R-marking: Referential Person Affixes in Classical Nahuatl Nouns

人文社会系研究科 修 士 学 位 論 文 R-marking: Referential Person Affixes in Classical Nahuatl Nouns （古典ナワトル語における名詞の指示対象人称接辞について） 2012 年 12 月 提出 言語学専門分野 21-116003 佐々木 充文 (Mitsuya SASAKI) R-marking: Referential Person Affixes in Classical Nahuatl Nouns Japanese title: 古典ナワトル語における名詞の指示対象人称接辞について A thesis submitted to the Graduate School of Humanities and Sociology in partial fulfillment of the requirements for the degree of Master of Letters By Mitsuya SASAKI Department of Linguistics The University of Tokyo Tokyo, Japan December, 2012 CONTENTS Contents Resumen en español / Spanish summary v Preface and acknowledgments vi Abbreviations ix List of tables xii List of figures xii 1 Introduction 1 1.1 The aim of this paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Theoretical importance of R-marking . . . . . . . . . . . . . . . . . . . 1 1.3 On the language, data, and notation . . . . . . . . . . . . . . . . . . . . . 6 1.3.1 Language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.3.2 Data and sources . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.3.3 Transcription, glosses, and notation . . . . . . . . . . . . . . . . 10 1.3.4 Grammatical terms . . . . . . . . . . . . . . . . . . . . . . . . . 12 2 Morphology and syntax of R-marking 2.1 2.2 2.3 15 Noun–verb distinction and omnipredicativity . . . . . . . . . . . . . . . 15 2.1.1 Noun–verb distinction . . . . . . . . . . . . . . . . . . . . . . . 15 2.1.2 Omnipredicativity . . . . . . . . . . . . . . . . . . . . . . . . . 16 Morphology of R-marking . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2.1 An overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2.2 Subject person marking on verbs . . . . . . . . . . . . . . . . . . 18 2.2.3 R-marking on nouns . . . . . . . . . . . . . . . . . . . . . . . . 20 2.2.4 Person prefixes and pronouns . . . . . . . . . . . . . . . . . . . 22 Cross-reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 ii CONTENTS 2.4 2.5 2.6 Distribution and usage of R-marked nouns . . . . . . . . . . . . . . . . . 28 2.4.1 R-marked nouns as principals . . . . . . . . . . . . . . . . . . . 29 2.4.2 R-marked nouns as non-principals . . . . . . . . . . . . . . . . . 32 2.4.3 R-marking in multi-word nominal construction . . . . . . . . . . 34 2.4.4 Nouns without R-marking . . . . . . . . . . . . . . . . . . . . . 35 R-marking in modern Nahuan languages . . . . . . . . . . . . . . . . . . 38 2.5.1 Nahuan languages with uniform R-marking . . . . . . . . . . . . 39 2.5.2 Nahuan languages without uniform R-marking . . . . . . . . . . 42 Comparable phenomena in other languages . . . . . . . . . . . . . . . . 45 2.6.1 Nominal predicates with person markers . . . . . . . . . . . . . . 45 2.6.2 Noun–verb uniformity for predicates . . . . . . . . . . . . . . . . 47 2.6.3 R-marking on non-predicate nouns . . . . . . . . . . . . . . . . . 50 3 Theoretical background 3.1 3.2 58 Language-internal analyses . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.1.1 Launey’s omnipredicative analysis . . . . . . . . . . . . . . . . . 58 3.1.2 Andrews’ omniclausal analysis . . . . . . . . . . . . . . . . . . . 59 Cross-linguistic issues . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 3.2.1 Typological status of omnipredicativity . . . . . . . . . . . . . . 61 3.2.2 Typological and theoretical importance . . . . . . . . . . . . . . 63 4 Discussion 65 4.1 Towards a non-omnipredicative account . . . . . . . . . . . . . . . . . . 65 4.2 Terminology and notation . . . . . . . . . . . . . . . . . . . . . . . . . . 66 4.3 Classification of R-marked nouns and R-marking . . . . . . . . . . . . . 68 4.3.1 Principal, intraeventive, and appositional . . . . . . . . . . . . . 68 4.3.2 Single-word vs. multi-word . . . . . . . . . . . . . . . . . . . . 72 iii CONTENTS 4.3.3 Section summary . . . . . . . . . . . . . . . . . . . . . . . . . . 74 4.4 Multi-word R-marking . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 4.5 Intraeventive R-marked nouns . . . . . . . . . . . . . . . . . . . . . . . 77 4.5.1 Problems with the simple omniclausal analysis . . . . . . . . . . 77 4.5.2 Omnipredicative hypothesis . . . . . . . . . . . . . . . . . . . . 81 4.5.3 Non-omnipredicative hypothesis . . . . . . . . . . . . . . . . . . 84 4.5.4 Section summary . . . . . . . . . . . . . . . . . . . . . . . . . . 88 Appositional R-marked nouns . . . . . . . . . . . . . . . . . . . . . . . 89 4.6.1 Preliminary notions . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.6.2 Counterexamples to the omnipredicative analysis . . . . . . . . . 91 4.6.3 Non-omnipredicative analysis . . . . . . . . . . . . . . . . . . . 97 4.6.4 Section summary . . . . . . . . . . . . . . . . . . . . . . . . . . 104 4.6 4.7 The status of R-marking . . . . . . . . . . . . . . . . . . . . . . . . . . 104 4.7.1 R-marking as a realization of a semantic feature . . . . . . . . . . 105 4.7.2 Status of principal R-marking . . . . . . . . . . . . . . . . . . . 109 4.7.3 R argument and R-marking . . . . . . . . . . . . . . . . . . . . . 114 4.7.4 Section summary . . . . . . . . . . . . . . . . . . . . . . . . . . 115 5 Conclusion 116 References 119 iv CONTENTS Resumen En esta tesis analizaremos los llamados “prefijos sujetos”, que denominaremos a partir de ahora prefijos R, y que se anteponen los nombres en náhuatl clásico. El náhuatl clásico, junto con algunos de sus parientes modernos, tiene la particularidad de que los “prefijos sujetos” se tienen que anteponer a casi todos los nombres, independientemente de que éstos pertenezcan al predicado o al argumento, si bien en la tercera persona el prefijo es nulo (Ø-). Por ejemplo, la frase yo fui rico se dice ni-koko:kawa’ ni-katka (1SGR-rico 1SGS-fui). Por lo tanto, el náhuatl clásico ha sido calificado como idioma omnipredicativo en algunos trabajos teóricos recientes como los de Launey (1994) y Andrews ([1975] 2003), quienes afirman que todos los nombres en este idioma son originalmente predicados, y que su uso como argumentos es secundario. Si esta suposición es correcta, se llegaría a una sorprendente conclusión teórica y tipológica, y los nuevos problemas que aparecerían serían numerosos. Contrariamente, en esta tesis proponemos un análisis no omnipredicativo de los nombres con prefijos R. Un examen sintáctico más atento nos revela que dichos nombres no funcionan como un binomio preformado del tipo sujeto-predicado, sino como una unidad nominal. Después de presentar los datos básicos con respecto a este fenómeno (marca R), clasificaremos los nombres con prefijos R en diversos tipos: se distinguen los nombres principales, intraeventivos y aposicionales, y tambien se identifican la marca R unipalabra y multipalabra. Algunas de estos tipos no se podrían explicar según la hipótesis omnipredicativa. También se muestra que la marca R afecta a todo el sintagma nominal, y no a cada nombre individual. Aquí proponemos considerar los prefijos R como una expresión de los rasgos semánticos de cada sintagma nominal, y trataremos de dar una explicación unificada de todos los casos de aparición de marca R. Nuestro análisis es que los prefijos R, a pesar de su apariencia engañosa, realmente no participan en la estructura de las frases. Por último nos gustaría señalar que este fenómeno necesita una investigación más profunda, y que aún queda pendiente evaluar cuál sería su área de expansión interlingüística, además de su clasificación tipológica. v CONTENTS Preface and acknowledgments This thesis discusses so-called “subject” marking on nouns, here designated as R-marking, and proposes the non-omnipredicative interpretation of the phenomenon. Since the time of Sapir (1921), the universality of noun–verb distinction has been the subject of a great deal of controversy. However, the discussion has tended to focus on the presence or absence of a lexical distinction within a particular language, and the issue of the universality of predicate–argument dichotomy has been left with less attention. Classical Nahuatl, a Uto-Aztecan language formerly spoken in Central Mexico, provides a significant example in this respect. As pointed out by Launey (1994) and Andrews ([1975] 2003), Classical Nahuatl has uniform and obligatory “subject” marking on both predicate and argument nouns. Consequently, this language has non-third-person nonpronominal argument nouns. The uniformity of R-marking led Launey (1994) and Andrews ([1975] 2003) to assume that Classical Nahuatl is a language where every noun is primarily a predicate (omnipredicative in Launey’s terminology). Their frameworks state that the difference between nominal predicates and argument nouns in this language is simply whether they are embedded or not. Following their understanding, Classical Nahuatl should be characterized as a peculiar language which exhibits an extreme predicate–argument symmetry — a feature which even the well-known alleged “noun–verb symmetrical languages” do not possess. A closer look, however, reveals that some usages of R-marking cannot be explained by their omnipredicative approach. Many examples suggest that R-marked nouns are not treated as pre-formed subject–object pairs in the syntax of Classical Nahuatl. As an alternative, this thesis presents a non-omnipredicative analysis of R-marking. After giving an overview of the R-marking phenomenon in Classical Nahuatl and several other vi CONTENTS languages, I classify R-marked nouns into a few different types and argue that some of them can be explained more consistently by a non-omnipredicative approach. Instead of assuming pre-formed subject–object complexes, I attempt to interpret R-marking as the realization of certain semantic features which do not participate in sentence composition. Despite its theoretical and typological importance, the limited cross-linguistic distribution and the syntactic marginality of uniform R-marking makes it less known to linguists. While many typological works mention the subject person marking on nominal predicates across languages, the R-marking on non-predicate nouns are hardly noticed; we do not even know how to call it so far. Through the discussion of R-marking in Classical Nahuatl, this thesis intends to provide the data and a descriptive framework for the sake of further studies, and above all, to show that it is a phenomenon worth investigating. In writing this thesis, I received invaluable help from many people in and outside of the Department of Linguistics at the University of Tokyo. First of all, I am deeply grateful to my advisor Prof. Hiroshi Kumamoto. I appreciate the honor of being one of his last students at the Department of Linguistics. I am also sincerely grateful to Prof. Tooru Hayasi, Prof. Yoshiki Nishimura, and Prof. Masato Kobayashi. In particular, I received generous support from Prof. Kobayashi in citing the data from Malto and Kurux, and had useful comments from Prof. Hayasi and Prof. Kobayashi respectively on Turkic and Dravidian languages. I am also thankful to Dr. Itsuki Nagasawa and Ms. Rika Sekine for their daily support. I owe the data and comments on Kaqchikel, Nuuchahnulth, Nahuatl of Cuetzalan, Sakha, and Q’eqchi’ respectively to Prof. Yoshiho Yasugi, Prof. Toshihide Nakayama, Prof. Tomás Amaya Aquino, Dr. Fuyuki Ebata, and Mr. Ken Shibushita. In particular, the relevant parts of this thesis were totally rewritten according to the comments from Prof. Yasugi and Prof. Nakayama. Without their expert comments, they would have contained vii CONTENTS many more errors than they do now. I am also thankful to Prof. Nozomi Kodama for the comments on Dravidian languages. The theoretical part of this thesis is partly based on my presentation at ICLAA, Tokyo, on 21 July 2012. I gratefully appreciate the comments from the people who attended the workshop, especially from Prof. Shinjiro Kazama, Prof. Honoré Watanabe, Dr. Hiroyuki Umetani, Mr. Yasuhiro Kojima, Mr. Masayuki Ishizuka, and Mr. Shun Nakamoto. I would also like to say thanks to the people whom I had the chance to see at and after the Yale Nahuatl Conference 2012. In finding Classical Nahuatl examples, I was deeply benefited from the Nahuatl text searching software TEMOA provided by Sup-Infor (http://www.sup-infor.com/). I would like to express thanks to its authors. I am also grateful to Mark Rosa-san, Ignacio Quirossan, and Yumi Mimura-san for helping me correct my English and Spanish. Por último quisiera darles cenxiquipilli de gracias a todos los nahuatlatos, presentes o pasados. Mā cemihcac onnemi in amotlahtōltzin. ‘¡Que viva su idioma para siempre!’ (Lastly, I would like to express eight thousand thanks to all the speakers of Nahuatl, both contemporary and from the past. May your language live forever.) viii CONTENTS Abbreviations Glosses COMP Complementizer Person (1, 2, 3, UH, UN), number (SG, PL), and types COND Conditional of person markers (S, R, O, P) are used in combina- DIM Diminutive tion: 1SGS (first-person singular subject), EN Ethnic-group name EUPH Euphonic segment EXCL Exclusive UHP (un- specified human possessor), etc. 1 First person EXHRT Exhortative 2 Second person FUT Future 3 Third person GER Gerund A Antecessive particle HON Honorific ABS Absolutive state HONP Honorific possessor ACC Accusative IMP Imperative ADM Admonitive IMPF Imperfect AFF Affirmative particle IN Quasi-article in ANIM Animate INCL Inclusive APPL Applicative INCMP Incomplete ART Article IND Indicative B Set B person marker IPP Imperfect participle B2 Base 2 IRR Irrealis CAUS Causative L Linking suffix CISL Cislocative LN Place name ix CONTENTS LOC Locative SG Singular M Masculine TRSL Translocative NA Nonactive UH Unspecified human NEG Negative UN Unspecified nonhuman NM Non-masculine VOC Vocative O Object OPT Optative - Morpheme boundary P Possessor = Clitic boundary PASS Passive . Unanalyzed combination PD Possessor derivational suffix Ø Zero morph or morpheme PL Plural CAPITAL Loan in Spanish spelling PN Personal name ‘ ... ’ POS Possessive state PRF Perfect PRS Present PST Past hon. Honorific PUN Punctual lit. Literally PURP Purposive WeEXCL We (exclusive) Q Question particle YouSG You (singular) R R-marking YouPL You (plural) RDP Reduplication YourSG Your (singular) REFL Reflexive YourPL Your (plural) S Subject Free translation Translation x CONTENTS Sources BD Bancroft Dialogues (Karttunen and Lockhart, 1987) Cant. Cantares mexicanos (Bierhorst, 1985a) Chim. Codex Chimalpopoca (Bierhorst, 1992a) Crón. Crónica mexicáyotl (Chimalpahin, 1997) FC Florentine Codex (Sahagún, 1950–1982) Plát. Plática by Olmos (Olmos, [1547] 1875) Prim. Primeros memoriales (Sahagún, 1997) Psalm. Psalmodia christiana (Sahagún, [1583] 1993) [G] Grammarian source [V] Verse source [X] Christian source xi LIST OF FIGURES List of tables 1 R-prefixes in Classical Nahuatl . . . . . . . . . . . . . . . . . . . . . . . 17 2 Verbal subject prefixes in Classical Nahuatl . . . . . . . . . . . . . . . . 19 3 Example of R-marking in Classical Nahuatl . . . . . . . . . . . . . . . . 21 4 R-prefixes and independent pronouns in Classical Nahuatl . . . . . . . . 23 5 Malto pronominal suffixes and personal pronouns . . . . . . . . . . . . . 50 6 Middle Elamite person markers and pronouns . . . . . . . . . . . . . . . 54 List of figures 1 Verbal inflection template of Classical Nahuatl . . . . . . . . . . . . . . . 19 2 Nominal inflection template of Classical Nahuatl . . . . . . . . . . . . . 20 3 Structure of the sentence-word in (12a) . . . . . . . . . . . . . . . . . . . 26 4 Cross-referencing structure of (12b) (simplified) . . . . . . . . . . . . . . 26 5 Reanalysis of (11) according to the extension model (simplified) . . . . . 27 6 Pacific Northwestern type of noun–verb symmetry . . . . . . . . . . . . 48 7 Pacific Northwestern type of noun–verb symmetry (extended) . . . . . . 49 8 Classical Nahuatl type of noun–verb symmetry . . . . . . . . . . . . . . 49 9 Cross-referencing structure of (62a) and (63a) (simplified) . . . . . . . . 71 10 Cross-referencing structure of (87) (simplified) . . . . . . . . . . . . . . 95 11 Cross-referencing structure of (88) (simplified) . . . . . . . . . . . . . . 96 xii 1 INTRODUCTION 1 Introduction 1.1 The aim of this paper Uniform and redundant “subject” person marking (R-marking) on both predicate and argument nouns is a feature which distinguishes Classical Nahuatl and some of its modern relatives from many other polysynthetic languages. So far, however, disappointingly little attention has been paid to this phenomenon especially outside of the tradition of Classical Nahuatl grammatical studies. It has rarely been covered in descriptive grammars on modern Nahuan languages or cited in typological literature. Even the proper terminology to describe and discuss this phenomenon has not yet been provided. The purpose of this thesis is twofold. The first is to clarify the syntactic and semantic aspects of R-marking in Classical Nahuatl and provide a non-omnipredicative analysis of R-marking; specifically, this thesis argues that R-marked nouns in Classical Nahuatl are nouns in the syntactic sense rather than subject–predicate complexes as Launey (1991, 1994, 2003, 2004) and Andrews ([1975] 2003) assume. The second is to make this phenomenon available for further typological and theoretical studies by providing enough data and a tentative descriptive framework. 1.2 Theoretical importance of R-marking Since the contrast between nouns and verbs has been considered as one of the most fundamental properties of human language, the parallelism between these two categories has always attracted the interest of linguists from various fields. Formalists hatched the X-bar theory out of the cross-categorical symmetry between verbs and nouns; semanticists have tried to explain lexical categorization in semantics; typologists have discussed innumerable alleged counterexamples to the universality of noun–verb distinction. 1 1 INTRODUCTION As stated by Launey (1991, 1994, 2003, 2004) and Andrews ([1975] 2003), Classical Nahuatl exhibits a peculiar type of noun–verb symmetry, which Launey designates as omnipredicativity. Firstly, while this language has a rigid lexical distinction between nouns and verbs, both of them can in principle serve either as predicates or as arguments. A noun takes the same set of prefixes as the verbal subject markers and serves as the nominal predicate of a sentence. (1) a. Verbal predicate ka ti-k-mati AFF 2 SGS -3 SGO -know ‘YouSG know it.’ (FC VI, 110) b. Nominal predicate ka ti-pi:lli AFF 2 SGR -noble ‘YouSG are a noble.’ (FC VI, 94) More importantly, a noun takes these “subject” prefixes obligatorily even when it is not the predicate (principal) of a sentence. In (2), for example, non-predicate nouns are redundantly person-marked despite that the copula (anyeske’) in (2a) and the matrix verb (timoči:was) in (2b) themselves have subject person markers respectively. (2) a. aw in i:k an-λa’to’ke’ an-ye-ske’ and IN thus 2PLR-leader 2PLR-be-FUT ‘So youPL will be the leaders.’ (Crón. 74) b. ok seppa ti-pilto:nλi ti-mo-či:wa-s still once 2SGR-baby 2SGS-REFL-make-FUT ‘YouSG will become a baby again.’ 2 (FC III, 18) 1 INTRODUCTION This becomes clearer when compared with a cognate language Pipil. As shown by the examples in (3), Pipil lacks the redundant person marking of the type observed in (2). (3) Pipil a. n-yu ni-nemi deskalsoh 1SGS-go 1SGS-be barefoot ‘I am going to be barefoot.’ b. pal ti-nech-kwepa (Campbell, 1985, 110) Cristianoj so 2SGS-1SGO-turn Christian ‘in order that youSG make me a Christian’ (NBTN, 2012, Acts 26:28) Since this “subject”-like person marking in Classical Nahuatl is not limited to the typical subject of a clause, I shall henceforth neutrally call it R-marking, the “R” standing for “referential.” Likewise, I shall refer to the prefixes used in R-marking as R-prefixes. The definition of these terms will be discussed in Section 1.3.4. The obligatory R-marking on nouns enables this language to have first- or secondperson nonpronominal arguments, which are usually translated by relative clauses or parenthetical appositions in Spanish and English. (4) a. wel ši-k-kakika:n ši-k-a:naka:n in an-te:-pilwa:n in well 2PLS-3SGO-hear.OPT 2PLS-3SGO-catch.OPT IN 2PLR-UHP-child IN an-te:-išwi:wa:n ... 2PLR-UHP-grandchild ‘Listen and accept it well, youPL children and grandchildren.’ b. san ni-mic/-siyawkec/a ... ni-mo-we:iyo: only 1SGS-2SGO-salute (Crón. 65) san ni-ton-elnanto’ 1SGR-2SGP-greatness only 1SGR-PN-PN ‘Exactly I, yourSG greatness Don Hernando myself, salute youSG .’ [V] (Cant. 51r:27) 3 1 INTRODUCTION As shall be considered in detail in Section 3.1 below, the distribution of R-marking in Classical Nahuatl raises a radical theoretical question: if all nouns in Classical Nahuatl appear with “subject” markers, does it mean that all nouns in this language have subjects? Launey (1991, 1994, 2003, 2004) and Andrews ([1975] 2003) give affirmative answers to this question. Launey (1994) suggests that, in Classical Nahuatl, nouns as well as verbs are primarily predicative and that their argumental function is a secondary one derived by a process of actantialisation. In his model, R-prefixes retain their status as subject markers through actantialisation (Launey, 1994, 71–74). Andrews ([1975] 2003) presents a rather radical picture; in his analysis, all nominal words in Classical Nahuatl are really clauses (nominal nuclear clauses: NNCs) just as verbal words are (verbal nuclear clauses: VNCs). Following him, such a sentence as (5a) should be glossed as (5b); he analyzes the nominal words i:tew ‘its stone (egg)’ and to: tolin ‘turkey’ as two equational or attributive clauses (i.e. NNCs) which respectively mean ‘it is its egg(s)’ and ‘it is a/the turkey’ (italic indicates that the NNC is the principal of a sentence). His framework thus interprets all nouns and verbs in Classical Nahuatl as minimal clauses. (5) a. o:=ni-k-kwa’ Ø-i:-tew Ø-to:tolin A =1 SGS -3 SGO -eat. PST 3 SGR -3 SGP -stone 3 SGR -turkey ‘I ate the turkey’s eggs.’ b. I-ate-it/them it-is/they-are-its-egg/eggs it-is-a-turkey-hen (Andrews, [1975] 2003, 139)1 Either Launey’s (1994) or Andrews’ ([1975] 2003) theory would end up characterizing the grammar of Classical Nahuatl as a fundamentally different one from those of 1 Transcription and glosses for (5a) are modified by me, and the word-to-word translation in (5b) is quoted directly from Andrews ([1975] 2003, 139). 4 1 INTRODUCTION more familiar languages such as English and Spanish. Andrews’ ([1975] 2003) model is especially fatal in that it would make all the traditional notions adopted in other descriptive works and textbooks a mere approximation. Thus, the status of R-marking is a fundamental problem for the description of Classical Nahuatl on which depends its whole grammar. In addition to these language-internal theoretical issues, both Launey’s (1994) and Andrews’ ([1975] 2003) interpretations lead to more serious formal and typological questions, as discussed in Section 3.2 further below. For example, the analysis which posits a predicative structure within every noun would be a critical counterargument to a theory like Baker (2003), which characterizes nouns as identifying expressions. Furthermore, it is a problem for the typological study of polysynthetic languages, since it might result in a counterintuitive conclusion that the sentence structures in Classical Nahuatl are radically different from those of many other polysynthetic languages such as Iroquoian and Caddoan, or from those of other Nahuan languages such as Michoacán Nahual and Pipil, which have been argued to have many grammatical features in common with Classical Nahuatl but lack the R-marking pattern discussed here. Lastly, as far as I have noticed, obligatory R-marking on both predicate and nonpredicate nouns seems to be a rare feature cross-linguistically. Although partially similar phenomena are observed in a few other groups of languages, I have not encountered a language outside of the Nahuan group which is proved to have obligatory R-marking in all the environments discussed in this thesis. The study of R-marking is therefore important both language-internally and crosslinguistically, and has implications for many formal, functional, and typological issues. 5 1 INTRODUCTION 1.3 On the language, data, and notation 1.3.1 Language Nahuatl, also known as Aztec or mexicano, is a group of closely related languages which belong to the Nahuan subgroup of the Uto-Aztecan family. It has many linguistic features in common with other related and unrelated languages of the Mesoamerican linguistic area, and is under the continuous influence of Spanish. The number of speakers today is reported to be approximately 1.5 million.2 The delimitation of the term Classical Nahuatl is not clear; it roughly refers to a group of old dialects of Nahuatl which were spoken in the Valley of Mexico during the sixteenth and seventeenth centuries. Although what is uniformly referred to as “Classical Nahuatl” actually consists of many slightly different variations, they have most of the fundamental phonological, grammatical, and lexical features in common and have often treated together as a single language. Dialectologically, it probably belongs to what is called the Central Area (área central). The absence of native speakers notwithstanding, Classical Nahuatl has a large amount of Latin-script records, often produced by native speakers, and a long tradition of missionary grammar, which altogether make it the best documented language in the Uto-Aztecan family. Classical Nahuatl is often cited as a typical example of so-called polysynthetic languages. It is undoubtedly “polysynthetic” under the traditional criteria (Fortescue, 1994; Mattissen, 2006); it shows almost exclusively head-marking morphosyntax, highly agglutinative template morphology, obligatory person marking for both subjects and objects, and fairly productive noun incorporation. In addition, Classical Nahuatl has been argued to be “polysynthetic” in Baker’s (1996) formal sense, since it appears to require that every lexical head have the morphological marking for its arguments. 2 1999–2000 census (INEGI, 2005). 6 1 INTRODUCTION The language may also be classified as a nonconfigurational language; in addition to the relatively free order between syntactic constituents, a constituent itself is sometimes broken up and is intervened by other words.3 Classical Nahuatl is an accusative–absolutive language; although it has no overt case marking on nouns or DP/NPs, both transitive and intransitive verbs take the same set of subject person prefixes. There is no distinction between inclusive and exclusive within the first-person plural. The distinction of number is drawn between animate plural and all the others; inanimate nouns are grammatically treated as singular regardless of the physical number or countability of the entities which they denote. 