Tani languages

CHAPTER FIFTEEN TANI LANGUAGES Mark W. Post and Jackson T.-S. Sun 1 INTRODUCTION Tani refers to a compact cluster of Tibeto-Burman languages situated at the eastern end of the Himalayas, in a primarily mountainous area skirted on four sides by Bhutan, Tibet, Burma and the Brahmaputra River in Assam. The main concentration of Tani languages covers the central part of the Indian state of Arunachal Pradesh, comprising the bulk of the following districts: East Kameng, Upper Subansiri, Lower Subansiri, Upper Siang, West Siang, East Siang, Lower Siang and Dibang Valley. Tani languages are spoken by populations who either are currently or at one time were ‘tribally’ identified as Apatani, Nyishi, Na, Bangni, Tagin, Galo, Hill Miri or Adi (various subtribes), in addition to numerous smaller, usually clan-based or village-based self-identifications (such as Komkar). The single largest population of Tani language speakers, however, is the Mising tribe of upper Assam. The Tani languages are spoken in a basically continuous area, bordered in the west by Miji, Puroik, Koro and Hruso speakers, in the east by Idu speakers, in the south by speakers of Boro and Assamese, and in the north by speakers of Tibetic languages. Lingua francas spoken in the Tani area include Assamese in the south and Arunachali Hindi in the central area. Tibetic languages were used as lingua francas in earlier decades, although their use has recently waned. The 2011 Census of India reports a total of 1,380,878 individuals in Arunachal Pradesh and Assam who self-identified as belonging to a tribe whose mother tongue is a Tani language (Census of India 2011): these included 631,042 Tani tribespeople in Arunachal Pradesh, and 680,424 Mising in Assam (plus an additional 7,412 Mising in Arunachal Pradesh). These figures must be treated with caution, as it is not the case that all people who align tribally with a primarily Tani language-speaking population speak this language themselves; given the rapid spread of Hindi throughout Arunachal Pradesh (Modi 2005), and of Assamese among the Mising in Assam (Pegu 2011: 157), the number of fluent Tani language speakers as a percentage of this total must be considerably lower. Scattered Tani communities spill over the Sino-Indian border into adjacent areas in Motuo (Miguba and Misinba tribes), Milin (Bokar and Tagin tribes), and Longzi (Bengni, Na, Bayi, Dazu and Mara tribes) counties of Tibet (Ou-Yang 1985: 76), where they are lumped with certain linguistically non-Tani peoples (e.g. the Idu, Sulung and Bangru) to form the Luoba nationality. Very little current information regarding these groups is available (although see Huber 2012). Tani languages vary greatly in terms of number of speakers, adequacy of description, and degree of endangerment. Mising has perhaps half a million speakers, but lacks a modern, comprehensive dictionary and grammar, and is currently undergoing considerable retraction under the influence of Assamese. On the other hand, Tangam has only 150 speakers; however, it has a forthcoming modern (if relatively slight) comprehensive description, and all Tangam children seem – for the time being, at least – to be learning Tangam as a first language and speaking it fluently. Relatively urgent descriptive priorities include TANI LANGUAGES 323 Nyishi, a large and important language about which very little current information of certain reliability is available, Apatani, a relatively divergent language with a robust and intricate tone system which lacks a comprehensive and reliable description,1 as well as all Tani varieties still spoken in Tibet. 2 GENEALOGY AND SUBGROUPING Tani languages appear to constitute a distinct subgroup within Tibeto-Burman, as argued in Sun (1993). All Tani languages shown on the family tree in Figure 15.1 share a large amount of common core vocabulary, exhibit a very large number of regular sound correspondences, and in general support firm reconstruction of the phonology and vocabulary of their common proto-language ‘Proto-Tani’. Koro and Milang, spoken on the western and eastern fringes of the Tani area, show many similarities to Tani languages, and Milang in particular has previously been included within the Tani subgroup. Post and Modi (2011) argued against the inclusion of Milang in Tani proper, showing that the strongest resemblances to Tani languages were undoubtedly due to contact with Padam Adi. Around the same time, Post and Blench (2011) placed Koro and Milang in a distinct ‘Siangic’ subgroup. In addition to Siangic languages, Tani languages show many lexical and grammatical resemblances to the easterly Idu-Tawrã languages, and to Tawrã in particular. Unfortunately, however, the current state of description does not enable us to determine whether these resemblances are due to common inheritance or to earlier language contact. Sun (1993) divided Tani languages into two major branches, on phonological and lexical grounds: Western Tani (WT), clustered around the Subansiri and Kameng Rivers, and Eastern Tani (ET), clustered around the Siang/Brahmaputra. The WT/ET division almost Proto-Tani Idu-Tawrã Koro Western Tani Eastern Tani Tangam, Damu Apatani Subansiri Pailibo Bori Karko, Shimong Nyishi-Hill Miri-Lower Tagin Siang Bokar, Ramo Minyong Na-BangniUpper Tagin Milang Galo Pasi-Padam Mising (= “Lower Adi”) Central contact zone FIGURE 15.1 PROVISIONAL FAMILY TREE OF THE TANI LANGUAGES (REVISION OF SUN 1993: 297) 324 TIBETO-BURMAN: LANGUAGES OF NE INDIA certainly has at least some sort of genealogical basis as Sun proposed; however, language contact has played an at least equally powerful role in the evolution of Tani languages, and has led to convergence and shared innovation both within and across genealogical branches. Post (2015) reviewed 15 Tani phonological innovations exhibited by 15 Tani languages, and found that no two high-level innovations identified the same set of genealogical branches. In general then, it is best to view the Tani languages as a dialect continuum, with a reconstructible shared ancestor, but with a subsequent evolution marked at least as much by areal sharing of innovations as by sharp genealogical branching. Figure 15.1 provides a heuristic Tani family tree on the basis of what might be called ‘aggregate sharing of innovations’. Depending on which of these innovations are attributed to genealogical inheritance, and which are attributed to contact, a large number of alternative trees could be drawn (particularly with respect to geographically central Tani languages). 