม ฽ละ ฼สียงกัก฽บบเม຋พ຋นลมจะเม຋ปรากฏกับพยัญชนะตຌนทีไมีการกัก฼สຌน฼สียงจึงทา฿หຌเม຋มีการ฽ยกตัวของ วรรณยุกต์฿นขัๅนทุติยภูมิ ดังนัๅน กล຋องวรรณยุกต์ซึไงมี฽ถวทีไประกอบเปดຌวย ฼สียงพ຋นลม ฽ละ฼สียงกัก ฽บบเม຋พ຋นลมจึงเม຋จา฼ป็นสาหรับภาษา฿นกลุ຋มเท฼หนือ ดຌวย฼หตุนีๅ วิทยานิพนธ์ฉบับนีๅจึงมีขຌอ฼สนอ฽นะ ว຋า หากช຋วง฼วลาของภาษาเทหรือวิธภาษาเทนัๅน โ ฼ป็นทีไทราบชัด฼จน ฿หຌนากล຋องวรรณยุกต์ทีไบ຋งชีๅถึง ความ฼กีไยวขຌองกันระหว຋างภาษาเทจากหลากหลายช຋วง฼วลา฽ละหลายกลุ຋มมา฿ชຌพิจารณา฼ป็นอันดับ ฽รก อย຋ า งเรกใ ต าม กล຋ อ งวรรณยุ ก ต์ ฽ บบผสมผสานนีๅส ามารถน าเป฿ชຌ ฽ สดงความ฽ตกต຋ า งดຌาน วรรณยุกต์ของวิธภาษาต຋าง โ ฿นตระกูลภาษาเทเดຌดຌวย฼ช຋นกัน vi TABLE OF CONTENTS Acknowledgements ......................................................................................................... i Abstract ........................................................................................................................ iii บทคดยอ............................................................................................................................ v List of Tables ................................................................................................................. x List of Figures ............................................................................................................ xiii List of Abbreviations and Symbols............................................................................. xiv Glossary ...................................................................................................................... xvi Introduction ................................................................................................... 1 1.1 Background .......................................................................................................... 1 1.1.1 Tai and related languages.............................................................................. 2 1.1.2 Tai classification ........................................................................................... 3 1.1.3 ISO 639-3 language code applied to Tai languages...................................... 4 1.1.4 Tai tones ........................................................................................................ 7 1.2 Objectives of the research .................................................................................... 9 1.2.1 Conventional descriptions of Tai tones....................................................... 10 1.2.2 More pending issues on Tai tones ............................................................... 16 1.2.3 Outstanding problems ................................................................................. 25 1.2.4 Three hypotheses for solving problems ...................................................... 30 1.3 Overview of the thesis ....................................................................................... 30 1.3.1 Scope and limitations .................................................................................. 30 1.3.2 Contribution of the thesis ............................................................................ 32 1.3.3 Chapters outline .......................................................................................... 33 Literature Review ........................................................................................ 34 2.1 Historical and comparative Tai .......................................................................... 34 2.1.1 Linguistic changes in Tai ............................................................................ 35 2.1.2 Comparative method applied to Tai ............................................................ 37 2.1.3 Language classification associated to Tai ................................................... 39 2.2 Tonology ............................................................................................................ 45 2.2.1 Tonogenesis ................................................................................................ 47 vii 2.2.2 Tonal splits and mergers ............................................................................. 59 2.2.3 The conditioning factors of tonal splits ...................................................... 69 2.2.4 Tai classification related to tonological issues............................................ 74 2.3 Summary ............................................................................................................ 79 Methodology, data and notation .................................................................. 80 3.1 General historical and comparative method ...................................................... 80 3.1.1 Linguistic change ........................................................................................ 80 3.1.2 Borrowing ................................................................................................... 81 3.1.3 Comparative method ................................................................................... 82 3.1.4 Linguistic classification .............................................................................. 83 3.1.5 Areal linguistics .......................................................................................... 84 3.2 Library research ................................................................................................. 85 3.3 Fieldwork ........................................................................................................... 85 3.3.1 Data collection and the LRPs ...................................................................... 86 3.3.2 Data management and PRAAT ................................................................... 88 3.4 Notation.............................................................................................................. 89 Diachronic hierarchies of Tai tonal development ....................................... 92 4.1 Tai Tonogenesis and earlier tonal behaviors revisited ....................................... 92 4.1.1 Principled clarifications of Tai tonogenesis ................................................ 93 4.1.2 A hypothesis of Tai tonogenesis based on tonal correspondence between Tai and Sinitic languages ..................................................................................... 96 4.1.3 Speculation of specific phonetic shapes of PT tones .................................. 98 4.1.4 Conclusion about Tai tonogenesis ............................................................ 120 4.2 Primary tonal splits in Tai ................................................................................ 121 4.2.1 Phonation types of initial consonants at the PT level ............................... 121 4.2.2 The cause of the primary tonal split .......................................................... 124 4.2.3 Straightforward tonal split in Tai languages ............................................. 127 4.3 Secondary tonal splits in Tai ............................................................................ 131 4.3.1 Determinations of the diachronicity of secondary tonal splits.................. 131 4.3.2 Diachronic process of secondary tonal splits conditioned by syllable initial phonations .......................................................................................................... 140 4.3.3 Secondary tonal splits conditioned by vocalic length ............................... 173 4.4 Summary .......................................................................................................... 182 Tai classification based on tonology ......................................................... 184 5.1 The tonological criteria of subgrouping Tai languages ................................... 184 5.2 NT-YN and ST: Tai classification based on the primary tonal split................ 187 viii 5.3 NT/YN + CT/SWT: Tai classification based on secondary tonal splits .......... 189 5.4 Supplementary evidence for supporting Tai division based on tonology ........ 192 5.4.1 Initial behaviors ........................................................................................ 192 5.4.2 Vocalic behaviors...................................................................................... 193 5.4.3 Exclusive lexical items ............................................................................. 195 5.5 Summary .......................................................................................................... 197 A new perspective of Tai tone box............................................................ 199 6.1 From carrying capacity to practicability .......................................................... 199 6.2 PT tone box: from tonogenesis to the primary tonal split ................................ 202 6.3 Tai tone boxes based on secondary tonal splits in ST ..................................... 204 6.4 Tai tone box based on secondary tonal splits in NT-YN ................................. 208 6.5 Integrated Tai tone box and its application ...................................................... 210 6.6 Limitations of Tai tone box.............................................................................. 215 6.7 Summary .......................................................................................................... 217 Conclusion................................................................................................. 218 7.1 An overview of this study ................................................................................ 218 7.2 Conclusions related to the goals of this study .................................................. 220 7.2.1 A diachronic scheme of Tai tonal development ....................................... 220 7.2.2 Implications of the diachronic process of Tai tonal development ............ 222 7.3 Suggestions for further study ........................................................................... 223 Bibliography .............................................................................................................. 226 Appendix A Questionnaire ........................................................................................ 244 Appendix B Checklist ................................................................................................ 246 Appendix C Language information and tone data analysis from 42 Tai varieties..... 257 Appendix D The maps of the Tai varieties investigated ............................................ 423 Resume....................................................................................................................... 425 ix LIST OF TABLES Table 1 Li Fang-Kuei’s Tai tone pattern (summarized from Li 1977) ........................ 11 Table 2 Tai Tone correspondence between the Sinitic and Western Circles ............... 12 Table 3 Examples from tonal numbering presenting the contradiction to the “tonal category based principle” in the Sinitic circle ............................................................. 13 Table 4 Tai Tone box (adapted from Gedney 1989[1972]: 202-204) ......................... 14 Table 5 Gedney’s Tone box and Sinitic tone numbers together applied to Thai ......... 15 Table 6 Tonal correspondence among Li’s pattern, Sinitic method, and Gedney’s tone box................................................................................................................................ 16 Table 7 A revised version of Gedney’s tone box (adapted from Liao & Shen 2012) . 18 Table 8 A tonal split in voiceless friction sounds in Tone A ....................................... 19 Table 9 Comparison of aspirations and voiced sounds among SWT, CT and NT in Column A ..................................................................................................................... 21 Table 10 Comparison of aspirations and voiced sounds among SWT, CT and NT in Column C ..................................................................................................................... 21 Table 11 Two new rows splitting from primary aspirations (adapted from Liao & Shen 2012) ................................................................................................................... 23 Table 12 Comparison of unaspirated stops among Tai varieties in Column A (adapted from Liao & Shen 2012) .............................................................................................. 24 Table 13 Comparison of unaspirated stops among Tai varieties in Column DL (adapted from Liao & Shen 2012) ............................................................................... 24 Table 14 Two new rows splitting from Row 2 in Gedney’s tone box (adapted from Liao & Shen 2012) ....................................................................................................... 25 Table 15 Some diagnostic lexical items of two groups in Tai (adapted from Zhang et al. 1999: 9-10) .............................................................................................................. 42 Table 16 Vietnamese tonogenesis (summarized from Haudricourt 1954a)................. 47 Table 17 MC tones and their OC source (adapted from Norman 1988: 57)................ 51 Table 18 Correspondences between PT and MC tones (adapted from Pittayawat Pittayaporn 2009: 240) ................................................................................................. 52 Table 19 Phonetic characteristics of PT tones (Pittayawat Pittayaporn 2009: 271) .... 56 Table 20 Vietnamese tones (adapted from Norman 1988: 55) .................................... 60 Table 21 The MC eight tonal categories (adapted from Norman 1988: 54) ................ 60 Table 22 Register tonal splits of Thai (adapted from Gedney 1989[1972]: 201) ........ 63 Table 23 Three-way register split in A tone in Kam (adapted from Haudricourt 1972: 68) ................................................................................................................................ 64 Table 24 Three-way register splits in Kam (adapted from Haudricourt 1972: 70) ..... 65 Table 25 Tonal correspondence of nasals between two groups of Kam-Sui languages (adapted from Weera Ostapirat 1994: 79) ................................................................... 72 Table 26 Four groups of Kam-Sui nasals proposed by Weera Ostapirat (1994: 80) ... 73 Table 27 Voicing alternation among Cao Lan and NT and CT representative varieties (Gregerson & Edmondson 1989: 160) ......................................................................... 77 Table 28 Voicing alternation among Nung An and the representative varieties of NT and CT (adapted from Edmondson 2002: 59).............................................................. 77 Table 29 Disagreement on the phonetic characteristics of PT tones *B and *C ......... 93 x Table 30 Tonal correspondence between Cantonese and Debao Urban Yang Zhuang (DB) ............................................................................................................................. 97 Table 31 Phonetic shapes of tones and their voice qualities in Tai varieties investigated .................................................................................................................. 99 Table 32 The correspondence among the voice qualities defined by different proposals .................................................................................................................... 103 Table 33 Basic tonal patterns in Huishui Bouyei (NT) .............................................. 107 Table 34 Basic tonal patterns in Debao Dongling Zhuang (NT) ............................... 108 Table 35 Some shared vocabularies with the B tones (Tones 5 and 6) in Dai and with syllable final -h in Wa (adapted from Liang & Zhang 1996: 816) ............................ 113 Table 36 Basic tonal patterns in Debao Urban Yang Zhuang (CT) ........................... 118 Table 37 Phonetic characteristics of PT tones in the two diachronic stages ............. 120 Table 38 Different speculations of initial phonation types at the PT level ................ 122 Table 39 The speculated primary register tonal split on post-PT level ..................... 130 Table 40 Type 1 of secondary tonal splits in the Zhuang dialects (adapted from Zhang et al. 1999: 243-244) .................................................................................................. 133 Table 41 Type 2 of secondary tonal splits in the Zhuang dialects (adapted from Zhang et al. 1999: 244) ......................................................................................................... 134 Table 42 Type 3 of secondary tonal splits in the Zhuang dialects (adapted from Zhang et al. 1999: 244-245) .................................................................................................. 135 Table 43 Type 4 of secondary tonal splits in the Zhuang dialects (adapted from Zhang et al. 1999: 245) ......................................................................................................... 136 Table 44 Series 2/1 of voicing alternation in Tai ....................................................... 142 Table 45 Correspondence involving labial clusters by monosyllabic and sesquisyllabic views (adapted from Pittayawat Pittayaporn 2009: 48-49)........................................ 143 Table 46 Etyma with PT *C̬.t- (Pittayawat Pittayaporn 2009: 168) .......................... 145 Table 47 Etyma with PKS *h'-w- (Weera Ostapirat 2006: 1085) ............................. 146 Table 48 Etyma with the proposed PT *ɬ.w- which causing voicing alternation in Tai .................................................................................................................................... 148 Table 49 Developing tracks of PT *ɬ.w- in modern Tai varieties .............................. 149 Table 50 Reflexes of *ʔd-, *ʔb-, and *ʔj- and their tones .......................................... 152 Table 51 PT monosyllabic onsets induced by aspiration in ST and YN ................... 155 Table 52 Reflexes of Li’s*kr- in some modern Tai varieties .................................... 158 Table 53 Etyma with PT *k.r- (adapted from Pittayawat Pittayaporn 2009: 169, 345) .................................................................................................................................... 160 Table 54 The reconstruction of PT ASO-CY ............................................................ 161 Table 55 Etyma with ASO-YN .................................................................................. 162 Table 56 Reflexes of aspiration involving in different modern Tai groups ............... 165 Table 57 Voicing alternation between Suogan of NT and other Tai varieties........... 166 Table 58 Voicing alternation in Tai and KS languages ............................................. 167 Table 59 Developing tracks of PT *C̥.N- in modern Tai varieties ............................ 168 Table 60 Basic tonal patterns in Daxin Baoxu Zuojiang Zhuang (cf. L23) ............... 170 Table 61 Basic tonal patterns in Huashan Min Zhuang (cf. L16) .............................. 172 Table 62 Tonal splits in checked syllables in Guangnan Nong Zhuang .................... 174 Table 63 Tonal splits in checked syllables in Debao Dalong Yang Zhuang ............. 175 Table 64 Basic tonal patterns in Debao Dalong Yang Zhuang (cf. L9) .................... 178 Table 65 Basic tonal patterns in Jingxi Anning Yang Zhuang (cf. L13) ................... 178 Table 66 Vowel-length alternation between ST and NT/YN .................................... 179 Table 67 The progressing vowel-length alternation between ST and NT/YN .......... 181 xi Table 68 Different initial behaviors of the single r- segment among NT, YN, CT and SWT ........................................................................................................................... 193 Table 69 Vowels involving Gedney’s Puzzles .......................................................... 194 Table 70 Some diagnostic items of NZ and SZ division (adapted from Edmondson 1994: 152-153) ........................................................................................................... 195 Table 71 Some diagnostic lexical items of ST and NT-YN (adapted from Zhang et al. 1999: 9-10, also cf. Table 15 in §2.1.3) ..................................................................... 196 Table 72 PT tone box: from tonal precursor to tones ................................................ 202 Table 73 Integrated PST & PNT tone box: from PT to PST (→) and PNT (→)....... 203 Table 74 PST tone box............................................................................................... 204 Table 75 SWT tone box (cf. Gedney 1989[1972]) .................................................... 205 Table 76 Ancillary box of the SWT tone box ............................................................ 205 Table 77 CT tone box ................................................................................................ 206 Table 78 Ancillary box of the CT tone box ............................................................... 206 Table 79 Different VASO-1/2 merging directions involving different tonal splits among Tai varieties .................................................................................................... 207 Table 80 PNT tone box .............................................................................................. 208 Table 81 NT tone box ................................................................................................ 209 Table 82 Ancillary box of the NT tone box ............................................................... 209 Table 83 YN tone box ................................................................................................ 210 Table 84 Ancillary box of the YN tone box .............................................................. 210 Table 85 Integrated Tai tone box ............................................................................... 211 Table 86 Ancillary box of the integrated Tai tone box .............................................. 211 Table 87 Integrated Tai tone box applying to Debao Dalong (cf. Table 64) ............. 213 Table 88 The merging direction of particular series in Debao Dalong...................... 213 Table 89 Examples of merging of initials and vowels in Debao Dalong .................. 214 xii LIST OF FIGURES Figure 1 Tai-Kadai language family tree (adapted from Diller 2008: 7) ....................... 2 Figure 2 Distribution of the Tai–Kadai language family (from wikipedia.org) ............ 3 Figure 3 Tripartite division of Tai languages (Li 1977) ................................................ 4 Figure 4 Two plus one taxonomy of Tai division (Gedney 1989b) ............................... 4 Figure 5 Dong-Tai (TK) family tree (adapted from Zhang et al. 1999: 3) .................. 39 Figure 6 Chao’s (1930) tone letter system ................................................................... 45 Figure 7 Voice qualities in PT tone summarized from Gedney (1989b) and Liang & Zhang (1996) ................................................................................................................ 55 Figure 8 PT tone features suggested by Sagart (1988: 89) .......................................... 55 Figure 9 Chinese tonal development scheme (summarized from Norman 1988: 52 ff.) ...................................................................................................................................... 62 Figure 10 Hengxian checked tones (Bi 1982: 21 cited in Fu 1995: 83) ...................... 66 Figure 11 Type of tonal mergers (Pan 1982) ............................................................... 68 Figure 12 Classification of SWT dialects based on tonal splits (adapted from Chamberlain 1975: 50) ................................................................................................ 75 Figure 13 Voice analysis diagram of ˀba:nC1ʔ ‘village’ in Debao Dalong by PRAAT .................................................................................................................................... 104 Figure 14 Sound analysis diagram of tʰa:n332 ‘charcoal’ and pʰa:332 ‘to chop’ in Debao Dalong Yang Zhuang (CT) by PRAAT ..................................................................... 115 Figure 15 Speculation of shared innovations on different diachronic orders for subgrouping languages............................................................................................... 185 Figure 16 Tai diachronic division based on tonology ................................................ 198 Figure 17 Tai family tree on the primary and secondary levels ................................ 198 xiii LIST OF ABBREVIATIONS AND SYMBOLS 1-A Aspirated sounds in the high register 1-C Continuant sounds in the high register 1-G Glottal sounds in the high register 1-U Unaspirated stops in the high register 1-UC Voiceless unaspirated stop + *r clusters 2 Voiced sounds in the low register 1/2 Alternation of tonal series between Northern Tai and Southern Tai AMO-ST Aspirated induced by monosyllabic onsets in Southern Tai AMO-SY Aspirated induced by monosyllabic onsets in Southern Tai and Yongnan ASO-CY Aspiration induced by sesquisyllabic onsets in Central Tai and Yongnan ASO-SY Aspiration induced by sesquisyllabic onsets in Southern Tai and Yongnan ASO-ST Aspiration induced by sesquisyllabic onsets in Southern Tai ASO-YN Aspiration induced by sesquisyllabic onsets in Yongnan CT Central Tai GB Guibei Zhuang KS Kam-Sui language branch L language MC Middle Chinese MSEA Mainland Southeast Asia xiv NT Northern Tai NZ Northern Zhuang OC Old Chinese PT Proto-Tai ST Southern Tai SWT Southwestern Tai SZ Southern Zhuang T Tone TK Tai-Kadai language family YN Yongnan Zhuang VASO-1/2 Series 1/2 of voicing alternation induced by sesquisyllabic onsets VASO-2/1 Series 2/1 of voicing alternation induced by sesquisyllabic onsets VASO-GB Voicing alternation induced by sesquisyllabic onsets in Guibei Zhuang xv GLOSSARY Diachronic process A linguistic process or evolution of a language considered to have undergone the development through history. Historical linguistics is typically a diachronic study (Saussure 1983). Gedney’s tone box The Tai tone box designed by Gedney (1989[1972]: 202) consists of five tone categories (A, B, C, DL, DS) crossed with four groups of initial consonants (Group 1=Voiceless friction, Group 2=Voiceless unaspirated, Group 3=Glottal, and Group 4=Voiced.) conditioning possible tone splits. Macrolanguage “Multiple, closely related individual languages that are deemed in some usage contexts to be a single language.” “In addition to defining three-letter codes for individual languages, the ISO 639-3 standard also defines codes for macrolanguages. Macrolanguages were introduced into the standard in order to handle cases in which varieties would be considered distinct languages by the criterion of nonintellgibility as described above, but had already been given a code as a single language by the previously existing ISO 6392 standard. For instance, Arabic [ara] and Chinese [zho] were already defined in ISO 639-2 on the basis of literature shared across many spoken varieties (and a shared writing system in the case of Chinese)” (https://www.ethnologue.com/about/problem-languageidentification). In the case of Tai, Zhuang (ISO 639-3:zha) is a macrolanguage including at least sixteen single languages in China (https://www.ethnologue.com/language/zha). xvi Middle Chinese From the language of the 7th century AD rhyme dictionary, the Qìyùn, to the language of the Song dynasty rhyme tables (the mid 12th-century Yùnjīng) (Norman 1988: 23-24). Monosyllable It includes syllables with single initial consonants or tautosyllabic initial clusters (like pl-, tr-, and gr-) (summarized from Pittayawat Pittayaporn 2009). Old Chinese The language of the Shījīng (The Book of Poetry), c. 1000 BC (Norman 1988: 23). Proto “Existing or coming before other things of the same type” (Pearson Education Asia Limited & Foreign Language Teaching and Research Press 2009: 1834). It is normally seen as a combining form meaning “first,” “foremost,” “earliest form of” used in the formation of compound words. A protolanguage refers to an ancient language (for example, ProtoIndo-European) diversifies and develops to several daughter languages (such as Proto-Germanic, Proto-Celtic and so on, in the case of Indo-European) (Campbell 2004: 187). In the case of Tai-Kadai language family, proto-Tai, proto-initials, proto-vowels and proto-tones, specialized in Tai linguistics to denote the first stage of the Tai parent language, and the initials, vowels and tones presenting in the Tai parent language. Proto-Tai Diller (1998:14) points out that most linguists working in the field accept a time depth of less than two thousand years for Li’s Proto-Tai. Register A register or pitch register is a prosodic feature of syllables in certain languages, in which tone, vowel phonation, glottalization, or similar features depend upon each other. In tonology, register means the subdivisions of the tone space (Fu 1995: 30-31). xvii Sesquisyllable “Literally ‘one-and-a-half syllable’, was originally coined by Matisoff (1993) to refer to words that consist of an unstressed ‘minor’ syllable followed by a fully stressed ‘major’ syllable. Crucially, the minor syllable is characterized as lacking a contrastive vowel, e.g. Palaung /k.tɛ/ ‘earth’, and /kr̩.taʔ/ ‘tongue’ ” (Pittayawat Pittayaporn 2009: 41). In Pittayaporn’s definition, sesquisyllable in Proto-Tai refers to “a prosodic word that consist of two syllables, the first of which is unstressed and lacks a phonological vowel”, and “the consonants in a sequence at the beginning of a sesquisyllable are not all linked to a single syllable node” (Pittayawat Pittayaporn 2009: 42). Tautosyllable Syllable structure presents as that the consonants in a sequence at the beginning of a tautosyllable are all linked to a single syllable node (Pittayawat Pittayaporn 2009: 42), such as syllables with initial clusters pl-, tr-, and gr-. Tonal development A theory of how tones develop in a specific language, such as tonogenesis, tonal splits, tonal mergers, and etc (summarized from Fu 1995). Tonal flip-flop An explanation of the voiced-high phenomenon, which refers to that in some specific tonal languages, the proto-voiced initial consonant group conditions high pitch tones, while proto-voiceless initial consonant group conditions low pitch tones. It assumes that “all register-splits start as voiced-low, and the voiced-high correlation is caused by a later development whereby high tones become low and low tones become high” (Fu 1995: 82). Tonal split A tone separated into different tones because of the effect of secondary changes in initials or vowel lengths (Fu 1995: 69). Tone category The category or class of different tones in a language. Words having the same tone category in a language have the same xviii tone value, but the same tone category in different varieties of the same language family does not always have the same tone value. This concept has been used in the tones of Chinese poetry and dialectology (四聲), which have been called four traditional tone categories of Chinese words for a long time (You 2004: 44). Tone value The tone (or value) in a language is a way of assessing how high or low or what contour it is (You 2004: 44). Toneme The smallest unit of tone which represents meaningful contrast (Yip 2002). Tonogenesis A hypothesis of how tones emerged in the history, such as tone arises as an effect of the loss or merger of final and initial consonants (Matisoff 1973). Tonology The science of tones or of speech intonations, proceeding historically and comparatively (Matisoff 1994). xix Introduction This study aims to analyze the tonal development in three modern sub-branches of Tai languages distributed in Southern China and some countries in Mainland Southeast Asia, to apply insights from the analysis of tonal development to the classification of the Tai languages, and to propose a revised tone box which may be more applicable for analyzing maximal Tai varieties. The descriptions are primarily based upon tone data from seven varieties from Southwestern Tai (hereafter SWT), sixteen varieties from Central Tai (hereafter CT), fourteen varieties from Northern Tai (hereafter NT), and five varieties of Yongnan Zhuang (hereafter YN) whose classification as CT or NT is open to debate1. The following sections provide some background information about the research topics and the rationale of the thesis, as well as the motivation for the thesis research subsequently presented. Section 1.1 introduces and discusses the background of Tai languages, Tai classifications, the definition of Tai languages and varieties, and Tai tones. Section 1.2 firstly discusses the main descriptions and some puzzles of Tai tones, and then lays out the objectives and goals of the research. Section 1.3 gives an overview of the study, including the scope, the limitations, the benefits and a description of the chapters in this thesis. Tai speaking groups are mainly found in the Guangxi Zhuang Autonomous Region of China, Northern Vietnam, Laos, and Thailand in Mainland Southeast Asia (hereafter MSEA), as far north as Guizhou Province of China, as far south as halfway down the Malay Peninsula, as far east as Guangdong Province of China, and as far west as Assam and the surrounding areas of India. With the most numerous groups, such as the Thai of Thailand with a population of 55 million, the Zhuang of Guangxi and Yunnan in China with 15 million, the Shan of Myanmar with 2.8 million, and the Bouyei of Guizhou in China and Vietnam with 2.5 million, “the total number of Tai speakers is greater than 80 million” (Edmondson & Solnit 1997: 1). 1 These Tai varieties and their numbering are arranged as in Appendix C. For example, the first language is Bangkok Thai and is labeled as Language 1 with its abbreviated number L1. Others are similarly hereinafter arranged as L2 (Songkhla Southern Thai) to L42 (Zhenning Bouyei). 1 Just as its name implies, Tai speaking groups generally speak Tai languages as their first language. The Tai language branch, which is also called the Zhuang-Dai branch ( 傣語支) in China, is grouped along with other related language branches like Kam-Sui (or Dong-Shui in the Sinitic circle), Hlai (or Li in the Sinitic circle), and Kra (or Ge-Yang in the Sinitic circle) to form a language family termed Tai-Kadai (hereafter TK) used by most Tai linguists, or other names, as Daic in Shafer (1940: 302-337), and as Kra-Dai in Pittayawat Pittayaporn (2009a: 2-5). Some linguists have different opinions of grouping branches of TK on different levels of the family tree diagram. A typical example of the classification of TK language family proposed by Diller (2008: 7) is shown in Figure 1. In this proposal, Kam-Sui and Tai are grouped as sisters of Lakjia to form a branch named Kam-Tai of which the Hlai and Kra branches are sisters, since Kam-Sui and Tai share much more genetic characteristics on cognates, phonology, tonology and grammar. There is a controversial hypothesis mainly in China that proposes all TK languages to form a language group under Sino-Tibetan language family, while there is another disputable opinion named Austro-Tai language family standing that TK is a subgroup of which Austronesian is a sister proposed by Benedict (1942) prevailing in Western circle (Zhang et al. 1999: 2-3). Despite different opinions of classifying the whole TK language family, most linguists have an agreement that certain varieties are grouped together to form a particular group as Tai, which is different from other groups. Tai-Kadai Language Family Kam-Tai (Zhuang-Dong) Kam-Sui Lakjia Hlai (Li) Kra (Geyang) Tai (Zhuang-Tai) Gelao Be Lachi Kam Laha Sui Buyang … NT CT SWT … Figure 1 Tai-Kadai language family tree (adapted from Diller 2008: 7) The geographical distributions of TK languages are shown in the following Figure 2, and Li’s (1977) three Tai sub-branches (Northern Tai in yellow, Central Tai in red, and Southwestern Tai in orange) shows a big vertical Y shape upper left to Guangxi 2 and Guizhou provinces of China, upper right to Northern Burma and Northeastern India, and down to Southern Thailand. Figure 2 Distribution of the Tai–Kadai language family (from wikipedia.org) Basically, there are two main positions about the primary classification of the Tai branch. They are the two plus one taxonomy proposed by Haudricourt (1956) and Gedney (1989) and the tripartite division of Li Fang-Kuei (1977). Li’s scheme of three branches, SWT, CT, and NT, has long been accepted as the standard one in the field of comparative Tai linguistics. 3 Tai Southwestern Tai Central Tai Northern Tai Figure 3 Tripartite division of Tai languages (Li 1977) However, Haudricourt (1956) insisted on a two-term system, dividing the proto-Tai (PT) first into Dioi or ʔyai (corresponding approximately but not exactly to NT) vs. Tai proper. Gedney (1989b) also tends to support this hypothesis, and proposes that Li’s CT sub-branch and SWT sub-branch can form a sub-group of which the NT subbranch is a sister. Gedney also admits that there are basic differences between the CT and SWT. He has proposed to divide these two sub-branches by lexical and phonological criteria, but proposes that the scale of these differences is different from those between NT and other Tai languages, from the evidence of some intermediate dialects between SWT and CT (Gedney 1989b: 65-66). Tai Northern Tai Central Tai Southwestern Tai Figure 4 Two plus one taxonomy of Tai division (Gedney 1989b) The argument of the classifications of the Tai languages is an important topic in the discussions subsequently presented because it is related to one of the objectives of this thesis. The analysis of tonal development is crucial to the explanation of the insights to the classification of the Tai languages. The two main primary classifications of the Tai branch aforementioned will be tested by the consequence of this study. There are many languages and dialects within three Tai sub-branches, such as Central Thai, Northern Thai, Southern Thai and Northeastern Thai in SWT, Southern Zhuang (hereafter SZ), Nung, and Tay in CT, and Northern Zhuang (hereafter NZ) and Bouyei in NT discussed in previous research materials, but actually the criteria of defining these “languages” and “dialects” are inconsistent and complicated. For 4 example, in Chinese publications, the word “the Zhuang language” is always assumed to denote an individual language. This “language” is first divided into Northern and Southern two big dialects, and NZ and SZ are furthermore divided into eight sub-dialects and five sub-dialects respectively (Zhang et al. 1999: 12-13). But just as in Edmondson (1994: 150), it has been pointed out that the SZ is included in Li’s (1977) CT sub-branch, while NZ is included in Li’s NT sub-group, “thus, in some sense the Zhuang speak at least two languages, one belonging to Northern Tai and the other to Central Tai” (Edmondson 1994: 150). Despite Edmondson’s definition of two languages of Zhuang, many vernaculars and varieties inside NZ and SZ are actually not intelligible with each other at all, and they are treated as “dialects” or “sub-dialects” of Zhuang, or even different “languages” of the Tai family in different materials. In this thesis, the definition of a “language” only follows the principles in part 3 of ISO 639, which is a code that aims to define three-letter identifiers for all known human languages. The standard of ISO 639-3 was published by ISO in 2007, and it attempts to provide codes for the representation of names of languages “as complete an enumeration of languages as possible, including living, extinct, ancient, and constructed languages, whether major or minor, written or unwritten” (http://www01.sil.org/iso639-3/). As a result, ISO 639-3 deals with a very large number of lesser-known languages. In this part of ISO 639, most identifiers are assumed to denote distinct individual languages. For instance, ISO 639-3 contains 16 identifiers designated as individual language identifiers for distinct varieties of Zhuang by 2013, while ISO 639-1 and ISO 639-2 each contain only one identifier for Zhuang (including NZ and SZ which belong to NT and CT respectively), “za” and “zha” respectively, which are designated as individual language identifiers in those parts of ISO 639. It is assumed in ISO 639-3 that the single identifiers for Zhuang in parts 1 and 2 of ISO 639 correspond to the many identifiers collectively for distinct varieties of Zhuang. Just as in the treatment of ISO 639-3, these varieties such as Yang Zhuang, Nong Zhuang, and Central Hongshuihe Zhuang are treated as individual languages in this thesis. On the other hand, Zhuang or NZ/SZ is treated as a macrolanguage with ISO 639-3 code “zha” which contains at least sixteen individual languages closely related (https://www.ethnologue.com/language/zha). The necessity of using ISO 639-3 code for defining a language in this thesis can be illustrated by the following aspects. 5 First, a language discussed in different sources may have different names, and it is necessary for this thesis to use a representative name in a consistent way. For example, “Central Thai” refers to the language including Tai varieties spoken in the Central areas of Thailand, and “Standard Thai” is the standardized variety of this language. This language has been more exclusively named “Siamese” in many sources such as Li (1977). Some sources also follow what Thai people call this language as Phasa Klang, which means “Central Language” in Thai. For uniformly addressing this language which has served different names mentioned above, this thesis follows its formal name in ISO 639-3 code to use only the name “Thai” when addressing the entirety of this language, except when citing other names of it from other sources. In a similar way, in this thesis “Northern Thai” represents the entirety of the language having other names such as “Phasa Nuea”, “Lanna”, “Kam Mueang”, and “Thai Yuan”, “Isan” represents the language having the other name “Northeastern Thai”, and “Southern Thai” represents the language with the other names “Pak Tai” (Thai: ภาษาปักษ์฿ตຌ) and “Dambro” (Thai: ภาษาตาม฾พร) (https://en.wikipedia.org/wiki/Southern_Thai_language). Second, it is required to have a consistent definition of a language when discussing “dialects”, and “sub-dialects” under a specific language. The linguistic varieties denoted by each of the identifiers in ISO 639-3 are presumed to be distinct languages and not dialects of other languages, even though for some purposes some users may consider a variety listed in ISO 639-3 to be a “dialect” rather than a “language”. If we follow the standard of ISO-639-3, the term “dialect” is used as in the field of linguistics where it simply identifies any sub-variety of a language that might be based on geographic region, age, gender, social class, time period, or the like, which are all included within the denotation represented by the identifier for that language. Thus, Yang Zhuang with the code “zyg” in ISO 639-3 represents the complete range of all the spoken or written varieties of this language, including Jingxi County Yang Zhuang, and Debao County Yang Zhuang, which are treated as dialects of Yang Zhuang, and the Iang-Noengz orthography recommended to be a unified writing system as proposed by Liao (2010) for all varieties of Yang Zhuang. Additionally, Debao Urban, Debao Suburb, Debao Ma’ai, and Jingxi Hurun varieties are treated as sub-dialects of the Debao County Yang Zhuang Dialect, as they share more phonological similarities than those varieties of Jingxi County Yang Zhuang Dialect, but differ from each other in tonal splits. Also, consistently using terms like “languages” and “dialects” can guide readers to understand what the term “variety” refers to in this thesis. Compared with the 6 specific scope of languages and dialects, there is not an obvious boundary to define a variety of a language or a language family. As a general concept of a variation of a language, a variety can address the entirety of a language under a language group or family, a dialect of a language, a sub-dialect of a dialect, or a specific geographical point investigated in a linguistic survey. For example, we can say Yang Zhuang is a variety of CT, Debao County Yang Zhuang is a variety of Yang Zhuang, and Debao Urban is a variety of Debao County Yang Zhuang. In this thesis, only “variety” but not “language” or “dialect” is used to address all linguistic points investigated for specifically illustrating the circumstances of tonal development. For example, when illustrating the tonal development of Thai Language (ISO 639-3: THA), Bangkok is chosen to be a specific “variety” to represent that language. For another example, Debao Urban is registered to be one of the specific varieties of Debao County Yang Zhuang. In summary, languages and dialects are defined by following the standard of ISO 639-3 code in this thesis, and all specific linguistic points investigated or chosen for analyzing tonal development are treated as varieties of Tai languages. It is generally known that all Tai languages are tonal just like the majorities of the languages in East Asia and MSEA. Tones in these languages are the use of pitch and other laryngeal features to distinguish lexical or grammatical meaning. A distinct tone normally has a fixed pitch, or has similar pitch on syllables with complementary distribution. A tone in a language is a way of assessing how high or low or what contour it is. In Chinese phonology, this kind of assessing the pitch of a tone is called tone value or tonal value (調值). The same tonal value on different syllables are normally classified to be of the same category, and this categorical behavior of tones is called tone category or tonal category (調類). These two terms are commonly used by linguistics working on tonal languages in the Sinitic circle. For example, in the discussion of the tonal development of the Chinese languages by Norman (1988: 52-57), there were four tonal categories píng, shǎng, qù, and rù in Middle Chinese, and they presented different tonal values like ‘level and non-abrupt’ for the tone of píng recorded in the Chinese dictionary Qièyùn compiled in AD 601 by Lù Fǎyán (Norman 1988: 24). Basically, the vast majority of Tai languages have five to six tones on live syllables ending in a continuant (vowel, semivowel, or nasal), or smooth syllables used by Gedney (1989[1972]: 192), and two to four tones on dead syllables ending with a 7 stop (-p, -t, -k, or -ʔ), or checked syllables used by Gedney (1989[1972]: 192). Neither all five to six tones can be found in all the live syllables, nor two to four tones can be found in all the dead syllables, since tonal behaviors are conditioned by different features of initial consonants. That is to say, only a part of initials can collocate with all tones. For example, there are five distinct tones in Thai (L1) with different tonal values 33 (mid level), 21 (low-falling), 41ʔ (high-falling, with glottal constriction), 453ʔ (high-rising-falling, with glottal constriction), 24 (low-rising), and three of them (21, 41, and 453) can occur in dead syllables, while the tone 453 present as another similar tonal value 45 (high-rising) in this situation (Gedney 1989[1972]: 192)2. When a syllable begins with the initial m-, we can find all five tones on live syllables and three tones on dead syllables, as ma:33 ‘to come’, ma:21 ‘to soak’, ma:j4ʔ1 ‘widowed’, ma:453ʔ ‘horse’, ma:24 ‘dog’, ma:k21 ‘betel nut’, ma:k41 ‘very much’, and mak45 ‘often’. But when a syllable begins with the initial d-, normally we can only find three tones on live syllables, and one tone on dead syllables, as di:33 ‘good’, da:m41 ‘handle’, da:21 ‘to scold’, and dɔ:k21 ‘flower’3 . Different Tai varieties present different tonal behaviors and different types of tonal splits. For example, in Thai (L1) syllables with aspirated stops can have all five tones, while in Yang Zhuang varieties, aspirated stops can only have three tones of all six distinct tones, as the following examples in Thai: kʰa:33 ‘to stick’, kʰa:21 ‘galangal’, kʰa:41 ‘to kill’, kʰa:453 ‘trade’, and kʰa:24 ‘leg’ vs. the examples in the Debao Urban variety of Yang Zhuang (L8): kʰa:353 ‘leg’, kʰa:24 ‘to kill’, and kʰa:33 ‘galangal’, but words like kʰa:31, kʰa:213 , and kʰa:55 which have the other three tones of the all six tones can be never found in this variety, except some modern loans from Mandarin Chinese. The limitation of the collocation between initials and tones originates from tonal register splits triggered by initial voicing on post-PT level (cf. §4.2.2), such as the voicing of proto-voiceless continuant initials (like *m̥ - > *m-, *n̥- > *n-, and *l ̥- > The tonal values (in Chao’s 5-pitch scale) of Thai tones in Gedney’s original text are not provided, instead the described diction like “low level”, “falling, with glotttal constriction” and etc. are provided. There are some specific phonation types or phonetic shapes accompanying with some specific tones in many Tai varieties (cf. §4.1.3.3). For example, tones 41ʔ and 453ʔ in Thai (L1) and 24ʔ and 213ʔ in Debao Urban Yang Zhuang (L8) have a final glottal constriction, and tones 51 and 11 in Yizhou Suogan Guibei Zhuang (L36) have creakiness accompanying with these tones (cf. Appendix C). Because these phonation types or phonetic shapes are not phonemic or tonemic, hereafter they will be omitted if the discussion is not related to tonogenesis or other aspects which need to probe into these phonation types or phonetic shapes. 3 Some exceptions can be found in some informal terms of Thai. For example, I have heard that some Thai young people use the English word ‘dog food’ as d ɕ fu:t , whose the first syllable with the initial d- which never has a high rising tone (45) on dead syllable in Thai words. This kind of collocation is treated as later development and is not counted as an inherent tonal behavior in Thai. 2 8 *l-) and the devoicing of proto-voiced stops (like *b- > *p-, *d- > *t- and *g- > *k-) which caused the PT four tonal categories to split into two series (original high and low registers) for preserving semantic distinctions (Theraphan L-Thongkum 1997: 207). During the time of tonal splits, in some Tai varieties like Debao Urban Yang Zhuang (L8) some of the series of initial consonants like aspirated plosives (pʰ-, tʰ-, and kʰ-) only developed from proto-voiceless sounds which conditioned high register tones at times of tonal split. This has made items with aspirated initials only collocate with high register tones in Debao Urban Yang Zhuang. In contrast, in Thai aspirated sounds developed from both proto-voiceless and proto-voiced sounds which conditioned high and low register tones respectively, and this has made items with aspirated initials collocate with both high and low register tones in Thai. This is why all five tones can be found on items with aspirated initials in Thai but only three of the all six tones can be found on items with aspirated initials in Debao Urban Yang Zhuang. The complicated situation of tones in Tai differs from variety to variety, from dialect to dialect, and from group to group. This complexity is a result of different situations of tonal development in different Tai daughter languages. For expressly probing into the cause and the effect of this complexity, this thesis is going to study the objectives involving Tai tonal development in the following section. On the basis of the previous research of Tai tones and the Tai classification based on Tai tones, this thesis aims at clarifying or solving the following issues. 1) To analyze the tonal development in three modern sub-branches of Tai languages, especially those in Li’s CT and NT languages. 2) To apply insights from the analysis of tonal development to the classification of the Tai languages. 3) To propose a revised tone box which may be more applicable for analyzing maximal Tai varieties. These three objectives are related to each other by their associated characteristic, which is the study of Tai tones, or Tai tonology. There are a number of competing descriptions of Tai tones. In this section, I first outline three most wildly adopted system of describing Tai tones. They are Li Fang-Kuei’s pattern, the Sinitic pattern, and Gedney’s tone box. 9 In Tai linguistics, the treatments of counting the number of tones are different within different linguists in linguistic circles. For example, as in §1.1.4, normally Western linguists follow the Thai traditional method to count that Standard Thai has five distinct tones, which can all occur on live syllables. Those three tones in dead syllables are treated as allotones of those tones having the same or similar pitch and occurring with live syllables. That is why the tone in the word mak45 ‘often’ is seen as an allotone of the tone in the word ma:453 ‘horse’, since they are similar in pitch (of high tones) and have a complementary distribution – the former tone only occurs on dead syllables and the latter tone only occurs on live syllables. However, scholars in the Sinitic Tai linguistic circle, which include Liang and Zhang (1996) and Zhang and colleagues (1999), count that there are ten tones in Thai, since they treat tones on dead syllables as different tones from those on live syllables, and count tones developed from different origin (proto-tones) as divided tones even though some of them may have the same tone. For example, Liang and Zhang (1996: 962) list ten tones in Thai as follows (with “T” as the abbreviation of a sorted tone): T1 (24), T2 (33), T3 = T6 (41), T4 (453), T5 (22), T7 (22), T8 (55), T9 (22), and T10 (41). Within these tones, T1 – T6 are on live syllables, and T7 – T10 are on dead syllables. T3 and T6 are separately counted even they are the same pitch (high-falling 41) and are both on live syllables (and they are actually of the same tone counted in Thai traditional method and in Western circle), since they developed from different prototones, such as in Thai the tones of the words ฆ kʰa:⁴¹ ‘to kill’ (T3 developing from proto-tone C) and ค kʰa:⁴¹ ‘value’ (T6 developing from proto-tone B) are treated as two tones even they are actually homophonous. T10 is also separately counted from T3 and T6 although it has the same tonal value as T3 and T6, since it is on dead syllables. Tones 5, 7 and 9 are listed with the same tonal value (low level), but are separately counted since they are on live and dead syllables respectively, or collocate with different vowel length even though two of them are both on dead syllables. That is to say, scholars in the Sinitic circle normally define tone based on the tonal category principle, in contrast scholars in the Western circle normally define tone based on tonal value principle. In the situation of explicitly distinguishing the concept of “tone” in the Sinitic and Western circles, we use “contrasting tone” for referring to the later one. To further illustrate the problem, it is necessary to compare the different styles of numbering the tones of the Tai languages or dialects used by Li Fang-Kuei, scholars 10 in the Sinitic circle, and Gedney’s tone box. In this issue, the concepts of “prototone” and “tonal category” in Tai languages play an important role. When illustrating tones in PT or comparing tones among modern Tai languages, Li Fang-Kuei’s system of numbering the proto-Thai tones is most commonly used, and that convention will be followed in this thesis as well. It is generally believed that PT had a system of three tonal categories on smooth syllables, and a fourth tone on checked syllables. These tones are designated by Li Fang-Kuei as Tones *A *B *C and *D. Later these tones split into two series based on the voicing of the initial consonants, and Li designated 1 for those from a proto-voiceless initial and 2 for those from a proto-voiced initial. Furthermore *D was divided into DL for a dead syllable with a long vowel or cluster of two vowels, and DS for a checked syllable with short or simple vowel respectively, since vocalic length is crucial to explaining the varying developments in different dialects (Li 1977: 25-28). Thus, in modern Tai varieties, there are six tonal categories A1, A2, B1, B2, C1, and C2 on live syllables, and four tonal categories DL1, DL2, DS1, and DS2 on dead syllables. Li’s tone pattern which is shown in Table 1 has been widely used by linguists working on Tai languages in Western circle. Table 1 Li Fang-Kuei’s Tai tone pattern (summarized from Li 1977) Proto-Tone *A *B *C Voiceless A1 B1 Voiced A2 B2 Proto-Initial *D (*DL) (*DS) C1 DL1 DS1 C2 DL2 DS2 In this tone pattern, normally the same tone category presents the same tonal value in a specific Tai variety. For example, the ten tonal categories in Debao Urban variety of Yang Zhuang (L8) present as the following tonal values: A1=453 (highrising-falling), A2=31 (mid-falling), B1/DL1/DS1=55 (high level), B2/DL2/DS2 =33 (mid level), C1=24ʔ (low- rising with final glottal constriction), and C2 =213ʔ (low-falling-rising with final glottal constriction), as in the following examples: ma:453 (A1) ‘dog’, ma:31 (A2) ‘to come’, ma:55 (B1) ‘to soak’, ma:33 (B2) ‘to wait (a minute)’, ma:24ʔ (C1) ‘to grow up’, ma:213ˀ (C2) ‘horse’, ma:k55 (DL1) ‘fruit’, ma:k33 (DL2) ‘to be plenty’, mak55 (DS1) ‘beautiful’, and mak33 (DS2) ‘ink’. Some exceptions also occur along with further tonal splits and mergers in some modern Tai varieties, such as in Thai tonal categories B2 and C1 have merged to present a high-falling pitch 41 to become a single contrasting tone. 11 In a similar way, in the Sinitic circle scholars have become accustomed to describing tones of Kam-Tai languages using the same numbering system and tonal categories that they use to describe the Chinese languages, the Miao-Yao languages and Vietnamese. To that end, they employ both by cardinal numbers (1-10) and traditional Chinese terms (píng for A, shǎng for C, qù for B, rù for D, yīn for voiceless, and yáng for voiced) related to traditional Chinese phonology. So, viewed in Chinese terms related to “tonal category based principle”, there are ten tones, which include six tones occurring with live syllables and four tones occurring with dead syllables, in Tai languages. This is in contrast to the system used by linguists mostly seen in the Western circle, in which the tones that occur with dead syllables are actually seen as allotones of tones that occur with live syllables, on account of “tonal value based principle”. In all, when being viewed in the Western method, the Debao Urban variety of Yang Zhuang has six contrasting tones which present the following tone values: 453, 31, 24ʔ, 213ʔ, 55, and 33. In spite of the different viewpoints of how many tones there are in a Tai variety, there is a specific correspondence of Tai tonal categories between the Western circle and the Sinitic circle which is demonstrated in Table 2. Table 2 Tai Tone correspondence between the Sinitic and Western Circles Chinese yīn yáng yīn yáng yīn yáng yīn yáng yīn yáng Terms píng píng shǎng shǎng qù qù rù rù rù rù Sinitic 1 2 3 4 5 6 7 8 9 10 A1 A2 C1 C2 B1 B2 DS1 DS2 DL1 DL2 33 41ʔ 453ʔ 21 41 21 45 21 41 31 24ʔ 213ʔ 55; 33 55 33 Pattern Western or Li’s Pattern Thai tone 24; value 33 Debao Urban tone 453; 31 33 55; 33 33 value In the tonal categories shown in Table 2, all the odd numbers correspond to those tones with proto-voiceless initials and all the even numbers correspond to those with proto-voiced initials, both in the Western and Sinitic circles. In the Sinitic circle a tone with proto-voiceless initials is often called “odd tone” (單數調), while a tone 12 with proto-voiced initial is often called “even tone” (雙數調). It is in contrast to some previous works out of these two systems numbering the tones of a Tai variety by filling the cardinal numbers without a consistent order of odd or even. For example, the traditional Thai teaching methodology numbers 1 – 5 tones in Thai to correspond to Li’s tonal categories A2/A1, B1/DL1/DS1, B2/C1/DL2, C2/DS2, and A1 respectively, and as Hanna (2012: xiv) lists 1 – 6 tones in Dai Lue with the order corresponding to Li’s tonal categories A1, B1/DL1/DS1, C1, A2, B2/DL2/DS2, and C2 respectively. Despite the correspondence between the tonal description methods of the Western and Sinitic circles, they do not always present the tones in the same way when applied to different Tai varieties. That is, for presenting tonal splits in those Tai languages and dialects which exhibit differently from most common situation, a simple split conditioned by the voicing of the initial consonant giving, in many cases, six contrasting tones instead of the earlier three, the description of the tonal splits is more complicated and the consistent correspondence is lost. For example, in Table 3 Tone A1 in Thai presents tonal splits to two tone values 24 and 33, and the later one merges with the tonal value of Tone A2. Table 3 Examples from tonal numbering presenting the contradiction to the “tonal category based principle” in the Sinitic circle Gloss Li’s tonal categories Debao Ma’ai of Yang Zhuang Thai Tonal Sinitic tonal values numbering 24 Tonal values 53 Sinitic tonal numbering dog A1 ma: ma:¹ ma: ma:¹ leg A1 kʰa:24 kʰa:1 kʰa:31 kʰa:² year A1 pi:33 pi:² pej53 pej¹ leaf A1 33 baj² 31 A2 kʰa:n² ˀbɔj² pole kʰa:n33 ˀbɔj ka:n31 ka:n² have A2 mi:33 mi:² mej31 mej² baj This kind of tonal splits and mergers result in a problem of numbering the tones in the Sinitic circle. This is when numbering two tones being in contrast and splitting from the same proto-tonal category with proto-voiceless initial, if one of them merge with an even tone’s tonal value, it is always numbered to be the same tone as that even tone, even it does exactly not develop from a proto-tone with a proto-voiced initial and should be categorized to an odd tone due to its proto-tonal category. In Table 3, the examples of Thai and the Debao Ma’ai variety of Yang Zhuang are 13 adapted from Liang & Zhang (1996: 283, 461, 335, 244, 449, 308) and Zhang et al. (1999: 150-151) respectively. Both in Thai and Debao Ma’ai variety of Yang Zhuang tone category A1 splits into two contrasting tonal values, and one of them merges with Tone A2. The tonal numbering of the words ‘year’ and ‘leaf’ in Thai, and the words ‘leg’ and ‘leaf’ in Debao Ma’ai is T2, and is actually contradictory to the “tonal category based principle” in the Sinitic circle since they all developed from prototones with voiceless initials. To solve the problem, Gedney’s tone box (Gedney 1989[1972]) was used to explain the complicated situation of tonal splits. It adapts Li’s tonal numbering of Proto-Tai and further defined four groups of voiceless proto-initial consonants by their shared phonetic features at the time of the tonal splits. These four groups are voiceless friction, voiceless unaspirated, glottal, and voiced respectively, and condition possible tone splits among different Tai languages. Within these four groups of proto-initials, the former three consist of voiceless sounds in Li’s pattern. Gedney’s tone box which is shown as in Table 4, supplemented by more lexical items, has been widely used by linguists working on SWT languages. Table 4 Tai Tone box (adapted from Gedney 1989[1972]: 202-204) Proto-Tai Tones Initials at time of tonal splits Voiceless friction sounds, *s-, *hm-, *ph-, etc. Voiceless unaspirated stops, *p-, *t-, *k-, etc. Glottal, *ʔ-, *ʔb-, etc. Voiced, *b-, *m-, *l-, *z-, etc. A hu: ‘ear’; B C DL kʰa:t ‘torn’; DS kʰa: ‘leg’ pʰa: ‘to split’ kʰaj ‘egg’; kʰa: ‘to kill’; sɨa ‘shirt’ ŋɨak ‘gums’ pʰak ‘vegetable’ mat ‘flea’; pi: ‘year’; pa: ‘forest’; hu: ‘ear’; pɔ:t ‘lung’; kop ‘frog’; ta: ‘eye’ kaj ‘chicken’ tom ‘to boil’ tɔ:k ‘to pound’ bin ‘to fly’; da: ‘to scold’; ba: ‘shoulder’ ba: ‘crazy’; ba:n ‘village’ dɛ:t ‘sunshine’; bet ‘fishhook’; pʰɔ: ‘father’; na:m ‘water’; mi:t ‘knife’; nok ‘bird’; dɛ:ŋ ‘red’ mɨ: ‘hand’; na: ‘rice field’ raj ‘dry field’ ma:j ‘wood’ ʔa:p ‘to bathe’ lɨat ‘blood’ tap ‘liver’ ʔok ‘chest’ lak ‘to steal’ The designation of Gedney’s tone box more reasonably presents tonal splits in those Tai languages and dialects which exhibit the deviation of the most common situation. In these Tai varieties, a single proto-tone has in many cases split into two or even three tones on the basis of conditioning phonetic features of the initial consonants other than the simple voiced-voiceless contrast. For example, Table 5 plots the tones in Thai using Gedney’s tone box and presents the tone numbers of the 14 Sinitic circle method in the box. In some cases, formerly contrastive tones have merged into one tone, and some tones in dead syllables are actually allotones of those in live syllables. In these cases, we use an equal sign “=” between two Sinitic tone numbers for illustrating they are the same tone in the box. In all, Thai has five contrasting tones. Table 5 Gedney’s Tone box and Sinitic tone numbers together applied to Thai Proto-Tai Tones Initials at time of tonal splits A B C DL DS 1 5 3 9=5 7=5 2 5 3 9=5 7=5 Glottal 2 5 3 9=5 7=5 Voiced 2 6=3 4 10=3 8=4 Voiceless friction sounds Voiceless unaspirated stops In Table 5 we see the tonal splits in Thai are irregular between Tone A and other tone categories. The split in Tone A is between voiceless friction and other initial groups, which cross the dividing line of voiceless and voiced proto-initials, while the splits in Tones B, C, DL, and DS are between voiceless and voiced sounds. For the tonal split in A in Thai, a scholar in the Sinitic circle might use the number 2, an even number, which would normally correlate with a voiced initial (the fourth row), to represent the tones found with voiceless unaspirated stops and glottal sounds (the second and the third rows), since the tones on these two rows are the same as T2 in the voiced row. This practice is already shown as the numbering of the Thai words in Table 3. These tones, however, are actually in the rows, which correspond with voiceless proto-initials. On the other hand, Li (1977: 28-29) describes the split in Tone A in Thai with a more consistent explanation: A1 can be mid level (33) or lowrising (24), depends on the different initial features, and A2 is mid level (33). That is to say, in Li's system, a tonal category can have more than one tone value depending on the proto-initial. Compared to Sinitic tone numbers and Li’s tone pattern, Gedney’s tone box explains this kind of special case of tonal development more consistently. Its purpose is to display “a maximum of possible tonal distinctions resulting from the various types of tonal splits that have been described. In any given Tai dialect there will be a division of each column of the chart into two or three tones, or in some cases no such division at all in one column or another. Most dialects will also show coalescence or 15 syncretism between two or more boxes belonging to different columns” (Gedney 1989[1972]: 202). Furthermore, a curious feature of the tone box is that, when using it to plot the tonal splits for any of the many Tai varieties, the set order of the horizontal rows, from top to bottom, is always the same. This is true for all SWT varieties that have been described, and “one never finds the same new tone associated with, say, consonants of the first or second category, and also the fourth, skipping the third”. This is called a single ‘contiguous’ tone box or a fixed order phonological spectrum (Gedney 1985: 120). The tonal numbering correspondence among Li’s pattern, Sinitic circle method, and Gedney’s tone box is demonstrated in the following Table 6. Table 6 Tonal correspondence among Li’s pattern, Sinitic method, and Gedney’s tone box Traditional Chinese terms Li’s pattern Proto-Tai Tones Initials at time of tonal splits A B C DL DS 1 5 3 9 7 2 6 4 10 8 píng qù shǎng Voiceless friction sounds yīn 1 Voiceless unaspirated stops Glottal yáng 2 Voiced Traditional Chinese terms rù In practice, Gedney’s tone box (Gedney 1989[1972]) has been proved to be a good way for solving the problems of complicated tonal splits in SWT varieties. However, when applied to varieties of CT, such as Debao Zhuang (Yang Zhuang in ISO 639-3), Wenma Zhuang (Dai Zhuang in ISO 639-3), and Bac Va Nung, and most NT varieties, Gedney’s tone box has several shortcomings, the most noticeable being the so-called voicing alternation between NT and CT/SWT, and the splits among the voiceless friction sounds and the voiceless unaspirated stops. Some scholars, including Gedney himself, have also found there are some problems with using the tone box for explaining the tonal splits and mergers in some CT and NT varieties. For example, Gedney (1989[1972]: 204) himself has alluded to this issue in his seminal article detailing the famous tone box. He mentions that there are two languages that split Box 1 of Column A into two series; a variety of Nung (a CT language, which is called Bac Va as collected by Gedney (Hudak 2008: 37-39)) in 16 Northern Vietnam, and Saek in Thailand (a Northern Tai language). He further states, “It is possible that further research may show that our four horizontal tiers are not enough… When someday we fully understand the historical reasons for the tonal irregularities found in languages of the Northern Branch of the Tai family, we may find it necessary to make still further horizontal subdivisions in our chart” Gedney (1989[1972]: 204). Although he believes that it is inefficient to further refine the chart since the tone box is already more refined than necessary for the vast majority of Tai languages (Gedney 1989[1972]: 204), it has been pointed out that his tone box can only capture all the tonal distinctions in SWT languages, but cannot deal with the problems of complex tonal correspondence among SWT, CT, and NT varieties (Liao & Shen 2012). As a suggestion for solving these problems, Liao and Shen (2012) have provided below in Table 7 a revised and expanded version of the Tai tone box which can be used beyond SWT. In their chart, every box in the proto-voiceless sounds category is given a fixed name based on Li’s tone designations (A1, B1, C1, DL1, DS1, and etc.) plus a letter corresponding to the first letter of the descriptive word accompanying “voiceless” in the proto-voiceless rows. For example, A1-A would represent the tone for a word which had a voiceless initial consonant in PT (1) and that voiceless initial consonant was also Aspirated (A). A2, B2, C1, DL1, and DS2 are used for the protovoiced Row, and A1/2, B1/2, C1/2, DL1/2, and DS1/2 are used for the protobreathy Row4. This revised tone box is supplemented by more lexical items commonly found in most languages from the three branches of Tai. Some etyma have different semantic changes among different languages. For example, tʰa:B2 ‘wharf’ in Bongkok Thai and ta:B2 ‘river’ in Debao Dalong Yang Zhuang both develop from PT *da:B, but have different and related meanings which are commonly found in SWT and CT/NT/YN respectively. In such situation, an different semantic option is bracketed after the main lexical definition which is more commonly found in more Tai language varieties, such as ‘river (wharf)’ in Box B2 in Table 7. 4 For clearly illustrating the initial changes at time of tonal split, in Table 7 I add some of the examples of common initial forms from modern Tai varieties into their box. Some forms of initial change in some Tai varieties are shown in a blacket, such as pʰ- (f-) in Aspirations referring to the aspirated initial consonant pʰ- in CT/YN varieties which normally presents as f- in SWT/NT varieties. For instance, the word for ‘rain’ is pʰ :nA1 in Debao Dalong Yang Zhuang (CT), pʰonA1 in Nanning Shuangding (YN), fonA1 in Bangkok Thai, and fɯnA1 in Wuming Shuangqiao Yongbei Zhuang (NT). Examples of initials in their so-called proto-voiced aspirations & breathiness are separated by a slash “/” to show different forms (normally aspirated in SWT/CT but unaspirated in NT) which cause voicing alternation, like the intial pair pʰ-/p- in CT/SWT pʰowC1/pʰu:C1 vs. YN/NT powC2/pu:C2 ‘person (CLF for person)’. 17 Table 7 A revised version of Gedney’s tone box (adapted from Liao & Shen 2012) Initial Tone categories Groups at Time of A B C DL DS Tonal Split Aspirations: A1-A B1-A C1-A DL1-A DS1-A pʰ- (f-), tʰ-, kʰ-, leg; to chop; cloth torn; carry vegetable; mountain egg; (sheet); to on poles; hailstone; to (cliff); to charcoal wait; to kill astringent shut pʰj-, kʰj-, h-, ɕ- … add Continuants: A1-C B1-C C1-C DL1-C DS1-C n-, m-, ŋ-, w- dog; thick; to soak; face; grass; fruits flea; heavy; three tired; to sour (areca); color (tattoo) (ʋ-), l-, j-, s- (ɬ-) Initials from ProtoVoiceless … Unaspirated Stops: p-, t-, k-, tɕ-, send A1-U B1-U deaf; pestle C1-U DL1-U DS1-U fish; door; to blow; aunt; mouth; cliff duck; liver; crow chicken; seedling; (waterfall); frog low establish mustard pl- (pj-), kl(kj-) … Unaspirated Stop + *r Clusters: A1-UC B1-UC C1-UC DL1-UC DS1-UC eye; die; to hunt near; CLF to expose; grasshopper; for stone broken; to seed put up bamboo *pr-, *tr-, *kr- strips Glottal Sounds: A1-G B1-G C1-G DL1-G DS1-G ˀb- (b-, m-, w-), to fly; shoulder; village; to scar (time); to pinch; to star; to scold; at get; hot extinguish; sugarcane (boiling); to one (above want tens digit) ˀd- (d-, n-, l-), ˀj- (j-, ð-), ˀw- medicine (w-, b-), ʔ- (hungry) Initials from Proto-voiced Sounds: p- (pʰ-), t- (tʰ-), k- (kʰ-), tɕ- (tɕʰ-), pj- (pʰl-), kj- (kʰl-), A2 B2 C2 DL2 DS2 ricefield; river crutch; to leech; rope; tired (rest); hand; (wharf); chew; blood bird; ant buffalo father horse (male); n-, m-, ŋ-, l-, w-, j-, s- (ɬ-), f-, h- dry … farmland Initials from Proto-voiced A1/2 B1/2 C1/2 rice; DL1/2 DS1/2 Aspirations & Breathiness: ear; to bean; to person to hit the to bite; ten; arrive; ride on; to (CLF of mark cooked (ripe) rightside pull down person); (correct) pʰ-/p-, f- (pʰ-)/p-, tʰ-/t-, kʰ-/k-, kʰ-/h-, h- (kʰj-)/r-, s- (ɬ-, θ-)/ ɕ-… bowl 18 For illustrating this revised Tai tone box, Liao and Shen (2012) separately enumerate the further tonal splits found in CT and NT as follows. The following Table 8 firstly displays a split inside Row 1 of Column A in several Central Tai varieties. Table 8 A tonal split in voiceless friction sounds in Tone A5 Initial Type Protoinitials *f- The 1st row *th- SWT CT Debao Thai Dalong Wenma Gloss Bac Va pʰɔ:n³¹ 24 fon tʰɛ:w24 tʰe:w³¹ in Gedney’s *kh- kʰa: kʰa: tone box: *h- ha:w24 ha:w³¹ voiceless *hm- ma:24 ma:53 4 53 24 31 rain column kʰɔ:⁴¹ 1 kha: leg yawn mɔ:²¹ ma:2 dog friction *hn- na:² sounds *hŋ- ŋa:j24 ŋa:j53 supine *hl- la:j24 la:j53 many *s- sɔ:ŋ θo:ŋ na: 24 thick 53 two As in Table 8, one of the most noticeable problems with Row 1 in Gedney’s tone box is a split among the voiceless friction sounds (Row 1) in several CT varieties, which separates two series of initials with different phonetic features. These two series are voiceless aspirated sounds (including proto-voiceless aspirated stops and *h-, *x-, *f-, and etc.) in shaded, and voiceless continuants (*hm-, *hn-, *hɲ-, *hŋ-, *hw-, *hj-, *hl-, and *s-). Liao and Shen (2012) have pointed out the split among voiceless friction sounds in Gedney’s tone box is clear as shown in Table 8, and have proposed a tone box with four rows of the tonal categories associated with proto-voiceless initials instead of three. They furthermore point out that actually Li (1966) has noted that there should be four tonal categories associated with voiceless initials from Proto-Tai, based on the evidence of tonal development in T’ien-pao (nowadays Debao Zhuang or Debao variety of Yang Zhuang in ISO 639-3): 1) voiceless unaspirated stops, 2) voiceless continuants, 3) glottalized consonants, and 4) voiceless aspirated stops. Li once again emphasizes “the Proto-Tai initials may be roughly divided into five groups according to the influence they may have on tone” (Li 1977: 43) and lists 1) 5 The sources of the tone data in the Table 8 are as follows, Thai and Debao Dalong variety of Yang Zhuang are from my own research, Wenma which is of a Dai Zhuang (CT) variety in Yunnan is from Theraphan L-Thongkum (1997), and Bac Va which is a CT variety in Vietnam is from Hudak (2008). 19 voiceless aspirated stops, 2) voiceless continuants, 3) voiceless unaspirated stops, 4) glottalized consonants, and 5) voiced consonants. Gedney’s tone box does not follow Li’s proposal of four groups of voiceless initials, but divides voiceless initials into only three rows, merging voiceless aspirated stops and voiceless continuants into ‘voiceless friction’. In Table 8, we have seen the evidence supporting Li’s proposal of four groups of voiceless initials over Gedney’s three. As in the following Table 8, Liao and Shen (2012) separate the first row of Gedney’s tone box to two rows naming ‘voiceless aspirations’ (new Row 1) and ‘voiceless continuants’ (new Row 2) in their revised Tai tone box. In addition to the split above in Row 1 of Gedney’s tone box, Liao and Shen (2012) also point out there is another problem with this row. In their additional row “voiceless aspirations” in CT and SWT varieties are actually lost the aspiration in NT varieties (modern NT varieties normally do not have aspirated sounds), and this series of initials also condition splits into two tonal series in NT varieties. One of these two tonal series always merges with the tone with voiced initials which is of the fourth row in Gedney’s tone box, in contrast this merger is never found in CT and SWT varieties. This series of initial is called “voicing alternation” by some linguists, such as Diller points out the so-called “voicing alternation” of proto- voiceless and proto-voiced among different Tai languages (Diller 1998: 7), which is a series of initials behaving an original low tone in NT languages but behaving an original high tone in both SWT and CT languages. The following Tables 9 and 10 use examples filled in Columns A and C of Gedney’s tone box from SWT, CT, and NT varieties respectively to show that such tonal splits and mergers are only found in NT varieties6. Note that in Tables 9 and Table 10, shaded areas refer to examples of tonal categories with proto-voiced initials and those split from tonal categories with proto-voiceless initials to merge with tonal values of voiced series. 6 The sources of the data in these two tables are as following: Siamese and Jingxi urban variety of Yang Zhuang (Jingxi for short in the tables) from my own research, and Donglan from Zhang et al. (1999). The proto-initials are adapted from Li (1977) and Liang & Zhang (1996). 20 Table 9 Comparison of aspirations and voiced sounds among SWT, CT and NT in Column A Examples of tonal splits and mergers Column A in Gedney’s tone box Gloss *x- white *kh- to sell Aspirations *th- in addition in Gedney’s *ɢɦ- bitter Row 1 *gɦ- *dɦ- son-in-law SWT CT NT Thai Jingxi Donglan kʰa:w²⁴ kʰa:w⁵³ ha:w⁵³ kʰa:j²⁴ tʰɛ:m²⁴ kʰom²⁴ kʰə:j²⁴ reach *bɦ- tʰɨŋ²⁴ sharpen fon²⁴ Voiced in *b- expensive Gedney’s Row pʰɛ:ŋ³³ *d- paint 4 *g- stuck tʰa:³³ kʰa:j⁵³ tʰe:m⁵³ kʰam⁵³ ka:j⁵³ tem⁵³ ham²³¹ kʰy:⁵³ kɯ:j²³¹ pʰɔ:n⁵³ pan²³¹ pe:ŋ³¹ peŋ²³¹ ta:³¹ ta:²³¹ ka:³¹ ka:²³¹ tʰaŋ⁵³ kʰa:³³ taŋ²³¹ Table 10 Comparison of aspirations and voiced sounds among SWT, CT and NT in Column C Examples of tonal splits and mergers Column C in Gedney’s tone box *xAspirations in Gedney’s Row 1 Voiced in Gedney’s Row 4 Gloss enter *kh- kill *ph- sweep *ɣɦ- *dɦ- rice bowl *bɦ- person *b- bride7 *d*g- SWT CT Thai Jingxi kʰaw⁴¹ kʰa:⁴¹ pʰɛ:w⁴¹ kʰa:w⁴¹ tʰuaj⁴¹ NT kʰaw 33 kʰa: 33 haw³³ ka:³³ pʰe:w33 pe:w³³ 33 haw¹⁴ 33 tu:j¹⁴ kʰaw tʰu:j pʰu:⁴¹ pʰow stomach tʰɔ:ŋ⁴⁵³ to chew kʰiaw⁴⁵³ to:ŋ (sa)pʰaj⁴⁵³ Donglan 33 pɔ:j33 33 33 ke:w pu:¹⁴ paɰ¹⁴ tuŋ¹⁴ ke:w¹⁴ Liao and Shen (2012) point out that there is no split inside the aspirations in Thai (SWT) and Jingxi (CT), while there is a split in Donglan (NT). Actually all described NT varieties exhibit the same split as in Donglan. That is to say, in the vast NT varieties, one of the two series of “proto-aspirated sounds” (Li 1977) always presents with the same tone as that found in Gedney’s Row 4 (the voiced sounds). This series of proto-initials is made up entirely of voiceless stops with a voiced aspiration (ɦ-) in Li Fang-Kuei’s reconstruction (Li 1989), such as *kɦ-, *tɦ-, and *pɦ-, while it is made 7 In SWT varieties, this etymon normally refers to ‘daughter-in-law’. 21 up entirely of voiced stops with a voiced aspiration (ɦ-) in Liang and Zhang’s reconstruction (1996), such as *ɢɦ-, *gɦ-, *dɦ-, and *bɦ-. The later one is adopted in this thesis due to its more reasonable collocation of “voiced stop + voiced fricative” than “voiceless stop + voiced fricative”. Obviously the tonal development of protovoiced aspirated sounds have gone along two different paths of development in the tone box: in CT-SWT varieties, they maintain the aspiration but lost the voiced feature, and merge into voiceless aspirated sounds (Row 1) in the tone box; while in modern NT varieties, they keep the voiced feature but lost the aspiration, and merge into voiced sounds (Gedney’s Row 4) in the tone box. Liao and Shen (2012) furthermore indicate that Gedney had actually earlier reconstructed “a series of six additional initial consonants for Proto-Tai, three stops and three spirants, which would have to form a fifth category in the chart of initial consonant types, contiguous with the fourth category, the voiced sounds, because in Tai languages of the Northern branch they behaved like voiced initials, and also with the first category, the voiceless friction sounds, because in all other Tai languages they behaved like these, so that our chart becomes a closed loop, just as in the color spectrum infra-red at one end meets ultra-violet at the other” (Gedney 1985: 120). And, furthermore, Gedney also demonstrates the evidence for this series of voiced initial in Proto-Tai, as *B, *D, *G, *V, *Z and *ɣ- (Gedney 1989). However, he does not make any addition to the tone box at the end of the discussion. Though Gedney’s and Li’s reconstructions differ in the details, it is conceivable that there is a series of special voiced sounds (voiced aspirated sounds in Li and another series of voiced initials in Gedney), which condition the different tonal developments between common aspirated sounds and common voiced sounds. In this sense, the “tonal split” of aspirations reflecting in NT languages is actually not an “split” anymore, but is a merger from another source – in NT varieties, it merges with voiced sounds, but in CT/SWT varieties it merges with voiceless aspirations. Liao and Shen’s (2012) solution for dealing with the two series of aspirated sounds (the aspirated feature only reflects in CT and SWT) is to divide them into two rows. They call these two rows “voiceless aspirations” (new Row 1 including proto- voiceless aspirated stops *ph-, *phl/r-, *th-, *kh-, *khl/r-, *khw-, *tɕʰ-, etc. and other spread sounds such as *h-, *x-, *f-, etc.), and “voiced aspirations and breathies (new Row 7 including proto-initial sounds reconstructed by them as *kɦ-, *tɦ-, *pɦ-, *s̤-, *f̤-, *ɦ-, and *r̤-), and sort the latter to be the last row – Row 7 under the row of “voiced sounds” (new Row 6 in Liao & Shen 2012) in their revised Tai tone box as 22 the following Table 11. This order first makes the new Row 7 “voiced aspirations and breathiness” associate with the new Row 6 “voiced sounds” reflecting in NT varieties, and then also links the new Row 7 to the new Row 1 when the chart becomes a “closed loop” called by Gedney (Gedney 1985: 120), for keeping the single ‘contiguous’ tone box or a fixed order phonological spectrum (Gedney 1985: 120). This treatment is shown in Table 11. Table 11 Two new rows splitting from primary aspirations (adapted from Liao & Shen 2012) Initials Initial type Initial feature Proto- 1. Voiceless voiceless aspirations Protobreathiness Tonal category A B C DL DS A1 B1 C1 DL1 DS1 A1/A2 B1/B2 C1/C2 DL1/DL2 DS1/DS2 7. Voiced aspirations & breathies In order to designate tone category, Liao and Shen (2012) follow Li’s pattern to use 1 and 2 only as indicators of proto-tone, since in different Tai varieties they exhibit different vocalic features. Therefore, they cannot designate the voiced aspirations as either proto-voiceless sounds or voiced sounds, but instead they designate a new name “proto-breathy” for this special category, since the proto segment*ɦ-, common to all of them, carries breathiness, and they have determined there were non-stop breathy sounds (*s̤-, *f̤-, *ɦ-, and *r̤-) in the same row, too. Besides the noticeable tonal splits or mergers in Row 1 in Gedney’s tone box, Liao and Shen (2012) also discuss a split in Row 2 of Gedney’s tone box. They point out that Johnson (2011: 30, 36-38) has proposed the split in Row 2 which is voiceless unaspirated stops in Gedney’s tone box, by listing only two examples (‘to die’ and ‘eye’ only in A tone) from three Dai Zhuang (of CT) varieties to demonstrate this split. Liao and Shen (2012) furthermore list more examples to illustrate the split, as in the following Tables 12 and 138. The split demonstrated in Table 12 and Table 13 is dropped shadows. Row 2 of Gedney’s tone box refers to proto-unaspirated stops. However, some items like the shaded items in Tables 12 and 13 which are put into Gedney’s Row 2 indeed present as unaspirated stop initials in all modern varieties of SWT and NT, but present as aspirated stop initials or h- in all modern CT varieties. 8 In Tables 12 and 13, the proto-initials *tr- and *pr- are adopted from Li (1977), and *p.t- and *p.r- are adopted from Pittayawat Pittayaporn (2009). Ma’ai is a Yang Zhuang variety. 23 Table 12 Comparison of unaspirated stops among Tai varieties in Column A (adapted from Liao & Shen 2012) Tone box categories Unaspirated stops in Row 2 Examples of tones Gloss of Column A CT NT Thai Ma’ai Po-ai tʰa:³¹ ta:²⁴ to die ta:j³³ eye ta:³³ to go paj³³ *t- door 21 (praʔ ) tu:³³ *k- I ku:³³ *tr- or *p.tProto-initials SWT *p- tʰa:j³¹ paj⁵³ tow⁵³ kaw⁵³ ta:j²⁴ paj²⁴ tu:²⁴ ku:²⁴ Table 13 Comparison of unaspirated stops among Tai varieties in Column DL (adapted from Liao & Shen 2012) Tone box categories Unaspirated stops in Row 2 Examples of tones Gloss of Column DL1 CT NT Thai Ma’ai Po-ai to dry ta:k²¹ break tɛ:k²¹ *p- mouth pa:k²¹ *t- answer tɔ:p²¹ *k- to hug *pr- or *p.rProto-initials SWT kɔ:t²¹ tʰa:k³³ tʰe:k³³ pa:k⁵⁵ ta:p⁵⁵ ko:t⁵⁵ ta:k²² te:k²² pa:k²² ta:p²² ko:t²² The initials of this series of items has been reconstructed by Li (1977: 86-87, 118119, 225) as a cluster consisting of an unaspirated stop (p-, t-, k-) followed by -r-, and in all CT varieties, the proto-unaspirated stop is aspirated on account of the drop the following -r-. In contrast, in Pittayawat Pittayaporn (2009) which considers that there were no aspirated stops in PT, this series of aspirated stops in CT varieties develop from so-called “sesquisyllabic clusters” (Pittayawat Pittayaporn 2009: 5961), such as *p.t- and *p.r- in the two tables above. No matter whether there were aspirated stops in PT or not (we will discuss this in §4.2.1 since it is crucial to the tonal development of Tai languages), this aspiration in Gedney’s Row 2 causes its tonal development to behave the same as other voiceless aspirated initials in Row 1. This split is not easily perceived due to the relative rarity of examples of this split’s conditioning factors, and only several CT varieties such as most Dai Zhuang varieties and some varieties of Yang Zhuang always present this split – the tonal value of the aspirated stops is different from those unaspirated stops. That is, other CT varieties do not present this split because tones conditioned by voiceless aspirated stops and voiceless unaspirated stops are the same. 24 From the discussion above, we see that some CT varieties need further refinement in Gedney’s Row 2. Similar to the management of dividing Gedney’s Row 1 to new Rows 1 and 2, and furthermore new Row 1 to new Row 7, Liao and Shen’s (2012) solution for managing two series of voiceless unaspirated sounds is to divide it into two new rows. They adapt Li’s (1977) reconstruction of proto-initials in this row and call these two new rows – the new Row 3 being voiceless unaspirated stops (including Li’s proto-voiceless unaspirated plosives and affricates), and the new Row 4 being “voiceless unaspirated stop + *r clusters” (*kr-, *pr-, and *tr-), which are shown in the following Table 14. Table 14 Two new rows splitting from Row 2 in Gedney’s tone box (adapted from Liao & Shen 2012) Tonal categories Proto-initials A B C DL DS A1 B1 C1 DL1 DS1 3. Voiceless unaspirated stops 4. Voiceless unaspirated stops + *r As the result of the illustration of tonal splits found in CT and NT varieties above, the revised tone box has been suggested by Liao and Shen as shown in Table 7. On my fieldwork research of this thesis, the practice has proved that this revised Tai tone box provides a more suitable framework as a research tool when researching the tones in CT and NT varieties. However, they have not been involved in the classification of Tai languages by their discovery on CT and NT tones. Moreover, there are also some exceptions of the cases of tonal irregularities lacking reasonable explanation among the three sub-branches of the Tai or even among different varieties of Tai languages, and the revised tone box still has limitations in its ability to describe these phenomena, as which are shown in the last part of the discussion of the following §1.2.3. All these unsolved issues are involved in the objectives of the research in this thesis. When centering on the objectives of this thesis and integrating them with the related discussions in §1.2.1 and §1.2.2, there are several outstanding problems apparent in the literature. First of all, the tonal development needs a clearer frame to explain its primary and secondary sequences in each Tai variety. Although Gedney’s tone box (1989[1972]) and the revised tone box found in Liao & Shen (2012) differ from the division of the 25 horizontal rows on account of their different initial grouping, the two boxes show the same synchronic division of the initials at times of tonal splits. That is to say, the two tone boxes do not describe the historical development of these tonal splits, such as two-way or three-way splits, and which ones occur initially and which ones are secondary splits. Liao & Shen’s revised tone box firstly divides proto-initials into three groups – voiceless, voiced and breathy, and then divides proto-voiceless one into five sub-groups based on the evidence of tonal splits in NT and CT varieties. These two levels of grouping the proto-initials show a hint for the sequence of tonal development especially those in voiceless sounds, but still do not provide a proper diachronic scheme. One of the most noticeable problems for grouping the proto-initials which conditioned tonal splits in history is whether the aspirated sounds should be put into that early stage of PT. As in the aforementioned discussion of the tonal split in Gedney’s row of “unaspirated stops” accompanying aspiration in a number of CT languages in §1.2.2, aspiration has been viewed as contrastive in PT in Li FangKuei’s reconstruction (Li 1977), which is widely accepted by Tai scholars. However, Pittayawat Pittayaporn (2009: 88-89) argues that PT lacked contrastive aspirations and the aspirated stops in modern CT and SWT languages developed mainly from clusters with medial *-r- as post-PT innovations. He mentions that Haudricourt (1963: 157) and Ferblus (1990) claim that aspirated sounds were of secondary development in SWT and CT varieties, and Liang and Zhang (1993, 1996) believe voiceless aspirated initials developed independently in different Tai varieties from various initial consonant clusters, such as *pr-, *tl-, *tr-, *pw-, *xp-, xpl-, and *xt-. Despite their different proposal on the specific origin of the rise and the development of voiceless aspirated initials in Tai languages, if voiceless aspiration truly secondarily developed in SWT and CT varieties, the treatment of the tonal split conditioned by voiceless aspirations in those varieties should be adjusted to be a post-PT development, but not a condition by “proto-initial”. In a similar way, although there is no dispute that there were pre-glottalized initial consonants in PT, Liang and Zhang (1996: 817-828) and Zhang and colleagues (1999: 243-245) suggest the tonal splits conditioned by this series of initials were secondary developments in a number of Tai varieties since they have various tonal behaviors that differ from each other. In contrast, in the remaining Tai varieties this series of initials uniformly present the same tonal behaviors with other voiceless initials, and this is why Li Fang Kuei and some scholars treat pre-glottalized initials as a sub-group among voiceless initials. The implication is if Liang and Zhang’s 26 proposal can be proved to be true, although pre-glottalized initial consonants may exist in PT, they might have the same tonal behavior with voiceless initial consonants at the earliest time of tonal splits, and then conditioned tonal re-split later independently in different Tai varieties. The treatment of the tonal split conditioned by pre-glottalized initial consonants in those Tai varieties should be adjusted to be a post-PT development. On the contrary, it appears that aforementioned “proto-breathy” sounds (Liao & Shen 2012) which conditions the so-called “voicing alternation” (Gedney 1989a, Diller 1998: 7, Pittayawat Pittayaporn 2009: 77) have conditioned primary tonal splits at the earliest division of the chronological stages of post-PT. Although only Liao and Shen (2012) explicitly provide a special initial group named “proto- breathy” sounds, this series of initials was said to exhibit “voicing alternation” by some linguists. For example, Diller points out the “voicing alternation” of proto- voiceless and proto-voiced among different Tai languages (Diller 1998: 7), which is a series of initials behaving like an original low tone in NT languages but behaving like an original high tone in both SWT and CT languages. In Li’s hypothesis, the reconstruction of this series of proto-initials is made up entirely of stops with a voiced aspiration (ɦ), such as *pɦ-, *tɦ-, *kɦ- (Li 1989). Obviously the tonal development of proto-voiced aspirated sounds has gone along two different paths of development in the tone box: in CT-SWT varieties, they maintain the aspiration and merge into voiceless aspirated sounds; while in modern NT varieties, they keep the voiced feature and merge into voiced sounds. This is very similar to Gedney’s (1985: 120) hypothesis of “a series of six additional initial consonants for Proto-Tai, three stops and three spirants” and furthermore his reconstruction of a series of voiced initial in Proto-Tai, as *B, *D, *G, *V, *Z and *ɣ (Gedney 1989). Furthermore, Zhang and colleagues (1999: 174) also modify this series of stops causing voicing alternation as voiced aspirated stops *bɦ-, *dɦ-, and *gɦ-, which behave original low tones (tones developing from proto-voiced initials) in NT but original high tones (tones developing from proto-voiceless initials) in CT and SWT. Although Pittayawat Pittayaporn (2009: 77-88) argues that so-called voicing alternation is not a unified phenomenon in PT, but consists of post-PT borrowings from Chinese which must not be reconstructed for PT, and positing so-called sesquisyllabic clusters. For interpreting this series of voicing alternation, Liao and Shen (2012) propose there was a series of proto-breathy initials out of the dichotomy of proto-voiced and protovoiceless initials, and have added a new row into the revised tone box for this series of initials (Liao & Shen 2012: 9-12). That is to say, initials in PT may be divided into a new trichotomy: proto-voiceless, proto-voiced, and proto-breathy. These “proto- 27 breathy” initials regularly condition tonal split going with original low tone with those developed from proto-voiced initials in all NT dialects, but condition tonal split going with original high tone with those developed from proto-voiceless initials in the vast majority of CT and SWT. If this series of “proto-breathy sounds” really existed in history, they should be treated as a group of sounds which should be put into the early stage of PT. The second outstanding question is whether the tonology of Tai languages can become evidence to support any of the theories of Tai classification either at the time of the formation of different Tai branches (or the early stage of post-PT) or at the stage of modern Tai languages. The tonology criterion of Tai division is complicated when being presented by different linguists. Some scholars such as Chamberlain (1975), and Theraphan L-Thongkum (1997) have proposed that the different tonal splits types can be seen as an important criterion of the classification of Tai branches (cf. §2.2.4). There is one thing that has been clarified earlier in this section is some tonal development conditioned by some types of initials in different Tai varieties may occur much later than the time of tonal split in PT, such as preglottalization and aspiration might condition secondary tonal splits like original high tones going with low tones observed by Edmondson (1994: 166-176). Then at least these secondary tonal splits should be reliably treated as a basis of Tai classification at variety level. If pre-glottalization and aspiration did condition secondary tonal splits in some specific Tai varieties, these kinds of tonal development should not be treated as a distinctive feature of classifying the main branches of Tai, since they occur far later than the time of the formation of different Tai branches, which are NT-CT/SWT being two plus one taxonomy (Gedney 1989) or NT-CT-SWT being the tripartite division (Li 1977). That is to say, at least some tonological phenomena such as tonal splits conditioned by pre-glottalized and aspirated initials should not be treated as evidence to support the Tai primary classification at the early stage of post-PT. However, this clarification does not mean that all the tonological phenomena of Tai varieties cannot be treated as distinctive features of the classification of the primary Tai branches. Since the tonal split of aforementioned “voicing alternation” conditioned by so-called “proto-breathy” sounds generally conditions different direction of tonal split between NT and CT/SWT, and it is supposed to be treated as a primary tonal development as in the aforementioned discussion, it should become one of the distinctive features which can determine the primary Tai branches. In the subsequent discussion, more phonological and lexical evidence will be referenced to 28 support this hypothesis. Since this hypothesis directly supports Gedney’s (1989b) two plus one taxonomy division – Li’s CT sub-branch and SWT sub-branch forming a sub-group of which the NT sub-branch is a sister, and Gedney (1989a: 231) also uses the term Southern Tai (hereafter ST) to refer to all the non-NT languages, it is proposed that before CT and SWT divided to two different groups, they were of the same proto-CT/SWT group. I adapt Gedney’s terminology to call this group ProtoSouthern Tai (hereafter PST) based on its geographical distribution relatively to Proto-NT (hereafter PNT). This division of PNT and PST is very similar to the twoterm system that Haudricourt (1956) insisted on, to divide the PT first into Dioi or ʔyai (corresponding approximately to NT) vs. Tai proper (corresponding to ST). As a matter of convenience, in this thesis North-South division is used to refer to this possible historical Tai primary classification on the early post-PT stage. The third outstanding problem is whether we can provide a more refined Tai tone box to be a supplement to Gedney’s tone box and Liao and Shen’s revised tone box, and to be able to capture the tonal distinctions in any Tai varieties. As in the discussion in §1.2.2, Gedney’s tone box can be treated as a suitable research tool when studying SWT varieties, but when working with CT and NT varieties, it has several shortcomings, and Liao and Shen (2012) has designed a revised tone box which aims at providing a more suitable framework as a research tool when studying Tai varieties out of SWT. However, they also admit the revised tone box still has limitations in its ability to describe every Tai language since there are cases of tones lacking regular correspondence prescribed by their tone box among different Tai varieties. For example, the word for ‘become; to be’ presents with a proto-voiceless initial in SWT (e.g. *penA1-U > pen³³ in Thai), but with a protovoiced initial in both CT and NT (e.g. *benA2 > pan³¹ in Debao Yang Zhuang and *ben A2 > pan³¹ in Wuming Zhuang), raising questions about Column A (Liao & Shen 2012). So far the conditions of such irregular tonal correspondence still lack reasonable explanation. If there are more potential rows of initials conditioning these seemingly irregular tonal behaviors, they should be added to Liao and Shen’s revised tone box for being able to capture these kinds of tonal distinctions. Actually Gedney (1989b: 25-33, 1989[1972]: 204) himself has alluded to this issue based on the tonal split on Box 1 of Column A of some varieties of CT an NT, and has pointed out that when the historical development of tones in NT is fully understood, more horizontal rows may have to be drawn in his Tai tone box (cf. §1.2.2). From this standpoint, to draw more lines or even columns to the tone box is not a problem if it is necessary. Based on the study of tonal development of Tai varieties in this thesis, I 29 will test whether a more accurate Tai tone box can be designed to reasonably explain these unsolved issues. In relation to the objectives outlined in §1.2 and discussed in §1.2.3, the following hypotheses are proposed: 1) After tone arose, three groups of proto-initials (voiceless-voiced-breathy) conditioned primary tonal splits whereas splits conditioned by other pharyngeal features, such as aspiration and pre-glottalization represent stages of secondary development. 2) The “Two plus one taxonomy” of Tai classification is more reasonable than the Tripartite Division at least in the early stage of Tai languages, because of different tonal behaviors of “proto-breathy” initials in Proto-Northern and ProtoSouthern Tai languages. 3) The proposed tone box will be able to capture the tonal distinctions in any Tai varieties. In relation to the background and the objectives of this thesis, the following sections are expanded for giving a brief overview to this study. The selection of language varieties from SWT, CT and NT are based on my own knowledge and my library research of tonal diversity of each Tai sub-group, for striving to cover all types of tonal splits in Tai. Tone data from the following fortytwo Tai varieties are used to determine the specific tonal developments: 1) Seven SWT varieties: Bangkok Thai (L1), Songkhla Southern Thai (L2), Khon Kaen Isan (L3), Chiang Rai Northern Thai (L4), Kho Lam Shan (L5), Rong Maet Tai Lue (L6), and Muong Quan Son Tai (L7). All the data of these languages have also been collected by me. Some published materials of the fully researched varieties such as Thai, Southern Thai, and Northern Thai are also referenced to. 2) Seventeen CT varieties: Six Yang Zhuang varieties including Debao urban (L8), Debao Dalong (L9), Debao Lüliu (L10), Jingxi Hurun (L11), Jingxi urban (L12), 30 and Jingxi Anning (L13); Debao Nalong Myang Zhuang (L14); Debao Tuoxin Pyang Zhuang (L15); Jingxi Huashan Min Zhuang (L16); as well as four Zuojiang Zhuang varieties including Tiandeng Xiangdu (L17), Daxin Naling (L18), Lingding Zuozhou (L19), and Daxin Baoxu (L23) which are collected by myself, but some of them are also compared with data of the same varieties in published resources. Xiaoguangnan Nong Zhuang (L20), Wenshan Dazhai Dai Zhuang (L21), Daxin Leiping Zuojiang Zhuang (L22), and Bac Va Nung (L24) are referred to published resources. 3) Thirteen NT varieties: Debao Longsang (undescribed L31), Debao Dongling (of Youjiang Zhuang, L32), Shanglin Yunling (of Central Hongshuihe Zhuang, L34), Du’an Jiaren (of Central Hongshuihe Zhuang, L35), Yizhou Suogan (of Guibei Zhuang, L36), Huishui Bouyei (L41), and Zhenning Bouyei (L42) are collected by myself, some of them are also compared with the data of the same languages in published resources. Wuming Shuangqiao (of Yongbei Zhuang, L30), Tianlin Lizhou (of Guibian Zhuang, L33), Liujiang Baipeng (of Liujiang Zhuang, L37), Donglan Urban (of Guibei Zhuang, L38), Qiubei Gehan (of Qiubei Zhuang, L39), and Lianshan Xiaosanjiang (of Lianshan Zhuang, L40) are referred to published resources. 4) Five varieties of YN whose classification as CT or NT is open to debate: Yongning Baiji (of Sanxiang YN, L25), Nanning Shuangding (L26), and Jingxi Daqiu Nung An (L28) are collected by myself, and Sanxiang and Nung An are also compared with the data of the same languages in published resources; Long’an Xiaolin (L27) and Yongning Xialeng (L29) are referred to published resources. The language names are cross-referred in §3.3.1 and Appendix C, a chart of the information of each language is also attached in Appendix C, and the map of the Tai varieties investigated in this thesis is shown in Appendix D. The numbers of the Tai varieties have the same ordering in this map. The tone data selected are analyzed by Liao and Shen’s (2012) revised tone box which is revised from Gedney’s (1989[1972]) tone box, and an overview of tones in each Tai variety is given. More attention will be given to the analysis of the tonal development in history and trends of tonal development in the current situation in different Tai varieties. Language data from varieties from CT and NT languages have been collected in Debao, Jingxi, Napo Counties, and Nanning City of the Guangxi Zhuang Autonomous Region, and Guangzhou of Guangdong Province of China via three times of fieldwork there. Transcription, translation and checking were done in 31 Guangxi and Chiang Mai. In addition, data of Thai, Northern Thai, Shan and Tai Lue were collected in Thailand. Tone data from another SWT language spoken in Thanh Hoa Province in Central Vietnam has been collected in the U.S. More language data will be referenced from published materials. The detail of data collection is crossreferred in §3.3.1 and is also listed in Appendix D. Sometimes the data collected from varieties selected are also limited because of several reasons. First, the Tai language branch is distributed widely in South China and MSEA, and has numerous varieties. It is impossible to work out a full-scale data collection of tones in a short time. Some special issues related to the tonal development may be hidden in those undescribed or unstudied varieties. Second, some data selected from published sources may be imprecise, and may give the study a negative result. This kind of issue has been found a lot in the previous readings. This is unavoidable and needs further research to remedy in the future. Furthermore, the wordlist itself also has limitations. The wordlist designated is only limited to cognates including Old Chinese loans inherited from PT, but is excluded to those modern loans from other languages in each variety, although such modern loans like English loans in Thai and modern Mandarin loans in Yang Zhuang varieties have been observed to cause some changes on tonal behavior. Besides, the wordlist also conditions selected data to be limited to speech environments. That is, I could only acquire the tone data from single words when the language resource person (LRP) read and repeated them, but could not observe the tone expression from the natural conversation. That means some tone changes (in natural speech) such as tone sandhi might be missed. Since tone changes in natural conversation especially tone sandhi may be accounted to be the newest part of tonal development, the limitation of lacking tone data from this occasion is regrettable but unmodifiable because of the limited fieldwork condition. For remedying these regrets, in §6.6 I also analyze tonal behavior changes due to Mandarin loans in Debao Urban Yang Zhuang (L8) and Thai loans in Tai Lue (L6) as additional examples of tone changes out of the tone expressing occasions limited by the wordlist. In relation to the hypotheses of the research in §1.2.4, the following contributions are expected: 32 1. This study will contribute to Tai linguistics by providing more detailed research on tonal diversities of Tai languages, especially those varieties beyond SWT. 2. By providing a more refined system of describing tone systems of Tai varieties, it will provide for a more comprehensive inventory of tonal split patterns that can be used as criteria for subdivision of the Tai language family. 3. A new tone box applied to the vast majority of Tai varieties will solve some of the shortcomings of Gedney’s (1989[1972]) tone box and Liao & Shen’s (2012) revised tone box. The remainder of this thesis consists of eight chapters. Chapter 2 presents a brief review of the literature relating to historical and comparative Tai and Tai tonology, including less well-known Chinese materials of Tai tone research. Chapter 3 describes the theoretical foundations of methodology used in this thesis, the fieldwork relating to the research methods, as well as notation for this thesis. Chapters 4, 5 and 6 deal with the discussions for establishing the three hypotheses of this thesis, namely diachronic hierarchies of Tai tonal development, Tai classification based on tonology, and a comprehensive new Tai tone box respectively. Chapter 7 as the last chapter provides the summarized conclusion of this thesis, as well as suggestions for further studies of Tai tones. 