Finite Structures from Clausal Nominalization in ...

10 downloads 57 Views 239KB Size Report
substantially centered on verb serialization (Matisoff 1969; DeLancey 1991) and nominalization (Matisoff ... syntax (Matisoff 1972; Mazaudon 1978; DeLancey 1999, 2005; Genetti 1992a; Genetti et. al. ms.; Noonan ..... Bauman, James. 1975.
Finite Structures from Clausal Nominalization in Tibeto-Burman Scott DeLancey University of Oregon 1. Introduction Tibeto-Burman languages illustrate a typological syndrome found in a number of families throughout the world, characterized by verb-final constituent patterns and syntax, which is substantially centered on verb serialization (Matisoff 1969; DeLancey 1991) and nominalization (Matisoff 1972). The use of nominalized clause constructions to form relative clauses has been a topic of considerable interest to students of Tibeto-Burman syntax (Matisoff 1972; Mazaudon 1978; DeLancey 1999, 2005; Genetti 1992a; Genetti et. al. ms.; Noonan 1997, 2008, inter alia). In this paper, I will discuss another manifestation of the importance of nominalization in Tibeto-Burman syntax, which is the regular development of new finite structures from clausal nominalization constructions. We will see that clausal nominalization in construction with a copula is, indeed, a major–source of new finite clause constructions throughout the family. 2. Phenomenon The frequent phenomenon of nominal morphosyntax associated with the finite verb has been noted since the earliest days of Tibeto-Burman studies. Konow (1908, 1911) and other early investigators were insistent that the verbs of Tibeto-Burman were not truly verbs in the sense found in Indo-European, but essentially nominal: The different classes of words are not clearly distinguished, and many instances occur in which a word can be used at will as a noun, as an adjective or as a verb. The verb can, on the whole, be described as a noun of action, and we find phrases such as "my going is" instead of "I go." (Konow 1911:929)

Konow was referring in particular to phenomena like the use of possessive proclitics as verb agreement in Kuki-Chin, which I will discuss below. As we will see, the stigmata of nouniness which these scholars perceived in the morphosyntax of Tibeto-Burman verbs reflect the fact that in many Tibeto-Burman languages the finite construction of the verb reflects an earlier construction in which the sentence or verb phrase is nominalized. The construction often includes a copula, of which the nominalized sentence is then an argument, but the copula may be dropped over time, and sometimes nominalizations are simply used as finite clauses. The relationship between nominalization and finiteness in Tibeto-Burman was re-introduced to the modern field by Matisoff’s seminal article (1972), in which he points out that the Lahu nominalizer ve occurs not only in familiar nominalization contexts, but marking relative clauses, and sometimes even occuring in non-subordinated main clauses. This phenomenon is very widespread in Tibeto-Burman; in fact I would

1

impressionistically claim that languages which show clear examples of a nominalized clause, with or without a final copula, functioning as a main clause are far more common than those (if in fact any could be found) which do not. To begin with a concrete example, most of the verb forms in the modern Tibetan dialects (for the standard Lhasa dialect see, e.g., DeLancey 1991 or Denwood 1999) obviously originated as NOM + COP constructions (see Saxena 1992, 1997). The general structure of a finite verb construction in modern Tibetan is the lexical verb with a nominalizer and a copula, as in these Lhasa Tibetan examples in table 1:1 Table 1: examples of the general structure of a finite verb construction in modern Tibetan STEM NOM COPULA nga zas -pa yin ‘I ate.’ nga za -gi yod ‘I am eating.’ nga za -gi yin ‘I will eat.’ nga za -rgyu yin ‘I still have to eat.’ nga za -mkhan yin ‘I am going to / intend to eat.’