3 PHONOLOGY Syllable and word are both important organizational units in Tani phonology. The ProtoTani (PT) syllable canon is (Ci)(G)V(X), in which Ci is an optional initial consonant, V is an obligatory nucleus, G is an optional approximant r or j, and X is an optional coda, either a nucleus-identical V or a final consonant Cf (generally drawn from a relatively restricted set). In at least one language – Apatani – the X constituent may also be represented as nasalization over the nuclear vowel. Tani syllables with an X constituent are bimoraic, or heavy; syllables without an X constituent are monomoraic, or light. Many phonological processes in Tani languages depend on this weight distinction, including syncope and prominence assignment in Galo (Post 2007: §4.2.3), and the realization of tones in Apatani (Post and Tage 2013). The PT syllable canon is well-preserved by the modern languages, albeit in different ways; generally speaking, onsets (including clusters) are better preserved in WT languages, while ET languages are more conservative with rhymes. Tani segment inventories are relatively simple. Consonants are found at labial, alveolar, (alveolo-)palatal, velar and glottal places of articulation (Table 15.1). Dental and retroflex consonants are not found in Tani languages – nor indeed are they found in either the Arunachali variety of Hindi or in the neighbouring Eastern Indo-Aryan language Assamese; occasional transcriptions of dentals and retroflexes in the sources would accordingly seem to be errors. All Tani languages contrast voiceless and voiced plosives. Affricates pattern with plosives in languages which retain them; in Siang-area languages such as Pugo Galo and Lower Adi, tɕ historically merges to ɕ and thence s or h. PT contrasted voiced and voiceless fricatives, but this distinction is maintained by very few modern languages other than Na/Bangni; in most modern languages, fricatives have merged to a two-way distinction between a glottal h, and a medial fricative, either the conservative ɕ (mostly in northerly lects) or the innovative, Indo-Aryan-influenced s (towards the south). A small number of languages, including Apatani, have also innovated a velar fricative x. A breathy/glottal onset distinction among vowel-initial words is found in some languages, with the breathy onset ʱ- reconstructible to a subset of PT fricatives, and the glottal onset ˀ- reflecting vowel-initial morphemes at the PT stage; for example, compare Northwestern Galo ʱáa‘come; set (sun)’ (< PT *vaŋ, cf. Bangni vaa-) with ˀàp- ‘shoot’ (< PT *(ˀ )ap, cf. Bangni ap-). Neither onset is found word-medially: compare the cognate first and second formatives in Northwestern Galo ʱag-jàa ‘fermented soya bean’ and pej-àk ‘soya bean’. Other Tani languages, such as lower dialects of Lare Galo and Lower Adi, have merged these categories completely, retaining only a single fricative s or h. TANI LANGUAGES 325 TABLE 15.1 PROTO-TANI CONSONANT INVENTORY, BASED ON SUN (1993: 56) Plosive [-VD] Plosive [+VD] Nasal Fricative [-VD] Fricative [+VD] Lateral Rhotic Semivowel Labial Alveolar (Alveolo-)Palatal Velar Glottal *p *b *m *f *v *t *d *n *tɕ *dʑ *ɲ *ɕ *ʑ *k *g *ŋ (*ˀ) *h *ɦ *l *r *j TABLE 15.2 PROTO-TANI VOWEL INVENTORY, BASED ON SUN (1993: 67) High Mid Low Front Central Back *i *e *ɨ *ə *a *u *o Most Tani languages preserve the PT seven vowel inventory (Table 15.2); although Apatani split and merged PT *ə to the polar vowels (e.g. Apatani já-mù ‘fire’ < PT *mə ‘fire’), the same vowel ə has re-emerged due to a vowel centralization rule in the environment of rhotics (e.g. Apatani tà-də́r ‘worm’ < PT *dor ‘worm’). An unusually large ten vowel system has been reported by Ou-Yang (1985: 77) for the Bodic-influenced and possibly mixed language Damu: a, i, u, e, o, ə, ɨ, y, ø, ɿ (= syllabic z); this report remains unconfirmed. Although most instances of contrastive vowel length in underlying forms of morphemes may be secondary, vowel length is a robust phonological feature found in nearly all Tani languages:2 compare Galo banám ‘to vomit’ and baanám ‘to move the head’. While contrastive word-medial vowel length is usually easy to detect, underlying wordfinal vowel length is more difficult to detect in context-free utterances, and is absent from the majority of descriptions. However, word-final vowel length can typically be discerned in phrase-medial contexts: for example, Galo aló gó ‘some salt’ vs alóo gó ‘a day’. True diphthongs are largely absent from the underlying forms of PT morphemes; diphthongs in the modern languages are typically either recent loans or contracted compounds. Tani languages vary greatly in their inventories of syllable codas. The language that preserves the largest number of the original proto-codas is (Adi) Padam, with -p, -t, -k, -m, -n, -ŋ, -r, and -l. Milang is the only other