33 Literature Review This chapter presents a review of the literature topics relating to the subjects of the thesis, namely tonology of Tai, Tai classification based on tonology, and Tai tone boxes. In order to put these topics in the broader context, §2.1 and its subsections give brief reviews of the literature relating to historical and comparative linguistics, providing theoretical foundations to this study. §2.2 describes the previous works on tonology of languages in China and MSEA by dividing the section into four subsections, namely tonogenesis, tonal splits and mergers, the conditioning factors of tonal splits, and Tai classification related to tonological issues. Within these subsections, sub-sections §2.2.1 and §2.2.2 are both related to the processes of tonal development; §2.2.3 reviews the previous works on the method of analyzing tonal splits based on the conditional factors, since this will help us to explore a new Tai tone box which will capture more tonal distinctions in any Tai varieties than the Tai tone boxes proposed by Gedney (1989[1972]) and Liao and Shen (2012); §2.2.4 gives a brief review of Tai classification based on the issues related to tonology, and this will give clues for testing our “Two plus one taxonomy” of Tai classification which is proposed to be more reasonable than the Tripartite Division at least in the early stage of Tai languages. Because the topic of this thesis – tonal development of Tai languages is based on Tai tonology, which is mainly due to the comparative method applying to Tai tones, the theory of historical and comparative linguistics is bound to become the basis of the research tool of this thesis. Historical linguistics including comparative methods has been becoming an important linguistic subject for a long time since it first developed in the 19th century for studying the historical linguistic changes among IndoEuropean languages, hence there are many precedents in historical linguistics like Hoenigswald (1960), Arlotto (1972), Lass (1997), Crowley (1994), and Campbell (2004), and to a large extent Jeffers & Lehiste (1982) and Hock (1991), and in other areas as well, e.g. Harris (1960) and Lyons (1977). In the Sinitic Circle, Historical 34 Linguistics by Xu (1991) is a summative work with reference value. The following respects of previous works on historical linguistics, comparative methods, and historical and comparative Tai are referenced to in this thesis. Historical and comparative method applying to Tai languages has been occupying a significant part in the Tai linguistics. Within these achievements, Li Fang Kuei’s (Li 1977) articles on Tai linguistics and his remarkable work A handbook of Comparative Tai, Gedney’s (1989[1972]) tone box and his other historical and comparative Tai linguistics works, and other works by other Tai linguists have provided abundant materials for subsequent research. In this section, the discussions will focus on some Tai linguistics topics related to linguistic change, comparative method, language classification, and areal linguistics except the content related to tonology which will be discussed in the subsequent sub-sections of §2.2. In previous studies, there are some specific cases related to linguistic changes in Tai languages as in the following examples. The discussions of the reconstruction of the Tai word ‘maternal grandmother’ have given a characteristic example for sound changes in Tai languages. This Tai word presents an irregularity on its modern forms ta:jB1 in most NT varieties, ta:jA1 in most CT varieties, na:jA2 in most SWT varieties, and ja:jA2 in Thai (Li 1971). Either the initial consonants or the tones of this word among three Tai sub-branches are different. Li (1971) and Liao’s (2013) opposite reconstructions of this word in PT, namely *na:jA2 and *ta:jA respectively, containing the contents of irregular sound change especially analogical change and a specific factor of sound change – “avoidance of pernicious homophony” – a term used by Campbell (2004: 322). Different forms of the aberrant development of this word within Tai sub-branches have been reflecting the dialectologists’ slogan ‘each word has its own history’9 attributed to Jules Gillíron (Campbell 2004: 212). The dialectologists’ slogan reflects the thinking of Johannes Schmidt’s ‘wave theory’ (1872), which deals with linguistic changes involving in language contact. Based on Schmidt’s wave model, linguistic changes spread outward concentrically like waves, and farther to the central point of the change the weaker influence of the change is presented. Therefore, there is no sharp boundaries between neighboring dialects or languages. According to this theory, the dialectologists’ slogan actually reflects that “a word’s history might be the result of various influences from various directions, and these might be quite different from those involved in another word’s history, hence each word has its own history”. (Campbell 2004: 213) 9 35 The historical linguistic change of a numeral word nɨŋB1 ‘one’ in Thai discussed by Li (1987) is an illustrative example for indicating syntactic change in Tai languages. Different from other numerals placed before the noun classifiers or measures, the numeral nɨŋB1 ‘one’ may occupy both the pre-classifier and the post-classifier positions in a noun phrase in Thai. He gives three hypotheses of the peculiar behavior of this numeral: 1) the pre-classifier is the original position, 2) the postclassifier is the original position, and 3) both of them are original positions just as in Thai today. He tries to find the evidence firstly from comparing cognate or corresponding meaning word of this numeral in other Tai languages. For example, the cognate of this numeral nɨŋB1 in Lao (most closely related language to Thai) and nɨŋB1 in Lungchow (a CT language), can only occur as a post-classifier or occur after the classifier. All other numerals in Lungchow are placed before the classifier as in Thai. Other words having corresponding meaning of nɨŋB1 in other Tai dialects, such as ʔdiaw in Wuming and ʔee in Po-ai (both of NT) also are placed after the classifier. Li states, “Thus the numeral one, whether it is nɨŋB1 or some other word, has definitely an association with the post-classifier position” (Li 1978: 142). From the documentary evidence, Li also finds that the earlier documentary evidence in Thai shows that nɨŋB1 was only used in the post-classifier position, while the other numerals were in the pre-classifier position. Finally, Li proposes that nɨŋB1 in modern Thai behaving in pre-classifier position was formed by being influenced by the usage of other numerals. Its original meaning also changes as “a descriptive word” after the syntactic change from the post-classifier position to the pre-classifier position. Hock (1991: 560) talks about the cases of borrowing like the English/French/ German relationship which furnishes people with insights that “make it possible to detect borrowing in other cases, where we do not have earlier, direct historical evidence indicating a borrowing relationship” to interpret the borrowings from Sinitic languages to TK languages by citing Greenberg’s (1957) argument of “triangular relationship between Thai10, Sino-Tibetan, and Kadai/Malayo- Polynesian11”, which concludes “the relationship between Thai and Sino-Tibetan must be one of borrowing”, since the “specific resemblances” between Thai and Chinese always “cluster in a few semantic spheres”, while Thai and Kadai/MalayoPolynesian may have genetic relations because those resemblances between Thai “tend to recur throughout the family” (Hock 1991: 560-561). In Hock’s discussion, “Thai” actually refers to the Tai language branch of TK. In Hock’s citation from Greenberg (1957), Kadai and Malayo-Polynesian seem to be sub-grouped together and to be different from the classification as TK in this thesis. 10 11 36 A similar point of view is also seen in the discussion by Xu (1991: 64) about the relationship among Kam-Tai, Miao-Yao and Sinitic languages. Xu cites Benedict’s (1984) discussion about the resemblances of the numerals in Kam-Tai, Miao-Yao, and Sinitic languages to demonstrate that even basic vocabularies like numerals can be borrowed, and calls a question about the traditional hypothesis of “Sino-TibetanTai-Miao/Yao genetic relationship” proposed by most scholars in the Sinitic circles. In the field of applying the comparative method to Tai languages, Li (1977) has done a most widely accepted PT reconstruction in his famous work A handbook of comparative Tai. His reconstruction has become a basic source on the comparative method of the Tai languages, has established ‘Tai’ in one particular sense, and has delineated a particular family tree of Tai languages. Li firstly enumerates fourteen Tai languages in the introduction section being the presentation of the basic materials of the comparative Tai, and chooses three of them, Siamese, Lungchow, and Po-ai, to be the representative of SWT, CT and NT languages respectively in the following sections, to reconstruct the tones, the consonants, and the vowels of PT. Data from other eleven Tai languages are also referenced to. The comparative method applied to Tai languages in the book cogently demonstrates the process of the reconstruction of PT morphemes. For example, when reconstructing the protoinitials, Li infers a proto-form by comparing the initials of cognates determined among the sample languages of SWT, CT, and NT. The proto-initials inferred are not always inherited in any of modern languages in these three Tai sub-branches. For example, the proto *b- has developed to ph- in Siamese (SWT) and p- in both Lungchow (CT) and Po-ai (NT). The reason that Li reconstructs the proto-initial as *b- is because the tones of this series of cognates are proposed to be developing from PT tones which always occurred with proto-voiced initials (Li 1977: 65-66). For the proto-voiceless series, Li also indicates that “it is also assumed that the voiceless nasals and liquids, written *hm-, *hn-, etc., remained voiceless, although all are voiced in the modern dialects” (Li 1977: 254-255). In conclusion, Li gives four PT tone categories *ABCD, sixty-seven initials, six final consonants, and thirty-three vowels as the integrated reconstruction for PT12. 12 Li’s (1977) reconstruction of PT consonants and vowels are as follows. 1) Sixty-seven initials include: 10 labials *p-, *pʰ-, *b-, *ʔb-, *m-, *hm-, *f-, *v-, *w- and *hw-; 9 labial clusters *pl-, *pr-, *phl/r-, *bl- (*br-), *ʔbl/r-, *ml/r-, *fr-, *vr-, and *vl-; 6 dentals *t-, *tʰ-, *d-, *ʔd-, *n- and *hn-; 8 dental clusters *tl-, *tr-, *tʰl-, tʰr-, *dl-, *dr-, *ʔdl/r- and *nl/r-; 4 liquids *l-, *hl-, *r- and *hr-; 9 sibilants *s-, *z-, *t -, *t ʰ-, *dʑ-, *ɲ-, *ɒɲ-, *j- and *ʔɔ-; 7 velars *k-, *ɕʰ-, *g-, *ŋ-, *ɒŋ-, *x- and * -; 37 Pittayawat Pittayaporn (2009) has also accomplished a comprehensive work on reconstructing the phonological sound system of PT based on a systematic application of the comparative method and an appreciation of the force of contact. Opposite to the generally accepted assumption that Proto-Tai was monosyllabic, he claims that PT was “a sesquisyllabic language that allowed both sesquisyllabic and monosyllabic prosodic word” (Pittayawat Pittayaporn 2009: 2). As the argument in the reconstruction of the PT prosodic word, he states that the monosyllabic view of PT fails to account for the range of comparative data and suggests in favor of the sesquisyllabic hypothesis (Pittayawat Pittayaporn 2009: 39-40). For example, he gives an initial consonant with sesquisyllabic structure *t.n- for PT, for the set of initials of the word list includes ‘star’, ‘to transplant’, and ‘earthworm’ which developed into *d- in Siamese (da:wA1, damA1, dɯənA1), Lungchow (da:wA1, damA1, dɤ:nA1), and Yay (da:wA1, damA1, dɯənA1), but into *tr- in Saek (tra:wA1, tramA1, truəlA1), because Li’s (1977) reconstruction for this set cannot be accounted for since it is limited to monosyllables (Pittayawat Pittayaporn 2009: 44-45). For the onset, consonants can combine to form either tautosyllabic clusters (e.g. *kr-, *tr-, and *pr-) or sesquisyllabic structures (e.g. *p.t-, *q.s-, and *k.r). For the reconstruction of initials, in contrast to the previous reconstructions, he argues that PT lacked the aspirated stop series (Pittayawat Pittayaporn 2009: 88-89), but only had the three contrastive phonation types, namely plain voiceless, implosive, and voiced. As for place of articulation, he adds a distinctive uvular series, in addition to the other five series (labial, alveolar, palatal, velar, and glottal) previously reconstructed (Pittayawat Pittayaporn 2009: 75-76). For the rime, PT had seven vowel qualities that contrasted in height, backness, rounding and length contrast for each quality (Pittayawat Pittayaporn 2009: 192-193). For the codas, PT had palatal and lateral as final consonants in addition to the ones generally reconstructed (Pittayawat Pittayaporn 2009: 193). Finally, for the tones PT was a tonal language contained four tone categories *ABCD contrasted both in terms of pitch and voice quality (Pittayawat Pittayaporn 2009: 285). He also points out that many of PT traits newly reconstructed, such as, sesquisyllabic and uvular initial series, are not attested in modern Tai languages because of the changs overridden by waves of convergent trends that transformed PT (Pittayawat Pittayaporn 2009: I-II). 8 velar clusters *kl-, *kr-, ɕʰl-, *ɕʰr-, *gl-, *gr-, *ŋl/r-, and *xr-; 6 labio-velars *kw-, *ɕʰw-, *gw-, *ŋw, *xw-, and * w-; 2 laryngeals *ʔ- and *h-. 2) Six final consonants include *-p, *-t, *-k, *-m, *-n, and *-ŋ. 3) Thirty-three vowels include: 9 simple vowels *-i, *-ɨ, *-u, *-e, *-ə, *-o, *-ɛ, *-a, and *- ; 20 diphthongs *-ie, *-ɨe, *-ue, *-əɓ, *-ei, *-ɛɓ, *-ai, *-ɨai, *- ɓ, *-əɨ, *-eɨ/ɛɨ, *-oɨ, *-əu, *-eu, *-ou, *-ɓəu, *-au, *-ɛu, *-iu and *-ɨɓ, and 4 triphthongs *iau, *-ɓəu, *uai, and *uəɓ. 38 In the Sinitic circles, Zhang and colleagues (1999: 168-246) have done an enormous comparison of sound systems and vocabularies among many dialects of NZ (of NT) and SZ (of CT) and some SWT languages. They choose cognates from 36 dialects (of which 22 are from NZ and 14 are from SZ) as the comparative materials, to clarify the correspondence of the sounds among Zhuang dialects. They state that they do not aim to reconstruct all the proto-sounds of so-called proto-Zhuang since SZ and NZ inherit from different specific ancestral groups (Luoyue and Xi’ou), and the sounds and vocabularies of SZ are closer to Dai (Lue), Thai, Lao, Shan and etc., on the contrary those of NZ are closer to Bouyei, but they do need to discuss some specific proto-initials and proto-vowels when analyzing some correspondences. Some proto-forms are adopted from Li’s (1977) reconstruction but some are modified by themselves (Zhang et al. 1999: 169). For example, the voicing alternation of protovoiceless and proto-voiced stops among different Tai languages which lack explanations in Li’s work (Li 1977), Zhang and colleagues (1999: 174) propose that they are of a series of voiced aspirated stops **bɦ-, **dɦ-, and **gɦ-13, and developed to be voiced stops with odd tones in pre-NT dialects but voiceless stops with even tones in pre-SZ dialects (cf. §1.2.2, §1.2.3, §2.2.3, §2.2.4). As the Tai language classification which has been briefly discussed in §1.1.2, there are less debates of grouping the branches of Tai (Zhuang-Dai), Kam-Sui (Dong-Shui), Hlai (Li) and Kra (Ge-Yang) together to form a language family named TK language family, which is also called Zhuang-Dong ( 侗) or Dong-Tai (侗台) in the Sinitic circles. In this section, more theoretical principles of classifying Tai languages and their genetic related languages are presented, except the Tai classification based on tonology which will be reviewed in §2.2.4. Tai Dong-Tai Family Dong-Shui Zhuang, Bouyei, Dai, Lingao, Thai, Lao, Shan, Black Tai, White Tai, Khamti, Saek, Tho, Nung, Tay, Ahom Dong, Mulao, Shui, Maonan, Mo, Yanghuang, Lakkja, Biao Li Li, Cun Ge-Yang Gelao, Laji, Pubiao, Buyang, Laha Figure 5 Dong-Tai (TK) family tree (adapted from Zhang et al. 1999: 3) 13 In Zhang et al. (1999), the single asterisk (*) is for labeling the forms in the node of Proto-Tai, the double asterisk (**) is for labeling the forms in the node of Proto-Kam-Tai. 39 Similar to Diller’s (2008: 7) TK language family tree in Figure 1 in §1.1.1, Zhang and colleagues (1999: 2-3) also provide an illustration of the family tree to the TK languages, which they refer to Dong-Tai language group (侗台语族) as shown in Figure 5. The illustration groups different TK branches, namely Tai, Dong-Shui (Kam-Sui in the Western circles), Li (Hlai in the Western circles), and Ge-Yang (Kra in the Western circles) together to form the Dong-Tai language family. In Figure 5 above, some language names follow the Chinese official items which are based on the national ethnic identification. Some of the languages like Zhuang actually consists of numerous “dialects” which are actually defined as single languages in ISO 639-3 language code (see §1.1.3). When subgrouping the Tai branch of TK Family, Li’s (1977) scheme of classifying the three sub-branches of Tai languages, namely SWT, CT, and NT, has long been accepted as the standard one in the field of comparative Tai linguistics. His definitions for the criteria of classifying the Tai languages are 1) phonological features. For example, PT *tr- gives different modern form in the dialects of these three branches, as in that it presents the initial of t- in the Southwestern dialects (Siamese taa ‘eye’), th- or h- in the Central dialects (Lungchow ha: ‘eye’, Tho ta: ‘eye’), and t- or r- in the Northern dialects (Po-ai ta: ‘eye’, Wuming ra: ‘eye’); 2) certain lexical items showing phonological variation, such as, the word for ‘to be’ indicates an original voiceless initial in SWT varieties, but indicates an original voiced initial in both CT and NT varieties (cf. §1.2.3); and 3) lexical items containing, as ‘there are lexical items which are lost in one of the subgroups but kept in the other two’ (Li 1977: xiii-xv). Since then the argument of the classification of Tai languages is mainly revolved in the relationship between Li’s CT and SWT sub-branches of this language family. Even though Li’s (1977) proposal of a tripartite division of Tai has long been accepted, Gedney (1989b: 62-66) and Chamberlain (1975) propose that the CT and SWT can form a sub-group of which NT is a sister. Nevertheless, Gedney also admits that there are basic differences between Li’s CT and SWT sub-branches, but argues that the scale of these differences is different from those between NT and other Tai languages, from the evidence of some intermediate dialects between SWT and CT languages. He points out that “(t)here is general agreement that the Northern Tai languages form a separate branch” (Gedney 1989: 62), supposes that “the Central and Southwestern groups form a continuous dialect area, with only gradual transition throughout and no real language boundary anywhere…” (Gedney 1989b: 40 66), and suggests using the cover term ST to refer to all the non-NT languages (Gedney 1989a: 231). As mentioned previously in this section, some “languages” in the Tai branch like Zhuang consist of more than one single language, and this leads to a more complex position of language subgrouping in Tai. The classification of Zhuang languages has been shown to be a complicated situation associated with political national identification (cf. Zhang et al. 1999). The population of the Zhuang is about 16,178,800 in 2000, making the Zhuang the biggest minority group in China (Office of Population and Social Science Statistics et al. 2003). However, as Edmondson (1994) points out that “(d)espite a large population, despite relatively accessible settlement areas on flat land, and despite a sizable urban population, the Zhuang are not well-known even among Tai specialists and other SE Asian cognoscenti” (Edmondson 1994: 147), the Zhuang is disproportionally not well known out of China when comparing with its large population. Besides, many scholars have found that several languages are very closely related to Zhuang. For instance, Tay (1 million) and Nung (70,000) in northern Vietnam are connected with SZ, and the language of 2.5 million Bouyei people in Guizhou province of China and Giay/Yay/Bo-Y in northern Vietnam are very closely related to NZ (Edmondson 1994: 149). Primarily, according to the classification established by Chinese scholars, there are two main dialects of the Zhuang language—NZ and SZ. Approximately two-thirds of the Zhuang people speak varieties of NZ, while the remaining one-third, about six million people speak varieties of SZ (Jackson et al. 2011: 1). For solving the contradiction between the linguistic classification and political ethnic identification of the Zhuang and Bouyei languages in China, Zhang and colleagues (1999: 11-12) first put SZ (linguistically of Li’s CT sub-branch) with Dai (Lue), Thai, Lao, and Shan (all of Li’s SWT sub-branch) together, and NZ with Bouyei (both linguistically of Li’s NT sub-branch) for clarifying the first stage of divergence time of PT. The evidence for establishing these two historical branches are both phonology and vocabulary. First, Thai, Dai, Lao, and Shan have aspirated initials shared with SZ, while NZ and Bouyei do not have these aspirated initials. Second, NZ and Bouyei have a segment r- initial occurring with both even and odd tones. Finally, there are also a number of diagnostic lexical items separate NZ/Bouyei from SZ/SWT, and some of them are listed in the following Table 15. 41 Table 15 Some diagnostic lexical items of two groups in Tai (adapted from Zhang et al. 1999: 9-10) Southern Tai Group Southwestern Tai Gloss ‘wing’ ‘tiger’ ‘cloth’ ‘head’ ‘above’ 14 Thai Lao pi:k9 pi:k9 sɯə¹ sɯə¹ sɯə³ huə¹ nɯə¹ sɯə³ huə¹ nɯə¹ Dai Northern Tai Group Southern Zhuang Northern Zhuang/Bouyei Longzhou Jingxi Wuming Liujiang Bouyei pi:k9 pik7 sə¹ ɬɯ:¹ pək7 fɯət8 fɯ:t10 vɯ:t8 kuk7 kuk7 ku:5 pu:6 pu:6 ɣaw³ pɯə6 kjaw³ (Lue) sə³ ɬɯ:³ θy:¹ θy:³ ho¹ hu:¹ thu:¹ nə¹ nɯ:¹ ny:¹ kɯn² tɕau³ kɯn² kɯn² kwak8 kuə6 fa:ŋ² fa:ŋ² 6 ‘to do’ -- het7 het8 hit7 hat7 ‘ghost’ pʰi:¹ pʰi:¹ pʰi:¹ pʰi:¹ pʰej¹ ku: , kuək⁸ fa:ŋ² According to Zhang et al. (1999), languages of SZ and SWT developed from the same ancestor, Ancient Luoyue, and linguistically split further and further because of the historical migration of SWT language groups from the homeland of Luoyue (Guangxi, China) to Vietnam, Laos, Thailand, and Burma with the linguistic contact with Mon-Khmer languages, and under the influence of Pali and other Indian languages. On the other hand, SZ and NZ (which developed from Ancient Xi’ou) become an ethnic union by sharing more and more areal features under the strong influence from Sinitic languages in the late history. That is to say, if the political approach is excluded, SZ (including Tay and Nung in Vietnam and approximately equated with Li’s CT) are linguistically much more close to SWT, while NZ and Bouyei group together to form a single group which almost equals to Li’s NT. The treatment of their Zhuang and Bouyei classification is very analogous to Gedney’s Tai internal classification. Zhang and colleagues (1999: 12-13) finally distribute the Zhuang language into two major dialects Northern and Southern, and further divide NZ into eight sub-dialects, and SZ into five sub-dialects as follows. NZ includes Guibei, Liujiang, Hongshuihe, Yongbei, Youjiang, Guibian, Qiubei, and Lianshan sub-dialects, while SZ contains Yongnan, Zuojiang, Dejing, Yanguang, and Wenma sub-dialects. Nevertheless, just as Tai languages applied by ISO 639-3 codes which we have discussed in §1.1.3, Jackson and colleagues (2011: Ⅴ, 117-118) points out that varieties of the Dejing 14 The cognate for ‘above’ in Dai (Lue) and Southern Zhuang varieties means ‘north’ in Thai and Lao. 42 dialect of SZ should be assigned to at least two distinct ISO 639-3 codes, and they are [zyg] for Yang Zhuang and [zgm] for Min Zhuang. Therefore, it is perceived further research may assign more distinct languages inside Zhang and colleague’s (1999) eight sub-dialects of NZ and five sub-dialects of SZ in the future. A recapitulative work on the comparative method applied to the classification of the Tai language family has been provided by Diller (1998), as what he states, “(a) central purpose of this thesis is to distinguish and clarify which types of linguistic evidence have a core relevance to the operation of the comparative method in the Tai case and which do not – a distinction which I believe has led to misunderstanding in the past, including proposed questions of or reconstructions of the Tai language family.” (Diller 1998: 1) Diller explains how the comparative method has been applied to the cases of Tai languages, and how the Tai languages are related to non-Tai languages in the TK family or out of this family. For the relationships of the languages belonging to different families, such as Thai, Lao and Khmer, Diller gives an example of the relationship between Swedish and Finnish, which belong to Indo-European and Finno-Ugric families respectively but share many similar cultural and linguistic areal features, to demonstrate the genetic relationship among Thai/Lao and other Tai languages, and the cultural and linguistic relationship formed with Khmers by contact with each other in the history. He argues that the approach accepted of language families as valid reifications is plausible since it is only based on the comparative method, but “does not imply other claims about specifics of history, culture, race or the like” (Diller 1998: 3). He also emphasizes that the comparative method provides nodes, rather than names for nodes, such as, he suggests that Li (1977) provides a particular node of Tai language tree, rather than names “Tai” to this family since other names such as “Zhuang-Dai” also can label the same node (Diller 1998: 3-4). He explains that the linguistic reconstruction is based on the regularity, such as in Li’s (1977) work of a very firm hypothesis of PT established by showing the comparison of over 1,000 cognate forms with regular sound correspondences. However, he also points out that “there are a number of exceptions to regular sound changes evident or implicit in Li’s work”, such as the voicing alternation of proto-voiceless and proto-voiced initials among different Tai languages (Diller 1998: 7). For positioning PT, he introduces some different principles of classifications of Tai languages, such as Li’s (1977) three-way division, Luo’s (1997) possibility of a separate Northwestern branch, Haudricourt’s (1948) suggestion of restricting the term Tai to only those languages whose speakers typically refer to themselves as Tai, 43 etc. (Diller 1998: 9-10). In the contrast to some widespread Tai features, Diller points out “the certain lexical categories that one might regard as basic do not show good Pan-Tai cognates”, such as question words ‘who’, ‘what’, ‘why’, and ‘how’, and many conjunctions, prepositions, preverbal auxiliaries and common verbs like ‘speak’, ‘look at’, ‘return’ and even ‘go’ and ‘be’ (Diller 1998: 14). In Diller’s discussion, the Tai homeland problem cannot be ignored since the modern Tai languages taken as a whole share a high degree of similarity of shared lexicons and regular sound correspondences (Diller 1998: 14). Many scholars have made useful contributions to the homeland problem and Diller summarizes, “(p)hilological evidence such as how flora, fauna and agricultural practices are referred to would not rule out the region encompassing parts of the watersheds of the Xi River system in Guangxi extending to northern tributaries of the Red River in Vietnam” (Diller 1998: 15). This has surely eroded traditional support for the Altai Mountains origin or the Nan Chao theory that Tai speakers are the descendants from Altai Mountains or Yunnanese Nan Chao Kingdom (now Yunnan Province of China) in certain Thai circles (Diller 1998: 15-16). Diller’s (1998) discussion also contains other proper examples of areal traits among Tai languages. In the discussion of more ultimate Tai relationships, Diller clarifies the problems with typological comparison, such as languages from different language families can share similar structures phonetically and syntactically, even the shared vocabulary and regular sound correspondences cannot avoid facing the problem how to determine which are inherited words and which are borrowed ones since “regular sound correspondences can be the product of either borrowing or inheritance” (Diller 1998: 17-18). In the many hypotheses of Tai related classifications, Benedict’s Austro-Tai hypothesis is one of those resounding, since some basic vocabularies are shared between PT and Proto-Austronesian, but it may be disputed since Benedict (1975, 1990) presents many additional far less convincing relationships (Diller 1998: 22). However, Weera Ostapirat (2013a) defends the Austro-Tai hypothesis by “a sizable number of shared basic vocabulary items, which are justified by regular sound correspondences”. In his basic vocabulary test-list containing 24 basic vocabulary items, 14 words ‘eye’, ‘hand’, ‘nose’, ‘tongue’, ‘tooth’, ‘louse’, ‘fire’, ‘sun’, ‘water’, ‘I’, ‘thou’, ‘one’, ‘two’, and ‘die’ are determined to be cognate etyma between Austronesian and TK languages, and this confirms the Austro-Tai hypothesis. Based upon the innovation on the mergers of proto-Austro-Tai *-a: and *-ɤ: in Austronesian and the preservation of these two 44 vowels in TK, he furthermore points out that TK is a separate sister language rather than an Austronesian daughter language group. Before reviewing tonology as applied to Tai languages, it is necessary for this study to look at the study of tones in any language with common usages and general universality. The system of tone transcription employed in this thesis was introduced in Chao (1930). It consists of three sets of symbols as illustrated in the following Figure 6: straight tones circumflex tones short tones 11 131 1 13 153 2 15 242 3 22 313 4 24 315 5 Figure 6 Chao’s (1930) tone letter system In Chao’s tone letter system, the tone heights are separated into five levels, and the numeral 1 stands for the lowest pitch, while 5 stands for the highest pitch. Straight tones contain level and non-circumflex contour (falling or rising) tones. Contour tones subsequently discussed only exclude level tones like 11, 22 and all short tones in Chao’s system. This system of tonal transcription was sanctioned into International Phonetic Association (IPA) in 1989. Previously proposed studies of tones have captured some of the proposed systems of tone features. A full scheme of these studies has been concluded by Fu (1995: 2) as follows. “i. The number of tone heights: the maximum number of tone heights any language can distinguish is five (Wang 1967; Woo 1969; Anderson 1978; Maddieson 1978b; Edmondson 1992b). ii. Register bipartition: historically, voiceless syllable initials induce a higher pitch on the following vowel, while voiced syllable initials induce a lower pitch. Such consonant effect on vowels may result in a bifurcation of the tone space, known as register bipartition (Yip 1980; Clements 1981b; Hyman 1986; Bao 1990). 45 iii. Register overlapping: the upper and lower register tones often overlap in pitch (Duanmu 1990). iv. Tone-height compression: with the increase of the number of tone heights, the overall tone space tends to stay the same with the space between each tone height compressed (Pike 1948; Wang 1967). v. Tone-space expansion: a system with a larger number of tone heights tends to have a wider tone space than a system with a smaller number of tone heights (Maddieson 1970; 1978b). vi. The markedness of tone: the mid tone is the least marked (Maddieson 1970; Hyman 1986; Pulleyblank 1986). ” Besides, Fu (1995: 2-3) also provides “the construction of a new system of tone features” which “have never been incorporated in any previous system of tone features”, as the citation shown below: “i. Register tripartition: in some Southeast Asian languages, tone space can split into three registers (Haudricourt 1961; Edmondson 1990a). ii. Tonogenesis from syllable final consonants: not only can syllable-initial consonants induce pitch changes on the following vowel, syllable-final consonants can also give rise to pitch changes on the preceding vowel (Haudricourt 1954a, b). In particular, -h gives rise to a falling pitch; -ʔ gives rise to a rising pitch; sonorant ending give rise to a mid-pitch. iii. Simultaneous tone-space expansion and tone-height compression: while tonespace tends to be larger in a system with a larger number of tone heights than in a system with a small er number of tone heights, the distance between each adjacent tone height tends to be smaller (Hombert 1978b). iv....Tone merger: tone-height compression results in the reduction of the perceptual distance between adjacent tones, which may ultimately lead to the merger of originally distinct tones (Haudricourt 1961; Hombert 1978b).” Fu’s conclusion above is based on the analyses of previous studies on tones, which actually differ from each other due to different theoretical bases. Besides, many of tone systems analyzed by Fu are mainly focused on phonetic features, such as tone height, pitch, tone-height compression and tone-space expansion. However, since Tai tonal development especially tonogenesis, tonal splits, tonal mergers, and tone registers associated with consonantal elements and/or vowel length focused on in this thesis are more historical and phonological, I am not going to display the detail 46 of these different theories and to estimate which ones are more superior. Some phonetic tonal features will be only briefly concerned with when our discussion needs some parts of them to prove some specific hypotheses. The following sections present many of the previous studies on tones of languages in China and MSEA including Tai languages to expand the discussion about the process of tonal development, especially the topic that how tones have arisen in history and how they have developed after their emergence thus far, as well as some previous studies on Tai classification based on tonology which are the main subjects of the thesis. Rejecting some inferences proposed by some early linguists like Henri Maspero (1912) that tone is an inherent feature of tonal languages and cannot develop from non-tonal elements, tonogenesis is a hypothesis proposing that tone is a secondary development which induced by syllable finals, and this has been more and more popular in modern linguistics. Fu (1995) points out that Matisoff (1973) has originally referred tonogenesis to the processes “whereby tonal contrasts arise through the decaying of the consonantal system”, and “has recently been extended to include cases of tonal development from stress contrast and language contact” (Fu 1995: 61). One of the earliest hypotheses of tonogenesis is presented by Haudricourt (1954b) as shown in Table 16. He observes that Vietnamese words of category C (hỏi-ngã) frequently correspond to cognate words ending in -h or -s, and words of category B (sắc-nặng) to cognate words ending in a glottal stop -ʔ in other non-tonal MonKhmer languages. Table 16 Vietnamese tonogenesis (summarized from Haudricourt 1954a) Tonal Source -# -ʔ -h -p/ -t / -k Tonal Category A B C D Hence he argues that three tones on smooth syllables in Vietnamese were actually derived from sonorant endings, -ʔ ending and -h ending respectively, and the latter two endings dropped in history to give a rising tone and a falling tone correspondingly, while the sonorant ending remained to give an unmarked mid tone. Therefore, in history Vietnamese became a tonal language from a non-tonal language. 47 After Haudricourt’s hypothesis of tonogenesis of Vietnamese in 1954, this scheme was applied to Chinese in a number of tonology studies. Haudricourt (1954a) himself has firstly proposed that the departing tone (qù) in Old Chinese (hereafter OC) was like the Vietnamese counterpart hỏi-ngã tone (the falling tones), and was marked by an -s. Pulleyblank (1962) posits a -ʔ in earlier source of rising tone (shàng) in Middle Chinese (hereafter MC) based on analogy with Haudricourt’s hypothesis of Vietnamese tonogenesis. He moreover contributes evidence for validating Haudricourt’s *-s hypothesis, and proposes that the level tone (píng) in MC arose due to the drop of syllable finals *-ɦ and *-ð in OC. In spite of his inexplicable hypothesis of the arising of MC level tone, his MC rising tone has been agreed with by Mei (1970), based on the evidence from some modern Sinitic dialects from Wu and Min which preserve rising tones associating with syllable ending *-ʔ. Mei (1970: 86) furthermore argues that syllable endings *-ʔ and *-h had dropped off in the majority of Chinese dialects by the time of Early Middle Chinese (hereafter EMC) which refers to the system of the Qìyùn, a rime dictionary first published in 601 A.D. and followed by several revised and expanded editions, and affirms that tones arising due to the loss of these two syllable endings in EMC had manifested a phonetic shape containing pitch, contour and length. That is, contrastive tones had arisen in Chinese by the time of EMC. This conclusion is queried by Pulleyblank to be “excessively cautious” (1978: 174). He argues that these two syllable endings were preserved in the majority of Chinese dialects until the eighth century, and MC shēng categories were not tonal but quasi-tonal at most. Contradicting to Pulleyblank’s argument above, Mei (1982) convincingly provides more evidence for proving that MC tones could not have associated with the final consonants since *-ʔ must have been dropped from the northern Chinese dialects, the representative Chinese language varieties at that time, no later than 200 A.D. Besides, he argues that MC rising tone had remained stable no less than six centuries in Central Plain (Zhō ngyuán) in Northern China area, at least nine hundred years in Loyang (Luòyáng), and for more than one thousand years in Xi’an. This perspective demonstrates that Chinese tonal system has remained stable for a very long time, and challenges “the traditional point of view that tonal system is the most unstable and changeable aspect of phonology due to “its great temporal changes and wide regional variations” (Fu 1995: 63). More evidence for supporting the syllable final *-ʔ can cause a rising tone by some specific mechanism in Lahu is given by Matisoff (1970). Lahu is a language under the Loloish branch of Lolo-Burmese group of the Tibeto-Burman family. Matisoff points out that a glottal coarticulation in proto-Lolo-Burmese is the result of the 48 prefix*s- in proto-Tibeto-Burman fused with the initial consonant of the verb-root in history. By the time proto-Lolo-Burmese had merged all final stops *-p, *-t, and *-k into *-ʔ. Consequently, two ʔ’s had been created to the syllable’s initial and final respectively by these two separated processes. This was more than the phonology system could assume for Lahu, therefore a process “glottal dissimilation” called by Matisoff caused to drop the final *-ʔ and gave the syllable a rising tone. In addition, there are other cases opposing to the tonogenesis hypotheses above. For example, glottal stop is also reported to cause a pitch-lowering effect on the followed vowel in some languages like Hopi15 (Xu 1995: 64). For the hypothesis of the direct precursor of the falling tone in MC, opposing to Haudricourt’s (1954a) hypothesis (*-s > *-h > falling tone), Sagart (1986) suggests the MC departing-tone was not derived from a syllable final *-h, but from a creaky phonation. His argument is mainly based on the fact that Sanskrit words with a syllable final laryngeal spirant *-h borrowed in MC were not transcribed by words with departing tone which was expected to preserve a*-h in MC. Besides, there are no modern Sinitic dialect that has ever been reported to preserve a syllable final *-h, even though more and more evidence has suggested an earlier *-s that should have been an earlier stage of the departing-tone. He argues that after the weakening of syllable final *-s, the syllable was characterized by a high degree of glottal constriction throughout the final in OC and then MC, and this feature was actually a creaky phonation. Sagart (1988) furthermore expands this hypothesis to all the tonogenesis of the falling tone that occurred in the history of Chinese and other MSEA tonal languages. In his viewpoint, in this area languages preserving -h are all non-tonal, while tonal languages have no -h final, therefore the final -h played no tonogenetic role. He argues that the correspondence between Mon-Khmer cognates with *-h in non-Vietnamese languages and those with an original falling tone in Vietnamese cannot be taken as evidence for the origin of the falling tone. This correspondence is actually due to two different processes from *-s to modern languages. For Vietnamese it present as *-s > creaky phonation > falling tone, while for non-tonal Mon-Khmer languages it present as *-s > -h. Since “the effective mass of the vocal folds is increased, which results in a slowing down of the vibrations, in other words, a pitch fall” (Sagart 1988: 84), the creaky phonation should be the common origin of the falling tone in Chinese and MSEA languages. This has directly challenged Haudricourt’s (1954a, 1954b) hypothesis of tonogenesis 15 Hopi is an Uto-Aztecan language spoken by the Hopi people (a Pueblo group) of northeastern Arizona, USA. 49 of the falling tone. However, Fu (1995: 65-66) points out that Hopi, a language out of the East Asia and MSEA, has been given data to support the *-h hypothesis. A fuller comparison between Chinese and Vietnamese tonogenesis has been provided by Norman (1988: 52-57) based on some previous tonogenesis studies. He firstly points out that although the modern Chinese dialects are all tonal, this was not necessarily always the case, and some studies especially Haudricourt’s Vietnamese tonogenesis hypothesis have shown that some present-day tonal languages are actually derived from non-tonal ancestral languages (Norman 1988: 52). Then he raises a question of whether OC (about first millennium BC) had tone or not. Although there were some scholars argued that tone was an inherent feature of tonal languages like Chinese and could not be derived from non-tonal elements, as what is reviewed above, Haudricourt (1954b) refutes this hypothesis and shows his evidence from his hypothesis of tonogenesis of Vietnamese. This hypothesis suggests the Vietnamese is derived from non-tonal proto-Mon-Khmer language and became a tonal language by the influence from Chinese due to the abundant borrowings from Chinese. Back to the question about the tonal feature in Old Chinese, Norman gives the following discussion. Firstly, he states that “in the earliest layer of Chinese loanwords in Vietnamese the Chinese departing tone corresponds regularly to the Vietnamese hỏi-ngã (category C) tone, he (Haudricourt) proposed that the two tonal categories had the same origin – they are both the result of the loss of an original final -h, which in turn derives from a still earlier -s” (Norman 1988: 56). Secondly, he also references the rising tone precursor hypothesis by Pulleyblank (1962) reviewed above, and says that “the Old Chinese rising tone, like the corresponding sắc-nặng tones of Vietnamese, might also have derived from a glottal stop (ʔ)” (Norman 1988: 56). The píng (level) tone was the unmarked category consisting of those syllables which ended in plain vowels, and the rù (entering) tone was distinguished by its stop endings. Finally, he gives the answer as shown in the following statement: “(i)f the rising tone can be derived from such a glottal segment and the departing tone from a final -s, then Old Chinese was in effect a toneless language” (Norman 1988: 56). The correspondence between MC tonal categories and their OC tone precursors with segments pa-, pang-, and pan- as examples given by Norman are shown in the following Table 17. 50 Table 17 MC tones and their OC source (adapted from Norman 1988: 57) MC tonal categories píng (level) shǎng (rising) qù (departing) rù (entering) OC tone precursors pa[g] pa-ʔ pa-s pak pang pang-ʔ pang-s kap pan pan-ʔ pan-s pat The method of tonogenesis has also been applied to Tai language family without exception. Some scholars like Li Fang Kuei (1977: 24) believe that “it is impossible to recover the final consonants that are assumed to have been dropped” to give the arising of tone in PT period, and it is reasonable to assume that tonal categories had been done by the time of PT period, although he also mentions that tone system is suggested to have evolved “from dropping of certain final consonants” in the assumptions proposed by some ancient Chinese philologists and some modern linguists, like Vietnamese tonogenesis by Haudricourt (1954b: 69 ff.). Li furthermore suggests that if the proposal of tonogenesis is tenable, for Tai “it must refer to a stage of the language prior to Proto-Tai” and the origin of PT tones may have been in the “Pre-Tai” period (Li 1977: 24). However, in this thesis at least the process of the arising of tones in Tai should be treated as the first step of tonal development, and Tai tonogenesis has to be probed into in spite of the dispute of the period of tone emergence. The base of tonogenesis applying to Tai is that the Tai tone system shows a strong resemblance to systems found in other surrounding language families such as Chinese, Miao-Yao, Vietnamese, and etc. This has been discussed in a numerous of literatures. It is generally believed that Tai tone system *ABCD is not an accident to resemble those in Chinese (Karlgern 1964, Li 1980, Pulleyblank 1991, Baxter 1992, Liang & Zhang 1996, Pittayawat Pittayaporn 2009), and other language groups in MSEA (Haudricourt 1961), such as Vietnamese (Haudricourt 1954b), and Miao-Yao (Chang 1975)16. The correspondence of tone system among these languages historically formed from the undoubted intense language contact between those non-Sinitic languages and Chinese. The connections between Chinese and Tai languages are especially notable since their tones manifest a very robust correspondence among their shared vocabularies, which present a Chinese source 16 Notice that the conventional tone descriptions of Tai languages and other non-Tai languages above must be changed over between tone categories *B and *C, that is, Tai Tone *B corresponds to Tone *C, and Tai Tone *C corresponds to Tone *B in Chinese, Vietnamese, and Miao-Yao languages. 51 borrowed into Tai (Wulff 1934; Prapin Manomaivibool 1975; Li 1976). Pittayawat Pittayaporn (2009: 240-241) also summarizes this issue and sketches the correspondence by showing the following Table 18. Table 18 Correspondences between PT and MC tones (adapted from Pittayawat Pittayaporn 2009: 240) PT tones PT reconstructions MC glossaries MC tonal categories *A *so:ŋA ‘two’ ṣaŋ ̊ A ‘two’ (雙) Even / A tone *bwi: ‘fat’ bjwei ‘fat’ (肥) píng (平) *ha:nB ‘goose’ ŋanC ‘goose’ (雁) Departing / C tone A *B *ɣe:ŋ ‘shin’ B *C *ha:C ‘five’ ɣieŋ ‘shin’ (脛) qù (去) ŋuoB ‘five’ (五) Rising / B tone C *ma: ‘horse’ ma ‘horse’ (馬) shǎng ( *pe:tD ‘eight’ pǎtD ‘eight’ (八) Entering / D tone C *D A *ŋɯək ‘crocodile’ D B ŋǎk ‘crocodile’ (鱷) D ) rù (入) Since Tai tones and tones in Chinese, Vietnamese and Miao-Yao languages are structurally identical and show such an evident correspondence, Tai tonogenesis cannot avoid being explored after the tonogenesis hypothesis applying to these languages. Here we also focus on the arising of PT Tones *B and *C which correspond to the aforementioned Chinese/Vietnamese falling tone and rising tone respectively. Although Tone *B and Tone *C in PT have generally been characterized as plain syllables, in many modern Tai languages tone C’s (both C1 and/or C2) are actually characterized by glottalization or a creakiness addition to pitch. Li (1977: 11) has pointed out that Tone C2 in Lungchow, which is a CT dialect, is always accompanied by a glottal stop at the end of the syllable, and this is considered a special feature of the tone rather than a final consonant. Gedney (1989b, 1989c) also points out that there are a great number of CT and SWT varieties having tones developed from PT Tone *C to be characterized by glottalization. For example, in Western Nung, which is a CT dialect in Vietnam, both C1 and C2 tones have been described as having glottalized tones (Pittayawat Pittayaporn 2009: 250). In some transcriptions of Debao Zhuang, either C1 and C2 tones are characterized by glottalization, or C2 is characterized by creakiness such as lən¹²~ ‘tongue’ (Theraphan L-Thongkum 1997: 204). Since in these Tai varieties, Tone C’s with voice quality in addition to pitch display a striking parallelism to some modern dialects in Min and Wu of Sinitic languages and Vietnamese, this phenomenon suggests a commonality in tonogenetic processes having been proposed 52 in Vietnamese and Chinese. Therefore, Tone *C has been speculated in PT as a glottalized tone by Gedney (1989b: 89). As for PT Tone *B, although there are no modern Tai dialects which keep a voice quality like aspiration, or in other words a final *-h like the parallelism in tonogenetic processes proposed by Haudricourt (1954a, 1954b) and Pulleyblank (1962, 1978) in Vietnamese and/or Chinese, in the same papers Gedney (1989b: 91, 1989c: 208) speculates that it also had a final *-h mainly based on its counterpart DL tone, and on the evidence that Thai treatment of using tone B to borrow Indic loanwords ending in -ha, such as Sanskrit Loha > Thai lo:B2 ‘shield’. In the Sinitic circles, the hypothesis of Dong-Tai (Kam-Tai) tonogenesis proposed by Liang and Zhang (1996: 810-816) agrees with Gedney’s Tai tonogenesis, although they do not cite from Gedney in their discussion. Initially, they assume that the emergence of tones in Dong-Tai languages was about three thousand years ago, and before the split of Ge-Yang (Kra) and Li (Hlai) from Dong-Tai languages, there were no tones in pre-Dong-Tai languages. Then, they propose that tonal categories *ABCD in proto-Dong-Tai languages are derived from different syllable finals, namely sonorants, glottal stop (*-ʔ), pharyngeal fricative (*-h), and oral stops (*-p, *-t, *-k) respectively. It should be noticed that they follow the conventional Chinese tone ordering píng (even) shǎng (rising) qù (departing) rù (entering) to sort the corresponding *ABCD tones in Dong-Tai languages, therefore their Tones B and C respectively amount to Tones C and B in Li and Gedney’s pattern which is accounted as conventional tone descriptions in the Western circles. Moreover, they account the precursor of Tone *B was *-ʔ based on the following evidence: 1) the later forms of this tone characterized by glottalization or a creakiness in many modern Yongnan and Zuojiang Zhuang varieties of Southern Zhuang (of CT). For example, in Daqingshanxiang variety of Longzhou Zhuang the final glottal stop accompanying with T4 (Liang & Zhang’s Tone B2, or Li and Gedney’s Tone C2) is especially clearer than other dialects, so that syllables with T4 ending in finals -m, -n and -ŋ sound very like -p, -t, and -k respectively due to the assimilation with tonal concomitant -ʔ. The final -ʔ accompanying with T4 in other plain syllables with non-nasal finals are a more acoustical glottal stop, such as te:ʔ ‘to bear’, tawʔ ‘crutch’, and majʔ ‘tree/wood’ (Liang & Zhang 1996: 814); 2) Some words with checked syllable ending in a final -ʔ contrasting with -k in Lin’gao17 are cognates with words on Tone Lin’gao is also known as Ong Be (native pronunciation: [ʔ ŋ˧ e˧]), Bê, or Vo Limgao (臨高, Lin’gao in Chinese), and is a language spoken by 600,000 people in the counties of Lin’gao, Qiongshan, Chengmai, and Zhan on the north-central coast of Hainan Island of China (Liang & Zhang 17 53 B (Li and Gedney’s Tone C) in other Dong-Tai languages, such as Lin’gao keʔ7 vs. Zhuang ke3 ‘to untie, to undress’ (Liang & Zhang 1996: 814); 3) The orthography of the old Dai Le (Tai Lue) script remains two alternative conventional writing styles of Tone B (Li and Gedney’s Tone C), of which one is marked by a tone marker ้, and the other one is labeled by a final -ʔ marker ะ. For example, some words with tone marker ้ and with final -ʔ marker ะ are transcribed respectively into IPA with conventional tonal category transcriptions in Chinese circles as follows: bɛ³~pɛʔ⁸ ‘goat’, sɔ⁴~sɔʔ⁸ ‘to seek’, ja⁴~jaʔ⁸ ‘to separate’, lo⁴~loʔ⁸ ‘muddy’, ŋɤ⁴~ ŋɤʔ⁸ ‘silly’, ŋa⁴~ ŋaʔ⁸ ‘sound of crying’ (Liang & Zhang 1996: 815); 4) Some shared vocabularies between Dai (Tai) and Wa (Austroasiatic) in Yunnan Province of China have a B tone (Li and Gedney’s Tone C) in Dai and a -ʔ final in Wa respectively, such as Dai bɛ³/peʔ⁸ vs. Wa pèʔ ‘goat’, Dai la3 vs. Wa laɯʔ ‘girlfriend’, Dai sɤ3 vs. Wa s'bèʔ ‘clothes’, Dai lu⁴ vs. Wa làɯʔ ‘girlfriend’, Dai la:j4 vs. Wa kraiʔ ‘bad’, Dai fa4 vs. Wa pràiʔ ‘sky’, Dai nɤ4 vs. Wa nèʔ ‘meat’, and Dai pɛ4 vs. Wa pɛ̀ʔ ‘to overcome’ (Liang & Zhang 1996: 815). For their PT tonal category *C (Li and Gedney’s PT Tone *B), Liang and Zhang (1996: 815) also point out that although there are no modern Dong-Tai dialects preserving a final -h which they suggest to accompany with this PT tone, the final –h is suggested to exist in PT based on the evidence from some shared vocabularies between Dai and Wa, just as their discussion of the fourth piece of evidence for the final -ʔ to their Tone *B reviewed above. They list some of these words as follows: Dai pha:w⁵ vs. Wa phruah ‘to broadcast (seed)’, Dai kɔj⁵ vs. Wa kɔih ‘to draw out, to dig’, Dai thaj⁵ vs. Wa tauh ‘to redeem’, Dai bi:⁵ vs. Wa bih ‘to crack’, Dai xe:⁵ vs. Wa rɛ̀h ‘to harrow’, Dai haw⁵ vs. Wa ràuh ‘to bark’, Dai lɛ:⁶ vs. Wa lɛ̀h ‘to cut, to split, to peel’, and Dai vɛ:⁶ vs. Wa vɛ̀h ‘in passing, incidentally’ (1996: 816). Therefore, Liang and Zhang’s proposal of Dong-Tai tonogenesis totally agrees with Gedney’s (1989b: 91, 1989c: 208) speculation of the origin of four PT tonal categories, and agrees with Haudricourt and Pulleyblank’s account of tonogenesis, although their arguments are not always based on the same evidence. Their agreement on PT tone’s voice qualities is shown in the following Figure 7. 1996: 8). The Lin’gao language is classified as a single language under Tai branch of Dong-Tai languages by Liang & Zhang (1996), but is suggested to form a sub-branch of which the Tai and KamSui sub-branch is a sister under the Tai-Kadai family (Diller 2008: 7; Edmondson & Solnit 1997: 2), as shown in Figure 1 in §1.1.1. 54 Proto-Tone A B C D Proposed Reconstruction Syllables ending in a vowels or sonorant. Syllables ending in a glottal fricative *-h. Syllables ending in a glottal stop -ʔ. Syllables ending in an oral stop (-p / -t / -k). Figure 7 Voice qualities in PT tone summarized from Gedney (1989b) and Liang & Zhang (1996) On the other hand, Sagart (1988: 89) argues another interpretation of Tai tonogenesis containing the following reconstructions shown in Figure 8. Proto-Tone A B C D Proposed Reconstruction Syllables ending in a sonorant, modal voice. Syllables ending in a glottal stop, modal voice. Syllables ending in a sonorant, creaky voice. Syllables ending in an oral stop, modal voice. Figure 8 PT tone features suggested by Sagart (1988: 89) Notice that Sagart’s PT tone features are exactly opposite to his aforementioned hypothesis to Chinese and Vietnamese tonogenesis in the same paper and in his previous paper about Chinese tonogenesis (Sagart 1986). Once again, his argument of the tone features in MC and Vietnamese contains a falling tone (corresponding to PT Tone *B) with a creaky voice, and a rising tone (corresponding to PT Tone *C) with a glottal stop, but for PT he has given an opposite ordering on glottal stop and creaky voice to these two counterpart tones as shown in Figure 8. Sagart reasons that PT Tone *B usually patterned with *D on tonal splits and mergers, and it is often the case that these two tonal categories have the same contours and split conditioning in modern Tai dialects. In other words, Tones *B and *D might be in complementary distribution. This suggests that they must have ended on the same kind of syllable finals, namely stops. When comparing with tonogenesis in Vietnamese (Haudricourt 1954b), which includes that the tonal category sắc-nặng on syllables ending in an oral or glottal stop, PT*B must have ended in a glottal stop *-ʔ than a creaky voice. Hence, he argues that Thai treatment of using Tone B’s to borrow Indic loans with a long vowel ending in a final *-ha was a result of adaptation to using Tone B (with *-ʔ) to assign this final by dropping the -a, since both -h and -ʔ are laryngeal consonants. He furthermore points out that in modern Thai words ending in a glottal stop are all with a short vocalic nuclei and are all borrowed from other languages, in particular Mon-Khmer languages. Therefore “Tai may not have had syllables with short vocalic nuclei ending in a glottal stop” (Sagart 55 1988: 89) before this borrowing process, and it means the final -ʔ in modern Thai has nothing to do with the *-ʔ in tone B in PT. As for PT *C, he accounts that it must have ended in a creaky voice than a glottal stop *-ʔ by positing those final glottal stop *-ʔ as a residual feature of the creaky voice quality of PT *C in some modern Tai varieties. Querying either Gedney or Sagart’s hypotheses of PT tone features, Pittayawat Pittayaporn (2009) provides another possible scheme by proposing that PT Tone *A was again a modal tone with mid level contour, *B was a creaky tone with a lowrising contour on a relatively long vowel duration, *C was a high-falling tone ending in a glottal constriction, and *D was the same tone as *B in a checked syllable. The scheme is as shown in the following Table 19. Table 19 Phonetic characteristics of PT tones (Pittayawat Pittayaporn 2009: 271) *A *B *C *D Pitch height mid low high low Contour level rising falling rising long short creaky glottal constriction Vowel duration Voice quality modal He shows that Sagart’s viewpoint that PT *B was ending in a glottal stop could not have been the case by pointing out that a syllable with a glottal stop ending must be expected to have contrasted long and short vowels as CV:ʔ vs. CVʔ, which is no evidence for such contrast in PT (Pittayawat Pittayaporn 2009: 252). More evidence for proving that posting a laryngeal coda for PT *B by the correspondence of loanwords as proposed by Gedney and Sagart was untenable is found in modern Thai loanwords from Indic or Khmer. That is, in Thai a -ʔ has to be inserted to an open syllable to preserve the original short vowel in the borrowing language when the source is with short vocalic nuclei in the original language, such as pʰraʔ ‘monk’ in Thai is actually borrowed from brah ‘honorific prefix’ in Khmer, and this is due to dropping the -h first since Thai does not allow a final fricative (Pittayawat Pittayaporn 2009: 253). By comparing the development of PT *ABCD tones in seven representative daughter languages, Pittayawat Pittayaporn (2009) puts forward his insights as in the 56 following summaries: 1) He firstly proposes the PT tones concurrently associating with pitch height, pitch contour, vowel duration, and voice quality. 2) As for the phenomenon that Tone *D normally followed the same pattern of tonal splits and mergers with Tone *B in the majorities of Tai languages, and Tone *D further split into DL preserving the same patterns with *B and DS coming to be identified with *C instead of *B, he posits vowel duration involving in conditioning tonal behaviors, and assumes that vowel duration in Tone *B must be relatively longer than in Tone *C. 3) Because the comparison of seven daughter languages reveals that Tone *C had a higher pitch than *B, he proposes that the pitch of PT Tones *ABC was mid, low, and high respectively, while *D had no distinctive phonetic properties since it was solely on checked syllables, but occupied the lower part of the pitch range due to its variant on non-checked syllable Tone *B. 4) The contours of PT Tones *ABCD were level, rising, falling, and rising respectively due to his comparative data. Although either level or falling contours could be reconstructed for both *A and *C tones, the glottalization of *C indicates that it should have been a falling contour. Therefore, *A must have been a level contour. The left contour rising is assigned to *B because “languages that point to a rising contour is *B are spoken very far apart” to make this contour for *B seem to be a case of retention, and “both experimental typological studies have shown that vowels under rising tones tend to be longer than vowels on level and falling tones”. Therefore, Tone *D is also reconstructed as rising contour because it is on complementary distribution to *B (2009: 276). 5) As for the voice quality of PT tones, he proposes that PT Tones *ABCD had modal, creaky, final glottalized, and creaky voice qualities respectively. Since previous linguists’ hypotheses of straightforward of Tone *A are less argument, and glottalization of Tone *C is supported by many of reflexes in modern Tai languages, Pittayawat Pittayaporn’s discussion is focused on the voice quality of Tones *B and *D. His argument of creakiness in *B is firstly based on relatively longer duration on Tone *B than *C, and this is because a “final glottal constriction tends to end the syllable abruptly resulting in a shorter vowel duration” (2009: 278), and a creakiness extends “over a relatively long period of the duration of the vowel” (2009: 277). Additional support for creakiness of *B is from some voice qualities accompanying with Tone B’s in several modern SWT and CT varieties. Even though many of them are described as glottalization, Pittayawat Pittayaporn believes that the voice qualities may actually be creakiness since at least one of the descriptions Nung Fan Slihng from Freiberger (1964) specifically mentions that the Tone B2 “ends with a final glottal stop, but laryngealization may occur over all the voiced phonemes”. Pittayawat Pittayaporn claims that this laryngealization is actually “a remnant of the 57 creaky voice characteristic of PT *B” (2009: 278-280). The last evidence is the correspondence between PT *B and MC departing tone which is proposed to derived from OC -s or -h finals and became creakiness during the MC period (2009: 280281). As for *D, he proposes that it had no voice quality of the tone itself, but the vowel in syllables with Tone *D had a glottalization before final stops like what has been discussed by Hombert and colleagues (1979), Mazaudon (1977), and Thurgood (2002), and this “glottalization in *D resembled the creakiness characteristics of *B” (2009: 281-282). In addition, a hypothesis that Tai tonogenesis is related to the loss of the stress of Austronesian languages has been provided by Ni (1990). He suggests that Dong-Tai (TK) languages are genetically related to Austronesian languages. On this premise, Dong-Tai languages’ tones A corresponds to stressed syllables in Austronesian languages, like Tai / Zhuang ta1 vs. Indonesian mata ‘eye’, and Tai / Zhuang tu1 vs. Indonesian pintu ‘door’; Dong-Tai language’s Tones BC correspond to unstressed syllables like Dong-Tai ha:n5 vs. Indonesian gaŋsa ‘goose’, and Dong-Tai’s tone D corresponds to the same check syllables in Austronesian languages (1990: 262). He argues that Dong-Tai tones arose due to the language structure change from multisyllable to monosyllable due to the language contact with Sinitic languages in history, based on a typical example that five tones in an Austronesian language Sanya Hui18 arose due to the strong influence from surrounding languages Chinese and Hlai. This hypothesis of Tai tonogenesis is very different from other scholars’ that tone occurred due to the loss of final consonants or voice qualities. A comprehensive summary of the tonogenesis hypothesis has been concluded by Fu (1995: 67) as below: “… there have been considerable disagreements in the literature regarding the hypothesis of tonogenesis. For example, the time of tone emergence in Chinese has been suggested to be as early as in the eleventh century B.C., or as late as the ninth century A.D., with a difference of two thousand years. The source of the Departing tone has been posited as -h in some studies, but as creaky phonation in others. Glottal stop is known to have conditioned the development of rising tone in Asian languages, but falling tone in American languages. Creaky voice is found to correlate with high pitch in languages Sanya Hui is also known as Tsat, Utsat, Utset, Huíhuī, or Hainan Cham, and is a language spoken by 4,500 people in Yanglan and Huixin, two villages on the outskirts of near Sanya City in Hainan Island, China by the Utsuls. Tsat is a member of the Malayo-Polynesian group within the Austronesian language family, and is one of the Chamic languages originating on the coast of present-day Vietnam (Edmondson 1993). 18 58 like Burmese, but with low pitch in languages like Mam. The falling pitch is believed to have been conditioned by -h, -ɦ, -ʔ, creaky voice, voiceless sonorants, semi-vowels, and breathy voice. But in spite of the disagreements regarding the exact TIME and CONDITION of tonogenesis, the PATTERN of Haudricourt’s original three-tone schema (1954a), i.e., rising, level and falling, is left intact.” After the emergence of tone, tonal splits associating with tonal mergers have been acting as a main role on the stage of tonal development. Tonal splits have received much more attention in the literature than tonal mergers, as what Fu (1995: 85) points out that since the patterns of split are more clear and obvious due to its conditions of register, and in contrast “both the conditions and the patterns of tone merger are much more difficult to pinpoint”. Sometimes it is difficult to find out the evidence to prove whether a seeming tonal merger is a real merger or lack of split. But in spite of the difficulties, some patterns of tonal mergers have been explored. In this sub-section the reviews are going to be focused on the patterns of tonal splits, and sometimes also involving with the discussion of tonal mergers. Tonal splits are usually treated as tonal registers. For the relationship between register and tone feature, Fu (1995: 52) has pointed out that “(i)n terms of feature definition, they are in a hierarchical relationship, register dominating tone” to indicate this essence. Taking tonal splits after Vietnamese tonogenetic processes proposed by Haudricourt (1954a) as an example, the three tones rising, level and falling arose from different syllable finals to represent the tone space before register split. The influence from different types of initials had given register distinctions on the tone, that is, voiceless initial gave the upper register, while voiced initial gave the lower register. Later, the merger of voiceless and voiced initial produced the two contrasting registers to each of the original tonal categories. Therefore, the essence of tonal splits is actually registers. Just as what Norman states, “Vietnamese tones can be plotted using the same scheme as that used for Chinese. Tonal categories A, B and C correspond to the traditional Chinese categories píng, shǎng and qù respectively. Category D, like the Chinese rù tone, consists of all syllables which ended in a stop” (1988: 55), the modern Vietnamese has eight tonal categories from Chinese terms, and these tones correspond to MC tones which will be discussed subsequently. This is shown in the following Table 20. 59 Table 20 Vietnamese tones (adapted from Norman 1988: 55) Register Tonal category A B C D upper 1. ngang 3. sắc 5. hỏi 7. sắc lower 2. huyền 4. nặng 6. ngã 8. nặng Next concern is Chinese tonal splits. Since at least the time of the Qièyùn dictionary (compiled by L̀ Fǎyán in AD 601), the source on which the reconstruction of MC is based, “Middle Chinese possessed four tonal categories” (Norman 1988: 52), it is necessary to examine the situation of how these four tonal categories developed into LMC and Modern Chinese dialects. From a description of a Chinese dialect recorded by a Japanese monk Annen in the ninth century, it is certain that at least by the late Tang (AD 923-936), the original tonal categories of Middle Chinese had split into two registers (Mei 1970), as shown in Table 21. Table 21 The MC eight tonal categories (adapted from Norman 1988: 54) Initial class Voiceless Voiced Tonal category pi ́ng shǎng qù rù 1. yīn píng 3. yīn shǎng 5. yīn qù 7. yīn rù ‘upper level’ ‘upper rising’ ‘upper ‘upper 2. yáng píng 4. yáng shǎng 6. yáng qù 8. yáng rù ‘lower level’ ‘lower rising’ ‘lower ‘lower departing’ departing’ entering’ entering’ From Qièyùn, the four tonal categories of the MC were defined as below: píng: ‘level and non-abrupt’, shǎng: ‘rising and non-abrupt’, qù: ‘(probably) falling and non- abrupt’, and rù: ‘abrupt (ending in a stop -p / -t / -k)’ (Norman 1988: 52). For the development of tone from MC to Modern Chinese dialects, “(p)robably the single most important factor in the development of the Qièyùn tonal categories to those of modern dialects was the development of tonal register’ (Norman 1988: 53), when the voiceless and voiced initials became merger. Basically, voiced initials condition a lower pitch and voiceless initials condition a higher pitch. At this stage, a new eight-term tonal system was created from the tonal splits from four-term tonal categories of MC. However, the tonal split conditioned by sonorant initials (nasals, laterals and semivowels) sometimes goes with voiced obstruents but sometimes goes with voiceless obstruents, depending on the tonal category. The 60 more complicating factor is that today only a few modern Sinitic dialects preserve this eight-term system intact, and most modern dialects “have merged one or more of these categories, more rarely, one of the categories has split” (Norman 1988: 54). For example, in Norman’s comparison of twelve Sinitic varieties from seven main sub-groups of Chinese, eight of them have merged the most register splits and have only preserved register distinction in the píng tonal category (1988: 182). Most of Mandarin dialects have lost the final stops in MC, consequently many of them have merged tone rù (entering, which have been accompanying with final stops) into other tonal categories, or have developed it as a distinct tonal category without a final stop, and several dialects merge all final stops into a single -ʔ to retain a distinct tonal category. The Beijing (Peking in Norman’s description) dialect, the basic dialect of Standard Mandarin, presents a tonal pattern as follows: it retains the split of píng tone as yīn píng and yáng píng two registers due to the initial voicing in MC; the shǎng tone shows no register distinction, and words with voiceless obstruent initials and sonorant initials retain a single shǎng category, while those with voiced obstruent initials of this tone in MC have merged into qù tone; the qù tone has totally merged two registers in history, and the rù tone with final stops in MC has dropped all the stops and has merged into other four tones (yīn píng, yáng píng, shǎng, and qù) in a distribution with only partly predictable (1988: 193-195). As for the Guǎngzhō u variety, which is treated as the standard form of Cantonese (Yuè dialect of Chinese in Norman’s description), it preserves the MC tones as best since it shows the most conservation of the MC final consonants with “the only merest handful of exception”, and the upper register rù in Guǎngzhō u even splits “into two subtypes conditioned by phonetic vowel length” to make this dialect have nine tonal categories which is rarely found in other Sinitic languages (Norman 1988: 217-218). A scheme of Chinese tonal development summarized from Norman (1988) can be drawn out as shown in the following Figure 9, and it reasonably explains how Chinese has developed from a non-tonal proto-language to well-developed tonal daughter languages during at least three thousand years. 61 Stage 1: Old Chinese (first millennium BC): no tones but four phonetic syllable contrasts vl initial plain vd initial vowels vowels-ʔ vowels-s vowels-p/-t/-k Stage 2: Early Middle Chinese: phonemic four tones vl initial vd initial píng shǎng qù rù Stage 3: Tonal splits to phonemic eight tones in Late Middle Chinese Initial voicing mergers yīn píng yīn shǎng yīn qù yīn rù yáng píng yáng shǎng yáng qù yáng rù Stage 4: Modern Chinese: tones preserved, or merged, or split from 8-tone system. e.g. Guǎngzhō u (Standard Cantonese) Preservation & split of MC tones yīn píng yīn shǎng yīn qù yáng píng yáng shǎng yáng qù split > xià yīn rù shàng yīn rù yáng rù e.g. Běijīng (Standard Mandarin): Preservation & mergers of MC tones yīn píng yáng píng Merger Merger > shǎng >qù Merger > one of other four tones Figure 9 Chinese tonal development scheme (summarized from Norman 1988: 52 ff.) As for the tonal split in Tai languages, we have already reviewed the basic situation of it in §1.2.1. Here we only have a briefly reminding. Just like what Li Fang Kuei states that “from a comparative study of the tones of the modern dialects, it is apparent that the opposition of voicing and voicelessness of the initial consonant influences practically all the tones in all dialect” (Li 1977: 25), Tai tonal splits shown in Table 1 in §1.2.1 and the remained discussion have presented a robust comparability with the case of Chinese and Vietnamese tonal splits. With a detailed description of the tonal systems of PT tones (*ABCD), Gedney (1989[1972]) points out that after the period of PT unity, a wave of drastic sound changes swept this entire area which includes various new locations groups of Tai speakers had dispersed to. In Tai family, “these sound changes involved splits in the tonal system, 62 with the splits conditioned by the phonetic nature of initial consonants of the syllable” (1989[1972]: 196). The details of these changes differed from place to place, but in general “each of the original tones A, B, C, and D split into two or more tones, with, in some dialects, various coalescences of the resulting tones with other tones from the sources in the previous pattern” (1989[1972]: 196). For the checked syllable, tonal splits are not only conditioned by initials but also conditioned by vowel length. The Proto-Tai tones which were at time of split are described by Li (1977) as Table 1 in §1.2.1. Gedney gives an example of tonal splits and mergers in Thai for briefly indicating the common situation of Tai case. Test words of Thai applying to this proto-tone box are as below (1989[1972]: 197, 199): Column A: Box 1. pii ‘year’, naa ‘thick’; Box 2. phii ‘fat’; naa ‘ricefield’; Column B: Box 3. pii ‘flute’, naa ‘arrow of crossbow’; Box 4. phii ‘older sibling’ Column C: Box 5. naa ‘face’, maa ‘to increase’; Box 6. maa ‘horse’; Column D-short: Box 7. lak ‘stake’, mat ‘flea’; Box 8. lak ‘to steal’, mat ‘to tie up’; Column D-long: Box 9. maak ‘fruit’; Box 10. maak ‘classifier for tools’. For the tones in smooth syllables (A, B, and C), as a result of tonal splits, six-tone systems are very frequent throughout the modern Tai dialects speaking domain, while many modern dialects shows irregular splits and mergers patterns. For example, tonal splits in Thai are shown in Table 22. Table 22 Register tonal splits of Thai (adapted from Gedney 1989[1972]: 201) A Initials Friction 5th tone at time Voiceless 1st tone B C D-short D-long 2nd tone 3rd tone 2nd tone 2nd tone 3rd tone 4th tone 4th tone 3rd tone of split Voiced In the table above, we see that in Thai there is an imbalanced tonal splits between original Tone A and other tones, and lower register of Tone B and the upper register of Tone C have merged with each other. Although the special split that Thai has made in Column A is not common, many known dialects have made a split of this sort in one or another of the other columns of the diagram, and this is why it requires a test diagram which is to allow for a maximum of possibilities, include this type of split (Gedney 1989[1972]: 201). As a result of comparison the tonal splits and mergers of many modern Tai languages, the Tai tone box designed by Gedney (1989[1972]: 202) consists of five tone categories (A, B, C, DS, DL) with four groups of initial consonants conditioning possible tone splits, as what has been shown in 63 Table 4 in §1.2.1. As for some specific tonal splits in CT and NT languages, Liao and Shen (2012) have provided a revised Tai tone box for solving the noticeable shortcomings of Gedney’s original tone box when applying to Tai languages beyond SWT, as what has been discussed in detail in §1.2.2, and has been shown in Table 7 in the same sub-section. Just as what tonal splits presenting in Vietnamese, Chinese and Tai reviewed above, two-way register split which refers to “split conditioned by the simple opposition between voiced and voiceless initial consonants” (Fu 1995: 80), has been known for a long time. However, later studies have shown three-way register split, a tripartition of tonal split, can possibly occur in some languages even it is relatively rare and less well-known. A more famous three-way register split among preglottalized vs. aspirated vs. voiced sounds in Kam has been presented by Haudricourt (1961, 1972), and here we only cite the comparison in tonal category A as shown in the following Table 23. Table 23 Three-way register split in A tone in Kam (adapted from Haudricourt 1972: 68)19 Category A Sui Mak Glottalized ʔnaa 11 ʔbaaŋ ʔdaai Aspirated 11 11 hmaa11 naa ʔbaaŋ ʔdaai 24 24 hai 13 mya mii 13 lee11 Raan11 hnai 11 Glossary naa⁵⁵ ‘thick’ laai⁵⁵ ‘well’ maaŋ⁵⁵ maa³⁵ maa 11 Kam nwaa³⁵ maa13 13 -Voiced 24 ai³⁵ ‘thin’ ‘dog’ ‘come’ ‘open’ mya ‘hand’ lee13 lee212 ‘writing’ žaan13 yaan212 ‘house’ 212 Just like what Fu (1995: 80-81) points out that “the conditioning factor is not just limited to the simple voiced: voiceless opposition”, tonal splits can also be conditioned by other pairs of oppositional initial sounds after the tonal splits of the original two-way register. For example, some Miao dialects present a secondary tonal split between the pair of voiceless unaspirated vs. voiceless aspirated sounds, and pre-nasalized stops/affricates vs. stops/affricates to make a three-way register split among these pairs and the original voiced register (Li et al. 1959). Furthermore, 19 IPA symbols in this table are all adopted from Haudricourt’s (1972: 68). There may be some imprecise symbols, such as R- in Raan ‘house’. The initial of ai35 ‘open’ in Kam may be ʔ- which may have been omitted in Haudricourt’s transcription. 64 Pulleyblank (1978) has also argued a triple-split case from LMC to EMC. Wujiang variety of the Wu dialect of Chinese has been found to have a typical three-way register split by several scholars (Chao 1928; Yuan 1960; Ye 1983; Shi 1992). As shown in Table 23, although Kam has lost the pre-glottalized element in other sister-languages, and has merged voiceless nasals20 to voiced nasals, the tonal split conditioned by these two contrastive original voiceless initial registers is clearly presented. The Kam tonal split not only demonstrates that register splits in initial consonants can be conditioned from the same features that have conditioned tonogenesis, such as glottalization (-ʔ) and aspiration (-h), but also presents that these features condition differently when they are in the final position and in the initial position. Comparing with the voiceless aspiration in syllable-final position giving a falling tone (lowest pitch) in the tonogenesis process, this feature in syllable-initial position does not condition a lowest pitch in the tonal register split process at least in Kam. A summary of Kam three-way split is provided by Haudricourt (1961, 1972) as shown in the following Table 24. Table 24 Three-way register splits in Kam (adapted from Haudricourt 1972: 70)21 Tone Initial changes and retentions A B C ˀb- > m-, ˀm- > m-, ˀn- > n-, ˀd- > l-, ˀR- > y-, k-, t-, p- 55 53 323 hm- > m-, hn- > n-, hn̄- > ñy-, hṅ- > ø-, kh-, th-, ph- 35 453 13 m-, n-, ñ-, l-, b->p-, d->t-, g->k-, R->y- 212 33 31 Edmondson (1990) explains the Kam three-way register split in a different way. He argues that the triple division of the tonal splits in Kam was actually not a simultaneous tripartition, but two temporarily distinct bipartitions containing a primary split preceding a secondary split. He furthermore suggests the cause of the secondary tonal split was not aspiration but breathy voice. In spite of the real cause of the secondary split being aspiration or breathy voice, the most notable exploration is that three-way register split may be a tripartition resulting from two successive double split. 20 In Kam, the tonal split of original voiceless nasal goes with aspirated sounds, and is treated as a kind of aspirated sound in the tonal development process. 21 IPA symbols in this table are all adopted from Haudricourt’s (1972: 70). There may be some imprecise symbols, such as hn̄-. The tonal values in Haudricourt’s text are 5-level pitch tone letters, and here I transliterate to the Chao tone numerals based on the Chao 5-level pitch scale (Chao 1930). 65 In spite of the different views concerning the issue that three-way register split is a simultaneous tripartite split or two successive double splits, Fu (1995: 82) states that “tripartition as a current state-of-affairs has been confirmed in a number of tonal systems”. One thing is clearly pointed out by him that “secondary split occurs only in one of the register produced by the primary split” (1995: 82), such as in Nanchang the lower register further splits into two but no further split occurs in the upper register, but in Miao, Kam, and Songling-Tongli the opposite situation is shown as that the secondary split only occurs in the upper register but is never found in the lower register. That is to say, “no four-way split occurs” (1995: 83). Fu furthermore points out the counterexample in tone on checked syllables, and this time seeming four-way splits occur. He cites some previous studies on check tones in some Cantonese dialects in Guangxi like Hengxian (Bi 1982), Yulin and Bobai (Yang et al. 1985) to indicate this issue. The example of Hengxian is a typical example as shown in the following Figure 10. DIa kəp DIb 44 tɕət44 hək 44 ‘quality’ ‘black’ ‘attack’ kik44 kuk ‘urgent’ 44 ‘valley’ DIIa ka:p 33 tɕa:t33 ha:k 33 ke:k33 kɔ:k 33 ‘first’ ‘prick’ ‘guest’ ‘foot’ DIIb tɕuk 11 tɕik11 muk 11 ləp11 ‘country’ fət 11 ‘yesterday’ tɕɔ:k22 ‘muddy’ ‘eye’ ‘no’ ‘straight’ ‘stand’ ‘Buddha’ tɕe:k22 mɔ:k 22 la:p22 fa:t 22 ‘touch’ ‘bacon’ ‘punish’ Figure 10 Hengxian checked tones (Bi 1982: 21 cited in Fu 1995: 83) In this language, the earlier two-way register split on checked syllable has split in four ways due to the vocalic durations. The upper register (DI tone) and the lower register (DII tone) co-occurs with short vowels are labelled as DIa and DIIa respectively, while they are labelled as DIb and DIIb when co-occuring with long vowels. Fu (1995: 83) cites a tentative explanation from Hirayama (1987: 19) to suggest that the two different but pitch-closed tones splitting from the same register due to the vowel duration are allotones to one another more than being two distinct tones. He furthermore argues that this kind of “vowel-duration-induced tone split” cannot constitute a counterexample to the proposition that “the maximum number of register splits is three”, since this kind of tonal splits can be treated as splits within a single register, and does not result in register split just like the “syllablefinal-induced tonogenesis” (1995: 84). The last issue of tonal splits concerns the explanation of the “voiceless-low” and “voiced-high” correlation, which refers to that initial voiceless consonants correlate 66 with low tones, while initial voiced consonants correlate with high tones, and violates the normal situation of “voiceless-high” and “voiced-low” register split reviewed above. Fu (1995: 81) has cited Maddieson (1978a) to point out that the correlation of “voiceless-low” and “voiced-high” is found in some languages and “is not unusual cross-linguistically”. He furthermore cited Brown (1975) to give data to show that the situation of “voiced-high” can be found in 75% of tone data from Thai, Northern Thai, Lao, Shan, Phuan, and Phu Thai of SWT languages. Brown’s own explanation for this phenomenon is based on the “different strategies listeners use to process linguistic signals” (Fu 1995: 82). Brown argues that in proto-languages the syllables with a voiced initial presented a lower pitch on initial consonants but higher pitch on the rest following part, while syllables with a voiceless initial presented an opposite situation that presented a higher pitch on initial but a lower pitch on remaining part. This situation has allowed two opposite directions of tone development based on the concentration being initial or median portions of the syllables due to different fashions and prestige. Therefore, either “voiced-low” or “voiced-high” languages were developed by different linguistic communities. For Tai languages, Brown suggests that “voiced-high” splits antedated “voiced-low” splits by 400 years. A more compellent explanation is that “voiced-high” phenomenon is tonal flip-flop suggested by many scholars like Wang (1967, 1969), Matisoff (1973), Hashimoto (1972), and Hashimoto (1986). Just like what Fu (1995: 85) states that “(i)t has been pointed out by many scholars that tonal systems have a life of their own; once established, they develop without regarding to their etymological values (Haudricourt 1961; Hombert 1978), and there seem to be no guide-lines for development after tone split, with different dialects changing in different directions (Henderson 1982)”, at the time of register-splits tone pitch always started from “voiced-low”, but the later development may change from high to low and low to high after the establishment of tones. Edmondson (1994: 164) applies tonal flip-flop to “voiced-high” phenomenon in SWT languages instead of the explanation from Brown (1975) above. He turns over a theme about the different tonology among NT (NZ), SWT, and CT (SZ). In this case, “voiced-low” principle, “which basically states that original voiceless initial consonants conditioned low tones and original voiceless initial consonants conditioned high tones” (Edmondson 1994: 164), becomes a distinctive feature. In the vast majority of NZ and SZ, the historically high tone set generally possesses a higher pitch than the historically low tone set, 67 while in SWT, “this situation is sometimes obscured by subsequent tonal flip-flop of highs and lows”, and becomes a characteristic for SWT. Though there has no tonal flip-flop in CT, there is a “slight tendency to tone reversal in the extreme southwestern locations” of SZ area and this also becomes a distinguishing feature from NZ (Edmondson 1994: 164). The other three explanations for “voiced-high” correlation are also briefly summarized by Fu (1995: 82) as follows. 1) “consonants had changed before register split”, such as that the initial b- had changed to an implosive ɓ- before tonal split suggested by Hombert and colleagues (1979). 2) “‘voiceless-low’ and ‘voiced-high’ are regarded as resulting from the effects of tone on consonant”, like what Maddieson (1978a) shows that besides the initial consonants, syllable-final voicing contrast, places or manners of articulation, and the insertion of consonantal segments “may all be attributable to tonal effects”. 3) The last explanation is “by posting more consonant types for the phonological system before register split”, and Fu also cites proto-Min’s situation that a series of voiceless sonorants caused sonorant initials in modern Min dialects to correlate with upper register tones suggested by Norman (1973, 1974); this is quite similar to that the voiceless sonorants have given upper register tones in Tai languages reviewed above. As for tonal mergers, one thing need to be formatted is the pattern of merger. Fu (1995: 85) cites the patterns of tonal merger from Pan (1982) and Lien (1986), and they are shown in the following Figure 11. In Pan’s figure of tonal merger pattern, Type A and Type B are called “vertical merger” and “horizontal merger” respectively by Lien (1986), and they represent mergers between the upper and lower registers as well as mergers across tonal categories within the same register respectively. Type C refers to mergers across both tonal categories and upper-lower registers. Pan points out that Types A and C are not found in languages where voicing distinction in initial obstruents is still preserved, and Type B is not found in languages preserving obstruent endings. A B C D yin IA IB IC ID yang IIA IIB IIC IID Type A merger Type B merger Type C merger Figure 11 Type of tonal mergers (Pan 1982) In addition, Fu (1995: 86) has reminded that “(a) true Type A merger must be carefully distinguished from lack of split”, but a true Type A merger and lack of split 68 is hard to judge. However, Fu (1995: 88) also points out that “the more tones split, the more likely merger would happen, especially mergers of Types B and C”. In the Thai tonal patterns shown in Table 22 above, tones lower B and upper C which should have been two distinct tones have merged as what Type C refers to. The identification of the conditioning factors of register split is concerned when talking about the tonal splits. In spite of some phenomenon of tonal flip-flop mentioned in §2.2.2, the traditional “voiced-low” and “voiceless-high” principle has been verified in a great number of cross-linguistic phonetic studies. However, just like what we have reviewed in the three-way register splits and Gedney’s analysis of Thai tonal splits in §2.2.2, more specific phonetic factors can also cause tonal splits. Therefore, it is necessary to review the previous studies on the conditioning factors of tonal splits, especially those factors beyond the simple voiceless-voiced opposition. The following contents centre on several phonation types, namely the aspiration, unaspiration, glottalization, and breathy voice, to expand the discussion. The first issue concerns the traditional voiceless frictions, and we focus on the situation in Tai languages since other languages such as Kam, Miao and Chinese dialects have been discussed on this issue in §2.2.2. Voiceless frictions have been further divided into several phonation types by Li (1977: 28-29). As the point of penetration, his description of the splits in Tone A in Thai is with a consistent explanation: A1 can be mid level (33) or low-rising (24), depends on the different initial features, and A2 is mid level (33). Furthermore, an example of the comparison among tonal development in different dialects is given, that is, A1 tone (A upper register) in Thai, Lungchow and Po-ai shows three different types of tonal development. Thai divides two tones between the line of aspiration and other types, Po-ai divides two tones between the line of glottalization and other types, and Lungchow does not divide tones in A1 (Li 1977: 29). Thus, voiceless initials at least can be treated as three phonation types by the comparison among these three languages: unaspiration, aspiration, and glottalization. In Li’s discussion of the relationship between initials and tonal development, proto-initials can be divided into three laryngeal types: voicing, glottalization, and aspiration (Li 1977: 26-27). Altogether, Li finds “the Proto-Tai initials may be roughly divided into five groups according to the influence they may have on tone” (Li 1977: 43) and lists 1) aspirated voiceless stops, 2) voiceless continuants, 3) unaspirated voiceless stops, 4) glottalized consonants, and 5) voiced consonants. Li’s five groups of initial can 69 actually serve for wider range of Tai varieties than Gedney’s Tai tone box, which only divides PT initials into four groups (cf. §1.2.1 and §1.2.2). In the progress of tonal development of the Zhuang languages, Zhang (1980), Liang and Zhang (1996: 818-828) and Zhang and colleagues (1999: 243 ff.) also assume that several specific laryngeal types of initial consonants have conditioned secondary tonal splits after the primary voiceless-voiced register split. This presents a temporal sequence of tonal developing event. Register splits firstly conditioned by the initial voicing gave eight tones from original four proto-tones. After that different laryngeal features of voiceless initials caused different secondary tonal splits and mergers in nearly half of Zhuang dialects. The processes of secondary tonal splits in different dialects can be divided into four types, which will be enumerated in the discussions of the determinations of the diachronic secondary tonal splits (cf. §4.3.1). In addition, Liang and Zhang (1996: 826-828) also analyze similar secondary tonal development conditioned by these specific phonation types of initials in other TK languages like Yanghuang (Then), Lakjia, Lin’gao, and Biao. However, some conditioning factors of secondary tonal development in these languages are different from those in Tai languages. For example, in Yanghuang the continuants (nasals, sonorants, and semivowels) development from proto-voiceless continuants have conditioned some of the words from odd tone (upper register) to merge with even tone (lower register), such as mjɛ:ŋ² ‘irrigation canals and ditches’, mje² ‘bear’, and ra² ‘thick’ with a lower register tone corresponding to an upper register tone in other sister-languages of TK family. In Lakjia, initials containing continuants like semivowels, labialized stops (stops + -w-) or consonant clusters, as well as some voiceless unaspirated stops also condition some words from an odd tone to an even tone, like jau² ‘green’, jum² ‘amaranth’, com² ‘hair’, kou² ‘horn’, pliŋ² ‘leech’, lɔŋ² ‘griddle’, ka:m² ‘to ask’, wa:n ‘sowing’, hwai⁶ ‘new’, and ja² ‘rice field’, which normally have an even tone in other sister-languages of the same family. The conditioning factor which is a continuant (from a PT voiceless-continuant) causing a secondary tonal split is not found in Tai languages at least in the previous studies. In short, Zhang (1980), Liang and Zhang (1996), and Zhang and colleagues (1999) also divide the initials which have influence on secondary tonal development into four groups, namely glottalization, aspiration, unaspiration, and the remaining continuants (from PT voiceless continuants) which is not found to condition secondary tonal development. That is to say, the PT voiceless continuants tend to mostly keep its tonal behaviors as in its primary tonal splits. This is somehow related 70 to but different from the suggestion from Theraphan L-Thongkum (1997) that “the merging of proto-voiceless sonorants with proto-voiced sonorants is the cause of primary tone split in Tho” (1997: 215) when she concludes her findings of tonal splits of Dai Tho and Tai Tho discussed in the following §2.2.4. The division of initial groups from these scholars above agrees with the four groups of PT voiceless initials divided by Li (1970) reviewed above. However, the noticeable problem mentioned in §1.2.3 that PT might not have aspirated sounds also reminds us to carefully avoid using the term “proto-initial” when dividing these initial consonants which are conditioning factors in the processes of tonal development. Besides, aspiration in relation to pitch lowering in some historical and dialectal studies (Haudricourt 1961, Brown 1975, Henderson 1982, Ye 1983) also presents a more complicated effect on tones, although phonetic studies have not yet confirmed that aspiration has a defining pitch-lower effect (Hombert 1976). The next issue concerns the original voiced initials or lower register. Although both plain and breathy voiced initials are assumed to give a pitch lowering effect, Fu (1995: 81) has pointed out that “recent studies tend to attribute the real cause of pitch lowering to breathy voice rather than plain voice”, such as the discussions in Pulleyblank (1978), Shi (1981) and Edmondson (1990). This hypothesis is conflictive with the suggestion that there was a contrast between breathy initials (or voiced aspirated, as well as a series of special voiced initial) and plain voiced initials in PT proposed by Gedney (1985), Li (1989), Liang and Zhang (1996), Zhang and colleagues (1999) and Liao and Shen (2012). The breathy voice suggested here is also commonly addressed as “voicing alternation”, which has been discussed in §1.2.2 and §1.2.3 as a discrepancy generally described to have disagreement between SWT/CT on one hand and NT on the other. For the vast majority of TK languages, breathy voice is not found, and it is just suggested to have had in history by the aforementioned scholars to explain the “voicing alternation” or “register alternation” among different TK daughter languages. We have focused on this hypothesis applying to Tai languages in §1.2.2 and §1.2.3, and this time Kam-Sui languages as another example to be given by Weera Ostapirat (1994). He also discusses about a hypothesis that breathy sounds (sounds with a ɦ-) in proto Kam-Sui (hereafter PKS) conditioned the different tone developments into different modern Kam-Sui languages. Since there are no breathy sounds in modern Kam-Sui languages and the data for reconstructing these kinds of breathy sounds from modern Kam-Sui languages are still not enough, Weera Ostapirat states “(t)he proposal here is not, however, a definite treatment but rather 71 a call for discussion” (Weera Ostapirat 1994: 79). The main evidence that proves there were breathy sounds in Kam-Sui is the irregularity and non-straightforward development of tones (sometimes including initial features) of PKS voiced nasal *N(N represents nasals in general). For example, Sui has nam³ while Kam has nam⁴ ‘water’ (Weera Ostapirat 1994: 79). This kind of evidence distinguishes two groups of modern Kam-Sui languages as shown in the following Table 25. Table 25 Tonal correspondence of nasals between two groups of Kam-Sui languages (adapted from Weera Ostapirat 1994: 79) Group 1 Gloss Kam Mulam Group 2 Then Sui Mak Maonan Maonan (Xianan) (Shangnan) ‘bamboo na:ŋ² na:ŋ² --- na:ŋ¹ na:ŋ¹ na:ŋ¹ n̥a:ŋ¹ ‘water’ nam⁴ nəm⁴ nam⁴ nam³ nam³ nam³ --- na:m⁵ n̥am³ --- --- --- shoot’ ‘mud’ ‘hand’ na:m⁶ na:m⁶ mja² nja² mja² mja¹ mii¹ --- --- ‘ghost, --- --- ma:ŋ² ma:ŋ¹ --- ma:ŋ¹ m̥ a:ŋ¹ ‘fog’ mun² --- --- --- --- mu:n¹ m̥ u:n¹ spirit’ Because in most situations “the reflexes of PKS voiced nasals *N- are very regular and straightforward, showing plain or voiced nasals with low tones in all daughter languages” (Weera Ostapirat 1994: 79), the irregularities of the *N- in some examples of two groups of PKS’s daughter languages in the table above show that the initials of these examples may not belong to the common *N-, and Weera Ostapirat proposes for these sounds the breathy nasals (*ɦN-)’ (Weera Ostapirat 1994: 80). Weera Ostapirat explains how *ɦN- develops into either N- or voiceless hN- by the following splits: breathiness = voiced (vocal cords vibrating) + aspirated (arytenoids apart), and it ‘then can be assumed to be either voiced (low) or voiceless (high) as to assigning tones’ (Weera Ostapirat 1994:80). Thus, Weera Ostapirat adds breathy nasals as an independent group of nasals into PKS, and demonstrates four different groups of nasals as shown in the following Table 26. 72 Table 26 Four groups of Kam-Sui nasals proposed by Weera Ostapirat (1994: 80) Sui Maonan Mak Kam Mulam Then *ØN- N-31 N-231 N-31 N-11 N-121 N-35 *ɦN- N-11 N-42 N-13 N-11 N-121 N-35 N̥ -11 N-42 N-13 N-35 N-13 ʔN-42 N̥ -42 N-24 N-55 N-42 N-13 *hN*ʔN- ʔN-11 Furthermore, Weera Ostapirat points out that there are not only nasals but also lateral and approximants have the same “tone alternation” between two groups (Group 1 and Group 2 in Table 25) of the daughter languages of PKS, and the conditional situations are the same. Therefore, he concludes a series of “PKS breathy liquids” as *ɦN-, *ɦl-, and *ɦr-. Moreover, Weera Ostapirat points out that there is also another kind of breathiness (PKS pre-nasalized stops with -r-) which evidences the so-called “alternation of tonal series”, as in his statement that -r- “together with the intervoiced environment, may have special influence in including breathiness during the development of these pre-nasalized stops (e.g. *mpr- > mphr- > mbr-) and bring about tone behavior similar to that of *ɦr-” (Weera Ostapirat 1994: 85). Rejecting the hypothesis that breathy voice causes voicing alternation, Pittayawat Pittayaporn (2009) has provided another explanation for this phenomenon. He suggests that most of words with “voicing alternation” are loanwords from Old Chinese during the post-PT period (Pittayawat Pittayaporn 2009: 77-87). For example, he suggests the word for ‘bean’ in Old Chinese was a pre-nasalized form *N-tho[k]-s citing from Sagart (1999: 185-187), and the CT/SWT dialects simply dropped the *N- and borrowed this etymon as *tʰuəB1 while the NT dialects changed the pre-nasalized stop into a plain stop, and borrowed this word as *duəB2 (Pittayawat Pittayaporn 2009: 86)22. However, Pittayawat Pittayaporn also admits some words with “voicing alternation” should be part of PT or Kra-Dai (TK) cognates. He also finds that the voicing alternation items that may have been part of PT cognates reflect proto-voiceless initials but those may have been borrowed from Chinese at post-PT period reflect proto-voiced-initials in MC (Pittayawat Pittayaporn 2009: 87). Finally, he argues that voicing alternation is not a unified phenomenon but “consists of two distinct sets of data”. One set is a result of different arrangements for post-PT borrowings between NT and CT/SWT languages, and the In Pittayawat Pittayaporn’s original text this word has the B1 tone (2009: 86), but this should be wrong since words developed from *d- should collocate with a lower register Tone B2, and all NT languages have shown a B2 tone for this item in the previous transcriptions. 22 73 other set which may be from PT cognates is accounted to have sesquisyllabic clusters onsets which have given different voicing effects between NT and CT/SWT languages (Pittayawat Pittayaporn 2009: 87-88). In summary, previous studies have proved that the conditioning factors of tonal splits is not only limited to the voiceless: voiced opposition, but is also involved with some other phonation features of initial consonant, like aspiration, de-aspiration, continuants, glottalization, breathiness, and pre-nasalization. Besides the aforementioned classification reviewed in §1.1.1, §1.1.2 and §2.1.3 of the Tai languages, there are several opinions on classifying the whole branch or a specific sub-group of the Tai languages by some principles established by tones. Chamberlain (1975) uses different tonal splits based on Gedney’s (1989[1972]: 202) tone box to classify SWT languages as shown in Figure 12. He suggests that the Tai languages should be divided first into Northern and South-Central branches, and that the South-Central branch should be divided into two groups, P and PH23. Finally, the further classification under P and PH should follow the basis of the tonal splits, that is to say, “certain types of tone system splitting and coalescing were associated only with the PH group and certain others only with P group” (Chamberlain 1975: 49). For example, in the P group tone system is *ABCD 123-424 with the possibility in some dialects of A12-34, while in the PH group two types of tone splits are also shown as *A 1-23-4 / BCD 123-4 and *ABCD 1-23-4. However, there are also rare exceptions in group P which show the splits like those in PH group. For Li’s Central Tai dialects, the tone system variation correlates in precisely the same way, A123-4 or A12-34 in P group and A 1-23-4 in the PH group. This means that there is no reason for separating the CT and SWT. Chamberlain then lists the hierarchy of criteria of classifying the Tai languages: 1) P/PH, 2) *A column, 23 Chamberlain (1975) uses P and PH to refer to the initial plosives developing from proto-voiced plosives (conditioning the low register tones). In some SWT dialects like Northern Thai, Tai Lue, and Shan, most of CT varieties, as well as all NT varieties, initials developing from proto-voiced plosives are all unaspirated plosives (e.g. p-, t-, k-, and t -), which is represented by his Group P. In many SWT dialects like Thai, Isan, and Southern Thai, as well as some CT varieties like Zuozhou, Leiping, and Baoxu, initials developing from proto-voiced plosives are all aspirated plosives (e.g. pʰ-, tʰ-, ɕʰ-, and t ʰ-), which is represented by his Group PH. 24 1, 2, 3, and 4 here refer to four initial groups allocated in the four horizontal rows of Gedney’s tone box which has been introduced in Table 4 in §1.2.1. These four groups of initial consonants are voiceless friction, voiceless unaspirated, glottal, and voiced respectively which condition possible tone splits among different Tai languages. 74 3) *BCD columns, and 4) B-DL coalescence, and divides Li’s SWT languages into the chart of four types (two under P and two under PH) as well. PSWT P PH (*A 1-23-4) *A1-23-4 ABCD 123-4; B=DL *BCD 123-4; B=DL BCD 1-23-4; B≠DL Tse Fang, Black Tai, Siamese, Phu Tai, Lao, Southern Thai. Tai Mao, Red Tai, Neua, Phuan, etc. Muang Ka. White Tai, Lue, Shan, Yuan, Ahom, etc. Figure 12 Classification of SWT dialects based on tonal splits (adapted from Chamberlain 1975: 50) Edmondson (1994) describes some different tonal behaviors among NT, CT, and SWT. In his statement, the voiced-low principle, “which basically states that original voiced initial consonants conditioned low tones and original voiceless initial consonants conditioned high tones” (Edmondson 1994: 164), becomes a distinctive feature. In the vast majority of NT and CT languages, the historically high tone set generally possesses a higher pitch than the historically low tone set, while in SWT, “this situation is sometimes obscured by subsequent tonal flip-flop of highs and lows”, and becomes a characteristic for SWT just as what we have reviewed in previous §2.2.2. Though there is no tonal flip-flop in CT, there is a “slight tendency to tone reversal in the extreme southwestern locations” of SZ area (Edmondson 1994: 164). Furthermore, Edmondson uses Gedney’s tone box to analyze the tonology of NZ and SZ, which are of NT and CT languages respectively (Edmondson 1994: 165-169). The situation of tonal development in SZ is generally much more complicated than in NZ. “If one can say that the theme of tone splitting in NZ is voiced-low with variations of pre-glottalized B and C going with the low, then the theme in SZ is voiced-low with additional changes in the A tone.” This is also a distinguishing feature between the CT and the NT branches pointed out by Gedney (1989c cited in Edmondson 1994: 169). From the standard system of pure voiced-low splitting (such as Guangnan), to quadripartition of splitting in some columns (such as Daxin), SZ shows all possibilities of splitting among all rows in Gedney’s tone box. That is to say, in the SZ areas there is a pattern of tonal splitting involving aspiration and involving pre-glottalization. Therefore, SZ is “a laboratory in which tonogenetic affinities of various phonological features can be studied” (Edmondson 1994: 176). 75 Based on the previously discussed “voicing alternation” in Tai languages, both Li (1977: 36-39) and Gedney (1989a) claim that there is a correlation between this phenomenon and the classification of Tai languages, although their division of Tai languages are different from each other. Li points out that the tonal behaviors related to voicing alternation show regularity that CT languages agree in most instances with SWT languages, but occasionally with NT languages. This supports his three groups or sub-branches of Tai languages. Gedney also proposes that reflexes as earlier voiceless onsets in ST (including Li’s SWT and CT) cognates and those as earlier voiced onsets in NT counterparts support his NT-ST division of Tai languages. In his point of view, the original phonetic feature of voicing alternation was “voiced” due to those voicing alternation forms borrowed from Chinese voiced initial forms. Therefore, these forms have undergone a process of initial devoicing to become voiceless aspirations and have gone with the Row 1 of his tone box in ST languages (1989a: 256). The exceptional examples that voicing alternation forms in some ST varieties (limited to Li’s CT in particular) with “tones reflecting an original voiced initial” agreeing with NT forms only indicate that “such cases the word failed for some reason to undergo devoicing of the initial in that particular Southern language” (1989a: 257). Rejecting to both Li and Gedney’s claims above, Pittayawat Pittayaporn (2009: 85) cites items regarding voicing alternation from Luo (1996) to show that “many cases in which some non-NT dialects also show reflexes of earlier voiced onsets” to supplement his aforementioned explanation to voicing alternation (see §2.2.3). However, this could not tenably deny Li and Gedney’s hypotheses since both Li and Gedney have mentioned about this kind of exceptions and have given their own explanation for it. Besides, Pittayawat Pittayaporn’s data cited from Luo (1996) show some unvalued information. For example, tɤjB2 ‘thick, dense’ in Debao (CT) must be a wrong transcription since it is experientially found that all varieties of Debao Zhuang (of Yang Zhuang in ISO 639-3) have the form tʰejB1 for this word25, and this corrected form agrees with the regular rule of voicing alternation among Tai languages. Further discussions about voicing alternation applying to the Debao Zhuang (or Debao County Yang Zhuang) is my mother tongue, and the word for ‘dense’ in my mother tongue is tʰeɔ33 and is transcribed as tʰejB1 according to the notation of this thesis. I have never found a form t ɔB2 for this word in any other Yang Zhuang varieties or even any other Zhuang varieties in Debao County, Guangxi due to my personal experience and previous materials provided by other linguists. For example, in Zhang et al. (1999: 775) the word for ‘thick, dense’ is tʰeɔ5 in Jingxi (of Yang Zhuang), and tʰeɔ6 in Debao (of Yang Zhuang). Both of these two forms are transcribed as tʰejB1 in this thesis due to their regular correspondence which will be discussed in the new tone box to be provided in this thesis. 25 76 classification of Tai languages present as one of the phonological “puzzles” of Cao Lan called by Gregerson and Edmondson (1998) and the similar tonal phenomenon in Nung An (Edmondson 2002a) as shown in Table 27 and Table 28 respectively. Table 27 Voicing alternation among Cao Lan and NT and CT representative varieties (Gregerson & Edmondson 1989: 160) Gloss Cao Lan Giay (NT) Tay Cao Bang (CT) ‘excrement’ həi⁴ ʔe⁴ khi³ ‘rice’ hau⁴ hau⁴ khau³ Table 28 Voicing alternation among Nung An and the representative varieties of NT and CT (adapted from Edmondson 2002: 59) Gloss Nung An Giay (NT) Nung Giang (CT) ‘excrement’ həi⁴ ʔe⁴ khjai³ ‘rice’ hau⁴ hau⁴ khau³ ‘bowl’ tui⁴ tui⁴ thui³ ‘bean’ tu⁶ -- thu⁵ ‘bitter’ ham² ham² kham¹ ‘son-in-law’ kɯi² kɯi² khi¹ Both Cao Lan and Nung An are spoken in Northern Vietnam, in which the majorities being CT speaking groups live. Unlike most languages of NT without aspirated initials, Cao Lan and Nung An have the same series of aspirated initials like languages of CT does. Nung An is actually a migration from Long’an (in Guangxi) (Edmondson 2002: 52), which is traditionally classified as one dialect of YN of CT. However, unlike other CT languages showing typical Central Tai distinctive features, Cao Lan and Nung An show more Northern-Like phonological and lexical forms, as shown as what Gedney states that for a few Central languages having “sporadic Northern-like forms” (Gedney 1989b: 63). One of the phonological puzzles of these languages is a number of common items with voicing alternation behaving an original low tone in Cao Lan, just like all NT languages do, but opposite to behaving an original high tone in both SWT and CT languages (Gregerson & Edmondson 1998: 160). In the comparison in the two tables above, both Cao Lan and Nung An clearly go with Giay, which present the typical feature of “proto-breathy” initials going with 77 low tone in NT as sufficiently discussed in previous reviews. As for that whether Cao Lan should be put into NT or CT groups, Gregerson and Edmondson (1998: 162) state that “the Cao Lan people were once a NT people moving southward who stopped and interfused with a CT speaking group at a time early enough for some late CT rules still to operate”. As for Nung An, Edmondson however states that it “brought these NT features with them when they immigrated from further north in an area located on the border between NT and CT areas” (Edmondson 2002: 61). These two different but ambiguous claims raise a question of the classification of NT-like purported CT languages (or in particular YN varieties and Cao Lan), which will be analyzed and discussed in §4.3.2.1 and 5.2. Theraphan L-Thongkum (1997) proposes that some other specific tonal split patterns can be treated as an important criterion of the classification of Tai branches. She firstly provides a research on tonal splits in Dai Tho (Dai Zhuang in ISO 639-3) and Tai Tho (Debao County Yang Zhuang in ISO 639-3) of CT in Yunnan and Guangxi of China. A special tonal behavior of these languages is that proto-voiceless sonorants did not go with aspirated plosives like other Tai languages did, but went with voiceless unaspirated plosives. This tonal behavior of Wenma (Dai Zhuang) and Debao has actually been mentioned in the discussion of more pending issues on Tai tones in §1.2.2 and has been partly shown in Table 7. However, her treatment is to put proto-voiceless sonorants into Row 2 of Gedney’s tone box to be with voiceless unaspirated plosives, and to keep only proto-aspirated plosives in Row 1 (Theraphan L-Thongkum 1997: 194). This is different from Gedney’s original tone box which has Row 1 containing both voiceless aspirated plosives and sonorants, and is different from Liao and Shen’s revised Tai tone box which separates more horizontal rows to solve the problems of more tonal split patterns. Just as mentioned in §2.2.3 that she argues that the main factor of primary tonal split is “the merging of proto-voiceless sonorants with proto-voiced sonorants” in Tho (Theraphan LThongkum 1997: 215) which includes her Dai Tho and Tai Tho, the tonal behaviors of these two languages show a specific pattern different from other Tai languages. Based on this, she tends to support Li’s (1977) tripartite division of Tai since she believes that the CT sub-branch has “its own interesting history of phonological development”, and “it should be kept apart instead of being grouped together with the Southwestern branch” (Theraphan L-Thongkum 1997: 215). In addition, Pranee Kullavanijaya and Theraphan L-Thongkum (1998) propose a sub-classification of CT to divide “Nong-Tay” and “Budai” by using tonal splits of PT tone A as the main factor, especially the tonal development conditioned by the dental consonant clusters *tr- and *tʰr- in Li’s reconstruction (1977). This time Dai Tho in Theraphan L-Thongkum (1997) is put into their Budai group of CT, of which Nong-Tay which presents is a sister. 