Downloaded From

V. Heeschen The position of the Mek languages of Irian Jaya among the Papuan languages; History, typology, and speech

In: Bijdragen tot de Taal-, Land- en Volkenkunde 148 (1992), no: 3/4, Leiden, 465-488

This PDF-file was downloaded from http://www.kitlv-journals.nl

Downloaded from Brill.com10/02/2021 03:38:53PM via free access VOLKER HEESCHEN

THE POSITION OF THE MEK LANGUAGES OF IRIAN JAYA AMONG THE PAPUAN LANGUAGES; HISTORY, TYPOLOGY, AND SPEECH'

Introduction This article sets out to evaluate the unity and diversity of the Papuan languages from different points of view. I wil1 tackle this genera1 task by giving an account specifically of what differentiates the Mek languages from, and subsequently what relates them to, the other Papuan languages. Until approximately forty years ago, the Papuan languages had 'been thought to be mostly not related to each other ... The term "Papuan languages" or "non-Austronesian languages" was, in consequence, only employed as a negative classificatory term describing languages referred to by that name, as distinct from Austronesian ... languages, without presuming the existence of any genetic link between them' (Wurm 1982:3). Over the past four decades an impressive number of descriptive works on these languages, including complete grammars, wntten mostly by members of the Summer Institute of Linguistics, have been published. Thanks to language surveys conducted by that institution, and, above all, thanks to the work of scholars like Anceaux, Cowan, Wurm, Laycock, Voorhoeve, Dutton2, and others, our insight into the classification of the

1 The main linguistic and anthropological field work for this study was camed out for a penod of about one and a half years in Eipomek and for about three years in Kosarek. Wordlists and sample texts were collected in almost al1 parts of the area of the Mek language family, however. The research was funded by the Deutsche Forschungsgemeinschaft, the Max-Planck- Institut fur Psycholinguistik, Nijmegen, and the Forschungsstelle fur Humanethologie in der Max-Planck-Gesellschaft, Andechs. Dunng my longer stays in Kosarek I was the guest of the Protestant Church of Irian Jaya and was financially supported by the Vereinigte Evangelische Mission of Wuppertal. The article was prepared for the International Conference on the Mek and their Neigh- bours held in Seewiesen, Germany, in October 1990. The ideas underlying it were first presented in a lecture given at the University of Bern on January 18th, 1988. 2 A complete bibliography is to be found in Wurm 1982.

VOLKER HEESCHEN is a research assistant at the Forschungsstelle fur Humanethologie in der Max-Planck-Gesellschaft, who obtained his Ph.D. at the Ruhr-Universitat in Bochum. Specialized in Papuan linguistics, the philosophy of language and ethnolinguistics, he has previously published Die Sprachphilosophie U! v. Humboldts and A dictionary of rhe Yale language (wirh a sketch of grammar and English index). Dr. Heeschen's address is Prinzenweg 22b, D-8 130 Starnberg, Germany.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 466 Volker Heeschen

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Position of the Mek Languages of Iriun Jaya 467

Papuan languages has improved dramatically. The classification work was based generally on lexicostatistical surveys, but was also supported by dialect studies, work on language families, and typological studies, as wel1 as by research into the problems of substrata and language mixing. It culminated in the establishment of five major and a nurnber of minor phyla, each including languages that are genetically related to each other, but themselves without any genetic interconnection. The Mek languages are spoken in an area in the eastern highlands of Irian Jaya. The Mek language family was first identified by Bromley (Bromley 1966-7; compare Voorhoeve 1975, Heeschen 1978). Wurm (1982: 187) decided 'to give the family provisionally the status of a subphylum' within the Trans-New Guinea phylum. Foley (1986), adher- ing to the more rigorous methods of comparative linguistics and taking the criticisms directed against the method of lexicostatistics int0 account, refuses to accept the division into phyla, but divides the Papuan languages into some 65 language families; he does not mention the Mek languages. The first data concerning a Mek language were published as early as 19 12 by De Kock, who, as a member of a Dutch expedition, had penetrated to the southern borders of the Mek language area, near Mount Goliath, and compiled a wordlist and analysed the counting system. Thus, the languages were known for a long time as the 'Goliath languages'. The boundaries of this family are: in the west, the mountains between the Yaholi and Obahak rivers and Kosarek, and the mountains between the Solo and Erok rivers; and in the east, the mountains between the Kloof and Sobger rivers, and the area between Okbap and Apmisibil. In the lowlands to the north and to the south of this area, hamlets are smaller than in the high valleys, and settlements are scattered more and more widely apart in seemingly uninhabited regions. Lines that could be drawn from Keleka to the mountains between the Kloof and Sobger rivers, and from the confluence of the Solo and Erok rivers to the mountains south of the Be and Ok Cop nvers might approximately indicate the northern and southern boundaries respectively. The main groups of languages (or dialect clusters) are: 1. in the north-west, the Yale language spoken around Kosarek, the dialects spoken in the Iluk, Obahak, Sesom, and Yamek valleys; 2. in the south-west, the dialects spoken around Korapun, Sela, and, probably, Nalca; 3. in the centre, Eipo, Una (spoken around Langda and Larye), and the dialect spoken in the Tanime valley; 4. in the east, al1 the dialects spoken between Bime and Okbap. The linguistic work done on the Mek languages so far has been in the fields of phonology, grammar, texts, and lexicography (Heeschen 1978, 1985, 1990b, 1992; Heeschen and Schiefenhövel 1983; Louwerse 1978 and 1988; Sims and Sims 1982). To the west of the Mek language area, languages of the Great Dani family are spoken (Bromley 1966-7, 1973; Voorhoeve 1975:35-7, 68).

