Working Papers in Linguistics

Hawaiian kapa design WO R KI NG P A P E RS IN LI NG UI ST I C S The notes and articles in this series are progress reports on work being carried on by students and faculty in the Department. Because these papers are not finished products, readers are asked not to cite from them without noting their preliminary nature. The authors welcome any comments and suggestions that readers might offer. Volume 51 2020 (December) DEPARTMENT OF LINGUISTICS UNIVERSITY OF HAWAI‘I AT MĀNOA HONOLULU, HI 96822 USA http://ling.hawaii.edu An Equal Opportunity/Affirmative Action Institution Fijian masi design DEPARTMENT OF LINGUISTICS FACULTY 2020 Victoria B. Anderson Andrea Berez-Kroeker Robert A. Blust Kamil Deen Katie K. Drager Shelece Easterday Gary Holton Bradley McDonnell William O’Grady Amy J. Schafer James Woodward, Jr. (Adjunct) A special mahalo to Charles-James Bailey for introducing our Departmental Working Papers in 1969 ii DELAYS WITHOUT FILLERS: THE EXAMPLE OF WESTERN SUBANON A. L. BLAKE According to Himmelmann (2014), western Austronesian languages do not have fillers comparable to English uh; instead they use other vocal hesitations, such as lengthened segments within lexical items. This paper explores hesitation phenomena in one western Austronesian language, Western Subanon (ISO639-3: suc), and shows that speakers use a number of additional delay devices besides segmental lengthening. These include a typologically unusual series of partially reduplicated case markers. These findings allow us to expand upon a typology of delay devices proposed by Fox (2010) and indicate that further research on hesitation phenomena in diverse languages is needed. Keywords: Western Subanon, Subanen, Philippine, Austronesian, filler, delay, hesitation 1. INTRODUCTION. Everyday speech rarely approximates the written word. Rather, it is riddled with starts, stops, and asides. Hesitation phenomena (Maclay and Osgood 1959) are an inevitable and essential part of communication. In (1), several hesitation phenomena are shown, such as lengthened vowels (marked by the colon), silence (marked by the period), and semantically-empty expressions (well, you know, uh): (1) well . I mean this . uh Mallet said Mallet was uh said something about uh you know he felt it would be a good thing if u:h . if Oscar went (Clark 1996:259) Aside from empty silence, hesitation phenomena consist of words and other vocalizations which do not add content to the primary message (Clark 1996). As a result, the remainder of the primary message is delayed; for this reason, hesitation phenomena are also referred to as delay devices (Schegloff 1987a). A closely related but conceptually broader notion is that of repair, which subsumes all mechanisms relating to perceived trouble in conversation (Schegloff 1987a). Whether viewed primarily as symptoms of cognitive difficulties, or as helpful listener-oriented signals, the study of delay devices has been informative for many theories of language processing (e.g. GoldmanEisler 1968) and conversational interaction (e.g. Clark 1996). Within linguistics, interest in delay devices arose in the middle of the 20th century. Goldman-Eisler, who developed tools to measure pauses in speech, is often credited with inaugurating the field of “pausology” – a predecessor to the study of hesitation phenomena – in the early 1950s (Griffiths 1990; Dechert and Raupach 1980). Maclay and Osgood (1959), looking at academic conference data, investigated the syntactic positions of English filled and silent pauses, as well as the choice of delay device (their categories of repeat, false start, filled pause, and unfilled pause) in relation to the duration of delay. Their project was later expanded by Schegloff, Jefferson, and Sacks (1977), who situated delay devices and repair devices within an overall description of turn-taking and conversation. These interactional investigations have predominantly been English-centric, with some studies of other major world languages (Finnish, German, Indonesian, Japanese, Spanish, and Swedish) (Kärkkäinen, Sorjonen, and Helasvuo 2007). Even within major world languages, linguists once considered delay devices such as pause vowels to be marginal, or even defective, parts of language (O’Connell and Kowal 2004). This view impacted language description. O’Connell and Kowal (2004) found that out of eleven major English dictionaries published between 1930 and 1990, only one contained an entry for er, two had entries for uh, and two had entries for um. Only in dictionaries published the 1990s or later were these fillers more likely than not to have entries. 1 This lag in adequate description is even more so the case for delay devices in less-studied languages. Although best practices in language documentation call for the inclusion of hesitation phenomena in primary transcripts (Gippert, Himmelmann, and Mosel 2006; Grenoble and Furbee 1 er: seven, uh: five, um: six, out of nine dictionaries between 1990 and 2003. 1 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 2010), the fillers of underdocumented languages are still often overlooked when it comes to language description (Watanabe 2010). In more recent decades there have been attempts to account for hesitation phenomena in a larger variety of languages, such as Kickapoo (Garcia 1994), Wichita (Mirzayan 2008), and Sochiapam Chinantec (Fox, Maschler, and Uhmann 2010), and to typologize these (as in Amiridze, Davis, and Maclagan’s 2010 edited volume). Different languages may prefer different strategies for delaying the next portion of a message (Fox, Maschler, and Uhmann 2010), with not all options preferred by, or possibly even open to, every language. The reasons for these preferences may be phonological, syntactic, or cultural (Fox, Hayashi, and Jasperson 1996). For instance, Schegloff (1987a), references Irene Daden’s report that while Quiche speakers may use “overlong” sound stretches to initiate conversational repair, they do not use brief sound stretches for this purpose. Schegloff, and possibly Daden, imply the phonemic status of “brief” sound stretches in Quiche prevents their co-option for repair. 2 Another example is described by Fox, Maschler, and Uhmann (2010), who examine the strategy of repeating (called recycling) in English, German, and Hebrew. They find all three languages recycle function words, such as prepositions, to delay uttering content words, but only English extensively recycles subject pronouns. They conclude that the variations in frequencies of recycling destination (that is, the first word which is repeated) are due to morphosyntactic differences across the languages examined. The English propensity to repeat subject pronouns is explained by conspiring particularities of English: obligatory expression of the pronominal subject, phonological linking between pronominal subject and verb, and typical occurrence of subject pronouns at the beginning of conversational turns and at the beginning of clauses. Delay devices thus do not haphazardly arise, but are correlated with the other structural elements of a language. While these researchers have focused on discovering language-specific parameters, others have sought universal explanations for repair devices and delay devices. For example, Dingemanse, Torriera, and Enfield (2013) surveyed ten languages and concluded that English-like huh? is a universal repair initiator used when an interlocutor has not clearly heard a speaker. Although he appears to have not done any comparable cross-linguistic investigation, Levelt (1983) proposes universal properties for English-like uh: ‘Uh’ has a special status…[it] is probably also the only interjection, if not the only lexical item, which is universal across languages. There are, surely, phonetic variations in the sound of ‘uh’ between languages, but these may be largely due to the neutral position of the oral cavity for different languages. The conclusion thus seems to be that ‘uh’ is…a neutral sound produced when speech is interrupted at or close to trouble. This, of course, does not exclude the possibility that ‘uh’ acquires some form of derived lexical status related to this basic phenomenon…We may even have to allow for the possibility that this lexical status leads to phonological change, and generalizes to most uses of ‘uh’. This may have happened in Swedish