The various nominalizers contribute their own semantics to the tense/aspect/modality constructions formed with them. The nominalizer -pa is the oldest Bodish, and Tibeto-Burman, nominalizer, and is associated with perfectivity in most of Bodish (DeLancey 2005). The non-perfective gi / kyi and rgyu are apparently related. Kyi does not occur anywhere outside of the tense aspect paradigm, but rgyu is used productively as a nominalizer for irrealis event predictions. Finally, mkhan is originally an agentive nominalizer (DeLancey 1999), so that the last sentence could be literally (or, at least, etymologically) parsed as ‘I am an eater’. This construction–a nominalized clause finitized by a copula (which in many languages may be optional or even absent altogether)–is one of the two basic sources of clausal structure in Tibeto-Burman languages. I will refer to it as a nominalized clause construction; its schematic manifestation in a verb-final language is STEM-NOMINALIZER (COPULA). Several scholars have discussed the common Tiberto-Burman phenomenon in which a formally nominalized clause is used as a main clause, i.e. where a construction which is formally non-finite has acquired a finite function (see especially Bickel 1999). I will be discussing primarily examples of grammaticalization of the construction with the copula. Such a construction is, of course, already finite, and it originates as a means of expressing a marked pragmatic or semantic category, such as aspect, evidentiality, register or style, or strength of assertion. My purpose, however, is not to examine the functions of such constructions in the beginning of their careers, but to document how, over time, they work themselves into the core finite system of the language, often eventually displacing and replacing older finite constructions. Obviously the development of new finite constructions from simple nominalizations, with no copula, represents the same diachronic phenomenon. 1.

These are presented in transliterated orthographic form; in speech -pa yin, in particular, is subject to substantial phonological reduction, to the point where a non-literate speaker may not recognize the etymology. Educated speakers, however, are quite able to explain the from structures, which are a recent development in the language, as we will discuss below.

2

3. Three case studies In this section, we will examine three examples of the development of finite structures from nominalized clause constructions. Two recently-innovated finite constructions from the Western Kiranti language Sunwar have shown how easily a nominalized clause construction can find a place in the finite tense/aspect system of a Tiberto-Burmen language. And the history of the modern Kuki-Chin finite clause gives an illustration of how easily a nominalized clause construction can completely replace earlier finite morphosyntax. 3.1 Sunwar: Nominalization in action I will present here a set of data originally reported in DeLancey (1992b), which show that Sunwar turns to nominalized clause constructions repeatedly as it innovates new tense/aspect/evidentiality categories. 3.1.1 A recently-innovated finite construction The Sunwar copula system has involves to form four etyma: locative/existentials tsha and 'baak, and an equational na, with a suppletive allomorph ho. The equational na is the oldest of all, with a robust Tiberto-Burmen etymology (Matisoff 2003). Tsha and ho, on the other hand, are transparent borrowings from Nepali. The last existential, 'baak, has the semantics of a recently-grammaticalized locative copula, in that it still occurs with the lexical sense ‘stay, live at’. Moreover, it has no evident copular cognates elsewhere in Bodic or Tiberto-Burmen. Sunwar’s nearest relatives, Hayu and Bahing, each has only a single copula, which is clearly cognate to Sunwar na. Thus its copula function appears to be a relatively recent development. The Sunwar verb has three paradigms, illustrated with 1st person singular forms of the verb la ‘go’ in table 2: Table 2: Paradigms of the Sunwar verb “go” Tense/aspect Sunwar verb “go” simple la-ŋa past là-ti non-past laî-nu-ŋ

Note that, with a vowel stem verb like la, the non-past has an expanded stem form, adding an off-glide /i/ and a falling tone, which elsewhere in the language occurs only when an underlying coda consonant is deleted. Consonant-final verb stems are identical in all three conjugations. The simple conjugation occurs in negated verbs, and a few other constructions. Both this and the past paradigm are widely shared throughout Kiranti, and reconstruct for Proto-Kiranti. The third paradigm, the non-past, is a very new development, not shared even with Bahing or Hayu, Sunwar’s closest Northwest Kiranti kin. Its status as a recent innovation is equally evident on language-internal grounds, as we will see directly. The non-past paradigms of the Sunwar non-past verb endings are given in table 3:

3

Table 3: Non-past paradigms of the Sunwar non-past verb endings DUAL SINGULAR INCLUSIVE EXCLUSIVE -nuŋ -nase -nasku 1st -neye -nisi -nisi 2nd -ba -nise -nise 3rd