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 468 Volker Heeschen

The languages from this family to be mentioned in the present article are: 1. the Angguruk language, a member of the Ngalik-Nduga sub-family, spoken in the Yaholi valley, from Welarek to Angguruk, and 2. Grand Valley Dani, a member of the Dani sub-family, spoken in the west of the Angguruk area.3 To the east of the Mek language area, languages of the Ok family are spoken. Languages from this family to be mentioned in this article are: l. Nalum, spoken around Apmisibil, and 2. Telefol, spoken across the border in Papua New Guinea. Both these languages are members of the Mountain Ok sub-family.4 To the south of the Mek language area, Awyu languages of the Awyu- Dumut family are spoken, and to the north, in pockets separated by uninhabited areas, languages of the Kaure stock.5

History To give an idea of the kind of difficulties faced by the linguist when doing research int0 Papuan languages, I will deal briefly with the general social and cultural context. Here the conditions of life in smal1 communities, and more especially intra- and inter-village marriages, trading partnerships, warfare, language attitudes and cults, are responsible for the preservation, mixing and diffusion of an incredible number of languages. Then I will discuss the eastern and western dialects of the Mek languages, in the case of which borrowing from and intermixing with the eastern and western neighbours respectively has occurred. These areas provide minor examples of the above-mentioned genera1 context, which will be described in the next paragraphs. The 750 Papuan languages and, in addition, more than 250 non-Austro- nesian languages are spoken in an area of about 900,000 square km. Thus, the average size of a language area is 900 square km, and even in the times before Western contact a Papuan could easily have walked from the centre to the periphery of such an area. The average number of speakers of any single language may come close to 3000 O have no exact population figures for Irian Jaya). If we assume that each language area comprises seven to ten hamlets, then we have here speech communities of 300 to 450 speakers.6 According to Forge (1972), 450 is the maximum number of inhabitants an egalitarian community can handle; when such a community exceeds that number, it becomes subject to fission. Thus there seems to be a close correlation between the number of speakers per dialect or language

3 The comparative notes in this article relating to these languages are based on Bromley 1981, Zöllner s.a. (a) and (b). 4 Wurm 1982:14-7. Comparative notes in this article relating to these languages are based on' Hylkema 1969, 1974, and s.a., Healey 1965a-c, Healey and Healey 1977. Voorhoeve 1975:27-28,45,69; Heeschen 1978:41-45. More recent works, such as Silzer and Heikkinen 1984 and Louwerse 1988:3, give no better or more precise information. 6 Similar figures are given in Foley 1986 and Sankoff 1977.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Position of the Mek Languages of Irian Jaya 469 on the one hand, and the size of communities, their communicative behavi- our, and their means of handling conflicts on the other. Endemic warfare tends to separate speakers of the same language, whereas exogamic, virilocal rules of mamage as well as trading partnerships relate speakers of different dialects or languages to each other. Children first learn the language of their mothers, who often come from another dialect or language area, while their 'mother tongue', that is, the language of their native area, is only taught them later by their fathers. Multilingualism is not an exception, but is the natura1 thing in a sociolinguistic environment of people living in smal1 tribes. Furthermore, we find conscious language manipulation in certain special groups, such as those of boys initiated at the same time, and in types of speech beyond the normal daily verba1 interaction. Fission, as one of the usual methods of solving conflicts within settlements, leads to people learning the language of their new settlement area and to language mixing (Cape11 1976; Foley 1986; ~eeschen1984 and 1990a; Laycock 1982; Thurston 1982 and 1987).

The founding myths of the eastern Mek speakers as well as those of the westernmost Ok speakers in Apmisibil suggest the presence of mountains in the centre of the Mek area (Hylkema 1974; Heeschen 1990b). The Apmisibil people borrowed the terms for mother's brother, mother (term of address), brother-in-law and sister-in-law from speakers of a Mek language. Correspondences like this link even the Telefol language in the centre of the Ok language area to the Mek languages.7

Example 1 Telefol Eipo baásim wibro base, basi sihu, fasihu nek cross-cousin (Y nik 6cross-cousin')8 nábinang, kábinang ka fasihu (among other rneanings) younger si, rnosi, fasihu (Y nabi 'the second-bom') mom mobro mam mobro, mobroso

Other words also point to the existence of 'cultural' and 'social' chains, such as, among others, Telefol (T) ket kon 'decoration', Eipo (E) kite 'paraphernalia, adornment for dancing', T mók 'type of stone adze', Yale (Y) mok 'sacred stone', T waa 'fence of vertical boards', E wa 'garden', T amém, abém 'forbidden, sacred', E mem 'no, taboo', Y mem 'taboo', T nong am 'men's house', dialects east of Eipomek nongnang oupe 'sacred stories',

7 For data on Telefol see Healey and Healey 1977. For al1 the words mentioned here, the Telefol language has synonyms not related to any words of a Mek language. 8 Henceforth the abbreviation Y wil1 be used for Yale, E for Eipo, T for Telefol, and A for Angguruk.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 470 Volker Heeschen

E nong 'body, the edible part of a fruit, the most important thing of something', T káyaam 'wild dog', E and Y kam 'dog', T kumak 'ginger', E kuma 'ginger', T mán dét 'boys' house song', T mufuúm det 'hair- initiation song7,E dite 'song7. The relationship of the westernmost Mek language, the Yale language, to the easternmost Dani language, the language of the Angguruk people, presents the Same picture. The Angguruk people and their western and northern neighbours have borrowed from the Mek languages the words for father-in-law, mother-in-law, and son-in-law and about half the words used in sacred formulas and incantations. The Yale speakers have borrowed from the west the words pabe 'incest' and ngei 'owner'. The word for initiation is historically the Same in al1 the Mek languages and in some of the Dani languages, viz. E kwit, Y wes, Angguruk (A) wyt. Throughout the whole of the area, that is to say from Ok to Dani, the counting system is the Same, and one counts from the small finger of the left hand, up the left arm, along parts of the head, and down the right arm to the small finger of the right hand. The unit word for the numerals from 1 to 25 or 27 is Ok deeng, Mek deng, Dani teng. Given these similanties, we can assume that these speech communities maintained intensive social and cultural contacts in a not-too-distant past. Other similarities are less easily explicable. The Ok languages, the central and eastern Mek languages, and the Dani languages west of Angguruk have two series of stops. The Yale language and the Angguruk language have only one series. In Eipomek and in the dialects east of Eipomek, accent is phonemic. Angguruk places the accent on the last syllable, and the Yale language tends to place it on the last syllable. Angguruk has borrowed many verb roots from a Mek language which must have been the precursor of the Yale language and which is much more closely related to the centra1 dialects, judging from the phonetic shape of the words.

Example 2 Angguruk Eipo, Yale

fala-ruruk to break (in the middle) E fak- to split, to open Y ak~ub-~to bend fok-tumk to sow E fok- to push open the soil (in order to plant), to copulate Y hok- to dig, to hollow out kun-turuk to be shady E kuna shadow E kunukmal it is getting dark Y kumb- to make dark, to hide Y kuna a- to grow dark

9 Aksub- is a verba1 compound from *ak- 'to bend, to break' and sub- 'to saw, to cut'.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Position of the Mek Languages of Irian Jaya 47 1 lebeg-turuk to flash E (liman) lebmal there is a flash of lightning Y (limang) ielamla there is a flash of lightning (Eipo f is retained in Angguruk, and becomes h or is lost in Yale.)