where ‘uh’ is not realized as schwa, but as [ɛ]. (74) In contrast to Levelt’s proposal, Himmelmann (2014) asserts that “in western Austronesian languages there are no conventional pause fillers such as uh (superlengthening of grammatical elements being the standard alternative)” (958). Using Himmelmann’s data from Tagalog and Lauje, in addition to his own Ilokano research, Streeck (1996) posits a “Philippine-type” interactional grammar which differs somewhat from that of Indo-European languages. Also noting the absence of uh in repair initiation, Streeck speculates that the presence of lengthened function words, as well as the use of placeholders (defined in §1.1), may have been grammaticized early in the evolution of Philippine-type languages. If Himmelmann’s claim is accurate, the structure of western Austronesian languages could explain why these languages achieve delay using these features rather than that of Levelt’s proposed universal. Building 2 I am not entirely sure what is meant by a “brief” sound stretch. Schegloff writes, “[The glottal stop is] phonemic in Quiche. Therefore it is not surprising that Daden reports that Quiche speakers do not use the cutoff or stop as a same-turn repair initiator. When English speakers do not use a stop, they often used a sound stretch; they prolong some sound in a word they are producing and then proceed to the repair. Like cutoffs, sound stretches are not phonemic in English. However, brief sound stretches are phonemic in Quiche. They are not used as same-turn repair initiators in Quiche. Quiche speakers do primarily use overlong sound stretches (which are not phonemic for them) to initiate same-turn repair” (1987a:213). 2 A.L. Blake: Delays without fillers: The example of Western Subanon on Fox, Hayashi, and Jasperson’s (1996) comparative work on Japanese and English, several investigations (Fincke 1999, Huang and Tanangkingsing 2005, Wouk 2005) introduce structural explanations for repair in a handful of western Austronesian languages (Bikol, Cebuano and Tsou, and Indonesian, respectively.) These three studies focus on destination of recycling rather than different strategies of repair. Huang and Tanangkingsing’s (2005) investigation note the following as being relevant to determining repair location in these languages: constituent components such as Predicate-Clitic and Verb-Object, consistency of clause structure such as rigidity of word order, and position of the verb (thought to be key in projectability of later elements). These four languages comprise less than one percent of the western Austronesian languages, leaving over 99% undocumented in this area. Although more thorough documentation of delay devices in western Austronesian languages is sorely needed, available evidence indicates that these languages may indeed be typologically unusual in this respect. In this paper, I investigate delay devices in one underdescribed western Austronesian language, Western Subanon, spoken in the southern Philippines. I show that Western Subanon has several different delay devices, including a rather surprising series of partially reduplicated case markers. Western Subanon in turn challenges a typology of delay devices developed by Fox (2010), which overly focuses on the syntactic word at the expense of phonological motivations. The study is structured in the following way: In §1.1, I outline Fox’s (2010) typology of delay devices. Following this, I briefly review the existing literature on western Austronesian delay devices in §1.2. Next, in §1.3, I give a brief overview of the Western Subanon language and introduce this study’s data sources. I describe Western Subanon delay devices in §2 as they relate to Fox’s typology. I investigate Subanon delay devices which do not fit into Fox’s typology in §3. In §4, I relate Subanon delay to that of other languages from a phonological perspective, and §5 calls for more extensive cross-linguistic documentation of delay devices, especially in western Austronesian languages. 1.1 FOX’S TYPOLOGY. Hesitation forms are used to delay the next item of an utterance, but there are many different causes of, or reasons for, delay. Speakers may postpone their speech due to performing other ongoing actions (Clark 1996), or they may still be planning the remainder of their message (Clark and Wasow 1998). Different hesitation forms may be employed at different stages of planning. Fox (2010) characterizes two broad categories of delay devices: 1) those formed by processes which operate on the current word in progress (hereafter Category #1) and 2) those formed by processes which operate between words (hereafter Category #2). Table 1 illustrates the typology of these devices. TABLE 1. Fox’s typology of devices for accomplishing delay delay devices 1. 2. operate on word in progress operate between words cut-offs sound stretches silence discourse markers interjective hesitators non-lexical sounds lexical items but conventional placeholders Although repeating a word or phrase is a common means of “marking time,” Fox does not include simple repetitions, also known as recycling without replacement, within the typology. This is possibly due to the fact that recycling is not a departure from the primary message – rather, it is a reinforcement of some portion of it. Fox reasons that there are only two devices which act on the word which is underway. The current word can either be cut-off in production, or it can be “stretched” (i.e. one or more segments can be lengthened.) These form the first category of delay devices. The second category contains various members: silence, discourse markers, interjective hesitators, and placeholders. Interjective hesitators are paralinguistic signals which are outside of the syntactic frame and can show up at any location in an utterance (Hayashi and Yoon 2006; 2010). Interjective hesitators are 3 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 further divided into 1) “non-lexical but nonetheless conventionalized sounds” such as uh, and 2) lexical items such as demonstratives and interrogative pronouns. Fox uses interjective hesitator rather than filler because she finds that the term filler has been applied too broadly, to the point of it nearly being synonymous with delay devices as a whole. Fox calls the “non-lexical but nonetheless conventionalized sounds” such as uh pause vowels (1). The distinction between the two types of interjective hesitators is therefore one of prototypical filler (e.g. English uh) versus filler word (e.g. Mandarin nàge (那个) ‘that’). Finally, placeholders are special forms which are incorporated into syntactic constructions (Hayashi and Yoon 2006; 2010). Placeholders do more than delay the next item due; they temporarily fill the appropriate syntactic slot for the delayed item. English does not make much use of placeholders, though it does have lexicalized clauses such as whatchamacallit and thingamajig. Serving as nominal placeholders, these can be inflected (i.e. whatchamacallits, thingamajigs). Other languages allow placeholders of different word classes, as well as placeholder stems with derived affixes. In (2), the Italian placeholder stem cos is affixed and temporarily holds the slot for a masculine noun, and for an infinitival verb. (2) a. c’-è un cos-o c’-è un here-be.PRES.3 INDEF.M PH-M here-be.PRES.3 INDEF.M contadin-o in Ponte d’Arbia. farmer-M in Ponte d’Arbia ‘There was a PH [whatchamacallit] … there was a farmer in Ponte d’Arbia’. b. si REFL.3 è mess-o a…a cosare be.AUX.PRES.3 start-PASTP.M PREP PREP PH.INF come si dice a provocare uno zingar-o how REFL.3 say.PRES.3 PREP provoke.INF INDEF.M Gypsy-M ‘[He] started to PH [whatchamacallit] … how to say … to provoke a Gypsy. (Podlesskaya 2010:15) Not included in Fox’s typology are certain alternate forms of words. Examples of these are English a, to, and the produced with unreduced vowels. The function words a, to, and the are usually pronounced with a reduced central vowel; in addition to other purposes, the unreduced forms can be used to signal delay (Clark 1996; Fox Tree and Clark 1997). The unreduced hesitation forms have been termed prolongations (Schnadt and Corley 2006) or elongated forms of words (Arnold et al. 2004). Although the longer duration is the feature which defines a successful delay device, this terminology obscures the additional change in vowel quality. In Clark 1996 and Fox Tree and Clark 1997, this type of hesitation is simply referred to as [vowel] nonreduction. Note that these differ from the aforementioned sound stretches, which are only characterized by longer duration. Although Clark (1996) does not attempt a cross-linguistic typology, his five common “suspension devices” for English include several of Fox’s categories, as well as nonreduction. 