78 However, Johnson (2011: 48-55) calls both the specific group of “Tho” from Theraphan L-Thongkum (1997) and the sub-classification of CT from Pranee Kullavanijaya and Theraphan L-Thongkum (1998) in question. He points out that although Dai Zhuang and Debao County Yang Zhuang which belong to Theraphan LThongkum’s Tho group seem to present a similar tonal split patterns to each other in PT A tone, they have totally different tonal split patterns in other tonal categories. The overall phonological systems of these two languages do not suggest that they have more shared development between these two groups than between other CT groups. He also points out that some “preliminary analysis of recent data from an SIL International East Asia group and Guangxi Region Language Commission survey of the Zhuang languages” has indicated linguistic grounds for grouping Debao Zhuang together with the Yang Zhuang of Jingxi County, and this agrees with the grouping of Debao together with Jingxi in previous studies on Zhuang dialects in the Sinitic circles like Wei & Qin (1980) and Zhang et al. (1999). Therefore, he feels that “the grouping of Debao and Tianbao, together with Dai Zhuang, is not a strong argument” (2011: 54). That is to say, from Johnson’s viewpoint some tonal split patterns conditioned by some specific conditioning factors like aspiration cannot tenably suggest being a criterion for classifying a sub-group of Tai languages. In this chapter, a review of the literature related to historical and comparative method applied to Tai languages and Tai tonology is drawn on to provide an overview of the theoretical framework for this study. Historical and comparative method is the basis of tone data analysis from forty-two Tai varieties. There were numerous previous studies on Tai tonology, and they are divided into four parts, namely tonogenesis, tonal splits and mergers, the conditioning factors of tonal splits and Tai classification related to tonological issues. The first three parts have close correlation to Tai tonal development scheme, which is the first objectives and is the main body of this thesis. In addition, the conditioning factors of tonal splits actually constitute the theoretical basis of the design of Tai tone box, which is one of the objectives of this thesis. Finally, Tai classification related to tonological issues is the precursor of Tai classification based on tonology, one of the targets of this study. Based on the previous studies, the following chapter describes the principles of methodology, and three chapters after that successively probe into the three objects of this studies, namely diachronic hierarchies of Tai tonal development, Tai classification based on tonology, and a new perspective of Tai tone box. 79 Methodology, data and notation Given the objectives outlined in §1.2, the problems discussed in §1.2.3, and the hypotheses in §1.2.4, it is imperative to assess the assumptions about Tai tonology. Therefore, this thesis approaches the reconstruction of tones in PT and the diachronic developmental stages of tones in Tai languages by using a large set of tone data from 42 Tai varieties. This section mainly presents a brief description of the theoretical basis of the methods to test the hypotheses of this thesis, and falls into four parts, viz. general comparative method, library research, fieldwork, and notation. As mentioned in §2.1, since Tai tonology is based on the study of tone changes among different Tai varieties, this study has to be based on the theory of historical linguistics which is first developed in the 19th century and mainly deals with language change among Indo-European languages (Campbell 2004). Several aspects of historical and comparative linguistics such as linguistic change, borrowing loanwords, comparative method and linguistic reconstruction, linguistic classification, and areal linguistics are emphasized in the discussion of this thesis. When talking about linguistic change, we have to consider how to explain this phenomenon, or what makes the change occur on sounds, meaning, lexicon, and grammar. In this thesis, the main linguistic changes that we will be concerned with are sound changes especially the changes of tones, such as non-tonal prosodic words in PT generated tones, tonal value changes presenting like original high tone goes to low tone in many varieties in SWT, or different tonal splits on those prosodic words with specific initials. With respect to the discussion of semantic change, in this thesis, cognates having derived different meanings among different modern Tai varieties are involved in the discussions for interpreting the tone changes actually occurred to the same prosodic word in PT. For example, the word for ‘river’ in CT 80 and NT varieties and the word for ‘wharf’ in SWT varieties both ascend to the same proto-form *da:B in PT, and these “two words” are actually the results of semantic changes from the same root and have to combine to be the same lexical item in the research wordlist, for interpreting the sound changes especially the tone changes of this prosodic word among different modern Tai varieties. In short, the discussions of sound changes occupy the main domain of linguistic change in this thesis, while semantic changes often supplement to sound change in related discussions. The causes of linguistic change includes internal causes and external causes, just like the discussion of the literature of linguistic change in §2.1.1. With respect to use of internal factor, in this thesis, a sound change normally is considered to be due to an internal factor, such as a tonal split is normally treated to be conditioned by a specific group of initials in history. However, it does not mean that external factors can be ignored even we have already known about that the factor was of internal on the one hand since there might be external factor on the other hand. For example, when talking about tonogenesis of Tai languages, many scholars have claimed there were no tones in PT and Tai tones produced due to the avoidance of meaning confusion because of the loss of final consonants which should be treated as an internal factor, but we still cannot ignore that the loss of final consonants might be as a result of being influenced by similar linguistic phenomenon on languages outside of Tai such as Middle Chinese due to areal linguistic method, and this factor is obviously external. The role of borrowing or loanwords is also a factor that cannot be ignored in applying the comparative method. Certain lexicons that are shared widely among Tai varieties might be loans from Old Chinese, and many lexicons shared among CT and NT varieties might also be loans from Middle Chinese and modern Chinese dialects (Prapin Manomaivibool 1975, 1976; Zhang et al. 1999: 246-286). Old Chinese loans give a hint for how tones developed in Tai. For example, except few loans with aberrant tonal development, in Tai languages most of loanwords borrowed from Old Chinese present a very regular correspondence between Tai tonal categories *A, *B, *C, and *D and Chinese tonal categories yīn (level), qù (departing), shǎng (rising), and rù (entering) respectively (Li 1976; Prapin Manomaivibool 1975). Since some scholars such as Norman (1988: 52-57) have tried to prove that tones qù and shǎng in Sinitic languages produced due to the loss of the fricative and glottal codas in Old Chinese, it is speculated that when PT 81 borrowed words from Old Chinese, the sound correspondence has to be arranged with the same or similar traits on coda, meaning in PT Tone B was with fricative codas, and tonal category C was with glottal codas. The tonal genesis of PT must be very similar to that in Chinese, meaning in the early stage of PT, before tonal splits occurred the loss of fricative and glottal codas produced Tones B and C respectively. The comparative method and the reconstruction of PT segments especially prototones is crucial to this study. Many linguists such as Li (1977), Gedney (1989[1972]) and Edmonson and Solnit (1997) have discussed in great detail the PT tonal system, and have also provided a successful reconstruction of tonal categories in PT by applying comparative methodology, as the discussion in §1.2.1. In this thesis, the comparative method is mainly applied to the discussion of Tai tonal development. For example, the “proto-breathy sounds” reconstructed by Liang and Zhang (1996) and further discussed by Liao and Shen (2012) is based on the so-called voicing alternation due to the comparison of different direction of tonal split between NT and ST varieties. In the related discussion is in §1.2.3, we have known that these “proto-breathy” initials regularly condition tonal split going with original low tone with those developed from proto-voiced initials in all NT dialects, but condition tonal split going with original high tone with those developed from proto-voiceless initials in the vast majority of ST. This specific tonal developmental phenomenon led to the addition of an extra horizontal row in Liao and Shen’s revised Tai tone box (2012). On the other hand, when we compare more data from NT and ST varieties, a contrastive tonal developmental phenomenon is observed, and that is some specific cognates such as ‘man’ and ‘to blow’ having tonal split going with original low tone with those developed from proto-voiced initials in all ST varieties (e.g. tɕa:jA2 ‘man’ and patDS2 ‘to blow’ in Jingxi County Yang Zhuang; tɕʰa:jA2 ‘man’ and pʰatDS2 ‘to blow’ in Thai), but condition tonal split going with original high tone with those developed from proto-voiceless initials in all NT varieties (e.g. θa:jA1 ‘man’ and patDS1 ‘to blow’ in Wuming Zhuang). The specific condition of this kind of tonal developmental phenomenon is still unstudied, and we need further discussion to prove the causes of it by comparative method (cf. §4.3.2.1). When discussing the comparative method, the regularity of correspondence among data from different Tai varieties is emphasized. However, aberrant development lacking regular tonal correspondence in some specific examples should not be ignored. Liao (2013) proposes that the proto-Tai reconstruction of the word 82 ‘maternal grandmother’ should be *ta:jA, which presenting a total irregularity in modern NT, CT and SWT varieties to reflect proto-form *ta:jB, *ta:jA and *na:jA respectively. Liao interprets that the irregularity is due to sound changes which served to dissimilate this word from the otherwise homophonous word ‘to die’ in NT and SWT varieties. This proto-form is supported by historical evidence and other non-Tai languages in the TK family. Similar phenomena like tone changes by analogy with words that are semantically similar should be also noted. For example, in my checklist of lexicon for investigating tone data for this thesis, the word for ‘maternal grandfather’ is designated in the column of *A tone since this tone is reflected by the vast majority of Tai varieties, while several CT varieties present this word as B1 tone which is different from other Tai varieties. As what has mentioned in §2.1.1, Liao (2013) has also discussed this phenomenon and has proposed that the original Tone A1 of the word ta:B1 ‘maternal grandfather’ in these CT varieties has been replaced with the tone of the word for ‘maternal grandmother’ ta:jB1 in these varieties, and is a case of analogical change with words of paired semantic contents in these CT languages, such as in Debao and Jingxi Yang Zhuang ta:B1 ta:jB1 ‘maternal grandparents, wife’s parents’. In short, these phenomena give a clue that when we apply to the comparative method, we should not ignore some external factors may hide behind irregularities of sound correspondence among different daughter languages. Language classification especially how to sub-group Tai varieties based on tonology is one of the main topics of this thesis. Campbell (2004: 188-191) gives a specific determination of subgrouping methodology as being an internal classification which determines the most closely related sister languages of the same language family, and points out that “the only generally accepted criterion for subgrouping is shared innovation” (Campbell 2004: 190), which is a linguistic change departing from a specific trait of proto-language and is shared by the same subset of the daughter languages. Because an innovation normally occurs in a single daughter language and subsequently diversifies into daughter languages of its own, normally this innovation is not shared by languages in other sub-group of the same family (Campbell 2004: 191). When applying to Tai classification based on tonology, the criterion is also given to determine which tonal innovations are shared and are not shared by which languages, and then to subgroup different subsets of daughter languages. A typical 83 Tai classification based on tonology is displayed by Chamberlain (1975). He suggests the Tai languages should be divided earlier into two branches, Northern and SouthCentral, and in the latter the South-Central branch was divided into two groups, P and PH, and the further classification under P and PH should follow the basis of the tonal splits to finally divide Li’s SWT into four types, as in the discussion in §2.2.4. Whether Chamberlain’s Tai classification can stand scrutiny, the criterion of shared innovation is indeed used to determine his viewpoint. Chamberlain’s claim will also be tested in this thesis since one of the hypotheses of this thesis is to determine the Tai classification based on tonology. The method of areal linguistics which is related to borrowing is emphasized here by the term ‘diffusion area’. It refers to a linguistic phenomenon which describes languages within a certain geographical area coming to share certain structural features including loanwords and “shared elements of phonological, morphological or syntactic structure” due to language contact, such as borrowing and diffusion (Campbell 2004: 330-331). An example of areal linguistics applying to Tai languages can be analyzed in the case of Gedney’s (1989b) viewpoint about the difference between Li’s (1977) CT and SWT languages. Although Gedney suggests CT and SWT groups “form a continuous dialect area, with only gradual transition throughout and no real language boundary anywhere” (Gedney 1989b: 66), he still admits that there are basic differences between CT and SWT when being determined by phonological and lexical criteria, such as CT languages do share certain words with NT but not found in SWT, and CT languages are conspicuous for tending to retain bizarre reflexes of original consonant clusters like the words ‘eye’ and ‘to die’ with initial t- in all SWT and NT languages but with such initials as tʰ-, h-, pʰ- or pʰj- in CT languages (Gedney 1989b: 64). If CT and SWT really form a continuous dialect area with no real language boundary, the basic differences between these two sub-groups can be interpreted by areal feature: Since CT and SWT languages have geographically separated from each other for a long time, the “bizarre reflexes” of original consonant clusters retained in CT languages can be supposed to be an areal trait due to independent development within their own area for a long period, and that is why SWT languages do not share this areal feature even though they are phonologically and lexically much more closely related to CT languages than NT to CT languages; and since CT and NT languages are still adjacent neighbors, CT languages may share certain inherited Tai 84 words with NT, but SWT languages might have dropped these inherited Tai words due to lexical shift being an areal trait within their area, or CT languages just directly borrowed these words from NT languages due to language contact, and no matter which reason is true, they both can interpret that why we cannot find these certain words in SWT languages. Likewise, when we talk about different tonal behaviors in different sub-groups of Tai, areal linguistics can explain many causes and reasons. For example, in the subsequent discussions, we will see although the CT and SWT sub-groups show a unified trait in the level of primary tonal development which is different from that of NT, the CT sub-group shows a much more complicated secondary tonal development while the SWT sub-group does not due to their own independent development forming different areal feature. Library research is necessary for completing the literature review for this study. The literature review has been done before doing data collection. The literature review is laid the emphasis on previous research of Tai tones including Chinese linguistics materials. In the past Chinese materials of Tai tonogenesis, Tai tonal development and other related research are easily neglected because of the language barrier. Therefore, I placed particular emphasis on Chinese materials which acquire more CT and NT linguistic information since most CT varieties and the vast majority of NT Tai varieties are found in China. The details of this process have been listed in Chapter 2. For doing a comprehensive study of Tai tonal development as much as possible, it is important to minimize gaps in our knowledge of tones in modern Tai varieties. Therefore, in this thesis forty-two Tai varieties which are identified as representatives of SWT (seven), CT (seventeen), YN (five) and NT (thirteen) languages respectively are chosen for data analysis. These are listed in §3.3.1, and are also filled in the chart of the information of Tai varieties selected attached in Appendix D, as well as in the map of the Tai varieties investigated with the same varieties numbering shown in Appendix D. Except some languages have been fully researched, such as Thai, Northern Thai, and Wuming Zhuang in published materials, the data of the other varieties is mainly based on my own fieldwork research for this thesis. 85 The selection of the Tai varieties investigated is based on tonal diversity. Previous studies have obtained much achievement on the research of tones in SWT. This thesis relatively puts more attention to the tones of the less studied CT and NT languages (including the YN group whose classification as CT or NT is open to debate). Within the forty-two Tai varieties, the number of the CT languages is seventeen to be most numerous. This is because when comparing with tonal development in NT languages, the situation of tonal development in CT languages is “generally far more diverse and complex” (Edmondson 1992a: 169). In the two maps of the Tai varieties investigated in Appendix D, seventeen languages including fifteen CT varieties and two NT varieties are concentrated in the narrow triangle zone which is on the southwestern corner of Guangxi, since that area contains one of the most tonal diversities in Tai languages. The tone data from these forty-two Tai varieties have been collected by the wordlist based on Liao & Shen’s (2012) revised Tai tone box (see §1.2.2), which is revised from Gedney’s tone box (1989[1972]). For investigating these Tai varieties, the wordlist with 195 words is designed as shown in Appendix B. The wordlist takes the following two checklists as the original templates: Gedney’s checklist for his Tai tone box (1989[1972]: 202-204) and the supplemental checklist made by Dr. Phinnarat Akharawatthanakun for collecting tone data in Thai dialects in AL701 Field Methods in Linguistics course, Linguistics Department, Payap University. Nevertheless, the wordlist in this thesis has abundant supplements for the higher precision of attesting the tonal splits and mergers. Certain words (normally five to seven words) are designated for one box for insuring the confidence of establishing the tone for that box. However, when there are not enough words found in one box, such as in B1UC, C1-UC, and DL1/2 boxes where I had only found 1-2 words for appropriately investigating before going to the fieldwork, these boxes will be designated to have the same blanks as other boxes (other tones) fully filled in with lexical items on the same horizontal rows. For example, although B1-UC and C1-UC boxes have only two lexical items (‘to hunt’ and ‘cage (box)’ for B1-UC as well as ‘near’ and ‘CLF for stone’ for C1-UC) found respectively, both of them contain five blanks since in the same horizontal level box of A tone (A1-UC box), there are plenty lexical items are found, of which five items (‘eye’, ‘to die’, ‘gizzard’, ‘cucumber’, and ‘to put up’) are selected to be put in the wordlist. These remainder blanks in boxes B1-UC and C1UC are reserved for filling in when I find that other appropriate words should be put in them in the future. 86 The criterion for designing this wordlist is that all etyma selected should be commonly found in all the three Tai branches. For example, some etyma commonly found in SWT languages like tu:C1-U ‘cabinet’ and kʰuətDL1-A ‘bottle’ in Thai are never found in CT, NT as well as the debatable affiliation YN varieties. Therefore, they are not considered to be listed in the preferential items in the wordlist of this thesis, even though they are commonly designated to be attested items in some checklist of Gedney’s tone box in the fieldworks of SWT. Nevertheless, when an etymon is not found in a specific Tai variety investigated, an optional substitute or a succedaneous item which was provisionally designated at that time will be used, like wa:wB2 ‘kite’ (in Thai) is sometimes used for investigating SWT even though it is not found in CT and NT. The implement of data collection was accomplished during my own fieldwork carried out on a number of Tai varieties from China, Thailand, and Vietnam. Altogether I accomplished fieldwork on thirty Tai varieties. The data collection periods are as below: During March 2012 – April 2012, I finished data collection from five varieties of Napo County Yang Zhuang, Debao County Yang Zhuang, and Min Zhuang around my home areas in Guangxi; In October 2012, I recorded five Tai varieties of Thai, Northern Thai, Northeastern Thai, Southern Thai, and Shan in Chiang Mai, Thailand; When I studied English course in the University of Arkansas, the U.S. from January to March 2013, I got to know a fellow who is a Tai native speaker from Central Vietnam, and asked for his permission to record his Tai variety, which is called Tai Dam by himself and is of SWT sub-group; From April – May 2013, I carried out the second time fieldwork to my home area Guangxi, China and finished data collection from eighteen varieties of NT and CT; In June 2014, I carried out fieldwork on the Tai Lue variety of Rong Maet Village, Jun District, Phayao Province of Thailand; After thesis proposal defense, during the winter of 2013, I found that tone data collection for SWT varieties should also be done by using the same wordlist with the one used for CT and NT varieties but not by the previous wordlists with only 60-90 words used for SWT only, for more convincingly interpreting the new revised tone box, I rerecorded tone data from Thai, Northern Thai, Southern Thai, Isan, Shan, and Tai Lue varieties. At the end of August, 2014, I carried out my last fieldwork for this thesis on the Daxin Baoxu variety of Zuojiang Zhuang in Guangxi. The detailed information of languages investigated by myself is also filled in the chart of the information of Tai varieties selected is attached in Appendix D. 87 Certain tone data of some Tai varieties are also referenced to other research materials, which are listed on the chart of the information of Tai varieties selected attached in Appendix C. For those Tai varieties only referenced to other research materials, I could only fill in the lexical items that are found in those materials, and left the symbol of blank space “--” when that cognate was not recorded. The transcriptions are all in IPA standard (International Phonetic Association 2015), and tones are transcribed by using pitch level and contours based on the Chao 5level pitch scale (Chao 1930), which has been previewed in detail in §2.2. Exceptionally, short tones in Chao’s system (tone letter 1, 2, 3, 4 and 5) are transliterated as long level tones (11, 22, 33, 44, and 55) to avoid the superposition of tonal categories (tones 1, 2, 3…10) in the Sinitic circles. In this situation, short tones can only be identified according to the transcriptions of short vowels, such as tɕat⁵⁵ ‘seven’ in Yang Zhuang varieties. For example, the six tones on live syllables in the Urban variety of Debao County Yang Zhuang contains one falling (31), one rising with glottalization (24ʔ), two levels (55 and 33), and two circumflexes (353 and 213ʔ). In addition, three level tones (including short tones) on dead syllables are 55, 33, and 11 respectively (for details see L8 in Appendix C). For acquiring the accurate tone data from each language, the selection of LRPs is limited to mother tongue speakers of the Tai varieties collected. I tried to select LRPs all above 30 years old and at least three LRPs for each variety for acquiring stable tone data, but sometimes failed to achieve this target because of limited time and the complex geographical situation of some remote locations, even though I have finished four times of fieldwork trips. When I met such a situation, I also guaranteed the selection of 1-2 LRPs being at least above 20 years old, and having wholesome vocal organs. The information of LRPs and each Tai variety is listed in the charts of language information in Appendix C. Data analysis has been done according to the general historical and comparative linguistics methods, which are discussed from §3.1.1 to §3.1.5. The tonal development has been analyzed mainly by comparing different tonal behaviors among different Tai varieties. When a tonal split or merger occurs, the condition of this tonal behavior such as initial feature, vowel length on dead syllable, or other related factors need to be determined. If there are two tones separating according to a conditional feature within one row, and that they are phonetically distinct in a certain number of Tai varieties, an extra row of the tone box needs to be added. 88 Tone analysis in this thesis has mainly been done by perception of auditory sense; however, acoustic analysis is used to analyze tone data when it is necessary. If the tones are phonetically quite distinct, e.g. falling compared to rising, then that is clear evidence that they are different tones. Acoustic analysis by using PRAAT (Version 5.1.34) is only applied for analyzing tones from two different rows of tone box that are almost identical and the LRPs claim there is a difference in pronunciation between these two tones. If examples from X tone and Y tone always show a fixed high-low frequency in Hz respectively in PRAAT analysis, they should be separated into two tones even though they are phonetically quite similar to each other, otherwise they should be treated as a merger. A typical example of tone analysis by having recourse to PRAAT is the definitions among tones B1-U, B2, and B1-G in the Jingxi Hurun variety of Yang Zhuang (cf. L11 in Appendix C). In the charts of tonal split and merger patterns of each variety (cf. Appendix C), some tones in smooth syllables and in dead syllables present different tonal values but are treated as allotones of the same tone, such as B1 (242) and DL1 (24) in Debao Dongling (L32, NT) in Table 34 (in 4.1.3.3.1, also cf. L32 in Appendix C), based on their similarities of the pitch on the main part. When a tone in smooth syllables is a complex-contour or so-called circumflex tones (including dipping e.g. 213, convex e.g. 453, delayed falling e.g. 553, and delayed rising e.g. 112), the criterion of determining its allotone in the dead syllables is based on the homologous pitch of the front half part of the circumflex on smooth syllables. This is because the back half part of the circumflex is always dropped off in a non-final position of a carrier sentence in normal speech, whereby this front part must be treated as the main contour of the tone. It means, 242 in Debao Dongling Zhuang actually presents 24 in a non-final position in normal speech, to be the same as its allotone 24 in dead syllables. This criterion of determining allotones will be throughout implemented. Although this thesis mainly analyzes data collected and described by myself, certain data is incorporated from diverse sources by different researchers. For facilitating comparisons different Tai varieties, it is necessary to establish notational conventions. All modern Tai varieties forms as well as reconstructed PT forms that appear in this thesis except those cited have been re-transcribed according to the IPA standard (International Phonetic Association 2015) as below. 89 1. IPA symbols are all in italic rather than in “[ ]”, e.g. Nong Zhuang tʃʰu:A1 ‘ear’ rather than [tʃʰu:A1]. 2. The dash “-” is put before an IPA symbol to indicate a following segment, e.g. -a: means a low-mid unrounded vowel following some consonant, and -m means a bilabial nasal final consonant or coda; while it is put after an IPA symbol to indicate a followed segment, e.g. m- means a bilabial nasal initial consonant followed by some vowel, and a- means a low-mid unrounded vowel followed by some coda. 3. The prepositive asterisk “*-” means its following form is a reconstructed form in a proto-language, e.g. PT *pajA ‘to go’ rather than pajA. 4. Tone value markers for modern Tai varieties: tones are transcribed by using superscripted pitch level and contours based on the Chao 5-level pitch scale (Chao 1930), like the regulation of transcribing tone data in §3.3.1. The tone markers must be tightly put next to the syllable transcribed by IPA. Creaky tones are marked under the superscripted pitch value, while glottalized tones are marked by a superscripted -ʔ next to the superscripted pitch values, e.g. Debao Urban (L8) ma353 ‘dog’, ma:213ʔ ‘horse’, and Yizhou Suogan (L36) paj51̰ ‘to go’. 5. Tonal category markers for PT and modern Tai varieties: PT tone categories *A, *B, *C, and *D and modern Tai tone categories A1, A2, B1, B2, C1, C2, DL1, DL2, DS1 and DS2 are italic and keep the normal capital forms in the sentences of the text, but are shrunk to the upper right corner when following the transcriptions of lexical items. The tone markers must be tightly put next to the syllable transcribed by IPA; when the PT tone markers are transcribed with consonant and vowels, the prepositive asterisk should be tightly put before the initial consonant, e.g. PT *da:B ‘river’; Debao ta:B1 ‘river’ rather than *da: B and ta: B1. 6. Superscripted symbols: aspiration is ⁻ʰ rather than a normal -h, and pre- glottalization is ˀ⁻ rather than a normal ʔ- except the situation of PT, e.g. Jingxi tʰa:A1 ‘eye’ rather than tha:A1, Nalong Zhongyi *ʰla:A1 ‘to seek’ rather than *hla:A1, Debao ˀja:A1 ‘medicine’ rather than ʔja:A1, but PT *hna:A ‘thick’ and *ʔjɯaA ‘medicine’. 7. The palatal semivowel is transcribed as j rather than y which actually represents another sound – the high front rounded vowel, e.g. Debao ja:B2 ‘husband’s mother’ rather than ya:B2. 90 8. Final semivowels of a syllable are treated as glides -j, -ɥ, -ɰ, and -w, e.g. Jingxi kai45 ‘chicken’, mou53 ‘pig’, Wuming θaɯ41 ‘to buy’ (cf. Zhang et al. 1999) and Debao ɬøy²¹³ ‘to buy’ (cf. Liao 2010) are re-transcribed as kaj45, mow53, θaɰ41, and ɬøɥ²¹³ respectively. 9. Non-IPA symbols should be re-transcribed as IPA symbols: č- represent the voiceless palatal affricate and re-transcribed as tɕ-; similarly, ǰ-, š- and ȵ- are re- transcribed as dʑ-, ɕ- and ɲ- respectively. 10. Long vowels are indicated by “-:” rather than double vowels, e.g. Daxin kʰa:A1 ‘leg’ rather than kʰaaA1. Note that the notation related to the usage of IPA described above only applies to forms in TK language family. When we reference to source from languages from another family, the forms of IPA are only transcribed in their original notation. Besides, in this thesis Chinese Pinyin which means Chinese Phonetic Alphabet formulated by the Chinese government from 1955 is used generally for those names of person and place, title of Chinese works in th literature, and some terminology with a Sinitic linguistic source. But some names have been well-known by academic circles for a long time, such as Li Fang-Kuei, and the Tai dialect of Lungchow researched by Li (1947; 1977), I will only follow the original notation rather than Chinese Pinyin which presents them as Li Fanggui and Longzhou respectively. However, if I reference Zhang et al. (1999), one of the Southern Zhuang dialects will be cited as Longzhou following the notation of Chinese Pinyin since the book is a new publication in contemporary China. Last, although simplified Chinese characters are officially and generally used in Mainland China, the Chinese characters in this thesis are only transcribed in traditional characters, because many discussions of Ancient Chinese loans in Tai varieties needs to regard etymology linking to the original forms of Chinese character. 91 Diachronic hierarchies of Tai tonal development This chapter discusses the process of Tai tonal development, from the arising of tones, tonal splits including the conditioning factors of tonal splits, to the trends of Tai tonal development in some specific modern Tai varieties. Since tonogenesis or the arising of tones in Tai languages is viewed as the first step of tonal development of these languages, it is necessary to give a full screening of different arguments of Tai tonogenesis when we discuss the very first scales and levels of Tai tonal development. Although all modern TK languages have welldeveloped tonal systems, Tai tonogenesis cannot avoid being explored after the tonogenesis hypothesis applying to their surrounding languages, because Tai tones and tones in Chinese, Vietnamese and Miao-Yao languages are structurally identical and show an evident correspondence. This is because those tonal languages like Vietnamese have genetically related non-tonal languages like Mon-Khmer languages, as shown in Vietnamese tonogenesis presented by Haudricourt (1954b) in §2.2.1. Moreover, in Sino-Tibetan Language Family, some languages especially the Sinitic languages have well-developed tonal system, but more languages like most TibetoBurman languages have less-developed tonal system, and some languages like Amdo Tibetan even have no tonal development at all (Dong 2014: 38). Tai tonogenesis due to the loss of syllabic stress of original Austronesian multisyllables proposed by Ni (1990) reviewed in §2.2.1 is rejected in this thesis. This is because the relationship between TK languages and Austronesian languages is still unclear, although recent studies such as Weera Ostapirat (2013a) also support the Austro-Tai hypothesis proposed by Benedict (1942). Even though the Austro-Tai hypothesis can be established, the arising of tones in PT should be on the stage of monosyllabic or sesquisyllabic structures due to the previous studies on PT reconstruction, and the time should be far later than the time when the language structure changed from multisyllables to monosyllables or sesquisyllables. Except Ni’s proposal, the following points are commonly agreed within the different points of view of PT non-tonal elements or tonal features in the literature review in 92 §2.2.1) in a specific period PT had a tonal system containing four tonal categories *A, *B, *C, and *D; 2) proto-tone *A was a modal tone without affiliated consonantal endings or particular voice quality; 3) proto-tone *B and*C arose by dropping some specific syllable finals or specific phonation voice; 4) proto-tone *D was on checked syllables without tonal contrast as they are reflected in the vast majority of the Tai daughter languages. In contrast, the phonetic characteristics especially the voice qualities or affiliated consonantal endings of PT tones *B and *C lack of a full agreement among different proposals. Table 29 summarizes some of the disagreement. Table 29 Disagreement on the phonetic characteristics of PT tones *B and *C Proto-tones *B Sources Pitch height Contour *C Vowel Voice Pitch duration quality height Contour Vowel Voice duration quality Pittayawat Pittayaporn rising long creaky high falling short no no -- -h no no -- -ʔ no no -- -h no no -- -ʔ -- -- -- -ʔ -- -- -- creaky (2009: 271) Liang & Zhang (1996: 815) Gedney (1989b: 89; 91) Sagart (1988: 89) glottal low constriction Before the comparison of the four arguments about the Tai tonogenesis process and the phonetic shapes of PT tones in the chart above, the following two clarifications are preferentially enumerated. First of all, the process of Tai tonogenesis should be treated as the very first step of the arising of tones in PT. This clarification rejects the treatment that tone in Tai is an inherent feature and cannot develop from non-tonal elements. The very first proposal of non-tonal structure at a specific period of PT is provided by Matisoff (1973), and it is supported by the tonogenesis hypothesis that all other tonal languages surrounding TK language family have been commonly proposed to develop from non-tonal languages, such as Vietnamese (Haudricourt 1954b; Matisoff 93 1973; Benedict 1972b, 1973), Chinese (Haudricourt 1954a; Pulleyblank 1962; Benedict 1972b, 1973; Matisoff 1973; Norman 1989), and Miao-Yao (Matisoff 1973; Benedict 1972b, 1973), and for the early period of PT this cannot be a single exception. That is to say, typologically the treatment of tone in Tai as an inherent feature and cannot develop from non-tonal elements is shown to be isolated. Although Li (1977: 24) believes that “it is impossible to recover the final consonants that are assumed to have been dropped” to give the arising of tone in PT period, he has furthermore suggested that if the proposal of tonogenesis is tenable, for Tai “it must refer to a stage of the language prior to Proto-Tai” and the origin of PT tones may have been in the “Pre-Tai” period (Li 1977: 24). Here we still consider the suspected non-tonal period to be of an early stage of PT instead of Li’s “Pre-Tai” period, since we have reviewed that in most previous studies, PT tones are always considered to arise by dropping the tonal-source finals -ʔ and -s/-h, or by reducing specific phonations like creakiness or glottal constriction. Second, at a later period of PT non-tonal structure of the earlier stage should have changed to tonal structures due to the decaying or the weakening of the specific consonantal finals, like -ʔ and -h. The typological evidence for this statement is provided by the previous studies especially tonogenesis of Vietnamese, Chinese, and Miao-Yao, although different studies have different points of view about the arising of Tai tones and the tonal features in PT, like the weakening of marked phonations, final glottal constriction and creakiness, suggested by Pittayawat Pittayaporn (2009). We will discuss below to demonstrate that the earliest origin of tones should have been some specific consonantal finals, which we treat as “cause” giving rise to tones, instead of those marked phonations or voice qualities, which we treat as the “effect” triggered by the cause. The most direct reason need to be emphasized is that the method of tonogenesis suggested by Matisoff (1973) is based on a hypothesis proposing that tone is a secondary development which induced by the decaying of the consonantal system especially syllable finals, and this method has been applied to Vietnamese, Chinese, and Miao-Yao, which have very clear tonal correspondence to Tai. In this thesis we only focus on the evidence from the sound rules of historical Chinese loanwords (OC and MC loans) in Tai languages. These Chinese loans present a clear regular tonal correspondence to the source language, and this indicates that to a certain extent, the non-tonal elements in the earlier period of PT and the phonetic shape of tones in the later period of PT should have been similar to those in OC and MC respectively. 94 The phonetic shapes of PT tones suggested by Liang and Zhang (1996) and Gedney (1989b) have the same shape, but based on different evidence. Their common agreement is that PT might not have been a tonal language, but might have a structure with four types of syllable finals, namely plain, fricative, glottal, and nonglottal stops. This might be why they do not aim to reconstruct the non-consonantal phonetic shapes of tones like pitch height, contour, and vowel duration. Another reason is that most previous studies do not reconstruct the phonetic shapes of PT tones because of the difficulty. Just as Haudricourt states that “once it is established, the tonal system evolves without regard for its old etymological pitch levels” (Haudricourt 1972: 63), Fu (1995: 85) has pointed out that tonal systems have a life of their own, and “there seem to be no guide-lines for development after tone split, with different dialects changing in different directions”. Although the four tonal categories of PT have been reconstructed based on the regular tonal correspondence among Tai daughter languages, the specific phonetic characteristics of PT tones which include pitch height, contour and vowel length seems very difficult to reconstruct by the comparative method because the absence of unity in phonetic shape in Tai tones among different daughter languages faces a huge challenge of reconstructing. Just as Li (1977: 24) points out the difficulty and only gives the four tonal categories without the specific phonetic shape, most Tai linguists do not aim to reconstruct the phonetic shape of PT tones. As what has been reviewed in §2.2.1, in spite of the difficulty Pittayawat Pittayaporn (2009) has provided a breakthrough on reconstructing the phonetic shape of PT tones by internal reconstruction in specific daughter languages and comparative method among those selected Tai varieties, and has given the shapes “low, rising, long, and creaky” to PT Tone *B and “high, falling, short, and glottal constriction” to PT Tone *C. However, we have indicated in the second clarification above that in different periods PT must have had the similar non-tonal and similar tonal shapes with their counterparts in OC and MC respectively. In other words, unlike previous studies that treat phonetic shapes of tones in PT as a synchronic feature and do not reach an agreement, in this thesis the tonogenesis in PT is seen to undergo diachronic processes and is divided into at least two different periods. In the earlier period it is speculated to have had non-tonal structure, and in the later period it is proposed that tones arose to present as the four tonal categories *ABCD under MC influence. Therefore, the hypothesis of phonation type (or voice quality), vowel duration of Tai tones (or their pioneers), the pitch height and contour of PT should be treated as applied to the later period of PT. The following sub-sections will demonstrate the theme of the two principled clarifications above. 95 Combining with the previous studies on PT tones especially the tonal correspondence among Tai, Chinese, Vietnamese, and Miao-Yao languages, the evidence of tonal comparison among thirty Tai varieties in this thesis suggests a result that PT tonal contrast arose by decaying of the syllable final consonantal system, namely the weakening or dropping of *-h and *-ʔ. For keeping the semantic distinction from the original plain syllable (without obstruent syllable finals), the decaying of syllable final -h gave an original falling tone *B, and the weakening of *-ʔ gave an original rising tone *C, to contrast to the least marked level tone *A on smooth syllables. Since *D was on checked syllables with oral stop finals preserved, it was not necessary for it to have any tonal feature to contrast to the *ABC tones. Thus, on the basis of this hypothesis, the origin of PT tones is as the following speculations: plain syllables with modal voice gave an unmarked level tone *A, syllables with final aspiration -h decayed to give an high-falling tone *B on relatively long vowel duration, syllables with final glottal stop -ʔ decayed to give a low-rising tone *C on relatively short vowel duration, and syllables with final oral stops -p/-t/ -k were preserved to consist the fourth tonal category *D with a high-falling contour like its counterpart tone *B. The most direct evidence of this hypothesis is the external comparison among Tai and those surrounding tonal languages. PT Tones *A, *B, *C, and *D correspond to the above-mentioned Chinese/Vietnamese/Miao-Yao tones *A (level/ngang-huyền), *C (falling or departing/hỏi-ngã), *B (rising/sắc-nặng), and *D (entering/sắc-nặng) respectively, as shown in Table 20 in §2.2.2. The typological evidence for this statement is provided by the previous studies although different studies have different points of view about the arising of Tai tones and the tonal features in PT. Here we firstly focus on additional evidence from the sound rules of historical Chinese loanwords (OC and MC loans) in Debao Urban Yang Zhuang (a CT variety) as a representation of the Tai languages, as shown in Table 30. For comparing the OC and MC loans in Yang Zhuang with their source from the borrowing language, the Chinese words are demonstrated by Cantonese instead of Mandarin, because Cantonese integrally preserves the tonal system from MC, in contrast Mandarin has developed its new tonal system beyond clear recognition. Note that the difference between OC and MC loans in Zhuang are ignored in the chart because they share regular tonal correspondence, and there is not yet a full study on this. 96 Table 30 Tonal correspondence between Cantonese and Debao Urban Yang Zhuang (DB) Tonal category A1 A2 B1 B2 C1 C2 DS1 DL1 DS2 DL2 Tai tone No. in 1 2 5 6 3 4 7 9 8 10 Chinese terms yi ̄n yáng yi ̄n yáng DB tonal value 453 31 55 Cantonese value 55 21/11 33 Gloss Series 1 open time to Cantonese hɔi¹ ʃi² kwɔ5 pun6 DB urban kʰaj¹ ɬej² Gloss Series 2 book Cantonese DB urban Sinitic circle pi ́ng pi ́ng yi ̄n yáng qù qù shàng xià yi ̄n yáng rù shǎng shǎng yi ̄n rù rù 33 24ʔ 213ʔ 55 55 33 33 22 35 24 55 33 22 22 horse seven deliver Buddha wipe kwɔŋ3 ma4 tʃʰat7 fat9 fat8 mat8 kwa:5 pu:n6 kwa:ŋ³ ma4 tɕat7 fa:t9 pat8 ma:t10 first save use to fry morning north eight ten candle ʃy¹ tʰau² kau5 jʊŋ6 tʃʰau³ tʃou4 pak7 pat9 ʃap8 lap8 ɬøɥ¹ taw² kjaw5 jʊŋ6 ɕa:w³ tɕaw4 pak7 pe:t9 tɕəp8 la:p10 pass company wide (October) Just like in the chart above, the correspondence of tones between Yang Zhuang and Cantonese are clearly regular. This regularity is not built on the similarity of phonetic shape of their modern tones at all (since their tonal phonetic shapes are quite different as shown in the chart), but on the same tonal roots from Ancient Chinese, namely OC and MC. Actually in the chart all the Yang Zhuang glosses are of OC and MC loanwords, which are actually not directly borrowed from Modern Cantonese but from OC or MC via the ancient Pinghua in history according to the previous studies of the ancient Chinese loans in the Zhuang languages, such as Zhang (1982) and Lan (2001). It is assumed that at the time when PT borrowed the earlier layer of Chinese loanwords, it adopted or adapted those loans with the similar phonetic shapes of non-tonal elements in OC and tonal categories in MC. In other words, it is proposed that the two corresponding counterparts in PT and OC/MC had similar original non-tonal elements/tonal forms, so that their daughter languages such as Cantonese developing from OC/MC and Yang Zhuang developing from PT keep this correspondence of tones on shared vocabularies due to the tonal development together in similar patterns as an areal feature. 97 This indicates that to a certain extent, in the earlier period of PT the phonation types of the voice qualities, the non-tonal phonetic shapes of the later tonal categories, should have been similar to their counterparts in OC and MC. For this reason, in different periods PT must have had the similar non-tonal and similar tonal shapes with their counterparts in OC and MC respectively, namely non-tonal structure in the period coexisting with OC, and the four tonal categories *ABCD in the period under MC influence. In the later period of PT, four tonal categories were established by being influenced by MC tonal development, and Tone *A might be level or unmarked like MC level tone píng, Tone*B might be falling like MC departing (falling) tone qù, *C might be rising like MC rising tone shǎng, and*D should be with final stops -p, -t, and -k like MC entering tone rù and as what the most majority of Tai languages still preserve. For this reason, Tai tonogenesis should have been very similar to that in OC and MC, namely the decaying of syllable final -h giving a falling tone *B, and the weakening of *-ʔ giving a rising tone *C, as what will be demonstrated in the following sections. In this sub-section, the specific phonetic shapes of PT tones are discussed by comparing the three arguments among Pittayawat Pittayaporn (2009), Liang & Zhang (1996), and Gedney (1989b) shown in Table 29, and by analyzing the tone data from thirty Tai varieties recorded by me as shown in the following Table 3126. 26 Since marked phonation types or voice qualities found with tones is crucial to the reconstruction of phonetic characteristics or shapes of PT tones, any specific phonation types or voice qualities found accompanying with tones are labeled with tonal value scales in Table 31, like C2 tone in Bangkok Thai is described as 453ʔ where the final -ʔ refers to a final glottal constriction. In Table 31, voice qualities on tones like creakiness and final glottal constriction are mainly determined by auditory identification. Some special issues are determined by PRAAT (Version 5.1.34), like the released final glottal stop in Dalong (analyzed in Figure 13 in §4.1.3.3), and the final-aspiration-like soft offset in Debao Urban (see Figure 14 in §4.1.3.3.2). In Table 31, -ʔ following the tonal numerical value indicates final glottal constriction (like 453ʔ in Bangkok Thai), or even an audible unreleased or slight released final glottal stop before pause or open juncture (like 24ʔ and 213ʔ in Debao Urban). Creakiness is marked as ~ put right after the lowest number within the tonal value since creakiness always occurs as of the characteristics of low pitch. Since almost all tones on checked syllables have glottalizations (glottal constrictions and/or creakiness), the voice qualities on the D tones are omitted in the chart. The shaded boxes in the chart refer to those tonal splits with “tonal flip-flop” or “voicehigh” phenomenon. Series 1 including all tones conditioned by proto-voiceless sounds (A1-A, A1-C, A1-U, A1-UC, and A-G). Therefore in some Tai varieties Series 1 contains more than one tonal value, such as A1 in Bangkok are 24 (A1-A and A1-C) and 33 (A1-U, A1-UC and A1-G) due to different groups of initial. 98 Table 31 Phonetic shapes of tones and their voice qualities in Tai varieties investigated Language SubGroup SWT CT Variety Name NT *B C* D* A2 B1 B2 C1 C2 DL1 DL2 DS1 DS2 Bangkok 24; 33 33 21 41ʔ 41ʔ 453ʔ 21 41 21 45 Songkhla 41; 453 31 41; 453 213 55; 44 22 55;44 213 55;45 21 Khon Kaen 24 35 33 33 41ʔ 21 41 23 45 Chiang Rai 13 33 22 31 41ʔ 453ʔ 22 31 24 35 Tai Yai 24 45 11̰ 33ʔ 33ʔ 41ʔ 11 33 55 41 Tai Lue 55 31/335 35 33 13ʔ 11ʔ 35 33 55 33 Quan Son Tai 453 322 55 35ʔ 35ʔ 41ʔ 55 35 41 41 Debao Urban 453; 31 31 55; 33 33 24ʔ 21̰3ʔ 55; 33 33 55 33 Debao Dalong 53; 31 31 454; 332 332 24ʔ 33 45 21 Debao Lüliu 453; 31 31 55; 33 33 24ʔ 33 55 33 Jingxi Hurun 51̰; 31 31 334 223 45ʔ 21̰3ʔ 55; 33 22ʔ 33 22 55 13 Jingxi Urban 53 31 453 131 33ʔ 21̰3ʔ 45 21 33 21 Jingxi Anning 51̰ 31 453 131 33ʔ 21̰3ʔ 45 13 33 21 Myang 55 42 35 33 24ʔ 31ʔ 35 33 55 42 Pyang 13 33 212 42 45ʔ 53ʔ 21 42 55 33 Min 24, 35 55 32 42 11ʔ 53ʔ 32 42 32; 55 55 Xiangdu 453, 31 31 33 11̰ 24ʔ 24ʔ 33 11 33 11 453 31 33 11̰ 24ʔ 11̰3ʔ 33 11 55 32 13ʔ, 34ʔ 11̰ʔ 34; 33 33 13; 34 11 Daxin Baoxu NT? *A A1 Daxin Naling CT?/ Tonal categories and their tonal values on smooth syllables 453, 53 Lingding 353, Zuozhou 554, 11 21ʔ; 41ʔ 21̰3ʔ 45; 33 31 34, 33 33 11 33, 31 31 24ʔ 42ʔ 33; 31 31 55 11 Baiji Sanxiang 53 331 343 31 445ʔ 445ʔ 445 31 55 33 Shuangding 33 31 24 22 35ʔ 44ʔ 35 22 35 44 Nung An 554; 33 33 11̰2 31 24ʔ 42ʔ 11 31 55; 33 33 Longsang 35 55 31 53 213ʔ 33ʔ 21 33 55 33 Dongling 11̰ 553 242 53 44ʔ 24 53 22; 44 44 Shanglin 35 221 33 53 55 31 33 53 55 32 Du’an 51̰ 231 33 31 553ʔ 24ʔ 33 31 55 35 Yizhou Suogan 51̰ 231 33 11̰ 42 24ʔ 33 23 55 23 Huishui Bouyei 24ʔ 11̰3ʔ 55ʔ 41ʔ 33; 22 22 55 41 24 113 33 11 55ʔ 24ʔ 31 31 55 24 55 24 Zhenning Bouyei 99 224ʔ; 44ʔ Since PT Tones *A was undisputedly speculated to be unmarked on plain syllables and *D was on checked syllables, the discussion of the PT tonal voice qualities of these two tones will not be given unnecessary details. Therefore, the following discussions will focus on the hypothesis of PT tonogenesis mainly surrounding the B and C tones. Concerning Pittayawat Pittayaporn’s hypothesis of the phonetic shapes of PT tones, we first look at the pitch heights which are speculated by him as low for *B and high for *C. The results of tonal development among different Tai daughter languages present complex diversity with common “voiced-low” situation in the most NT and CT varieties but “voiced-high” phenomenon in some dialects especially those in SWT. The “voiced high” phenomenon is caused by “tonal flip-flop” suggested by many linguists (see §2.2.2 and Glosarry), and it explains that at the time of registersplits tone pitch always started from “voiced-low”, but the later development may change from high to low and low to high after establishment of tone. Since “voiced- high” phenomenon is found in some languages and is not usual “cross-linguistically” (Fu 1995: 81), this explanation is treated as the most plausible one within the many explanations of “voiced-high” phenomenon reviewed in §2.2.2. In Table 31, nine Tai varieties out of thirty have tonal flip-flop (cf. §2.2.2), and four of them are found in the SWT areas which are treated as the areas of Tai emigration in history. Tonal flip-flop found in the two CT and the three NT varieties is also in the high lands at the western corner of Guangxi. These areas are historically treated as of new territories of Tai emigrations from their Tai homeland in the east, which is “the region encompassing parts of the watersheds of the Xi River system in Guangxi extending to northern tributaries of the Red River in Vietnam” (Diller 1998: 15). This suggests that tonal flip-flop phenomenon is of secondary development in the new Tai territories. In addition, only four from these nine Tai varieties have a total tonal flip-flop, the other five still retain some of the remnants of the original “voicelow” splits on some of the tonal categories, like the Tones A in Bangkok Thai (SWT) as well as the Tones DS in Fuping (CT) and in Longsang (NT). Shanglin (NT) even only has tonal flip-flop on the B column. This supports the idea that tonal flip-flop is not usually “cross-linguistically” and is of secondary development. Therefore, the internal reconstructions based on the phonetic shapes of tones in modern Tai varieties, specifically pitch height which may be a result of “tonal flip- flop” in some specific Tai varieties, may be untenable. The pitch height of PT Tones 100 B* and C* reconstructed by Pittayawat Pittayaporn (2009) based on the comparison among these internal reconstructions are considered to be false in this thesis. Opposed to Pittayapon’s hypothesis, when PT tones initially arose in the later stage of PT, the pitch height of Tones B* and C* should be reconsidered as high and low respectively according to the following discussion in §4.1.3.2 and §4.1.3.3 about the other aspects of phonetic shapes of these two tones, since pitch height may be influenced by contours and phonation types. The contour and the vowel duration of PT tones have to be discussed together since in Pittayawat Pittayaporn’s reconstruction these two shapes are related with one other. The reason that Pittayawat Pittayaporn (2009) gives the shapes “rising and long” to PT Tone *B and “falling and short” to PT Tone *C is that Tone *D normally followed the same pattern of tonal splits and mergers with Tone *B in the majorities of Tai languages, and Tone *D further split into DL preserving the same patterns with *B and DS coming to be identified with *C instead of *B. He posits vowel duration being involved in conditioning tonal behaviors, and assumes that vowel duration in Tone *B must be relatively longer than in Tone *C, and states that “both experimental typological studies have shown that vowels under rising tones tend to be longer than vowels on level and falling tones” (Pittayawat Pittayaporn 2009: 276). Pittayawat Pittayaporn’s observation is a plausible explanation if we only focus on the common collocation of vowel duration and pitch contour. However, if we consider other aspects, some flaws can be perceived. First, his additional evidence for Tone *C which is that although either level or falling contours could be reconstructed for both *A and *C tones, the glottalization of *C indicates that it should have been a falling contour. Second, the left contour rising is assigned to *B because “languages that point to a rising contour in *B are spoken very far apart” (Pittayawat Pittayaporn 2009: 276) to make this contour for *B seem to be a case of retention (cf. §2.2.1). These two points are treated to be arbitrary because we have discussed the untenable reconstruction of phonetic shapes of proto-tones including *A *B, and*C. Above all, many scholars have provided previous studies on tonogenesis, which suggests that glottal segments can cause a rising tone especially in Asian languages. For example, in the early step of the process of tonogenesis in East and Southeast Asian languages, a rising tone normally arose by dropping the syllable final -ʔ, such 101 as the sắc-nặng tone in Vietnamese (Haudricourt 1954a), shǎng ‘rising tone’ in Chinese (Pulleyblank 1962, 1978; Mei 1970, 1982; Norman 1988), and the Lahu rising tone caused by dropping the initial and the final -ʔ’s (Matisoff 1970) (cf. §2.2.1). Therefore, PT tone *C which arose by dropping or weakening the final -ʔ or a glottal constriction should not always be a falling tone even if its duration was actual shorter than Tone *B in history. For the second point, as we have discussed that tonal unpredictable development could have occurred in history, the rising contour for tones developing from *B in specific modern Tai varieties might not have been a retention, therefore the point of view that *B was a retentive rising tone is also untenable. Therefore, in this thesis PT proto-tones *B and *C are proposed to be falling and rising respectively due to above-mentioned MC counterparts influencing as well as the typological collocation in MSEA languages between tone pitch and phonation voice (fricative final causing falling tones and -ʔ causing rising tones) during the period that tones originally arose. The premise of this proposal is the voice quality of these two tones, which is discussed below. The rest of PT tonal phonetic shape is the phonation type or voice quality. Because previous hypotheses of straightforward of Tones *A and *D are less argument, the following discussion focuses on the voice quality of Tones *B and *C. Before continuing discussing the voice qualities of the tones B and C in modern Tai varieties and to prove the speculation of the phonetic characteristics of PT Tones *B and *C, it is necessary to distinguish the terms glottal stop, glottal constriction, and creaky voice, which are all found in tones in modern Tai varieties. Note that Gedney uses the term “glottal constriction” loosely without distinguishing between creakiness and final glottal constriction found in tones, and he even treats them as the same phonetic phenomenon by the following statement: “… the tones (usually two) that have developed from the earlier C tone are characterized by glottal constriction, sometimes called ‘creaky voice’”, and this is “manifested by extreme tension of the glottis throughout the vowel, increasing gradually, with closure in glottal stop at the end of the syllable before pause of open juncture” (Gedney 1989c: 209-210). Referencing previous studies such as Gordon & Ladefoged (2001) and Michaud (2004), Pittayawat Pittayaporn (2009: 277) distinguishes creakiness from final glottal constriction, by defining the former as an alternate name 102 ‘laryngealization’, which refers to “the irregular, low-frequency vibration of the vocal folds extended over a relatively long period of the duration of the vowel”, and by defining the latter as “a complete or almost complete closure of vocal at the end of rime”. On the basis of this distinction he has provided the aforementioned hypothesis giving “creakiness” and “final glottal constriction” to PT Tones *B and *C respectively (Pittayawat Pittayaporn 2009: 277-281). However, the types of voice quality defined by Michaud (2004: 120) are actually three. He gives the three definitions as follows: (1) ‘(Final) glottal stop’ is a gesture of vocal fold closure that limitedly arouses specific voice quality on the syllable. (2) ‘Glottal constriction’ refers to “a tense gesture of adduction of the vocal folds that extends over the whole of a syllable rhyme”. (3) ‘Laryngealization (creaky voice)’ is an “irregular vocal fold vibration” and “is not tense in itself”. He furthermore points out that the term ‘glottalization’ covers ‘glottal constriction’ and ‘laryngealization’ (Michaud 2004: 120). This shows that Pittayawat Pittayaporn’s definition of ‘final glottal constriction’ is used as a cover term for Michaud’s ‘final glottal stop’ and ‘glottal constriction’. The relationships of the definitions above are concluded as in the following Table 32. Table 32 The correspondence among the voice qualities defined by different proposals Proposer Correspondence (vertically) Gedney (1989) Pittayawat Pittayaporn (2009) Michaud (2004) Glottal constriction or glottalization final glottal constriction final glottal stop glottal constriction creakiness laryngealization (creaky voice) glottalization Combining with the previous studies, in this thesis we synthesize the definitions above to define as follows: ‘Final glottal constriction’, which is also called ‘glottal constriction’ as its shortened form, refers to a tense gesture over the whole syllable rhyme ending in a complete closure of the vocal folds. ‘Final glottal stop’, ‘creaky voice/creakiness/laryngealization’, and ‘glottalization’ are in accordance with the definitions provided by Michaud (2004: 120). Final glottal stop -ʔ abundantly found in the C tones in modern Tai varieties normally has coarticulatory effects on the preceding vowels to give a tense gesture 103 that extends over the whole of a syllable rhyme, and this tense gesture is called ‘glottal constriction’. This may be why Pittayawat Pittayaporn uses ‘final glottal constriction’ to name this glottal phenomenon on Tai tones, and it is used in the same way in this thesis. However, the glottalization mainly found in the C tones in my fieldwork for this thesis has its own symbolic feature, which presents to always end in a complete closure of vocal fold or rather a final glottal stop -ʔ, and this glottal stop is even slightly released before pause or open juncture in some varieties recorded from some LRPs, like in Debao Dalong (CT), Jingxi Nung An (CT/NT?), and Debao Longsang (NT). Take the term ˀba:nʔ24 ‘village’ with the C1 tone in Debao Dalong Yang Zhuang (CT) as an example shown in Figure 13. Figure 13 Voice analysis diagram of ˀba:nC1ʔ ‘village’ in Debao Dalong by PRAAT The selected part of the diagram indicates the audible released -ʔ even following a nasal coda -n, and this shows a counterevidence of that syllable-final obstruents (like -p, -t, -k and -ʔ) in tonal languages in East Asian and MSEA are known to be nonreleased, although the released segment actually disappears along with the final -ʔ at a non-final position in a carrier sentence. There is no indication that medial glottalization throughout the whole syllable can exist by itself without ending in a final glottal stop before pause or open juncture. This is very like the common presence of a glottal stop or a glottal constriction in association with a final oral stop in languages in China and MSEA, like in Fujian (Iwata et al. 1979), Cantonese (Iwata et al. 1981), Thai (Henderson 1964; Harris 2001), and Vietnamese (Thurgood 2002). It indicates that the glottalization manifested by extreme tension of the glottis 104 throughout the vowel on the C tones may be an effect triggered by the final glottal stop. Therefore, this thesis suggests the final glottal stop -ʔ to be the most symbolic feature of historical residual from PT *C instead of the medial glottalization over the whole of a syllable rhyme. In a similar way, creakiness found in some tones will be proved to be an effect of a low pitch, a dipping tone, or a rapid falling tone, instead of an inherent feature of some specific tones. Since there is more obvious evidence for proving the syllable final -ʔ accompanying with Tone *C than those evidence for proving the phonetic characteristics PT Tone *B, we discuss firstly from Tone *C. The voice qualities found in modern Tai varieties from my fieldwork are summarized as shown in Table 31 in §4.1.3. Among the different hypotheses summarized in Table 29, Sagart’s argument is totally different from the others on voice quality of Tones *B and *C, and this has been rejected by Pittayawat Pittayaporn (2009) (cf. §2.2.1). In this thesis, the points suggested by Gedney (1989c), Liang and Zhang (1996), and Pittayawat Pittayaporn (2009) that the phonetic characteristics of PT tone *C having a final -ʔ or being final glottal constriction are agreed with, since the data personally selected and analyzed by me in this thesis have shown that there are many modern Tai varieties that still preserve robust residues of the glottalized features in the C tones as shown in Table 31. Within the thirty languages personally recorded by me, twenty-six have final glottal constriction ending in a glottal stop on the C tones (C1 and/or C2) no matter what the pitch level or contour of the respective tone. The phenomenon that this glottal constriction with a glottal stop ending is mainly found on C tones has been discovered by Gedney (1989c: 210-211). He points out that many Tai varieties from Li’s CT and SWT such as Thai, Shan, Lao, White Tai, Lungchow, Lungming, and most other CT/SWT languages researched by him, have this glottal constriction on the tone that have developed from PT Tone *C, regardless of pitch level or contour such as the two C tones in Thai (falling and high), in White Tai (low-rising and lowfalling), and in Lungming (mid level and low-falling-rising) (Gedney 1989c: 210). The tonal correspondence among Tai, Sinitic languages, Miao-Yao, and Vietnamese also supports that tone *C in Tai (which corresponds to Tone B in the others) used to have a glottal final. This indicates that the final -ʔ should have accompanied with PT Tone *C but not with Tone *B. Therefore, Sagart’s argument is not supported in this thesis. 105 It is notable that Gedney also points out that the studied NT varieties except Saek, which may be influenced by surrounding SWT languages, do not have this final glottal constriction on C tones (Gedney 1989: 212-213). However, at least our data shown in Table 31 have accounted for the more complicated situation on C tones in NT varieties. At least four varieties within seven NT varieties and all the three varieties whose classification as CT or NT is open to debate have a clear final glottal constriction on C tones. Our LRP from Longsang Zhuang of NT even release the glottal stop after the syllables on C tones when they repeat the terms alone. The remaining three NT varieties which have no glottalization on C tones are Shanglin Zhuang, Huishui Bouyei, and Zhenning Bouyei. Shanglin Zhuang may reflect the situation pointed out by Gedney. The one LPR of this dialect produces none of the tones with a consistent phonation type. However, in the two Bouyei dialects which are spoken far to the north of the concentrated Tai areas, the situation shows secondary tonal reversal on smooth syllables as follows. In Huishui Bouyei (L41), items with A and B tones unexpectedly have a slight or clear glottal constriction, like pja24ʔ (A1) ‘hill’, ma:24ʔ (A1) ‘dog’, kɨn24ʔ ‘to eat’, wa:j113ʔ (A2) ‘buffalo’, si:55ʔ ‘four’ (B1), and ta:41ʔ (B2) ‘river’. In Zhenning Bouyei (L42), items with tones in the B column normally have a slight final glottal constriction, like (ˀda:ŋ33) ka:j55ʔ ‘egg’, ɬej55ʔ ‘four’ (B1), and (ʔə33) ta:24ʔ (B2) ‘river’. These phenomena are very unique because final glottal constriction is mostly found on C tones in many other Tai varieties, as what mentioned above. The problem is readily solved when comparing the tones in the whole tonal system in these varieties. For Huishui Bouyei (L41), the final glottal constriction on A and B tones may have been triggered by the mergers between Tones A and DS, as well as between Tones B and DL (as shown in Table 33). 106 Table 33 Basic tonal patterns in Huishui Bouyei (NT) Phonation types Smooth Syllable Checked Syllable A B C DL DS Proto- Aspirations A1-A: 24ʔ B1-A: 55ʔ C1-A: 33 B1-A: 55ʔ A1-A: 24ʔ voice- Continuants A1-C: 24ʔ B1-C: 55ʔ C1-C: 33 B1-C: 55ʔ A1-C: 24ʔ Unaspirated stops A1-U: 24ʔ B1-U: 55ʔ C1-U: 33 B1-U: 55ʔ A1-U: 24ʔ A1-UC: 24ʔ B1-UC: 55ʔ B1-UC: 55ʔ A1-UC: 24ʔ A1-G: 24ʔ B1-G: 55ʔ C1-G: 22 B1-G: 55ʔ A1-G: 24ʔ Proto-voiced sounds A2: 11̰3ʔ B2: 41ʔ C2: 22 B2: 41ʔ A2: 11̰3ʔ Proto-voiced aspirations & A1/2: 11̰3ʔ B1/2: 41ʔ C1/2: 22 B1/2: 41ʔ A1/2: 11̰3ʔ less Unaspirated stop + *r clusters Glottal sounds breathy sounds C1-UC: 33 The loss or weakening of the syllable final -k is commonly found in some NZ varieties such as Nandan and Tian’e in northwestern Guangxi, and Qiubei in Yunnan (Zhang et al. 1999: 240), as well as in their neighboring Bouyei languages, probably because of the language contact with Southwestern Mandarin and the Loloish languages (according to my fieldwork research)27. In Huishui Bouyei, *-k after a long vowel in PT had been changed to a slight non-phonemic final glottal constriction to merge into their counterpart Tones B (B1 and B2), like pa:55ʔ (DL1>B1) ‘forehead’, jia55ʔ (DL1>B1) ‘guest’, pa:55ʔ (DL1>B1) ‘mouth’, ma:55ʔ (DL1>B1) ‘bean’ (semantic change from ‘fruit’), ta:55ʔ (DL1>B1) ‘to expose under the sun’, tiɑ55ˀ (DL1>B1) ‘cracking’, ˀjiə55ʔ (DL1>B1) ‘hungry’, and ða:41ʔ (DL2>B2) ‘root’; *-k after a short vowel in PT has been changed to a clear non-phonemic glottal constriction to merge into their counterpart Tones A (A1 and A2), like to:24ʔ (DS1>A1) ‘thin bamboo stripes’, pja24ʔ (DS1>A1) ‘vegetable’, ðua:113ʔ (DS2>A2) ‘bird’, ða:113ʔ (DS1>A2) ‘to steal’, and ɕja:113ʔ (DS2>A2) ‘cooked’; this may be why the tonal value of tones on the vertical column of DL and DS has merged into the tonal value of their counterpart Tones B and A; the reason why items with tones on the columns A and B normally have a slight final glottal might be because in this dialect proto-final *-k from Tones DL and DS might have brought about an effect, a slight glottal final -ʔ, into Tones B and A. For Zhenning Bouyei (L42), the glottal constriction on B tone may have been triggered by the mergers between Tones B and DL, as the following analyses: on original dead syllables, *-k after a long vowel in PT been changed to a slight nonphonemic glottal constriction to merge into their counterpart Tones B (B1 and B2), 27 However, most of these languages well preserve the finals -t and -p. 107 like (na:31) pa:55ʔ ‘forehead’, (pou31) ɕiə55ʔ ‘guest’, (ˀda:ŋ33) pa:55ʔ ‘mouth’, (ˀdam33) ma:55ʔ ‘fruit’, ta:55ʔ ‘to expose under the sun’, piə55ʔ ‘cracking’, (tʊŋ31) ʔi:55ʔ ‘hungry’, la:24ʔ (va:j31) ‘root’ and (tiaw33) tɕa:24ʔ ‘rope’; like Huishui Bouyei, proto-final *-k from Tones DL might have brought about an effect, a slight glottal final -ʔ, into Tones B (cf. Appendix C). Therefore, the original final glottal constriction on Tones C in these two Bouyei varieties might have been dropped due to the extrusion of the later development of the final glottal constriction on Tones A and/or B. Furthermore, tonal splits and mergers on the C column in many NT languages indicate the lack of glottalization on C tones may also be a natural quality of the mergers of C1-G (the tonal category conditioned by initial glottal sounds) and C2 tones, which are very rarely found in CT and SWT varieties. The previous Table 33 and the following Table 34 show two examples of tonal mergers of C1-G and C2 in the two NT varieties investigated. Table 34 Basic tonal patterns in Debao Dongling Zhuang (NT)28 Phonation types Smooth Syllable Checked Syllable A B C DL DS Proto- Aspirations A1-A: 11̰ B1-A: 242 C1-A: 224ʔ DL1-A: 24 DS1-A: 22 voice- Continuants A1-C: 11̰ B1-C: 242 C1-C: 224ʔ DL1-C: 24 DS1-C: 22 Unaspirated stops A1-U: 11̰ B1-U: 242 C1-U: 224ʔ DL1-U: 24 DS1-U: 22 less Unaspirated stop + *r clusters Glottal sounds A1-UC: 11̰ B1-UC: 242 C1-UC: 224ʔ DS1-UC: DL1-UC: 24 22 A1-G: 11̰ B1-G: 242 C1-G: 44ʔ DL1-G: 24 DS1-G: 44 Proto-voiced sounds A2: 553 B2: 53 C2: 44ʔ DL2: 53 DS2: 44 Proto-voiced aspirations & A1/2: breathy sounds 553 B1/2: 53 C1/2: 44ʔ DL1/2: 53 DS1/2: 44 In the tonal patterns of Huishui Bouyei (L41) shown in Table 33 and Debao Dongling (L32) shown in Table 34, the same colors in the same chart indicate tonal mergers crossing tonal categories or registers. In Debao Dongling Zhuang and Huishui Bouyei, all the columns except the column C (and its counterpart checked syllable column) show straightforward tonal development, which splits their registered tones between original voiceless sounds (including glottal sounds) and voiced initial sounds. On the other hand, the column C shows non-straightforward 28 The management that treats B1 (242) on smooth syllable and DL1 (24) on checked syllable as well as C1 (224ʔ) and DS1 (22) to be allotones of the same toneme is explained in §3.3.2. Hereafter in the similar situation in all other Tai varieties, this criterion of determining allotones of a specific tone will be throughout implemented. 108 tonal development, which splits its registered tones between original voiceless sounds (excluding glottal sounds) and glottal plus voiced sounds. C1-G going with C2 is quite common in the western half part of NT areas. In Kam-Tai languages, glottal or pre-glottalized initials often have a pitch-lowering effect on the following syllable. Previous studies have found that CT and SWT varieties often have tonal mergers between glottal register and original lower register, namely A1-G and A2 as well as B1-G and B2 (and between their counterpart checked syllable tones), but rarely have this merger on the category C (except in Isan/Lao). The intensive final glottal constriction on the C tonal category in CT/SWT varieties may be the factor preventing the tonal mergers between C1-G and C2. In the early step of the process of tonogenesis in East and Southeast Asian languages, a rising tone normally triggered by the syllable final -ʔ, such as the sắc-nặng tone in Vietnamese (Haudricourt 1954a), shǎng ‘rising tone’ in Chinese (Pulleyblank 1962, 1978; Mei 1970, 1982; Norman 1988), and the Lahu rising tone caused by dropping the initial and the final -ʔ’s (Matisoff 1970). This indicates that the original double glottal segments (initial and final) on the *C tone in CT/SWT may condition a higher pitch on tonal behaviors than the pitch conditioned by the single glottal segment (initial) on the tonal categories *A and *B, therefore in most CT/SWT varieties C1-G always keeps the merger into the high register to be with the tone conditioned by other voiceless sounds, while A1-G and B1-G are often lowered or even merge into A1 and B2 respectively, such as ˀdam31 (A2<A1-G) ‘black’ and ˀjow33 (B2<B1-G) ‘at’ in Debao Yang Zhuang varieties. However, glottal stop sometimes has a pitchlowering effect on the preceding vowel in some Uto-Aztecan languages, like in Hopi (Fu 1995: 64). This is also reflected in NT varieties, which normally do not have the tonal merger between glottal registers and original lower-register on tonal categories A and B, but often have this merger on the C category which should have had been given a pitch-lowering effect on the preceding vowel, perhaps due to the weakening or dropping of the original final glottal stop on the C category. This may interpret the disappearance of the final glottal constriction on C tones in the NT varieties investigated by Gedney (1989c). Hereto, the phonetic characteristics of Tone *C in the earlier stage of PT can be concluded to have been very similar to its counterpart Chinese shǎng (rising) tone to have an affiliated syllable final glottal stop, which is still preserved as a final glottal constriction, sometimes even being a slight released final glottal stop before a pause or an open juncture, in the vast majority of modern CT/SWT and those varieties whose classification as CT or NT is open to debate, and in quite a few considerable 109 modern NT varieties. In the later stage of PT, this final glottal stop was weakened to cause a low-rising contour on relatively short vowel durations due to its inherent glottal constriction. As described in §2.2.1 and §4.1, various authors have proposed different characteristics for PT Tone *B. Gedney (1989c) and Liang and Zhang (1996) propose that it had a final *-h; Sagart (1988) argues that it had a modal voice rime ending with a glottal stop *-ʔ; Pittayawat Pittayaporn (2009) speculates that it was a lowrising tone with a creaky voice. However, the major problem with these proposals is that modern Tai dialects do not have a consistent voice quality on the tones developing from *B. This is apparent when being compared with the commonly preserved glottal constriction on the tones from *C, as shown in Table 31 in §4.1.3. In Table 31, the situation of the voice qualities of the non-C tones on smooth syllables appears as an untidy distribution – some final glottal constriction and creakiness irregularly accompanying one of the tones developing from *B in some modern Tai varieties or from *A tones in others. Therefore, all these hypotheses about PT Tone *B are quite speculative compared to the clear evidence in support of the phonetic characteristics of PT Tone *C. Even so, we have proposed in §4.1.2 that PT Tone *B might have been a falling tone like its MC counterpart departing tone qù, and the falling contour is believed to have been conditioned by the previous syllable finals -h (or -s) at the time29 in the languages in China and MSEA (see §2.2.1), this thesis proposes that this tone should have developed from a final fricative -h (or -s) which agrees with its OC counterpart in an earlier stage of PT to a creaky phonation which was with its MC counterpart in an later stage of PT. The earlier form of this tone agrees with the forms reconstructed by Gedney (1989b) and Liang and Zhang (1996), and the later form agrees with Pittayawat Pittayaporn’s (2009) hypothesis. However, the evidence for supporting hypothesis in this thesis of Tone *B are different from Gedney, Liang & Zhang, and Pittayawat Pittayaporn. In the following discussion, I will determine the residue of the phonetic characteristics of PT Tone *B by a process of elimination. The first proposal to be eliminated is final glottal constriction (cf. Sagart 1988). Glottalization found in A and B tones in the Tai languages are scattered, as shown in 29 MC departing tone is proposed to derive from OC -s or -h finals and to become creakiness during the MC period (Pittayawat Pittayaporn 2009: 280-281). 110 Table 31. In the thirty Tai varieties investigated during my fieldwork, only Thai, Tai Yai, and Quan Son Tai of SWT, and Huishui Bouyei and Zhenning Bouyei of NT have clear final glottal constriction on tones developing from *B. The final glottal constriction of B1 and B2 tones in the two Bouyei varieties has been clarified above as an effect by the mergers from original dead syllables. Moreover, the final glottalized Tone B2 in Thai, Tai Yai, and Quan Son Tai is actually spread from the original *C tone due to the coalescence or merger between C1 and B2 in these varieties. This phenomenon has been pointed out by Gedney, and he further pointed out the reverse that in some Tai varieties such as Khorat and the Yo dialect of Sakon Nakhon glottalization has disappeared from both of B and C tones due to the tonal coalescence involving the C tone (Gedney 1989: 214-215). However, Sagart (1988) accounts that the final of Tone *B should be a glottal stop -ʔ instead of -h based on that Tone *B patterns with Tone *D on tonal split and mergers, and the final of Tone *B should be a complementary stop (-ʔ) since*D always ends on non-glottal stops. Sagart’s argument is not supported in this thesis because there is more evidence supporting that the final -ʔ should be the accompanied feature of Tone *C instead of Tone *B. Moreover, his argument that glottalization in reflexes of Tone B in modern Tai varieties is common is not substantiated. The data from my fieldwork show that the glottalization in reflexes of the B tones is far less than in the C tones, and some of the final glottal constriction on the B tones is actually spread from the C tones due to tonal mergers as discussed above. Besides, the ‘glottalization’ regarded by him in reflexes of the B tones might not have always been the same ‘final glottal constriction’ like in reflexes of the C tones in our definition, but might be ‘creakiness’ which will be verified below. Pittayawat Pittayaporn (2009) also rejects Sagart’s proposals of the phonation type of Tone *B. He suggests that Sagart’s viewpoint that PT *B was ending in a glottal stop could not have been the case by pointing out that a syllable with a glottal stop ending must be expected to have contrasted long and short vowels as CV:ʔ vs. CVʔ, which is no evidence for such contrast in PT30 (Pittayawat Pittayaporn 2009: 252). To sum up, PT Tone *B need not have had glottal constriction even though this feature is found in the tones developing from PT *B tone in some of the modern Tai varieties. Pittayawat Pittayaporn’s clarification is neither supported in this thesis because final glottal stop accompanying with Chinese shǎng tone or Vietnamese sắc-nặnɑ tone (which correspond to Tai tone *C) has been suggested by many previous studies and the vowel length contrast is not always the case. This final glottal stop might be actually more than a simple syllable final since it even occurs after a syllable final nasal before pauses or open junctures in many modern Tai varieties like in Figure 13. Besides, so-called ‘short vowel’ on open syllables in modern Thai is actually followed by a glottal stop before pauses and open junctures, and these vowels are always short and do not have length contrast. 30 111 The second clarification is that creakiness on B tones in some of the modern Tai varieties should not be treated as the earliest origin of the tones themselves, or should not be the phonetic characteristics of *B tone in the earlier stage of PT. Pittayawat Pittayaporn suggests that creakiness found on *B in several modern SWT and CT varieties is inherited, even though many of them are described as glottalization (Pittayawat Pittayaporn 2009: 278). He believes the voice qualities may actually be creakiness instead of final glottal constriction since at least one of the descriptions Nung Fan Slihng from Freiberger (1964) specifically mentions that the tone B2 “ends with a final glottal stop, but laryngealization may occur over all the voiced phonemes”, and Pittayawat Pittayaporn claims that this laryngealization is actually “a remnant of the creaky voice characteristic of PT *B” (2009: 278-280). He even gives additional evidence – the correspondence between PT *B and MC departing tone which is proposed to derive from OC -s or -h finals and became creakiness during the MC period (2009: 280-281). However, creakiness is known as a characteristic of low pitch in tonal languages due to the previous studies, such as Northern Chinese dialects in Shanxi, Beijing, Hebei, Henan, Shandong, and Jiangsu provinces commonly have creaky voice as the feature [+low] (Zhu & Yang 2010). A typical example that indicates creaky voice as the feature [+low] in Tai languages is provided by Mai (2011). He points out that creakiness can be found in the sixth tone (B2) in Qingtang variety of NT from Zhongshan County, Guangxi. The sixth tonal category in this NT variety do not have a stable pitch, but always accompanies with a creaky voice variously presenting as follows: when the creaky voice appears at the central of the syllable, the tone presents a falling-rising contour (31̰3 or 31̰4); when it appears at the initial of the syllable, the tone presents a rising contour (1̰5 or 1̰55 and 3̰ 5 or 3̰55 due to different environment); and when it appears at the end the syllable, or appears during all the duration of the syllable, the tone presents a midfalling or low level contour (31̰ or 11̰) (Mai 2011: 22-23). That is to say, the creaky voice always accompanies with the low pitch or the so-called “shallow tone” (a tone with falling first then rising contour) in this dialect. In Table 31, we can also see the creakiness occurring ubiquitously on the low fundamental frequency of the pitch on a syllable, no matter any tonal categories (including the C and D tones). For example, in Yizhou Suogan Zhuang (NT), Tone A1 (51̰) has a strong creakiness triggered by its rapid falling contour from the highest pitch 5 to the lowest pitch 1, and B2 (11̰) also accompanies a weak creakiness due to the lowest pitch. Huishui Bouyei Tone A2 (11̰3) normally has a creakiness triggered by its extremely low level pitch 11 before rising to a higher pitch level 3. If we admit the method of tonogenesis, we must accept that tonal languages have developed from non-tonal 112 proto-languages; consequently, an original pattern of manifestation of the voice quality in a proto-language must not have been a fixed low pitch to automatically have coarticulatory effects to produce a creaky voice. Therefore, the treatment that the creakiness found on B tones is account for the genetic characteristics of the voice quality of PT *B tone is like to put the incidental before the fundamental. Moreover, the creakiness confirmed in my fieldwork are even more found in A and C tones than in B tones as shown in Table 31, because low tones, dipping tones, and rapid high falling tones are more concentrated in A and C tones than in B tones in the Tai varieties investigated in my fieldwork. In short, creakiness may be an effect of low pitches, dipping tones or rapid falling tones, and should not be treated as a cause of tonogenesis. Therefore, it is indeed hardly to be proved to be a genetic feature of PT *B tone. The rest of the discussion comes back to the hypothesis of final -h for the origin of PT tone *B. Although there are no modern Tai dialects which keep a voice quality like final aspirations, or in other words a final *-h like the parallelism in tonogenetic processes proposed by Haudricourt (1954a, 1954b) and Pulleyblank (1962, 1978) in Vietnamese and/or Chinese, Gedney (1989b: 91, 1989c: 208) speculates that it also had a final *-h mainly based on its counterpart DL tone, and on the evidence that Thai treatment of using Tone *B to borrow Indic loanwords ending in -ha. Liang and Zhang (1996) also propose a hypothesis totally agreeing with Gedney’s speculation, although their arguments are not based on the same evidence, but on their own evidence from the tonal correspondence of some shared vocabularies between Tai Lue and Wa, as has been reviewed in §2.2.1. Their detailed evidence is partly shown in Table 35. Table 35 Some shared vocabularies with the B tones (Tones 5 and 6) in Dai and with syllable final -h in Wa (adapted from Liang & Zhang 1996: 816) Dai (Tai Lue) Wa Glosses pha:w phruah kɔj kɔih ‘to broadcast (seed)’ thaj tauh bi: bih xe: rɛ̀h haw ràuh lɛ: lɛ̀h vɛ: ‘to draw out, to dig’ ‘to redeem’ ‘to crack’ ‘to harrow’ ‘to bark’ ‘to cut, to split, to peel’ ‘in passing, incidentally’ vɛ̀h 113 However, Gedney’s hypothesis of -h for the origin of PT Tone *B has been refuted by Pittayawat Pittayaporn (2009) by being pointed out that the evidence of Indic loanwords may have been processed by loanword adaptation. Loanword adaptation in Thai can assign Indic loans ending in a final -ha to an open syllable with tone *B, like Thai pʰa:B2 < Indic vāh(a) ‘bearer’ or Thai ʔutDS1 sa:B1 < Indic utsāh(a) ‘to make an effort’. This is due to the same pattern as Indic loans ending in other obstruents including the stops -p, -t, -c, and -k and the fricatives -s, -š, and -s were regularly assigned to the tonal category *D which is the check-syllable counterpart of *B. In those Indic loans with tonal category *B, the final -h in Indic source was in fact dropped in the process of loanword adaptation because final fricatives are not permissible in Thai (Pittayawat Pittayaporn 2009: 252-253). Pittayawat Pittayaporn’s point of view is convictive since loanword adaptation is commonly believed to cause loans in a borrowing language to change its original phonetic shapes in the borrowed language to accommodate to the sound system of the borrowing language. Therefore, Liang and Zhang’s argument is also not always reliable, because the reliability of their proposal is queried to be unstable by the loanword adaptation. The most crucial point is whether Liang and Zhang’s evidence is Tai Lue loans in Wa or Wa loans in Tai Lue. Unfortunately, they only illustrate a comparison of “corresponding words” shared between Dai and Wa, but do not indicate the original source of these words, thus loanword adaptation seems to fall into a confusion. If it is assumed that these words are Wa loans in Dai, the process can still assign those words with final -h in Wa to tonal category *B on an open syllable in Dai due to loanword adaptation just like Indic loans in Thai pointed out by Pittayawat Pittayaporn. However, because at least the words tʰajB1 ‘to redeem’ and hauB1 ‘to bark’ in Dai are from typical Tai source, their counterparts tauh ‘to redeem’ and ràuh ‘to bark’ in Wa should be treated as Dai loans. Therefore, the direction of loans from Dai to Wa is plausible. The final -h of those counterparts in Wa indicates that the source in Dai may have had a special laryngeal feature since -h is a laryngeal fricative. Loanword adaptation can hardly explain how Wa has adapted Tai Lue B tones (with a modal voice) by a syllable final -h. Therefore, tone *B in a specific history of Tai Lue should have had at least a special laryngeal feature since -h is a laryngeal fricative. So far, the final -h for the phonation type of *B is the most probable hypothesis after eliminating the possibilities of glottalized feature (final glottal constriction or final glottal stop) and creakiness, based on the tonal correspondence among Tai, Chinese, Vietnamese, and Miao-Yao, as well as on the loanword adaptation in Tai Lue-Wa shared vocabularies. 114 Therefore, even if at a specific period PT did have creakiness on its tonal category *B like its counterpart MC q̀ (departing) tone which is proposed to derive from OC s or -h finals and became creakiness during the MC period (Pittayawat Pittayaporn 2009: 280-281), this specific period of PT should be later than the exact tonogenesis period due to the regional typological tonal correspondence among Tai, Sinitic, Vietnamese, and Miao-Yao language families. The real cause of this proposed “creakiness” on *B tone should be an original high-falling tone, which easily gave a creakiness added to the pitch. This original falling tone typologically developed from the decaying of syllable final -h. To confirm the hypothesis of PT *B as from the decaying of syllable final -h, it is necessary for us to discuss a series of tones with complex-contour like dipping (e.g. 213ʔ in Debao Urban), convex (e.g. 453 in Daxin Baoxu), delayed falling (e.g. 554 in Zuozhou), and delayed rising (e.g. 112 in Nung An). When being individually read, these kinds of complex tones always cause the duration of the syllable longer than a simple tone namely level, falling and rising. In the Tai varieties investigated as shown in Table 31, the complex-contour tones mainly found in the A and B categories normally present as a modal voice over the whole syllable rhyme, but the complex-contour tones found in the C categories normally accompany with a final glottal constriction, like the C2 tone as 213ʔ in Debao Dalong. However, if we dig into the complex tone found in the A and B categories, we still find that at the end of the syllable before pause and open juncture they tend to have an aspiration-like offset or ending. This offset actually consists of the final falling point or the final rising point of the complex-contour, such as the final point (2) of a mid falling-rising tone (332, the B1-A or B2 tone) in Debao Dalong, and the final point (3) of a highrising-falling tone (453, the A1 tone) in Daxin Baoxu. It is this kind of offset makes the tone always being a waved contour. Figure 14 shows an example of complexcontour in B tone in Debao Dalong Yang Zhuang. Figure 14 Sound analysis diagram of tʰa:n332 ‘charcoal’ and pʰa:332 ‘to chop’ in Debao Dalong Yang Zhuang (CT) by PRAAT 115 In Figure 14, the voice quality of B1-A (332) tone in Debao Dalong Yang Zhuang presents as modal voice with a very slight aspiration-like final before pause or open juncture, and I will call this “soft offset” instead of a final -h because it is too slight to be audible, and is not phonemic or tonemic. The important point is that the soft offset and the last falling point (2) of the tonal value (332) are actually integrated. When the syllable is on a non-final position of a carrier sentence, the falling point (2) and the soft offset (-ʰ) both disappear to make the tone be 33 instead of the full form 332. It means the last falling point (2) and the soft offset of the full form are actually the same thing, which presents as a falling point (2) continuing the previous mid level contour (33) from the viewpoint of tonal value, and it presents as a soft offset from the viewpoint of voice quality. Such kind of dipping, convex, delayedrising and delayed-falling contour tones are also found in the C tones in some Tai varieties, like C2 (453ʔ) in Bangkok Thai. In the environment of being on a non-final position of a carrier sentence, both the final glottal offset -ʔ and the falling point of the contour (3) disappear since they are also integrated, or are the same thing divided into two concepts due to the different points of views. This indicates that the soft offset of A/B tones contrasts to the final glottal constriction presenting on the C tones in many modern Tai varieties. Many complex-contour tones found in the B and A tones in Tai varieties investigated from my fieldwork are shown in Table 31. This final aspiration-like soft offset of a syllable has not been found to be discussed in previous studies, and it needs more discussion and demonstrations to be proved. Nevertheless, it deserves further attention because this thesis considers it is crucial to the hypothesis of the phonetic characteristics of PT Tone *B. A hypothesis built on the basis of the correspondence among PT Tone *B, OC/MC tonal category qù (departing tone or original falling tone), and Vietnamese tonal category C (hỏi-ngã) is supported by the following mutual-correlative speculations involving complex-contour tones. 1) It can be speculated that in the period after tone arose, the original -h final collapsed to cause a falling tone, but the duration of the syllable with original -h final are lengthened to make up the decaying of the final -h, since syllables with a final -h are phonetically longer than modal syllables (Tone *A), syllables with a final glottal -ʔ (Tone *C), and syllables with an oral stop (Tone *D). 2) Later, the lengthened back part (a soft offset inherited from original -h) of the duration are reflected on tonal contour to be a rising point, to make the tonal value present a high-falling-rising contour, or in other words a dipping tone; the 116 important point of this process is that the first complex-contour tone in Tai is speculated to appear on *B. 3) During the periods of the secondary development of tones in different Tai daughter languages, the original contours of all tonal categories might be changed to other contours, since “once it is established, the tonal system evolves without regard for its old etymological pitch levels” (Haudricourt 1961); consequently, in some Tai varieties, the original dipping tone (falling-rising) developing from *B might have been changed to other complex-contours, like convex (rising-falling), delayed rising (level-rising), and delayed falling (levelfalling) which have been discussed above and showed in Table 31, or even have changed to other simple-contour such as level, rising and falling (also showed in Table 31) during later development especially at the period of register splits because tonal contours in a language are limited. 4) Any change in a tone may push another tone to change its original contour for avoiding the tonal merger to keep the tonal distinctions; because the change sometimes presents as a contour-exchange between two different tones, the changes of the original complex contour *B tone in certain Tai varieties may urge the complex-contour to spread to the other tonal categories, especially the A and C tones which are also on smooth syllables. 5) However, C tones in most modern Tai varieties (especially in CT and SWT) still robustly preserve the remnant or the residual of the phonetic characteristics of PT tone *C, a final glottal constriction or even a clear accessary final glottal stop, even though some of these C tones have been changed to the complex-contour spread from original Tone *B, such as the C2 tone (213ʔ) which is a dippingcontour ending in a rising point (also a simultaneous glottal stop) in Debao Dalong. This makes the C tones have the highest distinguishing degree to recognize its original phonetic characteristics among all the smooth-syllable tonal categories in most modern Tai varieties. 6) On the other hand, the A and B tones are more similar to each other on voice quality (normally modal) especially after the complex-contour spread from *B to the other tones, because the decaying of syllable final -h is more thorough. Glottalizations (including final glottal constriction and creakiness) found in the A and B tones as in the discussion above, are not inherited but are of one of the following situations: a feature spread from *C due to tonal mergers, a feature 117 spread from the weakening of final oral stops from the D tones due to tonal mergers, or an effect of a low pitch or a rapid falling contour. However, we should be cautious to attribute a modal voice to *B based on the elimination of glottalizations and on the similarity between the phonetic shapes of the A tones and the B tones in modern Tai varieties. The correspondence among *B in Tai and *C in Chinese/Vietnamese/Miao-Yao shows that *B in Tai should have had a specific phonation type/voice quality which is phonetically similar to voice quality of *C (a final aspiration like -s or -h) in Chinese/Vietnamese/Miao-Yao, instead of a modal voice. In addition, in traditional poetry well preserved in many Tai varieties (like Yang Zhuang and Tai Lue), the same syllable-rhyme with the tones developing from *B and *C can rhyme with one another but cannot rhyme with the same syllable-rhyme with tones developing from *A (Liao 2015). This also indicates that *B and *C tones in PT constituted an associated pair on voice qualities, and both of them should not have had a modal voice because they forbid being companions with *A which is treated to have a modal voice without dispute in previous studies. Tone *C has been frequently argued to have a syllable final -ʔ in many previous studies and in this thesis, thus its paired companion tone *B should also have had a counterpart syllable final -h, which has been favorably assumed in the speculations above. Note that a final -s which is speculated to be the phonetic shape of tone *C (qù or departing) in OC and to change to -h in MC in previous studies. Here we only consider the final -h as of the phonetic shape of tone *B in PT level, since it is also of lower laryngeal features to be paired with -ʔ in *C within the meters of traditional Tai poetry. Table 36 Basic tonal patterns in Debao Urban Yang Zhuang (CT) Phonation types Protovoiceless Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 453 B1-A: 33 C1-A: 24ʔ DL1-A: 33 DS1-A: 55 Continuants A1-C: 453 B1-C: 55 C1-C: 24ʔ DL1-C: 55 DS1-C: 55 Unaspirated stops A1-U: 453 B1-U: 55 C1-U: 24ʔ DL1-U: 55 DS1-U: 55 A1-UC: 453 B1-UC: 33 C1-UC: 24ʔ DL1-UC: 33 DS1-UC: 55 A1-G: 31 B1-G: 33 C1-G: 24ʔ DL1-G: 33 DS1-G: 55 A2: 31 B2: 33 C2: 21̰3ʔ DL2: 33 DS2: 33 A1/2: 453 B1/2: 33 C1/2: 24ʔ DL1/2: 33 DS1/2: 55 Unaspirated stop + *r clusters Glottal sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 118 Table 36 above shows the basic tonal patterns in Debao Urban Yang Zhuang, which provides a typical example for demonstrating the speculations above. The discussion will skip over the tones C and D since they straightforwardly reflect their original phonetic characteristics in PT. The tones A and B afford much food for thought although both of them have modal voice. It seems that they have interchanged their earlier tonal contours because of the following two reasons: (1) Both high level (55) and mid level (33) in the B tones reflect an original level tone, which should have been the original tonal contour of PT tone *A due to most previous studies. (2) The convex tone (453) and the mid-falling tone (31) in the A column should be originally from the B column, since we have previously discussed about the tonal developing processes of Tone *B after the decaying of -h: a falling tone (with long duration ending in a soft offset) > a dipping tone (to make up the decaying of -h) > other complex-contour tones like a convex tone (453). However, now we see that the tonal contours of A1/A2 and B1/B2 are interchanged. This indicates that a change in a tone can push one of the other tones to change to its original tonal shape. In the case of this dialect, a possible process is that at a specific historical period, a change of B1 from 453 to 55 causes A1 to switch from 55 to 453, as playing a matching action their counterparts also switch to each other (B2 from 31 to 33, and A2 from 33 to 31), for keeping the tonal distinctions. The speculations above together with evidence from the example analyzed indicate that complexcontours as a possible reflex of original syllable final -h is crucial to the hypothesis of the phonetic characteristics of PT Tone *B, although complex-contours may have also been spread to all the other tones in some modern Tai varieties. Only if the speculation that complex-contour firstly occurred in PT tone *B after the original falling tone arose is actually true, can the speculation that *B with a highfalling contour on a long vowel duration (cf. §4.1.3.2) be reasonably explained. The original high-falling contour can also explain Pittayawat Pittayaporn’s proposal that PT *B had creakiness like its counterpart qù in MC, and this creakiness is also found on the B tones in many modern Tai varieties. In addition, just like what Pittayawat Pittayaporn (2009) points that DL is the counterpart of the B tones, PT tone *D should have been like PT tone *B on pitch height, contour and voice quality after tone arose. My data (shown in Table 31) also support this hypothesis. The voice quality of Tone *D may have been glottalized since almost checked syllables found in modern Tai varieties always have glottal constriction in association with the final oral stops. The situation that DS is the 119 counterpart of the C tones in many modern Tai varieties suggests that it should be a secondary development due to their similar short vowel durations (cf. 4.3.3). To sum up, from the evidence of the tonal correspondence among Tai, Sinitic languages, Miao-Yao languages, and Vietnamese due to that all these non-Sinitic languages may have been influenced by OC and MC in history, and the Tai internal evidence demonstrated above, this thesis proposes that PT should be divided into two stages: an earlier non-tonal period and a later tonal period. In the earlier stage of PT, tonal categories *B and *C were on syllables with a final -h and a final -ʔ respectively, like their counterparts in Chinese, Vietnamese, and Miao-Yao languages. This indicates that the earlier phonetic characteristics of the *ABCD tonal categories were respectively identical to the *ACBD tonal categories in Chinese, Vietnamese, and Miao-Yao languages. Therefore, the argument of tonogenesis is rationally supported in Tai languages in this thesis. In the later stage of PT, under the influence of surrounding languages especially Chinese which may have undergone its own tonogenesis process, syllables with final -h and -ʔ in PT tended to decay. For keeping the semantic distinctions from the original plain syllables without obstruent syllable finals, the decaying of syllable final -h gave an original high-falling tone *B on relatively long vowel duration, and the weakening of *-ʔ gave an original low-rising tone *C on relatively short vowel duration, to contrast to each other and to the least marked level tone *A on smooth syllables. Syllables with final oral stops (-p, -t, -k) were preserved to comprise the fourth tonal category *D with a high-falling contour like its counterpart tone *B. The diachronic process of Tai tonogenesis is proposed as in Table 37. Table 37 Phonetic characteristics of PT tones in the two diachronic stages Earlier stage of PT plain -h -ʔ -p, -t, -k Tonal categories *A *B *C *D Pitch height mid high low high Contour level falling rising falling long short creaky glottal constriction non-tonal structures Later Stage of PT Vowel duration Voice quality modal 120 glottalization After tone arose, tonal splits and mergers have been playing a main role on the stage of tonal development. Just as tonogenesis initiated the four tonal categories at the PT level, register tonal splits account for the more complex situation of tonal categories in modern Tai varieties. As discussed in §1.2.1, the modern Tai tonal categories are illustrated by Li (1977) and this led to a scheme of ten tonal categories A1, A2, B1, B2, C1, C2, DL1, DL2, DS1, and DS2, which correspond to 1, 2, 5, 6, 3, 4, 9, 10, 7, and 8 tonal categories in the Sinitic circles. These schemes are actually based on the view that modern Tai tones are residual evidence of segments that have been lost. It indicates that the loss of the voicing contrast caused PT tones *ABCD to split into two series. Series 1 or high frequency was conditioned from the original voiceless initials, and series 2 or low frequency was usually conditioned by the original voiced initials. The *D tone on checked syllables split further depending on vowel length. This scheme is treated as the primary tonal split model as shown in Table 1 in §1.2.1. On the other hand, Gedney’s scheme (cf. Table 5) and Liao & Shen’s scheme (cf. Table 7) of Tai tonal categories indicate the more complex tonal development including secondary development, which will be discussed in the following sections. Before discussing the primary Tai tonal splits, it is necessary to briefly review the arguments of the different phonation types of initial consonants at the PT level, since they are the conditioning factors of the subsequent tonal splits (cf. §1.2.1, §1.2.2, §1.2.3 and §2.2.3). There are fewer disagreements about the following four phonation types at the PT level: voiceless stops, voiceless continuants, glottal sounds, and voiced sounds (including voiced stops and voiced continuants). The main outstanding disagreement for grouping the initial consonants at the PT level is whether there were aspirated stops at that stage. Li treats voiceless aspiration as a contrastive group and lists it to be with the four phonation groups above in his reconstruction of PT initials (Li 1977: 43). In spite of the different influence on tone between aspirated voiceless stops and voiceless continuants in some Tai dialects, Gedney puts them together under his definition “voiceless frictions” in his Tai tone box (cf. Table 4 in §1.2.1). Pittayawat Pittayaporn (2009) rejects Li’s reconstruction of aspiration at the PT level and argues aspiration should be of post-PT innovation mainly developing from clusters with medial *-r-. Liang and Zhang (1993, 1996) also suggest voiceless aspirated 121 stops developed on post-PT level from original initial clusters, like *pr-, *tl-, *tr-, *pw-, *xp-, *xpl-, and *xt- (cf. §1.2.3). All these different points of view on phonation types of PT initial consonants are summarized as in Table 38. Table 38 Different speculations of initial phonation types at the PT level Proposer Li (1977) Gedney (1989[1972]) Liang & Zhang (1993, 1996) Pittayawat Pittayaporn (2009) Phonation Types voiceless continuants voiceless aspirated stops voiceless unaspirated stops glottal sounds voiced sounds voiceless unaspirated stops glottal sounds voiced sounds × voiceless unaspirated stops glottal sounds voiced sounds × voiceless unaspirated stops glottal sounds voiced sounds voiceless frictions voiceless continuants voiceless continuants Although voiceless aspiration has long been viewed as contrast in PT (Li 1977), this thesis agrees with the viewpoint suggested by Liang and Zhang (1996) and Pittayawat Pittayaporn (2009) that voiceless aspirated stops are a post-PT innovation, since only this can explain the irregularities of aspiration corresponding within modern Tai varieties with fewest problems. Liang and Zhang’s argument is based on the observation that the correspondence between aspirated sounds in modern Kam-Tai languages are very uneven and irregular. They point out that the vast majority of modern NT languages, some Southern Thai varieties, and Northern Kam lack aspirated initials, whereas the other Kam-Tai languages all have aspirated initials. However, the situations of these aspirated initials are very complicated. In NT, aspirated initials only found in T3 (C1) in the third vernacular of Bouyei, but it is not contrastive. Within more than one thousand cognates in their data, very few cognates all consistently have aspirated initials within those Tai varieties which have aspirated initial sounds in their sound system. For example, the initial of the word ‘arm’ is aspirated in CT and SWT varieties which have well-developed aspirated initials, but in YN varieties (of those whose classification as CT or NT is open to debate) which also have robust aspirated initial sounds in their own sound system, this word is ke:n1 (A1) with an unaspirated initial sound, agreeing with NT. In Kam-Sui languages which also have 122 robust aspirated initials, the initial of this word is also unaspirated. On the other hand, some cognates have aspirated initials in Kam-Sui languages but have unaspirated initials in Tai languages which have robust aspirated initials. Even within Tai languages which have robust aspirated initials, the aspirated sounds in some varieties do not always present as aspirated in the other varieties. Therefore, they assert that aspirated initials are a secondary development after the split of the Kam-Sui and Tai branches, and even after the splits of different dialects of Tai varieties. They furthermore list the original segments of aspirated initials in modern Kam-Tai languages, like that PT voiced stops become voiceless aspirated stops in Thai, Lao, and Daxin Zhuang, voiced aspirations (in their own reconstruction) become voiceless aspirated in CT and SWT, and voiceless unaspirated stops plus -w-, -l-, or -r- clusters become aspirated initials in some of the modern Tai varieties. However, they also mention that these segments do not always cause aspiration in all the Kam-Tai languages which have robust aspirated initials (Liang & Zhang 1996: 75-77). Pittayawat Pittayaporn also points out that the view that aspiration was an original characteristic and NT lost it was problematic for the following reasons. First, the loss of aspiration is not a common sound change, and it can hardly explain the situation of NT since it has been heavily influenced by Chinese where aspiration is robust. Second, many of Li’s simple aspirated stops actually develop from voicing alternation which must be considered to be later loans or reflexes of complex onsets. Third, etyma with invariable aspirated onset in Li’s list are mostly questionable or irregular. Furthermore, the comparison of aspirations between Tai and Kam-Sui show disagreements, and the crucial evidence is that aspirated initials has been claimed not to be an inherited trait in Kam-Sui. Last, he mentions Liang & Zhang’s viewpoint that modern aspirations have developed from PT clusters (Pittayawat Pittayaporn 2009: 88-89). Data from my fieldwork also show some of the same evidence put forward by Liang and Zhang as well as Pittayawat Pittayaporn. For example, unaspirated stops with Tone C1 in the vast majority of NT varieties normally present aspirated in Zhenning Bouyei, like kʰa:31 ‘to kill’, lok44 kʰa:31 ‘seedling’, tʰã31 ‘to establish’ and kʰou31 ‘nine’. However, aspirated stops are not found in the other tonal categories in this dialect. It is crucial that C1 and C2 tones have been merged to be one single tone in this dialect, but the aspirated initial stops with original C1 tone contrast to unaspirated initial stops with original C2 tone. This indicates that aspirated stops in this NT variety are of secondary development in the C column for keeping the semantic 123 distinctions after the merger of the tones C1 and C2. Besides, some aspirated initials in those varieties whose classification as CT or NT, Sanxiang, Shuangding, and Nung An in my own fieldwork, present as unaspirated initials in the vast majority of CT/SWT varieties, like kʰa:ŋ35 (C1-U) ‘fishbone’ in Shuangding YN contrasting to ka:ŋ24 (C1-U) in Debao Yang Zhuang and ka:ŋ41 (C1-U) in Thai. This also supports that some aspirations are of independent development. Furthermore, in some specific varieties, some of the words still preserve unaspirated stops plus -r clusters which have become simple aspirated initials in most CT/SWT varieties, such as pra:33 (A1-A) ‘stone mountain/hill’ in Shuangding YN contrasting to pʰja:453 ‘stone mountain/hill’ in Debao Yang Zhuang and pʰa:24 ‘cliff’ in Thai. This also supports the viewpoint that some of the origins of aspiration are voiceless unaspirated stops plus -r clusters. Therefore, this thesis admits the four phonation types, voiceless stops, voiceless continuants, glottal sounds, and voiced sounds (including voiced stops and voiced continuants) were at the PT level, but considers voiceless aspirated initials as of post-PT innovation. Just as in the previous studies reviewed, after tone arose in the vast majority of PT voicing contrast was speculated to be lost. In this speculation, in the wave of the two series of mergers of the PT initial consonants (mergers from voiceless into voiced continuants due to the voicing process, as well as from voiced to voiceless stops due to the devoicing process), PT tones felt into two series of primary registers – original high tones and original low tones conditioned by original voiceless stops/continuants and original voiced stops/continuants respectively. As has been repeatedly mentioned, this process is treated as the primary register split, and formed a system of eight tonal categories, namely tones A1, A2, B1, B2, C1, C2, D1, and D231. However, there are some disagreements on the direct cause of the primary tonal split in some of the previous studies. The disagreement mainly lies in that the cause was the devoicing of proto-voiced stops or the voicing of proto-voiceless continuants. As Theraphan L-Thongkum (1997: 208) points out, previous studies normally select the voiceless-voiced distinction in initial plosives/stops to illustrate 31 Tonal splits in dead syllables are also conditioned by vocalic length to result tonal categories DL1, DL2, DS1 and DS2, which are treated as secondary development in this thesis. Therefore, they will be discussed in the following §4.3.3. 