The Papuan Languages of Timor, Alor and Pantar: Volume 2 9781614519027, 9781614519065

Citation preview

Antoinette Schapper (Ed.) The Papuan Languages of Timor, Alor and Pantar: Volume 2

Pacific Linguistics

Managing editor Bethwyn Evans Editorial board members Wayan Arka Mark Donohue Nicholas Evans Gwendolyn Hyslop David Nash Bill Palmer Andrew Pawley Malcolm Ross Jane Simpson

Volume 655

The Papuan Languages of Timor, Alor and Pantar: Volume 2 Sketch Grammars

Edited by Antoinette Schapper

ISBN 978-1-61451-906-5 e-ISBN (PDF) 978-1-61451-902-7 e-ISBN (EPUB) 978-1-5015-0033-6 ISSN 1448-8310 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. 6 2017 Walter de Gruyter Inc., Boston/Berlin Cover image: Emilie Wellfelt Typesetting: RoyalStandard, Hong Kong Printing and binding: CPI books GmbH, Leck ♾ Printed on acid-free paper Printed in Germany www.degruyter.com

Table of contents Abbreviations

vii

1

Antoinette Schapper 1 Introduction

2

Louise Baird 55 Kafoa

3

Glenn Windschuttel & Asako Shiohara 109 Kui

4

Nicholas J. Williams 185 Kula

5

Juliette Huber Makalero and Makasae

Index

353

267

Abbreviations Leipzig Glossing Rules (http://www.eva.mpg.de/lingua/resources/glossing-rules.php) are used throughout this volume. The table below summarizes the glosses used. 1 2 3 4 A ACT ADDR ADJ ADV AFF AGT AGR ALIEN AN

AP APPL ARG ART ASP ASSOC ATTR CAUS

CEA CLF CLS CMN CMP COM COMPL COND CONJ CONT CONTR DAT

DC DEF DEM DET DIR

first person second person third person fourth person agent-like argument of transitive verb actor addressee adjective adverbial affectedness agentive agreement alienable animate Alor-Pantar languages applicative argument article aspect associative plural attribute causative Central-Eastern Alor languages classifier clausal pro-form common person complement comitative completive conditional conjunction continuative contrastive dative deictic center definite demonstrative determiner direction

DISC DIST DISTR DU

EA EMPH

ET EXCL EXT. NEG EVID FIN FOC GEN HIGH HUM IMM IMP INAL INCP INCL INCORP INT INTS INTER INV IPFV LEVEL LOC LOW

MKL MKS MOD

N NEG NFIN NHUM NMLZ NOM

NP NUM

discourse marker distal distributive dual Eastern Alor languages emphatic Eastern Timor languages exclusive external negation evidential final focus genitive elevation higher than DC human immediate future imperative inalienable inceptive inclusive incorporated intentional intensifier interjection inverse imperfective elevation level with DC locative elevation lower than DC Makalero Makasae modifier noun negative non-final non-human nominalizer nominative noun phrase numeral

viii

OBJ OBL

P PAT PFV PL POSS POT PRED PRIOR PROSP PROG PROH PROX PSR QM QUANT

R RC RDP

Abbreviations

object oblique patient-like argument of transitive verb patientive perfective plural possessive potentive predicate priorative prospective progressive prohibitive proximal possessor question marker quantifier recipient argument of ditransitive verb relative clause reduplicant

RECP REC . PST REFL REL REM . PST

S SBJ SEQ SG SIM SPEC SPKR STAT

T TAP TOP

TNG TRUNC

V VBLZ

reciprocal recent past reflexive relative remote past single argument of intransitive verb subject sequential singular simultaneous specific speaker stative theme/gift argument of ditransitive verb Timor-Alor-Pantar languages topic Trans-New Guinea languages truncated verb verbalizer

Antoinette Schapper

1 Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2 1 Overview The Papuan languages of Timor, Alor and Pantar are one of the best described non-Austronesian language families of Melanesia. There are today more than fifteen linguists actively working on Timor-Alor-Pantar (TAP) languages. As documentary work on TAP languages continues, data from them is also increasingly coming to the attention of comparative linguists. Nonetheless, published descriptions of TAP languages still only cover a small number of languages and do not capture the full extent of the diversity within the family. These volumes bring together previously unpublished descriptions of TAP languages, thus providing a comprehensive overview of the variation across the entire TAP family and bringing to light the many noteworthy features of individual TAP languages. In this second volume, four new descriptions are presented. Like its predecessor, this volume bears witness to the great flourishing of TAP linguistics. Given the amount of work being carried out on TAP languages, any work that attempts to summarize the field is bound to be in need of updating within a short time. In the introduction to the first volume of TAP sketches (Schapper 2014a), I provided a short overview of the history of the study of the family and a brief survey of the state-of-the-art from typological and historical perspectives. Inevitably, in the introduction to this, the second volume of TAP sketch grammars, I present updates to the understanding of TAP languages offered by some of the most recent work. Section 2 presents the TAP languages, their locations and the number of speakers. Section 3 outlines the most recent works on the TAP languages. Sections 4 and 5 on the history and typology of the family have been particularly expanded since Volume 1 with the aim of giving a sense of the emerging issues in TAP linguistics. Section 6 overviews the chapters in the present volume.

2 The language scene The term “Papuan” is used to designate any language that is found in the area of New Guinea that does not belong to the Austronesian or Australian language families. Papuan languages do not form a single genealogical entity, but comDOI 10.1515/9781614519027-001

2

Antoinette Schapper

Map 1: The extent of Papuan languages (on and around New Guinea)

prise anywhere between 30 and 60 distinct languages families, depending on the individual researcher’s perspective. Lying some eight hundred kilometers from the New Guinea mainland as the crow flies, Timor-Alor-Pantar languages are the most westerly of the Papuan languages (Map 1). Their great distance from New Guinea means that the TAP languages are the most outlying of all the Papuan “outliers”, that is, Papuan languages that are not spoken on New Guinea. The TAP family comprises around 30 languages spoken by an estimated 300,000 people in south-eastern Indonesia and Timor-Leste (East Timor). Members of the family predominate on the islands of the Alor archipelago, consisting of Alor, Pantar and numerous small islands in the Pantar Strait. Other members are found interspersed among Austronesian languages on the islands of Timor and Kisar (Map 2). On Timor, there are four TAP languages: Bunaq straddling the border between Indonesian West Timor and independent East Timor, and Makasae, Makalero and Fataluku occupying a contiguous region at the eastern tip of the island. Close by to the north, again in Indonesia, Oirata is a TAP language spoken on Kisar, an island dominated by the Austronesian language, Meher. The greatest concentration of TAP languages is in the Alor archipelago. On current counts, there are ten TAP languages on Pantar and in the Pantar Straits and 16 on Alor (Map 3). The maps of these languages are constantly evolving as the linguistic geography of the islands becomes more apparent, not only with better surveys but also with in-depth studies. For example, two previously unlabeled languages have in the last few years been identified on Pantar island,

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

3

Map 2: The extent of the Timor-Alor-Pantar family

Map 3: The Alor-Pantar languages

despite its small size: Kaera, recently assigned the ISO-639 code [jka], is first mentioned in Klamer (2010a) then in Ino (2013) and Klamer (2014); Kroku, spoken in Modebur and surrounding villages in northwest Pantar, has recently emerged as its own language on the basis of recordings made by Hein Steinhauer in the

4

Antoinette Schapper

1970s (see preliminary data in Steinhauer 1995), but still lacks its own ISO-639 code.1 The changing numbers of languages on Alor and Pantar reflect, at least in part, the problem of determining what constitutes a separate language in this complex linguistic environment. Mutual intelligibility is typically difficult to distinguish from passive bilingualism. For example, according to native speakers, Teiwa and Sar are partially mutually intelligible, but Di’ang and Nedebang are not normally understood by Teiwa speakers (Amos Sir pers. comm.). At the same time, most Nedebang speakers understand Teiwa, as the dominant neighboring language (Karel Lamma pers. comm.), despite the significant phonological differences between the languages. Consequently, the independent language status of these four languages may not be equally justified, but capturing the exact relationships between them is not easily done without targeted socio-linguistic and dialectal study. Similarly, in central Alor, the languages of Kamang, Suboo, Moo, Tiee and Manet are closely related and together form the Northcentral Alor language area (Schapper & Manimau 2011). Speakers typically claim partial passive competence in one or more of the other group’s speech. Still the differences between them are not superficial in the details of the segmental phonology and morphology (such as in the number and functions of agreement paradigms) that appear to warrant each of them being regarded as an independent language (compare, for instance, Schapper 2014b on Kamang and Han 2015 on Suboo). Contributing to the problem of what constitutes a distinct language is the lack of unique glottonyms that are widely recognized and accepted for AlorPantar (AP) languages. A language may be referred to by the name of the village where it is spoken or the geographical location of its speakers: for instance, the Wersing language is often designated locally as Bol, though this is a general label that is used by speakers of montane Alor languages (e.g., Kamang, Kula etc.) in reference to any coastal group, while Wersing speakers themselves tend to refer to their language as Bahasa Kolana ‘the language of Kolana’, Bahasa Pureman ‘the language of Pureman’ etc. after their respective villages of origin. In other cases, a language may be designated simply as our language, as is the case with Blagar (Steinhauer 2014) or Teiwa (Klamer 2010b). Alternatively, a language may be referred to with the name of the clan or lineage group that speaks it; for instance, Lamma, a name used by Stokhof (1975) and Nitbani et al. 1 There may be still more languages in Pantar. Amos Sir, a local of Pantar, for instance, reports the existence of a language named Luul still spoken by a handful of people in the villages of Lelafang, Tonte, Adiabang, and Debu. At this stage, however, we have no data with which to assess whether Luul is its own language or a dialect of Sar.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

5

(2001), is a single clan within the larger West Pantar language area (Holton 2014), or Tanglapui, used by Stokhof (1975) and Donohue (1996), is the name of one of four Kula-speaking clans (Wellfelt 2016: 76–78). A common pattern on Pantar is to use the name of the royal lineage group (Indonesian suku raja) as the name for the language spoken by the whole group: many of the remaining Nedebang speakers prefer their language to be referred to as Klamu, the name for the royal lineage, rather than Nedebang, the name of the group’s origin village which has been adopted in the linguistic literature since Stokhof (1975). The TAP languages spoken on Timor do not typically have the same issues with glottonyms. In general, there appears to be an accepted name for each language, and this will typically be recognized both by speakers themselves as well as others outside the speech community. Individual dialect groups often have a distinct name for themselves and their language, but still identify with the wider cover label when used. For instance, the name Gaiʹ is used by the Bunaq communities between Bobonaro and Maliana in reference to themselves, but they consider themselves part of the wider Bunaq-speaking group. The main issue in Timor is that we know very little of the full extent of dialect variation in the TAP languages spoken there. For instance, there appear to be several intermediate language varieties between Makasae and Makalero. Most notable is Saʹani spoken around Luro. It is not clear whether this variety is closer to Makasae or Makalero, or whether it is different enough from both to be considered a separate language. Similar questions may be asked of divergent dialects of Bunaq and Fataluku. There is a sharp divide between the TAP languages of Timor and those of Alor and Pantar in terms of their vitality. The Timor languages are by far and away the largest in the family: Bunaq, Fataluku and Makasae together make up seventy per cent of TAP speakers. Although existing under the shadow of more dominant Austronesian languages, these languages are vigorous and still being learnt by children in the villages. The largest and most vibrant of the AP languages are Abui and West Pantar with 16,000 and 10,000 speakers respectively and active child acquisition. Other languages such as Nedebang are nearmoribund with all really fluent speakers over 50 years of age. For the majority of TAP languages, the situation is not so dire, but language shift is progressing. The typical situation in villages may be broadly characterized as follows: many adults still use the local languages regularly, but switching to the lingua franca, Malay/Indonesian in the Indonesian areas or Tetun in Timor-Leste, is common particularly in addressing children; some but not all children acquire the local language, but often only passively, and prefer to use the lingua franca with their peers.

6

Antoinette Schapper

3 Recent studies in TAP linguistics The tremendous boom in TAP linguistics which took place in the first decade of the 21st century (described in Schapper 2014a: 10–11 in brief and in Schapper & Huber 2012 in detail) has not abated. Significant advances in the description and documentation of the TAP languages continue to be made by teams of scholars around the world. Perhaps most notable is the increasingly prominent contribution of scholars from Indonesia to the field of TAP linguistics. Since 2012, the Lembaga Ilmu Pengetahuan Indonesia [Indonesian Institute of Sciences], or LIPI, has engaged in documenting endangered Papuan languages across the archipelago, including smaller languages belonging to the TAP family. This work has resulted in written materials for at least four TAP languages:2 for Oirata, a dictionary (Nazarudin 2013), a short film on language endangerment (see: https://www.youtube.com/ watch?v=t6rgw-17-EI), several article-length treatments (Nazarudin 2015, van Engelenhoven & Nazarudin 2016), as well as a thesis focussing on contact with Austronesian languages (Nazarudin in preparation); for Kafoa, several treatments of language as cultural heritage (Santosa 2012, Humaedi et al. 2013, Humaedi et al. 2014, Patji 2014), a short film (http://www.youtube.com/watch?v=PUdHUog8UGE), and a short dictionary (Widhyasmaramurti 2013), and; for Kui, a sociolinguistic study (Katubi & Thung Ju Lan 2014) and a dictionary (Katubi, Ju Lan & Akoli 2013). Historical linguistic contributions have been made by Ino (2013) and Mandala (2010), both theses produced at Udayana University in Denpasar. At the same time, archival materials on TAP languages are becoming available online. The Endangered Language Archive (ELAR) has digital collections for Kula (https://elar.soas.ac.uk/Collection/MPI971878) and the Bunaq oral literature genre zapal (https://elar.soas.ac.uk/Collection/MPI1029743). The Pacific And Regional Archive for Digital Sources in Endangered Cultures (PARADISEC) has deposits for several west Alor languages (http://catalog.paradisec.org.au/collections/LRB4) as well as more significant corpora for Klon (http://catalog.paradisec. org.au/collections/LRB3), Kafoa (http://catalog.paradisec.org.au/collections/LRB1) and West Pantar (http://catalog.paradisec.org.au/collections/GH2). Materials for numerous TAP languages are found in The Language Archive (https://corpus1. mpi.nl/): among the Lund corpora are materials for the three TAP languages of Eastern Timor (http://hdl.handle.net/10050/00-0000-0000-0003-7BE0-7@view), 2 This documentation built on a pilot project conducted by LIPI on Hamap, a TAP language spoken in western Alor. Publications from this initial work include Katubi et al. (2004), KledenProbonegoro, Katubi & Tondo (2005), Kleden-Probonegoro, Katubi & Tondo (2007), KledenProbonegoro (2008).

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

7

and; among the LAISENG corpora are multiple wordlists and text collections for AP languages (https://hdl.handle.net/1839/00-0000-0000-0000-0000-4@view). Whilst these collections are made by and for linguists, there are also archival collections of TAP languages by researchers from other fields. The archive of Cora Du Bois at Harvard University (http://oasis.lib.harvard.edu/oasis/deliver/ ~toz00001) contains the earliest texts and description of an Alor language, but unluckily no recordings.3 The collections of music, song and ritual speech amongst the Bunaq made in the 1950s, 1960s and 1970s by Claudine Friedberg (http://archives.crem-cnrs.fr/archives/collections/CNRSMH_I_1973_003/) and Louis Berthe (http://archives.crem-cnrs.fr/archives/collections/CNRSMH_I_1970_068/; http://archives.crem-cnrs.fr/archives/collections/CNRSMH_I_1970_069/) together represent an unparalleled record of dying verbal arts in the region. Ros Dunlop’s collection of music from Timor-Leste (http://catalog.paradisec.org.au/collections/ RD1) also includes performances by Bunaq, Fataluku and Makasae groups. Theses at a range of degree levels continue to be produced on TAP languages. In 2015 and 2016 alone, there were no fewer than seven theses on TAP topics brought to completion: on the morpho-syntax of Suboo (Han 2015), on the grammaticalization of serial verbs in AP languages (Willemsen 2015), on Fataluku phonology (Heston 2015), on socio-linguistics in the Fataluku-speaking region (Savio 2016), on Abui phonology and phonetics (Delpada 2016), on oral traditions in Alor (Wellfelt 2016), and on place reference in Kula (Williams 2016). Half a dozen more student projects are underway such that these will soon be followed by more studies.

4 Historical linguistics of the TAP family Papuan historical linguistics has been notoriously mired in controversial classifications based on a mix of typological and lexicostatistical arguments (Pawley 2005). The availability of reliable descriptive data for many TAP languages means that the application of historical linguistic methods is now possible. Today, historical linguistics has perhaps replaced language description as the busiest area of TAP linguistics. Still much careful reconstructive work needs to be done and many critical questions of TAP prehistory remain unanswered.4 3 Du Bois took a recording device with her to Alor on her pioneering fieldwork from 1938 to 1939, but she was never able to get it functioning in the field. 4 The following sources have been used in this and subsequent sections: Bunaq, Schapper 2010, Schapper fieldnotes; Makalero, Huber 2011, Huber this volume; Makasae, Huber 2008, Correia 2011, Huber this volume; Fataluku, van Engelenhoven forthcoming, Heston 2015; Oirata, Nazarudin 2013, 2015; Nedebang, Schapper 2017, Schapper forthcoming, Schapper fieldnotes;

8

Antoinette Schapper

4.1 Internal relationships For a long time many scholars assumed from the geographical proximity of the TAP languages that they must be related. However, it was only recently, with the publication of Schapper, Huber and van Engelenhoven (2014), that we have the first proof of the relatedness of the TAP languages using the comparative method. There are three primary subgroups of the TAP family: the Alor-Pantar subgroup, the Eastern Timor subgroup and Bunaq (Figure 1). Schapper, Huber & van Engelenhoven (2014) assumed two primary subgroups, namely Proto-Timor (Bunaq and Eastern Timor languages) and Proto-Alor-Pantar (PAP), because they were comparing the results of separate reconstructive work on the Timor languages on the one hand (Schapper, Huber & van Engelenhoven 2012), and on the AP languages on the other (Holton et al. 2012). However, Bunaq shares no innovations with either Eastern Timor or Alor-Pantar languages, and must be regarded as its own primary subgroup.

Figure 1: Tree diagram of Timor-Alor-Pantar languages

While the internal constituency of the Eastern Timor subgroup is already clear (Schapper, Huber & van Engelenhoven 2012, Huber 2016), the internal relationships of the languages in the AP subgroup are only being worked out now. Figure 2 presents a provisional tree of the AP languages based on ongoing reconstructive work.

West Pantar, Holton 2014; Teiwa, Klamer 2010b, Schapper fieldnotes; Kaera, Klamer 2014; Blagar, Steinhauer 2014; Reta, Willemsen forthcoming; Adang, Haan 2001; Klon, Baird 2008; Kafoa, Baird this volume; Kui, Windschuttel & Shiohara this volume, Katubi, Ju Lan & Akoli 2013; Kiraman, Schapper fieldnotes; Abui, Kratochvíl 2007, Delpada 2016, Schapper fieldnotes; Kamang, Schapper 2014b, Schapper fieldnotes; Sawila, Kratochvíl 2014, Kratochvíl, Bantara & Malaikosa 2014; Kula, Williams this volume; Wersing, Schapper & Hendery 2014, Schapper fieldnotes.

Figure 2: Possible tree diagram of the Alor-Pantar languages† † The position of the Straits languages remains problematic. Some innovations suggest they group with the Bird’s Head languages (e.g., PAP *g > ʔ), while other features suggest they don’t (e.g., not reflecting the change PAP *do ‘LOW ’ > pAlor *to).

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

9

10

Antoinette Schapper

While the complex phonological history of the AP languages still needs careful dissection (see Schapper 2017 for a statement of this), major divisions in the tree in Figure 2 increasingly find support on morphological grounds. For instance, the division of the AP subgroup into two primary branches, Alor and Pantar, is supported by irregular innovations in morphology. On the one hand, the Alor languages share the irregular change of *d > *t in the PTAP morpheme *d[o/e] ‘LOW ’ > PAP *do > pAlor *to. This morpheme is part of a paradigm of deictics encoding three elevationals. Table 1 sets out the PTAP morphemes and their reflexes (extended from Schapper 2014c). Table 1: Reflexes of PTAP elevationals PTAP

Reflexes

Level

*m[o/e]

Eastern Timor: Makalero -me-; Alor-Pantar: West Pantar mau, Blagar mo, Adang mɔ(ŋ), Kamang mu(ŋ), Wersing mo(na), Sawila ma(na)

High

*d[o/e]

Eastern Timor: Makalero -de- Makasae de-/do-; Alor-Pantar: West Pantar dau, Kaera de, Blagar do, Adang tɔ(ŋ), Klon ta, Kamang tu(ŋ), Wersing to (na), Sawila ta(na)

Low

*p[o/e]

Eastern Timor: Makalero -fe-, Makasae he-/ho-; Alor-Pantar: West Pantar pau, Kaera pe, Blagar po, Adang pɔ(ŋ), Kamang fu(ŋ)

On the other hand, the Pantar branch is supported by the irregular loss of *d from the PTAP reflexive prefix *dV-, which appeared on both nouns and verbs, as reflected by Bunaq d-, Abui da-, do-, de-, Adang sa-, Kafoa do-.5 Proto-Pantar dropped *d and simply used the vowels *a- and *i- (following the vowel grade pattern typical of TAP pronominals (see section 5.2, Tables 12 to 17) to crossreference singular and plural subjects, respectively. This is illustrated for two Pantar languages, Teiwa in (1) and Nedebang in (2).

(1)

Teiwa reflexive prefixes a. A sapatu ma a-ɸat shoe OBL REFL . SG -leg 3SG ‘S/he put the shoes on.’ b.

miaʔ. put

Iman sapatu ma i-ɸat miaʔ. shoe OBL REFL . PL -leg put 3PL ‘They put their shoes on.’ (Schapper fieldnotes)

5 In many TAP languages, reflexive prefix(es) have additional functions marking 4th persons (Holton 2010, 2014) and “middle”-type events in which there is little semantic differentiation between the role of the actor and undergoer (see, e.g., the description of Bunaq d- in Schapper 2010: 377–391).

11

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

(2)

Nedebang reflexive prefixes a. Gaŋ a-ɸaːra. 3SG REFL . SG -kill ‘S/he killed him/herself.’ b.

Giŋ i-ɸaːra. 3PL REFL . PL-kill ‘They killed themselves.’ (Schapper fieldnotes)

Greater descriptive coverage of the grammars of TAP languages has also allowed for the identification of grammatical innovations. For instance, thanks to detailed descriptions of ‘give’ constructions across the TAP family, Klamer & Schapper (2012) identified the “double take” give construction as a shared innovation in Eastern Timor languages. Similarly, innovative dual pronouns involving the suffixation of the numeral PAP *nuku ‘one’ onto a pronominal base (C)Vna- are found in Kamang, Wersing, Kula and Sawila (Table 2). This construction replaced the PTAP duals formed with the morpheme *-l[e/i] (see Table 3) and can be used, along with other innovations, to place these languages together in a Central-Eastern Alor (CEA) subgroup. Table 2: Reflexes of innovative PCEA duals *(C)Vna-nuku Kamang

1EXCL 1INCL 2 3

ninnok sinnok innok gannok

Wersing

1EXCL 1INCL 2 3

ɲinuk tinuk jinuk ginuk

Kula

1EXCL 1INCL 2 3

ngɐnɐku inɐku jɐnɐku (g)ɐnɐku

Sawila

1EXCL 1INCL 2 3

nine(a)ku ine(a)ku janaku ganaku

Continuing such reconstruction of the TAP subgroups will clarify the grammatical structure of PTAP (cf. section 5 on the typological divides in the family particularly in the area of morphology) and allow for the higher affiliations of the TAP family, if any, to be established.

4.2 External relationships Since 1975, Timor-Alor-Pantar languages have been repeatedly connected with the Trans-New Guinea (TNG) family, a large and essentially putative language family spanning the full length of the island of New Guinea. Wurm, Voorhoeve

12

Antoinette Schapper

and McElhanon (1975) were the first to assert this affiliation, albeit with the caveat that the TAP languages contained strong substratum elements.6 The association between TAP and distant languages classified as TNG on the New Guinea mainland has persisted on two grounds: (i) apparent lexical similarities between TAP and various western TNG subgroups, and (ii) resemblances in the form and structure of pronominal paradigms.7 On the basis of lexicon, Stokhof (1975) suggested a relationship between Alor-Pantar languages and languages of the South Bird’s Head of New Guinea, which had just been classified as TNG by Voorhoeve (1975). Hull (2004) observed lexical similarities between TAP and the West Bomberai languages, also classified as TNG by Voorhoeve (1975). Recently, both Holton & Robinson (2014) and Cottet (2015) have echoed Hull’s observations, but without significant addition of data to the claim and with little or no attempt to establish the regularity of sound correspondences. Pawley (2001, 2012) and Kratochvíl (2007: 6–11) offer TAP look-alikes for around a dozen PTNG lexemes. These lexical similarities are not enough to establish the regular sound correspondences demanded by the comparative method, but they do seem to point to some connection between TAP and mainland TNG languages. The weakness of the TNG signal in TAP in all likelihood reflects the great time-depth that presumably separates TAP languages from PTNG, and means that the TNG hypothesis as it touches on TAP may ultimately be unfalsifiable. The pronominal evidence of a TAP-TNG link has shown more promise, though it cannot be considered conclusive in itself. Ross (2005: 35–36) used a comparison of independent pronoun forms to preliminarily classify TAP languages as belonging to TNG.8 A TNG classification for TAP was initially suggested by the presence 6 This is confirmed by the rigorous study using an admixture model by Reesink, Singer & Dunn (2009). In this, TAP languages do not cluster with Trans-New Guinea, but fall within a separate cluster, containing the Papuan languages of the Bird’s Head and the Bismarck archipelago as well as Austronesian languages of eastern Indonesia. This indicates that TAP languages are the result of millennia long interaction of non-TNG Papuan and Austronesian speakers. 7 A relationship between West Papuan and TAP has also been claimed in the literature (e.g., Anceaux 1973, Capell 1975), but this has been largely abandoned (see Holton & Robinson 2014 for the latest consideration). The absence of even tentative lexical links between TAP and these languages combined with improbable pronominal relationships (contra Donohue 2008) has proved the TAP-West Papuan affiliation to be of little substance. The typological features which are similar in TAP and West Papuan are frequently also found in Austronesian languages in the region, pointing to shared substrate(s) in the region and/or wide diffusion of features (Schapper 2015). 8 The validity of using pronoun resemblances to establish preliminary groupings of Papuan languages into genealogical units is shown by Hammarström (2012). Using pronominal data from over 3000 languages he demonstrates that there is a tendency for the pronouns of Papuan languages to use certain consonants more often than in the rest of the world.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

13

of the PTNG *a~*i ablaut for singular~non-singular pronouns in TAP languages, and the common appearance of /n/ for 1st persons in both TAP and TNG pronouns. Other person forms in TAP, however, show little correspondence with TNG and a flip between TNG 2nd and 3rd person pronouns was suggested to explain the velar of TAP 3rd person pronouns: PTNG *ka ‘2SG ’ > PTAP *ga ‘3SG ’. Ross (2005) includes TAP within his “West Trans New Guinea linkage” which takes in the Dani, Wissel lakes and West Bomberai languages on account of the Eastern Timor subgroup apparently reflecting the innovative pronoun *ani ‘1SG ’. TAP and West Bomberai languages are further grouped together because of the shared innovation of a 1st person plural inclusive pronoun *bi (Ross 2005: 36), reflected as PTAP *pi ‘1PL . INCL ’ (Schapper, Huber & van Engelenhoven 2014), and in West Bomberai languages as Iha mbi ‘1PL . INCL ’ (Cottet 2015) and Karas pi ‘1PL ’9 (Visser 2016). Advances in the reconstruction of PTNG and PTAP pronominal morphology add to the case for TAP being included within TNG. Most significantly, the PTNG dual suffix *-li attaching to pronominal bases is continued as PTAP *-l[e,i], with pronominal paradigm reflexes in two TAP primary subgroups, Alor-Pantar and Bunaq (dual morphemes bolded in Table 3). Table 3: Reflexes of dual PTAP *-l[e,i] in pronominal paradigms Blagar

1INCL 1EXCL 2 3

pole nole ole ʔole

Klon

1INCL 1EXCL 2 3

plε nglε ~ nlε eglε glε ~ εlε

Kafoa

1INCL 1EXCL 2 3

pel nijel jel del

Kui

1INCL 1EXCL 2 3

palε ~ pilε ɲilε jalε alε

Bunaq

1INCL 1EXCL 2 3

ili neli eli halali

Suter’s (2012) PTNG verbal P agreement prefixes offer a similar picture to that presented by independent pronouns when compared to the reconstructed Alor-Pantar verbal agreement prefixes for P and non-active S (Table 4). Here, the first persons look related, but other persons do not reconstruct straightforwardly at this stage. 9 Visser (2016) gives two first person plural pronouns, ni and pi, for Karas, but is uncertain about whether the distinction between them is one of clusivity.

14

Antoinette Schapper

Table 4: Reconstructed verbal agreement prefixes

1SG 2SG 3SG 1PL . INCL 1PL . EXCL 2PL 3PL

PTNG

PAP

*na*ga*wa-, Ø – – – *ya

*na*ha*ga*pi*ni*hi*gi-

Comparison of verbal prefixes in the West Bomberai language Iha and the TAP language Bunaq is, at least superficially, more promising (Table 5), with a regular relationship between forms in the respective paradigms: Iha n regularly corresponds to Bunaq n, while Iha initial k regularly corresponds to zero in Bunaq (Usher nd.). It remains to be seen whether the PAP paradigm is conservative and reflects a continuation of the PTAP agreement prefixes. This would mean that the similarities in the Bunaq and Iha paradigms are merely the result of parallel, but independent reductions in an earlier more elaborate system. Table 5: Verbal agreement prefixes in Iha and Bunaq

1EXCL 1INCL /2 3

Iha (West Bomberai)

Bunaq (TAP)

nikiØ

nig-, Ø

The wider affiliation of the TAP languages remains an open question for now. However, a recent surge in work on West Bomberai languages means that proper consideration of the most credible claims of TAP affiliations with mainland New Guinea languages will become possible. Only with careful reconstruction of the proto-languages of the respective TAP and West Bomberai groups will we bring some serious historical perspective – most importantly the application of the comparative method – to the validity of TNG at its claimed fringes.

4.3 Interactions with Austronesian languages The outlying position of the TAP languages means that they have had a different history from that of many mainland New Guinea Papuan languages. In particular, they have been in close contact with Austronesian languages for at least 2,000 years and this has wrought many mutual changes on the languages that have only begun to be explored.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

15

A major question in the prehistory of TAP-Austronesian interaction is how the speakers of TAP languages resisted the incoming tide of Austronesian speakers which obliterated almost every other pre-Austronesian language group in Island South East Asia. The until recently accepted understanding of Austronesian history involves the expansion across eastern Indonesia, overwhelming pre-existing sparse populations of Papuan hunter-gatherers and transforming them into speakers of Austronesian languages (Bellwood 2006). Yet the TAP languages survive today, and we may speculate that their speech communities possessed some characteristics that allowed them to maintain themselves in the face of the Austronesian arrival. Significant population numbers sustained by agriculture is immediately suggested by the fact that New Guinea, where the TAP languages are thought to have originated according to the TNG hypothesis, is known as a major plant domestication center in which nobilization of taro (Colocasia esculenta), sugarcane (Saccharum officinarum) and at least one species of banana (Eumusa), among others, took place (Denham 2011). Three preliminary lexical reconstructions relating to crops and agricultural practice are suggestive of the speakers of PTAP having some kind of agriculture:

(3)

Agricultural vocabulary in PTAP a. PTAP *huˈbua ‘sugarcane’ Bunaq up; Alor-Pantar: Nedebang ɦuɸa, West Pantar habua, Kaera ub, Klon əba, Abui fa, Sawila ipua, Wersing upa; Eastern Timor: Makasae, Makalero ufa, Fataluku upa, Oirata uha. b.

PTAP *maˈgu[l] ‘banana’ Bunaq mok; Alor-Pantar: Nedebang mai, West Pantar maggi, Teiwa muħui, Kaera mogoi, Klon mgol, Kamang moːi, Wersing mlul; Eastern Timor: Makalero, Makasae, Fataluku muʔu, Oirata muː.

c.

PTAP *maˈħad[a] ‘garden’ Bunaq mar; Alor-Pantar: Nedebang maxara, West Pantar maggar, Teiwa maħar, Klon mgad; Eastern Timor: Makalero, Makasae ama.

Schapper & Huber (2012) suggest that a highly organized maritime culture amongst the speakers of PTAP would perhaps have facilitated resistance against the incoming Austronesians. This is, however, not readily apparent from the vocabulary reconstructed to PTAP. There are no reconstructions for items of maritime technology or practice, and only two items relating to marine life, but notably these can refer to fresh or salt water creatures:

16

(4)

Antoinette Schapper

Maritime vocabulary in PTAP a. PTAP *ˈħapi ‘fish’ Alor-Pantar: Nedebang ɦaːɸi, Teiwa ħaɸ, West Pantar hap, Blagar ab, Klon əbi, Abui afui, Kamang, Sawila, Wersing api; Eastern Timor: Makalero, Makasae afi, Fataluku api, Oirata ahi. b.

PTAP *ˈasi ‘prawn, shrimp’ Bunaq o; Alor-Pantar: Blagar asi(ŋkol), Klon ah(kol), Abui eti, Kamang oh; Eastern Timor: Makalero, Oirata asi.

More maritime vocabulary is reconstructable to PAP, indicating that the speakers of PAP perhaps had greater engagement with the sea. Examples include:

(5)

Maritime vocabulary in PAP a. PAP *ħei ‘canoe’ Nedebang ɦeː, Teiwa ħai, West Pantar hai, Blagar ʤe, Kui, Klon, Kafoa ei. b.

PAP *taˈma ‘sea’ Nedebang taː, Teiwa taʔ, Blagar taŋ, Kui, Klon tan, Kafoa tama, Abui tama, Kamang tama, Sawila tama, Wersing tma.

c.

PAP *ineˈmaŋ ‘shellfish gathered in the intertidal zone’ Nedebang nammaŋ, Teiwa namaŋ, Kui naman, Adang nemaŋ, Abui nemaŋ, Kamang inamaŋ, Wersing inem.

For the speakers of PTAP to journey from New Guinea to the Timor region they would have needed maritime expertise. The absence of reconstructable maritime vocabulary to PTAP most likely reflects a loss of maritime technology and practice in the Timor subgroups, rather than its innovation in PAP. Timorese culture is heavily land-oriented, with the sea regarded as dangerous and subject to many taboos. Timorese TAP groups for the most part live high in the mountains, and even when found in lowland areas, they tend to settle inland away from the sea. By contrast, the same taboos around the sea do not exist in Alor and Pantar, and whilst not everyone will sail, knowledge of the sea is in general greater and the diets of most people include a significant amount of food foraged in the intertidal zone. The restriction of TAP groups in Timor to inland lifestyles may itself have been caused by incoming Austronesians taking over lowland areas. The loss of maritime vocabulary in TAP languages of Timor also means that the role of any maritime culture in originally maintaining TAP languages against the Austronesian influx cannot be ascertained.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

17

The relative timing of the TAP and Austronesian arrival in the Timor region also remains an open question. Recent archaeological work suggests that the arrival of the domesticated dog and the cuscus species Phalanger orientalis in the Timor region was almost contemporaneous (O’Connor 2015). Across ISEA dogs are associated with the Austronesian expansion (Savolainen et al. 2004, Brown et al. 2011, Oskarsson et al. 2011), while Phalanger orientalis is known to have been transported by humans from the New Guinea area (Heinsohn 2010). This indicates that populations moving in from the west (Austronesian speakers) and the east (PTAP speakers) arrived in Timor at around the same time. The presence of reconstructable terms that have ultimately Austronesian etymologies in PTAP would show that the speakers of PTAP were in contact with Austronesians at the time of their arrival in the Timor region. Yet, beyond PTAP *[j]asir ‘salt’ < Proto-Malayo-Polynesian *qasiRa whose appearance in TAP languages may be the result of separate borrowing events into the three primary branches of the family, there is as yet no Austronesian lexical influence detectable in PTAP. Thus, it seems likely at this stage that TAP languages arrived in the Timor area before speakers of Austronesian languages, though perhaps not by much. The nature and extent of influence of Austronesian languages on TAP languages varies depending on the local contact situations between individual clusters of TAP languages and the proximal Austronesian languages. As is to be expected from geography, the TAP languages of Timor show the most influence from Austronesian languages. Bunaq, surrounded on all sides by Austronesian languages, shows the most extensive changes. Though it has retained the headfinal, head-marking syntax typical of its family, Bunaq has developed in isolation from the other TAP languages and has many unique characteristics that can only be explained by historical processes of contact and adaptation with its Austronesian neighbors (Schapper 2011a, 2011b). Bunaq has been a recipient of a very large number of Austronesian loans, most obviously from Tetun, but also Kemak and Mambae and other, as yet unidentified, Austronesian sources. Among the Eastern Timor languages, Austronesian lexical influence is rife. Makasae and Makalero have a sizeable number of borrowings from the Austronesian languages of the Waima’a-Naueti linkage. Fataluku has similar numbers of Austronesian borrowings to Makasae-Makalero, but the source of many appears to be in the Austronesian languages in the islands to the east of Timor. For example, both Makasae-Makalero and Fataluku have borrowings of reflexes of the Proto-Malayo-Polynesian term *qaRta, glossed variously as ‘slave’, ‘outsider’, or ‘alien’. Fataluku akan ‘slave’ is a borrowing of akan ‘slave’ from Makuva, the Austronesian language of Tutuala that has the sound change of PMP medial

18

Antoinette Schapper

*t > k, while Makasae-Makalero ata ‘slave’ is from the neighboring Waima’aNaueti languages ata ‘slave’, which preserve PMP *t as t. In the AP languages, Austronesian influence is overall more limited (see word list calculations of Robinson 2015). In the central and eastern areas of Alor, contact has historically been with the Austronesian-language speaking groups of northern Timor, in particular, the Tokodede people of north-central Timor. Tokodede songs are sung throughout central Alor (Wellfelt & Schapper 2013) and rulers are said to have been brought from Timor to the south coast of Alor (Wellfelt 2016). At this stage, around two dozen words in Central-East Alor languages can be sourced from Tokodede or other Timor languages. On Pantar, the Nedebang have a ritualized, so-called bela relationship with the Kedang, the Austronesian-language speaking group in the north of Lembata, the island immediately west of Pantar. A bela relationship entails that Kedang and Nedebang people have reciprocal rights to go to each other’s territories and take whatever they like for themselves, without risking raising the ire of the other group. Traditionally, such arrangements were only exploited in times of extreme hunger and need on the part of the one group, but would entail regular visits to each other’s villages in order to reaffirm the ties of the friendship. In the past when bela was still an important part of the inter-island networks around Pantar, Nedebang people knew how to speak Kedang. Today, however, there are few linguistic relics of this past bilingualism, with only a few loan words identified. Whether TAP languages have had an influence, either through contact or substrate, on Austronesian languages is yet to be systematically studied. Donohue and Schapper (2008) argue that the indirect-direct possessive contrast in Austronesian languages possibly originated through contact with PTAP (or antecedents thereof) which had the same contrast in possession. Schapper (2009) argues that the isolating structure of the Austronesian languages of Timor-Leste is the result of restructuring on a TAP substrate. Klamer (2012) makes a similar suggestion, positing that the morphologically reduced nature of the Austronesian language Alorese is the result of TAP language speakers on Pantar switching to speaking the Austronesian language. Edwards (2016) shows that a bottom-up reconstruction of the Austronesian language Uab Meto in western Timor reveals that a large amount of the Uab Meto lexicon is from a non-Austronesian source, possibly TAP. While these studies all present a picture of Austronesian languages replacing TAP languages, Austronesian languages can in some places also be shown to have lost ground to the TAP languages. The study of placename etymologies in Timor reveals that at least two members of the family, Bunaq and Makasae, have significantly expanded in the early modern or immediately premodern

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

19

period to take over areas previously occupied by speakers of neighboring Austronesian languages (Schapper 2011a, Huber, Schapper & van Engelenhoven 2015). The mechanisms that have allowed the Papuan groups in Timor subsequently to turn the tables and successfully expand and assimilate established Austronesian peoples remain to be investigated.

5 Typological overview of the TAP family In the introduction to Volume 1, I provided a typological sketch of the TAP family in general. In this volume, which includes the first comparative sketches of TAP languages in Timor, the treatment I present aims to bring out some salient typological divides within the family. TAP languages tend to show differences in typological patterns depending on the geographical area in which they are spoken. The Alor-Pantar branch shows a high degree of phonological and morphological diversity, while the TAP languages of Timor are typically more syntactically varied.

5.1 Phonology The most common vowel system in the TAP family is a simple inventory of five cardinal vowels, with the appearance of a length distinction varying throughout the family. Five cardinal vowel systems with no length distinction are found in disparate TAP languages including West Pantar, Wersing and Bunaq. Five cardinal vowel systems with a robust length distinction for all qualities of vowel are common across the AP languages, including Nedebang, Teiwa, Kaera, Abui and Kamang. Five cardinal vowel systems in which long vowels are only marginal are found in Blagar, Makalero and Fataluku. The greatest diversity in vowel systems is found in West Alor-Straits languages where (often marginally) contrastive heights of mid-vowels, phonemic central vowels, and length distinctions for only some vowel phonemes sporadically appear. Table 6 sets out the variety of these systems. The other region with unusual features in their vowel systems is East Alor Montane (Table 7). Sawila notably has the front-rounded vowel /y, yː/ contrasting with five short and long cardinal vowels, while Kula has two short central vowel phonemes contrasting with five cardinal vowels of regular length.

20

Antoinette Schapper

Table 6: West Alor-Straits vowel systems Adang

Klon

i

u e

o ɛ

ɔ

i iː e ɛ ɛː

u uː o oː ɔ

a

ə a aː

Kui

Reta

i iː e ɛ ɛː

u uː o oː

Kafoa i iː

u uː o

ʊ ʊː ɛ ɛː

ɔː a aː

i iː eː

u uː oː ɛ

ɔ

a aː

a aː

Table 7: East Alor Montane vowel systems Sawila i iː y yː

Kula u uː

ɛ ɛː

i ɛ

o oː a aː

u

ɪ̆ ɐ̆ a

o

In TAP languages, surface diphthongs are underlyingly sequences of vowels. The exception is Bunaq which has three phonemic diphthongs that contrast with sequences of the same vowels, as illustrated with the following minimal pairs: (6)

Bunaq minimal pairs for diphthongs versus vowel sequences a. /sai͡ / [saj] ‘exit’ ≠ /sai/ [saʲi] ‘be amused’ b.

/tei͡ / [tej] ‘dance’

≠

/tei/ [teʲi] ‘stare at’

c.

/poi͡ / [poj] ‘choice’

≠

/loi/ [loʷi] ‘good’

Inventories of consonant phonemes in TAP languages vary considerably, but are typically moderately small with an average of 13–16 consonant phonemes. Makalero has the smallest consonant inventory with only 11 phonemes (Table 8). Together with Fataluku (Table 9), it stands out in the family in lacking straightforward voicing distinctions in the plosives (of native words). The discrepancy in place of articulation of /t/ and /d/ in Makalero is also found in Blagar, which has /t ̪/ and /d/.

21

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

Table 8: Makalero consonant phonemes Bilabial Plosive Fricative Nasal Trill Lateral

Labio-Dental

Alveolar

f

t s

p

Post-Alveolar

Velar

d

m

Glottal ʔ

k

n r l

Table 9: Fataluku consonant phonemes Bilabial Plosive Affricate Fricative Nasal Approximant Trill Lateral

Labio-Dental

Alveolar

f

t ʦ s

p v

Palatal

Velar

Glottal ʔ

k z n

m

h j

r l

By far the largest consonant inventories in the family are found in Pantar. They are characterized by more elaborate fricative inventories and frequent phonemic gemination. The most elaborate is that of Nedebang, which has 21 singleton consonant phonemes over seven different places of articulation (Table 10) and 14 geminate consonant phonemes (Table 11). Nedebang is followed by West Pantar with 16 singleton consonant phonemes plus 10 geminate consonant phonemes, and Teiwa with 20 (only non-geminate) consonant phonemes. Table 10: Nedebang singleton consonant phonemes

Plosive Fricative Nasal Approximant Trill Lateral

Bilabial

Dental

Alveolar

Palatal

Velar

p ɸ

t s

c

θ

k x

b

d

m w

Uvular g

ʔ

q ɦ

ŋ

n

Glottal

j r l

Table 11: Nedebang geminate consonant phonemes Bilabial Plosive Fricative Nasal Trill Lateral

pp

bb mm

Alveolar tt ss

dd nn rr ll

Palatal cc

Velar kk ŋŋ

Uvular gg

qq

22

Antoinette Schapper

While there is little variation in the family for phonemes such as /r/ and /l/ or /m/ and /n/, the number of fricatives, affricates and places of plosive and nasal articulations varies widely. All languages have /s/, including those with just one fricative phoneme such as Sawila, Wersing, Kula and Kui. The second most common fricative is /h/, but in languages with just two fricative phonemes its appearance is intermittent: /s/ and /h/ are the only fricatives in West Pantar and Klon, but Makalero has /s/ and /f/, Kamang /s/ and /ɸ/, Kaera /s/ and /x/ (in each, the glottal fricative features as a marginal phoneme or an allophone of another phoneme). In languages with three or more fricatives, /s/ and /h/ are always present, while the identity of the other fricative(s) varies from language to language: for instance, Bunaq /s/, /h/, /z/, Makasae, Kafoa, Adang, Abui /s/, /h/, /f/, and Blagar /s/, /h/, /v/, /z/. Affricate phonemes occur sporadically throughout the family, but frequently have only marginal phonemic status: one affricative /ʦ/ in Fataluku and Bunaq (both also have [ʤ] as an allophone of /z/), marginal /ʤ/ in Reta, Kula and Kui, and /ʦ/ and /ʤ/, both marginal, in Adang. Similarly sporadic is the appearance of /q/ in Abui, Teiwa and Nedebang, /v/ in Teiwa, Blagar, Kroku and Fataluku, and weakly phonemic /ɲ/ in Kui, Adang and Wersing. Some consonant segments show regional patterning. Glottal stop is universally present in the phoneme inventories of Timor languages (although lost in some individual dialects), but is only sporadically found in Alor-Pantar. Velar nasals are absent in Timor languages, as is consistent with the surrounding Austronesian languages, but are frequently present in AP, though they often have dubious phonemic status (e.g., West Pantar, Kaera) or are clear allophones of /n/ (e.g., Wersing). The appearance of /ʈ/ in Oirata is under the influence of the neighboring Austronesian language which also has the retroflex. Implosive /ɓ/ is found in the Straits languages Blagar and Reta and may be the result of influence of Austronesian languages further to the east where implosives are common. The velar fricative /x/ is limited to three languages of north-eastern Pantar, Kroku, Kaera and Nedebang, while /ћ/ is only found in Teiwa. Consonant phonemes limited to just one member of the family are the marginal phonemes /β/, /kʷ/ and /gʷ/ in Kula, /θ/ and /c/ in Nedebang, and /ʎ/ in Kui. In TAP languages monosyllabic and disyllabic roots are most common, with polysyllabic words typically being limited to borrowings, historical compounds or (reduplicative) onomatopoeias. Across the family the most common syllable structures are (C)V and (C)VC, frequently with restrictions on the consonants permitted in codas. For example, in Bunaq, the only consonants able to appear in codas are voiceless obstruents, /n/, the two liquids and the glottal stop phoneme. Similarly, Fataluku does not allow /p/, /m/, or /f/ in word final position. In most TAP languages, consonant clusters are limited, even across syllable

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

23

boundaries. Whilst almost unknown in Timor languages, many AP languages (including Teiwa, West Pantar, Kula) allow clusters of an obstruent followed by a liquid in word initial onsets. Languages such as Klon, Kui and Wersing allow more types of underlying consonant clusters, but have surface processes of vowel epenthesis to break them up. Unusual features are the extensive appearance of echo vowels in Wersing to avoid final codas and in Makasae-Makalero on phrase-final words, and the productive morphophonemic process of high vowel metathesis found Bunaq and Wersing to maintain a CV structure in certain affixation environments. There are also noticeable differences in preferred word shapes across TAP languages. Words in Fataluku and Oirata overwhelmingly end in an open syllable. By contrast, in Klon, a language of Alor, there are few words that end in a vowel. Historical reconstruction shows that dropping historical final vowels, consonantvowel metathesis and glottal stop prosthesis have variously been applied to create consonant-final roots in Klon (7). Similar changes to consonant final word shapes can be shown to have occurred independently in Teiwa. (7)

Consonant final root creation in Klon a. Final vowel dropping ap ‘walk’ < PAP *ˈapa ‘walk’ luk ‘bow’ < PAP *ˈluku ‘bend, bow’ mot ‘back’ < PAP *ˈmota ‘back’ naŋ ‘not’ < PAP *ˈnaŋa ‘not’ b.

Consonant-vowel metathesis puin ‘hold’ < PAP *puˈni ‘hold’ muin ‘nose’ < PAP *muˈni ‘smell, nose’ moːn ‘snake’ < pre-Klon *muan < PAP *muˈna ‘snake’ daːr ‘verse’ < pre-Klon *daːr < PAP *daˈra ‘song’

c.

Final glottal stop prosthesis adaʔ ‘fire’ < PAP *ˈħade ‘fire, firewood’ ɛtɛʔ ‘tree’ < PAP *aˈte ‘tree’ nɛʔ ‘name’ < PAP *ne ‘name’ weʔ ‘blood’ < PAP *we ‘blood’

A range of stress systems are found in TAP languages, but most descriptions only treat suprasegmentals perfunctorily (exceptions are Heston 2015 and Delpada 2016). Timor languages have non-phonemic stress systems: Bunaq has regular penultimate stress, while Eastern Timor languages such as Fataluku and Makalero have stress systems based on right-aligned moraic trochees. AP languages have both non-phonemic and weakly phonemic stress systems. Blagar and Wersing

24

Antoinette Schapper

have stress systems with regular penultimate stress, while Kaera has regular final stress. Phonemic stress systems bearing low functional loads are found widely in AP languages, including Teiwa, Kamang, and Sawila. The most common pattern among these is for stress to regularly fall on heavy syllables, but where a disyllabic word consists only of light syllables stress is lexicalized and may fall on the first or second syllable of the root. Schapper (2017) argues that such a system of stress was ancestral in the AP languages and its regularization into a penultimate stress system led to the emergence of phonemic geminates in West Pantar and Nedebang. Tone is found in only two TAP languages. A pitchaccent system is reported for one dialect of Fataluku (Stoel 2007), but is not known in other Fataluku dialects. Abui has been described as a low-density lexical tone language, in which tone is associated only with certain syllables and tone minimal pairs are relatively sparse (Delpada 2016).

5.2 Morphology Morphological profiles show significant variation across the family in the extent, locus and type of morphology. Much TAP morphology is unproductive, in many cases having fossilized and fused with original roots. In order to make the connections between morphological patterns in TAP languages clearer, this discussion will have frequent recourse to historical arguments. Verbal agreement morphology in the AP languages and in Bunaq is entirely prefixing, whereas Eastern Timor languages have traces of both prefixing and suffixing agreement morphology. Within the AP subgroup, there is significant variation in the number of agreement paradigms: Pantar and the Pantar Straits languages basically have a single paradigm of agreement prefixes (Table 12); West Alor languages typically have three paradigms (e.g., Adang, Table 13); Central Alor languages have the most elaborate systems, with five paradigms in Abui (Table 14) and seven in Kamang (Table 15); East Alor languages have one main paradigm and one (e.g., Sawila, Table 16) or two (e.g., Wersing, Table 17) marginal paradigms that are found with a few verbs.10 Bunaq has a single paradigm of prefixes, but without the number distinction typical of AP languages (see Table 5 above). Makalero has a cognate third person agreement prefix k- that appears on a small number of verbs, and that is clearly fossilized on several Makasae roots (Schapper, Huber & van Engelenhoven 2014). A remnant of prefixal agreement is also present in Fataluku, which prefixes a 10 Some languages also have reflexive and reciprocal prefixes that belong to these paradigms. See section 4.1 for a discussion of the reflexives.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

25

vowel V- to verbs where a 3rd person referent is the anaphorically retrievable object of a restricted set of verbs. The vowel is not entirely predictable: if the first syllable of the root contains a mid vowel, the prefixed vowel is typically a copy; in other cases, the prefixed vowel is unpredictable, see (8).

(8)

Fataluku pah > nam > lur > me > nof > har > ruka >

verbs plus anaphoric 3rd person vowel i-pah ‘disseminate’ u-nam ‘catch in the hands’ u-lur ‘sweep’ e-me ‘take’ o-nof ‘die’ a-har ‘send’ a-ruka ‘multiply’

Table 12: Pantar and Straits agreement prefixes

1SG 2SG 3SG 1PL . EXCL 1PL . INCL 2PL 3PL

West Pantar

Nedebang

Blagar

nahaganipihigi-

nahaganipihigi-

nØʔnipiigi-

Table 13: Adang agreement prefixes

1SG 2SG 3 1PL . EXCL 1PL . INCL 2PL

I

II

III

naaʔanipii-

nεεʔεniεpiεiε-

nɔɔʔɔniɔpiɔiɔ-

Table 14: Abui agreement prefixes

1SG 2SG 3 1PL . EXCL 1PL . INCL 2PL

I

II

III

IV

V

naahanipiri-

neehenipiri-

noohonupuru-

neːeːheːniːpiːriː-

noːoːhoːnuːpuːruː-

26

Antoinette Schapper

Table 15: Kamang agreement prefixes

1SG 2SG 3 1PL . EXCL 1PL . INCL 2PL

I

II

III

IV

V

VI

VII

naaganisii-

neegenisii-

noowoniosioio-

neːeːgeːniːsiːiː-

noːoːwoːnioːsioːioː-

naoaogaonioːsioːioː-

neʔeʔgeʔniʔsiʔiʔ-

Table 16: Sawila agreement prefixes

1SG 2SG 3 1PL . EXCL 1PL . INCL 2PL

I

II

naagani(j)iji-

niegini(j)iji-

Table 17: Wersing agreement prefixes

1SG 2SG 3 1PL . EXCL 1PL . INCL 2PL

I

II

III

nagɲtj-

naaganitaji-

neegeniteji-

Eastern Timor languages also have a set of verbs which display initial consonant mutations that go back to an earlier prefix in the agreement paradigm. Examples of these mutations are provided for Fataluku and Makalero in Table 18. The consonant mutation is triggered when a verb occurs with an incorporated locational (Huber 2011, van Engelenhoven forthcoming) and goes back to the assimilation of the earlier locative *n- prefix (Schapper 2009), that is still preserved on a few verbs and nouns in Bunaq (for example: il ‘water’, n-il ‘LOC -water’ referring to the internal juices or sap of a fruit or tree; h-ilin, g-ilin ‘half way mark (e.g., in a journey)’, n-ilin ‘LOC -part’ referring to the center of location (e.g., of a garden).11 11 This prefix is ultimately derived from the PTAP locative postposition *na (reflexes include Bunaq no ‘LOC ’, Abui =ŋ ‘LOC ’, and Fataluku na-e ‘LOC -VBLZ ’), which became an applicative through incorporation into the verb phrase, eventually entering into a paradigmatic relationship with the other agreement prefixes.

27

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

Table 18: Examples of verbs with initial consonant mutations in Eastern Timor Fataluku

Makalero

h → c mutation her -cer hina -cina hura -cura

‘dry’ ‘plait’ ‘take from fire’

h → s mutation haka -saka heil -seil huri -suri

‘search’ ‘pull’ ‘shoot’

t → c mutation teku -ceku taia -caia tomok -comok

‘stir’ ‘sleep’ ‘soft’

t → d mutation teri -deri tia -dia tina -dina

‘cut’ ‘sleep’ ‘cook’

f → p mutation fetil -petil fulutete -pulutete faʔi -paʔi

‘stumble’ ‘spit’ ‘make’

f → p mutation† faniʔ -paniʔ

‘be like’

† This is the only known item displaying this mutation in Makalero

Suffixal agreement is limited to the Eastern Timor languages, where it is found in fossilized and often highly suppletive form on intransitive verbs. Agreement is with the number (singular or plural) of the subject. In Makalero and Makasae only a handful of these verbs are found and most are suppletive, though the plural forms are consistently associated with an ending in -Vr (Table 19). Number agreement in Fataluku on verbs is much more robust and involves less suppletion (Table 20). Whilst these agreement forms show irregularity also in Fataluku, it is clear that an historical suffix -re marked plural agreement. This is only preserved in all Eastern Timor languages on the verb ‘stand’.12 Table 19: Eastern Timor singular and plural forms of intransitive verbs Makalero

‘sit’ ‘stand’ ‘run’ ‘sleep’ ‘big’

Makasae

Fataluku

SG

PL

SG

PL

SG

PL

mit nat riaʔ tia pere

diar naser titar rou helar

mi na riaʔ – –

diar nahar ditar – –

mire nate tifare taia lafai

cuare nehere helere taiatere laficare

12 Cognates of both the singular and the plural forms of this ‘stand’ verb are found in the other subgroups (e.g., Bunaq net > ‘stand.SG ’; West Pantar natar ‘stand.PL ’), indicating that suffixal number agreement is not an innovation in Eastern Timor, but goes back to PTAP.

28

Antoinette Schapper

Table 20: Examples of Fataluku number agreement verbs (van Engelenhoven forthcoming)

‘dead’ ‘alive’ ‘laugh’ ‘eat’ (intr.) ‘fly’ ‘disappear’

SG

PL

umu lauhe kele mace ipile molu

umunoro lauhoro kelere macere ipilere molure

‘come’ ‘slip’ ‘hide’ ‘be.at’ ‘be.inside’ ‘be.on’

SG

PL

mau suke palake nae mucune micane

mauere sukoro pelere naere mucuere micaere

Verbal inflections other than agreement are unknown in Bunaq and Eastern Timor, but appear sporadically in AP languages. A verbal realis suffix is found in both Teiwa and Wersing. Aspectual morphology is the most common form of non-agreement verbal inflection. Known in seven AP languages, unlike agreement morphology, the use of aspectual affixes is never obligatory. Aspectual prefixes a- ‘INCP ’ and i- ‘PROG ’ are found in West Pantar, while the other AP languages with aspectual inflection have suffixes. Blagar (-t) and Kaera (-i ‘PFV ’, -(i)t ‘IPFV ’) have aspectual suffixes which only occur on verbs in a non-final clausal position; Kaera also has -aŋ ‘CONT ’ which can occur on verbs irrespective of clausal position.13 In Alor languages aspectual suffixes are optional and are hosted on the final verb of a clause. Kamang has three aspectual suffixes -si ‘IPFV ’, -ma ‘PFV ’ and -ta ‘STAT ’, the last of which is of only limited productivity; Kui and Kiraman, two closely related languages, have aspectual suffixes -i ‘PFV ’ and -a ‘IPFV ’. Cognates of these are found in Abui, but they are not fully productive, patterning differently depending on the lexical class of verb. The identity of the suffixes has also been partially obscured by a range of sound changes on the one hand, and fusion with other morphemes on the other. For example, the Abui verb ‘carry’ occurs without any suffixation as wik (9a), but when suffixed with -i ‘PFV ’ the root mutates to become wit- (9b). This alternation led Kratochvíl (2007) to speculate that the root was simply wi- and that the /t/ was suffixal. In fact, the root alternation is the result of PAP *t becoming /k/ in Abui non-initially before vowels other than *i (cf. PTAP *wita ‘carry’: Bunaq wit ‘take, fetch’; Kamang wita ‘carry hanging from forehead’; Wersing wis ‘carry hanging from forehead’, Sawila wise ‘carry on one’s back, carry on head’, etc.). Historical changes such as this may account for the aspectual alternations observed in some Kafoa verbs (Baird this volume).

13 Note that according to Steinhauer (2014: 205–209) aspectual marking is only one of a range of functions of Blagar -t.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

(9)

29

Abui aspectual root mutation with perfective suffix a. Na ara wik. 1SG . AGT fire(wood) carry ‘I carry firewood’ b.

Na ara wit-i. 1SG . AGT fire(wood) carry:PFV- PFV ‘I carried firewood’ (Schapper fieldnotes)

Abui verbs of the stative class have the suffix -a ‘STAT ’ when they have a stative reading (10a), but either -da ‘IPFV ’ (10b) or -di ‘PFV ’ (10c) when they have dynamic readings. While the stative suffix is a straightforward cognate of the Kui/Kiraman imperfective, the dynamic aspectual suffixes appear to be fusions of the -a and -i with another suffix, possibly related to Kamang -da which is a verbalizer giving dynamic meaning to non-verbal roots: ap-da fish-VBLZ ‘become a fish’. (10)

Abui stative verb aspectual suffixation a. Ja lil-a. water hot-STAT ‘The water is hot.’ b.

Ja lil-da. water hot-IPFV ‘The water is becoming hot.’

c.

Ja lil-di. water hot-PFV ‘The water has become hot.’ (Schapper fieldnotes)

Inflections for phrasal position are also found in several languages. Kaera inflections for phrasal position are found on certain verbs and numerals: on the clause level, -o ‘FIN ’ occurs on a verb/numeral in clause-final position; within the NP, -o ‘FIN ’ occurs on a verb/numeral in NP-final position, while -i ‘NFIN ’ marks a verb/numeral in a non-final position in the NP. In the East Alor Montane languages Sawila and Kula, alternations in word shape depending on phrasal position are found not only on verbs, but also on nouns and to a lesser extent on post-verbal grammatical morphemes such as negators. Phrase-finally, words which show the alternation end in a vowel; non-finally, they end in a consonant. The Sawila and Kula word alternations for final phrasal position can historically be attributed to an inflectional suffix only in part. Consider the forms in Table 21. For words with an historical open syllable finally, non-final forms show loss of the final vowel, while phrase-final forms retain the vowel. For words with an

30

Antoinette Schapper

Table 21: Sources of the final-nonfinal word alternations in Kula and Sawila Sawila

Kula

Final

NonFinal

Final

NonFinal

jaːti

jaːt

jati

jat

PAP *ˈjaci (cf. Nedebang jeːci, Teiwa jaːs, West Pantar jasa, Blagar jasi)

taːni

taːn

tani

tan

PCEA *tani (cf. Wersing -taiŋ, Kamang -tan)

‘hand’

-tana

-taŋ

-tɐna

-tɐn

PTAP *ˈ-tana (cf. PET *tana; PAP *tana: Nedebang -θaːŋ, Teiwa -tan, Kamang -taŋ)

‘NEG’

naːnu

naːn

nanu

nan

PEA *nanu (cf. Wersing nanu ~ nauŋ)

Final ‘bad’ vowel deletion ‘fall’

Reconstruction

‘senior’

saːku†

saku

sak

PCEA *ˈsaku (cf. Wersing saku, Kamang saːk)

‘soil’

ataku†

atɐku

atɐk

PEA *aˈtoku (cf. Wersing atoku)

mɐdɪŋ

PAP *muˈdiŋ (cf. Nedebang muddi, Teiwa midan, Blagar mudiŋ, Kamang mit).

masɪŋ

PTAP *mata[n]in (cf. Bunaq maʔas, Kamang -maitaŋ Wersing -masniŋ)

Addition ‘plant (v.)’ madine madiŋ mɐdɪna of *-a ‘hungry’

masine masiŋ masɪna

‘five’

joːtine

joːtiŋ

jawatɪna jawatɪn PAP *jiwasiŋ (cf. Nedebang jeːsiŋ, West Pantar jasiŋ, Teiwa jusan, Adang ifihiŋ, Abui jetiŋ, Kamang iwesiŋ)

‘fly (v.)’

irine

iriŋ

irɪna

irɪŋ

PAP *jiraŋ (cf. Nedebang jilla, West Pantar hilaŋ, Teiwa jiran, Kaera jir-)

‘person’

anine ~ aniŋ anina‡

anina

aniŋ

PAP *aniŋ (cf. Kaera -nam, Adang -naŋ, Kamang -aniŋ)

† Sawila has not extended the pattern of final vowel deletion to create final-nonfinal forms with these lexemes ending in /u/, while Kula has. ‡ Final forms with both -a and -e are found for Sawila ‘person’. The -e final forms in Sawila appear to be part of a sound change in progress in which the original /a/ of the suffix is raised where the preceding vowel is high as in these examples. It appears to have applied to verbs in the first instance.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

31

historical closed syllable finally, final forms have a vowel -a added, while nonfinal forms retain the historical closed syllable. Nominal inflection is limited in TAP languages. Prefixes for possessor are the only nominal inflections found across the whole family. In TAP languages of Timor, Pantar and the Straits, there is typically only one paradigm of possessive prefixes. In Bunaq and many languages of Pantar and the Straits, these prefixes are identical to the agreement prefixes on verbs and occur on the so-called inalienable class of nouns, a lexically determined subset of nouns, typically made up of kinship and body part nouns; the possessor of alienable nouns is marked by a free possessive marker (e.g., different 3rd person possessor encoding in Nedebang in (11). Both Teiwa (12) and Makalero (13) use one paradigm of prefixes to mark all possessors. Nouns differ in that, while alienable nouns are only optionally marked with the possessor prefix, inalienable ones are obligatorily marked by it.

(11)

Nedebang a. Inalienable ga-leːɸu 3SG -tongue ‘his/her tongue’

b. Alienable gei seː 3SG . POSS house ‘his/her house’ (Schapper fieldnotes)

Teiwa (12) a. Obligatorily possessed b. Optionally possessed ga-liɸi ga-yaɸ 3SG - house 3SG -tongue ‘his/her tongue’ ‘his/her house’ (Schapper fieldnotes) Makalero (13) a. Obligatorily possessed b. Optionally possessed ki-ifil ki-lopu 3-tongue 3-house ‘his/her/their tongue’ ‘his/her/their house’ (Huber pers. comm.) Western and Central Alor languages have two paradigms of agreement prefixes for possessors, an inalienable one and an alienable one (e.g., different 3rd person possessor encoding in Abui in 14). East Alor Montane languages have multiple, lexically-specified prefixal paradigms for inalienable nouns, and a single paradigm for alienable nouns (e.g., different 3rd person possessive forms in Sawila in 15). Fataluku is the only Timor language to have contrastive possessive prefixes for alienable and inalienable on vowel-initial nouns (16). Unlike

32

Antoinette Schapper

other possessive prefixes in the family, however, the use of these prefixes in Fataluku is entirely optional and alternative possessive constructions not involving morphology can be used (see examples of these other Fataluku constructions in 30).

(14)

(15)

Abui a. Inalienable ha-lifi 3-tongue ‘his/her/their tongue’ Sawila a. Inalienable I ga-limpuru 3-tongue ‘his/her/their tongue’ c.

(16)

Inalienable III ge-pa 3-father ‘his/her/their father’

Fataluku a. Inalienable n-ina INAL -eye ‘eye of . . .’

b.

Alienable he-fala 3-house ‘his/her/their house’ (Schapper fieldnotes)

b.

Inalienable II gaŋ-kise 3-back ‘his/her/their back’

d.

Alienable gi-araːsiŋ 3-house ‘his/her/their house’ (Kratochvíl 2014)

b.

Alienable h-ipar ALIEN - dog ‘dog of . . .’ (Heston 2015)

The only other nominal inflections found in the TAP languages are plural suffixes, albeit with a very limited appearance. A range of plural suffixes, often only on a small number of human and kinship nouns, are found in the Eastern Timor languages, Makalero and Fataluku. The most restricted suffix is the plural suffix -r found on a single noun, ‘child’, in Makalero, and a handful of nouns in Fataluku (examples in Table 22; see also Heston 2015: 21). Less restricted are Makalero -raː and Fataluku -ara which mark plurality on predominantly kinship nouns, though there is some loss of this meaning in Makalero but not in Fataluku. Makalero also has two productive plural suffixes -laː for plurality on non-kin nouns and -ara for associative plurality. The only known plural affix outside of Eastern Timor is Kamang -leː marking associative plurality.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

33

Table 22: Examples of inflections on human/kinship nouns in Eastern Timor Makalero

‘child’ ‘ySb’ ‘man’ ‘woman’

Fataluku

Uninflected

Inflected

Uninflected

Inflected

mata noko nami tufur

mata-r noko-raː nami-raː† tufur-aː†

moko noko nami tupur

moko-r noko-r nami-r tupur-ara

† The plural meanings have been exceptionally lost from these forms. The plural marked forms have instead taken on generic reference: while tufur means ‘sister’, tufuraː refers to a woman generically; while nami means ‘male (animal)’ or ‘husband’, namiraː refers to a man generically.

Derivational morphology is most developed in Eastern Timor languages, but is found scattered in various forms with limited productivity throughout the family. Fataluku and Oirata are distinct for having a large number of precategorical roots that must be marked with one of two suffixes (-(n)u ‘NMLZ ’ and -(n)e ‘VBLZ ’) depending on whether their use is nominal or verbal. Suffixes deriving nouns from verbs are found in the Eastern Timor languages (Makalero -ini and -r, and Fataluku -(n)ana) and in the Pantar Straits language, Blagar -(na)ŋ. Multiple languages in the family also show traces of a final *-k deriving verbs from nouns or transitive verbs from intransitive verbs, but the number of tokens in each language is small. For example, in Kamang just two instances are known (pida ‘hole’ and pida-k ‘bury’, tapu ‘be intoxicating’ and tapu-k ‘intoxicate’), while Makalero has around a dozen (e.g., huri ‘brush (n)’ and huri-ʔ ‘brush (v)’, lafu ‘life’ and lafuʔ ‘(a)live’). A causative suffix -(na)ŋ can be added to many intransitive verbs in Blagar and Kaera. Prefixes marking cause and intentionality are found in Straits-Western Alor languages: Blagar V-, Adang a-, Kui a-. Whilst lacking in many languages (e.g., West Pantar, Kaera, Blagar, Abui and Bunaq), applicatives are perhaps the most widespread form of derivational morphology in the family: a general applicative prefix PAP *un- is reflected in Teiwa un- and in most West Alor languages (Adang, Klon, Kafoa, Kui u-). Grammaticalization of postpositions and argument adding serial verbs into applicative prefixes has taken place widely in Alor languages: applicative prefixes from PAP *mia ‘be in’ are Klon, Kafoa, Kui, Kamang, Wersing and Kula mi-, and Sawila ma-; applicative prefixes from PAP *le ‘be on’ are Kui la-, Sawila li-, and Kula and Wersing le-, and; applicative prefixes from PEA *we ‘use’ are Sawila wiː-, Kula we-, and Wersing wa-. Much more extensive applicativization processes can be found in Eastern Timor languages, with productive incorporation into the verb phrase used to create dozens of applicative prefixes. Because such

34

Antoinette Schapper

incorporation in Eastern Timor languages has largely replaced the typical TAP verb serializations in the encoding of complex events, this phenomenon is discussed in section 5.4.

5.3 Word order All TAP languages have SV/APV and postpositions, word order features that set them apart from the SV/AVP and prepositions of their Austronesian neighbors. There are, however, differences, often geographical, in the details of word order patterns in TAP languages that make them less uniform than they superficially appear. A few examples of the word order differences in TAP are discussed here. Whilst all TAP languages have P before the verb, the extent to which the verb must occur in the final clausal position differs considerably. Eastern Timor languages are strictly verb-final, with free clausal negators and free aspectual markers consistently occurring preverbally, even when multiple such elements occur. This is illustrated in (17) on the basis of Fataluku.

(17)

Fataluku verbal finalness a. Negator – Verb Kuca akam nawar-e. NEG know-VBLZ horse ‘The horse didn’t know.’ b.

Aspectual particle – Verb Ipar hai nemer-e. ASP first-VBLZ dog ‘The dog went first.’

c.

Negator – Aspectual particle – Verb Ana akam hai fuleh-e. ASP return-VBLZ 1SG . SBJ NEG ‘I won’t come back.’ (Heston 2015)

At the other extreme, Bunaq, in central Timor, permits numerous elements to follow the verb, with strings of up to eight post-verbal elements encoding aspect, temporal duration, manner and polarity being observed. The simultaneous postverbal appearance of the negator and aspectual particles in Bunaq is illustrated in (18).

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

(18)

35

Bunaq verbal non-finalness a. Verb – Negator – Aspectual particle Inel masak niʔ taʔ. NEG ASP rain big ‘It hasn’t rained a lot yet.’ b.

Verb – Adverb – Aspectual particle – Aspectual particle Inel teni gie oa. rain again ASP ASP ‘It’s about to rain again.’ (Schapper fieldnotes)

AP languages occupy the middle ground between Bunaq and the Eastern Timor languages. They typically only allow one free clausal element to occur after the verb and have strategies in place to maintain this. For instance, in West Pantar, all negators occur after the verb, including the negative imperative marker gajaŋ (19a) and the negative polarity incompletive aspect marker yadda ‘not yet’ (19b). However, when these co-occur in the same clause yadda gets bumped to a preverbal position (19c).

(19)

West Pantar post-verbal position a. Verb – Negator Horaŋ gajaŋ. make.noise NEG . IMP ‘Don’t make noise.’ b.

Verb – Aspectual particle Naŋ hori yadda. 1SG . ACT eat NEG . ASP ‘I haven’t eaten yet.’

c.

Aspectual particle – Verb – Negator Yadda maijaŋ gajaŋ. NEG . ASP place NEG . IMP ‘Don’t put them down yet.’ (Holton 2014)

The result of the limitation of the postverbal slot to a single element is that there is often a discrepancy in AP languages between the position of imperfective and perfective aspect markers. Wersing is typical in this respect: the perfective aspectual particle kana which cannot co-occur with the negator is postverbal (20a), while imperfective de which frequently combines with the negator to express ‘not yet’ is preverbal (20b), even when no negator appears postverbally (20c).

36

(20)

Antoinette Schapper

Wersing postverbal & preverbal aspectual particles a. Verb – Perfective Ne-tamu ga-paŋ kana. 1SG -grandparent 3-dead PFV ‘My grandparents are already dead.’ b.

c.

Imperfective – Verb – Negator Ne-tamu de ga-paŋ 3-dead 1SG -grandparent IPFV ‘My grandparents aren’t dead yet.’

nauŋ. NEG

Imperfective – Verb Ne-tamu de go-ko. 3-stay 1SG -grandparent IPFV ‘My grandparents are still alive.’ (Schapper fieldnotes)

In the vast majority of TAP languages, standard negation is post-verbal. The exceptions are the Eastern Timor languages (as illustrated above with 17 in Fataluku) and the Alor language Kafoa. Whilst most TAP languages have distinct negators for declarative and imperative clauses, Kafoa uses the same negator (compare 21a and 21b).

(21)

Kafoa negation a. Declarative: Negator – Verb Ne bilau na ka karai, . . . 1SG food 1SG NEG eat ‘I won’t eat the food, . . .’ b.

Imperative: Negator – Verb Ka naŋ-karai, . . . NEG 1SG -eat ‘Don’t eat me, . . .’ (Baird this volume)

Kafoa preverbal negation in declaratives appears to be due to expanding an original imperative negator to cover negation in standard clauses. Whilst imperative negators are frequently in the same postverbal position as standard negators (e.g., in Nedebang, Kamang, and Wersing in 22), sporadically TAP languages have post-verbal standard negation, but preverbal negative imperatives. Preverbal negative imperatives are illustrated for Bunaq, Kaera and Blagar (23).

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

(22)

(23)

37

Postverbal imperative negator a. Nedebang See ga-mmi mali pulumma olli. NEG house 3SG -inside spit spit ‘Don’t spit inside the house!’ (Schapper fieldnotes) b.

Kamang Ala gobalee posan-si gee. arrogant-IPFV NEG 2SG . SPEC too ‘Don’t you be so arrogant!’ (Schapper 2014)

c.

Wersing A-soroŋ siniŋ. 2SG -angry NEG ‘Don’t be angry.’ (Schapper & Hendery 2014)

Preverbal imperative negator a. Bunaq Hani holon. NEG cry ‘Don’t cry!’ (Schapper 2010) b.

Kaera Tu gu aki er max-o. milk DEM NEG make bitter-FIN ‘That milk, don’t make (it) bitter.’ (Klamer 2014)

c.

Blagar Ana ʔake naiŋ bue. 1SG .OBJ hit 2SG . SBJ NEG ‘Don’t hit me.’ (Steinhauer 2014)

The original postverbal position of negation in Kafoa can still be observed in an optional double negation construction in which postverbal ke co-occurs with the preverbal ka (24).

(24)

Kafoa optional double negation Ka nan ke. NEG thing NEG ‘It’s nothing.’ (Baird this volume)

38

Antoinette Schapper

Elsewhere in the family optional double negation involves an emphatic preverbal negator being used alongside the standard postverbal negator. This is found in Bunaq (25a), in West Alor languages such as Klon (25b), and in Kamang (25c), but is not attested in Abui, East Alor, Pantar and the Straits.

(25)

Optional double negation: preverbal negator reinforces postverbal negator a. Bunaq I ozol minaʔ sael g-utu niʔ. pig 3-COM NEG 1PL . INCL NEG oil ‘We don’t (cook) with pig’s oil.’ (Schapper 2010) b.

Klon Mih, hok muŋ naŋ. NEG fall NEG sit ‘It’s sitting, it hasn’t fallen.’ (reanalyzed from Baird 2008)

c.

Kamang Bei auh ga-tak-si naː. NEG deer 3. PAT-see-IPFV NEG ‘There was no deer in sight.’ (Schapper fieldnotes)

Another example of differences in verb finalness between TAP languages can be observed in distransitive clauses, albeit with much less variation in the family. Just as PV is consistent in transitive clauses, the same word order of ditransitive clauses is found across the TAP languages, with only a single exception. The common TAP ‘give’ construction has free (pro)nominal elements for both R (recipient argument of the giving) and T (gift which is given) in preverbal position, as in:14

(26)

Common TAP ‘give’ clause word order a. Nedebang T R APPL-R-V Gaŋ innuku ga-naŋ ma-g-eːna. 3SG . SBJ something 3SG -younger.sibling APPL-3SG -give ‘He gave something to his younger sibling.’ (Schapper fieldnotes)

14 The R argument is co-indexed by a prefix on the verb in all languages except Makasae and Fataluku, while the encoding of T varies widely across the TAP languages. See Klamer and Schapper (2012) for a detailed treatment of the variation in ‘give’ constructions in the family.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

b.

39

Kui T R R-V Oːl omo doi gala na-maː g-ei. child DIST money gala 1SG -father 3-give ‘That child gave money to my father.’ (Windschuttel & Shiohara this volume)

c.

Makasae T R V Markus ira ma ani gini. Markus water OBL 1SG give ‘Markus gave me water.’ (Huber 2008)

Bunaq is the only language which deviates from the common TAP word order, placing the T argument after the give verb (27). This gives rise to a crosslinguistically highly unusual word order (Klamer & Schapper 2012).

(27)

Bunaq ‘give’ clause word order R R-V T Neto Markus g-ege paʔol. Markus 3AN -give corn 1SG ‘I gave Markus corn.’ (Schapper 2010)

Within the NP, word order differences are again minimal in the broadest terms: N ATTR 15, N DEM /ART, N NUM and GEN N word orders are consistently found across the whole family. Differences are, however, observable in the details of individual constructions, notably possessives and deictic elements other than demonstratives. Bunaq and Fataluku, two TAP languages of Timor, show the most variation in the NP and will be the focus of the treatment here. In most TAP languages adnominal possession is GEN N, that is, the possessor noun precedes the possessed noun. The possessive relation is variously marked by a free or bound possessive marker, as in:

(28)

GEN N a. Nedebang Karel gei seː Karel 3SG . POSS house ‘Karel’s house’ (Schapper fieldnotes)

15 The label “attribute” is used here rather than “adjective” because many TAP languages lack a distinct class of adjectives. Relative clauses are also often absent.

40

Antoinette Schapper

b.

Kiraman ol ge-kui child 3SG . POSS -dog ‘the child’s dog’ (Schapper fieldnotes)

In Timor languages, however, both GEN N and N GEN orders exist. The GEN N is the most common order and the N GEN order is an alternative. Where N GEN occurs, the whole possessive phrase is postposed to the possessed nominal, not simply the possessor nominal. This is illustrated for Bunaq (29) and Fataluku (30).

(29)

Bunaq a. GEN N mone gie man 3.POSS ‘a man’s cloth’ b.

(30)

tais cloth

N GEN tais mone gie cloth man 3.POSS ‘cloth of a man’ (Schapper fieldnotes)

Fataluku a. GEN N Timor i=cau-hafa=ere Timor POSS =head-bone=PL ‘Timor’s leaders’ b.

N GEN cau-hafa Timor=hin=i=ere head-bone Timor=POSS =DEM =PL ‘the leaders of Timor’ (van Engelenhoven forthcoming)

This alternative order is an areal feature. Many Austronesian languages of eastern and central Timor and nearby islands (for instance, Kemak, Tetun, Naueti, Leti) similarly have GEN N as the basic possessive word order and N GEN as an alternative word order that is available in particular semantic, pragmatic and phonological contexts. Deictic elements such as locationals,16 demonstratives and articles within the NP in the vast majority of TAP languages follow the head noun. For instance: 16 Locationals denote a location which acts as the ground for the referent of the NP head as figure. See Schapper (2014c) and Schapper (2010: 95–97) for a discussion of different types of locationals in TAP languages.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

(31)

41

N Deictic/DEM/ART a. Adang bel mɔŋ hɛmɔ dog LEVEL DEM . LEVEL ‘that dog over there’ (Haan 2001) b.

Abui fu do nu areca.nut DEM ART ‘that areca nut’ (Schapper fieldnotes)

c.

Makalero sefar=eː dog=ART ‘the dog’ (Huber 2011)

Sporadically, we find exceptions to the noun-initial order of NP. In Fataluku and Kula, whilst determiners marking distance or number follow the noun, emphatic topic markers in each language precede the noun head (32).

(32)

Preverbal emphatic topic markers a. Fataluku fan=tupur EMPH =woman ‘that woman’ (van Engelenhoven forthcoming) b.