Yale and Angguruk speakers can add the suffixes -na, (A) -no to the deictic particles, whereby these assume anaphoric functions. Furthermore, these suffixes integrate disjointed phrases into a clause. Yale speakers have borrowed at least one inflectional suffix from Angguruk, as witness the following example.

Example 3 (A) palukag they will cut, (Y) kadebaukang they will or want to cut.

At this point of the verba1 inflection systems, in the 3. plural future tense, the Yale language differs from the other Mek languages. In our days Yale speakers have introduced A an 'I' and hut 'you' for Yale na 'I' and an 'you' into their language. To my knowledge, in a good many verba1 forms in the Dani languages (as wel1 as the Ok languages) the verba1 root is separated as the carrier of the meaning from the auxiliary complex containing the suffixes for tense, aspect, and so on (Bromley 1981:23-30; Healey 1965b:30-40; Zöllner s.a. (a); cf. examples 9, 27, and 28). Adjunct plus pro-verb constructions are common in both language families. Again, the Yale language is the only one in the Mek language area to have developed similar verba1 phrases.

Example 4 (Y) de-do a-ok eat-sim./inf. make-3sg.Ipa.111~~ he ate (E) dib-uk eat-3sg.Ipa.111 he ate

These phenomena are indicative of social relationships, cultural chains, extensive borrowing, area1 features, and possibly re-lexification. Before tracing the evidence for a genetic relationship, I will demonstrate the unity and diversity among the Mek languages themselves. That these languages form a closely related family is beyond any doubt. The 1owest.percentage of shared cognates found was 59% (Heeschen 1978). Subsequent studies

'0 For an explanation of the abbreviations used in this article, see the list of abbreviations at the end of the 'text.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 472 Volker Heeschen identified increasingly more cognates and eliminated errors, so that the percentage for the relationship between Yale and Eipo ultimately was 70; by now it should even be between 75 or even 80 % (Heeschen 1978 and 1992).

The Mek languages A survey of the deictic systems wil1 illustrate the unity of the Mek languages.

Example 5 Bime Eipo Yale here a-tara a-tam ane up there i-tara ei-tam ani down there u-tara u-tam anu across there or-tara or-tam aned,ano

The bound morphemes -tara and -tam have the meaning 'side, at the side of. For Y an- I have not been able to establish a meaning. The unity of the Mek languages is also illustrated by the verba1 tense- person-number suffixes, as set out in the following example.

Example 6 Eipo Yale lsg. present durative yan-ma-n ya-lam-ne I come l sg. today's past ya-n ya-n I have come 1pl. present durative yan-mab ya-lam-ab we come I sg. past (yesterday) ya-nwe ya-no I came (yesterday) lpl. far past durative yan-ma-upe ya-lam-ubu we came.

Identifiable morphemes are: the root ya- 'to come' (in E a morpheme -n- is suffixed to the root, occasionally denoting repetitive action), the durative aspect suffixes -ma- and -lam-, lsg./present-future suffix -n, today's past indicator labialized -n-, Ipl./present-future suffix -ab, and Ipl. distant past indicator labialized -b- (in E /p/ is always labialized). The presence of different sets of connectives is the main cause of the lack of mutual comprehensibility of closely related 'languages', which according to the percentage of shared cognates between them should be classed as different 'dialects', as the following example testifies.

Example 7 (E) ya-lye-ora, kwaning areb-lyam! come-3sg.lmed.-then sweet potato . give-imper.12sg. (Y) ya-le-ba, kwaneng dad-lu-lam! come-3sg.lmed.-thenlds. sweet potato give-imper.-2sg. If he comes, give sweet potatoes to him.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Position of the Mek Languages of Iriun Jaya 473

The interrogative pronouns are supposed to belong to the oldest layers of the lexicon of a language. Compared with the usual degrees of unity among the Mek languages, the diversity between them as regards the interrogative 'what' is striking.

Example 8 Bime Lary e Eipo Yale ware una yate saa

There is no way of reconstructing common forms, unless one disregards the bulk of the words and their correspondences from one language to the next (for initia1 y with the forms for 'to come' cf. example 7 above). Two explanations for this diversity are: 1. different Mek languages that are not closely related underwent re-lexification by just one superstratum, which accounts for the otherwise proven unity; 2. pragmatic speech rules proscribing direct questions in intimate verba1 interaction have caused rapid changes in the set of interrogative pronouns (in the Eipo as wel1 as the Yale speech community I learnt to avoid and to paraphrase 'why' questions; posing a direct question like 'why have you come?' here is an offence).

The Mek languages and other Papuan languages In view of the above-described sociolinguistic conditions and in accordance with the standards of comparative linguistics, a comparison of words and a check of different comparative wordlists alone produce no evidence for a genetic relationship between languages. This is true also with respect to the Mek languages and the Trans-New Guinea phylum languages (compare, among others, McElhanon and Voorhoeve 1970, Bee 1973, Voorhoeve 1975). When consulting the available dictionaries, one is sometimes struck by similarities, such as, for example, Wahgi (Ramsey 1975) yu 'word, language', E yupe 'word, language'. Eipo (and 'common' Mek) ya- 'to come' and bi- or ba- 'to go' led me to look for correspondences even among the middle Sepik languages; Foley (1986:212) has made the reconstruction ya- 'to come' and pu- 'to go'. The virtually omnipresent root na- 'to eat' (Wurm 1982:92) is absent in the Mek languages. More reliable traces may be found in the Ok languages, such as: T iip 'air', E and Y im 'sky, air'; T kaal 'bark (of tree)', E ka1 'ficus species', Y ka1 'tree'; T mán 'child', E and Y me 'child'; T tiin 'eye', E asing 'eye'; T yuan 'foot', E and Y yan 'foot'; T tang 'smell', E tong 'smell'; T sít 'tooth', E and Yale si 'name, tooth', this latter item being in line with reconstructed Proto-Eastern Highland si- 'to say' (Foley 1986:257). As soon as we leave the lexica1 domain to examine grammatica1 sub- systems, any relationships seem to either fade away or to be much less easily proven. Functionally and etymologically, the system of the deictic particles is different in Ok, Mek, and Dani. The Yale language is the only