124 the cause of tonal split, like in her citations Haudricourt (1961, 1972), Matisoff (1973), Brown (1975), Mazaudon (1970), Li (1977), Henderson (1981, 1982), and Gedney (1985). However, in some specific modern Tai varieties, initial consonants developing from the PT voiceless/voiced continuants present as the only cause of their primary tonal split. In Cao Bang Tho (of CT) which still preserves the voiced stops, “the loss of aspiration in the sonorants” seems to be the only cause of tonal split (Haudricourt 1972: 65). Dai Zhuang (Dai Tho) in which the voicing contrast between stops is also still preserved, the merger of voiceless sonorants with voiced sonorants from PT is suggested to be the cause of the primary tone split (Theraphan L-Thongkum 1997: 215). These two CT dialects are very crucial to a hypothesis that the cause of the primary tonal split in Tai languages is actually the voicing of protovoiceless continuants, because at least they very clearly indicate that while the devoicing process of the proto-voiced stops has not yet started, the voicing process of proto-voiceless continuants has long been completed and tonal register split has been established since the early days. For further interpreting this phonetic plausibility, Theraphan L-Thongkum (1997: 209) points out that the sound change from *m̥ - (partially voiced) to m- (*m̥ -[m̥ m] > m-) takes shorter than that from *b(fully voiced) to p- (*b- > *b̥ - > p-) or to pʰ- (*b- > *b̤ - > *bh- > pʰ-) by citing the point that voiceless sonorants are actually “partially voiced” when preceding vowels (Ladefoged 1971: 11). Haudricourt (1972: 65) and Theraphan L-Thongkum (1997: 212) cautiously indicate the merger of proto-voiceless and proto-voiced sonorants to be the independent cause of tonal split in Cao Bang Tho and Dai Zhuang respectively. However, because voiceless continuants are actually not found in modern Tai varieties at all but voiced plosives are completely or partially preserved in some specific Tai varieties, this thesis suggests that the voicing of proto-voiceless continuants (including sonorants and the sibilant *s-) was the only cause of the primary tonal split cover all Tai varieties beyond the two varieties above. The following evidence give support to this suggestion. First, documentary evidence indicates that in specific Tai varieties that have completely merged proto-voiced-voiceless plosives, the merging was later than the tonal split period. Theraphan L-Thongkum cites Brown (1975) and Gedney (1991) to point out that the low-register plosives (from proto-voiced plosives) in Thai orthography are written as the original voiced sounds, even though they present voiceless aspirated in modern Thai. This implies that the devoicing of proto-voiced 125 plosives are quite recent and may be far later than the tonal splits period (Theraphan L-Thongkum 1997: 207-208). Second, the merger of proto-voiced-voiceless plosives present as in progress in some specific Tai varieties which have straightforward tonal split between earlier voiceless and voiced registers. We have seen that voiced stops (from PT voiced or low register) are still completely preserved in some modern Tai varieties, like Dai Zhuang and Cao Bang Tay discussed above. However, there are some Tai varieties that preserve incomplete voiced stops from PT voiced or low register. In the data from my own fieldwork, some proto-voiced initial obstruents are preserved as voiced in Yizhou Suogan Zhuang (Language 36 in my data), like (kaj33) ɣa:ŋ11 (B2 < B1/2) ‘young chicken’ and bəɰ24 (C2) ‘bride’, but some are merged into voiceless sounds, like tɯŋ24 (C2) ‘crutch’. This implies that the merger of proto-voiced-voiceless plosives is unfinished. In Yongning Baiji (L25), earlier voiced stops*b- and *d- have been merged into b̤- and d̤- which developed from earlier *ʔb- and *ʔd- to complementarily distribute on the higher register and the lower register respectively. For example, d̤wa:331 (A2) ‘to spread on’, d̤wa:31 (B2) ‘river’, b̤ wa:k31 (DL2) ‘insane’, and b̤wak33 (DS2) ‘very tired’. However, unlike that b̤ - and d̤- on the higher register sometimes present as p̤- and t ̤-, b̤ - and d̤- on the lower register always present as breathy voiced sounds. In brief, breathy voice in this dialect is a signature of the partially voiced in the process of devoicing earlier voiced and glottal plosives, in contrast voicing of proto-voiceless continuants has much earlier completed to give the straightforward register tonal split this dialect. These cases all suggest that devoicing of voiced stops causing the merger of voiced-voiceless stops must be much later than the voicing of voiceless continuants in Tai languages. Furthermore, as a cross-linguistic phenomenon found in TK, Mon-Khmer, Miao-Yao, and Tibeto-Burman languages, voiceless sonorants play an important role in the process of pitch raising to cause tonal split (Theraphan L-Thongkum 1997: 209-212). This confirms the plausibility that voicing of proto-voiceless continuants cause primary tonal split in Tai languages. The last piece of evidence is just like what has been reviewed in §2.2.3 that the PT voiceless continuants tend to mostly keep their tonal behaviors as in their primary tonal splits since they have not been found to have the case that other types of initials, namely glottalization, aspiration, unaspiration all have influence on secondary tonal development (Zhang 1980, Liang & Zhang 1996, Zhang et al. 1999). 126 This phenomenon will be furthermore discussed in the process of secondary tonal splits in the following §4.3.1. In brief, this thesis suggests that the devoicing of proto-voiced stops or the merger of voiced-voiceless stops was not the first trigger of register tonal split. Instead, the voicing of voiceless continuants which led the merger of voiceless-voiceless continuants was the actual trigger of register tonal split in all Tai languages. This clarification is considered to be very important in this thesis, since it reveals one more diachronic tonal developing process in Tai tonal development soon after the process of tonogenesis. On the basis of the two clarifications involving the primary tonal split in §4.2.1 and §4.2.2 above, we get a new structure of initial phonation types in the stage of the primary tonal split in the post-PT level: voiceless unaspirated stops, glottal sounds, voiced continuants and voiced stops. The primary register split caused by the voicing of earlier voiceless continuants formed an eight tonal category system, namely tones A1, B1, C1, and D1 on high register as well as A2, B2, C2, and D2 on low register, as shown in Table-1 in §1.2.1. In this stage, it is remarkable that phonation types of initial consonants coordinated with register split to provide a regular collocation in the process of tonal development. The merger of voicelessvoiced continuants must lead a result that the remaining voiced continuants collocated with high or low register tones respectively, at the same time the other phonation types of initials must also have to choose to be preceding syllables with either the high or the low register tones. Thus, the choice of high-low register tones for non-voiced continuants in this stage is essential to the pattern of the primary tonal split. In the previous studies on Kam-Tai tones like Weera Ostapirat (1994), the pattern that voiceless unaspirated stops and glottal sounds went with original voiceless continuants to condition high register tones (A1, B1, C1, DL1, and DS1), and voiced plosive went with original voiced continuant to condition low tones (A2, B2, C2, DL2, and DS2) is called straightforward tonal development32. The choice that protoglottal sounds went with the original voiceless stops/continuants to condition their 32 Straightforward tonal developing pattern in the previous studies normally also includes that voiceless aspirated stops go with original voiceless continuants, voiceless unaspirated stops, and glottal sounds to have original high pitches. However, in §4.2.1 we have claimed that voiceless aspirated stops were of secondary development in post-PT level, therefore they are omitted here and will be discussed in the following §4.3.2.1 and §4.3.2.2. 127 syllables to have a tone going together with the original high register may also be explained as that the glottal initial stop -ʔ or the pre-glottal segment are phonetically voiceless (Li 1977: 44). Although the straightforward tonal register split pattern it is not the most common pattern in SWT varieties in Thailand, it is most common found in the other Tai varieties out of Thailand, as some scholars in the Sinitic circles point out that it is found in more than half of the Zhuang dialects (Zhang et al. 1999: 245), which are linguistically of CT and NT respectively. Li Fang-Kuei also treats the straightforward pattern as his most common Type I of initial-type and register-tone combinations since it is found in all his three main sub-branches of Tai (Li 1977: 44-45). Within the forty-two Tai varieties investigated for this thesis, nearly half of them (the following twenty varieties) have the straightforward tonal register pattern: Kho Lam Tai Yai, Rong Maet (Tai Lue), and Muong Quan Son Tai in SWT, Jingxi Urban (Yang Zhuang), Jingxi Anning (Yang Zhuang), Nalong (Myang Zhuang), Xiaoguangnan (Nong Zhuang), Tuoxin (Pyang Zhuang), and Daxin Naling (Zuojiang Zhuang) in CT, Baiji (YN) and Shuangding (YN) in those whose classification as NT o CT is open to debate, as well as Wuming Shuangqiao (Yongbei Zhuang), Debao Longsang (undescribed), Du’an Jiaren (Central Hongshuihe Zhuang), Shanglin Yunling (Hongshuihe Zhuang), Xiaosanjiang (Lianshan Zhuang), Yizhou Suogan (Guibei Zhuang), Donglan (Guibei Zhuang), Baipeng (Liujiang Zhuang), and Zhenning (Bouyei) in NT, as shown in Appendix C. The straightforward pattern seems to reflect the primary tonal split triggered by the voicing of proto-voiceless continuants after tonogenesis, since this pattern is reflected in the majority of modern Tai varieties which do not reflect further secondary tonal developing patterns. However, as the following two reasons point out, the straightforward pattern cannot be treated as identical with the primary register tonal split pattern in that early stage. First, as Li points out that not all of this type of straightforward pattern has six tones33, some of the two tonal categories within the six ones have been merged into each other. There are five sub-types in this type, namely Ia (B2 = C1) represented by Shan and Red Tai, Ib (A1 = B2) represented by Tay, Ic (A2 = B2) represented by Nung, Id (C1 = C2) represented by Chenning Pu-i and Shuicheng Pu-i, as well as Ie (A2 = B1 and B2 = C2) represented by Phu Thai, make the six tonal categories 33 In Li’s definition, tonal categories (DL1, DL2, DS1, DS2) on checked syllables are treated as allotones of tones on smooth syllables. Thus, there are at most six register splitting tones (A1, A2, B1, B2, C1 and C2) on smooth syllables in the straightforward type. 128 result a system with five (Ia, Ib, Ic, and Id) or four (Ie) distinct tones (Li 1977: 45). In my data, except those having six distinct tones on smooth syllables, the following Tai varieties also have a five-tone system with the mergers from some of the two tonal categories. Tai Yai and Quan Son Tai belong to Li’s type Ia (B2 = C1), and Baiji YN and Zhenning Bouyei belong to type Id (C1 = C2) (cf. Appendix C). Therefore, at least the sub-types of this straightforward type do not reflect the speculated pattern in the stage of the primary tonal split. Second, tonal splits and mergers still lack of agreements on the diachronic process in previous studies. In the Sinitic circles, scholars like Zhang (1980), Liang and Zhang (1996: 816-817), and Zhang and colleagues (1999: 243-246) consider the primary register tonal split was straightforward in all Tai varieties, and those Tai varieties having non-straightforward tonal development patterns on some of the tonal categories actually reflect secondary tonal development influenced by some specific initial phonation types or laryngeal features, like aspiration, glottalization and preglottalization, and voiceless unaspirated stops. Li also doubts that tonal development influenced by some of the laryngeal initial features like glottalization and aspiration can be considered a PT phenomenon, and the tone 33 (author note: B1-A) conditioned by aspiration like in tʰa:ʔ33 ‘to expose to the sun’ and tʰe:ʔ33 ‘to break, crack’ in his T‘ien-pao (Debao Urban Yang Zhuang) is speculated to be secondary development when compared with the original tone 42 (author note: B1-U) in such words pa:ʔ42 ‘mouth’ and pe:t42 ‘eight’ (Li 1977: 26-27). However, some other scholars have different viewpoints. In interpreting his designation of Tai tone box (cf. Table 4 in §1.2.1), Gedney uses ‘initials at time of tonal splits’ to define the four initial groups which condition possible tonal splits (Gedney 1989[1972]: 202). This hints that different initial phonation types have chosen to collocate with one of the two or three registers when the primary tonal split occurred. Thus in this viewpoint even the primary tonal split cannot be straightforward in many Tai varieties which do not reflect the straightforward pattern, like the other half of the forty-two Tai varieties investigated in this thesis. Theraphan L-Thongkum (1997: 212) definitely points out two possible types of primary tonal split in her Tho (author note: Dai Zhuang and Debao Yang Zhuang) of CT. The first type basically equals to the straightforward pattern because the split is between voiced sounds and all other sounds, although the voiced plosives (from *b-, *d-, *g- and etc.) are still preserved in Dai Zhuang. The second type has the split between voiceless plosives (aspirated and unaspirated) and those developing from glottal sounds and voiced sounds in PT. This division of primary tonal split indicates that the proposer treats the primary tonal split not to be a unified phenomenon in history, but to be separated patterns in 129 individual Tai varieties. However, it seems that she does not account for the diachronic process of tonal splits since her ‘primary tonal split’ has covered all the modern tonal patterns in Dai Zhuang and Debao Yang Zhuang. In spite of the different viewpoints on the diachronic process of register tonal splits, this thesis somehow agrees with the primary-secondary division of tonal splits suggested by Zhang (1980), Liang and Zhang (1996), Zhang and colleagues (1999) and Li (1977). It considers the primary tonal split in Tai languages to be a common phenomenon in the predecessors of modern Tai varieties in the post-PT period. It further proposes that the voicing of PT voiceless continuants caused high-low two series of primary register split to occur on the PT four tonal categories. The primary register split is speculated to have been a straightforward pattern that grouped voiceless unaspirated stops and glottal/pre-glottal sounds together to condition tones in the high register from the voicing of proto-voiced continuants, and grouped voiced stops to condition tones in the low register depressed by proto-voiced continuants. Thus, an eight-tone system with Tones A1, B1, C1, and D1 on the high register and Tones A2, B2, C2 and D2 on the low register tones has been established in this stage, as shown in Table 39. In this table, the tone pitches and contours are speculated by combining the speculation of the phonetic characteristics of PT tones (cf. Table 37). Note that the primary split pattern is speculated to be an eight tonal category pattern with fixed pitch heights and contours, and is different from the straightforward split pattern on some of the tonal mergers in some specific modern Tai varieties. In other words, the straightforward split pattern covers the primary split pattern, but the latter does not conversely cover the former. Table 39 The speculated primary register tonal split on post-PT level Proto-Tone Register *A *B *C D* mid level high-falling low-rising high-falling A1 B1 C1 D1 high level high-falling high-rising high-falling A2 B2 C2 D2 low level low-falling low-rising low-falling High *m̥ -, *l ̥-, *s- … *p-, *t-, *k- … *ʔ-, *ʔb-, *ʔd- … Low *m-, *l-, *z- … *b-, *d-, *g- … We will prove this hypothesis by the steps in the following sections about the secondary tonal splits, which form the counterevidence to indicate that all the nonstraightforward tonal developing patterns in Tai languages are secondary. 130 Different from the unified pattern of the primary tonal split, secondary tonal splits in Tai differ from dialect to dialect, or even from variety to variety within the same language. The most typical example is the totally different secondary tonal split patterns among different varieties in Debao Yang Zhuang and Jingxi Yang Zhuang, which are under the same distinct language with “zyg” as the ISO 639-3 code (cf. Languages 8-13 in the Appendix C). This phenomenon shows that secondary tonal splits occurred after the formations of different hierarchies of languages and dialects in Tai. On account of interpreting the whole scheme of Tai tonal development, it is necessary to outline the diachronic hierarchies of secondary tonal splits in different groups and varieties of Tai languages. Within the following sub-sections of this section, §4.3.1 aims to prove that the non-straightforward patterns of tonal splits in Tai are all secondary development, §4.3.2 is separated into five sub-sections to indicate the diachronic hierarchies of secondary tonal splits on smooth syllables, and the last §4.3.3 discusses the secondary tonal split patterns in checked syllables in Tai languages. As pointed out in §4.2.2, initials developing from PT voiceless continuants tend to mostly keep their tonal behaviors as in their primary tonal split since they are of the only type that has only conditioned straightforward pattern of register split. That is, non-straightforward tonal splits have not been found to be conditioned by initials developing from proto-voiceless continuants. Moreover, it has been suggested that initials developing from proto-voiceless continuants was the only cause of the primary tonal split in Tai. Based on the reasoning above, it is apparent to receive a motivation that non-straightforward tonal splits conditioned by all the other types of initials, namely glottalized sounds, aspirated stops, and unaspirated stops are all secondary in the history of Tai languages. Previous studies, like Zhang (1980), Liang and Zhang (1996), and Zhang and colleagues (1999) have already provided sufficient discussions to achieve this suggestion, which is agreed with in this thesis. The reasons are summarized as follows. First, the straightforward pattern of register tonal split is cross-dialect within the Kam-Tai branch above the Tai group and the Kam-Sui (KS) group, while the non- 131 straightforward patterns are in an irregular distribution. The Zhuang dialects (of NT and CT) from the most areas, Tai Lue (of SWT), Shan (of SWT), as well as Sui (of KS) and Mulam (of KS) all reflect this pattern, and the tonal correspondence is basically regularly agreed with each other among these Kam-Tai varieties (Liang & Zhang 1996: 817). However, some of the Zhuang dialects (of NT and CT) from the western and northwestern parts of Guangxi, and from some counties around Nanning City, the majority of the Bouyei dialects (of NT), the Dai dialects (of WT) from the prefectures of Dehong and Honghe from Yunnan, Thai and Lao, as well as Kam, Maonan, Ai-Cham, and Mak (which are all of KS) all have non-straightforward patterns of tonal splits on the high register. The tonal correspondence among these non-straightforward tonal splits is irregular. Besides, within these languages some have splits on the high register on all the proto-tones, but some have splits only on one to three proto-tones (Liang & Zhang 1996: 818). Only if that the straightforward pattern reflects the primary split, and the non-straightforward patterns reflect secondary splits, can these phenomena be reasonably explained. Second, the causes of non-straightforward splits on the high register in Tai are found to be restricted to glottalized sounds, aspirated stops, and unaspirated stops. These non-straightforward patterns are foresightedly treated as secondary tonal splits by Zhang and colleagues (1999: 243-245). On the basis of previous studies (Zhang 1980, Liang & Zhang 1996), they divided these secondary tonal splits into four types as follows. (1) Type 1: only glottal initials condition secondary tonal splits. These glottal initials are ʔ-, ˀb-, ˀd-, ˀj- and those developing from proto-glottal initials such as j- (< *ʔj-), m- (< *ʔb-), and n- (< *ʔd-). This type occurs in the three NZ languages Youjiang, Guibian, and Qiubei, as well as in some varieties in some other Zhuang languages such as Donglan, Tian’e, Nandan of NZ (of NT) and Tiandeng of SZ (of CT) (Zhang et al. 1999: 243). As shown in the following glossary comparisons in Table 40, the first four represent those having Type 1 of secondary tonal splits, and Wuming represents those having the straightforward tonal split pattern. 132 Table 40 Type 1 of secondary tonal splits in the Zhuang dialects (adapted from Zhang et al. 1999: 243-244)34 Examples Modern Tai varieties Situations initials of tone causing changes secondary (secondary ‘good’ ‘leaf’ ‘cough’ ‘get’ ‘village’ ‘scold’ ‘at’ ‘out’ A1 A1 A1 C1 C1 B1 B1 DL1 ˀdi:1 ˀbaj1 ʔaj1 ˀdaj4 ˀba:n4 ˀda:5 ʔu5 ʔo:k9 ˀdi:1 ˀbaj1 ʔaj1 ˀdaj4 ˀba:n4 ˀda:6 ˀji:6 (6<10<9) ˀdej1 ˀbaɰ1 ʔaj1 ˀdaj3 ˀba:n3 ˀda:6 ˀjow6 ʔo:k10 naj2 maɰ2 ʔaj2 naj3 ma:n3 na:5 jow5 ʔo:k9 ˀdaj1 ˀbaɰ1 ʔaj1 ˀdaj3 ˀba:n3 ˀda:5 ˀjaw5 ʔo:k9 tonal splits tonal splits) Tian-dong ʔ-, ˀb-, ˀd-, (NT) Tian’e (NT) Donglan (NT) ˀj- ʔ-, ˀb-, ˀd-, ˀj- ʔ-, ˀb-, ˀd-, ˀj- Tiandeng ʔ-, m- (< (CT) *ʔb-), n- (< 3→4 3→4 5→6 9 → 10 5→6 9 → 10 ʔo:6 1→2 *ʔd-), j- (< *ʔj-) Wuming (NT) ʔ-, ˀb-, ˀd-, ˀj- Straightforward Take the word ʔo:6 ‘out’ in Tian’e (NT) in Table 40 as an example to interpret Type 1 demonstrated. This word contains complicated tonal changes – secondary tonal split associating with tonal merger due to the loss of final -k of the original checked syllable. The original tonal category T9 (DL1) was changed to T10 (DL2) due to the secondary tonal split conditioned by the initial ʔ-, and T10 (DL2) merged into T6 (B2) due to the loss of the syllable final -k (Zhang et al. 1999: 240). Within the twenty-two languages that have non-straightforward tonal development patterns from my fieldwork, Chiang Rai (L4) of SWT, Hurun Yang Zhuang (L11) and Xiangdu Zuojiang Zhuang (L17) of CT, as well as Dongling Youjiang Zhuang (L32), Lizhou Guibian Zhuang (L33), Gehan Qiubei Zhuang (L39), and Huishui Bouyei (L41) of NT are of this type. Lingding Zuojiang Zhuang (L19) is partly of this type but partly of Type 3 (see the detail in Type 3 below). (2) Type 2: Only aspiration conditions secondary tonal splits. These initials are all aspirated stops and this type is only found in Wenma Dialect (Dai Zhuang) of SZ 34 The numeral style of tones in Table 40 (similarly hereinafter) follows the style in the Sinitic Circles. The correspondence among it and the other styles of tonal numerals has been explained in §1.2.1. The shaded boxes refer to those cases being tone changes (secondary tonal splits) occurring. 133 (CT). The situations of secondary tonal splits present as in that T1 → T2, and T7 → T2 (Zhang et al. 1999: 244). In the following glossary comparisons, Wenma represents those having Type 2 of secondary tonal splits (in shaded), and Jingxi represents those having the straightforward tonal split pattern as shown in Table 41. Table 41 Type 2 of secondary tonal splits in the Zhuang dialects (adapted from Zhang et al. 1999: 244) Gloss Wenma (Dai Zhuang) Jingxi (Yang Zhuang) ‘dam’ ‘eye’ ‘leg’ ‘warehouse’ ‘vegetable’ ‘six’ pʰ :2 tʰo:2 kʰo:2 tsʰɛ:2 pʰe:2 tsʰa:2 pʰa:j1 tʰa:1 kʰa:1 ɕa:ŋ1 pʰjak7 kʰjɔ:k7 Within the data from my fieldwork, only Dazhai Dai Zhuang (L21) of CT is directly of this type. Huashan Min Zhuang (as shown in the tone box in L16 in Appendix C) represent a special case of Type 2. The pitches in tones *A and *DS are depressed by aspirated initials (both the rows 1-A, 1-UC, and 1/2-A) to form a secondary tonal split. In Tone *A this split causes a three-way split pattern (1’-1-2), but in *DS, in spite of this split all other rows (1-C, 1-U, 1-G, and 2) are merged together or lack of split to keep the two-way split pattern. The situations of Bac Va Nung (as shown in the tone box in L24 in Appendix C) is more complicated. In Bac Va, the secondary splits in Tones *A and *DL have Type 2, in which the aspirated sounds (both the rows 1-A and 1-UC) conditioned a secondary tonal split, but the secondary splits in Tone *B have Type 3 (which is discussed below) with aspirated stops and glottal sounds being the double conditioning causes. This dialect can be treated as having a mixed Type 2/3. (3) Type 3: Both aspirated stops and glottal sounds condition secondary tonal splits. This type is found in the Zhuang varieties like Yongnan, Fubei, Long’an (all of those whose classification as CT or NT is open to debate) and Debao Mai’ai (of Yang Zhuang of CT) (Zhang et al. 1999: 244). As shown in the glossary comparisons in Table 42, the first four represent those having Type 3 of secondary tonal splits, and Longzhou represents those having the straightforward tonal split pattern. 134 Table 42 Type 3 of secondary tonal splits in the Zhuang dialects (adapted from Zhang et al. 1999: 244-245) Examples Modern initials Situations of Tai causing varieties secondary tonal splits Yongnan Glottal Sounds: ʔ-, m- (<*ʔb ), n- (< *ʔd), j- (< *ʔj-); Fubei tone changes (secondary ‘stone ‘sugar- mountain’ cane’ ‘spring’ ‘at’ ‘carry by pole’ ‘one’ tonal splits) A1 A1 A1 C1 C1 B1 B1 DL1 naj1' ʔaw1' pʰla:1' ʔo:j5 mo:6 jow6 tʰa:p10 ʔət7 naj1' ʔaw1’ pʰa:1’ ʔo:j3 mo:5 jəw5 tʰa:p9 ʔit7' naj1 ʔaw1 pʰla:1 ʔɔ:j3 mo:5' jow5' tʰa:p9' ʔit8 ˀdɔ:j2 ʔaw2 pʰja:2 ʔo:j3 ˀbo:6 ˀjow6 tʰa:p10 ʔat7 ˀdaɰ1 ʔaw1 pʰja:1 ʔo:j3 ˀbo:5 1 →1' 3 →5 5 →6 Aspirated stops: 9 →10 pʰ-, tʰ-, kʰ-, ... 1 →1' 7 → 7' Long’an ‘inside’ ‘take’ 5 → 5' 7→8 9 → 9' (Ma’ai) Glotal Sounds: ʔ-, ˀb-, ˀd-, ˀj-...; Aspirated stops: pʰ-, tʰ-, kʰ-, ... Longzhou Glotal Sounds: ʔ-, ˀb-, ˀd-, j- (< *ʔj-) ...; Aspirated stops: pʰ-, tʰ-, kʰ-, ... 1→2 5→6 9 → 10 straightforward Debao ju:5 ha:p9 ʔit7 In Table 42, tonal category markers with an inverted comma like 1', 5', 7', and 9' refer to the so-called “derived tones” (派生調) in the Sinitic circles working on KamTai languages. Such kind of derived tones actually refer to one of the splits of the socall “three-way split”. The details will be discussed in the following §4.3.2.5. In this thesis, we also adapt this convention to mark “derived tones” when referencing to the previous studies in the Sinitic circles. The correspondence among the style in the Sinitic circles and the other styles of tonal numerals has been explained in §1.2.1. The shaded boxes refer to those cases being secondary tonal splits occurring. Within the data from my fieldwork, Debao Dalong Yang Zhuang (L9), and Debao Luliu Yang Zhuang (L10) of CT, as well as Long’an YN (L27), Daqiu YN (L28), and Xialeng YN (L29) of those whose classification as CT or NT is open to debate are of 135 this type. The situation of Lingding Zuojiang Zhuang (L19) and Debao Urban Yang Zhuang (L8) are complicated. In Lingding, the secondary splits in Tones *B and *DL are only conditioned by glottal sounds to be of Type 1, but in Tone *A the splits are conditioned by aspirated and glottalized initials to be of Type 3. However, the split conditioned by aspirated is distinct from the split conditioned by glottalized initials. The former (A1-1, A1-UC, A1/2-A) is depressed to be a derived tone 353 lower than the original A1 tone 554, but the latter (A1-G) is depressed to merge into A2 to be an extreme low tone 11. This results a three-way split in Tone *A. In Debao Urban, the secondary splits have Type 1 in Tone *A, but have Type 3 in Tones *B and *DL. These two dialects can be treated as having a mixed Type 1/3. (4) Type 4: Both glottal sounds and unaspirated stops condition secondary tonal splits. This type is only found in Houyi of Daxin Zuojiang Zhuang in SZ. The situation of secondary tonal splits present as in that T1 → T1’, T5 → T6, and T7 long (or 9) → T8 long (or 10) as shown in Table 43. The conditions of this type of tonal splits is the same as Thai and Lao (Zhang et al. 1999: 245). In the following glossary comparisons, Daxin Houyi (of Zuojiang Zhuang) represents this Type 2 of secondary tonal splits (in shaded), and Longzhou Urban (of Zuojiang Zhuang) represents those having the straightforward tonal split pattern. Table 43 Type 4 of secondary tonal splits in the Zhuang dialects (adapted from Zhang et al. 1999: 245) Gloss Daxin Houyi (Zuojiang Zhuang) Longzhou Urban (Zuojiang Zhuang) ‘good’ ‘year’ ‘eat’ ‘salt’ ‘shoulder’ ‘pass’ ‘at’ ‘out’ ˀdaj1' pi:1' kin1' tsɯə1' ˀba:6 kwa:6 ˀju:6 ʔo:k10 ˀdaj1 pi:1 kin1 kɯ:1 ˀba:5 kwa:5 ju:5 ʔo:k9 Within the data from my fieldwork, Bangkok Thai (L1), Songkhla Southern Thai (L2), and Khon Kaen Isan (L3) of SWT, as well as Leiping Zuojiang Zhuang (L22) and Baoxu Zuojiang Zhuang (L23) of CT are of this type. Based upon to the presentations of the four types of non-straightforward tonal splits above, we see that within the three phonation types that condition the claiming secondary tonal splits, glottalized sounds and aspirated stops can condition nonstraightforward tonal splits alone (Type 1 and Type 2) or together with one another (Type 3). However, unaspirated stops have not been found to condition nonstraightforward tonal splits by itself or together with aspirated stops, but only 136 condition the splits together with glottalized sounds (Type 4) (cf. Zhang 1980: 38, Liang & Zhang 1996: 826), perhaps due to that unaspirated stops share the same features [+voiceless, -contituant, -aspirated] with the segment ʔ- of glottalized sounds. This adequately indicates that at least unaspirated stops must secondarily condition the tonal split going after the split triggered by glottalized sounds. Since voiceless aspirated stops is suggested to be secondary development on post-PT level (cf. §4.2.1), the tonal splits on high register conditioned alone by aspirated stops have to be secondary. Consequently, tonal splits conditioned together by aspirated stops and glottalized sounds also have to be secondary. Third, non-straightforward tonal splits in some specific Tai varieties give a suggestion of the diachronic orders. The tonal splits in Southern Thai (cf. the tonal pattern of L2 in Appendix C) provide a good example. Only if the following diachronic process of secondary tonal splits in Southern Thai proposed by Zhang (1980: 38) is true, can the tonal split patterns in this language be reasonable explained. Zhang describes that after the primary tonal split, in Southern Thai Tone 5 (B1) with all proto-voiceless initials (voiceless stops, voiceless continuants, and glottal sounds) merged into T1 (A1) before the non-straightforward tonal splits occurred. After the secondary tonal splits conditioned by glottal sounds and voiceless unaspirated stops (Type 4), T1 (A1 and those merged from B1) derives T1' (A1-U/G as well as B1-A/G merging into A1-U/G), T3 (C1) derives T3' (C1-U/G), and T7 (DS1-U/G) is merged into Tone 8 (DS2)35. Therefore, voiceless frictions (including aspirated stops and continuants) only precede syllables with original T1 (including original T5 or B1 which merged into T1 or A1), T3, and T7, while voiceless unaspirated stops and glottal sounds only precede syllables with T1', T3', and T8. If the tonal splits conditioned by glottal sounds and voiceless unaspirated stops (Type 4) occurred on the time of the primary tonal split before the merger between original T1 (A1) and T5 (B1), at least the neat mergers between T1 (A1A/C) and T5 (B1-A/C) as well as T1' (B1-U/G) and T5' (B1-U/G) will not be reasonably explained. Zhang furthermore provides an example from Yongning Xialeng of YN to indicate the diachronic order of secondary tonal splits. Xialeng has a non-straightforward 35 In Zhang’s data the Southern Thai tones are only transcribed in tonal categories like T1, T3, T5, T7, T8, T1', and T3' with no tonal values. She does not give the specific location to her Southern Thai. The secondary tonal splits conditioned by glottal sounds and voiceless unaspirated stops are identical to my data of Hat Yai of Southern Thai (see L2 in Appendix C) except the merger from T7 (DS1-U and DS1-G) to T8 (DS2), which presents as a derived T7' (DS1-U and DS1-G) with high-rising contour (45) different from T8 (DS2) with low-falling contour (21). 137 tonal split pattern of Type 3, which presents the split condition being glottal sounds together with aspirated stops (cf. the tonal pattern of L29 in Appendix C). Under this condition, T1 (A1) derives T1' (A1-A/G), T3 (C1-A/G) changes to merge into T5 (B1C/U), T5 (B1-A/G) changes to merge into T6 (B2). The latter two changes both involve in T5 (original B1). One explanation making sense is that the original T5 conditioned by aspirated and glottal initials depressed to change to merge into T6 firstly, and this process gave the original tonal space of T5 (B1-A/G) for Tone 3 (C1A/G) conditioned by aspirated and glottal initials to change itself to take over (Zhang 1980: 38). Some aberrant examples in Xialeng give more hints for interpreting this proposal. In this variety, the term haɰ6 ‘to give’ with the aspirated initial (h- has the same tonal behaviors with aspirated stops in this dialect) should have had T5 (=B1-C/U) being changed its original T3 (C1-A) when being conditioned by aspirated initials (vs. the cognate reflecting *haɰC in the majorities of Tai language) due to the regular rule. However, this term actually has T6 (merging into B2). It is possible that when the process that B1-A/G changed to merge into B2 was not yet completely accomplished, the process that C1-A/G changed to occupy the tonal space of original B1-A/G already started, and some single terms from original C1-A/G like haɰ5<3 ‘to give’ felt into original tonal space of T5 (B1-A/G) first and then felt into T6 (B1-A/G > B2) along with all other terms with T5 conditioned by A/G to give a result haɰ6. After the process that B1-A/G merging into B2 was completely accomplished, most other terms with original T3 (A/G) were afterward changed to original T5 but do not fall into B2 anymore (Zhang 1980: 38). The proposal confirms that sound changes are gradually from some single words to spread to all other words. That is to say, there were different diachronic orders even within the non-straightforward tonal split patterns or the claiming secondary tonal splits. The last evidence is in the irregularities of tonal correspondence among some closely related varieties within a distinct language with the same ISO 639-3 code like Yang Zhuang (of CT). A survey of nineteen locations crossing the area of the language varieties below found evidence “from intelligibility, from similarity of wordlists (as determined by a string edit distance algorithm), and from speaker attitudes” has assigned them to the distinct ISO 639-3 language codes [zyg] “Yang Zhuang” (Jackson et al. 2011: V). In the data from my fieldwork, the tone boxes from six Yang Zhuang varieties (L8 – L13 in Appendix C) present the patterns as follows. When tones are conditioned by initials developing from proto-voiceless continuants and unaspirated stops, all these Yang Zhuang varieties agree on the primary split between voiceless-voiced registers. However, when tones are conditioned by 138 aspirated initials (including h- and ɕ-) and initials developing from proto-glottal sounds, they all have different patterns of tonal splits: Jingxi Urban (L12) and Jingxi Anning (L13) have straightforward patterns but are different in the tonal mergers on checked syllables, Jingxi Hurun (L11) has Type 1 of non-straightforward patterns (only conditioned by glottal initials), Debao Dalong (L9) and Debao Lüliu (L10) have Type 3 of non-straightforward patterns (conditioned by both glottalized and aspirated initials) but are all different in the details of tonal mergers in checked syllables, and Debao Urban (L8) is of a mixed Type2/3. When being conditioned by glottalized and aspirated initials, Debao Dalong and Debao Lüliu both merge their high register tones into the tones in the counterpart low register in the tonal categories A, B, and DL, but disagree on the DS2 tone – the former one merge its DS2 into C2, while the latter one merge its DS2 into B2. For Debao Urban, the situation is more complicated (just like in the discussion of Type 2 above). When being conditioned by glottalized initials, it agrees with the other two ones to merge its high tones into the low tones in Columns A, B, and DL; however, when being conditioned by aspirated initials, it only merges its high tones into the low tones in Columns B and DL, but keep the high tones in A. That is to say, even in the same distinct language, the closed related Yang Zhuang varieties36 have different patterns on non-straightforward tonal splits which are conditioned only by glottal sounds or by aspirated sounds together with glottal sounds. This phenomenon indicates that the non-straightforward tonal splits must be of quite late secondary development in the Yang Zhuang varieties, since the split among these varieties cannot be early as the time of primary tonal splits. In conclusion, the non-straightforward tonal splits only conditioned by glottalized sounds, aspirated stops, and/or unaspirated stops in Tai languages cannot be primary. This directly denies that the primary tonal split can be conditioned by the merger of glottal sounds and voiced implosives together with the mergers of voiceless-voiced sonorants in Wenma Tho (Dai Zhuang) suggested by Theraphan LThongkum (1997: 209, 212). 36 These Yang Zhuang varieties are not only linguistically closed related to each other, but also geographically closed to one other. For examples by straight-line distance, Debao Urban, Debao Dalong and Debao Lüliu distribute as a triangle with about three kilometers from one another. Jingxi Urban is about 35 kilometers from Debao Urban. Jingxi Anning is about 35 kilometers from Jingxi Urban, and Jingxi Hurun is about 40 kilometers from Jingxi Urban. The distribution of these languages is shown on the map in Appendix D. 139 As what has been discussed in §4.3.1 above, within the secondary tonal splits (on smooth syllables), the periods of that conditioned by different phonation types of initials must be different. That is, secondary tonal splits must be divided into different diachronic sequences. In this section, the determination of the diachronic sequences of secondary tonal splits is based on the degree of the dialectal uniformity that secondary tonal splits occur. When determining these diachronic sequences, the following discussions will show that some of the secondary tonal split patterns need to be discussed under the three Tai sub-branches altogether, but the other patterns are better to be divided into different language groups to discuss because most of these secondary tonal patterns have individually developed in different languages, dialects, and/or varieties. As shown in the discussions in §4.3.1 above, glottalized sounds are found to condition secondary tonal splits in Li’s all three Tai sub-branches, namely NT, CT, and SWT. The other two groups of initials, voiceless aspirated stops and voiceless unaspirated stops, are found to only condition secondary tonal splits in CT and in CT/SWT respectively. This seemingly suggests that glottalized sounds earliest condition secondary tonal splits within the three groups of initials. However, secondary tonal splits conditioned by glottalized initials present quite different features between NT and CT/SWT languages, as what has been partly discussed in §4.1.3.3.1 that these splits are normally found in the *C tone and are sometimes found in the *B tone in NT languages, but mostly found in the *A and *B tones in CT/SWT languages. In other words, glottalized initials do not uniformly condition tonal splits in the Tai languages like the merger of voiceless-voiced continuants conditioning the primary tonal split. The tonal splits conditioned by glottalized initials may have developed individually in different Tai sub-groups, just like voiceless aspirated and unaspirated initials. Aspirated initials, on the other hand, present to condition secondary tonal splits only in CT and those classification as CT or NT is open to debate. Besides, aspirated initials are treated to be of secondary development which still does not develop in the most majority of NT. In this sense, secondary tonal splits conditioned by aspirated initials could not arise earlier than those conditioned by glottalized initials. However, aspirated initials involving voicing alternation has provided a specific case that a series of tonal splits cover all the three Tai sub-branches. Based 140 upon their widest dialectal uniformity in the secondary tonal patterns, voicing alternation is suggested to arise later than the primary tonal split (since the splits follow the high-low registers), but earlier than the other secondary tonal splits including those conditioned by glottalized initials. In the following sub-sections of this section, voicing alternation uniformly covering all the Tai varieties will be firstly discussed since it is treated as the earliest secondary tonal developing pattern. It is impossible to determine the diachronic order of the splits conditioned by glottalized initials and those conditioned by aspirated initials, because they are all individual development in different languages, dialects, and/or varieties. However, because splits conditioned by glottalized initials are more commonly found in all three sub-branches of Tai, the discussion of it as well as its continuation, splits conditioned by voiceless unaspirated initials, is probed into right after the discussion of voicing alternation. Subsequently, tonal splits conditioned by aspirated initials only found in CT and voicing alternation on the low register only found in NT are successively discussed. Finally, we will revisit the two-way split and three-way split patterns in some specific Tai varieties. In the previous discussions, we have defined that voicing alternation covering all the Tai varieties has the following two distinctive features: 1) the conditioning initial group presents as unaspirated initials in NT but mostly as aspirated (including h-) in CT/SWT; 2) the tones conditioned by this conditioning group fall into the low register in NT, but fall into the high register in CT/SWT (cf. §1.2.2, §1.2.3 and §2.2.3). However, there is another series of voicing alternation attracting less attentions in the previous studies, and the situation of tonal splits/mergers is opposite to the aforementioned one. In this series of voicing alternation, tones fall into the high register in NT, but fall into the low register in CT/SWT, as shown in the following Table 44. 141 Table 44 Series 2/1 of voicing alternation in Tai Gedney’s ST Gloss ‘come’ ‘man’ ‘listen’ ‘be back’ ‘blow’ Li’s SWT CT/NT? Li’s CT NT Bangkok Tai Jingxi Daxin Shuang- Nung Thai Lue Urban Baoxu ding An ma: A2 ma: A2 ma: A2 ma: A2 ma: Wuming Dongling A1 ma:A1 ma:A1 ma:A1 θa:jA1 θa:jA1 ɬa:jA1 tɕʰa:jA2 tɕa:jA2 tɕa:jA2 ɕa:jA2 sa:jA1 - - təŋB2 tʰiŋB2 tʰəŋB1 tiŋB1 - - ta:wB2 tʰe:wB2 ta:wB1 ta:wB1 ta:wB1 ta:wB1 pʰatDS2 patDS2 patDS2 pʰatDS2 pʰatDS1 patDS1 patDS1 patDS1 tiŋB1 tiŋB1 Comparing with the common voicing alternation whose examples are plenty found, this series of voicing alternation is very rare found, and do not provide a specific hints of conditioning feature. For distinguishing these two opposite voicing alternation, the former one is called Series 1/2 (following Liao & Shen 2012) and the latter one is named Series 2/1 in this thesis. We have reviewed that in the previous studies the conditioning causes of Series 1/2 of voicing alternation are considered to be early breathy sounds (Liao & Shen 2012), proto-voiced aspiration (Li 1989, Liang & Zhang 1996, Zhang et al. 1999), a series of voiced initials *B, *D, *G, *V, *Z and *ɣ- (Gedney 1989), as well as a series of sesquisyllabic initial clusters (Pittayawat Pittayaporn 2009). Pittayawat Pittayaporn considers that except those developing from proto-sesquisyllabic initial clusters, many of the voicing alternation are actually of those borrowed from Chinese in history and have nothing to do with Tai source. The first three ones share a common agreement on that they all have voiced and aspirated features which are inherited in NT and CT/SWT respectively. Pittayawat Pittayaporn’s viewpoint may be related to Weera Ostapirat (2006), which firstly systematically suggests the “disyllabic” initial cluster (consonantal onsets plus another consonantal medial) to indicate the cause of voicing alternation in Proto-Kam-Sui (PKS). Although Pittayawat Pittayaporn and Weera Ostapirat do not have an agreement on the sesquisyllabic structure and disyllabic structure in PT and PKS respectively, their different structures agree on that consonantal onsets plus another consonantal medial (different from monosyllabic consonantal clusters) should have existed in these two related language branches in their respective proto-level. This thesis agrees on Pittayawat Pittayaporn’s viewpoint that sesquisyllabic and monosyllabic structures coexisted at the PT level, since it “best accounts for the wide 142 range of sound correspondences observed among the daughter languages” (Pittayawat Pittayaporn 2009: 162). He points out that the growing consensus on that monosyllabic structure has shortcomings on accounting the wide range of onset correspondence within cognates from Proto-TK, on the other hand the sesquisyllabic structure can straightforwardly capture these data (Pittayawat Pittayaporn 2009: 40, 44). He furthermore enumerates several sets of onset correspondence of Tai languages to indicate that the monosyllabic view is problematic when interpreting these sets of data, while sesquisyllabic view can account for these problems (Pittayawat Pittayaporn 2009: 44-52). For example, Li’s reconstructions *pʰr-, *tr-, and *pr- (Li 1977) are challenged by Pittayawat Pittayaporn as in the following Table 45. Table 45 Correspondence involving labial clusters by monosyllabic and sesquisyllabic views (adapted from Pittayawat Pittayaporn 2009: 48-49) Gloss Li SWT (1977) Pittayawat Pittayaporn Saek piəkDS1 pʰja:jC1 pʰrakDS1 pja:jC1 - pʰomA1 pʰjumA1 piəmA1 pʰramA1 ta:A1 ha:A1 ta:A1 pra:A1 ta:jA1 ha:jA1 ta:jA1 pra:jA1 ‘grasshopper’37 takDS1 - takDS1 - ‘to exposed’ ta:kDL1 pʰja:kDL1 ta:kDL1 pra:kDL1 te:kDL1 pre:kDL1 tawB1 - ‘to walk’ *pʰ- *pʰr- (2009) pʰakDS1 *pr- ‘head hair’ ‘eye’ ‘to die’ 2 ‘to burst’ 3 *t- *t- *tr- *pr- Thai Lung- Yay ‘vegetable’ 1 Proto- *p.t- *p.r- ‘to hunt’ pʰa:jC1 tɛ:kDL1 - chow pʰjakDS1 pʰe:kDL1 pʰjawB1 Pittayawat Pittayaporn first points out that Li’s reconstruction on Set 2 and Set 3 cannot account for the same containing of labial consonant suggested by Saek. Set 2 of Saek cannot be treated as innovation because they are attested by the cognate forms in some Tai or KS varieties like pʰja:A1 ‘eye’ in Bao Yen of CT and pla:A1 ‘eye’ in Lakkja of KS38. Li’s set 1 is also not supported by the view that PT lacked of aspirated sounds. As Li’s *pr- must be reconstructed by his another viewpoint on aspiration of original PT clusters (cf. Pittayawat Pittayaporn 2009: 93-101), he gives two sesquisyllabic clusters *p.t- and *p.r- to replace Li’s *tr- and *pr- respectively. He In Table 45, the full form of the word for ‘grasshopper’ in Thai is actually takDS1 ka-tɛ:nA1. Quite a few NT varieties also support the labial consonant contained in Set 2 of Saek, like pja:A1 ‘eye’ in Liujiang, Huanjiang, Rong’an, Longsheng, and Hechi (cf. Zhang et al. 1999: 640). 37 38 143 explains that *p.r- merged into *p.t- then later simplified to t- in Thai and Yai, on the contrary *p.t- merged into *p.r- then later changed to pr- in Saek. In contrast, Lungchow keeps the distinct by the modern reflexes h- and pʰj- (Pittayawat Pittayaporn 2009: 48-49). On the development from p.t- to h- and from p.r- to pʰj- in Lungchow, he suggests the processes *p.t- > *p.r- > *ʰr- > h- (Pittayawat Pittayaporn 2009: 163) and *p.r- > *pr- > *pʰj- respectively (Pittayawat Pittayaporn 2009: 171). In short, his sesquisyllabic structure of PT can successfully accounts for both labial reflexes on Set 2 and Set 3 without losing the distinction, which cannot be explained by the monosyllabic view. Besides, as we have suggested that there was no contrasting voiceless aspiration in PT level (cf. §4.2.1), so do Liang and Zhang (1996: 106-107) and Zhang and colleagues (1999: 211). This is crucial to query that the breathy sounds (cf. Liao & Shen 2012) or voiced aspirated sounds (cf. Li 1989, Liang & Zhang 1996, Zhang et al. 1999) which cause Series 1/2 of voicing alternation existed at the PT level since it was uncommon that the aspiration isolatedly existed in the voiced group if there was no contrasting aspiration existing in the voiceless group, thus Liang and Zhang (1996) and Zhang and colleagues (1999) may have a self-contradiction in this point. If we furthermore explore the more compellent suggestion for explaining voicing alternation occurring in PKS (cf. Weera Ostapirat 2006), we will see Pittayawat Pittayaporn’s sesquisyllabic onset can also account for the factors causing voicing alternation via the later breathy or voiced-aspirated procedure. As what has been reviewed in §2.2.3, Weera Ostapirat (1994) suggests that a series of breathy/voicedaspirated sounds were the actual factor for causing voicing alternation on sonorants among KS languages. In his new scheme for the cause (disyllabic onsets) of the voicing alternation in KS languages, he suggests that these sonorants were originally medials, the non-initial segment of the disyllabic onset or Pittayawat Pittayaporn’s sesquisyllabic onset, giving a byproduct breath/aspiration feature which finally give a tone in tonal series 1 (the high register) in some KS languages (Weera Ostapirat 2006: 1113-1114). That is, the breathy/voiced aspiration which is proposed to appropriately account for the voicing alternation between NT and CT/SWT may have secondarily appeared in history following their predecessors – a specific series of sesquisyllabic onsets. Thus, the main different theories of the conditioning cause of voicing alternation, breathy/voiced aspirated sounds (Gedney 1989, Li 1989, Liang & Zhang 1996, Zhang et al. 1999, Liao & Shen 2012) and a series of sesquisyllabic onsets (Pittayawat Pittayaporn 2009), become consistent on different diachronic sequence in the method from Weera Ostapirat’s explanation. We therefore believe that there were no breathy or voiced aspirated sounds at the PT 144 level, instead sesquisyllabic onsets including voicing alternation between the later NT and ST (the Proto-CT/SWT) did exist at the PT level. This proposal can also explain that many items with voicing alternation were actual OC and MC loans terms originally with pre-nasalized initials (cf. Pittayawat Pittayaporn 2009: 86), since the process of dropping of pre-nasalized segment might also cause voiced aspiration or breathiness. Those loans were borrowed as with aspirated initials (conditioning a high register tone) into CT/SWT. On the other hand, the voiced property in the breathiness of these loans became plain voiced stops (later conditioning a low register tone) in NT which do not have aspirated sounds. Table 46 Etyma with PT *C̬.t- (Pittayawat Pittayaporn 2009: 168) Gloss PT Thai Lungchow Yay ‘to arrive’ *C̬.tɤŋA tʰɯŋA1 tʰɤŋA1 taŋA2 *C̬.tɯ:kD tʰu:kDL1 tʰukDS1 tɯkDS2 *C̬.tɯəlB tʰɯənB1 tʰɤ:nA1 (tone change) tɯənB2 *C̬.tuəmB tʰuəmB1 tʰu:mB1 tu:mB2 ‘to come into contact’ ‘wild’ ‘to flood’ 39 The next concern is the determination of the phonetic characteristics of the two series of sesquisyllabic onsets that caused voicing alternations or alternations of tonal series. Previous studies have focused on Series 1/2 because this series have plenty data for comparing. For example, the vocabularies in Table 46, show a set of Series 1/2 of voicing alternation among Thai, Lungchow and Yay representing SWT, CT and NT respectively. Pittayawat Pittayaporn speculates a “voiced obstruent + voiceless stop” *C̬.C̥- for this series. Although he thinks that the initial voiced consonants of *C̬.C̥- were unclear and might be different due to different etyma, he The fourth gloss ‘to flood’ in Table 46 may be of data errors. This term is actually reflected as tʰuəmC1 in Thai and tʰu:mC1 in Longchow (cf. Zhang et al. 1999: 700) instead of the two forms with tone B1 in the chart. Its Thai orthographical form ทวม reflecting the B2 tone which is still different from the B1 in the chart, and this may be an incorrect normalization in Thai due to the merger of B2 and C1 in modern Thai. The forms in other CT/SWT dialects like tʰuəmC1 in Chiang Rai and tʰo:mC1 in Tai Lue (SWT), as well as tʰu:mC1 in Debao, Jingxi, and Daxin (CT), and tʰumC1 in Guangnan and Yangshan (CT) all reflect *tʰuəmC for this term. In contrast, many NT forms all reflect a *dumB tone for this term, like tumB2 in Wuming, Pingguo, Tianlin, and Lingle, as well as təmB2 in Qiubei (cf. Zhang et al. 1999: 700). Although *tʰuəmC reflected in CT/SWT and *dumB reflected in NT have similarities on the phonetic shape, their different tonal categories give a doubt to their cognate relation. It is crucial that in Debao Yang Zhuang (CT), there are two terms tʰu:mC1 (reflecting *tʰuəmC) and tʊmB2 (reflecting *dumB) for ‘to flood’, and the later one is normally used as idiomaticity like namC2 ɹʊmB2 tʊmB2 mɤ:ŋA2 ‘the Great Flood submerging the world’. If the two terms in this dialect are actually from two different etyma, the formulation using them to give an example for voicing alternation will be lost efficacy. 39 145 interprets that in NT the non-initial segment (voiceless *.C̥-) was assimilated to the initial segment (voiced *C̬.-) and become voiced to condition tones into the low register, on the other hand in CT/SWT the loss of the initial segment (voiced *C̬.-) caused its following voiceless *.C̥- become aspirated to condition tones into the high register (Pittayawat Pittayaporn 2009: 167-169). As ‘to arrive’, one of the examples in the chart above, presents a diachronic developing process as *C̬.tɤŋA > *dɤŋA > taŋA2 in Yay (NT), and *C̬.tɤŋA > *tʰɤŋA > tʰɯŋA1 in Thai (SWT). Pittayawat Pittayaporn furthermore explains that his reconstruction of this onset is based on that the etymon *C̬.tɤŋA may be related to Proto-Austronesian *dateŋ ‘to arrive’ (cf. Thurgood 2007: 255) “which shows the expected sequence” (Pittayawat Pittayaporn 2009: 168). Moreover, he gives some examples with other *C̬ .C̥- onsets reconstructed like *C̬ .k- and *C̬ .q- (Pittayawat Pittayaporn 2009: 168-169) to indicate that they likewise cause Series 1/2 of voicing alternation. On the other hand, Weera Ostapirat gives another explanation on the origin of voicing alternation. Although his reconstructions are for interpreting the situation of voicing alternation in KS languages, we still treat them as efficient for those in Tai because the situations in KS and in Tai are structurally similar, and some of them are even the same since KS and Tai are closely related branches in the TK family. He suggests that in some disyllabic onsets, the series with accented initial (like *k'-m-) and another series with unaccented initial segment (like *k-m-) have given different register splits in KS languages (Weera Ostapirat 2006: 1082). In the following Table 47, he gives some examples with the proposed accented initial disyllabic onsets to interpret the voicing alternation in KS languages. Table 47 Etyma with PKS *h'-w- (Weera Ostapirat 2006: 1085)40 Gloss Then Mak Sui Mulam Kam Tones ‘right (side)’ waa2 phaa1ʰ faa1 faa1 waa1ʰ *A ‘late’ wee2 -- fee1 fɛɛ1 wee1ʰ *A ‘garden’ wjaan2 fiin1ʰ fjaan1 fjen1 jaan1ʰ *A In his purpose, both accented and unaccented initials can give the following medial an aspirated or a devoicing effect in some daughter languages, but just simply loses 40 The -h on the top right corner of the tonal maker 1 in this chart refers to the derived tone conditioned by aspirated sound (including fricatives and voiceless sonorants) in Kam and Mak of the KS languages, following the tonal labeling system of Weera Ostapirat (2006: 1078). 146 in some daughter languages. Therefore in the chart above, PKS *h'-w- has become initial w- in Then with tonal Series 2 according to the Then voiced initial, has become f- in Sui and Mulam with tonal Series 1 according to their voiceless initials, has become pʰ- (and f- when following by -j-) in Mak with tonal series 1ʰ according to Mak voiceless aspirated initials, and has become w- (and j- when following by -j-) in Kam with tonal series 1ʰ indicating early Kam hw- due to the Kam tone conditioned by voiceless aspirated initials. He furthermore points out that the terms in this chart are related to some terms with proto-initial *s- in Tai, like Thai sa:jA1 ‘late’ and suənA1 ‘garden’; and some of the Tai varieties have labio-velar features for this set, namely Tho (CT) swa:jA1 ‘late’ and swa:A1 ‘right (side)’. He even suggests that this extra-KS evidence indicates a source *s- to the accented *h'- as in PKS *h'-w- < *s-w- (Weera Ostapirat 2006: 1085). Gedney (1989a: 245) suggests the initial of the term *sa:A1 ‘right (side)’ reflected in the majority of CT (agreeing with the Lungchow form in Table 48) to be contaminated by the initial of the counterpart term *za:jC ‘left (side)’, e.g. Lungchow ɬa:A1 ‘right (side)’ vs. ɬa:jC2 ‘left (side)’. Pittayawat Pittayaporn agrees on this point and on that this term has a “clear Chinese origin” (右 jəuB) by citing from Thurgood (2007) (Pittayawat Pittayaporn 2009: 82-83). For the other term ‘late (in the morning)’, Pittayawat Pittayaporn agrees on Thurgood (2007: 246-248) that among Thai, Lungchow and Po-ai it “shows irregularities in vowel length, and tone, as well as onsets” and consider that the NT term which suggests earlier *gw- is not related to the SWT/CT term (Pittayawat Pittayaporn 2009: 81), although in fact the vowel length and the tone (reflected) do not show irregularities at all. If we compare more data of other varieties from SWT, CT, and NT (like shown in Table 48) and the data from extra-Tai evidence (in Table 47), the viewpoints provided by Gedney, Pittayawat Pittayaporn as well as Thurgood above cannot be true. By comparing the terms ‘right (side)’ and ‘late’ with labio-velar features in other Tai varieties like Chiang Rai (Northern Thai), Rong Maet (Tai Lue), Debao Yang Zhuang varieties (L8, L9, and L10) and Debao Ronghua (L14) as well as their cognates presenting voicing alternation in NT, we will partly agree to Weera Ostapirat’s explanation involving voicing alternation on this set, as shown in Table 48. 147 Table 48 Etyma with the proposed PT *ɬ.w- which causing voicing alternation in Tai41 SWT Gloss Thai (L1) ‘right (side)’ ‘late’ Chiang Rai (L4) CT Tai Lue Lung- (L6) chow Tho NT Ronghua Debao (L14) (L8) Po-ai Wuming (L30) kʰwa:A1 kʰwa:A1 kʰwa:A1 ɬa:A1 swa:A1 θwa:A1 ɬɔ:A1 kwa:A2 kwa:A2 sa:jA1 kʰwa:jA1 kʰwa:jA1 ɬa:jA1 swa:jA1 θwa:jA1 ɬɔ:jA1 kwa:jA2 kwa:jA2 In Table 48, the additional SWT/CT Tai forms with labio-velar features (which are all in shaded) are added to indicate that both of these two terms share the same sesquisyllabic proto-onset which caused voicing alternation (of Series 1/2) between NT and ST (CT/SWT), especially when being compared with their cognates with labio-velar features in KS languages (shown in Table 47). All the forms reflect a PT tone *A to indicate that the term ‘right (side)’ has nothing to do with the claimed Chinese origin with a qù (departing) tone which corresponds to PT *B. The reason that I reconstruct the initial *ɬ.w- for this set is as follows. (1) The method of sesquisyllabic initial segment which cause voicing alternation in daughter languages agrees on Pittayawat Pittayaporn (2009: 167-169). Because Weera Ostapirat’s method that the accented and unaccented initials affecting on the following medials lacks of a consistent tonal behavior in a specific modern language, e.g. his PKS *h-p-, *k'-p-, *h'-t-, *k'-t-, *p'-t-, *p'-k- and *t'-k- condition tonal series (high-1 and low-2) 2, 1, 1, 1, 1, 2 and 2 in Then respectively (cf. Weera Ostapirat 2006: 1086, 1088, 1091, 1103, 1104), we cannot judge whether the proposed accented or unaccented initial could give the effect on the following medial consonant in Tai languages. Besides, the situations of voicing alternation among Tai languages are quite straightforward when being compared with those in KS languages, as both Series 1/2 and Series 2/1 occur between NT and CT/SWT (or Gedney’s ST). Therefore, Pittayawat Pittayaporn’s structure like C̬ .C̥- can be more advisable for the situation of Tai. (2) PT *s- commonly reconstructed (cf. Li 1977: 152, Pittayawat Pittayaporn 2009: 115) always exhibits the same tonal behavior as voiceless sonorants (*hm-, *hn-, *hɲ-, *hŋ-, and *hl-) and voiceless approximants (*hr- and *hw-) to form a unified group proto-voiceless continuants but appears to be quite out of place in this series phonetically. Since in many CT and NT varieties, such as Lungchow and Po-ai, the 41 Data in this chart are collected by me except those of Lungchow and Po-ai cited from Pittayawat Pittayaporn (2009: 80, 82) and Tho cited from Weera Ostapirat (2006: 1085). 148 speculated *s- presents as a voiceless lateral fricative ɬ-, which contains a feature of lateral shared with the sonorant l-, Liao and Shen (2012) has suggested the protoform of it may be *ɬ- instead of *s-. This accounts for its unusual phonetic characteristics when compared with the rest of the voiceless sonorants and approximants that it patterns with and provides a reasonable explanation for why these sounds pattern together as they do. (3) The medial *.w- is reconstructed due to the labio-velar features reflected in some Tai varieties and their related KS languages. The following initial and tonal development speculated shown in Table 49 give the assurance for this reconstruction. Table 49 Developing tracks of PT *ɬ.w- in modern Tai varieties42 PT Developing process Modern forms *k.w- (POA *gw- (voicing assimilation) assimilation) most CT *ɬw- *ɬ- Tho (CT) *ɬw- sw- Ronghua (CT) *ɬw- θw- NT (Series 2) ɬV[+rounded] *ɬw- *ɬwThai (SWT) *k.w- (POA most SWT low ɬ-/θ-/s- *ɬ.wDebao (CT) kw- Tone assimilation) *ɬ (MOA high assimilation) (Series 1) s- *kʰw- (aspirated due to the syllabic kʰw- structure change) The situation of those CT varieties preserving labio-velar features indicates that after voicing alternation process, the loss of the labio-velar segment in -w- of this set in CT was quite late. The most crucial evidence is that in different Yang Zhuang varieties which are of the same distinct language with the same ISO 639-3 code (zyg), most of Debao varieties have ɬɔ:A1 and ɬɔ:j1, but all Jingxi varieties and some Debao varieties have ɬa:A1 and ɬa:jA1 for these two term respectively. In contrast, the situation in Thai present two different tracks with the two terms, and this perhaps suggests the individual development of these two terms were quite early in this dialect. 42 The terms POA and MOA in Table 49 are the abbreviations of the phonetic terms “place of articulation” and “manner of articulation” respectively. 149 However, Pittayawat Pittayaporn’s C̬ .C̥ - which is suggested to cause voicing alternation still cannot cover this set since the medial *.w- has to be a voiced segment against to his suggesting *.C̥ -. Another possible reconstruction meeting his criterion is *ɮ.p- which can be supposed to have different development, namely *ɮ.p> *ɮ.b- > *ɮ.w- > *g.w- > *gw- > kw- (with low register tone) in NT, *ɮ.p- > *ɮ.b- > *ɮ.w- > *ɬ.w- > *ɬw- > ɬ- (with low register tone) in most CT, and *ɮ.p- > *ɮ.b- > *ɮ.w- > *g.w- > kʰw- (with low register tone) in most SWT. But this proposal contains more complicated sound changes and is left to further discuss in the future since the reconstruction of initial forms is not the main target of this thesis. Therefore, here we only give a concise and comprehensive proposal on the reconstruction of the onsets causing Series 2/1 of voicing alternation (shown in Table 44). Since all items in this series have an initial stop (including the nasal stop m- and the affricate tɕ-/ tɕʰ- which phonologically belong to stops, as well as their variations ɕ-/s-/θ-/ɬ-), I temporarily suppose their initials are of a sesquisyllabic cluster *C̥.C̬- (voiceless stop + voiced obstruent), to be relevant to the situation (low tone in CT/SWT and high tone in NT) which is opposite to Series 1/2. Hereafter we will leave the arguments of the voicing alternation induced by onsets reconstructed, but only focus on the result of alternation of tonal series, to name these two groups of onsets “Series 1/2 of voicing alternation induced by sesquisyllabic onsets” (VASO- 1/2) and “Series 2/1 of voicing alternation induced by sesquisyllabic onsets” (VASO2/1) respectively. In conclusion, after the primary tonal split, VASO-1/2 onsets became voiced stops to condition the tone into the low register in NT, and became voiced aspiration or breathy sound due to the loss of the previous initial segment (Weera Ostapirat’s accented/unaccented segment and Pittayawat Pittayaporn’s voiced segment) and then merge with the secondary voiceless aspiration to condition the tone into the high register in SWT/CT. On the other hand, VASO-2/1 onset became voiced stops to condition the tone into the low register in CT/SWT, and became voiceless obstruents due to assimilation of the initial *C̥.- segment to condition the tone into the high register in NT. Tonal behaviors involving voicing alternation between NT and CT/SWT suggests that Li’s CT and SWT were of the same group with PNT as a sister when splitting from PT, and this group is suggested to be ST by Gedney (1989a: 231). This will be further discussed in Chapter 5. 150 Tonal splits secondarily conditioned by glottalized sounds has been pointed out to be exist in Li’s all NT, CT, and SWT groups (cf. §4.3.1 and §4.3.2). In NT languages, these splits are concentrated in the *C and/or *B tone, but are very rarely found in the *A tone. In contrast, in CT/SWT languages the splits are normally found in the *A and *B tones, but are rare in the *C tone. These phenomena may be related to that the final glottal constriction on the C tones are very robust in CT/SWT, but are reduced in NT (cf. §4.1.3.3.1). Tones secondarily conditioned by glottalized sounds are normally depressed from the original high tone due to that glottal sounds are known to have a pitch-lowering effect on the vowel (Fu 1995: 64). Actually all tones of secondary splits are lower than the original one (Liang & Zhang 1996: 818), except in those Tai varieties that have developed tonal flip-flop (cf. §4.1.3.1). There is no lack of exceptions. Po-ai provides a single NT example that has a secondary split in the *A tone conditioned by glottalized sounds in the previous studies. The six tones A1, A2, B1, B2, C1, and C2 in smooth syllables in Po-ai are 24, 55, 22, 31, 44, and 33 respectively (Li 1977: 28). Both *B and *C present the straightforward pattern splitting between voiceless-voiced registers. However, *A has a three-way split pattern – initials developing from PT voiceless unaspirated sounds and voiceless continuants condition the 24 tone, initials from PT glottalized sounds condition the 31 tone (which is merged into B2), and initials from PT voiced sounds condition the 55 tone (cf. Li 1977: 29). Isan/Lao of SWT also merge their C1G tone together with C1-U into C2 as shown in the tone box of Khon Kaen (L3) in Appendix C, to indicate that secondary tonal splits conditioned by glottalized sounds also can occur in the *C tone in SWT, although such phenomenon is rarely found. At this point, secondary tonal splits conditioned by glottalized sounds are mostly common found in all Tai languages. Thus, glottalized sounds may earliest cause secondary tonal splits before aspirated stops and unaspirated stops. Data from my fieldwork show that glottalized initials have undergone kinds of sound changes in many modern Tai varieties except the glottal stop ʔ-. Pre-glottalized stops ʔ b- and ʔd- weakened to merge into plain voiced stops b- and d- are common in SWT. In NT and CT varieties as well as Shan in SWT, pre-glottalized stops ʔb- and ʔd- are merged into nasals m- and n-/l- (like in Shuangding YN, Jingxi Urban Yang Zhuang and Shan). In addition, the pre-glottalized glide ʔj- has been merged into j- in most SWT and CT varieties, but is preserved in many NT varieties and Debao Yang Zhuang varieties of CT. In Quan Son Tai (L7 in Appendixes C) of SWT, all preglottalized initial consonants have merged into their homorganic sonorants or 151 fricatives, and the processes are as *ʔb- > ʋ-, *ʔd- > l-, and *ʔj- > ð-, like in ʋa:nC1-G ‘village’, la:wA1-G ‘star’, and ða:A1-G ‘medicine’. However, in most situation these initial changes are later development than the time of tonal splits conditioned by their predecessors glottalized sounds. Po-ai of NT has also merged their glottalized stops into nasals, like min31 ‘to fly’, ni:31 ‘good’, and nɯ:n31 ‘month’, but their tone (A1-G) is still preserved to be the same as in ʔaw31 ‘to take’, to be different from the tones of paj24 (A1-U) ‘to go’, ka:j24 (A1-A) ‘to sell’ and na:55 (A2) ‘rice field’ (cf. Li 1977: 29). The phenomenon indicates that the glottalized stops in Po-ai conditioned the secondary tonal split in Tone *A before they are merged into nasals. The following Table 50 shows the similar situations. Table 50 Reflexes of *ʔd-, *ʔb-, and *ʔj- and their tones43 PT initials *hm- *k- *ʔd- *ʔb- *ʔj- Gloss ‘dog’ ‘crow’ ‘star’ ‘to fly’ ‘medicine’ Tonal categories A1-C A1-U ma:1 ka:1 la:w1 ʋin1 ða:1 na:2 xwa:j2 ma:1 ka:1 na:w1 man1 ja:1 na:2 ʋa:j2 ma:1 ka:1 na:w2 min2 ja:2 na:2 ʋa:j2 ma:1 ka:1 ˀda:w2 ˀban2 ja:1 na:2 ʋa:j2 Quan Son Tai (L7) Jingxi Urban (L12) Tiandeng Urban Jingxi Hurun (L11) *n- *ɣw- ‘rice ‘buffalo’ field’ A1-G A2 In the chart above, the first two varieties have straightforward tonal splits patterns in the tone *A, thus their tones conditioned by the original glottalized initials are preserved as T1, even though the glottalized sounds have been merged into voiced continuants in these two varieties. Tiandeng has a secondary tonal split conditioned by the original glottalized initials, therefore the tones conditoned by these sounds had split from T1 to merge into T2, before the glottalized sounds were merged into voiced continuants. Hurun presents an interesting counter-example. In this variety, pre-glottalized stops are still preserved and condition the secondary tonal split to merge into T2, like in the terms ‘star’ and ‘to fly’. However, pre-glottalized glide ʔj- has been merged into 43 For indicating the tonal splits and mergers, the tone markers in the Sinitic circles is followed to use in this chart. Secondary tonal splits conditioned by original glottalized sounds are in shaded. Data from Tiandeng Urban which is not in the list of the forty-two languages is collected by me. 152 j-, and this time the tone conditioned by this j- (< *ʔj-) is preserved the original T1 of the primary tonal split. This example accounts for the following speculations for this dialect. (1) Before conditioning the secondary tonal split, all the glottalized sounds did have the same tonal behaviors with the other voiceless groups to condition tones into Series 1 (the high register). This can be an additional evidence for indicating that tonal splits conditioned by glottal sounds are secondary. (2) Later, pre-glottalized glide *ʔj- merged into voiced glide j- but kept its original tonal behavior (Series 1). (3) After that, the preserved pre-glottalized stops conditioned the secondary tonal split in the tone *A (1 > 2). The initial j- (< *ʔj-) does not play a part in this process since its glottal feature has been lost. In short, glottalized sounds are given their tonal behavior when they have the glottalized feature. This tonal behavior can be secondarily changed (normally depressed or pitch-lowering) due to their glottal feature. The tonal behavior conditioned by original glottalized sounds, both primary or secondary, is preserved even though the glottalized sounds are merged into other sounds in some modern Tai varieties. To make a passing remark, secondary tonal splits conditioned by voiceless unaspirated stops all follow the patterns of those conditioned by glottalized sounds due to their phonetic similarities, and are only found in some SWT and CT varieties (cf. 4.3.1). They can be treated as the byproduct of secondary tonal splits conditioned by glottalized sounds. As in the previous discussions, aspirated sounds in CT/SWT and YN varieties (of those whose classification of CT or NT is open to debate) are secondary in post-PT level (cf. §4.2.1). So far, NT varieties still do not develop contrasting aspirated sounds. Therefore, tonal splits conditioned by aspirated sounds on the original high register have to be secondary. As what has been emphasized in the previous discussions (cf. §4.3.1 and §4.3.2), although aspirated sounds are robust in SWT, CT and YN, they are not found to condition secondary tonal splits in SWT, but only found in CT and YN. This phenomenon suggests that on a certain level CT languages have their own phonological development to separate them from SWT varieties (cf. L-Thongkam 1997: 215). Note that YN varieties such as Yongning Baiji (L25), Naning Shuangding (L26), Long’an (27), Nung An (28), and Yongning Xialeng (29) 153 investigated in this thesis are conventionally treated as dialects of SZ which is traditionally grouped into CT, because these varieties also have plenty aspirated initials which condition possible secondary tonal splits as well. But at the same time they present more NT features on voicing alternation, vocalic behaviors, and some of the initial behaviors. The secondary tonal splits conditioned by aspirated sounds in many of these dialects make their classification as CT or NT more complicated to debate (cf. Chapter 5). Secondary tonal splits conditioned by aspirated sounds (on the high register) are summarized as in Type 2 (aspiration condition splits alone) and Type 3 (aspiration condition splits together with glottal sounds) by Zhang and colleagues (1999), and we have also illustrated some of the mixed Type 2-3 and Type 1-3, such as Bac Va Nung, Lingding Zuojiang Zhuang, and Debao Urban Yang Zhuang (cf. §4.3.1). Within the forty-two Tai varieties investigated in this thesis, nine from CT and those of YN confirm tonal splits conditioned by aspirated sounds. In other words, these nine varieties have a split between voiceless aspirated sounds and voiceless continuant, which are both included in Gedney’s voiceless frictions. The significant inspiration of tonal splits due to aspirated sounds is to add an extra row (1-A) into Gedney’s Tai tone box (cf. Liao & Shen 2012, §1.2.2, and Chapter 6) at least when applying it to CT and YN. However, it is necessary to distinguish different origins of aspirated initials in PT since different groups of these aspiration induced by onsets do not always present as aspirated in different groups of modern Tai languages. For example, VASO-1/2 normally present as aspirated stops (including h- and ɕ- in some specific Tai varieties like Yang Zhuang) in both CT and SWT varieties, but never present as aspirated initials in any varieties of YN even though YN is treated as a SZ dialect in the Sinitic circles because they all have robust aspirations. Thus VASO-1/2 must be separated from the 1-A group to be an extra row in the ideal tone box, otherwise the tone box would not capture this tonal distinction between CT/SWT and YN/NT. The following discussions in this sub-section will focus on different groups of aspirated initials beyond VASO-1/2. The most significant group of aspirated sounds is the 1-A group in Liao & Shen’s Tai tone box (cf. Table 7 in §1.2.2), which is considered to commonly present as aspiration in all CT/SWT and YN varieties. This group is treated as monosyllabic onsets which includes those considered to be “plain stop + *-r-” clusters, *q-, *x-, *χ-, *h-,*ʰw-, and *ʰr- in PT by Pittayawat Pittayaporn (2009) as in Table 51. 154 *pʰr- *tʰr- *tʰl- ST (2009) Pittayaporn Pittayawat Li (1977) Table 51 PT monosyllabic onsets induced by aspiration in ST and YN Gloss ‘vegetable’ *pr- *tr- *kʰr*kʰ-/ *x- *kr- *qr- hawA1 ‘tail’ ‘plough’ ‘to wait’ ‘six’ *χ- *f- *ʰŋ-/ ʰr- 44 *h- tʰawA1 tʰa:ŋA1 hinA1 tʰənA1 tʰajA1 ‘white’ ‘to cross’ ‘fang’ pʰja:kDL1 pʰjamA1 tʰawA1 tʰɯ:ŋA1 Po-ai Zhongyi pʰlakDS1 Yay pjakDS1 piəkDS1 pja:kDL1 pja:kDL1 pʰlamA1 pjɔmA1 piəmA1 tʰawA1 lawA1 rawA1 lɯ:ŋA1 rɯəŋA1 pʰla:kDL1 tʰu:ŋA1 hinA1 rinA1 tɕɔ:jA1 tɕajA1 tɕajA1 sajA1 sajA1 tʰa:C1 tʰa:C1 tɕa:C1 - sa:C1 sa:C1 ha:A1 kʰja:A1 ʰɹa:A1 ʰla:A1 la:A1 ra:A1 lɔkDS1 rokDS1 ʰli:wA1 li:wA1 riəwA1 ʰlajB1 tɕajB1 tɕajB1 hokDS1 kʰjɔ:kDS1 kʰjajB1 ‘torn’ Nalong- Nung An pʰjakDS1 NT tʰənA1 kʰajB1 ‘to sift’ YN - ‘egg’ kʰɯŋA1 kʰa:C1 kʰa:tDL1 kʰɛ:nA1 kʰiəwA1 kʰa:wA1 kʰa:B1 kʰa:mC1 kʰiəwC1 kʰjaŋA1 ʰɹɔkDS1 ʰɹi:wA1 ʰɹajB1 ʰlɔkDS1 - ʰlaŋA1 tɕaŋA1 - ka:C1 ka:C1 ka:C1 ka:C1 ka:tDL1 ka:tDL1 ka:tDL1 ka:tDL1 kʰe:nA1 ke:nA1 ke:nA1 tɕe:nA1 tɕe:nA1 kʰi:wA1 he:wA1 je:wA1 he:wA1 hewA1 ha:wA1 ha:wA1 ha:wA1 ha:wA1 kʰa:B1 - - - ha:B1 kʰa:mC1 ha:mC1 ha:mC1 ha:mC1 ha:mC1 he:wC1 he:wC1 he:wC1 hewC1 kʰa:C1 kʰa:tDL1 kʰa:wA1 kʰe:wC1 kʰiŋA1 kʰəŋA1 həŋA1 həŋA1 hiŋA1 hiŋA1 ha:C1 ha:C1 ha:C1 ha:C1 ha:C1 ha:C1 hajC1 hɔ:jC1 hajC1 hajC1 haɰC1 haɰC1 ‘fragrant’ hɔ:mA1 ho:mA1 ho:mA1 ho:mA1 ho:mA1 homA ‘lid’ fa:A1 pʰa:A1 pʰa:A1 pʰa:A1 fa:A1 va:A1 fɯ:C1 - A1 - ‘ginger’ ‘to give’ *ʰw- ‘cloud’ *ʰr- pʰjamA1 kʰu:A1 ‘five’ *h pʰja:kDL1 huəA1 ‘galangal’ *x- ha:ŋA1 pʰjakDS1 ‘to laugh’ ‘arm’ *x- pʰakDS1 ‘headlouse’ ‘green’ *x- Debao pʰomA1 ‘to kill’ *q- Thai ‘head hair’ ‘to seek’ *xr- CT ‘forehead’44 pʰa:kDL1 ‘stone’ *cr- SWT fa:C1 pʰa:C1 fa:j ‘to yawn’ ha:wA1 ha:wA1 - ha:wA1 hawB1 hawB1 ʰɹ awB1 ʰlawB1 hiəwB1 he:wB1 ‘to bark’ ‘withered’ A1 pʰa:j pʰy:C1 ‘dam’ A1 pʰa:j pʰɯ:C1 A1 ʰɹ e:wB1 The full form of the word ‘forehead’ in Thai is na:C1 pʰa:kDL1. 155 pʰa:j A1 ʰle:wB1 fa:j ŋa:wA1 ra:wA1 lawB1 rawB1 - re:wB1 Li’s reconstructions in Table 5145 are rejected by the viewpoint that PT lacked of aspirated sounds. In contrast, Pittayawat Pittayaporn (2009: 93) supposes that some etyma with aspirated onsets in CT and SWT must have had an earlier media *-rwhich caused PT voiceless plain stops to become aspirated in the first five sets in Table 51. In his proposal, Debao has implemented the effect of the medial *-r- to cause all the preceding stops to become aspirated, but the other three dialects do not always carry out the aspirated process. On the sets *pr- and *cr- the medial *-r- had an effect on the preceding stop to cause aspiration in Thai, but had just been simply lost in Po-ai. On the sets *tr- and *kr-, the following processes may have occurred respectively: *tr- > *tʰr- > *ʰr- > h- and *kr- > *kʰr- > *ʰr- > h- in Thai, *tr- > *tʰr- > *ʰr- > r- as well as *tr- > *ʰr- > *r- > l- and *kr- > *ʰr- *r- > l- in Po-ai (Pittayawat Pittayaporn 2009: 94-95). We can also speculate *kr- > *kʰr- > ʰr- in Nung An, and furthermore change into ʰl- in Nalong Zhongyi. An example illustrated in §4.2.1 has directly supported Pittayawat Pittayaporn’s proposal. Shuangding (L26) of YN partly preserves the cluster pr- which have become aspirated pʰj- or pʰin most CT/SWT varieties, but the neighbouring Nalong Zhongyi variety (from my fieldwork) has changed the preceding plain stop into aspirated stops and has changed the medial -r- to -l- to make the initial become pʰl- at the same time, such as pra:33 (A1-A) contratsing to pʰla:453 ‘stone mountain/hill’ in Shuangding and Nalong Zhongyi respectively. For this reason, this thesis also supports the viewpoint that the earliest origin of aspiration in CT and SWT should be “voiceless unaspirated stops + -r-” clusters, namely *pr-, *tr-, *cr-, and *kr- in Table 51. Only if aspiration arose due to “voiceless unaspirated stops + -r-”, can other series of initial such as rich dorsal series (as illustrated in the latter six sets in Table 51) result the same phonation type in CT/SWT and YN. Note that h- in Debao as well as h-, ʰr- or ʰl- in Nung An and Nalong Zhongyi are grouped into aspirated sounds (1-A) because they always condition tones together with aspirated sounds due to the [+spread] feature shared among them. Sets *cr- and *q- are noteworthy on the 45 In Table 51, data of Thai, Debao (the former four sets), and Yay are mainly adopted from Pittayawat Pittayaporn (2009: 94, 104-105, 117-119, 133-135, 144-145). Po-ai Data are adopted from Li (1977: 148, 208-209, 229, 250). Data of Debao (the latter seven sets), Nung An and Nalong Zhongyi are collected by me. Pittayawat Pittayaporn’s original data of the first four sets are not provided the tonal series (1 or 2) for those on smooth syllables, but all terms are actually with Series 1 since these sets are from PT voiceless initials. Set 1 to 4 of Debao in the chart should be of Yang Zhuang due to the data provided, but some terms originally provided by Pittayaporn are also problematic, like (1) tʰajA1 ‘plough’ is actually not found in any Yang Zhuang variety in Debao but is found in Jingxi Yang Zhuang varieties, instead t :ɔA1 is found in Debao; (2) the vowel of tʰ nA1 ‘stone’ is -ə in all other transcriptions (cf. Li 1970, Zhang et al. 1999, Liao 2010, Jackson et al. 2011); and (3) the vowel of ɕʰɔoɕDS1 ‘six’ should be a long - : (this long vowel rather reflects a short *-o in PT, vs. Debao -o: reflecting the long *-o: in PT) due to the Yang Zhuang sound system (cf. Zhang et al. 1999, Liao 2010). 156 irregularity of induced by aspirated sounds. First, *cr- did not always merge into *tʰin Debao of CT, but changed to *c- and then tɕ- in the term ‘plough’; this PT initial also merged into *c- and then tɕ- in Nung An and Shuangding, which are of YN. Second, PT *q- has been merged into the aspirated kʰ- in SWT/CT (= Gedney’s ST), but into the unaspirated k- in YN varieties. Thus these two initials have been separated from the common “Aspirated induced by Monosyllabic Onsets in ST and YN” (AMO-SY) to be a single sub-group named “Aspirated induced by Monosyllabic Onsets in ST” (AMO-ST) in this thesis. In other words, Liao & Shen’s (2012) 1-A row will be divided into at least two sub-rows AMO-SY and AMO-ST to designate when dealing with the ideal Tai tone box (cf. Chapter 6). There is another significant peculiarity on aspiration in Tai. This is a series of words that have aspirated initial consonants in CT and YN, but have unaspirated stops in SWT and NT. Gedney treats these certain words to be “conspicuous for tending to retain bizarre reflexes of original consonant clusters” that distinguish Li’s CT from SWT (Gedney 1989b: 64). These certain words are reconstructed to have a series of “voiceless unaspirated stops + *r-” clusters, namely *pr-, *tr-, and *kr- by Li (1977: 86, 118, 225). As we have discussed in §1.2.2 (cf. Table 11), this series of initial has been noticed by Johnson (2011: 30, 36-38) and has proposed a split in Row 2 of Gedney’s tone box. For capture the tonal split conditioned by this series of aspirated sounds in CT, Liao and Shen (2012) adopt this proposal and add this series as the newly added row 1-UC (cf. Table 13 and Table 7). However, there are different hypotheses on the reconstructions of this series of PT initials that develop to unaspirated stops in SWT and NT but aspirated stops in CT and YN. First, the split is not easily perceived due to the relative rarity of examples of this split’s conditioning factors. Johnson (2011: 30-31) lists only two examples (‘to die’ and ‘eye’) with the PT *tr- speculated by Li (1977: 118) to demonstrate this split in Dai Zhuang. Liao and Shen (2012) list two more examples (‘to dry’ and ‘break’) with the PT *pr- speculated by Li (1977: 86) to confirm their new row 1-UC. Li lists six terms under the other one *kr- in his PT reconstruction (Li 1977: 225). In addition, the wordlist designated for collecting the examples from Tai varieties investigated in this thesis contains five, two, two, three, and two examples in A1-UC, B1-UC, C1-UC, DL-UC, and DS-UC respectively (cf. Appendix C). Nevertheless, at least the term ‘cage’ reconstructed as having Li’s *kr- (cf. Li 1977: 225) and adopted into Row B1-UC of the wordlist (cf. Appendix B) in this thesis should be a wrong design after comparing more data collected. Firstly, the Thai term 157 krɔ:ŋB1 46 listed in this set reflects a long vowel *-o: in PT, but forms in other dialects like huŋB1 in Lungchow (CT), lɔŋB1 in Po-ai (NT), kʰjɔ:ŋB1 in all Yang Zhuang varieties (CT), hʊŋB1 in Daxin Baoxu (CT), ɹɔŋB1 in Longsang (NT), hjɔŋB1 in Shanglin (NT) and the cognate forms in all other CT and NT varieties all regularly reflects Li’s another initial *xr- (cf. Li 1977: 233)47 and a short vowel *-o in PT. Besides, the Thai term klɔ:ŋB1 actually does not mean ‘cage’ but ‘box’. In addition, in other SWT varieties other forms with the meaning ‘cage’ and with regular sound correspondence to Li’s *xroŋB (or Pittayawat Pittayaporn’s *kroŋB) reflected in CT and NT varieties are not found. Therefore I suggests that the Thai term klɔ:ŋB1 ‘box’ may not be cognate with this etymon. Furthermore, in Li’s the other five terms ‘sieve’, ‘near’, ‘husk of rice’, ‘to polish, sooth’, and ‘to snore’ with his *kr- (cf. Li 1977: 225), I found that only ‘near’ actually reflects his reconstruction with an unaspirated initial stop in both SWT and NT but an aspirated kʰj- or its variation ɕ- (<tɕʰ- < kʰj- < *kr- (Li 1977: 225) or *k.r- (Pittayawat Pittayaporn 2009: 169) ) in CT varieties. The other four may confirm to his other reconstructions *xr- and *kl- due to sound correspondence among modern Tai varieties. Other terms that may be related to his *kr- may be ‘to put up (like an umbrella)’ and ‘CLF of a stone’ which have been listed in the wordlist designated (cf. Appendix B), as shown in Table 52. Table 52 Reflexes of Li’s*kr- in some modern Tai varieties48 Thai (L1- Debao Urban Xiangdu Shuangding (L26- Du’an (L35- SWT) (L8-CT) (L17-CT) YN-CT/NT?) NT) ‘near’ klajC1 kʰjɔ:jC1 ɕaɯC1 kʰjoɥC1 (Chongzuo) kjaɯC1 ‘to put up’ ka:ŋA1 kʰa:ŋA1 kʰa:ŋA1 kʰa:ŋA1 ka:ŋA1 kɔ:nC1 kʰo:nC1 kʰo:nC1 kʰo:nC1 ko:nC1 Gloss ‘CLF of a stone’ In the chart above, the two terms ‘to put up (like an umbrella)’ and ‘CLF of a stone’ are still different from ‘near’ on that lacking of the glide (-l- or -j-) which is found in ‘near’ in modern forms. However, it is still possible that they share the same PT initial and each developed into different paths due to their following vowels, otherwise this PT initial is rather tentative because it is attested only by one 46 The Thai form is cited from Li (1977: 225), but there is not such a form with this meaning in modern Thai. Li’s Thai form should be an accidental form of kl :ŋB1 ‘box’ in modern Thai. 47 Pittayawat Pittayaporn (2009: 142-143) reconstructs *kr- for Li’s *xr-, and k.r- (2009: 169, 345) for Li’s *kr- (cf. the term ‘near’) respectively. 48 In Table 52, Shuangding lacks of the Tai cognate ‘near’, instead the Chinese loan kanB2 (< MC *gjənC) is used. For replacing the borrowed term, the Tai term from Chongzuo which is also of Yongnan Zhuang and is closely related to Shuangding is cited from Zhang et al. (1999: 777). 158 example ‘near’. Therefore, I presume that these three items share the same PT initial, which is reconstructed as *kr- by Li (1977: 225) to be a partner of his *pr- and *tr-, which likewise induce voiceless aspirated initials in CT and YN on one hand and voiceless unaspirated initials in SWT and NT on the other hand. As what Li’s reconstruction of this series of PT initial (*pr-, *tr-, and *kr-) presents a consistent structure, namely “voiceless unaspirated stops + *-r-” clusters. This may also consistently explain the regular sound changes in different Tai daughter languages. However, we have discussed that PT lacked of contrasting aspirated initials, and aspirated stops (on the high register) arising in CT and SWT have several sources, such as VASO-1/2 (cf. §4.3.2.1) and PT clusters containing plain stops plus a medial -*r-, namely *pr-, *tr-, *kr-, and *cr- (Pittayawat Pittayaporn 2009: 93-94) to be rejective to Li’s reconstruction of this series. Note that Pittayawat Pittayaporn’s proposal that aspiration in CT and SWT developed mainly from *pr-, *tr-, *kr-, and *cr- has been shown in Table 51 and has been agreed with in this thesis. Therefore, Li’s reconstructions *pr-, *tr-, and *kr- as another series of modern aspiration only found in CT and YN must in fact be reconstructed differently. For this series, we have actually partly discussed in §4.3.2.1 (cf. Table 45) as that Pittayawat Pittayaporn gives proposals based on sesquisyllabic onsets. These onsets are *p.r-, *p.t-, and *k.r- corresponding to Li’s *pr-, *tr-, and *kr- respectively. Pittayawat Pittayaporn’s reconstructions are more proper if focusing on the individual sound changes in different Tai varieties, because they avoid the contradiction with his *pr-, *tr-, and *kr- which account for that PT lacked of aspirations. For example, it can be explained that *p.r- merged into *p.t- then later simplified to t- in Thai and Yai, on the contrary *p.t- merged into *p.r- then later changed to pr- in Saek. In contrast, Lungchow keeps the distinct by h- and pʰj- (cf. §4.3.2.1). Extending the explanation on the aspirated results in some other CT and YN varieties such as Debao and Chongzuo, *p.r- > *p.t- > *ʰ.t- > tʰ- like in tʰa:kDL1 ‘to expose’, *p.t- > *ʰ.t- > tʰ- like in tʰa:A1 ‘eye’, and *k.r- > *kʰr- > kʰj- like in kʰjɔ:jC1 or kʰjoɥC1 ‘near’ can be processed. On the development from his *k.r- to modern forms in Tai varieties, Pittayawat Pittayaporn suggests the following two different paths as shown in Table 5349. 49 In Table 53, data of Tiandeng Xiangdu Zuojiang Zhuang (L17) and Debao Urban Yang Zhuang (L8) in this chart are collected by me. 159 Table 53 Etyma with PT *k.r- (adapted from Pittayawat Pittayaporn 2009: 169, 345) Gloss ‘to imprison’ ‘to drive away’ PT *k.raŋA SWT CT NT Thai Lungchow Xiangdu Debao Yay kʰaŋA1 haŋA1 laŋA1 kʰjaŋA1 tɕa:ŋA1 - lapDS1 kʰjapDS1 - ha:ŋB1 la:ŋB1 kʰja:ŋB1 tɕa:ŋB1 hajC1 lajC1 kʰjajC1 tɕajC1 *k.rapD kʰapDS1 ‘top for spinning’ *k.ra:ŋB kʰa:ŋB1 ‘illness, fever’ *k.rajC kʰajC1 ‘to beg’ *k.ro:A kʰɔ:A1 hɔ:A1 lo:A1 kʰjo:A1 - klajC1 kʰjaɰC1 ɕaɰC1 kʰjɔ:jC1 tɕaɰC1 (2009: 169) ‘near’ *k.raɯC (2009: 345) These reconstructions are also problematic because of the following two reasons. (1) They are structurally different on that *p.r- and *k.r- as “voiceless stop + medial” on one hand and *p.t- as “voiceless stop + voiceless stop” on the other hand, and this makes the explanation on the induced by process of aspiration in CT and YN become less consistent than Li’s reconstruction. Nonetheless, this problem can still be solved by Pittayawat Pittayaporn’s explanation on different diachronic hierarchies of initial changes, like *p.t- > *p.r- or *p.r- > *p.t- depending on different languages. (2) The explanations of the sound changes from PT *k.r- are inconsistent. Although Pittayawat Pittayaporn claims that his PT *k.r- “either became a true cluster *kr-, or simply lost the initial *k- to become *ʰr-, depending on the language” (Pittayawat Pittayaporn 2009: 170), thus *k.r- > *kr- > kʰ- in Thai, *k.r- > *ʰr- > h- in Lungchow, *k.r- > *ʰr- > l- in Xiangdu, *k.r- > *kr- > kʰj- in Debao, and *k.r> *kr- > kj- > tɕ- in Yay can be drawn. However, data from Thai, Lungchow and Xiangdu in Table 53 show that the term ‘near’ should have a different PT initials from his *k.r-, otherwise the explanation cannot account for the different paths of initial changes in Thai, Lungchow, and Xiangdu. Therefore, another possible PT *k.lwhich accounts for *k.l- > kl- in Thai, *k.l- > *k.r- > kʰr- > kʰj- in Lungchow and Debao, *k.l- > *k.r- > kʰr- > kʰj- > tɕʰ- > ɕ- in Xiangdu, and *k.l >*kl- > kj- > tɕ- in Yay is proposed in this thesis, for distinguishing *k.r- proposed by Pittayawat Pittayaporn and for reasonably explaining the aspirated initial induced for this term in CT varieties. Hence, we finally get a special series of PT sesquisyllabic onsets consisting *p.r-, *p.t- and *k.l- on explaining the aspiration only found in CT and YN, as shown in the following Table 54. 160 Table 54 The reconstruction of PT ASO-CY Proposer Li (1977: 86, 118, 225) Pittayawat Pittayaporn (2009: 170, 163, 169, 345) Suggestions in this thesis ASO-CY reconstructed *pr- *tr- *kr- *p.r- *p.t *k.r- *p.r- *p.t- *k.l- Example Tai varieties ‘to expose’ ‘eye’ ‘near’ Thai (SWT) ta:kDL1 ta:A1 klajC1 Debao Urban (L8, CT) tʰa:kDL1 tʰa:A1 kʰjɔ:jC1 Lungchow (CT) pʰja:kDL1 ha:A1 kʰjaɰC1 Chongzuo (YN, CT/NT?) tʰa:kDL1 tʰa:A1 kʰjoɥC1 Yay (NT) ta:kDL1 ta:A1 tɕaɰC1 This series of initials is called “Aspiration induced by sesquisyllabic onsets in CT and YN” (ASO-CY), for distinguishing the other series “Aspiration induced by sesquisyllabic onsets in YN” (ASO-YN), in the following discussion. This ASO-CY will replace the formulation 1-UC in the revised Tai tone box (cf. Liao & Shen 2012) because they are not treated to be real “unaspirated stop + *-r-” in PT (cf. §6.5). Different from ASO-CY induced by aspiration in both CT and YN, the other series of onsets ASO-YN only induce aspiration in the most varieties of YN. In other words, ASO-YN in CT and SWT which containing robust aspiration are not found to be aspirated, as shown in Table 5550. 50 Data from Thai, Lungchow, Long’an and Yay are mainly adopted from Pittayawat Pittayaporn (2009: 164, 171) except data of the term ‘to touch’ collected by me. Data from Debao Lüliu (L10) and Shuangding (L26) are collected by me. 161 Table 55 Etyma with ASO-YN Li Pittayawat (1977: Pittayaporn 118, 121, (2009: 164, 187) 171, 330) SWT Gloss Debao Lung- Lüliu chow temA1 tamA1 totDS1 ‘big leaf’ tɔ:ŋA1 NT Shuang- timA1 tʰimA1 rimA1 tatDS1 tɤtDS1 tʰɯtDS1 tʰəmA1 tʰɔtDS1 ratDS1 to:ŋA1 to:ŋA1 - tʰo:ŋA1 roŋA1 tʊnB1 tɯ:nB1 tʰɤnB1 tʰənB1 rɯnB1 C1 ‘to touch’ tɔ:ŋ to:ŋC1 - - tʰo:ŋC1 - ‘fart’ *k.t- YN Long’an Thai ‘full’ *tl- CT ‘wake up’ tɯ:nB1 ding Yay *thr- *t.r- ‘to break’ hakDS1 takDS1 takDS1 tʰakDS1 tʰakDS1 rakDS1 *k- *k- ‘fishbone’ ka:ŋC1 ka:ŋC1 ka:ŋC1 kʰa:ŋC1 kʰa:ŋC1 ka:ŋC1 In the chart above, data from the first set are plenty to attest the phenomenon that the proposed sesquisyllabic onset *k.t- causes aspiration only in the varieties of YN whose classification as CT or NT is to debate, as drawn in the processes *k.t- > t- in Thai, Debao and Lungchow, *k.t- > *t.r- > *ʰr- > r in Yay, but *k.t- > *t.r- > *tr> tʰ- in Long’an and Shuangding of YN. The different developing paths of PT *k.t- indicates that NT and YN share closer innovation. The second and the third sets are both only provided a single example, but the rule of sound correspondence attested by them is still effective. Pittayawat Pittayaporn gives the first two sets sesquisyllabic onsets, but leave the third one ‘fishbone’ under PT *k- in the wordlist (Pittayawat Pittayaporn 2009: 330), probably because the initials for this term in all the Tai varieties investigated by him are all unaspirated. Li (1977: 187) also put this term under his *k- due to the fact that there is no aspirated initial for this term in any Tai dialect listed. However, at least Shuangding (L26) of YN from my fieldwork, and Fusui of YN (cf. Zhang et al. 1999: 626) both have kʰa:ŋC1 for this term. In Qinzhou of YN and Lianshan Zhuang (in Guangdong) which presents similar aspirated rules with YN, the term for ‘(throat) stucked by fishbone’ is also the aspirated kʰa:ŋC1 (cf. Zhang et al. 1999: 702), and this term has probably undergone a semantic change from ‘fishbone’. Data of these YN varieties suggest that the PT initial for ‘fishbone’ has to be in fact reconstructed differently. One may argue that this term may be a post-PT loan from Chinese, namely *kaŋB (鯁) ‘fishbone’ in MC (cf. Karlgren 1954). However, all the reconstructions for this term in MC agree on the unaspirated initial stop *k-, and this hardly explain the aspirated stop *kʰreflected in the varieties in YN. Therefore, I temporarily reconstruct a sesquisyllabic onset *C̥ .k- for this term, for the possibility that the initial voiceless segment might 162 give an aspirated effect on the following *.k- when it was lost in YN. Based upon the adoption of Pittayawat Pittayaporn’s *k.t- and *t.r- for the first two sets, and my own reconstruction *C̥ .k-, the AMO-YN row is established to be added into the designation of the new revised of Tai tone box in this thesis (cf. Chapter 6), since secondary tonal splits conditioned by aspiration are robust in many YN varieties. Before finishing the discussion of secondary aspiration in this sub-section, we briefly review the remaining sets of sesquisyllabic onsets proposed by Pittayawat Pittayaporn (2009). For example, *p.q- which is commonly reflected as the aspirated initial kʰ- in SWT and CT varieties, but is reflected as kw- in Saek like kwaA1 ‘leg’, and the labio-velar feature -w- is the basis of reconstructed the initial segment *p.-. Another similar one is *p.q- which is supposed to become *q- > kʰ- in SWT/CT, p- in NT, but *ʰ.p- > pʰ- in YN varieties, like on the term ‘body air, feather’ as kʰonA1 in Thai (SWT) and Lungchow (CT), punA1 in Yay (NT), but pʰunA1, pʰɤnA1, and pʰɯnA1 in Qinzhou, South Yongning, and Fusui (all of YN) respectively (Pittayawat Pittayaporn 2009: 164-165). All these proposed sesquisyllabic onsets are similar to AMO-SY to induce aspirated initials in both ST (SWT/CT) and YN varieties, and they are named as “Aspiration induced by sesquisyllabic onsets in ST and YN” with the abbreviation ASO-SY. Therefore, they should be grouped together under the 1-A row on the Tai tone box designated by Liao and Shen (2012) when applying to the possible secondary tonal splits conditioned by aspirated sounds. In contrast, PT *c.r- is proposed to have undergone the following processes *c.r- > *cr- > tʰ- in Thai (SWT) and Lungchow (CT), *c.r- > *cr- > s- in Yay (NT), but *c.r- > *k.r- > *kr- > *kʰr> kʰj- in Debao and Jingxi and furthermore becomes tɕʰ- in Western Nung and Guangnan Nung. Therefore, the word ‘to ask’ is tʰa:mA1 in Thai and Lungchow, sa:mA1 in Yay, kʰja:mA1 in Debao and Jingxi, as well as tɕʰa:mA1 in Western Nung and Guangnan Nung (Pittayawat Pittayaporn 2009: 171). However, in YN varieties, this PT initial developed into unaspirated sounds like in NT, like *c.r- > *cr- > tɕ- in Shuangding. Another similar one is *t.h-, as in PT *t.hajC ‘to weep, to cry’ > hajC1 in CT/SWT, but tajC1 in Yay (NT) (Pittayawat Pittayaporn 2009: 184-185) and YN (agreeing with NT). One more example that confirms aspiration in CT/SWT but unaspirated in YN/NT is PT *q.s-, which presents as in the single example *q.sepD ‘centipede’ > kʰepDS1 in Thai (SWT), kʰipDS1 in Lungchow (CT), and kʰjapDS1 in Jingxi (CT), but lypDS1 in Qinzhou (YN) and θipDS1 in Wuming (NT) (Pittayawat Pittayaporn 2009: 178). Therefore, a new series named “Aspiration induced by sesquisyllabic onsets in ST” with the abbreviation ASO-ST is also established. 163 In conclusion, secondary tonal splits conditioned by aspirated sounds has to be divided into different diachronic hierarchies. The earliest hierarchy should be VASO1/2, since it indicates that the sesquisyllabic onsets “voiced obstruent + voiceless stop” which induced by Series 1 of voicing alternation between NT/YN and CT/SWT (cf. §4.3.2.1) were preserved at least at the time of CT/SWT splitting from PT. This supports that CT and SWT should be grouped into the proposed ST group soon after the post-PT period (cf. Chapter 5). Aspiration induced by “plain stop + *-r-” clusters is commonly found in CT/SWT and YN, and this indicates that they may be earliest factor that cause aspirated sounds to arise, resulting a new phonation group (AMOSY) in which other PT initials (like *q-, *x-, *χ-, *h-, *ʰw-, and *ʰr-) were allowed to merge. After that, VASO-1/2 lost their sesquisyllabic structure to merge into aspirated sounds to condition tones into the original high register in CT/SWT, but this process is not found in YN. This reminds that VASO-1/2 has earlier lost its sesquisyllabic structure to become plain stops in YN, just like what processed in NT. This point is very crucial for grouping YN with NT together on a post-PT level before they split from each other. Since aspirated sounds in YN should have secondarily developed as an areal trait, they do not support to group YN with CT/SWT (cf. Chapter 5). Other sesquisyllabic groups, namely ASO-CY, ASO-YN, ASO-SY have to be later development after the arise of aspirated sounds which is induced by the AMO-SY group. However, they have to be separated to rank into different diachronic hierarchies in different language groups or varieties due to the different tonal development as follows. (1) As an areal feature, ASO-SY were commonly merged into aspirations in both ST and YN probably due to their adjacent distribution in history. (2) After SWT split from ST due to the migration to MSEA from the Tai homeland, ASO-CY as an areal feature were merged into aspirations in both CT and YN. (3) Finally, AMO-YN independently developed in YN as presenting to be merged into aspirations. VASO-1/2, AMO-SY, AMO-ST, ASO-SY, ASO-CY, AMO-YN, and AMO-ST which result in aspirated sounds (1-A) have consistently condition tonal splits in individual languages. In other words, in spite of the different sources, aspirated sounds (1-A) in a single language always condition the same tonal behavior (cf. Chapter 6). The situation of aspirated sounds developing from different PT initial groups is summarized as in Table 56. 164 Table 56 Reflexes of aspiration involving in different modern Tai groups51 Initial groups involving Directions of merger in modern Tai groups SWT CT YN NT VASO-1/2 1-A 1-A 2 2 AMO-SY 1-A 1-A 1-A 1-U AMO-ST 1-A 1-A 1-U 1-U ASO-SY 1-A 1-A 1-A 1-U ASO-CY 1-U 1-A 1-A 1-U ASO-YN 1-U 1-U 1-A 1-U ASO-ST 1-A 1-A 1-U/1-C 1-U/1-C aspiration Different from secondary tonal splits amply found in the high register or tonal series 1, tonal splits have not been found in the low register or tonal series 2 until the data from some varieties of Guibei Zhuang (of NT, hereafter GB) are collected. In the wordlist of Suogan (see L36 in Appendix C) which is of GB (NT), several terms with original voiced initials commonly reconstructed have an unexpected high register tone (tonal series 1). In the lexical comparison in Table 57, a phenomenon of voicing alternation between Suogan (NT) and all other Tai varieties is shown. The shaded items which are expected to have PT voiced initials rather have a high register tone to reflect having PT voiceless initials in Suogan. These items normally have initial consonants developing from the proposed *mw- (vs. Li 1977: 74, Liang & Zhang 1996: 97, Pittayawat Pittayaporn 2009: 154) like in məŋA1 ‘hand’ and majC1 ‘tree’, *nr- (vs. Li 1977: 131, Liang & Zhang 1996: 97) or *C̬ .n (vs. Pittayawat Pittayaporn 2009: 177) like in namC1 ‘water’ and nɔkDS1 ‘bird’, and *g- (cf. Li 1977: 199, Pittayawat Pittayaporn 2009: 359) or *gl- (Liang & Zhang 1996: 226) like hu:B1 ‘pair’52. This phenomenon is also found in Huanjiang (also of GB) (cf. Zhang et al. 1999: 192), which is geographically closed to Suogan. 51 1-A, 1-U, 1-C and 2 in the chart are adopted from the tone box revised by Liao and Shen (2012) (cf. §1.2.1), and refer to voiceless aspirated sounds, voiceless unaspirated sounds, initials developing from proto-voiceless continuants and initials developing from proto-voiced sounds respectively. 52 The term hu:33 (B1) ‘pair’ has an unexpected initial (k- in NT and most CT/SWT but ɕʰ- in Thai (SWT), Zuozhou (CT) and Baoxu (CT)) and tone (expected to have Tone B2 in NT), and it is not certain that it is cognate with *gu:B ‘pair’ reflected in other Tai varieties. If it is indeed the cognate, the changes of the initial and the tone need more counterpart data to be determined with. 165 Table 57 Voicing alternation between Suogan of NT and other Tai varieties Initial reconstructed Li Pittayawat (1977) Pittayaporn NT CT SWT Suogan Du’an Wuming Debao Thai (L36) (L35) (L30) (L8) (L1) ‘hand’ məŋA1 fəŋA2 fɯŋA2 mʊŋA2 mɯ:A2 ‘tree’ majC1 fajC2 majC2 majC2 ma:jC2 ma:C2 Gloss (2009) *mw- *mw- *m- *m- ‘horse’ ma:C2 ma:C2 ma:C2 ma:C2 *d- *d- ‘ant’ mɔtDS2 matDS2 motDS2 mɔ:tDS2 motDS2 *nr- *C̬.n ‘water’ namC1 ɹam C2 ɣam C2 namC2 na:mC2 ‘bird’ nɔkDS1 ɹɔk55 ɣok DS2 nɔ:kDS2 nokDS2 field’ na:A2 na:A2 na:A2 na:A2 na:A2 ‘rot’ nawB2 nawB2 nawB2 nawB2 nawB2 ‘pair’ hu:B1 kowB2 kawB2 kowB2 kʰu:B2 ga:A2 ka:A2 ka:A2 ka:A2 kʰa:A2 ‘rice *n- *g- *n- *g- ‘get stuck’ In this thesis, this phenomenon is temporarily proposed to be a split from the low register in some specific NT varieties (like Suogan and Huanjiang) to merge into the high register, since these terms all reflect a PT voiced initial in the vast majority of the Tai languages. Therefore, the conditioning factors that cause the tones on these terms to fall into the high register in these specific NT varieties afford for thought. It is interesting that the first two sets *mw- and *nr- (or *C̬ .n-) present as the same to m- and n- respectively in CT/SWT, but different from f- and r- respectively in the vast majorities of NT, like fəŋA2 ‘hand’, fajC2 ‘tree’, ɹamC2 ‘water’ and ɹɔkDS2 ‘bird’ in Du’an. This series of initials reconstructed by Li and Pittayawat Pittayaporn both can hardly account for this voicing alternation between Suogan and other Tai varieties. Therefore, these sets of PT initial must be differently reconstructed. Although we do not see that this series of initials in the low register causes tonal split in any other Tai varieties out of GB (NT), we do see that the same phenomenon is commonly found in the KS languages. Geographically these GB varieties (like Suogan and Huanjiang) distribute in the area of Tai and KS border, therefore this voicing alternation may be an areal trait shared by these GB varieties and some of 166 the KS languages. For that reason, the determination of the reconstruction of this series of initials may get help from the comparison between these GB varieties and the KS languages, as shown in the following Table 58. Table 58 Voicing alternation in Tai and KS languages53 Gloss NT KS Suogan Huanjiang Wuming Kam Then Mulam Mak Sui ‘hand’ məŋA1 mɯŋA1 fɯŋA2 mja:A2 mja:A2 nja:A2 mi:A1ʰ mja:A1 ‘tree’ majC1 majC1 majC2 majC2 majC2 majC2 majC2 majC2 ‘ghost’ ma:ŋA1 ma:ŋA1 fa:ŋA2 - ma:ŋA2 - ma:ŋA1 ma:ŋA1 ‘water’ namC1 nam C1 ɣam C2 namC2 namC2 nəmC2 namC1 namC1 ‘bird’ nɔkDS1 nok55 ɣok DS2 mokDS2 nokDS2 nɔkDS2 nokDS2 nokDS2 In Table 58 above, all items with the proposed secondary tonal split from the original low register are shaded. The voicing alternation present as that Suogan and Huanjiang of NT as well as Mak and Sui of KS share similar tonal split patterns in some items, in contrast Wuming (representing the vast majority of NT) as well as Kam, Then, and Mulam of KS are supposed to preserve the original tonal behaviors of these items. Just like what he reconstructs for other series of voicing alternation among KS languages (cf. §4.3.2.1), Weera Ostapirat gives the explanation that the accented initial (like *'-m-) was indeed the origin of voicing alternation of this series of nasals in KS languages. In his explanation, the sonorant m- in this series was the original medial of *'-m-, and the drop of the accented initial gave a byproduct, namely a breath/aspiration feature which finally gave a tone in tonal series 1 (the high register tone) in Mak and Sui, in contrast other KS languages just simply dropped the accented initial and the medial -m- became the initial which gave a tone in tonal series 2 (the low register tone) (Weera Ostapirat 2006: 1113-1114). Therefore, we also simply speculate a counterpart *'-n- which has processed a similar way in different KS languages for the set of ‘water’ in Table 58. Because Li (1977), Liang and Zhang (1996), and Pittayawat Pittayaporn (2009) all do not provide any reasonable reconstruction accounting the voicing alternation of this series of nasals between Suogan/Huanjiang and other NT (as well as CT/SWT) varieties, and the viewpoint of accented initial proposed by Weera Ostapirat is limited to the application of PKS, here I gives a provisional series of sesquisyllabic 53 Data from Huanjiang are adopted from Zhang et al. (1999: 596, 615, 626, 632, 644), and data from KS languages are adopted from Weera Ostapirat (2006: 1114) and Liang & Zhang (1996: 313, 314, 319). 