Kula gɐna

sen money ‘that money’ (Williams this volume)

EMPH

In Bunaq, there is a NP slot for locationals indicating elevation that precedes the head noun (33a), but this can also be postposed to a position between the head and a determiner, if any (33b). (33)

Bunaq a. esen

lolo ba mountain DEF ‘the mountain up there’ HIGH

b.

lolo esen ba mountain HIGH DEF ‘the mountain up there’ (Schapper fieldnotes)

42

Antoinette Schapper

Many more features of word order variation in TAP languages await closer consideration. Some are discussed in section 5.4 in relation to the ordering of verbs in serializations.

5.4 Serial verb constructions and complex predications Serial verb constructions are present in all TAP languages, but the type of serial verb constructions found and the degree to which they are available differ across the family. The most notable division is that, compared to Bunaq and AP languages, Eastern Timor languages have few serial verb constructions, preferring complex predications created by way of incorporation to encode events captured with serialization in other TAP languages. Motion serialization is found in all subgroups of the TAP family in two structurally uniform forms: (i) motion-action serialization, involving an open choice of motion verb followed by a verb denoting an action following on from the motion (e.g., go work, walk reach X) and (ii) directional serialization, consisting of an open choice of verb denoting a dynamic event followed by a motion verb indicating the direction in which the event denoted by the preceding verb occurs (e.g., throw descend = throw down, walk come = arrive). These types are found in Papuan and Austronesian languages across the whole of Eastern Indonesia and may perhaps be considered (near-)universal constructions in serializing languages. It is when we begin considering other kinds of serialization that the distinctiveness of the Eastern Timor languages becomes apparent. Causative (34) and resultative (35) serializations are found in Bunaq and AP languages, but not Eastern Timor languages. Both encoding cause-effect relations, the two types of serialization are closely allied and share the same structure across languages: both involve iconic ordering of a transitive verb denoting an action followed by an intransitive verb denoting the state or event resulting from the action. The only difference between the two is that the transitive action verb in a causative serialization is limited to the causative verb, typically ‘make’, while in a resultative serialization the choice of this verb is open.

(34)

Causative serialization a. Bunaq Neto buku g-ini topol. book 3AN - CAUS fall 1SG ‘I made the book fall down.’ (Schapper fieldnotes)

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

b.

(35)

43

Kula Aniŋ-jɐku gi-ape ajɐmu. NFIN . person-two 3-make FIN . die ‘The two of them killed it.’ (lit. ‘The two of them made it die.’) (Williams this volume)

Resultative serialization a. Bunaq N-ol uen oto g-eze heser. 1EXCL-child one car 3AN -crush dead ‘A car crushed one of my children dead.’ (Schapper 2010) b.

Western Pantar A-ule pai hinna kanna gaːta. finish already 4SG -neck slice dead ‘They sliced his neck and killed him.’ (lit. ‘They sliced his neck dead’) (Holton 2014)

Manner serializations are again found only in Bunaq and AP languages. This time, however, iconicity does not have a role in the ordering of serial verbs and we find different orderings of the manner verbs in the languages. In CentralEastern Alor languages, such as Wersing and Kamang, the manner verb precedes the action verb (36). Elsewhere in Alor and Pantar, we find the manner verb following the action verb, as in Nedebang and Adang (37).

(36)

Manner – Action serialization a. Wersing Gaida poko er. 3.TOP little crawl ‘He crawled a little way.’ (Schapper & Hendery 2014) b.

(37)

Kamang A deʔtaŋ maː-si? walk-IPFV 2SG . AGT limp ‘Are you walking with a limp?’ (Schapper 2014)

Action – Manner serialization a. Nedebang Wei gwaqa suʔu apa dia. DEM walk quick child little ‘That child is walking quickly.’ (Schapper fieldnotes)

44

Antoinette Schapper

b.

Adang Roni lamɛ hanɔʔaŋ. Roni walk hurry ‘Roni walked hurriedly.’ (Haan 2001)

In Bunaq, both orderings of manner verb and action verbs in serialization are possible, with slightly different semantics. Compare the serializations in (38). Where the manner verb precedes the action verb, the quickness on the part of the agent in the act of eating is emphasized (38a). Where the manner verb follows the motion verb, emphasis is on the fact that the act of eating happened quickly.

(38)

Bunaq manner serialization a. Manner – Action Neto laun bai a. quick thing eat 1SG ‘I am quick at eating.’ b.

Action – Manner Neto bai a laun. thing eat quick 1SG ‘I eat quickly.’ (Schapper 2010)

In Eastern Timor languages, verb serialization is not used to encode manner and result. Instead, incorporation of the result (39) and manner (40) verb into the preverbal slot is consistently used (see Huber this volume for a detailed description of the behavior of this slot in Makasae-Makalero). The incorporation is seen in the use of bound verb forms and/or the movement of theme NPs into oblique phrase introduced by the verb ‘take’.17

(39)

Resultative incorporation a. Makasae Gi bai guta-duri. 3SG pig kill-shoot:BOUND ‘He shot the pig dead.’ (Huber pers. comm.)

17 Note that a ‘take’ verb is not used in (39a) because the incorporated verb is transitive, making the verb complex as a whole transitive.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

b.

(40)

45

Makalero Kiloo pai u k-umu-suri. pig one 3-dead-shoot:BOUND 3SG ‘He shot a pig dead.’ (Huber pers. comm.)

Manner incorporation a. Makasae Gi fergu ere ma neinei-base. DEM take slow-hit 3SG nail ‘He hit the nail slowly.’ (Correia 2011: 223) b.

Makalero Kiloo seur mei=ni nainai-dina. meat take=CONJ slow-cook:BOUND 3SG ‘He cooked the meat slowly.’ (Huber pers. comm.)

Causatives in Eastern Timor languages are also normally constructed by means of incorporation of the caused verb into the preverbal slot, as seen in (41). In Oirata, however, the incorporated causative construction (42a) is being replaced by a serialization construction (42b), likely due to calquing of the causative structures of neighboring Austronesian languages.

(41)

Causative incorporation a. Makalero Kiloo ni-lopu mei=ni hare-nini. REFL -house take=CONJ clean-do:BOUND 3SG ‘She is cleaning the house.’ (Huber this volume) b.

Fataluku Tapa e-m

kaparu-pai. bad:NMLZ-do:BOUND ‘Don’t destroy it.’ (lit. ‘Don’t bad-make it’) (van Engelenhoven forthcoming) NEG . IMP

(42)

ANAPH -take

Oirata a. Causative incorporation ira ar-pai water hot-do ‘heat water’

46

Antoinette Schapper

b.

Causative serialization pai titlene do dry ‘dry (something)’ (Nazarudin 2015)

Not all constructions encoded with serial verbs in Bunaq and AP languages are matched by incorporation constructions in Eastern Timor. For instance, aspectual serialization is widely attested in Bunaq and AP languages, but aspectual markers are not incorporated into the preverb slot in Eastern Timor languages and instead are encoded by preverbal markers (see Huber this volume, Heston 2015). Aspectual serializations are also a lot more varied in form and function across those languages which do have them. Specific types of aspectual serialization tend to be geographically limited. For example, imperfective serialization using the verb ‘stay’ after the main semantic verb is specific to East Alor languages, while use of the existential verb (< PAP *waˈna ‘be, exist’) in serializations to express continuative aspect is specific to Pantar languages. In sum, the reliance on serial verbs for encoding complex events is not absolute in TAP languages. In those languages where serialization is central, areal differences in the orderings of serial verbs without any clear iconic relationship and in the lexical verbs associated with certain serialization types are readily apparent. This section has provided a taster of just a few variations that are found in complex predications in TAP languages, but many more remain to be articulated by future research.

6 This book The sketch chapters in this volume follow the same highly structured format as found in Volume 1. Each author provides information on the same important topics, namely: phonology, clause structure, noun phrases, verbal morphology in particular person-number prefixes, independent pronouns, serial verb constructions, and aspect marking. Readers with interests in diverse topics such as stress systems, reduplication, demonstratives and deixis, numeral systems, agreement and alignment, as well as verb serialization and more, will discover aspects of the TAP languages captivating and stimulating. TAP linguistics is a relatively new field and its terminology still shows flux (e.g., pronominal prefixes versus agreement prefixes). The largely parallel structures of chapters, however, allow the reader to directly compare phenomena between languages. Detailed tables of contents at the beginning of each chapter will also assist the reader in navigating quickly to treatments of desired topics. IPA has been used in the

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

47

Map 4: Major field sites and places of study in this volume

phonology sections of all sketches, but other sections use the individual orthographies of the different authors. Readers should keep in mind these potential differences. This volume, Volume 2, presents new descriptions of five languages over four chapters. Three previously undescribed languages of Alor are covered in the first three chapters: Kafoa (Baird this volume) located in the interior of western Alor, Kui (Windschuttel & Shiohara this volume) spoken in coastal settlements in western Alor, and Kula (Williams this volume) located in the inland mountainous region of eastern Alor. The final chapter (Huber this volume) presents a comparative sketch of Makalero and Makasae, two closely related languages of Eastern Timor. Sketches are primarily based on original fieldwork in Alor and Timor. Map 4 presents the major fieldwork sites and places of study in this volume cartographically. Taken together, these volumes represent a significant advance for descriptive Papuan linguistics, providing an authoritative and comprehensive overview of the grammars of the TAP family. There is still ground to be covered before we have a complete picture of the TAP languages, but with the appearance of these volumes we now have descriptions for around three quarters of the family. That said, almost all aspects of the TAP languages are still in need of analytic work. Investigations of phonetics, (lexico-)semantics, discourse and sociolinguistics

48

Antoinette Schapper

remain all but non-existent in the TAP languages. These volumes will also aid in the comparison of TAP with other Papuan languages, and contribute to the construction of a better picture of Island Southeast Asian and Melanesian prehistory.

Acknowledgments My research and editing of this volume was supported by a Netherlands Organisation for Scientific Research VENI project “The evolution of the lexicon. Explorations in lexical stability, semantic shift and borrowing in a Papuan language family”. Many thanks to Juliette Huber and Aone van Engelenhoven for their help with understanding the Eastern Timor languages of the TAP family better. I am indebted to Lourens de Vries, Ger Reesink, Malcolm Ross and Juliette Huber for providing comments on this introduction. Finally, Angela Terrill and Bethwyn Evans provided invaluable assistance in getting this volume ready for printing. All errors are, of course, my own.

References Anceaux, J. C. 1973. Naschrift. Bijdragen tot de Taal-, Land- en Volkenkunde 129: 345–346. Baird, Louise. 2008. A grammar of Klon: A non-Austronesian language of Alor, Indonesia. Canberra: Pacific Linguistics. Baird, Louise. this volume. Kafoa. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 2. Berlin: De Gruyter Mouton. Bellwood, Peter. 2006. Austronesian prehistory in Southeast Asia: Homeland, expansion and transformation. In Peter Bellwood, James J. Fox & Darrell Tryon (eds.), The Austronesians: Historical and comparative perspectives, 103–118. Canberra: ANU Epress. Brown, S.K., N.C. Pedersen, S. Jafarishorijeh, D.L. Bannasch, K.D. Ahrens, J.T. Wu, M. Okon, B.N. Sacks. 2011. Phylogenetic distinctiveness of Middle Eastern and Southeast Asian village dog Y chromosomes illuminates dog origins. PLoS One 6:e28496 Capell, Arthur. 1975. The West Papuan phylum: General, and Timor and areas further west. In Stephen A. Wurm (ed.), New Guinea area languages and language study, Papuan Languages and the New Guinea linguistic scene. Volume 1, 667–716. Canberra: Pacific Linguistics. Correia, Adérito José Guterres. 2011. Describing Makasae: A Trans-New Guinea language of East Timor. PhD thesis, University of Western Sydney. Cottet, Fanny. 2015. A phonology of Mbahám: Reduction and contrast. PhD thesis, The Australian National University. Delpada, Benidiktus. 2016. Abui phonology. MA thesis, Nanyang Technological University. Denham, T. 2011. Early agriculture and plant domestication in New Guinea and Island Southeast Asia. Current Anthropology 52: 379–395.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

49

Donohue, Mark. 1996. Inverse in Tanglapui. Language and Linguistics in Melanesia 27: 101– 118. Donohue, Mark. 2008. Bound pronominals in the West Papuan languages. In Claire Bowern, Bethwyn Evans & Luisa Miceli (eds.), Morphology and language history: In honour of Harold Koch, 43–58. Amsterdam: John Benjamins. Donohue, Mark & Antoinette Schapper. 2008. Whence the Austronesian indirect possessive construction? Oceanic Linguistics 47: 316–327. Edwards, Owen. 2016. Parallel sound correspondences in Uab Meto. Oceanic Linguistics 55.1: 52–86. Engelenhoven, Aone van & Nazarudin. 2016. A tale of narrative annexation. Stories from Kisar Island (Southwest Maluku, Indonesia). Wacana. Journal of the Humanities of Indonesia 17: 191–231. Engelenhoven, Aone van. forthcoming. Fataluku. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 3. Berlin: Mouton de Gruyter. Haan, Johnson. 2001. The grammar of Adang, a Papuan language spoken on the island of Alor, East Nusa Tenggara, Indonesia. PhD thesis, University of Sydney. Hammarström, Harald. 2012. Pronouns and the (preliminary) classification of Papuan languages. Language and Linguistics in Melanesia: Special issue on the history, contact & classification of Papuan languages, 428–539. Han, Tan Jian. 2015. A sketch grammar of Suboo. BA dissertation, Nanyang Technological University. Heinsohn, Thomas E. 2010. Marsupials as introduced species. In Simon Haberle, Janelle Stevenson & Matthew Prebble (eds.), Altered ecologies: Fire, climate and human influence on terrestrial landscapes, 133–176. Canberra: Terra Australis, ANU Epress. Heston, Tyler M. 2015. The segmental and suprasegmental phonology of Fataluku. PhD thesis, University of Hawai’i at Mānoa. Holton, Gary. 2010. Person-marking, verb classes, and the notion of grammatical alignment in Western Pantar (Lamma). In Michael C. Ewing & Marian Klamer (eds.), Typological and areal analyses: Contributions from East Nusantara, 101–121. Canberra: Pacific Linguistics. Holton, Gary. 2014. Western Pantar. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 23–96. Berlin: De Gruyter Mouton. Holton, Gary, Marian Klamer, František Kratochvíl, Laura Robinson & Antoinette Schapper. 2012. The historical relations of the Papuan languages of Alor and Pantar. Oceanic Linguistics 51: 86–122. Holton, Gary & Laura Robinson. 2014. The linguistic position of the Timor-Alor-Pantar languages. In Marian Klamer (ed.), Alor-Pantar languages: History and typology, 155–198. Berlin: Language Science Press. Huber, Juliette. 2008. First steps towards a grammar of Makasae. München: Lincom. Huber, Juliette. 2011. A grammar of Makalero: A Papuan language of East Timor. PhD thesis, Leiden University. Huber, Juliette. 2016. Blurring the boundaries: Identifying foreign etyma in the Papuan and Austronesian languages of Eastern Timor. Talk presented at the Workshop on contact and substrate in the languages of Wallacea, December 1–2. Huber, Juliette. this volume. Makalero and Makasae. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 2. Berlin: Mouton de Gruyter. Huber, Juliette, Antoinette Schapper & Aone van Engelenhoven. 2015. Homelands and dispersals in eastern Timor: A study of placename etymologies. Presented at the 13th International Conference on Austronesian Linguistics, Academia Sinica, Taiwan, 18–23 July 2015.

50

Antoinette Schapper

Hull, Geoffrey. 2004. The Papuan Languages of Timor. Studies in Languages and Cultures of East Timor 6: 23–99. Humaedi, M. Alie, Abdul Rachman Patji, Sudiyono, Sihol Farida Tambunan & Widhyasmaramurti. 2013. Mekanisme internal pelestarian bahasa & budaya Kafoa. [Internal preservation mechanisms for the Kafoa language and culture.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Humaedi, M. Alie, Abdul Rachman Patji, Sihol Farida Tambunan, Sudiyono, Anastasia Melati & Widhyasmaramurti. 2014. Strategi pelestarian bahasa dan budaya Kafoa: sistem sosial budaya dan kebijakan. [Preservation strategies for the Kafoa language and culture: Social system, culture and wisdom.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Ino, La. 2013. Protobahasa Modebur, Kaera dan Teiwa: Bahasa kerabat non-Austronesia di Pulau Pantar, Nusa Tenggara Timur. [Proto-language of Modebur, Kaera and Teiwa: Related non-Austronesian languages on Pantar Island, Nusa Tenggara Timur.] PhD thesis, Universitas Udayana. Katubi, Ninuk Kleden-Probonegoro, Datang & Frans Asisi. 2004. Bahasa & kebudayaan Hamap, kelompok minoritas di Alor. [Hamap language and culture, a minority group in Alor.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Katubi, Thung Ju Lan & Marcel Akoli. 2013. Kamus kecil bahasa Kui. [A short dictionary of Kui.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Katubi & Thung Ju Lan. 2014. Vitalitas etnolinguistik dan agen pemertahanan bahasa Kui di Alor, Nusa Tenggara Timur. [Ethnolinguistic vitality and agents for retention of the Kui language in Alor, Nusa Tenggara Timur.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Klamer, Marian. 2010a. One item, many faces: ‘come’ in Teiwa and Kaera. In Michael C. Ewing & Marian Klamer (eds.), Typological and areal analyses: Contributions from East Nusantara, 205–227. Canberra: Pacific Linguistics. Klamer, Marian. 2010b. A grammar of Teiwa. Berlin: Mouton de Gruyter. Klamer, Marian. 2012. Papuan-Austronesian language contact: Alorese from an areal perspective. In Nicholas Evans & Marian Klamer (eds.), Melanesian Languages on the Edge of Asia: Challenges for the 21st Century. Language Documentation & Conservation Special Publication No. 5, 72–108. Klamer, Marian. 2014. Kaera. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar: Volume 1. 97–146. Berlin: De Gruyter Mouton. Klamer, Marian & Antoinette Schapper. 2012. The history of ‘give’ constructions in the Papuan languages of Timor-Alor-Pantar. Linguistic Discovery 10: 174–207. Kleden-Probonegoro, Ninuk, Katubi & Fanny Henry Tondo. 2005. Identitas etnolinguistik Orang Hamap: Kode etnisitas dan bahasa symbol. [Ethnolinguistic identity of the Hamap people: Ethnic code and symbolic language.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Kleden-Probonegoro, Ninuk, Katubi, Fanny Henry Tondo & Imelda. 2007. Identitas etnolinguistik orang Hamap: Perubahan dalam divergensi dan konvergensi. [Ethnolinguistic identity of the Hamap people: Changes in divergence and convergence.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

51

Kleden-Probonegoro, Ninuk. 2008. The ethnolinguistic identity of the Hamap people in change. Journal of Indonesian Social Sciences and Humanities 1: 193–206. Kratochvíl, František. 2007. A grammar of Abui: A Papuan language of Alor. PhD thesis, Leiden University. Kratochvíl, František. 2014. Sawila. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 351–438. Berlin: De Gruyter Mouton. Kratochvíl, František, Isak Bantara, and Anderias Malaikosa. 2014. Sawila–English dictionary. MS, Nanyang Technological University. Mandala, Halus. 2010. Evolusi fonologis bahasa Oirata dan kekerabatannya dengan bahasabahasa non-Austronesia di Timor-Leste. [Phonological evolution of Oirata and its relationship with the non-Austronesian languages of East Timor.] PhD thesis, Universitas Udayana. Nazarudin. 2013. Kamus Kecil Bahasa Oirata. [Small dictionary of the Oirata language.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Nazarudin. 2015. Causative constructions in Woirata, Kisar Island (Southwest Maluku, Indonesia). Wacana, Journal of the Humanities of Indonesia 16: 27–41. Nazarudin. in preparation. Language contact on Kisar Island, Southwest Maluku, Indonesia; The case of Woirata. PhD thesis, Leiden University. Nitbani, Semuel H., Jeladu Kosmas, Sisila Wona & Hilda Naley. 2001. Struktur Bahasa Lamma. Jakarta: Pusat Bahasa, Departemen Pendidikan Nasional. O’Connor, Susan. 2015. Rethinking the Neolithic in Island Southeast Asia, with particular reference to the archaeology of Timor Leste and Sulawesi. Archipel 90: 15–47. Oskarsson, M.C.R., C.F.C. Klütsch, U. Boonyaprakob, A. Wilton, Y. Tanabe, P. Savolainen. 2011. Mitochondrial DNA data indicate an introduction through Mainland Southeast Asia for Australian dingoes and Polynesian domestic dogs. Proceedings of the Royal Society B: Biological Sciences 279: 967–974. Patji, Abdul Rachman. 2014. Bahasa, kebudayaan dan pandangan tentang kebahasaan masyarakat etnik (lokal) Kafoa di Alor, Nusa Tenggara Timur. [Language, culture and outlooks for popular linguistics amongst the Kafoa in Alor, Nusa Tenggara Timur.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Pawley, Andrew K. 2001. The Proto Trans New Guinea obstruents: Arguments from top-down reconstruction. In Andrew Pawley, Malcolm Ross & Darrell Tryon (eds.), The boy from Bundaberg: Studies in Melanesian linguistics in honour of Tom Dutton, 261–300. Canberra: Pacific Linguistics. Pawley, Andrew. 2005. Introduction to the chapters on historical linguistics. In Andrew Pawley, Robert Attenborough, Jack Golson & Robin Hide (eds.), Papuan Pasts: Cultural, linguistic and biological histories of Papuan-speaking peoples, 1–14. Canberra: Pacific Linguistics. Pawley, Andrew. 2012. How reconstructable is Proto Trans New Guinea? Problems, progress, prospects. Language and Linguistics in Melanesia: Special issue on the history, contact & classification of Papuan languages, 88–164. Reesink, Ger, Ruth Singer & Michael Dunn. 2009. Explaining the linguistic diversity of Sahul using population models. PLoS Biol 7: e1000241. Robinson, Laura. 2015. The Alor-Pantar (Papuan) languages and Austronesian contact in East Nusantara. In Malcolm Ross & I Wayan Arka (eds.), Language change in Austronesian languages, 19–33. Canberra: Pacific Linguistics Ross, Malcolm. 2005. Pronouns as a preliminary diagnostic for grouping Papuan languages. In Andrew Pawley, Robert Attenborough, Robin Hide & Jack Golson (eds.), Papuan pasts:

52

Antoinette Schapper

Cultural, linguistic and biological histories of Papuan-speaking peoples, 15–66. Canberra: Pacific Linguistics. Santosa, Puji. 2012. Kearifan budaya dan fungsi kemasyarakatan dalam sastra lisan Kafoa. [Cultural wisdom in and social functions of oral traditions of the Kafoa.] Meta Sastra 5: 67–82. Savio, Edegar da Conceição. 2016. Studi sosiolinguistik bahasa Fataluku di Lautem. [A sociolinguistic study of the Fataluku language in Lautem.] PhD thesis, Leiden University. Savolainen, P., T. Leitner, A.N. Wilton, E. Matisoo-Smith, J. Lundeberg. 2004. A detailed picture of the origin of the Australian dingo, obtained from the study of mitochondrial DNA. Proceedings of the National Academy of Sciences 101: 12387–12390. Schapper, Antoinette. 2009. Isolating Timor: The origins of isolating word structure in Timor. Talk given at the Isolating Languages Panel, International Conference of Austronesian Linguistics, Aussois France, 22–26 June. Schapper, Antoinette. 2010. Bunaq, a Papuan language of central Timor. PhD thesis, The Australian National University, Canberra. Schapper, Antoinette. 2011a. Crossing the border. Historical and linguistic divides among the Bunaq in central Timor. Wacana, Journal of the Humanities of Indonesia 13: 29–49. Schapper, Antoinette. 2011b. Finding Bunaq: The homeland and expansion of the Bunaq in central Timor. In Andrew McWilliam & Elizabeth G. Traube (eds.), Life and land in Timor: Ethnographic papers, 163–186. Canberra: ANU Epress. Schapper, Antoinette. 2014a. Introduction. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 1–22. Berlin: De Gruyter Mouton. Schapper, Antoinette. 2014b. Kamang. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 285–350. Berlin: De Gruyter Mouton. Schapper, Antoinette. 2014c. Elevational systems in Alor-Pantar languages. In Marian Klamer (ed.), The Alor-Pantar languages: History and typology, 247–285. Berlin: Language Science Press. Schapper, Antoinette. 2015. Wallacea, a linguistic area. Archipel 90: 99–151. Schapper, Antoinette. 2017. Stress and gemination in Alor-Pantar languages. Revising Heston (2016). Oceanic Linguistics 56.1: 257–266. Schapper, Antoinette. forthcoming. Nedebang. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 3. Berlin: De Gruyter Mouton. Schapper, Antoinette & Rachel Hendery. 2014. Wersing. In Antoinette Schapper (ed.), Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 439–504. Berlin: De Gruyter Mouton. Schapper, Antoinette & Juliette Huber. 2012. State-of-the-art in the documentation of the Papuan languages of Timor, Alor, Pantar and Kisar. Wacana, Journal of the Humanities of Indonesia 14.2: 370–404. Schapper, Antoinette & Marten Manimau. 2011. Kamus Pengantar Bahasa Kamang-IndonesiaInggris [Introductory Kamang – Indonesian – English Dictionary.] Kupang, Indonesia: UBBGMIT. Schapper, Antoinette, Juliette Huber & Aone van Engelenhoven. 2012. The historical relations of the Papuan languages of Timor and Kisar. Language and Linguistics in Melanesia: Special issue on the history, contact & classification of Papuan languages, 194–242. Schapper, Antoinette, Juliette Huber & Aone van Engelenhoven. 2014. The relatedness of TimorKisar and Alor-Pantar languages: A preliminary demonstration. In Marian Klamer (ed.), The Alor-Pantar languages: History and typology, 99–154. Berlin: Language Science Press.

Introduction to The Papuan Languages of Timor, Alor and Pantar. Volume 2

53

Steinhauer, Hein. 1995. Two varieties of the Blagar language (Alor, Indonesia). In Connie Baak, Mary Bakker & Dick van der Meij (eds.), Tales from a concave world: Liber amicorum Bert Voorhoeve, 269–296. Leiden: Department of Languages and Cultures of South-East Asia and Oceania. Steinhauer, Hein. 2014. Blagar. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 147–220. Berlin: De Gruyter Mouton. Stoel, Ruben. 2007. The Prosody of Fataluku. Paper presented at the First Workshop on the Languages of Papua, Universitas Negeri Papua, Manokwari. August 8–10. Stokhof, W.A.L. 1975. Preliminary notes on the Alor and Pantar languages (East Indonesia). Canberra: Pacific Linguistics. Suter, Edgar. 2012. Verbs with pronominal object prefixes in Finisterre-Huon languages. Language and Linguistics in Melanesia: Special issue on the history, contact & classification of Papuan languages, 23–58. Usher, Timothy. West Bomberai. URL: https://sites.google.com/site/newguineaworld/families/ trans-new-guinea/ Visser, Eline. 2016. A grammar sketch of Kalamang with a focus on phonetics and phonology. MA thesis, University of Oslo. Voorhoeve, C. L. 1975. The central and western areas of the Trans-New Guinea phylum: Central and Western Trans-New Guinea phylum languages. In Stephen A. Wurm (ed.), New Guinea area languages and language study. Volume 1, 345–460. Canberra: Pacific Linguistics. Wellfelt, Emilie. 2016. Historyscapes in Alor. Approaching indigenous histories in eastern Indonesia. PhD thesis, Linnaeus University. Wellfelt, Emilie & Antoinette Schapper. 2013. Enriching the archival picture: The use of local sources in assessing the nature and impact of historical contact. Talk presented as part of the Panel on East Timor, the Portuguese, and the Indonesian Archipelago- Historical Sources and Methodological Reconsiderations at the International Convention of Asia Scholars 8, Macau, 24–27 June. Widhyasmaramurti. 2013. Kamus kecil bahasa Kafoa. [Short dictionary of the Kafoa language.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Willams, Nicholas. 2016. Place reference and location formulation in Kula conversation. PhD thesis, University of Colorado. Willams, Nicholas. this volume. Kula. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 2. Berlin: De Gruyter Mouton. Willemsen, Jeroen. 2015. Function and grammaticalization in Alor-Pantar asymmetrical verb serialisation. MA thesis, Leiden University. Willemsen, Jeroen. forthcoming. Reta. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 3. Berlin: De Gruyter Mouton. Windschuttel, Glenn & Asako Shiohara. this volume. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 2. Berlin: De Gruyter Mouton. Wurm, Stephen A., C. L. Voorhoeve & Kenneth A. McElhanon. 1975. The Trans-New Guinea phylum in general. In Stephen A. Wurm (ed.), New Guinea area languages and language study. Volume 1, 299–322. Canberra: Pacific Linguistics.

Louise Baird

2 Kafoa 1

The language scene

2 2.1 2.2 2.3 2.4 2.5 2.6 2.6.1 2.6.2 2.6.3

58 Phonology 58 Consonants 61 Vowels 62 Phonotactics 63 Syllable structure 65 Stress 66 Morphophonemics 67 Reduplication 69 Compounding 70 Prefixal allomorphy

3 3.1 3.2 3.3 3.4 3.5 3.6 3.7

70 Basic clausal syntax 70 Constituent order 72 Non-verbal predicates 73 Locative verbs 75 Predicative possession 76 Negation 78 Interrogatives 79 Imperatives

4 4.1 4.2 4.3 4.4 4.5 4.6

81 Noun phrases Noun phrase structure 81 Relative clauses 82 Adjectives 83 Numerals Non-numeral quantifiers 85 Possession

5 5.1 5.2 5.3 5.4 5.5 5.6

88 Pronouns Pronoun paradigms 89 eed pronouns 90 a pronouns 90 o pronouns 91 e pronouns 92 el pronouns

DOI 10.1515/9781614519027-002

56

81

84

88

56

Louise Baird

6 6.1 6.2 6.3 6.4

Person prefixes and valence adjustment 94 Person prefixes The third / fourth person distinction 97 Reciprocal 98 Applicative prefixes u- and mi-

7

Serial verb constructions

8 8.1 8.2 8.3

102 Aspect marking 103 Completive yai 103 Imperfective dara 104 Stem alternations

9

Discussion

References

94 96

99

105

107

1 The language scene Kafoa (ISO 639–3 code: kpu) is a small language primarily spoken in the hamlet of Habolat (alternatively spelt Habollat and Hifolat) in the village of Probur Utara in the west of the island of Alor. According to local government figures, in 2004 Habolat had a population of 600 people. A small number of Kafoa speakers are also found in the multi-ethnic, linguistically diverse, coastal hamlet of Lola, 5km to the West of Habolat (Simpadat PMB LIPI 2014). Speakers reported that the name ‘Kafoa’ is used to refer both to their ethnic group and the language they speak. Other names for Kafoa listed in Ethnologue (Lewis, Simons & Fennig 2014) include Aikoli, Fanating, Jafoo, Pailelang and Ruilak.1 Kafoa divide themselves into twelve clans. Kafoa Kilelang is a term used to refer to all of these twelve Kafoa-speaking clans. In the past, these twelve clans have been separated and scattered throughout the western part of Alor. They have all resided together in Habolat since 1961. Kafoa regard themselves as refugees. Reportedly, they have in the past lived amongst Kui speakers on the south-west coast of Alor, amongst Abui speakers in central Alor, amongst Alorese speakers on the west coast of the Alor Bird’s Head, and amongst Kui, Hamap and Abui speakers on the southern side of Pearl Bay. Currently the Kafoa speech

1 I did not check knowledge of these names with speakers.

Kafoa

57

community borders with nearby Klon speech communities to the south and west and with more geographically distant Abui speech communities to the north and east. Speakers say the reason they have moved so much is due to being the victims of wars and famine. The intense contact with other local languages has led to Kafoa speakers being locally renowned for their multilingualism. Due to the location of Habolat on the main road running from the north to the southern part of west Alor, speakers of other regional languages, especially Klon speakers, frequently pass through the hamlet. Fewer Abui speakers pass through the hamlet than Klon speakers. However, prior to setting up the hamlet of Habolat, most Kafoa speakers resided in the Abui speaking area, so that all Kafoa over the age of 60 are fluent Abui speakers, and most have ongoing relationships with Abui speakers, some Kafoa speakers frequently go to Abui speaking areas, and Abui continues to be learnt by children. Speakers said that parents speak Indonesian2 almost exclusively to small children, but once children reach school age, and start attending communal gatherings they have extensive exposure to both Kafoa and other regional languages, and soon become proficient in them. It was explained that by about age ten most Kafoa children were fluent speakers of Malay, Kafoa, Klon and Abui. Depending on their family circumstances and the path that their lives take, Kafoa speakers may additionally become fluent in Alorese and/or Kui and/or Blagar (spoken on the nearby island of Pura). Recently, The Indonesian Institute of Sciences (LIPI) established research teams to investigate several endangered languages in Indonesia, including Kafoa. A team with expertise in the fields of culture, philosophy, religion and the Javanese language collected a variety of data from Kafoa speakers in Habolat and Lola, resulting in a YouTube video (Simpadat PMB LIPI 2014) and several publications including Santosa (2012), Widhyasmaramurti (2013) and Humaedi et al. (2013, 2014). Their research as presented in the YouTube video confirms a strong preference for parents to speak to children in Malay, and a teacherimplemented prohibition on the use of Kafoa at school. The reasons provided by Kafoa speakers for this preference, either to myself or in the video, include: mixed marriages, where mothers typically came from the nearby island of Pura; the idea that children would just ‘pick up’ their mother tongue as they grew up; and most commonly, because Indonesian is seen as a prestige language, and important for education and by implication a better future.

2 Parents used the term “Bahasa Indonesia” to refer to the language they speak to their children. The variety they speak is local Alor Malay.

58

Louise Baird

The data used in the preparation of this sketch was collected by the author in 2003 and 2004, and has been subsequently archived in PARADISEC (Baird 2003). It consists of approximately 30 minutes of recorded texts, supplemented by some elicited sentences and a word list consisting of 262 items. In total the corpus consists of just 228 sentential utterances. In 2003 Mr. Ibrahim Lapaibel, who at that time was in his mid-forties, provided a word list, elicited sentences and the Frog Story. In 2004 all of the remaining recorded texts were provided by Mr. Karel Mohar, who at that time was 67 years old. The texts consist of a historical overview of the Kafoa people, a list of items required for brideprice, and two folk stories.3 Needless to say the analysis here, based on such a small corpus, is preliminary and much further research is required.

2 Phonology 2.1 Consonants The consonant phonemes of Kafoa are presented in Table 1. Orthographic symbols appear in angle brackets, where different from the IPA symbol. The orthography is that used by speakers with the author in the field. It is the same as Indonesian orthography, with two exceptions in this sketch for long vowels, and the back vowel /ʊ/ (see section 2.2). Table 1: Consonant phonemes Labial Plosive Fricative Nasal Approximant Rhotic Lateral

p f

Alveolar b m w

t s

Palatal

Velar

d

k

n

ŋ

Glottal g h

j r l

3 The corpus consists of: Adat Kafoa Kilelang ‘Kafoa Kilelang customary law’ (AKK); elicited story Kafoa Frog Story (KFS); narrative Kafoa gi Wom 1 ‘Kafoa’s history 1’ (KGW1); narrative Kafoa gi Wom 2 ‘Kafoa’s history 2’ (KGW2); narrative Kafoa gi Wom 3 ‘Kafoa’s history 3’ (KGW3); story Kupai gi Amanuka ‘The Jungle People’ (KGA); story Murei Rarewang ‘Sweet Lemon’ (MR); isolated elicited sentences (ES); and elicited survey sentences (ESS).

Kafoa

59

Some minimal and near-minimal pairs of consonant phonemes are presented in (1). (1) /b/ ≠ /p/: /bai/ ‘wake up’ /d/ ≠ /t/:

/da-/ ‘3.INAL ’

/karlowɛn pai/ ‘male ancestor’s name’ /ta-/ ‘1PL . INCL . INAL ’

/g/ ≠ /k/: /ga-baŋ/ ‘3SG -shoulder’

/kabaŋ/ ‘sacred, taboo’

/m/ ≠ /n/: /ma/ ‘use, move’

/na/ ‘1SG ’

/n/ ≠ /ŋ/: /irin/ ‘circular dance, sing’ /irɛŋ/ ‘how much?’ /f/ ≠ /h/:

/fakaŋ/ ‘open’

/tahakaŋ/ ‘our back’

/s/ ≠ /h/: /isʊi/ ‘steep ravine’

/ihi/ ‘fruit’

/r/ ≠ /l/:

/ril/ ‘climb’

/lil/ ‘cuddle’

/w/ ≠ /j/: /wakafoka/ ‘body’

/jaka/ ‘to search’

/w/ ≠ /b/: /wala/ ‘to swell’

/bala/ ‘wall’

As noted, the word list of 262 items was provided by a single speaker. He was recorded repeating each item three times. In many cases, the three repetitions of a single word were phonetically different. Within texts (spoken by a different speaker to the one who recorded the word list) differences were also found in the pronunciation of individual words. From this it appears there is some free variation between certain pairs of sounds: three phonemic stops have phonetic fricative variants, two phonemic fricatives have different phonetic fricative variants, and a phonemic approximant has a different phonetic approximant variant. Sociolinguistic factors that may influence the choice of one sound over another have yet to be established. Three of the consonant phonemes can be realized by two allophones each. The allophones have the same voicing and same place of articulation, with one realized as a stop and the other as a fricative. These pairs are [k] / [x]; [b] / [β]; and [g] / [ɣ]. The distinction between [k] / [x] and [b] / [β] was only found wordmedially, while the [g] / [ɣ] alternation was found both word-initially and wordmedially. The use of the stop variant is more common, with relatively few items containing the fricative variant. Therefore the stop variant is considered the phoneme.

60

(2)

(3)

(4)

Louise Baird

Phoneme /b/ /b/ [b]

~

[β]

/tabibai/

[ˈta.biˌbai]

[ˈta.biˌβai]

‘our lips’

/tɛbuku/

[ˈtɛ.buˌku]

[ˈtɛ.βuˌku]

‘our knees’

Phoneme /k/ /k/ [k]

~

[x]

/palaka/

[ˈpa.laˌka]

[ˈpa.laˌxa]

‘deaf’

/taka/

[ˈta.ka]

[ˈta.xa]

‘stab’

Phoneme /g/ /g/ [g]

~

[ɣ]

/gamuku/

[ˈgaˌmu.ku]

[ˈɣaˌmu.ku]

‘horn’

/gawaikai/

[gaˈwaiˌkai]

[ɣaˈwaiˌkai]

‘tail’

/gɛːŋ/

[gɛːŋ]

[ɣɛːŋ]

‘at’

/asaganuku/

[ˈa.saˌga.nu.ku]

[ˈa.saˌɣa.nu.ku]

‘one hundred’

The phoneme /f/ has two variants: a bilabial allophone and a labiodental allophone, as seen in (5). Some words can be pronounced using either the voiceless labiodental fricative [f] or the voiceless bilabial fricative [ɸ]. Instances of this alternation occurred word-initially. Due to the low number of such pairs, and the overwhelming preference for its use, the voiceless labiodental fricative is regarded as the underlying phoneme.

(5)

Phoneme /f/ /f/ [f] ~

[ɸ]

/fagai/

[faˈgai]

[ɸaˈgai]

‘crocodile’

/fʊiha/

[ˈfʊiˌha]

[ˈɸʊiˌha]

‘chew betel nut’

The phoneme /w/ has two allophones: the voiced labiovelar approximant [w] and the voiced labiodental approximant [ʋ], as seen in (6). There are many instances of this alternation, and it occurs both word-initially and word-medially. The alternation always occurs when the approximant is followed by a front vowel. However, there are instances of words in the word list where the approximant [w] is followed by a front vowel and no alternation is present (for example [tawɛi] meaning ‘our ear’). There are instances in the data of [w] occurring in a word without a [ʋ] alternation, but there are no instances of the reverse situation, and so the labiovelar approximant is taken to represent the underlying phoneme.

Kafoa

(6)

Phoneme /w/ /w/ [w]

~

61

[ʋ]

/wɛː/

[wɛː]

[ʋɛː]

‘blood’

/wɛl/

[wɛl]

[ʋɛl]

‘child’

/wihoko/

[ˈwi.hoˌko]

[ˈʋi.hoˌko]

‘here’

/wɛiu/

[ˈwɛi.u]

[ˈʋɛi.u]

‘below’

/wɛːla/

[ˈwɛː.la]

[ˈʋɛː.la]

‘to bathe’

/iwɛhɛŋ/

[ˈi.wɛˌhɛŋ]

[ˈi.ʋɛˌhɛŋ]

‘five’

In the word list a glottal stop was pronounced at the beginning of some vowel-initial words, and was occasionally used to separate vowel sequences, but does not appear to be phonemic.

2.2 Vowels Kafoa has six pairs of long and short vowels, as presented in Table 2. Orthographically long vowels are represented in this sketch by a doubled vowel graph. The high front vowels are represented by the symbols for the short vowel and for the long vowel. The front mid vowels are realized phonetically as open. The short variant is represented by the symbol and the long vowel is represented by the symbol . The high back vowels are represented as for the short vowel and for the long vowel. The phonemes /ʊ/ and /ʊː/ are written as and respectively. Orthographically the short mid back vowel /o/ is represented by the symbol and the symbol is used for the long vowel /ɔː/. The low central vowels /a/ and /aː/ are represented by the symbols and respectively. Table 2: Vowel phonemes Front High

Central

Back

i iː

u uː ʊ ʊː o

Mid ɛ ɛː Low

ɔː a

aː

The short high front vowel /i/ is realized by the two allophones [i] and [ɪ], which are in complementary distribution. The allophone [i] has the widest distribution, and is more common, therefore was chosen to represent the phoneme.

62

Louise Baird

The allophone [ɪ] occurs word-initially (when not followed by a palatal approximant) and between consonants (when the preceding consonant is not /h/ and the following consonant is not the palatal approximant /j/). The allophone [i] occurs word-finally, as the second part in diphthongs (see section 2.3 below), following the glottal fricative /h/, before the palatal approximant /j/, and before a vowel-initial syllable. (7)

/i/ →

[ɪ] / # __ [-palatal] C [-/h/] __ C [-palatal] [i] / elsewhere

The high central vowel /ʊ/ is slightly forward of /o/ and /u/, less rounded, and occurs between them in height, that is, between close /u/ and close-mid /o/. The mid back vowel /o/ when realized as a short vowel is realized as the midclose vowel [o], but when lengthened becomes more open and is realized as [ɔː]. Some minimal and near-minimal pairs for /u/, /ʊ/, and /o/ are: (8)

/tarʊl/ ‘flat, smooth’

≠

/rol/ ‘mountain’

/taulʊl/ ‘to float’

≠

/loloŋ/ ‘to say’

/upʊi/ ‘to dry in the sun’

≠

/ʊpʊ/ ‘fruit’

/uhu/ ‘to drink’

≠

/oho/ ‘that’

Some minimal pairs and near minimal pairs based on vowel length are: (9)

/tut/ ‘beach’

≠

/tuːt/ ‘breast, milk’

/gamʊn/ ‘rotten’

≠

/mʊːn/ ‘snake’

/atoko/ ‘cassava’

≠

/tatɔːko/ ‘our stomach’

/hulaŋ/ ‘sky’

≠

/alaːŋ/ ‘clothing’

/tɛŋ/ ‘day’

≠

/tɛːŋ/ ‘our eyes’

/kikai/ [kɪkai] ‘male animal’

≠

/kiːka/ ‘red’

2.3 Phonotactics The distribution of Kafoa consonant phonemes within words is summarized in Table 3. There is only a single example of a consonant cluster within a syllable in the data: /maŋ.grik/ ‘feeling, emotion’. The velar nasal /ŋ/ only occurs in codas. The consonants that are typically found in codas are nasals or liquids, but there are a few instances of stops in final codas. However, there were no examples of /b/ and /g/ in this position.

63

Kafoa

Table 3: Distribution of consonant phonemes p

b

t

d

k

g

m

n

ŋ

f

s

h

r

l

w

j

Word-initial onset

+

+

+

+

+

+

+

+

–

+

+

+

+

+

+

+

Word-medial onset

+

+

+

+

+

+

+

+

–

+

+

+

+

+

+

+

Medial coda

–

–

–

–

–

–

+

+

+

–

–

–

+

+

–

–

Final coda

+

–

+

+

+

–

+

+

+

–

+

–

+

+

–

–

Kafoa vowels are frequently found in diphthongs consisting of rising sequences. Diphthongs typically occur word-finally, but there were a few found in the first syllables of disyllabic words. Examples are presented in Table 4. Kafoa speakers sometimes wish to orthographically represent the second element in diphthongs with the symbols used for approximants, namely and . Here diphthongs are written using vowel symbols. Table 4: Diphthongs /ai/ /ˈa.fɛˌnai/ /ˈlʊː.tai/ /kuˈpai/ /kaˈrai/ /baˈlai/

/ɛi/ ‘other’ ‘man’ ‘forest’ ‘eat’ ‘to hit’

/ʊi/ /uˈpʊi/ /ˈta.nʊi/ /fʊi/ /ˈkʊi/

/ˈa.tɛi/ /ˈna.tɛi/ /ˈka.bɛi/ /ˈgɛi.ja/ /maˈlɛi.pʊn/

‘tree’ ‘stand’ ‘few’ ‘burn’ ‘to spit’

/au/ ‘dry in sun’ ‘right’ ‘betel nut’ ‘to hunt’

/biˈnau/ /nau/ /ˈpa.nau/ /ˈtau.lʊl/

‘dirty’ ‘what?’ ‘dark’ ‘float’

2.4 Syllable structure Kafoa has both open and closed syllables, with a syllable minimally consisting of a vowel. Open syllables are more common than closed syllables. Heavy syllables are created when the nucleus contains a second vowel, creating either a long vowel or a diphthong, or when a syllable contains a coda. As noted above (in section 2.3) only a single instance of a consonant cluster was found in the data. Below, the possible syllable types and combinations are presented for one to four syllable mono-morphemic words.

64

Louise Baird

Nine syllable types were found in the data for mono-morphemic words consisting of one syllable.

(10)

Monosyllabic mono-morphemic words V /a/ ‘2SG ’ Vː

/iː/ ‘vomit’

VV

/ɛi/ ‘canoe’

VːC

/ɛːd/ ‘2SG ’

VVC

/aur/ ‘family’

CV

/mi/ ‘sit’

CVː

/taː/ ‘sleep, lie down’

CVV

/jau/ ‘2PL ’

CVːC

/nɛːd/ ‘1SG ’

Thirteen syllable types were found for mono-morphemic words containing two syllables in the data.

(11)

Two syllable mono-morphemic words V.CV /ˈa.ma/ ‘person’ Vː.CV

/ˈaː.ku/ ‘two’

V.CVː

/aˈjɛː/ ‘leaf’

V.CVV

/aˈfʊi/ ‘fish’

V.CVC

/aˈfɛŋ/ ‘village’

Vː.CVC

/ˈaː.kun/ ‘night’

CV.CV

/ˈba.la/ ‘wall’

CV:.CV

/ˈnaː.na/ ‘older sibling’

CV.CVV

/baˈlai/ ‘to hit’

CV.CVC

/ˈmɛ.hɛl/ ‘betel vine’

CV:.CVV

/ˈlʊː.tai/ ‘man’

CVC.CV

/ˈkal.ta/ ‘old person’

CVC.CCVC

/maŋˈgrik/ ‘feeling’

Thirteen syllable types were found in the data for mono-morphemic words containing three syllables.

Kafoa

(12)

65

Three syllable mono-morphemic words V.CV.CV /ˈa.bɛˌhɛ/ ‘young’ V.CV.CVV

/ˈa.fɛˌnai/ ‘other’

V.CVC.CVC

/ˈa.kɛŋˌfal/ ‘breathe’

V.CVC.CVV

/ˈa.rɛŋˌgau/ ‘threshold’

V.CV.CVC

/ˈa.raˌmaŋ/ ‘clan’

VC.CV.CV

/ɛlˈpa.ka/ ‘frog’

CV.CV.CV

/faˈla.ka/ ‘bright’

CV.CV.CVC

/ˈhi.lʊˌwaŋ/ ‘grandchild’

CV.CVC.CVC

/jɛˈnaŋ.mab/ ‘war’

CV.CV.VV

/ˈbi.luˌai/ ‘blue, green’

CV.V.CV

/hiˈo.pa/ ‘bone’

CVC.CV.CVC

/ˈkar.loˌhoŋ/ ‘hair’

CV.CVV.CVC

/maˈlɛi.pʊn/ ‘to spit’

Two syllable types were found for mono-morphemic words containing four syllables.

(13)

Four syllable mono-morphemic words CV.CV.CV.CV /ˈka.siˌbo.ta/ ‘shoes’ CV.CV.CV.CVV

/ˈhi.ja.kaˌtau/ ‘land’

2.5 Stress Kafoa has lexicalized stress. There is a common, but unpredictable, tendency for heavy syllables to bear stress. There is also a strong tendency for prefixes to bear stress. In unaffixed words there is a tendency for stress to appear on the first syllable. However, this is only a tendency, as stress is lexicalized in unaffixed words containing only light syllables, as evidenced by the data in Table 5. Table 5: Stress placement in words containing only light syllables Stressed 1st syllable /ˈʊ.pʊ/ /ˈmʊ.ra/ /ˈta.puˌhi/

Stressed 2nd syllable ‘fruit’ ‘afraid’ ‘narrow’

/iˈro/ /taˈki/ /aˈnaˌha/

‘egg’ ‘run’ ‘taro’

66

Louise Baird

There is a tendency for heavy syllables to bear primary stress in disyllabic words. However, as with light syllables these are tendencies only, as illustrated in Table 6, which shows that stress is also lexicalized in words containing heavy syllables. Table 6: Stress placement in words with heavy syllables Stressed heavy syllable /buˈrɛl/ /kaˈrai/ /iˈkoŋ/

Unstressed heavy syllable

‘worm’ ‘eat’ ‘louse’

/ˈlo.loŋ/ /ˈna.tɛi/ /ˈa.lum/

‘say’ ‘stand’ ‘cook’

In tri-syllabic words primary stress may fall on either the first syllable or second syllable.

(14)

Primary stress placement in tri-syllabic words /taˈbɛ.kɛl/ ‘fog’ /kaˈfa.ka/

‘spear’

/ˈma.hi.naŋ/

‘cold’

/ˈa.bɛ.hɛ/

‘young’

/ˈa.ja.ka/

‘rice’

/ˈa.to.ko/

‘cassava’

Prefixation can reassign primary and secondary stress. Primary stress typically falls on the possessive prefix on possessed nouns, and the person prefixes and valency increasing prefixes on verbs. For example: stress falls on the possessive prefix ta- ‘1PL . INCL . INAL ’ in /ˈta.puˌhai/ ‘our navels’ and /ˈta.taŋ/ ‘our arms’; and stress falls on the applicative prefix u- in /ˈu.laːm/ ‘to wash something’. If there is a heavy syllable in the first syllable of the root word then primary stress tends to fall on the heavy syllable. Once again this is a tendency only. For example, when the word /ˈwɛː.la/ ‘bathe’, which contains a heavy first syllable, is prefixed by a person prefix (in this case third person), the prefix takes primary stress, and the heavy syllable takes secondary stress, resulting in /ˈgaˌwɛː.la/ ‘wash him/her’.

2.6 Morphophonemics The two main morphophonemic processes in Kafoa are reduplication and compounding. There is also allomorphy in possessive prefixes.

Kafoa

67

2.6.1 Reduplication Nouns, adjectives, and verbs can all be reduplicated in Kafoa. Both full reduplication and partial reduplication occur. Reduplicants are prefixing, and have left to right copying. In most instances primary stress remains on the root word. In the examples in the corpus stress fell on the first syllable of the root word, except for verbs prefixed by the u- applicative, in which case primary stress fell on the applicative. All instances of reduplication found in the data are presented here. Nouns undergo full reduplication to indicate plurality. However, it is not obligatory to mark plural nouns in this way. It is highly likely that reduplication with this function has been borrowed from Alor Malay or Indonesian. Examples can be seen in (15). (15)

/aˈfɛŋ/

‘village’

/a.fɛŋ~ˈa.fɛŋ/

‘villages’

/ˈa.ra.maŋ/

‘clan’

/a.ra.maŋ~ˈa.ra.maŋ/

‘clans’

/kan/

‘child’

/kan~ˈkan/

‘children’

Adjectives can be reduplicated to either intensify their meaning or to derive adverbs. Contextual examples of this are presented in section 4.3. A process of full reduplication was used in the data for adjective to adverb derivation, as seen in (16). (16)

/ˈki.laŋ/

‘slow’

/ki.laŋ~ˈki.laŋ/

‘slowly’

/ˈbɛ.ka/

‘bad’

/bɛ.ka~ˈbɛ.ka/

‘badly’

The second function of adjective reduplication is to intensify the quality denoted by the adjective. Either full reduplication or partial reduplication is used to achieve this, as can be seen in (17). It is not clear what the motivation is for using both full and partial reduplication to perform this single function. In the case of full reduplication, the reduplicant is a phonetic copy of the root word. In the case of partial reduplication, the first syllable is reduplicated, preceding the root word, and the vowel in the reduplicant is reduced to schwa.

68 (17)

Louise Baird

/nu.ka/

‘good’

/nu.ka~ˈnu.ka/

‘very good, very beautiful’

/lɛ.ja/

‘big’

/lɛ~ˈlɛ.ja/

‘very big’

The corpus also contains examples of verbs either fully or partially reduplicated. Full reduplication of verbs is used to indicate durativity or iterativity, as in (18). (18)

/la.ka/

‘walk’

/la.ka~ˈla.ka/

‘walk for a long time’

/mi/

‘sit’

/mi~ˈmi/

‘sit a long time, stay, reside’

/mi.ga.ma/

‘dip’

/mi.ga.ma~ˈmi.ga.ma/

‘repeatedly dip’

As with the partial reduplication of adjectives, the reduplicant in instances of partial reduplication of verbs consists of the first syllable of the root, with the vowel reduced to schwa, as in (19). All of the instances in the data of partial reduplication involved root words containing a high-front vowel in the first syllable. Therefore, it isn’t yet clear whether the vowel reduction is specific to this particular vowel or a feature of partial reduplication more generally. The function of the partial reduplication of verbs and the impact on semantics also requires investigation. (19)

/ˈki.ra/ /ki~ˈki.ra/

‘play’ [kə~ˈkira]

/ˈu.hid/ /ˈu.hi~hid/

‘arrive’ [ˈuhə~hid]

‘arrive’ ‘hang’

/ˈu.lil/ /ˈu.li~lil/

‘play’

[ˈulə~lil]

‘hang’

Two additional reduplicated items were found in the data, presented in (20). (20)

/hama~hama/

‘together’, ‘middle’

/tara~tara/

‘each’

Kafoa

69

2.6.2 Compounding In the Kafoa corpus there are several examples of noun-noun compounding, where the first noun is a place name, and the second noun denotes a person or thing belonging to that place, as in (21). (21)

/ˈka.foa ˌkal.ta/

‘Kafoa old person’

/ˈmor ˌko.fa/

‘Moru wife’

/ˈmor iˌsui/

‘Moru cliff’

Other items were explained by speakers in semantic terms of being compounded words. They include both nouns and verbs, as in (22). (22)

/ˈi.rin-ˌmar/

‘circular dance’

/ˈdo.ma-ˌru.a/

‘to fight’

/ˈfi.al-kiˌli.kil/

‘starving’

/ˈso.jaŋ-ˌfa.jaŋ/

‘to dance’

/ˈwa.ka-ˌfo.ka/

‘body’

/jeˈnaŋ-mab/

‘war’

/ˈmi.tu.fuŋ-ˌmi.fa.riŋ/

‘many’

We know that these words are compounded, rather than instances of one noun modifying another, due to the stress placement. The first part of a compound contains primary stress, while the second part will be either unstressed, or take secondary stress, as marked in (22) above. The sequence of words ama ‘person’ followed by nuku ‘one’ is sometimes used as a compound, and at other times used as a numeral modifying a noun. Stress placement and semantics disambiguate the two possibilities. When the sequence ama nuku is an instance of a numeral modifying a noun, it means ‘one person’ and both ama ‘person’ and nuku ‘one’ take primary stress [ˈa.ma ˈnu.ku]. When the sequence ama nuku is an instance of compounding, it means ‘people’. Primary stress appears on ama only [ˈa.maˌnu.ku], and the stress on the first syllable of nuku ‘one’ is relegated to secondary stress. Kafoa speakers always wrote the compounds meaning ‘people’, that is amanuku and amanuka, as single words. I have followed their convention in order to avoid ambiguity between the compound reading and the possible analysis of noun + modifier. Speakers used hyphens between the components of some compounds and not others. Again I have followed their convention. If the two words that make up a

70

Louise Baird

compound frequently occur outside of the compound, then a hyphen is not used, for example, the noun-noun compounds in (21). Conversely if the words that make up a compound only occur in compounds, or rarely occur outside of a compound then a hyphen is used between them, as with the compounds in (22).

2.6.3 Prefixal allomorphy The inalienable possessive prefixes (see section 4.6) have two allomorphs each. This allomorphy is based on whether the nominal root to which the prefix attaches is vowel or consonant initial. As can be seen in Table 7, if the root is vowel initial then no vowel is used in the prefix. Otherwise the vowel /a/ is used in this prefix type. Second person singular, being unmarked, is impervious to the allomorphy. Table 7: Inalienable possessive prefix allomorphy

1SG 2SG 3 4 1PL . INCL 1PL . EXCL 2/3PL

Inalienable prefix

Allomorph /_V

Allomorph /_C

/na-/ /Ø-/ /ga-/ /da-/ /ta-/ /nija-/ /ija-/

[n-] [Ø-] [g-] [d-] [t-] [nij-] [ij-]

[na-] [Ø-] [ga-] [da-] [ta-] [nija-] [ija-]

3 Basic clausal syntax 3.1 Constituent order Intransitive clauses always have the structure of SV, that is, the single argument always precedes the intransitive verb. The single argument of an intransitive verb may be realized: (i) by a NP, as in (23); (ii) by a free pronoun, as in (24); (iii) by a bound pronoun, as in (25); (iv) by a combination of a NP and a free pronoun as in (26), or a NP and a bound pronoun, as in (27). (23)

Kan~kan

tut mi kira. beach be.at play ‘The children play at the beach.’ [ESS17.033a]

RDP ~child

Kafoa

(24)

Yayo weela. bathe 2PL ‘You bathe’ [ESS10.007]

(25)

Ama ga-boho ya do-taa do-mi=te. people 3-order go 3-sleep 3-stay=first ‘People ordered them to stay and sleep first.’ [KGA023]

(26)

Ama oho gi-kofa del ya. person that 3.ALIEN -woman 3DU go ‘That person goes with his wife.’ [ESS12.028]

(27)

Rabo luutai mang do-taa. finally man just 3-sleep ‘Finally the man he just slept.’ [KGA010]

71

The three constituent orders of APV, AVP and PAV can be found in transitive clauses. Examples of these three orders are presented in (28) and (29). A count of transitive clauses containing two nominal arguments was made for four of the texts. APV is overall the most frequent order, and based on this can be declared as basic.

(28)

(29)

A P V A V P Amanuku buhu karai mul, do mi nuku. people mushroom eat die 3 take one ‘People ate the mushrooms and died, (when) they took one.’ [KGW2:007] P A V Seng walanda ribu-kar-suai da ma. money Dutch thousand-ten-three 3 use ‘Thirty thousand of Dutch money he used.’ [KGW2:018]

As with the S argument in an intransitive clause, the A and P arguments in transitive clauses can also be expressed in multiple ways. The A argument can be realized by a NP, as with amanuku ‘people’ in (28) above; a free pronoun, as with third person da in (30) and (31); or a combination of NP and free pronoun, as with gi-kuy da ‘3.ALIEN -dog 3’ in (32). The P argument can be realized by a NP, as with awi ‘house’ in (30); a NP and a person prefix, as with kil g- ‘door 3-’ in (31) and niwel ga- ‘1SG . ALIEN -child 3-’ in (33); or just a person prefix.

72

Louise Baird

A P V (30) Da awi nati. 3 house build ‘He built a house.’ [ES10.002] A P P-V (31) Da kil g-aber. 3 door 3-open ‘He opened the door.’ [ES8.003] A P V (32) Gi-kuy da tukas de elpaka miya oho mi pukuy gahola. inside that be.at head move 3.ALIEN -dog 3 container REL frog ‘His dog it put its head into that container that the frog was inside.’ [KFS009] A P P-V (33) Na ni-wel ga-weela. 1SG 1SG . ALIEN -child 3-bathe ‘I bathe my child.’ [KS008]

3.2 Non-verbal predicates Equational clauses contain a non-verbal predicate, which can be nominal or adjectival, and a single argument, which precedes the predicate. In (34) the argument is ohong gom ‘the inside of the inside (referring to a cave)’, and the adjectival predicate is leya falaka ‘big (and) bright’. (34)

Ohong g-om leya falaka. inside 3.INAL-liver big bright ‘The inside of the inside (Lit. its liver4) was big and bright.’ [KGA007]

The expression ‘their name was. . .’ uses an equational clause, with g-anei ‘his/her name’ as the argument, and the person’s name as a nominal predicate, 4 In Kafoa the core meaning of the word -om refers to the physical organ ‘liver’. This sense is expanded upon so that -om can be used to also mean ‘inside’. With this sense it has cognates in other TAP languages (Antoinette Schapper pers. comm.). As can be seen from the example, there is another word ohong, also ‘inside’. The ‘liver’ as seat of emotions is expressed by another Kafoa word -hei.

Kafoa

73

as in (35). Extra identifying information about the person in question can be provided as a possessor preceding ganei ‘his/her name’, as in (36). (35)

[G-anei] [Karlowen Pai]. 3.INAL-name Karlowen Pai ‘His name was Karlowen Pai.’ [KGW2:006]

(36)

Kafoa kalta nuku g-anei Anderias Malailo. Kafoa old.person one 3.INAL -name Anderias Malailo ‘One old Kafoa person his name was Anderias Malailo.’ [KGW2:018]

3.3 Locative verbs Locative concepts commonly expressed through adpositional phrases in many languages are expressed through verbal clauses in Kafoa. A list of the locative verbs can be seen in Table 8.5 Table 8: Locative verbs Locative Verb

Meaning

miya mi tau itang hota weiu

‘stay, inside, be at, live’ ‘be at’ ‘be above’ ‘be below’ ‘to be (here)’ ‘to be below’

Sentential examples of some of the verbs can be seen in (37) to (40). (37)

Ama ge oho mai hiya miyal ir miya. person there that come descend banana tree be.at ‘That person there came down from being at the banana tree.’ [KGA005]

(38)

Da ya rol tau miyal maru. 3 go mountain be.above banana plant ‘He went to the top of the mountain to plant bananas.’ [KGA001]

5 The exact relationship between mi ‘take’, mi ‘be at’ and miya ‘stay, inside, be at’ is yet to be established. The form weiu ‘to be below’ appears to be derived from existential wei ‘exist, to be’, but further investigation is required.

74

Louise Baird

(39)

Rabo da kafurung ma pehe tau. finally 3 arrow use bow be.above ‘So he put an arrow on top of the bow.’ [KGA003b]

(40)

Bage afeng bal gal ipa gena oya afeng itang. but village there follow go until there village be.below ‘He went through the village until the lower part of the village.’ [KGA010]

The form mi is found in many of the Alor-Pantar languages, most often functioning as a locative verb or postposition, and in some cases is multi-functional. For example, in neighboring Klon the form mi is used as a comparative, as a verb in both mono-verbal clauses and serial verb constructions, as an applicative prefix and in temporal expressions (Baird 2010: 193). In Kafoa mi is primarily a verbal form, but is also used as an applicative (section 6.4). As a verb, mi can mean ‘take’ or it can have the locative meaning of ‘be at’.6 It typically occurs in serial verb constructions. The locative verb mi ‘be at’ is transitive, with the P argument expressing location. It occurs with either intransitive or transitive verbs in serial verb constructions (see section 7). Typically, when co-occurring with intransitive verbs, mi ‘be at’ and its P argument precede the intransitive verb, as in (41) and (42). The English translation is variable based on the semantics of other verbs that mi co-occurs with. Therefore, if a verb like hoka ‘come’ occurs with mi ‘be at’, as in (41), the English preposition ‘from’ is used to express the meaning, while if a verb like ya ‘go’ is used, as in (42), an English translation using the preposition ‘to’ may be more appropriate. These variable inferred meanings come from the deictic verbs, not the more neutral mi ‘be at’. (41)

Ama oho afeng mi hoka. person that village be.at come ‘That person comes from the village.’ [ESS11.012]

(42)

Pel ni-ayahang mi ya=te. 1DU. INCL 1SG . ALIEN -house be.at go=first ‘Let us go to my house first.’ [KGA005]

Mi ‘be at’ also occurs with transitive verbs, in which case mi and its P argument typically precede the other transitive verb and its P argument. This can be 6 Because mi is always preceded by a NP, it may appear to function as a postposition. However, there are examples in the data of mi taking person prefixes, as other verbs do, eg. ge-mi ‘to be there’.

Kafoa

75

seen in (43), which contains the transitive verb and its P argument pitey ukereyang ‘work our garden’, and (44), which contains the transitive verb and its P argument afói yap ‘search for fish’. (43)

Ama ya rol mi pi-tey u-kereyang. go mountain be.at 1PL . INCL . ALIEN -garden APPL-work 3PL ‘They went to the mountain to work our garden.’ [ESS17.032]

(44)

Niya hiya tut mi afói yap. 1PL . EXCL descend beach be.at fish search ‘We went down to the beach to search for fish.’ [ESS15.025]

There is one instance in the data where mi is used in a temporal expression, which is presented in (45). This example is unusual for two reasons. Firstly, in all other combinations of noun preceding mi location is expressed, rather than time. Secondly, the numeral one nuku occurs outside of the NP. (45)

Teng day

mi MI

nuku one

karal serrated.arrow

da 3

ya go

u-mahal APPL -guard

pehe, bow

kafurung arrow

lehe arrow

do-pón. 3-hold

‘One day he went to guard (them) with bow and various types of arrows.’ [KGA002]

3.4 Predicative possession Predicative possession is expressed using the transitive verb gowa ‘to have’.7 The A argument is the possessor and the P argument is the possessed item. Examples of this can be seen in (46) and (47). The clause in (46) is commonly used in the historical texts to indicate the passage of time. (46)

Wel kal gowa. child grandchild have ‘Children had grandchildren.’ [KGW1003]

7 It isn’t clear from the data whether the /g/ at the beginning of the verb is a pronominal prefix or not. There are no instances in the data of a different consonant at the beginning of the verb.

76 (47)

Louise Baird

Murai oho ihi gowa. lemon that fruit have ‘That lemon (tree) has fruit.’ [MR004a]

3.5 Negation Alor-Pantar languages typically have post-verbal negation. Kafoa is highly unusual in that it has preverbal negation. The negator ka immediately precedes the predicate that it negates. This can be seen in (48)–(50). (48)

Ne bilau na ka karai, yau yo 2PL EMPH 1SG food 1SG NEG eat ‘I won’t eat the food, you eat.’ [KGA009]

(49)

Wel child

kofa female

ung

ka

EMPH

NEG

ge there

wei exist

oho kasa that urinate

karai. eat

u-nuka APPL -good

kesa defecate

bisa. able

‘The woman there could urinate, (but) couldn’t defecate.’ [KGA016] (50)

Makong Makong nuka good

afeng village

gaten, make

Naugeng Naugeng

remaru plant

geia place

miya, live

awi house

ka

gowa. have

NEG

nuka good

gaten make

deeki bench

‘(They) lived in Makong village and Naugeng, good houses were made, good benches were made, no planting place was had.’ (Lit. planting place wasn’t had) [KGW2021] The negator ka can be used either in declarative utterances, as in (48)–(50), or as a prohibitive in imperative utterances, as in (51), in which case it still immediately precedes the verb it negates. (51)

Gi-el pa u-busara ka nang-karai! 3.ALIEN -brideprice 1PL . INCL APPL-discuss NEG 1SG -eat ‘Let us discuss her brideprice, don’t eat me!’ [KGA018b]

Although ka can be used in prohibitive utterances, Kafoa also has a separate prohibitive form dei ‘don’t’. There is only one use of it in the corpus, but there are multiple examples of its use in Widhyasmaramurti (2013). It is presented

77

Kafoa

in context in (52). Dei ‘don’t’ is repeated six times, in response to another participant’s words. It is used as an exclamation – it isn’t used to negate other constituents in an utterance, or to negate a whole utterance. In her section on grammar, Widhyasmaramurti (2013:6) provides two examples of sentences containing negation (c15 and c16). The first sentence contains prohibitive dei, the second example contains ka used preverbally, in the manner of the examples in (48)–(50). In the prohibitive example, dei occurs in the clause dei Indonesia lolong ‘don’t speak Indonesian’, with the constituent order of PROHIBITIVE P V, with no overtly expressed A argument. The example sentences for various dictionary entries in Widhyasmaramurti (2013) illustrate that the prohibitive dei can be used either preverbally or it can stand alone to be used as an exclamation. (52)

Karal serrated.arrow he CONJ

na 1SG

ama person hiya descend

pa 1PL . INCL

da 3

wai take

“Dei don’t

u-fan: APPL -say

demo so.that

mang all

ma use

pehe bow dei don’t

po 1PL . INCL

tau: be.above dei don’t

miyal banana

“Na 1SG dei don’t

ga-fiki”, 3-pull dei don’t

gi-el 3.INAL-price

dei, don’t

oho that

u-busara=t.” APPL -talk=first

‘He took a serrated arrow and put it on the bow: “I’ll pull it”, the person said “Don’t, don’t, don’t, don’t, don’t, don’t, I will come down so that we can just talk about the price of the bananas.”’ [KGA004] In the elicited word list kake was provided as the word for ‘no’, but it only occurs once in the corpus. It appears in its own clause, possibly as an exclamation, in (53). (53)

Gi-kasibota 3.ALIEN -shoe He

kake.

CONJ

NEG

da 3

go-pón 3-hold

bodan then

migawai, return

elpaka frog

gom inside

miya? be.at

‘He held his shoes then turned them upside down, was the frog inside? No.’ [KFS008] There is one instance of the form ka co-occurring with the form ke, on either side of a noun, in the idiomatic expression ka nan ke ‘it’s nothing’, in (54).

78

Louise Baird

(54)

Ka

nan ke. thing NEG ‘It’s nothing.’ / ‘Don’t worry.’ [KGA011] NEG

3.6 Interrogatives There are seven interrogative words in Kafoa, which can be seen in Table 9. Table 9: Interrogative words Interrogative

Function ‘what’ ‘who’ ‘where’ ‘when’ ‘how much/many’ ‘how’ ‘why’ ‘which’

nau ino yatu irentei ireng tówatai nadu tedo

used to question non-animate entities used to question animate entities used to question location used to question time used to question quantity used to question method used to question purpose or reason used to question animate and inanimate entities

Interrogative words/phrases occur in the place of the constituent being questioned, and cannot be moved around a clause in ways that the equivalent declarative constituent could not be moved. This is evidenced by the ungrammaticality of (55b) and (56b), but the grammaticality of the possessive example (57a–b). (55)

a.

Hi kikai yatu miya? chicken male.animal where stay ‘Where is the rooster?’ [ES14.001]

b. *Yatu miya hi kikai? where stay chicken male.animal *‘Where is the rooster?’ [ES14.002] (56)

a.

Hi kikai tedo? chicken male.animal which ‘Which rooster?’ [ES14.003]

b. *Tedo hi kikai? which chicken male.animal *‘Which rooster?’ [ES14.004]

Kafoa

(57)

a.

Hi kikai kika leya aaku oho ino chicken male.animal red big two that who ‘Those two big red roosters are whose?’ [ES12.011]

b.

Ino gi hi kikai who 3.ALIEN chicken male.animal ‘Whose rooster is it?’ [ES12.012]

79

gi? 3.ALIEN

do-hota? 3-to.be

Only a single polar question was found in the data, which is presented in (58). It has a declarative form, with rising intonation on the first syllable of lehe ‘arrow’, which falls again on the second syllable. The response consists of a short clause, using the verb in the question. (58)

a.

Del u-fan: “Niyi lehe nuka a 3DU APPL-say 1PL . EXCL . ALIEN arrow good 2SG ‘Those two said: “Do you know of our good arrow?”

b.

Kalta kofa: “Na gen”. old woman 1SG know ‘Old woman: “I know”.’ [MR003]

gen?” know

3.7 Imperatives Syntactically, imperatives appear to have the same form as declaratives. An illustration of this is that second person pronouns can be used in imperatives, as with yel ‘you two’ in (60). The verbal enclitic =t(e) is often found in imperative utterances, as in (59). Its use is not obligatory in all imperatives, as can be seen by its absence in (60). Both examples (59) and (60) contain unexpected structural features, which do not appear to be linked to their function as imperatives. Example (59) contains a very interesting use of the free pronoun ne ‘me’ (see section 5.5), where it occurs as the left-most constituent with the semantic role of beneficiary. Example (60) contains an instrumental SVC (see section 7) containing the verb ma ‘use’. Such SVCs typically have the structure P1 ma P2 (other) V. Therefore we would expect the order awai ilaka ma uwa ‘lime sign use rub’. However, in this example we find what appear to be two P arguments adjacent to one another followed by two verbs. There are two possible explanations for this: firstly ilaka ‘sign’ may actually be a verb and part of the SVC; secondly, if awai ‘lime’ and ilaka ‘sign’ are both nominal, in this context they may both be conceived of as arguments of both ma ‘use’ and uwa ‘rub’. While it is not possible

80

Louise Baird

to resolve this based on the recording and only a single occurrence of the word ilaka ‘sign’ in the data, it is clear from the recording that awai ma ilaka uwa ‘rub lime to make a sign’ meets the criteria for being a SVC. (59)

Ne kil g-aber pas=te. 1SG door 3-open later=first ‘Please open the door for me.’ [KGA010]

(60)

Yel mira murei wei, oho awai ilaka ma uwa. 2DU ascend lemon exist this lime sign use rub ‘You two climb the lemon tree there, use and rub lime to make a sign.’ [MR004b]

The verbal clitic =t(e), as seen in (59), often appears in utterances with illocutionary force that could be interpreted as being imperative, for example, requests and suggestions. It is possibly used to soften requests for politeness reasons. This usage has parallels in nearby languages. In Kui (see Windschuttel and Shiohara this volume) the verbal enclitic =tê is also used in this way. In Makalero, spoken in East Timor, the clitic =teʹe is described as a clause linker with the gloss ‘after’. It is also said to often be “. . . used as a de facto imperative marker, in which case it stands at the end of an utterance. . .” (Huber 2011: 253). Sometimes the verbal clitic =t(e) occurs just as =t, as in (61), in which it is cliticized to the verb busara ‘talk’. Otherwise the clitic takes the form =te, as with the verb bana ‘carry’ in (62). These two forms were always translated by Kafoa speakers using Malay dulu, roughly meaning ‘first’. It is unclear what motivates the difference in form between =t and =te. (61) Na hiya demo po miyal gi-el oho pa 1SG descend so.that 1PL . INCL banana 3.ALIEN -price that 1PL . INCL mang u-busara=t. APPL -talk=first just ‘I will descend so that we talk about the bananas’ price, we just talk.’ (Said to placate someone threatening to shoot him.) [KGA004] (62) Neheng woi hoko nuku o bana=te ate tuka woi mortar here this one 2SG carry=first wood short here garub hoko nuku o bana=te. many.people.sitting this one 2SG carry=first ‘You carry this one mortar here first, this one short piece of wood where many people are sitting here, you carry first.’ [KGA012]

Kafoa

81

4 Noun phrases 4.1 Noun phrase structure The structure of Kafoa noun phrases is presented in (63). (63) [(PSR-)N (N) (REL) (ADJ)* (NUM) (DEM)]NP The elicited noun phrase in (64) contains multiple modifiers, which illustrate the order presented in (63). The first noun hi ‘chicken’ is the head noun, with everything that follows it being post-modifiers.

(64)

N N ADJ ADJ NUM hi kikai kika leya aaku chicken male.animal red big two ‘those two big red roosters’ [ES12.010]

DEM oho that

4.2 Relative clauses Relative clauses introduced by the relativizer de follow the nominal head. There are only three examples in the data, all presented below. The NP containing the relative clause is bracketed, while the relative clause itself is bolded. In (65) the A argument kofa ‘woman’ is modified by the relative clause de kalta ‘that was an old person’ and the numeral nuku ‘one’. (65)

[Kofa de kalta nuku] hoka balai go-pón. come eldest 3-hold woman REL old.person one ‘One woman that was an old person came and held the eldest son.’ [MR006]

In (66), the noun phrase that occurs as the P argument of the locative verb mi ‘be at’ (section 3.3), contains the head noun tukas ‘container’ modified by the relative clause de elpaka miya ‘that the frog was inside’, followed by a demonstrative. (66) Gi-kuy da [tukas de elpaka miya oho] mi pukuy gahola. inside that be.at head move 3.ALIEN -dog 3 container REL frog ‘His dog it put its head into that container that the frog was inside.’ [KFS009]

82

Louise Baird

In (67) the S argument amanuka ‘people’ is modified by a relative clause containing the clause wei ye ohong miya ‘to be living inside’. (67) Ge [amanuka de wei ye ohong miya] nang mul paratai. REL exist EMPH inside live PL die complete thus people ‘Thus the people who were living inside they all died.’ [KGA024]

4.3 Adjectives There are four pieces of evidence to suggest that Kafoa has adjectives. Words with adjectival meaning behave in ways that prototypical members of the word classes of verbs and nouns do not. Firstly, words with adjectival meaning function in one of two positions. They either occur attributively within NPs as post-modifiers, modifying the head noun, as in (68), or alternatively they can occur predicatively, without affixation, in equational clauses, as seen in section 3.2. Notably verbs cannot occur in these positions. (68)

Bui Himor taki mai, [Woimahang afeng kiding] mimi. Bui Himor run come Woimahang village small live ‘Bui Himor ran away and lived in the small hamlet of Woimahang.’ [KGW2:002]

Secondly, words with adjectival meaning can be intensified by reduplication in attributive position, as in (69) (see section 2.6.1). (69)

[Kofa nuka~nuka nuku hoka. one come woman RDP ~good ‘One really beautiful (Lit. good) woman came.’ [MR006]

Thirdly, words with adjectival meaning can be used adverbially when reduplicated and used post-verbally, as in (70). See section 2.6.1. (70)

Gi-nana mira beka~beka. 3.ALIEN -older.sibling ascend RDP ~bad ‘His older brother ascended badly.’ [MR005]

Fourthly, words with adjectival meaning are only used verbally when prefixed by u- or mi- (see section 6.4), as in (71).

Kafoa

83

(71) Wel kofa ge wei oho kasa u-nuka kesa ung ka bisa. child woman 3 exist that urinate APPL-good defecate EMPH NEG able ‘That young woman there did good urination, (but) couldn’t defecate.’ [KGA016]

4.4 Numerals Numerals follow a base-ten pattern, as can be seen from Table 10. Numerals denoting multiples of ten begin with kar ‘ten’, for example kar-suai, literally ‘ten three’, is ‘thirty’. Numerals higher than ten begin with wal ‘plus’, except for multiples of ten, a hundred, a thousand and a million. This includes numerals denoting a value in the teens, or within a higher number such as ‘three’ within ‘forty three’, which is expressed as kar-butai wal-suai, literally, ‘ten-four plusthree’. The large numbers of thousand and million are borrowed from Malay – ribu for ‘thousand’ and juta for ‘million’. Table 10: Numerals 1 2 3 4 5 6 7 8 9 10

nuku aaku suai butai iweheng talaama bisau tidaroko tikainuku kar-nuku

11 12 13 14 15 16 17 18 19 20

wal-nuku wal-yaaku wal-suai wal-butai wal-iweheng wal-talaama wal-bisau wal-tidaroko wal-tikainuku kar-yaaku

21 30 43 100 1000 1946 1961 1973 1,000,000

kar-yaaku wal-nuku kar-suai kar-butai wal-suai asaga ribu-nuku ribu-nuku asaga-tikainuku kar-butai wal-talaama ribu-nuku asaga-tikainuku kar-talaama wal-nuku ribu-nuku asaga-tikainuku kar-bisau wal-suai juta-nuku

As shown in the template for noun phrase structure in (63) in section 4.1, nouns can be directly modified by a numeral. That is, classifiers are not necessary for counting, as can be seen in (72). (72)

Gena lai hila butai oho ama butai pei natei. until finished rope four that people four near stand ‘Until finished, four people stood near those four ropes.’ [KGA024] Numerals can be qualified, as seen in (73).