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 474 Volker Heeschen

one arnong the Mek languages to have developed a systern of articles frorn the deictic particles by adding morphemes rnarking singular, dual, and plural. Exarnples are: nimianeko 'the man (who is known or who has been rnentioned in previous discourse)', nimi abiniko 'the two men up there (who have already been rnentioned)', nimi danuko 'the men down there (who have already been rnentioned)'. The systern of local deictics in the Ang- guruk language is sirnilar, but adds a fourfold differentiation along the scale 'near - distant'. The grarnrnaticalization of the Yale systern is not due to the influence of Angguruk, however, which does not use the local deictics for the indication of 'definiteness'. Dani differentiates between 'here' and 'there', other grades of difference being denoted by directional verbs. Cornpanng the interrogative pronouns, one is ternpted to reconstruct a rnorpherne *in-,an- for 'who' and 'what' - cornpare arnong other possible cognates Awa anepo 'what', insebo 'who' (see R. and A. Loving 1975); T intap 'what', waántá, wantap 'who'. Larye una could be related to the former of these forrns, Birne ware to T for 'who'. However, as long as no proto-language can be established for any single language farnily, reconstructed forrns seem rather unreliable. In conclusion of this section I will exarnine the verba1 rnorphology and the linear order of the rnorphemes incorporated in the verb. If one sirnply takes al1 the possible positions of the Ok, Mek, and Dani verb into account, one arrives at the following formula.

Example 9 object stem aspect indirect auxiliary aspect object tense person pronoun (bene- object verb for pronoun number factive) pronoun aspect

Ok+ + + + - + - + + Dani + + - + + - - + + Mek - + - - - ++ ++

Exarnples are:

9a, T (from Healey 1965b:30) bok0 - b - 'neé - 1 - an- - a térn speakl bene- me punct- future he punc- fac- tiliar til- tivel iar punc- tiliar He will say to me.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Position of the Mek Languages of Irian Jaya

9b, Grand Valley Dani (from Bromley 198 1:1 14) WO - n - et - h - e give me auxiliaryt factitive 3sg. give to someone He gave to me.

9c, E leb - areb - nama - ki - n speak give future YOU l sg. I will speak to you (in your interest).

The morphemes in each position, especially those for tense, person, and number, are quite different. Besides, the Mek languages have inflections for the singular, dual, and plural in al1 three persons, Ok and Dani have no dual, and Ok differentiates between masculine and feminine in the third person singular. However, the first person singular suffix -n can be related to corresponding suffixes in other Trans-New Guinea phylum languages (Wurm 1982:32; Foley 1986:259). The Mek suffixes for the first person singular near past, -nwe and -no (compare example 6), seem to be related to what Foley (1986:259) has reconstructed as Proto Eastern Highland -u. In Telefol, the morphemes for the first person plural of sets A and B (Healey 1965a and b) are -úb and -ulúb; compare E -ab, -upe, and Y -ab and -u. Wurm (1982:32) suggests as proto-forms -Unb, -anb, -a(i), or -mb. In Angguruk, a morpheme -t- for the remote past and the counterfactual can be isolated, viz. pal-t-en 'cut-counterfactual-2sg.': 'if you had cut', while in Eipo we find -t- (Yale -s-) as the characteristic morpheme for the counterfactual (or potential mode), viz. yan-ma-t-eibye 'come-durative- potential-lpl./potential': 'if we had come'. Another correspondence is found in the morpheme -da- for reflexivity, for example Anggurukpal-ta- hv-n 'cut-reflexive-future-2sg.': 'you will cut yourself, Eipo kareb-da-n 'cut-reflexive- lsg./today's past': 'I have cut myself; compare Enga táng 'self (from Lang 1973). Wurm, quoting Haiman, points to the Trans-New Guinea phylum morpheme for dual number, which should be either the glottal stop or flapped /r/. The characteristic morpheme for dual number second and third person throughout the Mek languages is flapped -d- or -r-; compare Eipo ya-r-uk, Yale ya-d-ang 'come-dual-3pl.ltoday's past': 'the two have come'. The proto-form of the prefixed singular object markers can be taken to be na-, ka-, and a- (Wurm 1982:32; compare example 9). Mek non-terminal suffixes for direct and indirect object pronouns are lsg. -ne-, -ni-, 2sg. -k-,-ki-. I do not know whether Eipo *a- in a-reb- 'to give' and Bime and Tanime *ka- in ka-reb- 'to give' are a remote reflex of prefixed object pronouns: a loss of k- would not be regular. Dani prefixes these morphemes only to the verb for 'to beat', while the Ok languages have direct reflexes of the reconstructed forms. The introduction of a second verb meaning 'to give', as in example 9, in order

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 476 Volker Heeschen to make ditransitive constructions, seerns to be more typological in kind. The above is al1 the evidence I arn able to adduce for the inclusion of the Mek languages in the Trans-New Guinea phylurn. The evidence is scant, b~it,being systernatic, fully supports Wurm's classification.

Ty~olog~ The literature rnentions some typological features of the Trans-New Gui- nea phylurn languages which, though only rneant as an aid in character- izing these languages, are nevertheless regarded as indicators that may suggest, or direct the search for, genetic relationships. Among these features are classificatory verbs, media1 verbs, verb serialization, switch reference, and ergativity. So far there are only hints pointing in the direction of interdependence of these features, and, leaving aside the problern of whether a partial typology is possible, it should be said that each language makes a different use of these grammatica1 rneans and that these means, taken .as a whole, are structured and cornbined in different ways.

Class$catory verbs. Lang (1 975) distinguishes two types of classificatory verb: existential verbs and pro-verbs. The first type cornprises a set of verbs roughly meaning 'to be', but each of thern associated with, or attracting, a certain semantically defined set of nouns; thus nouns denoting round, srnall, or fernale objects rnay be cornbined with just one of the several existential verbs. This system yields covert noun classes. No trace of this verba1 type can be found in the Mek languages. The second type involves the combination of a noun (adjunct) with a verb with a rather general meaning, and is accordingly terrned 'pro-verb'. Three very rudirnentary exarnples of this kind of expression are found in Eipo.

Example 10 1. Nouns referring to some inner state are associated with the verb ub- 'to be, to become', for example elel ub-ma-l 'fear be-durative-3sg.Ipres.': 'fear is to hirn, he is afraid'. 2. Words for concrete objects, but also kanye 'soul', and words for parts of the body are associated with the verb bob- 'to carry' in this construc- tion; however, the meaning of this verb then is sornewhat weakened, and can be rendered as 'to have'. Exarnples are: ya bo-lam-le'stone adze carry-habit.-3sg.Ipres.': 'he has or owns a stone adze'; amol bolamle 'he has ears', rneaning 'he wil1 be able to understand what you are going to say'. 3. The verb leb- 'to speak' is associated with yupe 'word, language' and liman 'thunder', for example, yupe le-lam-le 'word speak-habit.-3sg.l pres.': 'he is speaking', liman lelamle 'it is thundering'.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Positwn of the Mek Languages of Irian Jaya 477

However, these constructions are by no means obligatory. That is why I speak only of 'association'.