language documented that exhibits an -l coda, albeit usually in Padam loans. The velar nasal *-ŋ and the dental/alveolar stop *-t codas are most prone to loss, especially in WT. The tendency towards coda attrition is epitomized in Apatani, where only two PT codas remain: -ʔ (from the original stop codas) and -r. New final consonants, including -s, -l, and -m, have been imported via recent Indo-European loans, such as pas ‘five’ (< Indo-Aryan). In Nyishi, Galo and Hill Miri, non-original codas are created via the clipping of short final vowels from original disyllables, leading to an innovative coda voicing contrast: e.g. Nishi ab ‘father’, tap ‘pumpkin’ (Chhangte 1992: 4); cf. Bengni a-buu ‘father’, ta-pɨɨ ‘pumpkin’. Phonological words are minimally bimoraic, and tend strongly to be disyllabic in Tani; among languages in our Tani database for which more than 1,000 lemma have been 326 TIBETO-BURMAN: LANGUAGES OF NE INDIA recorded, no language has more than 0.5 per cent morphologically simplex, monosyllabic lexical words. When such words occur, they seem to be idiosyncratic retentions from a basically monosyllabic proto-language; for example, Galo ɲíi ‘person’ but Minyong ˀa-mí ‘person’ and Apatani mí-yù ‘person’. Rhythm tends strongly to be trochaic in Tani, and has a number of phonological outcomes, including coda-reductions and syncope processes, post-head clitic incorporation, and perhaps even the overall disyllabification of the basically monosyllabic proto-lexicon (Post 2006, 2011b). Phonemic tone is found in most relatively well-described Tani languages, including Apatani (all dialects), Galo (all dialects), Na/Bangni, Upper Minyong and Tangam, as well as in Milang. Generally speaking, tones are more robust and have a higher functional load in WT languages than in ET languages; however, it would appear that both tone and non-tone languages may be found in each branch. Analysis of tone in Tani languages requires reference to both the surface phonological word and its underlying morpheme constituents. In the case of underlying morphemes, no more than two contrastive tones have been attested for any Tani language: an unmarked H and a marked L. These ‘underlying tone-bearing units (TBUs)’ then interact with syllable and word structure to produce a surface word tone contour, rules for the derivation of which differ from language to language. In Galo, a ‘pitch peak’ over an L word is attracted to heavy syllables, thus interacting with rhythmic processes; for example, r ̀ -dùu ‘do-IPFV’ > [ˈr .́ dù] but r ̀ -dùu-kú ‘do-IPFV-COMPL’ > [r .́ ˈdûu.kù] (here and elsewhere in this chapter, á is a (mid-)high tone, à is a low tone, and â is a (rising-)high-falling tone). In Apatani, two underlying H and L tones interact with the structure of word-final syllables to produce three disyllabic word tone contours: HH or HL for words with final light syllables (depending on the underlying specifications of constituent morphemes), and LH for words with final heavy syllables (independent of morpheme constituency). Underlying tones of words with final heavy syllables in Apatani can be discerned via tone spreading to following articles or suffixes. For example: ˀámí (kê) ‘(an) elder sister’, ˀámì (kè) ‘(a) tail’, ˀàpı́ ̃ (kê) ‘(some) cooked rice’, ˀàmı́ ̃ (kè) ‘(a) name’. At least some Tani languages or dialects thereof seemingly lack lexical tones, including Lower Adi, Mising and Nyishi. Unfortunately, due to an ongoing lack of adequate description (particularly of ET languages), reconstruction of PT tone categories is problematic. Phonological and morphophonological alternations are extensive in Tani, particularly syllable-structure adjustments or assimilation. The rich allomorphy of the nominal prefix *a- in Bokar provides a typical example: a-ŋaa ‘child’; i-kii ‘dog’; u-puk ‘arrow’; e-tɕe ‘clothes’; o-ŋoo ‘fish’; ə-jək ‘pig’; ɨ-lɨŋ ‘stone’. Morpheme identification can be tricky without knowing the sandhi processes at work. For example, given Bokar luɣin ‘fingernail’, four morphophonemic rules have to be undone to retrieve the original shapes of the component morphemes lok- ‘hand’ and jin- ‘nail’: namely, initial j- simplification (→ lok-in), obstruent voicing assimilation (→ log-in), spirantization (→ loɣ-in), and vowel assimilation (→ luɣ-in), compare Bokar a-lok ‘hand’, lə-jin ‘bird’s claw’ (compare also Bengni lak-sin ‘fingernail’, from PT *lak-ʑin). In general, morphophonological processes are more intricate among WT languages than among ET languages, perhaps reaching an apex in Tangam. 4 MORPHOLOGY 4.1 Diachrony Tani morphology is best understood against the backdrop of its evolution. Most Tani morphemes are reconstructible as simplex monosyllables, and presumably reflect simplex, TANI LANGUAGES 327 monosyllabic lexical words at some early stage (whether PT or earlier). Similarly, most Tani grammatical morphemes are reconstructible to earlier lexical roots; for example, most tense-aspect morphemes reconstruct to an earlier set of clause-final existential or auxiliary verbs (e.g. Galo -dùu ‘IPFV’ < PT *duŋ ‘sit; be at (sitting)’, or Apatani -dàʔ ‘COS’ < PT *dak ‘stand; be at (standing)’), and most predicate derivations originated as constituents of earlier serial verb constructions (e.g. Minyong laa-bi ‘take-BEN’ < PT *laŋ-bi ‘take-give’). The modern languages are synthetic and agglutinating to varying degrees; most nouns and adjectives are disyllabic compounds (or further compounds thereof), while earlier serial verb constructions have amalgamated into a single, morphologically complex and expansive predicate word (1). (1) Galo tolo tɕàa-lèn-l ̀ ɨ-máa-dùu < *tə *lo *tɕaŋ *len *lɨŋ DST.UP.LOC ascend-OUT-DESD-NEG-IPFV DST.UP LOC ascend exit want ‘(I) don’t want to go up there.’ *maŋ *duŋ not.(have) sit/stay 4.2 Word classes PT appears to have had nouns and verbs, together with classifiers, postpositions, demonstratives, articles, conjunctions and particles. The modern languages have nouns, most often disyllabic prefixations or compounds of earlier nouns, adjectives, which are nounlike in structure and often derive from deverbal nominalizations, as well as classifiers, postpositions, demonstratives, articles, ideophones and particles. Adverbs are not well-represented in either the early or the modern languages; manner adverbials in the modern languages are typically derived from adjectives via a postposition or suffix *pə (prob. < PTB nominalizer *pa), as in Lower Adi kampo pə ‘beautiful AVZR’ ‘nicely; beautifully’, while temporal expressions occur as oblique noun phrases, such as Lower Adi jumə lo ‘in (the) evening’. A few particles and predicate derivations also have adverb-like meanings and/or functions, such as the Adi ‘versatile’ particle ruŋ ‘certainly; definitely’ or the Adi predicate derivation -man ‘playfully’. Verbs in the modern languages are best discussed in terms of predicate structure (section 4.3.2). Nouns are most easily distinguished by their ability to take articles, as in Adi ami ko, Apatani míyù kè (person IND) ‘a person’. Verb roots must be nominalized to be used in referential phrases, as in Adi do-nam ko (eat-NZR IND) ‘some food’. Conversely, nouns cannot usually take predicate morphology; Galo *ɲíi-dùu ‘person-IPFV’ is unacceptable. In many languages, adjectives can be distinguished from verb roots and nouns by their ability to occur in both an article-derived copula construction, like nouns, and in an inflected predicate, like verb roots (2). (2) Lower Adi apeŋ=ə! ŋo apee-duŋ. tired-IPFV tired=COP 1.SG ‘How tired (I am)!’; ‘I’m tired.’ Singular, dual and plural pronouns occur in three persons in the majority of Tani languages (Milang has an additional clusivity distinction, seemingly absent from Tani proper). However, only the first and second person forms appear to be reconstructible; third person forms often reflect earlier *ba-ɦɨ, being a composition of *ɦɨ ‘self’ with an 328 TIBETO-BURMAN: LANGUAGES OF NE INDIA TABLE 15.3 TANI FIRST AND SECOND PERSON PRONOUNS 1 2 SG DL PL *ŋo ‘I’ *no ‘you’ *ŋo ɲi ‘I + two’ *no ɲi ‘you + two’ *ŋo luŋ ‘I + group’ *no luŋ ‘you + group’ unknown morpheme (as in Pailibo maɨ ‘3.SG’), or a demonstrative *ɦə (as in Tagin ʱə ‘3.SG’) or else ko as in Bokar (possibly < *ko ‘one’). Tangam has an especially unusual third person singular form nodɨ – the second formative in fact cognate with *ɦɨ ‘self’, but the first formative oddly homophonous with the second person singular no, and different from the 3DL and 3PL forms daɲi and datəŋ (Table 15.3). Classifiers seem to be a PT innovation, not found in Milang (Post and Modi 2011). PT classifiers are monosyllabic and simplex: e.g. *tɨŋ ‘CLF: GROUP (OF ANIMALS)’, *soŋ ‘CLF: LONG, SLENDER’, *buŋ ‘CLF: LONG, HOLLOW’ and *tak ‘CLF: FRAGMENT OF SPHERE’. Modern Tani languages tend to have dozens of classifiers; in some languages, classifier roots are prefixed in their citation form, as in Galo ˀa-búu ‘CLF: LONG, HOLLOW’, or else compounded to numeral roots or a small number of adjectival roots, as in Galo búu-ɲí ‘CLF: LONG, HOLLOW-two’ or búu-tə̀ ‘CLF: LONG, HOLLOW-big’. In other languages, such as Apatani, classifier roots only occur in compounded forms. In Bokar, classifiers are reported to not occur with numerals higher than one (Sun et al. 1980: 128). Notably, all known Tani classifier systems lack a generic classifier (comparable to Mandarin ge), as well as a classifier for human beings. Instead, humans and unclassifiables are enumerated without the help of a classifier. Particles are a broad category, expressing a host of modal, evidential and illocutionaryforce meanings. In general, inventories are much larger in WT languages than in ET, sometimes by a factor of six or seven. Some of the best attested Tani particles are *ju(kə) ‘REP’, *dɨ ‘WONDER’, *pə ‘INFR’ and *la ‘CQ’; typically, particles follow the focus of a clause, occurring clause-finally in the majority of cases, or else after a focused NP in a cleft/focus construction (section 5.1). Postpositions and articles are reconstructible, and will be discussed in section 5.2. At least two topographical-deictic distal demonstratives *tə ‘DST.UP(RIVER)’ and *bə ‘DST.DOWN(RIVER)’ are reconstructible, with a third category ‘DST.SLEV’ also reconstructible, but of uncertain phonetic value (possibly *ə). A proximate *ɕi is also reconstructible; languages vary in the remainder of their demonstrative inventories, but generally include either a medial-distal or an addressee-proximate, both of which also track anaphora. Ideophones are found widely in Tani, but remain poorly described. Many have non-prototypical phonology, and typically occur in construction with the verb ‘say’ (or a cognate suffix). (3) Galo cubúk! ə́m-làa sound.of.silent.entry.into.water say-NF ‘He slipped “sploosh!” into the water.’ ó-l ̀ k-káa fall.from.height-INSERT-PRF 4.3 Word structures 4.3.1 Nouns and adjectives Tani nouns and adjectives generally have one of the structures [ROOT] or, more often, [PFXor [ROOT-ROOT], as in Galo ɲíi ‘person’, ˀa-d ́ r ‘tired’ and lə̀-cèk ‘numb, of feet ROOT] TANI LANGUAGES 329 (< leg-numb)’ or Apatani dʑìi ‘black’, ˀà-míʔ ‘eye’ and mìʔ-láa ‘tear (< eye-juice)’. The most commonly attested prefixes are *a-, *ɕ(j)a-, *ja-, *ta-, and *pa-;*ɕ(j)a- is attached to roots denoting higher animals; *ta- is used for lower animals, insects, trees, male proper names and small objects in general; *ja- is used with colours, diseases, other negative-connotation items and female proper names; *pa- is confined to bird names and related nouns, e.g. Bokar pɨ-tɨr ‘chicken carrying basket’. As for *a-, it is found on many basic nouns, including kinship terms, body parts and basic features of the natural world, as well as on many basic adjectives; presumably, *a- derives from the early TB nominalizer. The prefixes *ɕ(j)a- and *pa- appear to derive from earlier initial formatives of compounds, reflecting perhaps PTB *sja ‘flesh/meat/animal’ and *bja ‘bird’ (Benedict 1972: 46, 177). Compounds of the form [ROOT-ROOT] are common among nouns and adjectives, often following a classificatory template in which the first formative names a generic class or ‘type’ and the second a specific ‘exemplar’ (Figure 15.2). The exemplar formative may in turn stand as a new type; the resulting word networks and families can be large (Table 15.4). Semi-reduplicative expressive compounds of the form A-B A-C are also common among Tani nouns, adjectives, and verbs alike, as illustrated by these Bengni examples: a-tuŋ a-juŋ ‘garbage’, a-bak a-jak ‘thick (liquid)’, dʑi-sit dʑi-mit ‘to pass something around’, and jip-mii jip-maa ‘sleepy’. The familiar Tibeto-Burman voicing alternation in Basic Noun ?a- Type Exemplar Type ?a-pòm ?a-móo ‘cluster’ Exemplar Type moo-bàa ‘plateau’ Exemplar tum-pə ‘dryland jungle’ ‘place’ moo-tùm ‘jungle’ ?a-pə ‘globe’ moo-mèn ‘clearing’ tum-ràa ‘dense jungle’ FIGURE 15.2 CLASSIFICATORY WORD FORMATION IN GALO (FROM POST 2012) TABLE 15.4 WORD FAMILY BASED ON THE INITIAL ROOT l ̀ ɨ ‘STONE’ (FROM POST 2006) lɨɨtɕàk lɨɨm ̀ k lɨɨmìk lɨɨtàk lɨɨkàa lɨɨpùu lɨɨtɕ ̀ k ‘pebble’ ‘gravel’ ‘algae’ ‘giant boulder’ ‘igneous rock’ ‘marble’ ‘cooking tripod’ lɨɨnə̀ lɨɨpùm lɨɨtòr lɨɨjàa lɨɨkə̀r lɨɨtɕùm lɨɨpə̀ ‘boulder’ ‘stone pile’ ‘hard stone’ ‘soft stone’ ‘green bead var.’ ‘green bead var.’ ‘sharpening stone’ 330 TIBETO-BURMAN: LANGUAGES OF NE INDIA the formation of simplex-causative pairs is preserved in some measure, albeit unproductively: for example, Galo d ́ r- ‘break (of a long thing)’ vs t ́ r- ‘break (a long thing)’, or búk- ‘burst’ vs púk- ‘pop (knuckles)’. 4.3.2 Predicates Most Tani predicates have the basic structure [V.ROOT(-DERIVATIONS)-INFLECTIONS], an example of which can be found in (3); adjectives may also head a predicate (e.g. (2)), but take a smaller set of the available derivations and inflections. Many verb roots are cognate with noun roots, and are preferentially used together with a preceding cognate nominal (usually non-referential and unmarked for case), as in the following Bengni examples: (4) to lay egg to spit to dream to wear shoes pɨ-pɨ pɨta-tɕɨr tɕɨrjip-maa maalɨ-kjam kjam- (cf. pɨ-pɨ egg) (cf. ta-tɕɨr spittle) (cf. jip-maa dream) (cf. lɨ-kjam shoe) Derivation in the Tani predicate is prolific. Many predicate derivations are semantically rich, homophonous with verb roots, and in simple stems may be interpretable as compounded lexical roots: for example, Galo ín-mèn ‘go-AS.PLAY’ ‘stroll’. However, predicate derivations tend to be fully productive, and as many as four or five can be ‘stacked’ onto a single stem, as in the following Minyong example (5). Predicate derivations encode meanings related to manner, result, direction, aktionsart, modality, argument structure change (e.g. applicatives) and word class change (Post 2010). (5) ami ə=kom gok-ta-kɨ-ram-hɨ-kaa-to=î. call-INCP-TENT-FRUS-REFL-EXP-PFV=QTAG person IND=ADD ‘The guy also tried in vain to have a go at calling, eh.’ Nominalization is highly functional in Tani languages, forming relative clause constructions in addition to action and participant nouns (section 5.3). The most widely attested nominalizers are realis -na ~ -nə ‘subject/actor NMLZ’, -nam ‘action/undergoer NMLZ’ -nana ‘instrumental NMLZ’ and -ko ‘locative NMLZ’; some languages also have irrealis/purposive nominalizers, such as Galo -há. Tani is similar to most other eastern Himalayan Tibeto-Burman subgroups, but different from several groups to the east and west, in lacking grammaticalized person-indexing on the predicate. The relevant inflectional categories of the predicate are aspect and mode, much as in Tibetan. Some amount of variation is found among Tani inflectional systems; the most widely attested set of predicate inflections is given in Table 15.5, together with their lexical source values (when known); a few languages exhibit additional inflections with uncertain provenance, such as Apatani -cì ‘intentional’ and -ŋé ‘imperative’, and Milang -kaʎ ‘irrealis’. In most Tani languages, predicate formation is sensitive to a relationship between subject person and speaker knowledge, comparable in many respects to the ‘egophoric’ systems found in Bodic languages and their neighbours. In general, declarative statements with a non-first person subject in perfective aspects are unacceptable unless qualified, as in these Bengni examples: (6) a. ŋuu kanoo-pa 1SG hungry-PFV ‘I have become hungry.’ TANI LANGUAGES 331 TABLE 15.5 TANI PREDICATE INFLECTIONS AND THEIR LEXICAL SOURCES Domain PT Negation Imperfectives *maŋ *duŋ *do *dak *to *ka *ku *rje *ai Perfectives Irrealis Anterior Value Source NEG ‘not (have)’ ‘sit; be (at)’ ‘lie down; be (at)’ ‘stand; be (at)’ ‘have/be associated to’ ‘have/exist’ Pre-PT ‘live/exist’ PTB *waj ‘cop’ (?) IPFV STAT COS PFV PRF COMPL IRR ANTR b. tatɨk . kanoo-pɨ-dɨɨ (*kanoo-pa) Tatik . hungry-PFV-WONDER ‘Tatik has become hungry.’ In some paradigms, as in the Galo ‘direct’ perfective, dedicated suffixes index the ‘egophoric’ (a.k.a. ‘conjunct’, or ‘self-person’) category of first person in declaratives/ second person in interrogatives, and the ‘alterphoric’ (a.k.a. ‘disjunct’ or ‘other-person’) category of second/third in declaratives, first/third in interrogatives (7a–b). These are semantic rather than morphosyntactic indices, however, a fact which can be demonstrated by the acceptability of other persons when evidential particles which alter the construal are employed (Post 2013) (7b). Finally, note that unlike in Tibeto-Burman languages to the north and west, egophoricity in Galo is not sensitive to agentivity or volitionality. (7) Galo a. ŋó (*nó, *b ̀ ɨ) ˀatɕín dó-tó-bá 1.SG (2.SG, 3.SG) cooked.rice eat-PFV:EGO-PFV:DIR ‘I’ve just had my meal (I experienced it).’ b. nó/b ̀ ɨ (*ŋó) ˀatɕín dó-ggée-bá 2.SG/3.SG) (1.SG) cooked.rice eat-PFV:ALTER-PFV:DIR ‘You/he/she just had your/his/her meal (I witnessed it).’ c. b ̀ ɨ ˀatɕín dó-tó-bá jukə̀ cooked.rice eat-PFV:EGO-PFV:DIR REP 3.SG ‘He, it is said, “had my meal.”’ d. ŋó (*nó, *b ̀ ɨ) ò-lòo-tó-bá (*o-loo-gee-ba) 1.SG (2.SG, 3.SG) fall-DOWN-PFV:EGO-PFV-DIR fall-DOWN-PFV:ALTER-PFV-DIR ‘I fell down (I experienced it).’ 5 SYNTAX 5.1 Clause and phrase types and structures Tani information structure has the basic order [TOPIC FOCUS]. Syntactically, clauses can be divided into predicative and appositive subtypes. A predicative clause is headed by an 332 TIBETO-BURMAN: LANGUAGES OF NE INDIA inflecting predicate (4.3.2), which usually occurs last, in the FOCUS position (7); more topical noun phrases occur furthest from the predicate, with the most frequent resulting order among core arguments of multi-argument clauses being agent > recipient > patient. However, different orders are straightforwardly possible (8): (8) Galo (popular song by Marto Kamdak) nó-m̀ =ne ŋó… káa-tó-làa 2.SG-ACC=NAGT 1.SG look-PFV-NF ‘I saw you.’ A basic appositive clause consists of two noun phrases arrayed in the order [TOPIC FOCUS]. Appositive clauses generally handle equative and attributive functions. In many if not all Tani languages, the demonstrative *ɦə ‘DIST/ANAP’ has developed via a definite article into a copula in this construction, and can predicate an adjectival as well as nominal attributive focus. (9) Tangam a. hi apoŋ (e < *ɦə) PROX rice.beer (COP) ‘This is rice beer.’ b. ŋo kenoŋ e 1.SG hungry COP ‘I’m hungry.’ Posture verbs *duŋ ‘sit’, *dak ‘stand’ and *do(ŋ) ‘lie’ also developed locative existential copula functions in the same construction; however, languages vary in the extent to which this construction is retained. Galo appears to have merged this construction into the predicative clause type – choosing to inflect its locational posture verbs – and some Adi varieties have generalized *duŋ in all locative-existential functions. However, the full paradigm is retained in Apatani, in which dàʔ predicates entities with legs, or in a standing position, dó predicates other inanimate entities, and dùu predicates other animate entities (10). (10) Apatani a. n ́ -kà pàróʔ so dàʔ kèe PROX.LOC be.at.LOC(STANDING) POL 2.SG-GEN fowl ‘Your chicken is (standing) here.’ b. mòo-kà mj ́ ɨ dùu be.at.LOC(ANIMATE) 3.SG-GEN wife ‘His wife is here.’ c. ŋ ́ -kà póosá dó 1.SG-GEN money be.at.LOC(INANIMATE) ‘I have money (with me).’ Negation in appositive clauses is generally handled by post-modifying the focus with a negator ma(ŋ) or mo(ŋ) (the latter mainly found in north-central-area languages). Only Apatani is known to have a suppletive negative copula ɲímá. Illocutionary force is generally signalled via a wide range of particles. In most cases, particles occur clause-finally; however, in a marked focus construction, they mark the focus. When a marked focus construction is based on a predicative (not an appositive) TANI LANGUAGES 333 clause, and when the predicate is outside the focus, the predicate must either be nominalized, or else is marked by a nominalizer-derived suffix, as in the following Lower Adi example: (11) Lower Adi no kolo la gɨ-ye-n? 2SG where CQ go-IRR-Q(< NMLZ:SBJ) ‘Where are you going?’ Within the noun phrase, compound nominals are typically MODIFIER–HEAD, as in Lower Adi kalam diitə ‘Mt. Kalam’. Coordinative compounding is also common at this level, with each noun signifying an exemplar and the whole signifying a superordinate type, as in Galo donám-tɨɨnám ‘eat.NMLZ imbibe.NMLZ’ ‘provisions’. Possessors are pre-head (10), while classifiers and numerals follow the head in that order, as in Apatani s ́ ɨ dór-ɲì ‘cattle CLF-two’, and are followed by relator nouns, articles and/or postpositions, as in Galo ˀaɲ ́ ɨ kook ̀ ɨ lo ‘after a year (year back LOC)’. A striking characteristic of demonstratives is they can occur on both flanks of the head (e.g. Bengni sii əki sii ‘this here dog’). Nominalization-based relative clauses also occur on both sides of the head, as in Lower Adi dəm-nam ami ~ ami dəm-nam ‘person who was hit’ (section 5.3). Adjectival modifiers, if formed as relative clauses, in the same way may occur on both sides; in languages which allow direct adjectival modification (without a nominalizer), the adjective can only post-modify, as in Lower Adi mimum kampo ‘woman beautiful’. 5.2 Referential and relational marking Tani noun phrases are marked differently depending on whether they represent generic/ non-referential, indefinite (new, usually focal), or definite (given, usually topical) information, and on whether they are in subject, non-subject core or oblique functions. Indefinite NPs are typically marked by an article *(a)ko (< ‘one’), while definite NPs are marked either by an appropriate demonstrative, or by a demonstrative-derived article, usually *ɦə. Generic/non-referential NPs are typically unmarked. Relational marking follows a basically nominative–accusative pattern: subjects tend to be definite, but are unmarked for case, whether functioning as the single intransitive argument or as the actor of a transitive verb. It is common to encounter identifications of a ‘nominative case’ marker ə in early descriptions of Tani languages; in fact, this is a reflex of the definite/topic marker *ɦə (12). (12) Lower Adi a. bɨ si-kai 3SG die-PRF ‘He died.’ b. bɨ-k əsu ə si-kai 3SG-GEN mithun TOP die-PRF ‘His mithun died.’ Non-subject core arguments are marked by a form of the accusative case when definite. Objects are distinguished via an unmarked accusative *-m (for pronouns, see (8)) ~ *ɦəm (for full NPs; see (13b)); note that the latter incorporates an earlier demonstrative *ɦə, 334 TIBETO-BURMAN: LANGUAGES OF NE INDIA which explains its restriction to definite noun phrases. Non-referential undergoers are generally predicate-adjacent, and unmarked for case (as in (7)). Indefinite arguments are never marked for case (13a). (13) Bokar a. ŋoo pətaŋ ako(*-m) koŋpoŋ-pa INDEF(*-OBJ) see-PFV 1SG bird ‘I saw a bird.’ b. ŋoo pətaŋ-hə-m koŋpoŋ-pa 1SG bird-DEF-OBJ see-PFV ‘I saw the bird.’ In most languages, there is an additional ‘non-agentive’ object marker usually descended from *mɨ (possibly itself the source of our pronominal accusative), which functions to highlight the object status of a highly topical/animate participant; an example of this is in (8) (using the innovative Galo form nè). Recipients are marked in the same manner as undergoers, with animacy and context apparently sufficing to disambiguate. Relational marking of agents has not been attested, except for demoted agents which are marked by an instrumental postposition in the passive constructions of some languages (14): (14) Minyong bɨ hɨmjo kɨŋ do-ko-hɨ-to INST eat-PASS-REFL-PFV 3.SG tiger ‘He was eaten by a tiger.’ A genitive case marker *ka is widespread and marks adnominal noun subordination (10), as well as subjects of both intransitive and transitive nominalized clauses (section 5.3). The allative cases are often identical to, or built on, the locative; the ablative and comitative are likewise usually derived from the locative, e.g. lokə ‘from’ < LOC=GEN. An instrumental case marker *kɨŋ is found in some though not all languages (see (12)); in languages which lack it, the instrumental is typically built on the ablative and/or genitive (as in Galo lokkə̀). 5.3 Complex constructions Tani languages are particularly rich in the variety of complex constructions they exhibit. Nominalization-based constructions are especially prominent; complex constructions not based on synchronic nominalizations include clause-chaining, postpositional and/or adverbial subordination, and complementation. Serial verb constructions are absent from the modern languages. Nominalization-based constructions include action nominalizations, several types of participant nominalization, internally headed, externally headed and headless relative clause constructions, cleft/focus constructions ((11) earlier), and clausal nominalizations (see Post 2011a). In general, a set of three to five ‘primary’ nominalizers (4.3.2) mark the participant/action and often also the modality status of the nominalized clause. In (15a), the Galo nominalizer -há produces an irrealis action nominalization, while in (15b) the same nominalizer produces a gapped externally headed non-subject relative clause construction (note here the genitive subject marking). TANI LANGUAGES (15) 335 Galo (Post 2011a: 267–9) a. bulù-kə̀ ˀagóm=əm porì-há káa-máa speech=ACC study-NMLZ:IRR.EVENT have/exist-NEG 3.PL-GEN ‘There would be no (need/cause) to study their language.’ ŋó-kə dó-há jaráa=ə 1.SG-GEN Øi eat-NMLZ:IRR.NONSBJ goodsi=COP ‘This is the thing that I’ll eat (lit. my to-be-eaten thing).’ b. hɨgɨ SPRX.IND Various types of adverbial subordination, including manner, purposive and temporal clauses are marked by postpositions, typically *pə (< DAT), *lo (< LOC) and *(ɦə)m (< ACC), and/or by dedicated markers built at least partly on one of them. Similar to nominalized clauses, many postpositionally subordinated clause types also exhibit a genitive subject (16a); however, postpositionally subordinated clause heads are often inflected (unlike most nominalized clause heads), typically lose genitive subject marking when more advanced in their development, and more generally tend to more closely resemble finite clauses than do nominalized clauses (16b). (16) a. Galo (Post 2011a: 282) taníi=gə hobə́=ə̀m pá-dàk=lo ŋó ˀín-ɲíi-tó 3.SG=GEN mithun=ACC strike.to.cut-COS=LOC 1.SG GO-ABANDON-PFV ‘I left as Tani killed the mithun (lit. at Tani’s killing of the mithun).’ b. Galo (Post 2009: 84) nó hag ̀ ɨ-booló hag ̀ ɨ-rə́ sigh-IRR 2.SG sigh-COND ‘If you sigh, (it too) will sigh.’ Complex and chained events in a thematically united sequence are typically represented using clause chains, most often marked via a non-final suffix (< additive coordinator *laŋ). Depending on a variety of factors, non-final predicates may or may not be inflected, and often exhibit additional suffixes (such as completive or attainment suffixes) which serve to ‘finetune’ the temporal-episodic sequencing of events, as in example (17) from Upper Minyong: (17) Upper Minyong dəl ammo hɨdla… də iˀala, ŋolu məhaalaŋkula… dəlo ammo hɨt-la də i-pa-la ŋolu mə-haŋ-laŋ-ku-la ANAP.LOC paddy dibble-NF ANAP DO-ATTN-NF 1.PL put-UPWARD-NF-COMPL-NF ‘And then we planted paddy…and after doing that, we would leave it to grow…’ Clause chaining and action nominalization are also frequently interwoven to produce episodic breaks amid thematic continuity, as is widely the case in Tibeto-Burman (18). (18) Upper Minyong də bɨɲi jə-bo-hɨ-duŋ. jə-bo-hɨ-la i-nam=ə… ANAP 3.DL live-RECP-REFL-IPFV live-RECP-REFL-NF do-NMLZ:RLS.EVENT TOP ‘So, the two of them lived together. Because of that (lit. Living together and that being so)…’ ADDITIONAL ABBREVIATIONS ADD ANTR Additive Anterior ALTER ATTN Alterphoric Attainment ANAP AVZR Anaphoric Adverbializer 336 TIBETO-BURMAN: LANGUAGES OF NE INDIA COS EGO INCP NAGT QTAG SLEV STAT Change of state Egophoric Incipient Non-agentive Question tag Same (topographic) level Stative CQ EXP IND NF REP SPRX TENT Content interrogative Experiential Individuator Non-finite/non-final Reportative Speaker-proximate Tentative DIR FRUS INFR POL RLS Direct evidential Frustrative Inferential Polite Realis NOTES 1 Simon (1972) and Abraham (1985, 1987) have been field-checked and found to be incomplete and/or unreliable in very many respects. Apatani Language Development Committee (2015) is reliable, but far from comprehensive. 