(73)

Din asaga galehel mote wei mar. amount.of.people hundred more new exist ascend ‘There were more than one hundred people before they ascended.’ [KGA022]

84

Louise Baird

4.5 Non-numeral quantifiers It was not possible to identify a united class of non-numeral quantifiers from the limited data. Six items were found in texts with non-numeral quantifier-type meanings, but with varied functions, as can be seen in Table 11, and from the examples that follow. There were only two instances in the corpus of any of these items modifying a noun within a NP: mang ‘all’, and mitufung-mifaring ‘many’. In all other occurrences, the three items used to modify nouns (rowang ‘every’, mitufung-mifaring ‘many’ and mang ‘all’) floated off the nouns they were modifying and appeared clause-finally. Table 11: Items with quantifier-type meanings Function falupu tówang galehel rowang mitufung-mifaring mang

‘all’ ‘many’ ‘more than’ ‘every’ ‘many’ ‘all’

nominal function nominal function modifies numerals modifies nouns modifies nouns modifies nouns

In (74) falupu ‘all’ is used as a nominal argument. (74)

Falupu go-nuku Hifolat mi, afeng gaten. all 3-one Habolat be.at village make ‘All became one in Habolat, making a village.’ [KFW2:021]

From (75) it can be seen that tówang ‘many’ is treated as a noun, occurring grammatically as an alienably possessed item (see section 4.6 on nominal possession), and galehel ‘more’ is used to modify the numeral that precedes it, as also seen in example (73) above. (75)

Amanuku gi-tówang, ribu-nuku galehel. people 3.ALIEN -many thousand-one more ‘Many people, more than a thousand.’ [KGW1:010]

In (76) rowang ‘every’ is modifying the A argument, but floats off it, occurring clause-finally. (76)

Aramang~aramang

gi-awi kabang 3.ALIEN -house sacred ‘Every clan has a sacred house.’ [KGW3:002]

RDP ~clan

gowa have

rowang. every

85

Kafoa

In (77) mitufung-mifaring ‘many’ is modifying the P arguments wel ‘child’ and kal ‘grandchild’, but instead of immediately following a noun, as in (78), it floats off the nouns and occurs clause-finally. The form mi on the front of the words mitufung-mifaring, together with its typically clause-final position, makes these words look suspiciously verbal (see section 6.4). (77)

Bapa grandparent kal grandchild

Ba Ba gowa have

Soro Ba Soro Ba

del 3DU

Muna Muna

Selly Selly

miya, live

wel child

gowa have

mitufung-mifaring. many

‘Ancestors Ba and Soro Ba they lived in Muna Selly, (they) had very many children and grandchildren.’ [KGW2:001] (78) Amanuka mitufung-mifaring, oho do isung 1962 niya sakola people many that 3SG year 1962 1PL . EXCL school go-fakal. 3-open ‘So very many people, so, in 1962 we opened a school.’ [KGW2:025] In (79) mang ‘all’ follows the noun gi-amanuka ‘their people’ that it modifies. Given the similarity in form to periphrastic plural forms in other Alor-Pantar languages, mang could possibly be marking plurality (see Klamer, Schapper and Corbett 2014). (79) Aramang kar-nuku wal-yaaku gi-amanuka mang niy-om 1PL . EXCL . INAL-liver clan ten-one plus-two 3.ALIEN -people all nuku, niya-tooko, kirkir nuku, manggrik nuku, one 1PL . EXCL . INAL-stomach think one feeling one huhung nuku lolong nuku. talk one say one ‘All the people of the twelve clans have one liver, one stomach, one mind, one feeling, and speak one language.’ [KGW3:007]

4.6 Possession Two possessive classes – inalienable and alienable – can be distinguished based on the formal distinction that is made in the marking of possession.

86

Louise Baird

Possession can be expressed either nominally, as described in this section, or predicatively using the verb gowa ‘to have’, which is described in section 3.4. Possessive prefixes mark the number and usually the person (although there are exceptions, see below) of the possessor on the possessed noun, as in (80). They can co-occur with a NP denoting the possessor, as in (81), in which case the word order is always POSSESSOR POSSESSED. (80)

Pel ni-ayahang mi ya=t. 1DU. INCL 1SG . ALIEN -home be.at go=first ‘(Let) us two go to my home first.’ [KGA005]

(81)

kui gi-amur dog 3.ALIEN -hair ‘the dog’s hair’ [ES13.003]

Two possessive pronominal paradigms can be constructed: inalienable possessive prefixes and alienable possessive prefixes. These are presented in Table 12. The two sets of prefixes have a different distribution, namely the inalienable possessive prefixes are obligatory with one sub-group of nouns, while the alienable possessive prefixes are optionally marked on another sub-group of nouns. The inalienable nouns that obligatorily take a possessive prefix include most body parts, and the words -mau ‘voice’ and -anei ‘name’. Notably, kin terms are not obligatorily prefixed. Those nouns that are optionally prefixed (including kin terms) are alienable, and make up the bulk of nouns in Kafoa. A noun can only take one type of prefix, that is, it cannot be realized as inalienable in some situations, and alienable in others. The alienable pronominal forms all contain the vowel /i/. While prefixes in the alienable paradigm have the same form regardless of the phonological shape of the following noun, the forms of the inalienable prefixes depend on the initial sound of the host noun. If the noun is consonant-initial the vowel /a/ forms a part of the prefix. If the noun is vowel-initial then the vowel /a/ isn’t uttered. The second and third person plural forms are conflated, that is the same forms (yi- / iy(a)-) are used for both second and third person plural. These forms also occur as a part of the first person exclusive form, with the form ni (niyi- / niy(a)-), but not the first person inclusive form. No form for second person singular was attested for the inalienable possessive paradigm.

87

Kafoa

Table 12: Possessive Prefixes

1SG 2SG 3 4 1PL . INCL 1PL . EXCL 2/3PL

Alienable

Inalienable

niigidipini(yi)yi-

n(a)Øg(a)d(a)t(a)niy(a)iy(a)-

As with person prefixes (see section 6.2), there are forms for third and fourth person possessive prefixes. There are six instances of nouns that are possessed by fourth person possessive prefixes di-, d- or da-. All of the nouns that are possessed using the di- prefix denote relationships between people (di-kofa ‘his wife’, di-adól ‘his master’, and di-ka ‘his younger sibling’). This prefix patterns with the other alienable possessive prefixes in both form and the items to which it attaches. The d- and da- prefixes attach to obligatorily possessed body parts (d-awa ‘her mouth’, d-it ‘her bottom’ and da-tang ‘his arm’), and pattern with the other inalienable possessive prefixes. The functional difference between the third/fourth person distinction remains a topic for future research. In the textual data the inalienably possessed noun -tang ‘arm’ occurs with both the ga- prefix, as in (82), and the da- prefix, as in (83), confirming that there is a difference between the pronouns, and that the presence of the two forms is not due to noun classes. (82)

Makil big.male.crab

tau be.above

u-li~lil APPL- RDP ~hang

hoko this

ga-tang 3.INAL-arm natei stand

u-li~lil, APPL - RDP ~hang

ga-tang 3.INAL-arm

karia-tanei cry

‘A big male crab was on top hanging from her arm, hanging from her arm this way (she) stood and cried.’ [KGA019b] (83)

Kan child ma use

luutai male dada chin

nuku one

da 3

bangku bench

taomi sit

da-tang 4.INAL-arm

aaku two

mi take

tukei. stick

‘A boy he sat on a bench, using his two arms to prop up his chin.’ [KFS001] Paradigms for individual nouns illustrating the possessive prefixes can be seen in Table 13, containing alienable nouns, and Table 14 containing inalienable nouns.

88

Louise Baird

Table 13: Alienable nouns with possessive prefixes

1SG 2SG 3SG 1PL . INCL 1PL . EXCL 2/3PL

yai ‘mother’

wel ‘child’

ni-yai i-yai gi-yai pi-yai ni-yai yi-yai

ni-wel i-wel gi-wel pi-wel niyi-wel yi-wel

Table 14: Inalienable nouns with possessive prefixes

1SG 2SG 3SG 1PL . INCL 1PL . EXCL 2/3PL

-eeng ‘eye’

-tooko ‘stomach’

n-eeng Ø-eeng g-eeng t-eeng niy-eeng iy-eeng

na-tooko Ø-tooko ga-tooko ta-tooko niya-tooko iya-tooko

5 Pronouns 5.1 Pronoun paradigms It is possible to identify five paradigms of independent pronouns (Table 15). Gaps in paradigms for singular and plural forms in the eed, a, o and e series are due to lack of data, while the el series consists solely of dual forms. The independent pronouns are used for either A or S arguments, typically (but not exclusively) having the semantic role of actor. Table 15: Independent pronouns

1SG 2SG 3SG 1PL . INCL 1PL . EXCL 2PL 3PL

eed series

a series

o series

e series

el series†

need eed deed peed

na a da pa niya yau ama

no o do po niyo yayo

ne

– – – pel niyel yel del

yeed yeed

ge pi yage

†These pronouns are used to refer to sets of two, as a dual number would. However, where bound person prefixes coindex the referent of an el series pronoun, it is plural not singular number. See Table 16. Hence, here el pronouns are listed under the plural number.

Kafoa

89

There is overlap in forms for the free pronouns and bound person prefixes, see Table 17. Only one full paradigm (the eed series) has forms that are not also found bound to verbs. While the paradigms have different forms in some person number combinations, in others they are identical. I treat the forms as grammatically separate items for two reasons: (i) one form is bound and the other free; and; (ii) there is a difference in the semantic role based on whether the pronoun is bound or free. Namely, free pronouns typically denote actors, experiencers or locatives, while bound pronouns can denote a range of semantic roles including patient, theme, experiencer, recipient and goal (see section 6.1).

5.2 eed pronouns The eed series were the forms cited when the word list was elicited. There are only four textual examples of their use. At present there is no explanation for why they were picked for the word list over the other possibilities. They are not the most commonly occurring type of free pronoun. There are only two textual examples of third person deed, and one each of first person singular need and second person singular eed. Three of the four examples were followed by demonstratives, and appear to have a locative meaning. They are presented in (84) to (87). (84)

Ama uyala, uyala bo deed midawai ya. want want CONJ 3 return go 3PL ‘They agreed, agreed so he went back.’ [KGA019a]

(85)

Need hoko na imou this 1SG die 1SG ‘Me here, I will die.’ [KGA010]

(86)

Eed 2SG

hoko this

niya 1PL . EXCL

owe EMPH

niyi-aur 1PL . EXCL . ALIEN -family

takau, steal

wer now

hoko this

ang-karai. 2SG -eat

‘You here stole our family (member), now we eat you.’ [KGA018a] (87)

Deed oho kua hoka. 3 that yesterday come ‘He came yesterday.’ [KS006]

90

Louise Baird

Based on form, this set may be cognate with Kamang reflexive pronouns (Schapper 2014: 248–249), and based on form and function these forms may be cognate with the Abui paradigm in -do, which Kratochvíl (pers. comm.) is currently labeling ‘focus’ pronouns.

5.3 a pronouns The a series pronouns are the most commonly used series in the data, with 71 textual examples. All of the referents of the a series of free pronouns are semantically actors, as illustrated by (88) to (90). (88)

Da ayaka lum karai. 3 uncooked.rice cook eat ‘She cooks and eats rice.’ [KS010]

(89)

Niya hiya tut mi afói yap. 1PL . EXCL descend beach be.at fish search ‘We went down to the beach and looked for fish.’ [KS025]

(90)

E oho na hiya na mang a-fuud. 2SG -swallow hey that 1SG descend 1SG only ‘Hey I’ll come down and I’ll swallow you.’ [KGA003b]

The form ama was alternatively translated by speakers as a noun meaning ‘person, people’ or the third person plural pronoun ‘they’. The status of this item as a noun, pronoun, or both is currently unresolved. An example that illustrates the possibility of either interpretation can be seen in (91). The utterances prior to (91) spoke of two brothers. It is not clear whether in (91) ama refers to just these two brothers or to the wider community. Clauses following (91) do not help to disambiguate. If this utterance refers to just these two brothers, then ama is interpretable as the third person plural pronoun; if it refers to the wider community it could be interpretable as being the noun ‘people’. (91)

Gena pin aakun, ama irin. until time night 3PL /people circular.dance ‘At nighttime they/people danced.’ [MR006]

5.4 o pronouns The o series of pronouns was used when the imperative paradigm for the verb bana ‘carry’ was elicited. There are also 11 instances of their use in the textual

91

Kafoa

data, in which case they occur as semantic actors and experiencers. Four of these textual examples can be seen in (92)–(95). (92)

Yayo weela! bathe 2PL ‘You bathe!’ [KS007]

(93)

Hoko mang towatai de no hiya=te this only how so.that 1SG descend=first ‘How about this so I get down (from) here.’ [KGA010]

(94)

Móón do tang-war tang-balai. RECP-coil RECP-surround snake 3 ‘The snakes they coiled around each other and surrounded each other.’ [KGA010]

(95)

Niyo afói yaka. 1PL . EXCL fish search ‘We searched for fish.’ [ESS025]

hoko. this

5.5 e pronouns There are 19 instances of e series pronouns in the textual data. A single semantic label has not been established yet for this series, but the pronouns in this series frequently appear to have a locative sense, as with third person ge in (96), or are associated with location, as with second person plural yage in (97). (96)

Bo del kofa aaku ge oho gang mi. 3 that marry take then 3DU woman two ‘So those two married the two women there.’ [MR007]

(97)

Wer now

bisou seven

mar=te ascend=first

ya go

wei, exist

pa 1PL . INCL

ni-afeng 1SG . ALIEN -village

gede go

Inta Inta

yage 2PL

mi be.at

u-busara. APPL -talk

‘Seven days from now, go to my village Inta, you come up first, we’ll talk.’ [KGA018b]

92

Louise Baird

5.6 el pronouns The el series of pronouns is a set of independent dual pronouns. The initial consonants found on the forms are the same as those found on the other plural person and number combinations, and each of the dual forms ends in -el. The dual forms can occur on their own, following NPs, or in combination with other pronominal forms. Typically, the first use of a dual pronoun immediately follows nominal expression of its referents, and thereafter is used to refer to those referents without the need to continue to express them fully in a NP. An example of this can be seen in (98). (98) [Kan luutai uh-aaku] del ya do afói yaka, del afói leya child man CLF-two 3DU go 3 fish search 3DU fish big nuku ga-bat. one 3-shoot ‘Two men, those two went searching for fish, those two shot one big fish.’ [MR001] There are 14 instances of dual pronouns in the Kafoa textual data. In all cases the pronouns had the semantic role of actor. An example of the first person inclusive dual pronoun pel can be seen in (99). In two of the occurrences a dual form co-occurs with another pronoun. These can be seen in examples (100) and (101). In example (100) the first person inclusive dual pronoun pel is co-indexed by the first person inclusive free pronoun pi. In (101) the third person dual pronoun del is co-indexed by the third person verbal prefix do-. (99)

Pel ni-ayahang mi ya=t. 1DU. INCL 1SG . ALIEN -home be.at go=first ‘We two go to my home first.’ [KGA005]

(100)

nan ke, laka pel pi NEG thing NEG walk 1DU. INCL 1PL . INCL ‘It’s nothing, walk, us two will go.’ [KGA011]

(101)

Lai bo aakun del ge mi do-taa. be.at 3-sleep finish CONJ night 3DU 3 ‘Finished then, at night, those two they slept there.’ [KGA008]

Ka

ya. go

In elicited verb paradigms, speakers used the dual form alone for A arguments, but always combined it with a person prefix for P arguments. This is illustrated for the verb balai ‘hit’ in Table 16 and with a sentential example in (102).

Kafoa

93

Table 16: Pronominal paradigm for balai ‘hit’

1SG 2SG 3SG 1PL . INCL 1PL . EXCL 2PL 3PL 1DU . INCL 1DU . EXCL 2DU 3DU

(102)

A argument

P argument

na a da pa niya yani ama pel niyel yel del

namamgamtamyamyamama pel tamniyel yamyel yamdel gam-

Del pel tam-balai. 3DU 1DU. INCL 1PL . INCL-hit ‘Those two hit us two.’

The dual pronouns can also co-occur with possessive pronouns, as illustrated by example (80) in section 4.6 above, reproduced here in (103). (103)

Pel ni-ayahang mi ya=t. 1DU. INCL 1SG . ALIEN -home be.at go=first ‘(Let) us two go to my home first.’ [KGA005]

Based on examples (99)–(103), it seems that Kafoa dual forms can co-occur with all other types of pronouns, that is, free argument pronouns, bound person prefixes and possessive pronominal prefixes. It appears that by default the semantic role of dual pronouns is that of actor, but they can take on the semantic role of other pronominal or person prefix forms if they co-occur with them. This is also the case in neighboring Klon. That is, the same dual form is used regardless of the syntactic role of any other pronominal form that it co-occurs with. In combinations such as these, the dual form and other pronominal form provide complementary information: the dual form provides information about number absent in the other pronominal form, and the other pronominal form provides information about the syntactic role of the referents (i.e. Actor or Undergoer), lacking in the dual form (Baird 2008: 76). In Teiwa, spoken on Pantar, dual forms are only used as subject pronouns (Klamer 2010: 82). In Kamang (north-central Alor) dual pronouns may be co-indexed by additional plural pronominal forms (Schapper 2014: 249). In the example Schapper provides, the

94

Louise Baird

free dual pronoun is co-indexed by an agreement prefix. In Wersing, Schapper & Hendery (2014: 365) provide an example of a dual pronoun co-occurring with an agentive pronoun.

6 Person prefixes and valence adjustment 6.1 Person prefixes Kafoa has five paradigms of person prefixes. These forms are regarded as being pronominal rather than agreement markers, in line with Corbett’s discussion of the distinction (Corbett 2006: 100–102). Arguments may be expressed in one of three ways: (i) a person prefix may be the sole referential expression of a discourse participant, in the same manner as free pronouns; (ii) a NP may be the sole referential expression of a discourse participant, or; (iii) a discourse participant may be expressed by both a NP and a person prefix, in which case the person prefixes are used to cross-reference nominal arguments on verbs, agreeing in number and person. Table 17: Person prefixes

1SG 2SG 3 4 1PL . INCL 1PL . EXCL 2PL

am series

ang series

a series

o series

e series

namamgam-

nanganggangtangniyangyang-

noogodotoniyoyo-

neege-

tamyamyam-

naagadataniyaya-

tenye-

The am series of person prefixes were elicited for the verb balai ‘to hit’. They were used for the P argument, with a semantic role of patient, as can be seen in (104). They do not occur anywhere else in the data. Note that the form yam- was the one provided for first person non-singular exclusive, and first person dual exclusive (see Table 16 above). It does not fit the typical pattern for first person non-singular exclusive pronouns, and the form niyam- would be expected. This anomaly needs investigation. (104)

Da am-balai. 3 2SG -hit ‘He hit you.’

Kafoa

95

The ang series of person prefixes were elicited for the verb bana ‘carry’. They were used for the P argument with the semantic role of theme, as with gang- in (105). There are no textual examples of their use. (105)

Niya gang-bana. 1PL . EXCL 3-carry ‘We carried him/her/it.’

The e series of person prefixes were elicited for the verb huhung ‘tell’. They were used for the P argument with the semantic role of recipient or goal, as in (106). There are no textual examples of their use. (106)

Yani u-ne-huhung. Yani APPL-1SG -tell ‘Yani told me.’ [ES003]

In contrast to the other person prefixes the a series and o series of person prefixes occurred in textual data, and not in any elicited verb paradigms. The prefixes in the a series always had the semantic role of patient, as in (107). They were used to mark either P or S arguments. The prefixes in the o series had the semantic role of theme or experiencer. They were also used to mark either S arguments, as in (108), or P arguments, as in (109). (107) Del afói leya nuku ga-bat. 3DU fish big one 3-shoot ‘Those two shot one big fish.’ [MR001] (108) Rabo ama ya deki pakil, do-taa do-mi. finally person go platform climb 3-lie 3-sit ‘Finally the people climbed onto the platform, they lay, they sat.’ [KGA024] (109) Rabo usep, wel luutai ge oho neheng tuka leya do-bana. finally true child man 3 that mortar short big 3-carry ‘Finally it was true, the young man there carried the big, short mortar.’ [KGA013]

96

Louise Baird

6.2 The third / fourth person distinction There is currently insufficient data to draw conclusions about the difference between the third person and fourth person forms in Kafoa. From the textual data it is possible to see that the single verb pón ‘hold, catch’ can take different person prefixes according to context. This is presumably the case with other verbs. In example (110) the referent of do- in do-pón to ‘hold/use it’ is a kind of instrument or theme. The semantic role of the referent of go-pón ‘hold him’ in example (111) is debatable, while in example (112) the referent of ga- on ga-pón ‘to hold, catch’ is some kind of patient. (110) Teng mi nuku da ya uhumal, pehe, kafurung, lehe, day be.at one 3 go guard, bow arrow arrow karal, do-pón. serrated.arrow 3-hold ‘One day he went to guard (the bananas), holding a bow, arrows and serrated arrows.’ [KGA002] (111) Kofa nuka-nuka nuku hoka di-ka go-pón. woman good one come 3.ALIEN -younger.sibling 3-hold ‘One fine woman came and held the younger sibling.’ [MR006] (112) Wel luutai da kula gal, ehe tafói da ga-pón. child male 3 river follow shrimp crab 3 3-catch ‘The young man he followed the river, he caught shrimp and crabs.’ [KGA017] Presumably the difference between the third and fourth person forms is one that is applicable to both person prefixes on verbs and possessive prefixes. In the data there are six instances of nouns that are possessed by fourth person possessive prefixes, that is, di-, d- or da- (see section 4.6). In Adang and Abui there are two forms for third person prefixes. Haan (2001: 149) labels the two forms as obviative and proximal. Fedden et al. (2013: 39) label them alpha and beta. For Western Pantar Holton (2014) labels these two types as third and fourth person. For consistency within these volumes I adopt Holton’s terminology. The forms in Adang are quite different to those in Abui and Kafoa. “In Adang, the α-type prefix indexes the subject of a small number of verbs which can be interpreted as reflexives where the implicit object is coreferential with the subject. The β-type indexes objects only” (Fedden et al.

Kafoa

97

2013: 39). The third person forms begin with the consonant [s], while the fourth person forms begin with a glottal stop. In Western Pantar fourth person free pronouns and person prefixes are used with a reflexive function to indicate coreference between two core arguments, and also function as switch reference markers indicating that a referent is not co-referential with the tracked referent (Holton 2014). Notably one of the forms in Kafoa has the same form as that found in Abui, that is, dV-. This alternation in forms also appears in the possessive pronominal paradigms of the two languages (see Table 12 for the Kafoa forms). In Kafoa the third person forms begin with the consonant /g/ and the fourth person forms begin with the consonant /d/. In Abui the equivalent third person forms begin with the consonant /h/ and, as in Kafoa, the equivalent fourth person forms begin with the consonant /d/. Kratochvíl (2011: 609) identifies the difference in Abui as due to instigation. “The dV- prefixes refer to affected participants that are themselves the primary cause of the event, i.e. responsible for the onset of an event.” He continues, saying that: “The hV- prefixes refer to affected participants that are not responsible for the onset of the event.” (Kratochvíl 2011: 610).

6.3 Reciprocal There is a single example in the word list and only two examples in the corpus containing a first person plural inclusive person prefix tang- used in a distributive way to indicate a reciprocal. In both textual examples it occurs in SVCs preceding each of the verbs in the serial complex. These can be seen in (113) and (114). The form in the word list is tang-gena ‘fight each other’. (113)

Móón do tang-war tang-balai. RECP-coil RECP-surround snake 3 ‘The snakes they coiled around each other and surrounded each other.’ [KGA010]

(114)

Liliki nanamul tang-fatain tang-kokop. RECP-murder RECP-cut war evil ‘There was a very bad war with (people) murdering each other and cutting each other.’ [KGW2007]

This form is clearly cognate with forms in surrounding languages. In Abui there is a separate series of distributional prefixes that are used with a reciprocal sense, these are te-, to- and ta- (Kratochvíl 2007: 78 & 186). In Klon the first

98

Louise Baird

person plural inclusive Undergoer pronominal prefixes t-, to-, tin- and te- are used to mark reciprocals (Baird 2008: 106). In Teiwa there is also a dedicated distributive pronoun taʹan used to refer to plural, non-collective human referents and which functions as a reciprocal (Klamer 2010: 184).

6.4 Applicative prefixes u- and miKafoa has two applicative prefixes u- and mi-. The applicative prefix u- occurs on 27 out of 206 verbs in the data, and on adjectives to derive intransitive verbs, as in (71) in section 4.3. When the prefix occurs on verbs it introduces a P argument into the clause. In (115) the verb kareyang ‘to work’ is used intransitively, without an object, and therefore, the verb is unaffixed. By contrast, in (116) it is used transitively, taking the argument pi-tey ‘our garden’, and so it takes the u- prefix. (115)

Ama ya niyi-tey niya go 1PL . EXCL . ALIEN -garden 1PL . EXCL 3PL ‘They went to our garden we worked.’ [ES032]

(116)

Ama ya rol mi pi-tey u-kareyang. go mountain be.at 1PL . INCL . ALIEN -garden APPL-work 3PL ‘They went to the mountain to work our garden.’ [ES032/016]

kareyang. work

In example (117) the verb busara ‘talk’ occurs with the u- prefix indicating the topic to be discussed, while in example (118) the verb busara ‘talk’ is bare, not specifying the topic under discussion. (117) Na hiya demo po miyal gi-el oho 1SG descend so.that 1PL . INCL banana 3.ALIEN -price that pa mang u-busara=t. 1PL . INCL only APPL -talk=first ‘I will descend so that we just talk about the bananas’ price.’ [KGA004] (118) Rabo, ama ge oho mai hiya, miyal ir miya, del finally person 3 that come descend banana tree stay 3DU busara bage he ama ge da u-fan: . . . talk but SIM person 3 3 APPL -say: ‘Finally, the person there came down from the banana tree, those two talked, but that person said: . . .’ [KGA005]

Kafoa

99

The applicative prefix mi- only introduces arguments with the semantic role of instrument. In the following examples the u- prefix and mi- prefix are compared. Note that person prefixes can co-occur with both u- and mi-. The valenceincreasing prefix occurs first, followed by the person prefix followed by the root (u-/mi-PRON-V). The word leke means ‘clear’. When prefixed it comes to mean ‘count’, as in (119) and (120), in which leke ‘clear’ has been prefixed by both a person prefix and a valency increasing prefix. The difference in choice of prefix is based on the semantic role of the argument that is introduced. In (119) the applicative prefix mi- is used to introduce the instrument argument seng ‘money’. In (120) the u- prefix is used to introduce the P argument ‘bananas’ which is marked on the verb using the third person prefix ga-, but not expressed in a NP. (119)

Nan thing

kar-yaaku ten-two

wal-nuku plus-one

wei exist

hoko, this

seng money

mi-ga-leke

he,

APPL -3-count

SIM

juta-kar-suai. million-ten-three ‘These twenty one items counted with money equal thirty million (rupiah).’ [AKK023] (120)

Da ya u-ga-leke, ama takau lai. . . 3 go APPL-3-count person steal finish ‘He went to count them (bananas), someone had stolen them all. . .’ [KGA002]

7 Serial verb constructions Serial verb constructions (SVCs) contain two or more verbs and their arguments which are used to express a single, typically complex, conceptual event. The verbs share at least one argument, and come under a single intonation contour. The following two sentences illustrate how similar concepts can be expressed either in a SVC, as in sentence (121), or two clauses as in (122). In (121) all of the verbs share the first person plural pronoun niya as an argument. In (122), there are two separate clauses. The first clause contains the verbs hiya ‘descend’ and mi ‘be at’ being used in a SVC, but yaka ‘search’ in the second clause does not share any of the same arguments found in the first clause, and hence is not a part of any SVC.

100

Louise Baird

(121)

Niya hiya tut mi afói yap. 1PL . EXCL descend beach be.at fish search ‘We went down to the beach to search for fish.’ [ESS025]

(122)

Niya hiya tama mi, niyo afói yaka. be.at 1PL . EXCL fish search 1PL . EXCL descend sea ‘We went down to the sea, we searched for fish.’ [ESS025]

The section of narrative in (123) shows a combination of co-referential argument deletion and SVCs. Falling intonation indicates where the clause boundaries are, and hence which verbs belong to a single clause. (Full stops are used in this example to illustrate the clause boundaries.) Using intonation as a cue, it is possible to identify five clauses. Clause (123a) contains a common type of SVC consisting of a motion verb ya ‘go’ and locative verb tau ‘be above’. Clause (123b) contains a single verb maru ‘to plant’, for which the A argument has been co-referentially deleted. Clause (123c) contains a SVC consisting of three verbs; intransitive mai ‘come’, transitive gowa ‘have’ and the transitive locative verb tau ‘be above’, and the P argument of gowa, ihi ‘fruit’ and the P argument of tau ‘be above’, fuma ‘flower’. Clause (123d) contains the iconically ordered SVC ya ugaleke ‘go count’. The third person prefix ga- in u-ga-leke ‘to count’ refers to the topic, that is, the bananas. In clause (123e) the verb lai ‘finish’ is used in conjunction with takau ‘steal’ to form a completive SVC. The highly topical P argument of ‘steal’ (bananas) has been co-referentially deleted. (123)

a.

Da ya rol tau. go mountain be.above 3 ‘He went to the mountain.’

b.

Miyal maru. banana plant ‘(He) planted bananas.’

c.

Miyal mai ihi gowa fuma tau. banana come fruit have flower be.above ‘The banana grew to have fruit and flowers on top.’

d.

Da ya u-ga-leke. 3 go APPL-3-count ‘He went to count them.’

e.

Ama takau lai. person steal finish ‘Someone had stolen (them) all.’ [KGA001–002]

101

Kafoa

Various combinations of intransitive and transitive verbs are found in SVCs. Transitive verbs in a SVC frequently share both their arguments, as in (124), in which ayaka ‘rice’ is the P argument of both transitive verbs lum ‘cook’ and karai ‘eat’. In such cases, verbs are contiguous. (124)

Ama oho ayaka lum karai. person that rice cook eat ‘That person cooked and ate rice.’ [ESS005]

It is also common for transitive verbs in SVCs to have different P arguments. The P arguments of verbs immediately precede the verb, and so it is common for transitive verbs within a SVC to be non-contiguous. The SVC in (125) consists of three verbs. All verbs share the A argument ama oho da ‘that person, he’, the two transitive verbs mi ‘take’ and ma ‘move’ contiguous to each other share the P argument seng ‘money’, while the P argument nimang ‘my father’ of the third transitive verb gena ‘give’ precedes the verb, and hence occurs between the verbs. (125)

Ama oho da seng mi ma ni-mang person that 3 money take move 1SG . ALIEN -father ‘That person, he gave money to my father.’ [ESS001]

gena. give

As mentioned, locative verbs and motion verbs frequently co-occur within SVCs. Locative verbs are transitive, while (most) motion verbs are intransitive. The order of the two types of verbs is variable. This means it is possible for an intransitive motion verb to be followed by a transitive locative verb and its P argument, as in (126). It is also equally possible for the intransitive motion verb to follow the transitive locative verb and its P argument, as in (127) and (128). (126)

Ama ya rol mi pi-tey u-kareyang. go mountain be.at 1PL . INCL . ALIEN -garden APPL-work 3PL ‘They go to the mountain to work our garden.’ [ESS032]

(127)

Kan~kan tut mi ki~kira taki. RDP ~child beach be.at RDP ~play run ‘Children run and play at the beach.’ [ESS033b]

(128)

Ama oho afeng mi hoka. person that village be.at come ‘That person came from the village.’ [ESS012]

102

Louise Baird

Iconicity plays a role in the ordering of verbs. In the manner SVC in (129) the idea of ‘going walking’ occurs before the ‘searching’ can take place. In this construction the intransitive verb depun ‘walk’ is used to describe the manner in which the ‘going’ expressed by laka ‘go’ is done. Likewise in (130) the subevent of ‘eating’ precedes the sub-event of ‘dying’. (129)

Ni-yai da depun laka ara yaka. walk go wood search 1SG . ALIEN -mother 3 ‘My mother went walking searching for wood.’ [ESS018]

(130)

Amanuku buhu karai mul. people mushroom eat die ‘People ate mushrooms and died.’ [KGW2007]

Serial verb constructions can be used to increase the number of arguments in a clause, as is done with instrumental serialization. The verbs pón ‘to use, to hold’ and ma ‘use’ can be used this way, within a SVC to introduce an instrument argument. In (131) pedang ‘machete’ is an instrument introduced as the P argument of pón ‘to hold’ in the SVC containing the verbs pón ‘to use, to hold’ and tokon ‘to cut’. In (132), the instrument argument seng ribu-kar-suai ‘thirty thousand rupiah’ is introduced as the P argument of the verb ma ‘use’ in the SVC containing the verbs ma ‘use’ and el ‘buy’. (131)

Ama nuku oho da pedang pón ate person one that 3 machete hold wood ‘Someone cut wood with a machete.’ [KS003]

(132)

Kalta old.person

nuku one

ribu-kar-suai thousand-ten-three

g-anei 3.INAL-name da 3

ma use

Anderias Malailo Anderias Malailo

lahatal God

tokon. cut

seng money

el. buy

‘One old man named Anderias Malailo he used thirty thousand rupiah to buy God.’ [KGW1005]

8 Aspect marking There are several adverb-type words that are used in texts that were translated with aspectual meanings. There is often only a single occurrence of such words in the data, making analysis of them untenable at this point in time. However, it

Kafoa

103

has been possible to establish the role of two aspectual adverbs: yai (section 8.1) and dara (section 8.2). In addition to aspectual adverbs, it may be the case that alternations in verb stems play a role in expressing aspect, as discussed in section 8.3.

8.1 Completive yai Although the word yai does not have an entry of its own in Widhyasmaramurti’s (2013) short dictionary, it does appear in seven example sentences for other entries. Based on these sentences, together with one example from my corpus, the aspectual adverb yai can be identified as being completive. It is used to indicate that the activity, achievement or accomplishment in a clause is complete. It always follows the verb, occurring in clause-final position, as can be seen in (133–135). (133)

Elpaka do tera yai. frog 3SG jump COMPL ‘The frog had jumped away.’ [KFS006]

(134)

Ni-mang do-nuka yai. 1SG . ALIEN -father 3-good COMPL ‘My father is all better.’ (Widhyasmaramurti 2013: 22)

(135)

Ni-bapa luutai imon yai. 1SG . ALIEN -grandparent man die COMPL ‘My grandfather has died.’ (Widhyasmaramurti 2013:39)

In her short dictionary Widhyasmaramurti (2013: 50) has an entry for the word leyai, which she translates with Indonesian habis, selesai, sudah ‘completed, finished, already’. Based on meaning and form, this word clearly appears related to the completive adverb yai, but what the relationship is between these two words remains to be established.

8.2 Imperfective dara The imperfective aspectual adverb dara is concerned with the ongoing internal temporal situation of an event. It indicates that a situation is incomplete. The inception of the event is implied, with no mention or implication concerning the endpoint. It can be typically translated into English with ‘still’, as in (136)

104

Louise Baird

and (137). Dara can be followed by the negative ka, and this combination of dara ka is translatable as ‘not yet’, as in example (138). When dara occurs with the negative, there is no implication concerning the inception or endpoint of the situation. (136) Gi-wakafoka g-e nuku petang miya g-e nuku dara 3.ALIEN -body 3.INAL-leg one outside be.at 3.INAL-leg one IPFV do tukas miya. 3 container be.at ‘Its body, its one leg is out, (and) one leg is still in the container.’ [KFS004] (137) Jadi yahot gi-bukti tafa wer dara hota. so like 3.ALIEN -evidence metal.drum present IPFV to.be ‘So the evidence is (that) the metal drum still exists.’ [KGA025] (138) ama luutai nuku dara ka kofa gowa person man one IPFV NEG woman have ‘one man (who) didn’t yet have a wife’ [KGA001]

8.3 Stem alternations In addition to aspectual adverbs, it appears the morphological shape of a verb may also play a role in the expression of aspect. In neighboring Abui verb stems show similar alternations depending on aspect (Kratochvíl 2007: 82–83). At this stage it is unclear whether the alternations in the Kafoa data are verbal suffixes, or verb stem alternations. There are two verbs that appear with two different forms in the textual data: fakang / fakal meaning ‘to open’; and yap / yaka ‘to search’, as in examples (121) and (122) above, repeated here as examples (139) and (140). If the ‘search’ words yap and yaka are in a paradigmatic relationship, then other verbs containing the form ya may also be involved, such as ya ‘go’, hiya ‘descend’, dóya ‘go home’ and miya ‘stay, inside, be at, live’. (139) Niya hiya tut mi afói yap. 1PL . EXCL descend beach be.at fish search ‘We went down to the beach to search for fish.’ [ESS025] (140) Niya hiya tama mi, niyo afói yaka. be.at 1PL . EXCL fish search 1PL . EXCL descend sea ‘We went down to the sea, we searched for fish.’ [ESS025]

Kafoa

105

Both occurrences of fakang and fakal ‘to open’ in the data occurred with the third person prefix go- attached to them, which co-indexed the NP sakola ‘school’, as can be seen in (141)–(142). The reason for these alternations is unclear. The alternation could simply be synonyms with similar forms; the alternation could relate to TAM categories; or based on these single examples of fakang / fakal ‘to open’, it could even be a voice distinction. (141) Isung ribu-nuku asaga-tikai nuku kar-talaama wal-nuku year thousand-one hundred-nine one ten-six plus-one sakola go-fakang. school 3-open ‘In 1961 a school was opened.’ [KGW1008] (142) Isung ribu-nuku asaga-tikai nuku kar-talaama wal-yaaku year thousand-one hundred-nine one ten-six plus-two niya sakola go-fakal. 1PL . EXCL school 3-open ‘In 1962 we opened a school. [KGW2:025]

9 Discussion Based on a small dataset, this overview of Kafoa is aptly named a sketch. It has only been possible here to present an outline of Kafoa, with many areas in need of further data and investigation. Kafoa shares some grammatical and lexical similarities with other AlorPantar languages, especially Klon and Abui, and of course also contains its own unique features. Further research needs to be undertaken to ascertain Kafoa’s genetic affiliation, and tease apart which features are the result of intense contact with other languages, and which features are inherited. The phonology of Kafoa, and the size and constituency of its consonant and vowel inventory, is typical of western Alor languages. The basic constituent order in Kafoa is SV / APV and is found in all Timor-Alor-Pantar languages. The possibility of multiple constituent orders for transitive constructions, namely APV, AVP and PAV, is also common amongst the languages of the region. Further research is required to identify the motivating factors for each order. However, highly unusual – perhaps even unique – amongst Alor-Pantar languages, Kafoa has preverbal negation, like Eastern Timor languages, whereas other Alor-Pantar languages have postverbal negation. Due to the paucity of textual examples of negation, the exact distribution and status of the three apparent negators

106

Louise Baird

(preverbal negator ka, prohibitive dei, and the third negator kake) needs further investigation. Typical of Alor-Pantar languages, Kafoa has multiple free pronoun and personprefix paradigms. Much more textual and elicited data is needed to unravel the intricacies of these paradigms, including their distribution in denoting A, S and P arguments. From the scant data available it appears that there is a split system whereby A arguments are expressed using free pronouns, P arguments are expressed by person prefixes, while some S arguments are expressed by free pronouns and others are expressed by person prefixes. One notable feature of person prefixes and possessive prefixes in Kafoa is the presence of the distinction between third and fourth person. This distinction is characteristic of a subset of Alor-Pantar languages including Adang, Abui and West Pantar. Kafoa has two applicative prefixes u- and mi-. Both of these forms are found with similar applicative functions in nearby languages. Applicative u- is also found in Adang (Robinson & Haan 2014: 211) and Klon (Baird 2008: 95). Applicative mi- is also found in Kamang (Schapper 2014: 257), Wersing (Schapper & Hendery 2014: 372) and Klon (Baird 2008: 101). Other applicative forms in nearby languages include -r in Western Pantar (Holton 2014: 69), ma-, li-, and wii- in Sawila (Kratochvíl 2014: 309), and in addition to mi- Wersing has the applicative le- (Schapper & Hendery 2014: 372). While languages may share one applicative in common, it is notable that Kafoa’s valency increasing morphology appears to be almost identical to that found in neighboring Klon. It would be interesting to learn whether the distribution of the applicative prefixes is identical or different in the neighboring languages, which could in turn help understand the nature of the relationship between Klon and Kafoa. A final key area that requires investigation is the possibility that Kafoa verbs have stem alternations. Based on the scant data, it appears that verb stem alternations may be utilized for the expression of aspect, as in neighboring Abui (Kratochvíl 2007). More textual data is required to determine the productivity, distribution and semantics involved. (143) Ya geuwai. ‘That’s finished.’

Acknowledgments I would like to thank the people of Habolat who always provided a warm welcome and enthusiasm for teaching me, with special thanks to Mr. Ibrahim Lapaibel and Mr. Karel Mohar. Thanks also to Antoinette Schapper, Marian Klamer, František Kratochvíl and two anonymous reviewers for providing valuable feedback on

Kafoa

107

drafts of this chapter. Thanks also to Widhyasmaramurti for discussions on the sociolinguistic setting of Kafoa. Funding of the research reported here was provided by the Netherlands Foundation for Scientific Research (NWO), through the Vernieuwingsimpuls VIDI-project “Language variation in eastern Indonesia” (2002–2007), and through a fieldwork grant provided by the Endangered Languages Documentation Programme, SOAS.

References Baird, Louise (collector). 2003. Kafoa Corpus (LRB1). Digital collection managed by PARADISEC. DOI: 10.4225/72/56E97A0265669 Baird, Louise. 2008. A grammar of Klon: A non-Austronesian language of Alor, Indonesia. Canberra: Pacific Linguistics. Baird, Louise. 2010. Grammaticalisation of asymmetrical serial verb constructions in Klon. In Michael Ewing & Marian Klamer (eds.), Contributions from East Nusantara: Typological and areal analyses, 185–202. Canberra: Pacific Linguistics. Corbett, Greville. 2006. Agreement. Cambridge: Cambridge University Press. Fedden, Sebastian, Dunstan Brown, Greville Corbett, Gary Holton, Marian Klamer, Laura C. Robinson & Antoinette Schapper. 2013. Conditions on pronominal marking in the AlorPantar languages. Linguistics 51: 33–74. Haan, John. 2001. The Grammar of Adang: A Papuan language spoken in the island of Alor East Nusa Tenggara – Indonesia. PhD thesis, University of Sydney. Holton, Gary. 2014. Western Pantar. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 23–96. Berlin: De Gruyter Mouton. Huber, Juliette. 2011. A grammar of Makalero: A Papuan language of East Timor. PhD thesis, Leiden University. Humaedi, M. Ali, Abdu Rachman Patji, Sihol Farida Tambunan, Sudiyono, Anastasia Melati & Widhayasmaramurti. 2013. Mekanisme internal pelestarian bahasa dan budaya Kafoa. [Internal preservation mechanisms for the Kafoa language and culture.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Humaedi, M. Ali, Abdul Rachman Patji, Sihol Farida Tambunan, Sudiyono, Anastasia Melati & Widhyasmaramurti. 2014. Strategi pelestarian bahasa dan budaya Kafoa: system sosial budaya dan kebijakan. [Preservation strategies for the Kafoa language and culture: Social system, culture and wisdom.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan. Klamer, Marian. 2010. A Grammar of Teiwa. Berlin: De Gruyter Mouton. Klamer, Marian, Antoinette Schapper & Greville Corbett. 2014. Plural number words in the AlorPantar languages. In Marian Klamer (ed.), Alor-Pantar Languages: History and Typology, 375–412. Berlin: Language Science Press. Kratochvíl, František. 2007. A grammar of Abui: A Papuan language of Alor. PhD thesis, Leiden University. Kratochvíl, František. 2011. Transitivity in Abui. Studies in Language 35: 588–635. Kratochvíl, František. 2014. Sawila. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 351–438. Berlin: De Gruyter Mouton.

108

Louise Baird

Lewis, M. Paul, Gary F. Simons & Charles D. Fennig (eds). 2014. Ethnologue: Languages of the World, Seventeenth edition. Dallas, Texas: SIL International. Online version: . Robinson, Laura C. & John Haan. 2014. Adang. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 221–284. Berlin: De Gruyter Mouton. Santosa, Puji. 2012. Kearifan budaya dan fungsi kemasyarakatan dalam sastra lisan Kafoa. [Cultural wisdom and social function within Kafoa oral literature.] Meta Sastra 5:67–82. Schapper, Antoinette. 2014. Kamang. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 285–350. Berlin: De Gruyter Mouton. Schapper, Antoinette & Rachel Hendery. 2014. Wersing. In Antoinette Schapper (ed.), Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 439–504. Berlin: De Gruyter Mouton. Simpadat PMB LIPI. 2014. Kafoa Untuk Selamanya (Kafoa Gena Wai Legaya). Video. Viewed 11 July 2014. . Widhyasmaramurti. 2013. Kamus kecil bahasa Kafoa. [A short dictionary of Kafoa.] Jakarta: Lembaga Ilmu Pengetahuan Indonesia, Pusat Penelitian Kemasyarakatan dan Kebudayaan.

Glenn Windschuttel & Asako Shiohara

3 Kui 1

The language scene

2 2.1 2.1.1 2.1.2 2.2 2.3 2.4 2.5 2.5.1 2.5.2 2.5.3 2.5.4 2.5.5 2.5.6

112 Phonology 112 Consonants 113 Minimal pairs for consonant phonemes 115 Consonant allophony 115 Vowels 118 Stress 120 Phonotactics and syllable structure 122 Morphophonemics 122 Allomorphy of agreement prefixes 123 Blending 124 Compounds and monophthongization 124 Metathesis 125 Voicing assimilation 126 Reduplication

3 3.1 3.1.1 3.1.2 3.1.3 3.1.4 3.2 3.3 3.4

127 Basic clausal syntax 127 Predicate types 128 Verbal predicates 130 Nominal predicates 131 Elevational predicates Predications expressing location and existence 135 Negation 136 Questions 138 Imperatives

4 4.1 4.2 4.2.1 4.2.2 4.3 4.4

139 Noun phrases Attributes and relative clauses 141 Quantification 141 Numerals 143 Plural marker 144 Demonstratives 144 Possession

DOI 10.1515/9781614519027-003

111

140

133

110

Glenn Windschuttel & Asako Shiohara

5 5.1 5.2 5.3 5.4 5.5

Pronouns 146 148 Basic pronouns 149 Subject pronouns 149 Possessive pronouns 150 Focus pronouns 151 Quantificational pronouns

6 6.1 6.1.1 6.1.2 6.1.3 6.2 6.2.1 6.2.2

6.3 6.3.1 6.3.2 6.3.3 6.4 6.5

152 Agreement, applicative and other prefixes 153 Agreement 154 Agreement on intransitive verbs 155 Agreement on transitive verbs 157 Agreement on ditransitive verbs 158 ‘Derivational’ uses of agreement prefixes 158 Intransitive uses of obligatorily prefixed transitive verbs Intransitive verbs taking valency-increasing patientive agreement 159 prefixes Intransitive verbs taking valency-increasing dative agreement 160 prefixes Intransitive verbs taking both forms of valency-increasing 161 agreement prefixes Transitive verbs taking valency-increasing agreement 162 prefixes 162 Applicative prefixes 164 u166 la166 mi168 Emphatic prefix 169 Noun incorporation

7 7.1 7.2 7.3 7.4 7.5 7.6

170 Serial verb constructions 170 Motion SVCs 172 Causative SVCs 172 Instrumental SVCs 174 Comitative SVCs 174 Aspectual SVCs 175 Resultative SVCs

8 8.1 8.2

175 Aspect marking Aspectual suffixes Aspectual enclitics

6.2.3 6.2.4 6.2.5

176 178

Kui

8.3 8.3.1 8.3.2 9

Aspectual particles 179 dé ‘already’ 180 lei ‘COMPL ’ Discussion

References

111

179

180

182

1 The language scene Kui (ISO 639-3 code: kvd), or in the language itself, Masin lak,1 is spoken in south-western Alor, traditionally centered on the village of Lerabaing. It is still spoken there and also now in Buraga, a neighboring village, and in Moru on the south side of Kalabahi Bay. Apart from Malay it is the only language in Lerabaing, while in Buraga, communities of Kui speakers are interspersed with speakers of Klon. The situation in Moru is more complex. Moru is a mixed community with Kui, Klon, Hamap and Abui quarters. The language spoken further east in Kiraman is recognized by some of its speakers as the same language, while many Kui speakers see it as a distinct, yet similar, language. While it is certain the two are closely related, the exact language/dialect status of Kui and Kiraman remains to be determined. Kui speakers are almost exclusively Muslim. Unlike many groups now living on the coast, it is their traditional home and they have lived there since first contact with the Dutch (Hägerdal 2010: 21). Today, the Kui are undergoing language shift, the native language being replaced by Alor Malay. This, exacerbated by the fact that Kui has always been a small language, has put Kui in serious danger of extinction. Katubi (2013) estimates the ethnic group to number 833 people; 399 in Moru and 315 in Buraga with only 119 still living in Lerabaing. However, the number of competent speakers is likely to be much smaller. According to Katubi (2011) only in Lerabaing is Kui still the main language even in the private sphere. The first author observed that there were still many children who did know a substantial quantity of Kui vocabulary in Buraga but many lacked the ability to confidently use more complex grammatical structures. Often they used calqued Indonesian structures, apparently being unable to generate the native patterns used by older speakers.

1 Speakers accept the use of the label ‘Kui’ when not speaking the language itself and generally refer to it by this name when speaking Malay.

112

Glenn Windschuttel & Asako Shiohara

Prior research consists of wordlists in published and unpublished sources (e.g. Stokhof 1975, Holton 2010) and initial investigations of the agreement system (Donohue 1997, Shiohara 2010), phonology (Akoli 2013a) and morphology (Akoli 2013b). Intensive ethnolinguistic research under the auspices of Indonesian Institute of Sciences (LIPI) is underway (Katubi 2011, 2012, 2013). As part of this a small Kui dictionary has been produced (Katubi, Thung & Akoli 2013). While these works have been consulted, this sketch is primarily based on fieldwork collected by the second author in August 2001 and 2005 in Moru and the first author in August 2012 and 2013 in Moru and Buraga. This data consisted of: (i) two narratives from Djumat Sanga and Kadir Sira Kasong (Shiohara fieldnotes); (ii) two narratives from Djumat Sanga, a 300 entry lexicon, 2000 elicited sentences and responses to four stimulus tasks produced by the Max Planck Institute for Psycholinguistics in Nijmegen by speakers, Djumat and Mas Sanga (Windschuttel fieldnotes).

2 Phonology 2.1 Consonants Kui has 18 consonant phonemes, as presented in Table 1. The orthography used for each phoneme is provided in angle brackets where it differs from the IPA. This convention is also followed in Table 3 for the vowels. Of these consonants, the palatals (with the exception of /j/) and the glottal stop are of marginal phonemic status and are therefore bracketed. Table 1: Consonant phonemes Bilabial Plosive

p

Alveolar b

Fricative Nasal

t

Palatal d

k

Glottal g

(ʔ)

s m

n

Affricate Approximant

Velar

w

( ɲ) ͡ (dʒ)

ŋ

j

Rhotic

r

Lateral

l

(ʎ) †

† Akoli (2013a: 4) disagrees with this characterization, suggesting rather that this is a lateral retroflex.

Kui

113

2.1.1 Minimal pairs for consonant phonemes The main plosive phonemes are /p b t d k g/. Minimal pairs that illustrate the phonemic voicing contrasts are given in (1). The last series also distinguishes these consonants from some of the nasals and other phonemes. (1)

/k/ ≠ /g/: /kur/ ‘dog’, /gur/ ‘male slave’ /t/ ≠ /d/: /bad/ ‘shirt’, /bat/ ‘coconut’ ͡ /tai/ ‘CMN ’, /ɲai/ ‘1PL . EXCL ’, /t/ ≠ /ɲ/ ≠ /n/≠ /m/ ≠ /b/ ≠ /p/ ≠ /j/≠ /dʒ/: ͡ ‘2/3PL ’ /nai/ ‘1SG ’, /mai/ ‘come’, /bai/ ‘go’, /pai/ ‘1PL . INCL ’, /jai/ ‘go’, /dʒai/

The minimal pairs in (2) are evidence of some manner contrasts in similar places of articulation. (2)

/w/ ≠ /m/:

/walεl/ ‘here’, /malεl/ ‘there’

/w/ ≠ /g/:

/wor/ ‘stone’, /gor/ ‘throw’

/t/ ≠ /s/:

/bata/ ‘big’, /basa/ ‘read’

/r/ ≠ /l/:

/sεr/ ‘salt’, /sεl/ ‘rope’

/d/ ≠ /r/:

/bad/ ‘shirt’, /bar/ ‘stumble’

There are no examples of minimal pairs for the velar nasal phoneme that distinguish it from the alveolar nasal. However, we still hold it to be an independent phoneme. This is because it does contrast with the alveolar nasal in similar environments. In the words given in Table 2 it contrasts with the alveolar nasal both intervocalically and word finally (it is unknown word initially). Table 2: Contrasting examples of /n/ and /ŋ/ /n/

/ŋ/

Intervocalic

/akana/ ‘black’

/naŋa/ ‘NEG ’

Word final

/gamin/ ‘headland’

/kusiŋ/ ‘nail’, /siliŋ/ ‘glasses’

Although we only have three instances of it, we conclude that the glottal stop is a distinct, albeit very marginal phoneme. The first example in (3) shows the meaningfulness of the glottal stop and the second a minimal pair with /k/.

114 (3)

Glenn Windschuttel & Asako Shiohara

/Ø/ ≠ /ʔ/:

/wo/ ‘yeah’, /woʔ/ ‘hello’ /ta/ ‘CMN ’, /taʔ/ ‘pole’2

/k/ ≠ /ʔ/:

/sok/ ‘be located, put into’, /soʔ/ ‘oar with a shaped handle’3

The palatal phonemes (with the exception of /j/) are only marginal. The only common use of the palatal nasal and palatal affricate is in pronominals. Representative examples are given in (4).4 The 2nd/3rd plural bound and free pronominals ͡ The nasal does appear elsewhere but contain the only known instances of /dʒ/. it is confined to the coda. The lateral is entirely restricted to appearing in the coda. The only examples of these phonemes known are given in (4). (4)

a.

/ɲ/: /ɲai/ ‘1PL . EXCL ’, /namaɲ/ ‘bird sp.’, /bakaɲ/ ‘oldest child’, /teɲ/ ‘sibling in-law’

b.

͡ /dʒai/ ͡ /dʒ/: ‘2/3PL ’

c.

/ʎ/: /akoʎ/ ‘fish sp.’5, /saʎ/ ‘deep canyon’, /paʎ/ ‘black stingray’6

Minimal pairs for the palatal nasal and lateral compared with their alveolar equivalents are given in (5).7 (5)

a.

/n/ ≠ /ɲ/: /namaɲ/ ‘bird sp.’, /naman/ ‘shellfish.’

b.

/ʎ/ ≠ /l/: /akoʎ/ ‘type of fish’, /akol/ ‘sugar palm’8

All of these phones occur as regular allophones of the corresponding alveolars. This is as a result of the process of palatalization that will be explained in section 2.1.2 being applied to a consonant preceding the perfective suffix /-i/. Some examples will be given in (6). 2 This item is from Akoli (2013a: 12). 3 This item is from Akoli (2013a: 12). 4 See Tables 6, 13, 14 and 15 for the other paradigms of pronominals which contain these phonemes. 5 This item is from Akoli (2013a: 5). 6 These items are from Akoli (2013a: 11). ͡ in the pronominals orginate in sequences of ni- and di- respectively. 7 The segments, /ɲ/ and /dʒ/, An i vowel grade marks plural on the pronouns in many languages of Alor-Pantar and n is a ͡ are clear marker of first person throughout Alor-Pantar (Holton et al. 2012). Cognates of /dʒ/ not so widely exemplified: the Abui cognate ri- (Kratochvíl 2007: 7) displays an initial r which regularly derives from initial d’s (Holton et al. 2012: 93). 8 This item is from Akoli (2013a: 5).

Kui

115

2.1.2 Consonant allophony The alveolars, /n/, /d/, /l/ and /s/, can be palatalized before [i] as described below in (6). This is most commonly practiced by older speakers, in particular, before the perfective suffix /-i/ on verbs. This interacts with optional final high vowel deletion so that often the only evidence of the perfective suffix is final palatalization. This will described further in section 2.2. (6)

/s/ > [s] ~ [ ʃ ] / _i

/mainsino/ [majnʃɪno] ~ [majnsɪno] ‘well, so’

/l/ > [l] ~ [ʎ] / _i

/akaːli/ [akaːʎi] ~ [akaːli] ‘eat’

/n/ > [n] ~ [ɲ] / _i ͡ / _i /d/ > [d] ~ [dʒ]

/abani/ [abaɲi] ~ [abani] ‘wait’ ͡ ~ [podi] ‘climb’ /podi/ [podʒi]

Intervocalic plosive phonemes often surface as fricatives. Whether there are any exceptions among the plosives is as yet unknown. This is set out below in (7a) with examples. In particular phrases such as in (7b) this process has been known to take place across word boundaries (see section 2.5.5 for an explanation of the voicing of /t/). (7)

a.

b.

/b/ > [b] ~ [β] / V_V

/abaisi/ [aβajʃi] ~ [abajʃi] ‘well, so’

/g/ > [g] ~ [ɣ] / V_V

/ogo/ [oɣo] ~ [ogo] ‘this’

/k/ > [k] ~ [x] / V_V

/akan/ [axan] ~ [akan] ‘black’

[a

βodo=dε]

/a

bodo=tε/

2SG . SBJ

stupid=AND

‘You’re stupid!’

2.2 Vowels In our data there is evidence for six vowel phonemes. These are the five cardinal vowels and a mid-front vowel /ε/. All vowels except /e/ are known to display a length contrast which is orthographically represented by doubling the graph. The phonemes are displayed in Table 3. While the high mid front vowel /ε/ is far more common than its lower counterpart /e/, speakers preferred the diacritic on /ε/ for the language’s practical orthography (possibly due to influence from Indonesian).

116

Glenn Windschuttel & Asako Shiohara

Table 3: Vowel phonemes Front High Mid Low

Central

i iː e ɛ ɛː

Back u uː o oː

a aː

Minimal pairs and sets illustrating the contrastiveness of the six vowel phonemes are given in (8). (8)

/a/ ≠ /u/: /ar/ ‘fire’, /ur/ ‘moon’ /a/ ≠ /o/ ≠ /e/: /tan/ ‘sea’, /ton/ ‘side dish’, /t-en/ ‘CMN -eye’ /a/ ≠ /i/: /sok-a/ ‘contain-IPFV ’, /sok-i/ ‘contain-PFV ’ /u/ ≠ /o/: /gur/ ‘slave’, /gor/ ‘throw’ /iː/ ≠ /ε/ ≠ /e/ ≠ /u/: /biːr/ ‘cry’, /bεr/ ‘sun’, /ber/ ‘whip’, /bur/ ‘terrace’

We have found minimal pairs demonstrating phonemic length in two vowels while Akoli (2013a: 8) found pairs for the other cardinal vowels. They are given below in (9). All vowels are known to appear in all positions, i.e., initially, medially and finally. (9)

/aː/ ≠ /a/: /aːr/ ‘vagina’, /ar/ ‘fire’ /uː/ ≠ /u/: /duːr / ‘knife’, /dur/ ‘rat’ /oː/ ≠ /o/: /oːl/ ‘child’, ol/ ‘pestle’ /εː/ ≠ /ε/: /tεːki/ ‘clear’, /tεki/ ‘see, catch sight of’ /iː/ ≠ /i/: /biːk/ ‘respect’, /bik/ ‘be strong’9

A syllable nucleus may contain two vowels, i.e., diphthongs. All complex nuclei end in a high vowel, either /i/ or /u/, which are realized phonetically as the glides, [j] and [w] respectively. All attested diphthong possibilities are given in the examples in (10) below.

9 These last three minimal pairs are adapted from Akoli (2013a: 8). However, because Akoli (2013a, 2013b) does not find the quality distinction between /e/ and /ε/ to be meaningful we have substituted a transcription from our data where possible.

Kui

(10)

117

/bui/ [buj] ‘betelnut’ /sei/ [sej] ‘chicken’ /sεi/ [sεj] ‘descend’ /nai/ [naj] ‘drink, eat (soft foods)’ /moi/ [moj] ‘clay’ /mau/ [maw] ‘cat’ /wouk~wouk/ [wowk~wowk] ‘owl’ /doit/ [dojt] ‘bottle’ /jaintε/ [jajntε] ‘when’

There are three reasons to support this diphthong analysis. Firstly, this is the environment for the process of monophthongization that will be described in section 2.5.3. Secondly, a diphthong followed by a coda consonant (despite being a phonetic sequence of vowel-glide-consonant) is not eligible for schwa insertion (see section 2.4), that is, the last two examples in (10) do not appear as *[dojət] and*[jajnətε]. From this we can conclude the surface off-glide and the consonant do not form a consonant cluster underlyingly. Lastly, metathesis (see section 2.5.4) assumes the underlying identity of an off-glide and nuclear vowel. In metathesis a vowel that fills a nucleus on its own may alternatively be the second part of a diphthong with the vowel of the preceding syllable. Other sequences of vowels are analyzed by speakers as two syllables, as in the examples in (11). In many cases the appropriate glide is inserted phonetically, [w] following back vowels and [j] following /i/ (see examples in 12). (11)

/goa/ [gowa] ‘exist’ /suan/ [suwan] ‘type of fish’ /akui/ [akuwi] ‘attached’

Across morpheme boundaries vowel sequences always form part of separate syllables. This is visible in (12). Each example is made up of two morphemes: in the first a verb root and the imperfective suffix -a; in the second the applicative prefix u- and a root; in the third a patientive prefix pa- and the root; and in the last the applicative prefix mi- and the root, with what is likely to be an epenthetic glottal stop separating them.

118 (12)

Glenn Windschuttel & Asako Shiohara

/mi-a/ [mija] ‘inside of’ /u-akaːl/ [uwakaːl] ‘eat’ /pa-uban/ [pa.uban] ‘talk to us’ /mi-i/ [miʔi] ‘put into’

Even where the sequence of vowels separated by the morpheme boundary is a possible morpheme internal diphthong in the language, each vowel is pronounced as a separate syllable. For instance, the possible diphthong [aw] is not formed in the second last example of (12). Similarly, even identical vowels across morpheme boundaries form two separate syllables; in the last example a sequence of identical vowels /ii/ is realized as two syllables. The vowel /i/ has a common allophone [ɪ]. This is the typical pronunciation of /i/ in closed syllables as in (13). (13)

/kadir/ [kadɪr] ‘clan’ /anin/ [anɪn] ‘person’ /ipa/ [ɪpa] ‘go.west’

Final high vowels (/i/, /u/ and sometimes /o/ are optionally deleted, as in (14). The only exception is where they form part of a diphthong (and are realized phonetically as an off-glide). Words that end in /u/ are quite rare in the language but those that are frequent all display this behavior. They include dobu ‘straight, correct’ and the first two numerals (and the other numerals which are based on them). When marked with the perfective suffix -i a verb may also be reduced in this way. The presence of the suffix is often still clear from the palatalization of the preceding consonant, which was discussed in section 2.1.2. From this one may deduce that the two processes are ordered and palatalization precedes vowel reduction. (14)

/nuku/ [nuk] ~ [nuku] ‘one’ /aroku/ [arok] ~ [aroku] ‘two’ /dobu/ [dob] ~ [dobu] ‘straight, correct’ /akaːli/ [akaːʎ] ~ [akaːʎi] ‘eat’ /abani/ [abaɲ] ~ [abaɲi] ‘wait’

2.3 Stress Due to the preliminary nature of our data we do not yet have a complete picture of stress in Kui. Nonetheless, preliminary conclusions may be drawn: stress appears to be based on moraic trochees that are right aligned. However, even these preliminary trends have exceptions in our small dataset. Table 4 gives a sample of the possibilities.

Kui

119

In disyllabic words the weight of the final syllable is significant. Final heavy syllables attract stress: heavy syllables may be defined by the presence of long vowels, diphthongs or a coda. In other disyllables stress falls on the first syllable. This suggests a trochaic structure sensitive that is to syllable weight. These two groups can be seen in (15). However, there are exceptions, given in (16), which create stress differentiated minimal pairs with regularly stressed items. These are both somewhat exceptional items in other ways: /uˈti/ is vulgar while /iˈka/ is a specifically vocative form (cf: the regular form /-nana/ ‘older sibling’). (15)

/taˈkiː/ ‘run’, /buˈai/ ‘crocodile’, /maˈran/ ‘go up’ /ˈnaŋa/ ‘NEG ’, /ˈnuku/ ‘one’

(16)

/uˈti/ ‘penis (vul.)’, /ˈut-i/ ‘clean-PFV ’ /iˈka/ ‘older sibling (voc.)’, /ˈik-a/ ‘red-IPFV ’

Words with three or more syllables have right-aligned trochees. The second syllable in trisyllabic/trimoraic words like /aˈroku/ is stressed indicating that the trochee is made of the last two syllables and thus is right aligned. Nonetheless, as /ˈdawan-i/ shows this rule is not exceptionless. (17)

/aˈroku/ ‘two’ /ˈdawan-i/ ‘share’

The words in Table 4 with over three syllables provide further support for this analysis. These words, however, do not give clear indication of the placement of primary stress since on some words the rightmost foot appears to be stressed while on others it is the leftmost. Table 4: Different stress patterns Stressed Syllable Disyllabic

1st

/ˈnaŋa/ ‘NEG ’, /ˈnuku/ ‘one’

2nd

/naˈman/ ‘shellfish’, /aˈnin/ ‘person’, /taˈkiː/ ‘run’, /buˈai/ ‘crocodile’, /maˈran/ ‘go up’

1st

/ˈdawani/ ‘share’

2nd

/aˈroku/ ‘two’

3rd

/u-ga-ˈsir/ ‘push along’

Quadrisyllabic

3rd

/pa-uˈbani/ ‘talk to us’, /omaˈsiŋin/ ‘mosque’

5 – syllabic

2nd

/uˈnababan-i/ ‘forget’

Trisyllabic

120

Glenn Windschuttel & Asako Shiohara

The exact phonetic correlates of stress are unclear, particularly what role pitch plays. There appears to be a minimal pair differentiated by pitch in (18). This may be due to a difference in stress but further analysis is necessary before this may be decided either way. (18)

/arak/ LH ‘arrack, rice wine’ /arak/ HL ‘uncooked, husked rice’

2.4 Phonotactics and syllable structure The surface syllable template is (C)V(C) except one or two instances of complex surface onsets. The same template applies to syllables with diphthongs: (C)VV(C) except that codas only appear if the second vowel is /i/. However, the underlying syllable can be more complex with both complex onsets and codas known, as set out in Table 5. The maximal underlying syllable is CCVC or CVCC ignoring diphthongs. Table 5: Attested underlying syllable structures for one or two syllable words (C)V(C)(C)

/ε/ ‘leg’, /agl/ [agəl] ‘coral’, /batr/ [batər] ‘rice’

(C)(C)V(C)

/ble/ [bəle] ‘day’, /brεk/ [bərεk] ‘split, rip’

(C)(C)VV

/ai/ ‘2SG ’, /tai/ ‘CMN ’, /trai/ [tərai] ‘build’

(C)V.CV(C)

/ara/ ‘on fire’, /baka/ ‘bad’, /arak/ ‘rice’, /bunuk/ ‘smoke’

CVC.CV(C)

/luktu/ ‘knee’, /kinmas/ ‘breath’

CCV(C).(C)CV(C)

/bgura/ [bəgura] ‘yellow’, /skinpo/ ‘type of bird’, /mamplel/ [mampəlel] ‘mango’

CV.V(V)

/goa/ [gowa] ‘exist’, /buai/ [buwai] ‘crocodile’

(C)VV(CCVC)

/buit/ ‘shoot an arrow’†, /jaintε/ ‘when’, /gamain/ ‘make’

† This item is from Akoli (2013b: 15).

The distribution of a small number of consonants is restricted. The palatal lateral and glottal stop are only attested in the coda. These possibilities are exemplified below in (19). Velar nasal may be similarly restricted, only being known word finally and intervocalically. (19)

/soʔ/ ‘oar with a shaped handle’ /akoʎ/ ‘type of fish’ /naŋa/ ‘NEG ’, /kusiŋ/ ‘nail’

Kui

121

͡ is confined to morpheme initial position. It is confined The consonant /dʒ/ to pronominals as noted previously. The adaption of Malay loans demonstrates that this is not an accidental result but a phonotactic constraint. For instance, this consonant is not preserved intervocalically as in (20). (20)

͡ /meda/ < Malay /medʒa/ ‘table’ ͡ /bad/ < Malay /badʒu/ ‘shirt’

Across syllables many different phonemic sequences are allowed. A preliminary listing of the possibilities is given by the series below in (21).10 (21)

nasal + fricative:

/kinsεl/ ‘breath’11

nasal + nasal:

/kinmas/ ‘breath’

nasal + plosive:

/kumba/ ‘basket’

plosive + plosive:

/luktu/ ‘knee’

rhotic + plosive:

/bordol/ ‘black winged tern’12

nasal + rhotic:

/enra/ ‘tear’13

fricative + lateral:

/oslasin/ ‘octopus’

Diphthongs occur widely in open syllables with or without onsets of varying complexity. In closed syllables the second element of all diphthongs is /i/. It is quite possible that these restrictions are due to the relative scarcity of diphthongs and resulting limited data set. However, this does fit the trend that most instances of syllables made up of a diphthong and a coda are the result of metathesis of the verbal suffix /-i/ as described in section 2.5.4. However, this is not (synchronically) true of all examples, for example, /jaintε/ ‘when’. As observed above, almost all surface syllables are of the form (C)V(C). To achieve this Kui has the process of schwa epenthesis in (22) that obtains in homosyllabic complex consonant clusters. This process obtains in almost all onset clusters with the possible exception of one item: /skinpo/ [skin.po] ‘bird sp.’. (22)

C1C2 → C1əC2

10 The syllabification of /mamplel/ [mam.pəlel] ‘mango’ shows the preference for complex onsets over complex codas (even where the onset is later simplified). 11 Both this and the next word are likely to be compounds containing the formative /kin/ as their first element. 12 This is likely to be a compound whose second element is /dol/ ‘bird’. 13 This word is complex, made of two roots: /en/ ‘eye’ + /ra/ ‘semen’

122

Glenn Windschuttel & Asako Shiohara

Otherwise, even plosive-liquid clusters undergo epenthesis (e.g., /brεki/ [bərεki] ‘split, rip’), and without exception all complex codas are simplified. A number of examples are given in (23). (23)

/agl/ [agəl] ‘coral’ /batr/ [batər] ‘rice’ /brεki/ [bərεki] ‘split’ /mamplel/ [mampəlel] ‘mango’ /ble/ [bəle] ‘day’

2.5 Morphophonemics 2.5.1 Allomorphy of agreement prefixes One prefixal series shows allomorphic variation, namely the series with the vowel /a/ in the singular. These are the prefixes of certain obligatorily possessed nouns (see section 4.4) and patientive agreement verbs (see section 6.1). The vowel in these prefixes may be reduced to a schwa or even deleted. The use of these allomorphs is optional and no conditioning factors have been found that link together the words for which this alternation is possible. Thus we conclude that membership of this subclass is lexically determined. Even in this lexical class the underlying forms may still appear phonetically, that is, as the full vowel of the prefix. Which of the two allomorphs is available is phonologically determined and depends on the shape of the root. Before vowel-initial roots the alternation results in a prefix made up of only a consonant. Before consonants the vowel in the prefix may become a schwa. The allomorphs of the morphemes are given in Table 6. Table 6: Patientive and possessive agreement prefixes Irreducible

Consonantal

Schwa

1SG

/na-/

[na-]

[n-]

[nə-]

2SG

/a-/

[a-]

[Ø-]

[a-]

3

/ga-/

[ga-]

[g-]

[gə-]

1PL . INCL

/pa-/

[pa-]

[p-]

[pə-]

1PL . EXCL

/ɲi-/

[ ɲi-]

[ɲ-]

[ ɲə-]

2/3PL

/ji-/

[ji-]

[j-]

[jə-]

CMN

/ta-/

[ta-]

[t-]

[tə-]

Kui

123

Table 7 has examples of obligatorily possessed nouns and patientive agreement verbs beginning with vowels and consonants. The first column shows vowel initial examples where prefixes may not be reduced. In the second column vowel initial examples with and without the reduction are given. The last column presents consonant initial roots where the vowel in the prefix has been reduced to a schwa. Table 7: Reduced and unreduced prefixes Irreducible prefixes

Consonantal prefixes

Schwa prefixes

Verbs

/ga-ui/ [ga-ui] 3.PAT-see ‘see him’

/ga-ei/ [ga-ej] ~ [g-ej] 3.PAT-give ‘give to him’

/ga-lεl/ [ga-lεl] ~ [gə-lεl] 3.PAT-lift ‘lift it’

Nouns

/ga-alor/ [ga-alor] 3-side ‘his side’

/ga-en/ [ga-en] ~ [g-en] 3-eye ‘his eye’

/ga-nei/ [ga-nej] ~ [gə-nej] 3-name ‘his name’

These two allomorphs (schwa and consonantal prefixes) can be unified into one if schwa epenthesis is taken into account. In such an explanation the allomorphic prefix is simply a consonant, C-. When the prefix is added to a root beginning with a consonant the onset is now of the form CC; this is the environment for schwa epenthesis described in section 2.4.

2.5.2 Blending Other reductions have been observed. The affixation of some enclitics promotes the shortening of regular bases. The first part of the bipartite negative (see section 3.2) and the first person singular subject pronoun are contracted together in (24), while the second part of the negative and aspectual clitic =mo are combined in (25). Finally, (26) is different, with the contraction taking place inside a word. However, the phonological form of the change is similar to (25), /naŋa-/ to /na-/. (24)

[nap] < /na gap/ ‘1SG . SBJ NEG ’ [na ɪpa nap nawor naŋa] /na 1SG . SBJ

ipa go.west

na 1SG . SBJ

gap NEG

n-awor 1SG . PAT-return

‘I’m going to the West and won’t come back’

naŋa/ NEG

124 (25)

Glenn Windschuttel & Asako Shiohara

[namo] < /naŋa=mo/ ‘NEG = CONT ’ [waktu omo tamuku gap gowa namo] /waktu omo tamuku gap goa naŋa=mo/ time DIST chief NEG exist NEG = CONT ‘At that time there wasn’t a village chief’

(26)

[nal] /naŋal/ what ‘what?’

2.5.3 Compounds and monophthongization Noun-noun compounds are quite common in Kui. The vast majority are rightheaded as in (27a) while a small minority involving /nεn/ ‘male’ and /mei/ ‘female’ as in (27b) possibly exemplify left-headed compounding. (27)

a. b.

/adol rok/

‘bird feather’

/e kusiŋ/

‘toe nail’

/anin mei/

‘woman’

/anin nεn/

‘man’

Some compounds display a particular phonological reduction at the boundary. When a root with a final diphthong forms the first part of a compound the diphthong is simplified to a monophthong by losing its second element. This is clear in the examples in (28) where the diphthong /sei/ becomes /se/ and /bui/ /bu/. (28)

a.

b.

/sei/ ‘chicken’ /sero/

/serok/

‘chicken egg’

‘chicken feather’

/bui/ ‘betelnut’ /busai/

/buyεra/

‘chew betelnut’

‘betelnut tree’

2.5.4 Metathesis The perfective suffix /-i/ and the final consonant of the verb may optionally undergo metathesis. As set out in (29) the vowel /i/, when suffixed to the verb, may instead precede the final consonant of the verb rather than follow it.

Kui

125

(29) VCi → ViC where C = {n,l}, V = {a,o} This is only known to occur if the root’s coda is simple and either /n/ or /l/. The preceding vowel has in all cases in our corpus been one of the non-front vowels /o/ and /a/.14 Thus the vowel sequence now forms a diphthong. Despite these restrictions it is a common process and some examples are given in (30).15 (30)

/alon-i/

[aloni] ~ [alojn]

‘write-PFV ’

/akaːl-i/

[akaːli] ~ [akaːjl]

‘eat-PFV ’

/-taŋgani/

[taŋgani] ~ [taŋgajn]

‘ask-PFV ’

/uban-i/

[ubani] ~ [ubajn]

‘talk-PFV ’

/gatan-i/

[gatani] ~ [gatajn]

‘free-PFV ’

2.5.5 Voicing assimilation The first consonant of two ‘suffixal’ morphemes (in both cases /t/) assimilates to the voicing of the preceding vowel or consonant as in examples (31) to (34). These are the enclitic /=tε/ ‘and’ and the suffix /-tu/ ‘ALL ’. In both cases the assimilation is generally optional but it is particularly common in rapid speech. For that reason in the fixed phrase in (33) it is effectively obligatory. (31)

[pa mijel=dε] ~ [pa mijel=tε] /pa

miel=tε/

1PL . INCL . SBJ

rest= AND

‘Let’s have a rest!’ (32)

[manam=dε] ~ [manam=tε] /manam=tε/ eat= AND ‘Eat, will you!’

14 Akoli’s (2013b: 15) /buit/ ‘shoot an arrow’ is likely also to be a result of metathesis. 15 František Kratochvíl (pers. comm.) suggests a link to a final/nonfinal distinction found elsewhere in the family, namely, in East Alor: Sawila (Kratochvíl 2014) and Kula (Williams this volume), and even on Pantar, Kaera (Klamer 2014). A quick search through some of the corpus reveals that this is not the case in Kui. For example, example (166a) has metathesized gamain utterance-finally.

126 (33)

Glenn Windschuttel & Asako Shiohara

[a βodo=dε] /a

bodo=tε/

2SG . SBJ

stupid=AND

‘You’re stupid! (34)

a.

[aj-tu] ~ [aj-du] /ai-tu/ 3SG -ALL ‘all of them’

b.

[nyaj-tu] ~ [nyaj-du] /nyai-tu/ 3SG -ALL ‘all of them’

2.5.6 Reduplication Both full and partial reduplication exist in Kui. They can both be used to signify repetition. In full reduplication the whole root is reduplicated as in (35). This new root is treated as a single unit by the aspect suffix which follows the doubled root in (35a). In partial reduplication the onset and nucleus is repeated as exemplified by (36). Both processes are uncommon in our data and only a handful more instances are known. Nevertheless, tentative observations may be made. In some cases reduplication clearly alters the aspect of the predicate iconically: the action of the verb is repeated in (35a) and (36a) (one may observe that these examples represent different forms of reduplication which, nonetheless, are fulfilling the same function). In other examples the meaning is not so clear: it is possible that some form of durative aspect is being expressed in the atelic predicates of (35b) and (36b). (35)

a.

/mester

glen

adol~adol-i/

mister

Glenn

RDP ~jump-PFV

‘Glenn jumped (up and down)’ b.

/gur~gur

naŋa/

RDP ~misbehave

NEG

‘Don’t misbehave!’

Kui

(36)

a.

/mester

glen

kertas

ga-ta~tar/

mister

Glenn

paper

3SG - RDP ~fan

127

‘Glenn fanned the paper.’ b.

/oːl

manak

nun

mu~mul-i/

child

young

PL

RDP ~play-PFV

‘Children were playing.’

3 Basic clausal syntax Kui has a basic SV/APV order in its clauses, for example, (37) and (38). Only a few items can follow a predicate, such as an aspect particle (37) and clausal negator (38). Overt arguments are readily elided where the referent can be retrieved from the discourse context.

(37)

(38)

P V ASP Jadi malél aban gamani lei. COMPL then there village make ‘Then they finished building the village there.’ V to-mug

Dé gap already NEG CMN . DAT-meet ‘We’re yet to meet.’

NEG nanga. NEG

In the following subsections, the basic clause structure of Kui is introduced: predicates (section 3.1), clausal negation (section 3.2), and non-declarative clause types, namely questions (section 3.3) and imperatives (section 3.4). Discussion is here limited to mono-predicative clauses. Multi-predicative serial verb constructions (SVCs) are discussed in section 7.

3.1 Predicate types Basic predicate types in Kui can be differentiated on the basis of their ability to host aspect suffixes and applicative prefixes.16 These suffixes are discussed in section 8 and section 6.3 respectively. Non-verbal predicates do not host either 16 The dative agreement prefixes (section 6.1.3) are found on verbs and possibly numerals (section 4.2.1).

128

Glenn Windschuttel & Asako Shiohara

of these suffixes. In our corpus we identify three kinds of predicate: (i) verbal predicates (section 3.1.1), (ii) nominal predicates (section 3.1.2), and (iii) elevational predicates (section 3.1.3). (An additional category of locational predicates is discussed in section 3.1.4 but this is only really a semantic subclass of the verbs).

3.1.1 Verbal predicates Verbal predicates can be categorized according to valency, the number of arguments that may appear, and the pattern of agreement. Agreement will be discussed in section 6. A verb may be a one-place, two-place or three-place predicate. One- and two-place predicates are rendered syntactically by intransitive and transitive clauses respectively. Three-place predicates have two possible syntactic realizations depending on which arguments are overt. This will be explained below. Kui does not have a distinction between verbs and adjectives. Properties are simply expressed by one-place or intransitive verbs with adjective-like stative semantics. Both semantic categories of verbs may carry the full range of verbal morphology. For example, both stative takat ‘dry’ and dynamic kel ‘cough’ can host the perfective suffix -i ‘PFV ’, as seen in (39) and (40) respectively.

(39)

(40)

S V [Msin yéra]NP dé takat-i. mustard.stick tree already dry-PFV ‘The mustard stick tree is already dry.’ S Botak Botak ‘Botak

V kel-i. cough-PFV coughed.’

There are many transitive verbs in the language and they may also host the suffix -i ‘PFV ’, as can be seen on méd ‘plant’ in (41). This is a verb that cannot show agreement. The different agreement prefixes on transitive verbs are illustrated in (42) and (43). See section 6.1 for a description.

(41)

A P Na atako 1SG . SBJ rice ‘I planted rice.’

V méd-i. plant-PFV

Kui

(42)

(43)

A Na awoi 1SG . SBJ again ‘I’m waiting for

129

P-V ga-ban. 3. PAT-wait him again.’

A P-V Mun gai jo-talék-i? who 3 2PL . DAT-attack-PFV ‘Who attacked you?’

Three-place predicates have two possible syntactic realizations. In both cases the verb’s agreement prefix indexes R. When both R and T are present as NPs the form gala appears following T. The exact role of gala in the syntactic structure is unclear at this stage but the structure resembles the trivalent structures of other TAP languages (Klamer & Schapper 2012).17 The prototypical trivalent verb -ei ‘give’, in (44) below, has three overt arguments: the giver A, ool omo ‘that child’; the theme T, doi ‘money’; and the recipient R, namaa ‘my father’. Because both R and T are overt it uses the gala structure: gala appears in its regular position before R and after T.

(44)

A T R Ool omo doi gala na-maa child DIST money gala 1SG -father ‘That child gave money to my father.’

R-V g-ei. 3. PAT-give

Another pattern is available where R is not present as a NP. Gala need not be present in such a situation and the T argument may appear directly before the verb. In (45) below R is only present as a prefix on the verb. This alternative pattern is optional and gala may appear even where R is not overt as in (46).

(45)

A T R-V Na sen g-ei. 1SG . SBJ money 3. PAT-give ‘I give him money.’

17 There are arguments beyond the scope of this sketch that group gala and R in the VP and separate them from T (similar in this way to Makasae). The form of gala itself is interesting when compared with the other TAP languages because it is not a reflex of ‘come’ or ‘take’ (cf: Klamer & Schapper 2012).