Verb serialization. This term denotes a series ofiverbs governed by the Same subject and themselves usually governing the same object, which are usually not interrupted by any co-ordinating or subordinating connectives, and in which only the final verb is marked for tense, aspect, number, and person.

Example 11 . (E) teke di-lam-ak stand/inf./sim. look-habit.-3pl.lpres. They stand and look. (Y) de- - wadek- salel- ba-lam-ang eat- go around- grasp- go-dur.-3pl.lpres. They go around (an earth-oven), take this and that, and eat.

This conjoining or juxtaposing of ierbs is one of the favourite constructions in the Mek languages. It is a process which also serves several other grammatica1 purposes, as set out below. 1. Two or more verbs are compounded to form a new semantic unit (new 'word'), viz.:

Example 12 - (E) dol-bin~ma-n take-go-dur.- l sg.1pres. I fetch. ::

2. The meaning of one of the members of such a compound may become generalized, now being available for denoting aspect or aktionsart, as follows:

(Y) gekeb-lob-ma-l hear-let-dur.-3sg.lpres He does not listen carefully, he does not pay attention.

3. The compound may carry the notion of increased transitivity andl or causativeness, viz.:

Example 13 (E) me tek-eib-ma-l child stand-put-dur.-3sg.lpres. He makes the child stand.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 478 Volker Heeschen

4. Especially with verbs of movement as the second member of the compound, a notion of intention or consecutiveness is introduced, e.g.:

Example 14 (E) mal-bi-nu-n -sleep-go-fut.-lsg. I wil1 go in order to sleep.

The Mek languages make frequent use of al1 of these grammatica1 devices. To my knowledge, other languages restrict themselves to one or two of the possibilities provided by conjoining.

Medial verbs and ciause chaining. With the morphological marking of that rnember of a two-verb unit that is not marked for tense, person, and number, in combination with the appearance of connectives between the two members as wel1 as the introduction of new objects and of the notion of successiveness and simultaneity between the actions denoted by these members, the Mek languages seem to shift from pure verb serialization to clause chaining.11 Sirnultaneous actions are marked by -do in Y and by -nin or -de in E, as the following example shows.

Example 15 (Y) na do-do de-n I take-inflsim. eat- l sg.1pa.I I have taken (sweet potatoes) and have eaten. (E) buk-de mab-ma-l sit-inf./sim. sleep-dur.-3sg.lpres. He is sleeping while sitting.

The example below is one for successive actions.

Example 16 (E) ya dob-uka areb-m-uk adze take-inf./suc. give-dur.-3sg./pa.III He took the stone adze and gave it to him.

With the introduction of more than one object the chain remains tightly structured, as witness the example below.

11 The relationship between verb serialization, clause chaining and the occurrence of inde- pendent clauses in the grammar of a single given language is not always obvious; typo- logica1 comparison across languages makes such a relationship plausible, however; compare, for exarnple, Givón 1991.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Position of the Mek Languages of Irian Jaya

Example 17 (E) mem kongkonga leb-uka, YUF taboo word speak-inf./suc. cordyline yala koub-uka, yo si-tam tamub-uka seedling plant-inf.1suc. tree tip-side turn-inf.1suc. keb-ik. do-3pl.lpa.111 They uttered the taboo incantations, planted a cordyline seedling, twisted off the tips off a tree - that was what they did.

These participle-like sentence-media1 verba1 forms are characteristic of the Mek languages. Many other Papuan languages inflect them with a different set of person and number morphemes from that used for sentence-final verbs. In the Mek languages just one such set can be identified; it is used if the tense is present or future, and is rarely used without a connective.

Example 18 (E) dib-min-ye-ora, mab-lyam eat-2sg.lmed.-and-then, sleep-imper.12sg. If you have eaten, you may sleep.

Switch reference. Whenever in such tightly structured chains another subject is introduced, or whenever it is necessary to keep track of previously mentioned subjects (and sometimes objects), a considerable number of Papuan languages resort to a mechanism whereby it is indicated whether , or not the subject of the next clause will be the same. The Mek languages lack such an obligatory mechanism for indicating switch reference. How- ever, the participle-like forms of the verb in the examples above imply maintenance of the Same subject. The connective -ora also implies that no change of subject will occur. Nevertheless, two clauses may also be con- nected by -ora if a change of subject is evident from the use of fully inflected sentence-final verba1 forms or from a set of sentence-media1 forms inflected for person and number, as follows.

Example 19 (E) dib-ma-lam ora, na kwaning din eat-dur.-2sg.lpres. then, I sweet potato eat/inf./sim. ate ur-buk-nu-n because possibly-sit-fut.- lsg. I have eaten sweet potatoes, so while you are eating, I wil1 only sit.

(E) areb-min-ye-ora, areb-nam-ki-n give- 2sg.lmed.-and-then, give-fut.1far-you- l sg. If you give something to me, I will give sornething to you.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 480 Volker Heeschen

The connective -buk (Y -ba or -bok) usually implies switch reference, its original meaning being 'but' or 'though', thus implying a strong contrast.

Example 20 (E) areb-min-buk, gum areb-nama-ki-l . give-2sg.lmed.-though, not give-fut./immediate-you-3sg. Though you are giving something to him, he wil1 not return the gift.

In most cases the morphemes for person and number will indicate a change of subject. The prototype of clause-conjoining by means of -buk is found in sentences Iike the following, where the first clause refers to things not amenable to human control.

Example 21 (E) nakina taleb-ma-Ie-buk, gum ya-nam-Ie sickness grasp-dur.-3sg.lmed.-ds., not came-fut./far-3sg. Because he is sick, he will not come.

Only in some cases is -buk crucial to an understanding of a change of subject, as in the following.

Example 22 (E) ya-le-buk, bi-nam-Ie come-3sg.lmed.-ds., go-fut.1far-3sg. If this one cornes, the other one wil1 go.

Even so in such cases, favoured by grammarians, the fact that two different subjects are being referred to is most likely clear from the situational context or from the preceding discourse. There are two other constructions that draw the hearer's attention to the presence of a second subject. In the first of these the speaker introduces the second clause with a personal pronoun, as follows.