3 1.2 DELAY DEVICES IN WESTERN AUSTRONESIAN LANGUAGES. Little work has been done on either delay devices or repair in western Austronesian languages. Notable exceptions include Ewing 2005 for Cirebon Javanese, Wouk 2005 and Williams 2009 for Indonesian; Hsieh and Tanangkingsing 2006 for Kavalan; and Huang and Tanangkingsing 2005 for Tsou. Within Philippine languages, only Bikol (Fincke 1999), Cebuano (Tanangkingsing 2009; Hsieh and Tanangkingsing 2006; Huang and Tanangkingsing 2005), Ilokano (Streeck 1996; Rubino 1996) and Guinaang Kalinga (Cruz 2016) are represented. None of these works attempt a complete description of delay devices available to a language. Cruz 2016 and Rubino 1996 are articles which utilize some repair or delay device data to answer other research questions, while Ewing 2005 and Tanangkingsing 2009 are grammars with some discussion of interactional repair. Hsieh and Tanangkingsing 2006 and Williams 2009 are explorations of a single lexical item (a 3 Clark’s devices are pause, word cut-off, elongation, nonreduction, and filler. 4 A.L. Blake: Delays without fillers: The example of Western Subanon dummy root) or group of items (demonstratives), respectively. Only Fincke 1999, Huang and Tanangkingsing 2005, Streeck 1996, and Wouk 2005 are discussions centered on repair strategies. While is not possible to make conclusions from an absence of an attestation, reviewing these works allows us to postulate some preliminary generalizations. I discuss vowel lengthening, placeholders, and interjective hesitators in these languages. Vowel lengthening as a delay device is evidenced in Cebuano, Ilokano, and Guinaang Kalinga. Ilokano extended sound-stretches are strategies for delaying the next word due; they occur in articles, demonstratives, linkers, conjunctions, and occasionally in prefixes (Streeck 1996). In (3), two stretched Ilokano function words, the conjunction ngem and the demonstrative didiay, are shown. 4 (3) Landlord: Nge:::m sumbatak man didiay::: (.) panangpilosopo di liblibro. ‘But I will address myself to the philosophy of the books’. (Streeck 1996:195) Placeholder stems are common in many western Austronesian languages, including Philippine languages (Blust and Trussell 2015). Their morphosyntactic specifications vary by language; for instance, the Cebuano dummy root kuʔan can replace a NP constituent, verb, or predicate, but its Tagalog cognate kuan can only replace a NP (Tanangkingsing 2009). In Cirebon Javanese, anu is used as a placeholder in (4) The noun it holds the place for, Sipé, is uttered in the following intonation unit. 5 (4) W: .. (H) Dadi ceritané si motor:, therefore story-DEF DEF motorcycle apan tanjakan anu nu. FUT grade WS that.MED .. (H) .. tanjaka:n Sipé nu. grade Sipé that.MED ‘So the story goes the motorcycle was going to go up to the grade, the grade at Sipé.’ (Ewing 2005:105) In addition to their respective dummy stems kua and kuʔan, Ilokano and Cebuano also use distal demonstratives as placeholders. Tanangkingsing (2009) believes that Cebuano kuʔan as well as the demonstrative-plus-linker kanang (kana + -ng) can also be used as fillers. In (5), the speaker utters kanang kuʔan before continuing with the primary message. This appears to be something other than simple placeholder use, as placeholders fit within a specific place in the syntactic frame. (For example, nag-kuʔan could be used in a specific verb stem search.) This may be a strategy to delay while formulating the rest of the clause. It does not seem likely that this is merely a false start, e.g. ‘that whatchamacallit-- I arrived…’ since there is no cut-off or intonational break to signal that nag-abot=ko begins a restart. 6 (5) L: ngano naʔa=man=ka diri, student/ or why EXIST=PAR=2S.NOM here student or J: m: diliʔ, kanang FIL NEG FIL kuʔan nag-abot=ko ato-ng April lang kuan AV-arrive=1S.NOM that-LK April only 4 I have followed Streeck’s glossing and translation convention. The colons indicate lengthening. The period inside parenthesis could possibly indicate a pause. 5 The capital letter stands for the speaker’s initial. Two periods mark a short pause, and line breaks indicate new intonation units. Commas show ontinuing transitional continuity, and periods show final transitional continuity. The (H) indicates inhalation. Ewing follows the transcription convention of Du Bois et al. 1992. To avoid confusion with morpheme boundaries, I have modified Ewing’s transcript so that the colon indicates lengthening. 6 The capital letter stands for the speaker’s initial. The forward slash indicates a rise in terminal pitch. Line breaks indicate new intonation units. Commas show continuing transitional continuity, and periods show final transitional continuity. To avoid confusion with morpheme boundaries, I have modified Tanangkingsing’s transcript to use a colon for what I assume is lengthening. I have also removed one interlinear line. 5 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 L: ‘Why are you here? (Intended: What's your identity enabling you to stay here?) (You're) a student/ or J: ‘M:, no, er: I only arrived last April.’ (Tanangkingsing 2009:217) Similar examples of the dummy stem occurring before the clause, rather than holding a particular syntactic slot during a word search, are evidenced in Ilokano. The dummy word kua occurs in this position in (6). (6) HB: Ah, kua, vocal:hesitations bigla nga n-ag-idda ti kawayan. suddenly LIG PST-INTRANS-LIE:3SABS ART bamboo ‘ah, um, suddenly he lay down on the bamboo…’ (Rubino 1996:657) 7 Thus, Cebuano kuʔan and Ilokano kua seem to function not only as placeholder stems, but also as interjective hesitators. Although it may be that western Austronesian languages use prototypical fillers (pause vowels) infrequently, they are not absent from these languages entirely, contrary to Himmelmann’s (2014) claim. Rubino (1996) glosses Ilokano a as ‘vocal hesitation,’ which seems to indicate that it might be a pause vowel. However, Rubino emphasizes that a can occur mid-word, (as shown in (7)). Demonstratives, too, can be interjected between prefix and root. This is counter to Fox’s (2010) assertation that pause vowels and other interjective hesitators only operate between words. From Rubino’s description, it is unclear whether a can also occur in other syntactic positions. (7) Daytoy bisikleta ti us-usar-en-da this bicycle ART PROG-use-TRANS-3pERG ‘the (aforementioned) bicycle is what they use’ ngaLIG → [pang— nominal:instrum:prefix -a:hesitation marker in the middle of a word -... ala-an-da] iti in-aldaw nga pag-bi-biag-da. take-SUFF-3PERG OBL every:day LIG NOM-PROG:ASP-life-3PERG ‘what they take every day for their livelihood.’ (Rubino 1996:647) Additionally, Indonesian uses e as a pause vowel, as shown in (8). 8 In (8) the question word apa ‘what’ is also used as an interjective hesitator, and the demonstrative itu ‘that’ is used as a placeholder. The use of demonstratives as placeholders in Indonesian is further discussed in Williams 2009. (8) Terus mengenai hadiah hadiah-nya itu, apa dari e: e itu, then about gift gift:GEN DEM what from uh uh DEM e Karang Taruna Nana sendiri, uh Karang Taruna Nana self 7 In Rubino’s convention, three periods indicate a pause greater than half a second. The arrow points to the intonation unit most relevant to discussion. The em dash marks truncation. Line breaks indicate new intonation units. Commas show continuing intonation, and periods show final intonation (phrase final). The brackets appear to mark the word pangalaanda 'what they take.' 8 Wouk uses Conversation Analysis notation. 6 A.L. Blake: Delays without fillers: The example of Western Subanon ‘Then as for the presents, (were they) what from uh that, uh your own Karang Taruna (name of an organization).’ (Wouk 2005:247) Thus, following Fox’s typology, western Austronesian languages likely employ both strategies which operate on the word in progress (i.e. cut-offs and sound stretches). Since discourse markers are not clearly defined by Fox, we cannot say whether western Austronesian languages have them or not. Discourse markers withstanding, all of Fox’s strategies which operate between words (i.e. silence, both non-lexical “pause vowel” and lexical interjective hesitators, and placeholders) appear to be present within western Austronesian languages. These few examples attest that a full range of delay devices occur throughout the western Austronesian languages, including at least one which resembles Himmelmann’s supposedly nonexistent prototypical filler (i.e. Indonesian e) and one which challenges Fox’s notion of how a pause vowel ought to behave, by occurring mid-word (i.e. the a of Ilokano). Before describing the delay devices of Western Subanon, some background information on the language is given in the next section. 1.3 WESTERN SUBANON. Western Subanon (ISO 639-3 suc), or simply Subanon, is a member of the Western Malayo-Polynesian branch of the Austronesian language family. It is one of eight Subanen languages (Lobel 2013) spoken by approximately 125,000 people on the Zamboanga Peninsula on the island of Mindanao in the Philippines (Eberhard, Simons and Fenning 2020). Western Subanon exhibits the typological characteristics of a Philippine-type language. Himmelmman (2005) defines Philippine-type languages as having: 1) at least two undergoer voices, 2) at least one nonlocal phrase marking clitic for nominal expressions, and 3) pronominal second position clitics. Furthermore, Subanon is canonically predicate-initial. Constituent rearranging may be influenced by information structure and other factors. Philippine-type languages are known for their voice system in which the semantic role of one focused or topic NP determines the verbal morphology of a clause. This NP can also be referred to as the trigger or pivot (here glossed as PIV). In a clause with multiple arguments, if the semantic agent is marked as the pivot, the verb must be marked for Actor Voice. If a different argument rather than the agent is marked as pivot, the verbal morphology must instead indicate one of a few Undergoer Voices. Subanon has two different Undergoer Voices, which I call Object Voice and Directional Voice. Semantic patients, beneficiaries, and locations commonly occur as syntactic pivots. In (9a), the pivot og mompalam ‘the mango’ is the semantic patient and requires that the verb stem bogoy ‘give’ be marked with the infix <in>, resulting in the Object Voice form binogoy ‘gave.’ In (9b), the pivot og laki ‘the man’ is the goal or beneficiary, and it triggers the infix <in> and suffix -an to create the Directional Voice verb form binogoyan ‘gave.’ In (9c), the pivot og libun ‘the woman’ is the semantic agent, so the verb is prefixed with mig- to form the Actor Voice verb form migbogoy ‘gave.’ These affixes are fusional, indicating voice and aspect. (9) a. Object Voice: Binogoy nog libun koni og mompalam sog laki koni. NPIV woman this PIV mango LOC man this gave ‘The woman gave the mango to the man.’ b. Directional Voice: Binogoyan nog libun koni og laki koni nog mompalam. gave NPIV woman this PIV man this NPIV mango ‘The woman gave the mango to the man.’ 7 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 c. Actor Voice: Migbogoy og libun koni nog mompalam sog laki. PIV woman this NPIV mango LOC man gave ‘The woman gave the mango to the man.’ 9 All noun phrases are preceded by a noun phrase marker. In various Philippine languages these have been analyzed as determiners, case markers and relation markers, among others (see Himmelmann 2005). Personal and deictic pronouns do not take case markers but are similarly inflected for syntactic status (pivot, non-pivot, oblique). Verbal form and noun case are key to disambiguating meaning. In fluent speech, noun phrase markers in Subanon are unstressed and realized as simple clitics. In this paper, a practical orthography is used in which these clitics are written as separate words. For this study, audio recordings of one native speaker of the Molayal dialect of Western Subanon were made on two different days in February and March of 2016. The recordings were originally made for the purpose of investigating word order variation. Skopeteas et al.’s (2006) Questionnaire for Information Structure (QUIS) was used to ask questions in English, to which the speaker was instructed to respond in Subanon. Eighty stimuli (S1-1 through S1-19, and S1-32 through S1-92) from Field Manual 1 were used. The total length of recorded speech in Subanon from these sessions was around 30 minutes. 10 Added to this are a few examples produced by the same speaker in a Field Methods class at the University of Hawaiʻi at Mānoa. To supplement these data, examples in §2.2.2 are taken from Hall (1987). 2. SUBANON DELAY DEVICES WITHIN FOX’S TYPOLOGY. As shown in Table 1, Category #1 processes operate on the current word in progress, and Category #2 processes operate between words (Fox 2010). 2.1 SUBANON DELAY DEVICES OF CATEGORY #1. As expected, Subanon utilizes delay devices in Fox’s first broad category: those which operate on the current word. Subanon speakers may cut off words and may lengthen segments. Upon seeing the picture stimulus S1-14, in (10), the speaker first stretches og libu:n ‘the woman,’ then produces the relative clause marker nog but cuts it off, and then immediately further delays her speech by pausing (10) Ongon og libu:n, exist PIV woman nogREL (1.2) ompulapula og tuling nog logdoy non reddish PIV color LK clothes 3SG.POSS nog pokpanow dosop nog REL walking also REL % (0.3) bu lolingitan. and angry ‘There is a woman who-… dressed in a reddish-colored dress who is also walking who- and is angry.’ (S1-14) 11 9 Western Subanon is here written in a practical orthography; phonemes mostly correspond to their IPA graphemes with the exceptions of ‘ (for /ʔ/), ng (for /ŋ/), y (for /j/), and o (for /ɔ/). 10 These can be found at http://hdl.handle.net/10125/70151 and http://hdl.handle.net/10125/70152. 11 Transcription conventions are modified from Du Bois et al. 1992. 8 A.L. Blake: Delays without fillers: The example of Western Subanon Stretched sounds can be conjoined with the following word as in (11), where the final syllable of gotow is extended and adjoined to the following noun phrase marker nog as part of a larger prosodic unit. (11) Ongon og goto:w nog miglomot nog bula. exist PIV person REL playing NPIV ball ‘There is a perso:n who is playing with a ball.’ (S1-5) The utterance in (11), then, is produced smoothly and fluidly, despite containing a delay. According to Fox (2010) only the two processes of cut-off and segment lengthening act on the current word in progress as delay devices. In §3.2, I show that Subanon has an additional strategy that acts on the current word in progress. 2.2 SUBANON DELAY DEVICES OF CATEGORY #2. Among Fox’s delay devices which operate between words, Subanon is attested to have silence, discourse markers, placeholders and one kind of interjective hesitator: a pause vowel a. However, data for discourse markers and interjective hesitators is limited to one source. 2.2.1 PLACEHOLDER OGAN. Placeholders fall under Fox’s second broad category: those formed by processes which can operate between words. Western Subanon makes use of a placeholder ogan. Ogan can be treated like other stems; that is, it can take verbal affixes. This is illustrated in (12), where ogan takes the perfective Object Voice affix pig-, and (13), where it takes the Actor Voice affix m-. (12) Pigogan ku og ma'is kitu.' PFV.thingy 1SG.NPIV PIV corn that ‘I thingied the corn.’ (elicited) (13) Na’ a na mogan, na’ a mog- mog- mangoy sog tindaan, NEG 2SG.PIV already AV.thingy NEG 2SG.PIV AV- AV- go LOC store po’ gobi na. because dark already ‘Don’t do the thing, don’t g- g- go to the store because it’s dark already.’ (segment of story) An example of ogan from the QUIS data is shown in (14). Here, ogan temporarily holds the place for barko ‘ship,’ which is produced in the following intonation unit. (14) Yeah. Okay thereOngon dua' buk gotow, exist two CL people mangka migtelescope ilan, and.then telescoping 3PL.PIV Right? And the pAnd bangka dinumonggu’ ogan ken. thingy this and and.then docked Barko ken. ship this ‘Yeah. Okay there, there are two people who are telescoping, right? And the p- and then the thingy was docked. The ship.’ (excerpt from S1-58) 9 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 Ogan can be analyzed as the proclitic noun phrase marker og= (o= before a word beginning with g) combined with the root gan. Hall (1987) lists guan as the placeholder for the Siocon dialect of Western Subanon. It appears to derive from Proto-Western Malayo-Polynesian *kuan ‘to speak, talk, say’, *kua ‘whatchamacallit, filler for word that cannot be recollected’, or *kua-n ‘quotative’ (Blust and Trussel 2015). Like its Cebuano (Tanangkingsing 2009) and Ilokano (Streeck 1996) counterparts, Subanon ogan can take verbal affixes, but the full range of its morphosyntactic distribution has yet to be explored. Streeck (1996) reports that Ilokano kwa “secures the speaker the right for further talk, but for talk that meets specific constraints” (202). He goes on to say, “There are no instances in our materials where a speaker uses the opportunity space that kwa creates for unprojected talk” (202). In the Subanon QUIS data, too, all instances of ogan are followed by the word that ogan has temporarily held the place of. In (15) the very next word produced after og o:gan is the appropriate noun, sulu’ ‘lamp.’ (15) Tubus nion, after that (3.7) og libun kitu' nog sinumupa'- sinumipa' nog bula, woman that REL kicking kicking NPIV ball PIV sinipa' non nosop og o:gan, kicking 3SG.NPIV also PIV thingy sulu'. lamp ‘After that, the woman who was kicking a ball, she's kicking a lamp, too.’ (S1-37) In (16), the word sogan is used to temporarily hold the place for a location. She then follows with the location pogdogangan, market. (16) Po' onda' buan @pokosaluy og gulang bata' koni nog komantis, because not EMP able.to.buy PIV oldest child this NPIV tomato di' non sunan ain og- .. ogo:g, NEG 3SG.NPIV know where PIV PIV.OG dalan posungu' sogog road towards LOC.OG (1.9) sogan, to.thingy pogdogangan. market ‘Because the oldest child wasn't @able to buy tomatoes, he didn't know where the- .. thee street going to the-e ... thingy, market.’ (excerpt from S1-43ii) When ogan holds the place of a noun in these examples, the speaker does not recycle the noun phrase marker; she just produces the bare noun. These manifest as sulu’ ‘lamp’ (not og sulu’) in (15) and pogdogangan ‘market’ (not sog pogdogangan) in (16). The stem ogan always indicates exactly how the repair proper should fit into the original utterance. This cements its clear status as a placeholder. 2.2.2 DISCOURSE MARKERS AND INTERJECTIVE HESITATORS. In addition to silence and placeholders, discourse markers and interjective hesitators fall under Fox’s (2010) category of delay devices which operate between words. Fox gives little description of discourse markers, only noting, “what may be loosely termed ‘discourse markers,’ such as English y’know and like can be used after the current word has come 10 A.L. Blake: Delays without fillers: The example of Western Subanon to completion to delay next word.” (2010:1). Jucker and Ziv (1998) similarly acknowledge that there is no agreed-upon definition of discourse marker. Discourse markers have metalingual functions; their reference is not enclosed within the world described by the text itself. (Maschler 1994; 1998). Discourse markers instead refer to the text, the participant interaction, or cognitive processes (Maschler 1994; 1998). The inclusion of discourse markers in Fox’s (2010) typology of delay devices is warranted, but given that the category can include expressions as diverse as I think, because, and hey (Crible 2018; Jucker and Ziv 1998), practical identification is no small task. A complete description of discourse markers in Subanon is far beyond the scope of this paper. Despite the breadth of the discourse marker category, there are few Subanon candidate members in the QUIS data. There are several reasons for this paucity. Firstly, whereas the QUIS data are mostly short responses, discourse markers may be more likely to be employed within stories and topics which stretch over longer periods of time (Maschler 1998). Secondly, in bilingual conversations, discourse markers are often code-switched (Maschler 1994). In the QUIS data, there are several instances of the speaker codeswitching, using English discourse markers for interactional purposes. It is possible that the majority of discourse markers are primarily used to negotiate interpersonal distance. 12 In (17), the speaker uses the English word right as a discourse marker to check that the researcher is following her description. (17) Yeah. Okay thereOngon dua' buk gotow, exist two CL people mangka migtelescope ilan, and.then telescoping 3PL.PIV Right? And the pAnd bangka dinumonggu’ ogan ken. and and.then docked thingy this Barko ken. ship this ‘Yeah. Okay there, there are two people who are telescoping, right? And the p- and then the thingy was docked. The ship.’ (excerpt from S1-58) When the speaker is instructed with, “You see a fence, a tree, and a well. Give me a short description about what is going on in this scene,” for the static picture S1-10, she produces a longer utterance, as shown in (18). The speaker first introduces the three individuals depicted. Before going on to explain precisely where each person is standing, she produces tu nitu’ ‘and then.’ Here tu nitu’ serves as a discourse marker connecting stretches of speech, rather than furthering narrative action. (18) Ongon og konglangan nog pogindog REL standing exist PIV family sogog .. pili nog timba'. LOC.OG near LK well Og konglangan koni, family this PIV dion og gama', 12 Maschler (1998) divides discourse markers into interpersonal, referential, structural, and cognitive categories. Of the 63 Israeli Hebrew discourse markers Maschler identifies, 35 belong to the interpersonal category. 11 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 there PIV father og gina', mother PIV bu og bata' nilan sola kotow. and PIV child 3PL.POSS one person (1.1) Tu nitu', and then (H) og gina koni dia: PIV mother this exist.at (0.2) ion og dia pogindog sog timba' koni. 3SG PIV exist.at standing LOC well this (1.6) Og laki non koni, PIV man 3SG.POSS this dia pogindog .. sog pili nog kayu. exist.at standing LOC near LK tree ‘There is a family standing near the well. This family, their mother, their father are there, and their one child. ... And then, this mother i:s.. she's who is standing near the well. ... Her husband is standing .. near a tree.’ (excerpt from S1-10) Discourse markers are not only devices for delay; rather, they provide lexical and propositional content (Crible 2018). Discourse markers also have a structuring role beyond simple delay: they delineate boundaries between units of talk (Schiffrin 1987; Crible 2018). We can see that this is in the case in (18), where the speaker utters tu nitu’ ‘and then’ when shifting from introducing the participants to describing each one’s exact location in the scene. It is not at all evident that the primary function of discourse markers is for delay. However, the same can be said for recyclings, as recycled turn beginnings often occur when there is speaker overlap, indicating their use in floor management (Schegloff 1987b). Interjective hesitators, whether pause vowels or filler words, were not found in the QUIS data; the speaker instead used other means of delaying the next portion of the utterance. Hall (1987) considers ah to mark hesitation within sentences in Western Subanon, as shown in (19). (19) Dadi:::, potolu'on ta ini, og solabu so cause-speak (Ipntin) here (tdet) single gilug ta koni si Mister Hall,... brother (Ipntin) this (tper) Mister Hall mangka ta potolu'oy... ah::: gilug ta then (Ipntin) cause-speak (hes) brother (Ipntin) koni nog Timuoy. this (rel) Timuoy ‘Now then, let's have our brother here, Mr. Hall, speak; then, we'll have our, ah, brother here speak, who is the chief.’ (34-35: CA012A) The word dadi, as shown in (19), also appears to be used as a discourse marker in the formal speeches Hall documents. With the inclusion of Hall’s data, we see that Subanon has both discourse markers and at 12 A.L. Blake: Delays without fillers: The example of Western Subanon least one of the two types of interjective hesitators (i.e. pause vowels). Thus, aside from lexical interjective hesitators, all of Fox’s delay devices which operate between words are present in the language. 