130

(46)

Glenn Windschuttel & Asako Shiohara

A T Na dona bat gala 1SG . SBJ yesterday coconut gala ‘I gave you the coconut yesterday.’

R-V Ø-ei. 2SG . PAT-give

Not all verbs belong exclusively to one valency class: some verbs are labile. Salaak ‘hang’ with the applicative prefix u- is such a verb. It may be used as a two-place or three-place predicate as illustrated by (47a) and (47b) respectively. In (47a) we see a two place predicate; the two arguments are the hanging thing and the location. They are realized by the typical transitive pattern discussed above, the hanging thing as A and the location as P. In (47b) the same prefixed verb is a three place predicate; the three arguments are the ‘hanger’, the hanging thing and the location. This change in valency was created without any derivational morphemes. The new A argument represents the causative agent and the A argument of (47a) is demoted to T. This new predicate acts as any other three place predicate and as both R and T are overt gala is added.

(47)

a.

b.

A P V Atei bata sél u-salaak-a. tree big rope APPL-hang-IPFV ‘The rope hangs off the big tree.’ T R V Sél gala atei guwadin u-salaak-i. APPL -hang-PFV rope gala tree forked ‘(He) hung the rope on the forked stick.’

Five or so other verbs with similar positional meanings are known to act in the same way. The change in aspectual suffix is typical and will be discussed further in section 8. 3.1.2 Nominal predicates Nominal predicates take the form of equative sentences. There is no copula and as such the NPs are simply juxtaposed. The argument typically precedes the predicate. Examples of equative sentences are given below in examples (48) to (50) where the bounds of the argument NP are signified by square brackets. Such clauses do not contain aspect suffixes and, unlike verbal predicates, nominal predicates may host the interrogative enclitic =e as in (50). Interrogative sentences will be discussed in greater depth in section 3.3.

Kui

131

(48) [Ool [ga-yad-i]RC]NP salan-ool. child 3.PAT-give.birth-PFV Islam-child ‘The child she gave birth to is a Muslim.’ (49) [Aban yésan omo]NP Kuiwas, Kuiman, Abelwas, Gamera, Sedenta. DIST Kuiwas Kuiman Abelwas Gamera Sedenta village five ‘These five villages were Kuiwas, Kuiman, Abelwas, Gamera, Sedenta.’ (50) [Aban og g-nei]NP nal=e? village PROX 3-name what=DISC ‘What’s the name of this village?’ While the final noun as part of the NP does form the predicate in nominal sentences, in some cases the noun may become a true verbal predicate and host aspectual suffixes. This is after the lexical process of conversion has taken place. The presence of a lexical change is clear because the noun and verb share only an idiosyncratic and culturally specific semantic connection. In example (51) below a verb meaning ‘search for shells’ derives from a noun meaning ‘shell’ (in Buraga if the tide is out it is common to collect shells, the flesh of which is eaten). (51)

a.

Mester naman-i. mister shellfish-PFV ‘The white guy looks for shellfish.’

b.

Mester naman gorap-i. mister shellfish search-PFV ‘The white guy looks for shellfish.’

3.1.3 Elevational predicates Kui has a class of elevationals, distinguishing three levels: HIGH , LEVEL and LOW (for a definition of the terminology used here and a comparison of the elevational systems of the AP languages, see Schapper 2014). For the Kui living on the coast HIGH and LOW correspond, not only to uphill and downhill but also to mountainward and seaward respectively. These are set out below in Table 8 along with the adverbs and motion verbs which follow the same system. The motion verb séi is exceptional as it is neutral regarding the relative position of the deictic center (DC).

132

Glenn Windschuttel & Asako Shiohara

There is another deictic system that shares the same forms (with one exception). The elevationals place the referent east or west of the DC. HIGH forms denote east and LOW forms west. Séi is again an exception and cannot be used to convey cardinal direction; ipa replaces it in the paradigm space, its only meaning being ‘go west’.18 Thus the system is mostly ambiguous except for séi and ipa, which are exclusive to the elevational and cardinal systems respectively. Since the Kui traditionally live on the south coast the two systems typically have distinct reference. Which of the two possibilities is followed depends on the context. There may be some tendency over short distances to follow the elevational system while for reference over greater distances a cardinal reference is more likely. Table 8: Elevational terms Motion verbs Elevationals

Adverbs

High

aro

Level

balo

Low

iyo

iran

Neutral

From DC

To DC

aran

mira

maran

balan

bai

mai

séi

Elevationals form a distinct word class. Unlike demonstratives (see section 4.3), which they most closely resemble, they may be predicates in their own right. Where elevationals are used as one-place predicates, this expresses the height and position of the argument relative to the DC. In example (52) the elevational indicates the mosque’s distal location, away from both the speaker and addressee, and level with the DC. (52)

Omasingin balo. LEVEL mosque ‘The mosque is over there.’

In their non-predicative uses the elevationals show the same syntactic distribution as demonstratives (see section 4.3). They place an entity or an action in a position relative to the DC when used at the end of a NP or clause respectively. This is illustrated by (53) and (54): balo follows kabii ‘goat’ in (53) to indicate it is located level with the speaker while postpredicative aro places the action of sitting above the DC in (54). 18 In other TAP languages cognates of sêi and ipa conform to the rest of the paradigm denoting ‘come down/west’ and ‘go down/west’ respectively (see Schapper 2014).

Kui

(53)

Kabii balo gatan-i! goat LEVEL free-PFV ‘Let the goat over there go!’ (Akoli 2013:8)

(54)

Gé misa aro. 3.FOC sit HIGH ‘He sat up there.’

133

3.1.4 Predications expressing location and existence Kui has a number of verbs to express existence depending on the semantics desired. There is the verb goa that expresses general existence. This is commonly used in negative clauses as in (55) but is not confined to them as evidenced by (56). (55)

Waktu omo tamuku gap goa namo. DIST chief NEG exist NEG . CONT time ‘At that time there wasn’t a village chief.’

(56)

Aninnok goa=té na aban mi-lol. person exist=AND 1SG . SBJ village IN -stay ‘If there were people, I would’ve stayed in the village.’

It is more complex to express existence at a location or simply the location of a particular referent in Kui. A number of different verbs are available. Which verb to choose depends on the animacy, number and relative shape of the referent. This is set out below in Table 9 and examples follow. Table 9: Positional verbs misa

‘sit’

nasén

‘stand’

isa

‘lie’

taa

‘lie’/ ‘sleep’ (human)

Items that are longer along the vertical axis necessitate the use of nasén ‘stand’. A bottle is one such item. Nasén is used for this noun, for example, in (57). (57)

Botil nuku wor la-nasén. bottle one rock ON -stand ‘One bottle is on the rock.’

134

Glenn Windschuttel & Asako Shiohara

Referents which do not have such a prominent vertical axis and, according to the explanation of speakers, ‘have a bottom’, require the use of misa ‘sit’. Pots fit into this category as in (58). (58)

Kumba atei ga-tany la-misa. pot tree 3-branch ON -sit ‘The pot sits on the tree branch.’

All other items ‘lie’ and use the verb isa. Balls do not have a clear bottom or a longer vertical axis and thus use isa as exemplified by (59). Similarly a cloth does not have either of those features and uses isa as in (60). (59)

Bal lapan mi-isa. ball ground IN -lie ‘The ball is on the ground.’

(60)

Meda saben bgura meda la-isa. table cloth yellow table ON -lie ‘The yellow cloth is lying on the table.’

The verb isa is the catch-all verb for all other locational predicates. The classifications above are not inherent to the items themselves but are only valid when in their canonical position. If a bottle is no longer upright and the longer axis is no longer vertical, it no longer ‘stands’ but ‘lies’ as in (61). Similarly, if a pot is not sitting on its bottom, isa is used as in (62). (61)

Botil kalat mare mi-isa. bottle basket inside IN -lie ‘The bottle is on its side in the basket.’

(62)

Kumba atei tuk la-rok-a pot tree short ON -upside.down-IPFV ‘The pot is on the stump upside down.’

isa. lie

However, isa is only used if the number of the referent is clear (i.e. easily counted). Thus if there are three of the referent it may still be used as in (63). However, at some point when there are too many the verb sok is used instead. In example (64) seven was enough to prompt the use of sok. This is also the regular verb to locate liquids and not easily individualized items such as beans

Kui

135

as in (65). In addition it is the native Kui way to render gatherings of people as in (66). In (66) Malay kumpul ‘gather, group’ is used alongside the Kui expression. (63)

Bal siwa meda la-isa. ball three table ON -lie ‘The three balls are on the ground.’

(64)

Botil meda la-sok-a. bottle table ON -be-IPFV ‘A group of bottles is on the table.’

(65)

Utun bur mi-sok-a. bean courtyard IN -be-IPFV ‘The beans are on the ground of the courtyard.’

(66)

Nyai 1PL . EXCL

kumpul-i gather-PFV

aban village

Sirla Sirla

sok-a be.in-IPFV

omasingin mosque

u-gasam-i u-ga-bur-i. APPL -improve-PFV APPL -3.PAT-flat-PFV ‘We gathered together in Sirla village to fix the mosque.’ (Akoli 2013: 3) The chief exception are animates. Animates may stand, sit or be grouped as above; however, while an animate being is alive the verb isa may not be used with it. The verb, taa, which by extension may also mean sleep, is the only possible choice. An example is given below in (67). (67)

Ala akany a téma taa? last night 2SG . SBJ where lie ‘Last night where did you lie i.e. sleep?’

3.2 Negation Kui clauses are negated by the addition of nanga at the end. There is also the regular possibility of gap appearing somewhere before P (often immediately before) as in (68a). This may be doubled but is not obligatory. The reasons behind its presence or absence are currently unknown. This form gap is otherwise unknown in our corpus. Declarative (68a) and imperative (68b) sentences are somewhat distinct in their behavior under negation. Both use nanga but gap is unknown in imperative sentences, for example (68b). As mentioned in section 2.5.2, negative nanga

136

Glenn Windschuttel & Asako Shiohara

and the modal clitic =mo may be reduced to the form namo as in example (68c) below. (68)

a.

(Gap) anin (gap) ool blés (NEG ) person (NEG ) child hit ‘The person didn’t hit the child.’

nanga. NEG

b.

Gur~gur nanga! misbehave~RDP NEG ‘Don’t misbehave!’

c.

Waktu omo tamuku gap goa namo. time DIST chief NEG exist NEG . CONT ‘At that time there wasn’t a village chief.’

Other negative modalities, such as inability and aversion, can be expressed by baka ‘bad’ in the same postpredicative slot as nanga. An example is given in (69). (69)

Segel bik-a bata, na aban mi-bai wind strong-IPFV very 1SG . SBJ village IN -go ‘Because it’s very windy, I can’t go to the village.’

baka. bad

3.3 Questions Polar questions are created from declarative clauses without a change in word order. In the case of verbal sentences there is no syntactic change at all as in (70). In the case of nominal clauses the clitic =e (or =we following vowels) is appended to the predicate as in (71) (this has only one other use, see example (79) in section 3.4). Tag questions are routinely formed by the addition of dé nanga? ‘or not?’ to the end of the clause as exemplified by (72). (70)

Dé manam? already eat ‘Have you eaten?’

(71)

A masin 2SG . SBJ Kui ‘Are you Kui?’

(72)

A nyo-mon noka 2SG . SBJ 1PL . EXCL . DAT-help can ‘You can help us, can’t you?’

innok=e? person=DISC

dé or

nanga? not

Kui

137

Content questions are formed by the use of one of the interrogative words found in Table 10. Some of these words are seemingly complex. For instance, -wainté appears not only as part of téwainté ‘why’, but also ogwainté ‘thus’. These words, as well as yainté ‘when’, end with a form identical with the clitic =té. It is not clear whether these words are synchronically segmentable. Table 10: Interrogatives té

‘which’

té(ma)

‘where’

téwainté

‘why’

yainté

‘when’

nal

‘what’

mun

‘who’

tanina

‘how’

ren

‘how many’

The interrogative pro-forms usually appear in the sentence in the same position as the questioned constituent. This is clear from example (73) where nal is in the preverbal position typical for P arguments. The question word may also host the interrogative clitic itself even in verbal clauses as is evident from (74). Nominal sentences with question words act in much the same way as polar questions in that the clitic appears clause finally as in (75) and (76). Beyond their use as arguments question words may appear attributively as in (76). Where the question word is adverbial in meaning it normally occupies the first position in the clause as in (77). (73)

Norma, a nal u-yér-i? Norma 2SG . SBJ what APPL -laugh-PFV ‘Norma, what were you laughing about?’

(74)

Mun=e a-brét-i? who=DISC 2SG . PAT-inform-PFV ‘Who told you?’

(75)

Aban og g-nei nal=e? village PROX 3-name what=DISC ‘What’s the name of this village?’

(76)

Og

nal yéra=we? what tree=DISC ‘What tree is this?’ PROX

138 (77)

Glenn Windschuttel & Asako Shiohara

A ipa, yainté a maran? 2SG . SBJ go.west when 2SG . SBJ go.east ‘Once you’ve gone, when will you return?’

3.4 Imperatives Imperatives have the same word order as sentences in other moods with few syntactic differences. The addressee is even less likely to be overt and the difference in negatives has already been discussed in section 3.2. The aspectual suffixes differ in their permissibility in imperative clauses. The perfective -i ‘PFV ’ is not uncommon, as in (78), while -a ‘IPFV ’ is unknown (the non-dynamic semantics likely preclude a sensible imperative interpretation, see section 8). The clitic used in questions, =(w)e, does appear in one expression with an imperative type meaning in (79); the base is some form of interjection with no other current uses in the language. (78)

Kabii balo gatan-i! goat LEVEL free-PFV ‘Let the goat over there go!’ (Akoli 2013: 8)

(79)

Ara=we! show=DISC ‘Show me it’

There is one enclitic, =té, that is commonly found in imperative sentences, for example (81) and (82). Its primary use, however, is not related to illocutionary force. It is the most common means to connect clauses. The semantic link between the two clauses can be quite varied: the meaning of the second clause in (80a) is dependent on the first, while in (80b) the two are quite independent in meaning. This is also the native way to express conditional relations as in (80c).19 (80)

a.

Na uban=té na laak. 1SG . SBJ say=AND 1SG . SBJ walk. ‘I’ve already let you know that I’m going.’

b.

Na aban mi-yai=té n-awar-i. 1SG . SBJ village IN -go=AND 1SG . PAT-return-PFV ‘I’m back from the village.’ (lit. ‘I went to the village then returned.’)

19 Cognates in other TAP languages also have these two functions, e.g., Makalero =teʹe (Huber 2011: 448)

Kui

c.

139

Aninnok gowa=té na aban mi-lol. person exist=AND 1SG . SBJ village IN -stay ‘If there were people I would’ve stayed in the village.’

In imperatives, for example (81), it adds politeness, softening commands often being translated into Malay as dulu. First person imperatives, for example (82), are of the same form; the first person exclusive pronoun is likely to appear as well. (81)

Manam=té! eat=AND ‘Go ahead and eat!’

(82)

Pa miel=té! 1PL . INCL . SBJ rest=AND ‘Let’s have a rest!’

4 Noun phrases The template of the Kui noun phrase (NP) is presented in (83). The NP is head initial: modifiers follow the head noun. It is maximally composed of a head noun (NHEAD) followed by a verbal attribute (ATTR ) or relative clause (RC), a quantifier (QUANT ), and a demonstrative (DEM ). Adnominal possessors are expressed by an agreement prefix on the head noun. A possessor NP may optionally also precede the NP to explicitly identify the possessor. The elements filling these slots will be discussed in the following sub-sections. Template of the Kui NP (83) (PSRNP) [AGRPSR-NHEAD (ATTR /RC) QUANT DEM ]NP This template is approximate. For instance, relative clauses follow the head noun as in (84a). In one example (84b), however, relative clause precedes the quantifier. The remaining examples in (84) illustrate the relative position of other slots. (84)

a.

NHEAD RC [ya [aban mi-yai]RC]NP road village IN -go ‘road to the village’

140

Glenn Windschuttel & Asako Shiohara

b.

(NHEAD) RC QUANT [Ø [Ei la]RC nun]NP maran. PL go.up boat ON ‘They went up from the boat.’ (lit. ‘They who were on the boat went up.’)

c.

NHEAD QUANT DEM [muur yéra nukNU M P orange tree one ‘that one orange (tree)’

d.

mo]NP DIST

NHEAD ATTR QUANT [atako tuk arok]NP yam small two ‘two small yams’

4.1 Attributes and relative clauses Attributes are intransitive verbs which directly follow the head noun as in (85) and (86). These are normally stative verbs with adjective-like semantics. (85)

[Atako tuk]NP tai la-galuma. CMN ON -pile cassava short ‘The short cassava are piled on top of one another.’

(86)

Naman takat la-lol. ON -walk shellfish dry ‘They walked on the shore.’

Relative clauses in Kui are unmarked and follow the noun. The relative clause employs a gapping strategy (this is ignoring obligatorily bound pronominals). Sentence (87) contains an example of a relative clause. Where only a single verb follows the noun in the NP the appropriate analysis is often unclear; however, where an aspect suffix is present the verb is a predicate (see section 3.1.1) and thus heads its own clause inside a NP, that is, a relative clause. This is exemplified by (88) and (89). (87)

bek-i. [Ya [aban mi-yai]RC ]NP dé already break-PFV road village IN -go ‘The road to the village is in a sorry state.’

Kui

(88)

[Ool [ga-yad-i]RC]NP salan-ool. child 3.PAT-give.birth-PFV Islam-child ‘The child she gave birth to is a Muslim.’

(89)

Mainsino ei la nun maran [Ø [takat-a]RC]NP go.up dry-IPFV then boat ON PL ‘They went up from the boat onto the shore.’

141

la. ON

4.2 Quantification Number is not a syntactic category marked on nouns in Kui. However, a NP may optionally be marked for it by a numeral or plural marker (‘PL ’). Other strategies include the use of verbs such as the abaya ‘many’ and pronouns marked for number (see section 5.5). These other strategies will not be discussed further here. 4.2.1 Numerals Table 11 gives a list of the ten basic numerals in Kui. The numerals up to and including 6 are simple without internal structure while the numerals 7, 8, 9 are complex and are combinations of two simplex numerals (with the exception of 8, the historical structure of which is more complex; see Schapper & Klamer 2014). The higher numeral bases beyond kar are asaga ‘100’ and rib ‘1000’ (from Indonesian ribu). Table 11: Numerals 1

nuku

6

talama

2

aroku

7

yésaroku tad usa

3

siwa

8

4

usa

9

5

yésan

10

yésan usa karnuku

Numbers over ten are based on ten or higher bases multiplicatively: the base on the left with the other factor following, for example, ‘twenty’ in (90a). Numbers which are not multiples of ten are created with wail ‘plus’. This is illustrated by the numeral ‘twenty three’ in (90b).

142 (90)

Glenn Windschuttel & Asako Shiohara

a.

kar-aroku ten-two ‘twenty’

b.

kar-aroku wail ten-two plus ‘twenty three’

siwa three

A numeral may also be a predicate, as in (91). (91)

Aban mo nuk kola. village DIST one only ‘There is only one village.’ (lit. ‘That village is one.’)

Numerals can take the third person dative prefix go- (see section 6.1), for example (92), which modifies the meaning of a numeral X so that it means ‘1/X’, that is, it derives fractions. They are ordered after the verb in (92) giving the construction the flavor of a SVC (this is attested elsewhere; Klamer et al. 2014: 355–356). (92)

Nyi lei gawaring go-siwa go-usa. 3.DAT-three 3.DAT-four 1PL . EXCL . SBJ COMPL share ‘We already broke it (the village Lerabaing) up into thirds up to quarters.’

A numeral (NUM ) may either occur alone or with a classifier (CLF ). The classifier precedes the numeral in the NP. An example is given below in (93).

(93)

NHEAD CLF NUM o yéra siwa house tree three ‘three houses’

Classifiers form a distinct word class in Kui. The form of the classifier may have its source in another word class but once grammaticalized it is bleached of its semantics. The classifier yéra derives from a noun meaning ‘tree’. In (94) it is possible that this meaning is preserved: the classifier specifies that it is a tree being counted. However, in (95) and (96) it does not classify the nouns o ‘house’ or meja ‘table’ as trees but simply as inanimate items that stand. The classifier is always optional: for example, o does not need a classifier to be present to be counted as shown by (97).

Kui

(94)

Bat yéra ren nasén? coconut tree how.many stand ‘How many coconut trees are there?’

(95)

o yéra siwa house tree three ‘three houses’

(96)

Meja yéra ren? table tree how.many ‘How many tables are there?’

(97)

O ren nasén? house how.many stand ‘How many houses are there?’

143

Nouns may be classified in different ways depending on the exact nature of the referent at a given point in time. For example, the noun bat ‘coconut’ is polysemous between the fruit and the tree. It may be used with either of the classifiers, ub ‘fruit’ or yéra ‘tree’, to unambiguously identify what is being counted. The ‘tree’ classifier was used above in (94) and the ‘fruit’ classifier is instead used below in (98) to produce a clear difference in meaning. (98)

Bat ub ren? coconut fruit how.many ‘How many coconuts are there?’

4.2.2 Plural marker Kui has a plural word nun ‘PL ’. The Q UANT slot can be filled by the plural marker. The plural marker follows the attribute and precedes the demonstrative as in (99a). Further examples of PL are given in (99b) and (99c). (99)

a.

b.

anin kalet nun person old PL ‘these old people’

ogo PROX

[Ool manak20 nun]NP PL child young ‘Children were playing.’

mu~mul-i. RDP ~play-PFV

20 The stative verb manak ‘young’ is part of a lexicalized phrase here, ool manak, meaning ‘children’

144

Glenn Windschuttel & Asako Shiohara

c.

[Ei la nun]NP maran. go.up boat ON PL ‘They went up from the boat.’ (lit. ‘They who were on the boat went up.’)

4.3 Demonstratives Demonstratives usually occur at the right end of the NP. The demonstratives are of two forms: proximal, near the speaker and distal, including near the addressee. The forms are given in Table 12. As noted in section 3.1.3 and section 5 the elevationals and the pronouns may also fill this same slot. Table 12: Demonstratives PROX

og(o)

DIST

(o)mo

Examples of different uses of the demonstratives are given in (100). Demonstratives may form part of a NP as in (100a) and (100b). Alternatively they may head a NP entirely on their own as in (100c). (100)

a.

nal=e? [Aban og]NP g-nei village PROX 3-name what=DISC ‘What’s the name of this village?’

b.

[muur yéra nuk mo]NP orange tree one DIST ‘that one orange tree’

c.

[Omo]NP

artinya masin ler. DIST means Kui king ‘That means the king of the Kui.’

4.4 Possession Possession is marked in Kui by a prefix on the noun which indexes the personnumber features of the possessor. The possessor may also appear as an overt NP preceding the possessed NP. Nouns with such prefixes are exemplified in (101) and (102) with and without an overt possessor NP respectively; the NP may even be a pronoun as in (103).

145

Kui

(101)

Bapak Hajik ga-o mi-yai. Bapak Hajik 3-house IN -go ‘I’m going to Bapak Hajik’s house.’

(102)

na-maa 1SG -father ‘my father’

(103)

Norma jalé ja-min ta-bagar. Norma 2DU 2PL-nose CMN . PAT-same ‘Norma’s and your nose are the same.’

Kui nouns divide into two classes, those that require a possessive marking and those that do not. The obligatorily possessed class covers most body parts, kin terms and -nei ‘name’, all typical members crosslinguistically. Exceptions include aar ‘vagina’. The different forms of the prefix are given in Table 13. The form of the prefix is generally the same for both classes of nouns although as discussed in section 2.5.1 some obligatorily possessed nouns may undergo vowel reduction and no other nouns are known to. However, this is not a characteristic of the whole class; there are obligatorily possessed nouns for which vowel reduction in the prefix is unknown, as noted there. Table 13: Prefixes on optionally and obligatorily possessed nouns

1SG

Optionally possessed

Obligatorily possessed

o ‘house’

-en ‘eye’

-nei ‘name’

na-o

n-en

n(a)-nei

2SG

a-o

en

a-nei

3

ga-o

g-en

g(a)-nei

1PL . INCL

pa-o

p-en

p(a)-nei

1PL . EXCL

nyi-o

ny-en

ny(i)-nei

2/3PL

ja-o

j-en

j(a)-nei

CMN

ta-o

t-en

t(a)-nei

It should be noted that a noun’s need for possession is entirely grammatical, that is, it is only the requirement of a prefix. For example, the noun -min ‘nose’ has been extended to mean ‘cape, headland’. When used thus the possessive marking has no semantic referent; nonetheless, it appears obligatorily, as in (104).

146 (104)

Glenn Windschuttel & Asako Shiohara

Anin balo ga-min la-maran. person LEVEL 3-nose ON -go.up ‘The people over there go up onto the headland.’

5 Pronouns Kui has multiple paradigms of independent pronouns. Table 14 presents the different pronouns in Kui. Gaps in the table indicate simply that a form was not found in our corpus and in most cases are not likely to be the result of a systematic gap. Pronouns show the same feature distinctions as the agreement and possessive prefixes. That is, they distinguish first, second, third, and COMMON (CMN ) persons as well as having a clusivity distinction. These features will be explained below. All but the quantifying pronouns distinguish singular and plural numbers. The pronominal paradigms are not marked for grammatical relations, with a single exception. Some details of the different pronominal paradigms will be given in the following sections. As mentioned earlier the person and number categories of the pronouns are shared by the possessive and agreement prefixes. Furthermore, the forms of both these prefixes and the independent subject pronouns (see section 5.2) are almost identical. Thus it will be sensible to discuss these shared features here at the same time. With three exceptions, 2/3PL , 3 and CMN , their meanings are straightforward. The pronominal feature CMN always marks a plural referent and has two uses: a distributive reading where the predicate holds for each member individually or a reciprocal reading. The distributive reading is displayed by (105a): despite the possibility of the act of washing being reciprocal the verb -wel with the distributive prefix does not depict a reciprocal action. The alternative reading is present in (105b) and (105c). It may be simply that the distributive reading is general and a reciprocal reading is forced on predicates for which the distributive reading is semantically impossible. (105)

a.

Pa yai ta-wel-i. 1PL . INCL . SBJ go CMN . PAT-bathe-PFV ‘We are going to bathe.’

b.

Bangku arok tai la-i. seat two CMN ON -put ‘Put the two stools on top of one another.’

c.

Opung Kainun alé ta-burin. Opung Kainun 3DU CMN . PAT-fight ‘Opung and Kainun fought with each other.’

147

Kui

The 3rd person and 2nd/3rd plural pronominals have overlapping possible referents. The g- forms may be used for both third person singular (106a) and plural (106b). The singular usage is by far the most common. The j- pronominals also cover the third person plural as in (107a). In addition they are the only possible realization of the second person plural as exemplified by (107b). Instead of glossing the j-forms as 2/3PL they are glossed with the person feature appropriate to the context. (106)

(107)

a.

Mester Glenn tata ga-balel-i. Mister Glenn seating.platform 3.PAT-circle-PFV ‘Glenn circled the seating platform.’

b.

Kuiwas nun gé nyo-talék. Kuiwas PL 3.FOC 1PL . EXCL . DAT-attack ‘Kuiwas are the ones who attacked us.’

a.

Ende j-awar-i ji ei Ende 3PL . PAT-return-PFV 3PL . SBJ boat ‘The Ende people returned to the boat.’

b.

Mun gé jo-talék-i? who 3.FOC 2PL . DAT-attack-PFV ‘Who attacked you?’

la-yai. ON -go

Table 14: Pronouns attested in our corpus Basic

Subject

Focus

Possessive

Dual

Group

nalé

nainun †

All

1SG

nai

na

né

neg

2SG

ai

a

é

eg

3

gai

ga

gé

geg

1PL . INCL

pai

pa/pi

pé

pilé/palé

painun

paitu

1PL . EXCL

nyai

nyi

nyé

nyilé

nyainun

nyaitu

2/3PL

jai

ja/ji

jé

jalé

CMN

tai

ta

té

Only naningkol aningkol

alé

aitu

teg

† It is unknown whether this pronoun has the same meaning as the ones built from the plural pronoun

These pronouns are not pro-NPs in the sense that they replace a whole NP. As (108a) demonstrates a noun and a pronoun may appear as the same argument; the proper noun Botak and the third person pronoun both represent the P argument. In all cases the pronoun immediately follows the rest of the NP. Examples,

148

Glenn Windschuttel & Asako Shiohara

such as (108b), suggest that the pronoun fills the demonstrative slot following the numeral phrase. (108)

a.

Mester Glenn [Botak Mister Glenn Botak ‘Glenn hugged Botak.’

b.

[Kuiwas nun gé]NP nyo-talék. PL 3.FOC 1PL . EXCL . DAT-attack Kuiwas ‘Kuiwas are the ones who attacked us.’

gai]NP 3

bakak-i. hug-PFV

5.1 Basic pronouns BASIC pronouns can appear as any argument in a clause. In a number of situations this is the only discourse neutral pronoun possible. This is the only neutral pronoun for P, for example (109); an applicative object, introduced by la- (110) or u- (111) (see section 6.3); or a comitative participant introduced as part of a comitative SVC as in (112) (see section 7.4). (109)

Na gai karap. scratch 1SG . SBJ 3 ‘I scratch him.’

(110)

Bangku arok tai la-i. seat two CMN ON -put ‘Put the two stools on top of one another.’

(111)

Na gai u-mai. APPL -come 1SG . SBJ 3 ‘I remember him.’

(112)

Na ai to-pun 1SG . SBJ 2SG CMN . DAT-hold ‘I am sitting with you.’

misa. sit

The basic pronoun may be used for A and S where the subject pronoun is also available (see section 5.2). This use appears to generally have a pragmatic function, for example (113), where an argument is already present as a NP. However, there are other examples such as in (114) where there are no clear factors justifying its use.

Kui

(113)

Mun gai jo-talék-i? who 3 2PL . DAT-attack-PFV ‘Who attacked you?’

(114)

Nai na-maa ga-takaal 1SG . SBJ 1SG -father 3-hair ‘I cut my father’s hair.’

149

u-garot-i. APPL -cut-PFV

5.2 Subject pronouns The SUBJECT pronouns (SBJ ) are homophonous with the patientive agreement prefixes except for some additional forms (pi and ji, which are in free variation with pa and ja). This paradigm of pronouns is confined to A and S in our data. This is exemplified by (115a) and (115b) respectively below. (115)

a.

Na awoi ga-ban. 1SG . SBJ again 3.PAT-wait ‘I’m waiting for him again.’

b.

Ala akany a téma taa? last night 2SG . SBJ where lie ‘Last night where did you sleep?’

There is no semantic restriction on the referent of the pronoun. It is not confined to agentive arguments, as in (115) above, but may be used for patientive S, as in (116) below. In (116b) the referent of the subject pronoun is even indexed by a patientive agreement prefix. (116)

a.

A-yak ga dela. 2SG -body 3.SBJ white ‘Your body’s white.’

b.

Ga awoi ga-rik-i. 3.SBJ again 3.PAT-sick-PFV ‘He had been sick again.’

5.3 Possessive pronouns The POSSESSIVE pronouns (POSS ) are pronouns with meanings and functions in the manner of English mine, yours, etc. (cf. possessor prefixes in section 4.4). That is, they pick out a referent purely by the person and number of its possessor.

150

Glenn Windschuttel & Asako Shiohara

This is illustrated by the examples in (117) below. Note that the possessor may be overtly expressed by a preceding NP as in (117b) (just like possession in any other NP, see section 4.4). In addition to this they have non-possessive uses, for example (118a) and (118b), representing the person-number combination itself. In these uses the referent always bears the semantic role of theme as in these examples. Whether this restriction reflects a real conclusion about the semantics of this pronoun requires further investigation. (117)

(118)

a.

Neg nang. 1SG . POSS NEG ‘It’s not mine.’

b.

Anin geg=e.21 person 3.POSS = DISC ‘It’s someone’s.’

a.

Eg séi go-yom-i. 2SG . POSS descend 3.DAT-watch-PFV ‘You went and looked at him.’

b.

Mansino og geg u-g-awar-i. PROX 3.POSS APPL -3.PAT-return-PFV then ‘Then he returned to this place.’

5.4 Focus pronouns The FOCUS pronouns (FOC ) appear following a question word or the focused constituent that answers a content question. Examples are (119a) and (119b) respectively: (119b) was given as an answer to the question in (119a). In other situations it appears alone as in (120) (it is unclear from the context whether the referent of the pronoun is in focus). (119)

a.

Mun gé jo-talék-i? who 3.FOC 2PL . DAT-attack-PFV ‘Who attacked you?’

b.

Kuiwas nun gé nyo-talék. 3.FOC 1PL . EXCL . DAT-attack Kuiwas PL ‘The Kuiwas are the ones that attacked us.’

21 The force of the discourse marker =e ‘DISC ’ on geg in (117b) is not currently understood.

Kui

(120)

151

Gé naman go-rap. 3.FOC shellfish 3.DAT-search ‘They were looking for shellfish.’

5.5 Quantificational pronouns Quantificational pronouns are special pronominal paradigms which assign their referents a certain quantity. There are four paradigms: ONLY, PLURAL (PL), DUAL (DU ) and ALL . The ONLY pronoun straightforwardly expresses ‘only me, you, etc.’ depending on the prefix. It is complex, made up of -anin ‘person’ and kol ‘only’. (121)

Ø-aningkol séi? descend 2SG - ONLY ‘You went down on your own?’

The PLURAL pronouns are created quite transparently from basic pronouns suffixed with the plural marker nun as in (122) below. The plural marker is not acting as an independent quantifier here since the plural marker typically precedes the pronoun in the NP (see section 4.2.2). (122)

Gai-nun tosoi séi aban pat-i. send descend village found-PFV 3-PL ‘They sent them down to found the village.’

Kui also has a pronominal paradigm, DUAL , for referents that number precisely two. An example is given by (123). These pronouns are apparently built up from the subject pronouns and the cranberry morph -lé.22 (123)

Palé yai menegi. 1DU. INCL go fish ‘We (you and I) go fishing.’

The third person form is used following two NPs, for example (124a), in situations where English would employ a conjunction. In other persons similar conjunctive uses are found except that only one NP precedes the dual pronoun. 22 However, the expected form *galê does not appear. Instead, alê, which does not have the expected second person meaning, functions as the third person dual. The ALL form, aitu, is similarly third person (cf. for example, Teiwa third person short pronoun a, Klamer 2010: 79).

152

Glenn Windschuttel & Asako Shiohara

It groups together the NP and the person represented by the pronoun; for example, 1st in (124b) or 2nd in (124c) are equivalent to English ‘me and NP’ or ‘you and NP’ respectively. This is an inclusory construction of a form described in Lichtenberk (2000). (124)

a.

Opung Kainun alé ta-burin. Opung Kainun 3DU CMN . PAT-fight ‘Opung and Kainun fought with each other.’

b.

Monas nalé Kalimbas mi-yai. Monas 1DU Kalabahi IN -go ‘Monas and me are going to Kalabahi.’

c.

Norma, Builo jalé ja-tawom dé goa. Norma Builo 2DU 2PL-partner already exist ‘Norma, you and Builo already have boyfriends.’

ALL

pronouns have the meaning ‘all of us, them, etc.’ as in the examples

in (125). (125)

a.

Nanga,

nyi walél, nyaitu to-lul=po. 1PL . EXCL . SBJ here 1EXCL . ALL CMN . DAT-call=CONT ‘No, we were here, all calling to each other.’ NEG

b.

Aitu to-lul to-kar. 3.ALL CMN . DAT-call CMN . DAT-call ‘They were all calling to each other.’

6 Agreement, applicative and other prefixes The template of prefix slots on the Kui verb is given in (126). There are up to three prefixal slots on the verb although only two are known to be filled at the same time. The slot closest to the verb root is the AGREEMENT SLOT (AGR ). This slot may contain up to two prefixes as exemplified by (127a) and (127b). The APPLICATIVE SLOT (APPL) precedes this slot. This is where one of the three Kui applicatives can appear: u- ‘APPL ’, la- ‘ON ’, and mi- ‘IN ’. This is clear from many examples such as the one given as (127c) with the prefix u-. From one example (127d) we place as the outermost slot the INCORPORATION SLOT (INCORP ), the site on the verb for the incorporated noun. The EMPHATIC SLOT (EMPH ) has as its only member the emphatic prefix, a-. From examples like (127e) we can tell that its slot is before the agreement slot.

Kui

(126)

(127)

153

Verbal prefix template   AGR-V INCORP-APPLEMPH Illustration of the verbal prefix template a. AGR DAT-AGR PAT -V Mester Glenn ga-bad ga-tan go-ga-ber-i. Mister Glenn 3-shirt 3-hand 3.DAT-3.PAT-get.out-PFV ‘Glenn let his hand get out of his shirt’ i.e. ‘Glenn undid his sleeve.’ b.

AGR DAT-AGR PAT -V Nabad go-g-ak-i. door 3.DAT-3.PAT-open-PFV ‘I open the door for him.’

c.

APPL-AGR-V Er snam u-na-dak. APPL -1SG . PAT-press water hot ‘He shot hot water at me.’

d.

INCORP-APPL-V Nawol dagar akaal-i kopi banana fry eat-PFV coffee ‘Eat fried banana with coffee!’

e.

er-la-nai! water-ON -drink

EMPH -AGR-V A-g-ak. EMPH -3.PAT-open ‘(I) open (my mouth) wide.’

6.1 Agreement There are two possible paradigms for agreement in Kui. Patientive agreement (PAT ) is a lot more common than dative agreement (DAT ). Dative agreement is so called because it has generally less patientive semantics. The forms of the prefixes are given below in Table 15 (patientive agreement has a reduced form on some verbs, see section 2.5.1). The person-number features correspond to those of the pronouns and have already been described in section 5.

154

Glenn Windschuttel & Asako Shiohara

Table 15: The two paradigms of pronominal agreement prefixes Patientive (PAT )

DATIVE (DAT )

1SG

na-

no-

2SG

a-

o-

3

ga-

go-

1PL . EXCL

nyi-

nyo-

1PL . INCL

pa-

po-

2/3PL

ja-

jo-

CMN

ta-

to-

It should be noted that agreement prefixes in Kui are not pronominal in the sense that they exclude free nominal elements. A participant indexed by an agreement prefix can simultaneously be encoded by a NP or pronoun as in (128) and (129) respectively. (128) Mester Glenn tata ga-balel-i. Mister Glenn seating.platform 3.PAT-circle-PFV ‘Glenn circled the seating platform.’ (129) Aninnok aban mi-a, na lei gap n-awar nanga. person village IN - IPFV 1SG . SBJ COMPL NEG 1SG . PAT-return NEG ‘If there were people in the village, I wouldn’t have already returned.’ 6.1.1 Agreement on intransitive verbs There are two primary lexical classes of intransitive verbs in Kui: those that must index S and those that do not need to, such as the verbs in (130) and (131) respectively. On all intransitive prefixed verbs, the prefix is obligatory. This is so for the verb -rik ‘sick’. In (130a) the third person prefix appears coindexing the proper noun and the only argument, Cucu. Without it the sentence is ungrammatical as in (130b). The only type of prefix used on intransitive verbs is the patientive. However, there are so few examples that one cannot draw any meaningful conclusions from this. Indeed, alongside -rik only awar ‘return’, -men ‘fear’, -mola ‘don’t want’, and -miyaga ‘want’ have been found.

Kui

(130)

a.

155

Cucu ga-rik. Cucu 3.PAT-sick ‘Cucu’s sick.’

b. *Cucu rik. Cucu sick Not good for: ‘Cucu’s sick.’ The class of unprefixed intransitive verbs vastly outnumbers the other (many of these may be prefixed, but this creates a transitive predicate, see section 6.2.2–4). A selection of such verbs is given in Table 16. Table 16: Intransitive verbs without prefixes snam- ‘hot’

sar ‘drip, flow’

tai ‘swim’

baka ‘bad’

ak ‘open’

mai ‘come’

isa ‘lie’

ganingin ‘angry’

agug- ‘yawn’

laak ‘walk’

bat- ‘big’

lol ‘stay’

bik ‘strong’

luk ‘squat’

misa ‘sit’

min ‘die’

An example of yai ‘go’ and mop ‘sleep’ appear in (131): no prefix is present to agree with S in both cases. (131)

a.

Pi yai. 1PL . INCL . SBJ go ‘We’re going home.’

b.

Ø-en mop=mo. 2SG -eye sleep=CONT ‘You’re tired.’ (lit. ‘Your eyes are sleeping.’)

6.1.2 Agreement on transitive verbs There are two classes of underived transitive verbs in Kui: (i) those that cannot index P (with a change in valency, see section 6.2.5), and (ii) those that must. Only P on transitive verbs is ever prefixed in Kui. This section is largely based on Shiohara (2010).

156

Glenn Windschuttel & Asako Shiohara

Unprefixed transitive verbs cannot show P agreement. Alongside the A argument in the following examples another argument is manifest in a NP, in (132a), and a pronoun, in (132b). (132)

a.

Na atako 1SG . SBJ rice ‘I plant rice.’

b.

Na gai karap. scratch 1SG . SBJ 3 ‘I scratch him.’

méd-i. plant-PFV

A selection of unprefixed transitive verbs is given in Table 17. Table 17: Unprefixed transitive verbs akaal ‘eat’

méd ‘plant’

alon ‘write’

mira ‘enter’

blés ‘hit’

mitapai ‘stab’

baresi ‘tear’

magi ‘listen’

el ‘buy’

on ‘cook’

gakawai ‘work on’

pod ‘climb, ride’

karap ‘scratch’

purin ‘throw away’

gamain ‘make’

séi ‘go down’

garoti ‘cut’

téri ‘wear’

goi ‘pick (fruits)’

usahate ‘fight’

mani ‘wear’

langaberi ‘win’

In the case of prefixed transitive verbs the verb root is bound. The prefix for each verb is lexically specified uniquely as either patientive or dative. Verbs do not alternate in their form of agreement. An example of a verb from each class, dative and patientive, is given in (133a) and (133b) respectively. (133)

a.

Na Siti 1SG . SBJ Siti ‘I heal Siti.’

go-naborin. 3.DAT-heal

b.

Na Siti 1SG . SBJ Siti ‘I push Siti.’

ga-dom. 3.PAT-push

Kui

157

Some verbs of these two classes are presented below in Table 18. Table 18: Obligatorily prefixed transitive verbs Patientive -ban ‘wait’

DATIVE -domi ‘refuse’

-gayei ‘release, make free’

-baret ‘tell’

-yoki ‘shake’

-kaki ‘crash’

-barin ‘kill’

-leki ‘count’

-lul ‘call’

-bik ‘pull’

-muni ‘kiss’

-naborin ‘heal’

-bit ‘kick’

-pater ‘chase’

-rap ‘look for’

-bot ‘shoot’

-puni ‘catch’

-mon ‘help’

-dom ‘push’

-teki ‘bear (a baby)’

-es ‘bite’

-ui ‘see’

-yadi ‘bear (a baby)’

-kabil ‘roll’

6.1.3 Agreement on ditransitive verbs There are two underived ditransitive verbs in our corpus, -dingin ‘borrow’ and -ei ‘give’. These are both obligatorily prefixed, -dingin with a prefix from the dative paradigm and -ei with a prefix from the patientive paradigm. In both cases R is indexed by the prefix and not T. An example of -dingin appears as (134a) below and two examples of -ei are in (134b) and (134c). (134)

a.

Na sen go-dingin. 1SG . SBJ money 3.DAT-borrow ‘I borrow money from him.’

b.

Na sen g-ei. 1SG . SBJ money 3.PAT-give ‘I give him money.’

c.

Ool omo doi gala na-maa child DIST money gala 1SG -father ‘That child gave money to my father.’

g-ei. 3.PAT-give

Our understanding of the alignment of underived Kui verbs from the viewpoint of agreement can be summarized as follows. One of the obligatory ‘absolutive’ arguments (P, R or S) can be indexed on a verb by a prefix in an AGR slot. Whether agreement with one of these arguments does appear is lexically determined: verbs that do prefix always do and verbs that don’t don’t (without deriving

158

Glenn Windschuttel & Asako Shiohara

a predicate with a different valence pattern; see section 6.2). Thus they form two disjunctive sets, prefixed and unprefixed. For P and R there are two forms of agreement, patientive and dative, the presence of which also classifies each verb uniquely: each prefixed verb may only use one or the other. For S only patientive agreement is known to appear.

6.2 ‘Derivational’ uses of agreement prefixes The description in the preceding paragraph is only valid for the basic prefixation of verbs. Prefixes may be used to alter the transitivity of the verb by increasing its valency (or perhaps even decreasing the valency of the verb, see section 6.2.1). Unprefixed verbs, both intransitive and transitive, are prefixed to index an additional participant patientively or datively (partially according to the semantics of the participant). This agreement is generally optional, but in the same sense as above in section 6.1 it is obligatory in the new valence frame.

6.2.1 Intransitive uses of obligatorily prefixed transitive verbs There are two obligatorily prefixed transitive verbs which may be used intransitively. We know that an obligatory prefix references P where these verbs are used transitively. The verbs are -kabil ‘roll’ and -wel ‘flow, wash’. Their basic transitive uses are seen in (135a) and (136a) respectively. These verbs may be used intransitively: -wel in (135b) and -kabul in (136b). Note that in these cases a prefix still appears, ta- ‘CMN ’. Here, the prefix has no apparent referent and is apparently only a place holder to fulfill the requirement of the verb for a prefix. This sentence is truly intransitive, distinct from the reflexive pattern of the language. Indeed, we know how the verb -wel ‘flow, bathe’ behaves under reflexivization from example (135c). No special marking is used and the prefix simply encodes the person-number features of the patient-cum-agent; this is distinct from the ta- pattern of (135b). (135)

a.

Na ool ga-wel. 1SG . SBJ child 3.PAT-wash ‘I bathe a child.’

b.

Er ta-wel. water CMN . PAT-flow ‘The water flows (out of a hose).’

Kui

(136)

c.

Ool ga-wel. child 3.PAT-wash ‘A child bathes.’

a.

Wor gala wor siwa ga-naya ga-kabil-i. rock gala rock three 3-between 3.PAT-roll-PFV ‘(I) rolled the rock among the three rocks.’

b.

Wor ta-kabul nabanta la. ON rock CMN . PAT-roll bridge ‘The rock rolled over the bridge.’

159

6.2.2 Intransitive verbs taking valency-increasing patientive agreement prefixes Intransitive verbs may have their valency increased by adding an agreement prefix. This is only known for unprefixed intransitive verbs. Dative and patientive prefixes are both attested in this alternation depending, to a certain extent, on the semantics desired. The patientive prefix is the only agreement prefix that may causativize a verb. A in the new clause is the causative agent and P corresponds semantically to S of the basic intransitive clause. How productive the alternation is remains still to be seen. The examples below are pairs of clauses headed by the same verb: tai in (137) and mai in (138). The first sentence in each instantiates the intransitive use of the verb while in the second the verb has been causativized by the patientive prefix. (137)

(138)

a.

Ga tai. 3.SBJ swim ‘She swims.’

b.

Na Siti ga-tai. 1SG . SBJ Siti 3.PAT-swim ‘I make Siti swim.’

a.

Awe mai. flood come ‘It flooded.’

b.

Buk ga-mai. book 3.PAT-come ‘Pass me the book.’

160

Glenn Windschuttel & Asako Shiohara

However, with certain verbs patientive prefixation does not causativize the verb, but increases the valency of the predicate in other ways, i.e., some form of applicativization. In such cases, S of the intransitive verb fulfills the same semantic role in the expanded predicate as it does in the intransitive clause. The prefix introduces a new participant, often non-patientive in meaning representing a highly prominent semantic role such as benefactive. In (139) and (140), the (a) examples display the intransitive use of the verb, while the (b) examples have the transitive use of the prefixed verb. (139)

(140)

a.

Misa=té! sit=AND ‘Take a seat!’

b.

Na ga-misa. 1SG . SBJ 3.PAT-sit.down ‘I sit down waiting for him.’

a.

Yainté a sawai? when 2SG . SBJ go ‘When did you arrive?’

b.

Na ga-sawai. 1SG . SBJ 3.PAT-go ‘I go and see him.’

6.2.3 Intransitive verbs taking valency-increasing dative agreement prefixes On intransitive verbs the dative prefix introduces a semantically prominent participant. The referent of the S argument and the A argument are the same, i.e., the prefix does not causativize the verb. In this way, it is similar to the second set of uses of the patientive prefix discussed in section 6.2.2. In every case the dative participant is human. Sometimes it alternates with the applicative prefix u- when an inanimate participant performs the same semantic role (see section 6.3.1). The following two pairs in (141) and (142) are examples of two intransitive verbs, baka ‘bad’ and talék ‘fight’. In each example, an intransitive sentence is followed by a sentence where a dative prefix has expanded the predicate. The dative prefix in both cases represents a highly prominent thematic role: in (141b) an experiencer and in (142b) a patient/agent type role.

Kui

(141)

(142)

a.

G-en baka. 3-eye bad ‘He’s blind.’

b.

Tabak bunuk no-baka. tabacco smoke 1SG . DAT-bad ‘Cigarette smoke is bad for me.’

a.

Kuiwas Murwas alé gé talék. Kuiwas Murwas 3DU 3.FOC fight ‘Kuiwas and Murwas are fighting.’

b.

Kuyas nun nyo-talék-i. 1PL . EXCL . DAT-fight-PFV Kuyas PL ‘The Kuyas fought with us.’

161

6.2.4 Intransitive verbs taking both forms of valency-increasing agreement prefixes There are situations where both dative and patientive prefixes appear on the one verb in a single clause. In such situations the changes brought about by each type of prefix are simply compounded. The patientive prefix causativizes the verb as described in section 6.2.2 and a beneficiary type participant is added by the dative prefix. This alternation is displayed in (143): the stative situation denoted by ak in (143a) is causativized in (143b). Then in (143c) a beneficiary is added by the dative prefix. The intransitive use of ber ‘get out’ is displayed in (144a). In (144b) this is causativized by the patientive prefix and a beneficiary is added by the dative prefix. These are the only known examples of such agreement stacking in our corpus. (Interestingly, gala is not used in (144b) although three participants are present as NPs; the implications of this will be left to future research). (143) a. Nabad ak-a. door open-IPFV ‘The door is open.’ b. A-g-ak. EMPH -3.PAT-open ‘(I) open (my mouth) wide.’ c. Nabad go-g-ak-i. door 3.DAT-3.PAT-open-PFV ‘(I) opened the door for him.’

162

Glenn Windschuttel & Asako Shiohara

(144) a. Mester Glenn ga-tuk gala kadera ga-naya mi-i ber baka. Mister Glenn 3-leg gala chair 3-between IN -put get.out bad ‘Glenn put his foot in between the chair legs and can’t get out.’ b. Mester Glenn ga-bad ga-tan go-ga-ber-i. Mister Glenn 3-shirt 3-hand 3.DAT-3.PAT-get.out-PFV ‘Glenn let his hand get out of his shirt’ i.e. ‘Glenn undid his sleeve.’ 6.2.5 Transitive verbs taking valency-increasing agreement prefixes Analogous to the intransitive examples above there are transitive verbs which can have their valency increased to become ditransitive by the affixation of an agreement prefix. The only two verbs known to undergo this change are ulik ‘sell’ and takau ‘steal’ which are otherwise transitive of the unprefixed type. This is demonstrated by (145) and (146) respectively. The verb ulik ‘sell’, which is transitive in (145a), can have the recipient specified by a patientive prefix as exemplified by (145b). The transitive verb takau ‘steal’ in (146a) may take a dative prefix to index the malefactive or source as in (146b). This follows the pattern of basic ditransitives where it is always R that is indexed. (145)

(146)

a.

Ga nawol ulik-i. 3.SBJ banana sell-PFV ‘He sold bananas.’

b.

Ga nawol na-ulik-i. 3.SBJ banana 1SG . PAT-sell-PFV ‘He sold me bananas.’

a.

Na ga-sen takau. 1SG . SBJ 3-money steal ‘I steal his money.’

b.

Na go-takau. 1SG . SBJ 3.DAT-steal ‘I steal something from him.’

6.3 Applicative prefixes An applicative by definition introduces an extra participant to the clause. The semantics of this participant are dependent on the particular prefix. Typically the verb to which an applicative is prefixed is intransitive. The verb laak ‘walk’

Kui

163

is intransitive as illustrated by (147a). In (147b) it is prefixed with u- and has two arguments. In (148a) the same pattern is evident. There is an applicative prefix and two arguments. However, this is a labile verb and can be used as a threeplace predicate as well. This is seen in (148b) (in such a case gala is used). (147)

(148)

a.

Yainté a laak-i? when 2SG . SBJ walk-PFV ‘When are you going?’

b.

Mester Glenn mamplel dét u-laak, a-brék-i. Mister Glenn mango leaf APPL-walk EMPH -crunch-PFV ‘Glenn stepped on the mango leaf so it crunched.’

a.

Atei atobun kalat u-gab-a. tree cut basket APPL-lean-IPFV ‘A chopped down stick is leaning on the basket.’

b.

Mester Glenn bangku gala ga-tuk Mister Glenn bench gala 3-leg ‘Glenn leaned the bench on his leg.’

u-gab-i. APPL -lean-PFV

An applicative prefix may also be added to a transitive verb. In such a situation no extra participant is introduced. In this usage, the prefix is not, of course, a true applicative but it is similar in other ways. Beyond this, the behavior of the resulting predicates depends on whether the verb is obligatorily prefixed or not. If the verb is unprefixed (in the sense of section 6.1.2) the prefixes have a particular discourse/semantic function. The already subcategorized P argument acts as if it is an argument newly introduced: the argument has the same discoursively backgrounded/less affected position of applicative Ps. The verb nai ‘drink, eat soft foods’ is a transitive verb of the unprefixed type as exemplified by (149) below. In the examples in (150) there are two verbs, each with one NP as P. The verb nai in the bracketed clause is prefixed with an applicative prefix, u- and la- respectively. In both examples the two clauses were translated together as one clause with the second P as a prepositional object and thus backgrounded in the discourse context. (149)

Batarma nai. rice eat ‘Eat rice.’

164 (150)

Glenn Windschuttel & Asako Shiohara

a.

Lalama nai [masa u-nai.] rice.porridge eat sambal APPL -eat ‘Eat rice porridge with sambal.’ (lit. ‘Eat rice porridge, eat sambal.’)

b.

Nawol dagar akaal-i [kopi banana fried eat-PFV coffee ‘Eat fried banana with coffee!’

er-la-nai!] water-ON -drink

In the case of obligatorily prefixed verbs (see section 6.1.2) the agreement prefix can be replaced by an applicative prefix. The verb -ui ‘see’ is an obligatorily prefixed verb and a bound root as exemplified by (151a). In both examples in (151b) and (151c) the verb is prefixed with an applicative prefix and no agreement prefix is necessary. (151)

a.

Ga alela marei ga-ui [ah] ji 3.SBJ above go.up 3.PAT-see ‘uh’ 3PL . SBJ ‘They went up above, saw it and they went.’

b.

Mainsino angain nélabakan aro Kerikla aro HIGH Kerikla HIGH then tomorrow morning ‘Then tomorrow morning look up at Kerikla up there.’

c.

bei=mo. go=CONT la-ui. ON -see

Kalol mi-ui. new.sign in-see ‘They saw a new sign.’

These examples also provide evidence for the prefixhood of mi- and la(since they are fulfilling the need of the bound verb root -ui for a prefix), which, nonetheless, remains tentative. For la- one more argument is available. It may appear after the incorporated noun as in (152); the noun and verb in ernai are otherwise tightly bound (see section 6.5). (152)

Nawol dagar akaal-i [kopi banana fried eat-PFV coffee ‘Eat fried banana with coffee!

er-la-nai!] water-ON -drink

6.3.1 uThe applicative u- introduces a participant with no direct effect on the action of the predicate. The introduced argument can be quite varied, but it is always inanimate: an instrument in (153), a location in (154), or a reason in (155).

Kui

(153)

a.

Ga tai. 3.SBJ swim ‘She swims.’

b.

Wataluk u-tai. APPL -swim pillow ‘Swim with bouyancy device.’

(154)

Kubit Botak ga-ten u-sol-i. red.ant Botak 3-hand APPL-crawl-PFV. ‘Red ants crawled on Botak’s hand.’

(155)

WC u-dingan-i. toilet APPL-fight-PFV ‘(They) were fighting about the toilet.’

165

The animacy restriction is clear from example (156): in (156a) the verb -yom ‘watch’ has an applicative prefix to encode the stimulus. If the stimulus is a person the applicative prefix cannot be used as in (156b). The grammatical way of expressing (156b) is given in (156c) where the dative prefix encodes the stimulus. (156)

a.

Na televisi u-yom-i. 1SG . SBJ television APPL-watch-PFV ‘I watched television.’

b. *Na bapak u-yom-i. 1SG . SBJ father APPL-watch-PFV Not good for: ‘I watched father.’ c.

Na bapak go-yom-i. 1SG . SBJ father 3.DAT-watch-PFV ‘I watched father.’

There are two examples of u- prefixed verbs which do not fit the pattern thus far described. There are no selectional restrictions: it may be human as in (157). The verb shares only a relatively opaque semantic relation to the unprefixed verb. The prefix u- added to min ‘die’ produces umin ‘forget’ and umai ‘remember’ is derived from mai ‘come’. In both cases the semantic connection is quite unpredictable.

166 (157)

Glenn Windschuttel & Asako Shiohara

Na gai u-mai. APPL -come 1SG . SBJ 3 ‘I remember him.’

6.3.2 laThe applicative prefix la- ‘ON ’ originates in the locative verb la meaning ‘be on’, illustrated in (158). (158)

Mainsino ei la nun maran takat-a go.up dry-IPFV then boat ON PL ‘They went out of the boat onto the shore.’

la. ON

Applicative la- again introduces an argument similar to the P of the homophonous independent verb. For example, the motion undergone by the elided subject in (159a) starts ‘on the mountain’. Other examples of prefixation by lafollow in (159). The separation of the location from la- by the theme in (159c) provides evidence that la- is not a postposition in keeping with the prefix analysis followed here. There is also an instance of a non-physical use of la- (159d) which was already discussed in section 6.3 above. (159)

a.

Dol la-yai. mountain ON -go ‘Go from the mountain.’

b.

Kumba atei tuk la-roka pot tree short ON -upside.down ‘The pot is on the stump upside down.’

c.

Atei tuk sél ta-pola la-isa. tree short rope CMN . PAT-fold.up ON -lie ‘The rope is folded up on the stump.’

d.

Nawol dagar akaal-i [kopi banana fry eat-PFV coffee ‘Eat fried banana with coffee!’

isa. lie

er-la-nai!] water-ON -drink

6.3.3 miThe applicative prefix mi- ‘IN ’ is grammaticalized from the locative verb mi and denotes ‘be inside, in’. This verb is illustrated without aspectual suffixation in (160a) and with imperfective suffix in (160b).

Kui

(160)

a.

Alur mei Mor mi. IN part come Moru ‘Part of them came and were in Moru.’

b.

Atako kalat mi-a. cassava basket IN - IPFV ‘The cassava is in the basket.’

167

These same physical uses are represented when the morpheme appears as a verbal prefix. As an applicative prefix in (161a) mi- adds the sense that the applied object, g-en ‘his eye’, contains the subject palak ‘earth, dirt’. The preposing of the location, separated from mi- by atako tuk arok in (161b), shows that this is not a postposition, which supports a prefix analysis. (161)

a.

G-en palak mi-yai. 3-eye earth IN -go ‘He got dirt in his eye.’ (lit., ‘Earth went in his eye.’)

b.

Kalat atako tuk arok mi-isa. basket yam small two IN -lie ‘Two small yams are in the basket.’

In addition mi- has non-applicative uses which do not show a clear relation to its basic function. This prefix is the means by which one forms comparatives in Kui. The standard of comparison appears following the verb, as in (162). (162)

Na mi-bat-a bapak Djumat. 1SG . SBJ IN -big-IPFV father Djumat ‘I’m bigger than Pak Djumat.’

In other contexts the prefixation of mi- signals that the action is generally experienced, that is, the action is not limited to any specific set of individuals but is experienced by everyone around. This is the case below in (163): tawel meaning ‘flow’ in (163a) is prefixed in (163b) to indicate that blood was flowing everywhere, that all the people in the battle were bleeding; similarly, the dust was being thrown up into the air everywhere. The verb takun in (164a) indicates something that produces light is not giving out light i.e. it is ‘off’. With the prefix mi- and without S this verb forms the usual expression for ‘dark’ as in (164b) i.e. all lights are off.

168 (163)

(164)

Glenn Windschuttel & Asako Shiohara

a.

Er ta-wel. water CMN . PAT-flow ‘The water flows (out of a hose).’

b.

Arui mi-ta-bon-i; we dust IN - CMN . PAT-throw.out-PFV; blood ‘Dust was raised; blood was everywhere.’

a.

Lampu takun-i. light dark-PFV ‘The light is off.’

b.

mi-ta-wel-i. IN - CMN . PAT-flow-PFV

Mi-takun-a. IN -dark-IPFV

‘It’s dark.’

6.4 Emphatic prefix The emphatic prefix depicts a situation where the action denoted by the verb cannot be increased further: it is taken to the greatest degree possible. The following pairs of examples give a verb with and without the prefix. In (165a) the unprefixed verb simply states that the box had been opened while in (165b), because of the prefix a-, the mouth is not only open but wide open. The use of gamain without the prefix in (166a) simply denotes building or constructing something. The verb with the emphatic prefix has a lexicalized use meaning that the day is completely over in (166b). The verb séi means ‘descend’ as in (167a). With the emphatic prefix in (167b) the existence of a particular destination is implied and reached. In this case, the bottle has turned upside down so that it is resting on the mouth. (165)

a.

b.

Gapatei kadin g-ak-i. box small 3.PAT-open-PFV ‘He opened the small box.’ A-g-ak. EMPH -3.PAT-open

‘I open my mouth wide.’ (166)

a.

O gamain taka gamain. house make seating.platform make ‘They built a house and constructed a platform to sit on.’

Kui

b.

169

A-gamani. EMPH -do

‘It is done, at an end.’ (167)

a.

Eg séi go-yom-i. 2SG . POSS descend 3.DAT-watch-PFV ‘You went and looked at him.’

b.

Botil kalat mi-nasén ga-wa a-séi. bottle basket IN -stand 3-mouth EMPH -descend ‘The bottle is in the basket pointed downwards (its mouth having descended all the way down).’

6.5 Noun incorporation There are three verbs in our corpus which have a noun that regularly appears prefixed to the verb. The noun and verb form a phonological unit with only one syllable bearing word level stress e.g. busai [buˈsaj], ernai [erˈnaj]. In one case mentioned in section 2.5.3 further reductions are seen. This is busai ‘chew betelnut’ in (168) which incorporates the noun bui ‘betelnut’ (the verb sai is reported by speakers to also mean ‘chew betelnut’). Ernai in which the noun er ‘water’ is incorporated is the regular expression for drinking (the verb nai has a wider meaning covering eating soft foods as well as drinking). These verbs are exemplified in (168) and (169). The intransitive verb tai ‘swim’ in (170) can be prefixed with tan ‘ocean’ to specify that the swimming takes place in the ocean. (168)

Ga 3.SBJ

balan LEVEL

akaal-i eat-PFV

ga 3.SBJ

bu-sai betel.nut-chew

ga 3.SBJ

nok-i. finish- PFV ‘He ate over there, chewed betel nut and drank.’ (169)

Nawol dagar akaal-i kopi banana fry eat-PFV coffee ‘Eat fried banana with coffee!’

(170)

Tan-tai. ocean-swim ‘Swim (in the ocean).’

er-la-nai! water-ON -drink

er-nai water-drink

170

Glenn Windschuttel & Asako Shiohara

7 Serial verb constructions As in other TAP languages, serial verb constructions (SVCs) are common in Kui. Serial verbs occur under a single intonation contour. For some SVCs one aspectual suffix has scope over a whole SVC and the presence of two aspectual suffixes is not grammatical. An example of an aspectual SVC (see section 7.5) where the scope of the aspectual suffix extends to the whole SVC is given in (171a). The first verb maman cannot bear the perfective suffix despite its perfective semantics; the ungrammaticality of this is shown by (171b). (171)

a.

Maman nok-i? eat finish-PFV ‘Finished eating?’

b. *Maman-i nok-i? finish-PFV eat-PFV Not good for: ‘Finished eating?’ SVCs form a unit in the syntax of the language. The creation of a SVC can have effects on the syntactic behavior of its elements. The verb awor ‘return’ is an obligatorily prefixed verb (see section 6.1.1). An example is given in (172a). In (172b) it is unprefixed; it is only because it is part of a manner SVC that this is grammatical. (172)

a.

b.

Na gap n-awor 1SG . SBJ NEG 1SG . PAT-return ‘I’m not going to return.’

nanga. NEG

Anin awor u-tagan-i: . . . person return APPL-ask-PFV ‘The people called back, saying: . . .’

7.1 Motion SVCs There are two classes of motion SVCs in Kui: directional and sequential SVCs. They differ in the relative timing of the events of the individual verbs that form part of the SVC. The two verbs in directional SVCs present concurrent actions. The last verb belongs to a closed class, the motion verbs given in Table 8 in section 3.1.3. This verb specifies the direction of the other verb’s action. The other verb may be

Kui

171

intransitive, for example (173). The ‘returning’ is specified as toward the DC east of the source because maran ‘go east’ follows. The main verb can also be transitive. In the example below (174) the action of the transitive verb -bik ‘pull’ is directed downwards because of the presence of séi ‘descend’ in the SVC. (173)

Mainté Walanda g-awor maran. then Dutch 3.PAT-return go.east ‘Then the Dutch returned here from the west.’

(174)

Ei ga-bik séi. boat 3.PAT-pull descend ‘Pull the boat down.’

The other form of motion SVCs, sequential SVCs, represents a series of successive actions. The verbs are iconically ordered following the order in which the events take place. In every example in our corpus the first verb is a verb with motion as a component of its meaning; the other verb denotes some other action. Hence in every instance the SVC means as a whole ‘move somewhere and do something’. The second verb may be transitive or intransitive exemplified by (175) and (176) respectively. Where it is transitive the P argument is placed between the two verbs. For example, in (175) masin lak ‘Kui’, the P, follows yai ‘go’ and precedes the transitive verb uguri ‘study’. In (176) the verb séi ‘descend’ is similarly used followed by the intransitive verb misa ‘sit’. A particularly common subclass has the second verb as a motion verb too. This is exemplified by (177) where motion to the house is followed by motion to Moru. (175) Na yai masin lak u-gur-i. language APPL -study-PFV 1SG . SBJ go Kui ‘I went to study Kui.’ (176) Aban ogo pi namata wa séi misa nanga. village PROX 1PL . INCL . SBJ mountain.people here descend sit NEG ‘We mountain people didn’t descend to be here in this village.’ (177) Pi a-o mi-yai Mor mi-yai. 1PL . INCL . SBJ 2SG -house IN -go Moru IN -go ‘We go to your house and then go to Moru.’

172

Glenn Windschuttel & Asako Shiohara

7.2 Causative SVCs Causative SVCs involve serialization of transitive gamain ‘make’ with a following intransitive verb denoting the caused event. S of this verb corresponds semantically to P of the SVC and A is causative agent. Examples (178) to (180) illustrate this causative construction. (178)

Mester Glenn kadera gamain Mister Glenn seat make ‘Glenn dropped the chair.’

(179)

A-wes nilon pak gamain a-nalin-i. EMPH -be.clean-PFV 2SG -teeth floss use make ‘You used floss to clean your teeth.’

(180)

Gamain arai. make be.crushed ‘Crush something.’

mukséi. fall

7.3 Instrumental SVCs There are a number of verbs used to introduce instruments in Kui. The most common verb used in instrumental SVCs is pun ‘hold’. Other verb-like elements that are used are pak, ua and punmai. The main verb is typically transitive and the instrument and pun ‘hold’ appear after A and before P, as in (181).

(181)

A Instrument Mester Glenn ga alat pun mister Glenn 3.SBJ device hold ‘Glenn recorded us with the device.’

P nyai 1PL . EXCL

V rekam. record

The instrumental use of verb pun is quite general in meaning, able to introduce all types of instruments. The instrumental use derives from its independent meaning ‘hold’. This use is exemplified by (182) below. In addition it can be part of a directional SVC meaning ‘carry’ (see section 7.1). Examples are given in (183a) and (183b). The instrumental use sometimes involves the act of holding as in (184a) where one must hold the knife to cut something with it. However, other sentences, for example (184b), make it clear that the verb’s instrumental use has been grammaticalized. Being transported in a car does not literally

Kui

173

involve ‘holding’ it. More than one aspectual specification is possible in these SVCs; how this may be integrated with the ungrammaticality of the example (171b) will be left to future research. (182)

A pun-a nanga. 2SG . SBJ hold-IPFV NEG ‘You’re not holding it.’

(183)

a.

Na laptop pun-a aléla 1SG . SBJ laptop hold-IPFV above ‘I carried the laptop up there.’

b.

Ga pun séi u-tol-i 3.SBJ hold descend APPL-clean-PFV ‘He took it down and cleaned it.’

(184)

a.

Duur pun kertas u-garot-i. knife hold paper APPL-cut-PFV ‘He cut the paper with the knife.’

b.

Na oto pun-a 1SG . SBJ car hold-IPFV ‘I go to Kalabahi by car.’

kalimbas Kalabahi

mira. go.up u-ai. APPL -clean

yai. go

The other instrumental constructions involving ua, pak and punmai are restricted in their semantics and are less common as well. These items are not found as independent clausal verbs but do derive from verbs. The coding of the mode of transport is the only instrumental use found for ua, for example (185). Pak introduces instruments that are physical tools, for example (186). In this example there is a deviation from the pattern of word order in instrumental SVCs displayed in (181) above; the instrument follows P rather than preceding it. Punmai is only known from one example, (187). Pak is clearly loaned from Malay pakai ‘use’. Punmai is a compound of pun ‘hold’ and mai ‘come’. The use of compounds of ‘take’ and ‘come’ to introduce obliques has precedence in other TAP languages (met-ma ‘take-come’ in Blagar, albeit with an unrelated word for ‘hold’; Klamer & Schapper 2012: 199). (185)

Ei lar ua Mor mi-yai. boat sail use Moru IN -go ‘Take the sail boat to Moru.’

174

Glenn Windschuttel & Asako Shiohara

(186)

A-wes nilon pak gamain a-nalin-i. EMPH -clean-PFV 2SG -teeth floss use make ‘You use floss to clean your teeth.’

(187)

Duur pun-mai pensil u-gai. knife hold-come pencil APPL-sharp ‘Sharpen a pencil with a knife.’

7.4 Comitative SVCs Comitative serialization involves the inflected verb to-pun ‘CMN . DAT-hold’ (independent use of which is described in section 7.3) introducing a human accompanier into the clause. In (188) and (189) to-pun denotes a second participant who accompanies the subject in the action performed by the final verb in the serialization. (188)

Na ai to-pun 1SG . SBJ 2SG CMN . DAT-hold ‘I sit with you.’

(189)

A nai to-pun 2SG . SBJ 1SG CMN . DAT-hold ‘You fight with me.’

misa. sit

ta-burin-i. CMN . PAT-fight-PFV

Inaminate referents cannot be introduced by to-pun. Thus (190a) is unacceptable. Inanimate referents are typically grouped together by simple juxtaposition as in (190b). (190)

a. *Buk HP to-pun isa. book mobile CMN . DAT-hold lie Not good for: ‘The book is with the mobile.’ b.

mi-isa. [Bal bat-a siwa]NP [bal kadin siwa]NP bar ball small three ground IN -lie ball big-IPFV three ‘Three large balls and three small balls are on the ground.’

7.5 Aspectual SVCs One type of aspectual SVC is known in Kui. The verb nok gives a sense of completion to the verb it follows. This verb can be used independently with the

Kui

175

same completive meaning as in (191). The verbhood of nok is evident from the appearance of the aspect suffix -i ‘-PFV ’ on it in (192a) and (192b). (191)

Dé nok=po. already finish=CONT ‘I’ve already got it.’

(192)

a.

Dé alon nok-i. already write finish-PFV ‘I have already written it.’

b.

Ga er-nai nok-i. 3.SBJ water-drink finish-PFV ‘He’s finished drinking.’

7.6 Resultative SVCs Resultative serialization minimally involves one verb denoting an event and a following verb denoting the situation resulting from that event. A result is typically encoded with an intransitive verb and the event leading up to it by a transitive. For instance, brék in (193) and min in (194) denote the results of the walking (ulaak) and stabbing (tak) events respectively (the example (194) is a special case as the A and P arguments are coreferential). (193)

Mester Glenn mamplel dét u-laak a-brék-i. mister Glenn mango leaf APPL-walk EMPH -break-PFV ‘Glenn stepped on the mango leaf until it was crushed.

(194)

Leer Daing ga-nal gai tak min-i. king Daing 3-REFL 23 3 stab die-PFV ‘King Daing stabbed himself to death.’

8 Aspect marking Grammaticalized aspect marking in Kui is limited to: (i) suffixes marking imperfective and perfective aspect on final verbal predicates (section 8.1); (ii) 23 The exact function and distribution of this pronoun is unknown. This is its only clear use in our corpus.

176

Glenn Windschuttel & Asako Shiohara

clause final enclitics (section 8.2); and (iii) two aspectual particles, dé ‘already’, lei ‘finished’ (section 8.3). There do not appear to be grammaticalized morphemes marking tense or evidentiality in Kui. Time can be indicated by means of temporal adverbs (e.g. dona ‘yesterday’, angain ‘tomorrow’ etc.).

8.1 Aspectual suffixes Kui has two aspectual suffixes: (i) -i ‘PFV ’, and (ii) -a ‘IPFV ’. These suffixes are in a paradigmatic relationship and do not co-occur. The use of aspectual suffixes in defining verbal predicates has already been discussed in section 3. This relation between SVCs and aspect suffixes was briefly illustrated in section 7. In this section, the semantics of the suffixes will be summarized. Marking a clause with the perfective suffix presents a situation as a complete whole. This marking is incompatible with actions presented as ongoing. The difference between a verb with this marking and an unmarked verb is suggested by the following example. The action in (195a) is punctual and a complete whole: being born happens at only one point in time and was complete at the time of speaking. The action in (195b) is incomplete and still ongoing. The meaning of the verb is construed differently, meaning something like ‘be alive’ and takes place over an open-ended time period. (195)

a.

Masin aban mi-na-yad-i. Kui village IN -1SG . PAT-born-PFV ‘I was born in the village.’

b.

Na na-yad tun 1SG . SBJ 1SG . PAT-born year ‘I’ve been alive since 1931.’

1931 1931

sampai until

yaal. now

The imperfective suffix presents the action of the verb simply as a state without a necessary beginning or endpoint. A comparison of examples of verbs with this aspect and the perfective highlights this fact. The verb with the imperfective suffix corresponds to the state resulting from the action of the verb marked with the perfective suffix. (196a) and (196b) are both transitive clauses containing the verb ak ‘open’. The difference in suffix corresponds to a clear difference in meaning. In other cases there is also a difference in valency: a causative agent is present in the perfective clause while absent in the imperfective clause. In (197) and (198) this is the case in the last two pairs of clauses; each has the

177

Kui

same verb, salaak ‘hang’ in (197a) and (197b) and gab ‘lean’ in (198a) and (198b). The only change in the verb in each pair is the difference in aspectual marking. (196)

(197)

(198)

a.

Inmei ga-wa mi-ak-i. woman 3-mouth IN -open-PFV ‘The woman opened her mouth.’

b.

Inmei ga-wa mi-ak-a. woman 3-mouth IN -open-IPFV ‘The woman has her mouth open.’

a.

Sél gala atei guwadin u-salaak-i. rope gala stick forked APPL -hang-PFV ‘He hung the rope on the forked stick.’

b.

Atei bata sél u-salaak-a. tree big rope APPL-hang-IPFV ‘The rope hangs off the big tree.’

a.

Mister Glenn bangku gala ga-tuk Mister Glenn stool gala 3-leg ‘Glenn leant the stool against his leg.’

b.

Pitan atei u-gab-a. bamboo tree APPL -lean-IPFV ‘The stick of bamboo is leaning on the tree.’

u-gab-i. APPL -lean-PFV

The imperfective meaning of the suffix -a is also made clear when two clauses are linked together. In (199) the first clause is marked with -a ‘IPFV ’. Perhaps because of the transitory nature of injury and sickness, -rik ‘sick, sore’ does not usually appear with the imperfective suffix. However, in (199) it does, ostensibly because the injury has no beginning or end during the action of the following clause. The state ‘your hand is sore’ is taken as a given in as far as the action ‘going to the hospital’ is concerned. (199)

A-tan 2SG -hand dé or

a-rik-a, 2SG . PAT-sick-IPFV

a 2SG . SBJ

lei COMPL

rumah sakit hospital

mi-yai IN -go

dénanga? still

‘With your hand being sore have you been to the hospital or not yet?’

178

Glenn Windschuttel & Asako Shiohara

8.2 Aspectual enclitics In Kui there are two enclitics, (i) =mo ‘=CONT ’, and (ii) =po ‘=CONT ’, which both mark continuative aspect and are in complementary distribution with the perfective suffix -i. Although they appear clause finally they can be distinguished from the suffixes above because they are not verbal inflection. They may occur in nonverbal clauses (200a) and in some cases following the negative marker (200b) and the imperfective suffix (200c). (200)

a.

Og

walél=mo. here=CONT ‘This is here.’ PROX

b.

Nanga=mo. NEG = CONT

‘It’s not.’ c.

Bér snam-a=mo. sun hot-IPFV = CONT ‘The sun’s currently hot.’

The continuative semantics of the examples in (201) are clear: the clitic is often translated into Alor Malay by ada (this forms part of a continuative construction common in this variety of Indonesian). The enclitic =po has a similar meaning to =mo being used for situations that are in process; it is frequent in questions as in (202). (201)

(202)

a.

Ø-en mop=mo. 2SG -eye sleep=CONT ‘You’re sleepy.’

b.

Na u-sakola=mo. 1SG . SBJ APPL-school=CONT ‘I’m studying (it).’

c.

Mister ga-yér=mo. mister 3.PAT-laugh=CONT ‘They’re laughing at the white guy.’

Nal gaman=po? what do=CONT ‘What are you doing?’

Kui

179

8.3 Aspectual particles Kui has two aspect particles, (i) dé marking perfective aspect (section 8.3.1) and (ii) lei marking completive (section 8.3.2). These are used in different contexts though they sometimes co-occur. The individual properties of each aspect marker are discussed in the following sections. 8.3.1 dé ‘already’ The perfective aspect particle in Kui is dé ‘already’. It is used in reference to an event that has ‘already’ been achieved by a reference time. Typically dé occurs directly before the predicate, for example, a-mai ‘come’ in (203a). The predicate need not be verbal as in (203b) where it is nominal. (203)

a.

Bér dé a-mai. sun already EMPH -come ‘The sun had already set.’

b.

Anin dé malél. person already there ‘The person was already there.’

When dé occurs in a negatively marked clause, the overall meaning is ‘not yet’. Dé immediately precedes gap (the first part of the bipartite negative) in these cases. This is illustrated in (204). (204)

Dé gap to-mug already NEG CMN . DAT-meet ‘We haven’t met yet.’

nanga. NEG

Dé along with the second part of the negative forms the complex adverb dénanga. This compound word has the meaning ‘still’ and occurs preverbally as in (205a). In tag questions it is equivalent to English ‘not yet?’ as in (205b). (205) a. Dénanga bik-a=mo. still strong-IPFV = CONT ‘I’m still full.’ b. A-tan a-rik-a, a lei rumah sakit mi-yai IN -go 2SG -hand 2SG . PAT-sick-IPFV 2SG . SBJ COMPL hospital dé dénanga? or still ‘With your hand being sore have you been to the hospital or not yet?’

180

Glenn Windschuttel & Asako Shiohara

8.3.2 lei ‘COMPL’ The completive aspect marker in Kui is lei ‘COMPL ’. It is used in reference to an event that has been completed and finished. For instance, in (206a) lei denotes that the cooking was finished before the person ate. Lei is typically clause final, as in (206a) and (206b). However, it may also appear earlier in the clause typically after A and before P as in (206c). (206) a. Ga on-y lei nok~nok ga akaal-i lei. 3.SBJ cook-PFV COMPL finish~RDP 3.SBJ eat-PFV COMPL ‘He cooked and ate.’ b. Jadi malél aban gamani lei. then there village make COMPL ‘Then they finished building the village there.’ c. A-tan a-rik-a, a lei rumah sakit mi-yai IN -go 2SG -hand 2SG . PAT-sick-IPFV 2SG . SBJ COMPL hospital dé dénanga? or still ‘With your hand being sore have you been to the hospital or not yet?’ A comparison with other AP languages (e.g. Wersing) indicates the particle lei has been grammaticalized out of a verb, though no trace of its former word class remains in Kui. This particle is on its way to become a conjunction meaning ‘then’. It is often collocated with Malay terus of the same meaning as in (207). This is also an example where the particle co-occurs with dé, the aspect particle discussed in section 8.3.1. It is still unclear how the meanings of the two particles combine. (207)

Mainsino then

dé already

sawai arrive

lei, COMPL

terus then

anin person

bei go

damat door

u-tér. APPL -close

‘So they already arrived and then someone went and closed the door.’

9 Discussion This is a preliminary description relying on a small corpus. Nonetheless, it is complete enough to allow comparison with the other languages of the TAP family.