Example 23 (E) menteb-re u-lam-uk ora, el (Babyal) miss-inf./sim. be-habit.-3sg.Ipa.111 then, he (Babyal) marab-uk hit-3sg.Ipa.111 While one missed him, he (the other: Babyal) hit and wounded him

If the first clause postpones the postposition -ak 'at', the subject of the second clause is likely to be a different one, since two actions in the Same place are supposed to be performed by two actors.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Position of the Mek Languages of Irian Jaya 48 1

Example 24 (E) mab-ma-l-ak, bi-nam-le sleep-dur.-3sg.lpres.-at, go-fut./far-3sg. He wil1 go to the place where the other one is sleeping.

Thus, because of the presence of a multiplicity of constructions and of pragmatically used devices, the Mek languages have not developed an obligatory mode of switch reference. On the other hand, a brief look at other Papuan languages informs US that these modes are not at al1 uniform; nor are constructions such as serialization or verb compounding similar. We find here: 1. a complete range, with the type adjunct plus pro-verb occurring at one end and the type of verba1 compounds that are characteristic of the Mek languages at the other; 2. either serial verbs or adjunct plus pro-verb phrases; 3. sentence-media1 verba1 forms without any indication of switch reference (for Siroi, see Wells 1979, and Yessan-Mayo, Foreman 1974); 4. sentence-media1 verba1 forms with an added set of morphemes to indicate switch reference, as in:

Example 25, T (from Healey 1965b:30) baka - 'boom - 'bal - a ... speakl some time switch he ... durative After he had spoken for some time, another (did something else).

5. sentence-media1 verba1 forms which at the Same time indicate switch reference (Kâte, see Pilhofer 1933); 6. sentence-media1 verba1 forms which indicate switch reference by means of a special suffix and which anticipate the pronomina1 subject of the next clause through addition of one of a special set of anticipatory subject suffixes, as in:

Example 26, Fore (from Scott 1978:120) kana:gírá agause kana- a:'- ki- ta a- ka - US- e come helswitch connective we two it see we two indicative He is coming, and we both see it.

The rudimentary switch reference system in the Mek languages is most similar to that of the Dani languages. These suffix connectives to participle-like forms, which are then understood to refer to another subject.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 482 Volker Heeschen

Example 27, Grand Valley Dani (from Bromley 198 1:237) wesa ikho-kkolek nykky-laken taboo saying-whilelswitch eating-you.are While people say it is taboo, you are eating it.

Another similarity results from the postpositioning of -my, -ma, like Eipo and Yale -uk 'at', which serves as a means of subordinating and, from time to time, by implying a movement from one place to another, switch reference.

Example 28, Angguruk (from Zöllner s.a. (a)) jabuk wituk-mv jet-esehe garden working-at see-you/wellpl./pa. We saw them working in the garden.

Eqativity. The Mek languages, and quite a number of other Papuan languages besides, are classified among the ergative type (see, among others, Louwerse 1988:108). Judging from some grammatica1 sentences, the Eipo language seems to belong to this type.

Example 29 (E) Basam ban-ma-l ~ig go-dur.-3sg./pres. The pig is running. Baby al arye basam ob-ma-l Babyal agent pig kill-dur.-3sg.lpres. Babyal kills the pig.

The deviations from this system are so numerous that the classification of the Eipo language as an ergative one is irrelevant. In utterances made in real life, as opposed to elicited sentences, the following principles are dominant (for a complete overview see Heeschen 1990~): 1. The accusative type is quite prominent, as witness the following example:

Example 30 (E) sal sik kelape-ak winya-lam-ik men their women-at say-habit.-3pl./pa.III The men said it to their women.

2. The agent need not be marked if the patient is inanimate.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Position of the Mek Languages of Irian Jaya 483

Example 31 (E) Babyal yin dob-uk Babyal bow take-3sg.Ipa.111 Babyal took his bow.

3. If the agent of a transitive second clause is the Same as the agentlsubject of a first intransitive clause, there is no agent-marking.

Example 32 (E) Babyal bin-am-uk-e, basam ob-uk Babyal go-result.-3sg.Ipa.111-and, pig kill-3sg.Ipa.111 Babyal went there and killed a pig. 4. Dislocation and fronting of the theme override case-marking.

Example 33 (E) Babyal ara, Ningke marab-uk Babyal theme, Ningke wound-3sg.Ipa.111 It was Babyal who wounded Ningke. Thus, case-marking is handled pragmatically. Furthermore, the ideal grammatica1 unit is one verb plus one noun phrase. Restricting the number of noun phrases in a clause renders case-marking superfluous. But this brings US to the last province which helps to explain the common features of the Papuan languages.

Speech I have demonstrated how ditransitive clauses are inevitably broken down into more manageable units through right dislocation, theme-rheme structure, verb serialization and repetition (Heeschen 1990~).Though Dubois' research (1987) is focused primarily on case-marking in ergative-type languages, his conclusions are nevertheless in line with my analysis: discourse principles override or influence case sytems, or, to generalize, the preferred information structure in ongoing discourse is maximally one lexica1 argument in one clause (Lawrence 1972; Bisang 1986). I tried to show that the underlying discourse principle in the grammar of Eipo is to give only one piece of new information in each unit. This principle tran- scends the structure of verb plus noun phrase, and is also reflected in noun phrases and predicative structures. Below follow three examples illustra- ting this principle. In the first one we see a series of three noun phrases. In real utterances these are never linked together in the ideal structure of a ditransitive sentence (such as 'he has given sweet potatoes to his friend'), if 'ditransitive' is taken as a descriptive term for superficial structures. One may say that they are distributed over three clauses according to the above- mentioned discourse principle through repetition, verb serialization and retrievability of information from the preceding sentence. .

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 484 Volker Heeschen

Example 34 (E) kwaning yongob-uk. Yongob-uk-e, el neikye sweet potato dig-3sg.Ipa.111 dig-3sg.Ipa.111-and he fnend y ane buk-arn-le-ak areble cornelinf. sit-result.-3sg.Ipa.I-at give-3sg.Ipa.I He dug up sweet potatoes. After that he gave them to his friend where the latter had come and sat down.

In the second exarnple subject and object are distributed over two clauses.

Example 35 (E) boblan a-wek-m-ak anYe, kwitnang door here-go-dur.-3pl.lpres. here, initiates a-wek-m-ak anye ... here-go-dur.-3pl.lpres. here ... Here it is that they disappeared through the door, the initiates disappeared here.