167 onsets “voiceless stop + nasal” such as *C̥ .N- such as *C̥ .n- and *C̥ .m- in PT for accounting of this series of voicing alternation. The initial segments of proposed *C̥ .N- are reduced in all modern Tai varieties and may have given different effects on the following nasal medial due to different languages as shown in the following Table 59. Table 59 Developing tracks of PT *C̥.N- in modern Tai varieties PT *N- (initial simply ST dropping) *ʰN- (weakening/MOA The vast majority NT Developing process *C̥.N- changing) Modern forms Tone n- and m- low (Series 2) r-/ɣ-/ɹ-/l- and f- low (Series 2) n- and m- high (Series 1) Suogan, Huanjia *N̥ - (devoicing) ng, etc. In ST (CT/SWT), it may have been just simply dropped off to make the following voiced nasal medials become simple voiced nasal initials to condition the tones into the low register, as in *C̥.n- > n-2 and *C̥.m- > m-2. In the vast majorities of NT, it may have dropped off to give a weakening effect (with an pre-aspiration) first then cause the following nasal medials to change their manner of articulation become rand f-, but the tonal behavior is preserved in the low register probably because it has been earlier conditioned by the voiced nasal medials, as in *C̥.n- > ʰn- > r-2 and *C̥.n- > ʰm- > f-2. In some GB varieties such as Suogan and Huanjiang which may have been influenced by the neighboring KS languages, it may have been dropped off and make the following voiced nasal medials become voiceless nasal initials to condition the tones into the high register, as in *C̥.n- > *n̥- > n-1 and *C̥.m- > *m̥ > m-1. The proposed tonal split within the low register in NT is then clarified to be a special series of voicing alternation, due to another specific series of sesquisyllabic onsets *C̥ .N-. This series is hereafter called “Voicing Alternation induced by Sesquisyllabic Onsets in GB” (VASO-GB) to be put into the proposed ideal Tai tone box in this thesis (cf. Chapter 6). Tonal splits in tonal languages in China and MSEA are commonly known as two-way register split, which refers to that voiceless initials give rise to a high pitch, and 168 voiced initials give rise to a lower pitch, as previously discussed (cf. §4.2). However, in some languages including some Tai varieties, the tonal splits are not just in two ways but in three ways, forming a tonal system with three registers, which is relatively rare found and less-known. Three-way splits are not single phenomenon in Tai. Lili and Jinxing, both of Wu dialects spoken in Jiangsu Province of China, have respectively split their MC tones *C/*D and MC tone *B into three tones (Chao 1928). Kam and Mak of KS have three-way splits on tones ABC and A respectively (Haudricourt 1961: 68-69). Threeway splits are also reported to be found in some of the Miao languages (Li et al. 1959). Three-way tonal splits in Tai have actually involved in the discussions of the types of secondary tonal splits in §4.3.1. In the Sinitic circles working on TK languages, a specific notation “derived tone” (as the tone makers 1', 5', 7' and 9' in Table-44)54 is used to name the third register which is considered to derived from the original odd tones (= tonal series 1) conditioned by proto-voiceless initials. Data from my fieldwork in this thesis show different schemes of three-way split in the following Tai varieties investigated. Songkhla (L2, Southern Thai) of SWT has a thorough three-way split pattern on all the ABCD tones. Jingxi Huashan (L16, Min Zhuang), Jingxi Lingding (L19, Zuojiang Zhuang), and Daxin Leiping (L22, Zuojiang Zhuang) of CT has it only in the A tone. Daxin Baoxu (L23, Zuojiang Zhuang) of CT has it in A, C and DS. Bac Va Nung (L24) of CT has it in A, B, and DL. Long’an (L27) of YN has it in B and DL. Xialeng (L29) of YN has it in A, C, and DL. Three-way register split is rare found in NT varieties. The only NT variety having been reported to have three-way register split may be Po-ai (cf. §4.3.2.2). An example of three-way split pattern from Baoxu is provided as in the following Table 60. Actually, the so-called “derived tones” (派生調) in the Sinitic circles refers to two different phenomena of splitting tones: 1) Those secondary tonal splits (on the high register) which are conditioned by one of the specific phonation types of initials (aforementioned glottalized sounds, aspirated stops, or unaspirated stops) and do not merge into any other tones on smooth syllables, like na:1 (A1-C) ‘thick’ with the “original tone” 55 vs. na:1’ (A1-G) ‘swaddling clothes’ with a “derived tone” 33 vs. na:2 (A2) ‘rice field’ with another tone 31 in Xialeng Yongnan Zhuang. 2) Those unexpected tone changes without any specific conditions on checked syllables. For example, in Xialeng Yongnan Zhuang the original DL1-A (with aspirated initials) should be a high-falling tone 53, but some items with aspirated initials have a different tone 35, and this is also called “derived tone”. In the discussions in this thesis, the “derived tones” only refers to the first situation if without additional notes. Note that Edmondson (1990) use “prime tone” which correspondences “derived tone” here in his KS languages for adopting the system of tone lables in the Sinitic circles. 54 169 Table 60 Basic tonal patterns in Daxin Baoxu Zuojiang Zhuang (cf. L23) Smooth Syllable Phonation types Protovoiceless Checked Syllable A B C DL DS Aspirations A1-A: 45 B1-A: 34 C1-A: 13ʔ DL1-A: 34 DS1-A: 13 Continuants A1-C: 45 B1-C: 34 C1-C: 13ʔ DL1-C: 34 DS1-C: 13 Unaspirated stops A1-U: 53 B1-U: 33 C1-U: 34ʔ DL1-U: 33 DS1-U: 34 A1-UC:45 B1-UC: 34 C1-UC: 13ʔ DL1-UC: 34 DS1-UC: 13 A1-G: 53 B1-G: 33 C1-G: 34ʔ DL1-G: 33 DS1-G: 34 A2: 31 B2: 33 C2: 11̰ʔ DL2: 33 DS2: 11 A1/2: 45 B1/2: 34 C1/2: 13ʔ DL1/2: 34 DS1/2: 13 Unaspirated stop + *r clusters Glottal sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds As what we have discussed in §4.3.1, three-way splits must contain a secondary split conditioned by aspirated and/or glottalized sounds (as well as unaspirated stops in some varieties). This split leads a new tone to derive from the original odd tones (= tonal series 1) which is conditioned by proto-voiceless initials. Therefore, three-way split must be considered not to be of the primary splitting pattern. However, not all secondary tonal split patterns result three-way registers. As in many other Tai varieties which have secondary tonal splits, the register is still two-way. That is, the split from the original high register has been merged into the low register, such as the tonal patterns containing secondary splits in Thai (L1), Isan (L3), Northern Thai (L4), the three Debao Yang Zhuang varieties (L8, L9, L10), Hurun (L11), Dazhai (L21), Daqiu Nung An (L28), Dongling (L32), Lizhou (L32), Gehan (L39) and Huishui Bouyei (L41). However, the following discussion will speculate a diachronic sequence beginning with three-way split and ending in a two-way split due to the limited capability of the number of tones in a language. That is to say, Tai varieties with secondary splits resulting at a two-way register may have undergone a diachronic process: primary two-way register split > secondary three-way split > secondary two-way split. Just like in Baoxu in Table 60, although the A, C, and DS columns preserve the three-way register split, the columns B and DL may have merged their derived tone into the low register to end in a two-way register split. The starting point of this hypothesis is that most secondary tonal splits conditioned by glottalized and aspirated sounds have depressed the pitch of the syllable due to 170 the laryngeal tension (Zhang 1980: 38). This point is supported by that the derived tones in Tai varieties with three-way splits are normally pitch-lower than the original tone (except those varieties having tonal flip-flop). This suggestion is inspired by the situation of three-way split in Kam. As what we have reviewed in §2.2.2, three-way split in Kam (cf. Table 24 and Table 25 in §2.2.2) is treated as the existence of register tripartition suggested by Haudricourt (1961), to demonstrate that the third register split can result from syllable initial voiceless aspiration. However, Edmondson (1990) argues that the three-way split pattern results from two temporarily distinct bipartitions rather than one simultaneous tripartition (Edmondson 1990: 188). On the basis of a proposed complex interaction among three factors (breathy voice, aspiration, and deaspiration) in Kam, he furthermore insists the real cause of secondary split to be breathy voice instead of aspiration, since aspiration is unlike to be a candidate for lowering pitch of the syllable onset because it is usually assumed to cause an increase in F0, on the other hand breathiness is often treated as pitch depressor (Edmondson 1990: 191). Thus breathy voice > aspiration > deaspiration is his process of the unmarking to form the two splits on the high register. Note that the viewpoint that PT lacked of aspiration also successfully applied to PKS (Liang & Zhang 1996). As we have supposed that aspiration in Tai arose by being induced by by *C̥ r- clusters rather than any other group of initials, breathy sounds which may be induced by *C̬ .C̥ - clusters (or VASO1/2) finally merged into aspiration through a possible breathy process (cf. §4.3.2.1 and §4.3.2.3). If Edmondson’s viewpoint that breathiness may be the real cause of pitch-lowering in Kam is true, the secondary split conditioned by aspiration in many CT and YN varieties may also have undergone a pitch-lowering process when the breathy stops merging into aspirated stops. That is, the lower pitch of the derived tone (or the prime tone by Edmondson) is brought into aspiration by breathiness when it dropped the voiced feature off to be unmarked. In addition, glottalized sounds is known as a pitch depressor, as it frequently occurring as the conditioning factor of secondary tonal splits in many Tai varieties. The three-way register split in Huashan Min Zhuang (CT) (shown in the following Table 61) provides an example for indicating the speculation on the diachronic process from three-way to two-way. In this dialect, the only three-way register split is found in the A column. Aspirations (Rows 1-A, 1-UC, and 1/2) always condition a derived tone with a lower pitch (242) than the original odd tone (353). This pitch distinction is quite slight and is easily ignored if we do not observe that the same tonal depress is also found on the same rows with aspirated initials (Rows 1-A, 1-UC, 171 and 1/2) in the DS column. Different from the slight split in Tone A, the pitchlowering in Tone DS is quite clear (a mid-falling 32 vs. a high level 55). Table 61 Basic tonal patterns in Huashan Min Zhuang (cf. L16) Smooth Syllable Phonation types Protovoiceless Checked Syllable A B C DL DS Aspirations A1-A: 242 B1-A: 32 C1-A: 11ʔ DL1-A: 32 DS1-A: 32 Continuants A1-C: 353 B1-C: 32 C1-C: 11ʔ DL1-C: 32 DS1-C: 55 Unaspirated stops A1-U: 353 B1-U: 32 C1-U: 11ʔ DL1-U: 32 DS1-U: 55 Unaspirated stop A1-UC: B1-UC: 32 C1-UC: 11ʔ DL1-UC: 32 DS1-UC: 32 + *r clusters 242 Glottal sounds A1-G: 353 B1-G: 32 C1-G: 11ʔ DL1-G: 32 DS1-G: 55 A2: 55 B2: 42 C2: 53ʔ DL2: 42 DS2: 55 Proto-voiced sounds A1/2-A: Proto-voiced aspirations & breathy sounds 242 A1/2-C: DS1/2-A: B1/2: 32 C1/2: 11ʔ DL1/2: 32 32 DS1/2-C: 353 55 It is interesting that this dialect has undergone tonal flip-flops as the original low tones are all pitch-higher than the original high tone. However, the derived tones splitting from the odd tones are still lower than the lowered odd tone after tonal flip-flop. This suggests that tonal flip-flop had established first, then secondary tonal split conditioned by aspiration happened to cause a pitch-lowering derived tone. The derived tone in the A column shows that the pitch-lowered effected by aspiration may be too slight to cause a final split, and it may merge back to the original pitch if the deriving process stops. However, the derived tone in the DS column may have forced the original high tone to become higher to merge into the original low tone which has long been high pitch since tonal flip-flop, to end in a two-way register split: derived tone vs. original odd/even tone. All these suggest tonal split are progressing. We therefore can imagine that in those Tai varieties which secondary tonal splits ending in a two-way register split may have undergone a similar process as that the derived tone splitting from the high register form a three-way register first, and then the pitch gets lower and lower, at last merge into the low register to end in a new two-way register pattern. 172 In the previous sub-sections of §4.3, we have discussed different types of tonal splits conditioned by different groups of PT initials. All these types of tonal splits are focused on smooth syllables which develop from PT tonal categories *A, *B, and *C, although they are also found on checked syllables developing from PT tonal category *D. However, tonal patterns on checked syllables in modern Tai varieties have to be divided into two tonal categories DL and DS due to vocalic length, as frequently presenting in the majorities of Tai that Tone DL goes with the same patterns of Tone B, and Tone DS goes with the same patterns of Tone C based on the long vocalic duration of *B and the short vocalic duration of *C in PT level (cf. §.4.1.3.2). In this sub-section, secondary tonal splits conditioned by vocalic length is focused on. As previously mentioned, this type of secondary tonal splits is only found in checked syllables in Tai languages (cf. §1.2.1 and §1.2.2). The following observations are perceived during the library research, fieldwork and data analyses of this thesis. First, tonal splits conditioned by vocalic length (on checked syllable) present as horizontally to contrast to those vertically conditioned by initial features. It means that Tone D splits into DL conditioned by a long vowel or a cluster of two vowels and DS conditioned by a short vowel, in contrast Tones A, B, C split into original high Tones A1, B1, C1 conditioned by voiceless initials or into original low tones A2, B2, C2 conditioned by voiced initials, or even into derived tones A1', B1', C1' secondarily conditioned by a specific sub-group of voiceless initials. Gedney’s Tai tone box (cf. Table 4 in §1.2.1) and Liao & Shen’s revised Tai tone box (cf. Table 7 in §1.2.2) has briefly and clearly drawn the outline of these two different directions of tonal splits. Second, the stability of tonal splits conditioned by vocalic length on checked syllable support the reconstruction of the vocalic length contrast in PT. Previous studies have argued on that whether there was vocalic length contrast in PT. Li (1977) posits that PT lacked of vocalic length contrast and had rich diphthongs and triphthongs, based on his argument that in many modern Tai dialects long and short vowels do not contrast except the pair of *a: and *a (Li 1977: 259). Following his proposal, Luo (1997) also considers that vocalic length contrast have secondarily developed in Tai daughter languages. In contrast, Haudricourt (1948), Nishida (1954, 1955), Wang (1966), Sarawit (1973), Pittayawat Pittayaporn (2009), and Weera Ostapirat (2013b) hold that PT contained vocalic length contrast. Sarawit’s argument which is cited by Pittayawat Pittayaporn (2009: 196-198) is pertinent on that PT short *-e 173 and *-o were banned from open syllables due to that “only long vowels were allowed in open syllable” (vs. that *-e: > -ɛ: and *-o: > -ɔ: in SWT can be in open syllable) to indicate that Li’s proposal that Thai vowels *-e, *-ɛ:, *-o, *-ɔ: have PT origins cannot account for the case (Pittayawat Pittayaporn 2009: 197-198). This thesis supports the viewpoint that there was vocalic length contrast in PT due to the stability of tonal splits conditioned by vocalic length on checked syllable. In some Tai varieties, vocalic length contrast has been lost but the tonal splits conditioned by the historical vocalic length are still preserved. In some other Tai varieties, some vowels have been reversed the vocalic length but they still condition the horizontal tonal splits on checked syllables together with their belonging original groups of historical vocalic length. The following two tables illustrates these phenomena. Table 62 Tonal splits in checked syllables in Guangnan Nong Zhuang 55 DL Tonal values Gloss 1-A=13 ‘torn’ 1-C=13 DS Tonal values Gloss kʰat13 1-A=55 ‘vegetable’ pʰak55 ‘fruit’ mak13 1-C=55 ‘heavy’ nak55 1-U=13 ‘hug’ kɔt13 1-U=55 ‘frog’ kɔp55 1-UC=13 ‘to expose’ tʰak13 1-UC=55 ‘grasshopper’ tak55 1-G=13 ‘to love’ ˀdip13 1-G=55 ‘raw’ ˀdip55 2=31 ‘outside’ nɔk31 2=33 ‘bird’ nɔk33 1/2=13 ‘to hit/correct’ sok13 1/2=55 ‘to bite’ kʰɔp55 Vocalic length contrast including *-a: vs. *-a has been totally lost in Nung Zhuang as shown in Table 62 above. However, both horizontal splits and vertical splits are straightforward to make a neat scheme of four tonal categories DL1 (13), DL2 (31), DS1 (55) and DS2 (33) in this dialect. The vertical splits are obviously conditioned by initials as what we have previously discussed. The horizontal splits lack of conditioning factors if the synchronic vowel system without length contrast is only considered. Only if the long vowels contrast to short vowels in history, can the horizontal tonal splits on checked syllables in this dialect be reasonable explained. 55 Data from Guangnan Nong Zhuang are adopted from Zhang et al. (1999: 603, 615, 619, 621, 628, 693, 699, 701, 705, 711, 729, 737, 776, 782). The terms ‘to love’ in the DL column and ‘vegetable’ in the DS column are from Yanshan Nong Zhuang, for they are able to more directly indicate the situation of the loss of vocalic length contrast in Nung Zhuang. In the columns “tonal values”, tonal categories (with abbreviated initial group names) and their tonal values are jointed by a “=” between them. 174 The situation of vocalic length conditioning horizontal tonal splits in Debao Dalong Yang Zhuang are more complicated as shown in Table 63. Table 63 Tonal splits in checked syllables in Debao Dalong Yang Zhuang 56 DL Tonal values Gloss ‘torn’ 1-A=33 1-U=45 1-UC=33 1-G=33 ‘fruit’ ma:k45 ‘fog’ ‘mouth’ ‘to pound’ 1-A=45 tʰʊk33 mo:k45 pa:k45 to:k45 ‘to expose’ tʰa:k33 ‘to love’ ˀdi:p33 1-C=45 1-U=45 ‘bamboo stripe’ tʰo:k33 1-UC=45 ‘flower’ ˀdo:k33 1-G=45 ‘to wipe’ ‘outside’ ‘blood’ ‘child’ ‘surname’ ‘knife’ ‘to remove’ 1/2=33 kʰa:t ɕo:k33 ‘bone’ 2=33 Tonal values 33 ‘measuring cup’ ‘spinning’ 1-C=45 DS ‘pale’ ‘to hit/correct’ ˀdʊk33 ‘chrysanths’ kʰjʊk45 ‘six’ kʰjɔ:k45 ‘heavy’ nak45 ‘belly’ ‘to stab’ ‘to fall’ 2=21 kjo:k33 mət33 1/2=45 pak45 tɔ:k45 tʰak45 ‘raw’ ˀdəp45 ‘to seed’ tʰɔ:k45 ‘to suck’ ˀdɔ:t45 ˀdʊk45 mat21 nɔ:k21 ‘poison’ tʊk21 ‘motar’ kjɔ:k21 ‘room’ ɹʊk21 ‘to throw’ ʋət21 ‘neat’ lət33 mɔ:k45 ‘grasshopper’ ‘bird’ lu:t33 tʰʊk33 pʰjak45 ‘a grain’ no:k33 pʰa:k33 ‘vegetable’ ‘catfish’ ma:t33 lʊk33 Gloss ‘to bite’ ‘cooked’ tɕət21 kʰap45 θʊk45 In Table 63 above, items on the same horizontal rows are supposed to have a proposed PT long vowel (under the DL column) and its counterpart PT short vowel (under the DS column) respectively. For example, the pair kʰa:t33 ‘torn’ vs. pʰjak45 ‘vegetable’ indicates a vocalic length contrast *-a: vs. *-a in PT, and the pair ˀdip33 ‘to love’ vs. ˀdəp45 ‘raw’ indicates a vocalic length contrast *-iə vs. *-i in PT, as the vocalic length contrast is still preserved in these two pairs even some vowel shapes have been changed (*-iə > -i: and *-i > -ə). However, the shaded items present an opposite situation that all these items have unexpected vowel length as what horizontal tonal splits they have conditioned. We first focus on -ɔ:- in the DS 56 This dialect is my mother tongue. 175 column. For example, nɔ:k21 ‘bird’ presents with a long vowel in Debao Dalong Yang Zhuang (and also in all other Yang Zhuang varieties), but tonally it behaves as if it belongs in the short vowel column (DS2). Actually the vowel -ɔ:- in Yang Zhuang varieties always phonetically presents as a long vowel but corresponds to a short vowel (-o- or -ɔ-) in the vast majority of Tai varieties, such as kɔ:n³¹ (Debao) vs. kʰon³³ (Thai) ‘person’, kjɔ:n53 (Debao) vs. kron³³ (Thai) ‘to snore’, and tɔ:k45 (Debao) vs. tok21 (Thai) ‘to fall’. The fact that ‘bird’ in the vast majority of Tai languages presents as a short vowel confirms that the proto-vowel of it must have been the short *-ok (cf. Pittayawat Pittayaporn 2009: 230), thus *-o- > -ɔ:- (as well as their counterpart *-o:- > -o:-) must have processed in Yang Zhuang varieties. However, in Yang Zhuang tones on the items with this long -ɔ:- on checked syllables are preserved to behave as in DS since they had been conditioned by its predecessor *-oin history. Therefore, kʰjɔ:k45 ‘six’, tʰɔ:k55 ‘to seed’, and ˀdɔ:t45 ‘to suck’ in the high register, as well as kjɔ:k21 ‘motor’ and nɔ:k21 ‘bird’ in the low register do not have the same tone as their DL counterpart 33 even they have the same vocalic length now. Comparing this with the words tʰʊk33 ‘spinning’, ˀdʊk33 ‘bone’, lʊk33 ‘child’ and tʰʊk33 ‘to hit/correct’ reflecting a short vowel -ʊ-, as well as mət33 ‘knife’ and lət33 ‘to dismantle’ reflecting the other short vowel -ə-, which presents phonetically as a short vowel but belongs to the DL column in this dialect, the proto-vowel sometimes have been overturned their length in daughter languages. These two vowels always contrast to their counterparts -ʊ- and -ə- in the DS column respectively on their different ways of tonal behavior conditioning. This suggests -ʊ- and -ə- under the DL column must have been long vowels in history. Cognates from SWT like Thai du:kDL1 ‘bone’, lu:kDL2 ‘child’ and tʰu:kDL1/2 ‘to hit/correct’ and mi:tDL2 ‘knife’ all with a long vowel conditioning their tones into DL clearly confirm this speculation. Therefore, although they have been merged into their DS counterparts to lose the vowel length contrast in Yang Zhuang, their collocating tones are still preserved to help us to restore their historical shape. Both the cases of Nong Zhuang and Yang Zhuang show us that in modern Tai varieties, words do not always retain their proto-vowel length, and the loss of vocalic length contrast must have been secondary. Moreover, tonal splits conditioned by vocalic length must also be secondary although the contrast of vocalic length was proposed to exist in PT. This is because of the following two reasons. (1) As what previously pointed out that in the majorities of Tai that tone DL frequently goes with the same patterns of tone B, 176 within the forty-two Tai varieties investigated in this thesis, thirty-six have the same or similar tonal patterns between tones developing from *DL and *B. Only the following six varieties have different tonal patterns between DL and B, as Songkhla (L2, SWT) and Bac Va Nung (L24, CT) with DL going with A, as well as Khon Kaen (L3, SWT), Longsang (L31, NT), Baipeng (L37, NT) and Gehan (39) with DL going with C. The dominant tonal merger between DL and B tones supports that a coordination between Tones *D and *B have existed at the PT level. For Tone DS, the tonal mergers between smooth and checked syllables are scattered as in that nine varieties have DS going with B, sixteen have DS going with C, and seventeen have DS going with A. The situation that tonal mergers between DS and other tonal categories lack of a dominant direction indicates that tonal splits conditioned by PT short vowels were secondary. (2) Mergers between tones on checked syllables and smooth syllables can be speculated to be on which diachronic orders due to comparison. Taking the tonal patterns of three Yang Zhuang varieties in Table 36 (in §4.1.3.3.2), Table 64, and Table 65 (also cf. L8 to L13 in Appendix C) below as examples, this problem will be readily solved. We first look at Table 36. Debao Urban (L8) have a neat merger between DL and B, and an irregular merger between DS and B. If we understand that secondary tonal splits conditioned by aspiration (Rows 1-A, 1-UC, 1/2) and glottal sounds (Row 1-G) have caused tones in these rows on the B and DL columns to fall into the low register or to merge into Row 2, we will recognize that both DL and DS had neat mergers into B in history. That is to say, short vowels on checked syllables in Debao Urban do not condition horizontal tonal splits, and the DS column is preserved to have the original straightforward tonal pattern while the DL column has undergone vertical secondary tonal splits conditioned by aspirated and glottalized initials. Debao Dalong in Table 64 present a horizontal secondary tonal split in the DS2 box, going along with C2, but all the other boxes under the DS column are preserved to have the original *D pattern going along with the original patterns in B (although vertical secondary tonal splits in this column have undergone like in Debao Urban). Jingxi Anning in Table 65 vertically represents a straightforward primary tonal split pattern, but horizontally illustrates a full tonal split conditioned by the short vowels, to cause DS to go along with C. Only if the diachronic process above was true, can the historical changes of tone on dead syllables in these three Yang Zhuang varieties be reasonably explained. It further suggests that tones DL and DS which are still coordinated with B should be treated as preserving the primary tonal pattern of *D, and Tones DL and DS which have went along with Tones A and C are of later development. 177 Table 64 Basic tonal patterns in Debao Dalong Yang Zhuang (cf. L9) Smooth Syllable Phonation types Protovoiceless Checked Syllable A B C DL DS Aspirations A1-A: 31 B1-A: 332 C1-A: 24ʔ DL1-A: 33 DS1-A: 45 Continuants A1-C: 53 B1-C: 454 C1-C: 24ʔ DL1-C: 45 DS1-C: 45 Unaspirated stops A1-U: 53 B1-U: 454 C1-U: 24ʔ DL1-U: 45 DS1-U: 45 C1-UC: 24ʔ DL1-UC: 33 Unaspirated stop + *r clusters Glottal sounds Proto-voiced sounds A1-UC: 31 B1-UC: 332 DS1-UC: 45 A1-G: 31 B1-G: 332 C1-G: 24ʔ DL1-G: 33 DS1-G: 45 A2: 31 B2: 332 C2: 213ʔ DL2: 33 DS2: 21 B1/2: 332 C1/2: 24ʔ DL1/2: 33 DS1/2: 45 Proto-voiced aspirations & A1/2-A: 31 breathy sounds A1/2-C: 53 Table 65 Basic tonal patterns in Jingxi Anning Yang Zhuang (cf. L13) Smooth Syllable Phonation types Protovoiceless Checked Syllable A B C DL DS Aspirations A1-A: 53 B1-A: 453 C1-A: 33ʔ DL1-A: 45 DS1-A: 33 Continuants A1-C: 53 B1-C: 453 C1-C: 33ʔ DL1-C: 45 DS1-C: 33 Unaspirated stops A1-U: 53 B1-U: 453 C1-U: 33ʔ DL1-U: 45 DS1-U: 33 A1-UC: 53 B1-UC: 453 C1-UC: 33ʔ DL1-UC: 45 DS1-UC: 33 A1-G: 53 B1-G: 453 C1-G: 33ʔ DL1-G: 45 DS1-G: 33 A2: 31 B2: 131 C2: 213ʔ DL2: 13 DS2: 21 A1/2: 53 B1/2: 453 C1/2: 33ʔ DL1/2: 45 DS1/2: 33 Unaspirated stop + *r clusters Glottal sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds We finally probe into a series of “vowel-length alternation between ST and NT” (DL/S), which is the last concern of tonal splits conditioned by vocalic length on checked syllable. This series of vowels present to have short length in ST (SWT/CT) on one hand and having long length in NT and YN on the other hand, as shown in Table 66. 178 Table 66 Vowel-length alternation between ST and NT/YN Tai group ST SWT Debao Gloss NT CT Thai (L1) Urban (L8) Long-sang Dongling (L31) (L32) YN Du’an Shuang- Jiaren ding (L35) (L26) Daqiu (L28) Sanxiang (L25) ‘gums’ ŋɯəkDL1 ŋy:kDL1 ŋɯ:kDL1 ŋəkDS1 həkDS1 - - - ‘deaf’ nuəkDL1 nu:kDL1 nukDS1 nokDS1 nokDS1 nokDS1 nukDS1 nokDS1 stripes’ tɔ:kDL1 tʰo:kDL1 tukDS1 tokDS1 tukDS1 tʰok DS1 tʰʊkDS1 tʰɔkDS1 ‘sunshine’ dɛ:tDL1 ˀde:tDL1 ˀditDS1 ˀditDS1 ˀditDS1 nət DS1 ˀditDS1 ˀditDS1 tʰu:kDL1 tʰʊkDL1 təkDS2 təkDS2 təkDS2 tək DS2 tikDS2 - ‘child’ lu:kDL2 lʊkDL2 ləkDS2 ləkDS2 ləkDS2 ləkDS2 likDS2 likDS2 ‘knife’ mi:tDL2 mjətDL2 mitDS2 mitDS2 mitDS2 - - - ‘bamboo ‘right/to be’ Note that in Table 66 the phonetically short vowels of tʰʊkDL1 ‘right/to be’, lʊkDL1 ‘child’ and mjətDL2 ‘knife’ in Debao actually reflect PT long vowels due to their tonal behaviors (cf. Table 63 and its related discussions). Just like what is shown in Table 63, the items designated into the wordlist under the DL column due to SWT are normally agreed on vocalic length by CT varieties, but show a regular vocalic change along with the tone change to be coordinate with the counterpart rows in the DS column in both NT and YN varieties. For example, in Longsang (L31), as those in other NT varieties, the following terms expected to be under the DL column show regular vocalic and tonal changes to coordinate with their counterpart DS rows: nuk55 (DS1-C) ‘deaf’, tuk55 (DS1-UC) ‘thin bamboo stripes’, ˀdit55 (DS1-G) ‘sunshine’, mit33 (DS2) ‘knife’, and tək33 (DS1/2-A) ‘to be’57. Shuangding (L26) of YN, whose classification as CT and NT is open to debate, does not agree on the CT/SWT varieties but agree on NT varieties to have terms nok35 (DS1-C) ‘deaf’, tʰok35 (DS1-UC) ‘thin bamboo stripes’, nət35 (DS1-G) ‘sunshine’, and tək44 (DS1/2) ‘correct’ which all reflect short vowels. Therefore, an obvious boundary between ST and NT/YN on the vocalic length reflected is outlined. This is crucial to our discussion because the However, in Longsang the term ŋɯ:ɕ21 (DL1-C) ‘gums’ agreeing with that reflecting a long vowel in the DL1-C column in all CT/SWT varieties. This may be influenced by its neighboring CT varieties due to that it distributes on the NT and CT border. 57 179 supposed secondary horizontal tonal splits conditioned by vocalic length present different splitting directions and results when meeting these kind of items. Besides, any Tai tone box already designated cannot capture this tonal distinction by its DL and DS columns. On the basis of the hypothesis that PT had vocalic length contrast, this thesis agrees on the reconstructions arguing that these items originally had long vowels from PT. For example, the regular shortening of PT high long vowel *-i:-, *-ɯ:-, and *-u:preceding stop codas are frequently found in NT (Pittayawat Pittayaporn 2009: 215, 218-219, 220-221). The environments of vowel shortening may not be limited to stop codas, such as *-ɨ- > -ɨ:- /___alveolar and *-ɨ:- > -u:-/___velar in Thai (SWT), but *-ɨ:- > -ɨ-/___alveolar and *-ɨ:- > -u-/___velar in Lungchow (CT), as well as *-ɨ:- > -ɨ/___alveolar and *-ɨ:- > -ɨ-/___velar in Po-ai (NT), thus the terms ‘night’ and ‘firewood’, ‘child’ and ‘hit the mark’ which are assumed to have this PT vowel are kʰɨ:nA2, fɨ:nA2, lu:kDL2 and tʰu:kDL1 in Thai, kʰɨnA2, fɨnA2, lukDS2 and tʰukDS1 in Lungchow, as well as hɨnA2, fɨnA2, lɨkDS2 and tɨkDS2 in Po-ai respectively (Weera Ostapirat 2013b: 194)58. That is, in some specific environments original long vowels are speculated to be shortened in some CT varieties and in NT (as well as YN) on both smooth and checked syllables. Note that Pittayawat Pittayaporn (2009: 222) reconstructs a short *-uk for the rime of ‘bone’ which is also listed in Table 64. He explains that *-uk was lengthened to *-u:k in all SWT and some CT varieties, was lengthened and lowered to *-o:k in NT, and is preserved as a short -uk in some CT varieties. This explanation can hardly account for the situation that it has a phonetically short vowel rime -ʊk but reflects a historical long vowel by its tonal behavior in Debao Dalong Yang Zhuang (cf. the discussions of Table 62). Therefore, the vowel of this term is still considered to be originally long even though we will not furthermore reconstruct every single proto-vowel in this thesis. Such kind of “vowel-length alternation” may have accomplished in early history after the split of CT/SWT and NT/YN due to the clear boundary between these two groups (as in Table 63), but some of the items show that this process may still be progressing. This can be confirmed by the comparison among SWT, NT and some Yang Zhuang varieties that are very closely related, as in the following Table 67. Note that items with both vocalic and tonal changes from DL to DS are shaded. 58 Weera Ostapirat (2013b) does not provide the tones for these items. Tones on these items are replenished by me according to my own fieldwork or library research. 180 Table 67 The progressing vowel-length alternation between ST and NT/YN Tai group Gloss ST SWT Thai (L1) Yang Zhuang (CT) Debao Debao Jingxi Jingxi Urban Dalong Urban Anning (L8) (L9) (L12) (L13) NT YN Dongling Daqiu (L32) (L28) ‘gums’ ŋɯəkDL1 ŋy:kDL1 ŋy:kDL1 ŋi:kDL1 ŋi:kDL1 ŋəkDS1 - ‘deaf’ nuəkDL1 nu:kDL1 nu:kDL1 nu:kDL1 nu:kDL1 nokDS1 nukDS1 stripes’ tɔ:kDL1 tʰo:kDL1 tʰo:kDL1 tʰo:kDL1 tʰo:kDL1 tokDS1 tʰʊkDS1 ‘sunshine’ dɛ:tDL1 ˀde:tDL1 ˀde:tDL1 ne:tDL1 ˀde:tDL1 ˀditDS1 ˀditDS1 tʰu:kDL1 tʰʊkDL1 tʰʊkDL1 tʰʊkDS1 tʰʊkDS1 təkDS2 tikDS2 ‘child’ lu:kDL2 lʊkDL2 lʊkDL2 lʊkDS2 lʊkDS2 ləkDS2 likDS2 ‘mucus’ mu:kDL2 mʊkDL2 mʊkDL2 mʊkDS2 mʊkDS2 mukDS2 mukDS2 ‘knife’ mi:tDL2 mjətDL2 mjətDL2 mitDS2 mətDS2 mitDS2 - ‘bamboo ‘right/to be’ In the chart above, four Yang Zhuang varieties present interesting progress on the “vowel-length alternation”. They all agree with Thai (SWT) on the original vowel length and the original tonal behaviors of the first four terms (from ‘gums’ to ‘sunshine’), and also agree with each other on the vowel shortening of the latter four terms (from ‘right/to be’ to ‘knife’). However, different from the original tonal behaviors are still preserved on the latter four terms in two Debao varieties, the tones of these latter four terms in the two Jingxi varieties are also changed to be coordinating with their shortened vowels, thus DL > DS agreeing with NT and YN forms. The situation in the four Yang Zhuang varieties indicates that tonal behaviors can be preserved even though the vowels are shortened, on the other hand they can also be changed along with the vowel-length shortening. The different two groups of the terms in the two Jingxi varieties also show that vowel-length shortening may be progressing in the modern Thai varieties. This is crucial to the hypothesis of vowel shortening in the Tai history. In short, in contrast to secondary tonal splits conditioned by different groups of initials to be vertical on both smooth and checked syllables (ABCD), secondary tonal splits can be conditioned by vocalic length as horizontal only on checked syllables 181 (D). Tonal splits conditioned by vocalic length often present as that the original D tone changes from the original behavior coordinating with tone B, to change to be coordinating with tone C or A. Vowel-length alternation normally presents as long vowel-length and/or its tonal behaviors preserved in SWT and CT, but vowel shortening with tonal behavior changes in NT and YN. The phenomenon of vowellength alternation suggests that an extra column DL/S should be added into the proposed ideal Tai tone box when being applied to all Tai varieties (cf. Chapter 6). In this chapter, I have proposed the diachronic processes of tonal development in Tai languages as briefly summarized as follows. 1) Tonogenesis: Four tonal categories arose after the dropping or weakening of the syllable finals *-h and *-ʔ in the earlier stage of PT, for keeping the semantic distinction among syllables with earlier plain voice (> *A), *-h (> *B) and *-ʔ (> C*) finals respectively. 2) Primary tonal splits: Within the voiceless continuants, voiceless stops, glottal sounds, voiced stops and voiced continuants in PT, the mergers between voiceless and voiced continuants caused PT tones to fall into two series of primary registers – original high tones (odd tones) and original low tones (even tones). This process displayed original voiceless continuants, voiceless stops, and glottal sounds on the high register, as well as original voiced continuants and voiced stops on the low register. An eight-tone system, namely Tones A1, A2, B1, B2, C1, C2, D1, and D2 was established. 3) Vertical secondary tonal splits: After ST split from PT, a series of “voiceless plain stop + *r-” cluster in PT gave rise to aspirated sounds (AMO-SY) in both ST and YN as an areal trait. VASO-1/2 then merged into this series in ST via a breathy process to condition odd tones, in contrast in NT and YN VASO-1/2 had earlier changed to voiced stops to condition even tones. VASO-2/1 also went to opposite directions of tonal behaviors between ST and NT/YN. After that, ST splitting from CT due to Tai migration from the Tai homeland to MSEA. ASO-CY then developed to merge into AMO-SY in both CT and YN but excluding SWT. At the same time, ASO-YN independently developed to merge into AMO-SY in YN excluding CT. In both CT and YN, aspirated sounds separating from voiceless continuant condition possible secondary tonal split. Glottalized stops commonly condition secondary tonal splits in all Tai groups. However, it is found in tones C 182 and/or B but rarely A in NT, in contrast, it is found in tones A and/or B in ST. Secondary tonal splits conditioned by unaspirated stops are only found in ST as they are treated as the byproduct of secondary tonal splits conditioned by glottalized sounds. In addition, VASO-GB induced voicing alternation between the vast majority of NT and a small cluster of Guibei Zhuang of NT (including Suogan and Huanjiang). 4) From three-way to two-way register: Derived tones splitting from the original high register cause three-way splits in some CT/SWT varieties. But in more Tai varieties, the bearing capacity of the tone numbers in a language often cause the derived tone to merge into the original low tone to make two-way splits always be the mainstream. 5) Horizontal secondary tonal splits: The tone *D on checked syllable secondarily split due to different vocalic vowel length. Those D tones do not agree with the tonal behavior as Tone*B should be treated as secondary development. Going back to the first hypothesis of this thesis, the diachronic scheme of Tai tonal development is outlined. The proposed three groups of proto-initials (voicelessvoiced-breathy) which conditioned primary tonal splits are attested to be of a misapprehensive idea, since breathy sounds were even later development which merged into aspiration only in ST (CT/SWT). However, the source of VASO-1/2 which contains a specific group of sesquisyllabic onsets must still be considered to be on the PT level. The significance of it is still on the primary level, since two different developing tracks between ST and CT/YN has been destined by it before any types of secondary tonal splits occurred. The idea that splits conditioned by other pharyngeal features, such as aspiration and pre-glottalization represent stages of secondary development is verified to be attested. In addition, some other diachronic stages and detailed issues involving Tai tonal development make the scheme fuller and more accurate. 183 Tai classification based on tonology This chapter suggests a scheme of Tai classification based on the diachronic hierarchies of Tai tonal development. §5.1 provides a tonological criterion of Tai classification. §5.2 suggests a Tai primary classification that NT and YN together form a sub-branch of which the ST sub-branch (containing Li’s CT and SWT) is a sister. §5.3 indicates that some of the secondary tonal splits can be applied to Tai secondary classifications, but others can only be treated as criteria of Tai classification on the levels of dialects or even varieties in the same dialect. §5.4 and its sub-sections provide supplementary evidence from initial behaviors, vocalic features, and exclusive lexical items to support the viewpoints of Tai classifications based on primary and secondary tonal splits. §5.5 briefly summarizes the Tai classification based on tonology. Tai classification has been briefly discussed in §1.1.2, §2.1.3, and §2.2.4. Within the three sub-sections, §2.2.4 especially focuses on the previous studies grouping Tai varieties based on tonology. Tai classification based on tonology has never been effectively suggested to subgroup Tai languages by itself, but has to be used for accounting for subgrouping Tai varieties together with other criteria. For example, in one of Li’s three criteria for grouping his NT, CT, and SWT, he points out that certain lexical items show voicing alternation reflected in their tones, such as the word ‘to be’ indicating an original voiced initial *b- in CT and NT, but an original voiceless initial *p- in SWT (Li 1977: xiii, also cf. §2.1.3). Chamberlain’s Tai classification suggested by the tonal splitting criterion together with the low register initial p- vs. ph- criterion (cf. 2.2.4) is used for arguing on grouping the sub-types of SWT as well as CT, but it seems that it is not able to be used for dividing the primary Tai groups like NT, CT and SWT. Although Edmondson (1994: 164) points out that tonal flip-flop is a characteristic of SWT (cf. 2.2.4), we have seen that there is no tonal flip-flop in some SWT varieties such as Tai Lue (L6) and Quan Son Tai (L7). On the other hand, in certain CT and NT varieties investigated here tonal flip- 184 flop does occur. These are Tuoxin Pyang Zhuang (L15) and Huashan Min Zhuang (L16) in CT, as well as Longsang (L31), Dongling (L32) and Shanglin (L34) in NT (cf. §4.1.3, §4.1.3.1, and Appendix C). Therefore, the tonal flip-flop criterion cannot be individually used for grouping Tai varieties on any level except the variety level. The only effective tonological criterion relating to Tai primary classification in previous studies may be the so-called voicing alternation or the alternation of tonal series (cf. §1.2.2, §1.2.3, §2.2.3, §2.2.4), although as previously pointed out, it has been used for dividing Tai primary groups together with phonological and lexical criteria. This tonal phenomenon has been commonly used to suggest Li’s NT to be separate from his CT and SWT (or Gedney’s ST). However, the affiliation of the varieties of YN which has the same tonal behaviors with NT on this tonal phenomenon is still open to debate. This is because some previous studies arguing that NT languages do not have aspirated stops rules out that varieties of YN are considered of NT. However, just as Nicolson suggests that “other traditional indicators of tone systems, segmental development and lexical items” can suggest Nung An to be a NT member if these indicators can be applied to Nung An data (Nicolson 2000: 293), YN varieties may possibly be of NT rather than CT. This will be further discussed in the following §5.2. Before probing into the essential discussion of Tai classification based on tonology, it is necessary for us to establish the tonological criteria based on diachronic hierarchies as shown in Figure 15. A B (v, w, x…) D (v, w-1, x-1…) C (x, y, z…) E (v, w-2, x-2…) F (x-2, y, z-1…) G (x-2, y, z-2…) Figure 15 Speculation of shared innovations on different diachronic orders for subgrouping languages In spite of the different viewpoints on tonological criteria of Tai classifications, it is very crucial that a language family must be subgrouped into different branches by using the criterion of shared innovation (cf. §2.1.3) on diachronic orders. For instance, in Figure 15 Group B and Group C splitting from the same proto-language A must have developed their own innovations like ‘v’ and ‘w’ in B and ‘y’ and ‘z’ in C, and both B and C share the same retention ‘x’. Note that these innovations are on the primary level. As time moved on, Group B split into Groups D and E, and Group C split into Groups F and G. The original innovations ‘v’ in B may be preserved in 185 both of the groups D and E, but ‘w’ and ‘x’ may have developed into w-1 and x-1 as well as w-2 and x-2 in Group D and Group E respectively. On the other hand, the original ‘y’ in C may be preserved in both of the groups F and G, but original ‘z’ may have developed into z-1 and z-2 in Group F and Group G respectively. In contrast, original ‘x’ in C may have developed into x-2 in both the groups F and G due to the areal trait shared with Group E. However, the shared innovation ‘x-2’ among E, F, and G cannot be a reason to suggest E, F and G to group together to form a subgroup, because this shared innovation is secondary when being compared with the innovations ‘v’, ‘w’, ‘y’, and ‘z’ on the primary level (in the groups B and C). From the diachronic logic, only the innovation ‘v’ shared between D and E as well as ‘w-1’ in D and ‘w-2’ in E reflected ‘w’ in B can become evidence for proving a closer genetic relation between D and E. Similarly, only the innovation ‘y’ shared between F and G as well as ‘z-1’ in F and ‘z-2’ in G reflecting ‘z’ in C can suggest a closer genetic relation between F and G. In other words, even ‘x-1’ in D is different from ‘x- 2’ in E, F and G, but both ‘x-1’ and ‘x-2’ reflect ‘x’ which is a retention (from protoA) shared by both B and C. In short, it is not logical to suggest that E is closer to F and G by the secondary changes ‘x-2’ shared among them. Therefore, in this thesis the tonological criterion of Tai classification is based on the diachronic hierarchies of Tai tonal development illustrated in Chapter 4. That is, any Tai varieties classified with each other in the same group should be considered to share more innovations complying with the diachronic orders. The more earlier shared innovations the closer genetic relations should be considered to have between/among the Tai groups. Some shared innovations between some groups fail to agree on the diachronic orders, and they may be of areal traits (or even coincidence of independent development) rather than the evidence for proving the closer genetic relation. With this criterion, Tai classification based on tonology become a consistent rule to attest the grouping of Tai varieties, and other criteria such as initial and vocalic features as well as exclusive lexical items can be treated as its supplementary evidence, since tones have arisen and have developed conditioned by consonants and vowels (cf. Chapter 4). I will demonstrate this argument by applying the criterion to the Tai classifications based on the primary tonal split and secondary tonal splits in the following §5.2 and §5.3. 186 Because tonogenesis is consistently reflected in all Tai language groups (cf. §4.1), tonogenesis which is treated as the very first step of tonal development in Tai is invalid to be applied to Tai classification. Similarly, since the tonal scheme A1, A2, B1, B2, C1, C2, D1, and D2 induced during the primary split period can be reflected in all Tai language groups like Li’s CT, SWT, and NT as well as Gedney’s NT and ST (cf. 4.2), Tai language subdivision cannot be advanced by the primary tonal split. However, we have reviewed and discussed a couple of times voicing alternation or the alternation of tonal series (cf. §1.2.2, §1.2.3, §2.2.3, §2.2.4) which indicates that VASO-1/2 cause tones to fall into the low register in NT/YN and into the high register in CT/SWT, and on the contrary VASO-2/1 cause opposite tonal directions in the relevant language groups (§4.3.2.1). Because the sesquisyllabic onsets containing the source of these two series of voicing alternation must have been on the PT level, and VASO-1/2 and VASO-2/1 resulting in the different register merger chosen by different Tai groups must have been on the primary tonal split level, Tai varieties can be grouped into two primary branches on this level. This suggestion can refer to the indication in Figure 15. If B represents the common ancestor of both NT and YN, C then represents Gedney’s ST which is the common ancestor of Li’s CT and SWT. Subsequently, D, E, F, and G represent NT, YN, CT, and SWT respectively. As the linguistic features within these groups, ‘v’ can represent the low register tones conditioned by VASO-1/2 or the high register tones conditioned by VASO-2/1 in Group B, and ‘y’ can represent the high register tones conditioned by VASO-1/2 or the low register tones conditioned by VASO-2/1 in Group C. This statement is built on the primary tonal split level and rejects any diachronically later tonal innovations (or any tonal coincidence of independent development) like ‘x-2’ (aspiration which will be discussed in §5.3) shared between YN (E) and CT (F)/SWT (G) to suggest another contradictory grouping, namely YN-CT or YN-CT-SWT constitute a subgroup of Tai. In short, according to the primary tonal split pattern involving VASO1/2 and VASO-2/1, Tai primary classification should be a two-term system which suggests Li’s CT and SWT to form a primary sub-group ST agreeing with Gedney (1989a: 231), and suggests Li’s NT and the debatable YN to constitute the other primary sub-group of Tai. It is noteworthy that both VASO-1/2 and VASO-2/1 are put under secondary tonal splits in Chapter 4. It is because the diachronic stage of the merger from VASO-1/2 187 to aspirated sounds in ST must have been secondary since aspirated sounds arose secondarily on post-PT level. However, the VASO-1/2 and VASO-2/1 themselves must have been differentiated in PNT and PST on the primary sub-branch level, otherwise their directions of merger would not be different like in the current situation. There is another similar series of secondary tonal splits having been differentiated on primary PNT and PST level, and it also can indicate the two-term system on this primary Tai sub-branch level. It is the so-called “vowel-length alternation” involving secondary tonal splits between NT-YN and ST (cf. §4.3.3). Although the horizontal tonal splits conditioned by this “vowel-length alternation” are actually of secondary level, the earlier forms of this series of “vowel-length alternation” are neatly reflected as long in ST but as short in NT-YN (cf. §4.3.3). That is, the vocalic differentiation of the earlier forms of “vowel-length alternation” must have processed on the primary PNT and PST level. As a result, YN varieties such as Cao Lan (cf. Gregerson & Edmondson 1998) and Nung An (cf. Nicolson 2000; Edmondson 2002), as well as Yongning Baiji (L25), Nanning Shuangding (L26), Long’an Xiaolin (L27), Jingxi Daqiu Nung An (L28) and Yongning Xialeng (L29) investigated in this thesis should be all viewed as members of the NT subgroup. This statement simply declares to solve the puzzle of Cao Lan (Gregerson & Edmondson 1998), and to clarify that Nung An is neither an aberrant son of CT nor a CT stepchild from NT quizzed by Nicolson (2000), but a virtual NT member. In fact, Edmondson (2002) has pointed out that Nung An located in Cao Bang Province of North Vietnam has brought “NT features with them when they immigrated from further north in an area located on the border between NT and CT areas” around Long’an of Guangxi (Edmondson 2002: 60-61). In this thesis I suggest to follow the term YN to include these virtual NT varieties, instead of directly calling them NT varieties because of the following three reasons. (1) In the Sinitic circles linguists have been used to using YN (Yongnan Zhuang) which literally means “the Zhuang in the areas south to Yongjiang River” to include these Tai varieties with debatable features, and it is not necessary to break this tradition. (2) YN varieties geographically distribute in the southeastern part of the Zhuang area (cf. Appendix D), connected to the south and to the east of the usual line of division between NT and CT languages, and so it is misleading to use the term “northern” for these varieties. (3) The most crucial reason is that these virtual NT varieties are close to each other by their own shared innovations like abundant aspirations and their conditioning secondary tonal splits (cf. §5.3) which separate them from the common NT varieties. 188 Based upon the naming scheme above, the aforementioned terms PNT and PST (cf. 1.2.3) are used to name the proto-languages of NT-YN and ST (CT-SWT) respectively. That is, if D, E, F, and G in Figure 15 represent NT, YN, CT, and SWT respectively, B and C on the primary level represent PNT (the common ancestor of NT and YN) and PST (C the common ancestor of CT and SWT) respectively. In some sense, this thesis agrees on the two-term system which is suggested by Haudricourt (1956) and divides PT first into Dioi or ʔyai (corresponding to NT) vs. Tai proper on the primary level, because the autonym “Tai” (<*dajA) commonly found in SWT and CT varieties has not been found in languages developing from PNT, and the autonym “Yay” (<*ˀjajC) (or Dioi in the French/Vietnamese circles) also has not been found in languages developing from PST. As discussed in §4.3 and its sub-sections, secondary tonal splits are found to be conditioned by glottalized sounds, aspirated sounds (including their non-aspirated variants like ɕ- (<tɕʰ- <kʰj- < *kr-), r- (< ʰr- < *kr-), or l- (< ʰl- < ʰr- < *kr-)), and unaspirated stops in different Tai varieties. Different types of secondary tonal splits are determined to be able to become indicators for subgrouping Tai languages nor not as below. A secondary split conditioned by glottalized sounds is normally found in the B and C tones in NT varieties, but in the A and B tones in ST (cf. §4.3.2.2). However, this cannot be a valid criterion to determine the affiliation of a Tai variety because certain exceptions are found. For example, this split is found to exist in the A column in Po-ai of NT (cf. §4.3.2.2), and in the C column in Isan (L3) and in Phuan (cf. Phinnarat Akharawatthanakun 2010: 76) of SWT. Besides, secondary split conditioned by glottal sounds in the C column is normally found in the western part of NT areas (including Bouyei areas), but is not found in the eastern part of NT areas and the areas of YN which has been grouped with NT on the primary level. The distribution indicates an areal trait more than a genetic feature. Therefore, it is not tenable to use secondary split conditioned by glottalized sounds to determine the Tai affiliation on primary level. A secondary split conditioned by unaspirated stops is only found in Gedney’s ST (including CT and SWT) varieties, such as Bangkok Thai (L1), Songkhla Southern Thai (L2), Khon Kaen Isan (L3), Leiping Zuojiang Zhuang (L22) and Baoxu Zuojiang 189 Zhuang (L23) (cf. §4.3.1). It seems to be an effective indicator for separating ST from NT because NT varieties have not been reported to have this split. However, the determination must be careful because unaspirated stops never condition tonal splits by itself but always together with glottalized initials. A secondary tonal split conditioned by unaspirated sounds should be a byproduct of those conditioned by glottal sounds (cf. §4.3.1 and §4.3.2.2). In many SWT varieties and most CT varieties, unaspirated stops do not condition secondary tonal splits at all. Therefore, it is appropriate to treat it as a dialectal feature on variety level. The last type of secondary tonal split is conditioned by aspirated sounds. In §4.3.2.3 the complicated situations of aspirated sounds in Tai languages are discussed. Note that aspiration developing from VASO-1/2 in ST must be ignored here since it has been proved to be on the primary level to separate ST and NT-YN from each other. Because we have emphasized that aspirated sounds in Tai languages must be secondary on post-PT level (§4.2.1), secondary tonal splits conditioned by these aspirated sounds must be a subordinate indicator for subgrouping Tai languages when being compared with VASO-1/2. Again taking symbols in Figure 15 to illustrate the diachronic orders, ‘x-1’ of Group D can refer to the lack of aspirated sounds in NT, and ‘x-2’ shared in E, F, and G can refer to the robust aspiration shared in YN, CT, and SWT due to their close neighboring distribution in history. Thus, that the lack of aspirated sounds in NT and the robust aspirations in YN, CT and SWT cannot prove YN to be closer to CT/SWT and to be apart from NT, since ‘x1’ and ‘x-2’ on the secondary level cannot be used to refute the ‘v’ and ‘y’ on the primary level. However, the real situation of aspirated sounds in Tai languages are much more complicated than in Figure 15 because the aspirated sounds develop from different sources in PT, and that it is aspirated in one group does not mean that it must also be aspirated in another group. Therefore, the segmental development involving aspirations should be separated into different series to be discussed as to whether they can be a valid indicator for Tai classification and on which level. These series are AMO-SY, ASO-SY, AMO-ST, ASO-ST, ASO-CY, and ASO-YN (cf. §4.3.2.3) except VASO-1/2. In Table 56 we have summarized different directions of merger involving aspirations in different modern Tai groups. Aspirations developing from AMO-SY and ASO-SY are shared between YN and ST (CT-SWT), and this is very possibly of areal traits between YN and ST before the ancestors of SWT migrated from the Tai homeland. This determination is very 190 crucial because aspirations developing from AMO-SY and ASO-SY cannot challenge VASO-1/2 on the primary sub-group level to suggest YN-ST to compose a sub-group under Tai. Tonal splits conditioned by these two series are suggested to be an indicator not more than dialectal or variant level when being applied to ST (CTSWT), but to be the indicator of the YN sub-group when being applied to NT-YN. Aspirations developing from AMO-ST and ASO-ST are shared between CT and SWT, on the other hand YN and NT both reflect unaspirated sounds for these two series, agreeing with the suggestion of VASO-1/2 on the primary level. This phenomenon on the contrary indicates that secondary tonal splits conditioned by these two series should be lifted to be with those conditioned by VASO-1/2, VASO-2/1 and “vowel- length alternation” on the primary level. Nevertheless, they will be put in further discussion in the future because the examples of these two series are still fewer than other series. The remaining two series ASO-CY and ASO-YN are important indicators on secondary sub-group level. Aspirations developing from ASO-CY are only found in CT and YN, but are never found in SWT (as well as NT). This indicates that this series of sesquisyllabic onsets must have been preserved in PNT and PST, and must have differentiated during the period that YN separating from NT-YN and SWT separating from ST. The differentiation should be that it is simply merged into unaspirated stops in SWT and NT due to individual development (they do not seem to have shared areal features on post-PT level), but merged into aspirated sounds in CT and YN as an areal trait after the migration of SWT to MSEA. Thus, ASO-CY as well as secondary tonal splits conditioned by it should be treated as important indicators of dividing SWT from CT, and YN from NT on secondary sub-group level. The series ASO-YN only develops into aspirated in YN but into unaspirated stops in SWT, CT, and NT. This individual development suggests that ASO-CY as well as secondary tonal splits conditioned by it be treated as the exclusive indicators of YN on secondary sub-group level. To summarize, secondary tonal splits as well as their conditioning initials can be indicators of grouping Tai languages on secondary sub-group, dialectal and variant levels according to different groups, languages, and dialects. Comparing with Tai classification based on the primary tonal split which consistently divides the neat two term system NT-YN and ST, Tai classification based on secondary tonal splits are dispersive, because secondary tonal splits are not always found in all varieties in a specific group, a language, and even a dialect (cf. tonal patterns of different Yang 191 Zhuang varieties in Table 36 in §4.1.3.3.2 and Tables 64 and 65 in §4.3.3). However, some specific series of aspirated sounds involving possible tonal splits can be very important indicators to separate CT from SWT (by ASO-CY), as well as YN from NT (by AMO-SY, ASO-SY, and ASO-YN) on secondary sub-group level. For modern Tai languages, neither Li’s tripartite division (NT, CT, and SWT) (Li 1977) nor Gedney’s two plus one taxonomy (CT/SWT + NT) can reasonably capture all the initials and vowels involving tonal distinct within them. A more possible “two plus two taxonomy” of modern Tai division, namely NT/YN + CT/SWT is led out as a result of the discussion of secondary tonal splits and their conditioning initials. “Two term system” of primary Tai sub-groups and “two plus two taxonomy” of modern Tai division based on tonology can be supported by the non-tonal evidence from initial and vocalic behaviors as well as exclusive lexical items as in the following three sub-sections. Some specific series of segments especially the so-called single r- segment in modern NT Tai languages regularly present a reflex of *r- as the distinct feature separating them from CT-SWT. The single r- segment, which is often treated as a specific initial segment occurring with both even and odd numeral tones in NT languages to distinguish NZ/Bouyei (NT) from SZ/Tay-Nung (CT) in the Sinitic circles, is given to support the Tai division based on tonology. The segment r- includes the merger or preservation of some proto-initials listed by Gregerson and Edmondson (1998: 158) as *r-, *l- and Li’s proto-clusters *dl-, *hr-, *nl/r-, *tʰr- in modern Tai languages. Traditionally this r- segment is very often used to determine whether a Tai language belongs to NT, because all Li’s proto-initials listed above have merged into this single r- or its variants, like *ɣ- in Wuming Shuangqiao (L30) or *ð- in Giay (cf. Gregerson & Edmondson 1998). For example, as shown in Table 68, Giay, Cao Lan, Debao, and Bangkok Thai represent NT, YN, CT, and SWT respectively. The seven proto-initials *dl-, *dr-, *hr-, *hw-, *nl/r-, *tʰr- and *kʰr- have all merged to the r- segment (which presents as ðdeveloping from *r-) in Giay (NT). Five of them (*dl-, *dr-, *hr-, *hw-, and *kʰr-) 192 have merged to the r- segment (which presents as l- developing from *r-) but two (*nl/r- and *tʰr-) are differently reflected in Cao Lan (YN), showing more similarities with Giay (NT) and less similarities with Debao (CT) and Bangkok Thai (SWT). Only *dr- goes with *r- in Thai, showing fewer similarities with Giay (NT) and Cao Lan (YN). No other initials merge into ɹ- except *r- → ɹ- in Debao, showing the fewest similarities with Giay (NT) and Cao Lan (YN), but the highest similarities with Bangkok Thai (SWT). Table 68 Different initial behaviors of the single r- segment among NT, YN, CT and SWT59 Proto-initial (Li 1977) Gloss NT YN CT SWT Giay Cao Lan Debao Bangkok Thai *r- ‘house’ ða:nA2 la:nA2 ɹu:nA2 rɯənA2 *dl- ‘wind’ ðumA2 lomA2 lamA2 lomA2 *dr- ‘boat’ ðuaA2 lu:A2 ly: A2 rɯəA2 *hr- ‘to bark’ ðawB1 lauB1 hawB1 hawB1 *hw- ‘comb’ ðojA1 lojA1 ʋejA1 wi:A1 *nl/r- ‘water’ ðamC2 nom4C namC2 na:mC2 *tʰr- ‘to carry’ ða:pDL1 tʰa:pDL1 tʰa:pDL1 ha:pDL1 *kʰr- ‘to find’ ða:A1 la:A1 kʰja:A1 ha:A1 Obviously, in this case Giay presents a most typical NT feature, and Debao and Bangkok Thai presents typical ST feature, while Cao Lan goes with NT rather than ST. However, Cao Lan still keeps its own characteristics due to the two different sets which do not merge into the r- segment. Debao and Bangkok Thai also disagree with each other on the set *dr-. Initial behaviors clearly support the following two points. (1) NT and YN on one hand and CT and SWT on the other hand should be grouped into two primary sub-groups NT-YN and ST under Tai. (2) To some extent, YN and NT differ from each other, and CT also differs from SWT on secondary sub-group level under the primary groups NT-YN and ST respectively. The irregular correspondence of vowels between NT-YN and ST is not limited to the so-called “vowel-length alternation” in checked syllables (cf. §4.3.3 and §5.3), but is 59 Data from Giay and Cao Lan are adapted from Gregerson & Edmondson (1998: 158-159), and data from Debao and Bangkok Thai are collected by me. 193 also found in other aspects. For example, in Lungchow (CT) both ŋu:A2 ‘snake’ and mu:A1 ‘pig’ reflect to the same proto-vowel *u:, and so do other ST (both CT and SWT). However, in NT such as Giay, ŋɯ:A2 ‘snake’ and mu:A1 ‘pig’ reflect two different proto-vowels like *ɯ: and *u:. We also see the data of Nung An, Cao Lan, and other varieties of YN go with NT in this case. More examples are summarized to be compared as in Table 69, in which Yay, Long’an, Debao Urban, and Bangkok Thai represent NT, YN, CT, and SWT respectively. Table 69 Vowels involving Gedney’s Puzzles60 NT Proto-vowel (Pittayawat Pittayaporn Gloss Yay 2009) YN CT SWT Long’an Debao Bangkok (L27) Urban (L8) Thai (L1) *-i: ‘year’ pi:A1 pɯjA1 pejA1 pi:A1 *-ɯ: ‘book’ θɯ:A1 ɬɯjA1 ɬøyA1 (naŋA1) sɯ:A1 *-u: ‘pig’ mu:A1 mowA1 mowA1 mu:A1 *-aj ‘egg’ rajB1 ʰlajB1 kʰjajB1 kʰajB1 *-aɰ ‘leaf’ baɰA1 majA1 ˀbɔ:jA1 bajA1 *-aw ‘old’ kawB1 kawB1 kawB1 kawB1 *-ɤj ‘fire’ fi:A2 (>*i:) fɯjA2 (>*i:) fajA2 (>*aj) fajA2 (>*aj) *-ɤɰ ‘breath’ sɯ:A1 (>*ɯ:) ɕɯjA1 (>*ɯ:) tɕɔ:jA1 (>*aɰ) tɕajA1 (>*aɰ) *-ɤw ‘empty’ pju:B1 (>*u:) plowB1 (>*u:) pjawB1 (>*aw) pla:wB1 (>*aw) *-ɯj ‘long’ rajA2 (>*aj) ʰlajA2 (>*aj) rejA2 (>*i:) ri:A2 (>*i:) *-ɯw ‘crab’ pawA1 (>*aw) pawA1 (>*aw) powA1 (>*u:) pu:A1 (>*u:) In this table, the first six sets refer to those items unequivocally reflecting PT vowels *-i:, *-ɯ:, *-u:, *-aj, *-aɰ, and *-aw, and the later five sets (in shaded) refer to those different reflexes of vowels in different Tai groups. The irregular vowel correspondence in this five sets among Tai groups is called “Gedney’s Puzzle” and has been discussed in detail by Gedney (1972). Pittayawat Pittayaporn (2009: 198208) has cleared “Gedney’s Puzzle” up and has given a solution in the proto-vowels reconstructed (in shaded). In his new reconstructions *-ɤj, *-ɤɰ, *-ɤw, *-ɯj, and *-ɯw developed to merge into different PT vowels (the first six sets) in different language groups. Again, Long’an (YN) goes with Yay (NT) on one hand, Debao (CT) This chart is summarized from the solution for Gedney’s Puzzle (Pittayawat Pittayaporn 2009: 198-208). Data from Yay are adapted from Pittayawat Pittayaporn (2009: 199-202). Data from Long’an are from Zhang et al. (1999: 602, 605, 612, 622, 628, 683, 695, 769, 784). Data from Debao and Thai are collected by me. 60 194 goes with Thai (SWT) on the other hand, to support the two-term (NT-YN + CTSWT) system of the primary Tai division. Just as what Pittayawat Pittayaporn points out that YN varieties like Qinzhou, Yongnan, Fusui, and Shangsi merged PT *-ɯ:# with *-i:# reflexed by their own modern forms -uj, -ej, -ɯj, and -o:y respectively (Pittayawat Pittayaporn 2009: 216), YN varieties have their own shared innovations in vocalic behaviors which separate them from the common NT varieties. CT varieties such as Yang Zhuang varieties of Debao (L8-10) and Jingxi (L11-13), as well as Debao Nalong Myang Zhuang (L14) and Tiandeng Xiangdu (L17) also diphthongize those original high vowels *-i:#, *ɯ:#, and *-u:# in open rimes into -ej, -øy (or >ej)/-əɰ, and -ow respectively. These phenomena are not found in SWT. All these also confirm the modern Tai “two plus two taxonomy” division based on tonology. Languages developing from the same root should share numerous cognate items. However, some lexical items found in some daughter languages are not found in other daughter languages due to lexical shifts, semantic changes and/or aberrant development. This kind of lexical items may become diagnostic indicators to divide daughter languages into different groups. Although Li (1977: xiv) points out that the risks of using lexical items to classify dialects is that words often wander across dialects line chiefly in the border dialects, the exclusive items still can be treated as effective if a large numbers of lexical items in Tai varieties from the vast Tai areas can be concluded. Edmondson ( 1994: 152-153) cites from Zhuangyu Jianzhi to provide some of these diagnostic items separating NZ and SZ as in Table 70. Table 70 Some diagnostic items of NZ and SZ division (adapted from Edmondson 1994: 152-153)61 Gloss Northern Southern ‘sky’ bɯn¹ fa:⁴ na:m6 tum1 ‘cow’ ɕɯ:2 mo:2 ‘butterfly’ buŋ5 ba:3 kap7 fɯ:4 ‘dry ground’ ‘tiger’ kuk⁷ ɬɯ:¹ Edmondson’s original data should be wrong in the following items. The initial of ŋa:m6 ‘dry ground’, the initial of ŋo:5 maj4 ‘bamboo shoot’, and the rime coda of pa:ŋ6 ‘flax’ are modified as in Table 70. 61 195 Gloss Northern Southern ‘dragonfly’ pi:ŋ2 pej6 fi:4 ‘wing’ pik⁷ fɯ:t⁸ ‘black’ lap̞⁷ kaw1 ‘horn’ ‘bamboo shoot’ ‘flax’ ‘head’ ‘cloth’ dam¹ ko:k7 ɣa:ŋ2 no:5 maj4 da:j3 pa:n5 kjaw³ hu:¹/baw³ paŋ pʰa:j3 2 ‘clothing’ ‘black’ ‘blanket’ pu:6 ɬɯ:3 lap7 dam1 teŋ2 fa:2 Similarly, Zhang and colleagues (1999) have also provided other testing items in Table 7162. Table 71 Some diagnostic lexical items of ST and NT-YN (adapted from Zhang et al. 1999: 9-10, also cf. Table 15 in §2.1.3) ST Exclusive NT-YN SWT items CT YN NT Long’an Fusui mɔ:n² mɯn2 lo:n2 du:n¹ du:n¹ pa:n⁵ pa:n⁵ na:j3 na:j3 da:i³ da:i³ sɯə³ ɬɯ:³ θy:³ po:6 pu:6 pu:⁶ pɯə⁶ huə¹ huə¹ hu:¹ tʰu:¹ ʰlaw3 law3 kjaw³ tɕaw³ ‘above’63 nɯə¹ nɯə¹ nɯ:¹ ny:¹ kɯn2 nɯ:1 kɯn² kɯn² ‘below’ ta:j3 taj3 taɰ3 tɔ:j3 la:3 tɯj3 la:3 la:3 ‘to do’ -- het⁷ hit⁷ hat⁷ kuk8 kuk10 kwak⁸ kuə⁶ ‘ghost’ pʰi:¹ pʰi:¹ phi:¹ phej¹ ma:ŋ2 ma:ŋ2 fa:ŋ² fa:ŋ² ‘he/she’ man² -- min² te:¹ te:5 ti:1' te:¹ te:¹ Thai Lao Longzhou Jingxi ‘rounded’ mon² mon² mon² ‘ramie’ pa:n⁵ pa:n⁵ ‘cloth’ sɯə³ ‘head’ Liujiang Bouyei The original chart in Zhang et al. (1999: 9-10) does not provide the data of Long’an and Fusui. Data from these two YN varieties are adapted from the Appendix 2 in Zhang et al. (1999: 603, 632, 639, 664, 687, 739, 770, 800). 63 The cognate for ‘above’ in CT means ‘north’ in Thai and Lao. 62 196 Just like what is shown in Edmondson’s wordlist above, in Table 71 YN varieties Long’an and Fusui clearly go together with NT varieties Liujiang and Bouyei on most lexical items (in shaded). Only very rare testing items in Long’an and Fusui are of ST terms which are probably of areal traits due to their shared areas of distribution. On the other hand, CT and SWT varieties again go together with each other when being applied by these exclusive lexical items, except very few NT items probably due to the areal traits like te:1 ‘he/she’ in Jingxi64. Lexical items can be easily borrowed from neighboring languages, as in Li’s warning of the risk of using lexical items, and since some YN varieties like Cao Lan and Nung An located in the areas dominated by the majority CT groups in Southwestern Guangxi and Northern Vietnam, they may borrow more ST diagnostic lexical items from these typical Central Tai languages. This does lead a risk that Cao Lan and Nung An may use more ST diagnostic lexical items than other YN languages. As one of the puzzles in Cao Lan, the mix of NT and ST diagnostic lexical items has been discussed by Gregerson and Edmondson (1998: 161). In conclusion, YN varieties should not be treated as a language in that is “half way” between NT and CT because they do show more NT features rather than CT features even in lexical items. After determining the exclusive lexical items, the primary NTYN and ST division based on tonology should be a reasonable establishment. In conclusion, different from the hypothesis that “two plus one taxonomy” of Tai classification is more reasonable than the Tripartite Division at least in the early stage of Tai languages in Chapter 1, this thesis suggests two levels of Tai division as follows. First, a two-term system, NT-YN (<PNT) and ST (<PST), based on the primary tonal splits is suggested by tonal features of VASO-1/2, VASO-2/1, and horizontal tonal splits conditioned by “vowel-length alternation”. It agrees on the two-term system which is suggested by Haudricourt (1956) and divides PT first into Dioi or ʔyai vs. Tai proper. Second, this thesis further suggests a new “Two plus two” (NT/YN + CT/SWT) taxonomy of modern Tai division based on secondary tonal splits, namely ASO-CY separating CT from SWT, as well as AMO-SY, ASO-SY, and ASO-YN together separating YN from NT on secondary sub-group level. In other This item may be of OC or MC loan, cf. Mandarin tʰaA ‘he/she’. Some other OC loans in Tai also lost the original aspiration in OC since there might be no aspiration in PT during the borrowing period, cf. Bangkok Thai t etDS1 ‘seven’ and Debao Yang Zhuang t atDS1 ‘seven’ vs. Cantonese tʃʰatD1 ‘seven’ and Mandarin t ʰi:A (<D) ‘seven’. 64 197 words, it proposes that Li’s CT and SWT can be divided from each other under the ST sub-group, and NT and YN differ from one another under the NT-YN sub-group. Finally, Tai classification based on tonology is diachronically and genealogically illustrated as in the following Figure 16 and Figure 17 respectively. PT PNT NT PST YN CT SWT Figure 16 Tai diachronic division based on tonology Tai NT-YN NT ST YN CT SWT Figure 17 Tai family tree on the primary and secondary levels 198 A new perspective of Tai tone box This chapter suggests a new opinion on Tai tone box based on the hypotheses of diachronic tonal development in Tai. According to the diachronic process of Tai tonal development in different Tai groups, a revised tone box which must reflect diachronic sequences of Tai tonal splits is supposed to be more applicable for analyzing maximal Tai varieties except those situations involving aberrant development. However, it is not necessary to apply a multi-purpose Tai tone box to all the Tai languages according to the actuality that there are not all types of tonal splits in a specific Tai group. Therefore, as efficient research tools different tone boxes are suggested to different diachronic periods and different Tai sub-groups. The criterion of designating a Tai tone box will be established in §6.1. §6.2 provides a Tai tone box which indicates the splits from PT to PNT and PST. §6.3 suggests three Tai tone boxes which can be applied to the development from PST to SWT and CT respectively. §6.4 suggests two Tai tone boxes which can be applied to the modern varieties of NT and YN respectively. On the basis of these sections, §6.5 offers an integrated Tai tone box which is supposed to be able to capture all tonal distinctions in any Tai varieties. §6.6 points out the limitation of Tai tone boxes. §6.7 briefly summarizes the method of Tai tone boxes. When comparing Li’s primary tonal split pattern due to voicing mergers (Table 1), Gedney’s tone box (Table 5), and the revised tone box suggested by Liao and Shen (Table 7), the carrying capacity of capturing the tonal distinctions can be applied to more and more Tai varieties (cf. §1.2.1, §1.2.2). It is obviously observed that Tai varieties preserving the primary tonal split pattern can apply with Li’s simple tonal split pattern in Table 1. These languages include the following Tai varieties investigated in this thesis: Kho Lam Tai Yai (L5), Rong Maet Tai Lue (L6), and Quan Son Tai (L7) of SWT, Jingxi Urban Yang Zhuang (L12), Jingxi Anning Yang Zhuang (L13), Myang Zhuang (L14), Pyang Zhuang (L15), Daxin Naling Zuojiang Zhuang (L18), and Xiaoguangnan Nong Zhuang (L20) of CT, Baiji (L25) and Shuangding (L26) of YN, as well as Shuangqiao Yongbei Zhuang (L30), Longsang (L31), Shanglin 199 Hongshuihe Zhuang (L34), Du’an Hongshuihe Zhuang (L35), Baipeng Liujiang Zhuang (L37), Donglan Guibei Zhuang (L38), Xiaosanjiang Lianshan Zhuang (L40) and Zhenning Bouyei (L42) of NT. However, some ST languages such as Bangkok Thai (L1), Songkhla Southern Thai (L2), Khon Kaen Isan (L3), and Chiang Rai Northern Thai (L4) of SWT, as well as Jingxi Hurun Yang Zhuang (L11), Daxin Leiping Zuojiang Zhuang (L22), and Baoxu Zuojiang Zhuang (L23) of CT have to apply with Gedney’s tone box for indicating their secondary tonal splits due to the four different initial groups at time of tonal splits, namely voiceless frictions, voiceless unaspirated stops, glottalized sounds, and voiced sounds (cf. §1.2.1). The following CT varieties investigated in this thesis cannot be applied with Gedney’s tone box due to a tonal split between voiceless aspirated sounds and voiceless continuants (both included in Gedney’s voiceless frictions). These are Debao Urban Yang Zhuang (L8), Debao Dalong (L9), and Debao L̈liu (L10), Jingxi Huashan (L16), Jingxi Lingding (L19), Dazhai Dai Zhuang (L21), and Bac Va Nung (24). Besides, voicing alternations because of VASO-1/2 between ST and NT-YN are not answerable for by Gedney’s tone box. For solving these problems, the revised tone box suggested by Liao and Shen (2012) has been provided (cf. §1.2.2). When dealing with more details of tonal splits involving more complicated initial and vocalic changes, none of the schemes above including Gedney’s tone box and Liao & Shen’s revised version can capture these tonal complexities. These complicated tonal distinctions include but are not limited to the following series of tonal conditioning factors: (1) The specific series of sesquisyllabic onset ASO-YN which induces aspiration only in YN varieties but induces unaspirated initials in all other Tai varieties (cf. §4.3.2.3); (2) The specific series of sesquisyllabic onset VASOGB which causes voicing alternation between GB and all other Tai varieties (cf. §4.3.2.4); (3) The series of vowel-length alternation conditioning the alternation of tonal Series DL and Series DS between NT/YN and ST varieties (cf. §4.3.3). That is to say, when dealing with YN varieties which have secondary tonal splits involving aspirated sounds like Long’an Xiaolin (L27), Jingxi Daqiu Nung An (L28) and Yongning Xialeng (L29), as well as Yizhou Suogan (of NT, L36) which have voicing alternation (caused by VASO-GB) with other Tai varieties, the available Tai tonal schemes or Tai tone box have shortcomings for capturing their particular tonal distinctions. To fix up these kinds of tonal puzzles by a multi-purpose Tai tone box has become one of the motivations of exploring the Tai tonal development. This is also a hypothesis of this thesis (cf. §1.2.4). 200 From the improvement solving relevant problems in previous studies, we can speculate that it is possible to provide a complex Tai tone box which is able to capture the tonal distinctions in any Tai varieties if all types of tonal splits in Tai languages are studied. However, as we have supported the argument that PT lacked contrasting aspirated sounds which secondarily arose to condition secondary tonal splits in some CT and YN varieties, a horizontal row in a Tai tone box must be reconsidered to be designated from a specific synchronic initial group at time of tonal splits rather than from a supposed diachronic PT initial group. The crucial point is that secondary tonal splits must have been individually conditioned at a post-PT period in different Tai varieties. Therefore, it is inappropriate that any added row in a Tai tone box is supposed to refer to an initial group at the PT level. The most proper formulation should be “initial groups at time of tonal splits” suggested by Gedney (1989[1972]). If we consider the different situations of secondary tonal splits in different Tai sub-groups, namely NT, YN, CT, and SWT, a multi-purpose complex Tai tone box seems to become a burdensome research tool. For example, secondary tonal splits conditioned by aspirated sounds and unaspirated stops are never found in NT varieties due to the fact that NT varieties lack contrastive aspirated sounds and that unaspirated stops never go with glottalized initials to condition secondary tonal splits. In the case of NT, a tone box designated with the divisions containing the rows of aspirated sounds and unaspirated stops appear to be unnecessary. Similarly, secondary tonal splits conditioned by aspirated stops are never found in SWT varieties even though they have robust aspirated stops. This fact led Gedney’s to put aspirated stops and voiceless continuants together to constitute the row “voiceless frictions” which condition possible tonal splits in SWT, and putting voiceless unaspirated stops and ASO-CY (which merge into voiceless unaspirated stops in SWT) together to be another single row. Gedney’s Tai Tone box has been proved to be suitable for SWT varieties, although it has been proved to be not able to capture all the tonal distinctions in CT and YN. Therefore, in practice a well-suited Tai tone box should be designated for a specific sub-group of Tai, rather than for all Tai varieties. For that reason, ASO-CY developing into aspirated initials in CT and YN varieties may be designated as being a component in Voiceless Aspirated Sounds in the tone boxes of CT and YN, but ASO-CY developing into unaspirated stops in SWT varieties may be designated as being a constituent part in Voiceless Unaspirated Stops in the SWT tone box. All aspirated sounds, voiceless continuants and voiceless aspirated stops in SWT, CT, and YN should be put into a single group Voiceless Sounds in the 201 NT tone box since there is no more secondary split from it. In addition, the row “proto-breathy/voiced aspirations & breathy sounds” added in the Tai tone revised by Liao and Shen (cf. Table 7) or VASO-1/2 suggested in this thesis will be simply deleted from a specific Tai sub-group tone box. This is because different Tai tone boxes will be separately designated for SWT, CT, NT, and YN, and voicing alternations must be ignored when talking about a single sub-group of Tai. In brief, although a Tai tone box can be designed as a multi-purpose research tool applied to all types of tonal splits in any Tai varieties, it may become cumbersome because the complete splitting divisions based on the most complicated tonal split patterns in some of the Tai varieties will not be necessary for those having simple tonal split patterns. It is not necessary to always use a multi-purpose Tai tone box if the research scope does not cover varieties from all sub-groups of Tai. Tai tone boxes designated with specific pertinence to different Tai periods and different Tai subgroups may make the tonal solution easier and more convenient. Based on the principle discussed in §6.1 that Tai tone box can be designated according to the actuality and practicality of different diachronic Tai periods and Tai sub-groups, the first attempt of designing a PT tone box based on the tonogenesis hypothesis, and a Tai tone box based on the primary tonal split is firstly discussed in this section. Li’s primary tonal split pattern (Table 1) indicates that tones were induced by voicing mergers at post-PT period, but does not provide the phonation types of initial groups at the PT level and post-PT level. For illustrating this, a PT tone box is designed as in Table 72 based on the determination of PT initial phonation types (cf. §4.2.1). Table 72 PT tone box: from tonal precursor to tones Syllable Finals / Phonation Types Monosyllabic & Sesquisyllabic Initial Series A Primary Subordinate Proto- Continuant voiceless Plosive sounds Glottalized Proto-voiced Plosive sounds Continuant -# B C -h -ʔ (→ -# / -ˀ) (→ -ˀ/ -#) 202 D -p /-t / -k In this PT tone box, proto-voiceless sounds are divided into three phonation types, namely continuant, plosive, and glottalized, and proto-voiced sounds are divided into two phonation types, namely plosive and continuant respectively. The main purpose of this PT tone box is to indicate the tonogenesis process from the non-tonal structure on the earlier stage of PT to the tonal structure on the later stage of PT. Note that tones on the later period of PT have not split to the voiceless-voiced registers since the voicing of voiceless continuants and the devoicing of voiced plosives have not yet taken place. Therefore, the two primary initial groups (voiceless and voiced sounds) and the five subordinate initials groups only indicate the potential effects on the tones at the coming register splitting period on the postPT level. At one of the specific periods of post-PT, tonal register splits were triggered by the merger of proto-voiceless continuants and proto-voiced continuants due to the voicing of proto-voiceless continuants (cf. §4.2.2). This process has been repeatedly mentioned as the primary tonal split, which presented as the straightforward register pattern inherited in about half of modern Zhuang varieties (Zhang et al. 1999: 245), and in twenty Tai varieties within the forty-two ones investigated in this study (cf. 4.2.3). As an extended version of Li’s two tonal series, an integrated PST and PNT tone box designed based on this primary tonal split pattern is as shown in Table 73. In this process, PT tones have split into two series 1 and 2 according to the high and the low registers respectively. Table 73 Integrated PST & PNT tone box: from PT to PST (→) and PNT (→) Monosyllabic & Sesquisyllabic Initial Series Tonal Categories A B C D A1 B1 C1 D1 VASO-1/2 A1/2 B1/2 C1/2 D1/2 VASO-2/1 A2/1 B2/1 C2/1 D2/1 A2 B2 C2 D2 Primary Subordinate Continuant Proto-voiceless sounds Plosive Glottalized Proto-voiced Plosive sounds Continuant The most remarkable tonal phenomenon in this Tai tone box is that two series of sesquisyllabic onsets have parted their two different trends, which induced voicing alternations between PST and PNT in the later period, especially after aspiration 203 arose in PST (cf. §4.3.2.1). VASO-1/2 resulted to merge into voiceless sounds in PST, but into voiced sounds in PNT, in contrast VASO-2/1 became voiced sounds in PST, but became voiceless sounds in PNT. After the designations of the two Tai tone boxes on the PT and the post-PT levels, the next concern becomes the tone boxes of the single Tai sub-groups after the postPT level. In this section the tone boxes of the ST group are designed. First, the simple PST tone box extracted from the intergrated PST & PNT tone box (cf. Table 73) is shown in Table 74. Table 74 PST tone box Tonal Categories Subordinate initial Primary groups conditioning secondary tonal splits A B C A1 B1 A2 B2 D DL DS C1 DL1 DS1 C2 DL2 DL2 Frictions Proto-voiceless/ VASO-1/2 Unaspirated stops Glottal Sounds Proto-voiced sounds/VASO-2/1 Plosives + Continuants In this PST tone box, series VASO-1/2 and VASO-2/1 have been put together with voiceless and voiced sounds respectively since voicing alternation is not necessary to take into account without PNT varieties. The subordinate initial groups in the high register are treated to have the potential conditionings of secondary tonal splits in the two daughter groups of PST, namely CT and SWT. Aspiration in the high register may have arisen on the PST level due to the uniform aspirated data shared by its daughter languages CT and SWT. But aspiration is put together with voiceless continuants to be under voiceless frictions due to the fact that they need not be separate to condition different tonal behaviors which is still not found in modern SWT languages. Actually there is even no evidence for proving that secondary tonal splits have occurred at the PST level. Therefore, the three subordinate initial groups in the high register are still treated as of uniform proto-voiceless sounds to condition tones into Series 1. Because tonal splits within the low register are never found in ST languages, it is not necessary to divide proto-voiced sounds to plosives and continuants anymore, even though there are also different phonation types in the low register, like unaspirated plosives (in most Tai varieties), aspirated plosive in 204 some SWT varieties including Thai (L1) and Isan (L3), as well as some CT varieties including Lingding (L19), Baoxu (L23), and Leiping (L22), and sonorants. Therefore, Series 2 in Tai tone boxes hereafter refers to tones conditioned by initials developing from proto-voiced plosives or continuants at time of tonal splits. The very first practical Tai tone box serving for a modern Tai sub-group SWT is Gedney’s tone box (cf. Table 4), which is adapted here and is revised by filling in the terminology in this thesis as in Table 75. It is almost identical with the PST tone box (cf. Table 74) except the horizontal divisions of the three sub-groups of voiceless sounds, because they have been found to condition possible secondary tonal splits in SWT languages. Table 75 SWT tone box (cf. Gedney 1989[1972]) Tonal Categories Subordinate initial Primary groups conditioning A B C Frictions A1-F B1-F Unaspirated stops A1-U Glottal sounds Plosives + Continuants secondary tonal splits D DL DS C1-F DL1-F DS1-F B1-U C1-U DL1-U DS1-U A1-G B1-G C1-G DL1-G DS1-G A2 B2 C2 DL2 DS2 Protovoiceless (1) Proto-voiced sounds (2) For indicating the situations of initial mergers in SWT before the time of tonal splits, an ancillary box of the SWT tone box is shown as in Table 76. Table 76 Ancillary box of the SWT tone box Initial Directions of particular series of initial mergers in modern SWT varieties groups merged in VASO-1/2 1-F x VASO- AMO- AMO- ASO- ASO- ASO- ASO- 2/1 SY ST SY CY YN ST x x x 1-U 2 x x x x This Ancillary box can help to clarify the different results of initial mergers which may result in different horizontal rows to condition different possible tonal splits when comparing with other modern Tai sub-groups. For example, ASO-CY becomes aspirated sounds in CT and YN to condition a possible tonal split together with other 205 aspirated sounds from other sources, but becomes unaspirated stops in SWT to condition a possible tonal split together with those common unaspirated stops developing from proto-voiceless unaspirated stops. In the column of ASO-CY in the ancillary box of the SWT tone box, the merger into the row of 1-U (voiceless unaspirated stops) is selected by a “x”. Under this condition, items with proto-initial of Series ASO-CY will be put into the 1-U row in the SWT tone box. Therefore, the word ‘eye’ is ta:A1-U in SWT even though it is tʰa:A1-A, ha:A1-A or pʰja:A1-A in CT and tʰa:A1-A in YN. Note that the 1-G row is not included in the ancillary box, since it is not found that any specific series of initials would regularly merge into glottalized sounds to condition possible tonal splits in any modern Tai varieties. Similarly, the simple CT tone box and its ancillary box are designed in Tables 77 and 78 respectively. Table 77 CT tone box Tonal Categories Subordinate initial Primary groups conditioning A B C Aspirated A1-A B1-A Continuant A1-C Unaspirated stops secondary tonal splits DL DS C1-A DL1-A DS1-A B1-C C1-C DL1-C DS1-C A1-U B1-U C1-U DL1-U DS1-U Glottal sounds A1-G B1-G C1-G DL1-G DS1-G Plosives + Continuants A2 B2 C2 DL2 DS2 Frictions Protovoiceless (1) Proto-voiced sounds (2) D Table 78 Ancillary box of the CT tone box Initial groups Directions of particular series of initial mergers in modern CT varieties VASO- VASO- AMO- AMO- ASO- ASO- ASO- ASO- merged in 1/2 2/1 SY ST SY CY YN ST 1-A x x x x x 1-C x 1-U 2 x x x The most noticeable point which differs the CT tone box from the SWT tone box is the split between aspirations and continuants (cf. §1.2.2) when being compared with 206 the SWT tone box. It means that the situation of initial mergers and tonal splits in CT varieties are more complicated than in SWT. Another different point between the CT tone box and the SWT tone box is the direction of initial merger of the Series ASO-CY, which merges into aspirated initials (1-A) in CT but merges into unaspirated stops (1-U) in SWT, as shown in the CT ancillary box in Table 78. For example, the VASO-1/2 column in the ancillary box is selected by a “x” both in the rows 1-A and 1-C. It means that initials developing from voicing alternation Series 1/2 in CT are merged into aspirated sounds in some items, but are merged into continuants in other items. This can be shown in the example in Debao Dalong Yang Zhuang (L9) in where a tonal split between 1-A (voiceless aspirated sounds) and 1-C (voiceless continuant) in the A, B, and DL columns as in Table 79. Table 79 Different VASO-1/2 merging directions involving different tonal splits among Tai varieties Gloss Bangkok Thai Debao Dalong Wuming (NT) (SWT) (CT) ‘ear’ hu:24 (A1-F) kʰjow31 (A1-A) ɣɯ:31 (A2) ‘bean’ tʰuə21 (B1-F) tʰu:331 (B1-A) tu:33 (B2) ‘CLF for person’ pʰu:41 (C1-F) pʰow24 (C1-A) pow41 (C2) ‘to hit the mark’ tʰu:k21 (DL1-F) tʰʊk33 (DL1-A) tɯk33 (DS2) ‘to bite’ kʰop21 (DS1-F) kʰap55 (DS1-A) hap33 (DS2) ‘right side’ kʰwa:24 (A1-F) θɔ:53 (A1-C) kwa:31 (A2) ‘late in morning’ sa:j24 (A1-F) θɔ:j53 (A1-C) kwa:j31 (A2) ‘ten’ sip21 (DS1-F) θəp55 (DS1-C) ɕip33 (DS2) ‘to mince’ sap21 (DS1-F) θap55 (DS1-C) θap33 (DS2) Bangkok Thai (L1), Debao Dalong (L9) and Wuming (L30) represent SWT, CT and NT respectively for indicating voicing alternation between ST and NT. The shaded items of Debao Dalong all have a continuant initial θ-, contrasting with the other items all with aspirated initial stops. In contrast, although Bangkok Thai also agrees with Debao Dalong on aspirations and continuants in most items, it is not necessary for it to separate the 1-A and 1-C rows from 1-F because a tonal split between aspirated and continuant sounds is not found in SWT. 207 Having the diachronic Tai tone boxes from PST to SWT and CT for reference, tone boxes designed from PNT to NT, YN, and GB are easily understood. In this section the tone boxes of the NT group are described. First, the simple PNT tone box extracted from the integrated PST & PNT tone box (cf. Table 73) is shown in Table 80. In this PNT tone box, series VASO-1/2 and VASO-2/1 have also been put together with voiceless and voiced sounds respectively since voicing alternation is not necessary to take into account without PST as a reference substance. Different from the PST tone box, only two sub-groups of voiceless sounds are treated to have the potential conditionings of secondary tonal splits in the daughter groups of PNT, namely NT and YN. This indicates there is no tonal splits between voiceless plosives and continuants in modern NT-YN varieties, except aspiration secondarily developed in YN probably due to areal feature shared with ST. Table 80 PNT tone box Tonal Categories Subordinate initial Primary groups conditioning secondary tonal splits Proto-voiceless Plosives + Continuants sounds/VASO-2/1 Glottal Sounds Proto-voiced sounds/VASO-1/2 Plosives + Continuants A B C A1 B1 A2 B2 D DL DS C1 DL1 DS1 C2 DL2 DS2 As the direct modern successor of PNT, modern NT languages can be applied with the simple NT tone box which is the most concise Tai tone box as shown in Table 81. Only two sub-groups of voiceless sounds are designate. Common voiceless sounds and glottalized sounds are labeled as 1 and 1-G respectively. Note that contrasting aspirations are not found in NT. It is almost identical with the PNT tone box (cf. Table 80) except the thorough horizontal divisions of the two sub-groups of voiceless sounds, because they have been found to condition possible secondary tonal splits in NT languages. Different from the situation of SWT and CT, Series VASO-GB is added to be a new column in the ancillary boxes (shown in Table 82) for the NT tone box. This is because of VASO-GB which is the voicing alternation involving the low register between the common NT varieties and some of the BG varieties (cf. §4.3.2.4). 208 Table 81 NT tone box Tonal Categories Subordinate initial Primary groups conditioning A B C Plosives + Continuants A1 B1 Glottal sounds A1-G Plosives + Continuants A2 secondary tonal splits Proto- D DL DS C1 DL1 DS1 B1-G C1-G DL1-G DS1-G B2 C2 DL2 DS2 voiceless sounds (1) Proto-voiced sounds (2) Table 82 Ancillary box of the NT tone box Initial Directions of particular series of initial mergers in modern NT varieties groups merged VASO- VASO- AMO- AMO- ASO- ASO- ASO- ASO- VASO- in 1/2 2/1 SY ST SY CY YN ST GB x x x x x x x x 1 2 x x In the ancillary box of the NT tone box, VASO-GB is checked in both tonal series 1 and 2. This is because in some GB varieties like Yizhou Suogan (L36), the merging direction of VASO-GB is opposite to the vast majorities of NT (cf. §4.3.2.4). When carrying a fieldwork research to the NT areas, if this series of initial is checked in tonal series 1, the Tai variety should be of GB. If it is checked in tonal series 2, the Tai variety should be treated as a common member of NT. Nevertheless, the ancillary boxes of the NT tone box are simplest to show the least initial merging directions (voiceless 1 and voiced 2). In the vast majorities of NT varieties (including most GB varieties), all specific series of initial groups merge into voiceless sounds to condition a high register tone except VASO-1/2 which merge into voiced sounds to condition a low register tone. In a few GB varieties, such as Yizhou Suogan (L36) and Huanjiang Urban (cf. Zhang et al. 1999: 192), all specific series of initial groups have the same behaviors with all the other NT varieties, except VASO-GB which merge into voiceless sounds to condition a high register tone. Comparing with the NT tone box, the YN tone box and its ancillary box shown in Tables 83 and 84 respectively is more complicated because the secondary aspirations conditioning possible secondary tonal splits in some YN varieties. 209 Table 83 YN tone box Tonal Categories Subordinate initial Primary groups conditioning A B C A1-A B1-A C1-A secondary tonal splits Aspirated Proto-voiceless Unaspirated stops + (1) Continuants Proto-voiced sounds (2) D DL DS DL1-A DS1-A A1-U/C B1- U/C C1- U/C DL1- U/C DS1- U/C Glottal sounds A1-G B1-G C1-G DL1-G DS1-G Plosives + Continuants A2 B2 C2 DL2 DS2 Table 84 Ancillary box of the YN tone box Initial Directions of particular series of initial mergers in modern YN varieties groups merged VASO- VASO- AMO- AMO- ASO- ASO- ASO- ASO- VAS in 1/2 2/1 SY ST SY CY YN ST O-GB x x x 1-A x 1-U 2 x x x x x The YN tone box is similar to the CT tone box (cf. Table 77) because of the shared aspiration which may condition possible tonal split in both YN and CT. However, YN lacks of split between voiceless continuants and voiceless unaspirated sounds. It is not necessary for YN varieties to separate 1-C and 1-U like in the CT tone box. Because aspirated and unaspirated sounds are a contrasting pair, in the YN tone box 1-U is used instead of 1-C to emphasize this contrast. For YN, the most remarkable initial merging direction is ASO-YN resulting in aspiration, and this is not found in CT and SWT even though they both have robust aspirated sounds (cf. §4.3.2.3). In the previous sections, we have discussed all the possibilities of designing different tone boxes for different Tai sub-groups. However, as we have predicted that an integrated or unified Tai tone box can capture all tonal distinctions in any Tai varieties, it is useful when the affiliation of the Tai variety investigated is unknown. At all events, this integrated Tai tone box and its ancillary box are shown as in Tables 85 and 86 on the basis of the different Tai tone boxes described in the preceding sections and the previous discussions in Chapter 4. 210 Table 85 Integrated Tai tone box Primary Tonal Categories Phonation types at initial groups time of tonal splits A B C Aspirated sounds A1-A B1-A Continuant sounds A1-C Unaspirated stops DL DS C1-A DL1-A DS1-A B1-C C1-C DL1-C DS1-C A1-U B1-U C1-U DL1-U DS1-U Glottal sounds A1-G B1-G C1-G DL1-G DS1-G Plosives + Continuants A2 B2 C2 DL2 DS2 Protovoiceless sounds (1) Proto-voiced sounds (2) D Table 86 Ancillary box of the integrated Tai tone box 1-A 1 U/C VASO-GB ASO-ST ASO-YN ASO-CY ASO-SY 1-A 1-F 1- CT AMO-ST SWT AMO-SY YN VASO-2/1 NT Directions of mergers in any Tai variety VASO-1/2 Initial groups merged in 1-C 1-U 2 Merging direction of DL/S in modern Tai varieties Vowel-length alternation DL DS DL/S The design principle of this integrated Tai tone box is that initials at the time of tonal splits are maximally divided into five phonation groups in modern Tai varieties, namely voiceless aspirated sounds (1-A), voiceless continuants (1-C), voiceless unaspirated stops (1-U), glottalized sounds (1-G), and voiced sounds (2), because tonal split have been found to be maximally conditioned by these five phonation groups in Tai languages, diachronically and synchronically. The integrated Tai tone box is almost identical to the CT tone box, because the most complete tonal split patterns have been found in some CT varieties, like Debao Yang Zhuang varieties, Baoxu, and Bac Va Nung. The integrated Tai tone box is also similar to the revised Tai tone box suggested by Liao and Shen (2012), but is different from it by deleting the unnecessary rows 1-UC and 1/2 in Liao & Shen’s 211 version. The reason is that particular series of initial groups are not limited to 1-UC (ASO-CY) or 1/2, but are at least nine groups as shown in the ancillary box of the intergrated Tai tone box in Table 86. The supplementary ancillary box in Table 86 is much more facilitative than nine rows of particular series of initials and one column of vowel-length alternation (DL/S) added in the integrated tone box. This is because only the five initial phonation types in the main box can condition possible tonal splits in Tai due to the research of this thesis (cf. Chapter 4). If these nine initial groups as extra-rows and the column of vowel length alternation on checked syllable are all added, the integrated Tai tone box will become very unwieldy. Note that “1-G” is not shown in the ancillary box because there are no any particular series of initials to merge into glottal sounds to condition possible tonal split in any Tai variety. Before attesting the tonal behaviors in a specific Tai variety, the ancillary box can first be used for figuring out the merging directions of the specific series of initial groups in that variety, by finally reducing them to the five groups conditioning. It can also deal with the situation of vowel-length alternation on checked syllables (DL/S) between ST and NT-YN (cf. §4.3.3). For example, in NT varieties like Debao Longsang (L31), the following lexical items designated into the DL column in the wordlist for this thesis show a regular vocalic change (shortened) to be coordinate with the counterpart DS column: nuk55 (DS1-C) ‘deaf’, tuk55 (DS1-UC), ‘thin bamboo stripes’, ˀdit55 (DS1-G) ‘sunshine’, mit33 (DS2) ‘knife’, and tək33 (DS1/2-A) ‘to hit the mark’. For testing the vowel-length changes of these items in a specific Tai variety, this series of items can be listed under the ancillary box to investigate. As for Longsang, these items would be ticked in the DS column. As for most SWT varieties, these items would be simply ticked in the DL column since they all reflect a long vowel in SWT varieties. In some CT varieties like Debao Dalong (L9), these items would also be ticked in the DL column even though their vowels have been shortened, because the tones are still preserved as those in the DL column in this variety (cf. §4.3.3). A notable benefit of the design of the ancillary box is that we may find more particular initial and/or vocalic changes which may condition secondary tonal splits in some undescribed Tai varieties in further studies. These potential new findings will be added into the ancillary boxes, but the main box will not need to be modified because it should have reached its complete function by adding the maximal rows of tonal-conditioning phonation types which condition possible tonal splits. For illustrating the design of the Tai tone boxes, Debao Dalong Yang Zhuang (L9) of CT is taken here as an example for application, as shown in Tables 87, 88 and 89. 212 Table 87 Integrated Tai tone box applying to Debao Dalong (cf. Table 64) Tonal Categories Phonation types Primary at time of tonal initial groups A B C Aspirated sounds 31 332 Proto-voiceless Continuant sounds 53 sounds (1) Unaspirated stops Glottal sounds splits Proto-voiced Plosives + sounds (2) Continuants D DL DS 24 33 45 454 24 45 45 53 454 24 45 45 31 332 24 33 45 31 332 213 33 21 Table 88 The merging direction of particular series in Debao Dalong 1 1-U/C x x 1-U 2 VASO-GB x x ASO-ST 1-C x ASO-YN x ASO-CY 1-A ASO-SY 1-F AMO-ST 1-A CT SWT AMO-SY YN VASO-2/1 NT Directions of mergers in any Tai variety VASO-1/2 Initial groups merged in x x x x Merging direction of DL/S in modern Tai varieties Vowel-length alternation DL DL/S x DS The basic tonal patterns in Debao Dalong Yang Zhuang can be compared with the one applied by Liao & Shen’s revised tone box (cf. Table 64 or tonal patterns of L9 in Appendix C). The comparison shows that the tonal patterns in the integrated Tai tone box (Table 87) are more concise than those in Liao & Shen’s version. This is because the two extra rows in Liao & Shen’s version, 1-UC and 1/2 which correspond to ASO-CY and VASO-1/2 respectively in the ancillary box, are actually merged into voiceless aspirated sounds (and partially into voiceless continuants) to condition the same tonal split pattern, and they are redundant to be shown when the investigation is only focused on this variety. In addition, if the investigation is expanded to the comparison with other Tai varieties, these two extra rows are not really enough. At least another seven rows of particular initials and one column of vowel-length alternation on checked syllable are needed to compare with since they potentially have different merging directions into the five initial groups of phonation types which condition possible tonal splits. 213 Table 89 Examples of merging of initials and vowels in Debao Dalong Checklist of particular series kʰjow31 (>A1-A) ‘ear’ θɔ:53 (>A1-C) ‘right side’ VASO-1/2 tʰu:33 (>B1-A) ‘bean’ kʰaw24 (>C1-A) ‘rice’ tʰʊk33 (>DL1-A) ‘to hit the mark’ θəp45 (>DS1-C) ‘ten’ tɕa:j31 (>A2) ‘man’ VASO-2/1 ma: 31 (>A2) ‘to come’ təŋ33 (>B2) ‘to listen’ ta:w33 (>B2) ‘to overturn’ pʰja:31 (>A1-A) ‘cliff’ kʰjaj33 (>B1-A) ‘egg’ AMO-SY pʰja:j24 (>C1-A) ‘to walk’ pʰja:k33 (>DL1-A) ‘forehead’ pʰjak45 (>DS1-A) ‘vegetable’ kʰe:n31 (>A1-A) ‘arm’ haw33 (>B1-A) ‘to bark’ AMO-ST kʰa:24 (>C1-A) ‘to kill’ kʰa:t33 (>DL1-A) ‘torn’ tʰat45 (>DS1-A) ‘to jump’ kʰɔ:n31 (>A1-A) ‘body air’ ASO-SY kʰa:j31 (>A1-A) ‘to sell’ kʰjɔ:j24 (>C1-A) ‘near’ tʰak45 (>DS1-A) ‘grasshopper’ tʰa:31 (>A1-A) ‘eye’ tʰaw33 (>B1-A) ‘to hunt’ ASO-CY kʰjɔ:j24 (>C1-A) ‘near’ tʰa:k33 (>DL1-A) ‘to bask’ tʰak45 (>DS1-A) ‘grasshopper’ tam53 (>A1-U) ‘full’ tʊn55 (>B1-U) ‘to wake up’ ASO-YN to:ŋ24 (>C1-U) ‘to touch’ tak33 (>DS1-U) ‘to break’ tat45 (>DS1-U) ‘fart’ kʰja:m31 (>DL1-A) ‘to ask’ ASO-ST haj24 (>C1-A) ‘to cry’ kʰjap45 (>DL1-A) ‘centipede’ mʊŋ31 (>A2) ‘hand’ kow33 (>B2) ‘pair’ VASO-GB maj213 (>C2) ‘tree’ nam213 (>C2) ‘water’ nɔ:k21 (>DS2) ‘bird’ 214 Checklist of particular series nu:k55 (>DL1-C) ‘deaf’ DL/S tʰo:k33 (>DL1-A) ‘thin bamboo stripes’ ˀde:t55 (>DL1-G) ‘sunshine’ mət33 (>DL2) ‘knife’ tʰʊk33 (>DL1-A) ‘to hit the mark’ The checklist under the ancillary box (Table 89) provides some of the examples of these particular series in Debao Dalong Yang Zhuang. In the practical fieldwork research, the checklist of main tone box can be designated together with the consideration of ancillary box if the research covers Tai varieties from different Tai sub-groups, namely NT, YN, CT, and SWT. Before entering the summary section to conclude the method of Tai tone box, it is necessary to point out the limitations of the Tai tone box. The most noticeable limitation is that aberrant tonal development can neither be resolved in a practical Tai sub-group tone box, nor be explained in an integrated Tai tone box which can be able to capture the tonal distinctions in any Tai varieties. This kind of aberrant tonal development normally includes tonal changes due to areal traits, phonological contamination, and conditioning factors change before/after the time of tonal splits. For example, different from the merger into DS2 in other NT varieties, ɬap55 ‘to mince’ merges its tone with the rows of DS1 in Debao Longsang (L31, NT), to present a ST behavior. This may be of an areal trait shared with the neighboring the Debao Yang Zhuang varieties, which are of CT. Another example is the term ˀdaŋ45 ‘to make fire’ in Daxin Baoxu (L23, CT). Comparing with the cognate ˀdaŋC1-G in Debao Longsang (L31, NT) and Debao Urban Yang Zhuang (L8, CT), it has an unexpected Tone A1-G (45), which is expected to be C1-G (34) in the tone box designated. This term may have been undergone an analogical change or phonological contamination by replacing its original *C tone to *A tone from *vɛjA ‘fire’ (cf. Li 1977: 79, 286) or *wɤjA ‘fire’ (cf. Pittayawat Pittayaporn 2009: 207) according to the frequently used compound word *ʔdaŋC *vɛjA/*wɤjA ‘to make fire’. This indicates that the tonal category of this item in Daxin Baoxu (L23) has been changed much earlier than the time of tonal splits. Besides, some individual changes before/after the time of tonal splits also cannot be captured by tone box. For example, we have previously discussed that in Jingxi 215 Hurun Yang Zhuang (L11), pre-glottalized stops condition a secondary tonal split from the high register to merge into the low register in Tone A, like in ˀda:w31 (A2 < A1) ‘star’ and ˀban31 (A2 < A1) ‘to fly’. However, pre-glottalized glide ʔj- has been earlier merged into j-, and the tone conditioned by this j- (< *ʔj-) is preserved the original tone A1 of the primary tonal split, like in ja:51 (A1) ‘medicine’. The initial j(< *ʔj-) does not condition a secondary tonal split since its glottal feature has been lost before the time of secondary tonal splits (cf. §4.3.2.2). Modern loanwords borrowed from foreign languages into Tai varieties often do not follow the tonal patterns at time of tonal splits, and this determine that tone box cannot apply to modern loanwords in Tai. For instance, Rong Maet Tai Lue (L6) has borrowed abundant loanwords from the national language Bangkok Thai (L1). Normally tones collocating with unaspirated initial stops in Thai all distribute in the high register, which has a secondary tonal split together with the split conditioned by glottalized initials from A1 to merge into A2 in the A column. Therefore, A1-U is actually merged into A2 in Thai, in contrast A1-U is preserved as other high register rows in Tai Lue, like Thai pla:33 (A2 < A1-U) vs. Tai Lue pa:55 (A1-U) ‘fish’. According to this correspondence, the third syllable of the Thai word sin24 laʔ45 pin33 ‘artist’ should have been borrowed as pin55 (A1-U) in Tai Lue. However, in Rong Maet Tai Lue it unexpectedly has a A2 tone pin31, probably due to the correspondence between Thai A2 (33) and Tai Lue A2 (31) because A1-U has been merged into A2 in Thai. For another example, the tonal patterns of Debao Urban Yang Zhuang (cf. Table 36) determines that aspirated initials are never found in the low register (except B1-A > B2 and DL1 > DL2) due to that voiced sounds never develop to aspirations, but lots of modern Chinese loanwords in Yang Zhuang have low register tones, such as tʰa:j31 (=A2) ‘platform’ (< tʰa:i31 Southwestern Mandarin ), pʰa:j31 (=A2) ‘rank’ (< pʰa:i31 Southwestern Mandarin ), and tʰa:w213 (=C2) ‘cover’ (< tʰau21 Southwestern Mandarin). B1-A has been merged into B2 (33) to be different from B1-U/C (55) due to the regular rules, however many modern Chinese loanwords with aspirated initials have 55 tone like in kʰa:55 (=B1-U/C) ‘card’ (< kʰa55 Southwestern Mandarin), tʰa:n55 (=B1-U/C) ‘blanket’ (< tʰan55 Southwestern Mandarin), and pʰa:w55 (=B1-U/C) ‘to run’ (< pʰau55 Southwestern Mandarin). In contrast, this 55 tone is never found in Tai inherited words in Debao Urban Yang Zhuang. This phenomenon in both the cases of Tai Lue and Debao Urban Yang Zhuang indicates that after the tonal structure formed, the collocation of initial and tones may become flexible to break through the restrictions at time of tonal splits. 216 In short, any Tai tone box would never capture any kind of aberrant tonal development and unconventional tonal behaviors in modern loanwords borrowed from foreign languages. A tone box has been proved to be a convenient research tool in the fieldworks of Tai tones, especially Gedney’s tone box applied to SWT languages. On the basis of his tone box and the revised tone box (Liao & Shen 2012), an attempt of improving the designs for being able to capture the tonal distinctions in any Tai variety has been hypothesized. However, an integrated Tai tone box may become redundant because tonal development in Tai appears in a different diverse scene in different groups. Tai tone boxes designed with specific pertinence to different Tai periods and different Tai sub-groups have been suggested to use as practical research tools. Nevertheless, Tai tone boxes suggested in this chapter are supposed to be able to apply to different diachronic stages of Tai, specific sub-groups of Tai, and all Tai varieties. From a practical point of view, if the modern Tai subgroup (namely SWT, CT, NT, and YN) of a particular variety is known, a specific Tai tone box and its ancillary box should be used for investigating the tones in that variety. If the affiliation of that Tai variety is not known at all, the integrated Tai tone box and its ancillary box should be applied, since it is supposed to be able to capture all tonal distinct in any Tai varieties, except aberrant tonal development in some single items and the unconventional tonal behaviors in modern loanwords borrowed from foreign languages. 