Kui

181

Both Kui’s phonology and morphology share much in common with other languages of West Alor; the syntax is largely the same as that of other Alor and Straits languages. Kui’s array of four palatal phonemes is unusual among the TAP languages, with most other languages having only one or two in this place. Like palatals in other languages, their limited appearance in the languages indicates that they are recent phonemicizations. The inventory of vowel phonemes is similar to that of other West Alor languages in that it has contrastive mid-vowel phonemes, but this distinction is not as developed in Kui as in some of its neighbors, such as Klon or Kafoa. The phonological interactions between the perfective suffix and the root, resulting from palatalization, metathesis, and high vowel deletion are notable. The morphology is of typical complexity for western Alor: alongside the prefixal pronominal morphology on verbs and nouns (section 6.1 and section 4.4 respectively) shared by other languages of Alor-Pantar there are a number of applicative prefixes on verbs (section 6.3). Kui verbal agreement displays similar lexical splits in alignment to other languages in Alor. The transitive verbs fall into three mutually exclusive lexical classes: a dative, or patientive prefix coreferent with P, or none at all. Thus there is a split in the marking of P, as is common in all parts of Alor. Intransitive verbs separate into two classes according to whether a patientive prefix is needed or not. Kui is not known to have verbs roots that can change prefixal classes. Obligatorily dative prefixed verbs cannot appear with a patientive prefix and vice versa. Moreover, unprefixed verbs may not be prefixed and the reverse is also impossible in the same way. However, these prefixes do have valency increasing functions which allow their use on non-prefixing verbs to add an argument. While many other languages of Alor-Pantar share some of the applicative prefixes (u- in Klon, Baird 2008: 199; Adang, Haan 2001: 307; un- frozen in Teiwa, Klamer 2010: 112; mi- in Klon, Baird 2008: 211; la- in Adang, Haan 2001: 287) the existence of all three in the one language is unique to Kui.

Acknowledgments Thanks are owed from the first author to Pak Djumat Sanga, Pak Mas Sanga, Ibu Santri, Antoinette Schapper and Marcelinus Y.F. Akoli. Beyond this two anonymous reviewers provided valuable comments that informed both the analysis and presentation of the data in this sketch. The section on Kui stress was vastly improved through the help of Bill Palmer, Debbie Loakes and the attendees of the CoEDL summer school session on acoustic phonetics. Of course all mistakes remain our own.

182

Glenn Windschuttel & Asako Shiohara

References Akoli, Marcelinus F. 2013a. Fonologi Bahasa Kui, Alor – Indonesia [The phonology of Kui, Alor – Indonesia]. Paper presented at Annual Report Meeting of the Indonesian Institute of Science (LIPI), November 15. Akoli, Marcelinus F. 2013b. Morfologi Bahasa Kui, Alor – Indonesia [The morphology of Kui, Alor – Indonesia]. Paper presented at Annual Report Meeting of the Indonesian Institute of Science (LIPI), August 15. Baird, Louise. 2008. A grammar of Klon: A non-Austronesian language of Alor, Indonesia. Canberra: Pacific Linguistics. Donohue, Mark. 1997. Developments of affectedness marking in the languages of Alor (split, splat). Paper presented at University of California Berkeley, September 25. Haan, Johnson. 2001. The grammar of Adang, a Papuan language spoken on the island of Alor, East Nusa Tenggara, Indonesia. PhD thesis, University of Sydney. Hägerdal, Hans. 2010. Van Galen’s memorandum on the Alor Islands in 1946. An annotated translation with an introduction. Part 1. HumaNetten 25. 14–44. Holton, Gary. 2010. Kui wordlist. Manuscript. Holton, Gary, Marian Klamer, František Kratochvíl, Laura Robinson & Antoinette Schapper. 2012. The historical relation of the Papuan languages of Alor and Pantar. Oceanic Linguistics 51. 86–122. Huber, Juliette. 2011. A grammar of Makalero: A Papuan language of East Timor. PhD thesis, Leiden University. Katubi (ed.). 2011. Etnografi Kebahasaan dan Kebudayaan Orang Kui di Alor, Nusa Tenggara Timur [Ethnography of the language and culture of the Kui, Alor, East Nusa Tenggara]. Jakarta: LIPI Press. Katubi (ed.). 2012. Pemertahanan Bahasa Kui di Alor, Nusa Tenggara Timur: Kajian Vitalitas Etnolinguistik dan Agen Pemertahanan Bahasa [Language maintenance among the Kui, Alor, East Nusa Tenggara: Ethnolinguistic vitality and agents of language maintenance]. Jakarta: LIPI Press. Katubi. 2013. Lego-Lego of Alor people in East Nusa Tenggara, Indonesia: The expression of ancestor’s experience and language maintenance. In Proceedings of the International Workshop on ‘Special Genres’ in and around Indonesia. 59–76. Tokyo: Research Institute for Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies. Katubi, Thung Ju Lan & Marcelinus F Akoli. 2013. Kamus kecil bahasa Kui [A short dictionary of the Kui language]. Jakarta: Pusat Penelitian dan Kemasyarakatan dan Kebudayaan PMB, Lembaga Ilmu Pengetahuan Indonesia. Klamer, Marian. 2010. A grammar of Teiwa. Berlin: De Gruyter Mouton. Klamer, Marian. 2014. Kaera. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 97–146. Berlin: De Gruyter Mouton. Klamer, Marian & Antoinette Schapper. 2012. ‘Give’ constructions in the Papuan languages of Timor-Alor-Pantar. Linguistic Discovery 10. 174–207. Klamer Marian, Antoinette Schapper, Greville Corbett, Gary Holton, František Kratochvíl & Laura C. Robinson. 2014. Numeral words and arithmetic operations in the Alor-Pantar languages. In Marian Klamer (ed.), The Alor-Pantar Languages: History and typology, 337–373. Berlin: Language Science Press.

Kui

183

Kratochvíl, František. 2007. A grammar of Abui: A Papuan language of Alor. PhD thesis, Leiden University. Kratochvíl, František. 2014. Sawila. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 351–438. Berlin: De Gruyter Mouton. Lichtenberk, Frantisek. 2000. Inclusory Pronominals. Oceanic Linguistics 39. 1–32. Schapper, Antoinette. 2014. Elevational systems in Alor-Pantar languages. In Marian Klamer (ed.), The Alor-Pantar languages: History and typology, 247–284. Berlin: Language Science Press. Schapper, Antoinette & Marian Klamer. 2014. Numeral systems in the Alor-Pantar languages. In Marian Klamer (ed.) The Alor-Pantar languages: History and typology, 285–336. Berlin: Language Science Press. Shiohara, Asako. 2010. Verb types and semantic roles of person markers of Kui, Alor. Paper presented at Workshop on the Languages of Papua 2, 8–12 February. Stokhof, W.A.L. 1975. Preliminary notes on the Alor and Pantar languages (East Indonesia). Canberra: Pacific Linguistics. Willams, Nicholas. this volume. Kula. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 2. Berlin: De Gruyter Mouton. Windschuttel. ms. TAP Pronominals chapter in PhD thesis. University of Newcastle, Newcastle, NSW.

Nicholas J. Williams

4 Kula 1

The language scene

2 2.1 2.1.1 2.1.2 2.1.3 2.2 2.2.1 2.2.2 2.3 2.4 2.5 2.6 2.6.1 2.6.2 2.6.3 2.7

188 Phonology 188 Consonants 189 Consonant phoneme minimal pairs 190 Consonant allophony 190 Marginal consonant phonemes 193 Vowels 193 Vowel minimal pairs 194 Vowel allophony 196 Phonotactics 198 Syllable structure 200 Stress 201 Morphophonemics 201 Nasal assimilation 202 Word-shape alternations 204 Vowel harmony 205 Orthography

3 3.1 3.2 3.2.1 3.2.2 3.2.3 3.2.4 3.3 3.4

205 Basic clausal syntax 205 Verbal predicates 208 Locative and existential predicates 208 Locative/existential verb míya 209 Posture verbs 210 Other existential verbs 211 Specifying location 212 Nominal predicates (equational clauses) 213 Negation

4 4.1 4.1.1 4.1.2 4.2 4.2.1 4.2.2

213 Noun phrases 215 Attributes Adjective-like attributes 215 Relative clauses 217 Quantification 217 Numerals Non-numeral quantifiers

DOI 10.1515/9781614519027-004

187

215

218

186

Nicholas J. Williams

Demonstratives and determiners 221 222 Pronominal demonstratives 223 Adnominal demonstratives 224 Adverbial and manner adverbs 225 Prenominal topic marker gána 226 Possession

4.3 4.3.1 4.3.2 4.3.3 4.4 4.5 5 5.1 5.2 5.3 5.4

227 Pronouns 229 Nominative pronouns 230 Potentive pronouns 231 Focus pronouns 232 Possessive pronouns

6 6.1 6.1.1 6.1.2 6.1.3 6.2 6.2.1 6.2.2 6.2.3

Verbal prefixes: person marking and applicatives 235 Agreement prefixes 237 Agreement on intransitive verbs 242 Agreement on transitive verbs Valency-increasing uses of agreement prefixes 251 Applicative prefixes 252 le253 mí254 we-

7 7.1 7.2 7.3 7.4 7.5 7.6

255 Serial verb constructions 255 Argument adding serialization 256 Aspectual serialization 257 Causative serialization 258 Resultative serialization 259 Motion verb serialization 260 Posture verb serialization

8 8.1 8.2 8.3 8.4

260 Aspect marking 261 di ‘IPFV ’ 262 tenu ‘PFV ’ 263 gaya ‘PROSP ’ 264 giya ‘PROG ’

9

Discussion

References

266

264

233

249

Kula

187

1 The language scene Kula (ISO 639-3: tpg) is spoken by approximately 5,000 people between the Kamang area in central-eastern Alor and the Sawila and Wersing areas on the eastern coast. This includes the hamlets Maukuru, Takala, Peisaka, Waika and Kiralela on the north coast, Pureman, Irakena, Dukila, Managomo, Paitoko and Kewala on the south coast, and Lantoka, Moduda/Samuda, Kaipera, Kobra, Naumang (and possibly others) in the mountainous interior. These are all “new villages” (Indonesian kampung baru), where people have lived semi-permanently for only the past 50 years or so. The traditional lands in the mountains include hundreds of named places and “old villages” (Indonesian kampung lama), where some people maintain gardens and raise animals. Previous work (Haan n.d.) has identified two dialects, exemplified by the Kula spoken in Maukuru and Lantoka, respectively. Speakers do not distinguish particular dialects, but claim that the language varies significantly from village to village. Future research will be needed to sort out these issues through a survey of the area where Kula is spoken. Some Kula speakers are also fluent in Wersing and/or Sawila due to frequent contact and intermarriage. In the years since the resettlement, Kula has become endangered. While adults use a mixture of Kula and Malay among themselves, children are most frequently spoken to in Malay. Most children have a passive knowledge of Kula, understanding simple instructions, but regularly responding in Malay. At least at the present time, nearly all young adults know the language well enough to interact with elders and sometimes use Kula among themselves. Kula is still de facto forbidden in schools, a holdover from the 1970s when teachers punished students for using Kula in the classroom. Still, Kula culture remains strong and adat (an Indonesian term denoting ‘local custom’) plays an integral part of daily life in the communities I have observed. Published work on Kula is limited to a wordlist (Stokhof 1975) and an article on inverse morphology under the name Tanglapui (Donohue 1996). Other available unpublished work includes a partial preliminary description of Kula phonology (Haan n.d.), as well as a short dictionary and a number of texts translated into Kula from Indonesian (Johnston n.d.). This sketch is based almost entirely on primary data collected by the author during fieldwork in Alor from 2012–2014. Additional unpublished sources (Haan n.d., Johnston n.d.) were consulted but not used extensively. Data for the sketch were collected in the context of a documentation project focused on audio-visual documentation of conversation and social interaction in Kula. This project has produced an extensive corpus of video-recorded Kula language use (deposited at the Endangered Languages Archive, https://elar.soas.ac.uk/Collection/MPI971878). Examples in the sketch are drawn from this corpus, which also includes audio recordings of traditional narratives, responses to stimulus tasks produced by

188

Nicholas J. Williams

the Max Planck Institute for Psycholinguistics in Nijmegen, and other directly elicited material. Recordings were made primarily in Tanglapui village, which includes the hamlets Lantoka, Samuda, and Kaipera. A handful of recordings were made in other locations, including Kiralela, Pureman, and Kewala. Data were transcribed and translated with assistance from several Kula consultants from Lantoka, especially Penipius Mosali. What follows is a sketch of the phonology and some of the more prominent aspects of Kula morphosyntax, highlighting those features that distinguish Kula from other languages of Alor and Pantar, particularly the closely related languages, Sawila (Kratochvíl 2014) and Wersing (Schapper & Hendery 2014).

2 Phonology The phonology of Kula differs in several ways from the phonology of other languages of eastern Alor. The phonemic inventory is relatively small. Notable features of the consonant inventory include the numerous marginal phonemes (/r/, /β/, /gw/, /kw/, and /ʤ/) and the presence of word-initial /ŋ/. In the sevenvowel system, partial loss of a historical length contrast has resulted in a set of five peripheral vowels with approximately cardinal values and two shorter central vowels (see section 2.2).

2.1 Consonants There are 18 native consonant phonemes in Kula, including eight plosives, two fricatives, three nasals, two glides and two liquids (Table 1). Of these, the bilabial fricative /β/, the palatal affricate /ʤ/, the rhotic /r/ and the labiovelars /kw/ and /gw/ are marginal phonemes (given in curved brackets). Non-IPA orthographic symbols are given in angle brackets. Table 1: Consonant phonemes Labial Plosive

p

Fricative

(β)

Alveolar b

t

Palatal

d

Velar k

s

Affricate

(ʤ)

Nasal

m

Approximant

w

ŋ

n j

Trill

(r)

Lateral

l

Labiovelar g

(kw)

(gw)

Kula

189

2.1.1 Consonant phoneme minimal pairs This section provides minimal and near-minimal pairs to demonstrate phonemic contrast between articulatorily similar consonants. Minimal pairs for the nonmarginal plosive phonemes in Kula – /p b t d k g/ – are given in (1). Minimal pairs for plosives (1) /ˈpula/ ‘sleepy’ /ˈbula/ ‘blind’ /aˈbi/

‘strong’

/aˈpɪ/

‘NFIN . fish’

/ˈdama/ ‘liver’

/ˈtama/ ‘grandparent’

/ˈkana/ ‘FIN .good’

/ˈgana/ ‘FIN .hunt’

/aˈjɐk/

‘unhusked rice’ /aˈjɐg/ ‘NFIN . mouth’

Minimal pairs contrasting plosives with other non-plosive consonants – fricatives, nasals, and approximants – are given in (2). Minimal pairs for non-plosives (2) /ti/ ‘row’ /si/

‘come.LOW ’

/ˈjala/

‘woman’

/ˈwala/ ‘egg’

/pu/

‘NFIN . break’

/mu/

‘NFIN .run’

/ˈŋɐnu/ ‘FIN .1INCL . NOM ’ /ˈgɐnu/ ‘FIN .3.NOM /ˈwɐka/ ‘burn’

/gɐˈka/ ‘skin’

/me/

/ne/

‘come.LEVEL ’

‘drink’

The distinction between /l/ and /r/ is not fully phonemic in Kula. Minimal pairs do exist contrasting /l/ and /r/ in word-medial position, while in wordinitial position, /r/ only occurs in loanwords (3). (3) /ˈsara/ ‘broom’

/ˈsala/ ‘coral’

/ˈroti/ ‘bread’ (Malay) /ˈloki/ ‘FIN .wet’ Finally, several additional marginal consonant phonemes exist in Kula. Minimal pairs for each marginal consonant are given in (4). These marginal phonemes are restricted to a small number of lexical items and each exhibit allophony as described in section 2.1.2. (4) /βɛ/

‘go.LEVEL ’

/we/ ‘blood’

/gʷe/ ‘goat’ /we/ ‘blood /ʤa/ ‘NFIN .water’ /-ja/ ‘mother’

190

Nicholas J. Williams

2.1.2 Consonant allophony Consonant allophony in Kula is limited to the phonemes /j/ and /ʤ/, the phonemes /t/ and /g/ when followed by /i/, and the marginal phonemes /gw/ and /β/. The velar nasal also regularly assimilates to a following consonant, both morphemeinternally and across morpheme boundaries. Examples of the morpheme-internal allophony are given below (5), while other alternations involving the velar nasal are cases of allomorphy and described in section 2.6. (5) /ŋ/ > [ŋ] ~ [n] / _C[+alveolar] /ˈgwaŋsa/ ‘big’ [ˈgwaŋsa ~ ˈgwansa] In (6), allophonic rules for /j/ and the sequences /ti/ and /gi/ are given. All three rules can be described as palatalization due to a following high front vowel /i/. (6) a. /j/ > [j] ~ [ʤ] / _V[+high] e.g. /ˈjika/ ‘red’ [ˈjika ~ ʤika]1 > [j] / elsewhere b. /ti/ > [ti] ~ [ʧ] / _a > [ti] / elsewhere c. /gi/ > [gi] ~ [ʤ] / _a > [gi] / elsewhere

/ˈjati/ ‘FIN .bad’ [ˈjati] e.g. /tia/ ‘PROH ’ [tia ~ ʧa] /ŋɐ-ˈti/ ‘1EXCL . POSS . II -stomach’ [ŋɐˈti] e.g. /gi-ˈajaŋ/ ‘3.POSS . II - NFIN .top [ˈgiajaŋ ~ ˈʤajaŋ] /gi-iˈli/ ‘3.POSS . II -river’ [giˈli]

Finally, the marginal phonemes /ʤ/, /gw/, and /β/ exhibit allophony as set out in (7). (7) a. /ʤ/ > [ʤ] ~ [j]

e.g. /ʤo/ ‘NFIN .granary’ [ʤo ~ jo] /ʤa/ ‘NFIN .water’ [ʤa ~ ja]

b. /gw/ > [gw] ~ [w] e.g. /gwe/ ‘goat’ [gwe ~ we] /=ˈdugwa/ ‘PL ’ [ˈdugwa ~ ˈduwa ~ ˈdua] c. /β/ > [β] ~ [w]

e.g. /βe/ ‘go.LEVEL ’ [βe ~ we] /βiŋ/ ‘FOC ’ [βiŋ ~ wiŋ]

2.1.3 Marginal consonant phonemes In this section, I discuss the marginal status of the consonant phonemes /β/, /ʤ/, /r/, /gw/ and /kw/. 1 [ʤ] is most frequent before /i/, occasionally before other high vowels, e.g., /ʤusa/ ‘divide’.

Kula

191

Of the five marginal phonemes, /kw/ and /gw/ are the most frequent in the currently available data. All attested occurences of these phonemes are set out in (8) and (9). Both occur initially and medially. The voiceless [kw] occurs wordfinally in words like [atɐkw], which is underlyingly /atɐku/. Almost all instances of /kw/ and /gw/ occur before the vowels /a/ and /i/. While /gw/ is realized as either [gw] or [w], /kw/ is always realized as [kw]. Instances of /gw/ (8) a. Initial ‘NFIN .rock’ /gwɐ/ /ˈgwata/

‘coconut’

/ˈgwaka/

‘chin’

/ˈgwaŋsa/

‘big’

/ˈgwina/

‘hold’

/gwi/

‘eucalyptus tree’

/-gwiˈlana/

‘FIN .afraid’

/gwita/

‘call’

/gwe/

‘goat’

b. Medial /algwaˈtani/ ‘FIN .follow’ /ˈdugwa/

‘PL ’

/katuˈgwala/

‘FIN . dog’

/tugwa/

‘already’

/ˈtugwan/

‘owner’

/agwuˈlana/ ‘fishing hook’ Instances of /kw/ (9) a. Initial ‘NFIN .fruit’ /kwa/ /ˈkwila / b. Medial /ˈaikwa/ /ˈtukwa/

‘grass’, ‘basket’ ‘dry’ ‘basket’

/seˈruŋkwa/

‘chili’

/ˈgisukwa/

‘probably, maybe’

/jɐˈkwana/

‘FIN . two’

/-ˈnikwa/

‘eye’

/ˈlekwi/

‘tie up’

192

Nicholas J. Williams

The phoneme /r/ occurs initially only in two recent loan words, while medially it occurs in a handful of native and borrowed words (10). It is almost entirely absent finally, with the one attested exception /kur/ ‘comb’. Kula has apparently lost Proto-East-Alor *r in final position (Antoinette Schapper, pers. comm.). Instances of /r/ (10) a. Initial /ˈribu/ ‘FIN .thousand’ b. Medial /arˈpane/

‘chase’

/ˈsara/

‘broom’

/ˈroti/

‘bread’

/ˈtira/

‘close, shut’

/asuˈrana/ ‘FIN .angry’ /aˈrosa/

‘rub, scratch’

/ˈirɪna/

‘FIN .fly’

/maˈreka/

‘yellow’

/ˈlura/

‘straight’

/seˈruŋkʷa/ ‘chili’ /ˈmarɪna/

‘FIN .small’

/aˈrasiku/

‘FIN .four’

/molaˈkiri/ ‘monkey’ /buˈrana/

‘FIN .very’

Both /β/ and /ʤ/ are very marginal, each being found in only two monosyllabic items initially (11–12). Both have allophones that are identical with other phonemes ([w] in the case of /β/ and [j] in the case of /ʤ/), indicating that they may be the result of incipient phonemic splits in the original /w/ and /j/ in order to avoid homophonies between monosyllabic words. Instances of /β/ (11) /βe/ ‘go.LEVEL ’ /βiŋ/ ‘NFIN . FOC ’ Instances of /ʤ/ (12) /ʤa/ ‘NFIN .water’ /ʤo/ ‘NFIN .storage house’

Kula

193

2.2 Vowels Kula has a seven-vowel system with five cardinal vowels and two non-cardinal vowels (Table 2). The distinction between cardinal and non-cardinal vowels in Kula is one of both quality and length: in stressed syllables, cardinal vowels /a, e, i, o, u/ are long with a typical duration of about 100 milliseconds (but may be longer), while stressed non-cardinal vowels /ɪ, ɐ/ are shorter, approximately 50 milliseconds (but may be shorter). The length difference between cardinal and non-cardinal vowels in unstressed syllables remains to be investigated in detail, but initial analysis shows that cardinal vowels in unstressed syllables are shorter, in the 50ms range similar to stressed central vowels, /ɪ, ɐ/. This shows that length and stress coincide to a certain extent. Minimal pairs demonstrating phonemic contrast between the cardinal and non-cardinal vowels are limited to stressed syllables. There is no clear contrast between cardinal and non-cardinal or long and short vowels in unstressed syllables. The seven vowel phonemes in Kula are shown in Table 2, with non-IPA orthographic symbols given in angle brackets.2 Table 2: Vowel phonemes Front High

Central

i

Back u

ɪ Mid

e

o ɐ

Low

a

2.2.1 Vowel minimal pairs Minimal pairs demonstrating the phonemic status of the five cardinal vowels are given in (13). (13) /-pa/ ‘father’ /pu/ ‘NFIN .break’ /mi/ ‘take’

/me/ ‘come.LEVEL ’

/pe/ ‘pig’

/iˈpi/ ‘NFIN .areca nut’

/ti/

‘row’

/te/

/to/

‘also

/tu/ ‘NFIN .three’

/po/ ‘but’ /mu/ ‘NFIN .run’

‘sleep’

2 Alternative analyses of the Kula vowel system are possible. One reviewer suggests treating vowels in stressed and unstressed syllables separately. Future work will need to look at how stress interacts with the vowel system in Kula to sort this out.

194

Nicholas J. Williams

Minimal and near-minimal pairs for /i/ versus /ɪ/ and /a/ versus /ɐ/ are given in (14) and (15) respectively. All attested minimal pairs contrast these vowels in stressed syllables. (14) /mi/ /ˈpisi/

‘take’

/mɪ/

‘NFIN .be.located’

‘strike’

/ˈpɪsi/

‘mango’

/aˈkiki/ ‘choose’ /aˈkɪki/ ‘grasshopper’ (15) /-pa/ ‘father’

/pɐ/

‘NFIN .garden’

/tan/ ‘NFIN .arrive’ /-tɐn/ ‘NFIN .hand’ /ˈlali/ ‘FIN .play’

/ˈlɐli/ ‘FIN .light (a fire)’ (verb)

Minimal and near minimal pairs contrasting the non-cardinal vowels /ɪ/ and /ɐ/ with each other are provided in (16). (16) /gɐs/ /a-ˈtɐn/ /ˈapɐ/

‘NFIN .cooked.rice’ /gɪs/ ‘3.POSS . I - NFIN .hand’ /aˈtɪŋ/

‘NFIN .contents’ ‘NFIN .swollen’

‘NFIN .bamboo.wall’

‘NFIN .fish’

/ˈmadɐŋ/ ‘NFIN .worker’

/ˈapɪ/

/mɐˈdɪŋ/ ‘NFIN .plant’

2.2.2 Vowel allophony Vowel allophony in Kula affects unstressed vowels and the stressed short/ non-cardinal vowels (/ɪ/ and /ɐ/). First, unstressed vowels are frequently reduced to [ə] or eliminated entirely, resulting in initial consonant clusters (in surface forms only). This process is so frequent for disyllabic words that speakers are generally unaware of the unstressed vowel and prefer to write these words with a consonant cluster and no vowel in the current orthography. Some illustrative examples are given in (17). (17) /puˈka/ [pəˈka] ~ [puˈka]

‘FIN .small’

/pɐˈte/ [pəˈte] ~ [pɐˈte]

‘corn’

/puˈta/ [pəˈta] ~ [puˈta]

‘pray’

/muˈda [məˈda] ~ [muˈda] ‘go.HIGH ’ /kɪˈne/ [kəˈne] ~ [kɪˈne]

‘knife’

/muˈra/ [məˈra] ~ [muˈra] ‘inside’ /mɪˈde/ [məˈde] ~ [mɪˈde] ‘come.HIGH ’ /sɪˈka/ [səˈka] ~ [sɪˈka] ‘arrow’

Kula

195

As for the stressed non-cardinal vowels, both exhibit allophonic variation. First, the phoneme /ɪ/ has an allophone [ɨ] which appears before or after velar consonants. Examples are given in (18). (18) /ɪ/ > [ɨ] /

[+velar] __ __ [+velar]

/paˈdaŋkɪ/ [paˈdaŋkɨ] ‘crocodile’ /iˈnɪŋki/

[iˈnɨŋki]

‘FIN .stoveplace’

/ˈpɪŋki/

[ˈpɨŋki]

‘FIN .plate’

/aˈkɪŋ/

[aˈkɨŋ]

‘NFIN .emotion’

The [ɨ] allophone has been attested in some other environments as well, as in (19). (19) /ˈapɪ/ [ˈapɨ] ~ [ˈapɪ]

‘NFIN .fish’

/ˈmɪti/ [ˈmɨti] ~ [ˈmɪti] ‘FIN .sit’ The phoneme /ɐ/ has a fronted allophone [ɛ] that occurs in the environment immediately preceding /j/ and, in stressed syllables, following /j/. Illustrative examples are provided in (20). (20) /ˈpɐja/ [ˈpɐja ~ ˈpɛja] ‘FIN .garden’ /ˈgɐja/ [ˈgɐja ~ ˈgɛja] ‘FIN .laugh’ /ˈjɐnu/ [ˈjɐnu ~ ˈjɛnu] ‘FIN .2.NOM ’ In addition to these allophonic variations, there are a number of other alternations, some motivated by phonotactic preferences. First, unstressed word-final high vowels (/ɪ/, /i/ and /u/) are frequently devoiced or dropped entirely in non-final clause position (see section 2.6.2 for more on word-shape alternations). Note that devoicing only occurs after voiceless stops and fricatives, but not following nasals (e.g. /ˈtani/ and /ˈdeni/). I provide rules for these alternations in (21) and give examples for each vowel in (22). (21) V[+high] > Ø / __# V[+high] > V[-voice] / C[-voice] __# (22) a. /ˈapɪ/ /ˈbusɪ/

[ˈapɪ ~ ˈapɪ ̥ ~ ap]

‘NFIN .fish’

[ˈbusɪ ~ ˈbusɪ ̥ ~ bus]

‘sand’

196

Nicholas J. Williams

b. /ˈmɪti/

[ˈmɪti ~ ˈmɪti ̥ ~ mɪt]

‘FIN .sit’

/ˈpati/

[ˈpati ~ ˈpati ̥ ~ pat]

‘FIN .eat, bite’

c. /ˈtuku/ [ˈtuku ~ ˈtuku̥ ~ tuk]

‘FIN .cut’

/ˈgusu/ [ˈgusu ~ ˈgusu̥ ~ gus]

‘FIN .shoot arrow’

/ˈsaku/ [ˈsaku ~ ˈsaku̥ ~ sak]

‘FIN .old’

/aˈtɐku/ [aˈtɐku ~ aˈtɐku̥ ~ aˈtɐk] ‘FIN .earth’ Unstressed initial /i/ is commonly dropped. Examples of this are provided in (23). Note that this does not occur when the initial unstressed /i/ is the agreement prefix i- ‘1PL . INCL ’.

[ igɐ- ~ gɐ]

‘1INCL . POSS . II ’3

/iˈpi/

[iˈpi ~ pi]

‘NFIN .areca nut’

/iˈli/

[iˈli ~ li]

‘NFIN .river’

(23) /igɐ-/

/iˈnɪŋki/ [iˈnɨŋki ~ ˈnɨŋki] ‘FIN .stoveplace’ There is also some limited evidence for unconditioned raising of /o/ to [u] in some words. Examples are given in (24). Note that this variation has only been observed in a small number of words with /o/, and the full extent of this raising remains to be determined. (24) /po/

[po ~ pu]

‘but (CONJ )’

/moˈsi/ [moˈsi ~ muˈsi] ‘if (CONJ )’ However, many words with /o/ are never attested with a raised vowel (25). (25) /to/ [to] – *[tu] ‘also’

2.3 Phonotactics Consonant distribution in Kula is characterized by greater restriction in coda position than in onset position. Nearly all consonants occur in onsets, both initial and medial. Two phonemes, /r/ and /ŋ/, are restricted in word-initial onsets: 3 /igɐ-/ takes no primary stress. Stress occurs on a syllable in the root, e.g. /igɐ-ˈmama/ → [igɐˈmama] ~ [gɐˈmama] ‘our field’.

197

Kula

word-initial /r/ is limited to loanwords and word-initial /ŋ/ is restricted to agreement prefixes (which, however, make it pervasive). In word-final codas only /k/, /n/, /ŋ/, /r/ and /l/ are attested. Word-medial codas are attested with /k/, /s/, /m/, /n/, /ŋ/ and /r/, but not /l/. A summary of these restrictions is given in Table 3 (including marginal phonemes) and illustrative examples can be found in section 2.4. Table 3: Distribution of consonant phonemes ʤ

p

b

β

t

d

k

g

kw

gw

s

word-initial onset

+

+

+

+

+

+

+

+

+

+

+

+

word-medial onset

+

+

+

+

+

+

+

+

+

+

–

+

word-medial coda

–

–

–

–

–

+

–

–

–

+

–

+

word-final coda

–

–

–

–

–

+

–

–

–

–

–

–

m

ŋ

r

+

+

–

+

+

+

+

+

+

+

+

+

+

+

+

–

–

–

+

+

+

+

–

–

n

l

w

j

Consonant clusters are uncommon. Attested types are given in Table 4. Within a syllable, clusters are restricted to onsets where a plosive is followed by either /l/ or /r/. Across a syllable, clusters are limited to a nasal followed by a plosive of the same place, with the exception of a few clusters with /s/. Clusters appearing in surface forms only due to short vowel deletion in unstressed syllables are omitted from Table 4. Table 4: Consonant clusters Clusters in onsets

Clusters across syllable boundaries

/pl/

/plɪŋ/ ‘NFIN . accompany’

/m.p/

am.pa ‘so that’, a.jum.pa ‘board’

/bl/

/bla.ke/ ‘papaya’

/m.b

bom.bu.la ‘tomato’

/br/

/ta.brin/ ‘disintegrate’

/n.t/

an.ta ‘or’

/tr/

/ba.la.tra/ ‘piece of bamboo’

/ŋ.k/

wa.lɐŋ.ka ‘green’, luŋ.ku.ki.ta ‘frog’

/kr/

/krɪ.ki/ ‘FIN .finger’, /pu.kra/ ‘spurt’

/k.s/

wik.si ‘flood’

/ŋ.s/

gwaŋsa ‘big’

/s.p/

dis.pa.ka ‘squash’

Distribution of single vowel phonemes in Kula is unrestricted. All seven vowels are attested in both open and closed syllables. Sequences of vowels are more limited. The attested sequences are given in Table 5. Note that while sequences /ai/ and /au/ constitute a single nucleus, other sequences are typically syllabified as separate nuclei. For example, /laisoma/ ‘onion’ is realized as [ˈlai.so.ma], while /ˈtia/ ‘PROH ’ is realized as [ˈti.a].

198

Nicholas J. Williams

Table 5: Vowel sequences

i a o u

i

e

ia

ea

a

o

u

ai ua uo au

Examples of the sequences /ai/ and /au/ are given in (26) below. Both are common in Kula, while /ai/ is only attested in one word in Sawila and /au/ is apparently not attested (Kratochvíl 2014). /ai/ and /au/ sequences (26) /ˈlaiŋki/ ‘ant’ /ˈaikwa/ /ˈtau/

‘dry’

‘too’ /bauˈlana/ ‘FIN .Monday’

All other vowel sequences are less common in Kula. Attested sequences include /ia/, /ea/, and /ua/. Examples of each are given in (27). Less common vowel sequences (27) /ˈtia/ ‘PROH ’ /ˈgea/ ‘tree’ /moˈaja/ ‘sap from a k.o. tree’ /iˈpua/

‘sugar cane’

2.4 Syllable structure Kula syllable structure allows for both open and closed syllables. Words consist minimally of a nucleus and an onset (CV) or coda (VC). In words of more than one syllable, initial syllables may consist of a single vowel (V). Maximum syllable shape is CCVC or CCVV. Complex onsets and nuclei are all found, but restricted, as described in section 2.3. Words range from monosyllabic (CV) to complex three- and four-syllable words. Onset clusters occur both word-initially and word-medially, consisting of either a stop and a liquid (/bl/, /pl/, /br/, /tr/, /kr/) or an /s/ and a liquid or stop (/sl/, /sr/, /sp/). Another unique phonotactic feature among the Alor-Pantar languages is the existence of syllabic nasals in Kula (e.g. /nta/ ‘or’ and /mpati/ ‘I eat’). The observed syllable types are set out in (28). Short/central vowels (/ɐ/ and /ɪ/) are indicated by V̆ . Stress is marked with for primary stress and for secondary stress before the onset of the stressed syllable.

Kula

(28) CV̆

pɐ

‘NFIN .garden’

CV

pe

‘pig’

CVV CV̆ C

tau

‘too’

sɐk

‘torn’

CVC

kul

‘leg’

CCVV CCV̆ C V.CV̆

blai plɪŋ

‘papaya’4 ‘NFIN .accompany, take with’

ˈa.pɪ

‘NFIN .fish’

V.CV

a.ˈji

‘I don’t know’

V.CVV V.CV̆ C

i.ˈpua

‘sugarcane’

a.ˈtɪŋ

‘NFIN .swollen’

VC.CV

am.ˈpa

‘so that’

C.CV CV̆ .CV̆ CV̆ .CV

n.ˈta

‘or’

CV.CV̆

ˈbu.sɪ

‘sand’

CV.CV

ˈwa.la

‘egg’

CV.CVV

ma.ˈkai

‘salty (water)’

CVV.CV CV̆ .CVC

ˈmoa.ja

‘sap from a k.o. tree’

[not attested] ˈmɪ.ti ‘FIN .sit’

CV.CVC

[not attested] ˈŋe.kul ‘my foot’

CVC.CV

ˈwik.si

‘flood’

C.CVCV

m.ˈpati

‘1EXCL . I - FIN .eat’

CV.CCV

ˈpu.kra

‘spurt’

CV.CCVC

ˈta.brin

‘fall apart’

CCV.CV

ˈbli.na

‘write’

V.CV.CV

a.ˈla.ta

‘taro’

V.CVC.CV

a.ˈjum.pa

‘board’

VC.CV.CV

ar.ˈpa.ne

‘chase’

CV.CV.CV

ˈka.pi.ta

‘flood’

CV.CVC.CV

wa.ˈlɐŋ.ka

‘green’

CV.CV.CCV

ba.ˈla.tra

‘piece of bamboo’

4 An alternate form /ˈbla.ke/ ‘papaya’ is also attested.

199

200

Nicholas J. Williams

CVV.CV̆ .CV

wai.ˈkɪ.ki

‘candlenut’

CVV.CV.CV

bau.ˈla.na

‘FIN .Monday’

CVC.CV.CV

ˈbom.bu.la

‘tomato’

CV.CV.CV.CV

ˌgɐ.ka.ˈba.ku ‘cat’

CVC.CV.CV.CV ˌluŋ.ku.ˈki.ta ‘frog’ Preferred syllable structure in Kula is CV. While some onset clusters are allowed, these are highly restricted. Coda clusters are unattested and coda consonants are also highly restricted. Where attested, coda consonants are rare, as can be seen from the examples in (28) above.

2.5 Stress Initial analysis of stress shows that it interacts with the length/quality contrast in the vowels. The contrast between short/central and long/peripheral vowels, /i/ and /ɪ/, /a/ and /ɐ/, is only robustly attested in stressed syllables. In unstressed syllables, there is no clear evidence of this distinction. It is possible that the short vowels in Kula have become more centralized under stress and lost in unstressed syllables, resulting in consonant clusters. In Table 6 I provide examples of word level stress in Kula. Stress is generally penultimate (e.g. /ˈmana/, /disˈpaka/, etc.), but heavier syllable weight due to presence of a long vowel or coda consonant can attract stress to an antepenultimate or final syllable (e.g. /ˈasɐka/, /aˈda/). Table 6: Examples of stress placement

Disyllabic

Trisyllabic

Quadrisyllabic

Syllable

Examples

1st

/ˈmana/ ‘village’ [ˈma.na], /ˈmɪti/ ‘FIN .sit’ [ˈmɪ.ti], /ˈlula/ ‘go.DIST ’ [ˈlu.la], /ˈjɐku/ ‘FIN .two’ [ˈjɐ.ku]

2nd

/muda/ ‘go.HIGH ’ [mə.ˈda], /aˈjɐk/ ‘rice’ [a.ˈjɐk], /aˈda/ ‘fire’ [a.ˈda]

1st

/ˈasɐkaˌ/ ‘tree’ [ˈa.sɐ.ˌka], /ˈtatuku/ ‘talk about’ [ˈta.tu.ˌku]

2nd

/taˈtaku/ ‘meet, find’ [ta.ˈta.ku], /disˈpaka/ ‘squash’ [dis.ˈpa.ka]

3rd

/ˌbaluˈne/ ‘sweet potato’ [ˌba.lu.ˈne]

3rd

/ˌadaˈmasa/ ‘bird’ [ˌa.da.ˈma.sa], /ˌgɐkaˈbaku/ ‘cat’ [ˌgɐ.ka.ˈba.ku], /milaˈpiki/ ‘lightning’ [ˌmi.la.ˈpi.ki]

Kula

201

Stress apparently makes some limited phonemic contrasts, but further analysis is required to confirm this. The minimal pair in (31) appears to contrast only in terms of stress placement. (31) /ˈkula/ [ˈku.la] ‘split in two’ or ‘Kula language’ /kuˈla/ [ku.ˈla] ~ [ku.ˈlua] ‘foam’

2.6 Morphophonemics Morphophonemic processes in Kula include nasal assimilation in the first person possessor and person marking prefixes (section 2.6.1), word-shape alternations (section 2.6.2), and limited vowel harmony (section 2.6.3).

2.6.1 Nasal assimilation Nasal assimilation affects the first-person exclusive prefixes in person marking and possessor prefix paradigms (Set I for each, see sections 4.5 on possession and 6.1–6.2 on person marking). This prefix, a velar nasal /ŋ-/, undergoes assimilation in place to the following consonant, the first consonant of the verb or noun it attaches to. Examples are given in (32). (32) /ŋ-ˈpati/ ‘1EXCL . I - FIN .eat’ /ŋ-ˈlula/ ‘1EXCL . I - FIN -go.DIST ’

[mˈpati] [nˈlula]

/ŋ-ˈku/

‘1EXCL . I - NFIN .stay

/ŋ-ˈβe/

‘1EXCL . I - FIN -go.LEVEL ’ [ɱˈwe]

/ŋ-ˈjo/

‘1EXCL . POSS . I -wife’

[ŋˈku] [nˈʤo]

In some cases, the process of assimilation affects the initial consonant as well. The initial /m/ in (33a) is reduced to a [w], still maintaining the labial feature. In (33b), the initial /m/ and /n/ are both dropped entirely. (33) a. /ŋ-ˈmɪti/ ‘1EXCL . I - FIN .sit’ b. /ŋ-mu/ /ŋ-ne/

[ˈŋwɪti]

‘1EXCL . I - NFIN .run’ [ŋu] ‘1EXCL . I -eat’

[ŋe]

202

Nicholas J. Williams

2.6.2 Word-shape alternations A portion of Kula lexical roots occur in two forms, the ‘final’ (FIN ) and ‘non-final’ (NFIN ) forms (following Kratochvíl 2014 on Sawila). The majority of lexical items in Kula occur in only one form. As in Sawila, the final forms typically end in a vowel (/a/, /i/ or /u/ in Kula), while non-final forms end in a consonant (34a–b). One exception to this in Kula is the third set of lexical items undergoing this alternation (34c), in which the non-final roots are vowel-final and the final forms add a glide /j/ between the root-final vowel and the /a/ marking the words as final. Examples of each set are given in (34). Items with final-nonfinal alternations (34) a. Non-final Final /aˈmɐŋ/ /aˈmɐna/ ‘like.DIST ’ /buˈraŋ/

/buˈrana/ ‘very, just’

/kaŋ/

/kana/

‘good’

/dɐŋ/

/dɐna/

‘one, some’

/aˈnɪŋ/

/ˈanɪna/

‘person’

/ˈmarɪŋ/

/ˈmarɪna/ ‘small’

b. Non-final Final /aˈlɐk/ /aˈlɐki/ /pin/

‘mud’

/ˈpini/

‘nose’

/den/

/ˈdeni/

‘name’

/ler/

/ˈleri/

‘king’

/ŋɐn/

/ˈŋɐnu/

‘1EXCL . NOM ’

/aˈtɐk/

/aˈtɐku/

‘land’

/ujiˈan/

/ujiˈani/

‘exam’5

c. Non-final Final /ˈapɪ/ /aˈpɪja/

‘fish’

/mɪ/

/ˈmɪja/

‘be.located’

/pɐ/

/ˈpɐja/

‘garden’

/ile/

/iˈleja/

‘tomorrow’

/pu/

/ˈpuja/

‘break’

/gɐ/

/ˈgɐja/

‘laugh’

5 Borrowings may exhibit the alternation as well. Ujian is a borrowing from Indonesian/Malay. The final form for borrowings is also unpredictable. Compare jon ~ joni ‘John’, skol ~ skoli ‘(attend) school’, but Yusup ~ Yusupu ‘Yusuf’.

Kula

203

The final forms occur at the end of a phrase, while non-final forms occur elsewhere. Directly elicited citation forms are always given as the final form, providing an easy way to identify final and non-final forms. The alternation affects roughly 50% of the lexicon (compare ~25% in Sawila, Kratochvíl 2014) and is seen in all word types, including nouns, verbs, aspectual and mood markers, and demonstratives. Examples of the type shown in (34a) are given below in (35) and (36). In (35a), giana is the final element of a verb phrase. In (35b), giaŋ is followed by an additional verb, aku ‘stay’, in an aspectual serial verb construction (see section 7.2). Similarly, in (36a), tana is the final element of the verb phrase buraŋ tana, while in (36b) taŋ is followed by a progressive aspectual marker, gija. (35) a. gi-pa ai kɐlɐn opas [awa giana]V P 3.POSS . II -father INTER NFIN .now commander then FIN .leave] ‘His father (said), “Hey! Opas, get going now!”’ b. [opas awa giaŋ a-ku=ŋa commander then NFIN .travel 3.I - NFIN .stay=DEF ‘(After) the commander had been going for a while . . .’ (36) a. lɐmɐna [burɐŋ tana]V P immediately NFIN .just FIN .plant ‘(And they) then immediately just planted.’ b. lɐmɐna [burɐŋ taŋ gija]V P immediately NFIN .just NFIN .plant FIN . PROG ‘Then (let’s) just plant.’ The examples in (37a–b) illustrate the use of final and non-final forms of the verb tani ‘arrive’. Other lexical items in (34b) fuction similarly. Notice that in (37a) tani is the final element of both the verb phrase and the entire utterance, while in (37b) tan is followed by /kɐsa/ ‘finish’, another aspect-marking verb in a serial verb construction. (37) a. [sɪrata ape mi=ŋo mana~mana tani]V P letter make take=DEF. PROX RDP ~ village FIN .arrive ‘(They) made letters and sent them to every village.’ b. [mana~mana tan kɐsa gunamɐn=ŋa]V P . . . RDP ~ village NFIN .arrive finish afterwards=DEF ‘(The letters) having arrived then in all the villages . . .’

204

Nicholas J. Williams

The examples (38a) and (38b) contrast final and non-final forms of the noun /alɐki/ ‘mud’. In (38a), alɐk is the locative argument for the posture verb ate ‘sleep, lie’, while in (38b), alɐki functions predicatively to mean ‘be muddy.’ The name, Momo, is not part of the clause, but rather a vocative uttered after the final word in the clause alɐki. (38) a. gɐn

alɐk=mu

ate nɐka sleep usually ‘He’s used to sleeping in the mud.’ NFIN .3.NOM NFIN .mud=LOC

b. an=si alɐki Momo DIST =TOP FIN .mud Momo ‘That (over there) is muddy, Momo.’ In (39), the postverbal evidential/epistemic marker guna is used twice, first in its non-final form, gun, before the conjunction nta ‘or’ and then in its final form guna at the end of a verb phrase. (39) amɐn-da sen=nga a-mɪt gun nta gɪta we-pake like.DIST- SEQ money=DEF 3.I - NFIN .sit NFIN . EVID or 3.POT APPL- NFIN .use le

guna

NFIN .finish FIN . EVID

‘So, was the money still there or did he already use it?’ 2.6.3 Vowel harmony Finally, limited vowel harmony is also attested in Kula. The clearest examples involve the vowel /u/. Examples are given in (40). The vowel in the final syllable becomes a diphthong, /a/ > [ua]. The addition of the [u] vowel in the final syllable is allophonic. In most cases, tokens without the vowel harmony also occur. (40) /suka/ → [sukua] /tupa/ → [tupua] ~ [tupa]

‘like’ ‘close’

/puta/ → [putua] ~ [puta] ~ [pta] ‘pray’

Kula

205

2.7 Orthography Since the work of Neil Johnston (n.d.), there has been a working orthography used for Kula. This orthography is based on the Indonesian orthography. The differences in the phonology of Kula and Indonesian, particularly the existence of two short/central vowels in Kula, have resulted in inconsistencies in the use of this orthography. Short/central vowels are sometimes written, sometimes not. Short vowels are sometimes written with orthographic {e}. In this sketch, I use a modified version of this orthography. In orthographic representation, the three vowel phonemes without short counterparts (/e, o, u/) are represented with the graphs {e, o, u}. The short/central vowels are distinguished from their long/peripheral counterparts by means of diacritics. /a/ and /i/ are represented with single graphs {a, i}, while /ɐ/ and /ɪ/ are represented with {á} and {í}, respectively. Unstressed epenthetic schwas are not written. Consonant phonemes are written with the graph that is identical to the IPA representations of their underlying form, with three main exceptions, all following Indonesian orthographic conventions: {y} is used for the phoneme /j/, {j} is used for the phoneme /ʤ/, {ng} is used for /ŋ/. The marginal phonemes /β/ and /gw/ can both be written as {w} where they are realized [w]. The low functional load and restricted distribution of these phonemes mean that there is little possibility of their confusion with /w/. Where realized as [gw], /gw/ is written {gw}. This is also in line with native speaker preferences. Clitics are usually written as separate words, but marked as clitics using ‘=’ in the second gloss line.

3 Basic clausal syntax This section describes the structure of verbal (section 3.1), locative and existential (section 3.2), and non-verbal clauses (section 3.3). I also address clausal negation (section 3.4). Basic clause structure in Kula consists of a predicate preceded by its arguments and followed by aspect and mood markers. Arguments are frequently omitted when discernible from discourse context. Post-verbal aspectual markers are described in section 8.

3.1 Verbal predicates Verbal predicates in Kula take a maximum of two arguments. Intransitive predicates occur with a single argument (S) expressed as a noun phrase or pronoun

206

Nicholas J. Williams

preceding the predicate and sometimes with a person marking prefix on the verb. Transitive predicates allow for two arguments, one more agent-like (A) and one more patient-like (P). Transitive verbs take person marking prefixes for either the A or P arguments. Prefixes for both A and P arguments on a single verb occur rarely, but are attested. No true ditransitive predicates exist (see section 7.1). Arguments are frequently omitted when easily identifiable from preceding discourse context. The single argument of an intransitive verb (S) typically precedes the verb. This argument is frequently omitted when easily interpreted from preceding context, as in the second clause of (41). (41) Ngán dua to nguda, lámána giana. ngS-muda lámána giana [ngán=dugwa=to]S NFIN .1EXCL . NOM =PL =also 1EXCL . I -go.HIGH then FIN .leave ‘We also got on (the boat) and then left.’ Transitive predicates allow for two arguments, A (the more agent-like) and P (the more patient-like). The unmarked order in transitive clauses is APV (42). A focused P argument can also occur before the A argument through leftdislocation in the marked order PAV (43). Pronouns can be used to express A arguments, but never P arguments (44). Person marking prefixes are not always present, as can be seen in example (44). More details on the behavior of these prefixes are given in section 6.1. (42) Kumapu asáka gámi. asákaP gáA-mi [kuma-pu] A back-NFIN .break wood 3.II -take ‘The crippled one (‘Kumapu’) took the wood.’ (43) Pe nga ngápa Yusup gagusu. gáA-gusu [pe=nga [ngá-pa Yusup]A pig=DEF 1EXCL . POSS . II - father NFIN .Yusuf 3-FIN .shoot.with.arrow ‘Then Yusuf shot the pig.’ (44) Gánto gisaku dua dígíni, ngánto ngásaku, ngátukwa ing wísa. [gi-saku=dugwa]P dígíni [ngán=to]A [gán=to]A NFIN .3.NOM =also 3.POSS . II -sack=PL FIN .lift NFIN .1EXCL . NOM =also [ngá-saku ngá-tukwa=nga]P wísa 1EXCL . POSS . II - sack 1EXCL . POSS . II - basket=DEF carry.on.shoulder ‘He picked up his sacks, and I also picked up my sack and basket and carried them on my shoulders.’

Kula

207

Both A and P arguments of transitive verbs can be omitted when recoverable from the preceding context. In (45), the A and P arguments for the final verb le-lila are both omitted, given their recoverability from the explicit mentions of those shared arguments from the preceding clauses. (45) Waikíki nga bukula baleka dáng gunamannga, ngáya Ribka ngánáku wísa nji Petuka mu dasin lelila. waikíki=nga bukula baleka dáng gunamán=nga candlenut=DEF husk container NFIN .one afterward=DEF [ngá-ya Ribka ngánáku]A wísa ngS-ji 1EXCL . POSS . II -mother Ribka 1EXCL . DU carry.on.shoulder 1EXCL . I -go.LOW Petuka=mu dasin le-lila Peitoko= LOC scale APPL-hang ‘(I) husked candlenut (and got) one ‘container’, after that Ribka and I, the two of us carried (it) down to Peitoko and weighed (it). Only one marginally ditransitive verb is attested in Kula, the verb -yáni ‘give’. This verb always occurs with the applicative prefix we-. The T argument is typically introduced as the argument of a separate verb, mi ‘take’, making this more like a serial verb construction (46). (46) Apá le amáng laporan mi camat wegayán amána. apá le amáng laporan mi camat NFIN .evening NFIN .finish NFIN .like.DIST report take district.head we-gaA-yán amána APPL -3.II - NFIN .give FIN .like.DIST ‘Yesterday, the district head (Ind. camat) gave (them) a report like that.’ (47) Ngápte abu nga mi ngápa bayáku wegani. ngá-páte abu=nga mi ngá-pa ba-yáku 1EXCL . POSS . II -corn dust=DEF take 1EXCL . POSS . II -father CLF. HUM - FIN .two we-ga-yáni

APPL -3.II - FIN .give

‘I gave the ground corn to the two men.’

208

Nicholas J. Williams

However, T arguments may also be included in the clause without the addition of mi (48). Interestingly, in this example, ake, an instrument is also included without any explicit flagging or verbal agreement. (48) Pte ma buráng ake wegayáni. páte ma buráng ake we-ga-yáni corn cooked NFIN .only spoon APPL-3.II - FIN .give ‘They only (were able to) give him food with a spoon.’ This is potentially due to the lack of a NP expressing the R argument. Another case in which there is no full NP for the R argument and the T argument is not marked with mi is given in (49). Since the only NP is inanimate, the T argument does not require extra flagging with mi to distinguish it from a possible R argument. (49) Gimasing amít nga wegayáni. gi-masing a-mít=nga we-ga-yáni 3.POSS . II - NFIN .food 3.I - NFIN .sit=DEF APPL -3.II - FIN .give ‘Give (him) his food.’ Thus, this verb, -yáni, is not a typical ditransitive verb. When all three arguments are overtly expressed, the T argument is marked with mi, resulting in a serial verb construction. When the R argument is omitted, the T argument is not marked with mi.

3.2 Locative and existential predicates Location and existence are expressed in several ways involving the general locative verb míya ‘be located’, a set of posture verbs, the verb giya ‘put’, as well as deictic elevationals and positional nouns. This section describes the available constructions for expression of location and existence in Kula.

3.2.1 Locative/existential verb míya The general locative verb míya ‘be located’ is commonly used to express both location and general existence. There are two basic types of constructions as depicted in (50a–b). There are two slots for NPs, with the location either preceding the theme (50a) or the theme preceding the location (50b).

Kula

(50) a. NPL O C NPT b. NPT

209

míya

NPL OC míya

Examples of each construction are given in (51) and (52). In (51), sirusa ‘work’ is the theme located in the village Lantoka, while in (51), kne ‘knife’ is the theme located in tas ‘bag’. (51) Lantoka to sirusa míya. Lantoka=to sirusa míya Lantoka=also work FIN .be.located ‘There is work in Lantoka, too.’ (52) Kne tas míya. kíne tas míya knife bag FIN .be.located ‘The knife is in the bag.’ This verb is also used to express existence when no location is specified (53). (53) Yo inga gitatuku míya. yo=nga gi-tatuku míya NFIN .store.house=DEF 3.POSS . II - FIN .story FIN .be.located ‘This storage house has a story (about it).’ 3.2.2 Posture verbs In addition to the locative verb míya, posture verbs are used to express the location and existence of both human (54) and non-human (55) entities. These constructions express temporary existence at the time of utterance rather than a permanent sense of existence. (54) Kris míti nta nanu? Kris míti nta nanu Kris FIN .sit or FIN . NEG ‘Is Kris there or not?’ (55) Kursi amít muna. kursi a-mít muna chair 3.I - NFIN .sit FIN . EVID ‘There’s probably a chair (there).’

210

Nicholas J. Williams

Other posture verbs used in existential and locative constructions include duka ‘stand’ (56), lila ‘hang’ (57), and te ‘lie’ (58). While humans occur only with míti ‘sit’, choice of posture verb with non-humans depends on either the position of the entity in the situation described or some physical property of the entity. (56) Ngáles pe níjiwe ngu duka muna. ngá-les pe ní-ji-we=nga=mu duka muna 6 1EXCL . II - NFIN .think pig LOC -go.LOW-go.LEVEL =DEF =LOC stand FIN . EVID ‘I think the pig/deer is (lit. ‘standing’) down over there.’ (57) Payengke tan gayogu, ngátukwa dua lila nanu. payengke tan gayogu ngá-tukwa=dugwa lila nanu 1EXCL . POSS . II -basket=PL hang FIN . NEG Payengke NFIN .arrive CONJ ‘Having arrived in Payengke, my baskets were not (lit. ‘hanging’) there.’ (58) Kátual ng tau lika ateya, an dua to mi gámuna kátuala=nga tau lika ateya ang=dugwa=to mi ga muna NFIN .dog=DEF too many FIN .lie DIST = PL = also take NFIN . PROSP FIN . EVID ‘There are so many dogs lying around, they will be recorded (lit., taken) too.’ 3.2.3 Other existential verbs Other verbs used in existential constructions include kuya ‘FIN .stay’, giya ‘FIN . put’, and agátu ‘be absent’. (59) Mery, kta níweji ili gi wawa. Mery káta ní-we-ji ili gi wawa Mery NEG . MOD LOC -go.LEVEL-go.LOW NFIN .river NFIN .put NEG . MOD ‘Mery, there’s not a river down there, is there?’ (60) Níweji akuya. ní-we-ji a-kuya LOC -go.LEVEL -go.LOW 3.I - FIN .stay ‘(It’s) down over there.’ (61) Kawiyang agátu. kawiyang agátu NFIN .mosquito be.absent ‘There are no mosquitoes (here).’ 6 See section 3.2.4 for explanation of location specification.

Kula

211

3.2.4 Specifying location More specific locative relationships can be expressed by using positional nouns such as mra ‘inside’ with the basic locative verb míya (62). (62) Api nga mangkok gimra míya. api=nga mangkok gi-mura míya NFIN .fish=DEF cup 3.POSS . II -inside FIN .be.located ‘The fish is inside the cup.’ Other positional words are used in constructions with posture verbs, such as tuka ‘bottom’ used with míti ‘sit’ in (63) to express the location of a historic relic. (63) Ang numda sup gituka míti. ang nu-muda supu gi-tuka míti DIST LOC -go.HIGH NFIN .canoe 3.POSS . II -bottom FIN .sit ‘(It) is up there, below the canoe.’ (lit., ‘It sits up there at the end of the boat.’) An elaborate set of deictic elevational terms is also used with locative constructions to encode more specific locative relations. The elevational terms are summarized in Table 7. Table 7: Elevationals

there come go

low

level

high

gotín si ji

gomán me we

goyon míde muda

Elevationals are typically used with posture verbs and rarely attested with the general locative verb míya ‘be.located’. One example is found in (63) above, in which numda ‘up there’ occurs prior and in apposition to sup gituka ‘bottom of the canoe’. Another example involving gomán and níwe is given in (64). (64) Gomán níwe ja mít nta nanu? gomán ní-we ja mít nta nanu NFIN .there.LEVEL LOC -go.LEVEL NFIN .water NFIN .sit or FIN . NEG ‘Is there any water over there or not?’

212

Nicholas J. Williams

As shown in Table 7, the basic deictic verbs for ‘come’ and ‘go’ in Kula include a three-way distinction based on elevation (LOW, LEVEL , and HIGH ). These verbs combine with a prefix ní-/nu- to produce an extensive set of elevational terms, supplementing the basic set of three (gotín, gomán, goyon). They are used in the same way as the elevationals in (64) above. The attested forms with free translations are given in (65). (65) ní-si

‘down here’

ní-ji

‘down there’

ní-si-me

‘down over here’

ní-ji-we

‘down over there’

ní-we-ji

‘down over there’

nu-muda-we ‘up over there’ ní-we-muda ‘up over there’ ní-si-míde

‘across over here’ (e.g. across a river or valley)

ní-ji-muda

‘across over there’ (e.g. across a river or valley)

For more on the use of elevationals in Kula, see Williams (2016).

3.3 Nominal predicates (equational clauses) Equational clauses involve the use of nominal predicates. In these clauses, two NPs are simply juxtaposed with no other predicative element. Kula does not allow the use of a copula or any other marker for an equational nominal clause. Examples are given in (66) and (67). (66) Ayo to jawa yala pka. a-yo=to jawa yala puka 3.POSS . I -wife=also Javanese female child ‘His wife also (is) a Javanese girl.’ (67) Inga, ado yala nga, atam giskwa. inga a-do yala=nga a-tamu gisukwa PROX 3.POSS . I - NFIN .child female=DEF 3.POSS . I - FIN .grandchild probably ‘This one, the female child, (is) probably her grandchild.’

Kula

213

3.4 Negation The particle nanu is used for clausal negation in both verbal (68) and non-verbal clauses (69). (68) Ngku nanu. ng-ku nanu 1EXCL . I - NFIN .stay FIN . NEG ‘I didn’t stay.’ (69) Gán to gur nanu. gán=to gur nanu NFIN .3.NOM =also NFIN .teacher FIN . NEG ‘He is also not a teacher.’ Negative imperatives use the clause-final particle tia. (70) He, yágwilan tia! he yá-gwilan tia INTER 2.II - NFIN .afraid PROH ‘Hey, don’t be afraid!’ Other negative polarity items include tabila ‘may not, not permitted’ and the discontinuous káta . . . wawa ‘there’s not . . . is there?’. Tabila is typically used with the clause-final negator, nanu, though this is not obligatory. Káta . . . wawa adds a sense of surprise, i.e., that the proposition is contrary to the speaker’s belief or current knowledge. (71) Tabila dáng aku nanu. tabila dáng a-ku nanu PROH NFIN .one 3.I -NFIN .stay FIN . NEG ‘Don’t let any remain.’ (72) Mery, káta níweji ili gi wawa. Mery káta ní-we-ji ili gi wawa Mery NEG . MOD LOC -go.LEVEL-go.LOW NFIN .river NFIN .put NEG . MOD ‘Mery, there’s not a river down there, is there?!’

4 Noun phrases A noun phrase template is provided in (73). Noun phrases in Kula are headed by nouns, which are distinguished from verbs primarily by being able to host a

214

Nicholas J. Williams

possessor prefix. The noun phrase template consists of a head noun, optionally hosting a possessor prefix, and followed by slots for attributes, degree words, numerals or non-numeral quantifiers, and demonstratives. Aside from possessor prefixes, topic-marking gána is the only other pre-nominal element in the noun phrase. Post-nominal enclitics include =si ‘TOP ’ and =to ‘also’, which are mutually exclusive. Relative clauses are optionally marked by =nga, which occurs phrase finally and derives from the demonstrative determiner, =nga. (73) NP → [POSS -N ATTR DEGREE NUM /QUANT DET ]=ALSO / TOP Sections 4.1 through 4.5 describe each slot in more detail. E xamples of noun phrases with different slots occupied are given in (74) to (77). While no single noun phrase has all slots occupied, these examples provide evidence for each individual slot and their proposed order. Numerals and nonnumeral quantifiers occupy the same slot. (74) N QUANT DET pe du onga pe=dugwa=onga pig=PL = PROX ‘those pigs’ PSR-N ATTR DEGREE (75) gikwila banang burána gi-kwila banang burána 3.POSS . II -grass NFIN .forest FIN . INTS ‘his very thick grass’ gána N(ATTR ) NUM =to (76) gána saku yákwan to gána saku yákwan=to DIST.TOP FIN .old NFIN .two=also ‘Those two old (people) as well’ N ATTR QUANT (77) jala pka dua jala puka=dugwa woman small=PL ‘young women’

Kula

215

4.1 Attributes This section describes the use of both adjective-like attributes and relative clauses to modify head nouns in a noun phrase. Both always follow the head noun.

4.1.1 Adjective-like attributes There is no distinct category of adjective in Kula. Some attributes are expressed by nominal roots, like jala ‘female’ in (78) – compare with (79), where jala is used as a noun. Other attributive words are not distinct from verbs and can typically be used predicatively. For example, gwangsa ‘big’ is used as a predicate in (80), while it is used attributively to modify the noun ige ‘road’ in (81). Attributes always follow the nouns they modify. (78) ado jala a-do jala 3.POSS . I - NFIN .child female ‘his/her daughter’ (79) jala pka jala puka female small ‘little girl’ (80) Yánu wansa. yánu gwangsa FIN .2.NOM big ‘You are big/important.’ (81) Igá wansa mu giang lula ape. igá gwangsa=mu giang lula ape NFIN .leave go.DIST do road big=LOC ‘Take the main road.’ 4.1.2 Relative clauses Nouns in Kula can also be modified by relative clauses, a clause that modifies the head noun and any associated attributes. Relative clauses are typically unmarked, as in (82) and (83). The head noun may express any role (S, A or P) in both the relative clause and the main clause. In (82), the head noun expresses

216

Nicholas J. Williams

the S argument in both the relative clause and the main clause. In (83), the head noun expresses the S argument in the relative clause, but the P argument of the main clause. (82) Jala sak ado yákwang plíng gian to gita ki tau yat burána. a-do yákwang plíng [[jala sak]S woman NFIN .old 3.POSS . I - NFIN .child NFIN .two NFIN .accompany gian=to NFIN .travel=also

gíta ki tau yat burana 3.POT NFIN .cry too NFIN .bad FIN . INTS

‘The woman (who) brought her two kids along, she cried horribly.’ (83) Igákwila du nga atansi po asiape dáng níjimda si ingu adi, káta ayám adapa wawa! igá-kwila=dugwa=nga a-tan si po 1INCL . POSS . II -basket=PL =DEF CAUS -arrive come.LOW so.that [[asiape dáng]S ní-ji-muda si ingu]RC adi NFIN .one LOC -go.LOW-go.HIGH come.LOW here NFIN .see deer káta

ayám

NEG . MOD NFIN .die

adapa wawa close NEG . MOD

‘Put our baskets down so (we can) see the deer (that) came down here from over there, it might be almost dead.’ While most relative clauses are unmarked, some include an optional marker =nga. This marker is likely related to the distal demonstrative/definite determiner, =nga (cf. ba in Kamang (Schapper 2014)). Note that the head noun sen nga ‘the money’ in (84) is the P argument of both the relative and main clauses. In (85), pka ngo ‘that child’ is the S argument of the relative clause and the A argument of the main clause. (84) Sen nga tánágan-tánágan nga mi si legasámu. le-gasámu [[sen=nga]NP tánágan~tánágan=nga]RC mi si take come.LOW APPL-FIN .save money=DEF RDP ~ NFIN .gather=REL ‘We took down and saved the money that we had gathered.’ (85) Gipa, ‘pka ingo mít aku ingo ngáleso nága dáng ape aku gisukwa!’ a-ku=ngo]RC gi-pa [[puka=ngo]NP mít 3.POSS . II -father small=DEF. PROX NFIN .sit 3.I-NFIN .stay=REL . PROX ngá-leso nága dáng ape a-ku gisukwa 1EXCL . II - FIN .think thing NFIN .one do 3.I -NFIN .stay probably ‘His father (said) “That kid who is at home, I think, he probably did something (wrong).”’

Kula

217

4.2 Quantification Number is not obligatorily marked on nouns in Kula. However, numerals (section 4.2.1) and other non-numeral quantifiers (section 4.2.2) are used to express number and other quantity concepts in noun phrases. An optional plural enclitic is common, though not obligatory. Both types of quantifiers follow the head noun and any other attributes modifying the head noun.

4.2.1 Numerals Basic numerals are given in Table 8 below. The Kula numeral system is quinary. While some other AP languages have a separate monomorphemic word for 6, Kula uses the quinary numeral, 5 + 1. Numerals 1, 2, and 3 have two forms. The longer citation forms are given in Table 8. The shorter forms encountered in spontaneous speech are sona, jáku, and tu. Table 8: Numerals 1

6

sonadána

jawatín sona

2

jákwana

7

jawatín jáku

3

tukana

8

jawatín tu

4

arasiku

9

5

jawatína

10

jawatín arasiku adajáku

An example of a higher, more complex numeral is given in (86). There is a native Kula word for ‘100’, gasáka, while the term for ‘1000’, ribu, is borrowed from Indonesian. These complex numerals are typically used in referring to specific calendar years. Otherwise they are uncommon. Note that when counting, ‘one’ is expressed by dána alone without sona. The applicative prefix mí- here is used to express the tens place, literally ‘eight in ten’. Arasíng, ‘plus’, is used before the ones digit in any number above ‘10’. (86) Ile rib dána gasáka yawatín arasiku adayáku miyawatín tu arasíng arasik nga ngwe ujiani. ile NFIN .year

rib dána gasáka yawatín arasiku adayáku thousand FIN .one hundred NFIN .five FIN .four ten

mí-yawatín

tu

APPL- NFIN .five NFIN .three

arasíng arasik=nga ng-we ujiani plus NFIN .four=DEF 1EXCL . I -go.LEVEL FIN .exam

‘The year 1984, I went to take the exam.’

218

Nicholas J. Williams

Classifiers in Kula are limited to the prefix ba-, which functions as an optional human classifier. The prefix attaches to numerals in a construction denoting the number of humans for the preceding noun, as in (87). The prefix is optional, however, as can be seen by its absence for the noun phrase ado yákwang in (88). The example in (89) demonstrates counting in natural discourse and the absence of any classifiers for non-human nominals. (87) Saku batu nga, dáng abasáya wána, dáng atáke nga wána, dáng gitana wána. saku

ba-tu=nga

dáng

FIN .old CLF. HUM -three=DEF NFIN .one

abasáya wána dáng rib carry.on.head NFIN .one

atáke=nga wána dáng gi-tana wána head=DEF carry.on.head NFIN .one 3.POSS . II -thigh carry.on.head ‘There were three old men, one carried the ribs, one carried the head, and one carried the thighs (of a deer).’ (88) Jala sak ado yákwang plíng gian to gita ki tau yat burána. [[jala sak]S a-do yákwang plíng woman NFIN .old 3.POSS . I - NFIN .child NFIN .two NFIN .accompany gian]R C =to NFIN .travel=also

gíta ki tau yat burana 3.POT NFIN .cry too NFIN .bad FIN . INTS

‘The woman (who) brought her two kids along, she cried horribly.’ (89) Motra sálál aku muna, motra nga tukana nan yákwana. . . motra sálál a-ku muna motra=nga tukana motorbike NFIN .search 3.I - NFIN .stay FIN . EVID motorbike=DEF FIN .three nan

NFIN . NEG

yákwana FIN .two

‘(They) were searching for a motorbike, it seems, either three or two motorbikes. . .’ This lack of classifiers in Kula, other than the one optional human classifier, is in line with the observations of Klamer (2014) that numeral classifiers decrease as one moves east across Alor languages. 4.2.2 Non-numeral quantifiers Non-numeral quantifiers in Kula include the modifiers lika ‘many’, padiki ‘a bit’, anawe ‘all’, dindini ‘each one’, dána ‘one, some, INDEF ’, and the plural clitic =dugwa. These quantifiers all follow the head noun they modify.

Kula

219

First, Kula uses an optional plural enclitic, =dugwa ‘PL ’ to express both a vague plural sense (90) and an associative plural sense (91). The plural clitic is never used with numerals. (90) Jala sak dua jiada nga anawe ape akáni. jala sak=dugwa ji-ada=nga anawe ape akáni make FIN .extinguish female NFIN .old=PL 2.POSS . II -fire=DEF all ‘Ladies, put out all of your fires!’ (91) Lipidama du ga nan tenu. lipidama=dugwa ga nan tenu NFIN .say NFIN . NEG FIN . PFV Lipidama=PL ‘Lipidama and those guys haven’t said (anything).’ This plural clitic can also be used with inanimate referents (92). (92) Ada du nga anawe yuka akáni. ada=dugwa=nga anawe yuka akáni fire=PL = DEF all pour.water FIN .extinguish ‘Extinguish all the fires!’ The modifiers lika and padiki are used to express the quantificational concepts ‘lots, many’ and ‘a little, not much’. Lika (93) is much more common than padiki (94), with only one clear instance of padiki attested in the current corpus. (93) Doki lika mi su pa wáka. doki lika mi su pa wáka mouse many take NFIN .come so.that roast ‘Bring lots of mice to roast.’ (94) Seng padik dáng mi ka. seng padik dáng mi ka money NFIN .little NFIN .some take DISC ‘Take a little bit of the money.’ Another non-numeral quantifier, dindini, conveys the sense of ‘each one of many’ (95). This is a reduplication of the word dini ‘how much?’, used to question quantity.

220

Nicholas J. Williams

(95) Sirusa dindini yatáng mra mí. . . sirusa dindini ya-táng mura mí work each.one 2.POSS . II - NFIN .hand inside NFIN .be.located ‘All of / each one of the tasks is in your hands. . .’ The quantifier anawe expresses ‘all’, a notion similar to that expressed by dindini, but without enumerating individual members of the set. (96) Aningkang anawe lula gu gaya. aning kang anawe lula gu gaya NFIN .person NFIN .good all go.DIST hit.gong FIN . PROSP ‘Everyone went to hit the gong.’ The quantifier, dána ‘one, some’ has several uses. The form derives from the numeral sonadána ‘one’. However, this full form is uncommon. Sona alone is used when counting, as in (97). The non-numeral quantifier dána is never used for counting. (97) Ya! ngialiku, “sona, yáku, tua”, amántuanga, anawe jaluk iji. ya ngi-aliku sona yáku tua amán tuanga anawe INTER 1EXCL .V-count one FIN .two three NFIN .like.DIST CONJ all ja

luk

NFIN .water NFIN .hole

i-ji 1INCL . I -go.LOW

‘OK! I’ll count, 1, 2, 3, and then we all go down into the pool.’ Dána can be used to mean ‘one’ without sona when enumerating items. For example, ‘one container’ is expressed as bleka dána in (98). (98) Ayák ntana gigís bleka yáku. Pte to lepuku, bleka dána. ayák ng-tana gi-gís bleka yáku rice 1EXCL . I - FIN .grind 3.POSS . II - NFIN .content container FIN .two páte=to le-puku bleka dána corn=also APPL-husk container FIN .one ‘I pounded unhusked rice (to remove the husk), and the result was two containers. I also husked corn, and (got) one container.’ Note that sona can also be used on its own as a modifer without dána to enumerate (99).

Kula

221

(99) Talpi mana sona ngwíti. talpi mana sona=nga míti Talpi village one=DEF FIN .sit ‘Talpi (will) occupy one village.’ Dána may also be used as a nominal modifier but without the precise meaning of a numeral. In these cases, its function is to indicate an indefinite but specific referent. For example, in (100), sirusa dáng refers to a specific, but indefinite, activity or job that the speaker intends to do. (100) Anuna sirusa dáng ape giya. anuna sirusa dáng ape giya because work NFIN .one do PROG ‘Because there is something (we) are going to do.’

4.3 Demonstratives and determiners Kula has a complex set of demonstratives and determiners outlined in Table 9. This section offers an initial analysis of demonstrative forms in Kula and a description of their primary functions. The proximal/distal contrast is expressed through the vowels i in the proximal forms and a in the distal forms. This contrast does not hold for adnominal demonstratives, however. In fact, there does not appear to be a robust spatial contrast in adnominal demonstratives. I posit =onga as the proximal form and =nga as the distal form, though they are not used consistently this way in everyday speech (see below for examples and further discussion). Table 9: Kula pronominal demonstratives Proximal

Distal

Pronominal

inga

anga

Adverbial

ingu

angu

Manner

inína

amána

Adnominal

=onga

=nga

Pronominal and adnominal demonstratives can be modified with an additional enclitic, =o, producing the four additional forms in Table 10.

222

Nicholas J. Williams

Table 10: Additional demonstratives Proximal

Distal

Pronominal

ingo

ango

Adnominal

=ongo

=ngo

4.3.1 Pronominal demonstratives First, inga ‘PROX ’ and anga ‘DIST ’ are used only as third person pronouns and never modifying nouns. These pronominal demonstratives are often accompanied by a point when used to refer to something in the immediate physical environment. They can also be used to point to chunks of discourse. Examples of their use in the immediate environment are given in (101) to (103). These examples are all taken from a recording of two Kula speakers doing the Family Problems Picture Task (San Roque et al. 2012). In this task, the speakers are presented with a series of pictures and asked to order them and tell a narrative using the pictures. Each of the demonstratives used here refers to one of the pictures as the speakers are presented with it, discussing it, or handing it back to the researcher. (101) Inga maka gisukwa? inga maka gisukwa PROX banana maybe ‘This is maybe a banana?’ (pointing at picture of a person holding bananas) (102) Kála inga pá mu gasu lámána gámi su tugwa. kála inga pá=mu gá-su lámána gá-mi now PROX NFIN .garden=LOC 3.II - NFIN .come immediately 3.II -take su

tugwa NFIN .come already

‘Now, this one is him coming from the garden, then (in this other one) he’s already bringing (his produce).’ (pointing to a picture) (103) Anga giweyawatína. anga gi-we-yawatína DIST 3.POSS . II -APPL- FIN .five ‘That’s the fifth one.’ (referring to the fifth in a series of pictures)

Kula

223

Pronominal demonstratives can be used to point to elements of the preceding discourse as well, as in (104), a common phrase said at the beginning of a narrative text. (104) Anga inína. anga inína DIST FIN .like.PROX ‘That was like this.’ Both pronominal demonstratives can take the enclitic =o. The precise function of this enclitic requires further study, but some idea can be gathered from examples (105) and (106) below. In these examples, =o appears to indicate the initiation of a word search, a common function for demonstratives crosslinguistically. Because of its function and similarity with -o in Sawila, I here tentatively gloss it as proximal ‘PROX ’. (105) Ingo, mda lila gána mi tan si tenu. inga=o muda lila gána mi tan si tenu PROX = PROX go.HIGH hang DIST.TOP take NFIN .fall come.LOW FIN . PFV ‘This one here, (something) hanging, he made it fall already.’ (106) Ango, yikaku Wisal wina. anga=o yi-kaku wisal wina DIST = PROX 2.POSS . II -FIN .younger.sibling Wisal hold ‘That’s, uh, your younger brother Wisal holds (that area).’ 4.3.2 Adnominal demonstratives Of the two adnominal demonstratives, the distal =nga is very frequent and often used in elicitation contexts to mark specific, identifiable referents. In fact, =nga appears to be developing into more of a definiteness marker than a true distal demonstrative. For instance, in (107) =nga is used in a context in which a distal interpretation would make little sense. The proximal onga is frequently used for discourse deixis (108), though spatial uses are also attested (109). (107) ngáwáge inga7 ngá-wáge=nga 1EXCL . POSS . II -tooth=DIST / DEF ‘my tooth’ 7 Here inga is the surface realization of =nga, because the preceding noun is vowel-final.

224

Nicholas J. Williams

(108) Mi su bására su onga, seng ngáles gámi tugwa. mi su bására su=onga seng ngá-les gá-mi take NFIN .come market NFIN .come=PROX money 1EXCL . I -think 3.II -take tugwa already ‘(In) this one (where he’s) bringing (something) from the market, I think he’s taking money.’ (109) yáwáge onga yá-wáge=onga 2.POSS . II -tooth=PROX ‘this tooth of yours’ The clitic =nga as a definite determiner is used to modify pronominal demonstratives in (110) and (111) and personal names (112). (110) Inga nga, aningo mata kali. inga=nga aning=ngo mata kali PROX =DEF NFIN .person=DIST hurt sick ‘This one [looking at a picture], that person is sick.’ (111) Anga nga giwe yawatín sona. anga=nga gi-we-yawatín-sona DIST = DEF 3.POSS . II -APPL -five-one ‘That was the sixth one.’ (112) Kátuala pe alula gunamán=ngo, ngápa Yusup nga, gálemuyi. kátuala pe alula gunamán=ngo ngá-pa Yusup=nga dog pig bark afterward=DIST 1EXCL . POSS . II -father NFIN .yusuf=DEF gá-le-mu-yi 3.II -APPL-run-? ‘After the dog barked at the animal, Yusuf chased after it.’ 4.3.3 Adverbial and manner adverbs The meaning and use of adverbial (ingu and angu) and manner demonstratives (inína and amána) is more straightforward. Also, neither of these types can be combined with the =o enclitic. Examples of each are given between (113) and (116).

Kula

225

(113) Anawe yasu ingu tánágana. anawe ya-su ingu tánágana all 2.I - NFIN .come here FIN .gather ‘Everyone come here and gather together.’ (114) Angu ngwít ngku. angu ng-mít ng-ku there 1EXCL . I - NFIN .sit 1EXCL . I - NFIN .stay ‘I was sitting there.’ (115) Anga inína. anga inína DIST FIN .like.PROX ‘That (story) was like this.’ (116) Amáng-amáng aku gun amána, lula pilawáka dán letani. amáng~amáng a-ku gun amána lula pilawáka RDP ~ NFIN .like.DIST 3.I -NFIN .stay NFIN . EVID FIN .like.DIST go.DIST month dán

le-tani

NFIN .one APPL- FIN .reach

‘So it went like that for a month.’

4.4 Prenominal topic marker gána Finally, gána occurs in a prenominal slot separate from the rest of the noun phrase. Examples are given in (117a) and (117b). Identified as a topic marker in Sawila (Kratochvíl 2014), the difference from other ‘topic’ markers (=si in Kula and =si and =ba in Sawila) remains to be investigated. (117) a. Amána gána pe inga. . . amána gána pe=nga like.DIST DIST.TOP pig=DEF ‘Like that, that pig. . .’ b. Gána sen anto ngáwe pake nanu. gána sen ang=to ngá-we-pake nanu DIST.TOP money DIST =also 1EXCL . I -APPL -NFIN .use FIN . NEG ‘I didn’t use that money either.’ (lit., ‘That money, I didn’t use it either.’)