In the third example, the principal noun of a noun phrase is repeated. Such a step-by-step process of adding information occurs in predicate structures as wel1 as in noun phrases with more than one attribute.12

Example 36 (E) yurn ara, el si kuwul si tonok first theme, he name uninitiated name alone Previously (before his initiation) his name was that of one who is not initiated.

If we really take discourse principles as our starting-point, the typological features of the Papuan languages, I believe, allow of a definition of the system as one in which each component part depends on, and supports, the next. So verb serialization is an excellent rneans of keeping down the nurnber of noun phrases or argument roles in a single clause. Verba1 cornpounds, resulting frorn the conjoining of two verbs, tend to make for an increase in the nurnber of noun phrases in a given clause through the creation of transitive verbs. Increased transitivity runs counter to the discourse principle, however. Thus verb-compounding either yields new semantic notions (lexicalization) or produces a shift towards the rnarking of aspect, aktionsart or other verba1 features (grarnrnaticalization). On the other hand, the discourse principle renders rigorous case-rnarking superfluous (cornpare especially exarnples 32,33, and 34 above); wherever two noun phrases occur in one clause, case-rnarking is effected in accordance with information from the preceding discourse, along the scale of anirnacy, and by reliance on the principle that the therne introduced at the beginning of a piece of discourse is the agent of al1 the following actions specified

'2 Feldman (1986:130) coined the term 'nomina1 senalization' for this.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Position of the Mek Languages of Irian Jaya 485 in it unless there are explicit indications to the contrary. The problems which still remain and which the speakers have to cope with while plan- ning their utterances, result from the sequencing of actions and predica- tions, as well as from the difficulty of introducing a new agent or subject. The solutions to this which the speakers have come up with and are still finding are the use of sentence-media1 verbs and switch reference. But co- occurrence of this type of solution does not necessarily imply a genetic relationship between the individual languages concerned. It can represent a typological feature only if the interdependence of the relevant parts of these diverse grammars can be proven. This overall picture of the diffusion of seemingly identical solutions to problems of building up a coherent discourse segment may be corroborated by two genera1 arguments. 1. Verb serialization as well as switch reference seem to be area1 features. That is to say, contacts between the speakers of different languages and the possibility of language mixing must be assumed. Verb serialization was first described in grammars of Niger-Congo languages. Some of the Austronesian languages along the coasts of New Guinea and on the surrounding islands also make use of this economical device. Switch reference is found in high language density areas, in Centra1 Australian languages, Caucasian languages, and the Indian languages of Califor- nia (Austin 198 1; Stahlke 1973). Area1 features presuppose borrowing, simplification and the grammatica1 creation in discourse. Thus we are brought back to our initia1 assumptions concerning the history of Pa- puan speech communities. 2. Plank (1987:178), though discussing only nominal morphology, may be offering another genera1 argument where he says: 'Analytic languages and languages with agglutinative nominal morphology prefer op- tional marking, viz. economy of use'. If we assume specific phases of relexification and grammatica1 restructuring in the history of Papuan languages, and further that the resultant morphology was not opaque, but was subject to the principles of 'economy of use' as well as more or less to conscious manipulation, the implication is that the kind of solutions hit on by Papuan speakers in dealing with the intrinsic problems of speech are confined to particular phases in the history of language. Apparently, the conditions for these phases were present again and again in the history of Papuan languages. The superficial unity between the Papuan languages may not be due to any distant genetic relationships, but be the direct outcome of a situation in which typological and area1 features coincide, as far as they are explainable by basic discourse principles and by similarity in the type of their basic grammatica1 units, which have to be incorporated in discourse. The similarity between the Mek languages and other Trans-New Guinea phylum languages may be the outcome more of the psychology of speaking than of abstract typological principles or genetic relationships. Nevertheless, the segments which the speakers have to assemble

Downloaded from Brill.com10/02/2021 03:38:53PM via free access 486 Volker Heeschen

come from the Same sources and are an outcome of the speech situation which I described at the beginning of this paper.

ABBREVIATIONS AND SYMBOLS USED - indicates the break between bound morphemes 1 divides different kinds of grammatica1 information concerning the given morpheme in the Eipo or Yale text 1,2,3 indicates first, second, and third person respectively ANU Australian National University d. dual ds. . different subject dur. durative fut. future habit. habituative imper. imperative mode inf. infinitive (verba1 form) KITLV Koninklijk Instituut voor Taal-, Land- en Volkenkunde, Leiden med. sentence-media1 verb pa. I, 11, I11 today's past, near past, remote past respectively pl. plural PL Pacific Linguistics result. resultative sg. singular sim. simultaneous ss. Same subject SUC. successive

REFERENCES Austin, Peter, 1981, 'Switch referente in Australia', Language 57:309-334. Bee, Darlene L., 1973, 'Cornparative and historica1 problerns in East New Guinea Highland languages', in: H. McKaughan (ed.), The languages of rhe Eizsrern Family of rhe ktNew GuUiea Highland Stock, pp. 739-768, Seattle and London: University of Washington Press. Bisang, Walter, 1986, 'Die Verb-Serialisierung im Yabêm', Lingua 70: 131 - 162. Brornley, H. Myron, 1966-67, 'The linguistic relationships of Grand Valley Dani', Oceania 37:286-308. -, 1973, 'Ethnic groups in Irian Jaya', Irian 2-3:l-37. -, 1981, A grammar of Lower Grand Valley Dani, Canberra: ANU. [PL, C-63.1 Capell, Arthur, 1976, 'Austronesian and Papuan "mixed languages: General rernarks', in: S. A. Wurm (ed.), New Guinea area languages and languagestudy, Volume 2, Austronesian langmges, pp. 527-579, Canberra: ANU. [PL, C-39.1 Dubois, John W., 1987, 'The discourse basis of ergativity', Language 63:805-855. Feldrnan, Harry, 1986, A grammar of Awtuw, Canberra: ANU. [PL, B-94.1 Foley, Williarn D,,1986, The Papuan langmges of New Guinea, Carnbndge: Carnbridge University Press.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access The Position of the Mek Languages of Irian Jaya 487