2 Kepor Mara, a trained linguist and native Tagin speaker, has indicated that vowel length may be non-contrastive in Tagin, however this remains unconfirmed. REFERENCES Abraham, P. T. (1985). Apatani Grammar. Mysore: Central Institute of Indian Languages. —— (1987). Apatani–English–Hindi Dictionary. Mysore: Central Institute of Indian Languages. Apatani Language Development Commttee (2015). ‘Tanw Aguñ: An Audio Dictionary of the Apatani Language.’ Available at: language.apatani.org (accessed 15 August 2016). Benedict, Paul K. (James A. Matisoff, contributing editor). (1972). Sino-Tibetan: A Conspectus. New York: Cambridge University Press. Census of India. (2011). A-11: Individual Scheduled Tribe Primary Census Abstract Data and its Appendix: Arunachal Pradesh and Assam. New Delhi: Government of India. Available at: www.censusindia.gov.in/2011census/PCA/ST.html (accessed 4 April 2016). Chhangte, Thangi. (1992). Phonology of some Nishi (Dafla) dialects. Paper presented at the 25th International Conference on Sino-Tibetan Languages and Linguistics, Berkeley, University of California. Huber, Toni. (2012). Micro-migrations of hill peoples in Northern Arunachal Pradesh: Rethinking methodologies and claims of origin in Tibet. In: Origins and Migrations in the Extended Eastern Himalaya, ed. by Toni Huber and Stuart Blackburn. Leiden: Brill, 83–106. Modi, Yankee. (2005). Emergence and Impact of Hindi as a Lingua Franca in Arunachal Pradesh. New Delhi: Jawaharlal Nehru University M. Phil Thesis. Ou-Yang, Jue-Ya. (1985). ⨎ᐤ᯿䈝䀰ㆰᘇ˄ፙቬঊాቄ䈝) [Luobazu Yuyan Jianzhi (Bengni Boga’eryu)/Brief Description of the Luoba Nationality Language (The BengniBokar Language)]. Beijing: ≁᯿ࠪ⡸⽮ [Minzu Chubanshe/Nationalities Press]. Pegu, Jugendra. (2011). Morpho-syntactic variation in the pagro and sa:jaŋ dialects of the Mising community. In: North East Indian Linguistics Vol. 3, ed. by G. Hyslop, S. Morey and M. W. Post. New Delhi: Cambridge University Press India, 155–170. Post, Mark W. (2006). Compounding and the structure of the Tani lexicon. Linguistics of the Tibeto-Burman Area 29.41–60. —— (2007). A Grammar of Galo. Melbourne: La Trobe University PhD Thesis. —— (2009). The semantics of clause linking in Galo. In: The Semantics of Clause Linking: A Cross-Linguistic Typology, ed. by R. M. W. Dixon and A. Y. Aikhenvald. Oxford: Oxford University Press, 75–91. TANI LANGUAGES 337 —— (2010). Predicate derivations in the Tani languages: Root, suffix, both or neither? In: North East Indian Linguistics Vol. 2, ed. by S. Morey and M. W. Post. New Delhi: Cambridge University Press India, 175–197. —— (2011a). Nominalization and nominalization-based constructions in Galo. In: Nominalization in Asian Languages: Diachronic and Typological Perspectives, ed. by F. H. Yap, K. Grunow-Hårsta and J. Wrona. Amsterdam: John Benjamins, 255–287. —— (2011b). Prosody and typological drift in Austroasiatic and Tibeto-Burman: In: Against ‘Sinosphere’ and ‘Indosphere’. Austroasiatic Studies: Papers from ICAAL4. Mon-Khmer Studies Special Issue No. 3, ed. by S. Srichampa, P. Sidwell and K. J. Gregerson. Canberra: Pacific Linguistics, 198–221. —— (2012). The language, culture, environment and origins of Proto-Tani speakers: What is knowable, and what is not (yet). In: Origins and Migrations in the Extended Eastern Himalaya, ed. by T. Huber and S. Blackburn. Leiden: Brill, 153–186. —— (2013). Person-sensitive TAME marking in Galo: Historical origins and functional motivation. In: Functional-Historical Approaches to Explanation: In Honor of Scott DeLancey, ed. by T. Thornes, J. Jansen, G. Hyslop and E. Andvik. Amsterdam: John Benjamins, 107–130. —— (2015). Sharing innovations across branches: A case study of the Tani subgroup of Trans-Himalayan [Tibeto-Burman]. Paper presented at the International Conference on Historical Linguistics 22. Naples, Italy. Post, Mark W. and Roger Blench. (2011). Siangic: A new language phylum in North East India. Paper presented at the 6th International Conference of the North East Indian Linguistics Society, Tezpur University, Tezpur, Assam, India. Post, Mark W. and Yankee Modi. (2011). Language contact and the genetic position of Milang (Eastern Himalaya). Anthropological Linguistics 53.215–258. Post, Mark W. and Kanno Tage. (2013). Apatani phonology and lexicon, with a special focus on tone. Himalayan Linguistics 12.17–75. Simon, Ivan Martin. (1972). An Introduction to Apatani. Gangtok, Sikkim: Government of India Press. Sun, Hong-Kai, Ji-Chuan Zhang, Shao-Zun Lu and Jue-Ya Ou-Yang. (1980). 䰘ᐤǃ ⨎ᐤǃ‫܌‬ӪⲴ䈝䀰 [Menba, Luoba, Dengren de Yuyan/The Language of the Menba, Luoba and Dengren] Beijing: ѝഭ⽮Պ、ᆖࠪ⡸⽮ [Zhongguo Shehui Kexue Chubanshe/Chinese Social Sciences Press]. Sun, Tian-Shin Jackson. (1993). A Historical-Comparative Study of the Tani (Mirish) Branch of Tibeto-Burman. Berkeley: University of California PhD Dissertation.