PLURAL INCLUSIVE EXCLUSIVE -nai -niki -nini -nini -nim -nim

And table 4 shows the conjugation of the equational copula na: Table 4: conjugation of the equational copula na DUAL SINGULAR INCLUSIVE EXCLUSIVE na-ŋ na-se na-sku 1st 'na-ye na-si na-si 2nd ho na-se na-se 3rd

PLURAL INCLUSIVE EXCLUSIVE na-i na-iki na-ni na-ni na-m na-m

The correspondence of these two paradigms is evident on simple inspection; the non-past paradigm is manifestly a morphologization of the na paradigm. Note that the non-past paradigm, like that of na, has an intrusive form in the 3rd singular (originally the Bodic nominalizer *ba), which must reflect the same gap in the original paradigm as the borrowed form in the copular paradigm. This leaves the problem of the non-past stem, which, given the transparency of the suffixes, must likewise have a fairly shallow analysis. Indeed, the alternation which we see between la- and laî- has an exact parallel in the behavior of verbs with an underlying coronal stop final (Genetti 1992b, DeLancey 1992b). Compare the conjugated forms of ta‘see’ and sad- ‘kill’ in table 5: Table 5: conjugated forms of ta- ‘see’ and sad- ‘kill’ ta- ‘see’ tau Imperative taa-ta-ŋ Past (1sg → 3sg) ta-ti Past (3sg → 1sg) taî-nu-ŋ Non-Past (1sg → 3sg)

sad- ‘kill’ sadu saî-ta-ŋ saî-ti saî-nu-ŋ

The behavior of ta in the non-past would be explained if there were an underlying coronal obstruent coda *-t or *-s in the non-past stem. Thus, by quite shallow internal reconstruction, we can reconstruct the original form of the non-past construction as *STEM-t na-AGR, which instantiates our STEM-NOMINALIZER COPULA formula. 3.1.2 New perfects I have noted above the evidence that 'baak is a recent innovation as a copula. The Nepali borrowing tsha is clearly a very recent adoption into Sunwar, so fresh that it retains its full Nepali inflectional paradigm. The contrast between 'baak and tsha is epistemological, marking a distinction between what I have elsewhere (DeLancey 1986, 1990) called "old" and "new" information, a peculiar Tiberto-Burmen manifestation of the wider mirative category (DeLancey 1992a, 1997). Tsha represents the unmarked category of information, and 'baak is mirative, representing information which the speaker would have no way of knowing save through information channels to the outside world. However, for our present

4

purposes, the main point about this distinction is that it is an innovation in the language, so I will not go into its semantic content in detail. For further discussion, see DeLancey (1990, 1992a, b, 1997, 2001). Some versions of this distinction are quite widespread in the Himalayas and adjoining areas, but there is no basis for reconstructing it to any great time depth. It is widespread in Bodic, except where it appears to have been borrowed directly (as from Central Tibetan into Cuona Monpa). It is also expressed in the various languages, in which it occurs by forms which clearly are not cognate, and sometimes not even homologous (in that, for example, it is expressed in Kathmandu Newar by verb endings, in Tibetan dialects by copular distinctions). Thus the category as such doesn’t seem to be reconstructable to Proto-Bodic or even to the level of Proto-Tibetan (DeLancey 1992a). So when we find it expressed in Sunwar by an opposition between two copulas both of which are evidently new developments in the language, we can infer that the category itself is a relatively new introduction into Sunwar, representing that language’s somewhat belated entry into the mini-Sprachbund, which is characterized by the “egophoric” or “conjunct/disjunct” pattern (DeLancey 1992a). Again, this inference is supported by the lack of any such category in Hayu or Bahing. Sunwar, as a Tiberto-Burmen language, has a predilection for nominalized constructions of various kinds. Thus, in typical Tiberto-Burmen fashion, the existence of this semantic distinction in the copula automatically leads to its incorporation into the verb system. All three copulas occur with various non-finite forms of the verb in constructions which form a significant part of a complex system of tense/aspect/evidential categories. These include innovative constructions consisting of a clause marked with the nominalizer -šo 2 in constructions with 'baak or tsha, which bring their evidential/mirative senses with them. In example (2), we get the clearly mirative sense of 'baak, indicating direct visual evidence of a fact, which otherwise would have been unknown to the speaker. This is contrasted with the unmarked semantics of tsha in example(1): (1)

kyaršE 'saî-šo tshaa goat kill-NOM exist/3s/NPST 'He is killing a goat/goats.'