217 Conclusion This chapter summarizes and discusses the research findings of this thesis, and makes suggestions for related research which needs to be further studied. §7.1 gives a brief overview of this study. §7.2 provides the conclusions of this study by being divided into two sub-sections. The former §7.2.1 discusses the diachronic processes of Tai tonal development which is the first objective of this thesis, and concludes what the agreements and the disagreements between the hypothesis and the actual findings of Tai tonal development are. The latter §7.2.2 concludes the other two objectives, namely Tai classification and Tai tone box based on the findings in the diachronic scheme of Tai tonal development, and also discusses the agreements and the disagreements between these conclusions and the two relating hypotheses. The last section §7.3 suggests further research of this study. As the title suggests, Tonal Development of Tai languages, this thesis aims to provide a full scheme of tonal development of Tai, from tonogenesis in PT to different diachronic hierarchies of tonal splits in different Tai groups and varieties, and further suggests new viewpoints on Tai classification and Tai tone box based on tonology. As a matter of my Tai ethnic background and my experience, I am very interested in the language comparison among Tai languages, especially the tones. It is generally known all Tai languages are tonal just like the majorities of the languages in China and MSEA. My concern is that how tones in Tai languages have arisen, or in other words how Tai languages have developed from a non-tonal proto-language to become various tonal daughter languages. Driven by this interest, the main part of library research has been accomplished before the hypotheses of this thesis were proposed. My starting point is Gedney’s tone box (Gedney 1989[1972]). As a result of comparing the tonal splits of many modern SWT languages, the tone box designed by Gedney (1989[1972]: 202) consists of five tone categories (A, B, C, DL, DS) with four groups of initial consonants (Group 1=Voiceless friction, Group 2=Voiceless 218 unaspirated, Group3=Glottal, and Group 4=Voiced.) conditioning possible tonal splits. But when applied to varieties of CT and NT, Gedney’s tone box (1989[1972]) has several shortcomings. The most noticeable phenomena are tonal splits among the voiceless friction sounds and the voiceless unaspirated stops. For this reason, my colleague and I have suggested a revised version of Gedney’s tone box (Liao & Shen 2012). This background and the related literature review are discussed in Chapter 1 and Chapter 2 respectively. However, the revised tone box was based on an elementary tonal comparison among some of the Tai varieties, and it needs a more exhaustive tone data analysis to draw out the full tonal development scheme of Tai for improving its incomplete method. Therefore, based upon the wordlist designated in this revised tone box, this study chooses forty-two Tai varieties from the four Tai sub-groups, namely SWT, CT, NT and YN to collect tone data to analyze with, among them most are collected during my fieldworks in China and Thailand, and some are referenced to published sources. The criterion of selecting Tai varieties is based on the diversity of tonal patterns and the amounts of internal varieties. Therefore, CT varieties which contain the most complicated tonal split patterns are most numerous in the Tai varieties investigated. Some undescribed Tai varieties, such as Quan Son Tai (L7) of SWT, Debao Nalong Myang Zhuang (14), Debao Tuoxin Pyang Zhuang (L15) and Daxin Baoxu Zuojiang Zhuang (L23) of CT, Debao Longsang (L31) and Debao Dongling (L32) of NT, as well as Nanning Shuangding (L26) and Yongning Baiji (L25) of YN are particularly selected to study with in case of careless omission of special tonal split patterns (although finally most of them are proved to have common tonal split pattern except Baoxu). Historical and comparative linguistics provides the main methodology for this study. The details of the methodology are introduced in Chapter 3. The hypotheses of this thesis are as follows. (1) Three groups of proto-initials (voiceless-voiced-breathy) conditioned primary tonal splits whereas splits conditioned by other pharyngeal features, such as aspiration and pre-glottalization represent stages of secondary development. This is the basis of the other two hypotheses because it points out the diachronic hierarchies and stages of tonal development. (2) The “two plus one taxonomy” of Tai classification is more reasonable than the Tripartite Division at least in the early stage of Tai languages, because of different tonal behaviors of “proto-breathy” initials in Proto-Northern and Proto-Southern Tai languages. This hypothesis is based on that in PNT these initials conditioned the tones to merge with those tones conditioned by proto-voiced initials, while in PST they conditioned the tones to merge those conditioned by 219 proto-voiceless initials. (3) Based on the fuller scheme of Tai tonal development, a more comprehensive revised version of Gedney’s tone box will be able to capture the tonal distinctions in any Tai varieties. These three hypotheses are discussed in the three finding chapters, Chapter 4, Chapter 5, and Chapter 6 of this thesis respectively. After data analysis, the frame of the thesis has been fixed up. The conclusions which aim to test the three hypotheses are as shown in the following sub-sections. Before answering the first hypothesis, Tai tonogenesis is inevitably to discussed with because it is the first step of tonal development. Just like the hypotheses of tonogenesis which are applied to the majorities of languages in China and MSEA, tones in Tai is a secondary development which induced by syllable finals. It was that the decaying of syllable final -h gave an original high-falling tone *B on relatively long vowel duration, and the weakening of *-ʔ gave an original low-rising tone *C on relatively short vowel duration, to contrast to each other and to the least marked level tone *A on smooth syllables, for keeping the semantic distinctions from the original plain syllables without obstruent syllable finals. Syllables with final oral stops -p / -t / -k were preserved to consist the fourth tonal category *D with a highfalling contour like its counterpart tone *B. Tone data analysis suggests a full scheme of tonal splits as follows. After the four tonal categories arose in PT, they further developed to fall into two series to form eight tones due to the register splits conditioned by voiceless-voiced groups of initials. The trigger must have been the voicing of proto-voiceless continuants causing the merger of proto-voiceless-voiced continuants. In this earliest register tonal split process, a straightforward tonal split pattern was presented as all initial sounds with original voiceless features (voiceless stops and glottalized sounds) went together with original voiceless continuants to condition an original high tone, in contrast original voiced stops went together with original voiced continuants to condition an original low tone. This is the primary tonal split on the post-PT level. In all modern Tai sub-groups many varieties are found to preserve this primary tonal split pattern. Secondary tonal splits in Tai languages must have been conditioned by glottalized sounds, unaspirated stops, and aspirated sounds in the high register in some of the modern Tai varieties far later than the period of the primary tonal split., 220 because the crucial evidence that aspirated sounds are secondary development in post-PT level in ST and YN. Glottalized sounds and aspirated sounds are found to be able to condition secondary tonal split itself or together with one another, but secondary tonal splits conditioned by unaspirated stops are only found to be a byproduct of those conditioned by glottalized sounds. Secondary tonal splits normally present as being depressed by the conditioning initials to become the third register first, and in some varieties the third register is preserved to form a “three- way split” pattern, but in more varieties the third register may go to merge into the original low register or go back to merge into the original high register to finally form the “two-way split” pattern. This is why the two-way split pattern is still the main stream in Tai varieties. For the first hypothesis of this thesis, the most remarkable change of the main idea is the proto-breathy initials in the proposed three groups of proto-initials (voicelessvoiced-breathy) conditioning primary tonal splits (or in other words the voicing alternation between NT and CT/SWT) was actually not an actual existence on the PT level. After a fuller data analysis and a cautious comparison among the previous studies, this thesis agrees with the hypothesis of the sesquisyllable and monosyllable coexisting structure in PT (Pittayawat Pittayaporn 2009), and with that voicing alternation should have been induced by the different directions of initial mergers of “initial segment + medial” onsets (Weera Ostapirat 2006). After realizing that breathy sounds would have been induced as a transitional shape from sesquisyllable onset to aspirated or continuants sounds in the modern ST varieties, the hypothesis that proto-breathy at the PT level must be corrected to be VASO-1/2, which merged into unaspirated sounds in PNT, but merged into aspirated sounds in PST. In addition, voicing alternation between ST and NT-YN is not only limited to VASO1/2, but also suitable to VASO-2/1, which merge into voiced sounds in ST but into voiceless sounds in NT-YN. However, the main significance of this hypothesis is not changed, for that the result of voicing alternation between ST and NT-YN must have been at the PT level. The most important point is that voicing alternation between ST and NT-YN must have been on the earliest diachronic hierarchy of tonal split, and this point is the basis of that Tai has to be first divided into two primary branches ST and NT-YN, to attest the second hypothesis of this thesis. 221 On the basis of the full tonal development scheme suggested by the conclusion of the first hypothesis above, the other two hypotheses are attested as follows. First, the “Two plus one taxonomy” of Tai classification based on the primary tonal splits must have been clarified to be somewhat incomplete. Based on the analysis of tonal behaviors with supplementary evidence from initial behaviors, vocalic behaviors, and exclusive lexical items, this thesis suggests two levels of Tai division as follows. Primary, it agrees on the two-term system suggested by Haudricourt (1956) to divide PT first into Dioi or ʔyai65 vs. Tai proper, by another two-term system, NT-YN (<PNT) and ST (<PST), for avoiding confusing the conventional naming of Tai sub-groups. The division on this level is based on the primary tonal splits suggested by tonal behaviors of VASO-1/2, VASO-2/1, and horizontal tonal splits conditioned by “vowel-length alternation” between Tai varieties from NT-YN and ST respectively. From this level, YN whose classification as NT or CT is open to debate has been clarified to be get together with NT rather than CT. Secondarily, it further suggests a new “two plus two” (NT/YN + CT/SWT) taxonomy of modern Tai division based on the conclusion of secondary tonal splits. For example, ASO-CY shared by CT and YN as secondary areal feature can separate CT from SWT on secondary sub-group level. AMO-SY, ASO-SY, and ASO-YN as secondary innovations Dioi or ʔɔaɓ here is actually the autonym of many NT language members including many Northern Zhuang languages and Bouyei. The second syllable of the term Bouyei “Yei” (as well as “Yi” from Chinese Pinyin “Buyi”) is actually the same thing of Dioi or ʔɔaɓ (while the Bou/Bu is the Tai etyma ‘person, people, CLF of person’, like pʰu:C1-A in Bangkok Thai (L1), pʰowC1-A in Debao Urban Yang Zhuang (L8), and powC2 in Wuming Shuangqiao Yongbei Zhuang (L30)). I suppose that all of these different forms (with different notations of spelling) must have developed from the same PT etyma *ʔɔwaɔC (my reconstruction). According to my fieldwork research, most CT speakers, such as Yang Zhuang in the counties Debao, Jingxi, and Napo in Southwestern corner of Guangxi, normally call the speakers of their neighboring NT languages (e.g. Youjiang Zhuang, Guibian Zhuang and Yongbei Zhuang) pʰowC1-A ˀɔuɔC1-G (literally “people + NT”) and call these NT languages ɕa:ŋC1-U ʔ jujC1-G (literally “speak + NT”). The segment ʔjujC1-G is actually the cognate of Dioi, ʔɔaɓ, Yei and Yi. The reason that I reconstruct *ʔɔwaɔC for these different modern forms is because the proposed initial *ʔɔw- can explain all the irregular vocalic correspondence among different Tai varieties, as illustrated in the following examples. 1) The regular change *ʔɔw- > *ʔɔ- in many modern NZ and Bouyei languages like Tianlin Lizhou Zhuang (L33) and Wangmo Bouyei (Standard Bouyei) led *ʔɔwaɔC > ˀɔaɔC1-G ‘Bouyei’. 2) *ʔɔw- cause the following vowel to become rounded in all Yang Zhuang varieties, so that *ʔɔwaɔC > ˀɔuɔC1-G ‘Northern Zhuang/Bouyei’ (the similar process can also explain the Dioi form); 3) The same change as in Yang Zhuang may have processed in some extreme western NT varieties, such as Zhenning Bouyei (L42) and Qiubei Gehan (L39). Note that in these varieties, -uj has been merged into -i:, such as ti:C1/2 ‘bowl’ in Qiubei Gehan (vs. tʰuɔC1/2 in Debao Yang Zhuang (L810)). In Zhenning Bouyei (L42) and Qiubei Gehan (L39), *ʔɔw- > *ʔ- if the following vowel has the feature [+high, +front], so that *ʔɔwaɔC > *ʔɔuɔC > ʔɓ:C1-G ‘Northern Zhuang/Bouyei’. I suppose that only if the reconstruction *ʔɔwaɔC ‘NT (Northern Zhuang/Bouyei)’ is true, can all these irregular modern Tai forms be reasonable explained. 65 222 which are shared between varieties of YN suggest that YN is separated from NT on secondary sub-group level, too. Therefore, Li’s CT and SWT can be divided from each other under the ST sub-group, in contrast NT and YN differ from one another under the NT-YN sub-group. Second, the proposed tone box which is suggested to be able to capture the tonal distinctions in any Tai varieties can be obtained after all the possible conditioning phonation types of initials at time of tonal splits are studied. However, an integrated Tai tone box have been proved to be cumbersome if a Tai variety’s affiliation has been known. It is because secondary tonal splits must have been individually conditioned at a post-PT period in different Tai varieties. For instance, NT varieties lack of contrastive aspirated sounds and that unaspirated stops never go with glottalized initials to condition secondary tonal splits, therefore a tone box designated with the divisions containing the rows of aspirated sounds and unaspirated stops appear to be unnecessary for NT. For this reason, Tai tone boxes designated with specific pertinence to different Tai periods and different Tai subgroups are first suggested if the diachronic period of a Tai variety or the affiliation of a modern Tai variety are known. Different Tai tone boxes are designated based on the different situations of secondary tonal splits in different Tai sub-groups, namely NT, YN, CT, and SWT. Nevertheless, an integrated Tai tone box is further suggested to be able to capture the tonal distinctions in any Tai varieties. In this sense, the first two conclusions of the goals of this thesis are the two substantial basis of the conclusion of Tai tone box. This study has solved the three main knotty tonal problems as were presented above. However, as has been pointed out in §1.3.1, there are still some limitations of tonal research of this study. All these limitations should be put forward to further studies together with other areas for further research suggested by this study. The main suggestions for these areas are as follows. First, secondary tonal splits are so far only found in the high register in Tai languages, but are never found in the low register of Tai languages even though there are also different phonation types (such as unaspirated stops, aspirated stops, and continuants) on the low register. The reason deserves more attention. Further study may give a reasonable explanation for this phenomenon. 223 Second, Tai tonal development in natural speech environments should be further investigated. The limitation of lacking tone data from this occasion cannot be solved by our wordlist. That is, tone data acquired from single words do not reflect tone expressions from the natural conversation. Some tone changes in natural conversation should be systematically researched, since these changes such as tone sandhi or tone coarticulation may be found to be the newest part of tonal development. Furthermore, as discussed in §6.6, the phenomenon indicating that after the tonal structure formed, the collocation of initial and tones may become flexible to break through the restrictions at time of tonal splits like tones of modern loanwords in different Tai varieties should be further studied. Some tentative research has suggested that tones of loanwords in Tai varieties differ from different borrowing periods, and from different varieties. For example, among different Yang Zhuang varieties all OC and MC loans reflects regular correspondence of tonal categories, but modern Chinese loans borrowed from Southwestern Mandarin and Standard Mandarin all present very irregular correspondence of tonal categories. The factors and the trends of these tonal phenomena need more attention from researchers. Moreover, Tai classification based on tonology needs more investigation and verification. In this thesis the two levels of Tai classification have been suggested by the analysis of different tonal behaviors among different groups of Tai, and has been supported from initial behaviors, vocalic behaviors, and exclusive lexical items. It should be crucial to test some other recent Tai classifications based on newly methodological findings. For instance, Pittayawat Pittayaporn (2009: 298) has provided a new complex Tai subgrouping diagram which suggests that “SWT is the only one of Li’s three groups that is recognized as a valid subgroup” (Pittayawat Pittayaporn 2009: 302). However, the primary and the secondary divisions suggested by the tonal analysis of this thesis have called in questions that whether Pittayawat Pittayaporn’s Tai division has complied with the diachronic orders of shared innovations. This problem needs further comparisons and discussions. Finally, the new perspective of Tai tone box suggested in this thesis needs further extension to apply to other language groups which have internal tonal diversity, like the more complicated tonal patterns in Kam-Sui languages and some of the Sinitic languages (such as Cantonese dialects in Yulin, and some of the Southern Min varieties). The design of the ancillary box of Tai tone box has been suggested to be a flexible tool for collecting tone data from Tai varieties with the complicated tonal 224 correspondence based on complicated initial correspondence. The main tone box is succinct because in any specific Tai variety conditioning phonation types of initial groups cannot be more than five groups. When applying to Kam-Sui languages which have extremely complex patterns of voicing alterations (cf. Weera Ostapirat 2006), a succinct tone box with a relatively complex ancillary box (in which the source of voicing alternation will be able to be directly perceived) may be applicable. The method of tone box will be proved to be able to solve all the complicated tonal puzzles of languages with full-developing tones in further studies. 225 BIBLIOGRAPHY Anderson, Stephen R. 1978. Tone features. In Victoria A. Fromkin (ed.), Tone: a linguistic survey, 133-176. New York: Academic Press. Anttila, Raimo. 1989. Historical and comparative linguistics, 2nd edn. Amsterdam & Philadelphia: John Benjamin’s Publishing Company. Arlotto, Anthony. 1972. Introduction to historical linguistics. Lanham, Maryland: University Press of America. Bao, Zhining. 1990. On the nature of tone. Massachusetts Avenue: MIT dissertation. Baxter, William. 1992. A handbook of Old Chinese phonology. Berlin & New York: Mouton de Gruyter. Benedict, Paul K. 1942. Thai, Kadai and Indonesian: A new alignment in Southeastern Asia. American Anthropologist 44. 576-601. Benedict, Paul K. 1975. Austro-Thai: language and culture. New Haven, Conn: HRAF Press. Benedict, Paul K. 1990. Japanese-Austro-Thai. Ann Arbor: Karoma. Bi, Kechao. 1982. 入 . [Entering tone]. Wuhan: Hubei People’s Publishing House. Bloomfield, Leonard. 1933. Language. Chicago: University of Chicago Press. Blust, Robert. 1987. Lexical reconstruction and semantic reconstruction: the case of Austronesian ‘house’ words. Diachronica 4. 79-106. Brown, J. Marvin. 1965. From ancient Thai to modern dialects. Bangkok: Social Science Association Press. Brown, J. Marvin. 1975. The great tone split: did it work in two opposite ways? In Jimmy G. Harris and James R. Chamberlain (ed.), Studies in Tai linguistics in 226 honor of William J. Gedney, 33-48. Bangkok: Central Institute of English Language, Office of State University. Campbell, Lyle. 2004. Historical linguistics: an introduction, 2nd edn. Cambridge: The MIT Press. Chafe, W. L. 1959. Internal reconstruction in Seneca. Language 35. 477-495. Chamberlain, James R. 1975. A new look at the history and classification of the Tai Languages. In Jimmy G. Harris and James R. Chamberlain (ed.), Studies in Tai linguistics in honor of William J. Gedney, 49–66. Bangkok: Central Institute of English Language, Office of State University. Chang, Kun. 1966. A comparative study of the Yao tone system. Language 42.2. 303310. Chang, Kun. 1975. The reconstruction of Proto-Miao-Yao tones. Bulletin of the Institute of History and Philology. Academia Sinica 44. 541-628. Chao, Y. R. 1928. Studies in the modern Wu-dialects. Beijing: Tsing Hua College Research Institute. Chao, Y. R. 1930. A system of tone letters. Le Maître Phonétique 45. 24-27. Clements, G. N. 1981a (ed.). Harvard Studies in Phonology 2. Bloomington: Indiana University Linguistics Club. Clements, G. N. 1981b. The hierarchical representation of tone features. In G. N. Clements (ed.) Harvard Studies in Phonology 2, 108-177. Bloomington: Indiana University Linguistics Club. Crowley, Terry. 1994. An introduction to historical linguistics. Auckland: Oxford University Press. Diller, Anthony. 1998. The Tai language family and the comparative method. In Somsonge Burusphat (ed.), The international conference on Tai studies, 1-32. Bangkok: Institute of Language and Culture for Rural Development, Mahidol University. 227 Diller, Anthony. 2008. Introduction. In Anthony V. N. Diller, Jerold A. Edmondson & Yongxian Luo (ed.), The Tai-Kadai languages, 3-8. Routledge language family series. London & New York: Routledge. Dong, Hongyuan. 2014. A History of the Chinese Language. London & New York: Routledge. Downer, G. B. 1967. Tone-change and tone-shift in White Miao. Bulletin of the School of Oriental and African Studies, University of London 30.3. 589-599. Duanmu, San. 1990. A formal study of syllable, tone, stress, and domain in Chinese languages. Massachusetts Avenue: MIT dissertation. Dyan, Isidore. 1974. Lexical reconstruction. In Isidore Dyen & David F. Aberle (ed.), Lexical reconstruction: the case of the Proto-Athapaskan kinship system, 5-22. Cambridge University Press. Edmondson, Jerold A. 1990a. Kam tone splits and the variation of breathiness. In Jerold A. Edmondson, Crawford Feagin, & Peter Mühlhäusler (ed.), Development and diversity: Language variation across time and space, 187-202. Arlington: The University of Texas at Arlington and Summer Institute of Linguistics. Edmondson, Jerold A (ed.). 1990b. Kadai: discussions in Kadai and SE Asian linguistics. Arlington: Department of Foreign Languages and Linguistics, University of Texas at Arlington. Edmondson, Jerold A. 1992a. A study of the tones and initials in Kam, Lakkja, and Hlai. In Carol J. Compton & John F. Hartmann (ed.), Papers on Tai languages, linguistics, and literatures, in honor of William J. Gedney on his 77th birthday, 77-102. Illinois: Northern Illinois University, Center for Southeast Asian Studies. Edmondson, Jerold A. 1992b. On five-level tone system. In Shin Ja J. Hwang and William R. Merrifield (ed.), Language in context: Essays for Robert E. Longacre, 555-576. Dallas: Summer Institute of Linguistics and the University of Texas at Arlington. Edmondson, Jerold A. 1993. Tonality in Austronesian languages. Hawaii: University of Hawaii Press. 228 Edmondson, Jerold A. 1994. Change and variation in Zhuang. In Karen L. Adams & Thomas John Hudak (ed.), Papers from the second Annual Meeting of the Southeast Asian Linguistics Society, 147-185. Glendale: Arizona State University, Program for Southeast Asian Studies. Edmondson, Jerold A. 2002. Nung An: Origin of a species. In Robert S. Bauer (ed.), Collected papers on Southeast Asian and Pacific languages, 51–62. Pacific linguistics series 530. Canberra: Australian National University. Edmondson, Jerold A. & David B. Solnit. (eds.) 1988. Comparative Kadai: Linguistic studies beyond Tai. Summer Institute of Linguistics and the University of Texas at Arlington Publications in Linguistics, 86. Dallas: Summer Institute of Linguistics and the University of Texas at Arlington. Edmondson, Jerold A. & David B. Solnit. 1997. Introduction. In Edmondson, Jerold A. & David B. Solnit (ed.), Comparative Kadai: the Tai branch, 1-17. Summer Institute of Linguistics and the University of Texas at Arlington Publications in Linguistics, 124. Dallas: Summer Institute of Linguistics and the University of Texas at Arlington. Enfield. N.J. 2005. Areal Linguistics and Mainland Southeast Asia. Annual Review of Anthropology 2005. 34. 181-206. Ferlus, Michel. 1990. Remarques sur le consonatisme de proto Thai-Yay (révision du proto-Tai de Li Fangkuei). Paper circulated at the 23rd International Conference on Sino-Tibetan Languages and Linguistics, University of Texas at Arlington. Freiberger, Nancy. 1964. A phonemic description of Nong (Nung). Te Reo 7. 15-22. Fu, Baoning. 1995. A system of tone features and its implications for the representation of tone. Burnaby, PhD: Simon Fraser University dissertation. Gedney, William J. 1972. A puzzle in comparative Tai phonology. In J. G. Harris & R. B. Noss (ed.), Tai phonetics and phonology, 52-57. Bangkok: Central Institute of English Language, Office of State Universities. Gedney, William J. 1985. Confronting the unknown: total splits and the genealogy of Tai-Kadai, Linguistics of the Sino-Tibetan Area: The State of the Art, Pacific Linguistics Seriese-No.87. Canberra: Australian National University. 116-124. 229 Gedney, William J. 1989[1972]. A checklist for determining tones in Tai dialects. In Robert J. Bickner, John Hartmann, Thomas John Hudak, and Patcharin Peyasantiwong (ed.), Selected papers on comparative Tai studies, 191-206. Michigan: Center for South and Southeast Asian studies 29. Michigan: the University of Michigan. Gedney, William J. 1989a. Evidence for another series of voiced initials in proto-Tai. In Robert J. Bickner, John Hartmann, Thomas John Hudak, & Patcharin Peyasantiwong (ed.), Selected papers on comparative Tai studies, 229-270. Michigan Papers on South and Southeast Asian Studies 29. Michigan: The University of Michigan. Gedney, William J. 1989b. Future directions in comparative Tai linguistics. In Robert J. Bickner, John Hartmann, Thomas John Hudak, & Patcharin Peyasantiwong (ed.), Selected papers on comparative Tai studies, 7-116. Michigan Papers on South and Southeast Asian Studies 29. Michigan: The University of Michigan. Gedney, William J. 1989c. Speculations on early Tai tones. In Robert J. Bickner, John Hartmann, Thomas John Hudak, & Patcharin Peyasantiwong (ed.), Selected papers on comparative Tai studies, 207-228. Michigan Papers on South and Southeast Asian Studies 29. Michigan: The University of Michigan. Gordon, Matthews, & Peter Ladefoged. 2001. Phonation types: a cross-linguistic overview. Journal of Phonetics 29. 383-406. Gregerson, Kenneth J. & Edmondson, Jerold A. 1998. Some puzzles in Cao Lan. n Somsonge Burusphat (ed.), The international conference on Tai Studies, 151164. Bangkok: Institute of Language and Culture for Rural Development, Mahidol University. Haas, Mary R. 1969. Internal reconstruction of the Nootka-Nitinat pronominal suffixes. International Journal of American Linguistics 35. 108-124. Hanna, William J. 2012. Dai Lue-English dictionary. Chiang Mai: Silkworm Books. Harris, Z. S. 1960. Structural linguistics. Chicago: University of Chicago Press. 230 Harris, J.G. 2001. States of the glottis of Thai voiceless stops and affricates. In M.R. Kalaya Tingsabadh & A.S. Abramson (ed.), Essays in Tai linguistics, 3–11. Bangkok: Chulalongkorn University Press. Hashimoto, Mantaro J. 1972. The linguistic mechanism of flip flop. Unicorn 10. 119. Hashimoto, Oi-kan Yue. 1986. Tonal flip-flop in Chinese dialects. Journal of Chinese Linguistics 14(2). 161-183. Haudricourt, André-Georges. 1948. Les occlusives velaires en thai. Bulletin de la Société de Linguistique de Paris 48. 86-89. Haudricourt, André-Georges. 1954a. Comment reconstruise le chinois archai ̈que. Word 10. 351-364. Haudricourt, André-Georges. 1954b. De l’origine des tons en viêtnamien. Journal Asiatique 242. 161-183. Haudricourt, André-Georges. 1956. De la restitution des intitiales dans les langues monosyllabiques: le problème du thai commun. Bulletin de la Société de Linguistique de Paris 52 (1). 307-322. Haudricourt, André-Georges. 1961. Bipartition et tripartition des systèmes de tons dans quelques langues d’Extrême-Orient. Bulletin de la Société de Linguistique de Paris 56. 163-180. Haudricourt, André-Georges. 1963. Remarques sur les initiales complexes de la language Sek. Belletin de la Société de Linguistique de Paris 58 (1). 156-163. Haudricourt, André-Georges. 1972. Two-way and three-way splitting of tonal systems in some far eastern languages (translated by Christopher Court). In J.G. Harris & R.B. Noss (ed.), A conference on Tai phonetics and phonology, 5886. Bangkok: Mahidol University. Henderson, Eugéne J. A. 1964. Marginalia in Siamese phonetic studies. In D. Abercrombie, D.B. Fry, P.A.D. MacCarthy, N.C. Scott & J.L.M. Trim (ed.), In honour of Daniel Jones: papers contributed on the occasion of his eightieth birthday 12 September 1961, 415–424. London: Longmans. 231 Henderson, Eugéne J. A. 1982. Tonogenesis: some recent speculations on the development of tone. Transactions of the Philological Society 1982. 1-24. Hirayama, Hisao. 1987. The record of tones of the Tang Dynasty in Shintanzoo by the Japanesse monk Annen. Wang Li memorial volumes, Chinese volume. Hoenigswald, H.M. 1960. Languages change and linguistic reconstruction. Chicago: University of Chicago Press. Hock, Hans Henrich. 1991. Principles of historical linguistics, 2nd edn. Berlin: Mouton de Gruyter. Holm, David. 2010. Linguistic diversity along the China-Vietnam border. Linguistics of the Tibeto-Burman Area 33(2). 1-62. Hombert, Jean-Marie. 1976. The effect of aspiration on the fundamental frequency of the following vowel. BLS 1976. 212-219. Hombert, Jean-Marie. 1978. A model of tone systems. In Donna Jo. Napoli (ed.), Elements of tone, stress, and intonation. Washington D. C.: Georgetown University Press. Hombert, Jean-Marie, John J. Ohala, & William G. Ewan. 1979. Phonetic explanations for the development of tones. Languages 55(1). 37-58. Huang, Yuanwei. 1997. The interaction between Zhuang and the Yue (Cantonese) dialects. In Jerold A. Edmondson & David B. (ed.), Comparative Kadai: the Tai branch, 57-76. Texas: The Summer Institute of Linguistics and The University of Texas at Arlington. Hudak, Thomas John. (ed.) 1995. William J. Gedney’s Central Tai Dialects: Glossaries, Texts, and Translations. Michigan Papers on South and Southeast Asia, Center for South and Southeast Asian Studies, The University of Michigan. Hudak, Thomas John. (ed.) 1997. William J. Gedney’s Tai Dialects Studies: Glossaries, Texts, and Translations. Michigan Papers on South and Southeast Asia, Center for South and Southeast Asian Studies, The University of Michigan. Hudak, Thomas John. 2008. William J. Gedney’s Comparative Tai Source Book. Honolulu: University of Hawaii Press. 232 Hyman, Larry M. 1986. The representation of multiple tone heights. In K. Bogers, H. Van der Hulst, & M. Mous (ed.), The phonological representation of suprasegmentals, 109–152. Dordrecht: Foris. International Phonetic Association (Ed.). 1999. Handbook of the International Phonetic Association: A guide to the use of the International Phonetic Alphabet. Cambridge & New York: Cambridge University Press. Iwata, R., M. Sawashima, & H. Hirose. 1981. Laryngeal adjustments for syllable-final stops in Cantonese. Phoniat 15. 45–54. Iwata, R., M. Sawashima, H. Hirose. & S. Niimi. 1979. Laryngeal adjustments of Fukienese stops: initial plosives and final applosives. Phoniat 13. 61–81. Jackson, Eric M., Emily H. S. Jackson, & Lau Shuh Huey. 2011. A sociolinguistic survey of the Dejing Zhuang dialect area. SIL international, East Asia Group in cooperation with the Guangxi Minorities Language and Scripts Work Commission. https://mail.link77.net/[email protected]/ 20110629DRAFT -English-Dejing_survey_report-ESR.pdf (15 Dec., 2012.) Jeffers, Robert J. & Lehiste, Ilse. 1982. Principles and methods for historical linguistics. Cambridege, Massachusetts & London: The MIT Press. Johnson, Eric C. 2011. A lexical and phonological comparison of the Central Taic languages of Wenshan Prefecture, China. SIL Electronic Working Papers 2001005. http://www.sil.org/silewp/2011/silewp2011-005.pdf (17 Feb., 2012.) Karlgren, Bernhard. 1964. Grammata serica recensa. Stockholm: Museum of Far Eastern Antiquities. Lan, Qingyuan. 1982. 语中古汉语借词及汉越语与平话的关系 [The relationship among Middle Chinese loans in Zhuang, Sino-Vietnamese, and Pinghua]. Minority Languages of China 2001(3). 48-61. Lass, Roger. 1997. Historical linguistics and language change. In S.R. Anderson, J. Bresnan, B. Comrie, W. Dressler, C. Ewen, R. Huddleston, R. Lass, D. Lightfoot, J. Lyons, R. H. Matthews, R. Posner, S. Romaine, N. V. Smith & N. Nincent (ed.), Cambridge studies in linguistics. Cambridge: Cambridge University Press. 233 Li, Fang Kuei. 1940. The Tai dialect of Lungchow: Text, Translations, and Glossary. Bulletin of the Institute of Ethnology, Academia Sinica 16. Li, Fang Kuei. 1943. The Hypothesis of a pre-glottalized series of consonants in primitive Tai. Bulletin of the Institute of Ethnology, Academia Sinica 11. 177188. Li, Fang Kuei. 1944. 语喉塞音及带喉塞音 母对于剥隘 调系统之影响 [The influence of the primitive Tai glottal stop and preglottalized consonants on the tone systems of Po-ai]. The Journal of the Institute of Siniology of the National University of Peking 4. 59-67. Li, Fang Kuei. 1945. Some Old Chinese Loan Words in the Tai Languages. Harvard Journal of Asiatic Studies 8. 333-342. Li, Fang Kuei. 1948. The distribution of initials and tones in the Sui language. Language 24. 160–167. Li, Fang Kuei. 1949. Tones in the Riming System of the Sui Language. Word 5. 262267. [Li, Fang Kuei. 1956. The Tai dialect of Wu-ming (Text, Translations, and Glossary). Bulletin of the Institute of Ethnology, Academia Sinica 19. Li, Fang Kuei. 1959. Classification by vocabulary: Tai dialects. Anthropological Linguistics 1.2. 15-21. Li, Fang Kuei. 1960. A tentative classification of Tai dialects. In S. Diamond (ed.), Culture in history: Essays in honor of Paul Radin, 951-959. Li, Fang Kuei. 1965. The Tai and the Kam-Sui languages. Lingua 14. 148-179. Li, Fang Kuei. 1966. The relationship between tones and initials in Tai. In Norman H. Zide (ed.), Studies in comparative Austroasiatic Linguistics, 82-88. Li, Fang Kuei. 1970a. Some tonal irregularities in the Tai languages. In Roman Jakobson & Shigeo Kawamoto (ed.), Studies in General and oriental linguistics, 415-422. Tokyo: TEC Corporation for language and educational research. Li, Fang Kuei. 1970b. 天保土歌-附音系 [The songs of T'ien-Pao: With a phonological sketch]. Bulletin of the Institute of Ethnology, Academia Sinica 30. 1-21. 234 Li, Fang-Kuei. 1971. On Siamese jaai. Academia Sinica/Bulletin of the Institute of History and Philology (Tapei) 42(3). 337-340. Li, Fang Kuei. 1973. Some dental clusters in Tai. Bulletin of the School of Oriental and African Studies 36.2. 334-339. Li, Fang Kuei. 1976. Sino-Tai. Computational Analysis of Asian and African Languages 3. 39-48. Li, Fang Kuei. 1977. A handbook of comparative Tai. Honolulu: The University Press of Hawaii. Li, Fang Kuei. 1978. A syntactic change in Thai. In Mohamed Ali Jazayery, Edgar C. Polome & Werner Winter (ed.), Linguistic and literary studies in honor of Archioald A. Hill, volume 3: Historical and comparative linguistics, 141-145. The Hague: Mouton Publishers. Li, Fang Kuei. 1980. 古音研究. [Researches on Old Chinese phonology]. Beijing: The Commerical Press. Li, Fang Kuei. 1987. 广西僮语概况 中国地质学会专刊 8. 17-20. [The General Situation of the Chuang Language of Kuang-his. Memoir of the Geological Society of China 8. 17-20]. Li, Fang Kuei. 1989. Proto-Tai *kh and *x-. In J.H.C.S. Davidson (ed.), South-east Asian Linguisitics: Essays in Honour of Eugénie J.A. Henderson, 143-146. School of Oriental and African Studies, University of London. Li, Yongsui, Chen Kejiong, & Chen Qiguang. 1959. Several issues concerning the initials and tones of Miao. Yuyan Yanjiu (Language Study) 1959(4). 65-80. Liang, Min & Zhang Junru. 1993. 侗台语族送气清塞音 母的产生和发展. [The rise and development of voiceless aspirated initials in Kam-Tai languages]. Minzu Yuwen 5. 10-15. Liang, Min & Zhang Junru. 1996. 侗台语族概论. [An introduction to the Kam-Tai Languages]. Beijing: China Social Sciences Publishing House.] Liang, Min & Zhang Junru. 2004. 从汉台语言的数词是否同源词说起. [Discussion on whether Chinese and Taic numerals are cognates]. Minzu Yuwen 2004(2). 52-56. 235 Liang, Yougang. 1986. 海南島文昌方言音系. [The phonology of the Wenchang dialect, Guangdong Province]. Fangyan 1986(2). 123-132. Liao, Hanbo. 2010. 台语支中部组佒侬语 德靖土语 音系概况及其拼音方案详解 [An overview of the sound system of the Central Taic language Yang-Nong (Dejing vernacular) and description of a phonetic spelling system]. In Li Fuqiang and Gao Ya-ning (ed.), Chinese Zhuang Studies 4, 70-192. Beijing: Publishing House of Minority Nationalities. Liao, Hanbo. 2013. Proto-Tai reconstruction of ‘maternal grandmother’ revisited: *na:j A or *ta:j A ?. Paper presented at The 23rd Annual Meeting of The Southeast Asian Linguistics Society, Bangkok. Liao, Hanbo. 2015. Yang Zhuang poetry. Paper presented at The 25th Annual Meeting of The Southeast Asian Linguistics Society, Chiang Mai. Liao, Hanbo & Shen, Ruiqing. 2012. Gedney’s tone box revisited: Evidence from some varieties of Central and Northern Tai. Paper presented at The 45th International Conference on Sino-Tibetan Languages and Linguistics, Singapore. Lien, Chinfa. 1986. Tone merger in the dialects of northern Chinese. Journal of Chinese Linguistics 14(1). 243-291. Luo, Yongxian. 1994. Some auxiliary verbs in Zhuang. In Hajime Kitamura, Nishida Tatsuo, & Nagano Yasuhiko (ed.), Current issues in Sino-Tibetan linguistics, 876-882. Osaka: Organizing Committee of The 26th International Conference on Sino-Tibetan Languages and Linguistics. Luo, Yongxian. 1996. Tonal irregularities in Tai revisited. Mon-Khmer studies 25. 69102. Luo, Yongxian. 1997. The subgroup structure of the Tai languages: A historical – comparative study. Journal of Chinese Linguistics, monograph series number 12. Lyons, J. 1977. Semantics, 2 vols. Cambridge: Cambridge University Press. M.R. Kalaya Tingsabadh. 2001. Thai tone geography. In M.R. Kalaya Tingsabadh & Arthur S. Abramson (ed.), Essays in Tai Linguistics, 205-228. Bangkok: Chulalongkorn University Press. 236 Maddieson, Ian. 1970. The inventory of features required for handing tone in transformational phonology. Research Notes of Department of Linguistics and Nigerian Languages 3.2-3. 3-18. University of Ibadan. Maddieson, Ian. 1978a. Tone effects on consonants. Journal of Chinese Linguistics 6. 327-344. Maddieson, Ian. 1978b. Universals of tone. In Joseph H. Greenberg (ed.), Universals of human language: phonology. Stanford: Stanford University Press. Mai, Yun. 2011. 从广西钟山清塘 语第六调看嘎裂 [Creakiness: A case of the sixth tone in Qingtang Zhuang from Zhongshan, Guangxi]. Minority Languages of China 2011(1). 20-26. Maspero, Henri. 1912. Etudes sur la phonétique historique de la langue annamite. Bulletin de l’Ecole Française de l’Extrême Orient 12. 1-127. Matisoff, James A. 1970. Glottal dissimilation and the Lahu high-rising tone: a tonogenetic case-study. Journal of the American Oriental Society 90 (1). 13-44. Matisoff, James A. 1973. Tonogenesis in Southeast Asia. In Larry M. Hyman (ed.), Southern California occasional papers in linguistics 1, 72-95. LA: University of Southern California. Mazaudon, Martine. 1977. Tibeto-Burman tonogenetics. Linguistics of the TibetoBurman area 3(2). 1-123. Mei, Tsu-lin. 1970. Tones and prosody in Middle Chinese and the origin of the rising tone. Harvard Journal of Asiatic Studies 30. 86-110. Mei, Tsu-lin. 1982. On the rising tone. Tsing Hua Journal of Chinese Studies new series XIV (1&2). 233-241. McMahon, April M.S. 1994. Understanding language change. Cambridge University Press. Michaud, Alexis. 2004. Final consonants and glottalization: new perspectives from Hanoi Vietnamese. Phonetica 61(2-3). 119-146. Ni, Dabai. 1990. 侗台语概论. [An introduction to Dong-Tai languages]. Beijing: The Press of Minzu Universiy of China. 237 Nicolson, Beth. 2000. The Nung An language of Vietnam: stepchild or aberrant son? In The Fifth International Symposium on Languages and Linguistics, 266-295. Ho Chi Minh City: Vietnam National University, Ho Chi Minh City University of Social Sciences and Humanities. Nishida, Tatsuo. 1954. トンネイムによるタイ諸語比較言語学的研究 [Comparative linguistic studies of the Tai languages based on tonemes]. Gengo Kenkyū 25. 19-46. Nishida, Tatsuo. 1955. マック・スイ語と共通語比較言語学的研究 [Mak-Sui and Common Tai]. Gengo Kenkyū 28. 30-62. Norman, Jerry. 1988. Chinese. Cambridge: Cambridge University Press. Owen, Robert Wyn. 2008. Language use, literacy, and phonological variation in Khuen. Chiang Mai: Payap University MA thesis. Owen, Robert Wyn. 2008. A tonal analysis of contemporary Tai Khuen varieties. Journal of the Southeast Asian Linguistics Societ 5: 12-31. Pan, Wuyun. 1982. Concerning some questions in the development of Chinese tones. Journal of Chinese Linguistics 10. 361-385. Pearson Education Asia Limited & Foreign Language Teaching and Research Press. 2009. Longman Dictionary of Contemporary English. Beijing: Foreign Language Teaching and Research Press. Phinnarat Akharawatthanakun. 2010. Phonological variation in Phuan. MANUSYA: Journal of Humanities 13.1. 50-87. Pike, Kenneth L. 1948. Tone languages. Michigan: University of Michigan Press. Pittayawat Pittayaporn. 2007a. A chronological-approach to subgrouping: The case of Southwestern Tai. Paper presented at Leipzig Students’ Conference in Linguistics. Max-Planck Institute for Evolutionary Anthropology, Leipzig, Germany. March 25-26. Pittayawat Pittayaporn. 2007b. Directionality of tone change. In Jürgen Trouvain and William J. Barry (ed.), Proceedings of the 16th International Congress of Phonetic Sciences (ICPhS XVI), 1421-24. Saarbrücken, Germany: Saarland University. 238 Pittayawat Pittayaporn. 2009. The phonology of Proto-Tai. New York: Cornell University dissertation. Prapin Manomaivibool. 1975. Study of Sino-Tai lexical correspondences. Seattle: University of Washington dissertation. Prapin Manomaivibool. 1976. Layers of Chinese loanwords in Thai. In W. Gething J. G. Harris & P. Kullavanijaya (ed.), Tai linguistics in honor of Fang-Kuei Li, 179184. Bangkok: Central Institute of English Language, Office of State University. Pranee Kullavanijaya. 1990. Undoing Homonymy: Cases in Debao Zhuang and Thai. In Proceedings of The 4th International Conference on Thai Studies, 78-92. Pranee Kullavanijaya. 1991. Dictionary of Southern Zhuang-Thai. Bangkok: Chulalongkorn University Press. Pranee Kullavanijaya & Theraphan L-Thongkum. 1998. Linguistic criteria for determining Tai ethnic groups: case studies on Central and South-western Tais. In Proceedings of The International Conference On Tai Studies, July 29-31, 1998, Bangkok. Pranee Kullavanijaya et al. 1984. ศพท์ ท 6 ภ ษ . [The vocabulary of six Tai languages]. Bangkok: Faculty of Arts, Chulalongkorn University. Pulleyblank, Douglas. 1986. Tone in lexical phonology. Dordrecht: D. Reidel. Pulleyblank, E. G. 1962. The consonantal system of Old Chinese. Asia Major 9. 58114 and 206-65. Pulleyblank, E. G. 1978. The nature of the Middle Chinese tones and their development to early Mandarin. Journal of Chinese Linguistics 6(2). 173-203. Pulleyblank, E. G. 1991. Lexicon of reconstructed pronunciation in Early Middle Chinese, Late Middle Chinese, and Early Mandarin. Vancouver: Unversity of British Columbia Press. Ren, Min. 1999. 粤方言区的 调问题. [Issues on tones of Cantonese dialectal areas]. Journal of Foshan Univeristy (Social Science Edition) 17(2). 61-65. 239 Rose, Phil. 2006. Zooming-in on Oujiang Wu: tonal homogeneity and acoustic reconstruction in a small subgroup of Chinese dialects. Australian International Conference on Speech Science and Technology 2006. 58-63. Sagart, Laurent. 1986. On the departing tone. Journal of Chinese Linguistics 14 (1). 90-113. Sagart, Laurent. 1988. Glottalized tones in China and South-East Asia. In D. Bradley, E. J. Henderson & M. Mazaudon (ed.), Prosodic analysis and Asian linguistics: to honour R. K. Sprigg, 83-93. Canberra: Pacific Linguistics. Saussure, Ferdinand de. 1983. Course in General Linguistics. Eds. Charles Bally and Albert Sechehaye. Trans. Roy Harris. La Salle, Illinois: Open Court. Sarawit, Mary E. 1973. The Proto-Tai vowel system. Michigan: The University of Michigan dissertation. Sarawit, Mary E. 1975. Some changes in the final component of the Tai syllable. In J. G. Harris & J. R. Chamberlain (ed.), Studies in Tai linguistics in honor of William J. Gedney, 316-328. Central Institute of English Language. Sears, Andrew & Julie A. Jacko. 2007. The human-computer interaction handbook: fundamentals, evolving technologies, and emerging applications, 2nd edition. New York, NY: Taylor and Francis Group. Schmidt, Johannes. 1872. Die Verwandtschaftsverhältnisse De Indogermanischen Sprachen. Weimar: Böhlau. Shafer, Robert. 1940. The vocalism of Sino-Tibetan. Journal of the American Oriental Society 60(3). 302-337. Shafer, Robert. 1944. Problems in Sino-Tibetan phonetics. Journal of the American Oriental Society 64(3). 137-143. Shen, Ruiqing & Liao Hanbo. 2012. Acoustic-tonetic study of Pyang Zhuang: An undescribed Central Tai variety. Paper presented at The 45th International Conference on Sino-Tibetan Languages and Linguistics, Singapore. Shi, Cunzhi. 1981. An outline of the history of the Chinese speech sounds. Beijing: The Commerical Press. 240 Shi, Lin. 1992. Distinctive features of tone for the Dong language. Minority Languages of China 1991(5). 26-34. Somsonge Burusphat. 2011. Variation of initial consonants in Thai Song. Language and Culture 29(2). 81-113. Somsonge Burusphat. 2012. Tones of Thai Song varieties. Journal of the Southeast Asian Linguistics Societ 5. 32-48. Strecker, D. 2009. Tai languages. In B. Comrie (ed.), The world’s major languages, 653-659. Abingdon: Routledge. Tai Chung-pui. 2006. 宋代嶺南文獻中的侗台語詞 [Kam-Dai words in Lingnan manuscripts of the Song Dynasty]. Minority Languages of China 2006(3). 2125. Theraphan L-Thongkum. 1997. Implications of the retention of proto-voiced plosives and fricatives in the Dai Tho language of Yunnan Province for a theory of tonal development and Tai language classification. In Jerold A. Edmondson & David B. (ed.), Comparative Kadai: the Tai branch, 191-220. Texas: The Summer Institute of Linguistics and The University of Texas at Arlington. Thurgood, Graham. 2002. Vietnamese and tonogenesis: revising the model and the analysis. Diachronica 19(2). 333-363. Wang, Stephen S. 1966. Phonology of Chinese loanwords in a northern Tai dialect. Seattle: University of Washington dissertation. Wang, William S-Y. 1967. Phonological features of tone. IJAL 33. 93-105. Wang, William S-Y. 1969. Competing changes as a cause of residue. LG 45(1). 9-25. Weera Ostapirat. 1994. Speculations on proto Kam-Sui breathy sounds. Kadai: Discussions in Kadai & Southeast Asian Linguistics 4. 79-87. Weera Ostapirat. 1999. Proto-Kra. Berkeley: Unviversity of California. dissertation. Weera Ostapirat. 2006. Alternation of tonal series and the reconstruction of protoKam-Sui. In Dah-An Ho, Samuel Cheung, Wuyun Pan, & Fuxiang Wu (ed.), Linguistic studies in Chinese and neighboring languages: festschrift in honor of Professor Pang-hsin Ting on his 70th birthday, 1077-1121. Language and 241 linguistics monograph series, W-6. Taipei: Institute of Linguistics, Academia Sinica. Weera Ostapirat. 2013a. Austro-Tai revisited. Paper presented at SEAL23, Bangkok. Weera Ostapirat. 2013b. The rime system of proto-Tai. Bulletin of Chinese Linguistics 7(1). 189–227. Wei, Qingwen & Qin Guosheng. 1980. 壮语简志 [Brief overview of the Zhuang language]. Chinese minority languages brief overview series. Beijing: The People’s Publishing House. Woo, Nancy. 1969. Prosody and phonology. Bloomington: Indiana University dissertation. Wulff, K. 1934. Chinesisch und Tai: sprachvergleichende Untersuchungen. København [Copenhagen]: Levin und Munksgaard, Ejnar Munksgaard. Xu, Tongqiang. 1991. 历史语言学 (第一版) [Historical linguistics, first edition]. Beijing: The Commercial Press.] Yang, Huandian, Liang Zhenshi, Li Puying & Liu Cunhan. 1985. The distribution of Chinese dialect in Guangi. Fangyan 1985(3). Ye, Xiangling. 1983. The tones of the Wujiang dialect. Fangyan 1983 (1). Yip, Moira. 1980. The tonal phonology of Chinese. Massachusetts Avenue: MIT dissertation. Yip, Moira. 2002. Tone. Cambridge: Cambridge University Press. Yuan, Jiahua. 1960. An outline of Chinese dialects. Beijing: Wenzi Gaige Chubanshe. Zhang, Jie. 2002. The effects of duration and sonority on contour tone distribution. New York; London: Routledge. Zhang, Junru. 1980. 原始台语 母类别探索 [An exploration of the classification of initial consonants of proto-Tai]. Minority Languages of China 1980(2). 31-40. Zhang, Junru. 1982. 广西中南部地区 语中的老借词源于汉语古“平话”考 [A verification that the old Sinitic loans in the Zhuang languages in Central- 242 Southern Guangxi are from ancient Chinese Pinghua]. The Study of Languages 1982(1). 197-219. Zhang, Junru, Liang Min, Ouyang Jueya, Zheng Yiqing, Li Xulian & Xie Jianyou. 1999. 语方言研究. [Zhuang dialect research]. (Chinese minority language dialect research series.) Chengdu: Sichuan Ethnic Publishing House. Zheng, Yiqing. 1996. 靖西 语研究. [Jingxi Zhuang language research]. Beijing: Chinese Academy of Social Science, Nationalities Research Institute. Zhu, Xiaonong & Yang Jianfen. 2010. 嘎裂聲作為低調特徵：河 省方言的聲調考察 [Creaky voice as a feature of [+low]: A tone survey of Hebei dialects]. Hong Kong University of Science and Technology 2010(7). 134-147. http://en.wikipedia.org/wiki/Tai_languages (8 May, 2012.) https://en.wikipedia.org/wiki/Southern_Thai_language (11 June, 2016.) http://www-01.sil.org/iso639-3/ (17 July, 2013.) http://www-01.sil.org/iso639-3/scope.asp#I (3 July, 2013.) https://www.ethnologue.com/about/problem-language-identification (24 April, 2016.) https://www.ethnologue.com/language/zha (24 April, 2016.) 243 APPENDIX A QUESTIONNAIRE Information collecting of Language and Language Resource Person66 Dialect/Language: Dialect/Language Name Sub-district Place of data elicitation Date Village District Province Month Year LRP’s Name-Last name Age Gender Birth Place: Village Sub-district District Places lived: 1. Village Province 2. Village Province 3. Village Province 4. Village Province Sub-district Period of time Sub-district Period of time Sub-district Period of time Sub-district Period of time District Time period of living at the present residence year(s) District year(s) District year(s) District year(s) year(s) Occupation Education Dialects/Languages spoken: LRP can speak 1st dialect/language 3rd dialect/language Birth place of LRP’s father: Village District Province dialects/languages dialect/language 4th dialect/language 2nd Sub-district Province LRP’s father’s spoken language(s): LRP’s father can speak dialects/languages 66 This questionnaire is made by Dr. Phinnarat Akharawatthanakun for collecting tone data in Thai dialects in AL701 Field Methods in Linguistics course, Linguistics Department, Payap University. 244 1st dialect/language 3rd dialect/language 2nd dialect/language 4th dialect/language Birth place of LRP’s mother: Village Sub-district District Province LRP’s mother’s spoken language(s): LRP’s mother can speak 1st dialect/language 2nd dialect/language 3rd dialect/language 4th dialect/language Parents’ place lived: District Village dialects/languages Sub-district Province Spouse 1: Village Sub-district District Province Status:  living together  divorced  passed away Language spoken: 3 1st dialect/language 2nd dialect/language 3rd dialect/language 4th dialect/language Spouse 2: Village Sub-district District Province Status:  living together  divorced  passed away Language spoken: 1st dialect/language 2nd dialect/language 3rd dialect/language 4th dialect/language Note: 245 APPENDIX B CHECKLIST A revised version of Gedney’s tone box (Liao & Shen 2012) attaching a checklist with examples from three representative varieties from SWT, CT, and NT respectively Proto-Initial/Tone Categories ProtoVoiceless Sounds A B C DL DS Voiceless Aspirations A1-A B1-A C1-A DL1-A DS1-A Voiceless Continuants A1-C B1-C C1-C DL1-C DS1-C Voiceless Unaspirated Stops A1-U B1-U C1-U DL1-U DS1-U A1-UC B1-UC C1-UC DL1-UC DS1-UC Glottal Sounds A1-G B1-G C1-G DL1-G DS1-G Voiced Sounds A2 B2 C2 DL2 DS2 A1/2 B1/2 C1/2 DL1/2 DS1/2 Voiceless Unaspirated Stop + *r Clusters Proto-Voiced Proto-Breathy Voiced Aspirations & Breathiness Smooth Syllable Checked Syllable * In the following charts of checklist, Bangkok Thai (L1), Debao Urban Yang Zhuang (L8), and Wuming Shuangqiao Yongbei Zhuang (L30) are varieties representing Southwestern Tai, Central Tai, and Northern Thai respectively. In the charts, these three representative languages are abbreviated as Thai, Debao-1, and Wuming respectively. The tonal values of the lexical items are omitted in the checklist, but can be consulted in the correcponding languages in Appendix C. There are some limitation of lexical items in some cases. For example, words are not plenty found as A tones in the boxes of B1-UC, B1/2, and so on. 246 Checklist of test words in Tai languages: Column A Language A1-A A1-C A1-U A1-UC A1-G A2 Thai Debao-1 Wuming Gloss kʰa: kʰa: ka: leg 脚 huə tʰu: -- head 头 pʰa: pʰja: pla: cliff (SWT), hill (CT, NT)石山 ha: kʰja: ɣa: to seek fa: pʰa: fa: lid, cover, wall ha:ŋ tʰa:ŋ ɣiəŋ tail tʰɛ:m tʰe:m te:m to add on, in addition添加 ma: ma: ma: dog 狗 na: na: na: thick ŋa:j ŋa:j -- turn face up -- ɬa: θa: paper (CT, NT) sa:m ɬa:m θa:m three saj ɬɔj θaɰ clean 净 ka: ka: ka: crow 乌鸦 kin kin kɯn to eat 吃 tu: tow tow door ta: ta: (B1-U) ta: maternal grandfather pla: pja: pla: fish paj paj paj to go 去 ta: tʰa: ɣa: eye ta:j tʰa:j ɣa:j to die taj tʰɔj taɰ gizzard 胗 鸡～ tɛ:ŋ tʰe:ŋ tiəŋ cucumber ka:ŋ kʰa:ŋ ka:ŋ put up, spread out -- ˀda: ˀda: baby sling (CT, NT) ba:n ˀba:n -- bloom da:w ˀda:w ˀda:w star ʔa: ʔa: ʔa: father’s younger sister ja: ˀja: ˀjɯ: medicine bin ban bin to fly tʰa: ta: ta: spread on; paint; daub kʰa: ka: ka: get stuck; stick 247 山崖 找 盖子 墙 尾巴 等 再 厚 仰望 抬头 纸 门 外公 鱼 眼睛 死 黄瓜 ～雨伞 婴儿背带 开 花～ 星星 药 飞 涂～色 ～漆 Column A Language A1/2 Thai Debao-1 Wuming Gloss -- pa: pa: wife kʰwa:j ʋa:j wa:j buffalo 水牛 na: na: na: rice field ma: ma: -- to come mɨ: mʊŋ fɯŋ hand 手 kʰwa: ɬɔ: kwa: right side hu: kʰjow ɣɯ: ear fon pʰɔn pan to sharpen; to grind kʰom kʰam ham bitter tʰɨŋ tʰaŋ taŋ to arrive 到 ～达 想～ tʰiəŋ tʰi:ŋ tiəŋ field pavilion 亭 棚子 tɛ:n tʰe:n tin hornet; wasp 248 妻子 水田 稻田 来 右 耳朵 磨～刀石 苦 味道～ 黄蜂 ～尖 Column B Language B1-A B1-C B1-U Thai Debao-1 Wuming Gloss pʰa: pʰa: pa: chop, split kʰa: kʰa: -- galangal kʰaj kʰjaj kjaj egg 蛋 kʰaw kʰaw ho: knee ha:ŋ ha:ŋ ha:ŋ few and far between 稀疏 hom ham hom to cover up tʰa:n tʰa:n ta:n charcoal ma: ma: -- soak nɨəj nu:j na:j tired wa:n ʋa:n wa:n to snow, to scatter ～米 saŋ ɬaŋ θaŋ to blow one’s nose 擤～鼻涕 soŋ ɬɔ:ŋ θoŋ to send, conduct si: ɬej θej four pa: pa: -- forest, trees kaj kaj kaj chicken 鸡 taw taw -- turtle , tortoise tam tam tam low kɛ: ke: ke: old 老 paw paw po: to blow 吹 -- tʰaw taw to hunt 打猎 klɔŋ kʰjɔ:ŋ ɣoŋ cage, box 笼子 ba: ˀba: ˀba: shoulder da: ˀda: ˀda: to scold 骂 ba:w ˀba:w ˀba:w young man 男青 da:ŋ ˀdaŋ ˀdaŋ potash, lye, alkaline salt ju: ˀjow ˀjaw to be exist, on, at, in 在 ʔɔ:n ʔo:n ʔo:n tender, soft tʰa: ta: ta: river (NT, CT), wharf (SWT) pʰɔ: po: po: father, male kʰu: kow kow 开 南姜 良姜 膝盖 盖～被子 碳 ～米 累 衰弱 抛～秧 送 四 森林 龟 矮 B1-UC B1-G B2 249 肩膀 帅 软 父亲 性 pair (chopsticks, shoes) 双 ～鞋 对 碱水 ～筷子 河 Column B Language Thai Debao-1 Wuming Gloss raj ɹ̥aj ɣej farm, dry farmland, upland field 旱田 la:ŋ B1/2 la:ŋ ɣiəŋ 旱地 under (SWT), the ground floor 牛栏 老房子的第 容易 层 ŋa:j ŋa:j ŋa:j easy naw naw naw rot tʰua tʰu: tu: bean 豆 tʰi: tʰej tej thick, dense -- tʰow taɰ chopsticks kʰi: kʰwej kɯj to ride on (horse, bicycle) kʰɨəŋ kʰy:ŋ ha:ŋ young (chicken) 鸡膥 tʰuəŋ tʰu:ŋ tuəŋ pull down 往 tʰɨən tʰɨ:n (LC) tɯ:n wild 野外的、野性的 250 腐烂 浓密 频繁 筷子 扯 骑 小母鸡 Column C Language Thai Debao-1 Wuming pʰa: pʰa: -- Gloss cloth, a sheet 云 kʰa: C1-A C1-C kʰa: ka: to kill tʰa: ɕa: to wait, if 等候 ha: ha: ha: five kʰaj kʰjaj -- sick, fever tʰa: 病怏怏 发烧 to walk 走 haj hɔj haɰ to give 给 na: na: na: face -- ma: ma: to grow up, to branch out生长 ja: ŋja: ɳɯ: grass -- to swim 游泳 θaj intestine 肠子 ʋa:j wa:j ɬaj som ɬam pa: pa: θom 行 脸 sour 草 酸 aunt (parents’ older sister) pa: kla: kja: kja: 姨 父或母的 seedling 秧苗 ka:w kaw kow nine ka:ŋ ka:ŋ ka:ŋ fishbone 九 鱼刺 鱼骨头 taŋ taŋ to erect, establish ta:j tɔj -- south, below klaj kʰjɔj kjaɰ near kɔ:n kʰo:n kon(Du’an) ˀwa: ʔa: ʔa: ˀba:n ba:n ˀdaj da:j ʔɔ:j ʔo:j ˀba: 面 Piece (CLF for stone) crazy ʔa: 竖起来 近 石头的量词 ba: 67 五 pla:j C1-UC C2 杀 pʰja:j taŋ C1-G ～布条 -- saj C1-U ～云 片状的量词 ～石头 疯 to open 张开 ~嘴巴 ˀba:n ˀdaj village 村 to get 得到 ʔo:j sugarcane ˀdaŋ to make fire, build fire 甘蔗 daŋ (Isan) ˀdaŋ sa-pʰaj pɔj paɰ bride, daughter-in-law新娘 tʰa:w taw tɯŋ crutch 67 (Longsang) tʰa:wC2 in Thai means ‘foot’. 251 生火 拐杖 媳 Column C Language Thai Debao-1 Wuming Gloss kʰiəw ke:w ke:w to chew ma: ma: ma: horse 马 na:m nam ram water 水 ma:j maj faj tree, wood ra:j ɹa:j ɣa:j bad, very much haw rice kʰej haj excrement pʰow pow kʰa:w kʰi: pʰu: kʰaw 米 树 木头 烂，厉害，非常 饭 大便 person (CLF for person) 人或 C1/2 嚼，咀嚼 个群体的量词 ～人 ～族 tʰuəj tʰu:j tɯ:i (Xilin) bowl su: ɬow ɕow to fight, to be against (SWT); to load, 碗 to carry (CT, NT) 装，盛，放 252 Column DL Language Thai Debao-1 Wuming kʰa:t kʰa:t ka:t Gloss to be broken, worn out 断线～ DL1-A 烂衣服～ to carry at both ends of a pole ha:p tʰa:p ɣa:p (na:C1) pʰa:k pʰja:k pla:k 挑 forehead -- to entrust 寄，交付 fɨət (Tianlin) astringent 苦涩 kʰe:k he:k guest kʰu:p ho:p fa:k fa:t kʰɛ:k kʰuəp pʰa:k pʰa:t 头 客人 circumference, anniversary ～期 周～ ma:k ma:k ma:k fruits 果 ŋɨək ŋy:k hɯ:k gums 齿龈 nu:k nuk deaf 聋 听不见 ɬa:p θa:p cockroach 蟑螂 sa:k pestle to smart, pierchingly painful nuək DL1-C DL1-U sa:p sa:k ɬa:k sɛ:p ɬe:p θe:p pa:k pa:k pa:k 辣痛，刺痛 mouth 嘴 kɔ:t ko:t ko:t to hug 搂抱 pi:k pək -- wing 翅膀 tɔ:k to:k to:k to pound, to hammer in 捶打 ta:t ta:t ta:t cliff, waterfall 悬崖 ka:t ka:t ka:t mustard plant ta:k tʰa:k ta:k tɛ:k tʰe:k te:k to be broken 碎裂 芥菜 阳 tʰo:k -- thin bamboo-stripes竹篾，竹条 ba:t ˀba:t ˀba:t scar, time dɔ:k ʔa:p dɛ:t ˀdu:t ˀdo:k ʔa:p ˀde:t ˀda:t hot -- flower 花 ʔa:p to bath 洗澡， ˀde:t 热 sunshine 阳光 水，游泳 晴 to want, to be hungry 饿 肚～ ta:k ta:k dry blood sucker 蟥 tɕʰɨək tɕy:k ɕa:k rope -- pa:k pa:k crazy 疯，傻 lɨət lu:t lɯət blood 血 tʰa:k ˀja:k 次数 ʔiək ja:k DL2 瀑布 tɔ:k dɨət DL1-G 木杵 to expose under the sun 晒～ DL1-UC 寄付 253 绳 旱 Column DL Language Thai Debao-1 Wuming Gloss ra:k la:k ra:k root ma:k ma:k ma:k full, plenty mi:t mət mit knife tʰu:k tʰʊk tɯk pʰɨək pʰa:k pu:k DL1/2 254 根 饱满 匕首 to hit the mark, right, cheap 对 正确，击中 white 白色，苍白 Column DS Language Thai Debao-1 Wuming Gloss pʰak pʰjak plak vegetable pʰat -- hot, pungent kʰɔ:t kut pʰet kʰot DS1-A DS1-C DS1-U DS1-UC 辣 卷曲 偏差 口音~ het hat -- to do 做 hap hap hap to shut (tɕa-) kʰep (kaj-) kʰjap θip centipede 蜈 -- hailstone 冰雹 关闭 hep tʰap mat mat mat flea 跳 nak nak nak heavy ŋok ŋak ŋak nod, shaking, shaky lap lap lap 重 点头 to be late, get to sleep wat wat -- 黑 闭眼 have a cold 感冒 sak ɬak θak color kop kap kop frog 青蛙 kat kat kat to bite, to be very cold tap tap tap liver tak tak tak 色 田鸡 肝脏 舀～饭 打～饭 刺，插，熬煮 pak pak to stab pet pat pit duck tak tʰak tak 鸭 grasshopper, cricket tok tʰɔ:k tok 蚱蜢 蝗虫 to seed, to sow dap ˀdap ˀdap to extinguish ʔok dip ʔet ˀbat ʔak ˀdəp ˀdak ʔak ˀdip ˀdak fishhook chest 播种 灭 鱼钩 胸部 unripe, raw deep night 生～肉 睡着 深夜 ʔit one pak pak very tired, rest kʰrok kjɔ:k -- mortar rap ɹ̥ap -- to receive, to welcome me mat nat pʰak ʔat ˀbit 十 的“ ” 累 臼 石～ a grain (CLF) 粒 255 啃咬 scoop (rice, alcohol) pak dɨk DS2 蔬菜 to coil, coiled up, twisted 弯曲 bet DS1-G 菜 颗 量词 ～米 接 Column DS Language Thai Debao-1 Wuming Gloss mot mɔ:t mot ant lak lak ɕak to steal nok nɔ:k kʰop kʰap sip suk DS1/2 sap ɬəp ɬʊk ɬap ɣok bird hap to bite ɕip ten 十 偷 鸟 咬 ɕuk cooked, ripe 熟～肉 sap mince 砍碎 256 剁碎 APPENDIX C LANGUAGE INFORMATION AND TONE DATA ANALYSIS FROM 42 TAI VARIETIES The naming principle of Tai varieties investigated follows “location name + language name”, such as Bangkok Thai (L1) and Khon Kaen Isan (L3). The language names are all individual language names in the list of ISO 639-3, or a name given by me in the case that the language has not been listed in ISO 639-3 such as Muong Quan Son Tai (L7, “Muong” as a location and “Quan Son Tai” as a language name). For distinguishing some language varieties of the same individual language in the same region, the segment of location can be separated into two parts, a upper level and a subordinate level, such as Debao Dalong Yang Zhuang (L9, “Debao” as a location name at the county level, “Dalong” as a village in Debao County, and “Yang Zhuang” as an individual language name listed in ISO 639-3) and Tiandeng Xiangdu Zuojiang Zhuang (17, “Tiandeng” as a county level location, “Xiangdu” as a township in Tiandeng County, and “Zuojiang Zhuang” as an individual language name listed in ISO 639-3). In the text of this thesis, the language names can be abbreviated as the individual language names such as Thai (L1), and the location names such as Debao Dalong (L9). Tai varieties investigated in this thesis and their numbering are as follows. 1. Seven from SWT (L1-7): 1. Bangkok (Thai, Language 1, abbreviated as L1 and similarly hereinafter) 2. Songkhla (Southern Thai, L2) 3. Khon Kaen (Isan, L3) 4. Chiang Rai (Northern Thai, L4) 5. Kho Lam (Shan, L5) 6. Rong Maet (Tai Lue, L6) 7. Muong (Quan Son Tai, L7) All the data of these languages are collected by me. 2. Seventeen from CT (L8-L24): 8. Debao Urban (Yang Zhuang, L8) 9. Debao Dalong (Yang Zhuang, L9) 257 10. Debao Lüliu (Yang Zhuang, L10) 11. Jingxi Hurun (Yang Zhuang, L11) 12. Jingxi Urban (Yang Zhuang, L12) 13. Jingxi Anning (Yang Zhuang, L13) 14. Debao Nalong (Myang Zhuang, L14) 15. Debao Tuoxin (Pyang Zhuang, L15) 16. Jingxi Huashan (Min Zhuang, L16) 17. Tiandeng Xiangdu (Zuojiang Zhuang, L17) 18. Daxin Naling (Zuojiang Zhuang, L18) 19. Jingxi Lingding (Zuojiang Zhuang, L19) 20. Xiaoguangnan (Nong Zhuang, L20) (Zhang et al. 1999) 21. Wenshan Dazhai (Dai Zhuang, L21) (Zhang et al. 1999) 22. Daxin Leiping (Zuojiang Zhuang, L22) (Hudak 2008, Zhang et al. 1999) 23. Daxin Baoxu (Zuojiang Zhuang L20) 24. Bac Va (Nung, L24) (Hudak 1997) Most of them are collected by myself except those with reference, but some of them are also compared with data of the same languages in published resources. Nong Zhuang, Dai Zhuang, Leiping, and Bac Vac Nung are referred to published resources. 3. Five from YN (L25-29): 25. Yongning Baiji (L25) 26. Nanning Shuangding (L26) 27. Long’an Xiaolin (L27) (Zhang et al. 1999) 28. Jingxi Daqiu (Nung An, L28) 29. Yongning Xialeng (L29) (Zhang et al. 1999) YN varieties in this paper are conventionally treated as dialects of Southern Zhuang (SZ) which is traditionally grouped into CT, because they have plenty aspirated initials which condition possible secondary tonal splits as the same to those in CT. But at the same time they present more NT features on voicing alternation, vocalic behaviors, and some of the initial behaviors. Baiji, Shuangding and Nung An are collected by me; Long’an Yongnan Zhuang and Xialeng Yongnan Zhuang are referred to published resources. 4. Thirteen from NT (L30-42): 30. Wuming Shuangqiao (Yongbei Zhuang, L30) 31. Debao Longsang (undescribed, L31) 32. Debao Dongling (Youjiang Zhuang, L32) 33. Tianlin Lizhou (Guibian Zhuang, L33) (Zhang et al. 1999) 34. Shanglin Yunling (Central Hongshuihe Zhuang, L34) 35. Du’an Jiaren (Central Hongshuihe Zhuang, L35) 36. Yizhou Suogan (Guibei Zhuang, L36) 258 37. Liujiang Baipeng (Liujiang Zhuang, L37) (Zhang et al. 1999) 38. Donglan Urban (Guibei Zhuang, L38) (Zhang et al. 1999) 39. Qiubei Gehan (Qiubei Zhuang, L39) (Zhang et al. 1999) 40. Lianshan Xiaosanjiang (Lianshan Zhuang, L40) (Zhang et al. 1999) 41. Huishui (Bouyei, L41) 42. Zhenning (Bouyei, L42) Most of them are collected by me; some of them are also compared with the data of the same languages in published resources. Tianlin, Donglan, Liujiang, Qiubei, and Lianshan are referred to published resources. The following contents concludes the language information obtained from LRPs and the results of tone data analysis. The first chart under the third item (in all languages recorded by me) or under the second item (in all languages referenced to published sources) “Basic patterns of tonal development of this dialect” demonstrates the basic patterns of tonal splits and tonal mergers of that variety. For example, the first chart under ‘1.3’ of L1 and 4.3 of L4 demonstrate the tone boxes or the basic patterns of tonal splits and tonal mergers of Bangkok Thai and Chiang Rai Northern Thai respectively. In those charts, boxes shaded by the same color indicates that they are of the same toneme or tones in complementary distribution in smooth and checked syllables respectively. Basically, in the same vertical column, register tones with higher pitch are in warm colors (progressive increasing by yellow, orange, red, crimson and dark red from Columns A - DS respectively), while register tones with lower pitch are in cool colors (cyan, green, blue, mazarine, and dark blue from Columns A to DS respectively). However, because there must be tonal mergers between different tones in a Tai variety, all tonemes of the same tone must be in the same color with their allotone in the very first former box from higher rows to lower row and from left to right columns. Therefore, the dark red and dark blue are actually never found in any of the tone boxes because all tones on dead syllables must be an allotone of a tone on smooth syllables. For example, in 1.3 (L1 or Bangkok Thai), the orange color has been used to mark the B2 tone first in Column B, and its following allotones C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and DL2 have to be marked in the same orange color, in spite of that they should be marked in blue and mazarine respectively when being compared with the other register tones in the same columns. Note: I got permission from all the LRPs to open their biodata in my thesis which will be able to be downloaded online. 259 1. Language 1 Bangkok, Thai, SWT 1.1 Language information Language Thai ISO 639-3 Code tha Autonym Thai Exonym Central Thai; Standard Thai; Siamese Data Source 1 LRP Variety Name Bangkok, Thailand. 1.2 Information of data source Serial number of LRP LRP1 Date Place of data elicitation Betta House, Nimmanhaemin 5, Chiang Mai, Thailand. Wirawit Given Name November 24, 2013 Family Name วรวชญ์ Gender Phonwattanakaisert พรวฒน กร ลศ male Age 27 Birth Place Sathon District, Bangkok, Thailand. Places living Before he moved to Chiang Mai, he had been living in Bangkok from his birth. Time period of living at the present residence Occupation physician 3 years Education Languages spoken Level Bangkok Thai Mother Tongue English Intermediate bachelor Language Remarks The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in Bangkok. He speaks Bangkok Thai and Teochew. LRP’s mother was born in Bangkok. She can speak Bangkok Thai, Teochew and English. LRP’s parents are living together in Bangkok now. Remarks Although the LRP was born in a Chinese Teochew family, Bangkok Thai is his only mother tongue. In Thailand, particularly in Bangkok, Teochew is still spoken among older ethnic Chinese Thai citizens; however, the younger generation tends to learn Standard Chinese as a third language after Thai and English, and Teochew seems to has become an endangered language due to the inheriting gap. 260 1.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, tonal flip-flop runs through all tones except tone A. Smooth Syllable Checked Syllable Phonation types A B C DL DS Proto- Aspirations A1-A: 24 B1-A: 21 C1-A: 41ʔ DL1-A: 21 DS1-A: 21 voice- Continuants A1-C: 24 B1-C: 21 C1-C: 41ʔ DL1-C: 21 DS1-C: 21 less Unaspirated stops A1-U: 33 B1-U: 21 C1-U: 41ʔ DL1-U: 21 DS1-U: 21 A1-UC: 33 B1-UC: 21 C1-UC: 41ʔ DL1-UC: 21 DS1-UC: 21 A1-G: 33 B1-G: 21 C1-G: 41ʔ DL1-G: 21 DS1-G: 21 A2: 33 B2: 41ʔ C2: 453ʔ DL2: 41 DS2: 45/55 A1/2: 24 B1/2: 21 C1/2: 41ʔ DL1/2: 21 DS1/2: 21 Sounds Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are summarized as in the following two charts. Vertically: Two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1/2] [1-U, 1-UC, 1-G, 2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1/2, 1-U, 1-UC, 1-G] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. The nonstraightforward column is A, which has a split between A1-C and A1-U. Horizontally: one individuals and four mergers Individual A[1-A, 1-C, 1/2] A[1-U, 1-UC, 1-G, 2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger B[2] = C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[2] C[2] = DS[2] 261 1.4 Tonal characteristics & tone occurrences: five distinct tones Tone 1: Low-rising 24 occurs in A[1-A, 1-C, 1/2]. Tone 2: Mid level 33 occurs in A[1-U, 1-UC, 1-G, 2]. Tone 3: Low-falling 21 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2], as well as in DL[1A, 1-C, 1-U, 1-UC, 1-G, 1/2] and DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: High-falling 41 occurs in C[1-U, 1-UC, 1-G, 2] and in DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-rising-falling 453 occurs in C[2], and high-rising 45 or high level 55 occurs in DS[2] to form a complementary distribution in smooth and checked syllables respectively. 1.5 Phonation types or voice qualities of tones in this dialect: Tones C1 (as well as B2 which merges into C1) and C2 have glottal constrictions, like tʰa:41ʔ ‘wharf’, ha:41ˀ ‘five’, fa:j41ˀ ‘cotton’, sɨə41ˀ ‘shirt’, ka:ŋ41ˀ ‘fishbone’, tʰa:w453ˀ ‘foot’, and na:m453ˀ ‘water’. All the glottal constrictions are not phonemic. 1.6 Sound changes involving tones reflected in the wordlist of this dialect: 1) The term ja:33 ‘medicine’ is reflected as having a low register consonant j- (< *j-) by the Thai orthography ยา, but it should have had a high register consonant j(< *ʔj-) in Row 1-G due to the data comparison with other Tai varieties. 2) The term kʰa:41ʔ ‘to kill’ is reflected as having a low register consonant kʰ- (< *g) by the Thai orthography ฆ຋า, but it should have had a high register consonant kʰ- (< *q- (cf. Pittayaporn 2009: 105, 357)) in Row 1-G due to the data comparison with other Tai varieties. 262 2. Language 2 Songkhla, Southern Thai, SWT 2.1 Language information Language Southern Thai ISO 639-3 Code sou Autonym Pak Tai Exonym Dambro Data Source 1 LRP Variety Name Muang District, Songkhla Province, Thailand. 2.2 Information of data source Serial number of LRP LRP1 Date December 23, 2013 Place of data Betta House Café, Nimmanhaemin Soi 5, Chiang Mai Province, Thailand. elicitation Given Name non³³ tʰaˀ³³ wat45 นนทวัช Family Name kə:t21 tʰip45 เกิดทิพย์ Gender male Age 27 Birth Place Muang District, Songkhla Province, Thailand. Places lived When he was 22 years, he moved to Chiang Mai from his hometown. Time period of living at the present residence 5 years in Chiang Mai Occupation Personal business Education bachelor Languages spoken Level Language Remarks Southern Thai Mother Tongue Standard Thai Fluent This language is the national language of the LRP’s country. English Intermediate Mandarin A little This language is used for his personal business when he makes contact with his Chinese customers. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in Muang District of Songkhla Province. His mother tongue is Songkhla Southern Thai. He also speaks Standard Thai. LRP’s mother was born in Nakhon Si Thammarat Province. Her mother tongue is Nakhon Si Thammarat Southern Thai. She also speaks Standard Thai. 263 2.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there are no tonal flip-flops in this dialect. Phonation types Checked Syllable A B C DL DS Aspirations A1-A: 41 B1-A: 41 C1-A: 55 DL1-A: 55 DS1-A: 55 Continuants A1-C: 41 B1-C: 41 C1-C: 55 DL1-C: 55 DS1-C: 55 Unaspirated stops A1-U: 454 B1-U: 454 C1-U: 44 DL1-U: 44 DS1-U: 45 A1-UC: 454 B1-UC: 454 C1-UC: 44 DL1-UC: 44 DS1-UC: 45 A1-G: 454 B1-G: 454 C1-G: 44 DL1-G: 44 DS1-G: 45 A2: 31 B2: 213 C2: 22 DL2: 213 DS2: 22 A1/2: 41 B1/2: 41 C1/2: 55 DL1/2: 55 DS1/2: 55 ProtoVoiceless Sounds Smooth Syllable Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: three way splits Tonal category Register 1 Register 1' Register 2 A [1-A, 1-C, 1/2] [1-U, 1-UC, 1-G] [2] B [1-A, 1-C, 1/2] [1-U, 1-UC, 1-G] [2] C [1-A, 1-C, 1/2] [1-U, 1-UC, 1-G] [2] DL [1-A, 1-C, 1/2] [1-U, 1-UC, 1-G] [2] DS [1-A, 1-C, 1/2] [1-U, 1-UC, 1-G] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. Although the splits between voiceless and voiced sounds are straightforward, but there is another secondary split between voiceless frictions (aspiration and continuants) and voicelss unaspiration (unaspirated stops, unaspriated stop + *r, and glottal sounds) to make the third register split in all the five tonal categories. Horizontally: one individuals and six mergers Individual A[2] A[1-A, 1-C, 1/2] = B[1-A, 1-C, 1/2] A[1-U, 1-UC, 1-G] = B[1-U, 1-UC, 1-G] = DS [1-U, 1-UC, 1-G] Merger B[2] = DL[2] C[1-A, 1-C, 1/2] = DL[1-A, 1-C, 1/2] = DS [1-A, 1-C, 1/2] C[1-U, 1-UC, 1-G] = DL[1-U, 1-UC, 1-G] C[2] = DS[2] 264 2.4 Tonal characteristics & tone occurrences: seven distinct tones Tone 1: High-falling 41 occurs in A[1-A, 1-C, 1/2] and B [1-A, 1-C, 1/2]. Tone 2: High-rising-falling 454 occurs in A[1-U, 1-UC, 1-G] and B[1-U, 1-UC, 1-G], as well as high-rising 45 in DS [1-U, 1-UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Mid-falling 31 occurs in A[2]. Tone 4: Low-falling-rising 213 occurs in B[2] and DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High level 55 occurs in C[1-A, 1-C, 1/2], as well as in DL[1-A, 1-C, 1/2] and DS[1-A, 1-C, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 6: High level 44 occurs in C[1-U, 1-UC, 1-G] and in DL[1-U, 1-UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 7: Low level 22 occurs in C[2] and DS[2] to form a complementary distribution in smooth and checked syllables respectively. 2.5 Phonation types or voice qualities of tones in this dialect: There are no phonation types of voice qualities found in tones in this variety. 265 3. Language 3 Khon Kaen, Isan, SWT 3.1 Language information Language Isan ISO 639-3 Code tts Autonym Lao Exonym Isan; Northeastern Thai; Lao Data Source 1 LRP Variety Name Chun Phae, Khon Kaen Province, Thailand. 3.2 Information of data source Serial number of LRP LRP1 Date October 12, 2013 Place of data elicitation tʰuŋ⁴¹ luəŋ³³ tʰɔ:ŋ³³ village, Chun District, Phayao Province, Thailand. Given Name na:j³³ tʰaj³³ Family Name kʰiəw²⁴ mi:³³ Gender male Age 61 Birth Place Muang District, Pitsanulok Province, Thailand. Places lived When he was 3 months, he moved to Chum Phae District, Khon Kaen with his mother, and lived there for 10 years until 1962 (2505). After that he left to Rong Maet Village (next to the village of the present residence) in Chun District of Phayao Province which was a district of Chiang Rai Province at that time for four years. After that he moved to the present residence until now. Time period of living at the present residence Occupation farmer 51 years Education First year of senior high school Languages spoken Level Remarks Isan Mother Tongue Northern Thai Fluent This language is the lingua franca in the LRP’s area. Standard Thai Fluent This language is the national language of the LRP’s country. Thai Lue familiar LRP is familiar with Tai Lue since it is spoken in the next village Rong Maet. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in That village, Phu Khiaw District, Chai Ya Phum Province. He probably can speak Isan and Thai, and dead before he left Khon Kaen. LRP’s mother was born in Sam Phan village, Khua Liang Khet District, Chum Phae District, Khon Kaen Province. She can speak Isan and Thai. LRP’s mother is living with his younger brother and younger sister in Chaiya Phuek Village, Don Sila Sub-district, Wiang Chai District, Chiang Rai Province now. Remarks Although the LRP lives in Chun District of Phayao Province where local Isan/Lao speakers live, he insists his Isan variety is of Chum Phae, Khon Kaen Province. 266 3.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, tonal flip-flop runs through all tones. Smooth Syllable Phonation types A B C DL DS Aspirations A1-A: 23 B1-A: 33 C1-A: 21ʔ DL1-A: 21 DS1-A: 23 Continuants A1-C: 23 B1-C: 33 C1-C: 21ʔ DL1-C: 21 DS1-C: 23 Unaspirated stops A1-U: 23 B1-U: 33 C1-U: 42ʔ DL1-U: 21 DS1-U: 23 A1-UC: 23 B1-UC: 33 C1-UC: 42ʔ DL1-UC: 21 DS1-UC: 23 A1-G: 23 B1-G: 33 C1-G: 42ʔ DL1-G: 21 DS1-G: 23 A2: 45 B2: 33 C2: 42ʔ DL2: 42 DS2: 45/55 A1/2: 23 B1/2: 33 C1/2: 21ʔ DL1/2: 21 DS1/2: 23 ProtoVoiceless Sounds Checked Syllable Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits and no splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] C [1-A, 1-C, 1/2] [1-U, 1-UC, 1-G, 2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. The nonstraightforward columns are B and C. The column B has no spilts at all. The column C have a unique two wait split: voiceless aspirations, voicelss continuants and voicing alternation serials on one register, and voiceless unaspirations, voicelss unaspirated stops + *r-, golttal sound, and voiced register on the other register. The merger that golttal sound goes with voiced sounds on C tone is often found in NT varieties, but is very rare found in Tai varieties developing from ST (CT/SWT). Northeastern Thai as well as its most closed Lao varieties represent this unique pattern of tonal split and merger. Horizontally: one individuals and four mergers Individual B[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger A[2] = DS[2] C[1-A, 1-C, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[1-U, 1-UC, 1-G, 2] = DL[2] 267 3.4 Tonal characteristics & tone occurrences: five distinct tones Tone 1: Low-rising 23 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] as well as in DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 2: High-rising 45 occurs in A[2] and high-rising 45 or high level 55 in DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Mid level 33 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2]. Tone 4: Low-falling 21 occurs in A[1-A, 1-C, 1/2] and in DL[1-A, 1-C, 1-U, 1-UC, 1G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-falling 42 occurs in C[1-U, 1-UC, 1-G, 2], and in DL[2], to form a complementary distribution in smooth and checked syllables respectively. 3.5 Phonation types or voice qualities of tones in this dialect: Tones C1 and C2 have glottal constrictions, like ha:21ˀ ‘five’, fa:j21ˀ ‘cotton’, siə²¹ˀ ‘shirt’, ka:ŋ41ˀ ‘fishbone’, da:m41ˀ ‘handle’, and nam41ˀ ‘water’. All the glottal constrictions are not phonemic. 3.6 Sound changes involving tones reflected in the wordlist of this dialect: 1) The term muan42 ‘to gargle’ has an unexpected initial consonant m-, which is expected to be b- of the glottal sounds row (C1-G) in the tone box designated, vs. buan C1-G ‘to gargle’ in Thai and ʔunC1-G ‘to gargle’ in Debao Yang Zhuang. However, its tone is kept in the C1-G row. 3.7 Other linguistic changes reflected in the wordlist of this dialect: 1) The term ba:w 33 ‘kite’ has an unexpected initial consonant b-, which is expected to be w- of the voiced row (B2) in the tone box designated. 268 4. Language 4 Chiang Rai, Northern Thai, SWT 4. 1 Language information Language Northern Thai ISO 639-3 Code nod Autonym kamA2 mɨəŋA2 Exonym Tai Yuan; Lanna Data Source 1 LRP Variety Name Chiang Mai, Thailand. 4.2 Information of data source Serial number of LRP Place of data elicitation LRP1 Date December 12, 2013 Mae Tam Tai Village, Mae Tam Sub-district, Phaya Meng Rai District, Chiang Rai Province. Given Name Nawatchai นวตรชย Family Name Kabchan ก บจนทร์ Gender male Age 33 Birth Place Mae Tam Tai Village, Mae Tam Sub-district, Phaya Meng Rai District, Chiang Rai Province. บ น มต๋ ต ต. มต๋ อ.พญ ม็งร ย ชยงร ย Places lived He moved from Mae Tam Tai Village to the main town of Phaya Meng Rai District when he was 13. He went to Bangkok to study in Ramkhamhaeng University for 4 years. After graduating, he went back to his hometown to live with his parents until now. Time period of living at the present residence Occupation 20 years Personal business Languages spoken Level Northern Thai Mother Tongue Standard Thai Near Native Education bachelor Language Remarks This language is the national language of the LRP’s country. English Intermediate The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents were both born in Mae Tam Tai Village, Mae Tam Sub-district, Phaya Meng Rai District, Chiang Rai Province. His parents speak Northern Thai as their mother tongue and speaks Standard Thai to outlanders. They both live together with the LRP now. 269 4.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, tonal flip-flop runs through all tones. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 24 B1-A: 22 C1-A: 44ʔ DL1-A: 22 DS1-A: 24 Continuants A1-C: 24 B1-C: 22 C1-C: 44ʔ DL1-C: 22 DS1-C: 24 Unaspirated stops A1-U: 24 B1-U: 22 C1-U: 44ʔ DL1-U: 22 DS1-U: 24 A1-UC: 24 B1-UC: 22 C1-UC: 44ʔ DL1-UC: 22 DS1-UC: 24 A1-G: 335 B1-G: 22 C1-G: 44ʔ DL1-G: 22 DS1-G: 24 A2: 335 B2: 31 C2: 53ʔ DL2: 31 DS2: 35 A1/2: 24 B1/2: 22 C1/2: 44ˀ DL1/2: 22 DS1/2: 23 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1/2] [1-G, 2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. The nonstraightforward column is A, which has a split between A1-UC and A1-G. Horizontally: two individuals and four mergers Individual C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] A[1-A, 1-C, 1-U, 1-UC, 1/2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger A[G, 2] = DS[2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] B[2] = DL[2] 270 4.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Low-rising 24 occurs in A[1-A, 1-C, 1-U, 1-UC, 1/2] as well as in DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 2: Mid level-rising 335 occurs in A[1-G, 2] and mid-rising 35 in DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Low level 22 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Mid-falling 31 occurs in B[2] and DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High level 44 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 6: High-falling 53 occurs in C[2]. 4.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have glottal constrictions, like ha:55ˀ ‘five’, fa:j55ˀ ‘cotton’, sɨə55ˀ ‘shirt’, ka:ŋ55ˀ ‘fishbone’, paj53ˀ ‘bride’, and nam53ˀ ‘water’. All the glottal constrictions are not phonemic. 271 5. Language 5 Kho Lam, Tai Yai, SWT 5.1 Language information Language Shan ISO 639-3 Code shn Autonym Kam Tai Exonym Shan; Tai Yai Data Source 1 LRP Variety Name Kho Lam Village, Taunggyi City, Shan State, Myanmar 5.2 Information of data source Serial number of LRP LRP1 Date October 12, 2013 Place of data elicitation Betta House, Nimmanhaemin 5, Chiang Mai, Thailand. Given Name Lungta Family Name Lungkham Gender male Age 50 Birth Place kʰo:24 lam24 ( ขวด๋ in Thai) Village, Taunggyi District, Shan State, Myanmar. Places lived Before he moved to Chiang Mai, Thailand when he was 42, he only stayed in his village. Time period of living at the present residence Occupation Construction 8 years Education He has never gone to a school. builder Languages spoken Level Shan Mother Tongue Thai Fluent Language Remarks This language is the national language of Thailand, in where the LRP is working. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents were born in the same village of LRP’s. They spoke Shan and Thai. Both of them passed away. Remarks There are many Shan migrant workers in Thailand, especially in Chiang Mai which is the largest city in Northern Thailand and is relatively closed to their hometowns in Shan State of Myanmar. Most of them grow up in their villages, and come to Thailand for their migrant work when they are in teen-age youths. Although Shan is their mother tongue, most of them also speak Standard Thai as their main occupational language. Most of them have a chance to learn Thai in the vocational schools built by their communities in some cities in Thailand. It is not clear that Kho Lam is of what level of the administration of Myanmar. The LRP claimed that Kho Lam was a single village, and he did not know the relationship of administrative subordination 272 between Kho Lam and other cities and districts in Shan State of Myanmar. However, the map of Myanmar shows that Kho Lam is a transportation junction within Taunggyi District. According to Wikipedia, states and divisions in Myanmar are divided into districts (ခရု င;် kha yaing or khayaing, IPA: [kʰəjàɪɴ]). These districts consist of townships (မ န ့ ယ်; myo-ne, IPA: [mjo̰nɛ̀]) that include towns (မ ့; myo, IPA: [mjo̰]), wards (ရပ်ကွက်; yatkwet, IPA: [jaʔ kwɛʔ])) and village-tracts (ကက းရွ ာအုပ်စု; kyayywa oksu, IPA: [tɕ́ jwà ʔoʊʔ sṵ]). Village-tracts are groups of adjacent villages (ကက းရွ ာ; kyayywa, IPA: [tɕ́ jwà]). Nevertheless, the LRP did not provide any clear information that what level of these divisions Kho Lam should be of. It reflects that the common Shan people are not entitled to a regular or favorable education system, which would have given them the necessary knowledge of the administrative subordination of their own regions. 5.3 Basic patterns of tonal development of this dialect 1) As shown in the chart above, tonal flip-flop runs through all tones. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 24 B1-A: 11 C1-A: 33ʔ DL1-A: 11 DS1-A: 55 Continuants A1-C: 24 B1-C: 11 C1-C: 33ʔ DL1-C: 11 DS1-C: 55 Unaspirated stops A1-U: 24 B1-U: 11 C1-U: 33ʔ DL1-U: 11 DS1-U: 55 A1-UC: 24 B1-UC: 11 C1-UC: 33ʔ DL1-UC: 11 DS1-UC: 55 A1-G: 24 B1-G: 11 C1-G: 33ʔ DL1-G: 11 DS1-G: 55 A2: 45 B2: 33ʔ C2: 41ʔ DL2: 33 DS2: 41 A1/2: 24 B1/2: 11 C1/2: 33ʔ DL1/2: 11 DS1/2: 33 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of straightforward tonal development. 273 Horizontally: one individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] A[2] =DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]= DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger B[2] = C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[2] C[2] = DS[2] 5.4 Tonal characteristics & tone occurrences: five distinct tones Tone 1: Low-rising 24 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 2: High-rising 45 occurs in A[2] and High level 55 in DS[1-A, 1-C, 1-U, 1UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Low level 11 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] and DL [1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Mid level 33 occurs in B[2], C[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2], and DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-falling 41 occurs in C[2] and in DS[2] to form a complementary distribution in smooth and checked syllables respectively. 5.5 Phonation types or voice qualities of tones in this dialect Tones C1 (as well as B2 which merges into C1) and C2 have glottal constrictions, like pɔ:33ʔ ‘father’, ha:33ˀ ‘five’, fa:j33ˀ ‘cotton’, ka:ŋ33ˀ ‘fishbone’, paɰ33ˀ ‘bride’, and nam41ˀ ‘water’. All the glottal constrictions are not phonemic. 5.6 Linguistic changes reflected in the wordlist of this dialect 1) As in Tai Lue and Quan Son Tai which are of Group P of SWT (cf. Chamberlain 1975), this Shan variety reflect unaspirated stops in the low register (A2, B2, C2, DL2, DS2), to contrast to Group PH (including Thai, Southern Thai and Isan) (cf. §2.2.4). Comparing with Northern Thai, although both Shan and Northern Thai reflect the same unaspirated initials in B2, C2 and DL2, but different in A2, which allow aspirated in Northern Thai, e.g. Northern Thai kʰwa:jA2 ‘buffalo’, kʰonA2 ‘person’ and tʰajA2 ‘Thai’ vs. Shan kwa:jA2 ‘buffalo’, konA2 ‘person’ and tajA2 ‘Tai/Thai’. 2) PT *ʔb- and *ʔd- becomes w-/m- and l- in Shan in most cases, such as winA1 ‘to fly’ and liaŋA1 ‘red’. 274 6. Language 6 Rong Maet, Tai Lue, SWT 6.1 Language information Language Tai Lü ISO 639-3 Code khb Autonym tajA2 lɯ:C2 Exonym Thai Lue; Tai Le; Dai Le Data Source 1 LRP Variety Name Rong Maet, Chun District, Phayao Province, Thailand. 6.2 Information of data source Serial number of LRP LRP1 Date Place of data elicitation Rong Maet (rɔ:ŋ⁴¹ mɛt⁴¹) village, Chun District, Phayao Province Given Name sa:²⁴ taʔ²¹ nan³³ Family Name sɛŋ²⁴ si:²⁴ tɕan³³ Gender male Age 33 Birth Place Rong Maet village, Chun District, Phayao Province Places lived Muang District of Lam Pang Province (junior college 2 years). April 18, 2013 Bangkok (4 years) Seoul of South Korea (6 years). Chiang Mai (2 years) Time period of living at the present residence 2 years Occupation freelancer Education Junior College Languages spoken Level Tai Lü mother tongue Northern Thai native This language is the lingua franca in the LRP’s area. Standard Thai near native This language is the national language of the LRP’s country. Isan familiar LRP is familiar with Isan since it is spoken in the same village. Korean intermediate LRP is familiar with Korean since he used to live in South Korea Language Remarks for six years. He can also read and write Korean. English a little He can only read some written English. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. He can speak Tai Lue, Northern Thai, and Thai. LRP’s mother was born in the same village. She can speak Tai Lue, Northern Thai, and Thai. LRP’s parents live together in the same village now. Remarks LRP’s village Rong Maet is a multilingual village. Northern Thai, Tai Lue, and Isan are spoken in Moo 7, Moo, 4 and Moo 9 which are three different sub-villages of this village respectively. Northern and Lao speakers also live together in Moo 11 of the same village. Northern Thai is the LWC of this village, and is used in the conversation among speakers of the different aforementioned languages. 275 6.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there are no tonal flip-flops in this dialect. Smooth Syllable Phonation Checked Syllable types A B C DL DS Aspirations A1-A: 55 B1-A: 35 C1-A: 13ʔ DL1-A: 35 DS1-A: 45/55 Proto- Continuants A1-C: 55 B1-C: 35 C1-C: 13ʔ DL1-C: 35 DS1-C: 45/55 Voiceless Unaspirated A1-U: 55 B1-U: 35 C1-U: 13ʔ DL1-U: 35 DS1-U: 45/55 A1-UC: 55 B1-UC: 35 C1-UC: 13ʔ DL1-UC: 35 DS1-UC: 45/55 A1-G: 55 B1-G: 35 C1-G: 13ʔ DL1-G: 35 DS1-G: 45/55 B2: 33 C2: 11ʔ DL2: 33 DS2: 33 B1/2: 35 C1/2: 13ʔ DL1/2: 35 DS1/2: 45/55 stops Sounds Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds A2: 31/335 A1/2: 55 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of straightforward tonal development. Horizontally: three individuals and four mergers A[2] Individual C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] B[2] = DL[2] = DS[2] 276 6.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High level [55] occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2], as well as high level [55] or high-rising [45] in DS [1-A, 1-C, 1-U, 1-UC, 1G, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 2: Mid-falling [31] or its free variation mid level-rising [335] (probably due to Northern Thai influence) occurs in A[2]. Tone 3: Mid-rising [35] occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2], and in DL[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Mid level [33] occurs in B[2], as well as in DL[2] and DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low-rising [13] occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 6: Low level [11] occurs in C[2]. 6.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have glottal constrictions, like ha:13ˀ ‘five’, fa:j13ˀ ‘cotton’, siə13ˀ ‘shirt’, ka:ŋ13ˀ ‘fishbone’, da:m13ˀ ‘handle’, and nam11ˀ ‘water’. All the glottal constrictions are not phonemic. 6.6 Sound changes involving tones reflected in the wordlist of this dialect When the item with A2 tone pronounced individually, the tone has a mid-falling [31] contour, but when the item is on an utterance, it can preserve mid-falling [31], or change to a mid-level-rising contour [335]. The environment of this tone coarticulation needs more data to study with. 6.7 Other linguistic changes reflected in the wordlist of this dialect In this language, stops developed from earlier voiced stops mostly present as unaspirated, like ka:A2 ‘to stuck’ and ta:A2 ‘to spread on’. However, some items with expected unaspirated initial stops unexpectedly have an aspirated stop, like kʰunA2 ‘person’ provided by the LRP. The reason may be that some of his pronunciations are influenced by Standard Thai or Northern Thai, which have aspirated stops developed from earlier voiced stops, like kʰonA2 ‘person’ in Thai and Northern Thai. 277 7. Language 7 Quan Son Tai, SWT 7.1 Language information Language Quan Son Tai ISO 639-3 Code unknown Autonym tajA2 Exonym Tai Lam, Tai Dam, Black Tai, Phớ Tày, Kun Tày, Phú tày Data Source 1 LRPs Variety Name Muong (meaning: mango) Village, Quan Son District, Thanh Hoa Province, Vietnam. 7.2 Information of data source Serial number of LRP LRP1 Date November 18, 2015 Place of data elicitation Quan Son Town, Quan Son District, Thanh Hoa Province, Vietnam Given Name Anh Tuan Family Name Pham Gender male Age 32 Birth Place Muong village, Quan Son District, Thanh Hoa Province, Vietnam Places lived The Muong, Lam and Na Meo villages and Quan Son Town of Quan Son district, Thanh Hoa Province. Ha Noi (3 years). Hue city, Thua Thien Hue Province. Long Xuyen City, An Giang Province. Ho Chi Minh City . Bangkok Thailand (3 years) Time period of living at the present residence Occupation District Financial staff 9 years Education Master Languages spoken Level Language Remarks Quan Son Tai mother tongue Another Tày đắm Near native This language is his wife’s mother tongue. Vietnamese near native This language is the national language of the LRP’s country. Lao familiar LRP is familiar with Lao since his hometown is closed to the variety in Quan Son Vietnam-Laos border, and his ethnic group is also similar to Lao. Thai familiar LRP is familiar with Thai since he used to study for his postgraduation in Bangkok Thailand for two years. English intermediate He used to study his master program in English when he studied in Thailand. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in Muong village, Quan Son District, Thanh Hoa Province, Vietnam. He can speak Quan Son Tai and Vietnamese. LRP’s mother was born in Muong village, Quan Son District, Thanh Hoa Province, Vietnam. She can speak Quan Son Tai, Vietnamese and Lao (Her mother is a Laotian). 278 Remarks Together with most other SWT language varieties, this language is officially classified as Thai in Vietnam. Therefore, non-Tai ethnic group like Kinh and Muong usually call them Thái. The Thai speaking groups make up 83% of the population of this district, and the others are Kinh, Muong, and Hmong. Among the 13 Thai communities in that area, the sounds and vocabularies are little different. Autonyms Phớ Tày, Kun Tày, and Phú tày are commonly used. However, speakers of this language claim they are of Tày đắm (black Tai) since other Tai groups around them like Red Tai and White Tai (Thái đỏ, Thái trắng) call them in this way. In Thanh Hoa province Red Thai live in Lang Chanh district, next to this group. White Tai live in Nhu Thanh district, about 100 km from this group. 7.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there are no tonal flip-flops in this dialect. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 453 B1-A: 55 C1-A: 35ʔ DL1-A: 55 DS1-A: 41 Continuants A1-C: 453 B1-C: 55 C1-C: 35ʔ DL1-C: 55 DS1-C: 41 Unaspirated stops A1-U: 453 B1-U: 55 C1-U: 35ʔ DL1-U: 55 DS1-U: 41 A1-UC: 453 B1-UC: 55 C1-UC: 35ʔ DL1-UC: 55 DS1-UC: 41 A1-G: 453 B1-G: 55 C1-G: 35ʔ DL1-G: 55 DS1-G: 41 A2: 312 B2: 35ʔ C2: 41ʔ DL2: 35 DS2: 41 A1/2: 453 B1/2: 55 C1/2: 35ʔ DL1/2: 55 DS1/2: 41 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits and no splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development with a register merger on DS tone. 279 Horizontally: two individuals and three mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] A[2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger B[2] = C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[2] C[2] = DS [1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] 7.4 Tonal characteristics & tone occurrences: five distinct tones Tone 1: High-rising-falling 453 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 2: Mid-falling-rising 312 occurs in A[2]. Tone 3: High level 55 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] and DL[1-A, 1C, 1-U, 1-UC, 1-G, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Mid-rising 35 occurs in B[2], and in C1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-falling 41 occurs in C[2], and in DS[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2], to form a complementary distribution in smooth and checked syllables respectively. 7.5 Phonation types or voice qualities of tones in this dialect Tones C1 (including B2 which merges into C1) and C2 have final glottal constrictions, like ta:35ʔ (B2) ‘wharf’, ha:35ʔ (C1) ‘five’, saj35ʔ (C1) ‘intestine’, maj41ʔ (C2) ‘tree’, and nam41ʔ (C2) ‘water’. When being pronounced individually, terms with C2 tone have a clear the glottal final stop which is slightly released even after a nasal coda. All the glottal constrictions are not phonemic. 7.6 Sound changes involving tones reflected in the wordlist of this dialect 1) The term ʔa:55 ‘father’s younger sister’ has an unexpected tone B1-G, which is expected to be A1-G in the tone box designated, vs. ʔa:A1-G ‘father’s younger sister’ in Thai and in Debao Yang Zhuang. This tone may have undergone an analogical change according to the frequency of phonological contamination on Tai kinship terms. 2) The term la:35ʔ ‘to scold’ has an unexpected tone B2, which is expected to be B1-G in the tone box designated, vs. da: B1-G ‘to scold’ in Thai and ˀda:B1-G ‘to scold’ in Debao Yang Zhuang. The cause of the tone change needs more data to determine with. 280 3) The term la:ŋ55 ‘ground floor’ has an unexpected tone B1, which is expected to be B2 in the tone box designated, vs. la:ŋB2 ‘under, below’ in Thai and la:ŋB2 ‘ground floor’ in Debao Yang Zhuang. The cause of the tone change needs more data to determine with. 4) The term laŋ453 ‘to make fire’ has an unexpected tone A1-G, which is expected to be C1-G in the tone box designated, vs. ˀdaŋC1-G in Longsang (NT) and Debao Yang Zhuang (CT). This term may have undergone an analogical change or phonological contamination by replacing its original *C tone to *A tone from *vɛjA ‘fire’ (cf. Li 1977: 79, 286) or *wɤjA ‘fire’ (fl. Pittayaporn 2009: 207) according to the frequently used compound word *ʔdaŋC *vɛjA/*wɤjA ‘to make fire’. 5) The term fu:453 ‘CLF of person’ has an unexpected tone A1/2, which is expected to be C1/2 in the tone box designated, vs. pʰu:C1/2 in Thai (SWT), pʰowC1/2 in Debao Yang Zhuang (CT), and pu:C1/2 in Dongling Youjiang Zhuang (NT). This term may have been undergoing an analogical change or phonological contamination by replacing its original *C tone to *A tone from *ɣu̯ɨnA ‘person, human being’ (cf. Li 1977: 215, 272) or *ɢwɯnA ‘person, human being’ (cf. Pittayaporn 2009: 283) since it has frequently been being used to classify *ɣu̯ɨnA or*ɢwɯnA ‘person, human being’, cf. Debao Yang Zhuang (CT) θa:mA1 pʰowC1/2 kɔ:nA2 ‘three people’ and Dongling Youjiang Zhuang (NT) ɬa:mA1 pu:C1/2 hunA2 ‘three people’. 6) The term tʰɯ:k35 (DL1/2) ‘to hit the mark, correct’ has an unexpected tone 35 which is merged into DL2, to have the same merging pattern as NT varieties. This is different from other SWT and CT varieties, which have the tone merging into DL1 for this cognate, such as tʰu:k35 (DL1/2>DL1) ‘to hit the mark, correct’ in Thai (SWT) and tʰʊk33 (DL1/2>DL1) ‘to hit the mark’ in Debao County Yang Zhuang (CT). The cause of the tone change needs more data to determine with. 7) The term ʔaŋ453 (A1-G) ‘chest’ has an unexpected tone A1-G which is expected to be DS1-G and unexpected syllable final -ŋ which is expected to be -k according to the cognate in other Tai varieties, cf. ʔokDL1-G ‘chest’ in Thai (SWT), ʔakDL1-G ‘chest’ in Debao Yang Zhuang (CT), and ʔakDL1-G ‘chest’ in Du’an Hongshuihe Zhuang (NT). The term in Quan Son Tai may not be cognate with the counterpart terms in other Tai varieties, or has been underwent a process of aberrant development which need more data to determine with. 7.7 Other linguistic changes reflected in the wordlist of this dialect 1) All pre-glottalized initial consonants in this dialect have merged into their homorganic sonorants or fricatives, and the processes are as ʔb- > ʋ-, *ʔd- > l-, and *ʔj- > ð-, like in ʋa:nC1-G ‘village’, la:wA1-G ‘star’, and ða:A1-G ‘medicine’. 281 2) The meaning of the term ʋa:j35ʔ (C1-C) ‘to float’ seems to have processed a semantic change from ‘to swim’, because the cognate in all other Tai varieties investigated has the meaning ‘to swim’. Examples with this term are as ʋa:j35ʔ (C1C) hɯə322 (A2) ‘to float a boat’ and ʋa:j35ʔ (C1-C) pɛ:31 (A2) ‘to float a raft’. 3) The term he:35ʔ (C1-UC) ‘near’ has an unexpected initial consonant h-, which is expected to be k- according to the vast majorities of SWT. However, when comparing to CT and NT varieties like kʰjɔj24ʔ ‘near’ in Debao Yang Zhuang (CT), ɕaɰ24ʔ ‘near’ in Tiandeng Zuojiang Zhuang (CT), tɕaɰ213ʔ ‘near’ in Longsang Zhuang (NT), and kjaɰ55 ‘near’ in Wuming Shuangqiao Yongbei Zhuang (NT), as well as to the reconstructed forms *krəɰC by Li (1977: 225, 289) or *k.raɰC by Pittayaporn (2009: 452), the modern initial consonant h- in this dialect is not a single evidence and presents a unique historical change developing from *kr- or *k.r- within SWT varieties. 4) The term kə- ʋa:35ʔ (C1-G) ‘crazy’ has an unexpected prefix kə-, which makes the initial consonant sound very like a sesquisyllabic consonant cluster k.ʋ-. The LRP confirmed that this prefix-like kə- or k- cannot be taken out from this term, or it would be homophonous with ʋa:35ʔ (B2) ‘to talk’, since tone C1 and B2 are merged into each other in this dialect. The LRP also claimed that this prefix-like kə- or khas no any other meaning and there is no any other term in his language has this segment kə- or k-. This is very unique since all other SWT varieties investigated in this paper have a consonant b- or ˀb- reflecting a PT *ʔb- or *ɓ- for this term, such as ba:41ˀ (C1-G) ‘crazy’ in Thai, ba:55ˀ (C1-G) ‘crazy’ in Northern Thai and in ˀba:13ʔ (C1-G) ‘crazy’ Tai Lue. But when comparing with the cognate in some CT varieties like ˀwa:24ʔ (C1-G) ‘crazy’ in Debao Yang Zhuang and ʋa:33ʔ (C1-G) ‘crazy’ in Jingxi Yang Zhuang68, the reconstruction of the initial consonant of this term should be reconsidered to be *ʔ.w- instead of *ʔb- or *ɓ-. This is because the form kə-ʋa:35ʔ (C1-G) ‘crazy’ in Quan Son Tai accounts for a direction on initial consonant *ʔ.w- >*k.w- > kə.ʋ- or k.ʋ- since the change from a glottal stop segment ʔ- to a velar stop segment k- is possible, and only the PT form *ʔ.w- reflected by kə.ʋ- or k.ʋ- in Quan Son Tai and ˀw- in Debao Yang Zhuang is able to explain all the following directions: *ʔ.w- >*ʔw- > *ʔb- > b-/ˀb- in SWT, *ʔ.w- >*k.w- > kə.ʋ- in Quan 68 The initial consonant form ʋ- in Jingxi Yang Zhuang presents a merging direction from ˀw- to ʋ- when comparing with the form in Debao Yang. Other paralleled phenomena in these two Yang Zhuang varieties also suggest this merging process, like ʋa:nA1 (Debao) vs. ʋa:nA1 (Jingxi) ‘sweet’, ʋɔ:nA2 (Debao) vs. ʋanA2 (Jingxi) ‘day’, and ʋa: C2 (Debao) vs. ʋa:C2 (Jingxi) ‘and’, however ru:ŋA2 ˀwa:A1 (Debao) vs. lu:ŋ A2 ʋa:A1 (Jingxi) ‘rainbow’, ˀwe:B1 (Debao) vs. ʋe:B1 (Jingxi) ‘to turn round’, and ˀwi:nB1 (Debao) vs. ʋi:nB1 (Jingxi) ‘to blame’. 282 Son Tai, *ʔ.w- >*ʔw- > ˀw- in Debao Yang Zhuang, and *ʔ.w- >*ʔw- > *w- > ʋin Jingxi Yang Zhuang. Nevertheless, our wordlist is only designated to collect the tone data and is limited to the whole sound system of the languages investigated. It is hardly to believe there is only a single example with this structure in a language, therefore the property of kə- (sesquisyllable segment) or k(sesquisyllabic cluster segment) needs more paralleled phenomena to confirm with in the future. If it is confirmed to be a sesquisyllabic cluster segment, Quan Son Tai will be a very rare modern Tai variety to inherit some of the sysquisyllabic clusters from PT. 5) The initial of the term ʋɔ:k55 (DL1-G) ‘flower’ indicates that the trace of the development of PT *ʔbl/r- (cf. Li 1977: 91) or *ɓl- (cf. Pittayaporn 2009: 157) is *ʔbl/r- or*ɓl- > *ʔb/ɓ- > *w-> ʋ- in this language. This is different from that *ʔbl/r- or*ɓl- > *ʔd/ɗ- > d/ˀd- in the vast majorities of SWT, which have dɔ:k DL1-G or ˀdɔ:k DL1-G for ‘flower’. 6) The initial of the term ʔet41 (DS1-A) ‘to do’ presents an aberrant development that is expected to be h- according to all other Tai varieties which have this cognate, like hetDS1-A in Isan (SWT), hatDS1-A in Yang Zhuang varieties (CT), and hetDS1-A in Zuozhou Zhuang (CT). 283 8. Language 8 Debao Urban, Yang Zhuang, CT 8.1 Language information Language Yang Zhuang ISO 639-3 Code zyg Autonym ˀja:ŋA1-G; tʰo:C1-A Exonym Fu; Nong Fu; T’ienpao/ Tianbao; ka:ŋC1-U ka:jA1-U; ka:ŋC1-U ŋajB1-C Data Source 1 LRP Variety Name Debao Urban Yang Zhuang, which is spoken in Chengguan Town, which is the main town of Debao County, Guangxi Zhuang Autonomous Region, China. 8.2 Information of data source Serial number of LRP LRP1 Date May 17, 2013 LRP’s home, Chengguan Town, Debao County, Baise Prefecture, Place of data elicitation Guangxi Zhuang Autonomous Region, China. Given Name Lìyàn (丽雁) Family Name Ýn (言) Gender female Age 37 Birth Place Duolang (tejB2 la:ŋA2 in local Yang Zhuang) Village (多郎屯), Najia Township, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Chengguan Town, the main town of Debao County. Liuzhou City (study for 2 years). Time period of living at the present residence Occupation self-employment 29 years Education technical secondary school Languages spoken Level Language Remarks Debao Duolang Yang mother tongue Spoken in Duolang village, Najia Township, Debao County, Guangxi, China Debao Urban Yang native LRP has been speaking this language to her friends around her in the main town from childhood. Standard Mandarin fluent This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Debao Duolang Yang Zhuang as native, he can communicate in Guiliu Southwestern Mandarin, Standard Mandarin, and Baihua (a Cantonese variety). 284 LRP’s mother was born in Shangwang (pʰi:ŋA1-A ˀja:ŋB1-G in local Yang Zhuang) village, Najia Township, Debao County. Besides her mother tongue Debao Shangwang Yang, she can speak Guiliu Southwestern Mandarin, and Putonghua. Remarks Many Debao Urban Yang Zhuang speakers speak both their parents’ mother tongue, one of the Debao rural Yang Zhuang varieties, and the urban variety of Chengguan Town (the main town of Debao County) as their corporate mother tongues, since they are born at the town or came to the main town from Childhood. Both the two Yang Zhuang varieties are the same native level for them. 8.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flop in this dialect. Smooth Syllable Checked Syllable Phonation types ProtoVoiceless Sounds A B C DL DS Aspirations A1-A: 453 B1-A: 33 C1-A: 24ʔ DL1-A: 33 DS1-A: 55 Continuants A1-C: 453 B1-C: 55 C1-C: 24ʔ DL1-C: 55 DS1-C: 55 Unaspirated stops A1-U: 453 B1-U: 55 C1-U: 24ʔ DL1-U: 55 DS1-U: 55 A1-UC: 453 B1-UC: 33 C1-UC: 24ʔ DL1-UC: 33 DS1-UC: 55 A1-G: 31 B1-G: 33 C1-G: 24ʔ DL1-G: 33 DS1-G: 55 A2: 31 B2: 33 C2: 213ʔ DL2: 33 DS2: 33 A1/2: 453 B1/2: 33 C1/2: 24ʔ DL1/2: 33 DS1/2: 55 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1/2] [1G, 2] B [1-C, 1-U] [1-A, 1-UC, 1-G, 2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-C, 1-U] [1-A, 1-UC, 1-G, 2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. Secondary tonal splits conditioned by glottalized sounds and aspirated sounds are found in Columns A, B, and DL. Horizontally: four individuals and two mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1/2] A[1-G, 2] 285 C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] B[1-A, 1-UC, 1-G, 2, 1/2] = DL[1-A, 1-UC, 1-G, 2, 1/2] =DS[2] Merger B[1-C, 1-U] = DL[1-C, 1-U] = DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] 8.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High-rising-falling 453 occurs in A[1-A, 1-C, 1-U, 1-UC, 1/2]. Tone 2: Mid-falling 31 occurs in A[1-G, 2]. Tone 3: Mid level 33 occurs in B[1-A, 1-UC, 1-G, 2, 1/2], as well as in DL[1-A, 1UC, 1-G, 2, 1/2] and DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: High level 55 occurs in B[1-C, 1-U], as well as in DL[1-C, 1-U] and DS[1A, 1-C, 1-U, 1-UC, G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low-rising 24 occurs in C[1-A, 1-C, 1-U, 1-UC, G, 1/2]. Tone 6: Low-falling-rising 213 occurs in C[2]. 8.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have final glottal constrictions, like ha:24ʔ (C1) ‘five’, ɬaj24ʔ (C1) ‘intestine’, maj213ʔ (C2) ‘tree’, and nam213ʔ (C2) ‘water’. All the glottal constrictions are not phonemic. 8.6 Sound changes involving tones reflected in the wordlist of this dialect 1) Secondary tonal split conditioned by aspirated sounds (1-A, 1-UC, 1/2) along with glottalized sounds (1-G) are found in Columns B and DL, but not in Column A, which only has a secondary tonal split conditioned by glottalized sounds. Such kind of situation is quite rare found, because any kind of secondary tonal split found only in Column A or in both Columns A and B is more common than only in Column B in ST (CT and SWT) varieties. 8.7 Other linguistic changes reflected in the wordlist of this dialect 1) The initial consonant ɬ- (<*s- (cf. Li 1977: 152) or <*ɬ- (cf. Liao & Shen 2012)) is sometimes pronounced as a free variation θ-, such as ɬɔ:jA1-C = θɔ:jA1-C ‘clear’. 2) Glottalized sounds are robust. Besides the common ones ʔ-, ˀb-, and ˀd-, the other two ones ˀj- and ˀw- are also well preserved. This is quite different from the Jingxi Yang Zhuang varieties, which have normally merged ˀj- and ˀw- into j- and ʋ- respectively. For example, Debao Urban Yang Zhuang ˀjowB1-G ‘at’ vs. Jingxi 286 Urban Yang Zhuang jowB1-G ‘at’, as well as Debao Urban Yang Zhuang ˀwaC1-G ‘crazy’ vs. Jingxi Urban Yang Zhuang ʋaC1-G ‘crazy’. 3) Yodicization of the PT liquids in an initial consonant cluster refers to the original liquid -l- and/or -r- eventually becomes -j- (sometime transcribed as -i- or -̂- by different authors) (cf. Li 1977: 83) are commonly found in this variety and in all the following CT and NT varieties. For example, the following items in this variety contains yodicization of PT medial liquids: pja:453 ‘fish’, pʰjak55 ‘vegetable’, kjɔ:k33 ‘mortar’ and kʰja:435 ‘to seek’ . This is very different from the changes of PT medial liquids in SWT, in which some languages partly preserve the liquids such as pla:33 ‘fish’ in Thai, and some languages have simply dropped the liquids such as pa:55 ‘fish’ in Tai Lue. 4) PT short *-o- is changed to a long -ɔ:- (in most situation) and to a short -a(before -m and -p) in this variety, like pʰɔ:n453 (A1/2) ‘to sharpen’, kʰam453 (A1/2) ‘bitter’, ham33 (B1-A) ‘to cover up’, θɔ:ŋ55 (B1-C) ‘to send’, kʰɔ:t55 (DS1) ‘twisted’, kap55 (DS1-U) ‘frog’, nɔ:k33 (DS2) ‘bird’, and mɔ:t33 ‘ant’. This situation is also found in other Yang Zhuang varieties (L9-10, L12-13) except the more complicated situation in Jingxi Hurun Yang Zhuang (L11). 5) The word developing from PT Etymon *da:B refers to ‘wharf’ in SWT varieties (L1-7). However, ta:B2 developing from *da:B in this variety refers to ‘river’. Actually, hereafter all CT and YN, and NT varieties have the same meaning ‘river’ for this etymon. 6) The word developing from PT Etymon *pa:B refers to ‘forest’ in SWT varieties (L1-7). However, pa:B1 developing from *pa:B in this variety refers to ‘a clump of tree’ or ‘bush’. Actually, hereafter CT and NT varieties which remain this etymon all have the meanings ‘a clump of tree’, ‘bush’ or ‘grove’ more than ‘forest’. 287 9. Language 9 Dalong, Debao Ma’ai, Yang Zhuang, CT 9.1 Language information Language Yang Zhuang ISO 639-3 Code zyg Autonym ˀja:ŋA1-G; tʰo:C1-A Exonym Fu; Nong Fu; T’ienpao/ Tianbao; ka:ŋC1-U ˀba:nC1-G Data Source 1 LRP Variety Name Dalong (village name) of Debao Ma’ai Yang Zhuang, which is spoken in the Northwestern Part of Debao County, Guangxi Zhuang Autonomous Region, China. 