3. SUBANON PROCESSES OUTSIDE FOX’S TYPOLOGY. Although not specifically classed as a delay device in Fox’s typology, recycling is another cross-linguistic means of delaying the next item due (Fox, Maschler, and Uhmann 2010). Apparently unique to Subanon, and thus also not included in Fox’s typology, are reduplicated forms of noun phrase markers used as delay devices; these are ogog, nogog, and sogog. 3.1 RECYCLING. Subanon selects noun phrase markers as common destinations for recycling. The pivot marker og as well as the start of the NP it heads (i.e. og libun koni ‘this girl’) is recycled twice in (20). (20) Og ligirl PIV Og liPIV girl Og li.bun koni PIV girl this (3.2) bina:lu:ng. shoulder.carry ‘The g- The g- This girl ... is being shouldered.’ (excerpt from S1-11) A re-occurring destination of recycling in the Subanon QUIS data set is the existential word ongon. Ongon often forms a prosodic unit with material that follows it. The speaker recycles to ongon with each successive repair of her introduced NP in (21). She initially produces ongon og gotow: ‘there is a person,’ then replaces this with ongon o:g, … komot ‘there is a: … hand.’ Next, she recycles to ongon again and replaces her NP again, this time with an intended relative clause beginning with a nasal consonant. Ongon og gotow nog ‘there is a person who’. The speaker cuts off this relative clause and recycles to ongon yet again to produce the final repair proper, indicated in brackets. (21) Ongon og gotow:exist PIV person Ongon o:g, exist PIV (1.5) komot- (H) hand (1.3) >Ongon og gotow nog-< exist PIV person REL (4.0) Ongon og gotow nog, exist PIV person REL (3.5) posungu tumagu' nog:, about.to put NPIV 13 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 (0.9) mosolag nog tinidur sog lomisahan koyon. LK fork LOC table this big ‘There is someone- There is a ... hand- There is someone who- ... There is someone who ... is about to put a ... big fork on the table.’ (S1-45) Ongon recycling results in a repair of the whole syntactic clause rather than only the constituent NP; it can also result in a repair of a prosodic word as in (21). 3.2 OGOG, NOGOG, AND SOGOG. Og, as mentioned in §1.3, is the noun phrase marker which indicates the pivot NP. When a noun phrase marker becomes partially reduplicated, however, the result is a delay device. In (22), (23), and (24), the three noun phrase markers og, nog, and sog, have been partially-reduplicated as ogog, nogog, and sogog, respectively. (22) Ongon ogog libun nog exist PIV.OG woman REL (1.5) pokpano:w nog bu mama' nog lolingitan. walking REL and looks COMP angry ‘There is a-a woman ... walking that- and looks angry.’ (S1-12) (23) Ongon dosop og laki:, exist also PIV man nogog pokpanow bu mama' nog lolingitan. walking and looks COMP angry ‘There's also a man who-ah is walking and looks like (he’s) angry.’ (S1-13) REL.OG (24) Og rabbit da og .. gongon sogog dibabow nog gulu ni Molia. rabbit PART PIV existing.thing LOC.OG above LK head PNM.NPIV Maria ‘The rabbit is the .. thing that’s-uh above Maria’s head.’ (S1-73) In its distribution, ogog does not behave like a typical filler word, nor a discourse particle. It is restricted to appearing in the same contexts where the noun phrase marker og does. Likewise, sogog only occurs where sog does, and nogog only replaces nog. 13 Resyllabification takes place when these partially reduplicated noun phrase markers occur. Thus ogog is realized as /oˈgog/ or /o.goːg/, with the second syllable bearing prosodic prominence. Similarly, sogog is realized as /soˈgog/ or /so.goːg/, and nogog as /noˈgog/ or /no.goːg/. In (25) and (26) the speaker describes the picture S1-10, which shows three individuals near various outdoor objects. She begins with Ongon ogog… ‘There is a-a…’ After several IUs, nearing the end of her description, she says Og- og bata nilan ‘The- their child’. Ogog and og- og clearly sound different, as no resyllabification takes place in (26). The token of og- og bata nilan has a very short (<100ms) pause after the initial og-. Thus, one can have an utterance of two og’s in a row without forming ogog. PIV (25) Ongon ogog exist PIV.OG (1.8) 13 Nog has many functions, one of which is to indicate the non-pivot core argument. Nog is also used to link possessors to their possessions and modifiers to their nouns. Furthermore, nog introduces relative clauses and complement clauses. In the QUIS data, there are no tokens of possessive or complementizer nog=og; however, this is probably just due to the small amount of data. 14 A.L. Blake: Delays without fillers: The example of Western Subanon konglangan sog pili nog- (0.2) timba'.(Hx) LOC near LK well family ‘There is a-a … family near a .. well.’ (excerpt of S1-10) (26) Og- og bata' nilan koni dia:child 3PL.POSS this exists.at PIV PIV dia pogindog sog pili no:g exists.at standing LOC near LK (2.0) galad. fence ‘The- Their child i:s is standing near a: … fence.’ (excerpt of S1-10) The Subanon speaker consulted recognized ogog, nogog, and sogog as markers of disfluency. If asked to repeat a phrase containing one of these, she spontaneously corrected it. She also noted that ogog, nogog, and sogog are used by other speakers and are easily recognized characteristics of her language, such that speakers of other languages are aware of them. Ogog and variants are also found in the Siocon dialect as documented by Hall (1987). These delay devices can be lengthened just like other Subanon words. In (27), the speaker produces a cut-off og-, followed by a lengthened ogo:g. (27) Po' onda' buan @pokosaluy og gulang bata' koni nog komantis, because not EMP able.to.buy PIV oldest child this NPIV tomato di' non NEG 3SG.NPIV sunan ain og- (0.7) ogo:g, know where PIV PIV.OG dalan posungu' sog og road towards LOC OG (1.9) sogan, to.thingy pogdogangan. market ‘Because the oldest child wasn't @able to buy tomatoes, he didn't know where the- .. thee street going to the-e ... thingy, market.’ (excerpt from S1-43ii) 4. DISCUSSION. Western Subanon has been found to employ several different kinds of delay device. The most typologically unusual of these are the partially reduplicated noun phrase markers ogog, nogog, and sogog. In this section I discuss phonological motivations for these delay devices. This study is limited in that it only considers 1) auditory data from a single speaker under experimental conditions, and 2) written formal speech collected by Hall (1987). I hope that future documentation of Western Subanon will allow for greater analysis of natural discourse to supplement these preliminary findings. Nevertheless, reviewing the attested Subanon delay devices, we find that many, if not all, of Fox’s subcategories are present. However, the discourse markers and pause vowels (“non-lexical but conventional sounds”) are only found in Hall (1987), not the QUIS data. As discussed in §2.2.2, discourse markers more likely occur in lengthier interactions between speakers. Although Himmelmann’s (2014) claim that western Austronesian languages lack “conventional pause fillers” may be an overstatement, both the absence of interjective hesitators in the QUIS data, and the paucity of pause vowels in Hall’s transcriptions, indicates 15 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 that this type of delay device may indeed be rarely used in Western Subanon. Table 2 summarizes the attested Subanon delay devices. TABLE 2. Delay devices attested in Western Subanon as per Fox 2010’s categories˛ delay devices 1. operate on word in progress 2. operate words between cut-offs sound stretches silence discourse markers interjective hesitators ✓ ✓ ✓ ✓ placeholders ✓ non-lexical but conventional sounds lexical items ✓ Fox’s typology can be extended in several respects. First of all, as discussed in §1.1, recyclings could be included. Recyclings are assumed to be universal and will not be further discussed here. Secondly, the requirement that interjective hesitators operate between words has been shown to be incorrect, at least if we consider Ilokano a to be an interjective hesitator. If a can occur in many syntactic environments, we need to make a third major category (i.e. operates on word in progress and between words) just for it. 14 Next, ogog and the other partially reduplicated Subanon noun phrase markers cannot be placed within the typology. They carry grammatical meaning and fit within the syntactic frame, unlike interjective hesitators. They operate on the word in