(2)

kyaršE 'saî-šo 'baâ-tA goat kill-NOM exist-3sPST '[I saw] he was killing a goat/goats.' [example: when I discovered him]

Borchers (2008) describes a variety of Sunwar, in which we can already see further grammaticalization of this last construction. Her data include the mirative construction (which she calls “expression of unexpected action”) in (2), and also examples where 'baak is suffixed directly to the verb stem in (3): (3)

2.

goi das ghanta hir-baa-ti-ni you(polite) ten hour walk-UNEX.-PST-2pl ‘You walked for ten hours?!’

For the etymology of this form see Noonan 2008: 227-8. 5

Besides, she also reports the use of bare -šo-nominalized clauses as finite main clauses. In analyzing these constructions in Sunwar, I have consistently argued that they have to be quite recent developments: on comparative grounds, they must postdate the divergence of Sunwar and Bahing, which is a very shallow relationship; on internal grounds, the transparency of the relationship between the non-final paradigm and the copula, and the corresponding regularity of the paradigm (in contrast to the opacity and irregularity of the older past paradigm) show that the non-past construction is a recent innovation, while the perfect constructions can hardly be more than a few centuries old. The nominalized clause construction is not necessarily as fertile in other Tiberto-Burmen languages as it seems to be in Sunwar, but it is productive everywhere. 2.2 Kuki-Chin: Reconstructing a Nominalized Clause Construction A somewhat older Nominalized Clause Construction is illustrated by the Kuki-Chin languages, a branch of Tibeto-Burman spoken in western Myanmar and Northeast India. The Kuki-Chin verbal system is an anomaly within Tibeto-Burman. In general, Tiberto-Burmen languages tend toward suffixing rather than prefixing morphology. Suffixal verb agreement is found in a substantial number of languages, and can be reconstructed for Proto-Tiberto-Burmen (Bauman 1975; DeLancey 1989; van Driem 1993; inter alia). Some languages have agreement systems with only suffixes, others include a few prefixes in the paradigm. The Kuki-Chin languages are unique in the family in having innovated a new, exclusively prefixal agreement paradigm. The history of this development is written in the structure of the system. The paradigm is evidently a recent development. In at least some languages of the branch, the agreement morphemes are not fully morphologized, but remain phonologically somewhat independent of the verb, so that many authors describe them as clitics rather than prefixes. The first piece of evidence for their history is the clitic forms used for verb agreement are also the possessive proclitics used with nouns. Note the uses of the 1st person proclitic ka3 in the following examples from the Northern Chin language Sizang (Stern 1963): (4)

(ké:i) ká pài: hî: (I) 1st go FINAL ‘I go / went.’

(5)

kâ mei: 1st tail ‘my tail’

(6)

kà káp hî: 1st cry.I FINAL ‘I cry’

(7)

kà ká: nà: 1st cry.II NOM

3.

The tonal alternation in the proclitic is phonologically conditioned by the tone of the following verb or noun stem. 6

‘my crying’ The second piece of evidence is the sentence-final particle hi, which occurs in the verbal examples. This is homophonous with the equational copula. Both can occur together in (8): (8)

ama khua:-pui:-te a he villager 3rd ‘He is a villager.’

hi: be

hi: FINAL

Notice that, in this example, the first occurrence of the morph is preceded by a proclitic agreeing with the 3rd person subject, and is thus behaving as a verb. Thus, on grounds of morphology and syntactic distribution, the copula and the final particle are synchronically distinct. Nevertheless, the diachronic derivation of the latter from the former can hardly be doubted. We can see the system at work in (9) (Stern 1984): (9)

a. na-lá:i hong thák ka-ngá: a: 2nd -letter CIS send 1st-receive NF 'I having received your letter which [you] sent to me ...' b. k-ong thûk kí:k lâ-lê:u hî: reply again once.more FIN 1st-CIS 'I in turn reply to you.'