1.3.2 Data and sources The data for this thesis come from the sixteenth- and seventeenth-century colonial texts. Some other contemporary grammatical works such as Launey (1994) and Andrews ([1975] 2003), which are often cited as the data of Classical Nahuatl in non-Nahuatlist linguistic literature, appear to contain several examples composed by the authors. On the contrary, this thesis attempts to exclude any examples which are not attested in Classical texts. The primary sources for this thesis are as follows: (i) Florentine Codex (FC) Books I–XII: a manuscript of Historia general de las cosas de Nueva España, a mid/late sixteenth-century Nahuatl ethnography narrated by native informants and compiled by Fray Bernardino de Sahagún with help of his indigenous students.4 The number which follows the book 3 The classification of Classical Nahuatl as a nonconfigurational language is in fact controversial. Launey (1994) argues that Classical Nahuatl has a basic word order and observes that there are some word-orderdependent phenomena. The issue of (non)configurationality is also discussed as to its closely related modern Nahuatl languages (MacSwan, 1998; Pharao Hansen, 2010). 4 The research method used by Sahagún and the nativeness of the texts in Florentine Codex are overviewed 7 1 INTRODUCTION number is the page number in the reproduced version by Anderson and Dibble (Sahagún, 1950–1982). (ii) Primeros memoriales (Prim.): a mid sixteenth-century ethnographic manuscript composed by Sahagún. The number which follows is the page number in the reproduced version by Sullivan (Sahagún, 1997). (iii) Psalmodia christiana (Psalm.): a mid/late sixteenth-century Nahuatl translation of Psalmody edited by Sahagún. The number which follows is the page number in the reproduced version by Anderson (Sahagún, [1583] 1993). (iv) Codex Chimalpopoca (Chim.): a mid sixteenth-century codex by an anonymous author, the major part of which is Anales de Cuauhtitlán. The numbers which follow are the original side/line numbers provided in the reproduced version by Bierhorst (Bierhorst, 1992a,b). (v) Crónica mexicáyotl (Crón.): a late sixteenth-century chronicle by an indigenous historian,5 which is originally entitled Historia o crónica mexicana. The number which follows is the page number in the newest reproduced version by Anderson and Schloeder (Chimalpahin, 1997, vol. 1, 26–177). (vi) Plática by Olmos (Plát.): a mid sixteenth-century huēhuehtlahtōlli (instruction speech by the elders) collected by Fray Andrés de Olmos. The number which follows is the page number in the reproduction by Siméon (Olmos, [1547] 1875, 231–264). (vii) Bancroft Dialogues (BD): probably a late sixteenth-century collection of polite conversations. The number which follows is the page number in the reproduced version by Karttunen and Lockhart (Karttunen and Lockhart, 1987). in López Austin (1974). 5 The authorship of Crónica mexicáyotl has often been credited to Alvarado de Tezozómoc, but the editors of the version on which this thesis is based attribute it to another native historian Chimalpahin. 8 1 INTRODUCTION (viii) Cantares mexicanos (Cant.): a late sixteenth-century collection of Nahuatl songs and poems. The numbers which follow are the original folio/line numbers provided in the reproduced version by Bierhorst (Bierhorst, 1985a,b). These texts are partially the same as those which Launey (1994, 8–9) selected as his corpora, but the relative marginality of non-zero (i.e. first- or second-person) R-marked nouns required my data to contain some other sources which are not really ideal for the study of colloquial Classical Nahuatl. Firstly, it may be problematic to include verse sources such as Psalmodia christiana and Cantares mexicanos. Of course, poetry and songs should be carefully distinguished from prose; not only are they subject to the metrical restrictions, they often contain the archaic features which the spoken language had no longer kept. Nevertheless, since such texts contain interesting examples of the phenomena pursued here, this thesis intentionally includes some verse examples from these sources. Examples from verse texts are marked with “[V].” It is also important to note that the sources listed above include translated texts. Examples from Christian literature such as Psalmodia christiana and the Appendix to Florentine Codex I obviously have the expressions translated from Spanish or Latin.6 When the examples are cited from such Christian sources, they are marked with “[X].” In addition, in order to cover marginal expressions which are rare in natural narratives, this thesis also relies on the examples given in the literature of missionary grammarians, especially that of Carochi (1645).7 Such examples are marked with “[G].” 6 It should be mentioned that the texts compiled by Sahagún, such as Psalmodia christiana, Primeros memoriales, and Florentine Codex, were edited and corrected with help of his native students, and are said to be relatively free from unnatural Nahuatl expressions. 7 Its English translation by Lockhart (Carochi, [1645] 2001) is also consulted. 9 1 INTRODUCTION 1.3.3 Transcription, glosses, and notation In this thesis, Classical Nahuatl texts are given in near-Americanist transcription. The symbols ’, /c, č, λ, š, and y stand for /P/, /ts/, /tS/, /tl /, /S/, and /j/ respectively, and other symbols are used in the same way as IPA (although the transcription is a phonological, but not phonetic, one). The glossing convention in principle follows the Leipzig Glossing Rules.8 Borrowed items written in Spanish spelling are cited with modern Spanish orthography and in capital letters (e.g. ti-SACERDOTE ‘youSG priest’). Multi-word proper names of Spanish origin are treated as single words, since they behave as single stems in R-marking (e.g. ti-SANTA-MARÍA ‘youSG Saint Mary’). Examples from other languages (including non-Classical Nahuan languages) are cited in the original transcription employed in their sources.9 Morpheme segmentation, glosses, and translations are sometimes given or modified by me. Accordingly, all errors and misinterpretations are mine. For the notational simplicity, this thesis gives simplified glosses with irrelevant grammatical items and features omitted. Verbs are in the indicative mood and in the present tense unless otherwise noted. The number feature is consistently reflected in the glosses of argument prefixes. Plural suffixes are intentionally left unsegmented as if they were parts of the elements which precede them (i.e. stems), since the number feature is already reflected in the glosses of the subject prefixes or R-prefixes. Absolutive/possessive state suffixes10 are simply ignored for the sake of convenience. The distribution of these suf8 http://www.eva.mpg.de/lingua/pdf/LGR08.02.05.pdf. 9 This results in a lack of uniformity among the examples of the same language but cited from different sources. 10 Most unincorporated nouns in Classical Nahuatl take particular set of suffixes which change their forms according to whether the stem takes a possessor prefix (possessive state) or not (absolutive state). The state feature forms portmanteau suffixes with the number feature (Andrews, [1975] 2003). 10 1 INTRODUCTION fixes are totally predictable from the information given by other glosses. The tense/mood is indicated in the glosses of stems when a tense/mood marker cannot be segmented morphologically. Accordingly, the glosses given in this thesis are not always isomorphic. For example, the verb ši-k-kaki ‘(YouPL ) listen to it!’ is glossed as “2SGS-3SGO-hear.OPT” in this thesis, but these glosses do not always reflect the real functions of the individual morphemes. The subject prefix š(i)- (2SGS-) is really a number-neutral second-person optative subject prefix (therefore it should rigorously be glossed as “2S.OPT-”), and accordingly says nothing about the number of the subject. Similarly, the form of the mood-inflected stem -kaki (hear.OPT) itself does not serve to identify the optative mood, since this class of verbs coincidentally has the identical phonological forms for the present indicative singular and the nonpast optative singular. Therefore, strictly speaking, ši-k-kaki should be glossed as “2S.OPT-3SGO-hear.SGS.” However, this isomorphic system would make the glosses less readable, and would not serve the morphosyntactic understanding of this verb. For another example, the noun ča:ne’ke’ ‘owners of houses, habitants’ is inaccurately segmented as Ø-ča:n-e’ke’ (3PLR-home-PD) in this thesis. Here, the null R-prefix Ø(3PLR-) does not really indicate that this noun is plural, since this third-person R-prefix is number-neutral. Correspondingly, -e’ke’ is really not a single morpheme, but the combination of -e’ (PD) and a plural marker -ke’. Therefore, ča:ne’ke’ should be technically glossed as “3R-home-PD-ABS . PLR.” However, this glossing would miss the simplicity of R-marking. Although the tentative glosses provided in this thesis may look quite inaccurate in the eyes of Nahuatlists, I adopt this glossing system for the sake of readability, simplicity, and consistency. Zero symbols and glosses for third-person R-prefixes are omitted when unnecessary. The combination of a causative/applicative suffix and a reflexive prefix in a verb, often 11 1 INTRODUCTION used jointly as an honorific (reverential) form is glossed morpheme by morpheme, and its honorific meaning is reflected only in the free translation. The internal structure of compounds, incorporated forms, and particle clusters are sometimes ignored. Labels for grammatical categories are based on those given by Andrews ([1975] 2003) except that the traditional term preterit (pretérito) is replaced by the cross-linguistic term past in this thesis.11 Since Nahuatl has a number distinction only for animate nouns, the terms singular and plural henceforth refer to “animate singular or inanimate” and “animate plural” respectively when used for Classical Nahuatl and some other Nahuan languages. This thesis also ignores the Classical Nahuatl quasi-article in in both notational and theoretical levels. This particle is overwhelmingly attested in Classical Nahuatl texts with no substantial change in meaning. It is usually procliticized to other words and behaves sometimes like an article, other times like a filler, and appears to introduce a subordinate clause as a complementizer in some cases. Since any gloss which specifies its function would be inevitably artificial, this thesis uniformly glosses it as “IN” and excludes it from the discussions. 1.3.4 Grammatical terms This thesis replaces the traditional notion of the “subject marking on nouns” by the term R-marking.12 The “R” here stands for “referential”; this term is adopted after the notion 11 It is in fact more precise to use the terms preterit or perfect. The use of past is simply for the sake of cross-linguistic consistency. 12 The term R-marking is used as a label of “goal marking” by some researchers; however, since there appears to be little likelihood of confusion, it does not seem to present a serious practical problem to introduce this word as a newly defined term in this thesis. 12 1 INTRODUCTION of R argument and R θ-role introduced by Williams (1981),13 although this thesis does not directly associate the R-marking with R argument. The nouns with R-marking are referred to as R-marked nouns, and the person/number indicated by R-marking is designated as Rperson/number. The prefixes used in R-marking (i.e. “subject” person markers on nouns) are termed R-prefixes. Analogously, the discourse participants specified by R-marking are called R-participants. For example, the R-participant of a noun R-marked with the first-person singular R-prefix is typically the speaker. The type of R-marking observed in Classical Nahuatl, which obligatorily appears in both predicate and non-predicate nouns, is called uniform R-marking. In the examples in this thesis, R-prefixes are temporarily distinguished from the subject person prefixes on verbs and are glossed as “1SGR-,” “2SGR-,” etc., although they are by no means distinct lexically or morphologically from verbal subject person markers. Secondly, this thesis adopts the term principal, which refers to the syntactic matrix verb/noun of a clause, from Andrews ([1975] 2003) and Launey (1994, 72). This term serves to avoid useless arguments as to, for example, whether the noun mela:wake’ ‘they are right, they who are right’ in (6) is the predicate of the sentence or not; semantically it might be analyzed as the predicate whereas it seems to be the copula katka’ that serves as the syntactic head of the sentence. In Andrews’ ([1975] 2003) sence, it is clear that katka’, not mela:wake’, is the principal of the sentence. In this thesis, this term is also used to characterize the relationship between words; the word katka’ in (6) is the principal of the word mela:wake’. 13 According to Williams (1981), R θ-role is an external θ-role of a noun which is assigned to the subject of a nominal predicate. Baker (1996, 248–249) argues that R-marking in Classical Nahuatl is the agreement with the R argument in Williams’s (1981) sense. As shall be discussed in Section 4.7.3, this thesis does not support the assumption that R-marking corresponds to the R argument. 13 1 INTRODUCTION (6) wel Ø-mela:wake’ Ø-kat-ka’ well 3PLR-right 3PLS-be-PST ‘They were very righteous.’ (FC X, 169) This thesis also employs the newly coined terms reactive and inert to describe the cross-referential status of inflectional affixes. Classical Nahuatl argument affixes (Rprefixes, verbal subject prefixes, etc.) are classified into two groups: those which can be cross-referenced with another item outside of the word and those which cannot. The former are reactive in that they can interact with other items across the word boundary; the latter are inert since they are inaccessible from any items outside of the word. Reactive and inert argument affixes have traditionally referred to as “definite” and “indefinite/nondistinct” ones respectively, but this terminology is in fact incorrect since reactive affixes can be cross-referenced with other items regardless of whether they are semantically definite/specific or indefinite/nonspecific. Lastly, in the discussion on Classical Nahuatl in this thesis, the distinction between nouns, nominal phrases (NPs), and determiner phrases (DPs) cannot be sufficiently rigid although this thesis attempts to be as rigorous as possible in this respect. The lack of overt determiners, free word order, and the obscure distinction between adjectives and nouns in this language make it almost impossible to distinguish between nouns and noun phrases purely through observation. Accordingly, it is difficult to determine whether R-marking targets a noun or a noun phrase. Although this thesis assumes that R-marking applies at the level of DP/NPs and realizes morphologically at the level of individual nouns (zero-level categories), as shall be discussed in Section 4.4, this distinction is sometimes obscured in the discussions below. 14 2 MORPHOLOGY AND SYNTAX OF R-MARKING 2 Morphology and syntax of R-marking This section presents the basic data for the R-marking phenomenon on which this thesis focuses. As a preliminary step, Section 2.1 introduces the notions of lexical categories and omnipredicativity in Classical Nahuatl. Sections 2.2 through 2.4 sketch the outline of both the morphological and syntactic aspects of R-marking. Sections 2.5 and 2.6 present a cross-linguistic survey on R-marking. 2.1 Noun–verb distinction and omnipredicativity 2.1.1 Noun–verb distinction First of all, it should be noted that Classical Nahuatl strictly distinguishes between nouns and verbs at the lexical and morphological levels, as already pointed out in various works such as Launey (1984) and Andrews ([1975] 2003). It is usually impossible for a single lexical item to serve both as a noun and as a verb although the distribution of these two categories sometimes overlaps.14 In this respect, Classical Nahuatl is in a sharp contrast with such languages as Mundari (Evans and Osada, 2005) and some languages of Pacific Northwest (Mithun, 1999, 60–67), where it has been argued that the lexical distinction between these two categories is less solid. This distinction between nouns and verbs is reflected directly in morphology. Although the “subject” person marking of these two categories employs the same set of person prefixes (see Section 2.2), they differ in almost all other aspects of inflectional morphology. That is: (i) only verbs take reactive object prefixes while only nouns take reactive possessor prefixes; (ii) only verbs inflect for tense, aspect, and modality while only nouns inflect for state (see Note 10); (iii) only verbs undergo voice alternation. 14 The contigency of these two categories are discussed in detail by Launey (1984). 15 2 MORPHOLOGY AND SYNTAX OF R-MARKING Also the processes of derivation, compounding, and incorporation are sensitive to the noun–verb distinction. The status of adjectives is marginal here. Though functionally and syntactically it seems to be appropriate to posit a separate class of “adjectives” as a subclass of nouns, they behave in almost exactly the same way as substantive nouns in R-marking. This thesis employs the term adjective as a convenient label, but the distinction between adjectives and substantive nouns is theoretically not significant.15 2.1.2 Omnipredicativity Notwithstanding the lexical and morphological distinction between nouns and verbs, these two categories show somewhat parallel syntactic behavior. On a purely observational level, both nouns and verbs can serve either as principals (predicates) or as non-principals (arguments) with no considerable change in meaning. As mentioned earlier, R-marked nouns often serve as principals of clauses and have an equational or attributive function. In such cases, R-prefixes appear to serve as true subject person markers. (7) a. kwiš an-čičime’ Q 2PLR-dog ‘Are youPL dogs?’ (BD 158) b. ka ne’wa:λ ni-TESTIGO AFF 1 SG 1SGR-witness (Chimalpahin, 2006, 221)16 ‘I am the witness.’ 15 The status of adverbs and locatives are marginal as well. However, since they never have non-zero R-marking (because manner and location cannot be equated with the speaker or the addressee), their categorization is not relevant in this thesis. 16 This example is taken from an early seventeenth-century text by Chimalpahin, the assumed author of Crónica mexicáyotl. 16 2 MORPHOLOGY AND SYNTAX OF R-MARKING Much less frequently, verbs or VPs can serve as arguments of other verbs or nouns. (8) in o:=ni-k-kwal-itki-k a. ka a’mo: Ø-no-λatki AFF NEG 3SGR-1SGP-possession IN A=1SGS-3SGO-good-carry-PST ‘What I have joyfully carried is not mine.’ b. niman ye then i:k Ø-kin-yawaloa’ (FC IX, 31) in Ø-mi’to:tia’ already thus 3PLS-3PLO-surround in 3PLS-dance ‘After that, they surround [the people who] are dancing.’ c. in i:k Ø-miki-s (FC XII, 55) in Ø-ki-kokolia IN thus 3SGS-die-FUT IN 3SGS-3SGO-hate ‘so that [the person who] hates him will die’ (Prim. 213) Launey (1991, 1994, 2003, 2004) refers to this feature as omnipredicativity, on the grounds that both nouns and verbs are “predicative” or “predicable” in principle. 2.2 Morphology of R-marking 2.2.1 An overview The following sections give an overview of the morphology of R-marking. The set of prefixes employed in R-marking is presented in Table 1. Person Singular Plural First person n(i)- t(i)- Second person t(i)- am- Third person Ø- Table 1: R-prefixes in Classical Nahuatl R-marking and verbal subject marking employ the same set of person prefixes except that the former lacks the second-person optative form š(i)-. Third-person forms are always 17 2 MORPHOLOGY AND SYNTAX OF R-MARKING zero, and the first- and second-person subject/R-prefixes are portmanteau morphemes which also indicate number. The omnipredicativity in Classical Nahuatl mentioned in Section 2.1.2 is clearly related to the fact that both nouns and verbs are marked with the same set of “subject” prefixes. Although the referential status of person affixes is itself an important theoretical question, it is not persued in this thesis. Whether they are bound-form pronominal items (Jelinek, 1984; Mithun, 1986; Alexiadou and Anagnostopoulou, 1998) or mere morphological by-products of agreement (Baker, 1996; Evans, 1999) is temporarily not relevant in the arguments below.17 Therefore, although it is more common to refer to such boundform person/number markers as pronominal affixes in the tradition of Native American linguistics, they are referred to by the neutral term person affixes or person markers in the following discussions. 2.2.2 Subject person marking on verbs It will be useful to overview the pattern of subject person marking on verbs first. Following the classification by Mattissen (2006), the morphology of Classical Nahuatl can be classified as “a mixed organization of scope-ordered and templatic systems.” Specifically, the stem-internal composition (derivation, voice alternation, etc.) employs a scope-ordered strategy and the stem-external inflection (person marking, tense inflection, etc.) employs a templatic strategy. Both verbal and nominal roots can be extended by cyclic derivation, incorporation/compounding, and voice alternation (only for verbs) to form stems; once a stem is formed, it is surrounded by a limited number of morphological slots into which fit particular types of affixes. A stem has its own argument structure which is saturated by person affixes. The stem-external inflection is predominantly prefixing 17 This issue is overviewed in Corbett (2001); Ackema et al. (2006); etc. 18 2 MORPHOLOGY AND SYNTAX OF R-MARKING whereas the stem-internal derivation and voice alternation mainly employ suffixes. Subject person marking and R-marking, therefore, constitute a part of the stem-external template morphology. A simplified template of verbal inflection is given in Figure 1.18 Slot I II III IV V VI VII Function S O Dir. REFL UHO UHO REFL Example of affixes Reactive/inert n(i)- mic/- Reactive on- n(o)- te:- – I′ S -number STEM λa- TNS ne- -Ø Inert Reactive Figure 1: Verbal inflection template of Classical Nahuatl As illustrated here, subject prefixes occupy the leftmost slot in the template. The rightmost suffix (Slot I′ ) changes its form in concordance with the number of the subject prefix (Slot I). The set of subject person prefixes (Slot I) is presented in Table 2. Person First person Second person Non-optative Optative Third person Singular Plural n(i)- t(i)- t(i)- amš(i)Ø- Table 2: Verbal subject prefixes in Classical Nahuatl The epenthetic vowel i appears unless the subject prefix is followed by a vowel. The nasal m in am- becomes homorganic to the following consonant.19 In addition, at least in the dialect described in Carochi (1645), this m obligatorily undergoes a total assimilation to the following sibilant.20 Although the Slot I prefix have the same form for the first18 “Dir.” and “TNS” in Figure 1 stand for “directional” and “TENSE” respectively. 19 e.g. am- + -ki-tta- ‘see it’ → ankitta’ [aNkittaP] (Cron. 98). 20 e.g. am- + -sekmiki- ‘suffer from cold’ → assekmiki’ ‘youPL feel cold’ (Carochi, 1645, f. 115r). 19 2 MORPHOLOGY AND SYNTAX OF R-MARKING person plural and the second-person singular (non-optative), these two forms are rarely confused since the number suffix in Slot I′ have different forms according to the number of the subject. Similarly, in the second-person optative and the third person, the number feature of the subject is recovered by the number suffix. 2.2.3 R-marking on nouns Classical Nahuatl nouns inflect for the R-person/number and the person/number of the possessor. The set of R-prefixes is already shown in Table 1 above. The inflection template of nouns is presented in Figure 2. Slot Type of affixes I II R-prefix Possessor I′ State/number STEM Example of affixes Reactive/inert n(i)- m(o)-, te:- -w Reactive Reactive or inert Reactive Figure 2: Nominal inflection template of Classical Nahuatl R-marking uses exactly the same set of prefixes as those employed in verbal subject marking (Table 2). Their assimilation patterns and the distribution of epenthetic vowel are also the same as those of the verbal subject prefixes displayed above. The only difference is that the optative second-person prefix š(i)- never appears in R-marking; nouns take the non-optative second-person prefixes t(i)- and am- even when they are cross-referenced with the verbal prefix š(i)-. Adjectives also follow this pattern since they are essentially nouns in Classical Nahuatl.21 (9) a. in ma: t-oko:c/oλ ši-ye-ni IN IRR 2SGR-sap 2SGS-be-OPT. PST ‘as if youSG were sap’ 21 [V] (Cant. 16r:20) For the status of adjectives, see Section 2.1.1. 20 2 MORPHOLOGY AND SYNTAX OF R-MARKING The parallelism between R-marking and verbal subject marking is related to another important feature of R-marking. That is, R-marking occurs word-internally or morphologically. An R-prefix never crosses a word boundary and cliticizes to another word. This word-internal nature of R-marking is mentioned in Section 2.4.3 below. The examples of R-marking on unpossessed nouns are presented in Table 3. The stems λa:ka- ‘person’ and a’λa:ka- ‘bad person, inhuman person’ take R-prefixes (n(i)-, t(i)-, Ø-, ...) and state/number suffixes (-λ and -’) according to the R-person/number.22 1SGR ni-λa:ka-λ ‘I (am) a person’ n-a’λa:ka-λ ‘I (am) a bad person’ 2SGR ti-λa:ka-λ ‘youSG (are) a person’ t-a’λa:ka-λ ‘youSG (are) a bad person’ 3SGR Ø-λa:ka-λ ‘he/she (is) a person’ Ø-a’λa:ka-λ ‘he/she (is) a bad person’ 1PLR ti-λa:ka-’ ‘we (are) people’ 2PLR an-λa:ka-’ ‘youPL (are) people’ am-a’λa:ka-’ ‘youPL (are) bad people’ 3PLR Ø-λa:ka-’ ‘they (are) people’ Ø-a’λa:ka-’ ‘they (are) bad people’ t-a’λa:ka-’ ‘we (are) bad people’ Table 3: Example of R-marking in Classical Nahuatl It should be noted that the R-prefixes are zero for the third person. Virtually all dictionaries show the zero-marked third-person forms as the dictionary forms for nouns. Andrews ([1975] 2003) analyzes the combination of a Slot I prefix and a Slot I′ suffix as a dyad which jointly serves as a subject marker of verbal/nominal nuclear clauses (e.g. n(i)-...-λ (1SGR) in ni-λa:ka-λ ‘I (am) a person’). This appears reasonable from the morphosyntactic point of view, since these two groups of affixes almost always have the same value of number feature. However, for both practical and descriptive reasons, this thesis does not treat them as a dyad. Practically, positing a discontinuous combination of co-operating affixes makes the glosses and descriptions unnecessarily complicated.23 22 The examples of R-marked forms are taken from Lockhart (2001, 2–3). 23 For instance, it would require for the word ti-λa:ka-’ ‘we (are) people’ to be glossed as “1PLR-person- 21 2 MORPHOLOGY AND SYNTAX OF R-MARKING Descriptively, a Slot I′ suffix sometimes appear between the stem and the honorific suffix -c/in as in (10a), while the reverse order is also possible. Moreover, as in (10b), it appears even when the noun is in a vocative form, which is assumed not to be R-marked. Therefore, also in the descriptive point of view, it seems appropriate to deal with Slot I prefixes and Slot I′ suffixes separately. (10) a. Ø-i:-pil-wa:n-c/i-c/in 3PLR-3SGP-child-POS-RDP-HON ‘his children (hon.)’ (Crón. 24) b. no-pil-wa:n-é 1SGP-child-POS-VOC ‘My sons!’ (FC VI, 87) 2.2.4 Person prefixes and pronouns Cross-linguistically, bound-form person markers tend to have similar forms to independent pronouns. This is also true to Classical Nahuatl. For instance, the first-person singular subject prefix n(i)-, object prefix ne: č-, and reflexive prefix n(o)- have the initial consonant in common with the first-person singular free-form pronoun ne’ or ne’wa:λ ‘I.’ The issue of R-marking in Classical Nahuatl, however, raises another question: are independent pronouns R-marked? More specifically, if independent pronouns have phonologically parallel forms to R-prefixes, can they be analyzed as the R-marked forms of a particular person-neutral pronominal stem? Andrews ([1975] 2003, 126–128) analyzes the independent pronouns in Classical Nahuatl as R-marked forms of a person-neutral nominal stem (y)e’(wa:)- ‘entity.’ According ABS . PLR ,” where t(i)-...-’ corresponds to 1PLR. It is troublesome to speak of a discontinuous “R-prefix- suffix-combination,” whose second part forming a portmanteau suffix with the state feature. 22 2 MORPHOLOGY AND SYNTAX OF R-MARKING to him, such forms as ne’wa:λ ‘I’ and ye’wa:λ ‘he/she/it’ can be translated as ‘I am an entity’ and ‘he/she/it is an entity.’ In the third-person form, the stem-initial e is realized as ye. His interpretation is illustrated in Table 4. R-prefix Independent pronoun Andrews’ ([1975] 2003) analysis 1SG n(i)- ne’(wa:λ) n-e’(wa:λ) ‘I am an entity’ 2SG t(i)- te’(wa:λ) t-e’(wa:λ) ‘YouSG are an entity’ 3SG Ø- ye’(wa:λ) Ø-(y)e’(wa:λ) ‘He/she/it is an entity’ 1SG t(i)- te’wa:n(tin) t-e’wa:n(tin) ‘We are entities’ 2SG am- ame’wa:n(tin) am-e’wa:n(tin) ‘YouPL are entities’ 3SG Ø- ye’wa:n(tin) Ø-(y)e’wa:n(tin) ‘They are entities’ Table 4: R-prefixes and independent pronouns in Classical Nahuatl So far, there is neither positive nor negative synchronic evidence concerning his analysis.24 Nonetheless, two points should be noted here. First, although it is very common in Classial Nahuatl for a word-initial e to break into ye, it is not regular enough to account for the consistent presence of y in the third-person pronouns. Indepent pronouns ye’(wa:λ) ‘he/she/it’ and ye’wa:n(tin) ‘they’ rarely occur without the initial y- whereas many nouns such as esλi ‘blood,’ eλ ‘bean,’ and e:yi ‘three’ usually appear without y-. Second, the function of these independent pronouns does not differ drastically from those found in many other pro-drop languages such as Spanish. Just like Spanish yo ‘I’ and él/ella ‘he/she,’ they bear pronominal references rather than establishing a equational or attributive relationship between R-prefixes and ‘entity.’ The diachronic aspect of this issue is also unclear. The only point which can be made is that the indepent pronouns can be reconstructed as pronouns (*naha/*n h ‘I,’ *taha ‘youSG ,’ *yaha ‘he/she/it,’ etc.) in e 24 Proto-Nahuan (Dakin, 1982, 177). 23 2 MORPHOLOGY AND SYNTAX OF R-MARKING Accordingly, there is no considerable advantage with the radical hypothesis which interprets Classical Nahuatl independent pronouns as ‘I am an entity,’ ‘youSG are an entity,’ etc. For these reasons, this thesis glosses independent pronouns simply as “1SG,” “2SG,” “3SG,” etc. without any further word-internal analysis. 2.3 Cross-reference Apparently, R-prefixes play a key role in Classical Nahuatl syntax, since the person/number features of nouns appear to be important in the cross-reference between items in most cases. In Classical Nahuatl, with a few exceptions, cross-referenced nouns have the R-prefixes of the same person/number as the person prefix with which they are crossreferenced. As do many other polysynthetic languages, Classical Nahuatl has obligatory subject/object agreement on verbs. The overall characteristics of Classical Nahuatl syntax conform to the type characterized in Jelinek (1984) and Baker (1996). In traditional terms, it can be described as follows: (i) both nouns and verbs obligatorily indicate their core arguments (subjects and objects for verbs, “subjects” and possessors for nouns) and can generally stand alone as sentence-words; (ii) verbs/nouns can be extended by means of cross-referencing; and (iii) the reactive person markers on nouns and verbs are interpreted as indicating person-pronominal (therefore definite) arguments when not cross-referenced with full DP/NPs. Verbal subject/object markers never drop even when there are overt DP/NP arguments.25 In (11), for example, verbs are obligatorily and redundantly person-marked 25 As pointed out by Baker (1996), Classisal Nahuatl is in a strong contrast in this respect with many other anaphoric-agreement languages such as Chicheŵa (Bresnan and Mchombo, 1986), where object class markers can drop when an overt object fits in a particular syntactic position. 24 2 MORPHOLOGY AND SYNTAX OF R-MARKING despite that there are overt pronominal and nominal DP/NP arguments. (11) a. aw in ne’ te’wa:λ močipa ti-ne:č-kokolia and IN 1SG 2SG always 2SGS-1SGO-hate ‘... but youSG always hate me.’ b. niman onka:n Ø-k-a:ltia then [G] (Paredes, 1759a, 34) in ti:siλ in pilto:nλi there 3SGS-3SGO-bathe IN midwife IN baby ‘Then the midwife bathes the baby there.’ (FC VI, 201) Many Classical Nahuatl researchers such as Launey (1994) and Andrews ([1975] 2003) consider this as an cross-referential extension of the principal sentence-word. That is, the main verbs tine: čkokolia (2SGS-1SGO-hate) in (11a) and ka:ltia (3SGS-3SGO-bathe) in (11b) both constitute minimal predicational structures by themselves (‘youSG hate me’ and ‘she bathes him’), and are extended in turn by a subordinating process which is termed actantialisation and supplementation respectively by Launey (1994) and Andrews ([1975] 2003).26 Here, reactive person prefixes on verbs (and nouns as well) are cross-referenced with other nouns with the same values of person/number features. The independence of sentence-words and its extension are exemplified in (12). (12) a. λa: ši-k-kaki IRR (kwa:wλekec/ki-é a’noso kwa:wko:a:λ-é) 2PLS-3SGO-hear.OPT PN-VOC or PN - VOC ‘Listen to it (oh Cuauhtlequetzqui, or Cuauhcoatl).’ b. in ti-no-pil-c/in in ti-no-te:lpo:č (Crón, 108) λa: ši-k-kaki IN 2SGR-1SGP-child-HON IN 2SGR-1SGP-youth IRR 2SGS-3SGO-hear.OPT 26 The major difference between Launey (1994) and Andrews ([1975] 2003) is that Andrews ([1975] 2003) characterizes supplementation as the relationship between the person marker and the supplementary word (nuclear clause), while Launey (1994) describes the cross-referencing as that between the person marker of the principal and the “subject” person marker of another element. The reason for this analysis is stated in Launey (2003, 2004). 25 2 MORPHOLOGY AND SYNTAX OF R-MARKING in Ø-λa’to:lli IN 3SGR-word ‘YouSG , my child, my youth, listen to these words.’ (FC VI, 113) In (12a), the verb šikkaki ‘(YouPL ) listen to it!’ stands as a clause with no full-DP/NP argument. The person prefixes š(i)- (2SGS.OPT)27 and k(i)- (3SGO) themselves indicate the pronominal arguments of the verb. ši2SGS- k3SGO- -kaki hear Figure 3: Structure of the sentence-word in (12a) In (12b), on the other hand, both the subject and the object are “extended” by full DP/NPs. The nouns tinopilc/in ‘youSG who are my child’ and tinote:lpo: č ‘youSG who are my youth,’ both R-marked with the second-person singular R-prefix t(i)-, are crossreferenced with the reactive person prefix with the same person/number (i.e. š(i)-) on the verb šikkaki ‘(YouPL ) listen to it!’ Similarly, another noun Ø-λa’to:lli ‘word,’ which is R-marked as third-person singular, is cross-referenced with the object prefix k(i)-. Cross-referencing ti-nopil¢in 2SGR- my child 2SG ši2SGS- Cross-referencing k3SGOPrincipal -kaki hear Ø3SGR- -λa'toːlli word 3SG Figure 4: Cross-referencing structure of (12b) (simplified) Here, the verb šikkaki serves as the principal of the clause. These cross-referencing relationships appear to be justified by the identity of person/number features; cross27 Strictly speaking, š(i)- is a number-neutral second-person prefix and the singular value of the subject number is recovered by the absence of the plural suffix which follows the verb stem. 26 2 MORPHOLOGY AND SYNTAX OF R-MARKING referenced nouns in (12b) have the R-prefixes of the same person/number as that of the person markers on the principal. According to this analysis, the subject/object agreement in (11a) and (11b) should be reanalyzed as Figure 5. Cross-referencing ne’ 1SG te’waːλ 2SG ti2SGS- neːč1SGO- -kokolia hate Principal Cross-referencing Cross-referencing Ø3SGS- k3SGO- -aːltia bathe Ø3SGR- -tiːsiλ midwife Ø3SGR- 3SG Principal -piltoːnλi baby 3SG Cross-referencing Figure 5: Reanalysis of (11) according to the extension model (simplified) This model interprets the overt DP/NP arguments in (11) as cross-referenced nouns which extend the minimal structures of the principal verbs. Again, the person/number feature appears critical here. As illustrated in Figure 5, the argument nouns have the same person/number as that of the person prefix with which they are cross-referenced. Lastly, it should be noted that the person affixes are interpreted differently according to whether they are cross-referenced or not. The cross-referencing with person markers is not limited to definite or specific expressions. Reactive third-person prefixes on verbs (and nouns as well) can also be cross-referenced with indefinite or nonspecific objects.28 (13) a. ma: kana’ IRR iλa’ Ø-ki-či:w somewhere something 3SGS-3SGO-make.ADM ‘... lest he should do something [evil] somewhere’ 28 Evans (1999) discusses the cross-linguistic implications of this feature. 27 (FC VI, 127) 2 MORPHOLOGY AND SYNTAX OF R-MARKING b. a:k in ti-k-čiya a:k in ti-k-λa:kamati who IN 2SGS-3SGO-await who IN 2SGS-3SGO-obey ‘Who are youSG waiting for? Who do youSG obey?’ (FC VI, 80) c. aya:k Ø-ki-ma:wka:itta’ ... aya:k Ø-k-i:šitta’ ... nobody 3PLS-3SGO-respect nobody 3PLS-3SGO-show.respect ‘They respect nobody ..., show respect to nobody ...’ (FC IV, 50) In contrast, they behave like personal pronouns when not cross-referenced; they are never interpreted as indefinite, nonspecific, or impersonal in such cases. (14) a. kil in Ø-okλi Ø-ki-či:w it.is.said 3SGS-3SGO-make.PST IN 3SGR-pulque ‘It is said that he made pulque.’ b. Ø-ki-mo-teo:tia:-ya’ [X] (FC I, 74) in Ø-ti:-tisi’ in 3PLS-3SGO-REFL-worship-IMPF IN 3PLR-RDP-physician IN Ø-te:ikmi:nke’ ... 3PLR-phlebotomist ‘[As for this goddess,] physicians, phlebotomists, ... worshipped her.’ (FC I, 15) 2.4 Distribution and usage of R-marked nouns This section provides the basic data concerning the syntactic distribution of R-marking. Not only do they appear in the principal positions and form nominal clauses, they are also obligatory in many non-principal positions. 28 2 MORPHOLOGY AND SYNTAX OF R-MARKING 2.4.1 R-marked nouns as principals As cited in many typological works such as Stassen (1997), R-marked nouns in Classical Nahuatl themselves serve as nominal predicates. In such cases, they are interpreted as tense-neutral.29 In Classical Nahuatl, the overt copula ka’30 does not appear in tense-neutral equational or appositional clauses while it realizes phonologically in marked tenses/moods. This is reminiscent of the copular sentences in such languages as Russian, where overt copulae appear only in non-present tenses. (15) shows the typical usages of principal R-marked nouns. As a basic feature of Classical Nahuatl principals, R-marked principal nouns are not accompanied by a quasiarticle in in matrix clauses (contrainte de non-définition; Launey, 1994, 59). (15) a. ti-λaškalte:ka 1PLR-EN ‘We are Tlaxcalteca.’ b. ka t-i:-ko:a:w (FC XII, 28) ka t-i:-to:to:w AFF 2 SGR -3 SGP -snake AFF 2 SGR -3 SGP -bird ‘YouSG are his snake, his bird’ (FC VI, 171) c. ka te’wa:λ ti-λa’toa:ni AFF 2 SG 2SGR-ruler ‘YouSG are the ruler.’ (Chim. 33:36–37) 29 The tense-neutral nature of principal R-marked nouns is discussed in Section 4.7.2. 30 The copular verb ka’/ye is a suppletive verb whose root alternates between -kat- (e.g. indicative present/past) and -ye- (e.g. indicative future and optative). This thesis uses ka’ as the dictionary form of this verb. 29 2 MORPHOLOGY AND SYNTAX OF R-MARKING d. ka ye AFF ni-we:we’ already 1SGR-old.man ‘I am already old.’ e. aw ye’i:ka (FC III, 33) in a:ška:n ti-me:ši’ka’ ye:se’ a’mo: ti-me:ši’ka’ ka and therefore IN now 1PLR-EN but NEG 1PLR-EN AFF ti-me:ši’tin 1PLR-EN ‘So we, who are now Mexica, are in fact not Mexica, but Mexitin.’ (Chim. 79:1) f. kwiš ti-no-na:mik Q kwiš ti-n-okičwi 2SGR-1SGP-spouse Q 2SGR-1SGP-man ‘Are youSG my spouse? Are youSG my husband?’ (Prim. 297) They may constitute a simple affirmative expressions as in (15a) and (15b), while they can be accompanied by independent arguments as in (15c). Moreover, they may be modified by adverbs or particles as in (15d), and negated or interrogativized as in (15e) and (15f). As these examples show, principal R-marked nouns are very frequently preceded by the affirmative particle ka, which is distinct from the copular verb ka’.31 Although third-person nouns are phonologically not overtly R-marked, they show almost exactly the same behavior as non-third-person, overtly R-marked nouns. That is, nouns are never interpreted as coreferent with the speaker or the addressee as long as they do not have overt, non-third-person R-marking. Therefore, many researchers of Classical Nahuatl are in agreement that it is appropreate to posit a third-person R-prefix 31 Although the copular verb ka’ and the affirmative particle ka are both spelled as ⟨ca⟩ in most colonial texts, they are clearly distinguished linguistically. Unlike ka’, the affirmative marker ka is never inflected (i.e. they do not take person or tense markers) and appears in both nominal and verbal clauses. 30 2 MORPHOLOGY AND SYNTAX OF R-MARKING which is phonologically null, namely Ø-. As seen in (16), they can serve as the principal of a clause with no overt subject noun or pronoun. (16) a. kwiš Ø-to:to:λ kwiš Ø-paλa:ni-s Q 3SGR-bird Q 3SGS-fly-FUT ‘Is he a bird? Does he fly?’ b. ka Ø-i:n-λatki (FC XII, 32) in Ø-to-na:nwa:n AFF 3 SGR -3 PLP -possession in Ø-to-ta’wa:n ... IN 3PLR-1PLP-mother IN 3PLR-1PLP-father ‘It is the possession of our mothers and fathers ...’ (FC IX, 31) The third-person principal R-marked nouns also non-obligatorily take overt subject DP/NPs as in (17). In such cases, they often look like copulaless sentences in other, more familiar languages like Russian and Arabic. The particle in which precedes the second nouns in (17a) and (17b) indicates that it is the first nouns, but not the second nouns, that are the principals of these sentences. (17) a. ka Ø-λa’toa:ni in Ø-no-ta’-c/in AFF 3 SGR -ruler IN 3SGR-1SGP-father-HON (Chim. 23:44)32 ‘My father is the ruler.’ b. senka’ Ø-kwalli very in Ø-pa’λi 3SGR-good IN 3SGR-medicine (Quetzalcoatl y Titlacahuan, ¶ 54)33 ‘This medicine is very good.’ 32 Bierhorst (1992b, 66) translates this sentence as ‘The ruler ... is my father.’ However, it is clear from the use of a quasi-article in that the second noun is not the principal. 33 Garibay K. ([1940] 1961, 143). This example is taken from a mid sixteenth-century anonymous manuscript. 31 2 MORPHOLOGY AND SYNTAX OF R-MARKING 2.4.2 R-marked nouns as non-principals What is conspicuous about Classical Nahuatl R-marking is that it is obligatory in nonprincipal nouns as well. The most identifiable examples appear in the non-present/non-indicative copular sentences. While the copular verb ka’ does not appear in the unmarked tense/mood, it overtly appears in marked tenses/moods. In such cases, the complement nouns of the copular verb should be overtly R-marked when the subject of the copula is in the first or second person. Examples include: (18) a. λa:l-ti-kpak ti-ye-s t-okičλi ground-L-above 2SGR-man 2SGS-be-FUT ‘YouSG will be a man in this world.’ b. t-i:n-λa:kawa:n ti-kat-ka’ (FC VI, 74) in Ø-λa:-λa:katecolo’ 1PLR-3PLP-slave 1PLS-be-PST IN 3PLR-RDP-devil ‘We were slaves of devils.’ [V][X] (Psalm. 242) c. in ma: t-oko:c/oλ ši-ye-ni IN IRR 2SGR-sap 2SGS-be-OPT. PST ‘as if youSG were sap’ [V] (Cant. 16r:20) Similarly, the complements of many verbs such as či:w(a) ‘make’ and kwep(a) ‘turn’ are also obligatorily R-marked. The examples with moči:w(a) ‘occur, become’ (the reflexive form of či:w(a) ‘make’) are especially abundant in the text. (19) a. ni-te:lpo:čλi ni-no-či:wa 1SGR-youth 1SGS-REFL-make ‘I become a youth.’ b. kwiš ti-siwa:-pi:lli Q (FC I, 82) ti-mo-kwepa-s-neki 2SGR-woman-noble 2SGS-REFL-turn-FUT-want 32 2 MORPHOLOGY AND SYNTAX OF R-MARKING ‘Are youSG trying to turn into a noble woman?’ c. ... in i:k a’mo: ti-šolopi’λi t-i:štotomak ... i:pan IN thus NEG 2SGR-fool 2SGR-idiot (Prim. 298) ti-mač-o:-s 3SGP-on 2SGS-know-NA-FUT ‘... lest youSG should be considered as a fool, an idiot ...’ d. ka ti-ča:lčiwiλ ka ti-ma:ki:sλi ka ti-teo:šiwiλ AFF 2 SGR -jade AFF 2 SGR -bracelet AFF 2 SGR -turquoise ti-pi:c/a-lo:-k (FC VI, 122) in IN in ti-mamali:-wa-k 2SGS-mold-NA-PST IN 2SGS-perforate-NA-PST ‘YouSG were molded and perforated as a jade, a bracelet, a turquoise.’ (FC VI, 31) Cross-referenced non-principal nouns are also R-marked at least when they are crossreferenced with non-third-person markers. Put in other words, Classical Nahuatl has nonthird-person nonpronominal arguments. (20) a. aw ka is ti-ka’ in ti-te:-a:č in ti-yakapanλi ... and AFF here.is 2SGS-be IN 2SGR-UHP-elder IN 2SGR-firstborn ‘Here youSG are, youSG who are elder, youSG who were born first, ...’ (FC VI, 87) b. wel ši-k-kakika:n in am-i:škičtin in sema:na:wa-k an-λa:ka’ well 2PLS-3SGO-hear.OPT IN 2PLR-entire IN world-LOC ‘Listen to it well, all of youPL , people in the world.’ c. ka ni-mo-ma:se:wal o:mpa ni-wi:c/ AFF 1 SGR -2 SGP -vassal 2PLR-person [V][X] (Psalm. 56) in Ø-nonowalkatepe:λ there 1SGS-come IN 3SGR-LN i:-c/inλa:n 3SGP-bottom ‘I, yourSG servant, have come from the roots of Mount Nonohualcatepetl.’ 33 2 MORPHOLOGY AND SYNTAX OF R-MARKING (Chim. 5:43–44) While some of such examples can be interpreted as mere juxtaposition of two or more separate sentences with principal nouns, there are some examples which should clearly be analyzed as single sentences. This point will be discussed in Section 4.6.1. Similarly, when a non-third-person pronoun serves as the principal of a clause (e.g. ‘it is me,’ ‘it is youSG ’), the subject is also R-marked. (21) ka ne’wa:λ in n-amo-te:či:wka:w in ni-mote:kwso:ma AFF 1 SG IN 1SGR-2PLP-governor IN 1SGR-PN ‘It is me that is yourPL governor Moteuczoma.’ (FC XII, 31) 2.4.3 R-marking in multi-word nominal construction Another important feature of R-marking in Classical Nahuatl is that it does not cross a word boundary. When two or more nouns constitute a noun-phrase-like cluster, all of its nominal members are R-marked in principle. The most identifiable among them are adjective–noun sequences such as (22). This construction is relatively rare in purely native texts; most examples of this construction come from Christian texts or the literature of missionary grammarians. It should also be noted that such examples as (22) may well be interpreted as containing two juxtaposed principal nouns rather than an adjective–noun constituent of the type found in English or Spanish. (22) te’wa:λ ti-kwalli 2SG ti-λa’toa:ni 2SGR-good 2SGR-ruler ‘YouSG are a good ruler.’ [G] (Paredes, 1759a, 33) 34 2 MORPHOLOGY AND SYNTAX OF R-MARKING 2.4.4 Nouns without R-marking The examples presented so far illustrate that nouns in Classical Nahuatl generally have obligatory R-marking regardless of whether they are principals or non-principals. In principle, nouns should have overt R-prefixes whenever their referents include the speaker or the addressee. However, R-marking seems non-obligatory or impossible in some cases. Firstly, vocative nouns34 are never R-marked while they usually refer to the addressee. This absence of R-marking will be explained by the fact that vocatives are syntactically independent from the matrix clauses by nature. (23) a. λa:kaλ-é to-te:kw-é ... person-VOC 1PLP-lord-VOC ‘Oh master, oh our lord!’ (FC VI, 1) b. me:ši’ká’ ši-wa:l-ne’nemika:n EN . VOC 2PLS-CISL-walk.OPT ‘Mexica, march (youPL ) hither!’ (FC XII, 67) Secondly, pronominal items (personal pronouns, demonstratives, indefinite/negative pronouns, interrogative pronouns, etc.) are marginal in R-marking. For instance, the status of personal pronouns in R-marking is not clear. It is already noted in Section 2.2.4 that Andrews ([1975] 2003) analyzes them as the R-marked forms 34 Vocative forms are made whether by adding an accented vocative suffix -é (male speakers) or assigning high pitch to the last syllable of the noun (female speakers), although some texts such as Codex Chimalpopoca contain bare vocatives by male speakers. Andrews ([1975] 2003) analyzes these vocative forms as third-person R-marked nouns rather than non-R-marked. In addition, some grammarians characterize the cross-referenced second-person R-marked nouns as “vocatives” (in ti-/in am- vocatives); however, this thesis does not follow this analysis. 35 2 MORPHOLOGY AND SYNTAX OF R-MARKING of a person-neutral pronominal stem, though there is little descriptive evidence which supports this interpretation. Demonstratives are clearly not R-marked:35 (24) a. ma: ne’ =i:n IRR 1SG =this ‘Let me be this.’ (FC VI, 89) b. sa: te’wa:n =i:n aw sa: t-iwke’ only 1PL =i:n =this and only 1PLS-like =this ‘We are these, we are such as these’ (FC VI, 137) This absence of R-marking may be related to the fact that Classical Nahuatl demonstratives =i:n (proximal) and =o:n (distal) are probably enclitics, but not full noun stems. Also interrogative, indefinite, and negative pronouns36 are usually not R-marked.37 (25) a. a:k ame’wa:n who 2PL ‘Who are youPL ?’ 35 (FC XII, 13) The two instances of =i:n ‘this’ in (24) can alternatively read as ‘in this manner.’ If so, they are naturally not R-marked since adverbial expressions are never R-marked. However, R-marked forms of demonstratives are not attested so far, and Andrews ([1975] 2003, 290) too suggests that =i:n and =o:n cannot be R-marked. 36 Most indefinite and negative pronouns are morphologically related to or identical with interrogative words. For example, aka’ ‘someone’ and aya:k ‘no one’ are derived from a:k ‘who.’ Similarly, a’mo: aka’ (NEG someone) means ‘no one.’ In other cases, the same forms as interrogative words are used as indefinite pronouns. 37 Contrary to this, there are some R-marked interrogative, indefinite, and negative pronouns: am-a:ki’ke’ ‘who are youPL ?’ (Molina, 1571, 4); a’=ti-λein ‘youSG are nothing’ (Carochi, 1645, f. 16v); etc. In these forms, interrogative stems are probably incorporated into a hypothetical extistential verb i’ (Andrews, [1975] 2003, 93–94), but this issue cannot be pursued any further here. 36 2 MORPHOLOGY AND SYNTAX OF R-MARKING b. λe’ ti-ye-s what 2PLS-be-FUT ‘What will youSG be?’ (FC XII, 74) c. aw in nika:n am-on-oke’ and IN here in an-no-pilwa:n 2PLS-TRSL-lie IN 2PLR-1SGP-child maybe.not someone ame’wa:ntin an-ki-tta-ti:wi’ 2PL aka’somo: aka’ in o:=ni-k-i’to’ 2PLS-3SGO-see-PURP IN A=1SGS-3SGO-say.PST ‘Maybe none of youPL , my sons who are here, [will] live to see what I said.’ [G] (Carochi, 1645, f. 85v) Spanish loanwords, especially proper nouns, often lack R-marking. (26) a. aw in i:k no: ne’wa:λ DON-HERNANDO-DE- ... and IN thus also 1SG PN - PN - PN -... ‘And therefore also I, Don Hernando de ...’ (Crón. 65) It is very common that, when a Spanish-style personal name consists of more than one phonological word, only the first element takes an R-prefix. In other words, Spanish-style names are treated as single stems in R-marking. This fact may reflect the opacity of internal strucutres of Spanish loan items in Classical Nahuatl. (27) a. in ti-čpo:čλi in ti-SANTA-MARÍA ... IN 2SGR-young.woman IN 2SGR-PN-PN ‘youSG , a virgin Saint Mary’ b. ka ni-miki-s AFF 1 SGS -die- FUT [V][X] (Psalm. 24) in ne’wa:λ ni-PEDRO-TOZAN ... IN 1SG 1SGR-PN-PN ‘Because I, Pedro Tozan, will die...’ 38 (Rojas Rabiela et al., 2000, 273)38 This example is taken from a testament made in 1587. The name ni-PEDRO-TOZAN is originally spelled as ⟨ni Pedro Doçan⟩. 37 2 MORPHOLOGY AND SYNTAX OF R-MARKING The absence of R-marking in some environments can be properly explained by the traditional approach. Firstly, when personal names are referred metalinguistically (e.g. ‘my name is X’), it is usually not R-marked. Since a name and the person denoted by it are distinct, this can be properly accounted for by the principle of R-marking. (28) a. ka Ø-no-to:ka: ic/to:lpanšo:či AFF 3 SGR -1 SGP -name PN ‘My name is Itztolpanxochi.’ (Chim. 22:4–5) Secondly, nouns with so-called “postpositions” are never overtly R-marked. While such forms as λa:ka-ti-ka (person-L-by) ‘by human’ (Chim. 5:11, etc.) and siwa:pan (woman-LOC) ‘at (the place of) woman’ (FC IX, 16, etc.) are very common, their Rmarked counterparts *ti-λa:ka-ti-ka ‘by us humans’ and *ni-siwa:-pan ‘at (the place of) me, woman’ are not attested. This can be explained by the fact that Classical Nahuatl “postpositions” are technically nouns (relational nouns). They are incompletely grammaticalized noun stems which form compound nouns with their “complement” noun stems. Accordingly, the R-marking slot (Slot I) of the resulting nouns should be filled by an R-prefix which indicates the person/number of the compound itself (i.e. a place or relation), which can be neither in the first nor second person. Similarly, as pointed out and explained by Launey (1994, 55), such forms as *ni-me:ši’ko (1SGR-LN) ‘I am (in/at) [the city of] Mexico’ and *ni-no-ča:n (1SGR-1SGP-home) ‘I am (in/at) my house’ are not found, because a location is never coreferential with an entity which is located inside it. 2.5 R-marking in modern Nahuan languages The comparison of present-day Nahuan languages shows that some of them lack the uniform R-marking of the type observed in Classical Nahuatl. While the dialects of 38 2 MORPHOLOGY AND SYNTAX OF R-MARKING Tetelcingo, Cuetzalan, and Milpa Alta appear to have the uniform R-marking of the type found in Classical Nahuatl, it is consistently lacking in Michoacán Nahual and Pipil. 2.5.1 Nahuan languages with uniform R-marking The Nahuatl of Tetelcingo39 (Nuclear Subarea in the Central Area40 ) appears to preserve the uniform R-marking inherited from its ancestor language. Not only does it appear on nominal predicates as in (29), it is also found on non-indicative/present copulatic sentences as in (30). Unlike Classical Nahuatl, this dialect has optional overt present indicative copulae as in (30a). (29) Tetelcingo Nahuatl a. taha ok tı-pılalaktlı 2SG yet 2SGP-lad ‘YouSG are still a lad.’ b. ... que (Tuggy, 1979, 15) tejuatzi te-Cristo te-tie-Cone-tzi COMP 2 SG Deus 2SGR-PN 2SGR-HONP-child-HON God ... ‘... that youSG are Christ, the Son of God ...’ (La Liga Bíblica, [1980] 2009, John 11:27) (30) Tetelcingo Nahuatl a. taha ok tı-ka-ka tı-pılalaktlı 2SG yet 2SGS-RDP-be 2SGP-lad ‘YouSG are still a lad.’ b. ayekmo šı-ye-kO (Tuggy, 1979, 17) šı-tuntoh-te no.longer IMP-be-PL IMP-stupid-PL 39 State of Morelos, about 70 km south of Mexico City. 40 This thesis employs the dialectal classification labels used in Lastra de Suaréz (1986). 39 2 MORPHOLOGY AND SYNTAX OF R-MARKING ‘Stop (youPL ) being so stupid!’ c. ohalá ma OPT nı-ye-nı (Tuggy, 1979, 75) nı-riko EXHRT 1 SGS -good- IRR 1SGR-rich ‘Would that I were rich.’ d. nemiech-chihua-lti-s (Tuggy, 1979, 32) nen-ye-sque nen-pescadur-te de tlöca 1SGS.2PLO-make-CAUS-FUT 2PLS-be-FUT 2PLR-fisher-PL of men ‘I will make youPL fishers of men.’ (La Liga Bíblica, [1980] 2009, Matthew 4:19) The only major difference with Tetelcingo Nahuatl from Classical Nahuatl observed so far is that R-marking is non-obligatory in the non-first elements of multi-noun construction in Tetelcingo Nahuatl, as stated by Tuggy (1979, 11). (31) Tetelcingo Nahuatl a. nemehwa nen-ka-te 2PL nen-kwalı nen-tlOka 2PLS-be-PL 2PLR-good 2PLR-men ‘YouPL are good men.’ b. taha tı-kwalı (Tuggy, 1979, 11) tlOkatl 2SG 2SGR-good man ‘YouSG are a good man.’ (Tuggy, 1979, 11) Nahuatl of Cuetzalan41 (Oriental Periphery) seems to share similar features (Tomás Amaya Aquino, personal communication).42 (32) Nahuatl of Cuetzalan 41 State of Puebla, about 180 km east of Mexico City. 42 The examples in (32) were provided through e-mail and glossed by me. All errors and misinterpretations are mine. 40 2 MORPHOLOGY AND SYNTAX OF R-MARKING a. ni-tomineh ni-catca 1SGR-rich.person 1SGS-be.PST ‘I was rich.’ b. ti-tayecanque ti-ye-zqui 2SGR-leader 2SGS-be-FUT ‘YouSG will be [a/the] leader.’ c. ti-tayecanque ti-mo-chihua-z 2SGR-leader 2SGS-REFL-make-FUT ‘YouSG will become [a/the] leader.’ d. ti-nelli (in) ti-teopixcat 2SGR-real IN 2SGR-priest ‘YouSG are a real priest.’ Interestingly, in addition to the environments exemplified above, R-marking is also attested on cross-referenced non-principal nouns in some dialects. (33) Tetelcingo Nahuatl a. noche costumbres tli all customs tejua te-jorioj-te COMP 1 PL te-c-pea 1PLR-Jewish.person-PL 1PLS-3SGO-have ‘all the customs which we, Jewish people, have’ (La Liga Bíblica, [1980] 2009, Acts 26:3) b. naja ne-Pöblo nemiech-ejcuel-fea nönca saludos ca no-mö 1SG 1SGR-PN 1SGS.2PLO-write-APPL this belaja greetings by 1SGP-hand itself ‘I, Paul, am writing these greetings to youPL with my own hand ...’ (La Liga Bíblica, [1980] 2009, 1 Corinthians 16:21) 41 2 MORPHOLOGY AND SYNTAX OF R-MARKING A text of the Nahuatl of Milpa Alta43 (Nuclear Subarea in the Central Area) narrated by an indigenous storyteller Doña Luz Jiménez shows that her language preserved the Rmarking on cross-referenced non-third-person nouns. Unfortunately, her text (Horcasitas, [1968] 1979) does not contain any instance of non-third-person non-present/indicative copulatic sentences. (34) Milpa Alta Nahuatl a. nehuatl ni-Luz-Jiménez o-ni-c-pia-ya 1SG 1SGR-PN-PN A-1 SGS -3 SGO -have- IMPF ‘I, Luz Jiménez, was about ten years old’ b. tlen ti-c-nequi (Horcasitas, [1968] 1979, 38) (Horcasitas, [1968] 1979, 86) c. ihuan tla namehuan nan-ta-tatin amo namo-tequipachoa tehuan 2PLR-RDP-father NEG 2PLS.REFL-worry 1PL ti-temachtiquez coza 1PLR-teacher year 1PL-teacher ‘what we teachers want’ if 2PL about ten tehuan ti-temachtiquez what 1PLS-3SGO-want 1PL and cana matlactli xihuitl to-tequipachoa ... thing(?) 1PLS.REFL-worry ‘And we teachers worry about [those] things if youPL parents do not.’ (Horcasitas, [1968] 1979, 86) 2.5.2 Nahuan languages without uniform R-marking In contrast to these dialects, Nahual of Michoacán44 (Occidental Periphery) reported by Robinson and Sischo (1969) and Sischo (1979) lacks the R-marking in most non-principal positions. 43 The Federal Disrict, about 30 km south of Mexico City. 44 State of Michoacán, about 450 km west of Mexico City. 42 2 MORPHOLOGY AND SYNTAX OF R-MARKING In this language, the subject marking appears in principal nouns. (35) Michoacán Nahual a. newal ni-mo-siwal 1SG 1SGS-2SGP-woman ‘I am yourSG wife.’ b. damo (Robinson and Sischo, 1969, 70) ni-koneho not.really 1SGS-rabbit ‘I am not a rabbit.’ (Sischo, 1979, 378) Contrarily, R-marking is not required in other environments where it would be obligatory in Classical Nahuatl. (36) Michoacán Nahual a. amo ši-ye NEG IMP -be tonto silly ‘Don’t be (youSG ) silly.’ b. wan ti-ye-k (Sischo, 1979, 320) gwače and 2SGS-be-PRF soldier ‘And youSG were a soldier?’ c. ti-ka-te (Sischo, 1979, 320) mičeros 2SGS-be-PL fishermen ‘We are fishermen.’ (Sischo, 1979, 319) d. tel animal-ito ti-iš-la-kwa-li-ta-t-aya 2SG animal-DIM 2SGS-1SGO-UNO-eat-APPL-L.be-(?)-PST ‘YouSG little beastie who ate up my crop for me always, ...’ (Sischo, 1979, 314) 43 2 MORPHOLOGY AND SYNTAX OF R-MARKING The situation is reported to be similar in Pipil (endonymically Nawat), a Nahuan language of El Salvador (Campbell, 1985). Similarly to Michoacán Nahual, Pipil has the subject marking on nominal predicates (37) whereas the examples in (38) show that the uniform R-marking is consistently lacking in this language. (37) Pipil a. taha ti-mayordomoh 2SG 2SGS-steward ‘YouSG are a steward.’ (Campbell, 1985, 55) (38) Pipil a. n-yu ni-nemi deskalsoh 1SGS-go 1SGS-be barefoot ‘I am going to be barefoot.’ b. ni-[y]aw ni-metzin-kwepa (Campbell, 1985, 110) tajtaka-itzkiani-met 1SGS-go 1SGS-2PLO-change people-fisher-PL ‘I will make youPL fishers of men.’ c. maka shi-panu-kan ken ne NEG . IMP IMP -pass- PL (NBTN, 2012, Matthew 4:19) tensasal-met like ART hypocrite-PL ‘Do not be like the hypocrites.’ (NBTN, 2012, Matthew 6:5) Curiously, the Nahuan languages presented above and Classical Nahuatl share more or less similar grammatical features such as obligatory object marking, productive noun incorporation, and inflected nominal predicates. Here, it seems that the presence or absence of uniform R-marking does not correspond to a major typological division. 44 2 MORPHOLOGY AND SYNTAX OF R-MARKING 2.6 Comparable phenomena in other languages Subject or referential person inflection of nouns is found in languages from diverse areas. However, outside of the Nahuan group, I have not yet encountered an undoubted example of a language with the same type of uniform obligatory R-marking as Classical Nahuatl. This section gives the examples of comparable phenomena in unrelated languages. Examples include Mayan, Turkic, Wakashan, Dravidian, and Elamite. 2.6.1 Nominal predicates with person markers It is very common for a language to have bound-form pronominal copulae. They are attached to nouns to form nominal predicates, and are usually limited to the principal of a clause. Stassen (1997) gives various examples of this type. In Mayan languages, for example, nouns can directly take the person markers called Set B (juego B), which typically mark the intransitive subjects and transitive objects of verbs (i.e. so-called absolutive person markers), to form nominal predicates. In Yucatec, Set B markers appear regardless of whether there is an overt pronominal subject.45 (39) Yucatec a. koolnáal-en (teen) farmer-B.1SG 1SG ‘I am a farmer.’ (Armstrong, 2009) b. polok-o’ob (leti’o’ob) fat-B.3PL 3PL ‘They are fat.’ (Armstrong, 2009) Similarly, in Turkic languages, nouns take pronominal copulae and serve as nominal predicates (Johanson, 1998, 41). In Turkish, the person markers employed in nominal 45 It is so far not clear whether this person marking is required in other environments in Yucatec. 45 2 MORPHOLOGY AND SYNTAX OF R-MARKING marking are basically the same as those which appear in verbal person marking (Göskel and Kerslake, 2005, 87–90). (40) Turkish a. Necla öğretmen PN teacher ‘Necla is a teacher.’ (Göskel and Kerslake, 2005, 120) b. (ben) öğretmen-im 1SG teacher-1SG ‘I am a teacher.’ (Kornifit, 1997, 78) c. evde-yim home-1SG ‘I am at home.’ (Johanson, 1998, 41) Interestingly, in Turkish, nominal predicates with pronominal copulae can appear in non-neutral tense/modality as well. (41) Turkish a. (ben) öğretmen-d-im 1SG teacher-PST-1SG ‘I was a teacher.’ (Kornifit, 1997, 79) b. evde-yse-m home-COND-1SG ‘if I am at home’ (Göskel and Kerslake, 2005, 85) However, Turkic-type person-marked nouns are typically limited to predicates. Nonprincipal nouns are not person-marked at least in Turkish and Sakha (Yakut) (Tooru Hayasi and Fuyuki Ebata, personal communications). 46 2 MORPHOLOGY AND SYNTAX OF R-MARKING 2.6.2 Noun–verb uniformity for predicates Since the time of Sapir (1921, 133–134), it has been claimed that some languages of Pacific Northwest in North America exhibit an extreme noun–verb symmetry. They include the languages of the Wakashan (Jacobsen, 1979), Salishan (Kinkade, 1983; van Eijk and Hess, 1986), and Chimakuan families (Mithun, 1999, 60–67). Although it has often been argued that even those languages have a lexical noun– verb distinction, the examples from Nuuchahnulth (Nootka) and Makah (both from the Nootkan group of Wakashan) below show that a single root serves both as a predicate and as an argument. (42) Nuuchahnulth a. qo·Pas-ma Pi·h.-Pi· man-IND .3 SG large-ART ‘The large one is a man.’ b. Pi·h.-ma· qo·Pas-Pi large-IND .3 SG man-ART (Swadesh, 1939, 78)46 ‘The man is large.’ (43) Makah a. bačiλ’–its q’idi·l-iq bite.PASS.1SG dog-ART ‘The dog bit me.’ b. q’idi·l dog.IND.3 ‘It’s a dog.’ 46 (Jacobsen, 1979, 110) The glosses in (42) are given according to the interpretation by Jacobsen (1979). 47 2 MORPHOLOGY AND SYNTAX OF R-MARKING Dixon (2010–2012) illustrates this pattern as Figure 6. By this scheme, he displays the fact that both nouns and verbs can serve either as predicates or as arguments. CLAUSE STRUCTURE predicate NP as arguments WORD CLASS verb noun (Dixon, 2010–2012, II, 43) Figure 6: Pacific Northwestern type of noun–verb symmetry The scheme of Figure 6 is also applicable to Classical Nahuatl and some other Nahuan languages; as shown in Section 2.1 above, the distinction between nouns and verbs does not directly correspond to the syntactic function in that nouns can be predicates and verbs can be arguments without overt copulae or complementizers. However, there is an important difference between these two groups of languages. At least in many languages of Pacific Northwest, non-predicate nouns do not inflect in the same way as predicate nouns do. In the examples presented above, argument nouns do not inflect for person.47 In Nuuchahnulth, for example, a predication contains only one predicate, and only the predicate can take a peripheral (i.e. person/mood) suffix (Nakayama, 2001, 80).48 In the Nuuchahnulth example (44), a non-predicate noun k’a:yišk’in ‘sea shell’ cannot take a peripheral suffix. (44) Nuuchahnulth 47 The non-predicate nouns in (42) and (43) appear totally uninflected at first glance. However, Toshihide Nakayama (personal communication) points out that it is dangerous to assume that they are really uninflected, since the “articles” in these examples might be interpreted as nominalizers (cf. Jacobsen, 1979). Accordingly, this thesis simply states that they “do not inflect for person.” 48 As another important feature of Nuuchahnulth, when two or more nouns form a noun-phrase-like constituent, only the head noun is inflected predicatively and other nouns are usually left uninflected (Toshihide Nakayama, personal communication). 48 2 MORPHOLOGY AND SYNTAX OF R-MARKING Pumč’a:pPaλ–qun k’a:yišk’in Pumč’a:p-’aλ–-qw in k’a:yišk’in play-PUN-COND.1PL sea.shell ‘We used to play with shells.’ (Toshihide Nakayama, personal communication) As already shown above, Classical Nahuatl exhibits a completely different type of cross-categorical symmetry. While only predicates inflect for person in Nuuchahnulth, Classical Nahuatl requires that every noun inflect for person regardless of whether it is a predicate or not. In other words, Classical Nahuatl differs from Nuuchahnulth in that it shows an inflectional predicate–argument symmetry. In the terminology of this thesis, the noun–verb symmetry observed in the languages of Pacific Northwest is not a uniform R-marking. Accordingly, Dixon’s (2010–2012) scheme can be extended as in Figure 7 and Figure 8 respectively. INFLECTION personal non-personal CLAUSE STRUCTURE predicate NP as arguments WORD CLASS verb noun Figure 7: Pacific Northwestern type of noun–verb symmetry (extended) INFLECTION personal non-personal CLAUSE STRUCTURE predicate NP as arguments WORD CLASS verb noun Figure 8: Classical Nahuatl type of noun–verb symmetry 49 2 MORPHOLOGY AND SYNTAX OF R-MARKING 2.6.3 R-marking on non-predicate nouns In contrast, some languages appear to show uniform R-marking on both predicate and non-predicate nominals. This section discusses the examples from Malto and Kurux (Dravidian), Elamite (isolate), and Kaqchikel (Mayan). In a limited number of Dravidian languages such as Old Tamil (South Dravidian) and the languages of Kurux-Malto group (North Dravidian) are reported to have nominal person markers (Nozomi Kodama and Masato Kobayashi, personal communications). In Malto, a Kurux-Malto language of Jharkhand, India, the same set of pronominal suffixes as those employed in verbal subject marking are used for both predicate and non-predicate nouns (Kobayashi, 2012, 29–30). Suffix (V_) Suffix (C_) Pronoun (nominative) 1SG -n -en e:n 2SG . M -y -e < *-ey ni:n 2SG . NM -Ø -i ni:n 3SG . M -h -eh a:h 3SG . NM -d -id a:d 1PL . EXCL -m -em e:m 1PL . INCL -t -et na:m 2PL -r -er ni:m 3PL -r -er a:ber (Based on Kobayashi, 2012, 27–30) Table 5: Malto pronominal suffixes and personal pronouns Unlike Classical Nahuatl, Malto pronominal suffixes have overt phonological forms also for the third person. As Table 5 illustrates, they are at least partly related to indepen50 2 MORPHOLOGY AND SYNTAX OF R-MARKING dent pronouns phonologically. Person-marked nouns can either stand alone as nominal predicates as in (45) or serve as arguments as in (46), although the overt person marking is undoubtedly optional in inanimate nouns. (45) Malto a. majye-n e:ni village.chief-1 SG I.myself ‘I am the chief of the village people.’ (Kobayashi, 2012, 141) b. i:d essa mecg manu-d this very high tree-3SG.NM ‘This is a very tall tree.’ (Kobayashi, 2012, 30) (46) Malto a. ayo-n baric-tan mother-1 SG come.B 2- PST.1 SG ‘I, mother, have come!’ b. ababeãi-tar-bahni (Kobayashi, 2012, 272) ão:k-tam e:mu onãqal-am grandfather-PL-LOC stay-PST.1 PL . EXCL we.EXCL sibling-1 PL . EXCL ‘WeEXCL siblings stayed at our grandfather’s place.’ (Kobayashi, 2012, 119) c. ortu maPe-h mañja-h one boy-3SG.M be.B 2- PST.3 SG . M ‘There was a boy.’ (Kobayashi, 2012, 236) Contrarily, it is unclear whether the person marking on the complements of ‘be’ and ‘become’ are really obligatory in Malto. Kobayashi (2012) has both person-marked and non-person-marked examples, as in (47) and (48) respectively. (47) Malto 51 2 a. sat MORPHOLOGY AND SYNTAX OF R-MARKING mañja-tam unãPal-em seven sibling-1PL.EXCL be.B 2. PST.1 PL . EXCL ‘WeEXCL were seven siblings.’ (Kobayashi, 2012, 87) b. andi bey-no ão:k-no bey-no ão:k-no Poúro-h ra:ja-h then be-IPP stay-IPP be-IPP stay-IPP blind-3SG.M king-3SG.M mañj-ahay become.B 2- PST.3 SG ‘While they were living, a blind man became a king.’ (Kobayashi, 2012, 221) (48) Malto a. ike-h man-nah isaye who-3 SG . M become-PRS.3SG Christian ‘Someone becomes Christian.’ (Kobayashi, 2012, 169) b. úu:ãu meñj-a? tiger be.B 2- PST.3 SG ‘Was she a tiger?’ (Kobayashi, 2012, 316) Moreover, Kobayashi (2012) contains some examples of animate nouns which possibly lack overt person marking. These examples might suggest that person markers can drop at least in some environments, but it is difficult to decide whether they are really absent or simply phonetically inaudible. (49) Malto a. ta:nu anake i: orta-qadi pelmaqe then now this.NM one-only girl ‘Now she is the only girl.’ b. a:r ort (Kobayashi, 2012, 314) okk-i and one.person sit-PRS .3 SG 52 2 MORPHOLOGY AND SYNTAX OF R-MARKING ‘And the other woman was sitting.’ (Kobayashi, 2012, 211) The situation seems to be similar in its sister language Kurux (exonymically also called Oraon). Grignard (1924, 163–166) reports that nouns in apposition to personal pronouns must be overtly person-marked. (50) Kurux a. ēn pacgi-n ekā besē bēc-on 1SG old.man-1SG which like dance-FUT.1SG ‘I, an old man, how can I dance?’ b. ēm (Grignard, 1924, 165) innā ahr.a-n Khristān-am bār-dam 1PL.EXCL Christian-1PL.EXCL today meat-ACC abstain.from-PRS.1PL.EXCL ‘WeEXCL Christians abstain from meat today.’ c. ās Kũr.ukh-as Kũr.ukh kattha-n (Grignard, 1924, 165) bal-das 3SG.M Kurux.man-3SG.M Kurux word-ACC do.not.know-PRS.3SG.M ‘He, a Kurux man, does not know Kurux.’ (Grignard, 1924, 166) Interestingly, Kurux-Malto languages have morphological case marking unlike Classical Nahuatl. Pronominal suffixes and case suffixes seem to interact at least in Kurux. (51) is the only example presented in Grignard (1924). (51) Kurux nı̄n Kũr.ukh-ai-n kōr’-ā hō˜ mal ci’-or 2SG Kurux.man-2SG-ACC enter-GER too NEG allow-FUT.3 PL ‘YouSG , a Kurux man, they won’t even permit indoors.’ (Grignard, 1924, 166) Malto and Kurux seem to differ in the pattern of person–case interaction, but more systematic data are needed to discuss this issue any further. 53 2 MORPHOLOGY AND SYNTAX OF R-MARKING More mysterious data are found in another Dravidian language Old Tamil, where there are several different person markers for a single person/number (Agesthialingom, 1977; Rajam, 1992). Although the nominal person markers have been amply documented language-internally, their typological status needs further investigation. Another candidate for a language which has uniform R-marking is Elamite, where there is a system of person marking which has traditionally been called gender.49 Elamite, especially in its earlier stage, is reported to have the markers which indicate the person of nouns. They also participate in the index agreement between the possessor and its head.50 Noun “gender” Participle Verbal (Conjugation I) Pronoun 1SG -k -k -h u 2SG -t -t -t ni/nu 3SG -r/-Ø -r -š i/hi 1PL – – -hu < *-h-h nika/nuku 2PL – – -h-t num/numi 3PL -p -p -h-š ap/ap(p)i (Based on Reiner, 1969; Stolper, 2004; inanimate “gender” markers are omitted) Table 6: Middle Elamite person markers and pronouns Similarly to Classical Nahuatl, Elamite has non-predicative person-marked nouns such as sunki-k (king-1SG) ‘I, the king’ and nap-(i)r (god-3SG) ‘he, god.’ First-, second-, and third-person forms of nouns are sometimes referred to as locutive, allocutive, and delocutive respectively. 49 Bork (1934); Grillot (1978); Reiner (1969); Stolper (2004), etc. 50 In its later stages, the referential nature of the “gender” markers is obscured by their increasing use as possessor–head agreement markers (Grillot-Susini and Roche, 1987, 21). 54 2 MORPHOLOGY AND SYNTAX OF R-MARKING Unlike Classical Nahuatl and Kurux-Malto, Elamite nominal person markers are different from verbal person markers.51 Neither does it phonologically correspond to independent person pronouns at least synchronically. Examples follow:52 (52) Middle Elamite a. u Untaš-Napiriša šak Humbannumena-ki sunki-k Anzan Šušun-k 1SG PN son PN-1SG king-1SG LN LN -1 SG ‘I, Untaš-Napiriša, the son of Humbannumena, king of Anzan and Susa’ (Grillot-Susini and Roche, 1987, 50) b. Inšušinak tepti Alim-Elum-r PN lord nap-(i)r LN - LN -3 SG . ANIM u-ri god-3SG.ANIM 1SG-3SG.ANIM ‘Inšušinak, the lord of High City, my god’ (Grillot-Susini and Roche, 1987, 56) c. kukunnum sunki-p urpi-p u-pi temple.tower king-3PL.ANIM predecessor-3PL.ANIM 1SG-3PL.ANIM im-me kuši-hš u kuši-ih NEG -3 PL . INANIM build-3PL.ANIM 1SG build-1SG ‘I built the temple tower which my previous kings did not build’ (Grillot-Susini and Roche, 1987, 51) Interestingly, the examples (52) show that the nominal person marking in Elamite crosses a word boundary unlike that of Classical Nahuatl. For example, tepti Alim-Elum-r ‘he, the lord of High City’ in (52b) has a person marking on Alim-Elum ‘High City,’ but not on tepti ‘lord.’ While Classical Nahuatl R-marking occurs only word-internally, Elamite nominal person markers sometimes behave like clitics rather than inflectional 51 Participles take nominal person markers instead of verbal person markers. 52 These examples omit the determinatives and other scriptual information. 55 2 MORPHOLOGY AND SYNTAX OF R-MARKING affixes. Accordingly, it may be the case that Elamite “gender” markers are clitics with their own syntactic positions. Mayan languages are also informative in these aspects, but the possibility or obligatoriness of nominal person marking seems to differ from language to language. The cross-linguistic distribution of uniform R-marking in the Mayan family is not clear. As mentioned in Section 2.6.1 above, Mayan languages in general can form copularless nominal predicates by means of the person markers called Set B. In Kaqchikel, where Set B markers are procliticized to noun stems as in (53), non-present form of copular clauses have redundant person marking on noun stems as in (54b) and (54c) (Yoshiho Yasugi, personal communication). In this respect, Kaqchikel nominal person marking is reminiscent of that of Classical Nahuatl. (53) Kaqchikel a. in B .1 SG ak’wal boy ‘I am a boy.’ b. at B .2 SG (García Matzar et al., 1999, 89) ak’wal boy ‘YouSG are a boy.’ c. Ø B .3 SG (García Matzar et al., 1999, 89) ak’wal boy ‘He is a boy.’ (García Matzar et al., 1999, 89) (54) Kaqchikel a. in B .1 SG nimanel obedient ‘I am obedient.’ 56 2 b. in B .1 SG MORPHOLOGY AND SYNTAX OF R-MARKING nimanel xk-in-ux obedient FUT. INCMP-1SG.B-be ‘I will be obedient.’ c. at B .2 SG kow x-at-ux strong PST-2SG.B-be ‘YouSG were strong.’ Again, this issue needs more investigation. Since the non-third-person marking on non-predicate nouns is a marginal phenomenon which is rarely mentioned in reference grammars, there is little information available so far concerning this phenomenon. Further research is needed both language-internally and cross-linguistically.53 53 Ken Shibushita (personal communication) suggests that another Mayan language Q’eqchi’, which belongs to the Quichean subgroup along with Kaqchikel, appears to lack the person marking on non-predicate nouns. 57 3 THEORETICAL BACKGROUND 3 Theoretical background Despite that the principal R-marking in Classical Nahuatl has been cited in many crosslinguistic works such as Stassen (1997) and Baker (2003), and despite that there has been intensive debate on the noun–verb symmetry in the languages of Pacific Northwest, the predicate–argument uniformity of R-marking has been the subject of little attention outside of the tradition of Classical Nahuatl grammatical studies. This section introduces the arguments of Launey (1994) and Andrews ([1975] 2003), two major contemporary theoretical works on Classical Nahuatl which put emphasis on the obligatoriness and uniformity of R-marking, and lists some cross-linguistic issues which is relevant to the problem of R-marking. 3.1 Language-internal analyses Based on the legacy of pre-modern missionary grammar and their own textual study, two leading contemporary Nahuatlists present comprehensive frameworks of Classical Nahuatl morphosyntactic analysis. Both of them emphasize the omnipredicativity and the self-containedness of sentence-words, and characterize the grammar of Classical Nahuatl as a radically different one from that of other languages. 3.1.1 Launey’s omnipredicative analysis Launey (1991, 1994, 2003, 2004) gives a formal framework of Classical Nahuatl morphosyntax based on a dependency model. He characterizes the Classical Nahuatl by the concept of omnipredicativity; that is, the morphosyntax of Classical Nahuatl constitutes a system where “any concept (represented by a lexical item and assigned related grammatical categories if necessary) is a predicable one” (Launey, 1994, 60) in that both nouns and verbs can be the predicate of a clause. 58 3 THEORETICAL BACKGROUND In Launey (1994, 59–60), he introduces a typological axis of the degree of separation between lexical categories (verb vs. noun) and predicative status, at one end of which Classical Nahuatl is located. He assumes that R-marked nouns are primarily predicative (i.e. the prototypical use of nouns is that as a principal), and their argumental use is derived through the process of cross-reference, which he terms actantialisation (Launey, 1994, 70–74).54 Consequently, he analyzes the nouns in Classical Nahuatl as a kind of intransitive predicates (Launey, 1994, 50–56). According to him, the major difference between nouns and verbs in Classical Nahuatl is that the former cannot bear tense, aspect, and modality, and there is no radical difference in argument structure between them. Accordingly, he assumes that argument nouns preserve their predicative structure through the process of actantialisation. 3.1.2 Andrews’ omniclausal analysis A more radically omnipredicative vision is presented in the revised edition of Andrews’ ([1975] 2003) textbook. He claims that what have been traditionally called “nouns” and “verbs” are really not nouns and verbs, but word-size nominal and verbal clauses which obligatorily contain a subject and a predicate within single words (Andrews, [1975] 2003, 45–49). Thus, in his model, the traditional cross-linguistic notions of noun, verb, and word are of little importance in analyzing the grammar of Classical Nahuatl; rather, in Classical Nahuatl, every noun is a nominal nuclear clause (NNC) which is itself a sentence-word, or more precisely a word-sentence.55 54 “... [I]n an omnipredicative system, the predicative function is primitive and any argumental function is derived.” (Launey, 1994, 82) 55 Hill’s (2006) review criticizes this view as too extreme. 59 3 THEORETICAL BACKGROUND Accordingly, he assumes that the person affixes (among them are R-prefixes) in Classical Nahuatl are personal pronouns in the same way as the independent pronouns in English are. He states that, like the personal pronouns in any other language, they are always “deictic and/or anaphoric or cataphoric” (Andrews, [1975] 2003, 49). According to him, the (apparently) non-predicative use of noun words is explained by the process of supplementation. A non-principal nuclear clause is cross-referenced with a person affix on the principal nuclear clause (Andrews, [1975] 2003, 137–138). In other words, a minimal sentence (non-principal nuclear clause) serves as a supplement to a personal pronominal argument (person affix) of another minimal sentence. A simplified version of the supplementation analysis is already illustrated in Figure 5 further above. Thus, such an example as (55b) can be paraphrased as (55b) according to Andrews ([1975] 2003).56 (55) a. ni-kwi:ka ni-petolo’ 1SGS-sing 1SGR-PN ‘I, Peter, sing.’ b. nikwi:ka ‘I sing’ + nipetolo’ ‘I am Peter’ (Andrews, [1975] 2003, 138–139)57 In sum, he assumes that every noun and verb in Classical Nahuatl is a clause in the literal sense. In analogy to the term omnipredicative, Andrews’ ([1975] 2003) model might be characterized as omniclausal. Although the model of Andrews ([1975] 2003) is a more radical one than that of Launey (1991, 1994, 2003, 2004), they share the intuition that Classical Nahuatl nouns 56 Technically, as already mentioned at the end of Section 2.2.3, Andrews ([1975] 2003) analyzes the subject person markers as prefix–suffix dyads. 57 This example seems to be a sentence composed by Andrews ([1975] 2003). 60 3 THEORETICAL BACKGROUND are pre-formed, self-contained subject–predicate complexes, and they preserve their predicative structure even when they are not sentence principals. 