226

Nicholas J. Williams

4.5 Possession Possession in Kula is marked by a prefix on the possessed noun for the possessor. The possessor is also optionally expressed as a separate noun phrase immediately preceding the possessed noun, as in (118a) and (118b). (118) a. Edu gisuba Edu gi-suba Edu 3.POSS . II -house ‘Edu’s house’

b. gisuba gi-suba 3.POSS . II -house ‘his/her house’

Kula possessive prefixes belong to one of the three sets listed in Table 11. Nouns occur with only one set of prefixes. Set I and II prefixes are common, each occurring with a roughly equal proportion of nouns. Set III prefixes are used on a very small group of nouns, so far attested with kárík ‘finger’, kul ‘foot’ and kás ‘foot (sore)’. Table 11: Possessor prefixes Set I

Set II

Set III

-yo ‘wife’

-yo ‘shed’

-kárík ‘finger’

1EXCL

ng-yo

ngá-yo

nge-kárík

1INCL

i-yo

igá-yo

ige-kárík

2

ya-yo

yi-yo

ye-kárík

3

a-yo

gi-yo

ge-kárík

DISTR

tá-yo

ti-yo

te-kárík

Whether a noun takes prefixes from set I or set II is lexicalized and based loosely on the concept of alienability. Set I prefixes occur primarily with kinship and bodypart nouns (119), while set II prefixes occur with a wider range of nouns (120). Obligatorily possessed nouns (mostly kinship and bodypart terms) occur in both set I and set II. Thus, not all kinship and bodypart terms occur with set I/inalienable possessor prefixes (121). (119) a. atamu b. ngikwa c. ntán a-tamu ng-nikwa ng-tán 1EXCL . POSS . I -NFIN .eye 1EXCL . POSS . I - NFIN .hand 3.POSS . I - FIN .grandchild ‘his/her grandchild’ ‘my/our eye’ ‘my/our hand’

Kula

227

(120) a. gisuba b. giseng c. ngápte gi-suba gi-seng ngá-páte 3.POSS . II -money 1EXCL . POSS . II -corn 3.POSS . II -house ‘his/her house’ ‘his/her money’ ‘my/our corn’ (121) a. gipa b. ngáya c. giti gi-pa ngá-ya gi-ti 1EXCL . POSS . II -mother 3.POSS . II -stomach 3.POSS . II -father ‘his/her father’ ‘my/our mother’ ‘his/her stomach’

5 Pronouns Kula has multiple sets of pronouns, though fewer than some other AP languages (see, for example, Schapper 2014 on Kamang). Table 12 presents the five paradigms that appear in the current corpus. Note that the singular/plural distinction has been lost in Kula in all pronouns and pronominal prefixes. Table 12: Pronouns Nominative

Potentive

Focus

Possessive

Dual

1EXCL

ngánu

nta

ngawing

ngánggiya

ngánáku

1INCL

igánu

ite

iwing

igánggiya

ináku

2

yánu

ita

yawing

yánggiya

yánáku

3

gánu

gita

gawing

gánggiya

(g)ánáku

Number is optionally marked on pronouns with the plural clitic =dugwa [=dugwa ~ =dua ~ =du] ‘PL ’. For example, a first person pronoun may refer to a single individual (the speaker) or multiple individuals. The appropriate number, singular or plural, must be inferred by the hearer from the discourse context, if not explicitly marked with =dugwa ‘PL ’. The following examples show a singular first person pronoun referring to a singular individual (122), a plural first person pronoun referring to multiple individuals (123), and a first person pronoun unmarked for number used to refer to multiple individuals (124). (122) Gána ile angu to, ngán nga ngwe ujiani. gána ile angu=to ngán=nga ng-we ujiani TOP year there=also NFIN .1EXCL . NOM = DEF 1EXCL . I -go.LEVEL FIN .exam ‘And that year, too, I went (to take) the exam.’

228

Nicholas J. Williams

(123) Jala araman nga, ngán du nga anawe wetela. jala araman=nga ngán=dugwa=nga anawe we-tela APPL -speak female NFIN .family=DEF NFIN .1EXCL . NOM =PL = DEF all ‘The women’s group, we all discussed (it).’ (124) Ngán nga, ngáklompok mi nga, an tela klompok samudra. ngán=nga ngá-klompok mi=nga ang tela NFIN .1EXCL . NOM = DEF 1EXCL . POSS . II -group take=DEF DIST speak klompok samudra group ocean ‘Us, our group, it’s called the ‘Ocean group.’ Plurals for second and third person pronouns function the same as for the first person. Since plural marking is optional, a simple second or third person pronoun form (yán- or gán-) can refer to either singular or plural participants. The remaining form, igán-, refers to the speaker and addressee(s), i.e., first person inclusive. In order to explicitly refer to a non-singular referent, pronouns may be affixed with a plural marker. However, not all pronouns allow explicitly plural forms. While the nu-series plural forms are all attested in the corpus, the plural forms for the focus pronouns were only confirmed through elicitation. Explicitly plural forms of the te-series and the possessive pronouns are unattested. Attested pronoun forms with plural marking are given in Table 13. Table 13: Plural-marked pronouns

1EXCL

Nominative

Focus

ngán-dua

ngán-dua-wing

1INCL

igán-dua

igán-dua-wing

2

yán-dua

yán-dua-wing

3

gán-dua

gán-dua-wing

The same plural morpheme can be found suffixed to the verb when it is affixed by bound pronominal prefixes (125). This could be seen as a marker of pluractionality, indicating that many people engaged in many individual acts of drinking. (125) Tuák ne aku dugwa. tuák ne a-ku=dugwa NFIN .palm.wine drink 3.I - NFIN .stay=PL ‘They are drinking palm wine.’

Kula

229

The following subsections describe the functions of each individual pronominal paradigm.

5.1 Nominative pronouns The nominative pronouns are the most frequently used free pronouns in Kula. In their full, vowel-final form, these pronouns encode single arguments of nonverbal clauses, (126) and (127). (126) Wansa, yánu wansa. gwangsa yánu gwangsa FIN .2.NOM big big ‘Great, you are great.’ (127) Ang gánu. ang gánu DIST FIN .3.NOM ‘That’s it/him/her’ (lit., ‘That is s/he (NOM )’). In their reduced, consonant-final form, they encode both S, as in (128) and (129), and A, as in (131) and (132), but never P arguments. ngán for S argument (128) Ngán nga Tombang nsiyi. ngán=nga Tombang ng-si yi FIN .1EXCL . NOM =DEF Tombang 1EXCL . I -come.LOW go.LOW ‘I went down to Tombang.’ gán for S argument (129) Gán to duka. gán=to duka NFIN .3.NOM =also stand ‘He also stood up.’ ngán for SP argument (130) Ngán ngo . . . ngánakali. ngán=ngo ngá-na-kali NFIN .1EXCL . NOM =DIST 1EXCL . II - INV-not.want ‘I don’t want (that).’

230

Nicholas J. Williams

ngán for A argument (131) Ngán nga musa ska ngá-gwina. ngán=nga musa síka ngá-gwina NFIN .1EXCL . NOM = DEF bow arrow 1EXCL . II -hold ‘I held my bow and arrow’ yán for A argument (132) Tabila yán ngo pe wáka nanu. tabila yán=ngo pe wáka nanu PROH NFIN .2.NOM =DIST pig roast FIN . NEG ‘You are not allowed to roast the pig!’ While nominative pronouns are never used to encode standard P arguments in a regular transitive clause, they are attested for the P-like argument of a transitive affective verb using the na-prefix (see section 6 for more on na-). However, since there is no explicit A argument in these types of constructions, the single argument behaves more like an S argument – thus allowing a nominative pronoun. An example is given in (133). gán for SP (133) Stel mde gán to gánami da. . . stel míde gán=to gá-na-mi da install go.HIGH NFIN .3.NOM =also 3.II - INV-take SEQ ‘Set it up there so he gets recorded too. . .’ Also worth noting is the frequent practice of marking nominative pronouns with a demonstrative or determiner, most frequently the distal adnominal demonstrative → definite determiner, =nga (see examples 128, 130, 131, 132). This appears to encode a type of contrastive focus, but remains to be investigated further.

5.2 Potentive pronouns Potentive pronouns appear much less frequently in Kula texts than the nominative pronouns. The third person form, gíta, is relatively frequent, while ite and nte/nta are much less common. There are no attested examples of ita in the corpus, only from direct elicitation. There is no difference between potentive and nominative pronouns with regard to position in the clause. Examples of this series of pronouns are given in (134–136). These pronouns typically encode agentive arguments that have or had the potential or intention to

Kula

231

do some action (134–135), similar to their cognate paradigm in Sawila (Kratochvíl 2014). However, the Kula pronouns appear in a broader range of contexts, encoding the agentive arguments of completed actions (136) in addition to future actions. (134) Ngáya dua anto su to nsurán da nta gotela nanu. ngá-ya=dugwa=ng=to su=to ng-surán da 1EXCL . POSS . II -mother=DEF = PL NFIN .come=also 1EXCL . I - NFIN .angry SEQ nta go-tela nanu 1EXCL . POT 3.IV-speak FIN . NEG ‘Those women also came, (but) I was angry so I wouldn’t speak to them.’ (135) Aya si nunung goko pisi gaya? Nanu, ite giana. aya si nunung goko pis gaya nanu rain come.LOW how leaf.type NFIN .cut FIN . PROSP FIN . NEG ite giana 1INCL . POT FIN .leave ‘It’s raining, how (can we) cut ‘goko’ leaves? No, we’ll just leave.’ (136) Gíta muyi, anuna sen ong gíta mi legasám nanu, gíta pake agátu. gíta mu-yi anuna sen=ong gíta mi le-gasam 3.POT run-? because money=PROX 3.POT take APPL- NFIN .store nanu

gíta pake agátu FIN . NEG 3.POT NFIN .use gone ‘He ran away, because he hadn’t saved our money. He used it all up.’

5.3 Focus pronouns Focus pronouns consist of a pronominal prefix plus the focus particle, winga. The same focus particle is also present elsewhere, on non-pronominal NPs (137). (137) Kapala kantor winga bantuan mi me duk ng wegani. kapala kantor winga bantuan mi me duka=nga we-ga-yáni take come stand=DEF APPL-3.II -give head office FIN . FOC help ‘It’s the office head who, coming and standing here, gave him the help.’ The focus pronouns are typically glossed in Indonesian with a focus construction (e.g., saya yang . . . ‘It’s me who . . .’), implying that these forms encode a kind of identificational focus. Examples of the focus pronouns are given below.

232

Nicholas J. Williams

(138) Awing baba ayám nta ngáwing ngágusu? awing baba ayám nta ngáwing ngá-gusu 3.NFIN . FOC beat NFIN .die or 1EXCL . FOC 1EXCL . II - FIN .shoot ‘Did he1 beat him2 to death or did I shoot him2 with an arrow?’ (139) Yawinga yawe lona lika lewaka gaya. yawinga ya-we lona lika le-waka gaya 2.FIN . FOC 2.I -go.LEVEL snake big APPL-watch FIN . PROSP ‘You are the one who will go watch over the big snake.’

5.4 Possessive pronouns These pronouns are used in place of a typical possessed noun phrase construction. Examples are given in (140) and (141). (140) Talona, yali? Yángiya? Talona, yángiya kása? Talona ya-ali yánggiya Talona yánggiya kása Talona 2.POSS complete Talona 2.II -buy 2.POSS ‘Talona, did you buy (areca nut)? Is this yours? Talona, is yours gone?’ (141) Anawe igánggiya. anawe igánggiya all 1INCL . POSS ‘They (areca nut trees) are all ours.’

5.5. Dual and other quantificational pronouns Several sets of quantificational pronouns exist as well. The most common are dual pronouns, transparently derived from the nominative pronouns and the numeral yáku ‘two’. These function as typical non-numeral independent pronouns, as demonstrated in (142). (142) Martin mde gunamánnga ngánáku nguda lengwíti. martin míde gunamán=nga ngánáku ng-muda Martin come.HIGH GLOSS =afterward=DEF 1EXCL . DU 1EXCL . I -go.HIGH le-ng-míti APPL -1EXCL . I - FIN .sit ‘(The truck) having come up from Maritaing, the two of us got on (it).’

Kula

233

In elicitation, nominative pronouns can be combined with nearly any numeral. The forms attested in the current corpus, however, are restricted to numerals two through four. Examples with numerals three and four are given below. (143) Ngántu mpati. ngán-tu ng-pati 1EXCL . NOM -three 1EXCL . I - FIN .eat ‘The three of us eat (it).’ (144) Ngánáku kdas nanu ampo ngángarasiku. ngánáku kída=si nanu ampo ngán-arasiku 1EXCL . NOM -four 1EXCL . DU just=TOP FIN . NEG but ‘It wasn’t just the two of us, but the four of us.’ These numeral pronouns have two additional uses, as a pronoun in a complex noun phrase expressing the speaker and one other participant (145) and as a predicative element on its own (146). (145) Wanta dán nga ngápa Yusup ngánáku tasola. wanta dán=nga ngá-pa Yusup ngánáku ta-sola NFIN .one=DEF 1EXCL . POSS . II -father NFIN .Yusuf 1EXCL . DU DISTR-invite day ‘One day, Yusuf and I urged/invited each other [to] . . .’ (146) Ngánáku, awa Halena to gipa anáku ngángarasiku. ngánáku awa Halena=to gi-pa anáku ngán-arasiku 1EXCL . DU then Halena=also 3.POSS . II -father 3.DU. NOM 1EXCL-four ‘There were two of us, then with Halena and her father, there were four of us.’

6 Verbal prefixes: person marking and applicatives The basic Kula verb template is given in (147). There are at least four distinct slots in the Kula template, although typically no more than three are filled at the same time. There is also variation in which prefixes occur and in which order. The template in (147) constitutes a preliminary characterization. Verb template (147) AGR 2-INV-APPL-AGR 1-V

234

Nicholas J. Williams

AGREEMENT SLOT 1 (AGR 1) occurs closest to the verb and is occupied by agreement prefixes from each of the five sets (described in 6.1 below). Set I and II prefixes mark S, A and P arguments, depending on the verb, while set III and IV prefixes mark primarily P arguments. Set V prefixes mark only A arguments on a small set of transitive verbs. The APPLICATIVE SLOT comes next, between AGR 1 and the affectedness marker. This slot is filled by one of three applicative prefixes (le-, mí-, we-). The INVERSE SLOT (INV) is occupied by only one morpheme, na-. The outermost prefix slot is the AGREEMENT SLOT 2 (AGR 2). This second agreement slot can be occupied by prefixes from sets I, II or III and may co-occur with prefixes in the AGR 1 slot. Use of AGR 2 with no prefix in AGR 1 is also possible with use of the affectedness and applicative prefixes. This set of possibilities is shown in examples (148)–(154). Note that no single verb form has all four slots occupied at the same time. AGR 1-V (148) Nte ngkuya. n-te ng-kuya 1EXCL . I S-sleep 1EXCL . I S-FIN .stay ‘I/we are sleeping.’ APPL -AGR 1-V (149) Pá míngkuya. pá mí-ngS-kuya NFIN .planted.field APPL -1EXCL . I - FIN .stay ‘I’ve been living in the fields.’ AGR 2-APPL-V (150) Kálán nga ngápárenta gi nga, yálemagina. yáA-le-magina kálán=nga [ngá-párenta gi=nga]P NFIN .now=DEF 1EXCL . POSS . II -order NFIN . PROG =DEF 2.II -APPL- FIN .hear ‘Now, listen to my orders.’ AGR 2-AGR 1-V (151) Lámána ginana gagogwita. gáA-goP-gwita lámána [gi-nana]P then 3.POSS . II -older.sibling 3.II -3.IV-call ‘Then he called his older brother.’ AGR 2-INV-APPL-V (152) Latala inálesayána. latalaA iP-na-le-FIN .sayána god 1INCL . II - INV-APPL-care.for ‘God loves us.’

Kula

235

AGR 2-INV-V (153) Nganalosa. ngáP/S-na-losa 1EXCL . II - INV-tired ‘I/we are exhausted.’ AGR 2-INV-V (154) Kaweana ngánapati. kaweanaA ngáP-na-pati mosquito 1EXCL . II - INV-FIN .bite ‘The mosquito bit me.’ In the rest of this section I will present an outline of verbal person marking in Kula (6.1) and the use of three applicative prefixes in Kula (section 6.2).

6.1 Agreement prefixes Kula has five paradigms of person marking prefixes (Table 14). Morphological similarity across the sets is readily apparent. The primary difference across the sets is the quality of the vowel (/a/ or no vowel for set I, /á/ for set II, /e/ for set III, /o/ for set IV, and /i/ for set V). While possessor prefixes on nouns show morphological similarity to these person marking prefixes, the paradigms are not identical (see section 4.5). Table 14: Agreement prefixes Set I

Set II

Set III

Set IV

Set V ngi-

1EXCL

ng-/nga-

ngá-

nge-

ngo-

1INCL

i-

i- ~ igá-

ige-

igo- ~ gio-

igi-

2

ya-

yá-

ye-

yo-

yi-

gá-

ge-

go-

gi-

tá-

te-

to-

ta-

3

Ø-

DISTR

ta-

a-

As in pronouns (section 5), number is not marked in Kula agreement prefixes. Plurality is optionally marked with =dugwa ‘PL ’ encliticizing to the verb, as in (155). This plural marking expresses both plurality of the A arguments and pluractionality of the event. This example comes from a narrative problem solving task in which the speaker is referring to multiple actions of multiple participants across a series of pictures used in the task.

236

Nicholas J. Williams

(155) Nadua paku dua guna? guna na=dugwa aA-pe aA-ku=dugwa what=PL 3.I -do 3.I - NFIN .stay=PL FIN . EVID ‘What are they all doing?’ First person forms are marked for clusivity. Using set I forms to illustrate, the i- form marks first person inclusive, while the ng- form marks first person exclusive. An example of the first person inclusive, i-, is given in (156). (156) Kuma-pu tela ilula pegang gipo ngán nga ngakuma puya. pe gang gí=po kuma-pu tela iS-lula back-NFIN .break speak 1INCL . I -go.DIST pig NFIN .hunt NFIN . PROG =but ngán=nga nga-kuma puya NFIN .1EXCL . NOM = DEF 1EXCL . POSS . II -back FIN .break ‘Kumapu said, “Let’s go hunting, but I’m crippled (lit., my back is broken)”.’ The exclusive form, ng- can be interpreted as singular or plural. For example, in (157) below, the first person exclusive prefix ng- on the verb muda ‘go.HIGH ’ is interpreted as plural due to the plural marking on the first person exclusive pronoun, ngán. Without the independent pronoun, the verb could be interpreted as singular or plural, as in (158). (157) Ngándua to nguda lámána giana. lámána giana ngán=dugwa=to ngS-muda NFIN .1EXCL . NOM =PL =also 1EXCL . I -go.HIGH immediately FIN .leave ‘. . . and then we got up (onto the horses) and left.’ (158) Igá limpa nsu. igá limpa nS-su NFIN .path long 1EXCL . I - NFIN .come ‘I/we came a long way.’ Set I and Set II prefixes are similar in form, with major differences only in the third and first person forms. Set I prefixes are used to encode S arguments of intransitive verbs, while set II prefixes mark A and P participants of transitive verbs. Set III and IV prefixes are used to mark P participants when P is a nonprototypical patient, typically a locative goal (set III) or source (set IV) arguments. Set III prefixes are also used on at least one intransitive verb (giana

Kula

237

‘leave’, see section 6.1.1.4). Set V prefixes are used on a small set of /a/-initial transitive verbs. A few basic examples of each prefix set are given below. Set I (159) Nsu tenu. tenu ngS-su 1EXCL . I - NFIN .come FIN . PFV ‘I/we have come.’ Set II (160) Pte ngátana. páte ngáA-tana corn 1EXCL . II -plant ‘I/we plant corn.’ Set III (161) Ngegian letumpa. le-tumpa ngeS-gian 1EXCL . III -NFIN .travel APPL-continue ‘And then we continued traveling.’ Set IV (162) Ngayogwita yame. ngáA-yoP-gwita yaS-me 1EXCL . I -2.IV-call 2.I -come.LEVEL ‘I called you to come here.’ Set V (163) Waikíki ngialomana. [waikíki=dugwa=to]P ngiA-alomana candlenut=PL =also 1EXCL .V- FIN .choose ‘I/we chose candlenuts too (pick them from the ground).’ I present detailed descriptions of person marking for intransitive (section 6.1.1) and transitive verbs (section 6.1.2), as well as valency increasing uses of person prefixes (section 6.1.3). 6.1.1 Agreement on intransitive verbs Intransitive verbs can be divided into three classes, based on their combination with Sets I to V. The marking of S on the three basic classes is set out in Table 15.

238

Nicholas J. Williams

Person marking on intransitive verbs primarily involves the first agreement slot (AGR 1 ). Class I and II verbs use the second agreement slot (AGR 2 ) due to the presence of the affectedness marker. Table 15: Intransitive verb classes Prefixation pattern

Membership

Class I

Set I prefix for S (AGR 1 )

Largest class – many verbs

Section 6.1.1.1

Class II

Set II prefix for S (AGR 2 ) + affectedness marker

10+ verbs

Section 6.1.1.2

Class III

No prefix, or Set II prefix for S (AGR 2 ) + affectedness marker

four verbs

Section 6.1.1.3

In addition to these classes, two verb roots display idiosyncratic agreement patterns that cannot be captured with any agreement class. These are treated in section 6.1.1.4. 6.1.1.1 Class I intransitive verbs Class Intr-I8 is the largest intransitive verb class in Kula and includes many frequently used intransitive verbs, including verbs of motion, posture, and consumption. These verbs combine with set I prefixes to mark their single S arguments. A few basic examples are given in (164) to (166). (164) Yamíti. yaS-míti 2.I - FIN .sit ‘You sat (down)’ or ‘Sit (down)!’ (165) Ngwala. ngS-wala 1EXCL . I -intoxicated ‘I am drunk.’ (166) Iwe. iS-we 1INCL . I -go.LEVEL ‘We went.’ 8 References to classes in the proceeding discussion will use the abbreviations Intr-1, Intr-2, etc. and Tr-1a, Tr-1b, etc. to avoid confusion between transitive and intransitive verbs.

239

Kula

There are two subclasses within class Intr-I, based on observed variation in the 3rd person forms of verbs.9 While subclass Intr-IA verbs show no overt third person prefix, subclass Intr-IB verbs are marked with a prefix a-. Otherwise the agreement inflections are identical. Table 16 lists sample paradigms of verbs from each class, while Table 17 lists additional verbs from each class. Table 16: Inflections of subclasses IA and IB Subclass IA: null-prefixed

Subclass IB: a-prefixed

lula ‘go.DIST ’

te ‘sleep’

1EXCL

n-lula

n-te

1INCL

i-lula

i-te

2

ya-lula

ya-te

3

lula (*a-lula)

a-te (*te)

DISTR

ta-lula

ta-te

Table 17: Sample verbs from subclasses IA and IB IA Verbs

IB Verbs

su(ya)

‘come’

páma

‘consume’

lula

‘go.DIST ’

wala

‘intoxicated’

we

‘go.LEVEL ’

yámu

‘die’

me

‘come.LEVEL ’

te

‘sleep’

ji

‘go.LOW ’

ku

‘stay’

si

‘come.LOW ’

lilawa

‘return’

mda

‘go.HIGH ’

mde

‘come.HIGH ’

mu(ya)

‘run’

wawán

‘think’

surána

‘be.angry’

míti

‘sit’

The distinction between subclass Intr-IA (null-prefixed) verbs and subclass IntrIB (a-prefixed) verbs shows some semantic tendencies: most of the null-prefixed forms belong to the class of basic motion and positional verbs, while the a-prefixed verbs come from a range of semantic classes. Subclass Intr-IA is more frequent than Intr-IB, with most intransitive verbs falling under subclass Intr-IA. 9 In the first person, we also find variation, but this is limited to surface allomorphy of the first person exclusive prefix, as described in section 2.6.1.

240

Nicholas J. Williams

6.1.1.2 Class II intransitive verbs Verbs in class Intr-II occur with a set II agreement prefix and an obligatory inverse prefix na-. The function of this prefix has been characterized as marking inverse voice on a subset of transitive verbs (Donohue 1996), but on this class of verbs its presence is entirely lexicalized. The sentences in (167) to (169) provide examples of verbs in class II used with the obligatory na- prefix. (167) Ngánalosa. (*Ngálosa.) ngáS-na-losa 1EXCL . I - INV-tired ‘I’m/we’re exhausted.’ (168) Lala pu dápa kála, iji amáng gada inaparekang nanu. lala pu dápa kála iS-ji amáng gada wave NFIN .break PRIOR then 1INCL . I -go.LOW NFIN .like.DIST so.that iS-na-parekang nanu 1INCL . II - INV-soaked FIN . NEG ‘(We waited for) the waves to break first, before we got down (from the canoe), so that we wouldn’t get wet.’ (169) Níweji Lantukam skola si ngánakali. ní-we-ji Lantuka=mu skola=si ngáS-na-kali LOC -go.LEVEL -go.LOW Lantoka=LOC school=TOP 1EXCL . II - INV-not.want ‘I did not want to go to school down in Lantoka.’ Table 18 provides a non-exhaustive list of verbs in class Intr-II, based on currently available data used for this sketch. This verb class contains, for the most part, stative verbs with an “affected” S argument. Table 18: Sample Verbs in Class Intr-II Root

Gloss

Root

Gloss

bíki

‘FIN .strong’

masína

‘FIN .hungry’

kali

‘dislike’

mata

‘sick’

kon

‘small, young’

mulana

‘NFIN .not.want’

loki

‘FIN .wet’

parekang

‘soaking.wet’

losa

‘tired’

tan ( ji/si)

‘fall down’

Kula

241

6.1.1.3 Class III intransitive verbs Class Intr-III intransitive verbs code their S argument in one of two ways: S can be unmarked on the verb or can be marked with a set II agreement prefix plus na-, in the manner of a class II intransitive verb. In (170) and (171), the (a) constructions without an agreement or inverse prefix encode a more permanent state, while the (b) constructions imply a process and possible unexpressed agent. For instance, the verb in (170b) can be combined with the perfective aspect marker, tenu, to express “I have already grown up.” (170) a. Ngán wansa. ngán gwangsa NFIN .1EXCL . NOM big ‘I am big/important.’ (171) a. Ngán limpa. ngán limpa NFIN .1EXCL . NOM tall ‘I am tall.’

b. Ngánawansa. ngáS-na-gwangsa 1EXCL . I - INV-big ‘I am big/important.’

b. Ngánalimpa. ngáS-na-limpa 1EXCL . I - INV-tall ‘I am tall.’

The class is very small, with only four verb roots being attested in the corpus thus far: (i) limpa ‘tall’, (ii) mata ‘sick’, (iii) paniki ‘suffer’, and (iv) gwangsa ‘big’. 6.1.1.4 Intransitive verbs with idiosyncratic agreement patterns Yima ‘hot’ and giana ‘leave, go’ have idiosyncratic agreement patterns that are not consistent with any agreement class that we have been able to establish. The verb yima ‘hot’ can be used with set II prefixes and the affectedness prefix na- (like a Class II intransitive verb), or with a set IV prefix, as seen in (172). In (172a), the set II prefix encodes an experiencer S argument, while in (172b), the set IV prefix encodes a reflexive recipient. (172) a. Ngánayima. ngáS-na-yima 1EXCL . II - INV-hot ‘I have a fever.’

b. Ngoima. ngoA/P-yima 1EXCL . IV-hot ‘I warm myself (e.g., by the fire).’

Finally, the intransitive verb, giana ‘leave, go’ typically occurs without a prefix (173a), but can also take a set III prefix (173b), a paradigm otherwise reserved for certain transitive verbs (see section 6.1.3.1).

242

Nicholas J. Williams

(173) a. Nanu, ite giana. nanu ite giana FIN . NEG 1INCL . POT FIN .leave ‘No, we’ll just leave.’

b. Ngegiana. ngeS-giana 1EXCL . III - FIN .leave ‘I’m leaving.’

6.1.2 Agreement on transitive verbs Transitive verbs exhibit complex agreement patterns involving both AGR 1 and AGR 2 . The predominant pattern is to mark A arguments with a set II prefix. Some verbs are attested with the same agreement prefixes for non-third person P arguments as well. In a few cases, human third person P arguments are marked with a prefix. The three major classes of transitive verbs and their agreement marking patterns are laid out in Table 19, followed by examples of A and P marking with prefixes. Table 19: Basic transitive verb classes

Class IA

Prefixation pattern

Membership size

Set II prefix for A, and Set II prefix for P

many

Section 6.1.2.1

Class IB

Set II prefix for A, and Set II prefix + na- for P

~10

Section 6.1.2.1

Class II

Set II prefix for A, and set IV prefix for P

~10

Section 6.1.2.2

Class III

Set V prefixes for A

~5–10

Section 6.1.2.3

Most transitive verbs can occur without the agreement prefixes listed in Table 19. This is common with third person A arguments which are expressed by full noun phrases or pronouns in the clause or preceding clause(s). In (174) and (175), the verbs dage ‘fry’ and mi ‘take’ are used without any person marking prefix. In (174), ngáya ‘my mother’ is the A argument of dage ‘fry’. In (175) the verb mi ‘take’ has no explicit A argument. The A argument of mi is understood as ‘the driver’, identical to the S argument of the intransitive serial verb construction in the immediately preceding clause (in parentheses). (174) Ngáya pte dage. ngá-ya páte dage 1EXCL . POSS . II -mother corn fry ‘My mother fried corn.’

Kula

243

(175) Sufiri mu aku su Nailang tani, giya gitama mi, mi lula wejina. sufiri mu aku su Nailing tani FIN .driver NFIN .run 3.I - NFIN .stay NFIN .come Nailang FIN .arrive gi-ya gi-tama mi mi lula wejina 3.POSS . II -NFIN .water 3.POSS . II -ocean take take go.DIST FIN .fill ‘The driver was running until he came to Nailang, (where) (he) took oil (and) filled up (the tank of the truck).’ Both verbs also occur with person marking prefixes, as shown in examples (176) and (177) below. In (176), taken from a narrative text, the verb dage is used again, now with a third person prefix from set II. In (177), the verb mi occurs with a set II prefix marking the A argument. In this case, no independent nominal for the A argument is present, so gá- functions as the sole marker of the A argument. (176) Ngáya pte dage. Pte gádage kása gunamánnga, gápána ape abu. páte dage páte gáA-dage kása gunamán=nga [ngá-ya]A complete afterward=DEF 1EXCL . POSS . II -mother corn fry corn 3.II -fry gáA-pána ape abu 3.II -grind 3.I -make powder ‘My mother fried corn. After she finished frying the corn, then she ground it into flour.’ (177) Lagimoka gámi níngk ayang legamíti. lagimoka gáA-mi níngk ayang le-ga-míti cooking.pot 3.II -take NFIN .stove NFIN .top APPL- CAUS - FIN .sit ‘He took a cooking pot (and) put it on top of the stove.’ Some class Tr-IB verbs, which typically occur with a set II prefix for the A argument, can also occur with only one prefix marking the P argument. This occurs with third person A arguments acting on non-third person P arguments. Examples (178) and (179) show class Tr-Ib verbs with set II prefixes for their A argument, while examples (180) and (181) show the same verbs with marking of a human P argument and no prefix for the A argument. These verbs require use of the na- inverse prefix with a human P argument, following the pattern originally identified by Donohue (1996).

244

Nicholas J. Williams

(178) Tabak dáng yapati. tabak dáng yáA-pati tobacco NFIN .one 2.II - FIN .eat ‘Chew (lit. eat) some tobacco.’ (179) Ngán=nga musa ska ngágwina. ngán=nga musa síka ngáA-gwina NFIN .1EXCL . NOM =DEF bow arrow 1EXCL . II -hold ‘I hold my bow and arrow.’ (180) Kaweana ngánapati. kaweana ngáP-na-pati mosquito 1EXCL . II - INV- FIN .bite ‘The mosquito will bite me’ or, possibly, ‘I’ll get bitten by a mosquito’. (181) Ado, yanagwina tenu. ado yáP-na-gwina tenu INTER 2.II - INV-hold FIN . PFV ‘Oh, it’s got you now!’ Class Tr-II verbs may also occur with set IV prefixes for their P argument and set II prefixes to mark A arguments. In many attested uses the set II prefix for the A argument is absent, as in (182). (182) Gimni inga lapun dígín lula giya gipa gotatuku. gi-muni=nga lapun dígín lula gi-ya 3.POSS . II -fragrance=DEF wind NFIN .lift go.DIST 3.POSS . II -mother gi-pa goP-tatuku 3.POSS . II -father 3.IV- FIN .tell ‘The wind brought the (bean’s) fragrance and told his mother and father (that their son had cooked the beans).’ Aside from these cases, transitive verbs showing agreement for P arguments alone are rare. Class Tr-IA verbs never occur with a prefix marking only the P argument. Marking of P arguments, with no prefix for the A argument, is limited to class Tr-II and Tr-IB transitive verbs. It is worth noting that non-third person A arguments may also be unmarked on transitive verbs when the argument is expressed in the same clause or immediately preceding clause on another verb or by means of an independent pronoun. For example, in (183) the class Tr-IV transitive verb lata occurs with no

Kula

245

prefixes. The A argument is expressed on the verb we and the P argument is expressed as a full noun phrase, igámama inga. (183) Iwe pa igámama inga lata. igá-mama=nga lata i-weS/A=pa 1INCL . I -go.LEVEL =so.that 1INCL . POSS . II -field=DEF burn ‘Let’s go burn our fields.’ To summarize, agreement prefixes on transitive verbs in Kula are used predominantly for marking A arguments. The same agreement prefixes are used to mark non-third person P arguments and, in a few cases, human third person P arguments. In sections 6.1.2.1–6.1.2.3, I describe the use of agreement prefixes on transitive verbs in more detail. 6.1.2.1 Class I transitive verbs (IA and IB) Class Tr-IA is the largest set of transitive verbs. Verbs in this class most commonly occur with set II prefixes marking the A argument of the verb. Examples with the verb mi ‘take’ and each set II prefix are given in examples (184) to (186). (184) Wata dua to ngámi. wata=dugwa=to ngáA-mi coconut=PL =also 1EXCL . II -take ‘I took some coconuts, too.’ (185) Asáka gomán-gomán áma, ansi yámi. asáka gomán~gomán=áma an=si yáA-mi wood RDP ~ NFIN .there.LEVEL =DEM DIST = TOP 2.II -take ‘Those other kinds of wood, that you (can) take.’ (186) Pte gámi we tana. tana páte gáA-mi we corn 3.II -take go.LEVEL FIN .plant ‘They took corn there and planted (it).’ There is no instance of igá- with the verb mi in my corpus, so I provide an example with another class Tr-I verb, mádína ‘to plant’ in (187). (187) Ya, maka igámádína. yaʔ maka igáA-mádína INTER banana 1INCL . II - FIN .plant ‘OK, let’s plant bananas.’

246

Nicholas J. Williams

While most Tr-I verbs occur with a set II prefix for the A argument, human P arguments may also be marked with an agreement prefix. There are two distinct patterns for marking human P arguments with Tr-I verbs, here labeled class Tr-IA and Tr-IB. Class Tr-IA verbs are attested with two set II prefixes, one for A and one for P, while Class Tr-IB verbs require use of the na- prefix and, unlike Tr-IA verbs, allow just one prefix, which is for the P argument. Examples of Tr-IA are given in (188) to (190). Attested examples of Tr-IA verbs are restricted to these three verbs in the current corpus. (188) Yán ngáyasi yalula. yaS-lula yán ngáA-yáP-asi NFIN .2.NOM 1EXCL . II -2.II -command 2.I -go.DIST ‘You, I told you to go.’ (189) Ngáyátána kang-kang mawo. ngáA-yáP-tána kang~kang mawo 1EXCL . II -2.II -teach RDP ~ NFIN .good TAM ‘I’ve taught you well.’ (190) Tinale naduas ngáyayat nanu? tinale na=dugwa=si ngáA-yáP-ayat nanu last.night what=PL = TOP 1EXCL . II -1II -inform FIN . NEG ‘What didn’t I tell you (to do) last night?’ There are no examples of third person P arguments marked on Tr-IA verbs in the corpus. While two set II prefixes are allowed, this is only attested for first and second person arguments. Very few verbs are attested with P arguments explicitly marked by a pronominal prefix. In all existing cases, the P argument is human and 1st or 2nd person. Expression of the P argument seems to be allowed only for highly animate arguments. Class Tr-IB is a smaller set of transitive verbs. The set of attested Tr-IB verbs is listed in Table 20. Table 20: Sample Class Tr-IB verbs baba gwina láta mi ne pati plíng

‘hit’ ‘hold’ ‘slap’ ‘take’ ‘drink’ ‘eat, bite’ ‘accompany’

Kula

247

The distinguishing feature of Tr-IB verbs is that they require use of the affectedness prefix, na-, for marked P arguments. An example of a first-person P argument with a Tr-IB verb is given in (191). The verb pati ‘eat, bite’ occurs with a set II prefix marking the first person P argument, while the third person A argument is unmarked. (191) Ngánapati. ngá-na-pati 1EXCL . II P-INV-FIN .bite ‘It (will) bite me.’ Similarly, a human third person P argument can be expressed with no prefix for the A argument (192). Notice the presence of na- again and the distinct sense of the verb here. This patterns like other SP arguments. (192) Gán to gánami! gán=to gáP-na-mi NFIN .3.NOM =also 3.II - INV-take ‘Record (lit., take) him too!’ Like Tr-IA verbs, Tr-IB verbs mark both first and second person P arguments with set II prefixes, and both arguments may be marked on one verb. However, Tr-IB verbs require use of na- in each case. (193) Yángánami. yáA-ngáP-na-mi 2.II -1EXCL . II - INV-take ‘You take me.’ (194) Ngáyánaplíng giana. giana ngáA-yáP-na-plíng 1EXCL . II -2.II - INV-NFIN .accompany FIN .leave ‘I (will) accompany you (there).’ The prefix na- is attested with a third person P argument in a kind of reciprocal construction (195). Notice the absence of the distributive prefix, ta-, often used in the expression of reciprocal events along with the inverse marker, e.g. ta-na-baba ‘hit each other’.

248

Nicholas J. Williams

(195) Gánaplíng amáng we-we. gá-na-plíng amáng we~we 3.II - INV-NFIN .accompany NFIN .like.DIST RDP ~go.LEVEL ‘They went like that leading each other along.’ 6.1.2.2 Class Tr-II transitive verbs Tr-II verbs constitute a small class; all four instances attested in the corpus are given in Table 21. These verb roots occur with a Set IV prefix marking P (see section 6.1.1.4 for the distinct valence-increasing use of set IV prefixes with intransitive verbs). Examples are given in examples (196) to (199) below. In each case, the P argument is a non-prototypical patient. In (196) and (198), the P is a recipient, and in (197) a goal. Table 21: Sample Class Tr-II verbs gwita

‘call’

musu

‘help’

te

‘dig’

lata

‘burn (a field)’

(196) Ngápa nga ngágogwita. ngá-pa=nga ngáA-goP-gwita 1EXCL . POSS . II -father=DEF 1EXCL . II -3.IV-call ‘I called my father.’ (197) Lula kayubaka gote gisukwa. lula kayubaka goP-te gisukwa go.DIST turmeric 3.IV-dig probably ‘They’re probably digging up turmeric roots.’ (198) Wána ngku le ngángka ayám da yángomusu. le ngá-ngka ayám da wána ngS-ku carry 1EXCL . I - NFIN .stay NFIN .finish 1EXCL . POSS . II -shoulder NFIN .die SEQ yáA-ngoP-musu 2.II -1EXCL . IV-FIN .help ‘I carried (them) all until my shoulders were dead, so you (should) help me.’ (199) Doki winga gagolata. doki winga gáA-goP-lata mouse FOC 3.II -3.IV-burn ‘The mouse was the one who burned it (=the field).’

Kula

249

6.1.2.3 Class Tr-III transitive verbs Class Tr-III consists of a handful of transitive verbs, all with roots beginning in a- (Table 22). These verbs take set V prefixes to encode A arguments. Table 22: Sample class Tr-III verb roots ape

‘make, do’

alomana

‘pick, choose’

awanta

‘do in place of’

adi

‘see’

aluatan

‘follow’

The following examples illustrate the agreement pattern of this class of transitive verb. (200) Nadua pka dáng jiape tenu. jiA-ape tenu [na=dugwa puka dáng]P little NFIN .one 2.V-make FIN . PFV what=PL ‘What (little) thing have you done?’ (201) Waikiki dua to ngialomana. [waikiki=dugwa=to]P ngiA-alomana candlenut=PL =also 1EXCL .V- FIN .pick ‘I picked (through) the candlenuts too.’ 6.1.3 Valency-increasing uses of agreement prefixes In addition to their standard uses described in section 6.1.2, set III and IV prefixes can also be used with a valency-increasing function with otherwise intransitive verbs. 6.1.3.1 Set III prefixes Set III prefixes can be used with certain intransitive verbs to add a P argument. This use is observed primarily with motion verbs where it adds a human goal participant, as in (202). The prefixing pattern of the S argument in its basic intransitive use is unaffected by the valency-increasing Set III prefix. So in (202) on me ‘come’ the mover is unmarked while the added goal is marked by a prefix, while in (203) both are marked; the added goal argument occurs in AGR 2 while the mover occurs in AGR 1 .

250

Nicholas J. Williams

(202) Gán geme. geP-me gánA NFIN .3.NOM 3.III -come.LEVEL ‘He1 comes to him2 .’ (203) Genlula. geP-nA-lula. 3.III -1EXCL . I -go.DIST ‘I go to him.’ 6.1.3.2 Set IV prefixes With a handful of intransitive verbs listed in Table 23, set IV prefixes add a P argument. Table 23: Sample verbs occurring with set IV prefixes irína

‘fly’

tatuku

‘tell (a story)’

tela

‘speak’

With the verb irína, for example, the set IV prefixes add a source argument (204). With verbs of speaking (tatuku ‘tell’ and tela ‘speak’), the set IV prefix adds a goal argument (205). (204) Adamasa gogirína. adamasa go-irína bird 3.IV-fly ‘The bird flew away from him (go-).’ (205) Gimni inga lapun dígín lula giya gipa gotatuku. gi-muni=nga lapun dígín lula gi-ya 3.POSS . II -fragrance=DEF wind NFIN .lift go.DIST 3.POSS . II -mother gi-pa go-tatuku 3.POSS . II -father 3.IV-FIN .tell ‘The wind brought the (bean’s) fragrance and told his mother and father (that their son had cooked the beans).’ In some cases, a reciprocal construction requires the use of a prefix from set IV, rather than the more typical set II tá-. This occurs when the verb roots are intransitive and the set IV prefix to- encodes a non-prototypical P argument as

Kula

251

the simultaneous A and P of the reciprocal construction. Examples are given in (206)–(207). These verbs are not attested in the more common reciprocal construction with tá-, but they are attested robustly with other standard agreement patterns. (206) Towe tomeyi. to-we to-meyi DISTR . IV-go.LEVEL DISTR . IV- FIN .come.LEVEL ‘(They) went back and forth on each other (i.e., debating something).’ (207) Pi masa ngo toyágyág burána. pi masa=ngo to-yág~yág burána NFIN .areca betel=DIST DISTR . IV- RDP ~ NFIN .request FIN . INTS ‘We are always asking each other for areca nut.’ In the current corpus, set IV prefixes have also been attested on numerals, with a similar applicativizing effect (208). (208) Gágoyáku. gá-go-yáku 3.II -3.IV-FIN .two ‘Do again, do a second time.’

6.2 Applicative prefixes Kula has three productive applicative prefixes: le-, mí-, and we-. These prefixes combine with transitive and intransitive verb roots to add a participant to the clause. The most general applicative prefix, le-, combines with non-verbal roots as well. The examples below (209a–b) show how the applicative prefix le- adds a location argument to the clause. The third person prefix a- is dropped when the applicative is present (209b). Note that the location cannot be specified without the applicative prefix (209c). (209) a. Gákabaku nga amíti. gákabaku=nga a-míti 3.I - FIN .sit cat=DEF ‘There’s a cat’ or ‘The cat sits.’ b. Gákabaku nga parka le-míti. gákabaku=nga parka le-a-míti rug APPL -3.I - FIN .sit cat=DEF ‘There’s a cat on the rug’ or ‘The cat sits on the rug.’

252

Nicholas J. Williams

c. *Gákabaku nga parka míti. gákabaku=nga parka míti rug FIN .sit cat=DEF In the following sections (sections 6.2.1 to 6.2.3), I will discuss the applicativizing use of these three prefixes in more detail.

6.2.1 leThere are several distinct applicativizing uses for le-. One frequent use of this prefix is with posture verbs (e.g. míti ‘sit’, lila ‘hang’, duka ‘stand’, etc.) to indicate the location of the erstwhile S participant. The locative relations expressed by this function of le- include on top of (210), attachment to a vertical surface (211), and horizontal contact/support (212). However, there is some overlap between the locative relations expressed by le- and those expressed by mí- (see section 6.2.2 on mí-). (210) Gákabaku nga parka lemíti. gákabaku=nga parka le-a-míti rug APPL -3.I - FIN .sit cat=DEF ‘There’s a cat on the rug’ or ‘The cat sits on the rug.’ (211) Saliat nga giasáka lelila. saliat=nga gi-asáka le-lila flag=DEF 3.POSS . II -tree APPL-hang ‘The flag is hanging on its pole (lit., tree).’ (212) Adin nga tembok leisa duka. adin=nga tembok le-isa duka APPL -lean stand ladder=DEF wall ‘The ladder is leaning on the wall.’ Another applicativizing function of this prefix is to add an argument towards which the action of the verb is directed. In (213), with the addition of le-, nadua ‘what’ becomes the object of ki ‘cry’, an otherwise intransitive verb. (213) Gán nadua leki akuya. gán na=dugwa le-ki a-kuya NFIN .3.NOM what=PL APPL- NFIN .cry 3.I - FIN .stay ‘What is she crying about?’

Kula

253

In other cases, this prefix is not applicativizing, but functions to increase the discourse transitivity of the clause (in the sense of Hopper & Thompson 1980). For instance, the basically transitive verb magina ‘hear’ is used in (214) without le-. The same verb is used in (215) with le-, but with no increase in number of arguments. Since the clause in (215) does not display the typical features of an applicative construction, the presence of le- serves only to increase the semantic sense of transitivity. This can be captured roughly by the English translations magina ‘hear’ and le-magina ‘listen to’. (214) Aningkang gigís adi nanu, amák kda magina. aning kang gi-gís adi nanu amák kída NFIN .person NFIN .good 3.POSS . II - NFIN .body see FIN . NEG voice just magina FIN .hear

‘People didn’t see their bodies, (they) only heard (a) voice.’ (215) Kálán nga ngaprenta gi nga yálemagina. kalan=nga ngá-párenta gi=nga yá-le-magina NFIN .now=DEF 1EXCL . POSS . II -order NFIN . PROG =DEF 2.II -APPL - FIN .hear ‘Now, you listen to what I’m about to command!’ Finally, le- is the default applicative prefix used with non-verbal roots, including borrowed nouns, to create new verbs as in (216). (216) Angu SMP leskoli. angu SMP10 le-skoli there middle.school APPL-FIN .school ‘(She) did her middle school there.’ 6.2.2 míThis prefix has two main functions. The first is as a locative applicative prefix, similar to the first function of le-. The main topological relation expressed by mí- is containment (217). A looser sense of containment is captured in (218), in which the boat floats on top of the water, but is contained within the larger ‘ocean’ (cf. le-lila in (211) in section 6.2.1). 10 SMP is an abbreviation borrowed from the Indonesian Sekolah (school) Menengah (middle) Pertama (first) which denotes junior highschool.

254

Nicholas J. Williams

(217) Apel nga mangkok mímíti. apel=nga mangkok mí-míti APPL- FIN .sit apple=DEF cup ‘The apple is in the cup.’ (218) Kapla nga tama ayang mílila. kapla=nga tama ayang mí-lila boat=DEF ocean NFIN .top APPL-hang ‘The boat is floating (lit. ‘hanging’) in the ocean.’ This same prefix is used to indicate progression further into a state (219). (219) Pka-pka mílika, sak-sak mísaku. puka~puka mí-lika saku~saku mí-saku RDP ~child APPL -many RDP ~ FIN .old APPL- FIN .old ‘The (number of) young people is increasing, and the adults are getting older.’ 6.2.3 weThis applicative prefix derives from the independent verb we ‘use’. Its primary applicative use is as an instrumental applicative, as in (220) and (221). (220) Ang kula wape. ang Kula we-ape DIST kula.language APPL -do ‘Say (lit. do) it using Kula’ or ‘Use Kula to do it.’ (221) Giado madíma inga kás ng webaba weláta. gi-ado madíma=nga kás=ng we-baba we-láta. 3.POSS . II - NFIN .child oldest=DEF thorn=DEF APPL -hit APPL-beat ‘(She) hit and beat her oldest child with a thorny branch.’ Additionally, the we- prefix is used with the verb yáni ‘give’. Since Kula does not have fully ditransitive verbs (see section 3.1), it encodes the theme argument as the single argument of a separate verb, mi ‘take’, while the A and R (recipient) arguments are marked as typical A and P arguments of a transitive verb. Additionally, the verb, yáni, always occurs with the prefix we-. An example is given in (222), in which the theme argument is omitted, easily understood from the

Kula

255

immediately preceding clause. In (223), an example with mi flagging the theme argument is given. (222) A: Yisiwa tau aposin kda! yi-siwa tau aposin kída 2.POSS . II -clothing too NFIN .beautiful just ‘Your clothes are just too nice!’ B: Ngatama awing wenjáni. ngá-tama awing we-n-yáni 1EXCL . POSS . II -grandparent 3.FOC APPL-1-FIN .give ‘My grandparent (is the one who) gave it to me.’ (223) Ngápte abu inga mi ngápa bayáku wegani. ngá-páte-abu=nga mi ngá-pa ba-yáku 1EXCL . POSS . II -corn-dust=DEF take 1EXCL . POSS . II -father CLF. HUM -FIN .two we-ga-yáni APPL -3-FIN .give ‘I gave my ground corn to the two men.’

7 Serial verb constructions This section describes serial verb constructions (SVCs) in Kula. These are common in Kula, like in other Alor-Pantar languages, and have a range of different functions described in the following subsections. The data for this section comes primarily from a single retelling of the Frog Story in Kula. Further research may reveal additional types and functions of SVCs in Kula. The SVCs described below are grouped by function, including adding arguments (section 7.1), aspectual (section 7.2), causative (section 7.3), resultative (section 7.4), motion/directional (section 7.5) and posture (section 7.6).

7.1 Argument adding serialization Kula lacks truly ditransitive verbs (see section 3.1). Instead, additional arguments can be added to a clause with the verb mi ‘take’ in a type of serial verb construction. For example, in (224), mi marks the theme arguments yamsalo ‘cassava’ and maka ‘banana’. This type of construction is required with the verb yáni ‘give’, which is not a true ditransitive in Kula (225).

256

Nicholas J. Williams

(224) Nte dalani, yamsalo mi maka mi tukwa wejina. n-te dalani yamsalo mi maka mi tukwa wejina 1EXCL . I -sleep FIN .morning cassava take banana take basket FIN .fill ‘We slept until morning, (then) filled the basket(s) with cassava and bananas.’ (225) Ngákib mi wenján dápa, nguda mi di! ngá-kib mi we-n-yán dápa ng-muda 1EXCL . POSS . II -payment take APPL -1EXCL . I -NFIN .give PRIOR 1EXCL . I -go.HIGH mi di take IPFV ‘Give me my payment first, before I go up to take (the underwear).’

7.2 Aspectual serialization Three verbs are used in serial verb constructions to express aspectual notions. These verbs are listed in Table 24. Table 24: Aspectual serial verbs Verb

Gloss

Meaning

Position

ku

‘stay’

imperfective

After

le

‘finish’

perfective

After

kása

‘complete’

completive

After

The verb ku ‘stay’ is used as a serial verb to encode imperfective aspect for ongoing actions. It follows the main lexical verb in a serial verb construction. Examples are given in (226) and (227). (226) Sirusa ape iku. sirusa ape i-ku work do 1INCL . I - NFIN .stay ‘We are doing work.’ (227) Lewaka ngku awa gaka dáng gimra mi. le-waka ng-ku awa gaka dáng gi-mura mi APPL -watch 1EXCL . I - NFIN .stay then week NFIN .one 3.POSS . II -inside in ‘I stood watch for a week.’ The verb le ‘finish’ is used in serial verb constructions for situations that have occurred but are not necessarily complete (compare with kása for the completive

Kula

257

aspect). Like ku, it follows the main verb and occurs as the last verb in the serial verb construction. Examples are (228) and (229). (228) Aning mi lula pte pá mu lekwi ba pat leya. aning mi lula páte pá=mu lekwi ba pat NFIN .person take go.DIST corn NFIN .garden=LOC tie.up CONJ NFIN .eat leya FIN .finish ‘Someone took (the cow) and tied it up at the corn field and it ate (corn).’ (229) Ngáparka dua to lok parekang leya. ngá-parka=dugwa=to lok parekang leya 1EXCL . POSS . II -jacket=PL =also NFIN .wet soaked FIN .finish ‘Our jackets were also soaking wet.’ Kása ‘complete’ is used in serial verb constructions to express completive aspect. In contrast with le, kása indicates the completion of an event. Compare, for instance, pat kása in (230) with pat leya in (228) above. In (228), the cow has not eaten all the corn in the field, while in (230) below, all of the meat has been eaten. (230) Gipe ng pat kása. gi-pe=nga pat kása 3.POSS . II -meat=DEF NFIN .eat complete ‘All of the meat has been eaten.’ This verb, kása, also occurs with certain semelfactive verbs like ‘die’ which are inherently fully completed once they occur (231). (231) Giya gipa ayámu kása. gi-ya gi-pa ayámu kása 3.POSS . II -mother 3.POSS . II -father FIN .die complete ‘His mother and father have died.’

7.3 Causative serialization Causative SVCs in Kula involve the verb ape ‘do, make’ as the first verb in the sequence. The verb(s) following ape express the caused event. The causer is encoded as the A argument of ape, while the causee is encoded as the S of the intransitive verb, -yámu ‘die’ (232) and akáni ‘extinguish’ (233).

258

Nicholas J. Williams

(232) Pe gátataku dáma aniyáku giape ayámu. pe gá-tataku dáma aning-yáku gi-ape ayámu pig 3.II -find after NFIN .person-FIN .two 3.V-make FIN .die ‘After they found the pig, the two of them killed it.’ (233) Jala sak dugwa, jiada nga anawe ape akáni. jala sak=dugwa ji-ada=nga anawe ape akáni make FIN .extinguish woman NFIN .old=PL 2.POSS . II -fire=DEF all ‘Women, all of you put out your fires!’

7.4 Resultative serialization In resultative serialization, an intransitive predicate follows the main verb(s) and indicates the state resulting from the event expressed by the main verb(s). These intransitive verbs are most frequently posture verbs, for instance, ate ‘lie’ in (234), which indicates the end result of the falling (tan si) event. (234) Gákte manga lámána tansi ate. gákte manga lámána tan si a-te surprise CONJ immediately NFIN .fall come.LOW 3.I -lie ‘He was surprised and immediately fell down flat.’ Not all resultative serial verb constructions involve posture verbs. In (235), the verb ayámu ‘die’ expresses the state of the younger sibling that results from the action expressed by the main verb, baba ‘hit, beat’. (235) Ginana aniyaku asuráng akína gána pka nga gikaku baba ayámu. gi-nana aning-yáku asuráng akína gána 3.POSS . II -older.sibling NFIN .person-FIN .two NFIN .emotion FIN .anger TOP puka=nga gi-kaku baba ayámu little=DEF 3.POSS . II -younger.sibling hit FIN .die ‘His two older siblings were angry and beat their younger sibling to death.’ Similarly, in (236), the verb agátu ‘disappear’, expresses the resulting end state of the S argument (lungkukita) of the intransitive mu. (236) Lungkukita nga lámána si me, mu agátu. lungkukita=nga lámána si me mu agátu frog=DEF immediately come.LOW come.LEVEL NFIN .run disappear ‘The frog then immediately went out (of the jar) and ran away.’

Kula

259

7.5 Motion verb serialization There are two types of motion verb serialization in Kula. In the first type, motion-action serialization, a motion verb precedes the main verb expressing a separate non-motion event. The two verbs expressing separate events occur in succession, the motion event followed by the non-motion event (237). When the second verb is transitive, the optionally expressed P argument would occur between the motion verb and the non-motion verb, as in the second example given here (238). (237) Níwe pá suba nga iwe imíti. ní-we pá suba=nga i-we i-míti LOC -go.LEVEL NFIN .garden house=DEF 1INCL . I -go.LEVEL 1INCL . I - FIN .sit ‘We go sit over there (at) the garden house.’ (238) Pátára nlula lewaka giya. giya pátára n-lula [ØNP] le-waka APPL -watch PROG moment 1EXCL . I -go.DIST ‘In a moment, we’ll go stand guard (over sth).’ The second type of serialization with a motion verb involves the main verb followed by a motion verb, indicating the directional path of the main event. The first verb in these constructions is most frequently a non-deictic motion verb such as in (239), but may also be a deictic motion verb (240) or a non-motion verb (241). (239) Kátuala inga lámána koda mda. kátual=nga lámána koda muda NFIN .dog=DEF immediately leap go.HIGH ‘The dog then immediately leaped up.’ (240) Gikátuala to we mda asa leduka. gi-kátuala=to we muda asa le-duka 3.POSS . II -FIN .dog=also go.LEVEL go.HIGH hill APPL-stand ‘His dog also went up and stood on a hill.’ (241) Ginura nga adi si me. gi-nura=nga adi si me 3.POSS . II -owner=DEF look come.LOW come.LEVEL ‘His (the frog’s) owner looked down this way (at the empty jar).’

260

Nicholas J. Williams

7.6 Posture verb serialization Similar to the use of motion verbs in direction serialization, posture verbs may occur following the main verb to express the position of the A or S argument of the main verb as the event takes place. The posture verbs used in this serialization include míti sit’ (242) and (244), duka ‘stand’ (243), lila ‘hang’ (244). (242) Ginura nga adi míti. gi-nura=nga adi míti 3.POSS . II -owner=DEF look FIN .sit ‘(The frog’s) owner sat looking (at him).’ (243) Lungkukita lemnik duka. lunkukita lemnik duka NFIN .sniff stand frog ‘Standing, the frog sniffed (him).’ (244) Atakagus to ayang lila ayámu, Tukagus to lámána mít ayámu. atakagus=to ayang lila ayámu tukagus=to Atakagus=also NFIN .top hang FIN .die Tukagus=also lámána mít ayámu immediately NFIN .sit FIN .die ‘Atakagus died hanging from above, and then Tukagus died sitting.’

8 Aspect marking Kula clauses can be marked for a range of aspectual and epistemic categories. The majority of morphemes encoding these categories occur postverbally. In this section, I present a preliminary analysis of the most frequently occurring aspect markers (di, tenu, gaya, and giya). In Table 25 I provide the full inventory of postverbal elements in Kula. Several occur infrequently and remain less well understood. Table 25: Aspect-Evidential markers di tenu gaya giya guna muna tugwa wáti mawo

Imperfective (IPFV ) Perfective (PFV ) Prospective (PROSP ) Progressive (PROG ) Certain (EVID ) Uncertain (EVID ) ‘already’ hearsay (EVID) (unclear)

Kula

261

Kula aspect marking resembles that of neighboring Wersing (see Schapper & Hendery 2014), with a three-way contrast between imperfective, perfective and prospective aspect (section 8.1 to 8.3). Imperfective and prospective markers are cognate with their Wersing counterparts, while Kula utilizes a distinct perfective aspect marker. Additionally, Kula uses a progressive aspect marker, giya (section 8.4).

8.1 di ‘IPF V ’ The imperfective aspect marker, di ‘IPFV ’, always occurs before the predicate. The position of di relative to arguments is flexible. The marker may occur immediately before the verb (245), or before a NP preceding the verb (246). There is no evidence that there is any difference in meaning associated with these different positions. (245) Di limpa. di limpa IPFV long ‘(It’s) still (too) long.’ (246) Di ja mí gisukwa, Momo. di ja mí gisukwa Momo IPFV NFIN .water NFIN .be.located probably Momo ‘Momo, there’s probably still water (in those areca nuts).’ When di occurs in a clause marked with the negator nanu ‘NEG ’ postverbally, the interpretation is ‘not yet’, as in (247) and (248). The form in (248) is an alternate pronunciation, de, with the same function as di. (247) Di ga nanu. di ga nanu IPFV NFIN .speak FIN . NEG ‘(She) hasn’t spoken yet.’ (248) De siwe nan gisukwa. di si-we nanu gisukwa IPFV come.LOW-go.LEVEL FIN . NEG probably ‘(They) probably haven’t come out yet.’

262

Nicholas J. Williams

The sense conveyed by di is of a previous event continuing into the present, with something like the effect of English ‘still’. Another type of imperfectivity is also encoded in Kula by the progressive aspect marking serial verb -ku ‘stay, live’. This contrasts with di, instead expressing a current ongoing state that is true only for the present time, not a continuation of a previous state (249). See section 7 for more on serial verb constructions. (249) Angu oto lewaka ngwít ngku. angu oto le-waka ng-mít ng-ku there truck APPL-watch 1EXCL . I - NFIN .sit 1EXCL . I - NFIN .stay ‘We sat there waiting for the truck.’

8.2 tenu ‘PF V ’ The postverbal particle tenu marks perfective aspect in Kula, translated roughly as ‘already’. It refers to events that have already occurred with reference to some point in time, whether in the past or present time. It occurs with a range of predicate types, but primarily states and motion verbs. Examples with a variety of predicate types are given below, starting with states in (250) and (251), a motion verb in (252), and activities such as singing in (253). (250) Nsu Baumi tani, malen tenu. ng-su baumi tani malen tenu 1EXCL . I - NFIN .come Baumi FIN .arrive NFIN .evening FIN . PFV ‘(Once) we arrived in Baumi, it was already evening.’ (251) Ngátela nduka tenu. ngá-tela ng-duka tenu 1EXCL . II -speak 1EXCL . I -stand FIN . PFV ‘I’m already here speaking (to you).’ (252) Tama luk ji tenu. tama luk ji tenu ocean NFIN .hole go.LOW FIN . PFV ‘He’d already gone down (i.e. fell) into the ocean.’ (253) Tais ngadaya tenu! tais ngá-daya tenu NFIN .middle 1EXCL . II -sing FIN . PFV ‘I’m in the middle of singing already!’

Kula

263

In each of these examples, tenu marks the predicate as having already occurred, relative to some other reference point in time. For example, in (250), by the time the speaker arrived in Baumi, it was ‘already’ evening. The change of state into ‘evening’ happened prior to the event of arriving in Baumi, thus the use of tenu ‘already’. In (251), tenu marks that the event expressed by ngátela ngduka ‘I am speaking’ was already in progress before the time of speaking. The perfective tenu marks the event as having started already before the time of speaking. In example (253), the speaker, who is singing a song, is interrupted by another person singing in the background. To get the other person to stop singing she says ‘I’m already singing here!’ in (253) – using tenu because her singing had already started before the interruption and time of her utterance in (253).11 The perfective marker tenu is frequently used in combination with the verbs lea ‘finish, done’ and kása ‘finish off, complete’. Examples are given in (254) to (256). (254) Yimasíng yisera wísa nsi kása tenu. yi-masíng yi-sera wísa n-si kása tenu 2. POSS . II -NFIN .food 2. POSS . II -food carry 1EXCL-come.LOW complete FIN . PFV ‘I have already finished carrying down all your food.’ (255) Le ten tabak dán yapati. le ten tabak dán ya-pati finish NFIN . PFV tobacco NFIN .one 2-FIN .bite ‘That’s done (eating areca nut), so chew a piece of tobacco (too).’ (256) Kása ten e? kása ten e complete NFIN . PFV DISC ‘All done, huh?’

8.3 gaya ‘PROSP ’ The prospective aspect in Kula is marked with the postverbal particle gaya ‘PROSP ’. This is a frequent aspectual particle, contrasting with the perfective marker tenu, and often translated with an immediate future (mau) in Indonesian. Examples are given in (257) and (258).

11 On one reviewer’s suggestion, another possibility is that tenu marks ‘perspectival aspect’ (Reesink 2009).

264

Nicholas J. Williams

(257) Nungal yalula gaya? nung-al ya-lula gaya where-DIR 2.II -go.DIST FIN . PROSP ‘Where are you going?’ (258) Mi lula gomán nga basa gaya. mi lula gomán=nga basa gaya take go.DIST NFIN .there.LEVEL =DEF FIN .read FIN . PROSP ‘(He) is going to take (our language) over there and read it.’

8.4 giya ‘PROG ’ The postverbal partical giya (possibly related to the verb gi ‘put, place’) expresses progressive aspect, contrasting with the prospective aspect expressed by gaya. The progressive giya typically refers to an event that is ongoing and will continue into the immediate future. This can be used to describe an event in the past, as well as the present, as in (259). Occasionally, the events marked by giya have not yet begun but are imminent and the use of giya indicates that they are effectively ongoing from the speaker’s perspective. The use in (260) is a good example of this. This example is taken from a conversation in which the speaker is responding to another speaker’s insistence that they go to the house to eat areca nut from their current location near the areca nut trees. In response to her insistence, the speaker says yo iwe giya ‘yeah, we’re going’, implying that they are already moving and the event expressed by iwe is already ongoing. (259) Mana dáng angu we gaya onga, wiksi giya. mana dáng angu we gaya=onga wiksi giya village NFIN .one there go.LEVEL FIN . PROSP = PROX flood FIN . PROG ‘(I) was going to go to a certain village, (and) there was a flood.’ (260) Yo, yo, iwe giya. yo yo i-we giya INTER inter 1INCL . I -go.LEVEL FIN . PROG ‘Yeah, yeah, we’re going.’

9 Discussion This sketch presents the first description of the Kula language. With the facts presented here, it is possible to position Kula within the family and in relation

Kula

265

to its closest relatives, Sawila and Wersing. In many respects, Kula is intermediate between these two close relatives – sharing some features of each. Kula phonology is complex, with a unique seven-vowel system and numerous phonological processes, including final versus non-final word-shape alternations for a subset of the lexicon and extensive nasal assimilation. While most AlorPantar languages display a length contrast in their vowel systems, Kula exhibits only the remnants of a length distinction. Kula central vowels appear to correspond with short vowels in other languages of Eastern Alor. However, the historical relationship between Kula central vowels and the short-long vowel distinction in other AP languages remains to be investigated. The data from Kula also show that vowel length is closely tied to stress and syllable weight. While more remains to be investigated, one possibility is that vowels are treated differently depending on whether the syllable is stressed or unstressed. The stress appears to correlate with syllable weight, as well, and potentially pitch (though this remains to be investigated). It might be possible to analyze the system as having seven vowels (including long and short /a/ and /i/) in stressed syllables, but just five vowels (all of which are short and typically reduced) in unstressed syllables. This points to the need for a closer look at length contrasts in other Alor-Pantar languages, as well as a historicalcomparative study on the development of stress and vowel length in the family. With regard to morphosyntactic features, Kula exhibits an average number of pronominal paradigms and verbal person marking prefix series, compared with other Alor-Pantar languages. Several of these pronominal and prefix series appear more peripheral, pointing to Kula’s intermediate position between languages of eastern and western Alor. Unlike some western Alor languages, Kula allows prefixes to mark S, P and A. However, the agreement marking paradigms are relatively complex and still not fully understood. A unique feature of Kula person marking is the so-called ‘inverse’ marker, here analyzed as an affectedness prefix on certain verb classes. Kula also shares many features with its close relatives, Sawila and Wersing, including applicative prefixes and multiple postverbal aspectual markers. Kula’s system of elevationals and demonstratives is relatively complex compared to the simpler system of Wersing. This brief sketch has provided initial analyses of the demonstrative and deictic elements, including elevationals, aspectual and other postverbal operators, as well as valency-increasing operations, while indicating the need for further study to better understand each of these rich areas of Kula grammar.

266

Nicholas J. Williams

Acknowledgments I would like to thank Penipius Mosali for his enthusiasm, dedication, and patience over the years as my primary Kula teacher. Special thanks go to Antoinette Schapper for her encouragement, patience, and intellect which greatly improved the sketch. Mark Donohue made important contributions to a preliminary draft version of several sections of the sketch and offered valuable commentary on my initial analyses. Thanks also are due to František Kratochvíl, Gary Holton, David Rood, Barbara Fox, Andy Cowell, and two anonymous reviewers, all of whom commented on various versions of the sketch. I am grateful for the financial support provided by a Fulbright student research grant (2012–2013), a Dissertation Improvement Grant from the US National Science Foundation (BCS-1159510), and an individual graduate student fellowship from the Endangered Languages Documentation Programme (IGS0181).

References Donohue, Mark. 1996. Inverse in Tanglapui. Language and Linguistics in Melanesia 27: 101–118. Haan, John. n.d. Kula phonology. Unpublished manuscript. Hopper, Paul J. & Sandra A. Thompson. 1980. Transitivity in grammar and discourse. Language 56: 251–299. Johnston, Neil. n.d. Kula translations of SIL materials. Unpublished manuscripts. Klamer, Marian. 2014. Numeral classifiers in the Papuan languages of Alor and Pantar: A comparative perspective. In Marian Klamer & František Kratochvíl (eds.), Number and quantity in East Nusantara, 103–122. Canberra: Pacific Linguistics. Kratochvíl, František. 2014. Sawila. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 351–438. Berlin: De Gruyter Mouton. Reesink, Ger. 2009. A connection between Bird’s Head and (Proto) Oceanic. In Bethwyn Evans (ed.), Discovering history through language: Papers in honor of Malcolm Ross, 181–192. Canberra: Pacific Linguistics. San Roque, Lila, Alan Rumsey, Lauren Gawne, Stef Spronk, Darja Hoenigman, Alice Carroll, Julia Colleen Miller & Nicholas Evans. 2012. Getting the story straight: Language fieldwork using a narrative problem-solving task. Language Documentation and Conservation 6: 135–174. Schapper, Antoinette. 2014. Kamang. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 285–350. Berlin: De Gruyter Mouton. Schapper, Antoinette & Rachel Hendery. 2014. Wersing. In Antoinette Schapper (ed.), Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 439–504. Berlin: De Gruyter Mouton. Stokhof, W.A.L. 1975. Preliminary notes on the Alor and Pantar languages (East Indonesia). Canberra: Pacific Linguistics. Willams, Nicholas. 2016. Place reference and location formulation in Kula conversation. Ph.D. thesis, University of Colorado.

Juliette Huber

5 Makalero and Makasae 1

The language scene

2 2.1 2.2 2.3 2.4 2.5 2.6 2.6.1 2.6.2 2.6.3 2.6.4 2.6.5 2.6.6 2.6.7

271 Phonology 271 Consonants 275 Vowels 277 Semivowels 279 Phonotactics 281 Stress 282 Morpho(phono)logy 282 Initial consonant mutations 283 Truncation of verbs in the INCORP LOC slot 285 Clitic allomorphy in Makalero Vowel harmony in the deictic lV-prefix in Makasae 287 Reduplication 288 Derivational morphology 290 Nominal compounding

3 3.1 3.1.1 3.1.2 3.1.3 3.2 3.3 3.3.1 3.3.2 3.4 3.5 3.6 3.6.1 3.6.2 3.7

291 Basic clausal syntax 292 Verbal predicates 293 Intransitive verbs 294 Transitive verbs 295 Ditransitive verbs 296 Non-verbal predicates 297 The INCORP slots 297 The INCORP P slot 299 The INCORP LOC slot 304 Negation 305 Adverbs 306 Questions 307 Polar questions 308 Information questions 309 Imperatives

4 4.1 4.2

Noun phrases Modifiers Determiners

DOI 10.1515/9781614519027-005

310 311 314

268

287

268

Juliette Huber

Possession 316 318 Plural marking

4.3 4.4 5 5.1 5.2 5.3 5.4 5.5 5.5.1 5.5.2

Pronouns and argument marking 320 320 Personal pronouns 321 The reflexive pronoun 322 The reciprocal pronoun 323 Interrogative pronouns 323 Argument marking 323 A arguments 324 P arguments

6 6.1 6.2 6.3

Agreement and verbal morphology 326 Suppletive subject number agreement on verbs 327 [+HUM ] agreement on numerals 329 Object marking on verbs in Makalero

7 7.1 7.2 7.2.1 7.2.2 7.2.3 7.2.4 7.2.5 7.2.6

331 Serial verb constructions 332 Directional serialization 333 Serialization with ‘take’ verbs 333 Benefactive serialization 336 Instrumental serialization 337 Location and manner adjuncts 339 Causatives 340 Caused positions Makalero mei and the Makasae oblique marker ma

8 8.1 8.2

343 Aspect marking Aspectual particles and adverbs 346 Aspectual enclitics

9

Discussion

References

326

341

343

347

350

1 The language scene Makalero and Makasae are spoken in the eastern part of East Timor. According to the 2010 population census of East Timor’s Direcção Nacional de Estatística, Makasae (ISO 639-3 code: mkz) is the country’s third largest language, with some 100,000 speakers. Makasae speakers are the dominant linguistic group in the Viqueque and the Baucau districts. In the south-eastern part of the Baucau

Makalero and Makasae

269

district, Makasae gradually gives way to Makalero, which is the mother tongue of a population of almost 8,000 people in the adjacent south-western part of the Lautém district. Makasae, with its large number of speakers, is vibrant. Despite its much smaller size, the same currently holds for Makalero, probably due to the Iliomar subdistrict’s isolation and poor infrastructure. However, the gradual improvement of the infrastructure is likely to have negative effects on the language’s vitality in the near future. Makalero has often been treated as a dialect of Makasae1, and it has only been assigned its own ISO 639-3 code [mjb] in the most recent edition of the Ethnologue (Lewis, Simons & Fenning 2015). This chapter aims to support and substantiate this change by presenting a side-by-side comparison of the two languages, drawing attention not only to the far-reaching similarities, but also to the important differences setting them apart from one another. Table 1 sets out how closely related Makalero and Makasae are; cognate vocabulary is abundant and easily identified, and the same holds for language structures. However, its bottom rows also show that the number of non-cognate items increases noticeably as we move from lexical to more grammatical meanings such as modal verbs, temporal adverbials and negation. Furthermore, as will be highlighted at various points in this chapter, a number of morphological and syntactic differences exist between Makalero and Makasae. Due to the high number of cognates, mutual intelligibility between Makalero and Makasae is possible to a certain degree, although it is highly dependent on a speaker’s familiarity with the other language. Makalero speakers tend to have less trouble understanding Makasae than vice versa, a fact that can be explained through (i) the lower degree of morphological complexity of Makasae and (ii) the fact that Makasae is more widespread, so that most Makalero speakers have a higher degree of exposure to Makasae than vice versa. While Makalero speakers recognize the close connection of their language to Makasae, they consider Makalero to be a separate language, and it plays an important role in defining their sense of identity.2 1 There is significant dialect variation in Makasae. In older Portuguese sources, some of these dialects are listed as separate languages. Thus, Almeida (1994: 36) lists the following varieties, which are today considered dialects of Makasae: Màcassai, Nai-Damo, Ná-Ine, Sá-Âni. To date there has been no research exploring the extent of the diversity amongst these lects. 2 The same is true for speakers of varieties still considered dialects of Makasae. For instance, the inhabitants of the Luro subdistrict speak a variety called Saʔani (spelled Sá-Âni in Almeida 1994), which they consider to be a separate language. Saʔani has not been studied, but it is my impression that it shows a very high degree of similarity to Makalero. However, East Timor’s National Institute for Linguistics treats Saʔani as a dialect of Makasae. This highlights the degree of arbitrariness involved in delimiting languages from dialects, particularly in cases where the varieties in question are poorly known.

270

Juliette Huber

Table 1: Lexical comparison between Makalero and Makasae

‘day’ ‘night’ ‘child’ ‘younger sibling’ ‘river’ ‘earth, ground’ ‘house’ ‘meat’ ‘bone’ ‘be able’ ‘now’ ‘RECENT PAST ’ ‘NEG ’

Makalero

Makasae

watu kamu mata noko weir muʔa omar, lopu seur hafa hul aireʔ aiteʔ nomo

watu gamu mata noko wair maʔa, muʔa oma seu sapa beʔu ehani opoloi to, noto, tonai, notonai, noko

The earliest notes on Makasae were published in the 1940s (Capell 1944a; Capell 1944b; Capell 1944c), and unpublished materials dating from the 1950s are held at the Instituto de Investigação Científica Tropical (IICT) in Lisbon. Further pertinent studies did not appear until the late 20th century, when Marques (1990) published a Portuguese-language learner’s course of Makasae, and Sudiartha et al. (1998) a short sketch grammar. This was followed by Hull’s (2004) overview of the Papuan languages of Timor, which contains a great deal of information on Makasae. Hull also co-authored a sketch grammar of the Baucau dialect of Makasae (Hull & Costa 2005) and a short Makasae-Tetun dictionary (Hull & Correia 2006). In 2003 and 2004, two theses on Makasae came out of the Australian National University: Brotherson (2003) focuses on spatial expressions in Makasae, while Carr (2004) examines three traditional Makasae genres. In 2008, Huber published a sketch grammar of the Ossu dialect of Makasae (Huber 2008a). The most recent Makasae materials to appear are Correia’s PhD thesis, a grammar of the Laga dialect submitted at the University of Western Sydney (Correia 2011), and Fogaça’s MA thesis on Makasae phonology at the University of Brasília (Fogaça 2011).3 All of these materials are sketchy to varying degrees, and many aspects of Makasae remain underdescribed.4

3 Fogaça’s work on Makasae is ongoing; he is currently working on a PhD thesis. 4 At the time of writing, mini-sketches of Makasae including word lists and short texts written by native speakers were available from the Language Documentation Training Center of the University of Hawai’i (http://ling.hawaii.edu/ldtc/). As of December 2016, these materials can no longer be accessed.

Makalero and Makasae

271

Makalero has been described in a reference grammar (Huber 2011), as well as several papers (Huber 2008b, Huber 2014, Huber forthcoming, Huber & Schapper 2014). Furthermore, the Timor Loro Saʹe Nippon Culture Center has overseen the production of a number of materials by native speakers, among them a monolingual Makalero dictionary (Pinto 2004) and a Makalero language course with Tetun and English translations (Pinto 2007). The Makalero data used in this chapter come from my own fieldwork, conducted in and around Iliomar town between 2007 and 2012. The Makasae data are mainly from my work with Makasae speakers in Coimbra (Portugal) in 2005 and from Correia (2011) and Brotherson (2003). Note that Correia (2011) focuses on the Laga dialect (Baucau district) of Makasae, while Brotherson’s as well as my own data come from the Ossu dialect (Viqueque district). Reference to these dialects will occasionally be made in the text where differences exist between them; otherwise the information given holds for all the Makasae varieties I have had access to.5 Since the description of Makalero is more complete than that of Makasae, Makalero is taken as the basis for this comparative grammar sketch, and similarities and differences to Makasae are pointed out where known. When referring to specific items in the text, Makalero forms are marked with the abbreviation ‘Mkl’ and Makasae forms as ‘Mks’. Where an item is identical in the two languages, it is marked as ‘Mkl/Mks’.

2 Phonology Makalero and Makasae have fairly simple phoneme inventories. The main difference between the consonant inventories is the fact that Makasae retains a distinction between voiced and voiceless stops, which has been lost in Makalero. In this respect, the Makasae consonant inventory is thus more conservative than Makalero’s. The two languages’ vowel inventories are identical.

2.1 Consonants Table 2 shows the 11 native Makalero consonant phonemes, and Table 3 gives the 13 native consonant phonemes of Makasae. Although it is the Ossu dialect 5 In addition, the materials of the Language Documentation Training Center of the University of Hawai’i have been consulted. Examples with no indications of source come from my own data. In all other cases, the sources are indicated in brackets. The spelling and glosses of Makasae examples from other sources are adapted to the conventions used in this sketch.

272

Juliette Huber

which is represented in the rest of this chapter, the phoneme inventory shown in Table 3 is that common to the majority of Makasae dialects except the Ossu dialect. The choice of this phoneme inventory over that of the Ossu dialect is due to the fact that in the comparison of Makalero and Makasae, phoneme correspondences and mergers are more clearly visible; they are somewhat obscured in the Ossu dialect (see below for details).6 The orthography used in this chapter is straightforward and directly reflects the phoneme values.7 Table 2: Makalero consonant phonemes

Plosive Fricative Nasal Lateral Trill

Labial

Alveolar

Post-alveolar

Velar

Glottal

p f

t s

d

k

ʔ

m

n l r

Table 3: Makasae consonant phonemes Labial Plosive Fricative Nasal Lateral Trill

b f m

Alveolar

Velar

t s

k

d

Glottal g

ʔ h

n l r

Table 4 gives a list of (near-)minimal pairs illustrating the contrastiveness of Makalero and Makasae consonant phonemes, using cognate vocabulary where possible. Note the absence of a pair contrasting the glottal phonemes given in Table 3 for Makasae. Their status as phonemes is somewhat dubitable and is commented on below.

6 Loan phonemes found in both Makalero and Makasae are the alveodental and palatal affricates /ts/ and /dz/, and /ʧ/ and /ʤ/, respectively; the velar nasal /ŋ/; and the voiced fricatives /v/ and /z/. In addition, /p/ is used in Makasae in Portuguese and Tetun loans, while some Makalero speakers use the voiced stops /b/ and /g/ in loanwords. Apart from the affricates, all loan phonemes exhibit significant inter-speaker variation in realization, with some speakers replacing them by the most similar native phoneme. 7 Most previous sources have used the apostrophe to represent the glottal stop. In this chapter, however, I use the IPA symbol .