Here we see the proclitics functioning both as possessives (na-lá:i ‘your letter’) and as agreement markers (ka-ngá: ‘I receive’, k-ong thûk ‘I reply [to you]’). We can also see the synchronic function of the final particle: it contrasts with non-final markers like the a: which ends the first clause, and marks the end of a clause chain. In Tiberto-Burmen languages without a final particle construction this function is marked by a finite verb. These two facts together indicate that the synchronic finite construction of the Kuki-Chin verb is a reanalysis of a former nominalization construction. The use of possessive clitics with a verb stem implies nominal status for that stem. If we hypothesize that, in the structure ká pài: hî:, the verb stem was historically nominalized, and the structure reflects an original *’It is my going’, we neatly explain why the agreement proclitic and the final particle are homophonous with, respectively, the possessive and the copula. The original construction must have been a nominalized clause construction with a nominalized verb stem, its subject expressed as a grammatical possessor, and the copula as the highest verb. A further fact about the Northern Chin languages lets us fill in the story. In most of Kuki-Chin, the new finite construction with proclitic agreement and a copular sentence final particle has completely replaced the older, inherited Tibeto-Burman finite verb. In Northern Chin, however, some of the old suffixed forms remain as an alternate finite paradigm, so that the verb may have either agreement proclitics or agreement suffixes, but not both, as in (9) and (10): (10) pài: ke-ŋ go NEG-1st ‘I don’t go.’

7

(11) ká pài: kei hî: 1st go NEG FINAL ‘idem.’ As in the nearby Jinghpaw-Konyak languages (DeLancey 2008a, to appear), in Northern Chin, the agreement suffixes do not attach directly to the lexical verb stem, but to a set of highly grammaticalized elements derived from old auxiliary verbs. , are both well-attested in other branches of the The suffixes, 1st person -ŋ and 2nd -t family, and clearly represent inheritance from Proto-Tibeto-Burman (Bauman 1975; DeLancey 1989). Thus we have to reconstruct two distinct main clause constructions for Proto-Kuki-Chin: the older finite form, where the end of the sentence is marked by an auxiliary conjugated with agreement suffixes, and the newer, originally nominalized construction, where the lexical verb looks like a noun possessed by its subject, and the end of the sentence is marked by the grammaticalized copula. We have descriptions of the alternation for only two Northern Chin languages, Sizang (Stern 1963) and Tiddim (Henderson 1957, 1965). In both languages use of the nominalized construction is more formal, and the older suffixed forms informal. So, pending data from other members of the branch, we can tentatively reconstruct a Proto-Kuki-Chin sociolinguistic pattern in which the old, inherited finite construction is used in everyday informal situations, and the innovative, nominalized structure is used for more formal or public styles. (Henderson notes that in Tiddim the nominalized structure is used in storytelling and narrative). In Northern Chin the alternation remains; in the other branches of Kuki-Chin, the more formal construction has completely replaced the original. Proto-Kuki-Chin, then, was a language like those discussed by Noonan (1997, 2008) and Bickel (1999), in which a nominalized construction is used instead of the ordinary finite construction for some marked purpose. This interpretation is strengthened by the fact that the suffixes are in complementary distribution not only with the proclitics, but with the entire construction – the final particle occurs only on clauses with proclitic agreement, not with the suffixal paradigm.The available reports suggest that the function of the alternation between the old finite and the nominalization clause constructions may have been stylistic, with nominalization associated with more formal style. 3. The evidence of final particles As we have noted, the occurrence of a sentence-final particle in declarative sentences is a peculiarity of many Tibeto-Burman languages. In languages with such a particle, it marks the end of a clause chain; in languages which lack this construction, such as modern Central Tibetan, the final verb in a clause chain requires a finite verb (which may itself be a Nominalized Clause Construction). I cannot provide etymologies for all of the final declarative particles in the family, but where their history has been traced, they derive from old nominalizers or copulas, and represent the final stage of grammaticalization of a nominalized clause construction. We will look at only two examples here, from Classical Tibetan and Lolo-Burmese, but similar data can be found in many other branches of the family as well.