progress, but not by cutting it off or lengthening one of its segments. However, they function similarly to sound stretches in that they can add phonological weight to function words. At least for Ilokano, sound stretches most frequently occur in the terminal vowel of particles and function words (Streeck 1996). Function words and content words exhibit different phonological properties. English monosyllabic function words alternate between the “strong” stressed form and the “weak” unstressed form depending on their syntactic position; content words have only stressed form (Selkirk 1996). As noted in §1.1, Fox’s typology also does not include stressed forms of English words such as a, to, and the. These also operate on the word in progress, but not by cutting it off or only lengthening a segment (as there is also a change in vowel quality). There is a similarity between stressing English function morphemes and partially reduplicating Subanon function morphemes: namely, both strategies transform a grammatical particle into an independent phonological word. In English, a minimal qualification for wordhood is a primary stress. In Subanon and other Austronesian languages, a minimal word is canonically disyllabic; free morphemes experience pressure to become disyllabic (Blust 2007). As reduplication already exists within the Subanon “toolbox,” it is a natural strategy for the language to use to achieve canonical wordhood. Both English stress and Subanon reduplication iconically increase the amount of time taken to utter the function morpheme, so they work excellently as delay devices. These highly frequent transformed function words precede the speaker’s next content word, adding time before producing it. Their syntactic positions are conducive to conventionalization as exerted by the conversational pressure to delay next item due. Unlike stressed the though, the additional syllable in ogog is considered by Subanon speakers to be a marker of disfluency. Clark (1996) argues that English nonreduced articles produce more fluent-sounding utterances than other delay devices. Comparing English thiy: (unreduced the) to filled (um and uh) or empty 14 An alternate analysis would classify a as a conventional interjective hesitator and re-categorize Ilokano affixes as clitics due to their allowing an intervening syllable, despite this intervening syllable not being a syntactic element. Rubino’s (1996) observations that “significant pauses between prefix(es) and roots occur rather frequently in spoken Ilocano discourse” (646) and “sometimes whole words, such as demonstratives, can intervene between the prefix(es) and root” (649) may seem to strengthen this alternate analysis, until we recall that pauses and demonstratives could also qualify as delay devices which can operate on the word in progress and operate between words. 16 A.L. Blake: Delays without fillers: The example of Western Subanon pauses, he notes the nonreduced articles “give the added illusion of fluency, as if the speakers weren't truly disrupted” (268). From the perspective of metalinguistic awareness then, the additional syllable in ogog is similar to Fox’s subcategory within interjective hesitators, the “non-lexical but nonetheless conventionalized sounds” such as uh and um. We can conclude that English stress (as in thiy), Subanon partial reduplication (as in ogog), and Ilokano sound stretches (as in ti::, nga::, etc.) represent different means of adding phonological weight to function words which normally occur as clitics. Contrary to Fox’s assertion, cut-off and lengthening are not the only means of extending a word in progress to delay next item due. 5. CONCLUSION. It would be remarkable if Western Subanon were the only language in the world to partially reduplicate function words to delay the next part of a speaker’s message. More likely, other languages with similar phonological and prosodic profiles have produced similar innovations. However, this area of grammar is likely seriously underdocumented cross-linguistically. Although Subanon does contain the occasional pause vowel, such as ah, a cursory impression of the language is that there are far fewer of these than occur in languages such as English. Himmelmann’s assertation that western Austronesian languages do not have typical pause fillers may simply attest to their general infrequency, rather than their nonexistence. It is hoped that future corpora studies will confirm this. Furthermore, while the cues relating to perceived fluency have been examined in the fields of second language acquisition, language disorders, and machine processing, these have tended to look exclusively at major languages with well-established literary traditions. We know much less about how speakers of predominantly unwritten languages perceive and process utterances with different kinds of disfluencies. It is hoped that further investigations will address this, in part because ideas of what make a “good speaker,” are highly relevant to endangered language revitalization. Speakers have a universal need for conventionalized strategies to delay some portion of their utterance. With this paper I have attempted to draw attention to the unique ways in which Western Subanon speakers accomplish this task. 17 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 APPENDIX 1. TRANSCRIPTION CONVENTIONS . final intonation unit , continuing intonation unit ? appeal intonation unit - cut-off word . word pronounced as if broken in two .. short pause ... long pause % glottal catch, swallow, gulp, etc. (0.2) silent pause, time in seconds (H) inhalation (Hx) exhalation : lengthening @ laughter > < faster rate ° ° softer 18 A.L. Blake: Delays without fillers: The example of Western Subanon APPENDIX 2. Questionnaire for Information Structure The Questionnaire for Information Structure (QUIS) (Skopeteas et al. 2006) is an investigative tool for collecting data relating to information structure. For the purpose of this study, a section of Field Manual 1 (the picture stimuli S1-1 through S1-92) was used. The elicitor’s version of the manual contains both images and specific prompts which are read to the consultant, whereas the consultant’s version only contains images; a sample from the elicitor’s point of view is shown in Figure 1. The elicitor spoke English, while the consultant responded in Subanon. Sample stimuli are shown in table 3 and table 4. TABLE 3. QUIS sample stimuli S1-33 Elicitor view Consultant view TABLE 4. QUIS sample stimuli S1-12 through S1-15 Elicitor view Consultant view (consultant scrolls down to next image) (consultant scrolls down to next image) 19 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 (consultant scrolls down to next image) 20 A.L. Blake: Delays without fillers: The example of Western Subanon LIST OF ABBREVIATIONS complementizer; EMP emphatic; LK linker; LOC locative marker; NPIV non-pivot core argument marker; PART particle; PIV syntactic pivot marker; PNM personal name marker; POSS possessive marker; REL relativizer COMP 21 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 REFERENCES AMIRIDZE, NINO; BOYD H. DAVIS; and MARGARET MACLAGAN (eds.). 2010. Fillers, pauses and placeholders. Typological Studies in Language 93. Amsterdam: John Benjamins Publishing. ARNOLD, JENNIFER E.; MICHAEL K. TANENHAUS; REBECCA J. ALTMANN; and MARIA FAGNANO. 2004. The old and thee, uh, new: Disfluency and reference resolution. Psychological Science 15(9):578–582. BLUST, ROBERT. 2007. Disyllabic attractors and anti-antigemination in Austronesian sound change. Phonology 24(1):1–36. BLUST, ROBERT, and STEPHEN TRUSSEL. 2015. The Austronesian comparative dictionary, web edition. http://www.trussel2.com/acd. CLARK, HERBERT H. 1996. Using language. Cambridge: Cambridge University Press. CLARK, HERBERT, and THOMAS WASOW. 1998. Repeating words in spontaneous speech. Cognitive Psychology 37:201–242. CRIBLE, LUDIVINE. 2018. Discourse markers and (dis)fluency: Forms and functions across languages and registers. Pragmatics & Beyond New Series 286. Amsterdam: John Benjamins Publishing. CRUZ, LEVI CIRILO. 2016. Documenting conversations using Basic Oral Language Documentation (BOLD). Payap University Master’s Thesis. DECHERT, HANS W., and MANFRED RAUPACH. 1980. Temporal variables in speech: Studies in honour of Frieda Goldman-Eisler. Janua Linguarum. Series Maior 86. The Hague: Mouton de Gruyter. DINGEMANSE, MARK; FRANCISCO TORREIRA; and NICK J. ENFIELD. 