Makalero and Makasae

273

Table 4: Minimal pairs Makalero /t/ ≠ /d/

Makasae /tanen/ ‘right (side)’ /dane/ ‘move, manipulate’

/t/ ≠ /d/

/tane/ ‘right (side)’ /dane/ ‘give, hand over’

/k/ ≠ /g/

/kuʔul/ ‘warm’

/gur/ ‘blow’

/p/ ≠ /m/ /pane/ ‘wash’

/mana/ ‘hole’

/b/ ≠ /m/ /bane/ ‘wash’

/mana/ ‘hole’

/p/ ≠ /f/

/patan/ ‘log’

/fatu/ ‘swell’

/b/ ≠ /f/

/bata/ ‘log’

/fatu/ ‘swell’

/t/ ≠ /n/

/fatu/ ‘swell’

/fanu/ ‘face’

/t/ ≠ /n/

/fatu/ ‘swell’

/fanu/ ‘face’

/d/ ≠ /n/

/dane/ ‘move, manipulate’

/nana/ ‘snake’

/d/ ≠ /n/

/dane/ ‘give, hand over’

/nana/ ‘eye’

/t/ ≠ /s/

/teur/ ‘dust, smoke’

/seur/ ‘meat’

/t/ ≠ /s/

/teu/ ‘dust’

/seu/ ‘meat’

/d/ ≠ /s/

/dur/ ‘wake up’

/sura/ ‘rat’

/d/ ≠ /s/

/dura/ ‘rat’

/sura/ ‘count’

/l/ ≠ /r/

/walir/ ‘edge’

/wari/ ‘still’

/l/ ≠ /r/

/wali/ ‘edge’

/wari/ ‘wait’

/k/ ≠ /ʔ/

/k/ ≠ /ʔ/

/saka/ ‘search’

/saʔal/ ‘fry’

/wakal/ ‘pour’

/saʔal/ ‘fry’

/ʔ/ ≠ Ø

/raʔu/ ‘plate’

/rau/ ‘be good’ /ʔ/ ≠ Ø

/raʔu/ ‘plate’

/rau/ ‘be good’

[h] ≠ Ø

[hɛna] ‘cloth’

/ena/ ‘see’

/h/ ≠ Ø

/hena/ ‘cloth’

/ena/ ‘see’

/nana/ ‘snake’

/m/ ≠ /n/ /mana/ ‘hole’

/m/ ≠ /n/ /mana/ ‘hole’

/nana/ ‘eye’

The Makalero consonant phonemes are for the largest part stable in their realization. One relatively rare exception is that a two-syllable sequence /nV1rV2/ is sometimes reduced to [ndV2]; for instance, Mkl /nanaraː/ [ˌnanaˈraː] ‘elder siblings’ is often realized as [ˈnaːnda], and the place name /senira/ (/sen/ ‘pineapple’, /ira/ ‘water’) [ˌsɛːnˈira] as [ˈsɛːnda]. Makasae exhibits some allophonic variation in the velar stops: before back vowels, /k/ tends to be uvularized. /g/ and /b/ are frequently fricativized in intervocalic position.8 The differences between the Makalero and the Makasae consonant systems are mostly limited to the plosive series, where Makalero lost a voicing distinction which was retained in Makasae. This voicing distinction was part of the common ancestor language (Schapper, Huber & van Engelenhoven 2012). The presence of /d/ in Makalero seems to contradict this statement; however, given that /d/ is realized in a postalveolar rather than the alveolar position, it cannot be treated as the voiced counterpart to /t/ synchronically.9 Instead, it is part of a plosive series with five members distinguished by place of articulation rather than voicing (Huber 8 According to Brotherson (2003: 23), intervocalic fricativization can be found with all voiced stops. In my own data, I did not observe fricativization of /d/. 9 Asymmetries in place in voiceless-voiced plosive pairs are not unknown in the family. In Blagar, for instance, /t/ is dental, whereas /d/ is alveolar. The other plosive pairs are produced at the same place of articulation (Steinhauer 2014: 152). Asymmetries are also found in Austronesian languages of the region, e.g., Dawan (see Steinhauer 1993) or Leti (see van Engelenhoven 2004).

274

Juliette Huber

2011: 49). In Makasae, on the other hand, /d/ is straightforwardly analyzed as the voiced counterpart of /t/, from which it does not differ with respect to place of articulation (Fogaça 2011: 14f.).10 Table 3 shows a gap in the Makasae plosive series: there is no unvoiced bilabial plosive /p/. This is explained through the fact that proto-Timor *p and *f merged into /f/ in both Makasae and Makalero. The only exception to this is the Ossu dialect of Makasae, where *p and *f merged to /p/.11 For instance, in the Ossu dialect, /sapa/ ‘bone’ corresponds to /safa/ in all other Makasae dialects, and /hafa/ in Makalero. The phoneme /p/ in Makalero, on the other hand, is a reflex of proto-Timor *b. Another difference between the Makalero and the Makasae systems is the absence of the glottal fricative /h/ in the Makalero phoneme system in Table 2. Makalero does have a glottal fricative phone, but it appears in a largely complementary distribution with [ʔ] and is therefore treated as an allophone of a single glottal phoneme /ʔ/ (Huber 2011: 51). The distribution of these allophones is shown in (1).12 Both allophones are frequently pronounced so faintly that they are practically inaudible. Makalero allophony of [ʔ] (1) /ʔ/ → [h] /#_ e.g., /ʔeti/ [ˈheti] ‘ask’, /ʔiʔa/ [ˈhiʔa] ‘street’ → [ʔ] (~ [h]) / V _ V e.g., /raʔu/ [ˈraʔu] ‘plate’, /laʔa/ [ˈlaʔa] ‘walk’, but also /naʔa/ [ˈnaʔa] ~ [ˈnaha] ‘work’ / V _ # e.g., /isiʔ/ [iˈsiʔi] ~ [iˈsihi] ‘be at’, /fasaʔ/ [faˈsaʔa] ‘dirt’, but also /douʔ/ [douh] ‘six’ 10 Note, however, that according to Correia (2011: 37), Mks /t/ is alveolar, while /d/ is postalveolar. Though none of the other sources mention this asymmetry, it is possible that it exists in at least some dialects of Makasae. 11 The Makasae examples in the following sections mostly come from the Ossu dialect and thus use /p/ rather than /f/. 12 The occurrence of either [ʔ] or [h] word-medially appears to be lexically determined. In both word-medial and word-final position, the [ʔ] allophone is more common. -ʔ is also a fairly frequent but unproductive verbalizer (see section 2.6.6). In this case, the two realizations appear to be in free variation. The analysis shown in (1) captures the statistically dominant distributional tendencies of [ʔ] and [h], but cannot predict the occurrence of either [ʔ] or [h] word-medially in all circumstances and is thus somewhat unsatisfactory. In order to avoid this problem, one would have to posit two distinct glottal phonemes /ʔ/ and /h/. The latter would have to have a [ʔ] and a [h] allophone, to account for those cases where the two sounds appear to be in free variation word-finally. This analysis would be able to account for the difference between such lexemes as Mkl [ˈnaʔu] ‘just’, which is only ever realized with a glottal stop, and [ˈnaʔa] ~ [ˈnaha] ‘work’, which can be heard either with a glottal fricative or a stop. However, without clear distributional criteria for the two /h/ allophones, such an account would not explain those lexemes where internal or final [h] does not alternate with [ʔ]. Finally, [ʔ] and [h] cannot be shown

Makalero and Makasae

275

All of the Makasae sources consulted (Sudiartha et al. 1998, Brotherson 2003, Carr 2004, Huber 2008a, Fogaça 2011, Correia 2011) list /ʔ/ and /h/ as separate phonemes. However, none of them provides minimal pairs. Fogaça (2011: 57) actually gives two realizations of ‘thirsty’, [iɾaˈhaː] and [iˈɾaʔ], on the same page; he appears to treat them as separate items, giving the phonemic forms /iraʔ/ and /iraha/. According to Correia (2011: 38), the glottal stop does not occur initially; and in some dialects intervocalic /h/ is normally replaced by /ʔ/ (Correia 2011: 44). This strongly suggests that Mks [ʔ] and [h] may better be analyzed as allophones rather than separate phonemes.13

2.2 Vowels Both Makalero and Makasae exhibit five-vowel systems as depicted in Table 5. Long vowels have marginal phonemic status and are hence indicated in brackets. Table 5: Vowel phonemes Front High Mid Low

Central

i (iː) e (eː)

Back u (uː) o (oː)

a (aː)

Fogaça (2011: 46) distinguishes /e/ and /ɛ/, and /o/ and /ɔ/ in his analysis of the Makasae phoneme system. He provides a number of near minimal pairs to support this analysis, but no full minimal pairs. In Makalero, [ɛ] and [ɔ] appear as allophones of the phonemes /e/ and /o/, respectively, with the mid-high variants generally occurring in the vicinity of high vowels (Huber 2011: 55), and I suggest the same is the case for Makasae. Table 6 demonstrates the distinctiveness of the vowel phonemes. Table 6 shows that there is generally an exact vowel correspondence in cognate vocabulary. Exceptions are vowel sequences or vowels in the vicinity of /ʔ/, where irregular correspondences are frequent, as shown in Table 7. Though more research is needed, the comparative evidence suggests that in most of these cases, the Makalero forms are closer to the reconstructed proto-Timor forms (Schapper, Huber & van Engelenhoven 2012). to contrast in any context, thus failing the most important criterion for phonemic status. The Makalero orthography used in this chapter generally represents the phonetic values of the glottal phoneme’s allophones, and . Where the they appear to be in free variation, the glottal stop symbol is used. 13 But see the Makasae question words listed in section 3.6.2; Mks /naʔi/ ‘what’ and Mks /nahi/ ‘where, when’ appear to be a minimal pair. However, Mks /naʔi/ ~ /nahi/ is an interrogative root used in all of the Makasae interrogatives, and whether there is an actual semantic opposition between the two items translated above as ‘what’ and ‘where, when’ needs to be tested.

276

Juliette Huber

Table 6: Minimal pairs Makalero /i/≠/e/ /i/≠/u/ /u/≠/o/ /o/≠/e/ /o/≠/a/ /e/≠/a/

/ina/ ‘eye’ /ira/ ‘water’ /uful/ ‘fly’ (n.) /oroʔ/ ‘spear’ /oso/ ‘casuarina’ /ate/ ‘tree, wood’

Makasae /ena/ ‘see’ /ura/ ‘worm’ /ofo/ ‘snake’ /ore/ ‘foam’ /asa/ ‘bird’ /ata/ ‘fire’

/ina/ ‘mother’ /ira/ ‘water’ /ufulai/ ‘fly’ (n.)† /oro/ ‘spear’ /oso/ ‘casuarina’ /ate/ ‘tree, wood’

/ena/ ‘see’ /ura/ ‘vein’ /ofo/ ‘snake’ /ore/ ‘foam’ /asa/ ‘bird’ /ata/ ‘fire’

† lai is a fossilized suffix found most frequently on personal names and body part nouns, e.g., tanelai ‘right (side)’, lawalai ‘ankle’, fahilai ‘shoulder’, but also on a number of stative verbs, e.g., kaʔulai ‘small’, lumulai ‘green’. Cognate forms in Makalero as well as other TAP languages of Timor show that this element is not part of the root. Table 7: Irregular correspondences with vowel sequences and glottal stop

‘sew’ ‘sleep’ ‘two’ ‘name’ ‘blood’ ‘land, ground’ ‘laugh’

Makalero

Makasae

/faʔa/ /tia/ /loloi/ /nei/ /wei/ /muʔa/ /ʔiʔe/

/feiʔ/ /taʔe/ /lolaʔe/ /nai/ /wai/ /maʔa/ /hiʔa/

The long vowels are bracketed in Table 5 to signal their marginal phonemic status. In Makalero, vowel length is for the most part predictable on the basis of one of the following rules: (i) Lexical morphemes must be at least bimoraic, i.e., the nucleus of a monosyllabic lexical morpheme must be either a long vowel or a diphthong. There is no such requirement for grammatical morphemes. (ii) VʔV sequences where the vowels surrounding the glottal phoneme are identical are mostly realized as long vowels. In careful and slow speech, the glottal phoneme tends to be audible. (iii) At the end of a phonological phrase, an echo vowel is normally added to morphemes ending in closed syllables. In this process, the vowel nucleus of the erstwhile closed final syllable is commonly lengthened (see Huber 2011: 83f. for details).

Makalero and Makasae

277

In the list of apparent minimal pairs based on length in (2), rule (i) accounts for the first pair, rule (ii) for the second and rule (iii) for the third.14 (2)

Makalero /i/ ≠ /iː/

/ni/ [ni] ‘REFL ’ ≠ /ni/ [niː] ‘mother’

/a/ ≠ /aː/

/nana/ [ˈnana] ‘elder sibling’ ≠ /naʔan/ [ˈnaːna] ‘NEG . EXIST ’

/u/ ≠ /uː/

/ʔulu/ [ˈhulu] ‘spoon’ ≠ /ʔul/ [ˈhuːlu] ‘be able’

There is, however, a small number of Makalero lexemes with unpredictable long vowels; examples are shown in (3). The presence of these long vowels can be explained by assuming that they are originally compounds, made up of at least one monosyllabic element. This is supported by the fact that some of them contain consonant clusters, which are not generally found word-internally (section 2.4). Since they do not appear to be analyzable, the long vowels contained in them are treated in Makalero as phonemes of marginal status. (3)

Makalero /puːlata/ /paːrdufu/ /leteroː/

‘head’ ‘widow’ ‘coarse bag’

There is no detailed account of vowel length in Makasae. It appears that monosyllabic lexical morphemes also tend to have a long vowel nucleus, which suggests there is a similar bimoraicity requirement as in Makalero (cf. Correia 2011: 2). According to Fogaça (2011: 44f.), long vowels occur predictably in stressed word-final syllables.15 Furthermore, VʔV or VhV sequences with two identical vowels tend to be realized as long vowels (Fogaça 2011: 49). Though the issue requires more detailed analysis, the evidence suggests that vowel length in Makasae follows similar principles as in Makalero. Vowel length is not represented in writing if it can be predicted based on the rules above. Unpredictable long vowels are represented by the doubling of the vowel graph. Lastly, echo vowels are not written.

2.3 Semivowels Makalero has two semivowels, [w] and [j], which can function both as onglides in syllable onsets, and as offglides in complex nuclei. [w] is frequent in both 14 Mkl hul ‘be able’ is a monosyllabic lexical item and therefore has a long vowel nucleus per se. The length of this nucleus is not affected by the addition of the echo vowel. 15 I would suggest that it is the vowel length which attracts stress; in other words, that the causality is the reverse of that described by Fogaça (2011). See also section 2.5.

278

Juliette Huber

positions, as shown in (4). On the other hand, [j] is truly frequent only as an offglide, and is found as an onglide only rarely in native Makalero words (5). In Makalero, these glides can be analyzed as allophones of the corresponding high vowels occurring either as a syllable onset, or as the second element in a vowel sequence (Huber 2011: 60f.).16

(4)

Makalero /uata/ [ˈwata] /rau/ [raw]

(5)

Makalero /iar/ [jar] /dai/ [daj]

‘coconut’ ‘be good’

‘cry’ ‘stranger, king’

For Makasae, Fogaça (2011) and Correia (2011) list a semivowel phoneme /w/, which is found only as an onglide in syllable onsets (6). However, where the high vowels /u/ and /i/ are the second element in a vowel sequence, they may be realized as offglides [w] and [j] in quick pronunciation (Correia 2011: 33).

(6)

Makasae /wata/ [ˈwata] /lawa/ [ˈlawa]

‘coconut’ ‘money’

Table 8 demonstrates the phonemic contrasts of /w/ ([w]) with phonetically similar consonant phonemes. Table 8: Minimal pairs for /w/ ([w]) Makalero

Makasae

/p/ ≠ [w]

/patan/ ‘log’

[wata] ‘coconut’

/b/ ≠ /w/

/bata/ ‘log’

/wata/ ‘coconut’

/f/ ≠ [w]

/fatu/ ‘swell’

[wata] ‘coconut’

/f/ ≠ /w/

/fatu/ ‘swell’

/wata/ ‘coconut’

16 In keeping with their analysis as allophones of the high vowels, the glides were spelled as and in Huber (2011). In this chapter, as in other publications, however, the labiovelar approximant is spelled as rather than as if used as an onglide, although I continue to use for the offglide. The reasons for this change are that Makalero speakers objected to the use of for the onglide as being unintuitive and difficult to read; and that it obscures the similiarities to Makasae and other languages. In the case of the palatal approximant [j], on the other hand, the spelling has not proven problematic and is therefore maintained in all positions.

Makalero and Makasae

279

2.4 Phonotactics The most common syllable structure in both Makalero and Makasae is (C)V; (C)VC may also be found. In native words, closed syllables are predominately limited to the morpheme-final position, though they are disfavored in both languages even in this position. At the end of a phonological phrase, they are converted to two open syllables through the addition of an echo vowel; thus Mks /sisir/ ‘sick’ becomes [siˈsiːri].17 Similarly, within a phonological phrase, echo vowels are inserted in order to avoid consonant clusters. As a consequence, a final consonant is resyllabified as a syllable onset. Thus, even though producing the closed-syllable variants is considered more correct, especially with respect to Portuguese loans, they are in practice found rarely. Word-internally, consonant clusters are very uncommon in both languages. Medial consonants are thus generally syllable onsets. While Makasae is quite consistent in avoiding consonant clusters, they can be found in a limited number of native Makalero words. There is good evidence for an analysis of such lexemes as old compounds, with the cluster signaling an erstwhile morpheme break (Huber 2011: 78). In both languages, an epenthetic vowel is commonly inserted in consonant clusters in loanwords (e.g. Mkl /kaːlsa/ [ˌkaːˈlasa] ‘trousers’ from Portuguese calças). Table 9 shows the distribution of consonantal elements in both languages. A blank cell in the table signals that the phoneme in question is not present in the language. Because they are usually converted into onsets through the use of echo vowels, codas are shown in brackets. Table 9 shows that the distribution of the phonemes which are common to both languages is largely identical. All consonant phonemes can function as syllable onsets, both in initial and in medial position. Only some consonant phonemes can be found in the coda position, however. In Makalero, seven out of 11 consonants occur in this position. Among the stops, these are /t/, /k/ and /ʔ/; of the phonemes involving a diversion of the air stream, /n/ and /l/; and of the continuants /s/ and /r/. Out of the 13 consonantal phonemes of Makasae, eight appear in the coda position, namely the voiceless stops /t/, /k/ and /ʔ/, the fricative /h/, the nasal /n/, the liquids /l/ and /r/, and the approximant /w/. The only significant difference is /s/, which can be found word-finally in Makalero, but not, apparently, in Makasae. Both languages appear to have a restriction

17 See Huber (2011: 83f.) for an account of the compensatory lengthening that occurs in conjunction with echo vowels in Makalero. Fogaça (2011: 57) seems to suggest that the same process is at work in Makasae.

280

Juliette Huber

against labial sounds in the coda position, and additionally, voiced plosives are barred from that position in Makasae. Table 9: Phoneme distribution /p/†

Position #_V V_V V_#

Mkl Mks

Z

Mkl Mks

Z

Mkl Mks

−

/b/

/t/

/d/

/k/

/g/

/ʔ/

/f/

/s/

/h/‡

/m/

/n/

/r/

/l/

Z

Z Z

Z Z

Z Z

Z

Z −

Z Z

Z Z

Z

Z Z

Z Z

Z Z

Z Z

Z

Z Z

Z Z

Z Z

Z

Z Z

Z Z

Z Z

Z

Z Z

Z Z

Z Z

Z Z

−

(Z) (Z)

− −

(Z) (Z)

−

(Z) (Z)

− −

(Z) −

(Z)

− −

(Z) (Z)

(Z) (Z)

(Z) (Z)

† Most Makasae dialects do not have a phoneme /p/ (see the discussion in section 2.1). ‡ In section 2.1 I argue that [h] is an allophone of /ʔ/ in Makalero. While a similar analysis is likely for Makasae, this has not been rigorously demonstrated, and consequently /ʔ/ and /h/ are treated as distinct phonemes in Makasae for the purpose of this table.

The examples in (7) illustrate the vowel sequences found in Makalero and Makasae. In all cases, the high vowels which are the second element of the sequence are realized as offglides, [w] and [j], respectively. (7)

a.

b.

Makalero /au/ /rau/ /eu/ /leu/ /ou/ /toutou/ /ai/ /dai/ /ei/ /rei/ /ui/ /Puidau/ /oi/ /loloi/

‘be good’ ‘call’ ‘owl’ ‘king, stranger’ ‘outward’ ‘Puidau’ (a personal name) ‘two’

Makasae /au/ /rau/ /eu/ /seu/ /ou/ /toutou/ /ai/ /dai/ /ei/ /rei/ /ui/ /kui/ /oi/ /opoloi/

‘be good’ ‘meat’ ‘owl’ ‘king, stranger’ ‘pull’ ‘flute’ ‘REC . PST ’

Combinations of the two high vowels are restricted; /iu/ is not found at all, and Mkl /ui/ has been found in a few personal names only. While some instances

Makalero and Makasae

281

of Mks /ui/ can be found in lexical items, this, too, is relatively rare. Makasae additionally permits the mid-vowel /e/ as the second element in a vowel sequence. The combinations attested in the corpus are shown in (8). According to Correia (2011: 32), these vowel sequences are actually disyllabic in Makasae, although they are realized as diphthongs in rapid speech.

(8)

Makasae /ae/ /pae/ /oe/ /loe/

‘four (HUM )’ ‘ash’

2.5 Stress Information on stress in both languages, and in particular Makasae, is rudimentary, and a substantial amount of further research is needed. Stress appears to be non-phonemic in both languages.18 In Makalero, stress assignment is based on the moraic trochee: the main stress falls on the last foot, which is formed either by a heavy syllable on its own or by two light syllables, as exemplified in (9). Beyond the word-level, the domain of stress assignment is the phonological phrase, and a given word may be stressed on different syllables, according to the context of the utterance in which it occurs. Stress in the utterance follows a roughly trochaic pattern.

(9)

Makalero /ˈasa/ /ˌkamuˈnei/ /ˌpaːrˈdufu/

‘bird, chicken’ ‘tomorrow’ ‘widow’

The rules for stress assignment in Makasae are poorly known. In morphemes of the form CVCV, stress is placed on the penultimate syllable, with a secondary stress on the anteprepenultimate for longer lexemes. Evidence regarding stress assignment in words containing other syllable types comes largely from loan words: Mks /ˌaraˈbau/ ‘buffalo’ (an Austronesian loan) shows that syllables with diphthongs attract stress, and Fogaça (2011: 57) provides evidence demonstrating that the same holds for syllables with long vowels. Likewise, closed syllables, such as in the Portuguese loan /ˈsaːlva/ ‘save’, seem to attract stress. This is consistent with stress assignment on the moraic trochee as described for Makalero.

18 Although ‘tone’ (i.e., pitch accent) is lexical in Makasae, according to Correia (2011: 24).

282

Juliette Huber

2.6 Morpho(phono)logy Makalero and Makasae both tend towards isolating, although Makalero is somewhat more conservative and retains more morphological processes than Makasae. The majority of processes observed center on the applicative-like expression of locatives in the INCORP LO C slot (section 2.6.1 and section 2.6.2). Further morpho(phono)logical processes discussed include clitic metathesis in Makalero (section 2.6.3), a case of vowel harmony in the deictic system of Makasae (section 2.6.4), reduplication (section 2.6.5), derivational morphology (section 2.6.6) and nominal compounding (section 2.6.7).

2.6.1 Initial consonant mutations Makalero and Makasae have a variety of verbs which exhibit initial consonant mutation. The forms with the altered consonant, which will be referred to as bound forms, are used when the verb is combined with mostly locative information in the INCORP L OC slot (section 3.3.2). Some representative examples are shown in Table 10. Similar alternations are found with a subgroup of verbs in Fataluku (van Engelenhoven 2009). This appears to be an innovation in the Eastern Timor languages. Table 10: Examples of altered onset verbs Makalero

‘release, shoot’ ‘put, bury’ ‘see’

Makasae

Free

Bound

Free

Bound

/ʔuri/ /taru/ /ena/

/-suri/ /-daru/ /-kena/

/suri/ /taru/ /ena/

/-duri/ /-daru/ /-gena/

Table 10 illustrates the three groups of altered onset verbs: (i) those which alternate between Mkl /ʔ/ ([h]) and /s/ and Mks /s/ and /d/, respectively; (ii) those which alternate between Mkl/Mks /t/ and /d/; and (iii) a group of miscellaneous alternations, most of which are vowel-initial in the free form. Makalero has more than 40 altered onset verbs (Huber 2011: 135–136). There is no information on the number of verbs exhibiting this alternation in Makasae. It is not attested in all cases where the Makalero cognate alternates, and Correia (2011: 216) shows that in several cases the free and the bound forms alternate in the same syntactic context, suggesting that the distinction is in the process of being lost in Makasae. The use of the free (i.e., unalternated) and bound forms is illustrated in (10) and (11).

Makalero and Makasae

(10)

Makalero a. Free form /ani afi ʔifaʔ/ fish catch/ 1SG ‘I caught (a) fish.’ b.

(11)

283

Bound form /kiloː tali were seti-sifa/ rope DEM pass-catch.BOUND 3SG ‘He missed the rope (trying to catch it).’

Makasae a. Free form /ani seu-meti sipaʔ/ 1SG meat-sea catch ‘I caught (a) fish.’ b.

Bound form /ai gi ma nahigalu 2SG 3SG OBL where ‘Where did you catch it?’

isi-dipaʔ/ at-catch.BOUND

2.6.2 Truncation of verbs in the INCORP LOC slot A second group of verbs lose their final consonant if used in the INCORP LO C slot (section 3.3.2). In this position, they usually specify the location or manner of a following verb. Such truncated forms are fairly frequent in Makalero, with 40 verbs showing the alternation listed in Huber (2011: 132–134). In the majority of cases, the final consonant which is lost in the reduced form is a glottal stop. Some examples are given in (12a). A few Makalero verbs which lose final consonants other than the glottal stop are illustrated in (12b). Examples (13) and (14) show their use. (12)

Makalero a. Free form /mutuʔ/ /namaʔ/ /tauʔ/ b.

Free form /dikar/ /hul/ /turut/

Truncated form /mutu-/ ‘be inside’ /nama-/ ‘be higher up’ /tau-/ ‘be where’ Truncated form /dika-/ ‘be short’ /hu-/ ‘be able’ /turu-/ ‘continue’

284

(13)

(14)

Juliette Huber

Makalero a. /lopu-mutuʔ/ house-be.inside

‘be inside the house’

(free form)

b.

‘go inside the house’

(truncated form)

/lopu-mutu-laʔa/ house-inside-go

Makalero a. /were this.ADDR b.

turut/ continue

/turu-lolo/ continue.TRUNC -say

‘lengthen this’

(free form)

‘continue telling’

(truncated form)

In Makasae, truncated forms are much rarer. The cognates to the glottal-stop final Makalero forms in (12a) all lack this glottal stop in Makasae: Mks /mutu/ ‘be inside’, /gata/ ‘be in contact’, /nama/ ‘be higher up’.19 Mks /mutu/, without the glottal stop, appears both in contexts where it is the sole predicative element, as in (15a), as well as in cases where it modifies another verb, as in (15b). The cognates to the Makalero examples in (12b) are variably attested with or without a final consonant; for example, Mks /digar/ ‘be short’, but Mks /turu/ ‘follow, continue’. None of the Makasae examples shows productive variation like that found in Makalero. However, some other verbs have been found to alternate consonant-final with truncated forms in the expected contexts, as shown in (16); see also Correia (2011: 218f.). More data is needed to determine how frequent this variation is in Makasae, and whether it is also found in verbs cognate to the Makalero set exhibiting this alternation.

(15)

Makasae a. /seu naʔi mutu/ meat what be.inside ‘Where (inside what) is the meat?’ b.

/ni-oma

mutu-ria isa/ inside-run go.up ‘. . . (she) ran up into her house. . .’ REFL . POSS -house

19 Brotherson (2003) analyzes all of these as spatial postpositions. However, the fact that they can be found in predicative function suggests they may be stative verbs. In both languages, the glottal stop is associated with verbalization; while it is not productive in either Makalero or Makasae, examples are more frequent in Makalero (Huber 2011: 128) than in Makasae (Huber 2008a: 11).

Makalero and Makasae

(16)

285

Makasae a. /ua malen apa/ village be.near IMM ‘. . . about to (get) close to the village. . .’ b.

/ni-sirbisu

male-saʔi gini near-finish make ‘. . . about to finish one’s work.’ REFL . POSS -work

apa/ IMM

Truncated verbs (as well as other elements in the INCORP L O C slot, see section 3.3.2) form a morphosyntactic unit with a following verb. A subclass of verbs appears in special bound forms with initial consonant mutations if used as the final verb in such a unit (see section 2.6.1). In this chapter, this unit is signaled through a hyphen linking the verbs. Note that in a morphosyntactic unit consisting of a truncated verb and a main verb, the truncated verb receives a secondary stress, while the main stress is on the final verb. Thus the truncated verbs behave like elements of a compound, rather than like affixes in terms of stress. The majority of truncated verbs express locative notions and are glossed here as adpositions, i.e., where a free form Mkl isiʔ is glossed as ‘be.at’, the truncated form isi- is glossed as ‘at’.

2.6.3 Clitic allomorphy in Makalero A morphophonological variation present in Makalero, but not in Makasae, concerns a set of five clause-level enclitics expressing a variety of temporal and aspectual notions as well as conditional clause linkage, listed in Table 11. Representative of the set as a whole, (17) shows the distribution of the allomorphs of the immediate future clitic Mkl /=faʔa/: the vowel-initial form appears after consonant-final roots, while the metathesized consonant-initial form appears after vowel-final roots. The consonant-initial form is chosen as the underlying form of the clitic for two reasons: given that the majority of morphemes in Makalero are vowel-final, the consonant-initial forms are more frequent; and the conditional clitics in their vowel-initial forms contain consonant clusters, which is highly untypical for Makalero morphemes. A contextual example of the use of these clitic allomorphs is given in (18).

286

Juliette Huber

Table 11: Makalero clitics with metathesized forms

Completive Conditional Counterfactual Conditional Intentional Immediate future

(17)

Makalero /=faʔa/ → [=afa] → [=faʔa]

(18)

Consonant-initial form

Vowel-initial form

[=teʔe] [=fata] [=nata] [=naʔa] [=faʔa]

[=ete] [=afta] [=anta] [=ana] [=afa]

/ C#=_ / V#=_

Makalero a. [aɛ utafaʔa] /ae uta=faʔa/ rain fall=IMM ‘It is about to start raining.’ b.

[ira mutu liʔanafa] /ira-mutu-liʔan=faʔa/ water-inside-fall=IMM ‘(He) is about to fall into the water.’

Makasae has clear cognates for three of these clitics: the completive marker Mks /(d)ete/ corresponds to Mkl /=teʔe/ ‘COMPL ’; the intentional Mks /nana/ to Mkl /=naʔa/ ‘INT ’; and the immediate future marker Mks /apa/ to Mkl /=faʔa/ ‘IMM ’. However, the Makasae markers do not metathesize, as illustrated in (19) with the immediate future marker. While they are given here as free forms, they appear to be phonologically dependent, and it is possible that they cliticize to the preceding verb; however, more data are needed to determine their status.

(19)

Makasae a. [uwa malɛn apa] /ua malen apa/ village be.near IMM ‘. . . about to (get) close to the village.’ b.

[nisirbisu malɛ saʔi gini apa] /ni-sirbisu male-saʔi gini REFL . POSS -work near-finish make ‘. . . about to finish one’s work.’

apa/ IMM

Makalero and Makasae

287

2.6.4 Vowel harmony in the deictic lV-prefix in Makasae A morphophonological process found only in Makasae concerns the /lV-/ prefix found with nominal and verbal spatial deictics in this language. It takes the form [le-] or [lo-] depending on the first vowel of the deictic; thus, with the addressee-related demonstrative Mks /were/, it appears as [le-]: Mks [le-were]. With the distal same-elevation demonstrative Mks /wori/, on the other hand, it takes the form [lo-]: Mks [lo-wori] (see Table 16 in section 4.2 for the full paradigm). The function of this prefix requires further research. According to Brotherson (2003: 164f.), it encodes a distinction between ‘location known to hearer’ (without prefix) and ‘location unknown to hearer’ (with prefix). This analysis is questioned in Huber (2008a: 68f.), although no alternative analysis is provided. Correia (2011: 160) characterizes the /lV-/prefixed deictic forms as emphatic.

2.6.5 Reduplication The most common morphological process in Makalero and Makasae is reduplication. In Makalero, full reduplication is most common, while in Makasae, full and partial reduplication appear to be used in roughly equal proportions. In the case of full reduplication, the final consonant of a reduplicant ending in a closed syllable is dropped to avoid a consonant cluster (20). In partial reduplication, only the first syllable is reduplicated (21). In addition, in Makasae a combination of both full and partial reduplication is relatively frequent (22). As in simplex morphemes (see section 2.5), the main stress falls on the last foot. Secondary stresses are assigned following a trochaic pattern. (20)

a.

b.

(21)

a.

b.

Makalero /leʔur/ ‘be black (?)’

/leʔu~leʔur/ ‘be black-ish’

Makasae /masin/ ‘usually’

/masi~masin/ ‘usually’

Makalero /toton/ ‘watch’

/to~toton/ ‘keep watching’

Makasae /sapu/ ‘brush, slide’

/sa~sapu/ ‘brush repeatedly’

288

(22)

Juliette Huber

Makasae a. /maʔa/ ‘earth, ground’ b.

/nake/ ‘take’

/maʔa~ma~maʔa/ ‘(many different) regions’ /nake~na~nake/ ‘take along (habitually)’

Reduplication has a variety of functions. In the nominal domain, it is mostly used to derive nouns from verbs, as illustrated in (23). However, this is not very frequent, presumably because verbs can freely be used as arguments without any derivation. Other uses include the derivation of nouns from other nouns, and distributive pluralization.

(23)

Makalero a. /nua/ /nua~nua/

Makasae b. /nawa/ /na~nawa/

c.

d.

/(k-)ini/ /kini~kini/

/gini/ /gi~gini/

‘eat’ ‘food’ ‘do, make’ ‘(customary) behavior’

With activity verbs, reduplication is read as denoting a continuing or repeated action, as seen in (21) above. With stative verbs, reduplication is read as a high degree of the quality expressed by the verb, or a quality resembling that expressed by the verb (24).

(24)

Makalero a. /felun/ /felu~felun/

Makasae b. /pelun/ /pe~pelun/

c.

d.

/putir/ /puti~putir/

/butir/ /buti~butir/

‘be nice, pretty’ ‘be very nice, pretty’ ‘be white’ ‘be white-ish’

2.6.6 Derivational morphology Apart from reduplication, there is very little productive derivational morphology in Makalero or Makasae. The most productive morpheme found is Mkl/Mks /-doːr/, an agent noun suffix illustrated in (25). This is a borrowing from Portuguese found in many Timorese languages (see, e.g., Hajek & Williams-van Klinken 2003).20 20 Correia (2011: 66f.) gives a number of native Makasae suffixes to derive deverbal nouns with person reference, namely Mks /-sege/, /-geʔe/ and /-guli/. It is not quite clear how productive these are. No Makalero equivalents to these suffixes are known.

Makalero and Makasae

(25)

Makalero a. /tuaʔ keʔ/ palm.wine drink ‘drink palm wine’ Makasae c. /ta tuku/ RECP punch ‘punch one another’

b.

d.

289

/tuaʔ-keʔ-doːr/ palm.wine-drink-AGT ‘(palm wine) drinker’ /ta-tuku-doːr/ RECP-punch-AGT

‘boxer’

A few unproductive derivational suffixes can be identified in both Makalero and Makasae, listed and illustrated in Table 12. Mkl/Mks /-ʔ/ is used to derive verbs from nouns and Mkl/Mks /-ini/ derives nouns from verbs. Finally, Makalero also has Mkl /-r-/, which is associated with nominalization. In Makasae, this formative is found exclusively in the deictic paradigm (see 26). The Makasae equivalents of Mkl /umur/ ‘dead body’ and Mkl /watu-misar/ ‘east’ use the underived forms, i.e., Mks /umu/ ‘be dead, a dead body’, and Mks /watu misa/ ‘east’. Table 12: Unproductive derivational morphemes in Makalero and Makasae Verbalizing -ʔ

Nominalizing -ini

Nominalizing -r-

Makalero

/ʔuri/ ‘brush’ (n.)

/ʔuriʔ/ ‘brush’ (v.)

Makasae†

/wali/ ‘ear’

/waliʔ/ ‘hear’

Makalero

/nua/ ‘eat’

/nua-ini/ ‘food’

Makasae

/paʔ/ ‘sew’

/paʔ-ini/ ‘everything related to sewing’

Makalero

/umu/ ‘die’

/umur/ ‘dead body’

/watu misa/ ‘sun go.up’

/watu-misar/ ‘east’

† This example is from Hull (2004).

Nominalizing /-r-/ and verbalizing /-ʔ/ are used prominently in the deictic paradigms of both languages, as illustrated in (26), which gives two deictic roots and the derived nominal and verbal forms. The derived nominal forms, Mkl/Mks /ere/ and /were/, can be used nominally or attributively. The corresponding verbal forms, Mkl/Mks /eʔ(e)/ and /weʔ(e)/, predicate the location of the subject participant (see section 4.2 for the full deictic paradigm).

290

(26)

Juliette Huber

Makalero / Makasae a. /e-/ /ere/ near.SPKR this.SPKR b.

/we-/ near.ADDR

/were/ this.ADDR

/eʔ(e)/ be.here.SPKR /weʔ(e)/ be.here.ADDR

2.6.7 Nominal compounding The head of the compound follows the modifying noun, as shown in (27). Compounding is particulary frequent with obligatorily possessed nouns such as body parts (section 4.3), where the modifying noun expresses a generic possessor. The modifier does not have a specific referent and cannot be marked as definite or otherwise modified. However, all nominal compounds can be paraphrased as possessive phrases, as seen in (28); in this case the possessor is understood to be specific rather than generic, and can be combined with modifiers.

(27)

(28)

Makalero a. /muʔu banana

fasu/ skin

Makasae b. /muʔu banana

c.

/ia raka/ leg digit

d.

/iti buli/ leg digit

e.

/asa chicken

f.

/asa chicken

a.

Makalero /asa ki-seur/ chicken 3.POSS -meat ‘(the) chicken’s meat’

b.

Makasae /asa gi-seu/ chicken 3.POSS -meat ‘(the) chicken’s meat’

seur/ meat

uli/ skin

‘banana peel’ ‘toe’

seu/ meat

‘chicken (meat), poultry’

In Makalero, the main stress in a nominal compound falls on the last foot, i.e., on the compound’s head. Further research is necessary, however, to determine whether it is possible for the modifier to be stressed, e.g. in a contrastive context, or whether a possessive phrase such as in (28) would have to be used in such cases. There is no information on stress in Makasae compounds.

Makalero and Makasae

291

Nominal compounds are distinguished from nominal modifiers (i.e. a noun in the modifier slot of the NP, see section 4.1) through the order of elements. A nominal modifier follows the head noun, as in (29). (29)

a.

b.

Makalero /asa chicken

paru/ hen

‘hen’

Makasae /aramari wardrobe

falasatiku/ plastic

‘plastic wardrobe’ (Correia 2011: 85)

3 Basic clausal syntax Monopredicative clauses in both Makalero and Makasae minimally consist of a predicate and up to two arguments. The basic order of constituents is SV/APV. The majority of modifiers precede the predicate.21 A prototypical P argument in a transitive clause precedes the verbal modifiers, as shown in (30). An asterisk indicates that a given element may occur more than once.

(30)

Makalero/Makasae transitive clause template (ARG A ) – (ARG P ) – (ASP ) – (ADV )* – (NEG ) – (ADV )* – PRED

In multi-predicate clauses, there is no formal asymmetry between final and non-final predicates; this is a consequence of the isolating nature of Makalero and Makasae, which mostly lack the verbal agreement or derivational morphology found in other languages of the family (see section 7 for a discussion of the difficulties involved in the analysis of serial verb constructions and related phenomena). The predicates in a multi-predicate clause usually share at least one argument. Aspect, adverbials and negators precede the first predicate. The following predicate(s) cannot be independently modified. Arguments are optional and are readily omitted if the speaker assumes the hearer can infer the intended referent from either the linguistic or the extralinguistic context.

21 A few modifiers follow the predicate. In both languages, some elements expressing phasaland tense-like notions are attached to the right boundary of the verb phrase and thus follow the verb. See section 8.2 for a discussion.

292

Juliette Huber

There is relatively little variation to the basic SV/APV word order, though arguments from both monopredicative and multipredicative clauses can be leftdislocated. As in (31), this dislocation is usually topicalizing in function. (31) a. Makalero . . . ani were nua=ni heʔul. [Ki-resin ere]T O P ani mei=ni 3.POSS -rest DEM 1SG take=CONJ 1SG DEM eat=CONJ full ni-noko-raa kini. . . REFL . POSS -younger.sibling-PL give.to.3 ‘. . . I ate that (until I was) full. The rest I gave to my younger siblings. . .’ b. Makasae . . . era lolo seti lolo Fretilin lari? [Lari ere]T O P ani to mi-ger. 1SG NEG along-think 3PL say ask say Fretilin run run DEM ‘. . . they asked me (in Indonesian) ‘Fretilin lari?’. The (word) ‘lari’ I did not understand. . .’ A left-dislocated participant may also be in focus, though this is much rarer. Mostly, these are long and complicated constituents, as in (32) from Makalero. This sentence also shows that in such cases the position of the displaced constituent in the clause is usually occupied by a resumptive element. Makalero ini wereʔ nonton. . . (32) . . . [ filem tifi-mutuʔ ki=felu~felun]F OC movie TV-be.inside ATTR = RDP ~be.nice 1PL . EXCL DEM .VBLZ watch ‘. . . nice movies that were on TV, we watched those. . .’ Right-dislocation, on the other hand, is uncommon. In Makalero, it is found in rare circumstances with question words and afterthoughts. None of the existing sources mention right-dislocation in Makasae. The following sections discuss verbal predicates (section 3.1), non-verbal predicates (section 3.2), the INCORP slots (section 3.3), negation (section 3.4), adverbs (section 3.5), questions (section 3.6) and imperatives (section 3.7).

3.1 Verbal predicates Makalero and Makasae are largely isolating, and exhibit a high degree of word class flexibility. Basically any content word, including those that would prototypically be translated as nouns, can be used predicatively. As such, the definition of parts of speech such as verbs is not straightforward. Verbs are prototypically used predicatively. Used in argument position, they typically translate

Makalero and Makasae

293

as ‘the act of x-ing’, ‘the product/result of x-ing’, or ‘the instrument involved in x-ing’. The class of verbs includes both activities, such as Mkl nua / Mks nawa ‘eat’, Mkl/Mks laʔa ‘go’, or Mkl huri / Mks suri ‘shoot’, as well as statives, such as Mkl pere / Mks bere ‘be big’, Mkl mutuʔ / Mks mutu ‘be inside’ and Mkl akaʔ / Mks agaʔ ‘fear, be afraid’. Note that stative verbs include the translational equivalents of adjectives and adpositional meanings.

3.1.1 Intransitive verbs Intransitive verbs take a single S argument, as illustrated in (33). (33)

a.

Makalero hai hau haiʔ. . . . [ki-iskola]S 3.POSS -school INITIAL FINAL be.over ‘. . . his education is already complete.’

b.

Makasae saʔi. . . . . . iskolaS au school FINAL be.over ‘. . . (after) school was over. . . .’

Intransitive verbs can be combined with phrases indicating location and manner adjuncts. As the bracketing in the examples below indicates, these phrases are analyzed as standing in the INCORP LOC slot (see section 3.3.2 for arguments supporting this analysis); an example with an intransitive verb is shown in (34). With some position or movement verbs such as Mkl mit / Mks mi ‘sit’, Mkl nat / Mks na ‘stand’ or Mkl liʔan / Mks desar ‘fall’ such locative phrases are in fact obligatory, and their omission results in an ungrammatical sentence. This is illustrated in the Makalero sentence in (35a), the intended reading for which was ‘I already fell (over)’, with the initial boundary marker hai (see section 8.1). However, Makalero speakers consistently rejected this reading, and insisted that the only possible interpretation of this sentence is the one shown in (35a), where (h)ai is interpreted as a variant form of the truncated deictic verb e- (see section 2.6.2). There is no direct evidence that the Makasae equivalent desar ‘fall’ is ungrammatical without such a locative; however, this verb is never used without one in my corpus or any of the consulted sources. Instead, the truncated deictic verb ho- appears to be used as a default locative with this verb, used whenever no more specific information is given (35b).

294 (34)

(35)

Juliette Huber

a.

Makalero k-ata]L OC-numu. KilooS [ni-upa REFL . POSS -father 3.OBJ -near-die.BOUND 3SG ‘He died next to his father.’

b.

Makasae IniS beʔu [ii gata]L O C-umu. 1PL . EXCL can 2PL near-die ‘We could die next to you.’

a.

Makalero AniS aiL OC-liʔan. 1SG here-fall ‘I fell here.’

b.

Makasae hoL OC-desar. AiS 2SG here.below-fall ‘You fell (over / down).’

3.1.2 Transitive verbs The most common position for the P argument of a transitive verb is shown in the template in (30), repeated here in simplified form as (36). In this position, which will be referred to as the object position, the P argument precedes modifiers such as adverbials, aspect and negation.

(36)

Makalero/Makasae simplified transitive clause template (ARGA ) – (ARG P ) – (MOD ) – PRED

The object position is mainly associated with undergoers which are referential and expressed through lexical nouns rather than pronouns. The sentences in (37) and (38) illustrate the position of such P arguments with respect to the pre-predicative modifiers, an adverb in the case of (37) and a negation in the case of (38). (37)

a.

Makalero Mata ere [ni-pada]P child DEM REFL . POSS -friend ‘This child hit his friend.’

hau FINAL

pase. hit

Makalero and Makasae

b.

(38)

a.

b.

Makasae Mata wori [ni-bada]P child DEM REFL . POSS -friend ‘That child hit his friend.’ Makalero nomo Ani [ni-upa]P 1SG REFL . POSS -father NEG ‘I didn’t know my father.’

au FINAL

295

base. hit

hofe. know

Makasae noto Ani [ni-baba]P 1SG REFL . POSS -father NEG . EMPH ‘I didn’t know my father.’

sope. know

In specific contexts, arguments are expressed in the INCORP slots indicated in (39) and (40). These slots are directly adjacent to the verb, following the modifiers, and are associated with reduced transitivity. The INCORP P slot indicated in (39) is used when the P argument to a transitive verb is non-referential or pronominal; the INCORP LOC slot indicated in (40), on the other hand, is used to express objects of movement and positional verbs. The INCORP slots are discussed in detail in section 3.3.

(39)

Makalero/Makasae incorporated P slot (ARGA ) – (MOD ) – (INCORP P )-V

(40)

Makalero/Makasae incorporated locative slot (ARGA ) – (MOD ) – (INCORP L OC )-V

3.1.3 Ditransitive verbs There are no syntactically ditransitive verbs in Makalero and Makasae. A state of affairs involving three participants must be expressed in Makalero in a serial verb construction with Mkl mei ‘take’, and in a complex construction with an oblique marker, Mks ma, in Makasae, as illustrated in (41); see section 7.2 for a more detailed discussion. Such argument flagging is common in TAP languages, although it is not usually obligatory.

296 (41)

Juliette Huber

a.

b.

Makalero Ani dudu ira 1SG breast water ‘I gave him milk.’ Makasae Ani dudu ira 1SG breast water ‘I gave him milk.’

mei=ni take=CONJ

kiloo 3SG

ma

gini. give

OBL

gi 3SG

kini. give.to.3

3.2 Non-verbal predicates Adpositional meanings are expressed by transitive positional verbs (section 3.3.2); hence non-verbal predicates in Makalero and Makasae are essentially nominal. Simple nouns or full noun phrases can function as predicates if placed in the appropriate position in a clause. In these cases, the predicate is not a property, but an entity, which is equated with the entity referred to by the S argument. In other words, nominal predicates are normally read as ‘is (the/an) X’. Nominal predicates denote stative states of affairs. (42) gives examples of noun phrases used predicatively. (42)

a.

Makalero . . . ani [kiloo ki-halafi]P R E D . 3.POSS -enemy 1SG 3SG ‘. . . I am his enemy.’

b.

Makasae . . . ani ere [gi-soli]P R E D . 1SG DEM 3.POSS -enemy ‘. . . I am his enemy.’

In Makalero, the demonstratives are the only elements to require derivation in order to be used in the predicate slot. An example is given in (43), where Mkl wereʔ is a verbalized form of the near-addressee demonstrative were. Such verbalized forms are derived from the nominal forms by means of an unproductive but fairly frequent glottal stop verbalizer; they are distinct from the deictic verbs formed from the same stems (see sections 2.6.6 and 4.2). A comparable phenomenon is not attested in the Makasae corpus, where underived demonstratives are used as nominal predicates in this context as in (44).

Makalero and Makasae

(43)

(44)

Makalero . . . ki-nei ki-apalidu-ni 3.POSS -name 3.POSS -name ‘. . . their names are these.’

297

wereʔP R E D . DEM .VBLZ

Makasae Balde waʔa gi-noko ere geʔe le-ereP R E D . 3.POSS -younger.sibling DEM possession LE - DEM bucket REL ‘The bucket which belongs to his younger sister is this (one).’

3.3 The INCORP slots Section 3.1.2 briefly introduced the two slots shown in (45) and (46), which are used to express arguments to transitive verbs in specific contexts. Both of these slots are associated with reduced transitivity and are immediately adjacent to the verb. They are therefore reminiscent of the type of incorporation by juxtaposition frequently described in Oceanic languages (Mithun 1984: 849) and are accordingly referred to as incorporated slots.

(45)

Makalero/Makasae incorporated P slot (ARG A ) – (MOD ) – (INCORP P )-V

(46)

Makalero/Makasae incorporated locative slot (ARG A ) – (MOD ) – (INCORP L OC )-V

The two incorporated slots cannot co-occur, hence it is impossible to determine their relative sequence. However, they differ in one important property: the INCORP L OC slot (46) triggers the initial consonant mutations in the following verb discussed in section 2.6.1, while the INCORP P slot (45) does not. For this reason, they are treated as two distinct slots, despite their superficial similarity.

3.3.1 The INCORP P slot The INCORP P slot, which does not trigger initial consonant changes in the following verb, is associated with a specific type of P argument in clauses with transitive verbs in Makalero, namely the reflexive pronoun Mkl ni as shown in (47a), the reciprocal pronoun Mkl ta as shown in (48a), as well as personal pronouns with object-marking verbs (section 6.3). In Makasae, the use of the INCORP P slot

298

Juliette Huber

is more restricted. It appears not to be available for personal pronouns or the reflexive marker, which stand in the more frequent object position illustrated in (36), as seen in (47b). The only P argument which has been found in the INCORP P slot in Makasae is the reciprocal marker Mks ta, as shown in (48b) from the Ossu dialect. Note, however, that according to Correia (2011: 130), both Mks ni and Mks ta stand in the object position in the Laga dialect, suggesting that this dialect of Makasae does not have an INCORP P slot at all. (47)

(48)

a.

Makalero Ani hau niI N C O R P-waroh. 1SG FINAL REFL-bathe ‘I have finished showering.’

b.

Makasae waruʔ. Ani niO B J au 1SG REFL FINAL bathe ‘I have finished showering.’

a.

Makalero Lopu-laa ere hai taI N C OR P -mi-naser. house-PL DEM INITIAL RECP-along-stand.PL ‘These houses stand alongside one another.’

b.

Makasae mara. Anu mahe ai taI N C OR P -mi-e-da person two.HUM INITIAL RECP-along-here-pass go ‘The two are walking one after another.’ (Brotherson 2003: 77)

In Makalero, the INCORP P slot is used furthermore to express non-referential P arguments. This is the case in particular with object-marking verbs (section 6.3). There are a few cases where non-referential Ps stand in the INCORP P slot of non-object-marking verbs, as in (49) and (50), where naʔa ‘work’ and sa ‘wife’ do not refer to concrete entities, but form complex expressions together with the following verbs. Further research is required to determine whether this is actually a productive pattern, or whether we are dealing with a small number of lexicalized expressions only. P arguments in the form of referential lexical nouns are expressed in the object position (36). (49)

Makalero Kiloo hai nomo naʔaI NC OR P -mei. INITIAL NEG work-take 3SG ‘He is no longer working.’

Makalero and Makasae

(50)

299

Makalero Kiloo hai saI NC OR P -ena=naʔa. INITIAL wife-see=INT 3SG ‘He was going to get married.’

3.3.2 The INCORP LO C slot The INCORP L OC slot has two functions which are the same in Makalero and Makasae, namely the expression of objects of positional and motion verbs, and the expression of locative and manner adjuncts with all other kinds of verb. These functions are discussed below. In Makalero and some dialects of Makasae, the INCORP L O C slot is furthermore used to express caused actions; this is the topic of section 7.2.4. The first function of the INCORP L OC slot is the expression of objects of positional and motion verbs. Some examples of this verb class are shown in Table 13. Note that the positional verbs semantically correspond to adpositions. The position of such locative arguments is illustrated in (51) with a positional verb, and in (52) with a movement verb. Table 13: Transitive verbs taking locative arguments22

‘be at’ ‘be inside’ ‘go down, descend’ ‘jump’

(51)

Makalero

Makasae

isiʔ mutuʔ isa heʔel

isi mutu isi seʔel

a.

Makalero Ani hai [ni-lopu]L OC-mutuʔ. 1SG INITIAL REFL . POSS -house-be.inside ‘I am already in my house.’

b.

Makasae Ani hai [ni-oma]L OC-mutu. 1SG INITIAL REFL . POSS -house-be.inside ‘I am already in my house.’

22 Table 13 lists the Makasae equivalent of ‘be at’ as Mks isi. There is also a form Mks isiʔ, which is used as the main predicate in an utterance. Brotherson (2003: 101) analyzes Mks isiʔ as a verbal form and Mks isi as a postposition. However, Mks isi can also be used predicatively; I thus suggest that Mks isiʔ and Mks isi are not actually distinguished by their categorical status; rather, this variation is due to the ongoing loss of the glottal stop verbalizer in Makasae.

300 (52)

Juliette Huber

a.

Makalero Ani hai muʔaL OC-liʔan. 1SG INITIAL ground-fall ‘I have fallen to the ground.’

b.

Makasae Ani hai muʔaL OC-desar. 1SG INITIAL ground-fall ‘I have fallen to the ground.’ (Brotherson 2003: 152)

The second function of the INCORP LOC slot is the expression of locative and manner adjuncts. Such adjuncts can be combined with both intransitive and transitive verbs. Interestingly, these adjuncts are usually complex phrases made up of a positional verb and its argument as illustrated in (53). Example (54) shows that the locative adjuncts used in (53) can also function independently as full predicates. (53)

(54)

a.

Makalero Kaila fun un [isi-lopu-male]L O C-nat. papaya stem one 1PL . EXCL . POSS -house-near-stand.SG ‘A papaya tree was standing near our house.’

b.

Makasae Ate u [igreza-male]L OC-e-na. wood one church-near-here-stand.SG ‘A tree is standing near the church.’ (Brotherson 2003: 110)

a.

Makalero Kiloo laʔa=ni [isi-lopu-maleʔ]. go=CONJ 1PL . EXCL . POSS -house-be.near 3SG ‘He came close to our house.’

b.

Makasae Gi laʔa [igreza-malen]. church-be.near 3SG go ‘He got close to the church.’

Several locative phrases can be coordinated in the INCORP LOC slot, as shown in (55).

Makalero and Makasae

(55)

a.

Makalero . . . kilooraa . . . [ate k-ua]LO C-udeL O C-diar. . . tree 3.OBJ-on.top-up.there-sit.PL 3PL ‘. . . they (. . .) sat on top of a tree up there. . .’

b.

Makasae . . . [ate bata-gua]L OC-doL O C-namaL O C-diar. tree log-on.top-up.there-upwards-sit.PL ‘. . . (they) were sitting up on the log.’ (Brotherson 2003: 141)

301

The analysis of a complex phrase as standing in an INCORP slot is unusual, and more research is required in order to determine the exact syntactic status of these adjuncts. It is clear, in any case, that locative adjuncts and arguments of positional and movement verbs as illustrated in (51) and (52) are syntactically equivalent: they both trigger the use of bound forms with the verbs discussed in section 2.6.1. Furthermore, it is possible to coordinate locative argument NPs and complex locative phrases in the INCORP LOC slot, as seen in example (56), and nominal locative arguments and verbal locatives are found in this slot in parallel constructions, as shown in example (57). Finally, scope differences are observed depending on whether an aspect marker is placed within the verbal adjunct phrase or outside of it (see Huber 2011: 333).23 Makalero (56) . . . fi laʔa laʔa laʔa=ni [SMP-isi]L O C-muʔaL OC -isa=ni. . . 1PL . INCL go go go=CONJ junior.high.school-at-ground-go.down=CONJ ‘. . . we keep going and then go downwards at the junior high school. . .’ Makalero (57) . . . noko nana=ni . . . nomo fuliL OC-reiL O C-laʔa younger.sibling elder.sibling=CONTR NEG together-outwards-go omaL O C-laʔa. . . house.REDUCED -go ‘. . . siblings (. . .) don’t go out together and (don’t) go to (each other’s) house. . .’ 23 Locatives in the I NCORP L O C slot appear to be equivalent to the incorporated P arguments discussed in section 3.3.1 in that they are mutually exclusive with such arguments. Moreover, it could be argued that locatives stand in a closer relationship to the verb than incorporated P arguments, given that bound verb forms are found only with locatives, but not with incorporated P arguments. Finally, like incorporated P arguments, locatives are associated with reduced transitivity (see section 7.2.3). The incorporation of locatives is the best syntactic cognate of incorporated postpositions as found in Kamang (Schapper 2014) and Blagar (Steinhauer 2014).

302

Juliette Huber

Manner adjuncts as illustrated in (58) and (59) are rarer than locatives. These adjuncts are not generally complex phrases, but one-word expressions (although reduplication is permissible, as seen in (58b)). Manner adjuncts are predominantly verbal, although a few nominal expressions have been found in Makalero (59). (58)

(59)

a.

Makalero Kiloo ini hai mei rauM A N N E R-suma. . . 1PL . EXCL INITIAL take be.good-be.angry.BOUND 3SG ‘He was very (properly) angry with us. . .’

b.

Makasae sirbisu ere ma DEM OBL work ‘do the work well’

[rau~rau]M A N N E R-gini RDP ~be.good-do

Makalero Ani naʔu ateM A N N E R-nata. tree-stand.SG 1SG just ‘I’ll just stand upright’.

The INCORP LOC slot triggers the morphophonological changes discussed in sections 2.6.1 and 2.6.2. Mkl maleʔ and Mks malen ‘be near’ in (53) and (54) may serve as examples for verbs which appear in the INCORP LOC slot in truncated forms; Mkl/Mks taru ‘put, place, bury’ in (60) and (61) illustrate the verbal subgroup which exhibit initial consonant mutations if combined with an adjunct in the INCORP LOC slot. (60)

(61)

a.

Makalero . . . asi-upa hau taru. . . 1SG . POSS -father FINAL put ‘. . . after (we) buried my father. . .’

b.

Makasae . . . umu gapu rate-isi-laʔa to taru. . . dead with grave-at-go PURP put ‘. . . (they) went to the grave with the body to bury it.’

a.

Makalero ... fi pusi mei [ata k-ua]L O C-daru. . . 1PL . INCL pan take fire 3.OBJ-on.top-put.BOUND ‘. . . we put the pan on top of the fire. . .’

Makalero and Makasae

b.

Makasae . . . ai

303

laʔa [meza-gua]L OC-woiL OC -daru. INITIAL OBL move table-on.top-over.there-put.BOUND ‘. . . she moves it (and) puts it on the table.’ (Brotherson 2003: 68) ma

Where the INCORP LOC slot expresses an adjunct to a transitive verb, the difference between the two incorporated slots can clearly be shown. In the sentences in (62), with the reciprocal pronoun in the INCORP P slot, the verb appears in the free form; but in (63), where it is preceded by a locative adjunct in the INCORP LOC slot, the same verb appears in the bound form. (62)

(63)

a.

Makalero . . . amuni hai tepa ta-huri. . . person INITIAL constantly RECP-shoot ‘. . . people kept shooting (at) each other. . .’

b.

Makasae . . . ini mahe ai ta-ena gi=rau. . . 1PL . EXCL two.HUM INITIAL RECP-see ATTR =be.good ‘. . . we had (already) become friends. . .’ (lit., ‘we were looking at one another in a good way’)

a.

Makalero . . . ini . . . fatil mei nama-suri. . . 1PL . EXCL gun take upwards-shoot.BOUND ‘. . . we (. . .) will shoot a gun upwards (into the air). . .’

b.

Makasae . . . era . . . lari=ee gata-gena. mountain=DEF near-see.BOUND 3PL ‘. . . they (. . .) looked towards the mountains.’

With transitive verbs, the INCORP LOC slot interacts with valency in that the presence of an adjunct in this slot causes the verb’s logical object to be expressed as an argument to the light verb Mkl mei ‘take’ in the case of Makalero, or to be marked with the oblique marker Mks ma in the case of Makasae (see section 7.2.6); that is, the INCORP LOC slot reduces the verb’s valency. This construction is discussed in detail in section 7.2.3.

304

Juliette Huber

3.4 Negation Makalero and Makasae have two distinct negator positions, each filled with separate lexical negators. The first negator position is that of the internal (predicate) negation. While Makalero has one internal negator, Mkl nomo, there are several forms in Makasae: Mks to, noto, noko and noho. Mks to is the most common internal negator; Mks noto is used in the same contexts, and is somewhat more emphatic than to. Mks noko and noho are very rare and appear to be lexically limited to one verb each; the static Mks rau ‘be good’ in the case of Mks noko, and Mks beʔu ‘be allowed to’ in the case of Mks noho. The internal negator is one of the pre-predicative modifiers; its position is illustrated in the clause template in (30). It always follows the aspect marker Mkl hai / Mks ai (see section 8.1), while its position vis-à-vis the pre-predicative adverbs (section 3.5) depends on the intended scope, as illustrated in (64) and (65). The pre-predicative position of the internal negation is unique to the Eastern Timor languages, and is found also in Fataluku. In contrast, negation is post-predicative in the other TAP languages.

(64)

Makasae a. Asi-tumata to mega ani-isi-maʔu. 1SG -at-come 1SG . POSS -nephew NEG ever ‘My nephew never comes to see me.’ b.

...

situasaun ere mega to azuda. . . NEG help situation DEM ever ‘. . . the situation was entirely unsupportive. . .’

Makalero (65) a. Timor ki-renu hai naʔu nomo tafi se roual. T. 3.POSS -populace INITIAL just NEG true very be.many.NHUM ‘The population of Timor is already not truly very large.’ b. Timor ki-renu hai naʔu tafi nomo se roual. T. 3.POSS -populace INITIAL just true NEG very be.many.NHUM ‘The population of Timor is already truly not very large.’ The second negator position is clause-initial and is used for external (sentential) negation. In Makalero, the external negator is Mkl nomohaka, and Mks tonai and notonai in Makasae.24 As in the case of the internal negators Mks 24 Correia (2011: 48) further mentions the forms tosai and notosai, which appear to be characteristic of the Laga dialect of Makasae.

Makalero and Makasae

305

to and noto, the difference between these two forms appears to be one of emphasis.25 Example (66) illustrates the clause-initial position of the external negator. It is also very common for the subject NP to be left-dislocated, in which case it precedes the external negator, as seen in (67). (66) a. Makalero Nomohaka ani lopu teuh=eʔ tule. . . EXT. NEG 1SG house buy=CMP refuse ‘It is not the case that I don’t want to buy a house (but. . .)’ b. Makasae Tonai ani tula karak oma tehu. . . EXT. NEG 1SG refuse want house buy ‘It is not the case that I don’t want to buy a house (but. . .)’ (67) a. Makalero Dai ere nomohaka amu ki=selu wainpo ki-meestri. person ATTR =other but 3.POSS -teacher foreigner DEM EXT. NEG ‘This foreigner was no-one else but his teacher.’ b. Makasae Dai ere tonai ana gi=selu ba gi-meestri. foreigner DEM EXT. NEG person ATTR =other but 3.POSS -teacher ‘This foreigner was no one else but his teacher.’

3.5 Adverbs Among the pre-predicative modifiers are two slots in which a limited number of adverbial elements such as Mkl naʔu ‘just’, wari ‘still’, and teni ‘again’, and Mks naʔu ‘just’, teni ‘again’, and mega ‘ever’ can appear. These may either precede or follow the internal negator, according to the intended scope; examples are given in (64) for Makasae and in (65) for Makalero. The Makalero sentences in (65) also show that adverbial elements can stand on both sides of the negation in the same sentence. The most prominent member of this class of adverbs is Mkl hau / Mks au, which is grammaticalizing into an aspect marker and is discussed in detail in section 8.1.

25 Mks tonai and notonai are also used as negative sentence equivalents, while Makalero uses the negative existential Mkl naʔan in these contexts.

306

Juliette Huber

Makalero and Makasae also have a second type of adverbial, which can appear either clause-initially or following the subject, as illustrated in (68) and (69). In both languages, this type of adverbial expresses mostly temporal and aspectual as well as some modal notions.

(68)

Makalero a. Hanaʔe

ini Soibada-isiʔ. 1PL . EXCL Soibada-be.at ‘We used to stay in Soibada.’ REM . PST

b.

(69)

Ini hanaʔe Suai-isiʔ. 1PL . EXCL REM . PST Suai-be.at ‘We used to stay in Suai.’

Makasae a. Opoloi

gi oma-isi-laʔa. 3SG house-at-go ‘He has just gone home.’

REC . PST

b.

Gi opoloi oma-isi-laʔa. 3SG REC . PST house-at-go ‘He has just gone home.’

These adverbials can host conjunctive elements such as Mkl =(i)ni (the most general conjunction) or Mks (d)ete ‘after’, as seen in (70). This characteristic suggests that these adverbials are predicative; indeed, in Huber (2011), I have treated them as avalent verbs. (70)

a.

b.

Makalero Kiloo aiteʔ=ini REC . PST = CONJ 3SG ‘He only just arrived.’

maʔu. come

Makasae Gi opoloi dete maʔu. 3SG REC . PST CONJ come ‘He only just arrived.’

3.6 Questions Interrogative markers are used in Makalero and Makasae in a largely equivalent way. It is noticeable, however, that many of the items in use are non-cognate.

Makalero and Makasae

307

Section 3.6.1 discusses polar questions. Information questions are treated in section 3.6.2.

3.6.1 Polar questions Polar questions may be morphologically unmarked in both languages, as shown in (71). These questions are distinguished from declarative clauses only through their rising-falling intonation (see Huber 2011: 439 for an intonation contour of a polar question in Makalero). (71)

a.

Makalero Ei soʔot=ini ni-weri leuk? 2SG want=CONJ REFL . POSS -bottom drill? ‘Do you want your bottom to be drilled?’

b.

Makasae Ehani ai karak sirbisu? today 2SG want work ‘Do you want to work today?’

Both languages also have overt question markers, as listed in Table 14. The sentences in (72) show that the question markers stand at the right edge of the clause. While Mkl =wa is analyzed as a clitic, the Makasae markers are treated as free elements in all of the sources consulted. However, they are phonologically bound and attach to any clause-final element; so they might also be analyzed as clitics. Like Mkl =wa, Mks tai appears to be a dedicated question marker. Mks hani is originally a verb meaning ‘be like, alike’, which is also used as a marker of indirect evidence and epistemic modality (‘maybe’); Mks dawa, too, marks epistemic modality (‘perhaps’) and also functions as a conditional conjunction ‘if’. Next to these question markers, Table 14 also lists a tag question, Mkl =wai naʔan and Mks ini tonai, which contrasts with the dedicated question markers in that it has the form of a regular short clause including a conjunction. In both languages, the tag question translates literally as ‘or not’ and is used in a way equivalent to the other question markers. The tag question is commonly shortened to the conjunction ‘or’, Mkl =wai and Mks ini, respectively. Table 14: Question markers in polar questions Makalero

Makasae

Question markers

=wa

tai, hani, dawa

Tag question

=wai (naʔan)

ini (tonai)

308

(72)

Juliette Huber

a.

Makalero Iliomar ki-isiʔ hoʔo hofe=wa? Iliomar 3.POSS -belong some know=QM ‘(Did you) know some (people) from Iliomar?’

b.

Makasae Gi maʔu tai? 3SG come QM ‘Does he come?’

3.6.2 Information questions Table 15 lists the question words found in Makalero and Makasae. Again, few cognate elements are used in this domain. In both languages, the interrogative for ‘what’ is also used as a noun meaning ‘thing’. Table 15: Question words

‘what’ ‘who’ ‘where’ ‘which’ ‘why’ ‘when’ ‘how’ ‘how many’

Makalero

Makasae

saʔa(ni) muʔa(ni) tauʔ taureʔ saʔa-faniʔ tetepaneʔ taureʔ faniʔ tauropaʔ, teteropaʔ

naʔi nahireʔe nahi (galu), naʔi mutu nahi galu, nahi teʔe, nahi gaʔu naʔi gau nahi, naʔi-naʔi naʔi hani, netani, netalapu nahiroba

As shown in section 5.4, Mkl saʔa(ni) ‘what’ and Mkl muʔa(ni) ‘who’ are pronouns and can either head a NP or appear in adnominal position as possessors; the remainder of the Makalero interrogatives are used predicatively. An example is given in (73a). Mks naʔi ‘what’, too, is pronominal, and as seen in Table 15, this root (or a form Mks nahi) forms the basis of all other interrogatives. While the category of some of these (e.g., Mks nahireʔe ‘who’) requires more research, several of the elements which are combined with the base Mks naʔi are verbal (e.g. Mks gau ‘be for’, Mks hani ‘be like, alike’), and the sentence in (73b) shows that they are used predicatively like their Makalero equivalents. (73)

a.

Makalero Sefar=ee hai laʔa dog=DEF INITIAL go ‘Where did the dog go?’

tauʔ? be.where

Makalero and Makasae

b.

309

Makasae Depa ai laʔa nahigalu? INITIAL go be.where dog ‘Where did the dog go?’

3.7 Imperatives Imperatives may be either morphologically unmarked or marked with a special particle. In unmarked imperatives (74), Mkl hau / Mks au, an aspect marker associated with the completion of an action, is frequently used. (74)

a.

b.

Makalero . . . hoʔo mei ni-mata some take REFL . POSS -child ‘. . . give some to your child!’ Makasae Tali ere au rope DEM FINAL ‘Untie this rope!’

hau FINAL

kini! give.to.3

dasaʔ! untie

Imperatives can be marked in Makalero with the clause-final clitics Mkl =wai and Mkl =(e)lo (75). These markers soften the imperatives and are considered more polite. In Makasae, an unrelated particle Mks (m)ara is used (76). Mks mara, in the form with the initial m, is homophonous with Mks mara ‘go’, a verb that denotes movement away from the deictic center. It is unclear whether it is etymologically related to that verb.

(75)

(76)

Makalero Asi-ropa hau pane=lo! 1SG . POSS -clothes FINAL wash=IMP ‘(Please) wash my clothes!’ Makasae Tali ere au dasa rope DEM FINAL untie ‘Untie that rope (please)!’

mara! IMP

Prohibitives are formed with Mkl irau / Mks erau in clause-initial position, as shown in (77). If the addressee is overtly expressed, Mkl irau / Mks erau stands between it and the remainder of the clause.

310 (77)

Juliette Huber

a.

b.

Makalero Irau ani-asu osan PROH 1SG -for money ‘Don’t give me money!’

manini! give.to.1SG

Makasae Erau ma

ani gini! 1SG give ‘Don’t give (it) to me!’

PROH

OBL

Mks erau is always replaceable by to beʔu (‘NEG allowed.to’). No equivalent paraphrase is known for Makalero.

4 Noun phrases The structure of the noun phrase as represented in (78) is identical in both Makalero and Makasae. Except for the possessor, which stands to the left of the head, the noun phrase is head-initial. An asterisk signals that a given item can occur more than once. The Makalero and Makasae NPs in (79) illustrate this template.

(78)

Makalero/Makasae NP template (POSS ) – (HEAD ) – (MOD )* − (DET )

(79) a. Makalero [asiP OS S -padaH E A D [u ki-nei Karlitu]M O D wereD E T ]N P 1SG . POSS -friend one 3.POSS -name Karlitu DEM ‘this one friend of mine whose name is Carlito’ b. Makasae [asiP OS S -rakuH E AD [u gi-nai Armindu damu]M O D ereD E T ]N P 1SG . POSS -friend one 3.POSS -name Armindu be.named DEM ‘one of my friends who is called Armindo’ Prototypically, a noun phrase is headed by either a pronoun (see section 5), a simple noun or a nominal compound (see section 2.6.7). Note that the head is marked as an optional constituent in (78) above. Noun phrases without nominal heads are exemplified in (80).

Makalero and Makasae

(80)

a.

Makalero [[ki=pere]M O D hoʔoD E T ]NP ATTR =big.SG some ‘some big (ones)’

b.

311

Makasae [[gi=kuini]M O D woriD E T ]N P ATTR =big DEM ‘that big (one)’ (Brotherson 2003: 27)

A possessor can only be expressed if a nominal head is present. Furthermore, in Makalero, the head is the element to which the plural suffixes are attached. In Makasae, on the other hand, the cognate plural clitic can appear on elements other than noun phrase heads (see section 4.4). The following sections discuss various kinds of modifiers (section 4.1), determiners (section 4.2), possession (section 4.3), and plural marking (section 4.4).

4.1 Modifiers This section discusses three types of modifiers: nominal, verbal, and clausal. Nominal modifiers follow the head noun without any marking and specify sex, age, origin, location, or material of the referent of the head noun, as illustrated in (81). (81)

a.

Makalero [asa paruM O D bird hen ‘the one hen’

b. un=ee]NP one=DEF

Makasae [oma apaM O D ]N P house stone ‘stone house’

Verbal modifiers are most often stative verbs describing the head and may appear without a marker, or with the attributive marker Mkl ki= / Mks gi=, which is homophonous with and most likely related to the third person (possessive) pronoun (see section 4.3). The main difference between the use of an unmarked verbal modifier and one with the attributive marker in Makalero is that there is a contrastive notion attached to the modifier with Mkl ki=, as illustrated in (82). Mks gi= is used in a parallel construction (83). According to Correia (2011: 289), it is obligatory when the modifier is an inherited TAP stative verb, but optional when it is an Austronesian (or Portuguese) loan.26 It is unclear whether the use of the marker in Makasae also conveys a contrastive notion in those cases where it is optional. In Makalero, it may also be used with activities (84), although this 26 Note, however, that it is not always easy to determine whether a given root is Papuan (TAP) or Austronesian. Many roots are widespread in both Austronesian and Papuan languages of the island; and there are many ‘local’ Austronesian roots that are not found outside of East Nusantara, which could conceivably be old Papuan loans.

312

Juliette Huber

is rare. No comparable use of Mks gi= with an activity verb has been found. A similar attributive marker is found in Blagar (Steinhauer 2014: 179) as well as Teiwa (Klamer 2010: 207).

(82)

Makalero were]NP rei-liʔan. a. Wai=teʔe=fe [nana [ki=pere]M O D CLS =after=? snake ATTR =be.big.SG DEM outwards-fall ‘Only then did the big snake (as opposed to many small ones) fall out.’ b.

(83)

(84)

k-asu kilooraa [nana pereM O D ]N P mei=ni snake be.big.SG take=CONJ 3.OBJ-be.for 3PL ‘. . . they show him the big snake.’ (only one snake) ...

Makasae Ani ere [guru [gi=seluʔ]M O D ]NP =ini kimia 1SG DEM teacher ATTR =other=CONTR chemistry ‘Me, a different teacher teaches me chemistry.’

ma OBL

ani 1SG

isi-suma. at-show

pana. teach

Makalero ira [ki=keʔ]M O D ]N P water ATTR =drink ‘drinking water’

Clausal modifiers, too, may be marked or unmarked. The sentences in (85) illustrate unmarked clausal modifiers. In the Makalero example, it is a complete intransitive clause with its own S argument (Mkl ki-kor roual ‘its colors are many’), while in the Makasae example, it is a transitive verb with its P argument.

(85)

Makalero a. [ faru [ki-kor shirt 3.POSS -color ‘(a) colorful shirt’27 b.

roual]M O D ]N P be.many.NHUM

[dotor [ini war]M O D doctor 1PL . EXCL call ‘the doctor who called us’

ere]N P DEM

27 Mkl roual ‘many’ is restricted to non-human referents. For humans, Mkl rial ‘many’ is used. Note that in both languages, the lower numerals distinguish human from non-human forms (section 6.2).

Makalero and Makasae

313

Relative clauses are clausal modifiers which are overtly marked by the clitic marker Mkl =wa in Makalero, and the cognate particle Mks (w)aʔa in Makasae. Marking is centered on the subject, whether it is the relative clause’s head or not. In cases where the relative clause head is the subject, it is the locus of marking, as in (86). In a relative clause on a non-subject, the relative marker follows the subject, as in (87). A possible explanation for this typologically unusual marking pattern is that the relative marker is a second-position element; if the first position is empty, as in the case of a subject relative clause, it attaches to the immediately preceding element, i.e. the relative clause head. In relative clauses on non-subjects, the relative marker may occur not only on the relative clause subject, but can additionally be used after the head, as illustrated in (87). Note, however, that only the marker attached to the relative clause subject is obligatory. In instances of double relative marking, it appears to be good style in Makasae to not use the same form of the relative marker twice. The positions of the waʔa and the aʔa variants seem to be free. (86)

(87)

a.

Makalero tufuraa=wa ei soʔot ere woman=REL 2SG want DEM ‘the woman who likes you’

b.

Makasae tupurae aʔa ai karak woman REL 2SG want ‘the woman who likes you’

ere DEM

a.

Makalero ropa(=wa) ei=wa kutu~kutu=ee clothes=REL 2SG = REL RDP ~wear=DEF ‘the clothes which you were wearing’

b.

Makasae kaalsa (waʔa) ani aʔa gutu 1SG REL wear trousers REL ‘the trousers which I was wearing’

ere DEM

In a Makasae relative clause, if a non-head relative clause subject is identical with the subject of the matrix clause, it can be omitted, as (88) shows. This results in a transitive relative clause in which neither of the two participants is overtly expressed.

314

(88)

Juliette Huber

Makasae Ani [buku u [Ø waʔa watuʔ-isi Ø haga]M OD ere]N P tehu. REL past-be.at search DEM buy 1SG book one ‘I bought this one book which I have been looking for for a long time.’

In cases where the relative clause includes an adverbial in Makalero (see section 3.5), it is an option to cliticize the relative marker Mkl =wa to that adverbial, in addition to the marking on the subject argument, as shown in (89). This pattern is possible with both relative clauses on subjects and on objects, and supports the analysis of the relative marker as a second-position clitic in Makalero. It is not found in the Makasae sources. Makalero hau (89) . . . [ filem [aireʔ=wa ini=wa nonton=ana]M OD ere]N P hai DEM INITIAL FINAL movie now=REL 1PL . EXCL = REL watch=INT haiʔ. . . be.over ‘. . . the movie that we were going to watch earlier was already finished. . .’

4.2 Determiners The last slot in the noun phrase is that of the determiner. Long and complex noun phrases tend to have their determiner slot filled, likely to signal the end of the phrase. The determiner slot can contain one element at a time; possible fillers are demonstratives, the quantifier Mkl hoʔo / Mks oho ‘some’ and the definite marker Mkl/Mks =ee. Makalero has a series of five, Makasae of seven demonstratives as shown in Table 16. The analysis of the Makasae demonstratives is from Brotherson (2003: 164f.). All of these deictics also exist in a verbal form, characterized by the unproductive glottal stop verbalizer (see section 2.6.6), given in Table 16 for completeness. Makalero and Makasae are alone among the Timor and Kisar languages in marking elevation in their demonstratives, though this is also found in the languages of West Alor, Pantar and the Pantar Strait (see e.g. Holton 2014 on Western Pantar, Steinhauer 2014 on Blagar). The Makasae demonstrative system is more complex than its Makalero counterpart in two respects: in the demonstratives for higher and lower elevation, Makasae encodes a distinction in horizontal distance which is absent from Makalero. Furthermore, the Makasae demonstratives may be prefixed by a lV- element whose function, though not entirely clear, may be related to the addressee’s

315

Makalero and Makasae

knowledge (see section 2.6.4). A similar distinction, related to visibility, is described in Holton (2014) for Western Pantar and Kratochvíl (2014) for Sawila. Table 16: Demonstratives Makalero

Makasae

Elevation

Distance

Determiner

Deictic verb

Determiner

Deictic verb

Level

Near Spkr Near Addr Distant from Spkr and Addr

ere were umere

eʔ weʔ umeʔ

(le-)ere (le-)were (lo-)wori

eʔe weʔe woʔi

High

Horizontal distance No horizontal distance

udere

udeʔ

(le-)deri (lo-)dore

deʔi doʔo

Low

Horizontal distance No horizontal distance

ufere

ufeʔ

(le-)heri (lo-)hore

heʔi hoʔo

These demonstratives are used as determiners not only with noun phrases headed by nouns, as in (86), but also with personal pronouns. The demonstratives as well as the quantifier Mkl hoʔo / Mks oho ‘some’ can constitute full noun phrases of their own, as shown in (90). (90) a. Makalero ale teuh. . . Ani [hoʔoD E T ]N P mei=ni some take=CONJ hulled.rice buy 1SG ‘I used some (of the money) to buy rice. . .’ b. Makasae i tabaku tehu. . . . . . ani [ohoD E T ]NP mini ma isera again OBL hulled.rice and tobacco buy 1SG some ‘. . . I used some (of the money) again to buy rice and tobacco. . .’ Unlike the demonstratives and the quantifier, the clitic definite marker Mkl/ Mks =ee seen in (91) is not a free element. However, it cannot co-occur with them. Therefore, despite its different morphosyntactic status, it is analyzed as a determiner. This element is not discussed in any of the Makasae sources consulted. Thus, while it is taken here to be an enclitic like its Makalero equivalent, its exact status remains to be determined. (91)

a.

Makalero Ini wata=ee k-afa-laʔa. 1PL . EXCL coconut=DEF 3.OBJ-away.from-go ‘We left the (aforementioned) coconuts behind.’

316

Juliette Huber

b.

Makasae Ini wata=ee naʔu gapa-laʔa. . . away.from-go 1PL . EXCL coconut=DEF just ‘We just left the (aforementioned) coconuts behind. . .’

The issue of definiteness and anaphoric reference in discourse requires further research. Of the demonstratives, only Mkl/Mks ere (near speaker) and were (near addressee) are used in discourse function. The precise nature of this use and how it differs from that of =ee is unclear.

4.3 Possession The possessor phrase can consist either of a possessive pronoun or a noun phrase with the possessive marker Mkl ki- / Mks gi-. Makasae and Makalero have a single set of possessive pronouns, illustrated in Table 17. Example (92) illustrates their use. Note that the possessive pronouns of the 2nd person singular, the 1st person plural inclusive and the 2nd person plural are formally identical to the free personal pronouns (section 5.1). In Makasae, this also holds for the possessive pronoun of the 3rd person singular.28 Table 17: Possessive pronouns

1SG . POSS 2SG . POSS 3SG . POSS 1PL . EXCL . POSS 1PL . INCL . POSS 2PL . POSS 3PL . POSS REFL . POSS

(92)

a.

Makalero

Makasae

asieikiisifiiikini-

asiaigiisifiiigini-

Makalero asi-pada 1SG . POSS -friend ‘my friend’

28 The Makasae possessive pronouns are analyzed as unbound (Huber 2008a). However, due to the fact that they can never take a main stress and are phonologically dependent on the NP head, they are here treated as bound forms, like their Makalero counterparts. This distinction is quite clear even in cases where the free pronouns and the possessive pronouns are formally identical.