Foreman, Velma M., 1974, Grammar of Yessan-Mayo, Ukarumpa: Summer Institute of Linguistics. Forge, Anthony, 1972, 'Normative factors in the settlement size of neolithic cultivators', in: P. J. Ucko, R. Tnngham and G. W. Dimbleby (eds), Man, settlement, and urbanism, pp. 363-376, London: Duckworth. Givón, Talmy, 1991, 'Some substantive issues concerning verb senalization: Grammatical vs. cognitive packaging', in: Claire Lefebvre (ed.), Serial verbs; Grammatical comparative and cognitive approaches, pp. 137- 184, Amsterdam: Benjamins. Haiman, John, 1987, '0n some ongins of media1 verb morphology in Papuan languages', Studies in Language 11:347-364. Haiman, John, and Pamela Munro (eds), 1983, Switch-reference and universal grammar; Proceedings of a Symposium on Switch-Reference and Universal Grammar, Winnipeg,May 1981, Amsterdam: Benjamins. Healey, Phyllis, 1965a, 'Telefol clause structure', PL, A-51-26, -, 1965b, 'Telefol verb phrases', PL, A-5:27-51. -, 1965c, Telefol noun phrases, Canberra: ANU. [PL, B-4.1 Healey, Phyllis, and Alan Healey, 1977, Telefol dictionary, Canberra: ANU. [PL, C-46.1 Heeschen, Volker, 1978, 'The Mek languages of Inan Jaya with special reference to the Eipo language', Irian 7-2:3-46. -, 1984, 'Intuitionen; Grammatische Gesprache in nichtakkulturierten Sprachgemeinschaf- ten', Linguisrische Berichte 94:27-44. -, 1985, 'Die Yale-Sprache; Eine Papua-Sprache', Studium Linguistik 16:35-45. -, 1990a, 'Die ersten Stunden: Am Ursprung der Sprachwissenschaft', in: H.-J. Niederehe and K. Koemer (eds), History and historiography of linguistics; Papersfrom the Fourfh Internarional Conference on the Hisrory of the Language Sciences (ICHoLs IV), Trier, 24- 28 August 1987, Volume I, pp. 33-47, Amsterdam: Benjamins. -, 1990b, Ninye bun; Mythen, Erzählungen, Lieder und Märchen der Eipo im zentralen Bergland von Irian Jaya (West-Neuguinea), Indonesien, Berlin: Reimer. -, 1990c, 'Injunktive Grammatik', in: G. Rickheit and S. Wichter (eds), Dialog; Festschrift fur Siegfried Grosse, pp. 129-147, Tubingen: Niemeyer. -, 1992, A dictionary of the Yale (Kosarek-) language (with sketch of grammar and English index), Berlin: Reimer. Heeschen, Volker, and Wulf Schiefenhövel, 1983, Wörterbuch der Eipo-Sprache; Eipo - Deutsch - Englisch, Berlin: Reimer. Hylkema, S., 1969, Woordenldst opgenomen te Apmisibil. [Unpublished manuscript.] -, 1974, Mannen in her draagnet Mens- en wereldbeeld van de Nalum (Sterrengebergte), The Hague: Nijhoff. [KITLV, Verhandelingen 67.1 -, s.a., Untitled manuscript on the Grammar of the Nalum language. Kock, M. A. de, 1912, 'Eenige ethnologische en anthropologische gegevens omtrent een dwergstam in het bergland van Zuid Nieuw-Guinea', Tijdschrift van het Koninklijk Ne- derlands Aardrijkskundig Genootschap 29:387-400. Lang, Adrianne, 1973, Enga dictionary, Canberra: ANU. [PL, C-20.1 -, 1975, The semantics of classificatory verbs in Enga (and other Papua New Guinea languages), Canberra: ANU. [PL, B-39.1 Lawrence, Marshall, 1972, 'Oksapmin sentence structure', PL, A-34:17-46. Laycock, Don C., 1982, 'Melanesian linguistic diversity: A Melanesian choice?', in: R.J. May and H. Nelson (eds), Melanesia; Beyond diversity, vol. I, pp. 33-38, Canberra: ANU. Louwerse, Jan, 1978, 'A tentative Una phonology', Irian 7-3:43-90.

Downloaded from Brill.com10/02/2021 03:38:53PM via free access Volker Heeschen

Louwerse, John, 1988, The morphosyntax of Una in relation to discourse structure; A descriptive analysis, Canberra: ANU. [PL, B- 100.1 Loving, Richard, and Aretta Loving, 1975, Awa dictionary, Canberra: ANU. [PL, C-30.1 McElhanon, Kenneth A., and C. L. Voorhoeve, 1970, The Trans-New Guinea phylum: Explorations in deep-level genetic relationships, Canberra: ANU. [PL, B- 16.1 Pilhofer, G,, 1933, Grammatik der Kâte-Sprache in Neuguinea, Berlin: Reimer. Plank, Frans, 1987, 'Number neutralization in Old English: Failure of functionalism?', in: W. F. Koopmann et al. (eds), Explanation and linguistic change, pp. 177-238, Amsterdam: Benjamins. Ramsey, Evelyn M., 1975, Middle Wahgidictionary, Mount Hagen: Church of the Nazarene. Sankoff, Gillian, 1977, 'Multilingualism in Papua New Guinea', in: S. A. Wurm (ed.), New Guinea area languages and language study, Vol 3, Language, culture, society, and the modem world, Fascicle 1, pp. 265-307, Canberra: ANU. [PL, C-40.1 Scott, Graham, 1978, The Fore language of Papua New GuUiea, Canberra: ANU. [PL, B-47.1 Silzer, Peter J., and Helja Heikkinen, 1984, 'Index of Irian Jaya languages', Irian 12:l-24. Sims, Andrew, and Anne Sims, 1982, 'Ketengban phonology', Workpapers in Indonesian Linguistics 1:50-74. Stahlke, Herbert, 1973, 'Serial verbs', Studies in Afncan Linguktics 1:60-99. Thurston, William R., 1982, A cornparative study in Anêm and Lusi, Canberra: ANU. [PL, B-83.1 -, 1987, Processes of change in the languages of North-western New Bntain, Canberra: ANU. [PL, B-99.1 Voorhoeve, C. L., 1975, Languagesof lrian Jaya: Checklist,preliminary classijìcation, language maps, wordlists, Canberra: ANU. [PL, B-3 l.] Wells, Margaret, 1979, Siroi grammar, Canberra: ANU. [PL, B-5 l.] Wurm, Stephen A., 1982, Papuan languages of Oceania, Tubingen: Narr. Zöllner, Siegfned, s.a. (a), 'Verbformen der Angguruk-Sprache'. [Unpublished manuscript.] , -, s.a. (b), Wörterbuch Angguruk - Deutsch. [Unpublished manuscript.]

Downloaded from Brill.com10/02/2021 03:38:53PM via free access