2013. Is “Huh?” a universal word? Conversational infrastructure and the convergent evolution of linguistic items. PloS ONE 8(11):e78273. DU BOIS, JOHN W.; SUSANNA CUMMING; STEPHAN SCHEUTZE-COBURN; and DANAE PAOLINO (eds.). 1992. Discourse transcription. Santa Barbara Papers in Linguistics 4. Santa Barbara: UCSB Department of Linguistics. EBERHARD, DAVID M.; GARY F. SIMONS; and CHARLES D. FENNIG (eds.). 2020. Ethnologue: Languages of the world. Twenty-third edition. Dallas, Texas: SIL International. Online version: http://wwwethnologue-com. EWING, MICHAEL C. 2005. Grammar and inference in conversation: Identifying clause structure in spoken Javanese. Studies in Discourse and Grammar 18. Amsterdam: John Benjamins Publishing. FINCKE, STEVEN. 1999. The syntactic organization of repair in Bikol. In Cognition and function in language, ed. by Barbara Fox, Dan Jurafsky, and Laura Michaelis, 252–267. Stanford: CSLI Publications. FOX, BARBARA. 2010. Introduction. In Fillers, pauses and placeholders, ed. by Nino Amiridze, Boyd Davis, and Margaret Maclagan. 1–9. Amsterdam: John Benjamins Publishing. FOX, BARBARA A.; MAKOTO HAYASHI; and ROBERT JASPERSON. 1996. Resources and repair: A crosslinguistic study of syntax and repair. In Interaction and grammar, ed. by Elinor Ochs, Emanuel A. Schegloff, and Sandra A. Thompson, 185–237. Studies in Interactional Sociolinguistics 13. Cambridge University Press. FOX, BARBARA A.; YAEL MASCHLER; and SUSANNE UHMANN. 2010. A cross-linguistic study of selfrepair: Evidence from English, German, and Hebrew. Journal of Pragmatics 42:2487–2505. 22 A.L. Blake: Delays without fillers: The example of Western Subanon FOX TREE, JEAN E., and HERBERT H. CLARK. 1997. Pronouncing “the” as “thee” to signal problems in speaking. Cognition 62(2):151–167. GARCIA, JULE MARIE GOMEZ DE. 1994. Communicative strategies in conversational Kickapoo. University of Colorado at Boulder PhD dissertation. GIPPERT, JOST; NIKOLAUS P. HIMMELMANN; and ULRIKE MOSEL. 2006. Essentials of language documentation. Trends in Linguistics 178. Berlin: Walter de Gruyter. GOLDMAN-EISLER, FRIEDA. 1968. Psycholinguistics: Experiments in spontaneous speech. London: Academic Press. GRENOBLE, LENORE A., and N. LOUANNA FURBEE, (eds.). 2010. Language documentation: Practice and values. Amsterdam: John Benjamins Publishing. GRIFFITHS, ROGER. 1990. Pausology and listening comprehension: Theory, research and practice. JALT Journal 12(1):99–120. HALL, WILLIAM C. 1987. Aspects of Western Subanon formal speech. Publications in Linguistics Number 81. Dallas: Summer Institute of Linguistics. HAYASHI, MAKOTO, and KYUNG-EUN YOON. 2006. A cross-linguistic exploration of demonstratives in interaction: With particular reference to the context of word-formulation trouble. Studies in Language. International Journal sponsored by the Foundation “Foundations of Language” 30(3):485–540. HAYASHI, MAKOTO, and KYUNG-EUN YOON. 2010. A cross-linguistic exploration of demonstratives in interaction: With particular reference to the context of word-formulation trouble. In Fillers, pauses and placeholders, ed. by Nino Amiridze, Boyd Davis, and Margaret Maclagan, 33–66. Amsterdam: John Benjamins Publishing. HIMMELMANN, NIKOLAUS. 2005. The Austronesian languages of Asia and Madagascar: Typological characteristics. In The Austronesian languages of South East Asia and Madagascar, ed. by K.A. Adelaar and N.P. Himmelmann, 110–181. London: Routledge. HIMMELMANN, NIKOLAUS. 2014. Asymmetries in the prosodic phrasing of function words: Another look at the suffixing preference. Language 90:927–960. HSIEH, FUHUI, and MICHAEL TANANGKINGSING. 2006. The empty root in Cebuano and Kavalan: A cognitive perspective. Paper presented at the 10th International Conference on Austronesian Linguistics, Palawan, Philippines. HUANG, HUEI-JU, and MICHAEL TANANGKINGSING. 2005. Repair in verb-initial languages. Language and Linguistics 6(4):575–597. JUCKER, ANDREAS H., and YAEL ZIV (eds.). 1998. Discourse markers: Descriptions and theory. Amsterdam: John Benjamins Publishing. KÄRKKÄINEN, ELISE; MARJA-LEENA SORJONEN; and MARJA-LIISA HELASVUO. 2007. Discourse structure. In Language typology and syntactic description 2, ed. by Timothy Shopen, 301–371. Cambridge: Cambridge University Press. LEVELT, WILLEM J.M. 1983. Monitoring and self-repair in speech. Cognition 14:41–104. LOBEL, JASON WILLIAM. 2013. Philippine and North Bornean languages: Issues in description, subgrouping, and reconstruction. University of Hawai‘i at Mānoa PhD dissertation. MACLAY, HOWARD, and CHARLES E. OSGOOD. 1959. Hesitation phenomena in spontaneous English speech. Word 15(1):19–44. 23 University of Hawai‘i at Mānoa Working Papers in Linguistics 51 MASCHLER, YAEL. 1994. Metalanguaging and discourse markers in bilingual conversation. Language in Society 23(3):325–366. MASCHLER, YAEL. 1998. Rotselishmoa ké ta? ‘wanna hear something weird/funny [lit.‘a segment’]?’: The Discourse markers segmenting Israeli Hebrew talk-in interaction. In Discourse markers: Descriptions and theory, ed. by Andreas H. Jucker and Yael Ziv, 13–59. Amsterdam: John Benjamins Publishing. MIRZAYAN, ARMIK. 2008. A preliminary study of same-turn self-repair initiation in Wichita conversation. In Lessons from documented endangered languages, ed. by David K. Harrison, David S. Rood, and Arienne Dwyer, 317–354. Typological Studies in Language 78. Amsterdam: John Benjamins Publishing. O'CONNELL, DANIEL C., and SABINE KOWAL. 2004. The history of research on the filled pause as evidence of the written language bias in linguistics. Journal of Psycholinguistic Research 33(6):459–474. PODLESSKAYA, VERA I. 2010. Parameters for typological variation of placeholders. In Fillers, pauses and placeholders, ed. by Nino Amiridze, Boyd Davis, and Margaret Maclagan. 11–32. Amsterdam: John Benjamins Publishing. RUBINO, CARL. 1996. Morphological integrity in Ilocano: A corpus-based study of the production of polymorphemic words in a polysynthetic language. Studies in Language. International Journal sponsored by the Foundation “Foundations of Language” 20(3):633–666. SCHEGLOFF, EMANUEL A.; GAIL JEFFERSON; and HARVEY SACKS. 1977. The preference for self-correction in the organization of repair in conversation. Language 53:361–382. SCHEGLOFF, EMANUEL A. 1987a. Between micro and macro, contexts and other connections. In The Micro–Macro Link, ed. by Jefferey C. Alexander, Bernhard Giesen, Richard Münch, and Neil J. Smelser, 207–234. Berkeley: University of California Press. SCHEGLOFF, EMANUEL A. 1987b. Recycled turn beginnings: A precise repair mechanism in conversation’s turn-taking organization. Talk and social organisation 1:70–85. SCHIFFRIN, DEBORAH. 1987. Discourse markers. Studies in Interactional Sociolinguistics 5. Cambridge: Cambridge University Press. SCHNADT, MICHAEL J., and MARTIN CORLEY. 2006. The influence of lexical, conceptual and planning based factors on disfluency production. In Proceedings of the Annual Meeting of the Cognitive Science Society, 28. SELKIRK, ELISABETH. 1996. The prosodic structure of function words. In Signal to syntax: Bootstrapping from speech to grammar in early acquisition, ed. by James L. Morgan and Katherine Demuth, 187– 213. Mahwah, NJ: Lawrence Erblaum Associates. SKOPETEAS, STAVROS; INES FIEDLER; SAM HELLMUTH; ANNE SCHWARZ; RUBEN STOEL; GISBERT FANSELOW; CAROLINE FÉRY; and MANFRED KRIFKA. 2006. Questionnaire on information structure (QUIS): reference manual 4. Universitätsverlag Potsdam. Online: http://www.sfb632.unipotsdam.de/en/quis. STREECK, JÜRGEN. 1996. A little Ilokano grammar as it appears in interaction. Journal of Pragmatics 26:189–213. TANANGKINGSING, MICHAEL. 2009. A functional reference grammar of Cebuano. National Taiwan University PhD dissertation. WATANABE, HONORÉ. 2010. Fillers and their relevance in describing Sliammon Salish. In Fillers, pauses and placeholders, ed. by Nino Amiridze, Boyd Davis, and Margaret Maclagan. 173–187. Amsterdam: John Benjamins Publishing. 24 A.L. Blake: Delays without fillers: The example of Western Subanon WILLIAMS, NICHOLAS. 2009. Toward a linguistic anthropological account of deixis in interaction: Ini and itu in Indonesian conversation. Colorado Research in Linguistics 22(1):1–23. WOUK, FAY. 2005. The syntax of repair in Indonesian. Discourse Studies 7(2):237–258. 25