3.2 Cross-linguistic issues The problem of R-marking has many implications for language typology and theoretical linguistics. That is, Classical Nahuatl is a strong candidate for a language without the morphosyntactic distinction between predicates and arguments. 3.2.1 Typological status of omnipredicativity Although the universality of noun–verb distinction has been the subject of numerous discussions (Sasse, 1993; Croft, 2000; Evans, 2000; Evans and Osada, 2005, etc.), they are mainly on the presence or absence of the lexical distinction between nouns and verbs, in respect of which Classical Nahuatl is of little importance. For example, as mentioned in Section 2.6, some languages in Pacific Northwest have often been alleged to lack the lexical distinction between nouns and verbs. In contrast, as seen in the classic examples (56), these languages have rigid morphological distinction between predicates and their arguments. The root P·h.w - ‘large’ takes a peripheral (person/mood) suffix when it is used as a predicate whereas it is followed by an article when used as an argument. (56) Nuuchahnulth a. Pi·h.-ma· qu·Pas-Pi large-IND .3 SG man-ART ‘The man is large.’ b. qu·Pas-ma (Swadesh, 1939, 78) Pi·h.-Pi· man-IND .3 SG large-ART 61 3 THEORETICAL BACKGROUND ‘The large one is a man.’ (Swadesh, 1939, 78) As already mentioned in Section 2.6.2, Classical Nahuatl raises a completely different issue. Unlike the well-known examples from Pacific Northwest, Classical Nahuatl exhibits a morphological predicate–argument symmetry. Of course, Classical Nahuatl has a clear distinction between predicates (principals) and their arguments (non-principals) at the logical level, although many examples are ambiguous as to which word is construed as the principal of a clause.58 This logical distinction is also reflected in the consistent absence of quasi-article in on sentence principals. However, both principal and non-principal nouns are obligatorily marked with the same set of person affixes. In other words, Classical Nahuatl clearly does not distinguish between principals and non-principals at the morphological level. The problem is, then, whether this uniformity reflects the structural similarity between principals and non-principals. Does it mean that non-principal nouns are treated in the similar way as principals? More specifically, if principal nouns are intransitive predicates (I am a woman, it is a house, etc.), is it necessary to assume that non-principal nouns (woman, house, etc.) equally contain predicative structures (an entity which is a woman, an entity which is a house, etc.)? The answer to this question is yes according to the omnipredicative/omniclausal analyses of Launey (1991, 1994, 2003, 2004) and Andrews ([1975] 2003). If they are correct, Classical Nahuatl is a peculiar language where all full words are predicates, which would be a surprising conclusion for typologists. 58 For example, the example (55) can be construed either as ‘I, Peter, sing’ and ‘I, who am singing, am Peter.’ Nonetheless, the very fact that there is a principal–non-principal ambiguity proves that Classical Nahuatl distinguishes between these two functions. 62 3 THEORETICAL BACKGROUND 3.2.2 Typological and theoretical importance If their understanding is on the right track, it leads to many theoretical and typological problems for the study of language. Firstly, it appears to be inconsistent with the typological similarity between Classical Nahuatl and many other languages without uniform R-marking. For example, as extensively discussed in Baker (1996), Classical Nahuatl has many morphosyntactic features in common with other polysynthetic languages,59 most of which lack uniform R-marking of the type observed in Classical Nahuatl. Even within the Nahuan group, as overviewed in Section 2.5.2, there are languages such as Michoacán Nahual which lack uniform R-marking but preserve the overall polysynthetic characteristics. Secondly, it would be a critical blow to the theory which characterizes nouns by their referential property. Baker (2003) assumes that nouns should be characterized as the expressions which refer to some entity.60 Previous to him, Chierchia (1998) argues that bare nouns in some languages are expressions of type ⟨e⟩, contrarily to the standard analysis that both verbs and common nouns are universally of type ⟨e, t⟩. Baker (2003) extends this intuition and claims that noun is universally a lexical category which bears a referential index. If Classical Nahuatl non-principal nouns are really predicates, it would be a counterexample to his analysis. In addition, if Classical Nahuatl nouns should be interpreted as self-contained subject– predicate complexes, it would lead to many technical problems in any syntactic framework, be it formal, functional, or cognitive. For example, Baker (1996), which is a 59 Among the characteristics which Baker (1996) takes as examples are: (i) syntactic noun incorporation; (ii) possessor agreement in nouns; (iii) absence of free-form reflexives, “true” determiners, and infinitives; (iv) inclusive agreement; (v) free word order; etc. 60 Baker (2003, 103) points out that the notion of “criterion of identity” introduced in Geach (1962) and Gupta (1980) are in line with his hypothesis. 63 3 THEORETICAL BACKGROUND controversial but influential generative work on polysynthetic languages, does not provide detailed discussion on the word-internal structure of R-marked nouns. It is probably because he bases his argument largely on the data from Mohawk, a language which is typologically similar to Classical Nahuatl in many respects but lacks uniform R-marking on nouns. Although the polysynthesis–non-polysynthesis dichotomy and the syntactic analysis presented in Baker (1996) themselves need justification, his framework would also need refinement in this regard. 64 4 DISCUSSION 4 Discussion 4.1 Towards a non-omnipredicative account In opposition to the omnipredicative analysis held by Launey (1991, 1994, 2003, 2004) and Andrews ([1975] 2003), this section proposes a non-omnipredicative account of Rmarking. As pointed out in the last section, Launey (1994) and Andrews ([1975] 2003) agree in that R-marking should be interpreted as the realization of the omnipredicative nature of the language. According to them, Classical Nahuatl nouns are pre-formed subject– predicate complexes regardless of their syntactic positions and even non-predicational (non-principal) nouns preserve their predicative structures through the process of subordination (actantialisation in Launey (1994), supplementation in Andrews ([1975] 2003)). However, their omnipredicative interpretation leads to serious problems both languageinternally and cross-linguistically. Language-internally, the omnipredicative approach misses many usages of non-principal R-marked nouns, as is to be pointed out in Sections 4.5 and 4.6 below. Cross-linguistically, it leads to an unnatural conclusion that the grammar of Classical Nahuatl is a radically different one from that of many other languages, including its closely related languages such as Michoacán Nahual and Pipil, in that all nouns have clausal or predicative structures in Classical Nahuatl. This thesis puts forward another interpretation of R-marking: namely, that non-principal R-marked nouns do not behave as subject–predicate complexes in Classical Nahuatl syntax. Informally speaking, this thesis argues that R-marked nouns are nouns, not clauses. Specifically, virtually all examples of non-principal R-marked nouns can be unproblematically described once it is assumed that the R-prefixes on them are ignored in the process of sentence composition. Consequently, R-prefixes can be interpreted either as (i) 65 4 DISCUSSION appositional bound-form pronominal elements which do not affect the syntactic category of its host noun, or as (ii) the realization of an optional61 semantic feature. In either case, the presence of R-marking does not assign a clausal status to its host; R-marked nouns are not treated as pre-formed subject–object pairs in syntax. Section 4.2 introduces the terminology and notation necessary in the discussion of R-marking. Section 4.3 proposes to distinguish three types of R-marked nouns (principal, intraeventive, and appositional) according to their environments, and identifies a phenomenon of multi-word R-marking which occurs in all of these three environments. Their individual characteristics are discussed in Sections 4.4 through 4.6. Section 4.7 aims to provide a general understanding of R-marking. 4.2 Terminology and notation Four terms are distinguished in this thesis: principal, propositional, predicative, and predicate. The term principal has already been defined as “the syntactic matrix verb/noun of the clause” in Section 1.3.4 further above. The definition of propositional in this thesis follows the definition of proposition in logic and formal semantics (e.g. Gamut, 1991, 29); an expression is propositional iff it bears a truth value. Linguistically, a propositional expression should be finite, provided with tense and affirmative force. The term predicative is used after the generative literature such as Williams (1980) and Bowers (1993); an expression is predicative iff it contains both a predicate and its subject. A predicative expression, therefore, is not necessarily a propositional one. So-called small clauses in English, for instance, is predicative but not propositional. The traditional term predicate, which is used in the sense stated by Dixon (2010–2012, I, 79), is hereafter used only as a convenient informal label. 61 See Note 79 below for the definition of optionality of features. 66 4 DISCUSSION Henceforth, the R-prefix and its host (i.e. the stem plus the possessor prefix) are abbreviated as “R” and “S” respectively. In addition, the R-marking relationship between R and S is formalized as “[R, S].” For example, the nouns (57a) and (57b) can be written as (58a) and (58b) respectively. (57) a. ti-pi:lli 2SGR-noble ‘youSG are a noble; youSG , who are a noble’ (FC VI, 94 etc.) b. ni-mo-ta’ 1SGR-2SGP-father ‘I am yourSG father; I, who am yourSG father’ (58) (FC IV, 61 etc.) a. [2SG, noble] b. [1SG, yourSG father] When the predicative relationship [R, S] is itself propositional (i.e. the clause states that R is S), it is expressed as “R=S.” For example, the R-marked noun ti-me:ši’ka [1PL, Mexica] in (59a) is an R=S expression since this clause states that there is an attributive relationship between the R-participants (people in the speaker’s group) and the entity denoted by S (Mexica). Contrarily, the R-marked noun t-etikapo:l [2SG, lazy] in (59b) is not an R=S expression since the relationship [2SG, lazy] here states nothing about the actual relationship between R (2SG) and S (lazy). (59) a. ka ti-me:ši’ka’ AFF 1 PLR - EN ‘We are Mexica.’ (FC XII, 34) b. a’mo: t-etikapo:l ti-ye-s NEG 2SGR-lazy 2SGS-be-FUT ‘YouSG should not be lazy.’ (FC VI, 214) 67 4 DISCUSSION 4.3 Classification of R-marked nouns and R-marking This thesis divides the R-marked nouns in Classical Nahuatl along two axes. Firstly, they are classified into three types according to their syntactic environments: principal, intraeventive, and appositional. The term non-principal is used as a cover term for the latter two types. Secondly, a distinction is made between single-word R-marking and multi-word R-marking according to whether an R-marking involves a single nominal word or more. These two dimentions cross-cut each other; multi-word R-marking appears in any one of the three syntactic environments (principal, intraeventive, or appositional). 4.3.1 Principal, intraeventive, and appositional Principal R-marked nouns are used as sentence principals (principal NNCs in Andrews’ ([1975] 2003) terminology). Their characteristics have already outlined in Section 2.4.1. The nouns emphasized by bold-face type in (60) are principal R-marked nouns. (60a) is a typical example of principal R-marked nouns. Principal R-marked nouns may be negated (60b), interrogativized (60c), and embedded into another clause (60d). (60) a. ka ye AFF ni-we:we’ already 1SGR-old.man ‘I am already old.’ (FC III, 33) b. a’mo: sa: ti-ne:n-λa:kaλ NEG only 2SGR-useless-person ‘YouSG are not only a useless person.’ (FC VI, 118) c. kwiš t-otomiλ Q 2SGR-EN ‘Are youSG an Otomí?’ (FC X, 178) 68 4 DISCUSSION in am-me:ši’ka’ ... d. in λa: Ø-nelli IN if 3SGR-true IN 2PLR-EN ‘If it is true that youPL are Mexica, ...’ (FC XII, 6) Non-principal nouns (supplementary NNCs in Andrews’ ([1975] 2003) terminology) illustrated in Section 2.4.2 above are further divided into two subtypes in this thesis, namely intraeventive and appositional ones. Intraeventive R-marked nouns typically appear as complements of copulae and as resultative complements of other verbs such as či:w(a) ‘to make X’ and kwep(a) ‘to turn X.’ Although they are syntactically not principals, they constitute an essential part of an event. Launey (1994, 110–114) characterizes this type of constructions as a subtype of what he calls “attributive constrcutions.” (61) a. a’mo: no: ye NEG ti-tološočto:n ti-ye-s also already 2SGR-hypocrite 2SGS-be-FUT ‘Nor should youSG be a hypocrite.’ b. ... i:k ok seppa t-amo-pilwa:n thus still one (FC VI, 101) ti-to-či:wa’ 1PLR-2PLP-child 1PLS-REFL-make ‘... so that we become yourPL children again’ c. ka ti-we:i ti-malakayo’ ti-počoλ AFF 2 SGR -big 2 SGR -shade Ø-mic/-mo-či:wi-lia (FC VI, 145) t-a:we:we:λ 2SGR-ceiba 2SGR-cypress in Ø-to-te:kwyo: 3SGS-2SGO-REFL-make-APPL IN 3SGR-1PLP-lord ‘Our lord makes (hon.) youSG as a great shade, a ceiba, a cypress.’ (FC VI, 58) d. kwiš ti-siwa:-pi:lli Q ti-mo-kwepa-s-neki 2SGR-woman-noble 2SGS-REFL-turn-FUT-want ‘Are youSG trying to turn into a noble woman?’ 69 (Prim. 298) 4 DISCUSSION On the contrary, appositional R-marked nouns are less essential to the event denoted by the principals. They are typically cross-referenced with the person affixes of their principals and appear to serve as “supplements” to their pronominal arguments. The use of overtly R-marked appositional nouns constitutes the stylistic uniqueness of Classical Nahuatl texts and are often translated with relative clauses or parenthetical appositions (‘I, (who am) yourPL ancestor Axayacatl, ...’). (62) a. ... ni-no-čo:ki-lia in n-amo-ko:l in n-a:ša:yaka ... 1SGS-REFL-cry-APPL IN 1SGR-2PLP-grandfather IN 1SGR-PN ‘I, yourPL ancestor Axayacatl, lament and ...’ b. ka nika:n t-on-ka’ AFF here [V] (Cant. 74v:11) in ti-no-pil-c/in ... 2SGS-TRSL-be IN 2SGR-1SGP-child-HON ‘Here youSG are, youSG who are my child, ...’ c. nika:n Ø-ame:č-mo-senλa:li-lia here Ø-ame:č-mo-nečikalwia (FC VI, 93) in 3SGS-2PLO-REFL-gather-APPL 3SGS-2PLO-REFL-assemble.APPL IN am-a:-wa’ke’ in an-tepe:-wa’ke’ 2PLR-water-PD IN 2PLR-mountain-PD ‘He gathers (hon.) youPL , assemples (hon.) youPL , youPL who are the owners of water, owners of mountain (i.e. habitants of the city).’ d. ači wel iwki in Ø-to-λa’to:l (FC VI, 67) ti-λa:ka’ a.bit well like IN 3SGR-1PLP-word 1PLR-person ‘It is almost like we humans’ language.’ (FC XI, 26) In Launey’s (1994) and Andrews’ ([1975] 2003) frameworks, even third-person argument nouns, which do not have overt R-marking, are interpreted as R-marked. In the following examples, argument nouns are R-marked as third person, and thus are cross-referenced with the third-person markers on their principals. 70 4 DISCUSSION (63) a. aw in ye’wa:ntin Ø-ESPAÑOLES Ø-ki-wa:l-la:sa’ and IN 3PL 3PLR-Spaniard i:-wa:n Ø-tepos-mi:λ in 3PLS-3SGO-CISL-throw IN in Ø-λe-kikisλi 3SGR-iron-arrow 3SGP-with IN 3SGR-fire-trumpet ‘And the Spaniards shoot iron bolts and guns.’ b. in i:k Ø-kim-i’kalke’ (FC XII, 62) Ø-me:ši’ka’ in Ø-ESPAÑOLES ... IN when 3PLS-3PLO-fight.PST 3PLR-EN IN 3PLR-Spaniard ‘When the Mexica fought with Spaniards, ...’ (FC XII, 57) c. ka Ø-λa’toa:ni in Ø-no-ta’-c/in AFF 3 SGR -ruler IN 3SGR-1SGP-father-HON ‘My father is the ruler.’ (Chim. 23:44) Ø-mote:kwso:ma d. in Ø-i:-ti:λanwa:n IN 3PLR-3SGP-messenger 3SGR-PN ‘Moteuczoma’s messengers’ (FC XII, 17) For example, the relationship between the R-marked noun and its principal in (62a) and (63a) can be parallelly analyzed as Figure 9 below despite that the third-person nouns in (63a) are not overtly R-marked. Cross-referencing ni- -nočoːkilia n- Principal Non-principal Cross-referencing Ø- -ESPAÑOLES Non-principal -amokoːl Cross-referencing Ø- ki- -waːllaːsa’ Principal Ø- -teposmiλ Non-principal Figure 9: Cross-referencing structure of (62a) and (63a) (simplified) 71 4 DISCUSSION It must be noted that it is not obvious whether these “appositional” nouns are really appositional. Due to the absence of audio data and the inconsistent punctuation of traditional texts, such examples as (62) might be analyzed as sentence juxtaposition (‘I lament. I am yourPL ancestor. I am Axayacatl. ...’) rather than sentence-internal apposition. This problem will be covered in Section 4.6 below. 4.3.2 Single-word vs. multi-word Most of the examples presented in the last section involve what can be characterized as single-word R-marking. Although some examples have more than one juxtaposed Rmarked noun with an identical R-person/number, they seemingly do not form constituents except in (61c) and (62c). In contrast, there are some examples where two or more Rmarked nouns appear to form a constituent as in (61c) and (62c). Firstly, there are some examples which look similar to the adjective–noun construction in other languages such as English and Spanish. This type includes (61c), repeated here as (64b). This type of construction is found mainly in Christian literature and in the texts of missionary grammarians, and is attested both in principal (64a) and non-principal (64b) positions. In these examples, the adjectival nouns ti-nelli [2SG, true] and ti-we:i [2SG, big] appear to modify other R-marked nouns which follow them. (64) a. ka ti-nelli ti-teo:pišk-a:-c/inλi ti-nelli AFF 2 SGR -true 2 SGR -priest- L - HON ti-SACERDOTE 2SGR-true 2SGR-priest ‘YouSG are a real priest, a real sacerdote.’ [X] (La adoración de los Reyes, 308)62 b. ka ti-we:i ti-malakayo’ ti-počoλ AFF 2 SGR -big 2 SGR -shade 62 t-a:we:we:λ 2SGR-ceiba 2SGR-cypress Horcasitas (1974, I, 308). For the source of this example, see Note 71. 72 4 DISCUSSION Ø-mic/-mo-či:wi-lia in Ø-to-te:kwyo: 3SGS-2SGO-REFL-make-APPL IN 3SGR-1PLP-lord ‘Our lord makes (hon.) youSG as a great shade, a ceiba, a cypress.’ (FC VI, 58) It is indeed inherently difficult to distinguish them from mere noun juxtaposition, since such an example as (64b) can also be interpreted as ‘Our lord made youSG great, made youSG as a shade, ...’ Nonetheless, the order of nouns in (64a) implies that the two adjective–noun pairs form separate constituents, for it is the only way to understand a double occurrence of the adjectival noun nelli ‘true.’63 Another type of multi-word R-marking is found in figurative word pairing. In Classical Nahuatl, two juxtaposed nouns sometimes express a single concept. It has traditionally been called difrasismo and constitutes a stylistic feature of Classical Nahuatl, or of Mesoamerican languages in general (León-Portilla, 1985; Bright, 1990; Montes de Oca, 1997). For example, na:n-(λi) ta’-(λi) ‘mother and father’ in (65a) jointly means ‘parent(s).’ Similarly, kwa:wλi o:se:lo:λ ‘eagle and jaguar’ and kwe:iλ wi:pi:lli ‘skirt and blouse’ in (65b) respectively mean ‘warrior’ and ‘woman.’64 (65) a. ka a’=n-a:wiya a’=ni-wellamati AFF NEG =1 SGS -be.satisfied NEG =1 SGS -be.contented in ni-mo-na:n IN 1SGR-2SGP-mother in ni-mo-ta’ IN 1SGR-2SGP-father ‘I, yourSG mother and father (i.e. parent), am not satisfied, not contented.’ 63 Correspondingly, Launey (1994, 116–122) implies a close syntactic relationship between a prenominal modifier (“epithet”) and its head noun. 64 The pair a:wa’ tepe:wa’ ‘owner of water, owner of mountain’ (i.e. habitant of the city) in (62c) is also an example of difrasismo, which is based on a:-(λ) tepe:-(λ) ‘water and mountain’ (i.e. city). 73 4 DISCUSSION (FC VI , 107) b. nika:n ti-ka’ here in ti-kwa:wλi in t-o:se:lo:λ aw is ye te’wa:λ 2SGS-be IN 2SGR-eagle IN 2SGR-jaguar and here.is already 2SG in ti-kwe:iλ in ti-wi:pi:lli IN 2SGR-skirt IN 2SGR-blouse ‘Here youSG are, youSG eagle and jaguar (i.e. warrior); and here youSG are, youSG skirt and blouse (i.e. woman).’ (FC VI, 67) The examples in (65) imply that the two words in a difrasismo form a constituent and receive a joint R-marking. Such cases cannot be accounted for by assuming that they are separately R-marked. In (65a), both ‘mother’ and ‘father’ are R-marked with the first-person singular R-prefix n(i)-; similarly, in (65b), originally inanimate nouns are R-marked with the second-person R-prefix t(i)-. Thus, the examples in (65), along with those in (64), should be analyzed as examples of joint R-marking which targets two syntactically related nouns at a time. 4.3.3 Section summary This section proposes a three-way distinction between principal, intraeventive, and appositional R-marked nouns according to their environments. In addition, this section points out that multi-word R-marking can occur in both principal and non-principal noun clusters. The following sections discuss the characteristics of each type of R-marked nouns and show that they are syntactically not treated as subject–predicate complexes. 4.4 Multi-word R-marking Firstly, multi-word R-marking cannot be accounted for straightforwardly by the omnipredicative approach held by Launey (2004) and Andrews ([1975] 2003). 74 4 DISCUSSION As mentioned in Section 4.3.2, two nouns in a difrasismo have the identical R-prefixes, although each noun which constitutes a difrasismo is not always equated with the Rparticipant. In (66a), both ni-mona:n [1SG, yourSG mather] and ni-mota’ [1SG, yourSG father] are marked as first-person singular. Since (66a) is an instruction speech by a father, the first-person singular marking of ni-mona:n [1SG, yourSG mother] cannot be predicted by the omnipredicative analysis which assumes that ni-mona:n is a pre-formed subject–object complex. A similar inconsistency is observed in (66b), where both [1SG, yourSG mother] and [1SG, yourSG father] are marked as first-person plural although neither the adressee’s mother nor father can be equated to ‘we.’ (66) a. ka a’=n-a:wiya a’=ni-wellamati in ni-mo-na:n AFF NEG =1 SGS -be.satisfied NEG =1 SGS -be.contented IN 1SGR-2SGP-mother in ni-mo-ta’ IN 1SGR-2SGP-father ‘I, yourSG mother and father (i.e. parent), am not satisfied, not contented.’ (FC VI , 107) b. aw m-i:ško mo-kpak o:=ti-λačiš-ke’ in ti-mo-na:nwa:n and 2SGP-front 2SGP-above A=1PLS-watch-PST IN 1PLR-2SGP-mother ti-mo-ta’wa:n 1PLR-2SGP-father ‘And we, yourSG mother and your father, keep an eye on youSG , in front of youSG ’ (Plát. 231) Such examples as (66) can be properly accounted for only if the process of R-marking applies after sentence composition. That is, in order for the two nouns in a difrasismo to have a joint R-marking, they should be combined before the realization of R-marking. If each R-marked noun has a predicative structure before sentence composition, these 75 4 DISCUSSION examples would need an additional morphosyntactic or figurative explanation. The examples in (67) display this fact more clearly. The difrasismos kwa:wλi o:se:lo:λ ‘eagle and jaguar’ and kwe:iλ wi:pi:lli ‘skirt and blouse’ in (67a) respectively mean ‘warrior’ and ‘woman.’ Likewise, kwiλapilli a’λapalli ‘tail and wing’ in (67b) means ‘commoner.’ In these examples, each R-marked noun is not even a human, and means human only in a difrasismo. (67) a. nika:n ti-ka’ here in ti-kwa:wλi in t-o:se:lo:λ aw is ye te’wa:λ 2SGS-be IN 2SGR-eagle IN 2SGR-jaguar and here.is already 2SG in ti-kwe:iλ in ti-wi:pi:lli IN 2SGR-skirt IN 2SGR-blouse ‘Here youSG are, youSG eagle and jaguar (i.e. warrior); and here youSG are, youSG skirt and blouse (i.e. woman).’ b. in ti-čo:kili:lo:ni (FC VI, 67) in ti-kwiλapilli in t-a’λapalli ma:noso i:wiya:n IN 2SGR-lamentable IN 2SGR-tail IN 2SGR-wing if.only quietly ma:noso yo:ko:ška: in ši-mo-nemi-ti if.only peacefully IN 2SGS-REFL-live-CAUS.OPT ‘Live (hon.) quietly and peacefully, youSG lamentable [woman], youSG tail and wing (i.e. commoner).’ (Prim. 242) Thus, although the omnipredicative analysis presented by Launey (1994) and Andrews ([1975] 2003) can unproblematically account for typical examples of appositional Rmarked nouns, it fails to explain the examples where R-preson/number does not conform to the referent of S. In such examples, the R=S scheme does not apply. The behavior of multiple R-marking suggests another important feature of R-marking. That is, it implies that R-marking targets noun phrases, but not individual nouns (zerolevel categories) in reality. In other words, it is not the case that each noun has its own 76 4 DISCUSSION R-person/number pre-syntactically. 4.5 Intraeventive R-marked nouns This section discusses the syntactic and semantic aspects of intraeventive R-marked nouns and argues that they are syntactically treated as nouns, but not as subject–predicate complexes, in sentence composition. Specifically, although they have a dyadic structure [R, S] at the morphological level, only one of these two parts, namely S, is significant in syntax. First of all, they cannot be directly accounted for by Andrews’ ([1975] 2003) omniclausal analysis. Although Launey (1994) correctly points out that they do not conform to the simple cross-referencing model and that they form complex predicates with their principals, he either does not seem to provide an expilcit formal analysis of them. This section presents two possible hypotheses concerning the syntactic status of intraeventive R-marked nouns, namely omnipredicative and non-omnipredicative ones, and claims that the latter can account for the phenomenon more consistently. Following this second hypothesis, the syntactic status of the S of an intraeventive R-marked noun is parallel to that of the resultative nominals in other languages. 4.5.1 Problems with the simple omniclausal analysis In some cases, the relationship between an intraeventive R-marked noun and its principal verb does not conform to the cross-reference scheme presented by Andrews ([1975] 2003). As pointed out in Launey (1994, 110–114), intraeventive nouns form an essential and inseparable part of the event denoted by the principals. In addition, the exact meaning of intraeventive R-marked nouns themselves cannot be understood until they are assigned propositionality by their principals. These observations are inconsistent with Andrews’ ([1975] 2003) view that non-principal nouns are “supplements” to the person marker of 77 4 DISCUSSION the principals. The inseparability of intraeventive R-marked nouns from the events is shown by the fact that some verbs change their meanings according to whether they are accompanied by intraeventive nouns or not. For example, the copular verb ka’ is originally an existential verb as in (68a), and is also used as a support verb of locative constructions when it is not accompanied by a complement noun as in (68b). In contrast, it serves almost purely as a copula when combined with a complement noun and its existential meaning is inactivated, as shown in (68c) below. (68) ka’ ‘to be’ a. o:mpa Ø-kat-ka in Ø-i:-i:ši:pλa’ Ø-λa:lok there 3SGS-be-PST IN 3SGR-3SGP-representative 3SGR-PN ‘There was the image of Tlaloc there.’ b. a’mo: ye NEG (FC II, 179) o:mpa alačina’ ti-ka’ already there LN 2SGS-be ‘Wouldn’t youSG be in China?’ [G] Carochi (1645, f. 105r) c. ... in λatilo:lko Ø-λa’toa:ni Ø-i:-to:ka: IN LN Ø-kat-ka Ø-λa:kateo:c/in 3SGR-ruler 3SGR-3SGP-name 3SGS-be-PST 3SGR-PN ‘The name of the ruler of Tlatelolco was Tlacateotzin.’ (Chim. 32:23–24) The transitive verb či:w(a) ‘to make X’ and its reflexive counterpart mo-či:w(a) ‘to occur’ show a similar alternation. The verb či:w(a) ‘to make X’ is abundantly attested as a basic verb denoting producing or fabricating something as in (69a), and mo-či:w(a) ‘to occur’ most frequently means ‘for something to occur or come into being’ as in (70a)– (70b). However, they respectively mean ‘to change X into Y’ and ‘to change (oneself) into Y’ when they take resultative complements as in (69b) and (70c). (69) či:w(a) ‘to make X’ 78 4 DISCUSSION a. kil in Ø-okλi Ø-ki-či:w it.is.said 3SGS-3SGO-make.PST IN 3SGR-pulque ‘It is said that he made pulque.’ b. ma: Ø-mo-yo:llo: IRR [X] (FC I, 74) Ø-mo-na:n Ø-mo-ta’ 3SGR-2SGP-heart 3SGR-2SGP-mother 3SGR-2SGP-father ti-k-či:w 2SGS-3SGO-make.ADM ‘YouSG should not make yourSG heart yourSG mother and father (i.e. parent).’ (Prát. 245) (70) mo-či:w(a) ‘to become’ a. Ø-co:stik Ø-teo:kwiλaλ ka λa:l-la:n in Ø-mo-či:wa 3SGR-yellow 3SGR-gold AFF ground-at IN 3SGS-REFL-make ‘Gold: it is on the ground that it is yielded.’ b. niman ye then i:k Ø-mo-či:wa (FC XI, 233) in Ø-ya:o:yo:λ already thus 3SGS-REFL-make IN 3SGR-war ‘Thereupon the battle occurs.’ c. ... Ø-λa’toa:ni Ø-mo-či:w (FC XII, 61) tenočtiλan 3SGR-ruler 3SGS-REFL-make LN ‘... and he became the ruler of Tenochtitlan.’ (Crón. 38) As another corroboration, Launey (1994, 111) points out that the quasi-article in does not intervene between an intraeventive noun and its principal. The data for this thesis also support his observation. This fact may reflect the close syntactic relationship between these two items. Moreover, Andrews’ ([1975] 2003) omniclausal analysis often fails to account for the semantic interpretation of individual intraeventive R-marked nouns. Consider first the 79 4 DISCUSSION examples in (71), where intraeventive R-marked nouns are emphasized by bold-face type. (71) a. ni-koko:ka-wa’ ni-kat-ka 1SGR-property-PD 1SGS-be-PST ‘I was rich.’ b. in λa: o:=n-on-mik IN if [G] (Molina, 1571, 24) te’wa:λ ti-n-i:ši:pλa’ A =1 SGS - TRSL-die. PST 2 SG 2SGR-2SGP-representative ti-ye-s 2SGS-be-FUT ‘If I die (lit. died), then youSG will be my successor.’ (Chim. 33:21–22) In Andrews’ ([1975] 2003) analysis, (71a) would be paraphrased as ‘I-am-rich I-was,’ the first word ni-koko:kawa’ [1SG, rich] being an attributive clause which can itself stand as an entire sentence which means ‘I am rich.’ Similarly, according to him, the second part (consequent) of (73b) would be analyzed as ‘you-are-my-successor you-will-be.’ However, these two sentences do not state that the situations denoted by the intraeventive nouns are realized at the time of utterance. Though the R-marked noun ni-koko:kawa’ [1SG, rich] in (71a) means ‘I am rich’ when it stands alone, the two-word sentence (71a) as a whole cannot mean that the speaker is actually rich.65 Equally, ti-ni:ši:pλa’ [2SG, my successor] in (71b) never entails that the addressee is already the successor. Thus, intraeventive R-marked nouns does not entail R=S as such. More generally, intraeventive R-marked nouns cannot always be self-contained copulatic expressions. If intraeventive R-marked nouns are to be analyzed as copulatic expressions, it is almost impossible to specify their truth conditions. If an R-marked noun ni-koko:kawa’ [1SG, rich] is a copulatic clause which serves as the “supplement” to the subject marker on the copular verb nikatka ‘I was,’ it should mean ‘I, who am 65 Classical Nahuatl past (preterit) tense often implies that the event denoted by the stem has already ended. 80 4 DISCUSSION rich, existed’. However, this appositional interpretation does not conform to the overall meaning of (71a).66 Therefore, intraeventive R-marked nouns cannot be understood by assuming a simple process of supplementation. Rather, they appear to form complex predicates with their principals and play an essential role in the event denoted by them. The question arises, then, as to how the relationship between the R-prefix and the stem can be interpreted. If [R, S] does not imply R=S, what does the pair [R, S] exactly mean? Launey (1994, 110–114) is aware of the fact that these “attributive constructions” cannot be directly accounted for by a simple cross-reference model and points out a few characteristics which distinguish them from usual cross-referencing. Instead, he suggests that intraeventive R-marked nouns and their principals form complex predicates. There are at least two possibilities to interpret the syntactic and semantic status of intraeventive R-marked nouns. The first is to maintain the omnipredicative analysis syntactically and posit a special rule concerning the semantic interpretation of the pair [R, S]. The second is to ignore completely the relationship between R and S at the syntactic level. Although the first approach may appear intuitive at first sight, the discussions which follow show that there are some observations favorable to the second model. 4.5.2 Omnipredicative hypothesis The first, omnipredicative interpretation of intraeventive R-marked nouns is the extension of the omnipredicative analysis presented by Launey (1994, 110–114). He maintains that also intraeventive R-marked nouns have predicative structures (Launey, 1994, 54), while he states that they are combined with their principals and form complex predicates 66 One may assume that this appositional analysis is not completely impossible, since John was rich can be paraphrased as there was John who was rich in a broad interpretation. However, this interpretation misses the fact that overt copulae and principal R-marked noun are in complementary distribution and non-present/indicative copulae are often used in purely copulatic senses. 81 4 DISCUSSION (Launey, 1994, 111). Launey (1994, 60, 110) suggests that overt copulae in such examples as (71), repeated here as (72), serve as supporting verbs which supplement the lack of tense in nominal predicates, since nouns cannot inflect for tense. (72) a. ni-koko:ka-wa’ ni-kat-ka 1SGR-property-PD 1SGS-be-PST ‘I was rich.’ b. in λa: o:=n-on-mik IN if [G] (Molina, 1571, 24) te’wa:λ ti-n-i:ši:pλa’ A =1 SGS - TRSL-die. PST 2 SG 2SGR-2SGP-representative ti-ye-s 2SGS-be-FUT ‘If I die (lit. died), then youSG will be my successor.’ (Chim. 33:21–22) It seems possible to paraphrase Launey’s (1994) analysis as follows: an intraeventive R-marked noun has a predicative structure [R, S] and means that the equational or attributive relationship R=S is established (or is not established, in negative clauses) at some point in time. The supporting verb (i.e. principal) assigns the time or situation when the relationship R=S becomes true. R=S is true in the past in (72a) and in the future in (72b), for their principal verbs are in the past and the future tense respectively. Thus, intraeventive R-marked nouns [R, S] becomes propositional with the help of their principal verbs. This interpretation can also be applied to modal expressions. The word ni-kwalli [1SG, good] in (73a) does not mean that the speaker is actually good, but that the speaker is good in the imaginary situation mentioned by its principal, the optative copula o:niyeni. Similarly, (73b) does not mean that the addressee is actually sap (R=S), but the principal (optative past copula šiyeni) implies that R=S is established in an imaginary space. In both cases, the condition of R=S is assigned by the principals. 82 4 DISCUSSION (73) a. ma: ni-kwalli IRR o:=ni-ye-ni 1SGR-good A=1SGS-be-OPT. PST ‘Would that I had been good.’ [G] (Carochi, 1645, f. 38r) b. in ma: t-oko:c/oλ ši-ye-ni IN IRR 2SGR-sap 2SGS-be-OPT. PST ‘as if youSG were sap’ [V] (Cant. 16r:20) This analysis can be extended to other verbs than the copular verb ka’ with resultative intraeventive R-marked nouns, for example či:w(a) ‘make’ (Launey, 1994, 112). (74), for example, states that the predicative relationship encoded in the intraeventive R-marked noun ti-te:kwa:ni [2SG, beast] would become R=S as the result of the event denoted by the principal timoči:w ‘(youSG should not) become.’ (74) ma: ye IRR iwk=in ti-te:kwa:ni ti-mo-či:w already like=IN 2SGR-beast 2SGS-REFL-make.ADM ‘YouSG should not become like a beast.’ (FC VI, 52) Following the terminology defined in Section 4.2, this analysis assumes that an intraeventive R-marked noun is a predicative expression which is not propositional until it is provided with the condition of equation/attribution by its principal verb. In other words, an intraeventive R-marked noun is not itself an equational or attributive clause; rather, it forms a complex expression combined with its principals without which it cannot obtain a propositional force. This omnipredicative analysis is parallel to the “small clause” analysis developed in generative grammar. Following this analysis, the examples (75a) and (75b) can be analyzed as (76a) and (76b) respectively. The small clauses [SC R, S] themselves say nothing about their propositional status, which is to be assigned by their principal verbs. 83 4 DISCUSSION (75) a. ni-koko:ka-wa’ ni-kat-ka 1SGR-property-PD 1SGS-be-PST ‘I was rich.’ [G] (Molina, 1571, 24) b. ka ... ti-počoλ ... Ø-mic/-mo-či:wi-lia ... AFF 2SGR-ceiba 3SGS-2SGO-REFL-make-APPL ‘He (our lord) makes (hon.) youSG as ... a ceiba ...’ (76) a. [ [SC 1SG, rich ] b. [ [SC 2SG, ceiba ] (FC VI, 58)67 I-was ] he-makes-youSG ] 4.5.3 Non-omnipredicative hypothesis In contrast to the omnipredicative analysis presented above, this thesis proposes another hypothesis which does not posit a clausal structure inside intraeventive R-marked nouns. This analysis is a more purified version of Launey’s (1994) intuition that an intraeventive R-marked noun and its principal verb altogether form a complex predicate. That is, it assumes that an intraeventive R-marked noun, or at least its S, is a part of a predicate; in other words, it does not contain a subject. More specifically, this analysis hypothesizes that intraeventive R-marked nouns behave as nouns as though they were not accompanied by R-prefixes. Although they appear to be subject–predicate complexes ([R, S]) at the morphological level, it is considered that it is only S that is significant in sentence composition. Contrarily to the omnipredicative interpretation (76), this second approach interprets the examples (75a) and (75b) as (77a) and (77b) respectively. (77) a. [ I [VP rich be ] ] b. [ he youSG [VP ceiba make ] ] 67 This example is a simplified version of (61c) above. 84 4 DISCUSSION Of course, there is an obvious problem with the analysis of (76). (76) does not capture the fact that intraeventive nouns are redundantly marked with R-prefixes. Nonetheless, a closer look at intraeventive R-marked nouns suggests that the interpretation of (76) is not as unnatural and artificial as it appears at first glance. Firstly, as already illustrated above, R-marking in intraeventive nouns are redundant unlike in principal R-marked nouns. They always have the same person/number with their principals. In this respect, intraeventive R-marked nouns are in a strong contrast with appositional R-marked nouns discussed in Section 4.6 below, which sometimes have different R-person/number from their principals. Therefore, it is not difficult to explain intraeventive R-marking by some formal process analogous to the gender or number concord in other languages. Alternatively, an intraeventive R-marked noun might be analyzed as a noun phrase headed by S and accompanied by an appositional bound-form pronominal item, namely an R-prefix. In either understanding, it is only S, but not R nor [R, S], that is essential in sentence composition. Secondly, intraeventive R-marked nouns behave differently from clauses in crossreference. As has been repeated above, intraeventive R-marked nouns have the same person/number as the theme argument (subject/object) of their principals. In (78), the principals are marked as second-person singular, which is the person/number of the theme (someone who becomes S or someone who is made into S). If intraeventive R-marked nouns are clausal complements of principals, then these “intraeventive clauses” are crossreferenced with the second-person singular prefixes on the principals. (78) a. ti-kimil-e’ ti-ka:ka:š-e’ ti-ye-s 2SGR-package-PD 2SGR-packframe-PD 2SGS-beFUT ‘YouSG will become a bearer of a burden, bearer of a packframe.’ (FC VI, 185) 85 4 DISCUSSION b. ka ti-we:i ti-malakayo’ ti-počoλ AFF 2 SGR -big 2 SGR -shade t-a:we:we:λ 2SGR-ceiba 2SGR-cypress in Ø-to-te:kwyo: Ø-mic/-mo-či:wi-lia 3SGS-2SGO-REFL-make-APPL IN 3SGR-1PLP-lord ‘Our lord makes (hon.) youSG as a great shade, a ceiba, a cypress.’ (FC VI, 58) However, in Classical Nahuatl, clauses are generally cross-referenced with thirdperson singular markers as shown in (79). If intraeventive R-marked nouns are clauses, there is no reason for the principals in (78) to have second-person markers.68 (79) a. ni-k-neki ni-λa-po:wa-s 1SGS-3SGO-want 1SGS-UNO-read-FUT ‘I want to read something.’ b. ka ni-k-i’toa [G] (Carochi, 1645, f. 41v) λa: ši-k-kaki ... AFF 1 SGS -3 SGO -say IRR 2 SGS -3 SGO -hear. OPT ‘I say, “(YouSG ) listen!”’ c. ti-k-mati’ (Chim. 38:24–25) ka mo-λa:n-c/in-ko ti-kate’ 1PLS-3SGO-know AFF 2SGP-at-HON-LOC 1PLS-be ‘We know that we are with youSG .’ [X] (FC I, 62) More generally, the omnipredicative approach requires that a subject–predicate complex receive a semantic role as if it were a DP/NP or a pronoun, but not as a subject– predicate complex as such. Contrarily, if the intraeventive nouns in (78) are assumed to form complex predicates with their principals (‘to burden-PD-become,’ ‘to ceiba-make,’ etc.), the person marking pattern in question is successfully accounted for, since this 68 This problem may be avoidable under Launey’s (1994) assumption that a cross-referencing relationship is established between two person markers rather than between a person marker and a word. 86 4 DISCUSSION interpretation predicts that there is no real cross-referencing at all between intraeventive nouns and their principals. This non-omnipredicative view is also supported by the pattern of noun incorporation. In Classical Nahuatl, it is sometimes the case that only the stem of intraeventive nouns are incorporated into the principal verbs.69 The nouns which typically appear as intraeventive R-marked nouns are sometimes incorporated into the principals without R. If one excludes R from the syntactic structure, this alternation can be unproblematically predicted. (80) a. Ø-k-i’toa:-ya’ Ø-mo-teo:-kwepa’ 3PLS-3SGO-say-IMPF 3PLS-REFL-god-turn ‘They used to say that they [women who died in childbirth] become gods’ [X] (FC I, 71) b. Ø-ame:č-yo:llo’λawe:li:lo:k-a:-kwepa Ø-ame:č-čiči-kwepa in Ø-okλi 3SGS-2PLO-crazy-L-turn 3SGS-2PLO-dog-turn IN 3SGR-pulque ‘Pulque makes youPL crazy, makes youPL like dogs.’ [G] (Carochi, 1645, f. 113v) c. Ø-mo-λa:ka-kwepa’ in neči’či:walis-ti-ka in neši:n-ti-ka 3PLS-REFL-person-turn IN outfit-L-by λa’to:l-ti-ka Ø-kin-λa-e’e:kalwia’ in IN hair.style-L-by IN in Ø-ča:n-e’ke’ word-L-by 3PLS-3PLO-UNO-imitate IN 3PLR-home-PD ‘They disguise themselves; they imitate local habitants with outfit, hair style, and language.’ (FC I, 42) The non-omnipredicative account is also favorable from a cross-linguistic point of 69 In addition to the examples cited in (80), a frequently attested transitive verb ya:o:-či:w(a) (enemy-make) ‘to fight with X, to do battle with X’ might be an example of this type (i.e. ‘to make X into an enemy’); however, one must be careful in handling this example, since the noun root ya:o:- ‘enemy’ often means ‘war’ in compounds. 87 4 DISCUSSION view. In the majority of world’s languages, the complements of copulae and the resultative complements of verbs are assumed to be encoded by DP/NPs. As repeatedly mentioned above, even within the Nahuan group, some languages lack intraeventive R-marking despite that they have many other polysynthetic features in common with Classical Nahuatl. Michoacán Nahual, for example, systematically lacks intraeventive R-marking. (81) Michoacán Nahual a. amo ši-ye NEG IMP -be tonto silly ‘Don’t be (youSG ) silly.’ b. wan ti-ye-k (Sischo, 1979, 320) gwače and 2SGS-be-PRF soldier ‘And youSG were a soldier?’ (Sischo, 1979, 320) This fact can be accounted for if it is assumed that only S is significant in sentence composition, since this hypothesis minimizes the difference in morphosyntax between Classical Nahuatl and the languages without obligatory intraeventive R-marking. 4.5.4 Section summary It is not necessary to assume that intraeventive R-marked nouns have a predicative structure at the syntactic level. Rather, their meaning and their interaction with the principals can be accounted for more consistently by assuming that they are treated as nouns. Although morphologically they have a dyadic structure [R, S], it is likely that only S is relevant in sentence composition. Unlike in principal nouns, R-prefixes in intraeventive nouns do not seem to play a substantial role. R-prefixes in these environments can either be interpreted as a mere by-product of a formal process of concord, or as appositional pronominal items which do not participate in the basic syntactic process. 88 4 DISCUSSION 4.6 Appositional R-marked nouns The previous section proposed the interpretation that intraeventive R-marked nouns are not treated as pre-formed subject–predicate complexes in Classical Nahuatl syntax. This section, in turn, shows that appositional R-marked nouns are equally non-propositional at least in several cases. The omnipredicative assumption that they retain the predicative structure through the process of sentence composition sometimes fails to account for their distribution. 4.6.1 Preliminary notions Firstly, it should be clarified whether appositional R-marked nouns are really appositional. Yoshiho Yasugi (personal communication) suggests that such sentences as (82) can be analyzed as a mere sentence juxtaposition (‘I am not satisfied. I am not content. I am yourSG mother. I am yourSG father.’) rather than a syntactic configuration. (82) ka a’=n-a:wiya a’=ni-wellamati AFF NEG =1 SGS -be.satisfied NEG =1 SGS -be.contented in ni-mo-na:n in IN 1SGR-2SGP-mother IN ni-mo-ta’ 1SGR-2SGP-father ‘I, yourSG mother and father (i.e. parent), am not satisfied, not contented’ (FC VI , 107) Indeed, this alternative interpretation cannot be excluded in the majority of examples. Parallelism and sentence juxtaposition constitute the stylistic characteristics of Nahuatl native literature (León-Portilla, 1985; Bright, 1990). Nevertheless, there are some examples of appositional R-marked nouns which cannot be interpreted as juxtaposed sentences. In the examples below, for instance, object R-marked nouns intervene between the irrealis particle ma: and the principal verb. 89 4 DISCUSSION (83) a. ma: Ø-kokolisλi IRR ti-k-mo-kwi:-ti’ 3SGR-sickness 2SGS-3SGO-take-CAUS . ADM ‘Do not get (youSG , hon.) sick.’ b. ma: Ø-ti:sa-c/inλi IRR (Prim. 295) Ø-i’wi-c/inλi to-k-on-ti:λanika:n 3SGR-chalk-DIM 3SGR-feather-DIM 1PLS-3SGO-TRSL-send.OPT ‘Let us send chalk and feathers (i.e. sacrifice).’ (Anales de Tlatelolco, f. 8r)70 More generally, the quasi-article in in such examples as (82) above indicates that these appositional R-marked nouns are not separate sentences, since in does not appear in sentence principals (contrainte de non-définition; Launey, 1994, 59). Another preliminary problem is whether the use of non-third-person appositional R-marked nouns constitutes a significant part of the grammar of Classical Nahuatl. Since non-third-person marking on argument nouns may appear highly marked and unnatural from the viewpoint of other languages, it should be confirmed that it is not limited to a particular style or type of texts. Indeed, non-third-person appositional R-marked nouns are attested overwhelmingly in huēhuehtlahtōlli (some chapters in Florentine Codex VI, Plática by Olmos, etc.) and poems (Cantares mexicanos, etc.), both of which belong to formal, figurative styles of speech. They are, however, also amply attested in postconquest historical literature and even in twentieth-century narratives (as in (34)). It therefore seems natural to assume that it was still active as a part of the synchronic grammar at the period of Classical Nahuatl. Thus, this thesis considers appositional R-marked nouns as parts of sentences rather than self-contained sentences, and their use as a genuine part of the synchronic grammar of Classical Nahuatl. 70 Tena (2004, 62). This example is taken from a manuscript from the first half of the sixteenth century. 90 4 DISCUSSION 4.6.2 Counterexamples to the omnipredicative analysis As repeatedly mentioned earlier, both Launey (1994) and Andrews ([1975] 2003) analyze appositional R-marked nouns as subordinate predicative units, namely subject–object complexes. In Launey’s (1994) model, the appositional use of R-marked nouns is a derived one from the principal use. Andrews ([1975] 2003), too, considers appositional R-marked nouns as supplementary NNCs, which are themselves minimal clauses. Both models assume that those pre-formed subject–object complexes are combined into larger constituents to form multi-word sentences. However, while most appositional R-marked nouns can be interpreted as subordinate predicative units without any trouble, there are some examples which the omnipredicative analysis fails to predict. A clear counterexample to the omnipredicative analysis is the R-marking on quantificational nouns moči and i:škič (in plural forms moči:ntin and i:škičtin respectively), both of which mean ‘all,’ ‘entire,’ or ‘every.’ As intuitively expected, when they have first- or second-person R-prefixes, these nouns mean ‘all of us’ or ‘all of youPL .’ (84) a. ka san wel i:-sel-c/in AFF Ø-te:č-mo-maki-lia in only well 3SGP-alone-HON 3SGS-1PLO.3SGO-REFL-give-APPL IN t-i:škičtin sema:na:wak ti-λa:ka’ 1PLR-entire world.LOC Ø-kiyawiλ ... 1PLR-person 3SGR-rain ‘Only He [God] gives (hon.) rain to all of us people in the world.’ [X] (FC I, 68) b. ma:kamo: ti-to-λapolo:ltika:n IRR . NEG in t-i:škičtin ti-mo-λa:kawa:n in 1PLS-REFL-lose.mind.OPT IN 1PLR-entire 1PLR-2SGP-slave IN ti-mo-ma:se:walwa:n in ti-JERUSALÉN-λa:ka’ 1PLR-2SGP-vassal in 1PLR-LN-person 91 4 DISCUSSION ‘Let it be that all of us, yourSG slaves, yourSG vassals, people in Jerusalem, do [X] (La adoración de los Reyes, 302)71 not lose our mind.’ c. i:-wa:n in am-i:škičtin in am-me:ši’ka’ in an-tenočka’ nika:n 3SGP-with IN 2PLR-entire IN 2PLR-EN IN 2PLR-EN here an-ki-mati-ske’ ... 2PLS-3SGO-know-FUT ‘And here all of youPL , Mexica Tenochca, will know that ...’ d. ka o:=ti-senki:sa-ko in ye AFF A =2 SGS -assemble- PURP. CISL t-i:škič (Crón. 62) in ye IN already 2SGR-entire IN already ti-moči 2SGR-all ‘Every one of you (lit. youSG ) has come together.’ e. o: no-pilwa:n-é INTERJ t-i:škič in ye ti-moči in 1SGP-child-VOC IN already 2SGR-entire IN already 2SGR-all IN nika:n t-on-ok here in ye (FC IX, 52) in ti-siwa:λ 2SGS-TRSL-lie IN 2SGR-woman ‘Oh my daughters, every one of you women (lit. youSG woman) who are here’ (Prim. 242) If R-marked nouns are pre-formed as subject–predicate complexes before they are combined into larger constituents, the R-marked quantificational nouns in the examples (84) should be expected to yield the meaning ‘we are all’ or ‘youPL are all,’ as reflected in the glosses given by Andrews ([1975] 2003, 132–134). This omnipredicative intrepretation is inconsistent with the overall meanings of the sentences in (84). 71 Horcasitas (1974, I, 302). This example is taken from a sixteenth-century (according to the editor) Catholic play. 92 4 DISCUSSION In traditional terms, a quantificational expression states the relationship between two sets of entities: the proposition all jaguars have spots states that the set of entities which have spots contains the set of entities which are jaguars. Thus, if the expression t-i:škičtin [1PL, entire] is to be interpreted as ‘we are all,’ then it would mean that the group of people denoted by ‘we’ include all the people mentioned in the current context. This feature of i:škič ‘entire’ is observed in (85a), where i:škič states that (the speaker is asking if) the set of events which occurred so far contains every event which is expected in this context. In (85b), similarly, i:škič expresses that (the speaker is asking if) the gold appeared so far includes all the gold which is mentioned. (85) a. kwiš san aokmo: kwiš ye Q only no.more Q Ø-i:škič already 3SGR-entire ‘Is it to be no more? Is it all?’ b. san ye Ø-i:škič (FC VI, 3) in i:n Ø-teo:kwiλaλ in o:=Ø-mo-piya-ya only already 3SGR-entire IN this 3SGR-gold IN A=3SGS-REFL-keep-IMPF me:ši’ko LN ‘Is this gold all which was stocked in [the city of] Mexico?’ (FC XII, 125) The quantificational function of ‘all’-type nouns is different in (84a)–(84d) above. In (84a), for example, t-i:škičtin [1PL, entire] does not state the relationship between the R-participants (‘we’) and all the people available in this context; rather, it establishes the relationship between the set of R-participants, which equals to the set of ‘people in this world,’ on one hand, and the set of entities which is given rain by God on the other. Moreover, in (84c) and (84d), the noun i:škič ‘all, entire’ is in the singular form, which is by no means all of the people addressed in this situation. Thus, if R-marked nouns are pre-formed copulatic expressions, the examples in (84) cannot be accounted for. 93 4 DISCUSSION “Some”-type quantifiers also give a counterexample to the omnipredicative analysis. In Classical Nahuatl, such expressions as ‘some of us people’ or ‘one of us men’72 sometimes involve the non-third-person R-marking on the nouns which denote, in informal words, the “population” (in the sense of statistics) or the “universal set.”73 (86) a. in a:k in te:-i:špan Ø-c/ik-kwa’-kwa in t-okičtin IN who IN UHP-front 3SGS-chicle-RDP-eat IN 1PLR-man ‘someone out of us men who chews chicle in public’ b. in λa: aka’ t-okičtin i:-pan (FC X, 90) Ø-siwa:-koči-s IN if someone 1PLR-man 3SGP-LOC 3SGS-woman-sleep-FUT ‘if anyone out of us men should sleep with an woman ...’ c. Ø-se: Ø-siwa:λ nika:n ti-λa:ka’ 3SGR-one 3SGR-woman here (FC I, 31) in Ø-kin-wa:l-wi:ka-k 1PLR-person IN 3SGS-3PLO-CISL-lead-PST in Ø-wa:l-na:waλa’to’-ti-ya’ IN 3SGS-CISL-interpret-L-go.PST ‘a woman out of us people here (i.e. an indigenous woman) who led them hither, headed hither serving as an interpreter’ d. Ø-se:me’ as-siwa’ ši-k-mo-kwiλawi:ka:n (FC XII, 25) in 3PLR-one 2PLR-woman 2PLS-3SGO-REFL-take.care.OPT IN 72 Carochi (1645, f. 85r–85v) notes that the obligatory use of t-okičtin [1PL, man] ‘we men’ instead of Ø-okičtin ‘men,’ which is exemplified by (86a) and (86b), is a stylistic feature limited to the speech by male speakers. However, as (86d) shows, the use of R-prefixes in ‘one of us’ type of construction is not limited to this stylistic use of t-okičtin. Carochi’s (1645) observation should therefore be interpreted that, although the context-independent and obligatory use of t(i)- on okičλi is a mere stylistic custom of male speakers, the non-third-person R-marking itself is possible regardless of the style or the gender of the speaker when the speaker wants to imply that the noun in question is coreferent to the speaker’s or the addressee’s group. 