9.2 Information of data source Serial number of LRP LRP1 Place of data elicitation Date Xinlong Road 115, Chengguan Town, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Dānlú (丹 Gender male Birth Place May 18, 2013 ) Family Name Huáng (黄) Age 56 Dalong (ka:B2 ɹaŋA2 in local Yang Zhuang) Village (大龙屯), Chengguan Township, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived He lives in Dalong Village, but goes to the main town Debao County which is about four km from his village every day to deal in his trade. He has never left Debao County to other place to live more than three months since he was born. Time period of living at the present residence Occupation farmer, petty dealer 56 years Education junior high school Languages spoken Level Language Remarks Debao Dalong Yang mother tongue Standard Mandarin intermediate This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. It has been used as the language of instruction and education in the western part of Guangxi before Standard Mandarin was popularized from 1980s. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Debao Dalong Yang Zhuang, he can communicate in Guiliu Southwestern Mandarin and Cantonese with Yang Zhuang accents. LRP’s mother was born in the same village. She was monolingual in Yang Zhuang. Remarks Debao Dalong Yang Zhuang is actually often treated as a variety of Debao Ma’ai Yang Zhuang in local concept. The only difference between Dalong and Ma’ai may be the different tonal adaptation of 288 Mandarin Chinese loans. For example, the modern Mandarin loanword ‘Guangxi’ is borrowed as kwa:ŋB1 θi:B1 into Dalong but kwa:ŋA1 θi:B1 into Ma’ai. 9.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flop in this dialect. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 31 B1-A: 332 C1-A: 24ˀ DL1-A: 33 DS1-A: 45 Continuants A1-C: 53 B1-C: 454 C1-C: 24ˀ DL1-C: 45 DS1-C: 45 Unaspirated stops A1-U: 53 B1-U: 454 C1-U: 24ˀ DL1-U: 45 DS1-U: 45 A1-UC: 31 B1-UC: 332 C1-UC: 24ˀ DL1-UC: 33 DS1-UC: 45 A1-G: 31 B1-G: 332 C1-G: 24ˀ DL1-G: 33 DS1-G: 45 Proto-voiced sounds A2: 31 B2: 332 C2: 213ˀ DL2: 33 DS2: 21 Proto-voiced aspirations & A1/2-A: 31 breathy sounds B1/2: 332 C1/2: 24ˀ DL1/2: 33 DS1/2: 45 A1/2-C: 53 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-C, 1-U, 1/2-C] [1-A, 1-UC, 1G, 2, 1/2-A] B [1-C, 1-U] [1-A, 1-UC, 1-G, 2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-C, 1-U] [1-A, 1-UC, 1G, 2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. Secondary tonal splits conditioned by glottalized sounds and aspirated sounds are found in Columns A, B, and DL. Horizontally: three individuals and three mergers A[1-C, 1-U, 1/2-C] Individual A[1-A, 1-UC, 1-G, 2, 1/2-A] C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] B[1-A, 1-UC, 1-G, 2, 1/2] = DL[1-A, 1-UC, 1-G, 2, 1/2] Merger B[1-C, 1-U] = DL[1-C, 1-U] = DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] =DS[2] 289 9.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High-falling 53 occurs in A[1-C, 1-U, 1/2-C]. Tone 2: Mid-falling 31 occurs in A[1-A, 1-UC, 1-G, 2, 1/2-A]. Tone 3: Mid level-falling 332 occurs in B[1-A, 1-UC, 1-G, 2, 1/2] and DL[1-A, 1-UC, 1-G, 2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: High-rising-falling 454 occurs in B[1-C, 1-U], as well as High-rising 45 occurs in DL[1-C, 1-U] and DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low-rising 24 occurs in C[1-A, 1-C, 1-U, 1-UC, G, 1/2]. Tone 6: Low-falling-rising 213 occurs in C[2], as well as low-falling 21 occurs in DS[2] to form a complementary distribution in smooth and checked syllables respectively. 9.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have final glottal constrictions, like ha:24ʔ (C1) ‘five’, θaj24ʔ (C1) ‘intestine’, maj213ʔ (C2) ‘tree’, and nam213ʔ (C2) ‘water’. The final glottal constriction is quite clear, and the LRP almost releases it as a glottal final stop every time when reading word with these two tones. However, all the glottal constrictions are not phonemic. 9.6 Sound changes involving tones reflected in the wordlist of this dialect 1) Different from the Urban variety that has different tonal split patterns between Column A and Columns B/DL, secondary tonal split conditioned by aspirated sounds (1-A, 1-UC, 1/2) along with glottalized sounds (1-G) are neatly found in Columns A, B and DL. 2) In Columns A, B and DL, only initials developing from voiceless continuants (1-C) and voiceless unaspirated stops (1-U) tend to condition the original high register tones, other initial sounds in the high register (1-A, 1-UC, 1-G, 1/2) all condition a secondary tonal split to depress the original high register tone to merge into the low register tone. 3) Because in Row 1/2 (or VASO-1/2) there are two different initial sounds, aspirated sounds and continuants, two different register tones are found in this row. For example, kʰjow31 (A1/2-A) ‘ear’ vs. θɔ:53 (A1/2-C) ‘right side’ are put into the same Row 1/2 (because they are of the voicing alternation initial group), but have different register tones conditioned by aspirated and continuant sounds 290 respectively. This suggests that the designation of Tai tone box should be based on modern phonation types (aspirated and continuant in this case) rather than on historical sources, which may develop to merge into different modern phonation groups. 9.7 Other linguistic changes reflected in the wordlist of this dialect 1) The initial consonant ɬ- (<*s- (cf. Li 1977: 152) or <*ɬ- (cf. Liao & Shen 2012)) in the Urban variety is pronounced as θ- in this variety, such as θɔ:jA1-C ‘clear’ and θamC1-C ‘sour’. 2) Just as the Urban variety, glottalized sounds are also robust. Besides the common ones ʔ-, ˀb-, and ˀd-, the other two ones ˀj- and ˀw- are also well preserved. 3) Some of the PT long vowels on dead syllables are shortened in this variety, but the tonal behaviors of these shortened vowels are preserved to be in the longvowel column (DL). For example, tʰʊk33 (DL1/2 > DL1-A) ‘to hit the mark’, lʊk33 (DL2) ‘child’, and mjət33 ‘knife’ all have short vowels corresponding to long vowels reflected in SWT varieties, but the tonal behaviors are all preserved as in Column DL, which has different tones in Column DS in this variety. 291 10. Language 10 Lüliu, Debao Suburb, Yang Zhuang, CT 10.1 Language information Language Yang Zhuang ISO 639-3 Code zyg Autonym ˀja:ŋA1-G; tʰo:C1-A Exonym Fu; Nong Fu; T’ienpao/ Tianbao; ka:ŋC1-U ŋajB1-C ˀba:nC1-G Data Source 1 LRP Variety Name Lüliu (village name) of Debao Suburb Yang Zhuang, which is spoken at the rural areas surrounding the main town of Debao County, Guangxi Zhuang Autonomous Region, China. 10.2 Information of data source Serial number of LRP LRP1 Place of data elicitation Date May 18, 2013 Xinlong Road 115, Chengguan Town, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Q́xí (奇席) Family Name Ľ (李) Gender male Age 34 Birth Place Lüliu (lu:kDL2 nawB2 in local Yang Zhuang) Village (绿柳屯), Chengguan Township, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived He was born in Lüliu Village of Chengguan Township, Debao County. When he was 23 years old, he went to Zhongshan City of Guangdong Province for his migrate work for about 10 years. He just came back to his village from Guangdong 1 year ago. Time period of living at the present residence Occupation farmer 23 years Education junior high school Languages spoken Level Language Remarks Debao Dalong Yang mother tongue Standard Mandarin fluent This language is the national language of the LRP’s country. Cantonese fluent The LRP learned this language during the time he worked in Guangdong Province, where Cantonese is a lingua franca. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Debao L̈liu Yang as native, he could communicate in Guiliu Southwestern Mandarin with outlander. LRP’s mother was born in the same village. She is monolingual in Yang Zhuang. She lives in the same village with the LRP now. Remarks Yang Zhuang community members from the rural areas normally go to the neighboring Guangdong Province (where is one of the most economically developing regions in China) for their migrant works after junior high school, if their families are not able to support them to continue their studies. 292 10.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flop in this dialect. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 31 B1-A: 33 C1-A: 24ʔ DL1-A: 33 DS1-A: 55 Continuants A1-C: 453 B1-C: 55 C1-C: 24ʔ DL1-C: 55 DS1-C: 55 Unaspirated stops A1-U: 453 B1-U: 55 C1-U: 24ʔ DL1-U: 55 DS1-U: 55 A1-UC: 31 B1-UC: 33 C1-UC: 24ʔ DL1-UC: 33 DS1-UC: 55 A1-G: 31 B1-G: 33 C1-G: 24ʔ DL1-G: 33 DS1-G: 55 A2: 31 B2: 33 C2: 213ʔ DL2: 33 DS2: 33 A1/2: 31 B1/2: 33 C1/2: 24ʔ DL1/2: 33 DS1/2: 55 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-C, 1-U] [1-A, 1-UC, 1G, 2, 1/2] B [1-C, 1-U] [1-A, 1-UC, 1-G, 2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-C, 1-U] [1-A, 1-UC, 1G, 2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. Secondary tonal splits conditioned by glottalized sounds and aspirated sounds are found in Columns A, B, and DL. Horizontally: four individuals and two mergers A[1-C, 1-U] Individual A[1-A, 1-UC, 1-G, 2, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] Merger B[1-A, 1-UC, 1-G, 2, 1/2] = DL[1-A, 1-UC, 1-G, 2, 1/2]= DS[2] B[1-C, 1-U] = DL[1-C, 1-U] = DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] 293 10.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High-rising-falling 453 occurs in A[1-C, 1-U]. Tone 2: Mid-falling 31 occurs in A[1-A, 1-UC, 1-G, 2, 1/2]. Tone 3: Mid level 33 occurs in B[1-A, 1-UC, 1-G, 2, 1/2] and DL[1-A, 1-UC, 1-G, 2, 1/2] and DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: High level 55 occurs in B[1-C, 1-U], as well as DL[1-C, 1-U] and DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low-rising 24 occurs in C[1-A, 1-C, 1-U, 1-UC, G, 1/2]. Tone 6: Low-falling-rising 213 occurs in C[2]. 10.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have final glottal constrictions, like ha:24ʔ (C1) ‘five’, θaj24ʔ (C1) ‘intestine’, maj213ʔ (C2) ‘tree’, and nam213ʔ (C2) ‘water’. However, all the glottal constrictions are not phonemic. 10.6 Sound changes involving tones reflected in the wordlist of this dialect 1) All tonal values of the six tones are the same with the Urban variety (L8), but the basic tonal patterns are almost identical with the Dalong variety (L9) except that DS2 does not merge into C2 like Dalong does, but merges into B2 like the Urban variety (L8) does. 2) Just like the Dalong variety, there are two different initial sounds, aspirated sounds and continuants, are found in Row 1/2 (or VASO-1/2). For example, kʰjow31 (A1/2-A) ‘ear’ vs. θɔ:453 (A1/2-C) ‘right side’ are put into the same Row 1/2 (because they are of the voicing alternation initial group), but have different register tones conditioned by aspirated and continuant sounds respectively. This supports that the designation of Tai tone box should be based on modern phonation types (aspirated and continuant in this case) rather than on historical sources, which may develop to merge into different modern phonation groups. 10.7 Other linguistic changes reflected in the wordlist of this dialect 1) The initial consonant ɬ- (<*s- (cf. Li 1977: 152) or <*ɬ- (cf. Liao & Shen 2012)) in the Urban variety is pronounced as θ- in this variety, such as θɔ:jA1-C ‘clear’ and θamC1-C ‘sour’. 294 2) Unlike the Urban variety (L8) and the Dalong variety (L9), there are only four initial consonants ʔ-, ˀb-, ˀd- and ˀj- in the glottalized sound group. The one ˀw- in the Urban and Dalong varieties has been merged into ˀb- in this variety. For example, Lüliu ˀba:24ʔ ‘crazy’ vs. Dalong / Urban ˀwa:24ʔ ‘crazy’. 295 11. Language 11 Jingxi Hurun Yang Zhuang, CT 11.1 Language information Language Yang Zhuang ISO 639-3 Code zyg Autonym nʊŋA2; tʰo:C1-A Exonym ka:ŋC1-U nɔŋA2; huA1-A jʊnC2 Data Source 3 LRPs Variety Name Hurun (township name) Nong of Yang Zhuang. This variety is spoken in the urban and rural areas of Hurun Township, a town near Chinese-Vietnam border in Jingxi County, Guangxi Zhuang Autonomous Region, China. Hurun Nong of Yang Zhuang is an awkward name for this variety. From intelligibility, from similarity of wordlist and from speaker attitudes, this variety is of Yang Zhuang (ISO 639: zyg) beyond doubt. However, different from other Yang Zhuang varieties in both Jingxi and Debao counties (where Yang Zhuang varieties are the dominant language), speakers of this variety never call themselves Yang but Nong. Note that they are not of Nong Zhuang (ISO 639-3: zhn) distributing in the neighboring Yunnan Province of China. All speakers investigated are obviously unwilling to be called Yang during my fieldwork research on this variety. Nevertheless, I have to temporarily treat them as a variety of Yang Zhuang in this thesis due to the ISO 639-3 principle of the individual language naming in §1.1.3. 11.2 Information of data source Serial number of LRP Place of data elicitation LRP1 Date May 16, 2013 The 4th community, Hurun Town, Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Yù hóng (玉紅) Family Name Húng (黄) Gender female Age 48 Birth Place Hurun (hu:A2 jʊnC1 in local Yang Zhuang) Township (湖潤鎮), Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived She has been living in Hurun from her birth for her whole life. Time period of living at the present residence Occupation farmer 48 years Education Languages spoken Level Hurun Nong of Yang mother tongue junior high school Language Remarks Zhuang Standard Mandarin intermediate This language is the national language of the LRP’s country. Baihua intermediate This is a Cantonese variety in Guangxi. 296 Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in Shangbu Community ( 卜屯 ˀba:nC1 pʊkDS2 ni:A1 in local Zhuang), Xinqun Village (新群村), Hurun Township, Jingxi County. Besides Yang Zhuang as native, he could communicate in Guiliu Southwestern Mandarin, Standard Mandarin, and Cantonese with Yang Zhuang accents. LRP’s mother was born in the same village of her father’s. She was monolingual in local Yang Zhuang. Remarks Hurun is the east end of Jingxi County, and borders on Daxin County, where the main town Taocheng Town has Baihua which is Cantonese variety spoken in Guangxi to be a lingua franca. Therefore, the local Zhuang people in Hurun Township are normally familiar with Baihua. Serial number of LRP Place of data elicitation LRP2 Date May 16, 2013 Hurun (hu:A2 jʊnC1 in local Yang Zhuang) Township (湖潤鎮), Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Hóngḿi (紅梅) Family Name Sū (蘇) Gender male Age 47 Birth Place Hurun (hu:A2 jʊnC1 in local Yang Zhuang) Township (湖潤鎮), Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Xinjing Town, Jingxi County (10 years). She is now living in Hurun Township, Jingxi County. Time period of living at the present residence Occupation farmer 30 years Education Languages spoken Level Hurun Nong of Yang Zhuang mother tongue Jingxi Urban Yang Zhuang fluent Standard Mandarin intermediate junior high school Language Remarks This language is the lingua franca of Jingxi County. This language is the national language of the LRP’s country. Baihua a little This is a Cantonese variety in Guangxi. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents were born in Hurun Township. Besides local Hurun Nong of Yang Zhuang, they both can communicate in Baihua and Standard Mandarin. Both of them passed away. 297 Serial number of LRP Place of data elicitation LRP3 Date May 16, 2013 Hurun (hu:A2 jʊnC1 in local Yang Zhuang) Township (湖潤鎮), Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Gù ihuā (桂花) Family Name Ńng (農) Gender female Age 48 Hurun (hu:A2 jʊnC1 in local Yang Zhuang) Township (湖潤鎮), Jingxi County, Baise Birth Place Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived She has been living in Hurun from her birth for her whole life. Time period of living at the present residence Occupation farmer 48 years Education Languages spoken Level Hurun Nong of Yang Zhuang mother tongue Standard Mandarin intermediate Junior high school Language Remarks This language is the national language of the LRP’s country. Baihua intermediate This is a Cantonese variety in Guangxi. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents were born in Hurun Township. Besides local Hurun Nong of Yang Zhuang, they both can communicate in Baihua and Standard Mandarin. Both of them live with the LRP now. 11.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flop n this dialect. Phonation types ProtoVoiceless Sounds Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 51 B1-A: 445 C1-A: 45ʔ DL1-A: 44 DS1-A: 55 Continuants A1-C: 51 B1-C: 445 C1-C: 45ʔ DL1-C: 44 DS1-C: 55 Unaspirated stops A1-U: 51 B1-U: 445 C1-U: 45ʔ DL1-U: 44 DS1-U: 55 A1-UC: 51 B1-UC: 445 C1-UC: 45ʔ DL1-UC: 44 DS1-UC: 55 A1-G: 31 B1-G: 445 C1-G: 45ʔ DL1-G: 44 DS1-G: 55 A2: 31 B2: 334 C2: 13ʔ DL2: 33 DS2: 13 A1/2: 51 B1/2: 445 C1/2: 45ʔ DL1/2: 44 DS1/2: 55 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 298 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1/2] [1G, 2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. Secondary tonal splits conditioned by glottalized sounds are found in Column A. Horizontally: two individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1/2] A[1G, 2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger B[2] = DL[2] C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] =DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] = DS[2] 11.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High-falling 51 occurs in A[1-A, 1-C, 1-U, 1-UC, 1/2]. Tone 2: Mid-falling 31 occurs in A[1-G, 2]. Tone 3: High-level-rising 445 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and mid level 44 occurs in DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Mid-level-rising 334 occurs in B[2] and mid level 33 occurs in DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-rising 45 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and high level 55 occurs in DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Low-rising 13 occurs in C[2] and in DS[2] to form a complementary distribution in smooth and checked syllables respectively. 299 11.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have final glottal constrictions, like ha:45ʔ (C1) ‘five’, ɬaj45ʔ (C1) ‘intestine’, maj13ʔ (C2) ‘tree’, and nam13ʔ (C2) ‘water’. However, all the glottal constrictions are not phonemic. 11.6 Sound changes involving tones reflected in the wordlist of this dialect 1) The two register tones in Column B in Hurun Yang Zhuang are almost identical to be a relatively high level and rising tone. However, all the LRPs claimed that they can certainly distinguish minimal pairs with tones B1 and B2. For testing the pitch difference of the two register tones, I have used PRAAT (5.1.34) to define these two register tones as 445 and 334 respectively. For example, three words with these two register tones are as ta:445 (B1-U) ‘grandfather’, ˀda:445 (B1G) ‘to scold’, and ta:334 (B2) ‘river’ in the following figure. 2) Basically, glottalized sounds (ʔ-, ˀb-, ˀd-) in Column A (A1-G) condition a secondary tonal split to depress the original high register tone (51) to merge it into the low register tone A2 (31), such as in ˀbən31 ‘to fly’, ˀda:w31 ‘star’, and ʔaw31 ‘to get’. However, the initial ˀj- (<*ʔj-) preserved in the Yang Zhuang varieties in Debao County has been merged into j- in this variety, and the loss of the pre-glottalized feature does not affect the condition of this secondary tonal split, therefore the tone of ja:51 ‘medicine’ and ja:ŋ51 ‘Yang Zhuang people / language’ is preserved as the original high register pitch 51. 3) The term kaw13 ‘nine’ has a low-rising tone which is of C2. This is unexpected because all other Tai varieties all reflect C1-U, which is presented as a high rising tone 45 in this variety. 300 11.7 Other linguistic changes reflected in the wordlist of this dialect 1) Unlike the three Debao Yang Zhuang varieties (L8-10), only three glottalized sounds ʔ-, ˀb-, and ˀd- are preserved. The ˀj- and ˀw- has been merged into j- and ʋ- respectively in this variety. For example, Hurun Yang jowB1-G ‘at’ vs. Debao Urban Yang ˀjowB1-G ‘at’, and Hurun Yang ʋaC1-G ‘crazy’ vs. Debao Urban Yang ˀwaC1-G ‘crazy’. 2) Different from most other Yang Zhuang varieties (including L8, L9, L10, L12, and L13) which reflect that PT *-o- has become a long -ɔ:- (in most situation) and a short -a- (before -m and -p), like pʰɔ:n53 (A1/2) ‘to sharpen’, kʰam53 (A1/2) ‘bitter’, ham353 (B1-A) ‘to cover up’, θɔ:ŋ353 (B1-C) ‘to send’, kʰɔ:t33 (DS1) ‘twisted’, kap33 (DS1-U) ‘frog’, nɔ:k21 (DS2) ‘bird’, and mɔ:t21 ‘ant’ in Jingxi Urban Yang (L12), in this variety the development of PT short *-o- is more complicated. It has been changed to a short -ʊ before -ŋ, to -ə before -m and -n, to -ɔ:- (but sometimes -a-) before -k and -t, to -a- before -p. Therefore, the items above are pʰən51 (A1/2) ‘to sharpen’, kʰəm51 (A1/2) ‘bitter’, həm445 (B1-A) ‘to cover up’, θʊŋ445 (B1-C) ‘to send’, kʰɔ:t55 (DS1) ‘twisted’, nɔ:k13 (DS2) ‘bird’, kap55 (DS1-U) ‘frog’, and mɔ:t13/mat13 ‘ant’ in this variety. 301 12. Language 12 Jingxi Urban, Yang Zhuang, CT 12.1 Language information Language Yang Zhuang ISO 639-3 Code zyg Autonym ja:ŋA1-G; tʰo:C1-A; ɬəŋA2 Exonym Jingxi Zhuang; Nong Shun; Nung Giang Data Source 2 LRPs Variety Jingxi Urban of Yang Zhuang, which is spoken at the main town of Jingxi County, Name Guangxi Zhuang Autonomous Region, China. 12.2 Information of data source Serial number of LRP LRP1 Date May 16, 2013 No.768 Chengdong Road, Xinjing Town, Jingxi County, Baise Place of data elicitation Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Cáish́o (才韶) Family Name Ẃng (王) Gender male Age 33 Birth Place Xinjing Town, Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Chengguan Town, Debao County (high school study, 1 year) Shunde City of Guangdong Province (1 year). Nanning (1 year) He is now living in the Jingxi main town. Time period of living at the present residence Occupation unemployed Languages spoken Level Jingxi Urban Yang mother tongue Standard Mandarin fluent 30 years Education junior high school Language Remarks This language is the national language of the LRP’s country. Debao Urban Yang familiar He used to learn in Debao main town for half a year, and has many local friends from Debao. Cantonese intermediate He used to work in Guangdong Province, where Cantonese is a lingua franca. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in Zumei Community (足梅屯 ˀba:nC1 pja:C1 tʰa:ŋA1 in local Yang Zhuang), Shangba Village ( 壩村), Longbang Township (龍邦鎮), Jingxi County. Besides local Yang Zhuang, he speaks Southwestern to communicate with outlanders. The LRP’s mother was born in Gongtun Community (供屯), Wuping Village ( 村), Wuping Township ( 鎮), Jingxi County. Her mother tongue is local Yang Zhuang. She also speaks Standard Mandarin and Baihua. Both of them live with the LRP now. 302 Remarks The main town of Jingxi County, Xinjing Town, is the largest city in the southwestern corner of the Guangxi Zhuang Autonomous Region of China. Probably due to the language contact with Han Chinese immigrations since the recent history, the Jingxi Urban variety has the most simplified sound system (like the complete loss of pre-glottalization) in Yang Zhuang. Serial number of LRP LRP2 Date August 26, 2014 Tianyu Studio, Nanning City, Guangxi Zhuang Autonomous Region, Place of data elicitation China. Given Name B̄nbīn (彬彬) Family Name Liú ( Gender female Age 28 Birth Place ) Xinjing Town, Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Xinjing Twon, Jingxi (16 years); Baise City (high school study, 3 years); Guilin City (university, 4 years); Yizhou City (3 years). She is living in Nanning City. Time period of living at the present residence Occupation teacher Languages spoken Level Jingxi Urban Yang mother tongue Standard Mandarin fluent 3 years Education bachelor Language Remarks This language is the national language of the LRP’s country. Southwestern Mandarin familiar She used to work in Yizhou City, where Southwestern Mandarin is a dominant language. Baihua a little She is now work in Nanning City, where Cantonese variety Baihua is a lingua franca. English a little The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in Ronglao Township (榮勞鄉) of Jingxi County. LRP’s mother was born in the main town of Jingxi County. They both speak their own Yang Zhuang varieties as mother tongue, and communicate with outlanders in Standard Mandarin. LRP’s father can speak Guiliu 303 12.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flop in this dialect. Smooth Syllable Phonation types ProtoVoiceless Sounds Checked Syllable A B C DL DS Aspirations A1-A: 53 B1-A: 353 C1-A: 33ʔ DL1-A: 35 DS1-A: 33 Continuants A1-C: 53 B1-C: 353 C1-C: 33ʔ DL1-C: 35 DS1-C: 33 Unaspirated stops A1-U: 53 B1-U: 353 C1-U: 33ʔ DL1-U: 35 DS1-U: 33 A1-UC: 53 B1-UC: 353 C1-UC: 33ʔ DL1-UC: 35 DS1-UC: 33 A1-G: 53 B1-G: 353 C1-G: 33ʔ DL1-G: 35 DS1-G: 33 A2: 31 B2: 131 C2: 213ʔ DL2: 21 DS2: 21 A1/2: 53 B1/2: 353 C1/2: 33ʔ DL1/2: 35 DS1/2: 33 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 3) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of straightforward tonal development. Horizontally: three individuals and three mergers A[1-A, 1-C, 1-U, 1-UC, 1G, 1/2] Individual A[2] B[2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] =DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] = DL[2] = DS[2] 304 12.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High-falling 53 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 2: Mid-falling 31 occurs in A[2]. Tone 3: Mid-rising-falling 353 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and midrising 35 occurs in DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low-rising-falling 131 occurs in B[2]. Tone 5: Mid level 33 occurs in C[1-A, 1-C, 1-U, 1-UC, G, 1/2] and DS [1-A, 1-C, 1U, 1-UC, G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Low-falling-rising 213 occurs in C[2] as well as low-falling 21 occurs in DL[2] and DS[2] to form a complementary distribution in smooth and checked syllables respectively. 12.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have final glottal constrictions, like ha:33ʔ (C1) ‘five’, ɬaj33ʔ (C1) ‘intestine’, maj213ʔ (C2) ‘tree’, and nam213ʔ (C2) ‘water’. However, all the glottal constrictions are not phonemic. 12.6 Sound changes involving tones reflected in the wordlist of this dialect 1) Different from the Yang Zhuang varieties in Debao County and the Jingxi Hurun Nong variety of Yang Zhuang which all have non-straightforward tonal development on tonal splits, Jingxi Urban Yang Zhuang has neat tonal splits between proto-voiceless and proto-voiced sounds. 2) According to the previous study of Jingxi Urban Yang Zhuang in Zheng (1996), DL2 and DS2 in this variety are distinct tones, which are merged into B2 and C2 respectively just like the tonal merging pattern of Jingxi Anning Yang Zhuang (L13). However, in my data from the two LRPs, these two tones tent to be merged into C2. This may be a recent tone change, because Jingxi Urban data provided by Zheng (1996) are based on her personal linguistic background. Note that she was born in Jingxi main town in 1936, and her mother tongue accent may be of the old style, which preserves the original tonal pattern. 3) Initials in the groups of glottalized sounds are all merged into continuants (ˀd>n-, ˀb- > m-, ˀj- > j-, ˀw- > ʋ-) except ʔ-. However, tones conditioned by this initial groups are neatly preserved as in the high register. This indicates that the loss of pre-glottalization are much later than that glottalized sounds conditoned tones in the high register in this variety, since the neighboring Yang Zhuang 305 varieties (including those in Debao County and those in the rural areas of Jingxi County) still have robust pre-glotallized sounds. For example, ˀban31 ‘to fly’, ˀda:w31 ‘star’, ʔaw31 ‘to get’, ˀjow33 ‘at’, and ˀwa24ʔ ‘crazy’ in Debao Urban Yang Zhuang are man53, na:w53, ʔaw53, jow353, and ʋa33ʔ respectively in Jingxi Urban Yang Zhuang. 12.7 Other linguistic changes reflected in the wordlist of this dialect 1) Unlike the three Debao Yang Zhuang varieties (L8-10) and Jingxi Hurun Yang Zhuang, pre-glottalized sounds are least robustly preserved in this variety. 2) Like -ɔ: on dead syllables in other Yang Zhuang varieties, items with this long vowel always show a regular short tone (DS), like kʰɔ:t33 (DS1) ‘twisted’, nɔ:k21 (DS2) ‘bird’, and mɔ:t21 ‘ant’, to correspond to cognate with short *-o in other Tai varieties. 3) The vowels of tʰʊk33 (DL1/2 > DS1/2) ‘to hit the mark’ and mit21 ‘knife’ (DL2 > DS2) are shortened. 306 13. Language 13 Jingxi Anning, Yang Zhuang, CT 13.1 Language information Language Yang Zhuang ISO 639-3 Code zyg Autonym ja:ŋA1-G; tʰo:C1-A Exonym Jingxi Zhuang; Nong Shun; Nung Giang Data Source 1 LRP Variety Name Anning Jiaotun of Jingxi Yang Zhuang, which is spoken at the rural areas of Anning Township of Jingxi County, Guangxi Zhuang Autonomous Region, China. 13.2 Information of data source Serial number of LRP Place of data elicitation LRP1 Date August 27, 2014 Jingxi High School, Xinjing Town, Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Cháowěi (朝偉) Family Name Ľ (李) Gender male Age 42 Birth Place Jiaotun Community (叫屯, ˀba:nC1 ke:wB1 in local Zhuang) , Zuhuai Village (逐懷村), Anning Township (安寧鄉), Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Anning Township, Jingxi County (16 years); Baise City (high school + working 10 years). Nanning City (undergraduate study, 4 years); Pingguo County (13 years). He is now living in Baise City. Time period of living at the present residence Occupation civil servants Languages spoken Level Jingxi Anning Yang mother tongue 7 years Education on-job postgraduates Language Remarks Zhuang Jingxi Urban Yang fluent Zhuang Debao Ma’ai Yang fluent Zhuang Spoken in most rural areas of Debao County. This variety is treated as the representative accent of Debao Yang Zhuang. The LRP has many friends from Debao County, and he likes speaking Debao Ma’ai Yang Zhuang to his Debao friends. Standard Mandarin fluent This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. 307 Baihua familiar English a little The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents were both born in the same village. Both of them are monolingual in local Yang Zhuang. 13.3 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there is no tonal flip-flop in this dialect. Smooth Syllable Phonation types ProtoVoiceless Sounds Checked Syllable A B C DL DS Aspirations A1-A: 53 B1-A: 453 C1-A: 44ʔ DL1-A: 45 DS1-A: 44 Continuants A1-C: 53 B1-C: 453 C1-C: 44ʔ DL1-C: 45 DS1-C: 44 Unaspirated stops A1-U: 53 B1-U: 453 C1-U: 44ʔ DL1-U: 45 DS1-U: 44 A1-UC: 53 B1-UC: 453 C1-UC: 44ʔ DL1-UC: 45 DS1-UC: 44 A1-G: 53 B1-G: 453 C1-G: 44ʔ DL1-G: 45 DS1-G: 44 A2: 31 B2: 131 C2: 213ʔ DL2: 13 DS2: 21 A1/2: 53 B1/2: 453 C1/2: 44ʔ DL1/2: 45 DS1/2: 44 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits and no splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of straightforward tonal development. Horizontally: two individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] A[2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger B[2] = DL[2] C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] = DS[2] 308 13.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High-falling 53 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 2: Mid-falling 31 occurs in A[2]. Tone 3: High-rising-falling 453 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and high-rising 45 occurs in DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low-rising-falling 131 occurs in B[2], and low-rising 13 occurs in DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High level 44 occurs in C[1-A, 1-C, 1-U, 1-UC, G, 1/2] and DS [1-A, 1-C, 1-U, 1UC, G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Low-falling-rising 213 occurs in C[2] as well as low-falling 21 occurs in DS[2] to form a complementary distribution in smooth and checked syllables respectively. 13.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have final glottal constrictions, like ha:44ʔ (C1) ‘five’, ɬaj44ʔ (C1) ‘intestine’, maj213ʔ (C2) ‘tree’, and nam213ʔ (C2) ‘water’. However, all the glottal constrictions are not phonemic. 13.6 Sound changes involving tones reflected in the wordlist of this dialect 1) Just like Jingxi Urban Yang Zhuang, this variety has neat tonal splits between proto-voiceless and proto-voiced sounds. 2) Unlike Jingxi Urban merging DL2 and DS2 together to C2, this variety merges DL2 to B2, but DS2 to C2. This pattern seems to be more common, because Columns DL and B are of long vowel-duration pair, and Columns DS and C are of short vowel-duration pair (cf. §4.1.3.2). This makes the whole tonal pattern of this variety to be identical with the Jingxi Urban Yang Zhuang variety in the previous study Jingxi Zhuang Language Research (Zheng 1996) (cf. the point 2) of 12.6 in Appendix C). 3) Unlike Jingxi Urban Yang Zhuang which merges all pre-glottalized sounds into continuants (ˀd- >n-, ˀb- > m-, ˀj- > j-, ˀw- > ʋ-), this variety preserves ˀd- and ˀb- but merges ˀj- > j- and ˀw- > ʋ-. However, tones conditioned by this initial groups are neatly preserved as in the high register. Therefore, ˀban31 ‘to fly’, ˀda:w31 ‘star’, ˀjow33 ‘at’, and ˀwa24ʔ ‘crazy’ in Debao Urban Yang Zhuang are ˀban53, ˀda:w53, jow453, and ʋa44ʔ respectively in this variety. 309 13.7 Other linguistic changes reflected in the wordlist of this dialect 1) Unlike the three Debao Yang Zhuang varieties (L8-10) which most robustly preserve glottalized sounds, and unlike Jingxi Hurun Yang Zhuang (L12) which least robustly preserves glottalized sounds, this variety is like Jingxi Hurun (L11) to preserve the pre-glottalized stops (ˀb- and ˀd-) but to lose the pre-glottalization of minivowels (ˀj- > j- and ˀw- > ʋ-). 2) Like -ɔ: on dead syllables in other Yang Zhuang varieties, items with this long vowel always show a regular short tone (DS), like kʰɔ:t33 (DS1) ‘twisted’, nɔ:k21 (DS2) ‘bird’, and mɔ:t21 ‘ant’, to correspond to cognate with short *-o in other Tai varieties. 3) The vowels of tʰʊk33 (DL1/2 > DS1/2) ‘to hit the mark’ and mit21 ‘knife’ (DL2 > DS2) are shortened and the tone of this item is also changed to be in Column DS. This is unlike Debao Yang Zhuang varieties, in where the same item has been shortened the vowel but is preserved its tonal behavior conditioned by long vowels, like tʰʊk33 (DL1/2) in Debao Ma’ai. 310 14. Language 14 mja:ŋA2, Debao Ronghua Nalong, CT 14.1 Language information Language Myang ISO 639-3 Code undescribed Autonym mja:ŋA2; tʰo:C1-A Exonym Debao Ronghua Data Source 1 LRP Variety Name Nalong (village name) of Myang, which is spoken at the most areas of the Ronghua township, Debao County, Guangxi Zhuang Autonomous Region, China. As far as I known, varieties of Myang Zhuang are only found in Ronghua Township in Debao County. The LRP provided interesting information on their own autonym and some exonyms to other Zhuang varieties around them. They call their native language mja:ŋA2, which is labled as Myang Zhuang here. Members of Myang Zhuang call all Yang Zhuang varieties nʊŋA2 instead of ja:ŋA1-G which is the autonym of many Yang Zhuang variety. This is because the exonym ja:ŋA1-G called by Myang Zhuang members refers to Debao Longsang (L31), which is an undescribed NT variety. We will see the corresponding information in the following Language 31, Debao Longsang, which is called ja:ŋA1-G as an autonym by their own members. In fact, as far as I know, many NT varieties in the western part of Guangxi call some NT varieties distributing on the CTNT border ja:ŋA1-G. In other words, ja:ŋA1-G never refers to the Yang Zhuang varieties (CT) (which is put into ISO 639-3 code as “zyg”) by these NT varieties. Instead, they call Yang Zhuang varieties of Debao and Jingxi ɲaŋA2 or noŋA2. The term noŋA2 is the common autonym or exonym of CT varieties. The term ɲaŋA2 is actually adapted from the exclusive Southern Zhuang (CT) lexical item ‘still’ which is not found in NT varieties, to refer to the Yang Zhuang varieties (which robustly use this item) by NT members (Liao 2010: 77). In short, the exonyms of Debao Longsang and Yang Zhuang called by the Myang Zhuang members indicate the awkward situation of the naming Yang Zhuang again, to respond the situation that the members of Hurun Nong of Yang Zhuang reject the exonym Yang but prefer the autonym Nong (cf. the language information of the LRP1 of L10). The information indicates that the naming of Yang Zhuang in ISO 639-3 system needs more review and discussions. 14.2 Information of data source Serial number of LRP Place of data elicitation LRP1 Date May 15, 2013 Xinlong South Six-Road 13, Chengguan Town, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Fèngrén (奉仁) Family Name M̌ng (蒙) Gender female Age 60 Birth Place Nalong Community (那陇屯na:A2 lʊŋB2 in local Myang and Yang Zhuang), Nalong Village (那陇村), Ronghua Township, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. 311 Places lived She was born in Nalong Village of Ronghua Township, Debao County. When she was 18 years old, she moved to Chengguan Town to live until now. Time period of living at the present residence Occupation retired worker 42 years Education Languages spoken Level Myang Zhuang mother tongue Debao Urban Yang Zhuang familiar primary school Language Remarks This language is the lingua franca in Debao County, Guangxi. Standard Mandarin intermediate This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. Baihua intermediate The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. He was monolingual in local Myang Zhuang. LRP’s mother was born in Maomei Village (tʰu:A1-A ˀdɔ:ŋA1-G in local Yang Zhuang), Ronghua Township, Debao County. She was also monolingual in local Myang Zhuang. 14.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flop in this dialect. Phonation types Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 55 B1-A: 35 C1-A: 24ʔ DL1-A: 35 DS1-A: 55 Continuants A1-C: 55 B1-C: 35 C1-C: 24ʔ DL1-C: 35 DS1-C: 55 Unaspirated stops A1-U: 55 B1-U: 35 C1-U: 24ʔ DL1-U: 35 DS1-U: 55 A1-UC: 55 B1-UC: 35 C1-UC: 24ʔ DL1-UC: 35 DS1-UC: 55 A1-G: 55 B1-G: 35 C1-G: 24ʔ DL1-G: 35 DS1-G: 55 Proto-voiced sounds A2: 42 B2: 33 C2: 31ʔ DL2: 33 DS2: 42 Proto-voiced aspirations & A1/2: 55 B1/2: 35 C1/2: 24ˀ DL1/2: 35 DS1/2: 55 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds breathy sounds 312 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits and no splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of straightforward tonal development. Horizontally: two individuals and four mergers Individual C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger A[2] = DS[2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] B[2] = DL[2] 14.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High level 55 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 2: High-falling 42 occurs in A[2] and DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Mid-rising 35 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and DL[1-A, 1-C, 1-U, 1UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low level 33 occurs in B[2] and DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low-rising 24 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 6: Mid-falling 31 occurs in C[2]. 14.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have final glottal constrictions, like ha:24ʔ (C1) ‘five’, ɬaj24ʔ (C1) ‘intestine’, maj213ʔ (C2) ‘tree’, and nam213ʔ (C2) ‘water’. However, all the glottal constrictions are not phonemic. 313 14.6 Sound changes involving tones reflected in the wordlist of this dialect 1) Just like the Jingxi Urban and Jingxi Anning Yang Zhuang varieties, this variety has neat tonal splits between proto-voiceless and proto-voiced sounds. 2) The term ŋa:j42 ‘turn face up’ has an unexpected A2 tone, which is expected to be A1 tone in the tone box designated, vs. ŋa:j24 (A1-C) in Thai and ŋa:j453 (A1-C) in Debao Urban Yang Zhuang. 14.7 Other linguistic changes reflected in the wordlist of this dialect 1) Like the three Debao Yang Zhuang varieties (L8-10) which most robustly preserve pre-glottalized sounds, ʔ-, ˀb-, ˀd-, and ˀj- are all found in this Myang Zhuang variety, and they all condition a high register tone. Because of the limited data, we do not know whether it preserves ˀw- or not. 2) Labialized sounds are robuster than the neighboring Yang Zhuang. Not only kw-, kʰw-, and ŋw- (which are also found in Debao County Yang Zhuang) are found, but also ɬw- and tw- (which are not found in Debao County Yang Zhuang) are found due to my data, like ɬwa:55 (A1/2) ‘right side’ and twa: (A2) ‘to spread on’. 3) The term tʰjap55(DS1-A) ‘hailstone’ has an unexpected initial consonant tʰj-, which is expected to be tʰ- according to other Tai varieties. 314 15. Language 15 Tuoxin, Pyang Zhuang, CT 15.1 Language information Language Pyang Zhuang ISO 639-3 Code undescribed Autonym pja:ŋB2; tʰo:C1-A Exonym ˀjujC1-G; Rui Data Source 1 LRP Variety Name Tuoxin (驮信村 ˀba:n55 te:31 θin21 in local Pyang Zhuang), Debao Fuping Pyang Zhuang, which is spoken at the rural areas surrounding the main town of Debao County, Guangxi Zhuang Autonomous Region, China. 15.2 Information of data source Serial number of LRP LRP1 Date May 17, 2012 Xinlong Road 115, Chengguan Town, Debao County, Baise Place of data elicitation Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Chángchéng (長城) Family Name Ćn (岑) Gender male Age 34 Birth Place Tuoxin Community (驮信村 ˀba:n55 te:31 θin21 in local Pyang Zhuang), Fuping Village, Jingde Township, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Tuoxin (5 years); Tongkuang (Copper Mine Refinery) in Yandong Township, Debao County (10 years)y. Chengguan Town, Debao County (Senior high school 3 years + working 7 years). Nanning City (Undergraduate study 4 years + working 6 years). He is now living in Chengguan Town, Debao County. Time period of living at the present residence Occupation instrument technician 7 years Education bachelor Languages spoken Level Language Remarks Tuoxin Pyang Zhuang mother tongue Standard Mandarin intermediate This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides mother tongue Pyang, he spoke Debao Ma’ai Yang Zhuang which is the lingua franca of Debao County, Southwestern Mandarin and Standard Mandarin. LRP’s mother was born in an unknown village at where Nung An (of Yongnan Zhuang) was spoken. When she was 5, she moved to the LRP’s village. Her mother tongue is Nung An, but she states that her Pyang Zhuang is much better than Nung An because she speaks Pyang Zhuang in her family after she moved to Tuoxin Village. She also speaks Debao Ma’ai Yang Zhuang and Min Zhuang which is another 315 CT language spoken in the near villages in Fuping area. Besides, she can communicate in Standard Mandarin and Southwestern Mandarin in her Zhuang accent. Remarks This language is also called ˀjujC1 by the surrounding Yang Zhuang members, and is sometimes transcribed as Rui (瑞) in Chinese character. As far as I known, varieties of Pyang Zhuang are only found in Ronghua Township in Debao County and the neighboring Kuixu Township in Jingxi County. It may be closely related to Nong Zhuang (ISO 639-3: zhn) varieties in Yunnan Province of China due to some of their shared innovations, e.g. PT *xr- (Li 1977: 233) or *kr- (Pittayaporn 2009: 143) is reflectd as tɕʰ- both in Pyang Zhuang and Nong Zhuang (vs. kʰj- in Yang Zhuang, and h- in SWT varieites), such as tɕʰa:A1-A ‘to seek’ and tɕʰɔkDS1-A ‘six’, as well as some specific words like hɔkDS1-A ‘to do’ only found in Pyang Zhuang and Nong Zhuang (vs. hatDS1-A ‘to do’ in Yang Zhuang, hetDS1-A ‘to do’ in Zuojiang Zhuang and Isan). Besides, Gedney’s western Nung (cf. Hudak 2008) may be closed related to Nong Zhuang and Pyang Zhuang. 15.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there are tonal flip-flops on columns B and DL. Phonation types Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 24 B1-A: 212 C1-A: 45/55ʔ DL1-A: 21 DS1-A: 55 Continuants A1-C: 24 B1-C: 212 C1-C: 45/55ʔ DL1-C: 21 DS1-C: 55 Unaspirated stops A1-U: 24 B1-U: 212 C1-U: 45/55ʔ DL1-U: 21 DS1-U: 55 B1-UC: C1-UC: DL1-UC: DS1-UC: 212 45/55ʔ 21 55 A1-G: 24 B1-G: 212 C1-G: 45/55ʔˀ DL1-G: 21 DS1-G: 55 Proto-voiced sounds A2: 33 B2: 42 C2: 53ʔ DL2: 42 DS2: 33 Proto-voiced aspirations & A1-U: 24 B1/2: 212 C1/2: 45/55ʔ DL1/2: 21 DS1/2: 55 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds A1-UC: 24 breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of straightforward tonal development. 316 Horizontally: one individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] A[2] = DS[2] Merger B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] B[2] = DL[2] C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] 15.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Low-rising 24 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 2: Mid level 33 occurs in A[2] and DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Low-falling-rising 212 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and low-falling 21 occurs in DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: High-falling 42 occurs in B[2] and DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-rising 45 or high level 55 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and high level 55 occurs in DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: High-falling 53 occurs in C[2]. 15.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have final glottal constrictions, like ha:24ʔ (C1) ‘five’, θaj24ʔ (C1) ‘intestine’, maj53ʔ (C2) ‘tree’, and nam53ʔ (C2) ‘water’. However, all the glottal constrictions are not phonemic. 15.6 Sound changes involving tones reflected in the wordlist of this dialect 1) Just like other CT varieties investigated, the vowel and the tone of the term pik55 ‘wing’ is a short vowel which conditions the tone to be of DS1-U (55). This is different from the cognate of SWT varieties, which have this term with a long vowel and a DL1-U tone in the tone box designated, vs. pi:k 21 (DL1-U) ‘wing’ in Thai. 15.7 Other linguistic changes reflected in the wordlist of this dialect 1) In this Pyang Zhuang variety, yodicizaion of PT liquid is also found, such as pja24 and pʰjak55 ‘vegetable’. However, the yodicization kj- and kʰj- which are 317 commonly found in the neighboring Yang Zhuang varieties has been changed to tɕ- and tɕʰ- due to palatalization in this variety, such as tɕɔk33 ‘mortar’, tɕʰu:24 ‘ear ‘and tɕʰɔk33 ‘six’. 2) Some specific words like hɔkDS1-A ‘to do’ is found in this variety. This is different from the common ST cognate hatDS1-A ‘to do’ in Yang Zhuang, hetDS1-A ‘to do’ in Zuojiang Zhuang and hetDS1-A ‘to do’ in Isan). 318 Language 16 Jingxi Huashan, Min Zhuang, CT 16.1 Language information Language Min Zhuang ISO 639-3 Code zgm Autonym ɕu:ŋB1; minA1; tʰo:C1 Exonym Zhong; Min; Black Cloth Zhuang Data Source 1 LRP Variety Name Hushan(華山屯, mo:A2 ɹ̥ənC2 in local Min Zhuang) of Jingxi Sanhe Min Zhuang, which is spoken at the rural areas of Sanhe Township, Jingxi County, Guangxi Zhuang Autonomous Region, China. 16.2 Information of data source Serial number of LRP LRP1 Date May 20, 2012 Shangjie 65, Ande Township, Jingxi County, Baise Prefecture, Place of data elicitation Guangxi Zhuang Autonomous Region, China. Given Name Čipíng (彩萍) Family Name X̌ (許) Gender female Age 42 Birth Place A2 Huashan (華山屯, mo: Township ( ɹ̥ən C2 in local Min Zhuang), Sanxi Village ( 西村), Sanhe 合鄉), Baise Prefecture, Jingxi County, Guangxi Zhuang Autonomous Region, China. Places lived Hushan Village (22 years); Dajie Community (大節屯, ˀba:nC1 ˀdaɰA1 tɕe:tDL1), Dale Village (大樂村), Sanhe Township ( 合鄉) (16 years) . Guangdong Province (6 years). Hainan Province(1 year). Nanning City (1 year). Time period of living at the present residence Occupation farmer Languages spoken Level Huashan Min Yang mother tongue Jingxi Yang Zhuang familiar Standard Mandarin fluent Cantonese intermediate 16 years Education The fourth year of primary school Language Remarks This language is the national language of the LRP’s country. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Min Zhuang, he also spoke Yang Zhuang which is the lingua franca of Jingxi County, as well as Mandarin and Cantonese for communicating with outlanders. LRP’s mother was born in Dale Village (大樂村 ˀdaɯA1 lɔ:kDL1), Ande Township, Jingxi County. Besides Min Zhuang, she also spoke Yang Zhuang which is the lingua franca of Jingxi County. Remarks Min Zhuang is a second largest language in the Dejing area of Guangxi, next to Yang Zhuang (Jackson et al. 2011). The main body of Min Zhuang region includes the rural areas of all the townships of western part of Jingxi County, the most areas of Napo County, and the southwestern corner of Debao County. 319 They are well-known as Black Cloth Zhuang in Chinese media recently, due to their unique traditional dress. Although the LPR claims that she can speak Yang Zhuang, she insisted to communicate in her Min Zhuang with me even though I spoke my Debao County Yang Zhuang to her during the whole investigation. This does not mean that Yang Zhuang and Min Zhuang are intelligible with one another at a functional level (that is, can understand based on knowledge of their own variety without needing to learn the other variety). For most Min Zhuang speakers (as well as other non-Yang Zhuang languages) in Jingxi County, Yang Zhuang is a second language which can be acquired when they go to a school where Yang Zhuang children are the dominant group. For a Yang Zhuang mother tongue speaker, to understand Min Zhuang (as well as other non-Yang Zhuang languages in Jingxi County) at a functional level without learning Min Zhuang is quite difficult. However, most of the Yang Zhuang mother tongue speakers in Ande, Sanhe, and Longlin Townships can understand Min Zhuang very well because Min Zhuang is the largest Zhuang group in these townships. It is very commonly seen that a local Yang Zhuang and a local Min Zhuang speaker communicate with each other in their own languages with inherent understanding. 16.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, tonal flip-flop runs through all tones. Smooth Syllable Phonation types Checked Syllable A B C DL DS Proto- Aspirations A1-A: 242 B1-A: 32 C1-A: 11ʔ DL1-A: 32 DS1-A: 32 Voice- Continuants A1-C: 353 B1-C: 32 C1-C: 11ʔ DL1-C: 32 DS1-C: 55 less Unaspirated stops A1-U: 353 B1-U: 32 C1-U: 11ʔ DL1-U: 32 DS1-U: 55 A1-UC: 242 B1-UC: 32 C1-UC: 11ʔ DL1-UC: 32 DS1-UC: 32 A1-G: 353 B1-G: 32 C1-G: 11ʔ DL1-G: 32 DS1-G: 55 Proto-voiced sounds A2: 55 B2: 42 C2: 53ʔ DL2: 42 DS2: 55 Proto-voiced aspirations & A1/2-A: 242 breathy sounds A1/2-C: 353 B1/2: 32 C1/2: 11ʔ DL1/2: 32 Sounds Unaspirated stop + *r clusters Glottal Sounds DS1/2-A: 32 DS1/2-C: 55 2) Patterns of tonal split and merger are as in the following two charts: Vertically: three-way splits on A and two-way splits on the others Tonal category Register 1 Register 1ʼ Register 2 A [1-C, 1-U, 1-G, 1/2-C] [1-A, 1-UC, 1/2-A] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-UC, 1/2-A] [1-C, 1-U, 1-G, 2, 1/2-C] In conclusion, tones in this dialect are of non-straightforward tonal development. The nonstraightforward columns are A and DS. The splits on column A presents as three-way, with voiceless 320 aspirated stops (1-A, 1-UC, and 1/2-A) on one register, voiceless non-aspirated sounds (1-C, 1-U, 1-G, 1/2-C) on one register, and voiced on the other register. The others are all two-way, but the column DS has very unique and meaningful two-way splits: voiceless aspirated stops (1-A, 1-UC, and 1/2-A) on one register and other sounds including voiced sounds on the other register. The unique and important points are as: 1) the merger that voiced sounds go with all other voiceless sound except aspirated stops is never found in any other Tai varieties, and this may indicate a possible process accomplishing on the column DS: three-way splits [1-A, 1-UC, 1/2-A]-[1-C, 1-U, 1-G, 1/2-C]-[2] like in the column A occured first, and later the two registers [1-C, 1-U, 1-G, 1/2-C] and [2] merged into each other since they are all high tones, therefore a special two-way split on this column is established; 2) the split between 1-A and 1-C on column DS confirms that voicelss frictions must be divided into 1-A and 1-C, since previous studies have only found the split between 1-A and 1-C limited to the A, B, and DL columns in Debao Yang Zhuang varieties and Dai Zhuang. Horizontally: four individuals and three mergers A[1-A, 1-UC, 1/2-A] Individual A[1-C, 1-U, 1-G, 1/2-C] C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] A[2] = DS[1-C, 1-U, 1-G, 2, 1/2-C] Merger B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DS[1-A, 1-UC, 1/2-A] B2[2] = DL[2] 16.4 Tonal characteristics & tone occurrences: seven distinct tones Tone 1: Low-rising-falling 242 occurs in A[1-A, 1-UC, 1/2-A]. Tone 2: Mid-rising-falling 353 occurs in A[1-C, 1-U, 1-G, 1/2-C]. Tone 3: High level 55 occurs in A[2]. Tone 4: Mid-falling 32 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2], DL[1-A, 1-C, 1-U, 1UC, 1-G, 1/2], as well as DS[1-A, 1-UC, 1/2-A] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-falling 42 occurs in B[2] and DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Low level 11 in C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 7: High-falling 53 occurs in C[2]. 321 16.5 Phonation types or voice qualities of tones in this dialect: Tones C1 and C2 have glottal constrictions, like ha:55ˀ ‘five’, θɯ:11 ‘shirt’, ka:ŋ11ˀ ‘fishbone’, maj53ˀ ‘tree’, te:53ʔ ‘carry on the back’, and ma:53ʔ ‘horse’. However, all the glottal constrictions are not phonemic. 16.6 Sound changes involving tones reflected in the wordlist of this dialect: 1) Like some other CT varieties such as Jingxi Urban Yang, the items pik55 ‘wing’ in DL1 and mit55 ‘knife’ in DL2 designed to expect having a long vowel have merged their tone into DS1(44/55) and DS2 (44/55) respectively, probably triggered by the shortening of their vowels. 16.7 Other linguistic changes reflected in the wordlist of this dialect: 1) The terms (na:11) pʰjak 32(DL1-A) ‘forehead’ and pʰjak 32 (DL1-UC) ‘to expose under the sun’ both have an unexpected short vowel -a, which is expected to be a long vowel -a: according to the cognate in other Tai varieties. Therefore, they have become homophonous with the term pʰjak 33 (DS1-A) ‘vegetable’. 2) Like many other CT varieties, the term pik 55 ‘wing’ has a short vowel -i, which is expected to be a long vowel -i: in DL1-U designated. Therefore, its tone has been also changed to DS1-U. 3) The vowel of te:k42 ‘to measure’ is unexpected, vs. ta:kDL2 in all Yang Zhuang varieties and in NT varieties as well as tʰa:kDL2 in Jingxi Lingding and Daxin Baoxu. 4) The vowel of pʰo:32 ‘to chop’ is unexpected, vs. pʰa:B1-A in other CT and SWT varieties. 322 Language 17 Xiangdu, Tiandeng, Zuojiang Zhuang, CT 17.1 Language information Language Zuojiang Zhuang ISO 639-3 Code zzj Autonym tʰɔ:C1-A Exonym Tiandeng Tuhua; ka:ŋC1 həɰA1 na:jC1 Data Source 3 LRPs Variety Name Xiangdu Township (向都鎮, həɰA1 na:jC1 in local Xiangdu Zhuang) of Tiandeng County Zuojiang Zhuang, which is spoken at the town of Xiangdu Township, Tiandeng County, Guangxi Zhuang Autonomous Region, China. Actually, varieties of Zuojiang Zhuang and varieties of Yang Zhuang can more or less understand with each other in a certain extent. However, they are divided into two different individual languages based on intelligibility and similarity of wordlist. Although this Zhuang variety is divided into Zuojiang Zhuang, speakers of this variety have inherent understanding of Debao County Yang Zhuang at a functional level, that is Xiangdu Zuojiang Zhuang and Debao County Yang Zhuang can understand each other based on knowledge of their own variety without needing to learn the other variety. This indicates that Tiandeng County Zuojiang Zhuang including this variety may be a transitional dialect between Zuojiang Zhuang and Yang Zhuang. 17.2 Information of data source Serial number of LRP Place of data elicitation LRP1 Date May 18, 2013 Xinlong Road 115, Chengguan Town, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Guótuán (國團) Family Name Wáng (王) Gender Male Age 28 Birth Place Xiangdu Township (向都鎮, həɰA1 na:jC1 in local Xiangdu Zhuang), Tiandeng County, Chongzuo Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Xiangdu Township (23 years); The main town of Tiandeng County (3 years). Beijing (1 year). Shenzhen, Guangdong Province (1 year). Shanghai (2 years). He is living in Xiangdu now. Time period of living at the present residence Occupation 23 years personal business Education Languages spoken Level Tiandeng Xiangdu Zuojiang Zhuang mother tongue Standard Mandarin fluent junior high school Language Remarks This language is the national language of the LRP’s country. Baihua intermediate 323 A Cantonese variety spoken in Guangxi. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents were born in the same town. Besides their Xiangdu Tiandeng Zuojiang Zhuang, both of them can speak Standard Mandarin and Baihua. Serial number of LRP LRP2 Date May 18, 2013 Xinlong Road 115, Chengguan Town, Debao County, Baise Prefecture, Place of data elicitation Guangxi Zhuang Autonomous Region, China. Given Name Jiàn (劍) Family Name Ćn (岑) Gender male Age 35 Birth Place Xiangdu Township (向都鎮, həɰA1 na:jC1 in local Xiangdu Zhuang), Tiandeng County, Chongzuo Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Xiangdu Township (15 years). The main town of Tiandeng County (4 years). Nanning City (4 years). He is living in the main town of Debao County. Time period of living at the present residence Occupation social service Languages spoken Level Tiandeng Xiangdu Zhuang mother tongue Debao Urban Yang Zhuang familiar Standard Mandarin intermediate 12 years Education bachelor Language Remarks This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. Baihua intermediate A Cantonese variety spoken in Guangxi. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents were born in the same town. Besides his Xiangdu Tiandeng Zuojiang Zhuang, his father could can speak Baihua. His mother was monolingual in local Zhuang. Both of them passed away. Remarks The LRP2 speaks his own Tiandeng Xiangdu Zuojiang Zhuang to Debao Yang Zhuang speakers in his daily life. I observed that he sometimes tried to use some Debao Yang Zhuang pronunciations and lexical items which are different from his mother tongue since he has been living in Debao for a long time. However, his Debao Yang was still with strong Tiandeng Xiangdu accent. 324 Serial number of LRP LRP3 May 18, 2013 Xinlong Road 115, Chengguan Town, Debao County, Baise Prefecture, Place of data elicitation Guangxi Zhuang Autonomous Region, China. Given Name Tīnf́ ( Gender male Birth Place Date 福) Xiangdu Township (向都鎮, həɰ A1 C1 na:j Family Name D̀ (杜) Age 28 in local Xiangdu Zhuang), Tiandeng County, Chongzuo Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Xiangdu Township (23 years). Guangzhou City of Guangdong Province (1 year). Dongguan City of Guangdong Province (2 years). Other cities in Guangdong Province (several years). Time period of living at the present residence Occupation farmer 23 years Education Languages spoken Level Tiandeng Xiangdu Zhuang mother tongue Standard Mandarin intermediate junior high school Language Remarks This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. Cantonese intermediate He learned this language when he was in Guangdong Province. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents were born in the same town. Besides his Xiangdu Tiandeng Zuojiang Zhuang, both of his parents could can speak Baihua and Standard Mandarin. Both of them live with the LRP now. 17.3 Basic patterns of tonal development of this dialect are as below. 1) As shown in the chart below, there is no tonal flip-flop in this dialect. Phonation types ProtoVoiceless Sounds Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 453 B1-A: 33 C1-A: 24ʔ DL1-A: 33 DS1-A: 33 Continuants A1-C: 453 B1-C: 33 C1-C: 24ʔ DL1-C: 33 DS1-C: 33 Unaspirated stops A1-U: 453 B1-U: 33 C1-U: 24ʔ DL1-U: 33 DS1-U: 33 A1-UC: 453 B1-UC: 33 C1-UC: 24ʔ DL1-UC: 33 DS1-UC: 33 A1-G: 31 B1-G: 33 C1-G: 24ʔ DL1-G: 33 DS1-G: 33 A2: 31 B2: 11̰ C2: 24ʔ DL2:11 DS2: 11 A1/2: 453 B1/2: 33 C1/2: 24ʔ DL1/2: 33 DS1/2: 33 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 325 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits and no splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. Column A has a split between 1-UC and 1-G. The column C has no spilts at all. Horizontally: three individuals and two mergers A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Individual A[2] C[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] B1[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = Merger DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] B[2] = DL[2] = DS[2] 17.4 Tonal characteristics & tone occurrences: five distinct tones Tone 1: High-rising-falling 453 occurs in A[1-A, 1-C, 1-U, 1-UC, 1/2]. Tone 2: Low-falling 31 occurs in A[1-G, 2]. Tone 3: Mid level 33 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2], DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low level 11 occurs in B[2], DL[2] and DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low-rising 24 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2]. 17.5 Phonation types or voice qualities of tones in this dialect Tone C has glottal constrictions, like ha:24ˀ ‘five’, pʰja:j24ˀ ‘to walk’, ɬɯ:24ˀ ‘shirt’, ka:ŋ24ˀ ‘fishbone’, maj24ˀ ‘tree’, and nam24ˀ ‘water’. Tone B2 has creakiness triggered by its extreme low pitch. All the voice qualities are not phonemic. 326 17.6 Sound changes involving tones reflected in the wordlist of this dialect 1) Like in some CT varieties such as all Yang Zhuang varieties and Baoxu, the term ta:33 ‘maternal grandfather’ has an unexpected B1-U tone, which is expected to be A1-U in the tone box designated, vs. ta:33 (A1-U) in Thai (SWT) and ta:51 (A1-U) in Du’an Hongshuihe Zhuang (NT). The tone change is speculated as a tonal contamination by the tone of the term ta:jB1-U ‘maternal grandmother’ in these CT varieties (Liao 2016, in press). 2) The tone of the term ha:ŋ11 ‘young hen’ has an unexpected B2 tone, which is expected to be B1(33)<B1/2 in the tone box designated. Because the initial of this term is also unexpected to be h- which is expected to be kʰ- due to the correspondence among Xiangdu and other CT/SWT varieties (according to the comparison among kʰʊmA1<A1/2 ‘bitter’~kʰawC1<C1/2 ‘rice’~ha:ŋB2<B1/2 ‘young hen’ in This variety, kʰamA1<A1/2 ‘bitter’~kʰaw C1<C1/2 ‘rice’~kʰy:ŋB1<B1/2 ‘young hen’ in Debao Urban Yang Zhuang (CT), and kʰomA1<A1/2 ‘bitter’~kʰa:w C1<C1/2 ‘rice’~kʰɨəŋB1<B1/2 ‘young hen’ in Thai (SWT)), this term should be a loan word from the neighboring NT varieties, which have an initial reflecting *ɦ- and a tone B2 merged from B1/2 for this term, like ha:ŋB2<B1/2 in Wuming, and ɣa:ŋB2<B1/2 in Huishui Bouyei. 17.7 Other linguistic changes reflected in the wordlist of this dialect 1) The initial consonant developing from PT *xr- (cf. Li 1977: 233) or *kr(Pittayaporn 2009: 143-144) presents as l- in this dialect, like in la453 (A1-A) ‘to seek’, lou453 (A1/2) ‘ear’, and laj33 (B1-A) ‘egg’. However, the tonal behavior of this initial keeps in the A1-A Row as other aspirated sounds. The process of the initial development might have been like *xr- or *kr- > *kʰr- >*ʰr- > *r- > l-. 2) Short -a- preceding -k has been changed into a long -a:-, as in pʰja:kDS1 ‘vegetable’ = ‘forehead’, pa:kDS1 ‘to stab’ = ‘mouth’, pa:kDS2 ‘tired’ = ‘insane’, and la:kDS2 ‘to steal’ = ‘root’. 327 Language 18 Naling, Daxin, Zuojiang Zhuang, CT 18.1 Language information Language Zuojiang Zhuang ISO 639-3 Code zzj Autonym tʰɔ:C1-A Exonym Daxin Tuhua Data Source 1 LRP Variety Name Dongli (village name, tʊŋB2 lɛ:B2 in local Zhuang) of Daxin Naling Zuojiang Zhuang. 18.2 Information of data source Serial number of LRP LRP1 31 August, 2014 #4101 Dushi Jimei Residential Quarter, Jianshe Road, Nanning Place of data elicitation City, Guangxi Zhuang Autonomous Region, China. Given Name Yùb̄ (月 Gender female Birth Place Date ) Family Name Zhào (趙) Age 66 Dongli Community (洞利屯, tʊŋB2 lɛ:B2 in local Zhuang), Nalian Village (那廉村), Naling Township (那岭鄉, fa͏ɰB2 lɛ:ŋC2 in local Zhuang), Daxin County, Chongzuo Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived She was born in 1948. She grew up and married in her hometown. She moved to Taocheng Town, the main town of Daxin County, in 1977 with her husband, who worked as a teacher in Daxin High School. She lived in Taocheng Town from 18771999. From 2000 up to now, she moved to Nanning to live with her daughters. Time period of living at the present residence Occupation school worker Languages spoken Level Daxin Naling Zuojiang mother tongue 15 years Education primary school Language Remarks Zhuang Daxin Encheng Zuojiang familiar Zhuang This language is another variety of Daxin Zuojiang Zhuang. Baihua fluent This language is the lingua franca of Daxin County. Standard Mandarin intermediate This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. He could speak all the languages that the LRP can. LRP’s mother was born in Balan Village (巴兰村, faɰB2 ˀda:nB1-G in local Zhuang) of Naling Township, Daxin 328 County. Besides her mother tongue Daxin Balan Zuojiang Zhunag, she could speak Daxin Naling Zhuang, Guiliu Southwestern Mandarin, and Baihua. Remarks From the simplest tonal patterns in Daxin Naling to the very complicated tonal patterns in Daxin Baoxu (L23), Zuojiang Zhuang varieties in Daxin County have complicated diversities. 18.3 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there is no tonal flip-flop in this dialect. Phonation types Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 453 B1-A: 33 C1-A: 24ʔ DL1-A: 33 DS1-A: 55/45 Continuants A1-C: 453 B1-C: 33 C1-C: 24ʔ DL1-C: 33 DS1-C: 55/45 Unaspirated stops A1-U: 453 B1-U: 33 C1-U: 24ʔ DL1-U: 33 DS1-U: 55/45 A1-UC: 453 B1-UC: 33 C1-UC: DL1-UC: DS1-UC: 24ʔ 33 55/45 A1-G: 453 B1-G: 33 C1-G: 24ʔ DL1-G: 33 DS1-G: 55/45 Proto-voiced sounds A2: 31 B2: 11 C2: 113ʔ DL2: 11 DS2: 31 Proto-voiced aspirations & A1/2: 453ʰ B1/2: 33ʰ C1/2: 24ʔ DL1/2: 33 DS1/2: 55/45 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of straightforward tonal development. Horizontally: one individuals and four mergers Individual C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger A[2] = DS[2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] B[2] = DL[2] 329 18.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Low-rising-falling 453 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2], as well as high-rising 45 or high level 55 in DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 2: Mid-falling 31 occurs in A[2] and in DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Mid level 33 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low level 11 occurs in B[2] and in DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low-rising 24 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2]. Tone 6: Low-level-rising occurs in C[2]. 18.4 Phonation types or voice qualities of tones in this dialect Tone C has glottal constrictions, like ha:24ˀ ‘five’, pʰja:j24ˀ ‘to walk’, ɬɯ:24ˀ ‘shirt’, ka:ŋ24ˀ ‘fishbone’, maj113ˀ ‘tree’, and nam113ˀ ‘water’. All the glottal constrictions are not phonemic. 330 Language 19 Lingding, Jingxi Zuozhou, Zuojiang Zhuang, CT 19.1 Language information Language Zuojiang Zhuang ISO 639-3 Code zzj Autonym ɕu:ŋB2; tɕa:B1 tɕowA1; Exonym Zazhou; Zuozhou tʰo: C1 Data Source 2 LRPs Variety Name Lingding (凌 屯, θaŋA2 tɛ:ŋB1 in local Zhuang) Zuozhou of Zuojiang Zhuang, 19.2 Information of data source Serial number of LRP LRP1 Date Lingding Community (凌 Place of data elicitation Village (峒 May 16, 2013 屯, θaŋA2 tɛ:ŋB1 in local Zhuang), Dongping 村), Xinjia Township (新甲鄉), Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Míngchū n (明春) Family Name Zh̄u (周) Gender male Age 67 Birth Place Lingding Community (凌 屯, θaŋA2 tɛ:ŋB1 in local Zhuang), Dongping Village (峒 村), Xinjia Township (新甲鄉), Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived He lives in Lingding village during all his life. Time period of living at the present residence Occupation farmer 67 years Education Languages spoken Level Lingding Zuozhou Zuojiang Zhuang mother tongue Jingxi County Yang Zhuang intermediate primary school Language Remarks This language is the lingua franca of Jingxi County. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. Standard Mandarin a little The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents born in the same village. Besides his mother tongue Lingding Zuozhou, LRP’s father can communicate in Jingxi County Yang Zhuang and Jingxi Min Zhuang. LRP’s mother was monolingual in Zuozhou, but can understand Jingxi Yang Zhuang. Remarks Zuozhou is a specific group of Zuojiang Zhuang. In Jackson et al. (2011), Zuozhou is not classified in any studied individual Zhuang languages listed in ISO 639-3 because their data is limited. However, 331 according to my personal fieldwork research, Zuozhou is an immigration group migrating from Zuozhou Township, Jiangzhou District, Chongzuo Prefecture of Guangxi. The name Zuozhou is actually their ancestral home. Because the Zhuang dialect in Zuozhou Township is put under Zuojiang Zhuang in previous studies in China, I also treat this group as a variety of Zuojiang Zhuang, which is listed as “zzj” in ISO 639-3. However, Zuozhou is a specific group containing their own internal characteristics. Varieties of Zuozhou may consist of an individual language with its own ISO 639 code from intelligibility, from similarity of wordlist and from speaker attitudes in further studies. For convenience, in this thesis I use Zuozhou to refer to this specific group rather than Zuojiang Zhuang in my formulation. Therefore, Jingxi Lingding Zuozhou is used instead of Jingxi Lingding Zuojiang Zhuang. Serial number of LRP LRP2 Date Lingding Community (凌 Village (峒 Place of data elicitation May 16, 2013 屯, θaŋA2 tɛ:ŋB1 in local Zhuang), Dongping 村), Xinjia Township (新甲鄉), Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Given Name Guó’ān (國安) Family Name Ĺng (凌) Gender male Age 61 Birth Place Lingding Community (凌 (峒 屯, θaŋA2 tɛ:ŋB1 in local Zhuang), Dongping Village 村), Xinjia Township (新甲鄉), Jingxi County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived He was born, grew up, studied, and worked in the same village during all his life. The Dongping High School is where he studied, and Dongping Primary School is where he studied and taught. Time period of living at the present residence Occupation retired teacher 61 years Education senior high school Languages spoken Level Language Remarks Zuozhou mother tongue Jingxi County Yang Zhuang fluent This language is the lingua franca of Jingxi County. Standard Mandarin intermediate This language is the national language of the LRP’s country. Guiliu Dialect familiar A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. He spoke Zuozhou and Jingxi County Yang Zhuang. He passed away. LRP’s mother was born in Dazhong Community (大種屯), Daci Village (大慈村), Ronglao Township (榮勞鄉), Jingxi County. Besides her mother tongue Dazhong Zuozhou, she can speak Jingxi County Yang Zhuang. She lives with LRP now. 332 19.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flops except tones C in this dialect. Smooth Syllable Phonation types Checked Syllable A B C DL DS Aspirations A1-A: 353 B1-A: 33 C1-A: 24ʔ DL1-A: 33 DS1-A: 55 Proto- Continuants A1-C: 554 B1-C: 33 C1-C: 24ʔ DL1-C: 33 DS1-C: 55 Voiceless Unaspirated Sounds stops A1-U: 554 B1-U: 33 C1-U: 24ʔ DL1-U: 33 DS1-U: 55 A1-UC: 353 B1-UC: 33 C1-UC: 24ʔ DL1-UC: 33 DS1-UC: 55 A1-G: 21 B1-G: 31 C1-G: 24ʔ DL1-G: 31 DS1-G: 55 Proto-voiced sounds A2: 21 B2: 31 C2: 42ʔ DL2: 31 DS2: 21 Proto-voiced aspirations & A1/2-A: 353 breathy sounds A1/2-C: 554 B1/2: 33 C1/2: 24ʔ DL1/2: 33 DS1/2: 55 Unaspirated stop + *r clusters Glottal Sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: three-way splits and two way splits Tonal category Register 1 Register 1' Register 2 A [1-C, 1-U, 1/2-C] [1-A, 1-UC, 1/2-A] [1-G, 2] B [1-A, 1-C, 1-U, 1-UC, 1/2] [1-G, 2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1/2] [1-G, 2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. Horizontally: three individuals and four mergers A[1-A, 1-UC, 1/2-A] Individual C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[C2] A[1-C, 1-U, 1/2-C] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger A[1-G, 2] = DS[2] B[1-A, 1-C, 1-U, 1-UC, 1/2]= DL[1-A, 1-C, 1-U, 1-UC, 1/2] B[1-G, 2] = DL[1-G, 2] 333 19.4 Tonal characteristics & tone occurrences: seven distinct tones Tone 1: High level-falling 554 occurs in A[1-C, 1-U, 1/2-C] (continuants and unaspirated stops). Tone 2: Mid-rising-falling 353 occurs in A[1-A, 1-UC, 1/2-A] (aspirated sounds), and high level 55 occurs in DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Low-falling 21 occurs in A[1-G, 2] and DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Mid level 33 occurs in B[1-A, 1-C, 1-U, 1-UC, 1/2], and in DL[1-A, 1-C, 1-U, 1UC, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Mid-falling 31 occurs in B[1-G, 2], and in DL[1-G, 2], to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Low-rising 24 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 7: High-falling 42 occurs in C[2]. 19.5 Phonation types or voice qualities of tones in this dialect: Tones C1 and C2 have glottal constrictions, like ha:24ˀ ‘five’, θɯ:24ˀ ‘shirt’, ka:ŋ24ˀ ‘fishbone’, maj42ˀ ‘tree’, and nam42ˀ ‘water’. All the glottal constrictions are not phonemic. 19.6 Sound changes involving tones reflected in the wordlist of this dialect 1) The three-way split in Column A is similar to Huashan Min Zhuang (L16). Aspiration (1-A, 1-UC, 1/2) depress the tone pitch from the original high register (1) to change to a lower pitch to form a third register (1'), which is not merged into the low register (2) like in Debao Yang varieties. 2) Like many other CT varieties, the term pik33 ‘wing’ has a short vowel -i, which is expected to be a long vowel -i: in DL1-U designated. However, its tone (33) is preserved to be in DL1-U. This is very similar to Debao Dalong Yang Zhuang (L9). 3) Unlike other CT/SWT varieties, the terms nuk55 ‘deaf’ and hɯk55 ‘gums’ which are expected to be long vowels in Column DL1 designated have short vowels like NT varieties, and their tones have also been changed from Columns DL1 to DS1. 19.7 Other linguistic changes reflected in the wordlist of this dialect 1) The word (japDS1) nemA1 ‘in addition’ has an unexpected initial n-, vs. hɛ:mA1 ‘more’ in Rong Maet Tai Lue, (ʔejC1) he:mA1 in Debao Urban Yang Zhuang, and tʰe:m ‘in addition’ in Jingxi Anning Yang Zhuang. It may have undergone an initial change due to phonological contamination of the initial of the first segment of this item, namely the initial of nemA1 may have been changed from tʰ- (stop) to n- (continuant) due to the analogy of j- (continuant). 334 Language 20 Xiaoguangnan, Nong Zhuang, CT 20.1 Language information Language Nong Zhuang ISO 639-3 Code zhn Autonym nɔŋA2 Exonym Nong Dao Data Source Zhang et al. 1999: 47-48, 157-159, 595-808; Johnson 2011: 20-28 Data Info. The language data was collected from Xiaoguangnan Village, Xiaoguangnan Community, Liancheng Township, Guangnan County, Yunnan Province, China. 20.2 Basic patterns of tonal development of this dialect 1) As shown in the chart below, tonal flip-flops runs through B, C, and DL tones. Phonation types ProtoVoiceless Sounds Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 35 B1-A: 12 C1-A: 22 DL1-A: 13 DS1-A: 55 Continuants A1-C: 35 B1-C: 12 C1-C: 22 DL1-C: 13 DS1-C: 55 Unaspirated stops A1-U: 35 B1-U: 12 C1-U: 22 DL1-U: 13 DS1-U: 55 A1-UC: 35 B1-UC: 12 C1-UC: 22 DL1-UC: 13 DS1-UC: 55 A1-G: 35 B1-G: 12 C1-G: 22 DL1-G: 13 DS1-G: 55 A2: 33 B2: 31 C2: 53 DL2: 31 DS2: 33 A1/2: 35 B1/2: 12 C1/2: 22 DL1/2: 13 DS1/2: 55 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of straightforward tonal development. 335 Horizontally: one individuals and four mergers Individual C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] C[2] A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger A[2] = DS[2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] B[2] = DL[2] 20.3 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Mid-rising 35 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2], as well as high-level in DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 2: Mid level 33 occurs in A[2] and DS[2], to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Mid-level 33 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2]. Tone 4: Low-falling 21 occurs in A[1-A, 1-C, 1/2], and in DL[1-A, 1-C, 1-U, 1UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-falling 41 occurs in C[1-U, 1-UC, 1-G, 2], and in DL[2], to form a complementary distribution in smooth and checked syllables respectively. 20.4 Phonation types or voice qualities of tones in this dialect. We do not know the phonation types or voice qualities of tones in this dialect since they are not mentioned in the original text. 336 Language 21 Dazhai, Dai Zhuang, CT 21.1 Language information Language Dai Zhuang ISO 639-3 Code zhg Autonym dajA2 Exonym Tulao; Tuzu; Dai Data Source Zhang et al. 1999: 48-50, 164-165, 595-808 Reference Johnson 2011:28-38 Data Info. According to the original material, this language is commonly called Dai in local Zhuang, or Tulao in the local Chinese dialect in that area. The language data was collected from Dazhai Village, Heimo Township, Wenshan County, Yunnan Province, China. 21.2 Basic patterns of tonal development of this dialect 1) As shown in the chart below, only a tonal flip-flop runs through Tone A in this dialect. Phonation types Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 31 B1-A: 53 C1-A: 55 DL1-A: 53 DS1-A: 31 Continuants A1-C: 112 B1-C: 53 C1-C: 55 DL1-C: 53 DS1-C: 33 Unaspirated stops A1-U: 112 B1-U: 53 C1-U: 55 DL1-U: 53 DS1-U: 33 A1-UC: 31 B1-UC: 53 C1-UC: 55 DL1-UC: 53 DS1-UC: 31 A1-G: 112 B1-G: 53 C1-G: 55 DL1-G: 53 DS1-G: 33 Proto-voiced sounds A2: 31 B2: 42 C2: 33 DL2: 42 DS2: 31 Proto-voiced aspirations & A1/2-A:31 breathy sounds A1/2-C:112 B1/2: 53 C1/2: 55 DL1/2: 53 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds DS1/2-A: 31 DS1/2-C:33 2) Patterns of tonal split and merger are as in the following two charts: Vertically: two-way split Tonal category Register 1 Register 2 A [1-C, 1-U, 1-G, 1/2-C] [1-A, 1-UC, 2, 1/2-A] B [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-UC, 2, 1/2-A] [1-C, 1-U, 1-G, 1/2-C] In conclusion, tones in this dialect are of non-straightforward tonal development. Secondary tonal split patterns are found in Column A and DS. 337 Horizontally: two individuals and four mergers Individual A[1-C, 1-U, 1-G, 1/2-C] C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] A[1-A, 1-UC, 2, 1/2-A] = DS[1-A, 1-UC, 2, 1/2-A] Merger B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] B[2] = DL[2] C[2] = DS[1-C, 1-U, 1-G, 1/2-C] 21.3 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Low level-rising 112 occurs in A[1-C, 1-U, 1-G, 1/2-C]. Tone 2: Mid-falling 31 occurs in A[1-A, 1-UC, 2, 1/2-A] and DS[1-A, 1-UC, 2, 1/2A] to form a complementary distribution in smooth and checked syllables respectively. Tone 3: High-falling 53 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and DL[1-A, 1C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: High-falling 42 occurs in B[2], and in DL[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High level 55 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 6: Mid level 33 occurs in C[2] and DS [1-C, 1-U, 1-G, 1/2-C], to form a complementary distribution in smooth and checked syllables respectively 21.4 Phonation types or voice qualities of tones in this dialect We do not know the phonation types or voice qualities of tones in this dialect since they are not mentioned in the original text. 338 Language 22 Lei Ping, Zuojiang Zhuang, CT 22.1 Language information Language Zuojiang Zhuang Autonym po:n C1-U h B2 Data Source Hudak 2008: 27-30 Reference Zhang et al. 1999:44-45, 146-149, 595-808 Data Info. According to the original material, Gedney collected the Pei Ping data from Mr. Liang t i: ʋa: B2 ISO 639-3 Code zzj Exonym tʰo:C1-U Shao-lu (62-year-old) in 1966, in Hong Kong. The informant called the language as po:nC1-U thi:B2 ʋa:B2, which means ‘the native language’ in this language (Hudak 2008: 27). This language is spoken in Leiping Township (in nowadays Daxin County) in southwest Guangxi. The Zhuang dialect Daxin Houyi described in Zhang et al. (1999) is almost identical with Leiping investigated by Gedney according to the data comparison. Because Houyi is a village name in nowadays Leiping Township, it can be supposed that Houyi and Lei Ping are of the same variety even though the tonal values between Houyi and Lei Ping are a little bit different (probably due to different auditory sense between the researchers). As for tonal categories, the only difference in the descriptions between Lei Ping and Houyi is that there is no tonal split in Column C in Lei Ping (Hudak 2008: 28), but in Zhang et al. (1999: 146) Houyi contains some single items containing Tone 4 (C2) separating from Tone 3 (C1), such as naj44 ‘this’. Nevertheless, in Houyi the original Tone 4 (C2) has been merged into Tone 3 (C1) in most situation just like what Gedney’s observation. It can be assumed that Gedney did not observe the remnant of the original C2 tone in such as naj44 ‘this’ because of the data limitation. Here I adopt Gedney’s description of Lei Ping to draw the tone box, but data from Houyi in Zhang et al. (1999) are important supplement to Gedney’s Lei Ping because Houyi’s vocabularies are much more plenty. 22.2 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flops except Tones C in this dialect. Phonation types ProtoVoiceless Sounds Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 454 B1-A: 44ʔ C1-A: 13ʔ DL1-A: 44 DS1-A: 44 Continuants A1-C: 454 B1-C: 44ʔ C1-C: 13ʔ DL1-C: 44 DS1-C: 44 Unaspirated stops A1-U: 44ʔ B1-U: 11 C1-U: 13ʔ DL1-U: 44 DS1-U: 11 A1-UC: 454 B1-UC: 44ʔ C1-UC: 13ʔ DL1-UC: 44 DS1-UC: 44 A1-G: 44ʔ B1-G: 11 C1-G: 13ʔ DL1-G: 44 DS1-G: 11 A2: 21 B2: 11 C2: 13ʔ DL2: 21/11 DS2: 11 A1/2-A: 454 B1/2: 44ʔ C1/2: 13ʔ DL1/2: 44 DS1/2: 44 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 339 2) Patterns of tonal split and merger are as in the following two charts. Vertically: three-way split, two way split and no split Tonal category Register 1 Register 1' Register 2 A [1-A, 1-C, 1-UC, 1/2] [1-U, 1-G] [2] B [1-A, 1-C, 1-UC, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DS [1-A, 1-C, 1-UC, 1/2] [1-U, 1-G, 2] [1-U, 1-G, 2] In conclusion, tones in this dialect are of non-straightforward tonal development. There are secondary tonal splits found in Columns A, B, DS, and Column C has merged the two registers. Horizontally: three individuals and two mergers A[1-A, 1-C, 1-UC, 1/2] Individual A[2] C[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] A[1-U, 1-G] = B[1-A, 1-C, 1-UC, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] = Merger DS[1-A, 1-C, 1-UC, 1/2] B[1-U, 1-G, 2] = DL[2] = DS[1-U, 1-G, 2] 22.3 Tonal characteristics & tone occurrences: five distinct tones Tone 1: High-rising-falling 454 occurs in A[1-A, 1-C, 1/2-A] (aspirated sounds). Tone 2: High level 44 occurs in A[1-U, 1-G] as well as B[1-A, 1-C, 1-UC, 1/2], DL[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] and DS[1-A, 1-C, 1-UC, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Mid-falling 21 occurs in A[2]. Tone 4: Low level 11 occurs B[1-U, 1-G, 2] as well as in DL[2] and DS[1-U, 1-G, 2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low-rising 13 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2]. 22.4 Phonation types or voice qualities of tones in this dialect The tone on Column C is glottalized. The tone with tonal value 44 (A[1-U, 1-G]) and B[1-A, 1-C, 1-UC, 1/2]) has a final drop before pause (Hudak 2008: 28). 340 Language 23 Baoxu, Zuojiang Zhuang, CT 23.1 Language information Language Zuojiang Zhuang ISO 639-3 Code zzj Autonym tʰo:C1-A Exonym Nung Fan Sling Data Source 1 LRP Variety Name Naya Community (那 屯, ˀba:nC1-G na:A2 jɯ:A1-G in local Zhuang), Shangyi Village (尚益村), Baoxu Township (寶墟鄉, faɯB2 ɕɔ:kDL2 in local Zhuang), Daxin County, Chongzuo Prefecture, Guangxi Zhuang Autonomous Region, China. 23.2 Information of data source Serial number of LRP LRP1 Place of data elicitation Room 403, Nanning Gucheng Road Branch, City Express Inn Given Name Hǎihuá ( 海華) Family Name Nóng (農) Gender male Age 29 Birth Place Naya Community (那 ( Date August 31, 2014 屯, ˀba:nC1-G na:A2 jɯ:A1-G in local Zhuang), Shangyi Village 益村), Baoxu Township, Daxin County, Chongzuo Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived He grew up in the village in where he was born in 1985. When he was 15, he went to Taocheng Town, the main town of Daxin County, to study in high school for 6 years. From 2003 to 2010 he had his undergraduate and postgraduate studies in Nanning City. After postgraduate he has been working in Nanning City for four years. Time period of living at the present residence Occupation executive secretary 9 years Education M.A. Languages spoken Level Language Remarks Daxin Baoxu Zuojiang Zhuang mother tongue Baihua fluent This language is the lingua franca of Daxin County. Standard Mandarin fluent This language is the national language of the LRP’s country. English intermediate The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides his Daxin Baoxu Zhuang as native, he can communicate in Baihua and Guiliu Standard Mandarin with his Zhuang accents. LRP’s mother was born in Banzhou Community (板 屯, ˀba:nC1-G ɕowC2 in local Zhuang) village, Jingyang Village (景陽村), Baoxu Township, Daxin County. Besides her mother tongue Daxin Baoxu 341 Zuojiang Zhuang, she can communicate in Baihua and Guiliu Standard Mandarin with his Zhuang accents. Remarks Together with Lei Ping and other Zuojiang Zhuang varieties in Daxin County, Baoxu has complicated tonal split and merger patterns. The most noticeable point is the three-way split pattern found in Columns A, C, and DS, which has a secondary tonal split conditioned together by glottalized sounds (AG) and unaspirated stops (A-U). This kind of tonal split conditioning factor is commonly found in SWT varieties (like Bangkok Thai and Isan), but is very rare found in CT varieties. 23.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flops in this dialect. Smooth Syllable Checked Syllable Phonation types ProtoVoiceless Sounds A B C DL DS Aspirations A1-A: 45 B1-A: 34 C1-A: 13ʔ DL1-A: 34 DS1-A: 13 Continuants A1-C: 45 B1-C: 34 C1-C: 13ʔ DL1-C: 34 DS1-C: 13 Unaspirated stops A1-U: 53 B1-U: 33 C1-U: 34ʔ DL1-U: 33 DS1-U: 34 A1-UC:45 B1-UC: 34 C1-UC: 13ʔ DL1-UC: 34 DS1-UC: 13 A1-G: 53 B1-G: 33 C1-G: 34ʔ DL1-G: 33 DS1-G: 34 A2: 31 B2: 33 C2: 11̰ʔ DL2: 33 DS2: 11 A1/2: 45 B1/2: 34 C1/2: 13ʔ DL1/2: 34 DS1/2: 13 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: three-way splits and two way splits Tonal category Register 1 Register 1' Register 2 A [1-A, 1-C, 1-UC, 1/2] [1-U, 1-G] [2] B [1-A, 1-C, 1-UC, 1/2] C [1-A, 1-C, 1-UC, 1/2] DL [1-A, 1-C, 1-UC, 1/2] DS [1-A, 1-C, 1-UC, 1/2] [1-U, 1-G, 2] [1-U, 1-G] [1-U, 1-G, 2] [1-U, 1-G] In conclusion, tones in this dialect are of straightforward tonal development. 342 [2] [2] Horizontally: three individuals and four mergers A[1-A, 1-C, 1-UC, 1/2] Individual A[[1-U, 1-G] A[2] B[1-A, 1-C, 1-UC, 1/2] = C[1-U, 1-G] = DL[1-A, 1-C, 1-UC, 1/2] Merger B[1-U, 1-G, 2] = DL[1-U, 1-G, 2] C[1-A, 1-C, 1-UC, 1/2] = DS[1-A, 1-C, 1-UC, 1/2] C[2] = DS[2] 23.4 Tonal characteristics & tone occurrences: seven distinct tones. Tonal value 34 can be found in the following boxes: B[1-A, 1-C, 1-UC, 1/2], C[1-U, 1-G] and DL[1-A, 1-C, 1-UC, 1/2]. Column DL on checked syllables can be ignored when discussing the difference between the two columns in smooth syllables. It is very interesting that this tone 34 is not glottalized in Column B (1-A, 1-C, 1-UC, 1/2) but has clear glottal constriction in Column C (1-U, 1-G). It indicates that before tonal merger, B[1-A, 1-C, 1-UC, 1/2] and C[1-U, 1-G] were two different tones, and this tone in Column C still preserves the glottal constriction even its tonal value has been merged with B[1-A, 1-C, 1-UC, 1/2]. However, because its tonal value 34 complementarily distributes in B[1-A, 1-C, 1-UC, 1/2] (conditioned by voiceless frictions) and C[1-U, 1-G] (conditioned by voiceless unaspirated stops and glottal sounds) with different groups of initials, the glottalized feature of C[1-U, 1-G] cannot account for as phonemic or tonemic. Therefore, B[1-A, 1-C, 1-UC, 1/2] and C[1-U, 1-G] has to be treated as allotones of the same toneme. Tone 1: High-rising 45 occurs in A[1-A, 1-C, 1-UC, 1/2]. Tone 2: High-falling 53 occurs in A[1-U, 1-G]. Tone 3: Mid-falling 31 occurs in A[2]. Tone 4: Mid-rising 34 occurs in B[1-A, 1-C, 1-UC, 1/2] and C[1-U, 1-G], as well as in DL[1-A, 1-C, 1-UC, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Mid level 33 occurs in B[1-U, 1-G, 2] and in DL[1-U, 1-G, 2] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Low-rising 13 occurs in C[1-A, 1-C, 1-UC, 1/2] and in DS[1-A, 1-C, 1-UC, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 7: Low level [11] occurs in C[2] = DS[2] to form a complementary distribution in smooth and checked syllables respectively. 343 23.5 Phonation types or voice qualities of tones in this dialect: All the three tones in Column C have glottal constrictions, like ha:13ˀ ‘five’, ɕa:j13ˀ ‘to walk’, ɬɯ:13ˀ ‘shirt’, ka:ŋ34ˀ ‘fishbone’, ˀba:n34ˀ ‘village’, and nam11̰ˀ ‘water’. Tone C2 has a creakiness triggered from its extremely low pitch. All the glottal constrictions are not phonemic. 23.6 Sound changes involving tones reflected in the wordlist of this dialect 1) PT *pʰl/r- (cf. Li 1977: 87-88) or *pr- (cf. Pittayawat Pittayaporn 2009: 140-141, 230) is changed to be ɕ- in this dialect, like in the term ɕa:45 (A1-A) ‘hill; cliff’, and ɕa:j24ʔ ‘to walk’. PT *khr- (cf. Li 1977: 228-229) or *qr- (cf. Pittayawat Pittayaporn 2009: 144-145, 330) is also changed to be ɕ- in this dialect like in ɕaj34 (B1-A) ‘egg’. That is to say these two PT initials have been merged into one another. However, the tonal behavior of ɕ- in this dialect is preserved to be with aspirated sounds in the 1-A row, which is different from the tonal behaviors of the 1-U and 1-G rows. This indicates the process of the initial development of ɕ- may have processed a period of aspiration, as *pʰl/r- or*pr- > *pʰr- > *pʰj- > *kʰj- > *tɕʰ- > ɕ- and khr- or*qr- > *kʰr- > *kʰj- > *tɕʰ- > ɕ- due to the neighboring CT varieties which have pʰja:A1-A ‘hill; cliff’ (in all Yang Zhuang varieties) or kʰja:A1-A ‘hill; cliff’ (in some Zuojiang Zhuang varieties like Tiandeng Bahe which has merged PT *pr- and *kr- into kʰj-), pʰja:jC1-A ‘to walk’ as well as kʰjajB1-A ‘egg’ for these terms. Since PT *pl- (cf. Li 1977: 84-85; Pittayawat Pittayaporn 2009: 147148) in this dialect has been merged into *kl- as a modern tɕ- via a possible process *pl- > *pj- > *kj- > tɕ- like in the term tɕa:53 (A1-U) ‘fish’, the counterpart PT *pr- is suggested to process the transitional succession *pʰj- > *kʰj- > *tɕʰ- > ɕ-. However, Daxin Naling (Language 18), which is another Daxin Zuojiang Zhuang variety in the same county, has two optional variations pʰja:A1-A and ɕa:A1-A for ‘hill; cliff’. This also suggests that the initial development of this consonant in Baoxu may have been directly changed from pʰj- into ɕ-. 2) Initial consonants in the supposed 1-UC row which is to follow Li’s unaspirated stop + *r- clusters or ASC (Aspiration induced by Sesquisyllabic Clusters) row which is to be suggested as a series of sesquisyllable cluster following the hypothesis provided by Pittayawat Pittayaporn (2009) normally regularly present as aspirated sounds or a sound with spread feature (h-), like in ha:45 (A1-UC) ‘eye’, ha:j45 (A1-UC) ‘die’, pʰe:ŋ45 (A1-UC) ‘cucumber’, and kʰa:ŋ45 (A1-UC) ‘to put up’ like other CT and Yongnan Zhuang varieties. However, the item taɰ53 (A1-U<A1-UC) which is expected to be in this row due to the cognate in other CT varieties (like tʰɔjA1-UC in Yang Zhuang varieties and tʰaɰA1-UC in Tiandeng) has a unexpected unaspirated initial stop t- and a tone corresponding to be in the A1-U row. The 344 term ‘gizzard’ with an unaspirated initial stop t- is also found in some other Tai varieties like Yongnan Zhuang Varieties which have an additional tonal split due to the proposed ASC row. The irregularity involved in this term needs further investigation. 3) The term ˀdaŋ353 ‘to make fire’ has an unexpected Tone A1-G, which is expected to be C1-G in the tone box designated, vs. ˀdaŋC1-G in Longsang (NT) and Debao Yang Zhuang (CT). This term may have been undergoing an analogical change or phonological contamination by replacing its original *C tone to *A tone from *vɛjA ‘fire’ (cf. Li 1977: 79, 286) or *wɤjA ‘fire’ (cf. Pittayawat Pittayaporn 2009: 207) according to the frequently used compound word *ʔdaŋC *vɛjA/*wɤjA ‘to make fire’. 4) The collocation between tones and vowels on boxes DS1-A, DS1-C, DS1-UC, and DS1/2 is interesting as follows: boxes DS1-A, DS1-C, DS-UC, and DS1/2 have a low-rising tonal value 13, and since a short vowel hardly fulfils a low rising tonal value, items with these groups of tones have been lengthened their original short vowels for coordinating to this tonal value, like ɕa:k13 (DS1-A) ‘vegetable’, he:t13 (DS1-A) ‘to do’, ha:p13 (DS1-A) ‘to shut’, ma:t13 (DS1-C) ‘flea’, na:k13 (DS1-A) ‘heavy’, wa:t13 (DS1-A) ‘have a cold’, ɬe:p31 (DS1/2-C > DS2) ‘ten’, and ɬo:k13 (DS1/2-C> DS2) ‘cooked’, while other items like het11 (DS1-A > DS2) ‘centipede’, wak11 (DS1C > DS2) ‘to nod’, lap11 (DS1-C > DS2) ‘to close (eyes)’, and kʰɔp11 (DS1/2-A > DS2) ‘to bite’ have been changed their original tones to a low-level tone for keeping their vocalic length. 5) The item ma:t13 (DS1-C) ‘flea’ has a free variation mat11 (DS1-C > DS2). The item mat11 (DS2) ‘grain’ also has a free variation mat34 (DS1ʼ) ‘grain’. These phenomena may have internal relationship. The term mat11 (DS2) ‘grain’ may be changed its original DS2 tone to the derived DS1' tone (tonal value 34) for avoiding being homophonous with the item mat11 (DS1-C > DS2) ‘flea’ after being changed from DS1-A/C to DS2 as a free variation. That is to say, mat11 (DS1-C > DS2) ‘flea’ which is the free variation of ma:t13 (DS1-C) ‘flea’ might have urged the tone of mat11 ‘grain’ to change from DS2 (11) to DS1' (34) to form its free variation. However, this process is not completed; therefore, both mat11 and mat34 stand for ‘grain’. 23.7 Other linguistic changes reflected in the wordlist of this dialect: 1) Sound changes on the term ham45 (A1-A) ‘in addition’ are as follows: like some other CT and SWT varieties such as Debao Urban and Northern Tai, the initial of this term has been irregularly changed from *tʰ- to h- due to its grammaticalized process (to add (v.) > in addition (conj.)); probably due to the same 345 grammaticalized process, the term has an unexpected vowel -a, which is expected to be -e according to other Tai varieties, like he:m453 in Debao Urban, hɛ:m13 in Chiang Mai, and tʰɛ:m24 in Bangkok. 2) The term ma:j45 ‘to face up’ has an unexpected initial m-, which is expected to be ŋ- in the tone box designated according to the correspondence to the cognate in other Tai varieties, like ŋa:jA1-C in Thai and in Yang Zhuang varieties. 3) PT *-e (cf. Pittayawat Pittayaporn 2009: 223) is diphthongized as -ej like in kej33 (B1-U) ‘old’. This is very rare since PT *-e is generally preserved as a mid-vowel in most CT and NT varieties whereas is lowered as -ɛ in some CT and most SWT varieties. 4) PT *-o (cf. Pittayawat Pittayaporn 2009: 228) is diphthongized as -ow like in pʰow33 (B2) ‘male; father’. This is very rare since PT *-o is generally preserved as a mid-vowel in most CT and NT varieties whereas is lowered as -ɔ in some CT and most SWT varieties. 5) PT *-ʰɲ- (cf. Pittayawat Pittayaporn 2009: 228) is yodicized as nj- in this dialect like in nja:24ʔ (C1-C) ‘grass’. This is very rare since PT *-ʰɲ- is generally preserved as ɲ- or is changed to a yod j- in the vast majority of Tai varieties. 6) PT *r- has been merged into *s- (cf. Li 1977:152, Pittayawat Pittayaporn 2009:115) or *ɬ- (cf. Liao & Shen 2012) to be a modern ɬ- in this language, like in the terms ɬaj33 (B2) ‘dryfarmland’ and ɬa:j11 (C2) ‘bad’. 346 Language 24 Bac Va, Nung, CT 24.1 Language information Language Bac Va ISO 639-3 Code nut Autonym Nung Exonym Nung Fan Sling Data Source Hudak 1997: 1147-1186 Data Info. The data of Bac Va are adopted from Gedney’s study on the Tai dialect of Bac Va (Hudak 1997). According to the original material, Gedney collected the data from Mr. Tran Phuc Ky on July 19, 1964, in Dalat, Vietnam. This language is commonly called Nung, and is spoken in northeastern Vietnam near Phu Lang Thuong (Bac Giang). 24.2 Basic patterns of tonal development of this dialect 1) As shown in the chart below, a tonal flip-flops occurs on C in this dialect. Smooth Syllable Phonation types Checked Syllable A B C DL DS Aspirations A1-A: 14 B1-A: 21ʔ C1-A: 11ʔ DL1-A: 33 DS1-A: 14 Continuants A1-C: 45 B1-C: 14 C1-C: 11 DL1-C: 55 DS1-C: 14 Unaspirated stops A1-U: 45 B1-U: 14 C1-U: 11ʔ DL1-U: 55 DS1-U: 14 A1-UC: 14 B1-UC: 21ʔ C1-UC: 11ʔ DL1-UC: 33 DS1-UC: 14 A1-G: 45 B1-G: 21ʔ C1-G: 11 DL1-G: 55 DS1-G: 14 Proto-voiced sounds A2: 33 B2: 32 C2: 55ʔ DL2: 33 DS2: 32 Proto-voiced aspirations & A1/2-A: 14 breathy sounds A1/2-C: 45 B1/2: 21ʔ C1/2: 11ʔ DL1/2: 33 DS1/2: 14 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds 2) Patterns of tonal split and merger are as in the following two charts: Vertically: three-way splits and two way splits Tonal category Register 1 Register 1' Register 2 A [1-C, 1-U, 1-G, 1/2-C] [1-A, 1-UC, 1/2-A] [2] B [1-C, 1-U] [1-A, 1-UC, 1-G, 1/2] [2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] DL [1-C, 1-U, 1-G] [1-A, 1-UC, 2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] [2] In conclusion, tones in this dialect are of non-straightforward tonal development. Secondary tonal split patterns are found in Columns A, B, and DL. 347 Horizontally: three individuals and four mergers A[1-C, 1-U, 1-G, 1/2-C] Individual B[1-A, 1-UC, 1-G, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] A[1-A, 1-UC, 1/2-A] = B[1-C, 1-U]= DS[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] Merger A[2] = DL[1-A, 1-UC, 2, 1/2] B[2] = DS[2] C[2] = DL[1-C, 1-U, 1-G] 24.3 Tonal characteristics & tone occurrences: seven distinct tones. In the original material, Gedney determines eight tones to the open syllables. However, Tone 6 (occurring on Liao & Shen’s C1-A, C1-U, C1-UC, C1/2 tonal categories) and Tone 7 (occurring on Liao & Shen’s on C1-C and C1-G) are both low level 11, and the only difference is that there is a the glottalized feature on Tone 6. Just like Tones B1 and C1' in Baoxu, since Tones 6 and 7 are in a complementary distribution with different groups of initial consonants, the glottalized cannot be treated as phonemic or tonemic. Here I treat them to be the same tone, and reduce the tone numbers of this dialect from eight to seven. Tone 1: Low-rising 14 occurs in A[1-A, 1-UC, 1/2-A], as well as in B[1-C, 1-U] and DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 2: High-rising 45 occurs in A[1-C, 1-U, 1-G, 1/2-C]. Tone 3: Mid level 33 occurs in A[2] and DL[1-A, 1-UC, 2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low-falling 21 occurs in B[1-A, 1-UC, 1-G, 1/2]. Tone 5: Mid-falling 32 occurs in B[2] and DS[2] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Low level 11 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]. Tone 7: High level 55 occurs in C[2] and DL[1-C, 1-U, 1-G] to form a complementary distribution in smooth and checked syllables respectively. 24.4 Phonation types or voice qualities of tones in this dialect According to Gedney’s original text, Tone 4 (B[1-A, 1-UC, 1-G, 1/2]), Tone 7 (C2) and Tone 6 (C[1-A, 1-C, 1-U, 1-UC, 1-G, 1/2]) all have glottalized feature (Hudak 1997: 1147). We do not know which type of glottalization, final glottal constriction or creakiness, these tones are of since they are not mentioned in the original text. All the glottal constrictions are not phonemic. 348 Language 25 Yongning Baiji, Yongnan Zhuang, CT/NT? 25.1 Language information Language Sanxiang YN ISO 639-3 Code zyn Autonym ɬa:mA1 jiəŋA1; tʰo:C1-A Exonym Sanxianghua ( 鄉話); jajC1-G Data Source 2 LRPs Reference Zhang et al. 1999: 42-44, 119-127, 595-808. Variety Name Baiji of Sanxiang Yongnan Zhuang (YN). This variety is spoken in the rural and urban areas of Baiji Township (百濟鄉), Yongning District (邕 區), Nanning City, Guangxi Zhuang Autonomous Region, China. According to my own field trip, this language is commonly known as Sanxiang (ɬa:mA1 jiəŋA1), which literally means ‘three townships’ in Chinese and in Zhuang. The main areas of this language are the three townships Baiji (百濟), Nalou (那樓) and Zhonghe (中和) in Yongning District of Nanning City. Although Sanxiang is a language under Yongnan Zhuang (ISO 639-3: zyn), it is not intelligible with other YN varieties like Shuangding, Long’an and Nung An. Besides, speakers of Sanxiang have strong sense of internal identity towards to speakers of any varieties of Sanxiang, but lack of this identity with speakers of other YN varieties and other Zhuang languages. Varieties of Sanxiang may put together to consist a new individual language with ISO 639-3 code in further research and surveys. 25.2 Information of data source Serial number of LRP LRP1 Place of data elicitation Date Railway Hotel, Nanning City, Guangxi Zhuang Autonomous Region, China. Given Name Q̀ngtáng (慶唐) Gender male Birth Place Tunsu Community (屯蘇坡, ˀbua:n (屯林村, tʰuən lum tɕʰɔ:n A1 District (邕 Places lived May 13, 2013 A2 A1 C1 tʰuən ɬɔ: A1 Family Name Xì (謝) Age 36 A1 in local Zhuang), Tunlin Village in local Zhuang), Baiji Township (百濟鄉), Yongning 區), Nanning City, Guangxi Zhuang Autonomous Region, China. Tunsu Community (16 years). Pumiao Town (蒲廟鎮), the main town of Yongning County (nowadays Yongning District) (3 years). The main town of Daxin County (1 year). He is now in Nanning City. Time period of living at the present residence Occupation self-employed entrepreneurs Languages spoken Level Baiji Sanxiang YN mother 16 years Education Language Remarks tongue 349 junior high school Baihua fluent This language is a Cantonese variety in Guangxi and is a lingua franca in the main town of Yongning District and in Nanning City. Standard Mandarin fluent This language is the national language of the LRP’s country. Pinghua intermediate A Sinitic Language spoken in the Central part of Guangxi, China. It is closed to Cantonese, and is supposed to be the official language in Guangxi in history before the Guiliu Southwestern Mandarin came to Guangxi. Naknai intermediate A Cantonese variety spoken in Qinzhou City of Guangxi. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Baiji Sanxiang as native, he can communicate in Baihua. LRP’s mother was born in Nafeng Village (那豐村, nuəA2 fʊŋA1 in local Yang Zhuang), Naxiang Township (那鄉鎮), Qinzhou City, Guangxi. She is monolingual in her Sanxiang YN. She lives with LRP in Nanning City now. Serial number of LRP LRP2 Place of data Hengyang Xilu Nantie Er Jie 1, Nanning City, Guangxi Zhuang Autonomous elicitation Region, China. Given Name Yù ngāo (運高) Gender male Birth Place Date Tanhuai Community (壇槐坡, tʰəm tʰuən lum tɕʰɔ:n A1 A2 A1 A2 ma:j A2 May 19, 2013 Family Name Lù (陸) Age 39 in local Zhuang), Tunlin Village (屯林村, in local Zhuang), Baiji Township (百濟鄉), Yongning District (邕 區), Nanning City, Guangxi Zhuang Autonomous Region, China. Places lived Tanhuai Community (15 years). Hengyang City, Hunan Province (4 years). Pingguo County (2.5 years). Baise City (10 years). He is now living in Nanning. Time period of living at the present residence Occupation engineering supervision Languages spoken Level Baiji Sanxiang YN mother tongue Baihua fluent 6 years Education on-the-job undergraduate Language Remarks This language is a Cantonese variety in Guangxi and is a lingua franca in the main town of Yongning District and in Nanning City. Standard Mandarin fluent This language is the national language of the LRP’s country. Guiliu intermediate A Sinitic Language spoken in the Central part of Guangxi, China. It is closed to Cantonese, and is supposed to be the official language in Guangxi in history before the Guiliu Southwestern Mandarin came to Guangxi. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Baiji Sanxiang as native, he can communicate in Baihua and a little Standard Mandarin. LRP’s mother was born in Sanlexia Community ( 350 樂 坡), Shitang Village (石塘村), Lingli Township (伶俐鎮), Qingxiu District (青秀區), Naning City, Guangxi. Her mother tongue is of Yongbei Zhuang (a NT language). Besides, she can speak Baiji Sanxiang YN and Standard Mandarin, and a little Baihua. She lives with LRP in Nanning City now. 25.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flops in this dialect. Smooth Syllable Checked Syllable Phonation types A B C Aspirations A1-A: 53 B1-A: 343 C1-A: 445ʔ Continuants A1-C: 53 B1-C: 343 C1-C: 445ʔ Unaspirated stop A1-U: 53 B1-U: 343 C1-U: 445ʔ DL DS DL1-A: 34 DL1’-A: 445 DS1-A: 55 DL1-C: 34 ProtoVoiceless Sounds DL1’-C: 445 DS1-C: 55 DL1-U: 34 DS1-U: 55 DL1'-U: 445 Unaspirated stop + *r clusters A1-UC: B1-UC: C1-UC: 53 343 445ʔ A1-G: 53 B1-G: 343 C1-G: 445ʔ DL1-UC: 34 DS1-UC: 55 DL1-G: 34 Glottal Sounds DS1-G: 55 DL1'-G: 445 DL2: 31 Proto-voiced sounds A2: 331 B2: 31 C2: 445ʔ DS2: 33 DL2': 445 Proto-voiced aspirations & breathy sounds A1/2: 331 B1/2: 31 C1/2: 445ʔ DL1/2: 31 DS1/2: 33 2) The tonal complexity in the C and DL columns (the unexpected tones are all in dark-orange) does not suggest the tonal flip-flop or the non-straightforward development. In this dialect, most items with expected C2 tone have merged their original C2 tone into C1, a high-level-rising tone 445, to be treated as the merged C2. The derived tone DL1’ on DL1-A and DLA-C has been found from the wordlist, like tʰa:k445 (DL1'-A) ‘to prop’ and ɬa:k445 (DL1’-C) ‘pestle’. According to a supplementary investigation to this dialect by confirming with LRP2, I verified that this high-level-rising (DL1') are not only found on DL1-U and DL1-G, but also on found on DL2 as a derived tone of DL2. According to the examples provided by LRP2, the derived tones DL1’ and DL2’ are both a high-level-rising 445 (vs. DL1=223 and DL2=31), and are mainly used for Middle Chinese loans, like lwak445 (DL2’) ‘happy’ (<樂 vs. lɤ51 Mandarin, lɔk22 Cantonese)，ŋwa:k445 351 (DL2’) ‘prison’ (<獄 vs. y51 Mandarin, jʊk22 Cantonese), hwa:k445 (DL2’) ‘to learn’ (<學 vs. ɕyɛ35 Mandarin, hɔk22 Cantonese), lwa:p445 (DL2’) ‘dried meat’ (<臘 vs. la51 Mandarin, la:p22 Cantonese), tʰwa:p445 (DL2’) ‘to tread’ (<踏 vs. tʰa53 Mandarin, ta:p22 Cantonese), lo:p445 (DL2’) ‘green’ (<綠 vs. ly51 Mandarin, lʊk22 Cantonese), me:k445 (DL2’) ‘wheat’ (<麥 vs. mai53 Mandarin, mak22 Cantonese), and etc. These merged or derived tones are of secondary aberrant development due to MC loans and are out of the evaluation criteria of the conditioning of tonal splits. Therefore, they cannot be a reference substance of tonal flip-flops or the directionality of tonal development. Actually this kind of complicated tonal splits on dead syllable due to MC loans are found in all the YN varieties investigated in this thesis (L25-29). They are omitted in this study because they are beyond the tonal conditioning factors like initials or vowel-length. 3) Patterns of tonal split and merger are as in the following two charts. Vertically: two-way splits and no-split Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] DL DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] [1'-A, 1'-C, 1'-U, 1'-UC, 1'-G] [2'] [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development. Similar to the previous point 1), the derived tones on DL1-A and DL1-C are not conditioned by any homogeneous initial groups. Many of them are used for Middle Chinese loans. They are of secondary aberrant development and are out of the evaluation criteria of the conditioning of tonal splits. Therefore, they cannot be a reference substance of the directionality of tonal development. Horizontally: no individual and five mergers; there are five tones on checked syllables, and each of them is on the complementary distribution with those five tones on smooth syllable. A[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] A[2, 1/2] = DS[2, 1/2] Merger B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] B[2, 1/2] = DL[2, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] = DL[1-A', 1-C', 2'] 352 25.4 Tonal characteristics & tone occurrences: five distinct tones: Tone 1: High-falling 53 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G], as well as high-level 55 in DS [1-A, 1-C, 1-U, 1-UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively. Tone 2: Mid level-falling 331 occurs in A[2, 1/2], and mid level 33 in DS[2, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Low-rising-falling 343 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G], and mid rising 34 in DL[1-A, 1-C, 1-U, 1-UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Mid-falling 31 occurs in B[2, 1/2] and in DL[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-level-rising 445 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2], and in the derived DL[1'-A, 1'-C, 1'-U, 1'-G, 2'], to form a complementary distribution in smooth and checked syllables respectively. 25.5 Phonation types or voice qualities of tones in this dialect The tone on the C column has a glottal constriction, like ha:445ʔ ‘five’, pʰa:j445ˀ ‘to walk’, ɬom445ˀ ‘sour’, taŋ445ˀ ‘to establish’, b̤wa:n445ˀ ‘village’, and nam445ˀ ‘water’. Syllables with B1 and B2 Tones have a slight breathy final -ɦ. All tones on the lower register except those on the C column always accompany a slight breathy voice. All the glottal constrictions are not phonemic. 25.6 Sound changes involving tones reflected in the wordlist of this dialect 1) Labialization is very common seen before -a: and -a with initials on earlier voiced and glottal rows in this dialect, like nwa:331 (A2) ‘rice field’, dwa:331 (A2) ‘to spread on’, kʰwa:331 (A2) ‘to stuck’, d̤wa:31 (B2) ‘river’, and b̤ wa:tDL1-G ‘time’. Sometimes the labialized segment -w- presents as an unrounded -ɰ-, like in t ̤ɰak55 (DS1-G) ‘deep night, to fall in sleep’. The term lwa:31 (B2<A1) ‘to seek’ been has changed its tone from A1-A to B2 probably due to its labialized initial, which is commonly seen before -a: and -a on lower register tones69. On the contrary, the term ma:k34 (DL1<DL2) ‘plenty’ been has changed its tone from DL2 to DL1 probably due to 69 However, a change from A1 to A2 is more regular than from A1 to B2. The reason why the tone of lwa:31 (B2<A1) ‘to seek’ is not from A1 to A2 is not clear. Another possiblitiy is that the term lwa:31 (B2<A1) ‘to seek’ is actually not cognate with*xraA (cf. Li 1977: 233) or*kraA (cf. Pittayawat Pittayaporn 2009: 353) reflected in other Tai varieties. 353 the loss of its labialization, which is commonly seen on lower register tones. Therefore, it becomes homophonous with ma:k34 (DL1-C) ‘fruits’. In brief, labialization before -a: and -a in this dialect is one of the signatures of the earlier voiced and glottal registers. 2) Pre-glottalized stops ˀb- and ˀd- developed from proto-glottal sounds in other Tai varieties have been changed to breathy stops in this dialect, and the voicing of these stops are not phonemic since sometimes they are p̤- and t ̤- but sometimes they are b̤ - and d̤-, cf. the sound files of t ̤wa:343 (B1-G) vs. d̤wa:343 (B1-G) ‘to scold’. Here the forms b̤ - and d̤- are treated as the phonemes because they have developed from earlier glottalized voiced stops. The breathy voice is their phonemic feature to make them contrast with the plain voiceless stops p- and t- in the 1-U row. For exmaple, d̤wam53 (A1-G) ‘black’, b̤wa:343 (B1-G) ‘shoulder’, b̤ wa:n445 (C1-G) ‘village’, b̤ wa:t334 (DL1-G) ‘time’ and d̤wak44 (DS1-G) ‘deep night’. Moreover, earlier voiced stops*b- and*d- have been merged into b̤ - and d̤- which developed from earlier *ʔb- and *ʔd- to complementarily distribute on the higher register and the lower register respectively. For example, d̤wa:331 (A2) ‘to spread on’, d̤wa:31 (B2) ‘river’, b̤ wa:k31 (DL2) ‘insane’, and b̤wak33 (DS2) ‘very tired’. However, unlike that b̤- and d̤- on the higher register sometimes present as p̤- and t ̤-, b̤ - and d̤- on the lower register always present as voiced sounds. In brief, breathy voice (including voiced aspiration) in this dialect is another signature of the earlier voiced and glottal registers. 