Makalero and Makasae

b.

317

Makasae asi-bada 1SG . POSS -friend ‘my friend’

The possessive marker Mkl ki- / Mks gi- is originally the third person possessive pronoun, but is used with noun phrases in the function of a generalized possessive marker. (93) shows examples of its use with possessor NPs. (93)

a.

Makalero elehaa were ki-nei old.man DEM 3.POSS -name ‘that old man’s name’

b.

Makasae tupurae ere gi-nai woman DEM 3.POSS -name ‘this woman’s name’

Makalero and Makasae differ from most TAP languages in not having separate pronominal paradigms for alienable and inalienable possession. Nonetheless, Makalero distinguishes between two classes of nouns with respect to possession, namely obligatorily possessed nouns and non-obligatorily possessed nouns. Non-obligatorily possessed nouns can freely occur with or without a possessor expressed, as shown with Makalero examples in (94). In terms of their semantics, they correspond to alienable nouns. Obligatorily possessed nouns, on the other hand, must either occur with a possessive marker or as part of a compound, as illustrated in (95); they cannot be used on their own. This class includes part terms (i.e., body parts of animates and constituent parts of inanimates), kinship terms and days of the week, and thus corresponds to inalienably possessed nouns. There is no information on this distinction in Makasae.

(94)

Makalero a. Kiloo sefar=ee dog=DEF 3SG ‘He hit the dog.’ b.

...

pase. hit

ki-sefar potil=ee hai 3.POSS -dog bottle=DEF INITIAL ‘. . . his dog is looking into the jar. . .’

mutu-puna. . . inside-look

318

(95)

Juliette Huber

Makalero a. ki-fasu 3.POSS -skin ‘(its) skin, peel’

b.

muʔu fasu banana skin ‘banana peel’

4.4 Plural marking Makalero and Makasae can mark plural on nouns with the suffixes Mkl -laa and -raa and the clitic Mks =laa (spelled -la in Huber 2008a). Mkl -laa and Mks =laa derive from a free plural word Mkl/Mks laʔaneʔ(e). These plural words follow the determiner, suggesting that they stand outside the NP; hence no slot for the plural word is given in the NP template in (78). The Makalero suffixes only occur on nominal heads, as in (96a), while Mks =laa can be found both on nominal heads (96b) as well as, in some cases, other constituents of the NP if a nominal head is not overtly expressed (97).29 It thus looks more like a clitic than a suffix. (96)

(97)

a.

Makalero Namiraa-laa ni-lopu k-ue-laʔa. REFL . POSS -house 3.OBJ -around-go man-PL ‘The men surrounded their house.’

b.

Makasae Asukai=laa ani goe-robuk. 1SG around-stand.densely man=PL ‘The men surrounded me.’ (Huber 2008a: 14)

Makasae . . . gi naʔu wai [le-ereD E T =laa]NP only LE - DEM = PL 3SG just ‘. . . he bought these here only.’

heʔe be.here

nai only

tehu. buy

Plural marking is not obligatory in either language. It appears to be more frequent in Makasae, and pragmatically more marked in Makalero. The Mkl suffixes -laa and -raa are restricted to [+HUM ] nouns in spontaneous speech;30 Mkl -raa is used only with kinship terms, as shown in (98), while Mkl -laa appears with all other human referents, as seen in (96a). 29 The Makalero equivalent to the Makasae sentence in (97) would have to make use of the free plural word Mkl laʔaneʔ (Huber 2011: 210). 30 When specifically asked, speakers will also mark [-HUM ] and inanimate nouns with -laa.

Makalero and Makasae

319

Makalero (98) . . . hoʔo mei . . . ni-noko-raa meih=ee kini REFL . POSS -younger.sibling-PL two.HUM - DEF give.to.3 some take ‘. . . (I) gave some to my two younger siblings. . .’ In Makasae, this distinction is less clear-cut than in Makalero. Mks =laa, though most common with human nouns, is also found with [-HUM ] nouns as well as inanimates, as in (99), even in spontaneous speech. There is no special plural marker for kinship terms.31 (99)

Makasae Gi sari ma ni-arabau waka=laa 3SG fodder OBL REFL . POSS -buffalo cow=PL ‘He brought his cows fodder.’ (Huber 2008a: 14)

gau be.for

nake. take

The presence of plural suffixes/clitics appears to be a common trait of the Eastern Timor languages; Fataluku has three plural-marking clitics, =ere, =(a)ra and =afu. The presence of the bound plural markers distinguishes the Eastern Timor languages from the other TAP languages. The languages of Alor and Pantar (Western Pantar, Teiwa, Kamang, Abui and Wersing) use separate plural words, comparable to Mkl/Mks laʔaneʔ(e) (which could in fact be cognate to Abui loku, cf. Schapper, Huber & van Engelenhoven 2014: 108–109 for the correspondence between Mkl/Mks /ʔ/ and pAP *k). Mkl/Mks laʔaneʔ(e) is not, however, cognate with the reconstructed pAP form *non (Klamer, Schapper & Corbett 2014). Whereas these plural words are part of the NP in most AP languages, Makalero is comparable to Kamang, where the plural word nung is also found outside the NP (Klamer, Schapper & Corbett 2014: 387f.). In Makalero, there is furthermore an associative plural marker -ara (100). This is found rather infrequently and occurs in my Makalero data with personal names only. Note that the general Mkl plural suffix -laa can also be used in these contexts, as shown in (101). None of the available Makasae sources mention an associative plural marker. (100)

Makalero . . . Leni-ara naʔamuni Agustu mi-naser. . . Agustu along-stand.PL Leni-ASSOC in.turn ‘. . . Leni and them stood alongside Agusto. . .’

31 In both languages, there is a lexically restricted plural suffix associated with the noun Mkl/ Mks mata ‘child’: Mkl -niki and Mks -kini. This suffix is not used elsewhere. It can be combined with the more productive plural markers; for instance, Mkl mata ‘child’ can be pluralized as Mkl mata-niki, mata-niki-laa, or mata-niki-raa (see Huber 2011: 118 for even more combinations, and Huber 2008a: 12 for the combinations found in Makasae).

320

(101)

Juliette Huber

Makalero Sabina-laa hai tuʔ. Sabina-PL INITIAL be.first ‘Sabina and them went ahead.’

Plurality can also be expressed through the use of numerals or other quantifiers (see section 6.2 on numerals), or a plural verb form in the case of a small number of verbs exhibiting suppletive subject number agreement (see section 6.1). In both cases, a plural marker may additionally appear on the head noun, though this is not obligatory.

5 Pronouns and argument marking Pronouns can be used to head noun phrases of their own or as adnominal modifiers, in which case they are interpreted as possessive. They include personal pronouns (section 5.1), the reflexive pronoun ni (section 5.2), the reciprocal pronoun ta (section 5.3), and a subset of interrogatives (section 5.4). Section 5.5 discusses strategies for making syntactic relations explicit.

5.1 Personal pronouns Table 18 shows the Mkl/ Mks independent personal pronouns. The corresponding adnominal (possessive) forms are set out in Table 17 (section 4.3). Table 18: Personal pronouns32

1SG 2SG 3SG 1PL . EXCL 1PL . INCL 2PL 3PL

Makalero

Makasae

ani ei kiloo ini fi ii kilooraa, werlaa

ani ai gi ini fi ii era, mai

32 The form Mks mai for the third person plural appears to be characteristic of the Laga dialect of Makasae (Correia 2011: 99). It is not found in my data from the Ossu dialect.

Makalero and Makasae

321

The third person pronouns, in particular in the plural, have distinct nominal qualities, such as the use of the nominal plural markers -laa and -raa in Makalero. Furthermore, the free forms of the third person plural pronouns precede Mkl ki- / Mks gi- in a possessive construction. The result is a structure parallel to that found with nominal possessors (see section 4.3). The third person singular pronoun Mkl kiloo looks like a compound of the original third person pronoun Mkl ki (compare Mks gi) and an unknown, possibly lexical morpheme loo (see section 2.2 on long vowels). The free pronouns are used invariably as S/A and P arguments, as illustrated in (102) through (104). In contrast, Fataluku has distinct subject and object pronoun forms (van Engelenhoven forthcoming), and the languages of Alor and Pantar are characterized by multiple pronominal sets for different functions. The pronominal systems of Makalero and Makasae are thus the least complex of all TAP languages. (102)

a.

Makalero kilooP tuku. AniA 3SG punch 1SG ‘I punched him.’

b.

Makasae AniA giP tuku. 1SG 3SG punch ‘I punched him.’

(103)

a.

Makalero masan. KilooA aniP 1SG be.afraid 3SG ‘He is afraid of me.’

b.

Makasae GiA aniP agaʔ. 3SG 1SG be.afraid ‘He is afraid of me.’

(104)

a.

Makalero KilooS naʔu tepa just constant 3SG ‘He kept refusing.’

b.

Makasae GiS tula maʔu. 3SG refuse come ‘He doesn’t want to come.’

tule. refuse

5.2 The reflexive pronoun Mkl/Mks ni, the reflexive pronoun, is illustrated as noun phrase head and in adnominal function in (105) and (106). It signals co-reference with the A argument and can thus only function as a P argument (see section 5.5.2). (105)

a.

Makalero Ani weir-isi-laʔa=ni ni-waroh. 1SG river-at-go=CONJ REFL-bathe ‘I went to the river to wash (myself).’

322

(106)

Juliette Huber

b.

Makasae Ani laʔa wair-mutu ni waruʔ. river-be.inside REFL bathe 1SG go ‘I went to wash (myself) in the river.’

a.

Makalero Ani ni-lopu-isi-laʔa. 1SG REFL . POSS -house-at-go ‘I went to (my) home.’

b.

Makasae Ani ni-oma-isi-laʔa. 1SG REFL . POSS -house-at-go ‘I went to (my) home.’

As discussed in section 3.3.1, Mkl ni, if heading a NP of its own, is expressed in the INCORP P slot, whereas Mks ni stands in the more common object position.

5.3 The reciprocal pronoun Mkl/Mks ta, the reciprocal pronoun, differs from the other pronouns presented here in that it has only been found heading a NP of its own; no adnominal use is attested in either the Makalero or Makasae corpora. Mkl ta shares the property of standing in the INCORP P slot with Mkl ni. In at least the Ossu dialect, Mks ta is also used in this position (although Mks ni is not). The sentences in (48), repeated here as (107), show its position relative to the initial boundary marker Mkl hai / Mks ai. (107)

a.

Makalero Lopu-laa ere hai taI NC O R P -mi-naser. house-PL DEM INITIAL RECP-along-stand.PL ‘These houses stand alongside one another.’

b.

Makasae mara. Anu mahe ai taI N C O R P -mi-e-da person two.HUM INITIAL RECP-along-here-pass go ‘The two are walking one after another.’ (Brotherson 2003: 77)

Like Mkl/Mks ni, Mkl/Mks ta signals co-reference with the A argument and is thus found only as a P argument.

Makalero and Makasae

323

5.4 Interrogative pronouns The Mkl interrogatives muʔa ‘who’ and saʔa ‘what’ as well as Mks naʔi ‘what’ have the same distribution as the pronouns discussed in the preceding sections, namely that they can be used either heading NPs of their own, or adnominally as possessors, as shown in (108) and (109). Note that Mkl muʔa ‘who’ and saʔa ‘what’ are obligatorily used in the forms muʔani ‘who’ and saʔani ‘what’, respectively, if used as A/S arguments; the same is probably true for Mks naʔi ‘what’ and the contrastive marker Mks =ini/ene. The status of Mks nahireʔe ‘who’ is not quite clear. Other interrogatives such as Mkl tauʔ ‘where’ or Mkl tetepaneʔ ‘when’, and Mks naʔi mutu ‘where’ or Mks naʔi gau ‘why’ are analyzed as verbal (see section 3.6.2). (108)

Makalero a. . . . muʔani asi-minakuda who.SBJ 1SG . POSS -kuda.oil ‘. . . who took my kuda oil . . .?’ b.

(109)

hau FINAL

mei. . . ? take

Ere

muʔa-mata? DEM who-child ‘Whose child is this?’

Makasae a. Naʔi=ene mata wori gi-iti buli what=CONTR child DEM 3.POSS -foot digit ‘What bit that child’s toe?’ (Correia 2011: 141) b.

kiʔi? bite

Naʔi-karita=ene gali-toku? what-car=CONTR back-overturn ‘What car overturned?’ (Correia 2011: 142)

5.5 Argument marking NPs are not usually marked for grammatical roles in Makalero and Makasae. Given that arguments can readily be omitted (section 3), the interpretation of a given NP as either an A or a P argument is not always straightforward. Though they are not obligatory, there are elements helping to determine the correct interpretation. Argument marking focuses on the A argument (section 5.5.1). The P argument is discussed in section 5.5.2. 5.5.1 A arguments In a canonical transitive clause, the A argument precedes the P argument and the predicate (see section 3). In contexts where word order is not sufficient to

324

Juliette Huber

indicate a noun phrase’s grammatical role, e.g. when a participant is omitted or left-dislocated, Mkl =(i)ni / Mks =(i)ni ~ =(e)ne, a contrastive marker, is an important means of marking the A argument, as shown in (110). Its use is not limited to, but strongly associated with A/S arguments.33 Based on the use of this marker in sentences with non-canonical constituent order in Makalero, a topic-worthiness hierarchy for personal pronouns can be deduced in which first person pronouns are ranked highest, followed by second and then third person pronouns. Mkl =(i)ni / Mks =(i)ni ~ =(e)ne are also used to mark the end of long and complicated NPs. (110)

a.

Makalero Ki-upa=ni pase. 3.POSS -father=CONTR beat ‘(It was) his father (who) beat (him).’

b.

Makasae Gi-baba=ni base. 3.POSS -father=CONTR beat ‘(It was) his father (who) beat (him).’

Mkl =(i)ni is also a conjunction of very general meaning; thus (110) is reminiscent of a cleft construction. Mks =(i)ni ~ =(e)ne is not used as a conjunction, but only as a NP marker. 5.5.2 P arguments Makalero has no marker dedicated to P arguments. In cases where confusion is possible, the A argument is the one that is marked. Some Makasae dialects, on the other hand, interchangeably use the conjunctions Mks do and (d)ete as contrastive P argument markers (Correia 2011: 46), as seen in (111).34 (111)

Makasae a. Fara mata dete inimigu laʔaneʔe ataka? attack ship child CONTR enemy PL ‘Was it a boat which was attacked by the enemy?’ (Correia 2011: 46)

33 According to Correia (2011: 46), Mks =ene is used exclusively as a contrastive subject marker in the Laga dialect of Makasae. 34 In my own data from the Ossu dialect, Mks (d)ete is not used as a contrastive marker, but as a sequential conjunction only. There is, however, a sequential conjunction Mks to in the Ossu dialect (presumably cognate with Correia’s do from the Laga dialect of Makasae), which can function both as a conjunction and a contrastive marker. In the Ossu dialect, Mks to is used as a contrastive marker with A/S arguments only (see Huber 2008a: 83f. for examples).

Makalero and Makasae

b.

325

Kotemaa do Soru kuikui. Kotemaa CONTR Soru tickle ‘It is Kotemaa that Soru tickled.’ (Correia 2011: 46)

Furthermore, Mkl/Mks ni-, when used as a reflexive possessive (section 5.2), often functions as a means to disambiguate the reading of a sentence. Being co-referent with a preceding A argument, an NP containing Mkl/Mks ni- can only be interpreted as a P argument. As shown in (112) and (113), Mkl/Mks nithus contrasts with Mkl ki- / Mks gi-, which may be used in all argument roles.

(112)

Makalero a. Tufuraa ere=ini ni-nami naʔu tepa constant woman DEM = CONTR REFL . POSS -man just ‘This is the woman who often beats her husband.’ b.

(113)

Tufuraa ere=ini ki-nami naʔu tepa pase. constant beat woman DEM = CONTR 3.POSS -man just ‘This is the woman who is often beaten by her husband.’ (or: ‘This is the woman who often beats somebody else’s husband.’)

Makasae a. Tupurae ere=ini ni-asukai gahana woman DEM = CONTR REFL . POSS -man constant ‘This is the woman who often beats her husband.’ b.

pase. beat

base. beat

Tupurae ere=ini gi-asukai gahana base. woman DEM = CONTR 3.POSS -man constant beat ‘This is the woman who is often beaten by her husband.’ (or: ‘This is the woman who often beats somebody else’s husband.’)

In both languages, personal pronouns are prototypically used to express agents. The combination of a personal pronoun with the demonstrative Mkl/ Mks ere is associated with non-agentive participant roles, as shown in (114). Again, however, this correlation is not absolute. (114) a. Makalero b. Makasae Ani ere ei-mata. Ani ere ai-anu laʔida. 1SG DEM 2SG . POSS -person servant 1SG DEM 2SG . POSS -child ‘I am your child.’ ‘I am your servant.’

326

Juliette Huber

6 Agreement and verbal morphology Makalero and Makasae are the most isolating of the TAP languages, and there are only few and unproductive cases of agreement. A group of verbs agree with their subject in number (section 6.1), and the lower numerals have special [+HUM ] forms (section 6.2). Furthermore, in Makalero, a group of verbs take pronominal object prefixes (section 6.3).

6.1 Suppletive subject number agreement on verbs Most verbs in Makalero and Makasae are invariable. There is, however, a small class of intransitive verbs which use suppletive forms for singular and plural subjects, shown in Table 19.35 Table 19: Verbs inflecting for singular and plural subjects Makalero

‘sit’ ‘stand’ ‘run’ ‘sleep’ ‘big’

Makasae

Singular

Plural

Singular

Plural

mit nat riaʔ tia ~ -dia pere

diar naser titar rou helar

mi na riaʔ

diar nahar ditar

At the formal level, all of the plural forms end in -ar or -er, with the exception of Mkl rou ‘lie, sleep (PL)’.36 The majority of these inflectional forms are suppletive, showing no direct morphological relation of the plural form to the singular form. An example of the use of such verbs is given in (115) and (116).

35 Brotherson (2003: 29) lists two additional pairs for Makasae: daro and doen ‘sit’ (presumably ‘be positioned’), and wou and rai ‘lie’. These forms do not occur in my own field notes, and no examples are found in Brotherson (2003). A further singular-plural pair mentioned in both Brotherson (2003: 29) and Carr (2004: 19) is Mks lolo and lolini ‘say, talk’. Again, this is not found in my field notes. 36 This form is somewhat special in another respect, too. The singular form, Mkl tia ‘sleep’, can be used both as a positional verb, in which case it is obligatorily constructed with a locative adjunct in the INCORP LOC slot (section 3.3.2), or as an activity verb, where it is free to occur without a location. Mkl rou, the plural form, on the other hand, only has the positional reading, and a locative phrase must occur with it.

Makalero and Makasae

327

(115) a. Makalero Kiloo hai lopu sofu-isi-ume-nat. INITIAL house back-at-there-stand.SG 3SG ‘He is standing over there behind the house.’ b. Makasae Ani oma panukai-isi-e-na. 1SG house face-at-here-stand.SG ‘I am standing right in front of the house.’ (Brotherson 2003: 98) (116) a. Makalero Wata=ee lopu=ee k-ue-naser. coconut=DEF house=DEF 3.OBJ-around-stand.PL ‘The coconut trees stand around the house.’ b. Makasae Ate anu mahe gi-naruta-isi-hei-nahar isi. tree person two.HUM 3.POSS -back-at-down.there-stand.PL go.down ‘There are trees down behind the two of them.’ (Brotherson 2003: 61) Suppletive singular and plural forms are also found for a partially overlapping set of verbs in Fataluku (‘sit’, ‘stand’, ‘stand upright’, ‘extend’, ‘lay’, ‘hang’), but nowhere else in the family.

6.2 [+ HUM ] agreement on numerals Numerals and other quantifiers are analyzed as stative verbs.37 Table 20 shows that some of the lower numerals exhibit special forms for the quantification of human referents.38 In Makalero, this applies to the numerals ‘two’ and ‘three’ only, while in Makasae, it is the numerals from ‘two’ to ‘four’. In addition, Makalero distinguishes a human and a non-human ‘many’ quantifier: Mkl rial [+HUM ] and Mkl roual [‒HUM ]. The corresponding Mks quantifier baun ‘many’ is used for both human and non-human referents. 37 The only exception is Mkl hoʔo / Mks oho ‘some’, which is syntactically a determiner (see section 4.2). 38 There has been extensive borrowing from Austronesian languages in the numeral system. Of the numerals displayed in Table 20, Mkl/Mks lima ‘five’, fitu/pitu ‘seven’ and siwa ‘nine’ are reflexes of PAN *lima, *pitu and *siwa; furthermore, Mkl fat ‘four’ corresponds to PAN *Sepat. The last syllable of Mks loloha ‘four’ could possibly also be a reflex of this form (see Huber 2011: 179 for more information on Austronesian loans in the Makalero numerals).

328

Juliette Huber

Table 20: [+ HUM ] agreement on numerals Makalero Non-human 1 2 3 4 5 6 7 8 9 10 ‘many’

Makasae Human

Non-human

meih itu

lolaʔe lolitu loloha

u(n) loloi lolitu

u

fat lima douh fitu afo siwa ruru / ru roual

Human

mahe mitu pae lima daho pitu apo siwa ruru (u) baun

rial

The use of the [+HUM ] and [‒HUM ] forms in attributive function is shown in (117) and (118). Example (119) shows that the distinction is also made if the numerals are used predicatively in Makalero. I have not found an example of a predicative use in Makasae. (117) a. Makalero kareta loloi car two.NHUM ‘two cars’ (118) a. Makalero namiraa meih man two.HUM ‘two men’

b. Makasae kareta lolaʔe car two.NHUM ‘two cars’ b. Makasae asukai anu mahe man CLF two.HUM ‘two men’

Makalero (119) a. Asi-dada-raa ere meih=po. . . 1SG . POSS -grandfather-PL DEM two.HUM =but ‘My grandfathers were only two, but. . .’ b. . . . ni-pura

kini=oo nomo roual, tenki un loloi. give.to.3=too NEG much must one two ‘. . . giving them the bride price too, it doesn’t have to be many (animals), (but) there have to be one or two.’ REFL . POSS -bride.price

Makalero and Makasae

329

The [+HUM ] agreement on lower numerals appears to be an innovation exclusive to Makalero and Makasae. Note, however, that the numeral ec- ‘two’ in Fataluku may be combined with the [+HUM ] plural marker =afu into ec=afu ‘the two (humans)’.

6.3 Object marking on verbs in Makalero Makalero has a special class of transitive verbs which mark P arguments with pronominal prefixes. A few examples are given in Table 21 (see Huber 2011: 163– 164 for an exhaustive list). The majority of the Makasae cognates involve an initial g-, which is a fossilized 3rd person prefix; there are a few only which are vowel-initial (cf. Schapper, Huber & van Engelenhoven 2012: 214). Note that the Mks forms corresponding to Mkl (k-)afu ‘carry’ have undergone semantic differentiation: Mks apu, without the 3rd person prefix, translates as ‘carry’ (especially a child), while Mks gapu is a stative verb ‘to be with’. The velar 3rd person prefix is reconstructable to proto-TAP (Schapper, Huber & van Engelenhoven 2014: 106). It is a remnant of more extensive verbal person-number inflection, which is still present productively in Bunaq and Alor-Pantar languages, but entirely stripped from Fataluku. Table 21: Object-marking verbs

‘kill’ ‘cover, wear’ ‘be in contact, into, onto’ . . . ‘to death’ ‘carry’ ‘do’ ‘be on top’ ‘be for’

Makalero

Makasae

(k-)uta (k-)utu (k-)ataʔ (k-)umu(k-)afu (k-)ini (k-)uaʔ (k-)asu

guta gutu gata umu apu, gapu gini gua gau

The pronominal prefixes in Makalero correspond to the free personal pronouns (section 5). The only exception is the third person, where a k-prefix is employed in both singular and plural instead of the full pronouns Mkl kiloo and kilooraa.39 The prefix further differs from the other forms in the paradigm in that it can cooccur with the full pronoun. Unlike the other prefixes, it is thus not a pronominal prefix, but an agreement marker (see below). Table 22 shows a sample paradigm. 39 Note that Mkl kiloo is a complex form; the comparison with Makasae suggests that the original pronoun is ki. See section 5.1.

330

Juliette Huber

Table 22: Paradigm of Makalero (k-)uta ‘kill’

1SG 2SG 3SG 3PL 1PL . INCL 1PL . EXCL 2PL

Pronoun

Inflected -uta ‘kill’

ani ei kiloo kilooraa fi ini ii

ani-uta ‘kill me’ ei-uta ‘kill you (SG )’ k-uta ‘kill him, them’ fi-uta ‘kill us (INCL )’ ini-uta ‘kill us (EXCL )’ ii-uta ‘kill you (PL )’

The pronominal prefixes stand in the INCORP P slot illustrated in section 3.3.1, as seen in (120), where the first person plural inclusive pronoun follows the Mkl aspect marker hai. In all other cases, personal pronouns stand in the more common object position illustrated in (36) (section 3.1.2). (120) shows that the Makasae cognates behave like regular transitive verbs – they are invariable and a pronominal object stands in the object position. (120)

a.

b.

Makalero Amuni ki=selu hai person ATTR =other INITIAL ‘Someone else killed us.’

fi-uta. 1PL . INCL-kill

Makasae Ana pi saʔi guta. person 1PL . INCL be.over kill ‘People could kill us (dead).’

Referential nominal arguments appear in the object position and are crossreferenced on object-marking verbs in Makalero with the k-prefix, as seen in (121). Non-referential nominal arguments, on the other hand, behave like personal pronouns: they stand in the INCORP P slot and are combined with the unprefixed form of the verb, as seen in (122).40 Makalero hai k-asu lolo. . . (121) . . . ani hai maʔu papa=eeP 1SG INITIAL come Indonesian=DEF INITIAL 3.OBJ-be.for say ‘. . . I came and said to the Indonesian. . .’ 40 The distinction between the k-prefixed and the vowel-initial forms appears to be undergoing reanalysis in terms of a distinction between free and bound forms with initial consonant mutations (section 2.6.1; see Huber 2011: 356 for details).

Makalero and Makasae

331

Makalero (122) . . . ini ni-asu lopu-ini. . . 1PL . EXCL REFL-be.for house-make ‘. . . we build a house for ourselves. . .’

7 Serial verb constructions Eastern Timor languages show a much more restricted use of serialization than Bunaq and the Alor-Pantar languages. Most serializations in Makalero and Makasae involve the partially grammaticalized ‘take’ verbs, Mkl mei and Mks ma, discussed in section 7.2. As in the Alor-Pantar languages (see the sketches in Volume 1), the main function of the ‘take’ verbs in these constructions is to introduce an additional participant. ‘Take’ serializations are more pervasive in Makalero and Makasae because they are also used in contexts where locative and manner adjuncts as well as caused actions are expressed in the INCORP L OC slot (see section 3.3.2). The only serializations in Makalero and Makasae which do not involve the ‘take’ verbs are directional serializations (section 7.1); these are equivalent to directional serializations found in Bunaq and the AP languages. However, there are no direct equivalents in Makalero and Makasae to the aspectual, causative and manner serializations which are ubiquitous in Bunaq and the Alor-Pantar languages. In my previous work on Makalero, I have avoided the term ‘serial verb construction’ when referring to the multi-predicate constructions discussed in the following sections. The reason for this is the fact that it appears to be possible in all cases to use a semantically general conjunction, Mkl =ini, to join the verbs together, with no apparent difference in meaning or function (Huber 2011: 203, 402). Prototypical serial verb constructions do not allow “any overt marker[s] of coordination, subordination, or syntactic dependency of any sort” (Aikhenvald 2006: 1).41 On the other hand, the Makalero multi-predicate constructions behave like typical serial verb constructions in that the verbs cannot be independently marked for aspect or negation. The properties and status of these constructions require further research for both languages, but particularly for Makasae.

41 This criterion is emphasized in most descriptions of SVCs. However, there are languages which have a specialized SVC marker: for instance, in Kove, an Austronesian language spoken on New Britain, the marker gha obligatorily links verbs in constructions which look like typical SVCs in all other respects. While Kove gha is a specialized serial verb unifier (cf. Sato 2013: 413f.), Mkl =ini is more problematic because it is also a conjunction. Its semantics are very general, and it is associated with subject continuity. It is the conjunction most commonly used in the narration of a sequence of events (see Huber 2011: 457f.).

332

Juliette Huber

7.1 Directional serialization In directional serialization, a V1 implying undirected motion is combined with a directed motion verb, as shown in (123).42 The conjunction Mkl =ini is not common in this type of serialization, although it is found in some examples (124). The five basic directed motion verbs of Makalero and Makasae are shown in Table 23. Table 23: Basic directed motion verbs

Movement towards speaker, equal elevation Movement away from deictic center, equal elevation Movement towards deictic center other than speaker, equal elevation Movement downwards Movement upwards

(123)

a.

b.

(124)

Makalero

Makasae

maʔu mara laʔa isa misa

maʔu mara laʔa isi misa

Makalero Pipirusa=ee hai toil=ee k-ou-riaʔ INITIAL slope=DEF 3.OBJ -towards-run.SG deer=DEF ‘The deer runs downwards to the slope.’

isa. go.down

Makasae Kona ate gata-e-ria isi. monkey tree near-here-run.SG go.down ‘The monkey runs down the tree.’ (Brotherson 2003: 58)

Makalero . . . ei=ni. . . 2SG = CONTR asar=ini send=CONJ

ani-asu 1SG -for

sera~seran RDP ~thing

natal Christmas

k-isiʔ 3.OBJ-belong

maʔu. come

‘. . . it was you (. . .) who sent me a Christmas present over here.’ Mkl/Mks mara ‘go’ as a V2 may also indicate an area of dispersion around a deictic center (125). 42 An alternative analysis of these constructions is as hierarchical, as indeed described in Huber (2011: 345). In this version, the first verb is in the I NCORP LOC slot of the final verb. There is at this point no syntactic evidence to distinguish between these alternative analyses.

Makalero and Makasae

(125)

a.

Makalero . . . ate asa=ni ira-mutu-we-lafuʔ mara. tree leaf=CONTR water-inside-there-live go ‘. . . there were leaves living in the water around there.’

b.

Makasae Mata arai e-nahar mara. . . child small here-stand.PL go ‘Small children are standing around. . .’ (Brotherson 2003: 61)

333

7.2 Serialization with ‘take’ verbs All other serializations in Makalero and Makasae use the partially grammaticalized ‘take’ verbs, Mkl mei and Mks ma. These ‘take’ verbs introduce an additional participant, such as in the case of benefactive (section 7.2.1) and instrumental serializations (section 7.2.2). The ‘take’ verbs also appear in constructions where location and manner adjuncts and caused actions are expressed in the INCORP L O C slot (sections 7.2.3 and 7.2.4, respectively). As discussed in section 3.3.2, the INCORP L O C slot is associated with reduced transitivity, and its presence causes a transitive verb’s logical object to be expressed as an argument to a ‘take’ verb. The ‘take’ verbs are furthermore used to express caused position (section 7.2.5). Finally, section 7.2.6 discusses the differences between Mkl mei ‘take’ and its counterpart, Mks ma. While Mkl mei can also be used as a full verb, this is not the case for Mks ma, which is synchronically analyzed as an oblique marker.

7.2.1 Benefactive serialization A monopredicative clause in Makalero and Makasae allows for the expression of two arguments at most (see section 3). States of affairs which logically involve more than two participants must be expressed in a complex construction with Mkl mei / Mks ma, as shown in the sentences in (126), which involve an agent, a theme, and a recipient. The participant which is higher on the animacy hierarchy – generally the recipient – occupies the object position of the final and semantically main verb, while the other stands with Mkl mei / Mks ma. A reversal of this order is taken to be significant. That is, the Makalero sentence in (127) cannot be understood to be a variant expression of (126). Instead, the erstwhile recipient is interpreted as

334

Juliette Huber

an instrument in (127). If one of the two non-subject participants remains unexpressed, the other can be expressed as the main verb’s argument, irrespective of its semantic role (Huber 2011: 404f.). (126)

a.

b.

(127)

Makalero Meestri were matematika mei=ni teacher DEM mathematics take=CONJ ‘That teacher teaches me maths.’ Makasae Guru were matematika ma teacher DEM mathematics OBL ‘That teacher teaches me maths.’

ani 1SG

Makalero Meestri were ani mei=ni matematika teacher DEM 1SG take=CONJ mathematics ‘That teacher uses me to teach maths.’ *‘That teacher teaches maths to me.’

ani 1SG

fana. teach

pana. teach

fana. teach

The most frequent ditransitive construction involves the verb Mkl -ini / Mks gini ‘give’. In Makalero, this verb presents a somewhat special case in that it consists of a paradigm of forms which incorporate the recipient, as shown in Table 24. Mks gini is invariable; the initial g- is a fossilized 3rd person agreement prefix (see section 6.3). Table 24: Pronominal recipients in Makalero -ini ‘give’ Recipient

Form of ‘give’

Personal pronoun

1SG 2SG 3SG /PL 1PL . EXCL 1PL . INCL 2PL

m-ani-ini m-ei-ini Ø-k-ini m-ini-ini Ø-fi-ini m-ii-ini

ani ei kiloo / kilooraa ini fi ii

The Makalero ‘give’ verb forms are contracted versions of multi-predicate clauses with the light verb Mkl mei ‘take’, which is reflected in the m-prefix found in most forms (Huber 2011: 406f.; Klamer & Schapper 2012). The T argument can be introduced in different ways: it may appear directly as the argument to the complex verb of giving, as in (128), or with Mkl mei ‘take’, as in (129). The

Makalero and Makasae

335

clitic conjunction =ini is optional in such sentences. The construction is analyzed as biclausal with the conjunction, but as monoclausal without it in Klamer & Schapper (2012). Note that when Mkl mei ‘take’ is used to flag the T argument, we find two reflexes of this light verb in the construction. A construction such as (128), without the light verb, on the other hand, can be argued to express all three participants in a single monopredicative clause.

(128)

(129)

Makalero . . . osan tepa Ø manini. . . money constant give.to.1SG ‘. . . (he) kept giving me money. . .’ Makalero . . . papa naʔu tepa apoiu mei=(ni) father just constant support take=CONJ ‘. . . keep giving us (your) support, father.’

minini. give.to.1PL . EXCL

In Makasae, the recipient is always constructed as an argument of Mks gini ‘give’. The T argument is flagged with the oblique marker Mks ma.

(130)

Makasae Isera ma ini hulled.rice OBL 1PL . EXCL ‘(They) gave us rice. . .’

gini. . . give

Mks ma is usually present even if the T argument is not overtly expressed, as shown in (131). While there are not many comparative cases for Makalero, it appears that Mkl mei is less obligatory in such contexts (132).

(131)

(132)

Makasae Paʔ-ini dawa erau Ø ma ani gini. OBL 1SG give sew-NMLZ COND PROH ‘If it has anything to do with sewing don’t give it to me.’ Makalero . . . ani-asu surat hai weʔ=isi Ø 1SG -for letter INITIAL be.there=CONJ ‘. . . there is a letter there for me, so give it to me.’

manin=ana. give.to.1SG = INT

336

Juliette Huber

A third person recipient may be expressed by Mks gi, the third person singular pronoun, or may be left unexpressed, as in (133). This suggests that, although the g-prefix is fossilized, the form is still associated with the third person. (133)

Makasae . . . ani surat ai ma (gi) gini. 1SG letter INITIAL OBL 3SG give ‘. . . I have already given him the letter.’

An interesting aspect of give-constructions in Makasae is that Mks ma can be used to replace a third person recipient argument, as in (134).43 Doubling of the ‘take’ element is also found in certain contexts in Fataluku (Klamer & Schapper 2012: 193), and is also reminiscent of the double reflexes of Mkl mei. (134)

Makasae a. . . . ani susuwehe to ma NEG OBL 1SG milk ‘. . . I don’t give him any milk.’

gi 3SG

gini. give

b.

ma

gini. give

...

ani susuwehe to ma NEG OBL 1SG milk ‘. . . I don’t give him any milk.’

OBL

Makalero and Makasae give-constructions are structurally similar to those found in the Alor-Pantar languages. Distinguishing features are that a) there are no bare T constructions (although Makalero sentences such as example 128 synchronically look like it) and b) that the ‘take’ element used to flag the T argument can be doubled in some contexts. This characteristic is found also in Fataluku, but is not shared in other TAP languages. 7.2.2 Instrumental serialization The expression of instruments is one of the core functions of Mkl mei / Mks ma. In (135), without the ‘take’ verb, Mkl saʔa ‘what’ and Mks naʔi ‘what’ are interpreted as undergoers; in (136), however, where they are syntactically 43 Klamer & Schapper (2012: 194–195) analyze the first ma in the sequence as a placeholder for the R argument, and the second as a VP element. In my view, it is more parsimonious to assume that the first ma flags the T argument, while the second is a placeholder for the non-overt R. Under this analysis, (134b) is parallel to (134a); Klamer & Schapper (2012), on the other hand, assume a rather different construction for (134b), where the R, rather than the T, appears to be flagged. The fact that only third person pronominal R arguments have been found to be replaced by ma may be connected to the fact that Mks gini is originally a third-person marked form.

Makalero and Makasae

337

the arguments of Mkl mei / Mks ma, the same pronouns are interpreted as instruments. (135)

a.

Makalero Ei saʔa-tina? 2SG what-cook ‘What are you cooking?’

b.

Makasae Ai naʔi tina? 2SG what cook ‘What are you cooking?’

(136)

a.

Makalero Ei saʔa-mei tina-ini? 2SG what-take cook-do ‘What are you cooking with?’

b.

Makasae Ai naʔi ma tina? 2SG what OBL cook ‘What are you cooking with?’

7.2.3 Location and manner adjuncts A pervasive type of construction involving serialization with Mkl mei / Mks ma is the expression of locative and manner adjuncts in the INCORP LOC slot of transitive verbs. The transitive verb’s logical object must in this case be expressed with Mkl mei / Mks ma as in (137) and (138). (137)

(138)

a.

Makalero Ani ni-pada nomo 1SG REFL . POSS -friend NEG ‘I did not meet my friend.’

b.

Makasae Ani gi noto rataʔ. 1SG 3SG NEG meet ‘I did not meet him.’

a.

Makalero Ani ni-pada hai 1SG REFL . POSS -friend INITIAL ‘I met my friend in Lospalos.’

b.

rataʔ. meet

mei=ni take=CONJ

[Lospalos-isi]L OC-rataʔ. Lospalos-at-meet

Makasae Ini tentara Indonesia ma [Larigutu-e]L OC-rataʔ. 1PL . EXCL soldier Indonesian OBL Larigutu-here-meet ‘We met Indonesian soldiers here in Larigutu.’

338

Juliette Huber

These sentences show that the INCORP LOC slot, like the INCORP P slot, is associated with reduced transitivity. This is clearly the case in Makalero, where the INCORP P slot is used to express reflexive and reciprocal pronouns as P arguments, as well as non-referential Ps; but less prominent for Makasae, where only the reciprocal pronoun is found in the INCORP P slot. The interaction of locatives and transitivity in Makalero is discussed at some length in Huber (2011), but there is no comprehensive treatment of this phenomenon in Makasae. In the examples shown above, the Makasae constructions parallel those of Makalero; however, there are also instances where the two languages differ. An example is given in (139). In the Makalero sentence in (139a), the logical object of Mkl dane ‘raise’ appears as an argument to the light verb mei ‘take’, due to the presence of the locational verb namaʔ ‘be above, upwards’ in the INCORP LOC slot. In the equivalent Makasae sentence in (139b), however, the object of Mks dane ‘raise’ is expressed directly, without the expected oblique marker ma. Likewise, the use of the bound verb forms in Makasae does not always parallel that of Makalero. In (140a), for instance, Mkl hifaʔ ‘catch’ is combined with a locative in the INCORP LOC slot and consequently appears in the bound form -sifaʔ. In the same syntactic context, however, Mks sipaʔ ‘catch’ appears in the free form rather than the bound form ‑dipaʔ. Clearly, the properties of the INCORP LOC slot in Makasae require more research. (139) a. Makalero Mata kaʔu ni-saapatu hai mei=ni nama-dane. child small REFL . POSS -shoe INITIAL take=CONJ upwards-raise ‘The child lifts up his shoe.’

(140)

b.

Makasae Mata ni-bisikeleta Ø nama-dane. upwards-raise child REFL . POSS -bike ‘The child takes up his bike.’ (Brotherson 2003: 189)

a.

Makalero . . . were

mei [taure-isi]L O C -sifaʔ? DEM take which-at-catch.BOUND ‘. . . where (did you) catch this?’ b.

Makasae . . . gi ma [nahigalu-isi]L O C-sipaʔ? 3SG OBL where-at-catch ‘. . . where (did you) catch him?’

Makalero and Makasae

339

Some manner expressions have also been found in the INCORP LOC slot (see section 3.3.2). The syntactic consequences of this are the same as with locatives, as seen in (141) and (142): a transitive verb’s logical object is expressed with Mkl mei / Mks ma.

(141)

Makalero a. Ibu=ee fi Indonesian.woman=DEF 1PL . INCL ‘The Indonesian lady beats us.’ b.

(142)

pase. beat

Ibu=ee fi mei rau-pase. Indonesian.woman=DEF 1PL . INCL take be.good-beat ‘The Indonesian lady beats us thoroughly.’

Makasae a. sirbisu ere gini DEM do work ‘do the work’ b.

sirbisu ere ma DEM OBL work ‘do the work well’

rau~rau-gini RDP ~be.good-do

The applicative-like expression of locative adjuncts in the INCORP LOC slot is also found in Fataluku, where these adjuncts interact with transitivity in the same way as described here for Makalero and Makasae. It thus appears that we are dealing with a common innovation in the Eastern Timor languages.

7.2.4 Causatives Makalero and Makasae make use of the verb ‘do’ to form causatives, as shown in (143) through (145). Mkl -nini ‘do’ (or a variant form -ini) is the head of the Makalero causative construction; Mkl hareʔ ‘be clean’ occupies this verb’s INCORP LOC slot. Mkl ki-lopu ‘her house’, the erstwhile S argument, is a P argument in the derived causative construction and, like the P arguments of transitive verbs with locative adjuncts (section 7.2.3), it is introduced with Mkl mei ‘take’. Thus, although a causative is semantically transitive, the syntactic structure used to express it in Makalero is intransitive. Two causative constructions have been found in Makasae. The sentence pair in (144), taken from the Laga dialect of Makasae,

340

Juliette Huber

shows a causative construction which is parallel to that illustrated for Makalero: in the b. sentence, a truncated form of the intransitive verb Mks bokol ‘be wet’ stands in the INCORP LOC slot of Mks gini ‘do’ (see section 2.6.2). The logical object of this causative construction appears with the oblique marker Mks ma. In contrast, the oblique marker is not used in the sentence from the Ossu dialect in (145). This suggests that a different construction is used here, where Mks sareʔ ‘be clean’ does not stand in the INCORP LOC slot of gini ‘do’. More research is needed to determine its structure and use. Causatives in Fataluku, the third of the Eastern Timor languages, are equivalent to those shown for Makalero and the Laga dialect of Makasae.

(143)

Makalero a. Ki-lopu hai hareʔ. 3.POSS -house INITIAL be.clean ‘Her house is (already) clean.’ b.

(144)

hare-nini. clean-do.BOUND

Makasae a. Gi-lesu ai bokol. 3.POSS -handkerchief INITIAL be.wet ‘His handkerchief is (already) wet.’ b.

(145)

Kiloo ni-lopu mei=ni REFL . POSS -house take=CONJ 3SG ‘She is cleaning the house.’

Mai ni-lesu ma boko-gini. . . 3PL REFL . POSS -handkerchief OBL wet-do ‘They dampen their (own) handkerchief. . .’ (Correia 2011: 231)

Makasae a. Gi-omu ai sareʔ. 3.POSS -house INITIAL be.clean ‘Her house is (already) clean.’ b.

Gi leu gata oma sareʔ gini. 3SG sing while house be.clean make ‘While singing, she cleans the house.’

7.2.5 Caused positions Caused position serializations involve a static positional verb (V2) and Mkl mei / Mks ma, which semantically adds a motion component. (146) and (147) give

Makalero and Makasae

341

examples. These serializations differ from the ‘take’ serializations discussed in the preceding sections in that they are “switch subject”; that is, the theme participant, which the two verbs share, is the P argument of the ‘take’ verb, but the A argument of the positional verb following it. As with the previously discussed multipredicate constructions, it is possible to use the conjunction Mkl =(i)ni on the first verb in the sequence in Makalero. (146) a. Makalero Kiloo ni-riku mei=ni ki-tana-isiʔ. 3SG REFL . POSS -riches take=CONJ 3.POSS -hand-be.at ‘He handed his riches over to her.’ (lit. He took his riches and put them in her hands) b. Makasae Era sitaʔa ma asi-tana-isi. 3SG machete OBL 1SG . POSS -hand-be.at ‘They handed me a machete.’ (lit. They took a machete and put it in my hands) (147) a. Makalero Ki-ifil fanu=ee . . . hai mei ni-faru lode-mutuʔ. . . 3.POSS -tongue face=DEF INITIAL take REFL . POSS -shirt pocket-be.inside ‘(He) put the tip of its tongue in his shirt pocket. . .’ (lit. ‘took the tip of its tongue and put it in his pocket’) b. Makasae Ani surat ere . . . ma ni-bolsu-mutu. . . 1SG letter DEM OBL REFL . POSS -pocket-be.inside ‘I put the letter in my pocket. . .’ (lit. ‘I took the letter and put it in my pocket’) 7.2.6 Makalero mei and the Makasae oblique marker ma Mkl mei, which is used as a light verb in the serializations discussed in sections 7.2.1 through 7.2.5, can function as a full verb ‘take’, as shown in (148).

(148)

Makalero . . . ini=ni nomo mei=oo mei. take=too take 1PL . EXCL = CONTR NEG ‘. . . even (if) it wasn’t us who took (it), (they said we) took (it).’

342

Juliette Huber

Its counterpart Mks ma is cognate with with Mkl mei; both items derive from proto-Eastern-Timor *mej ‘take’ (Schapper, Huber & van Engelenhoven 2012). However, Mks ma is not used as a full verb, but appears to have the purely grammatical function of introducing peripheral semantic roles. Reductions of ‘take’ verbs used in the expression of three-participant events are common in the TAP languages (cf. Klamer and Schapper 2012 as well as the sketches in Volumes 1 and 2 of this work). Despite not being used as a full verb, Mks ma exhibits a variety of verbal properties. (149a) shows that it can appear with a Ø object. This example parallels (149b), where a transitive verb’s P argument is extracted from its normal position. Also, negation and aspect markers precede Mks ma, as seen in the a. examples of (150) and (151). The b. examples show that these markers are found in the same position with verbs such as Mks nake ‘take’ and Mks gaʔel ‘bite’. Lastly, like its Makalero counterpart mei ‘take’ (see, e.g., 129 and 139a above), Mks ma can be followed by a sequential conjunction (152).

(149)

Makasae a. Erau

ani gini. 1SG give ‘Don’t give (it) to me.’

PROH

b.

(150)

ma

OBL

Tupurae ere=ini gi-asukai Ø gahana base. constant beat woman DEM = CONTR 3.POSS -man ‘This is the woman who is often beaten by her husband.’

Makasae a. Se lisensa to ma liurai seti dawa. . . if permission NEG OBL king ask COND ‘If (they) don’t ask for the king’s permission. . .’ b.

(151)

Ø

Ini mahe bukae to nake. . . 1PL . INCL two.HUM provisions NEG take ‘The two of us didn’t bring provisions. . .’

Makasae a. . . . ani surat ai ma gini. 1SG letter INITIAL OBL give ‘. . . I’ve given the letter (to him).’ b.

Depa=ini seu ere ai dog=CONTR meat DEM INITIAL ‘It was the dog who ate the meat.’

gaʔel. bite

Makalero and Makasae

343

(152) Mai hiʔa gi=rau ma do alunusu laʔaneʔe gau kina. for show 3PL way ATTR =be.good OBL CONJ student PL ‘They (teachers) set a good example for the students.’ (Correia 2011: 324)

8 Aspect marking Makalero and Makasae share the aspectual markers Mkl hai / Mks ai and Mkl hau / Mks au discussed in section 8.1. Section 8.2 introduces a number of enclitics expressing aspectual notions. Tense and mood are expressed lexically through adverbs (see section 3.5) or modal verbs and are not further discussed in this sketch.

8.1 Aspectual particles and adverbs The only real aspect marker in Makalero is Mkl hai. The Makasae cognate of the same meaning is Mks ai. Their position in the clause is shown in section 3. Mkl hai / Mks ai mark an aspectual category termed ‘initial boundary’ in Huber & Schapper (2014) which expresses inception of an event or entry into a state. The latter corresponds to the crossing of the final boundary of an event. With state verbs, Mkl hai / Mks ai provide the state with an initial boundary, as seen in (153). (153)

a.

Makalero Ani hai laʔit. 1SG INITIAL be.old ‘I am (already) old.’

b.

Makasae Gi ai laʔir. 3SG INITIAL be.old ‘She is already old.’

With activity verbs, the readings of Mkl hai / Mks ai are more varied. With unbounded activities, it is read as inchoative (154). In activities with a prominent end point, however, it selects the final boundary (155). (154)

a.

Makalero Waini Agusto hai isine~neʔet. . . thus Agusto INITIAL RDP ~think ‘So Agusto started to think. . .’

344

Juliette Huber

b.

(155)

Makasae Dai ere=oo ai mini gali stranger DEM =too INITIAL again return ‘The foreigner too started thinking again. . .’

a.

Makalero Sefar=ee hai laʔa tauʔ? be.where dog=DEF INITIAL go ‘Where is the dog?’ (lit. Where has the dog gone?)

b.

Makasae Depa ai laʔa nahigalu? INITIAL go be.where dog ‘Where is the dog?’ (lit., Where has the dog gone?)

ger. . . think

Mkl hai / Mks ai is the most commonly used marker in the narration of a series of events, as seen in the extracts from narrative texts in (156). The termination of each event in this series is seen as setting the stage for the next event to happen. (156) a. Makalero . . . daihael hai mei maʔu. Ani kiloo hai k-asu lolo. . . INITIAL take come 1SG 3SG INITIAL 3.OBJ -be.for say yarn ‘. . . (he) brought the yarn. I said to him. . .’ b. Makasae . . . asa apa gi-ina ere mega ai rataʔ. bird be.wild 3.POSS -mother DEM ever? INITIAL meet Wori asa ere ai gau lolo. . . DEM bird DEM INITIAL be.for say ‘. . . (he) met a wild hen. He said to the hen. . .’44 Mkl hai is clearly related to the full verb Mkl haiʔ ‘end, finish’. The relation between Mks ai and the full verb Mks saʔi is less obvious. Mkl hau / Mks au, the second marker, has two functions: with activity verbs, it marks the crossing of the final boundary, i.e. the completion of an event, as in (157). With stative verbs, on the other hand, it marks the state as holding completely, all over, or to a high degree (158). 44 From the mini-sketch of the Osorua dialect of Makasae, available at the time of writing at http://www.ling.hawaii.edu/ldtc/languages/makasae_osor/.

Makalero and Makasae

(157)

(158)

a.

Makalero Kilooraa hau nua mata-niki hai lolo. . . FINAL eat child-PL INITIAL say 3PL ‘(When) they had eaten, the children said. . .’

b.

Makasae Ini au nawa era maʔu ini sasa. . . 3PL come 1PL . EXCL check 1PL . EXCL FINAL eat ‘(When) we had eaten, they came and searched us. . .’

a.

Makalero Surat ere hau hifal. . . DEM COMPLETE be.wet letter ‘The letter was completely soaked. . .’

b.

Makasae Surat au ipal. . . letter COMPLETE be.wet ‘The letter was completely soaked. . .’

345

Mkl hau / Mks au also functions as a universal quantifier (‘all’), and quantification appears to be the original function of this marker, while the aspectual functions are still in the process of grammaticalization (cf. Huber & Schapper 2014). The fact that it is originally a quantifier explains why Mkl hau / Mks au does not stand in the aspect slot of the clause template in section 3 like Mkl hai / Mks ai, but in the adverbial slot. The two markers are often combined. This is illustrated in (159) with a stative verb, where the reading of Mkl hau / Mks au is clearly distinct from that of Mkl hai / Mks ai. As the leftmost of all verbal modifiers, Mkl hai / Mks ai has scope over Mkl hau / Mks au. (159)

a.

b.

Makalero Ki-mata hai hau 3.POSS -child INITIAL COMPLETE ‘Her child is already very big.’

pere. be.big.SG

Makasae . . . gi-tupurae ai au 3.POSS -woman INITIAL COMPLETE ‘. . . his wife is already very old.’

laʔir. be.old

The effect of the combination of the two markers is less clear with activity verbs such as Mkl omu ‘ball (yarn)’ in (160a) or Mks gini ‘do’ in (160b). In these

346

Juliette Huber

cases, Mkl hau / Mks au denote the transgression of the activity’s final boundary. The additional meaning component that Mkl hai / Mks ai adds in this situation is that it presents this final boundary as the beginning of a new state. (160)

a.

Makalero . . . mara=ni . . . go=CONJ mara=ni go=CONJ

hai

hau

INITIAL

FINAL

omu, ball

omu=fata ball=COND

fi 1PL . INCL

lale. coil

‘. . . then, when we have balled (all the yarn), then we coil it up (around the length of the forearm).’ b.

Makasae . . . ana people

imboskada ambush

ai

au

INITIAL

FINAL

gini. do

Ana people

ai INITIAL

nati-hei-wari. . . ?-down.there-wait ‘. . . they had prepared an ambush. People were (already) waiting down there. . .’

8.2 Aspectual enclitics Makalero and Makasae have a number of enclitics or particles (their status in Makasae is not quite clear) which encode temporal/phasal notions and are found at the right edge of the verb phrase. In Makasae, these particles are invariable, but the corresponding Makalero enclitics display metathesis, depending on whether the preceding item ends in a consonant or a vowel (see section 2.6.3). The first of these, Mkl =naʔa ~ =ana / Mks nana marks the action expressed by the verb it is cliticized to as an intention. It appears that this intention was originally required to be unfulfilled, as implied by the sentences in (161), though this requirement is being lost. A second enclitic, Mkl =faʔa ~ =afa / Mks apa marks actions that are about to happen, i.e. an immediate future, as in (162). Mkl =teʔe ~ =ete and Mks (d)ete differ from these elements in that they connect two clauses; they cannot typically be used without a following clause. Mkl =teʔe ~ =ete signals that the action expressed by the verb it is cliticized to must be completed as a requirement for the action of the following verb to take place. Mks (d)ete, on the other hand, is a more general sequential conjunction.

Makalero and Makasae

(161)

(162)

a.

Makalero Ei soʔot=ini k-ua-isikola=naʔa 2SG want=CONJ 3.OBJ-on.top-school=INT ‘You want to continue schooling, but. . .’

b.

Makasae Ani karak iskola nana mas. . . but 1SG want school INT ‘I want to go to school, but. . .’

a.

Makalero Sefar ere hai umu=faʔa. DEM INITIAL die=IMM dog ‘This dog is about to die.’

b.

Makasae Depa ere ai umu DEM INITIAL die dog ‘This dog is about to die.’

347

wainini=po. . . CLS =but

apa. IMM

Mks rere is a clause-final element that does not have a Makalero counterpart. Mks rere is the most frequent way, after Mks ai and au, to mark an aspectual distinction and expresses progressivity, as shown in (163).

(163)

Makasae Kuandu gi maʔu ani televisaun when 3SG come 1SG TV ‘When he came, I was watching TV.’

ena watch

rere. PROG

9 Discussion Makalero and Makasae are very closely related, as is apparent from the examples provided throughout this chapter. Interestingly, while the proportion of lexical cognates is very high, non-cognate items are relatively frequent in the expression of more grammatical meanings. The two languages are characterized by simple phonologies and largely isolating morphological profiles. Phonologically Makasae is somewhat more complex than Makalero. For instance, it has retained the distinction between voiced and voiceless plosives, which has been lost in Makalero; it also has a larger inventory of vowel sequences than Makalero.

348

Juliette Huber

Conversely, in terms of morphology and morphosyntax, Makalero is more conservative than Makasae, having retained a few moderately productive processes such as object prefixation with a subclass of verbs. In both languages, up to two arguments can be expressed in a monopredicative clause. Where more than two participants are involved, Mkl mei, a light verb translatable as ‘take’, and Mks ma, a synchronic oblique marker which is not used as an independent verb, introduce the additional participant. Interestingly, this is also the case when the main verb is preceded by a phrase expressing peripheral information such as location or manner. It has been argued that the INCORP LOC slot in which such information is found is associated with reduced transitivity. The same slot is also used to express causation. This pervasive construction and its close association with valency is one of the most intriguing characteristics of Makalero and Makasae. Makalero and Makasae are in many ways the least complex languages of the family. This is most apparent in the domain of pronouns and agreement. Where most TAP languages have multiple pronominal paradigms to encode A, S and P arguments, including complex systems of split-S alignment, Makalero and Makasae have a single paradigm for all of these roles. Furthermore, most TAP languages have multiple sets of pronominal prefixes used to mark or crossreference verbal arguments. While Makalero retains a subset of verbs which combine with pronominal prefixes (and an agreement prefix in the case of the third person), this is completely absent in Makasae. Most of the Makasae cognates of Makalero object-marking verbs appear with a fossilized agreement prefix for the third person. Lastly, the majority of TAP languages distinguish separate possessive pronoun paradigms for alienable and inalienable possession; Makalero and Makasae, on the other hand, have a single set of pronouns. Nonetheless, I have argued that the distinction is relevant at least for Makalero, where kinship and part terms are either obligatorily possessed or expressed as the head of a nominal compound. Whilst being simpler than the average TAP language in many respects, Makalero and Makasae also exhibit some complexities not found in other languages of the family. Several of these are characteristic not only of Makalero and Makasae, but also of the other languages making up the Eastern Timor subgroup of TAP languages, i.e. Fataluku and Oirata (Schapper, Huber & van Engelenhoven 2012: 225f.). Among these are the suppletive number agreement with several stative verbs and the expression of nominal plurality with suffixes or clitics rather than free plural words. Furthermore, the applicative-like expression of locative adjuncts in the INCORP LOC slot, as well as the associated initial consonant mutations found with a subgroup of verbs, is a common characteristic of Makalero, Makasae and Fataluku. In all three languages, these adjuncts inter-

Makalero and Makasae

349

act with the verb’s transitivity such that a transitive verb’s logical object must be expressed with the light verb ‘take’ or a reduced form of it. This incorporation construction is pervasive in the Eastern Timor languages, and is linked to another striking property that sets them apart from the other TAP languages, namely the fact that verb serialization is much more restricted than in Bunaq and the Alor-Pantar languages. This is because the incorporated construction is also used for causatives and manner adjuncts, which are expressed through verb serializations in Bunaq and the Alor-Pantar languages. The incorporated construction of the Eastern Timor languages does often involve verb serialization because undergoers have to be expressed through ‘take’ serializations; however, these ‘take’ serializations are due to the reduced transitivity of the incorporated construction and are thus quite different from the causative and manner serializations found in the other TAP languages. The light verb ‘take’ is also obligatory in the encoding of events involving recipient arguments; interestingly, these reflexes are doubled in various contexts in Makalero, Makasae as well as Fataluku. Finally, a characteristic that appears to be unique to Makalero and Makasae is the [+HUM ] / [‑HUM ] distinction found in some lower numerals and, in the case of Makalero, in a quantifier. Throughout this sketch, I have aimed to highlight the differences between Makalero and Makasae, and thus to support their treatment as separate languages. What has also become apparent is how patchy the description of Makasae, the third largest language of East Timor, still is. In-depth research on this important regional language and its various dialects must therefore be a high priority of TAP linguistics.

Acknowledgments I am very grateful to my two main language assistants, Mafalda Rosa for Makalero and Eduardo da Costa Guterres for Makasae. I would also like to thank Carolino Assunção and his wife Joana da Silva, who always make me feel at home in Iliomar. Many thanks also go to two anonymous reviewers and, in particular, to Antoinette Schapper, through whose helpful comments have much improved both this sketch as well as my understanding of Makalero and Makasae grammar. Funding for fieldwork on Makasae (2005) was provided by the Swiss Society for Endangered Languages SGfbS; fieldwork on Makalero (2007 through 2013) was supported by the Leiden University Centre of Linguistics, the Leiden University Fund, the Swiss Society for Endangered Languages, as well as the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013)/ERC Grant agreement n_263512.

350

Juliette Huber

References Aikhenvald, Alexandra Y. 2006. Serial verb constructions in typological perspective. In Alexandra Y. Aikhenvald & R. M.W. Dixon, (eds.), Serial verb constructions: A cross-linguistic typology, 1–68. Oxford: Oxford University Press. Almeida, António de. 1994. O Oriente de expressão portuguesa. [The eastern Lusophone world.] Lisbon: Fundação Oriente e Centro de Estudos Orientais. Brotherson, Anna. 2003. 2003: A spatial odyssey: Referring to space in Makasai. Honours thesis, Australian National University. Capell, Arthur. 1944a. Peoples and languages of Timor. Oceania 14: 191–219. Capell, Arthur. 1944b. Peoples and languages of Timor. Oceania 14: 311–337. Capell, Arthur. 1944c. Peoples and languages of Timor. Oceania 15: 19–48. Carr, Felicita Mary. 2004. Pupuuk, Masu Moru and Data: Three genres of Makasai. MA thesis, Australian National University. Correia, Adérito José Guterres. 2011. Describing Makasae: A Trans-New Guinea Language of East Timor. PhD thesis, University of Western Sydney. Engelenhoven, Aone van. 2004. Leti: A Language of Southwest Maluku. Leiden: KITLV Press. Engelenhoven, Aone van. 2009. On derivational processes in Fataluku, a non-Austronesian language in East Timor. In W.L. Wetzels (ed.), The linguistics of endangered languages: Contributions to morphology and morphosyntax, 333–362. Utrecht: LOT. Engelenhoven, Aone van. forthcoming. Fataluku. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 3. Berlin: De Gruyter Mouton. Fogaça, Jessé Silveira. 2011. Fonética e fonologia da língua Makasae. [Phonetics and phonology of the Makasae language.] MA thesis, Universidade de Brasília. Hajek, John & Catharina Williams-van Klinken. 2003. Um sufixo românico numa língua austronésia: -dor em Tetum. [A Romance suffix in an Austronesian language: -dor in Tetum.] Revue de linguistique romane 67: 55–65. Holton, Gary. 2014. Western Pantar. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 23–96. Berlin: De Gruyter Mouton. Huber, Juliette. 2008a. First steps towards a grammar of Makasae: A language of East Timor. München: LINCOM. Huber, Juliette. 2008b. Preliminary remarks on object-marking in Makalero. In Marjo van Koppen & Bert Botma (eds.), Linguistics in the Netherlands 2008, 87–96. Amsterdam: John Benjamins. Huber, Juliette. 2011. A grammar of Makalero: A Papuan language of East Timor. PhD thesis, Leiden University. Huber, Juliette. 2014. Landscape in East Timor Papuan. Language Sciences 41: 175–196. Huber, Juliette. forthcoming. Natural locations and the distinction between “what” and “where” concepts: Evidence from differential locative marking in Makalero. Linguistics. Huber, Juliette & Antoinette Schapper. 2014. The relationship between aspect and universal quantification: Evidence from three Papuan languages of Timor and Alor. In Marian Klamer & František Kratochvíl (eds.), Number and quantity in East Nusantara, 152–169. Canberra: Asia-Pacific Linguistics. Hull, Geoffrey. 2004. The Papuan Languages of East Timor. Studies in Languages and Cultures of East Timor 6: 23–99.

Makalero and Makasae

351

Hull, Geoffrey & Adérito José Guterres Correia. 2006. Disionáriu badak Makasae-Tetun, TetunMakasae = Disionari gi digara Makasae-Tetun, Tetun-Makasae. [Short dictionary MakasaeTetun, Tetun-Makasae.] Dili: Instituto Nacional de Linguística, Universidade Nacional Timor Lorosa’e. Hull, Geoffrey & Mario Valdes Costa. 2005. Makasai. Dili: Instituto Nacional de Linguística, Universidade Nacional Timor Lorosaʹe. Klamer, Marian. 2010. A grammar of Teiwa. Berlin: De Gruyter Mouton. Klamer, Marian & Antoinette Schapper. 2012. The history of ‘give’ constructions in the Papuan languages of Timor-Alor-Pantar. Linguistic Discovery 10: 174–207. Klamer, Marian, Antoinette Schapper & Greville Corbett. 2014. Plural number words in the AlorPantar languages. In Marian Klamer (ed.), Alor-Pantar languages: History and typology, 375–412. Berlin: Language Science Press. Kratochvíl, František. 2014. Sawila. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 351–438. Berlin: De Gruyter Mouton. Lewis, M. Paul, Gary F. Simons & Charles D. Fenning (eds.). 2015. Ethnologue: Languages of the World. Eighteenth edition. Dallas, Texas: SIL International. Online version: . Marques, Tito Lívio Nunes. 1990. Método prático para apreender o Makasae. [Practical method for understanding Makasae.] Lisbon: self-published. Mithun, Marianne. 1984. The evolution of noun incorporation. Language 60: 847–894. Pinto, Tomas Mariano. 2004. Makalero Dicionário. [Makalero Dictionary.] Dili: Timor Loro Sa’eNippon Culture Center. Pinto, Tomas Mariano. 2007. Maʹu Fi Makalero Lolo. [Come let’s speak Makalero.] Dili: Timor Loro Sa’e-Nippon Culture Center. Sato, Hiroko. 2013. Grammar of Kove: An Austronesian language of the West New Britain province, Papua New Guinea. PhD thesis, University of Hawai’i at Manoa. Schapper, Antoinette. 2014. Kamang. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 285–350. Berlin: De Gruyter Mouton. Schapper, Antoinette, Juliette Huber & Aone van Engelenhoven. 2012. The historical relations of the Papuan languages of Timor and Kisar. Language and Linguistics in Melanesia: Special issue on the history, contact & classification of Papuan languages, 194–242. Schapper, Antoinette, Juliette Huber & Aone van Engelenhoven. 2014. The relatedness of TimorKisar and Alor-Pantar languages: A preliminary demonstration. In Marian Klamer (ed.), The Alor-Pantar languages: History and typology, 99–154. Berlin: Language Science Press. Steinhauer, Hein. 1993. Notes on verbs in Dawanese (Timor). In Ger Reesink (ed.), Topics in descriptive Austronesian linguistics, 131–158. Leiden: Vakgroep Talen en Culturen van Zuidoost-Azië en Oceanië, Rijksuniversiteit te Leiden. Steinhauer, Hein. 2014. Blagar. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and Pantar. Sketch grammars. Volume 1, 147–220. Berlin: De Gruyter Mouton. Sudiartha, I Wayan, I Nengah Budiasa, Ni Luh Partami & Anak Agung Putra. 1998. Struktur bahasa Makasai. [Structure of the Makasai language.] Jakarta: Pusat Pembinaan Dan Pengembangan Nasional.

Index Abui 5, 6–7, 111 – typological comparison, grammar 25, 28, 29, 31–33, 38, 41, 90, 96–97, 104–106, 319 – typological comparison, phonology 19, 22, 24 – witnesses for Proto-Timor-Alor-Pantar reconstructions 10, 15, 16, 28, 30 Adang – typological comparison, grammar 25, 33, 41, 43–44, 96, 106, 181 – typological comparison, phonology 20, 22 – witnesses for Proto-Timor-Alor-Pantar reconstructions 10, 16, 30 Adiabang 4 adjectives 39, 67, 68, 72, 82, 98, 128, 140, 215, 293 adverbs 137, 221, 224–225, 304–306 – and elevationals 131–132 – and reduplication 67, 82 – and word order 34–35, 291, 294, 304 – aspectual/temporal 102–103, 176, 179, 269, 305–306, 343–344 agriculture 15 Aikoli 56 alienability 31–32, 70, 85–86, 226, 317, 348 alignment, see morphological alignment allomorphy 66, 70, 122–123, 190, 239, 285– 286 Alor Malay, see Malay Alorese 18, 56, 57 animacy 78, 133, 135, 164–165, 174, 219, 246, 319, 341 applicatives 33, 98–99, 106, 152, 162–167, 181, 283–285, 251–254, 265 apposition 211 articles 40 aspect – and evidentiality 260–261 – completive 103–104, 180 – continuative 178 – durative 126 – enclitics 198, 346

– imperfective 29, 35–36, 103, 117, 166–167, 256, 260–261 – in serial verb constructions 170, 174–175, 256–257, 342 – language comparison 28–29 – particles 34–36, 102, 127, 343–345 – perfective 175, 176, 179, 241, 262–263 – progressive 264 – prospective 263–264 – stem alternations 28–29, 104, 106 – suffixes 130, 131, 138, 140 – see also serial verb constructions assimilation 125, 190, 201, 265 associative plurals 32, 190, 201, 265 Austronesian languages – comparison with 12, 22, 34, 40, 42, 273 – contact with 6, 14, 15, 16, 17, 18, 19, 45, 281, 311, 327 Baucau 268, 270, 271 benefactives 160, 333–336 bilingualism, see multilingualism Bird’s Head 9, 12, 56 Blagar – typological comparison 4, 19, 20, 22, 31, 25, 28, 33, 312, 314 – witnesses for Proto-Timor-Alor-Pantar reconstructions 10, 13, 16, 30 blending 123–124 Bobonaro 5 body part nouns 31, 86, 87, 145, 226, 276, 289, 290, 317 Bol 4 borrowing, see loans Bunaq 2, 5, 6, 7, 8, 17, 18, 27 – typological comparison, grammar 24, 28, 31, 33, 34, 35, 38–46, 329, 331, 349 – typological comparison, phonology 19, 20, 22, 23 – witnesses for Proto-Timor-Alor-Pantar reconstructions 10, 13, 14, 15, 16, 26, 30 Buraga 111, 112, 131

354

Index

causatives 33, 45, 130, 159, 160, 161, 176, 349 – see also serial verb constructions Central Alor languages 4, 18, 24, 31, 43, 56 Central-Eastern Alor languages 11, 18, 43, 187 change, see language change clans 5, 56, 67 classifiers 83, 142–143, 218 clitics – and phonological processes 123, 125, 285–286 – completive 286, 346 – continuative 178 – demonstrative 221–222, 224, 315 – immediate future 346 – imperative 79–80, 138, 309 – in relative clauses 313, 314 – intentional aspect 346 – interrogative 130, 137, 138, 307 – modal 136 – plural 217, 218, 219, 227, 235, 318, 319, 348 – proximal 223 – topic 214 Coimbra (Portugal) 271 common person 146–147 compounds 69–70, 121, 124, 173, 290–291, 317–318, 321, 348 consonant clusters – in coda 23, 62, 63, 122, 197, 200, 277, 279, 285, 287 – in onset 121, 194, 198, 200 constituent order 34, 38–42 – clausal 70–71, 105, 127, 134, 138, 206, 291–292, 323–324 – in NPs 81, 86, 214 – in serial verb constructions 101, 171, 173, 333 contact – with Austronesian languages 14, 17–18, 281, 311, 327 – with Malay 121, 173, 192 – with other languages 18, 57, 105, 187, 189, 192, 272, 279, 281, 311 coordination 300, 301, 331 copulas, lack of 130, 212

Dani 13 Dawan 273 Debu 4 definite marker 216, 221, 223–224, 230, 290, 314–316 deixis – and constituent order 40–41 – and vowel harmony 287 – deictic center 131 – deictic verbs 74, 259, 293, 296, 309, 314– 315, 332 – discourse deixis 223–224 – elevationals 10, 132, 208, 211–212 – nominal deixis 289–290 demonstratives, see deixis derivational morphology 73, 142, 165 – and reduplication 288 – deriving adverbs 67 – deriving intransitive verbs 98 – valency changing 158 – word class changing 33, 131, 289, 296 Di’ang 4 diphthongs 20, 63, 116–119, 120–121, 124, 125, 204, 276, 281 ditransitive constructions, see trivalent constructions Dukila 187 Eastern Alor languages 24, 38, 46, 47, 125, 188, 192, 265 Eastern Alor Montane languages 4, 19, 20, 29 31 East Timor languages – and contact 17, 40 – and Proto-Timor-Alor-Pantar 10, 282, 339 – region 6 – typological comparison, grammar 11, 21, 28, 32–36, 42, 44–46, 105, 304, 319, 331 – typological comparison, phonology 23, 27 – subgroup 8, 10, 11, 13, 15, 16 ELAR 187 elevationals – and Proto-Timor-Alor-Pantar 10 – in demonstratives 282, 314–315 – predicates 131–132, 211–212, 332 – see also deixis

Index

emphasis 41, 168–169, 287, 304–305 endangerment 5, 6, 57, 111, 187 Ethnologue 4, 269 evidentiality 176, 204, 260 existential predicates 46, 73, 208–209, 210, 313 Fanating 56 Fataluku – and contact with other languages 17 – and typological comparison 19–28, 31– 34, 36, 39, 40–41, 45 – as an East Timor language 2 – dialects of 5 – witnesses for Proto-Timor-Alor-Pantar reconstructions 15, 16 focus 90, 147, 150–151, 206, 227–228, 230, 231–232, 292 fourth person 87, 96–97, 106 Frog Story 58, 255 Gai’, see Bunaq gapping 140 gemination 21, 24 “give” constructions, see trivalent constructions glottonyms 4, 5 grammaticalization 142, 166, 172, 175, 176, 180, 305, 331, 345 Habolat (also Habollat, Hifolat) 56, 57 Hamap 6, 56, 111 head-final languages 17 Iha 13, 14 IICT (Lisbon) 270 Iliomar 269, 271 incorporation – and serialization 33–34, 42 – locational 295, 299–303 – noun 164, 169 – patient 295, 297–299, 303 – verbal 44–46 Indonesian 5, 103, 178, 187, 231, 263 – and orthography 58, 115, 205 – calques/loans from 67, 111, 141, 202, 217, 231, 263

355

– in language shift 57 – see also Malay initial consonant mutations 26, 27, 282– 283, 285, 297, 302, 348 interrogatives 78–79, 136–138, 275, 306– 309, 323, 150, 178–179 inverse marker 240–241, 243–244, 265 Irakena 187 Islam/Muslims 111 Island South East Asia 15, 17, 48 Jafoo 56 Kaera 3 – typological comparison 19, 22, 24, 28, 29, 33, 36, 37 – witnesses for Proto-Timor-Alor-Pantar reconstructions 10, 15, 30 Kaipera 187, 188 Kalabahi Bay 111 Kamang 4, 37, 38, 43, 90, 93–106 – typological comparison, grammar 26, 28– 29, 36 – typological comparison, phonology 19, 22, 24, 32, 227, 319 – witnesses for Proto-Timor-Alor-Pantar reconstructions 10, 11, 15, 16, 30, 33 – cognates with other languages 216, 301 Karas 13 Kedang 18 Kemak 17, 40 Kewala 187, 188 kinship nouns 31, 32, 33, 226, 317, 318, 319, 348 Kiralela 187, 188 Kiraman 8, 28, 29, 40, 111 Kisar 2, 17, 314 Klamu 5 – see also Nedebang Klon 2, 6, 8, 57, 111 – typological comparison 20, 22–23, 33, 38, 74, 93, 97, 105–106, 181 – witnesses for Proto-Timor-Alor-Pantar reconstructions 10, 13, 15, 16 Kobra 187 Kolana 4 Kroku 3, 22

356

Index

labile verbs 130, 163 Laga 271, 298, 304, 320, 324, 339, 340 LAISENG Corpora 7 language change 14, 17 – see also contact Lantoka 187, 188, 209, 240 Lautém 269 LDTC (Hawai’i) 270, 271, 344 Lelafang 4 Lembata 18 Lerabaing 111 Leti 40, 273 LIPI (Indonesian Institute of Sciences) 6, 56, 57, 112 loans 17, 18, 189, 197, 272, 281, 311, 327 – from Malay 121, 173 – see also contact, Indonesian locationals 40–41, 128, 133–134, 338 Lola 56, 57 Luro 5, 269 Màcassai 269 mainland New Guinea 1, 2, 11, 12, 14, 15, 16, 17 Malay 80, 139, 178, 180 – as lingua franca 5 – children speaking 57, 111, 187 – loans from 67, 83, 121, 135, 173, 189, 202 – shift to 111 – see also Indonesian Maliana 5 Mambae 17 Managomo 187 Manet 4 marginal phonemes 19, 22, 112–113, 114, 188–190, 191, 192, 197, 205, 275–277 maritime culture/vocabulary 15, 16 Masin Lak 111, 171 Maukuru 187 Meher 2, 17 Melanesia 1, 48 metathesis 23, 117, 124–125, 181, 282, 346 middle voice 10 Modebur 3 Moduda/Samuda 187 Moo 4

morphological alignment 106, 157, 181, 348, 106 morphological innovations 10, 11, 13, 27, 329, 339 Moru 111, 112 Max Planck Institute for Psycholinguistics Nijmegen 112, 188 multilingualism, bilingualism 4, 5, 18, 56, 57, 187 Muslims, see Islam/Muslims Ná-Ine 269 Nai-Domo 269 Naueti 17, 18, 40 – see also Waima’a-Naueti linkage Naumang 187 Nedebang – and language loss 5 – typological comparison 10, 11, 19, 21–22, 24–25, 31, 36–39, 43 – witnesses for Proto-Timor-Alor-Pantar reconstructions 15, 16, 30 negation – postverbal 36, 37, 135 – preverbal 36, 76, 105, 213, 304 New Guinea, see mainland New Guinea non-Trans New Guinea 12 Oirata 2, 6, 15, 16, 22, 23, 33, 45, 348 orthography 58, 63, 115, 194, 205 Ossu 270, 271, 272, 274, 298, 320, 322, 324, 340 Pailelang 56 Paitoko 187 Pantar Straits 2 – see also Straits languages PARADISEC 6, 7, 58 Pearl Bay 56 Peisaka 187 population movement 57, 111, 187 Portuguese 269, 270, 272, 279, 281, 288, 311 postpositions 33, 34, 74, 166, 167, 284, 299, 301 prepositions 34 prestige languages 57

Index

Probur Utara 56 prohibitives 76, 77, 106, 309 proto languages of Timor-Alor-Pantar 8, 10, 14, 17, 192, 274, 275, 329, 342 see also Proto-Timor-Alor-Pantar Proto-Trans-New-Guinea 12, 13, 14, 17 Proto-Timor-Alor-Pantar 10, 11, 13–17, 18, 27, 28, 30 Pura 57 Pureman 4, 187, 188 questions, see interrogatives reciprocals 97–98, 146, 247–248, 250–251, 297–298, 322, 338 reduplication 67–68, 82, 126–127, 219, 287–288, 302 reflexives 10–11, 96–97, 241, 297–298, 321– 322, 338 relative clauses 39, 81–82, 139–141, 215– 216 313–314 Ruilak 56 Sá-Âni 5, 269 Sar 4 Sawila 187, 188 – comparison with Kula 202, 203, 223, 225, 231, 265 – typological comparison, grammar 26, 29, 30, 31—32, 33, 106, 125, 315 – typological comparison, phonology 19, 20, 22, 24, 198 – witnesses for Proto-Timor-Alor-Pantar reconstructions 10, 11, 15, 16, 30 serial verb constructions 34, 99–102, 170– 175, 255–260, 333–343, 349 – aspectual 175, 203, 256–257 – causative 42, 46, 172–174, 257–258, 339– 341 – comitative 174 – instrumental 102, 173, 336–337 – manner 43–44, 337–339 – motion 101, 170–171, 259, 332–333 – posture verbs 260 – result 176, 258 – “take” serialization 255–256, 295–296, 333–336, 341–343, 349

357

– see also aspect – see also causatives – see also incorporation songs 7, 18 sound change 9, 10, 18, 23, 30, 123 split S systems, see morphological alignment stem alternations 104–105, 106, 282–283 Straits languages 2, 9, 19, 20, 22, 24, 25, 31, 33, 38, 181 stress – and vowel length 193–265 – non-phonemic 23, 118–119, 281–282 – penultimate 24, 200 – phonemic 24, 65–66, 200, 209 – phonetic correlates of 120 – primary/secondary 69, 285 Suboo 47 substrate elements 12, 18 suprasegmentals 23 Takala 187 Tanglapui 5, 187, 188 Teiwa – as a dominant language 4 – typological comparison, grammar 10, 28, 31, 93, 98, 181, 312, 319 – typological comparison, phonology 19, 21, 22, 23, 24 – witnesses for Proto-Timor-Alor-Pantar reconstructions 15, 16, 30, 33 Tetun 5, 17, 40, 270, 271, 272 The Language Archive 6 third vs. fourth person see fourth person Tiee 4 Timor Loro Sa’e Nippon Culture Center 271 Timor-Leste 2, 5, 7, 18 Tokodede 18 topic 41, 160, 225–226, 292, 324 Trans-New Guinea languages 11, 12 – see also Proto-Trans-New Guinea trivalent constructions 157, 162, 206–208, 254, 255, 295–296, 334 – “double take” give constructions 11 – “give” constructions 11, 38–39, 207, 334– 336 truncation 283–285, 293, 302, 340

358

Index

Uab Meto 18 valency changing – and verb structure 128, 130 – decreasing 158, 303 – increasing 66, 99, 106, 158, 159–162, 248–250 Viqueque district 268, 271 vowel epenthesis 23, 122, 123 Waika 187 Waima’a-Naueti linkage 17, 18 – see also Naueti Wersing 4, 187, 188 – typological comparison, grammar 24, 26, 28, 35–37, 43, 106, 180, 261, 265, 319 – typological comparison, phonology 19, 22, 23

– witnesses for Proto-Timor-Alor-Pantar reconstructions 10, 111, 15, 16, 30, 33 West Alor languages 14, 31, 33, 47, 105, 111, 181, 314 West Bomberai 12, 13, 14 West Pantar 5, 6 – typological comparison, grammar 19, 25, 28, 33, 35, 106 – typological comparison, phonology 21, 22, 23, 24 – witnesses for Proto-Timor-Alor-Pantar reconstructions 10, 15, 15, 30 West Trans New Guinea linkage 13 Wissel Lakes 1 word order, see constituent order