73 The plural marking on se: ‘one’ in (86d) is discussed in Section 4.6.3. 94 4 DISCUSSION Ø-kokošk-a:-c/inλi 3SGR-sick.person-L-HON ‘Let one of youPL women take care (hon.) of the sick person.’ [G] (Carochi, 1645, f. 85v) What is problematic with these examples is that the appositional R-marked nouns do not correspond to any particular person affix of their principals. In typical examples of appositional R-marking, R-marked nouns appear to be cross-referenced with a person marker with the same person/number. For instance, in a canonically cross-referenced example (87), the appositional R-marked noun ti-λa:ka’ [1PL, person] is properly crossreferenced with a first-person plural reactive object prefix te: č- on the principal verb te: čkokolia ‘he hates us.’ This canonical cross-referencing is shown in Figure 10 below. (87) in i:k Ø-te:č-kokolia in λa:l-t-ikpak ti-λa:ka’ IN thus 3SGS-1PLO-hate IN ground-L-above 1PLR-person ‘He hates us earthly people’ [X] (FC I, 76) Cross-referencing Ø- teːč- -kokolia 1PLO ti- -λaːka’ 1PL Figure 10: Cross-referencing structure of (87) (simplified) In contrast, in such examples as (86c) (partially repeated below as (88)), the appositional R-marked noun does not have a person marker with which it can be crossreferenced. While the appositional R-marked noun ti-λa:ka’ [1PLS-person] is marked as first-person plural, all other nouns and verbs in (88) are marked as third person. (88) Ø-se: Ø-siwa:λ nika:n ti-λa:ka’ 3SGR-one 3SGR-woman here 1PLR-person ‘a woman out of us people here (i.e. an indigenous woman)’ 95 (FC XII, 25) 4 DISCUSSION Thus, the omnipredicative analysis cannot capture the R-marking on both ‘all’-type and ‘one/some’-type quantificational nouns. This failure of cross-referencing is illustrated in Figure 11 below. × Ø- -seː 3SGR Cross-referencing Ø- × -siwaːλ ti- 3SGR -λaːka’ 1PL Figure 11: Cross-referencing structure of (88) (simplified) Similarly, the omnipredicative model collapses when a numeral expressions are themselves R-marked. As many other types of nouns, numerals in Classical Nahuatl may have R-marking. (89) aw in te’wa:ntin ka t-o:meštin ti-t-o:linia’ and IN 1PL AFF 1 PLR -two 1PLS-REFL-move ‘And as for us, we two are in trouble.’ [X] (El sacrificio de Isaac, 240)74 In contrast to (86c) and (86d) above, where ‘one’ does not have overt R-marking, Carochi (1645, f. 85r–85v) reports that the numeral nouns are sometimes non-obligatorily R-marked as first or second person in such expressions as ‘one of us,’ ‘one of youPL ,’ etc. (90) a. ma: ti-se:me’ te’wa:ntin ti-wia:n IRR 1PLR-one 1PL 1PLS-go.OPT ‘Let one of us go.’ [G] (Carochi, 1645, f. 85v) b. ma: as-se:me’ ame’wa:ntin an-λa’piške’ IRR 2PLR-one 2PL 2PLR-house.servant 2PLS-UNO-sweep.OPT ‘Let one of youPL house servants sweep.’ 74 ši-λa-čpa:naka:n [G] (Carochi, 1645, f. 85v) Horcasitas (1974, I, 240). This example is taken from a Catholic play in 1539. 96 4 DISCUSSION Although the non-obligatoriness of R-marking in such environments should also be accounted for, at least such examples as (90) cannot be explained by the equational model; in these examples, R-marked numerals such as ti-se:me’ [1PL, one] clearly cannot be paraphrased as ‘we are one’ etc.75 4.6.3 Non-omnipredicative analysis The problems with these examples come from the presupposition that appositional nouns are already R-marked (i.e. they form subject–predicate complexes) before sentence composition and thereafter put together into larger syntactic constituents. Contrarily to this omnipredicative approach, this thesis proposes an alternative interpretation which assumes that R-marking are morphologically realized after sentence composition. In other words, this thesis claims that appositional R-marked nouns are not treated as pre-formed subject–predicate complexes in syntax. Let us begin with the basic intuition that R-marking is related to some discoursereferential status of the S, but cannot be captured simply by assuming the R=S scheme. The overall meaning of an R-marked noun t-i:škičtin [1PL, entire] ‘all of us’ in (84a) is clearly related to its R-participant (‘we’) in some way or other, but this relationship is not the same as ‘we are all’ or ‘we, who are all.’ This relationship is not as easy as it looks to formalize. As shown in the previous section, there are some examples where non-third-person R-marked nouns do not have corresponding principals which have the person affixes of the same person/number. In order to account for their distribution, it is necessary to abandon the assumption that such nouns as t-i:škičtin [1PL, entire] are pre-formed subject–predicate complexes. 75 This reasoning is slightly unfair, since it seems to be a basic grammatical feature of Classical Nahuatl to assign plural marking to such nouns as ‘one,’ ‘some,’ etc. This issue is discussed in more detail in Section 4.6.3. 97 4 DISCUSSION The scenario is as follows: such a combination as t-i:škičtin ti-λa:ka’ ([1PL, entire] [1PL, person]) ‘all of us people’ cannot be explained omnipredicatively (‘we-are-entire we-arepeople’), but once the R-prefixes are ignored, there is no problem in deriving the meaning ‘all the people’ out of the combination of ‘entire’ + ‘people.’ Similarly, the meaning of se: siwa:λ nika:n tiλa:ka’ ‘a woman out of us people here’ in (91) can be correctly predicted if only S’s are relevant; it seems intuitively natural for a noun (or noun phrase) cluster ‘one’ + ‘woman’ + ‘indigenous people’ to mean ‘one woman out of the indigenous people.’ If it is assumed that R-person/number is assigned on semantic grounds, the first-person plural R-marking on ti-λa:ka’ [1PL, person] can be accounted for by the fact that the speaker belongs to the group of indigenous people. (91) Ø-se: Ø-siwa:λ nika:n ti-λa:ka’ 3SGR-one 3SGR-woman here in Ø-kin-wa:l-wi:ka-k in 1PLR-person IN 3SGS-3PLO-CISL-lead-PST IN Ø-wa:l-na:waλa’to’-ti-ya’ 3SGS-CISL-interpret-L-go.PST ‘a woman out of us people here (i.e. an indigenous woman) who led them hither, headed hither serving as an interpreter’ (FC XII, 25) Let us make this hypothesis more explicit. There are seemingly two theoretical options to this approach. The first is to consider that the R-person/number of appositional Rmarked nouns is absent in syntax, and is assigned after sentence composition. The second is to include R-prefixes in syntax and to assume that they do not affect the syntactic behavior of their S’s. Since it seems more natural to hypothesize that all items which participate in semantic interpretation are present in syntax, this thesis tentatively adopts the second option. Importantly, as displayed in Section 4.4, R-marking applies to a noun phrase rather than to an individual noun. It is therefore most straightforward to assume that [R, S] 98 4 DISCUSSION is simply S pre-syntactically, and that the noun cluster [R1 , S1 ]+[R2 , S2 ] in multi-word R-marking is underlyingly S1 +S2 . When R-marking applies to S or S1 +S2 , it realizes as [R, S] or [R1 , S1 ]+[R2 , S2 ], respectively. This relationship between R and S (or S1 +S2 , etc.) can be symbolized as “S←R” (or “[S1 S2 ]←R,” etc.) henceforth. For instance, a typical example of appositional R-marking (92a) can be analyzed as (92b).76 Quasi-article in is omitted here. (92) a. ka nika:n t-on-ka’ AFF here in ti-no-pil-c/in... 2SGS-TRSL-be IN 2SGR-1SGP-child-HON ‘Here youSG are, youSG who are my child, ...’ b. AFF here youSG .exist (FC VI, 93) my.child←2SG It is also necessary to exclude adverbs, locatives, and intervening nouns with different R-marking from phrasal R-marking, but it is not discussed in detail here. The counterexamples to the omnipredicative analysis presented in the last section, partially repeated below as (93)–(96), can be successfully accounted for in this model. Here, it is assumed that the appositional nouns are not pre-formed subject–predicate complexes (i.e. only S’s are relevant) and that they receive phrasal R-marking which is not relevant to sentence composition. The structures in (93a), (94a), (95a), and (96a) are abstracted as (93b), (94b), (95b), and (96b) respectively. (93) a. Ø-se: Ø-siwa:λ nika:n ti-λa:ka’ ... 3SGR-one 3SGR-woman here 1PLR-person ‘a woman out of us people here (i.e. an indigenous woman)’ b. [ one 76 woman ]←3SG [ (here) (FC XII, 25) person ]←1PL The status of the person markers on principal verbs is problematic. The (92b) can be assumed to be agreeing with the R-marked subject nouns, or to be itself a pronominal argument of the verb. This issue is not persued here. 99 4 DISCUSSION (94) a. ... in t-i:škičtin sema:na:wa-k ti-λa:ka’ ... IN 1PLR-entire world-LOC 1PLR-person ‘we, all the people in this world’ (95) b. [ entire (in.the.world) a. ... in ye t-i:škič [X] (FC I, 68) person ]←1PL in ye ti-moči IN already 2SGR-entire IN already 2SGR-all ‘every one of you (lit. youSG )’ b. [ (already) (96) entire (FC IX, 52) (already) all ]←2SG a. ti-se:me’ 1PLR-one ‘one of us’ [G] (Carochi, 1645, f. 85v) b. one←1PL The representations in (93b), (94b), (95b), and (96b) are of course not the only possible interpretations. In this section, identical R-marking is tentatively formalized as multi-word R-marking as far as possible. It is of course also reasonable, for example, to analyze (93a) as “one←3SG woman←3SG ...” if (93a) is interpreted as not containing a multi-word R-marking. As already mentioned, the overall meaning of (93) can be derived from the combination ‘one + woman + [here (i.e. indigenous) people].’ The meaning of (94) can also be predicted from the structure ‘entire + [people in this world].’ Similarly, (95) and (96) can be analyzed simply as ‘entire and all’ and ‘one’ respectively. The remaining problem is how the R-person/number are assigned. If it is not the case that each noun is pre-syntactically formed as a copulatic pair of R and S, where does R-marking come from? The criteria of R-marking is clear in such examples as (93), where the denotation 100 4 DISCUSSION of the noun cluster nika:n λa:ka’ ‘people here, indigenous people’ can unproblematically be equated with the R-participant. The first-person marking in (94) can also be accounted for, since the non-omnipredicative approach enables to analyze the cluster ti:škičtin sema:na:wak tiλa:ka’ as a nominal constituent which means ‘entire people in this world,’ whose total referent can be equated to the R-participant ‘we.’ The explanation of (95) and (96), on the other hand, is slightly more complicated. (95) does not state that the R-participant (i.e. the addressee) is the only person in this context; rather, (95) should be analyzed as a singular counterpart of such examples as (94), where ‘all’-type quantifiers are marked as plural. The fact that universal quantificational expressions sometimes behave as singular in Classical Nahuatl, just as every in English and cada in Spanish, seems to justify this view. (97) i:k Ø-moči Ø-λa:kaλ Ø-λama’se:wa-ya thus 3SGR-all 3SGR-person 3SGS-do.penitence-IMPF ‘Then everyone would do penitence.’ (FC IV, 6) The plural marking in (96a) is the most problematic of all. In a purely semantic respect, ‘one of us’ cannot be equated to ‘we.’ In order to account for the R-marking of the type (96a), the person/number system of Classical Nahuatl should be consulted. Carochi (1645, f. 85v) points out that the noun se: ‘one’ is sometimes marked as plural instead of singular when it is used as nonspecific. Thus, he gives such forms as Ø-se:me’ te’wa:ntin (3PLR-one 1PL) ‘one (lit. ones) of us’ and Ø-se:me’ as-siwa’ (3PLR-one 2PLR-woman) ‘one (lit. ones) of youPL women.’ Moreover, he reports that it is prescriptively favored putting the verb in the plural form in such cases. (98) ma: ti-se:me’ te’wa:ntin ti-wia:n IRR 1PLR-one 1PL 1PLS-go.OPT ‘Let one of us go.’ [G] (Carochi, 1645, f. 85v) 101 4 DISCUSSION This noncanonical plural agreement is not limited to se: ‘one.’ In (99), the verb is in the second-person plural form despite that aka’somo: aka’ ‘maybe none’ does not have any person/number marking. (99) aw in nika:n am-on-oke’ and IN here aka’somo: aka’ 2PLS-TRSL-lie IN 2PLR-1PLP-child maybe.not someone ame’wa:ntin an-ki-tta-ti:wi’ 2PL in an-no-pilwa:n in o:=ni-k-i’to’ 2PLS-3SGO-see-PURP IN A=1SGS-3SGO-say.PST ‘Maybe none of youPL , my sons who are here, [will] live to see what I said.’ [G] (Carochi, 1645, f. 85v) These examples show that Classical Nahuatl tends to assign to an nonspecific, themselves singular nouns the same person/number feature as the set of their potential referent (“universal set”). It should be noted that the omnipredicative model still cannot account for the Rmarking in (96) (see Note 75). The assumption that an R-marked noun [R, S] constitutes a pre-formed subject–predicate pair is inconsistent with the fact that such expressions as ti-se:me’ [1PL, one] ‘one of us’ are not interpreted as a nominal copulatic expression which establishes a predicative relationship between R and S. Rather, se: ‘one’ in these examples should be interpreted as a kind of quantifier which takes a set of entities (e.g. ‘we’) as an argument and outputs an expression such as ‘one of us.’77 In sum, it seems reasonable to assume that R-marking does not participate in sentence composition. The counterexamples to the omnipredicative model can be unproblemati77 In contrast, an eighteenth-century sermon (Paredes, 1759b, 81) has a phrase ka san ti-se:me’ (AFF only 1SGR-one) ‘we are united,’ which is glossed as somos todos unos, in the principal position. Correspondingly, Andrews ([1975] 2003, 308,428) argues that the original meaning of ti-se:me’ [1PL, one] is ‘we are one (in number),’ and the meaning is “strangely reversed” (Andrews, [1975] 2003, 428) and comes to mean ‘one of us.’ So far, I do not have enough data to discuss this issue. 102 4 DISCUSSION cally accounted for by ignoring R-marking in sentence composition instead of assuming R-prefixes as subject markers. A possible interpretation of R-marking is that they are assigned on semantic grounds just as the number feature in noun phrases in other languages. That is, noun phrases in Classical Nahuatl are semantically specified as to their discourse participant role in an analogous way that noun phrases are specified as to number in such languages as English and Spanish.78 This understanding is also consistent with the observation that the nouns of the type ti-se:me’ [1PL, one] ‘one of us’ have non-obligatory non-third-person R-marking. The non-obligatoriness of first- or second-person marking seems to correspond to the choice of the speaker as to whether s/he describes a nonspecific individual as a representative of the group of R-participants or not. The fact that R-marking does not really control agreement also supports the hypothesis that R-marking is controlled at a different level from syntax. Although appositional Rmarked nouns apparently trigger the agreement of verbs in the majority of examples, the examples in (100) show that the R-person/number and the person/number of their principals do not always coincide. (100) a. Ø-se:me’ te’wa:ntin ti-ya:-ske’ 3SGR-one 1PL 1PLS-go-FUT ‘One of us will go.’ [G] (Carochi, 1645, f. 85v) b. Ø-o:mentin Ø-te:lpo:po:čtin wel Ø-mo-λaso’λa-ya’ 3PLR-two 3PLR-youth 78 aw in Ø-se:me’ well 3PLS-REFL-love-IMPF and IN 3PLR-one This analysis follows the same line as the traditional characterization of Elamite nominal person marking (Reiner, 1969, etc.), which considers the person markers on nouns as “gender” markers (Section 2.6). 103 4 DISCUSSION in ok Ø-se: ye’wa:ntin Ø-ki-lwi’ 3PL 3SGS-3SGO.3SGO-tell.PST IN still 3SGR-one ‘Two young people loved each other very much, and one of them said to the other, ...’ [G] (Carochi, 1645, f. 86r) The lexical or morphological status of R-prefixes are temporarily irrelevant here. Whether they are bound-form pronominal elements or mere by-products of a postsyntactic morphological process is theoretically an important issue, but so far it seems to be only the matter of framework. 4.6.4 Section summary The omnipredicative analysis presented in Launey (1994) and Andrews ([1975] 2003) fails to account for some examples of appositional R-marking. Specifically, the omnipredicative approach cannot explain the examples with quantificational expressions. In contrast, their distribution can be properly predicted by assuming that R-marked nouns are treated as nouns, namely S’s, in the composition of sentences, and receive phrasal R-marking. 4.7 The status of R-marking The previous sections argued that the omnipredicative analysis held by Launey (1994) and Andrews ([1975] 2003) cannot capture the distribution and meaning of multi-word or non-principal R-marking. This section unifies the non-omnipredicative models presented above and tries to give a uniform account of R-marking in various environments. 104 4 DISCUSSION 4.7.1 R-marking as a realization of a semantic feature The core idea of omnipredicative analysis is that all R-marked nouns in Classical Nahuatl are pre-formed copulatic pairs of [R, S]. According to this approach, sentence composition is a process which puts together these subject–predicate pairs into larger units. Contrary to this, Sections 4.4 through 4.6 above argued that R-marked nouns do not behave as pairs of R and S in the syntax of Classical Nahuatl. Rather, it is only S that is significant in sentence composition and R-person/number does not participate in word-external syntax. The discussions above can be summarized as follows. (i) Principal R-marking usually behaves like pronominal clitics in other languages. (ii) The pattern of multi-word R-marking shows that R-marking is phrasal; a single marking process can involve two (or more) nouns at once. (iii) Intraeventive R-marking is probably a formal process which does not have any substantial meaning. (iv) Appositional R-marking is probably a semantic process which is analogous to the number marking on nouns in other languages, and it does not necessarily trigger cross-referencing or control agreement. This section intends to unify these facts by assuming that R-person/number is specified as optional semantic features.79 The notion of semantic feature has been used typically for intrinsic lexical-semantic components such as [+female] in sister and actress; in contrast, Svenonius (2007) suggests that some interpretable formal features such as the 79 In this thesis, the term optional is used in the sense of Chomsky (1995, 231) when it is used for features. A feature is optional when it is not lexically determined (i.e. not intrinsic). For example, the number feature of cat/cats in English is optional, since the lexical entry cat can occur either in the singular or plural form. In contrast, the person feature of you is not optional, since it is lexically specified as second person. Optional here does not mean that a feature can be unspecified or absent. 105 4 DISCUSSION number feature of noun phrases can be considered as optional semantic features. This thesis follows this understanding and intends to characterize R-marking as the realization of optional semantic features of noun phrases. The feature analysis proposed in this section is not the only possible account. As repeated in the previous sections (Sections 4.1, 4.5.3, and 4.6.3), R-prefixes can be considered either as the realization of optional semantic features or as appositional bound-form pronominal items. Either analysis can explain the exclusion of non-principal R-marking from sentence composition.80 Although the discussions below adopts the feature analysis, it is not intended to exclude the pronoun analysis. The feature account is followed in this thesis purely because it is the simplest way to capture the redundant, obligatory, and word-internal nature of R-marking without positing new grammatical constraints which justify the obligatory apposition of bound-form pronouns. This feature analysis enables to characterize different types of R-marking uniformly. As suggested in Section 4.6.3, appositional R-marking can be accounted for successfully by considering R-marking as the realization of semantic features which are analogous to the number or gender feature in other languages. This interpretation can be extended to intraeventive R-marking, which appears to be an almost mechanical concord; crosslinguistically, semantic features often trigger a concord between items. It is common that a single marker has both informative (interpretable) and non-informative uses81 according 80 Which hypothesis should be followed seems to depend on the interpretation of person markers in Classical Nahuatl in general. The pronoun analysis would be favorable if bound-form person markers are interpreted as pronouns as in (Jelinek, 1984; Mithun, 1986; Alexiadou and Anagnostopoulou, 1998). 81 This thesis does not use the term informative in the meaning of “functionally relevant.” Rather, the item X is informative iff the primary meaning of X is mapped onto the meaning of the sentence in a direct and systematic way. Accordingly, a non-informative marker is not necessarily functionally null; it is possible that a non-informative element participate in indexation or index agreement at the functional level (Croft, 2003; Svenonius, 2007). This characterization of informativeness follows the re-definition 106 4 DISCUSSION to the environment. For example, in consistent pro-drop languages such as Spanish (and also Classical Nahuatl), bound-form person markers seem to play a double role. In Spanish, person endings on verbs are usually not informative when there is an overt argument; they simply appear to agree with the person/number of the subject. In contrast, they acquire an informative status when an overt argument does not appear. The pronominal subject is recovered in turn from the person/number of the verb. Therefore, the observation that intraeventive R-marking does not have any substantial meaning does not contradict the assumption that R-person and R-number are primarily semantic features. Multi-word R-marking can also be accounted for as a concord. Thus, if R-marking is assumed to be primarily the realization of semantic features, both the mechanical concord in intraeventive or multi-word R-marking and the half-autonomous nature of appositional R-marking can be uniformly characterized. Seemingly, the contents of R-person/number features do not need intricate analysis. They may simply be assumed as [1], [2], [3], [SG], and [PL] following the traditional analysis, or may be decomposed into primitive features as in Harley and Ritter (2002). The only important thing which should be explained is that the nouns of the type ti-se:me’ [1PL, one] ‘one of us’ are non-obligatorily marked with the person/number of their “universal set” (Section 4.6.2). As already mentioned in Section 4.6.3, this non-obligatoriness of R-marking can be attributed to the speaker’s choice; accordingly, it can be unproblematically explained by the feature analysis. Considering R-marking as the realization of semantic features also simplifies the problem concerning the R-marking on pronouns. As pointed out in Section 2.2.4 further above, Classical Nahuatl R-prefixes and independent pronouns are phonologically related. Andrews ([1975] 2003) accordingly analyzes them as R-marked forms of a person-neutral of interpretability in Svenonius (2007). 107 4 DISCUSSION pronominal stem. However, a closer look at independent pronouns shows that the situation is slightly more complicated than it appears. In Classical Nahuatl, independent pronouns can serve either as principals (e.g. ‘it is I,’ ‘it is youSG ’) or as non-principals (e.g. ‘I am,’ ‘youSG do’). Independent pronouns have the same form in either cases. For instance, they serve as principals in (101) and as arguments in (102). (101) a. kwiš ye Q te’ in ti-mote:kwso:ma already 2SG IN 2SGR-PN ‘Is it youSG that are Moteuczoma?’ b. ka ne’wa:λ in n-amo-te:či:wka:w AFF 1 SG (FC XII, 31) in ni-mote:kwso:ma IN 1SGR-2PLP-governor IN 1SGR-PN ‘It is me that is yourPL governor Moteuczoma.’ (102) (FC XII, 31) a. aw in ne’wa:λ ka senka’ ni-kwala:ni and IN 1SG AFF very 1SGS-be.angry ‘And I am very angry.’ [X] (FC I, 76) b. aw in ne’ te’wa:λ močipa ti-ne:č-kokolia and IN 1SG 2SG always 2SGS-1SGO-hate ‘[I never hated youSG ] but youSG always hate me.’ [G] (Paredes, 1759a, 34) Although these examples are not problematic if independent pronouns are excluded from omnipredicativity, the model proposed here can account for these data more consistently. Since R-marking is the realization of certain semantic features, it can be assumed that some items have inherent values.82 82 The only counterexample to this analysis is a form ti-te’wa:n(tin) ‘we are us’ (i.e. you/they are from our tribe) (Carochi, 1645, f. 86r), which has an R-prefix in addition to a pronoun. However, since this seems to be a highly idiomatic expression, this thesis cannot pursue this issue any further. 108 4 DISCUSSION Similarly, under this hypothesis, the fact that some items such as demonstratives and interrogatives lack R-marking (Section 2.4.4) becomes unproblematic. The assumption that some lexical items are not specified as to particular semantic features seems more reasonable than Andrews’ ([1975] 2003) argument that they are consistently marked as third person regardless of their referents. The feature analysis is also typologically favorable. The inventory of semantic feature varies cross-linguistically; some languages have a morphological number distinction while other languages do not. Thus, under the non-omnipredicative analysis, it is not necessary to assume in Classical Nahuatl a radically different syntactic structure from those in other polysynthetic languages and other Nahuan languages without uniform R-marking. 4.7.2 Status of principal R-marking Let us turn to principal R-marked nouns. Although the status of principal R-marking is an open issue, a few examples appear to suggest that it can be considered as a special case of intraeventive R-marking. If this analysis is on the right track, all of the three types of R-marking can be uniformly explained. Apparently, R-prefixes in principal nouns behave like the pronominal person markers in other languages like Turkish (Section 2.6). This analysis assumes that R-prefixes on principal nouns are syntactically treated as arguments. It can be a problem for the unified account of R-marking proposed here, since it characterizes R-prefixes as the realization of particular semantic features which do not change the syntactic category of their hosts. This problem is related to the question whether a “zero copula” can be posited in nominal predicates in Classical Nahuatl. Clauses headed by a principal R-marked nouns can be analyzed either as (i) copular clauses with covert (zero) copulae, or (ii) tenseless intransitive clauses. Launey (1994) and Andrews ([1975] 2003), of course, argue that 109 4 DISCUSSION there are no “zero copula” in nominal clauses with principal R-marked nouns, for the semi-verbal nature of nouns constitutes the foundation of their omnipredicative models. Although the zero-copula analysis (i) is more consistent with the argument of this thesis, there is so far no decisive evidence for either approach. However, as displayed at the end of this section, there are a few phenomena which may support the view of (i). Let us first see how these two models work. As already mentioned in Section 2.4.1 and 2.4.2, principal R-marked nouns and overt copulae are in complementary distribution. In principle, an overt copula appears only when the clause is in a marked tense/mood, as shown in the examples (103) (neutral tense/mood) and (104) (other tenses/moods). (103) a. ka ni-towe:nyo: AFF 1 SGR - EN ‘I am a Huastec.’ (FC III, 20) b. ka a’mo: ni-pi:lli AFF NEG 1SGR-noble ‘I am not a noble.’ (104) (Chim. 52:38) a. n-ALCALDE o:=ni-kat-ka 1SGR-mayor A =1 SGS -be- PST ‘I was a mayor.’ b. ma: ni-kwalli IRR (FC VI, 232) o:=ni-ye-ni 1SGR-good A=1SGS-be-OPT. PST ‘Would that I had been good.’ [G] (Carochi, 1645, f. 38r) As mentioned by Launey (1994, 51–56), there are two possible approaches to explaining the distribution of principal R-marked nouns. The first is to posit a zero copula in such sentences as (103). In fact, Tetelcingo Nahuatl has a non-obligatory overt copula in corresponding environments. 110 4 DISCUSSION (105) taha tı-ka tı-sıhtlı 2SG 2SGS-be 2SGR-jackrabbit ‘YouSG are a jackrabbit.’ (Tuggy, 1979, 139) Although this thesis is not intended to support any particular theoretical framework, this zero-colpula analysis seems to be in line with the “Pred” analysis held by Bowers (1993). The second theoretical option is to interpret principal R-marked nouns as intransitive predicates. Launey (1994, 51–56) holds this position on the grounds that intraeventive nouns are redundantly R-marked (therefore R-prefixes are not pronominal copulae) and that locative construction requires an overt copula (therefore R-prefixes are different from the zero copula of such languages as Moroccan Arabic, where locative construction permits a zero copula). Parallelly, Stassen (1997) characterizes Classical Nahuatl principal nouns as “verbally encoded,” since they take the same “subject” person markers as verbal subject markers and are negated in the same way as verbs. Launey’s (1994) argument is not really critical here, however. Although Launey (1994) uses the distribution of R-prefixes in intraeventive and locative nouns as the corroborations of his argument, they can support his interpretation only if R-prefixes are treated as pronominal or copulatic syntactic items. Within the framework of this thesis, the presence/absence of an R-prefix can be assumed to be independent from syntactic structure. Rather, it enables to analyze R-person/number as redundantly specified. It should be pointed out that, in either approach, the absence of an overt copula should not be equated with present indicative tense/mood. The fact that principal R-marked nouns can appear in non-present/indicative environments suggests that they are tenseneutral rather than present indicative.83 83 This issue needs more investigation, since the present indicative tense/mood is often used for describing 111 4 DISCUSSION (106) a. in ma’an ti-šiw-c/inλi IN as.if in ma’an ti-kil-c/inλi 2SGR-grass-DIM IN as.if o:=ti-wa:l-i:š-e:wa-k 2SGR-leaf-DIM o:=ti-šoλa-k o:=ti-kwepo:n A =2 SGS - CISL -face-depart- PST A =2 SGS -burst-PST A =2 SGS -spring. PST ‘YouSG came out, youSG burst into bloom, youSG sprang out as if youSG were a little herb, little leaf.’ b. aw ka in Ø-λa:kaλ (FC VI, 94) Ø-no-λa’-c/in ... in i:k and AFF IN 3SGR-person 3SGR-1SGP-uncle-HON IN thus mi:l-la’ Ø-te:č-mo-mi:l-maki-li’ o:se:lo:tepe:k aw in 3SGS-1PLO-REFL-farm-give-APPL . PST farm-among LN i:k a’=ni-ye:kλi in i:k a’=ni-kwalli and IN ka senka’ thus NEG=1SGR-right IN thus NEG=1SGR-good AFF very ni-kwala:n 1SGS-be.angry.PST ‘But when the lord, my uncle ... gave (hon.) us a cultivated field in the fields of Ocelotepec, I was improper and bad; I was very angry.’ (Juan de San Antonio’s Letter, f. 130v)84 Similarly, Bierhorst (1985b, 185) points out that the irrealis particle ma:, which is usually used with optative or admonitive forms of verbs, is used with nouns several times in Cantares mexicanos. This is also attested in the prose texts in Florentine Codex. (107) a. ma: ni-kec/al-to:to:λ IRR 1SGR-quetzal-bird ‘Let me be a quetzal.’ (Cant. 64r:21) past events in Classical Nahuatl texts. 84 This example is taken from a letter in 1564 included in Chimalpahin (1997, vol. 2, 230–232). 112 4 DISCUSSION Ø-no-yo:llo: b. ma: Ø-šo:čiλ IRR 3SGR-flower 3SGR-1SGP-heart ‘Let my hearts be flowers.’ c. ma: Ø-iwki IRR (Cant. 36v:14) =i:n 3SGR-like =this ‘Let it be thus.’ (FC III, 63) d. ma: ne’ =i:n IRR 1SG =this ‘Let me be this.’ (FC VI, 89) In addition, the “ma: + noun” sequence is attested with other meanings than equation as in (108). Although such examples are infrequent and highly marginal, it may be possible to assert that these examples support the view that the absence of a copula is the result of an ellipsis of a verb.85 (108) a. Ø-pa’-pa:n-namaka-ya Ø-k-i’to’-ti-nen-ka ma: 3SGS-RDP-flag-sell-IMPF 3SGS-3SGO-say-L-live-PST IRR Ø-amo-pa:n-c/in 3SGR-2PLP-flag-DIM ‘She was selling flags, saying, “[Here are] yourPL little flags!”’ b. ma: nika:n in Ø-i:-pal IRR 85 here nem-o:wa Ø-i:-tlayo:col IN 3SGR-3SGP-palm live-NA (FC III, 24) ye’wa:-n 3SGR-3SGP-creation 3SG-EUPH The fact that the copular verb cannot be omitted in locative or existential constructions seems to be problematic here. Although it seems possible to assume that the copular verb drops iff it does not have a substantial lexical meaning, I must acknowledge that the omnipredicative model can explain the facts more elegantly in this respect. 113 4 DISCUSSION Ø-DIOS 3SGR-God ‘May the creations of God, by whom everything lives, be here.’ (Cant. 7v:12) Probably the strongest support for the zero-copula analysis is the fact that R-marking seems to target noun phrases rather than individual nouns. Multi-word R-marking in the principal position as in (109) suggests that it is difficult to analyze principal R-marked nouns as predicative pairs of predicates (S) and its subjects (R). Since such adjective–noun combination as ti-nelli ti-teo:piška:c/inλi ([2SG, real] [2SG, priest]) seems to be better analyzed as ‘[real priest]← 2SG’ rather than ‘youSG are a real one and youSG are a priest,’ such sentences as (109) can be analyzed more straightforwardly by positing a zero copula. (109) ka ti-nelli ti-teo:pišk-a:-c/inλi ti-nelli AFF 2 SGR -true 2 SGR -priest- L- HON ti-SACERDOTE 2SGR-true 2SGR-priest ‘YouSG are a real priest, a real sacerdote.’ [X] (La adoración de los Reyes, 308)86 Thus, although there is not enough evidence to exclude the non-zero-copula analysis completely, a couple of phenomena suggest that the zero-copula analysis can account for the behavior of nominal clauses in Classical Nahuatl more consistently. If this analysis is correct, principal R-marked nouns can be included in intraeventive R-marking. Accordingly, all of the three types of R-marking can be uniformly accounted for. 4.7.3 R argument and R-marking Lastly, it is important to point out that R-marking cannot be directly associated with the R θ-role discussed in Williams (1981) at either the descriptive or theoretical level. Williams 86 Horcasitas (1974, I, 308). For the source of this example, see Note 71. 114 4 DISCUSSION (1981) argues that nouns have referential arguments (R arguments) just as verbs have external arguments, and the argument structure of nouns is preserved also in non-predicate positions. Baker (1996, 248–249) suggests that the R-marking in Classical Nahuatl is an agreement with the R arguments. However, the view that R-marking represents the external arguments of nominal predicates is exactly what has been denied in the discussions above. Williams’ (1981) hypothesis that the predicate–argument relationship is a linking between two θ-roles parallels with Launey’s (1994) model where a cross-reference relationship is established between two person markers. Thus, following the previous discussions, R-marking cannot be interpreted as the agreement with R arguments. 4.7.4 Section summary This section attempts to provide a unified account of R-marking. In contrast to the omnipredicative analysis which assumes that R-marking directly reflects the subject– predicate relationship between R and S, this thesis considers R-marking as the realization of certain optional semantic features of noun phrases. The relationship between informative marking and non-informative concord accounts for the opposite characteristics of appositional and intraeventive R-marking. Although the exact characteristics of principal R-marking are not clear, some observational facts suggest that principal R-marking can be characterized as a subtype of intraeventive R-marking. 115 5 CONCLUSION 5 Conclusion This thesis places the characteristics of R-marking in Classical Nahuatl as a fundamental problem for both language-internal and cross-linguistic studies, and intends to provide a non-omnipredicative analysis of this phenomenon. Although many languages are reported to have nominal “subject” person marking, the uniform R-marking of the type found in Classical Nahuatl is highly rare across languages. Accordingly, R-marking has been paid little attention compared with other grammatical features of Classical Nahuatl. In the tradition of the study of Native American languages and/or polysynthetic languages, the description of nominal person marking mainly focused on possessor marking, and less on “subject” marking. Two recent comprehensive theoretical studies on Classical Nahuatl, namely Launey (1994) and Andrews ([1975] 2003), consider that the R-marking on nouns reflects the omnipredicative nature of the language. They assume the uniformity between verbal subject marking and R-marking as the evidence of the copulatic nature of Classical Nahuatl nouns, and describe the syntax of the language as a process of assembling pre-formed nominal predicative units into larger constituents. In contrast, in Section 4 it is shown that the various uses of R-marking cannot be accounted for by their omnipredicative analysis. Many examples suggest that R-prefixes are irrelevant to sentence composition and that R-marking does not assign syntactic predicativity to the nouns. Instead, Section 4 proposes a non-omnipredicative analysis which assumes that R-marking is the realization of certain semantic features analogous to the number feature of noun phrases in other languages, and compares the informative and non-informative uses of R-marking to the independent (interpretable) and concordal (uninterpretable) uses of features. Put in other words, Classical Nahuatl nouns are really nouns like those in many other languages are, and R-person/number is a semantic property 116 5 CONCLUSION inherent in each noun phrase rather than the syntactic subject of a nominal predicate. Although the characteristics of principal R-marked nouns are problematic, it seems also possible to account for the principal R-marking as a special case of intraeventive Rmarking. This analysis leads to a conclusion that it is not necessary to assume that Classical Nahuatl is an extraordinary language which has a radically different morphosyntax from that of many other languages. One does not have to assume that Classical Nahuatl sentences have essentially different structures from those of many other polysynthetic languages such as Mohawk, or from its sister Nahuan languages which preserve the overall polysynthetic morphosyntax but lack uniform R-marking. It also has some implications on linguistic theories. Launey’s (1994) assumption that Classical Nahuatl constitutes an extreme example of noun–verb symmetry is itself typologically significant, but the omnipredicative approach would lead to a more serious conclusion that all apparent non-predicates are really predicates in the language. Under the omnipredicative hypothesis, the uniformity of R-marking in Classical Nahuatl would be a critical counterexample to the arguments that nouns can be characterized as expressions which identify something (Baker, 2003). According to the conclusion of this thesis, on the other hand, the example of Classical Nahuatl is less problematic (therefore less interesting) in this respect. The cross-linguistic comparison of nominal “subject” marking and the typological status of Classical Nahuatl R-marking are open issues. Since non-principal overt Rmarking is a relatively marginal aspect of a grammar, it has hardly been covered in the grammatical works on other languages. The cross-linguistic distribution of R-marking in Nahuan and Mayan groups, or more broadly in Mesoamerica, is therefore open to further research. A careful study may show that uniform R-marking is cross-linguistically 117 5 CONCLUSION not as uncommon as it appears now. In addition, the status of nominal person marking in Dravidian, Elamite, and possibly in other languages, should be investigated in more detail. Further cross-linguistic study of R-marking will make the status of the R-marking in Classical Nahuatl clearer, and may shed light on the typological and theoretical study of agreement, predication, polysynthesis, or lexical categories. 118 REFERENCES References Ackema, Peter, Patrick Brandt, Maaike Schoorlemmer, and Fred Weerman. 2006. The role of agreement in the expression of arguments. In P. Ackema, ed., Arguments and Agreement, pages 1–32. New York and Oxford: Oxford University Press. Agesthialingom, S. 1977. A Grammar of Old Tamil with Special Reference to Patirruppattu: Phonology & Verb Morphology. Annamalai Nagar, Tamil Nadu, India: Annamalai University. Alexiadou, Artemis and Elena Anagnostopoulou. 1998. Parametrizing AGR: word order, V-movement and EPP-checking. Natural Language and Linguistic Theory 16:491– 539. Andrews, J. Richard. [1975] 2003. Introduction to Classical Nahuatl. Norman, Oklahoma: The University of Oklahoma Press, revised edn. Armstrong, Grant. 2009. On copular sentences in Yucatec Maya. In Proceedings of the Conference on Indigenous Languages of Latin America (CILLA) IV. Retrieved online at: http://www.ailla.utexas.org/site/cilla4/Armstrong_CILLA_IV.pdf. Baker, Mark C. 1996. The Polysynthesis Parameter. New York and Oxford: Oxford University Press. Baker, Mark C. 2003. Lexical Categories: Verbs, Nouns, and Adjectives. Cambridge: Cambridge University Press. Bierhorst, John. 1985a. Cantares Mexicanos: Songs of the Aztecs. Stanford, California: Stanford University Press. Bierhorst, John. 1985b. A Nahuatl–English Dictionary and Concordance to the Cantares Mexicanos. Stanford, California: Stanford University Press. Bierhorst, John. 1992a. Codex Chimalpopoca: The Text in Nahuatl with a Glossary and Grammatical Notes. Tucson, Arizona and London: The University of Arizona Press. 119 REFERENCES Bierhorst, John. 1992b. History and Mythology of the Aztecs: The Codex Chimalpopoca. Tucson, Arizona: The University of Arizona Press. Bork, Ferdinand. 1934. Die elamische Klammer. Archiv für Orientforschung 9:292–300. Bowers, John. 1993. The syntax of predication. Linguistic Inquiry 24:591–656. Bresnan, Joan and Sam A. Mchombo. 1986. Grammatical and anaphoric agreement. In Papers from the Parasession on Pragmatics and Grammatical Theory at the 22nd Regional Meeting of the Chicago Linguistic Society, pages 278–297. Bright, William. 1990. ‘With one lip, with two lips’: parallelism in Nahuatl. Language 66:437–452. Campbell, Lyle. 1985. The Pipil Language of El Salvador. Berlin: Walter de Gruyter. Carochi, Horacio. 1645. Arte de la lengua mexicana: con la declaración de los adverbios della. México: Juan Ruiz. Carochi, Horacio. [1645] 2001. Grammar of the Mexican Language: With an Explanation of Its Adverbs (1645). Stanford, California: Stanford University Press. Trans. by J. Lockhart. Chierchia, Gennaro. 1998. Reference to kind across languages. Natural Language Semantics 6:339–405. Chimalpahin, Don Domingo Francisco de San Antón Muñón, Quauhtlehuanitzin. 1997. Codex Chimalpahin, Volumes 1–2. Ed. and trans. by A. J. O. Anderson & S. Schroeder. Norman, Oklahoma: The University of Oklahoma Press. Chimalpahin, Don Domingo Francisco de San Antón Muñón, Quauhtlehuanitzin. 2006. Annals of His Time. Ed. and trans. by J. Lockhart, S. Schroeder & D. Namala. Stanford, California: Stanford University Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Massachusetts: The MIT Press. 120 REFERENCES Corbett, Greville G. 2001. Agreement: Canonical instances and the extent of the phenomenon. In G. Booij, J. DeCesaris, A. Ralli, and S. Scalice, eds., Topics in Morphology: Selected papers from the Third Mediterranean Morphology Meeting (Barcelona, September 20-22, 2001), pages 109–128. Retrieved online at: http://epubs. surrey.ac.uk/1075/. Croft, William. 2000. Parts of speech as language universals and as language-particular categories. In P. M. Vogel and B. Comrie, eds., Approaches to the Typology of Word Classes, pages 65–102. Berlin: Walter de Gruyter. Croft, William. 2003. Typology and Universals. Cambridge: Cambridge University Press, 2nd edn. Originally published in 1990. Dakin, Karen. 1982. La evolución fonológica del protonáhuatl. Universidad Nacional Autónoma de México. Dixon, R. M. W. 2010–2012. Basic Linguistic Theory, Volumes I–III. New York and Oxford: Oxford University Press. Evans, Nicholas. 1999. Why argument affixes in polysynthetic languages are not pronouns: evidence from Bininj Gun-wok. Sprachtypologie und Universalienforschung 52:255–281. Evans, Nicholas. 2000. Word classes in the world’s languages. In C. Lehmann, G. Booij, and J. Mugdan, eds., Morphologie: ein internationales Handbuch zur Flexion und Wortbildung, 1 Halbband, HSK 17.1, pages 708–732. Berlin: Walter de Gruyter. Evans, Nicholas and Toshiki Osada. 2005. Mundari: the myth of a language without word classes. Linguistic Typology 9:351–390. Fortescue, Michael D. 1994. Morphology, polysynthetic. In R. E. Asher and J. M. Y. Simpson, eds., The Encyclopedia of Language and Linguistics, vol. 5, pages 2600– 2602. Oxford, etc.: Pergamon Press. 121 REFERENCES Gamut, L. T. F. 1991. Logic, Language, and Meaning, Volume 1: Introduction to Logic. Chicago and London: The University of Chicago Press. García Matzar, Pedro Oscar, Valerio Toj Cotzajay, and Domingo Coc Tuiz. 1999. Gramática del idioma kaqchikel. La Antigua, Guatemala: Proyecto Lingüístico Francisco Marroquín. Garibay K., Ángel María. [1940] 1961. Llave del náhuatl. México: Editorial Porrúa, 2nd edn. Geach, Peter. 1962. Reference and Generality. Ithaca, New York: Cornell University Press. Göskel, Aslı and Celia Kerslake. 2005. Turkish: A Comprehensive Grammar. London and New York: Routledge. Grignard, A., S.J. 1924. A Grammar of the Oraon Language and Study in Oraon Idiom. Calcutta: A. Rome, Catholic Orphan Press. Grillot, Françoise. 1978. Les affixes nominaux et les pronoms indépendants de la langue élamite. Journal Asiatique 282:1–18. Grillot-Susini, Françoise and Claude Roche. 1987. Éléments de grammaire élamite. Paris: Éditions Recherche sur les Civilisations. Gupta, Anil. 1980. The Logic of Common Nouns. New Haven, Connecticut: Yale University Press. Harley, Heidi and Elizabeth Ritter. 2002. Person and number in pronouns: a featuregeometric analysis. Language 78:482–526. Hill, Kenneth C. 2006. Introduction to Classical Nahuatl and Workbook for Introduction to Classical Nahuatl by J. Richard Andrews. Anthropological Linguistics 48:88–94. Horcasitas, Fernando. [1968] 1979. De porfirio de Díaz a Zapata: Memoria náhuatl de Milpa Alta. México: Universidad Nacional Autónoma de México, 2nd edn. 122 REFERENCES Horcasitas, Fernando. 1974. El teatro náhuatl: épocas novohispana y moderna. México: Universidad Nacional Autónoma de México. INEGI. 2005. Perfil sociodemográfico de la población hablante de náhuatl. México: Instituto Nacional de Estadística, Geografía e Informática. Online version retrieved at: http://www.inegi.org.mx/prod_serv/contenidos/espanol/bvinegi/productos/censos/ poblacion/poblacion_indigena/Hablantes_Nahuatl.pdf. Jacobsen, William, Jr. 1979. Noun and verb in Nootkan. In B. Efrat, ed., The Victoria Conference on Northwestern Languages : Victoria, British Columbia, November 4-5, 1976, pages 83–155. Jelinek, Eloise. 1984. Empty categories, case, and configurationality. Natural Language and Linguistic Theory 2:39–76. Johanson, Lars. 1998. The structure of Turkic. In L. Johanson and É. A. Csató, eds., The Turkic Languages, pages 30–66. London: Routledge. Karttunen, Frances and James Lockhart. 1987. The Art of Nahuatl Speech: The Bancroft Dialogues. Los Angeles: UCLA Latin American Studies Publications. Kinkade, M. Dale. 1983. Salish evidence against the universality of ‘noun’ and ‘verb’. Lingua 60:25–39. Kobayashi, Masato. 2012. Texts and Grammar of Malto. Vizianagaram, India: Kotoba Books. Kornifit, Jaklin. 1997. Turkish. London and New York: Routledge. La Liga Bíblica. [1980] 2009. El Nuevo Testamento en el náhuatl de Tetelcingo. Crete, Illiois: La Liga Bíblica. Online version retrieved at: http://www.scriptureearth.org/ data/nhg/PDF/00-WNTnhg-web.pdf. Langacker, Ronald W., ed. 1979. Studies in Uto-Aztecan Grammar, Volume 2: Modern Aztec Grammatical Sketches. Dallas, Texas: The Summer Institute of Linguistics and 123 REFERENCES the University of Texas. Lastra de Suaréz, Yolanda. 1986. Las áreas dialectales del náhuatl moderno. México: Universidad Nacional Autónoma de México. Launey, Michel. 1984. Fonctions et catégories dans l’opposition verbo-nominale. Modèles Linguistiques 6:133–148. Launey, Michel. 1991. La logique omniprédicative dans la syntaxe nahuatl. In LALIES: Actes des sessions de linguistique et de littérature d’Aussois no 10, pages 215–231. Launey, Michel. 1994. Une grammaire omniprédicative : essai sur la morphosyntaxe de nahuatl classique. Paris: CNRS Éditions. Launey, Michel. 2003. Le type omniprédicatif et la morphosyntaxe du nahuatl. Faits de langues 21. 2:9–24. Launey, Michel. 2004. The features of omnipredicativity in Classical Nahuatl. Sprachtypologie und Universalienforschung 57:49–69. León-Portilla, Miguel. 1985. Nahuatl literature. In M. S. Edmonson, ed., Supplement to the Handbook of Middle American Indians, Volume 3: Literatures, pages 7–43. Austin, Texas: University of Texas Press. Lockhart, James. 2001. Nahuatl as Written: Lessons in Older Written Nahuatl, with Copious Examples and Texts. Stanford, California: Stanford University Press. López Austin, Alfredo. 1974. The research method of Fray Bernardino de Sahagún: the questionnaires. In M. S. Edmonson, ed., Sixteenth-Century Mexico: The Work of Sahagún, pages 111–149. Albuquerque, New Mexico: University of New Mexico Press. MacSwan, Jeff. 1998. The argument status of NPs in Southeast Puebla Nahuatl: Comments on the Polysynthesis Parameter. 17:101–114. 124 Southwest Journal of Linguistics REFERENCES Mattissen, Johanna. 2006. Ontology and diachrony of polysynthesis. In D. Wunderlich, ed., Advances in the Theory of the Lexicon, pages 291–294. Berlin: Mouton de Gruyter. Mithun, Marianne. 1986. When zero isn’t there. In Proceedings of the 12th Annual Meeting of the Berkeley Linguistics Society, pages 195–211. Mithun, Marianne. 1999. The Languages of Native North America. Cambridge: Cambridge University Press. Molina, Fray Alonso de. 1571. Vocabulario en lengua mexicana y castellana. México: Antonio de Spinosa. Montes de Oca, Mercedes, Vega. 1997. Los difrasismos en el náhuatl: un problema de traducción o de conceptualización. Amerindia 22:31–44. Nakayama, Toshihide. 2001. Nuuchahnulth (Nootka) Morphosyntax. Berkeley, California: University of California Press. NBTN. 2012. Ne Bibliaj Tik Nawat [the Bible in Nawat]. http://nebibliaj.org/. Olmos, André de. [1547] 1875. Grammaire de la langue nahuatl ou mexicaine. Paris: Imprimerie Nationale. Published by Rémi Siméon. Paredes, Ignacio de. 1759a. Compendio del arte de la lengua mexicana por el P. Horacio Carochi de la Compañía de Jesus. México: Imprenta de la Bibliotheca Mexicana. Paredes, Ignacio de. 1759b. Promptuario manual mexicano. México: Imprenta de la Biblioteca Mexicana. Pharao Hansen, Magnus. 2010. Polysynthesis in Hueyapan Nahuatl: The status of noun phrases, basic word order, and other concerns. Anthropological Linguistics 52:274– 299. Rajam, V. S. 1992. A Reference Grammar of Classical Tamil Poetry: 150 B.C.–Prefifth/sixth century A.D. Philadelphia: American Philosophical Society. Reiner, Erica. 1969. The Elamite language. In J. Friedrich, E. Reiner, A. Kammenhuber, 125 REFERENCES and G. N. A. Heubeck, eds., Altkleinasiatische Sprache, pages 54–118. Leiden and Köln: E. J. Brill. Robinson, Dow F. and William R. Sischo. 1969. Michoacán (Pómaro) Nahual clause structure. In D. F. Robinson, ed., Aztec Studies I: Phological and Grammatical Studies in Modern Nahuatl Dialects, pages 53–74. Norman, Oklahoma: The Summer Institute of Linguistics. Rojas Rabiela, Teresa, Elsa Leticia Rea López, and Constantino Medina Lima, eds. 2000. Vidas y bienes olvidados: Testamentos indígenas novohispanos, Vol. 2. México: Centro de Investigaciones y Estudios Superiores en Antropología Social. Sahagún, Fray Bernardino de. [1583] 1993. Psalmodia Christiana. Trans. by Arthur J. O. Anderson. Salt Lake City, Utah: The University of Utah Press. Sahagún, Fray Bernardino de. 1950–1982. Florentine Codex: General History of the Things of New Spain. Books I–XII. Trans. by A. J. O. Anderson & C. E. Dibble. Santa Fe, New Mexico: The School of American Research and The University of Utah. Sahagún, Fray Bernardino de. 1997. Primeros Memoriales: Paleography of Nahuatl Text and English Translation. Trans. by Thelma D. Sullivan and Henry B. Nicholson. Norman, Oklahoma: The University of Oklahoma Press. Sapir, Edward. 1921. Language: An Introduction to the Study of Speech. New York: Harcourt, Brace and Company. Sasse, Hans-Jürgen. 1993. Das Nomen — eine universale Kategorie? Sprachtypologie und Universalienforschung 46:187–221. Sischo, William R. 1979. Michoacán Nahual. In Langacker (1979), pages 307–380. Stassen, Leon. 1997. Intransitive Predication. New York and Oxford: Oxford University Press. Stolper, Matthew W. 2004. Elamite. In The Cambridge Encyclopedia of the World’s 126 REFERENCES Ancient Languages, pages 60–94. Cambridge: Cambridge University Press. Svenonius, Peter. 2007. Interpreting uninterpretable features. Linguistic Analysis 33:375– 413. Swadesh, Morris. 1939. Nootka internal syntax. International Journal of American Linguistics 9:77–102. Tena, Rafael. 2004. Anales de Tlatelolco. México: Consejo Nacional para la Cultura y las Artes (Conaculta). Tuggy, David H. 1979. Tetelcingo Nahuatl. In Langacker (1979), pages 1–140. van Eijk, Jan and Thom Hess. 1986. Noun and verb in Salish. Lingua 69:319–331. Williams, Edwin. 1980. Predication. Linguistic Inquiry 11:203–238. Williams, Edwin. 1981. Argument structure and morphology. The Linguistic Review 1:81–114. 127