3) The tone of paj331 (A2) ‘to go’ is expected to be A1, but it is actually A2 in this dialect. Similarly, in LRP1’s dialect the tone of mɛ:343 (B1) ‘mother’ is expected to be B2, but it is actually B1 in this dialect. These tone changes need more data to determine with. 4) The tone of kʱɰej331 (A2) ‘to ride on’ is expected to be B2 (<B1/2) due to the reflection in the vast majority of Tai varieties, but it is actually A2 in this variety. The reason may be that this etymon is actually a Middle Chinese loan (vs. tɕʰi35 (A2) Mandarin, kʰei21 (A2) Cantonese), and the tonal category is also adopted from its Sinitic source. 5) Unlike other CT/SWT varieties, the terms nok31 (DL2<DL1-C) ‘deaf’ and tʰɔk31 (DS1-UC) ‘thin bamboo stripes’ which are expected to have a long vowel in DL1 designated have a short vowel like NT varieties. However, the tones of these two terms do not merge into their counterpart DS1 tones (55) like other Yongnan 354 Zhuang varieties and all NT varieties, but merge into DL2 tone (31). In addition, their tones are slightly different from the common DL2 because of the difference between their vocalic lengths. Since items with expected DL2 tone normally have a long vowel, their tones present a full mid-falling 31. But the DL2 tone on these two items lower their pitch to be like 21 because their short vowel. The terms lit31 (DL2) ‘blood’, fet31 (DL2) ‘wing’ and met31 (DL2) ‘knife’ which are expected to have a DL2 tone are similar to these two terms to have a lower pitch on their tone to be like 21, because their vowels are also shorten. Nevertheless, this lower pitch is treated as a conditioned variation of DL2 and is not an independent tone. Therefore, it is transcribed as 31 as the common DL tone. 6) The original tonal value of DL1 tones is mid-rising 34 (=B1) like tʰa:p34 (DL1) ‘to carry’ and ɬa:p34 (DL1) ‘cockroach’, but some of the items has a derived tone DL1’ whose tonal value is high level-rising 445 (=C1), like tʰa:k445 (DL1’-A) ‘to prop’ and ɬa:k445 (DL1’-C) ‘pestle’. 7) The initial and the tone of the term tʰap55 (DS1-A) ‘to receive’ are unexpected. One may say that this term might not be cognate with *rapD ‘to receive’ reflected in other Tai varieties. However, if we think over this problem with deep consideration, we may see that this term may have been changed its tone from DS2 to DS1 probably due to the loss of its labialization, which commonly precedes -a: and -a on lower register tones. And then the initial may have been changed from *r- to the same POA tʰ- because there are no modern initial consonant developing from *r- on the higher register, and tʰ- is actually a common initial on the higher register. 25.7 Other linguistic changes reflected in the wordlist of this dialect. 1) All aspirated initial stops and fricatives on the lower register like tʱ-, pʱ-, kʱ-, and tɕʱ- have the voiced aspiration, which is phonetically breathy. For example, kʱwa:331 (A2) ‘to stuck’, kʱwa:331 (A2<A1/2) ‘right side’, tʱiaŋ331 (A2<A1/2) ‘pavilion’, pʱun331 (A2<A1/2) ‘to sharpen’, pʱu:31 (B2) ‘male’, tʱu:31 (B2<B1/2) ‘bean’, tʱi:31 (B2<B1/2) ‘dense’, kʱɰej331 (A2) ‘to ride on’, and pʱu:35 (C2<C1/2) ‘CLF (of a person)’. 2) The initial of ʔa:jA1-G ‘to turn face up’ is unexpected since it is designated into A1C. 355 3) The term kʰɔ:445 (C1-UC) ‘CLF (of stones)’ is expected to have a final consonant -n (vs. kɔ:n41ˀ in Thai of SWT, kʰo:n24ˀ in Debao Yang Zhuang of CT, and ko:n213ˀ in Longsang of NT), but it might have lost this final consonant. 4) As those in other NT varieties, the following terms put in the wordlist for the revised tone box (Liao & Shen 2012) has different etyma, like lomA1-A ‘mortar’, tawC1 ‘to come’, fe:t31 (DL2) ‘wing’, θe:pDL1 ‘fishhook’, and hatDS1-A ‘morning’. 5) The vowel of tʰɔ:kDL1-UC ‘cracking’ is unexpected. This sound change needs more data to determine with. 6) The meaning of the term ʔa:pDL1-G ‘to apply (medical ointment, powder or etc.), to spread on’ might be a result of semantic change from ‘to bath’. 7) The initial consonant of ɬak55 (DS1-UC) ‘grasshopper’ is unexpected. It reflects tin SWT and NT, and tʰ- in most CT varieties. 356 Language 26 Nanning Shuangding, Yongnan Zhuang, CT/NT? 26.1 Language information Language Yongnan Zhuang ISO 639-3 Code zyn Autonym tʰo:C1 Exonym ja:ŋA1;Tuhua Data Source 1 LRP Variety Shuangding (township name) Urban of YN is spoken at the urban area of Shuangding Name Township, Xixiangtang District, Nanning City, Guangxi Zhuang Autonomous Region, China. All Shuangding varieites including Urban have an exonym ja:ŋA1-G called by other Zhuang groups surrounding them. This is interesting because this exonym is found in Yang Zhuang and some NT varieties like Debao Longsang (L31), which are far away from Shuangding YN. This phenomenon further suggests that the definition of Yang Zhuang in ISO 639-3 has problems (cf. Language Information in L11 and L14, and L31). 26.2 Information of data source Serial number of LRP LRP1 Date May 5, 2013 Boluo Community (菠萝坡), Xingping Village (興平村), Shuangding Place of data elicitation Township (雙 鎮), Xixiangtang District (西鄉塘區), Nanning City, Guangxi Zhuang Autonomous Region, China. Given Name Xiángjiā (祥嘉) Family Name Liáng (梁) Gender male Age 50 Birth Place Boluo Community (菠萝坡) or Jiujie Street (舊街, literally Old Street), Xingping Village (興平村), Shuangding Township (雙 鎮), Xixiangtang District (西鄉塘區), Nanning City, Guangxi Zhuang Autonomous Region, China. Places lived He lives in Boluo Community over all his life. Time period of living at the present residence Occupation farmer Languages spoken Level Shuangding YN mother tongue Standard Mandarin intermediate Baihua intermediate Pinghua a little 50 years Education junior high school Language Remarks This language is the national language of the LRP’s country. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in Longshan Village (龍山村, pla:A1 plukDL2), Shuangding Township (雙 鎮), Xixiangtang District (西鄉塘區), Nanning City. Besides his Shuangding YN as native, he could communicate in Baihua. LRP’s mother was born in Jiujie Street (舊街), Shuangding Township (雙 357 鎮), Xixiangtang District (西鄉塘區), Nanning City. Besides her mother tongue Pinghua (of a Sinitic language), she could speak Shuangding YN. Remarks Boluo Community (菠萝坡) of Xingping Village (興平村) is actually the old street of Shuangding Township. It is contiguous with the new street of Shuangding Township. Local people treat the Zhuang variety in Boluo Community and in the urban area of Shuangding Township the same urban variety of Shuangding YN. In my investigation in Shuangding Township, three LRPs accepted my interview. I found that the other two LRPs are from other two villages Licun of Shuangding Township and Zhongyi of Nalong Township. Although they insist that they speak in the same language with Shuangding Urban, the sound systems of their own varieties are different. For example, both the varieties from Licun and Zhongyi has aspiration to correspondence to Shuangding Urban’s unaspirated stop + -l- clusters, like pʰa:33 vs. pla:33 ‘stone mountains’ in Licun/Zhongyi and Shuangding Urban respectively. Besides, PT *ɯa has been changed to a rounded long -y: in Zhongyi (like in Debao Urban and Dalong) but has been merged into the unrounded -i: in Shuangding Urban and Licun, such as my:B1-G vs. mi:B1-G ‘boring’ in Zhongyi and in Shuangding Urban/Licun respectively. Therefore, I finally decide to reduce the number of LRPs of Shuangding Urban from 3 to 1 because data from a single identical variety is needed to analyze. 26.3 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there is no tonal flip-flops in this dialect. Smooth Syllable Checked Syllable Phonation types A B C Aspirations A1-A: 33 B1-A: 24 C1-A: 35ʔ Continuants A1-C: 33 B1-C: 24 C1-C: 35ʔ Unaspirated stop A1-U: 33 B1-U: 24 C1-U: 35ʔ A1-UC: 33 B1-UC: 24 C1-UC: 35ʔ A1-G: 33 B1-G: 24 C1-G: 35ʔ A2: 31 B2: 22 C2: 44ʔ A1/2: 31 B1/2: 22 C1/2: 44ʔ DL DS DL1-A: 24 DL1’-A: 35 DS1-A: 35 DL1-C: 24 ProtoVoiceless Sounds DL1’-C: 35 DS1-C: 35 DL1-U: 24 Unaspirated stop + *r clusters DL1’-U: 35 DL1-UC: 24 DS1-U: 35 DS1-UC: 35 DL1-G: 24 Glottal Sounds DL1’-G: 35 DS1-G: 35 DL2: 22 Proto-voiced sounds Proto-voiced aspirations & breathy sounds 358 DL2’: 44 DL1/2: 22 DS2: 44 DS1/2: 44 2) The tonal complexity on the DL column (in dark-orange and blue) does not suggest the tonal flip-flop or the non-straightforward development. The derived tone DL1’ on DL1-A and DLA-C has been found from the wordlist, like he:k35 (DL1’-A) ‘to prop’ and ɬa:k445 (DL1’-C) ‘pestle’. According to a supplementary investigation to this dialect by confirming with LRP1, I verified the mid-rising 35 tone on DL[1'-A, 1'-C, 1'-U, 1'-G] (in dark-orange) as a derived tone of DL1, and a high-level 44 tone on DL2' (in blue) as a derived tone of DL2. According to LRP1, both the derived tones DL1’ (vs. DL1=24) and DL2’ (vs. DL2=22) are mainly used for Middle Chinese loans, like tʰi:t35 (DL1’-A) ‘iron’ (<鐵 vs. tʰiɛ214 Mandarin, tʰit33 Cantonese)，tʰa:p35 (DL1’-A) ‘pagoda’ (<塔 vs. tʰa:214 Mandarin, tʰa:p33 Cantonese), si:t35 (DL1’-C) ‘snow’ (<雪 vs. ɕyɛ214 Mandarin, syt33 Cantonese), wa:t35 (DL1’-C) ‘to dig’ (<挖 vs. wa55 Mandarin, wa:t33 Cantonese), ta:p35 (DL1’-U) ‘to answer’ (<答 vs. ta35 Mandarin, ta:p33 Cantonese), ʔa:p35 (DL1’- G) ‘duck’ (<鴨 vs. ia55 Mandarin, ʔa:p33 Cantonese), ʔa:k35 (DL1’-G) ‘ferocity’ (< 惡 vs. ɤ51 Mandarin, ʔɔk33 Cantonese), la:p44 ‘dried meat’ (DL2’) (<臘 vs. la51 Mandarin, la:p22 Cantonese), ŋi:t44 ‘month’ (DL2’) (<月 vs. yɛ51 Mandarin, jyt22 Cantonese) and etc. These derived tones are of secondary aberrant development and are out of the evaluation criteria of the conditioning of tonal splits. Therefore, they cannot be a reference substance of tonal flip-flops or the directionality of tonal development. 3) Patterns of tonal split and merger are as in the following two charts: Vertically: two-way split Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] [1'-A, 1'-C, 1'-U, 1'-UC, 1'-G] [2'] [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DL DS In conclusion, tones in this dialect are of straightforward tonal development. The complicated tonal split patterns are ignored here because they are aberrant development due to MC loans (cf. L25). 359 Horizontally: two individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G] A[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] Merger B[2, 1/2] = DL[2, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] C[2, 1/2] = DL[2, 1/2] 26.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Mid level 33 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: Mid-falling 31 occurs in A[2, 1/2]. Tone 3: Low-rising 24 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G] and DL[1-A, 1-C, 1-U, 1UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low level 22 occurs in B[2, 1/2] and in DL[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-rising 35 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G] as well as in DL[1'-A, 1'-C, 1'-U, 1'-UC, 1'-G] and DS[1-A, 1-C, 1-U, 1-UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: High-level 44 occurs in C[2, 1/2] as well as in DL [2'] and DS[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. 26.5 Phonation types or voice qualities of tones in this dialect. Tones C1 and C2 have a final glottal constriction, and the glottalization on C2 are clearer like to nearly assimilate the previous nasals to become stops, like ha:35ˀ ‘five’, na:j35ˀ ‘face’, pla:j35ˀ ‘to walk’, ʔɔ:j35ˀ ‘sugarcane’, luj44ˀ ‘bad’, and namʔ44 ‘water’. Tones B1 and B2 have slight breathy finals. However, all these voice qualities are not phonemic. 26.6 Sound changes involving tones reflected in the wordlist of this dialect. 1) The tone of the term lu:ŋ24ʔ (B1) ‘cattle pen (<ground floor)’ is expected to be B2, but it is actually B1 in this dialect. The reason needs more data to determine with. 2) Unlike other CT/SWT varieties, in this dialect the terms nok35 (DS1-C) ‘deaf’, tʰok35 (DS1-UC) ‘thin bamboo stripes’, nət35 (DS1-G) ‘sunshine’, and tək44 (DS1/2) ‘correct’ 360 which are expected to be long vowels in DLs designated have short vowels and have been merged their tones into their counterpart DS boxes like NT varieties. 3) The tone of the term ta:k24 (DL1) ‘dry blood sucker’ is expected to be DL2 (22), but in this dialect it is an unexpected DL1. The reason is still unknown. 26.7 Other linguistic changes reflected in the wordlist of this dialect. 1) All pre-glottal sounds in PT have been changed to sonorants in this dialect, the rules are *ʔb- > m-, *ʔd- > n-, and *ʔj- > j-. 2) The initial consonant of the term kʰa:ŋ35 (C1) ‘fishbone’ is expected to be an unaspirated stop k-, but in this dialect it is an aspirated stop kʰ-. 3) The initial consonant of the term law35 (C1) ‘head’ is expected to be an unaspirated stop k-, but in this dialect it is a lateral l-. 361 Language 27 Long’an Xiaolin, Yongnan Zhuang, CT/NT? 27.1 Language information Language Yongnan Zhuang ISO 639-3 Code zyn Autonym tʰo:C1-A Exonym Long’an Zhuang Data Source Zhang et al. 1999:42-44, 127-130, 595-808. Data Info. Long’an Xiaolin of YN. This variety is spoken in Xiaolin Township, Long’an County, Nanning City, Guangxi Zhuang Autonomous Region, China. 27.2 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flops in this dialect. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 34 B1-A: 213 C1-A: 55 DL1-A: 213 DS1-A: 21 Proto- Continuants A1-C: 34 B1-C: 35 C1-C: 55 DL1-C: 35 DS1-C: 55 Voiceless Unaspirated Sounds stops A1-U: 34 B1-U: 35 C1-U: 55 DL1-U: 35 DS1-U: 55 A1-UC:34 B1-UC: 213 C1-UC: 55 DL1-UC: 213 DS1-UC: 21 A1-G: 34 B1-G: 213 C1-G: 55 DL1-G: 213 DS1-G: 21 A2: 22 B2: 21 C2: 33 DL2: 33 DS2: 21 A1/2: 22 B1/2: 21 C1/2: 33 DL1/2: 33 DS1/2: 21 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts: Vertically: three-way splits and two way splits Tonal category Register 1 Register 1' A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-C, 1-U] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DL [1-C, 1-U] [2, 1/2] DS [1-C, 1-U] [1-A, 1-UC, 1-G] [1-A, 1-UC, 1-G] Register 2 [1-A, 1-UC, 1-G, 2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development. 362 Horizontally: two individuals and five mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G] A[2, 1/2] B[1-C, 1-U] = DL[1-C, 1-U] B[1-A, 1-UC, 1-G] = DL[1-A, 1-UC, 1-G] Merger B2[2, 1/2] = DS[1-A, 1-UC, 1-G, 2, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-C, 1-U] C[2, 1/2] = DL[2, 1/2] 27.3 Tonal characteristics & tone occurrences: seven distinct tones Tone 1: Mid-rising 34 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: Low level 22 occurs in A[2, 1/2]. Tone 3: Mid-rising 35 occurs in B[1-C, 1-U] and DL[1-C, 1-U] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low-falling-rising 213 occurs in B[1-A, 1-UC, 1-G] and DL[1-A, 1-UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low-falling 21 occurs in B[2, 1/2] and DS[1-A, 1-UC, 1-G, 2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: High level 55 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G] and DS[1-C, 1-U] to form a complementary distribution in smooth and checked syllables respectively. Tone 7: Low level 33 occurs in C[2, 1/2] and DL[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. 27.4 Phonation types or voice qualities of tones in this dialect. We do not know the phonation types or voice qualities of tones in this dialect since they are not mentioned in the original text. 363 Language 28 Daqiu, Nung An, Yongnan Zhuang, CT/NT? 28.1 Language information Language Yongnan Zhuang A2 ʔa:n A1-G ISO 639-3 Code zyn Exonym Nung An Autonym nɔŋ Data Source 1 LRP Reference Edmondson (2002); Zhang et al. (1999: 42-44, 127-130, 595-808) Variety Name Daqiu (Chinese) or te:B2 kʰjouB1-A (local Zhuang) (village name) of Nung An, which dispersed distributes at the rural areas of Jingxi County, Guangxi Zhuang Autonomous Region, China. 28.2 Information of data source Serial number of LRP Place of data elicitation LRP1 Date May 16, 2013 Qiulu Village (球路村), Xinjing Town, Jingxi County, Guangxi Zhuang Autonomous Region, China. Given Name Q̀ngyù (慶育) Family Name Lù (陸) Gender female Age 62 Birth Place Daqiu (te:C1 kʰjouB1-A in local Zhuang) Village (大秋屯), Longman Community (龙满 村), Longlin Township (龙临镇), Jingxi County, Guangxi Zhuang Autonomous Region, China. Places lived Daqiu Village of Longlin Township, Jingxi County (22 years). Da’en (大恩, or teC1 ŋjanC1 in local Zhuang) of the same township (6 years). Xinjing Town, the main town of Jingxi County (33 years). Time period of living at the present residence Occupation housewife 33 years Education junior high school Languages spoken Level Language Remarks Nung An mother tongue Daqiu Variety Jingxi Yang fluent This language is the wider communication language of Jingxi County. Standard Mandarin intermediate This language is the national language of the LRP’s country. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Nung An as native, he can communicate in Jingxi Yang, which is the WCL of Jingxi County. LRP’s mother was born in Daman (大满 or te:C1 mo:nC2 in local Yang Zhuang) village, Longlin Township, Jingx County. She could only speak her mother tongue Jingxi Shangjia Yang. Remarks According to Edmondson (2002), Nung An may be immigrant language from Long’an County of Guangxi. 364 28.3 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there is no tonal flip-flops in this dialect. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 33 B1-A: 11̰2 C1-A: 24ʔ DL1-A: 11 DS1-A: 33 Proto- Continuants A1-C: 554 B1-C: 11̰2 C1-C: 24ʔ DL1-C: 11 DS1-C: 55 Voiceless Unaspirated Sounds stops A1-U: 554 B1-U: 11̰2 C1-U: 24ʔ DL1-U: 11 DS1-U: 55 A1-UC:33 B1-UC: 11̰2 C1-UC: 24ʔ DL1-UC: 11 DS1-UC: 33 A1-G: 33 B1-G: 11̰2 C1-G: 24ʔ DL1-G: 11 DS1-G: 33 A2: 33 B2: 31 C2: 42ʔ DL2: 31 DS2: 33 A1/2: 33 B1/2: 31 C1/2: 42ʔ DL1/2: 31 DS1/2: 33 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-C, 1-U] [1-A, 1-UC, 1-G, 2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-C, 1-U] [1-A, 1-UC, 1-G, 2, 1/2] In conclusion, tones in this dialect are of non-straightforward tonal development. Secondary tonal split patterns are found in Columns A and DS. Horizontally: one individuals and four mergers Individual C[1-A, 1-C, 1-U, 1-UC, 1-G] C[2, 1/2] A[1-C, 1-U] = DS[1-C, 1-U] Merger A[1-A, 1-UC, 1-G, 2, 1/2] = DS[1-A, 1-UC, 1-G, 2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] B[2, 1/2] = DL[2, 1/2] 365 28.5 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High level-falling 554 occurs in A[1-C, 1-U] and DS [1-C, 1-U] to form a complementary distribution in smooth and checked syllables respectively. Tone 2: Mid level 33 occurs in A[1-A, 1-UC, 1-G, 2, 1/2] and DS[1-A, 1-UC, 1-G, 2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Low level-rising 112 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G] and low level 11 occurs in DL[1-A, 1-C, 1-U, 1-UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Mid-falling 31 occurs in B[2, 1/2] and DL[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low-rising 24 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 6: High-falling 42 occurs in C[2, 1/2]. 28.5 Phonation types or voice qualities of tones in this dialect Tones B1 have creakiness triggered by its extreme low pitch, like ʰɹaj11̰2 ‘egg’ and ˀba:11̰2 ‘shoulder’. Tones C1 and C2 have glottal constrictions, like ha:24ˀ ‘five’, pʰja:j24ˀ ‘cotton’, ʔɔ:j24ˀ ‘sugarcane’, maj42ˀ ‘tree’ and nam42ˀ ‘water’. All the glottal constrictions are not phonemic. 28.6 Sound changes involving tones reflected in the wordlist of this dialect 1) Initial consonants in the supposed 1-UC row which is to follow Li’s unaspirated stop + *r- clusters or ASC (Aspiration induced by Sesquisyllabic Clusters) row which is to be suggested as a series of sesquisyllable cluster following the hypothesis provided by Pittayawat Pittayaporn (2009) normally regularly present as aspirated sounds or a sound with spread feature (h-), like in tʰa:33 (A1-UC) ‘eye’, tʰa: j33 (A1-UC) ‘die’, and tʰe: ŋ33 (A1-UC) ‘cucumber’ like other CT and Yongnan Zhuang varieties. However, the item taɰ554 (A1-U<A1-UC) and tɕa:ŋ554 (A1U<A1-UC) ‘to put up’ which are expected to be in this row due to the cognate in other CT varieties (like tʰɔjA1-UC and kʰa:ŋA1-UC in Yang Zhuang varieties as well as kʰa:ŋA1-UC in Tiandeng) has unexpected unaspirated initial stops t- and tɕ- as well as a tone corresponding to be in the A1-U row. The term ‘gizzard’ with an unaspirated initial stop t- is also found in other Yongnan Zhuang Tai varieties and Daxin Baoxu (CT, Language 23) which have an additional tonal split due to the 366 proposed ASC row. The irregularity involved in this term needs further investigation. 2) The tone of lɯŋ112 (B1<B2) ‘grand floor’ is expected to be B2 due to the reflection in the vast majority of Tai varieties, but it is unexpected to be B1 in this variety. 3) Unlike other CT/SWT varieties, the terms nuk55 (DS1-C) ‘deaf’, tʰʊk33 (DS1-UC) ‘thin bamboo stripes’ and tik33 (DS1/2) ‘to hit the mark’ which are expected to have a long vowel in DL1 designated have a short vowel and are merged into their counterpart DS1 rows like NT varieties. 4) Just like other CT varieties, the term mit33 (DS2) ‘knife’ which is designated into DL2 due to its vocalic length in SWT has a short vowel. Its tone is also changed to DS2 (mid level). 28.6 Other linguistic changes reflected in the wordlist of this dialect. 1) The vowel of the term mowA2 ‘hand’ is unexpected. 2) The final consonant of kʰɔ:ŋ24 (C1-UC) ‘CLF of a stone’ is expected to be -n, but it is actually -ŋ in this dialect. 3) The final stops on checked syllables in Tai languages normally are not released, but the LRP often tends to release some of the final stops, like in the terms ma:k11 ‘fruits’, pa:k11 ‘mouth’, and kɔt11 ‘to hug’. However, the released stops are not phonemic since she also repeated some of those items with non-released final stops. 4) As those in other NT varieties, the following terms put in the wordlist for the revised tone box (Liao & Shen 2012) has different etymological sources, like ɹ̥umA1-A ‘mortar’, tawC1 ‘to come’, la:C1 ‘below’, fɯ:tDL2 ‘wing’, θe:pDL1 ‘fishhook’, and hatDS1-A ‘morning’. 367 Language 29 Xialeng, Yongnan Zhuang, CT/NT? 29.1 Language information Language Yongnan Zhuang ISO 639-3 Code Autonym tʰo: Exonym Data Source Zhang et al. 1999: 42-44, 121-127, 595-808 Data Info. Xialeng (township name) of YN. This variety is spoken at Xialeng Township, C1-A zyn Yongning District, Nanning City, Guangxi Zhuang Autonomous Region, China. 29.2 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there is no tonal flip-flops in this dialect. Smooth Syllable Phonation types Aspirations Proto- Continuants Voiceless Unaspirated Sounds stops Unaspirated stop + *r clusters Glottal Sounds Proto-Voiced Sounds Proto-voiced aspirations & breathy sounds Checked Syllable A B C A1-A: 33 B1-A: 53 C1-A: 13 A1-C: 55 B1-C: 13 C1-C: 35 A1-U: 55 B1-U: 13 C1-U: 35 A1-UC:33 B1-UC: 53 C1-UC: 13 A1-G: 33 B1-G: 53 C1-G: 13 A2: 31 B2: 53 C2: 11ʔ A1/2: 31 B1/2: 53 C1/2: 11ʔ DL DL1-A: 53 DL1’-A: 35 DL1-C: 13 DL1-U: 13 DL1’-U: 35 DL1-UC: 53 DL1-G: 53 DL1’-G: 35 DL2: 53 DL2’: 11 DL1/2: 53 DS DS1-A: 55 DS1-C: 55 DS1-U: 55 DS1-UC: 55 DS1-G: 55 DS2: 33 DS1/2: 33 2) Patterns of tonal split and merger are as in the following two charts. Vertically: three-way splits and two way splits Tonal category Register 1 Register 1' Register 2 A [1-C, 1-U] [1-A, 1-UC, 1-G] [2, 1/2] B [1-C, 1-U] C [1-C, 1-U] DL DS [1-A, 1-UC, 1-G, 2, 1/2] [1-A, 1-UC, 1-G] [2, 1/2] [1-C, 1-U] [1-A, 1-UC, 1-G, 2, 1/2] [1'-A, 1'-U, 1'-G] [2'] [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of non-straightforward tonal development. Secondary tonal split patterns are found in all columns except Column DS. 368 Horizontally: three individuals and four mergers A[2, 1/2] Individual C[1-C, 1-U] C[2, 1/2] A[1-C, 1-U] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] Merger A[1-A, 1-UC, 1-G] = DS[2, 1/2] B[1-C, 1-U] = C[1-A, 1-UC, 1-G] =DL[1-C, 1-U] B[1-A, 1-UC, 1-G, 2, 1/2] = DL[1-A, 1-UC, 1-G, 2, 1/2] 29.3 Tonal characteristics & tone occurrences: seven distinct tones. Just like in L25, L26, and L27, The complexity of tonal splits on dead Column DL are of secondary aberrant development due to MC loans and are out of the evaluation criteria of the conditioning of tonal splits. They are omitted in this study because they are beyond the tonal conditioning factors like initials or vowel-length. Tone 1: High level 55 occurs in A[1-C, 1-U] and DS[1-A, 1-C, 1-U, 1-UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 2: Mid level 33 occurs in A[1-A, 1-UC, 1-G] and DS[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 3: Mid-falling 31 occurs in A[2, 1/2]. Tone 4: Low-rising 13 occurs in B[1-C, 1-U] and C[1-A, 1-UC, 1-G], as well as in DL [1-C, 1-U] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-falling 53 occurs in B[1-A, 1-UC, 1-G, 2, 1/2] and in DL[1-A, 1-UC, 1-G, 2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Mid-rising 35 occurs in C[1-C, 1-U] and in DL[1'-A, 1'-U, 1'-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 7: Low-level 11 occurs in C[2, 1/2] and in DL[2'] to form a complementary distribution in smooth and checked syllables respectively.. 29.4 Phonation types or voice qualities of tones in this dialect According to the original materials, Tones C2 (including C1/2 which is merged into C2) have clear glottal constrictions, like maj11ʔ ‘tree’, nam11? ‘water’, and na:m11ʔ ‘abdomen’ (Zhang et al. 1999: 122). This glottal constriction is not phonemic. 369 29.5 Sound changes involving tones reflected in the wordlist of this dialect. 1) Like in NT varieties, the terms nuk55 (DS1) ‘deaf’ which is expected to have a long vowel in DL1 designated have short vowels. 2) There are many “derived tones” in this language. In the Sinitic circles, linguists use the specific punctuation ʼ to mark these “derived tones”, like na1 (A1-C) ‘thick’ vs. na1’ (A1-G) ‘swaddling clothes’ in Xialeng. In this thesis, we also adapt this convention to mark “derived tones”, like naA1 (A1-C) ‘thick’ vs. naA1’ (A1-G) ‘swaddling clothes’. 29.6 Notes on dead syllables According to the original material (Zhang et al. 1999: 122), tones on the DS column are regularly splits as “Tone 7” (DS1, high-level 55) and “Tone 8” (DS2 and DS1/2, mid-level 33) on earlier voiceless-voiced initials sounds as shown in the following Series 1. However, tonal splits in DL column are especially complex and present as follows: The tonal values of “Tone 9” (DL1) and “Tone 10” (DL2 and DL1/2) are low-rising 13 and high-falling 53 respectively, as shown in the following Series 2; however, aspirated initial stop (including aspirated initial clusters) and those initial consonants developing from proto-glottal sounds like ʔ-, m- (<*ʔb-), n- (<*ʔd-), j- (<*ʔj-), and w- (<*ʔw-) condition the Tone 9 to change to Tone 10, and it means that these two groups of initials condition tones DL1-A (as well as DL1-UC which also has aspirated initials) and DL1-G to merge into DL2 in our words in this thesis. The remained tones of DL1-C and DL1-U preserve the original Tone 9 (that is, original DL1 tone), as shown in the following Series 3. Nevertheless, this situation only indicates the regular rules of tonal splits and merger on the DL column. Some words in DL1-A, DL1-UC and DL2 unexpectedly have “Tone 9” (low-rising 13), which is the same tone as in DLC and DLU, as shown in the following Series 4. In addition, there are two other derived tones on the DL column, and they are Tone 9’ (DL1’, a mid-rising tone 35) and 10’ (DL2’, a low-level tone 11) respectively. These two tones normally occur on terms with Sinitic etymologic source, and the derived tone DL2’ is normally used for the cìzhuó (次浊) or ‘secondarily muddy’ initials (which refers to voiced sonorants in the terminology of the traditional Chinese phonology) of Middle-Chinese loans, as shown in the following Series 5. 370 Series 1: Original tones on the DS column (in the order of DL1-A, DL1-C, DL1-U, DL1-UC, DL1-G, DL2, and DL1/2) hat55 (DS1-A) ‘morning’ tʰat55 (DS1-A) ‘to run’ tʰot55 (DS1-A??) ‘fart’ hlət55 (DS1-A) ‘scabies’ me:t55 (DS1-C) ‘end’ nuk55 (DS1-C) ‘deaf’ (vocalic length agreeing with NT) lap55 (DS1-C) ‘dark’ lap55 (DS1-C) ‘tired’ tap55 (DS1-U) ‘liver’ tok55 (DS1-U) ‘to fall’ kok55 (DS1-U) ‘root’ tak55 (DS1-U) ‘to scoop’ tət55 (DS1-U) ‘to fillip (by fingers)’ kut55 (DS1-U) ‘fern’ klik55 (DS1-U) ‘lazy’ kuk55 (DS1-U) ‘tiger’ tip55 (DS1-U) ‘petal’ tsəp55 (DS1-U) ‘to kiss’ tʰuk55 (DS1-UC) ‘bamboo skin’ (vocalic length agreeing with NT but disagreeing with CT/SWT) nip55 (DS1-G) ‘raw’ ʔak55 (DS1-G) ‘chest’ ʔəp55 (DS1-G) ‘to close (with the door left unlocked or unlatched)’ ʔut55 (DS1-G) ‘to bend (a bamboo)’ ʔot55 (DS1-G) ‘to plug’ (ɬək55)ʔek55 (DS1-G) ‘to hiccup’ ʔuk55 (DS1-G) ‘brains’ nət55 (DS1-G) ‘sunshine’ (vocalic length agreeing with NT but disagreeing with CT/SWT) mok55 (DS1-G) ‘dry up’ (vs. Debao ˀbɔ:k55 (DS1-G) ‘dry up’ and Thai bok21 (DS1-G) ‘land’) ʔik55 (DS1-G) ‘benefit’ (<益 vs. i53 Mandarin, jɪk55 Cantonese) 371 lik33 (DS2) ‘child’ (vocalic length agreeing with NT but disagreeing with CT/SWT) jip33 (DS2) ‘to sew’ kap33 (DS2) ‘to catch’ kət33 (DS2) ‘indignant’ mat33 (DS2) ‘CLF (of a grain)’ mot33 (DS2) ‘ant’ nok33 (DS2) ‘bird’ ləp33 (DS2) ‘to stroke’ hot33 (DS2) ‘to stir’ kut33 (DS2) ‘thick (liquid)’ təp33 (DS2) ‘to smash’ hap33 (DS1/2) ‘to bite’ nak33 (DS1/2) ‘to press on’ tsʰip33 (DS1/2) ‘ten’ tik33 (DS1/2) ‘to be’ (vocalic length agreeing with NT but disagreeing with CT/SWT) Series 2: Original tones on the DL column (in the order of DL1-C, DL1-U, DL2, and DL1/2) ɬe:t13 (DL1-C) ‘to finish’ mo:k13 (DL1-C) ‘fog’ ɬe:p13 (DL1-C) ‘fishhook’ pe:t13 (DL1-U) ‘eight’ pa:t13 (DL1-U) ‘basin’ kwe:t13 (DL1-U) ‘to scrape’ ko:p13 (DL1-U) ‘hold in both hands’ kle:p13 (DL1-U) ‘husk (of rice)’ ko:t13 (DL1-U) ‘to hug’ ko:k13 (DL1-U) ‘corner’ te:p53 (DL2) ‘husk (of rice)’ ta:k53 (DL2) ‘to measure’ to:k53 (DL2) ‘to read’ (<讀 vs. tu35 Mandarin, tʊk35 Cantonese) ma:k53 (DL2) ‘CLF (of a knife)’ mo:t53 (DL2) ‘rice weevil’ 372 hlo:t53 (DL2) ‘outside’ na:k53 (DL2) ‘otter’ to:t53 (DL2) ‘to protrude’ ku:k53 (DL2) ‘to do’ pu:t53 (DL2) ‘to dial’ (<撥 vs. po55 Mandarin, put22 Cantonese) ho:p53 (DL2) ‘wound healing’ mo:p53 (DL2) ‘mouth-muffle (of an ox)’ nu:k53 (DL2) ‘infect’ (ɬow11) ɬə:k53 (DL2) ‘whacked’ te:p53 (DL2) ‘dish’ (<碟 vs. tiɛ35 Mandarin, tip22 Cantonese) pu:k53 (DL1/2) ‘white’ (tonal behavior agreeing with NT but disagreeing with CT/SWT) Series 3: expected tone changes (DL1 > DL2 conditioned by aspiration and glottal sounds) tsʰa:t53 (DL2<DL1-A) ‘brush’ (<刷 vs. ʂua55 Mandarin, tsʰa:t35 Cantonese) pʰla:k53 (DL2<DL1-A) ‘forehead’ pʰlu:k53 (DL2<DL1-A) ‘taro’ tʰu:k53 (DL2<DL1-A) ‘oil dipper’ hle:k53 (DL2<DL1-A) ‘wok’ ha:p53 (DL2<DL1-A) ‘box’ (<盒 vs. xɤ35 Mandarin, hap11 Cantonese) tʰe:k53 (DL2<DL1-UC) ‘cracking’ ʔi:p53 (DL2<DL1-G) ‘to salt’ ʔi:t53 (DL2<DL1-G) ‘to stretch’ ʔe:k53 (DL2<DL1-G) ‘yoke (of an ox)’ na:t53 (DL2<DL1-G) ‘hot’ ma:t53 (DL2<DL1-G) ‘times’ Series 4: unexpected tonal preservation on DL1-A and DL1-G he:k13 (DL1-C/U<DL1-A) ‘guest’ (<客 vs. kʰɤ51 Mandarin, ha:k33 Cantonese) tʰa:p13 (=DL1-C/U<DL1-A) ‘pagoda’ (<塔 vs. tʰa214 Mandarin, tʰa:p33 Cantonese) kʰwi:t13 (=DL1-C/U<DL1-A) ‘lack’ (<缺 vs. tɕʰuɛ55 Mandarin, kʰwit33 Cantonese) tʰa:t13 (=DL1-C/U<DL1-A) ‘to excoriate (skin)’ ʔe:p13 (=DL1-C/U<DL1-G) ‘to force’ 373 ʔa:k13 (=DL1-C/U<DL1-G) ‘fierce’ (<惡 vs. ɤ51 Mandarin, ʔɔk33 Cantonese) Series 5: unexpected derived tones on the DL column hlo:p35 (DL1’< DL1-A) ‘quaff (drinks)’ tsʰo:t35 (DL1’< DL1-A) ‘end’ ki:p35 (DL1’< DL1-U) ‘tough’ ku:t35 (DL1’< DL1-U) ‘uncover’ (<揭 vs. tɕiɛ55 Mandarin, kʰit33 Cantonese) tsi:t35 (DL1’< DL1-U) ‘festival’ (<節 vs. tɕiɛ35 Mandarin, tsi:t33 Cantonese) ka:p35 (DL1’< DL1-U) ‘to unite’ (<合 vs. hɤ35 Mandarin, hap33 Cantonese) ʔu:t35 (DL1’< DL1-G) ‘fool’ li:p11 (DL2’< DL2) ‘to hunt’ (<獵 vs. liɛ51 Mandarin, li:p11 Cantonese) la:p11 (DL2’< DL2) ‘dried meat’ (<腊 vs. la51 Mandarin, la:p11 Cantonese) 374 Language 30 Shuangqiao, Wuming, Yongbei Zhuang, NT 30.1 Language information Language Yongbei Zhuang ISO 639-3 Code zyb Autonym to:C1-U Exonym ˀba:nC1-G Data Source Zhang et al. 1999: 35-36, 50-55, 595-808. Data Info. According to the original material, This language is commonly called to:C1-U ‘local, native’, and is spoken in Shuangqiao Township, Wuming County, Nanning City, Guangxi, China. 30.2 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flops except tones C in this dialect. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 24 B1-A: 35 C1-A: 55 DL1-A: 35 DS1-A: 55 Continuants A1-C: 24 B1-C: 35 C1-C: 55 DL1-C: 35 DS1-C: 55 Unaspirated stops A1-U: 24 B1-U: 35 C1-U: 55 DL1-U: 35 DS1-U: 55 A1-UC: 24 B1-UC: 35 C1-UC: 55 DL1-UC: 35 DS1-UC: 55 A1-G: 24 B1-G: 35 C1-G: 55 DL1-G: 35 DS1-G: 55 A2: 31 B2: 33 C2: 42 DL2: 42 DS2: 33 A1/2: 31 B1/2: 33 C1/2: 42 DL1/2: 42 DS1/2: 33 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development. 375 Horizontally: two individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G] A[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] Merger B[2, 1/2] = DS[2, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] C[2, 1/2] = DL[2, 1/2] 30.3 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Low-rising 24 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: Mid-falling 31 occurs in A[2, 1/2]. Tone 3: Mid-rising 35 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G] and DL[1-A, 1-C, 1-U, 1-UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Mid level 33 occurs in B[2, 1/2] and in DS[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High level 55 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G] and in DS[1-A, 1-C, 1-U, 1UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively. Tone 6: High-falling 42 occurs in C[2, 1/2] and in DL[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. 30.4 Phonation types or voice qualities of tones in this dialect We do not know the phonation types or voice qualities of tones in this dialect since they are not mentioned in the original text. 376 Language 31 Longsang, NT 31.1 Language information Language Longsang Zhuang ISO 639-3 Code undescribed Autonym ja:ŋ Exonym ˀju:jC1-G Data Source 3 LRP Variety Name Debao Longsang. This is a undescribed NT language spoken at the urban and rural A1-G C1 ; to: areas of Longsang Township, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. This language’s autonym and exonym are both ja:ŋA1-G, which is contradictory to the definition of Yang Zhuang (L8-13) (cf. Language Information of L11, L14, and L 26). 31.2 Information of data source Serial number of LRP Place of data elicitation LRP1 Date May 6, 2013 Xinlong Road 96, Chengguan Town, Debao County, Guangxi Zhuang Autonomous Region, China. Given Name Xìcǎo (秀草) Gender female 33 Family Name Lù (陸) Age 72 31 Birth Sanhe Community ( 合屯, fa:k kaw in local Zhuang), Qiaotou Village (橋頭村), Place Longsang Township (隆桑鎮), Debao County, Guangxi Zhuang Autonomous Region, China. Places She lived in her own village until she was 37. After that she came to follow her husband to lived live in the main town of Debao County until now. Time period of living at the present residence Occupation retired worker Languages spoken Level Debao Longsang Zhuang mother tongue Debao Urban Yang fluent Zhuang 34 years Education primary school Language Remarks This language is the lingua franca of Debao County. However, as a native Yang Zhuang speaker, I could cognize that her Yang Zhuang had clear NT accent. For example, aspirated sounds in Yang Zhuang were always pronounced as unaspirated counterpart sounds when she spoke Yang Zhuang to me. Standard Mandarin intermediate This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Longsang Zhuang as native, he can communicate in Debao County Yang Zhuang. LRP’s mother was born in Qiaonan Community (橋南屯 tɕawC1 kɔ:ŋC1 in 377 local Zhuang) village, Qiaotou Village (橋頭村), Longsang Township (隆桑鎮), Debao County. She was monolingual in Longsang Zhuang. Remarks The LRP1 is the LRP2’s mother. Their pronunciations are obviously different in some tones (like tones A1 are 24 by LRP1 but 242 by LRP2), probably because the son has grown up in the main town of Debao County, where is far away from his mother’s village. The basic tonal descriptions are based on LRP1’s pronunciation because it is treated as the original pattern due to the fact that LRP1 is LRP2’s mother, and she grew up in the central area of this language. Data from the LRP3 also support this speculation. Serial number of LRP LRP2 Date May 15, 2013 Xinlong Road 6, Chengguan Town, Debao County, Guangxi Zhuang Place of data elicitation Autonomous Region, China. Given Name Wèiquán (衛權) Family Name Lù (陸) Gender male Age 40 Birth Place Sanhe Community ( 合屯, fa:k33 kaw31 in local Zhuang), Qiaotou Village (橋頭村), Longsang Township (隆桑鎮), Debao County, Guangxi Zhuang Autonomous Region, China. Places lived He lived in her own village until he was 6. After that he came to follow his parents to live in the main town of Debao County until now. Time period of living at the present residence Occupation deliveryman 34 years Education junior high school Languages spoken Level Language Remarks Debao Longsang Zhuang mother tongue Although he left his village to live in the main town of Debao from 6 years old, he always speaks his mother tongue with his family until now. Debao Urban Yang Zhuang native This language is the lingua franca of Debao County. He speaks Yang Zhuang almost as fluently as native speakers because he started to speak this language from his childhood. Standard Mandarin intermediate This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in Daji Village (大吉村), Longsang Township (隆桑鎮), Debao County. Besides his mother tongues Longsang Zhuang, he also speaks Debao Urban Yang Zhuang, Standard Mandarin, and Guiliu Southwestern Mandarin. LRP’s mother was born in the same village as the LRP’s. She also speaks Debao Urban Yang Zhuang, Standard Mandarin, and Guiliu Southwestern Mandarin. She is now living with the LRP. 378 Serial number of LRP LRP3 Date May 15, 2013 Xinlong Road 96, Chengguan Town, Debao County, Guangxi Place of data elicitation Zhuang Autonomous Region, China. Given Name Jí (傑) Family Name Lù (陸) Gender male Age 60 Birth Place Longtan Community (龍壇屯, luŋ31 ta:n55 in local Zhuang), Longtan Village (龍壇 村), Longsang Township (隆桑鎮), Debao County, Guangxi Zhuang Autonomous Region, China. Places lived Longtan Village (16 years). Tianlin County, Baise Prefecture (6 years). Baise City (2 years). He is now living in the Debao main town. Time period of living at the present residence Occupation retired cadre 35 years Education junior college Languages spoken Level Language Remarks Debao Longsang Zhuang mother tongue Although he left his village to live in the main town of Debao from 6 years old, he always speaks his mother tongue with his family until now. Debao Urban Yang Zhuang fluent This language is the lingua franca of Debao County. He speaks Yang Zhuang with NT accent. Standard Mandarin fluent This language is the national language of the LRP’s country. Guiliu Dialect fluent A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. Baihua intermediate This Cantonese variety is the lingua franca of the urban area of Baise City. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides his mother tongues Longsang Zhuang, he also speaks Debao Urban Yang Zhuang, Standard Mandarin, Guiliu Southwestern Mandarin, and Baihua. LRP’s mother was born in Longyuan Community (龍苑屯), Longtan Village (龍壇村), Longsang Township (隆桑鎮), Debao County, Guangxi Zhuang Autonomous Region, China. Besides her mother tongues Longsang Zhuang, she also speaks Debao Urban Yang Zhuang. The LRP’s parents are now living with the LRP. 379 31.3 Basic patterns of tonal development of this dialect 1) As shown in the chart above, tonal flip-flop runs through all tones except the column of DS, which preserves a higher pitch on tones DS1 and a lower pitch on tones DS2 and DS12. Smooth Syllable Phonation types Checked Syllable A B C DL DS Aspirations A1-A: 35 B1-A: 31 C1-A: 213ʔ DL1-A: 21 DS1-A: 55 Continuants A1-C: 35 B1-C: 31 C1-C: 213ʔ DL1-C: 21 DS1-C: 55 Unaspirated stops A1-U: 35 B1-U: 31 C1-U: 213ʔ DL1-U: 21 DS1-U: 55 A1-UC: 35 B1-UC: 31 C1-UC: 213ʔ DL1-UC: DS1-UC: 21 55 A1-G: 35 B1-G: 31 C1-G: 213ʔ DL1-G: 21 DS1-G: 55 Proto-voiced sounds A2: 55 B2: 53 C2: 33ʔ DL2: 33 DS2: 33 Proto-voiced aspirations & A1/2: 55 B1/2: 53 C1/2: 33ʔ DL1/2: 33 DS1/2: 33 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development with voiceless-voiced register splits. Horizontally: three individuals and three mergers A[1-A, 1-C, 1-U, 1-UC, 1-G] Individual B[1-A, 1-C, 1-U, 1-UC, 1-G] B[2, 1/2] A[2, 1/2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] Merger C[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] C[2, 1/2] = DL[ 2, 1/2] = DS[ 2, 1/2] 380 31.4 Tonal characteristics & tone occurrences: six distinct tones Tone A1 A2 B1 B2 C1 C2 DL1 DL2 DS1 DS2 LRP1, 3 24 55 31 53 213ʔ 33ʔ 21 33 55 33 LRP2 242/24 45/55 31 53 213ʔ 33ʔ 22 33 55 33 Tone 1: Low-rising 24 (LRP1, LRP3) or Low-rising-falling 242 (LRP2) occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]; some items with this tone from LRP2’s recording also present as low-rising [24], like ta:35 ‘eye’, ta:j35 ‘to die’, and taɰ35 ‘gizzard’ but they are treated as the free variation of [242], due to the fact that the LRP2 insists that ‘eye’ (recording ta:24) and ‘maternal grandmother’ (recording ta:242) are homophonous. Tone 2: High-level 55 (LRP1, LRP3) or high rising 45 (LRP2) occurs in A[2, 1/2], and high level 55 occurs in DS[1-A, 1-C, 1-U, 1-UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively; some items with this tone on A[2, 1/2] from LRP2’s recording also present as high level 55 which is the free variation of 45, such as na:55 ‘rice field’, ham55 ‘bitter’, and kwa:55 ‘right side’. Tone 3: Mid-falling 31 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 4: High-falling 53 occurs in B[2, 1/2] and DL[2, 1/2]. Tone 5: Low-falling-rising 213 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G], and low-falling 21 (LRP1) or low level 22 (LRP2) in DL[1-A, 1-C, 1-U, 1-UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Mid level 33 occurs in C[2, 1/2], DL[2, 1/2], and DS[2, 1/2]. 31.5 Phonation types or voice qualities of tones in this dialect Tones C1 and C2 have clear glottal constrictions, like fɯ:213ˀ ‘cloud’, tɕa:j213ˀ ‘to walk’, ʔa: 213ˀ ‘to open’, ma:33ˀ ‘horse’, nam33ˀ ‘water’, and maj33ˀ ‘tree’. All these voice qualities are not phonemic. 31.6 Sound changes involving tones reflected in the wordlist of this dialect 1) The initial of the term ʔa:24 ‘crow’ is of the initial group of glottal sounds, as that in other NT varieties. 2) Different from the merger into DS2 in other NT varieties, ɬap55 ‘to mince’ merges its tone with the rows of DS1. This may be of an areal trait shared with the neighboring the Debao Yang Zhuang varieties, which are of CT. 381 3) As those in other NT varieties, the following items designated into DL columns show a regular tone change to be coordinate with the counterpart DS columns: nuk55 (DS1-C) ‘deaf’, tuk55 (DS1-UC) ‘thin bamboo stripes’, ˀdit55 (DS1-G) ‘sunshine’, mit33 (DS2) ‘knife’, and tək33 (DS12-A) ‘to hit the mark’. However, the term ŋɯ:k21 (DL1-C) ‘gums’ agreeing with that reflecting a long vowel in the DL1-C column in all CT/SWT varieties. 4) Not only the tones DL2 and DS2 are totally merged, but all the long vowels in the DL2 column are also shortened to merge into short vowels, like that nɔk33 (DL2) ‘outside’ and nɔk33 (DS2) ‘bird’ are homophonous, and the vowels of the following terms ɕakDL2 ‘rope’, pakDL2 ‘insane’, lakDL2 ‘root’ and makDL2 ‘plenty’ are all shortened. 31.7 Other linguistic changes reflected in the wordlist of this dialect 1) As those in other NT varieties, the following terms put in the wordlist for the revised tone box (Liao & Shen 2012) has different etymological sources, like tawC1 ‘to come’, te:B1 ‘to wait’, la:C1 ‘below’, fɯ:tDL2 ‘wing’, θe:pDL1 ‘fishhook’, ɹʊmA1 ‘mortar’, and to:nC1 ‘to receive’. 2)The term te:m24 ‘in addition’ has a compound form ʔej213 te:m24, the optional form ˀjem24 for this item should be a liaison from this compound form. 3) The term ta:55 only refers to ‘to pump’ in this dialect, while in CT/SWT varieties, the cognate developing from *da:A refers to both ‘to spread on’ and ‘to pump’. 4) The vowel of faɰ55 (A2) is different from that reflecting *-ɯŋ in other NT varieties, but is the same to that reflecting *-ɯ: in the vast majorities of CT/SWT varieties. 5) The initial of nam33ˀ ‘water’, nɔk33 ‘bird’, and nɔk33 ‘outside’ is n- which does not agree with that reflecting *r- in the vast majorities of NT varieties, but agrees with that reflecting *n- in all CT/SWT varieties. In a similar way, the initial of maj33ˀ ‘tree’ is m- which doesn’t agree with that reflecting *v- in the vast majorities of NT varieties, but agrees with that reflecting *m- in all CT/SWT varieties. 6) The item tɯkDL1 only refers to ‘touch’ or ‘contact’, while in other Tai varieties, its cognate has several other meanings besides ‘touch’, like ‘to be’, ‘right’ or ‘to hit the mark’. 7) As those in other NT varieties, the initial of θipDS1-A ‘centipede’ which is designated into DS1-A reflects *s- or *ɬ- which should be considered as DS1-C. 382 Language 32 Dongling, Youjiang Zhuang, NT 32.1 Language information Language Youjiang Zhuang ISO 639-3 Code zyj Autonym to:C1-U; ja:ŋA1-G Exonym ˀju:jC1-G Data Source 2 LRPs Variety Name Dongling (township name) of Youjiang Zhuang. This variety is spoken at the plain paddy areas of Dongling Township, Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Dongling is an intermountain basin township. Dongling Youjiang Zhuang is spoken in the central plain paddy part of this basin, while Yang Zhuang varieties are spoken in the surrounding mountainous areas. The Yang Zhuang population is much more than Dongling Youjiang Zhuang. Besides, Yang Zhuang is the lingua franca of the whole Debao County. Therefore, many Dongling Youjiang Zhuang speakers are bilingual in their mother tongue and Debao County Yang Zhuang. The autonym ja:ŋA1-G provided by LRP2 again indicates the problem of the definition of Yang Zhuang (cf. Language Information in L11, L14, L26, and L31). 32.2 Information of data source Serial number of LRP LRP1 Place of data elicitation Date May 17, 2013 Xinlong Road 115, Chengguan Town, Debao County, Guangxi Zhuang Autonomous Region, China. Given Name Ľi (磊) Family Name Húng (黄) Gender male Age 31 Birth Place Xiadong Community ( 垌屯, tɔŋB2 la:C1 in local Zhuang), Dongling Village (東陵村), Dongling Township (東陵鎮), Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Xiadong Community (18 years). Wuzhishan City in Hainan Province (2 years). Shandong Province (0.5 years). He lives in the main town of Debao County now. Time period of living at the present residence Occupation worker 5 years Education Languages spoken Level Dongling Youjiang Zhuang mother tongue Debao County Yang Zhuang fluent senior high school Language Remarks This language is the lingua franca of Debao County. Standard Mandarin fluent This language is the national language of the LRP’s country. Baihua intermediate A Cantonese variety spoken in Guangxi, China. 383 The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Dongling Youjiang Zhuang as native, he can communicate in Debao County Yang Zhuang, Guiliu Southwestern Mandarin, Standard Mandarin (Putonghua hereafter), and Baihua. LRP’s mother was born in the same village. Besides her mother tongue Dongling Zhuang, she can speak Standard Mandarin. Serial number of LRP LRP2 Place of data elicitation May 18, 2013 Xinlong Road 57, Chengguan Town, Debao County, Guangxi Zhuang Autonomous Region, China. Given Name Chángcāo (常操) Gender male Birth Place Date Suoda Community (索達屯, θo:k DL1 Family Name Lú (羅) Age 57 B2 ta: in local Zhuang), Dongling Village (東陵村), Dongling Township(東陵鎮), Debao County, Baise Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Suoda Community (18 years). Baise City (3 years). Dongling Township (17 years). Now he lives in the main town of Debao County. Time period of living at the present residence Occupation administrative leader Languages spoken Level Dongling Youjiang Zhuang mother tongue Debao County Yang Zhuang fluent Standard Mandarin fluent 19 years Education junior college Language Remarks This language is the lingua franca of Debao County. This language is the national language of the LRP’s country. Guiliu fluent This language is a Southwestern Mandarin variety spoken in the western part of Guangxi. Baihua intermediate A Cantonese variety spoken in Guangxi, China. Yao intermediate The LRP often went to support the poor in the country side in some Yao communities in the mountain areas in Dongling Township, Debao County when he worked in Dongling Central Primary School. This language is called jiwA2 in Debao County Yang Zhuang. It is not clear that which Yao variety it is. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Dongling Youjiang Zhuang as native, he can communicate in Debao County Yang Zhuang, Guiliu Southwestern Mandarin, Standard Mandarin (Putonghua hereafter), Baihua, and Yao. LRP’s mother was born in the Qinke Community (琴克屯), Dongling Village (東陵村), Dongling Township(東陵鎮), Debao County. Besides her mother tongue Dongling Zhuang, she can speak Standard Mandarin, Debao County Yang Zhuang, and Nung An. 384 32.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, tonal flip-flop runs through all the tones in this dialect. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 11̰ B1-A: 242 C1-A: 224ʔ DL1-A: 24 DS1-A: 22 Continuants A1-C: 11̰ B1-C: 242 C1-C: 224ʔ DL1-C: 24 DS1-C: 22 Unaspirated stops A1-U: 11̰ B1-U: 242 C1-U: 224ʔ DL1-U: 24 DS1-U: 22 A1-UC: 11̰ B1-UC: 242 C1-UC: 224ʔ DL1-UC: 24 Glottal Sounds A1-G: 11̰ B1-G: 242 C1-G: 44ʔ DL1-G: 24 DS1-G: 44 Proto-voiced sounds A2: 553 B2: 53 C2: 44ʔ DL2: 53 DS2: 44 A1/2: 553 B1/2: 53 C1/2: 44ʔ DL1/2: 53 DS1/2: 44 ProtoVoiceless Sounds Unaspirated stop + *r clusters Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC] [1-G, 2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC] [1-G, 2, 1/2] In conclusion, tones in this dialect are of non-straightforward tonal development. The nonstraightforward columns are C and DS, with a spilt being voiceless register on one side and golttal/voiced register on the other side. Horizontally: two individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G] A[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] Merger B[2, 1/2] = DL[2, 1/2] C[1-A, 1-C, 1-U, 1-UC] = DS[1-A, 1-C, 1-U, 1-UC] C[1-G, 2, 1/2] = DS[1-G, 2, 1/2] 385 DS1-UC: 22 32.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Low-level 11 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: High level-falling 553 (LRP1) occurs in A[2, 1/2]. Tone 3: Low-rising-falling 242 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G], and low-rising 24 occurs in DL[1-A, 1-C, 1-U, 1-UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively. Tone 4: High-falling 53 occurs in B[2, 1/2] and DL[2, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Low level-rising 224 occurs in C[1-A, 1-C, 1-U, 1-UC], and mid level 22 in DS[1A, 1-C, 1-U, 1-UC], to form a complementary distribution in smooth and checked syllables respectively. Tone 6: High-level 44 occurs in C[1-G, 2, 1/2] and DS[1-G, 2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. 32.5 Phonation types or voice qualities of tones in this dialect A1 tone (11) has a creakiness triggered by its lowest pitch 1; Tones C1 and C2 have glottal constrictions, like fiə224ˀ ‘cloud’, pja:j224ˀ ‘to walk’, ʔa: 224ˀ ‘to open’, ma:44ˀ ‘horse’, lam44ˀ ‘water’, and maj44ˀ ‘tree’. The glottal constriction on Tone C1 is clearer than on Tone C2. All these voice qualities are not phonemic. 32.6 Sound changes involving tones reflected in the wordlist of this dialect 1) The initial of the term ʔa:11 ‘crow’ is of the initial group of glottal sounds, as that in other NT varieties. 2) As those in other NT varieties, the following items designated into DL columns show a regular tone change to be coordinate with the counterpart DS columns: nok22 (DS1-C) ‘deaf’, ŋək22 (DS1-C) ‘gums’, tok22 (DS1-UC) ‘thin bamboo stripes’, ˀdit22 (DS1-G) ‘sunshine’, mit44 (DS2) ‘knife’, and tək44 (DS12-A) ‘to hit the mark’. 3) LRP2 claimed that the term for ‘to pull down’ has two optional tones, tuəŋ53 (merging into B2) and tuəŋ242 (merging into B1). The former one agrees with NT varieties. The later one agrees with CT/SWT varieties, and should be an areal trait shared with its neighboring Debao Yang Zhuang varieties (of CT). 4) LRP2 provided ta:k24 ‘dry blood sucker’ which has DL1 tone agreeing with CT varieties but different from DL2 tone in other NT varieties. However, when I asked him to pronounce ‘to measure’ which should be the homophone of ‘dry blood sucker’, he provided two optional tones ta:k24 (DL1) and ta:k53 (DL2). The later one 386 should be the original form since it agrees with all other NT varieties, while the former one should be an areal trait shared with the neighboring Debao Yang Zhuang varieties (of CT). 32.7 Other linguistic changes reflected in the wordlist of this dialect 1) As those in other NT varieties, the following terms put in the wordlist for the revised tone box (Liao & Shen 2012) has different etymological sources, like tawC1 ‘to come’, te:B1 ‘to wait’, la:C1 ‘below’, fiətDL2 ‘wing’, and θe:tDL1 ‘fishhook’. However, the term lom55 ‘mortar’ has a C2 tone, which does not agree with its cognate reflecting *trumA or *krumA in other NT varieties. 2) The term ˀje:m53 ‘in addition’ provided by LRP1 has been changed not only the initial but also the tone (from A1 to C2), and this indicates a typical sound change triggered by grammaticalization. Its initial consonant change can be conferred to its cognate ˀjem24 in Longsang (Language 27), which has a compound form ʔej213 te:m24 being simplified to ˀjem24. 3) The term for ‘to put up’ is ka:ŋ11(A1) provided by LRP1, but is ha:ŋ11̰(A1) provided by LRP2. 4) As those in other NT varieties, the initial of θipDS1-A ‘centipede’ which is designated into DS1-A reflects *s- or *ɬ- which should be considered as DS1-C. 5) The vowel of kuəB2 ‘pair’ provided by LRP1 is unexpected (it is expected to be ku:B2 due to the sound correspondence, and LRP2 provides the expected form). This aberrance change might be influenced by the compound word kuəB2 tɯəB12-A ‘a pair of chopsticks’ provided by LPR1. However, the vowel of tɯ:B12-A ‘chopsticks’ is a long ɯ: when it is pronounced independently. 6) In the vast majority of Youjiang Zhuang varieties, the initial pj- has been merged into tɕ-. But in this variety, the process of this merger is not completed. For example, LRP1 only provides pja:jC1 for ‘to walk’, but LRP2 provides two variations pja:jC1 and tɕa:jC1. LRP2 prefers the later one, and he only read the form tɕa:jC1 for ‘to walk’ in the actual recording. In a similar way, LRP1 only provides pjakDS1for ‘vegetable’, and LRP2 provides pjakDS1 and tɕakDS1. For this term LRP2 read the both forms for the actual recording. 7) The term ɬe:t DL1 ‘fishhook’ has a different syllable final -t from -p shared by other NT varieties. 387 8) LRP2 unexpectedly provided lap55 for ‘to receive’, which is shared with CT/SWT varieties, because all other NT varieties do not share this etymon. 9) LRP2 provided ɕakDS2 (agreeing with NT) and lakDS2 (agreeing with CT/SWT) for ‘to steal’. But he claimed the meanings of these two terms are slightly different. 388 Language 33 Lizhou, Tianlin, Guibian Zhuang, NT 33.1 Language information Language Guibian Zhuang ISO 639-3 Code zgn Autonym ˀjajC1-G Exonym ˀjajC1-G; ˀjo:jC1-G Data Source Zhang et al. 1999: 37-38, 65-67, 595-808. Data Info. This language is spoken in Lizhou Township (利周鎮), Tianlin County (田林縣), Baise Prefecture, Guangxi Zhuang Autonomous Region, China. 33.2 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is tonal flip-flops in Columns A and C in this dialect. Smooth Syllable Checked Syllable Phonation types ProtoVoiceless Sounds A B C DL DS Aspirations A1-A: 13 B1-A: 24 C1-A: 45 DL1-A: 24 DS1-A: 55 Continuants A1-C: 13 B1-C: 24 C1-C: 45 DL1-C: 24 DS1-C: 55 Unaspirated stops A1-U: 13 B1-U: 24 C1-U: 45 DL1-U: 24 DS1-U: 55 A1-UC: 13 B1-UC: 24 C1-UC: 45 DL1-UC: 24 DS1-UC: 55 A1-G: 13 B1-G: 24 C1-G: 33 DL1-G: 24 DS1-G: 55 A2: 52 B2: 31 C2: 33 DL2: 22 DS2: 22 A1/2: 52 B1/2: 31 C1/2: 33 DL1/2: 22 DS1/2: 22 Unaspirated stop + *r clusters Glottal Sounds Proto-Voiced Sounds Proto-Voiced Aspirations & Breathy Sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC] [1-G, 2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of non-straightforward tonal development. Secondary tonal split conditioned by glottalized sounds is found in Column C. 389 Horizontally: three individuals and three mergers A[1-A, 1-C, 1-U, 1-UC, 1-G] Individual A[2, 1/2] B[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] Merger C[1-A, 1-C, 1-U, 1-UC] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] C[1-G, 2, 1/2] = DL[2, 1/2] = DS[2, 1/2] 33.3 Tonal characteristics & tone occurrences: six distinct tones. Tone 1: Low-rising 13 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: High-falling 52 occurs in A[2, 1/2]. Tone 3: Low-rising 24 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G] and in DL[1-A, 1-C, 1-U, 1UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Mid-falling 31 occurs in B[2, 1/2]. Tone 5: High-rising 45 occurs in C[1-A, 1-C, 1-U, 1-UC] and in DS[1-A, 1-C, 1-U, 1-UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Mid level 33 occurs in C[1-G, 2, 1/2], as well as in DL[2, 1/2] and DS[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. 33.4 Phonation types or voice qualities of tones in this dialect We do not know the phonation types or voice qualities of tones in this dialect since they are not mentioned in the original text. 390 Language 34 Shanglin Yunling, Central Hongshuihe Zhuang, NT 34.1 Language information Language Central Hongshuihe Zhuang ISO 639-3 Code zch Autonym ɕuəŋB2 Exonym Shanglin Zhuang Data Source 1 LRP Variety Name Yunling (village name) of Central Hongshuihe Zhuang. It is spoken at Xiyan Town, Shanglin County, Guangxi Zhuang Autonomous Region, China. 34.2 Information of data source Serial number of LRP LRP1 Date May 20, 2013 Chuangyi Art School, Nanning City, Guangxi Zhuang Autonomous Place of data elicitation Region, China. Given Name Ýng (瑩) Family Name Sū (蘇) Gender female Age 26 Birth Place Yunling Village (雲靈村), Xiyan Township (西燕鎮), Shanglin County ( 林縣), Nanning Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived When she was 8 years old, she moved to Nanning City with her family. She lives in Nanning City until now. Time period of living at the present residence Occupation dancing teacher 18 years Education bachelor Languages spoken Level Language Remarks Shanglin Yunling Central mother tongue Although she left her village to live in the Hongshuihe Zhuang Nanning from 8 years old, she always speaks his mother tongue with his family until now. Standard Mandarin intermediate This language is the national language of the LRP’s country. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Yunling Central Hongshuihe Zhuang as native, he can communicate in Standard Mandarin and Baihua. LRP’s mother was born in Hongtao Community village, Yunling Village (雲靈村), Xiyan Township (西燕 鎮), Shanglin County ( 林縣). Hongtao is a Yao community. Besides her mother tongue Yao, she can speak Yunling Central Hongshuihe Zhuang, Standard Mandarin, and Baihua. LRP’s parents live together with the LRP in Nanning City now. 391 34.3 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there is a tonal flip-flop on tones B, and no tonal flip-flop in other tones. Smooth Syllable Checked Syllable Phonation types ProtoVoiceless Sounds A B C DL DS Aspirations A1-A: 35 B1-A: 33 C1-A: 55 DL1-A: 33 DS1-A: 55 Continuants A1-C: 35 B1-C: 33 C1-C: 55 DL1-C: 33 DS1-C: 55 Unaspirated stops A1-U: 35 B1-U: 33 C1-U: 55 DL1-U: 33 DS1-U: 55 A1-UC: 35 B1-UC: 33 C1-UC: 55 DL1-UC: 33 DS1-UC: 55 A1-G: 35 B1-G: 33 C1-G: 55 DL1-G: 33 DS1-G: 55 A2: 221 B2: 53 C2: 31 DL2: 53 DS2: 32 A1/2: 221 B1/2: 53 C1/2: 31 DL1/2: 53 DS1/2: 32 Unaspirated stop + *r clusters Glottal Sounds Proto-Voiced Sounds Proto-Voiced Aspirations & Breathy Sounds 2) Patterns of tonal split and merger are as follows. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development with voiceless-voiced register splits. Horizontally: two individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G] A[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] Merger B[2, 1/2] = DL[ 2, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] C[2, 1/2] = DS[ 2, 1/2] 392 34.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Mid-rising 35 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: Low level-falling 221 occurs in A[2, 1/2]. Tone 3: Mid level 33 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G] and DL[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 4: High-falling 53 occurs in B[2, 1/2] and DL[2, 1/2]. Tone 5: High level 55 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G] and in DS[1-A, 1-C, 1-U, 1UC, 1-G]. Tone 6: A mid-falling with a wider decreasing amplitude 31 occurs in C[2, 1/2] and another mid-falling with a narrower decreasing amplitude 32 occurs in DS[2, 1/2], to form a complementary distribution in smooth and checked syllables respectively. 34.5 Phonation types or voice qualities of tones in this dialect According to the sound recording of the only one LRP, there are no phonation types or voice qualities of tones in this dialect. 34.6 Sound changes involving tones reflected in the wordlist of this dialect 1) The term hja31 ‘to seek’ has an unexpected C2 tone, while its cognate in other Tai varieties is always of A1 tone. 2) The initial of the term ʔa:35 ‘crow’ is of the initial group of glottal sounds, as that in other NT varieties. 3) As those in other NT varieties, the following items designated into DL columns show a regular tone change to be coordinate with the counterpart DS columns: nok55 (DS1-C) ‘deaf’, hək55 ‘gums’ (DS1-A), tuk55 (DS1-UC) ‘thin bamboo stripes’, ˀdit55 (DS1-G) ‘sunshine’, and mit32 (DS2) ‘knife’. 34.7 Other sound changes reflected in the wordlist of this dialect 1) The vowel of ke:33 ‘egg’ is unexpected. The sound change needs more data to determine with. 2) The initial of kom33 ‘to cover up’ is unexpected. The sound change needs more data to determine with. 3) The term ŋɔn33 ‘soft’ has been changed the initial from glottal -ʔ to nasal ŋ-, while its tonal category is contained to be at the B1-G column. 393 4) The initial of ɕəɯ53 ‘chopsticks’ is unexpected because all other NT varieties show an initial t-, but the tonal category of this term is contained to be at the B12-A column. The sound change needs more data to determine with. 5) As those in other NT varieties, the initial of θipDS1-A ‘centipede’ which is designated into DS1-A reflects *s- or *ɬ- which should be considered as DS1-C. 394 35. Language 35 Jiaren, Du’an Yong’an, Central Hongshuihe Zhuang, NT 35.1 Language information Language Central Hongshuihe Zhuang ISO 639-3 Code zch Autonym to:C1; ɕuəŋB2 Exonym Du’an Zhuang Data Source 1 LRP Jiaren (village name) of Du’an Hongshuihe Zhuang, which is spoken at Yong’an Variety Name Township, Du’an County, Guangxi Zhuang Autonomous Region, China. 35.2 Information of data source Serial number of LRP LRP1 Date May 20, 2013 Chuangyi Art School, Nanning City, Guangxi Zhuang Autonomous Place of data elicitation Region, China. Given Name Chū nyáng (春陽) Family Name Mó (磨) Gender male Age 42 Birth Place Jiaren Community (加仁屯, ka:B1 ɲanB2 in local Zhuang), Yong’an Village (永安村), Yong’an Township (永安鄉), Du’an County (都安縣), Hechi Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Jiaren Community (16 years). Nandan County, Hechi Prefecture (2 years). The main town of Du’an County (2 years). He is now living in Nanning City. Time period of living at the present residence Occupation vice-president of school 22 years Education on-the-job junior college Languages spoken Level Language Remarks Du’an Central mother Hongshuihe Zhuang tongue Standard Mandarin intermediate This language is the national language of the LRP’s country. Guiliu Dialect intermediate A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Du’an Central Hongshuihe Zhuang as native, he can communicate in Guiliu Southwestern Mandarin. LRP’s mother was born in Banke Community (板客屯), Anhua Village (安 安鄉), Du’an County (都安縣). She is monolingual in her Du’an Zhuang. 395 村), Yong’an Township (永 35.3 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there is no tonal flip-flop in this dialect. Smooth Syllable Checked Syllable Phonation types ProtoVoiceless Sounds A B C DL DS Aspirations A1-A: 51̰ B1-A: 33 C1-A: 553ʔ DL1-A: 33 DS1-A: 55 Continuants A1-C: 51̰ B1-C: 33 C1-C: 553ʔ DL1-C: 33 DS1-C: 55 Unaspirated stops A1-U: 51̰ B1-U: 33 C1-U: 553ʔ DL1-U: 33 DS1-U: 55 A1-UC: 51̰ B1-UC: 33 C1-UC: 553ʔ DL1-UC: 33 DS1-UC: 55 A1-G: 51̰ B1-G: 33 C1-G: 553ʔ DL1-G: 33 DS1-G: 55 A2: 231 B2: 31 C2: 24ʔ DL2: 31 DS2: 35 A1/2: 231 B1/2: 31 C1/2: 24ʔ DL1/2: 31 DS1/2: 35 Unaspirated stop + *r clusters Glottal Sounds Proto-Voiced Sounds Proto-Voiced Aspirations & Breathy Sounds 2) Patterns of tonal split and merger are as follows: Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development with voiceless-voiced register splits. Horizontally: two individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G] A[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] Merger B[2, 1/2] = DL[2, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G]=DS[1-A, 1-C, 1-U, 1-UC, 1-G] C[2, 1/2]=DS[2, 1/2] 396 35.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High-falling 51 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: Mid rising-falling 231 occurs in A[2, 1/2]. Tone 3: Mid level 33 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G] and DL[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 4: Mid-falling 31 occurs in B[2, 1/2] and DL[2, 1/2]. Tone 5: High level-falling 553 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G], and High-level 55 occurs in DS[1-A, 1-C, 1-U, 1-UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Low-rising [24] occurs in C[2, 1/2] and Mid-rising [35] occurs in DS[2, 1/2], to form a complementary distribution in smooth and checked syllables respectively. 35.5 Phonation types or voice qualities of tones in this dialect A1 tone has a creakiness triggered by its rapid falling contour from the highest pitch 5 to the lowest pitch 1; Tones C1 and C2 have a glottal constriction, like hu:553ˀ ‘cloud’, pja:j553ˀ ‘to walk’, ʔa:553ˀ ‘to open’, ma:24ˀ ‘horse’, ɹam24ˀ ‘water’, and faj24ˀ ‘tree’. All these voice qualities are not phonemic. 35.6 Sound changes involving tones reflected in the wordlist of this dialect 1) The initial of the term ʔa:51 ‘crow’ is of the initial group of glottal sounds, as that in other NT varieties. 2) The tone of the term ma:51 ‘to come’ is of A1 tone, to present another type of voicing alternation which reflects a proto-voiceless initial in NT and voiced in CT/SWT. 3) The tone of the term kon33 ‘CLF of stone’ is unexpectedly of B1-UC tone, while this word is expected to be of C1-UC tone in the tone box designed. 4) The tone of the term ˀdaŋ33 ‘to make fire’ is unexpectedly of B1-G tone, while this word is expected to be of C1-G tone in the tone box designed. 5) The tone of the term tuaj33 ‘bowl’ is unexpectedly of B1-U tone, while this word is expected to be of C1/2-A tone in the tone box designed. However, the LRP also claims that tuaj13ˀ ‘bowl’ which is of C12-A tone is also spoken in other villages around his area. 6) As those in other NT varieties, the following items designated into DL columns show a regular tone change to be coordinate with the counterpart DS columns: 397 nok55 (DS1-C) ‘deaf’, hək55 ‘gums’ (DS1-A), tuk55 (DS1-UC) ‘thin bamboo stripes’, ˀdit55 (DS1-G) ‘sunshine’, mit35 (DS2) ‘knife’, and tək35(DS12-A) ‘to hit the mark’. 35.7 Other sound changes reflected in the wordlist of this dialect 1) The vowel ɛ: in ˀdɛ:w51 ‘star’ is an aberrant development when being compared with a long a: reflected by its cognate (<*ˀda:wA) in other Tai varieties. 2) The term ŋɔn44 ‘soft’ has been changed the initial from glottal -ʔ to nasal ŋ-, while its tonal category is contained to be at the B1-G column. 3) As those in other NT varieties, the initial of θipDS1-A ‘centipede’ which is designated into DS1-A reflects *s- or *ɬ- which should be considered as DS1-C. 398 Language 36 Yizhou Suogan, Guibei Zhuang, NT 36.1 Language information Language Guibei ISO 639-3 Code zyb Autonym ɕuəŋB2 Exonym Man Data Source 2 LRPs Variety Name Suogan (village name) of Guibei Zhuang. This variety is spoken at some villages in the rural areas in Ma’an Township, Yizhou City, Hechi Prefecture, Guangxi Zhuang Autonomous Region, China. From the innovation voicing alternation induced by sesquisyllabic onsets between some Guibei Zhuang and all the other Tai varieties (VASO-GB) shared between this variety and Huanjiang Chengguan which distributes in the neighboring Huanjiang County, it can be speculated that some Guibei Zhuang varieties around that area may be able to be put under an individual language with a new ISO 639-3 code in further studies and surveys. 36.2 Information of data source Serial number of LRP LRP1 Place of data elicitation Qixing Road 13, Nanning City, Guangxi Zhuang Autonomous Region, China. Given Name F́zhèng (福正) Family Name Wéi (韋) Gender male Age 73 Birth Place Date May 19, 2013 Liangqiao Community (良橋屯, li:ŋA2 kiəwA2 in local Zhuang) Suogan Village (索敢村, ʰɹo:kDL1-A ka:mC1 in local Zhuang), Anma Township (安馬鄉), Yizhou City, Hechi Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Liangqiao Community (15 years). Yizhou City (12 years). Guilin City (4 years). Tiandong County, Baise Prefecture (1 year). Jinchengjiang City (3 years). He is now living in Nanning City. Time period of living at the present residence Occupation retired cadre 39 years Education bachelor Languages spoken Level Language Remarks Suogan Guibei Zhuang mother tongue Standard Mandarin fluent This language is the national language of the LRP’s country. Guiliu Dialect fluent A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. This language is the LRP’s hometown Yizhou City’s lingua franca. Russia intermediate His undergraduate major was Russian Language. 399 The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Suogan Guibei Zhuang as native, he can communicate in Guiliu Southwestern Mandarin. LRP’s mother was born in Daping Community (大平屯), Suogan Village (索敢村), Anma Township (安馬 鄉), Yizhou City, Hechi Prefecture, Guangxi. She was monolingual in Suogan Guibei Zhuang. Serial number of LRP LRP2 Place of data elicitation Changfu Road 35, Nanning City, Guangxi Zhuang Autonomous Region, China. Given Name Hù idǒu (慧斗) Family Name Q́n (覃) Gender male Age 31 Birth Place Date May 12, 2013 Liangqiao Community (良橋屯, li:ŋA2 kiəwA2 in local Zhuang) Suogan Village (索敢村, ʰɹo:kDL1-A ka:mC1 in local Zhuang), Anma Township ( 安 馬 鄉 ), Yizhou City, Hechi Prefecture, Guangxi Zhuang Autonomous Region, China. Places lived Liangqiao Community (12 years). Yizhou City (8 years). He is currently living in Nanning City. Time period of living at the present residence Occupation engineering management 10 years Education bachelor Languages spoken Level Language Remarks Suogan Guibei Zhuang mother tongue Standard Mandarin fluent This language is the national language of the LRP’s country. Guiliu Dialect fluent A Southwestern Mandarin variety spoken in the northwestern part of Guangxi, China. This language is the LRP’s hometown Yizhou City’s lingua franca. English a little The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village. Besides Suogan Guibei Zhuang as native, he can communicate in Guiliu Southwestern Mandarin and standard Mandarin. LRP’s mother was born in Xiadong Community ( 洞屯, ʰɹʊŋB2 la:C1 in local Zhuang), Dacai Township (大才鄉), Huanjiang County (環江縣), Hechi Prefecture, Guangxi. Her mother tongue is very closed to Suogan Guibei Zhuang. Besides, she also speaks a little Guiliu Southwestern Mandarin. 400 36.3 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there is no tonal flip-flop in this dialect. Phonation types ProtoVoiceless Sounds Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 51̰ B1-A: 33 C1-A: 42 DL1-A: 33 DS1-A: 55 Continuants A1-C: 51̰ B1-C: 33 C1-C: 42 DL1-C: 33 DS1-C: 55 Unaspirated stops A1-U: 51̰ B1-U: 33 C1-U: 42 DL1-U: 33 DS1-U: 55 A1-UC: 51̰ B1-UC: 33 C1-UC: 42 DL1-UC: 33 DS1-UC: 55 A1-G: 51̰ B1-G: 33 C1-G: 42 DL1-G: 33 DS1-G: 55 A2: 51̰ B2: 33 C2: 42 DL2: 33 DS2: 55 A2: 231 B2: 11̰ C2: 24ʔ DL2: 23 DS2: 23 A1/2: 231 B1/2: 11̰ C1/2: 24ʔ DL1/2: 23 DS1/2: 23 Unaspirated stop + *r clusters Glottal Sounds Proto-Voiced Sounds Proto-Voiced Sounds 2) Patterns of tonal split and merger are as follows: Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2] [2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2] [2, 1/2] Some of the items designated into the voiced sounds row ([2]) have the same tone as the one in higher register ([1-A, 1-C, 1-U, 1-UC, 1-G]), while most others in this row keep distinct tone in the lower register. This tonal split is unique because tonal splits at the row of voiced sounds are not found in the vast majorities of Tai. However, a similar tonal split pattern is found in the neighboring KS languages. Tonal split in the row of voiced sounds makes this dialect cannot share the common NT tone box with other NT varieties. A specific Guibei Zhuang tone box (see Chapter 6) derived from the common NT tone box is designated based on this dialect. Except this unique tonal split, tones in this dialect are of straightforward tonal development with voiceless-voiced register splits. Horizontally: two individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G, 2] A[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G, 2] = DS[1-A, 1-C, 1-U, 1-UC, 1-G, 2] Merger B[2, 1/2] = DS[ 2, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G, 2]=DL[1-A, 1-C, 1-U, 1-UC, 1-G, 2] C[2, 1/2]=DL[ 2, 1/2] 401 36.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High-falling 51 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G] and in tones of some items designated into A2 (like məŋ51 ‘hand’), as well as a higher level 55 occurs in DS[1A, 1-C, 1-U, 1-UC, 1-G] and in tones of some items designated into DS2 (like nɔk55 ‘bird’), to form a complementary distribution in smooth and checked syllables respectively. Tone 2: Low-rising-falling 231 occurs in A [2, 1/2], as well as low-rising [23] occurs in DL[2, 1/2] and DS[2, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 3: High level 33 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G] and in tones of some items designated into B2 (like hu33 ‘pair’) as well as in DL[1-A, 1-C, 1-U, 1-UC, 1-G, 2], to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low-level 11 occurs in B[2, 1/2], Tone 5: High-falling 42 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G] and in tones of some items designated into C2 (like nam42 ‘water’ and maj42 ‘tree’). Tone 6: Low-rising 24 occurs in C[2, 1/2]. 36.5 Phonation types or voice qualities of tones in this dialect A1 (51) has a creakiness triggered by its rapid falling contour from the highest pitch 5 to the lowest pitch 1; B2 (11) also accompanies a weak creakiness due to the lowest pitch; C2 has a glottal constriction, like bə:24ˀ ‘bride’, təŋ24ˀ ‘crutch’, and ma:24ˀ ‘horse’. All these voice qualities are not phonemic. 36.6 Sound changes involving tones reflected in the wordlist of this dialect 1) In the wordlist, the items which are designated into the voiced sounds row ([2]) but have the same tone as the one in the high register ([1-A, 1-C, 1-U, 1-UC, 1-G]) are as follows: məŋ51 (A2 > A1) ‘hand’, hu:33 (B2 > B1)‘pair’, nam42 (C2 > C1) ‘water’, maj42 (C2 > C1) ‘tree’, and nɔk55(DS2 > DS1) ‘bird’. These items normally have initial consonants developing from *mw- (vs. Li 1977: 74, Liang & Zhang 1996: 97, Pittayawat Pittayaporn 2009: 154) like in məŋ51 ‘hand’ and maj42 ‘tree’, *nr- (vs. Li 1977: 131, Liang & Zhang 1996: 97) or *C̬ .n (vs. Pittayawat Pittayaporn 2009: 177) like in nam42 ‘water’ and nɔk55 ‘bird’, and an undetermined *g- (vs. Li 1977: 198, Pittayaporn 2009: 113) or *gl- (Liang & Zhang:1996: 226) 402 like hu:33 ‘pair’70. It is interesting that these initials (except *g- > h- in hu:33 ‘pair’) present as the same to those in CT/SWT varieties (like mʊŋA2 ‘hand’, majC2 ‘tree’, namC2 ‘water’ and nɔ:kDS2 ‘bird’ in the Yang Zhuang varieties), but different from those in the vast majorities of NT, which have f- and r- for these items respectively, like fəŋA2 ‘hand’, fajC2 ‘tree’, ramC2 ‘water’ and rɔkDS2 ‘bird’ in Du’an. This phenomenon is also found in Huanjiang (also of Guibei Zhuang) (vs. Zhang et al. 1999: 192, 198). The items presenting a special voicing alternation between Suogan/Huanjiang and the vast majority of Tai varieties make the originally simple NT tone box not able to capture all the tonal splits. 2) The initial of the term ʔa:51 ‘crow’ is of the initial group of glottal sounds, as that in other NT varieties. 3) The tone of the term ma:51 ‘to come’ (provided by LRP2) is of A1 tone, to present another series of voicing alternation which reflects a proto-voiceless initial in NT and voiced in CT/SWT. 4) As those in other NT varieties, the following items designated into DL columns show a regular tone change to be coordinate with the counterpart DS columns: nuk55 (DS1-C) ‘deaf’, hɯk55 ‘gums’ (DS1-A), tuk55 (DS1-UC) ‘thin bamboo stripes’, and ˀdit55 (DS1-G) ‘sunshine’. Since both DL2 and DS2 have merged into one another to have a low-rising tone (23), it is not clear that mit23 ‘knife’ and tɯk23 ‘to hit the mark’ which are designated into DL columns but are normally coordinate with DS columns in other NT varieties have tones DS12-A and DS2 respectively or not. 36.7 Other sound changes reflected in the wordlist of this dialect 1) s-/θ-/ɬ- reflecting *s- (vs. Li 1977) or *ɬ- (vs. Liao & Shen 2012) in other Tai varieties has been changed to a voiceless approximant ɹ̥- (or sometimes a free variation r̥-) in this dialect, like r̥a:m51 ‘three’, ɹ̥i:33 ‘four’, and ɹ̥aj42 ‘intestine’. 2) Some proto-voiced initial obstruents are preserved as voiced in this dialect, like (kaj33) ɣa:ŋ11 (B1/2) ‘young chicken’ and bəɰ24 (C2) ‘bride’, but some are merged into voiceless sounds, like tɯŋ24 (C2) ‘crutch’. The term hu:33 (B1) ‘pair’ has an unexpected initial (k- in NT and most CT/SWT but ɕʰ- in Thai (SWT), Zuozhou (CT) and Baoxu (CT)) and tone (expected to have tone B2 in NT), and it is not certain that it is cognate with *gu:B ‘pair’ reflected in other Tai varieties. If it is indeed the cognate, the changes of the initial and the tone need more counterpart data to be determined with. 70 403 3) The term wun33 ‘soft’ has been changed the initial from a glottal -ʔ to an approximant w-, while its tonal category is contained to be at the B1-G column. 4) The term ŋa:42 ‘to open’ has been changed the initial from a glottal -ʔ to a nasal ŋ-, while its tonal category is contained to be at the C1-G column. 5) As those in other NT varieties, the initial of ɹ̥ipDS1-A (gɯ:A2) which is designated into DS1-A reflects *s- (vs. Li 1977) or *ɬ- (vs. Liao & Shen 2012) which should be considered as DS1-C. 6) The final consonant -k has a free variation -ʔ in this dialect, like pja:ʔDL1 ‘forehead’, pjaʔDS1 ‘vegetable’ and ʔaʔDS1 ‘chest’. 7) For the item ‘guest’‚ LRP1 provided hekDL1 but LRP2 provided kekDL1. 8) The initial ɣ- has a free variation g-, like ɣaw24 / gaw24 ‘rice’. 404 Language 37 Baipeng, Liujiang Zhuang, NT 37.1 Language information Language Liujiang Zhuang ISO 639-3 Code zlj Autonym tsu:ŋB2 Exonym tsu:ŋB2; su:ŋB2 Data Source Zhang et al. 1999: 33-34, 75-81, 595-808. Data Info. This language is spoken in Baipeng Township (百朋鎮), Liujiang Township (柳江縣), Liuzhou City (柳州市), Guangxi Zhuang Autonomous Region, China. 37.2 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there is no tonal flip-flops. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 42 B1-A: 33 C1-A: 53 DL1-A: 24 DS1-A: 55 Continuants A1-C: 42 B1-C: 33 C1-C: 53 DL1-C: 24 DS1-C: 55 Unaspirated stops A1-U: 42 B1-U: 33 C1-U: 53 DL1-U: 24 DS1-U: 55 A1-UC: 42 B1-UC: 33 C1-UC: 53 DL1-UC: 24 DS1-UC: 55 A1-G: 42 B1-G: 33 C1-G: 53 DL1-G: 24 DS1-G: 55 A2: 231 B2: 22 C2: 24 DL2: 22 DS2: 33 A1/2: 231 B1/2: 22 C1/2: 24 DL1/2: 22 DS1/2: 33 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds Proto-Voiced Sounds Proto-Voiced Aspirations & Breathy Sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development. 405 Horizontally: two individuals and four mergers Individual A[1-A, 1-C, 1-U, 1-UC, 1-G] A[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[2, 1/2] Merger B[2, 1/2] = DL[2, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] C[2, 1/2] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] 37.3 Tonal characteristics & tone occurrences: five distinct tones Tone 1: High-falling 42 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: Low-rising-falling 231 occurs in A[2, 1/2]. Tone 3: Mid level 33 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G] and DS[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low level 22 occurs in B[2, 1/2] and in DL[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High-falling 53 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G] and in DS[1-A, 1-C, 1-U, 1-UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. 37.4 Phonation types or voice qualities of tones in this dialect. We do not know the phonation types or voice qualities of tones in this dialect since they are not mentioned in the original text. 406 Language 38 Donglan, Guibei Zhuang, NT 38.1 Language information Language Guibei Zhuang ISO 639-3 Code zgb Autonym ˀjaiC1-G Exonym Donglan Zhuang Data Source Zhang et al. 1999: 32-33, 99-101, 595-808. Data Info. This language is spoken in the main town of Donglan County, Hechi Prefecture, Guangxi Zhuang Autonomous Region, China. 38.2 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there is no tonal flip-flops except tones C in this dialect. Smooth Syllable Checked Syllable Phonation types ProtoVoiceless Sounds A B C DL DS Aspirations A1-A: 53 B1-A: 35 C1-A: 33 DL1-A: 35 DS1-A: 55 Continuants A1-C: 53 B1-C: 35 C1-C: 33 DL1-C: 35 DS1-C: 55 Unaspirated stops A1-U: 53 B1-U: 35 C1-U: 33 DL1-U: 35 DS1-U: 55 A1-UC: 53 B1-UC: 35 C1-UC: 33 DL1-UC: 35 DS1-UC: 55 A1-G: 53 B1-G: 35 C1-G: 33 DL1-G: 35 DS1-G: 55 A2: 231 B2: 11 C2: 14 DL2: 11 DS2: 11 A1/2: 231 B1/2: 11 C1/2: 14 DL1/2: 11 DS1/2: 11 Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development. 407 Horizontally: three individuals and three mergers A[2, 1/2] Individual C[1-A, 1-C, 1-U, 1-UC, 1-G] C[2, 1/2] A[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] Merger B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] C[2, 1/2] = DL[2, 1/2] = DS[2, 1/2] 38.3 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High-falling 53 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G] as well as high level 55 occurs in DS [1-A, 1-C, 1-U, 1-UC, 1-G, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 2: Low-rising-falling 231 occurs in A[2, 1/2]. Tone 3: Mid-rising 35 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G] and DL[1-A, 1-C, 1-U, 1UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low level 11 occurs in B[2, 1/2], as well as in DL[2, 1/2] and DS[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Mid level 33 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 6: Low rising 14 occurs in C[2, 1/2]. 38.4 Phonation types or voice qualities of tones in this dialect We do not know the phonation types or voice qualities of tones in this dialect since they are not mentioned in the original text. 408 Language 39 Gehan, Qiubei Zhuang, NT 39.1 Language information Language Guibian Zhuang ISO 639-3 Code zgn Autonym ˀi:C1-G Exonym ˀjo:jC1-G; Shahua Data Source Zhang et al. 1999: 38-40, 72-75, 595-808. Data Info. The data of Gehan (township name) are adopted from Zhang et al. (1999). This language is spoken in Gehan Township (革寒鎮), Qiubei County (丘 縣), Wenshan Zhuang and Miao Autonomous Prefecture, Yunnan Province, China. 39.2 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there are tonal flip-flops on A and B tones, but are not on C, DL, and DS tones in this dialect. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 32 B1-A: 12 C1-A: 35 DL1-A: 55 DS1-A: 55 Continuants A1-C: 32 B1-C: 12 C1-C: 35 DL1-C: 55 DS1-C: 55 Unaspirated stops A1-U: 32 B1-U: 12 C1-U: 35 DL1-U: 55 DS1-U: 55 A1-UC: 32 B1-UC: 12 C1-UC: 35 DL1-UC: 55 DS1-UC: 55 A1-G: 32 B1-G: 12 C1-G: 33 DL1-G: 55 DS1-G: 55 A2: 53 B2: 31 C2: 33 DL2: 22 DS2: 22 A1/2: 53 B1/2: 31 C1/2: 33 DL1/2: 22 DS1/2: 22 ProtoVoiceless Sounds Unaspirated stop + *r clusters Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts: Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC] [1-G, 2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of non-straightforward tonal development. Secondary tonal split conditioned by glottalized sounds is found in Column C. 409 Horizontally: four individuals and two mergers A[1-A, 1-C, 1-U, 1-UC, 1-G] Individual A[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] B[2, 1/2] Merger C[1-A, 1-C, 1-U, 1-UC] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] C[1-G, 2, 1/2] = DL[2, 1/2] = DS[2, 1/2] 39.3 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Mid-falling 32 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: High-falling 53 occurs in A[2, 1/2]. Tone 3: Low-rising 13 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 4: Mid-falling 31 occurs in B[2, 1/2]. Tone 5: Mid-rising 35 occurs in C[1-A, 1-C, 1-U, 1-UC], as well as high level 55 occurs in DL[1-A, 1-C, 1-U, 1-UC, 1-G] and in DS[1-A, 1-C, 1-U, 1-UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 6 Mid level 33 occurs in C[1-G, 2, 1/2], as well as low level 22 occurs in DL[2, 1/2] and in DS[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. 39.4 Phonation types or voice qualities of tones in this dialect We do not know the phonation types or voice qualities of tones in this dialect since they are not mentioned in the original text. 410 40. Language 40 Xiaosanjiang, Lianshan Zhuang, CT 40.1 Language information Language Lianshan Zhuang ISO 639-3 Code zln Autonym tshy:ŋB2 Exonym -- Data Source Zhang et al. 1999: 40-42, 114-119, 595-808. Data Info. This variety is spoken in Xiaosanjiang Township (小 江鎮), Lianshan County (連 山縣), Qingyuan Prefecture, Guangdong Province. Lianshan Zhuang is the only Zhuang language distributing in Guangdong Province. 40.2 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there are no tonal flip-flops in this dialect. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 42 B1-A: 35 C1-A: 55 DL1-A: 35 DS1-A: 55 Continuants A1-C: 42 B1-C: 35 C1-C: 55 DL1-C: 35 DS1-C: 55 Unaspirated stops A1-U: 42 B1-U: 35 C1-U: 55 DL1-U: 35 DS1-U: 55 A1-UC: 42 B1-UC: 35 C1-UC: 55 DL1-UC: 35 DS1-UC: 55 A1-G: 42 B1-G: 35 C1-G: 55 DL1-G: 35 DS1-G: 55 A2: 132 B2: 13 C2: 214 DL2: 13 DS2: 13 A1/2: 132 B1/2: 13 C1/2: 214 DL1/2: 13 DS1/2: 13 ProtoVoiceless Sounds Unaspirated stop + *r clustersz Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development. 411 Horizontally: three individuals and three mergers A[1-A, 1-C, 1-U, 1-UC, 1-G] A[2, 1/2] Individual C[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] Merger B[2, 1/2] = DL[2, 1/2]= DS[2, 1/2] C[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] 40.3 Tonal characteristics & tone occurrences: six distinct tones Tone 1: High-falling 42 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: Low-rising-falling 132 occurs in A[2, 1/2]. Tone 3: Mid-rising 35 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G] and in DL[1-A, 1-C, 1-U, 1UC, 1-G]. Tone 4: Low-rising 13 occurs in B[2, 1/2], as well as in DL[2, 1/2] and DS[2, 1/2] to form a complementary distribution in smooth and checked syllables respectively. Tone 5: High level 55 occurs in C[1-A, 1-U, 1-UC, 1-G] and in DS[1-U, 1-UC, 1-G] to form a complementary distribution in smooth and checked syllables respectively. Tone 6: Low-falling-rising 214 occurs in C[2, 1/2] 40.4 Phonation types or voice qualities of tones in this dialect We do not know the phonation types or voice qualities of tones in this dialect since they are not mentioned in the original text. 412 Language 41 Huishui, Bouyei, NT 41. Language information Language Bouyei ISO 639-3 Code pcc Autonym ˀjʊjC1-G Exonym Bouyei Data Source 2 LRPs Variety Name This variety is of the Central Qian Group of Bouyei. This variety is spoken at the rural areas around Dima Township, Huishui County, Qiannan Bouyei and Miao Autonomous Prefecture, Guizhou Province, China. 4.2 Information of data source Serial number of LRP LRP1 Place of data elicitation Room 606, Panyu Hotel, Panyu City, Guangdong Province, China. Given Name Cháokuān (朝寬) Family Name Lú (羅) Gender male Age 49 Birth Place Dazhai Village (大 Date May 3, 2013 , ˀba:nC1-G wʊŋA1 in local Bouyei), Dima Township (抵麻鄉), Huishui County (惠水縣), Qiannan Bouyei and Miao Autonomous Prefecture, Guizhou Province, China Places lived Dazhai Village (10.5 years). Huaxi District of Guiyang City (the capitcal of Guizhou Province) (9 years). The main town of Huishui County (4 years). He is now living in Panyu City of Guangdong Province. Time period of living at the present residence Occupation teacher 25 years Education Languages spoken Level Huishui Bouyei mother tongue Standard Mandarin fluent bachelor Language Remarks This language is the national language of the LRP’s country. Southwestern Mandarin fluent This language is the lingua franca of the whole Guizhou Province. Cantonese Intermediate This language is the lingua franca of the cities in Pearl River Delta in Guangdong Province. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents were born in the same village as the LRP’s. Besides Huishui Bouyei as native, both of them can communicate in Southwestern Mandarin and Standard Mandarin. 413 Serial number of LRP LRP2 Place of data elicitation Room 606, Panyu Hotel, Panyu City, Guangdong Province, China. Given Name Cháoyǔ (朝宇) Gender male Birth Place Dazhai Village (大 , ˀba:n Date C1-G May 3, 2013 Family Name Lú (羅) Age 43 A1 wʊŋ in local Bouyei), Dima Township (抵麻鄉), Huishui County (惠水縣), Qiannan Bouyei and Miao Autonomous Prefecture, Guizhou Province, China Places lived Dazhai Village (11 years). Huaxi District of Guiyang City (the capitcal of Guizhou Province) (10 years). Xiamen City of Fujian Province. Shenzhen of Guangdong Province. He is now living in Panyu City of Guangdong Province. Time period of living at the present residence Occupation 20 years company employee Education bachelor Languages spoken Level Language Remarks Huishui Bouyei mother tongue Standard Mandarin fluent This language is the national language of the LRP’s country. Southwestern Mandarin fluent This language is the lingua franca of the whole Guizhou Province. Cantonese Intermediate This language is the lingua franca of the cities in Pearl River Delta in Guangdong Province. The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s parents were born in the same village as the LRP’s. Besides Huishui Bouyei as native, both of them can communicate in Southwestern Mandarin and Standard Mandarin. Remarks LRP2 is the younger brother of LRP1. 41.3 Basic patterns of tonal development of this dialect 1) As shown in the chart below, there are no tonal flip-flops in this dialect. Phonation types Smooth Syllable Checked Syllable A B C DL DS Aspirations A1-A: 24ʔ B1-A: 55ʔ C1-A: 33 DL1-A: 55 DS1-A: 24 Continuants A1-C: 24ʔ B1-C: 55ʔ C1-C: 33 DL1-C: 55 DS1-C: 24 Unaspirated stops A1-U: 24ʔ B1-U: 55ʔ C1-U: 33 DL1-U: 55 DS1-U: 24 A1-UC: 24ʔ B1-UC: 55ʔ C1-UC: 33 DL1-UC: 55 DS1-UC: 24 A1-G: 24ʔ B1-G: 55ʔ C1-G: 22 DL1-G: 55 DS1-G: 24 Proto-voiced sounds A2: 11̰3ʔ B2: 41ʔ C2: 22 DL2: 41 DS2: 113 Proto-voiced aspirations & A1/2: 11̰3ʔ B1/2: 41ʔ C1/2: 22 DL1/2: 41 DS1/2: 113 ProtoVoiceless Sounds Unaspirated stop + *r clustersz Glottal Sounds breathy sounds 414 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC] [1-G, 2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of non-straightforward tonal development. The only nonstraightforward column is C, with a spilt being voiceless register on one side and golttal/voiced register on the other side. Horizontally: two individuals and four mergers Individual C[1-A, 1-C, 1-U, 1-UC] C[1-G, 2, 1/2] A[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] Merger A[2, 1/2] = DS[2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, 1-C, 1-U, 1-UC, 1-G] B[2, 1/2] = DL[2, 1/2] 41.4 Tonal characteristics & tone occurrences: six distinct tones Tone 1: Low-rising 24 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G] and DS[1-A, 1-C, 1-U, 1UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively. Tone 2: Low level-rising 113 occurs in A[2, 1/2] and DS[2, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 3: High level 55 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G], and in DL[1-A, 1-C, 1-U, 1UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively. Tone 4: High-falling 41 occurs in B[2, 1/2] and DL[2, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Mid level 33 occurs in C[1-A, 1-C, 1-U, 1-UC]. Tone 6: Low level 22 occurs in C[1-G, 2, 1/2]. 415 41.5 Phonation types or voice qualities of tones in this dialect In this dialect, C tones do not accompany with a glottal constriction, while items with all other tones unexpectedly have a slight or clear glottal constriction, like pja24ʔ (A1) ‘hill’, ma:24ʔ (A1) ‘dog’, kɨn24ʔ ‘to eat’, wa:j113ʔ (A2) ‘buffalo’, si:55ˀ ‘four’ (B1), and ta:41ˀ (B2) ‘river’. This phenomenon is very unique because glottal constrictions are mostly found on C tones in many other Tai varieties. This may be triggered by the mergers between Tones A and DS, as well as between Tones B and DS, as the following analyses: on original dead syllables, *-k after a long vowel in PT been changed to a slight non-phonemic glottal constriction to merge into their counterpart tones B (B1 and B2), like pa:55ˀ ‘forehead’, jia55ˀ ‘guest’, pa:55ˀ ‘mouth’, ma:55ˀ ‘bean’ (semantic change from ‘fruit’), ta:55ˀ ‘to expose under the sun’, tiɑ55ˀ ‘cracking’, ˀjiə55ˀ ‘hungry’, and ða:41ˀ‘root’; *-k after a short vowel in PT has been changed to a clear non-phonemic glottal constriction to merge into their counterpart tones A (A1 and A2), like to:24ˀ ‘thin bamboo stripes’, pja24ˀ ‘vegetable’, ðua:113ˀ ‘bird’, ða:113ˀ ‘to steal’, and ɕɥa:113ˀ ‘cooked’; this may be why the tonal value of tones on the vertical column of DL and DS has merged into the tonal value of their counterpart tones B and A; the reason why items with tones on the columns A and B normally have a slight final glottal might be because in this dialect proto-final *-k from tones DL and DS might have brought about an effect, a slight glottal final -ˀ, into tones B and A; the glottalization on tones A is clearer than on tones B, probably because *-k after a short vowel in PT tends to be changed to a clearer glottalization than *-k after a long vowel does, and the former one has merged into tones on the A column, and the later one has merged into tones on the B column. The original glottalization on tones C might have been dropped due to the extrusion of the later development of glottalization on Tones A and B. In addition, Tone A2 (113) normally has a creakiness triggered by its extremely low pitch 11 before rising to a higher pitch 3. All these voice qualities in this dialect are not phonemic. 41.6 Sound changes involving tones reflected in the wordlist of this dialect 1) The initial of the term ʔa:11 ‘crow’ is of the initial group of glottal sounds, as that in other NT varieties. 2) The tone of ˀdaŋ55 ‘to make fire’ is unexpectedly of B1-G tone, while this word is expected to be of C1-G tone in the tone box designed. 416 3) As those in other NT varieties, the following items designated into DL columns show a regular tone change to be coordinate with the counterpart DS columns: no:24 (DS1-C) ‘deaf’, to:24 (DS1-UC) ‘thin bamboo stripes’, ˀdit24 (DS1-G) ‘sunshine’, ˀdat55 ‘hot’ and miat11 (DS2) ‘knife’. 41.7 Other linguistic changes reflected in the wordlist of this dialect 1) ʔa:j24 ‘turn face up’ is a glottal sound, while in most Tai dialects it is a nasal ŋ-. The final consonant of tiŋ113ˀ ‘hornet’ may have been changed from -n to -ŋ. 2) As those in other NT varieties, the initial of θipDS1-A ‘centipede’ which is designated into DS1-A reflects *s- or *ɬ- which should be considered as DS1-C. 3) Sound changes on dead syllables present as follows: (1) *-k after a long vowel in PT is completely lost, like pa:55 ‘forehead’, jia55 ‘guest’, pa:55 ‘mouth’, ma:55 ‘bean’ (semantic change from ‘fruit’), ta:55 ‘to expose under the sun’, tiɑ55 ‘cracking’, ˀjiə55 ‘hungry’, and ða:41‘root’; *-k after a short vowel in PT has been changed to a slight non-phonemic glottal final to merge with tones A (A1 and A2), and this means that tonal value of tones on the vertical column of DS has merged into the tonal value of their counterpart tones A, like to:24ˀ ‘thin bamboo stripes’, pja24ˀ ‘vegetable’, ðua:113ˀ ‘bird’, ða:113ˀ ‘to steal’, and ɕya:113ˀ ‘cooked’; the reason why items with Tone A always have a slight final glottal which we have previously pointed out might be because in this dialect proto-final *-k from tones DS might have brought about an effect, a slight glottal final -ˀ, into tones A; (2) *-t and *-p are preserved; however, all items with DL1 tones and *-t /*-p finals, proto-long vowels have been changed to short vowels, like wat55 ‘astringent’, ɣap55 ‘to carry’, ɣop55 ‘circumference’, kat55 ‘mustard plant’, and kwat55 ‘to hug’. 417 Language 42 Shitouzhai, Zhenning Bouyei, NT 42.1 Language information Language Bouyei ISO 639-3 Code pcc Autonym pow24 ʔi:31 Exonym Bouyei Data Source 1 LRP Variety Name Shitouzhai, Western Qian Group (or Zhenning Bouyei), Bouyei. This variety is spoken in Shitouzhai Village (石頭 ), Huangguoshu Township (黃果樹 鎮), Anshun City (安順市), Qiannan Bouyei and Miao Autonomous Prefecture, Guizhou Province, China. Although this language is put under the individual language Bouyei with ISO 639-3 code “pcc”, it is clear that varieties of this language are not intelligible with varieties of the other two Bouyei groups, namely Central Qian of Bouyei and Southern Qian of Bouyei. Its unique features like aspirated sounds, diphthongization, and the loss or weakening of final stops, together with speaker attitudes suggest that it should be an individual language with its own ISO 639-3 code. 42.2 Information of data source Serial number of LRP LRP1 Date May 2, 2013 Qifu Shijie (祈福食街), Zhongcun Village, Shiguang Road, Panyu Place of data elicitation City, Guangdong Province, China. Given Name Dēngyàn (登艷) Family Name Wǔ (伍) Gender male Age 38 Birth Place Shitouzhai Village (石頭 ), Huangguoshu Township (黃果樹鎮), Anshun City (安 順市), Qiannan Bouyei and Miao Autonomous Prefecture, Guizhou Province, China. Places lived Shitouzhai Village (15 years). Huangguoshu Township (3 years). Anshun City (3 years). Beijing (5-6 years). From 2003 up to now, he has been living in Panyu City, Guangdong Province. Time period of living at the present residence Occupation worker 10 years Education Languages spoken Level Huishui Bouyei mother tongue Standard Mandarin fluent senior high school Language Remarks This language is the national language of the LRP’s country. Southwestern Mandarin fluent This language is the lingua franca of the whole Guizhou Province. Cantonese Intermediate This language is the lingua franca of the cities in Pearl River Delta in Guangdong Province. 418 The main factors of the process of LRP’s language cultivating behavior like the language background of LRP’s parents LRP’s father was born in the same village as the LRP’s. Besides Zhenning Bouyei as native, both of them can communicate in Southwestern Mandarin and Standard Mandarin. LRP’s mother was born in Zhedou Village, Guanling County, Anshun City (安順市), Qiannan Bouyei and Miao Autonomous Prefecture, Guizhou Province. Her mother tongue is also of Zhenning Bouyei and is identical with the LRP’s one. She also speaks Southwestern Mandarin and Standard Mandarin. Remarks Because of the many mergers of phonemes, most nouns in this dialect are disyllable, like ti:55 ka:33 ‘leg’, ti:55 ʋoŋ11 ‘hand’, ʔə:33puə33 ‘hill’, and pe44 ɕi:ŋ44 ‘tail’, to distinguish the meaning of many homophones on single syllables. Many natural monosyllable nouns in other Tai varieties are added a prepositive classifier (often with aberrant sound change) in this dialect. For example, the LRP insists that the word for ‘dog’ must be tu:31 ma:33 but not ma:33, or people would not understand its accurate meaning. Besides the disyllable nouns, sometimes the LRP also added another segment to a verb or a classifier to make the item become a compound word. For example, the LRP provided haŋ55 fa:11 for ‘to cover’, but actually the segment fa:11 is actually ‘quilt’, and haŋ55 fa:11actually refers to ‘to cover up a quilt’. In a similar way, the LRP provides kow24 o:33 for ‘pair’, but the segment o:33 is actually a phonetic change of ‘one’. Therefore, he did not follow my expected scheme to pronounce all the monosyllable items during the whole recording. This is very important because some disyllabic words distinguish meanings from other words containing the homophonous segments of its nuclear syllable. For example, on the contrast ˀda:ŋ33 ka:j53 ‘egg’ vs. tu:11 ka:j53 ‘chicken’, the first syllables are CLF of substance and animal respectively, and they play important role of distinguishing the meaning of the following nuclear syllables which are homophonous with one another. Another example is (ˀdan33) taʔ ‘liver’ vs. taʔ55(ʔo31) ‘to scoop (rice)’. Some tone sandhi or tone coarticulation is found in the recording of disyllable items, but the specific situation need more data to research with. In the data analyses, only the syllables being cognate with those in other Tai varieties are employed. 42.3 Basic patterns of tonal development of this dialect 1) As shown in the chart above, there are no tonal flip-flops in this dialect. Smooth Syllable Checked Syllable Phonation types A B C DL DS Aspirations A1-A: 33 B1-A: 55ʔ C1-A: 31 DL1-A: 55 DS1-A: 55 Continuants A1-C: 33 B1-C: 55ʔ C1-C: 31 DL1-C: 55 DS1-C: 55 Unaspirated stops A1-U: 33 B1-U: 55ʔ C1-U: 31 DL1-U: 55 DS1-U: 55 A1-UC: 33 B1-UC: 55ʔ C1-UC: 31 DL1-UC: 55 DS1-UC: 55 A1-G: 33 B1-G: 55ʔ C1-G: 31 DL1-G: 55 DS1-G: 55 A2: 11 B2: 24ʔ C2: 31 DL2: 24 DS2: 24 A1/2: 11 B1/2: 24ʔ C1/2: 31 DL1/2: 24 DS1/2: 24 ProtoVoiceless Sounds Unaspirated stop + *r clustersz Glottal Sounds Proto-voiced sounds Proto-voiced aspirations & breathy sounds 419 2) Patterns of tonal split and merger are as in the following two charts. Vertically: two way splits and no splits Tonal category Register 1 Register 2 A [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] B [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] DL [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] DS [1-A, 1-C, 1-U, 1-UC, 1-G] [2, 1/2] In conclusion, tones in this dialect are of straightforward tonal development. The only specific column is C with no spilts. Horizontally: three individuals and two mergers A1[1-A, 1-C, 1-U, 1-UC, 1-G] Individual A2[1-G, 2, 1/2] C [1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2] B[1-A, 1-C, 1-U, 1-UC, 1-G] = DS[1-A, 1-C, 1-U, 1-UC, 1-G] = DL[1-A, Merger 1-C, 1-U, 1-UC, 1-G] B[2, 1/2] = DL[2, 1/2] = DS[2, 1/2] 42.3 Tonal characteristics & tone occurrences: five distinct tones Tone 1: Mid level 33 occurs in A[1-A, 1-C, 1-U, 1-UC, 1-G]. Tone 2: Low level 11 occurs in A[2, 1/2]. Tone 3: High level 55 occurs in B[1-A, 1-C, 1-U, 1-UC, 1-G], as well as in DL[1-A, 1-C, 1-U, 1-UC, 1-G] and DS[1-A, 1-C, 1-U, 1-UC, 1-G], to form a complementary distribution in smooth and checked syllables respectively. Tone 4: Low-rising 24 occurs in B[2, 1/2], as well as in DL[2, 1/2] and DS[2, 1/2], to form a complementary distribution in smooth and checked syllables respectively. Tone 5: Mid-falling 31 occurs in C[1-A, 1-C, 1-U, 1-UC, 1-G, 2, 1/2]. 42.4 Phonation types or voice qualities of tones in this dialect In this dialect, C tones do not accompany with a glottal constriction, while items with tones in the B column normally have a slight glottal constriction, like (ˀda:ŋ33) ka:j55ˀ ‘egg’, ɬej55ˀ ‘four’ (B1), and (ʔə33) ta:24ˀ (B2) ‘river’. This may be triggered by the mergers between tones B and DL, as the following analyses: on original dead syllables, *-k after a long vowel in PT been changed to a slight non-phonemic glottal constriction to merge into their counterpart tones B (B1 and B2), like (na:31) pa:55ˀ 420 ‘forehead’, (pou31) ɕiə55ˀ ‘guest’, (ˀda:ŋ33) pa:55ˀ ‘mouth’, (ˀdam33) ma:55ˀ ‘fruit’, ta:55ˀ ‘to expose under the sun’, piə55ˀ ‘cracking’, (tʊŋ31) ʔi:55ˀ ‘hungry’, la:24ʔ (va:j31) ‘root’ and (tiaw33) tɕa:24ˀ‘rope’; the reason why items with tones on the B column normally have a slight final glottal might be because in this dialect proto-final *-k from tones DL might have brought about an effect, a slight glottal final -ˀ, into tones B; the original glottalization on tones C might have been dropped due to the extrusion of the later development of glottalization on tones B. The glottal constriction in this dialect is not phonemic. 42.5 Sound changes involving tones reflected in the wordlist of this dialect 1) The A1 tones sometimes tend to have a higher pitch 44 than its original pitch 33 when it is pronounced individually. But when it precedes or follows a B1 tone (55, sometimes a free variation 53), it clearly presents a mid-level 33 for being easily distinguished from the high level 55 of B1 tone. 2) The tonal value of B1 has two free variations, high level (55) and high-falling (53 when being deliberately pronounced), such as (ˀda:ŋ33) ka:j53 and (ˀda:ŋ33) ka:j55 ‘egg’. Another example is that the first syllable in kɔŋ55 ˀda:33‘swaddling clothe’ has a B1 tone, and it was clearly pronounced as high level and high falling respectively. 3) The initial of the term ʔa:11 ‘crow’ is of the initial group of glottal sounds, as that in other NT varieties. 4) The second syllable of the item ʑa:24 nja:33 ‘medicine’ should be cognate with *ʔjɯəA reflected in other Tai varieties. Its initial consonant has been changed from a pre-glottalized sound to a yodicized nasal nj-. The environment of the sound change needs more data to be studied with. 5) The tone of kʰa:33 ‘get stuck’ has an A1 tone (while other Tai varieties have A2 tone for this item), to leave a question open. 6) The tone of ti:55 ʋoŋ11 ‘hand’ is clearly of A2, but in the item ʋoŋ33 kwa:11 ‘right side’, the first syllable refers to ‘hand’. The tone change from A2 to A1 is suspected to be a tone sandhi, but need more data to confirm it. 42.6 Other linguistic changes reflected in the wordlist of this dialect 1) Short -aj in other Tai varieties becomes long a:j, like pa:j33 ‘to go’; long -a:j in other Tai varieties becomes -e:, like te:33 ‘to die’ and ʋe:11 ‘buffalo’. The second 421 syllable of the item kʊŋ44 te:53 ‘maternal grandfather’ should be cognate with ‘maternal grandmother’ in other Tai varieties due to its vowel (-a:j > -e:) and its B1 tone (the vast majorities of NT reflect *ta:jB for ‘maternal grandmother’). In the investigation, the LRP also told that the word for ‘maternal grandmother’ is na:j53 te:53. The prefixes kʊŋ44 and na:j53 are actually Chinese loans ‘grandfather’ (<公 kʊŋ55) and ‘grandmother’ (<奶 na:i214) respectively, and they must be added to the etymon te:53 (<*ta:jB ‘maternal grandmother’) to distinguish the meanings. Therefore, this item is ruled out in this dialect because it is not the cognate with *ta:A ‘maternal grandfather’ reflected in other Tai varieties. 2) -a (open front unrounded vowel) contrasts with -ɑ (open back unrounded vowel), like kʰa31 ‘to kill’ vs. kʰɑ31 ‘near’. 3) Unaspirated stops with Tone C1 in other NT varieties normally present aspirated in this dialect. For example, kʰa:31 ‘to kill’, lok44 kʰa:31 ‘seedling’, tʰã ‘to establish’ and kʰou31 ‘nine’. The aspiration on C1 contrasts with unaspiration on C2. 4) Sound changes on nasals: -m in other Tai varieties has been changed to -ŋ in this dialect, like ɬaŋ33 ‘three’ and za:ŋ31 ‘water’; -ŋ in other Tai varieties has been changed to -n in this dialect, like lək24 tin33 ‚‘cucumber’, but some are nasalized, like tã ‘to arrive’. The change environments need more data to confirm with. The n in other Tai varieties seems not to be changed，like tu11 tən11 ‘hornet’ and kən33 ‘to eat’ 5) Proto-Tai *ɣ- is changed to the glottal initial ʔ- like in ʔo:31 ‘rice’ and ʔiə31 ‘excrement’. 6) Proto-Tai *tr- change to ɕ- in this dialect, like ɕən33 (<*trin A) ‘stone’ and ɕəʔ55 (<*trap DL) ‘to carry’ . 7) The sound files of (lok44) kʰa:31 ‘seedling’, (ˀbən33) ləʔ55 ‘get dark’, and (tu:11) taʔ55 ‘grasshopper’ are followed by modal final particle word ma31 or ɑ31 to be emphasized by the LRP. 8) Sound changes on dead syllables present as follows: 1) *-k after a long vowel is completely lost, like (na:31) pa:55 ‘forehead’, (ˀda:ŋ33) pa:55 ‘mouth’, (ˀda:m33) ma:55 ‘fruit’, ta:55 ‘to expose under the sun’, and za:24 (va:j31) ‘root’; *-k after a short vowel is preserved or is changed to -ʔ, like (zi:11) nok55 ‘deaf’, pak55 or paʔ55 ‘vegetable’, and naʔ55 ‘heavy’. 422 APPENDIX D THE MAPS OF THE TAI VARIETIES INVESTIGATED See the amplified detail in Map 2 MAP 1. Tai varieties investigated in South China and MSEA (cf. Map 2) *This map is produced based on the Google Map. The language names and numberings with ISO 639-3 codes are as in the chart below (also cf. Appendix C). 423 MAP 2. Investigated Central Tai and Northern Tai varieties concentrating in Guangxi and its neighboring areas (cf. Map 1) *This map is produced based on the Baidu Map. The language names and numberings with ISO 639-3 codes are as in the chart below (also cf. Appendix C). Seven from SWT (L1-7) 15. Debao Tuoxin (unknown) Thirteen from NT (L30-42) 1. Bangkok Thai (tha) 16. Jingxi Huashan (zgm) 30. Wuming Shuangqiao (zyn) 2. Songkhla Southern Thai (sou) 17. Tiandeng Xiangdu (zzj) 31. Debao Longsang (unknown) 3. Khon Kaen Isan (tts) 18. Daxin Naling (zzj) 32. Debao Dongling (zyj) 4. Chiang Rai Northern Thai (nod) 19. Jingxi Lingding (zzj) 33. Tianlin Lizhou (zgn) 5. Kho Lam Shan (shn) 20. Xiaoguangnan (zhn) 34. Shanglin Yunling (zch) 6. Rong Maet Tai Lue (khb) 21. Wenshan Dazhai (zhg) 35. Du’an Jiaren (zch) 7. Muong Quan Son Tai (unknown) 22. Daxin Leiping (zzj) 36. Yizhou Suogan (zgb) 23. Daxin Baoxu (zzj) 37. Liujiang Baipeng (zlj) 24. Bac Va (nut) 38. Donglan Urban (zgb) Seventeen from CT (L8-L24) 8. Debao Urban (zyg) 39. Qiubei Gehan (zqe) 9. Debao Dalong (zyg) Five from YN (L25-29) 40. Lianshan Xiaosanjiang (zln) 10. Debao Lüliu (zyg) 25. Yongning Baiji (zyn) 41. Huishui Bouyei (pcc) 11. Jingxi Hurun (zyg) 26. Nanning Shuangding (zyn) 42. Zhenning Bouyei (pcc) 12. Jingxi Urban (zyg) 27. Long’an Xiaolin (zyn) 13. Jingxi Anning (zyg) 28. Jingxi Daqiu (zyn) 14. Debao Nalong (unknown) 29. Yongning Xialeng (zyn) 424 RESUME Name: Liao Hanbo Date of Birth: 25 July 1977 Place of Birth: Chengguan Township, Debao County, Guangxi, China Institutions Attended: 1996-2000, Bachelor of Arts in Japanese Language (Tourism), Xi’an Foreign Languages University (西安外国语学 ). 2010-2011, Postgraduate Foreign Research Student (The phonological study of Zhuang), The University of Tokyo (東京大学). 2011-2016, Master of Arts in Linguistics, Payap University (มหาวิทยาลัยพายัพ). Academic Publications: Liao Hanbo. 2010. 台语支中部组佒侬语 德靖土语 音系概况及其拼音方案详解 [An overview of the sound system of the Central Taic language Yang-Nong (Dejing vernacular) and description of a phonetic spelling system]. In Li Fuqiang and Gao Ya-ning (ed.), Chinese Zhuang Studies 4:70-192. Beijing: Publishing House of Minority Nationalities. Liao, Hanbo. 2016a. Diachronic Hierarchies of Tai Tonal Development. Payap University Journal, 26(2), to appear. Liao, Hanbo. 2016b. Proto-Tai reconstruction of ‘maternal grandmother’ revisited: *na:jA, *ta:jA or *ta:jB? Language and Linguistics, to appear. 425 Academic Papers Presented: Liao, Hanbo. 2013. Proto-Tai reconstruction of ‘maternal grandmother’ revisited: *na:jA or *ta:jA? SEALS23, Bangkok, 29–31 May, 2013. Liao, Hanbo. 2015. Yang Zhuang poetry. SEALS25, Chiang Mai, 27–29 May, 2015. Liao, Hanbo & Shen, Ruiqing. 2012. Gedney’s tone box revisited: Evidence from some varieties of Central and Northern Tai. ICSTLL45, Singapore, 26–28 October, 2012. Shen, Ruiqing & Liao, Hanbo. 2012. Acoustic-tonetic Study of Pjang Zhuang: An undescribed Central Tai Variety. ICSTLL45, Singapore, 26–28 October, 2012. 426