8

3.1 Classical Tibetan Classical Tibetan has a quite different system, with a grammaticalized sentence-final particle ‘o, which geminates a preceding final consonant, as in (12): (12) nam.mkha'-la 'phur te 'gro ' o sky LOC fly NF go FINAL ‘[He] went flying (around) in the sky. (13) bya de-s srog thar-pa-r byas so bird DEM-ERG life save-NOM-LOC did FINAL ‘The bird acted to save his life.’ (14) khyim.bdag-gi bu-s 'di skad ces gsol to4 householder-GEN boy-ERG this word thus ask FINAL ‘The householder's son made this request.’ If we hypothesize that declarative final particles in general originate in the nominalizer or the copula of a Nominalized Clause Construction, as in Kuki-Chin, then it is significant that equational sentences frequently occur with no copula, i.e. that the final particle can serve duty as an equational copula (examples from Hahn 1974:39): (15) bram.ze de dbul.po zhig go Brahmin that pauper a FIN ‘That Brahmin was a pauper.’ (16) khyim de chen.po-’o house that big FIN ‘That house is big.’ By the time of its attestation in Classical Tibetan, ‘o is nothing but a final particle; there is no homophonous but distinct copula as there is in Kuki-Chin. It probably reflects a widespread Proto-Tibeto-Burman copular root *way which we will see below turning up in other branches. As far as I know, this etymon has no other reflex in modern varieties of Tibetan, in all of which the oldest copulas are forms of Classical Tibetan yin and yod. However, it is well attested elsewhere in the Bodish branch, and Bodic subbranch, to which Tibetan belongs. In Kurtöp, representing the East Bodish languages which are the closest cousins of Tibetan, the copulas are na and wen, the latter < *way-na (DeLancey 2008b). We find the root as a copula in West Himalayan, the closest to Tibetan of the non-Bodic branches, as in Rangpo hwə- (Zoller 1983:68), and in Kiranti, representing East Himalayan, as in Limbu existential wa· (van Driem 1987). Since *way then clearly reconstructs for each of the ancestors of Tibetan, from the nearest node all the way back, its loss in Tibetan calls for some explanation; the explanation is that it persisted into Classical Tibetan as the sentence final particle, and 4.

The initial t in to in ex. (14) is a fossil of a lost perfective suffix on the verb (Przyluski and Lalou 1933). 9

subsequently abandoned its function as a copula. The idea of obligatory final particles which do nothing but mark the end of a sentence may seem a bit redundant to speakers of European languages. But they actually do have a function in Tibeto-Burman languages, all of which have clause-chaining structure. A literary example is (17): (17) a. tad.mo lta-ba- 'i lam-du rgyal.po-'i bu.mo dang phrad nas sights see-NOM-GEN way-LOC king-GEN daughter with meet ABL ‘on the way to see the sights [he] met the king's daughter.’ b. rgyal.po-'i bu.mo la chags te king-GEN daughter LOC desire NF ‘desiring the king’s daughter.’ c. bu.mo dang nyal nas daughter with lie.down ABL ‘[he] slept with the girl.’ d. phyi gsang shor te after secret come.out NF ‘later the secret getting out.’ e. rgyal.po-s zin nas king-ERG seize ABL ‘the king seized him.’ f. gsod-du bcug-pa las kill-DAT/LOC command-NOM ABL ‘and as he ordered him to be killed.’ g. khyim.bdag gi bu-s 'di skad ces gsol to householder GEN boy-ERG this word thus ask FINAL ‘the householder's son asked thus.’ The non-final constructions are the equivalent of conjunctive participles in Indic languages; the to at the end of this sequence of clauses effectively marks the end of an episode. This sort of thing is normal in Tibeto-Burman narrative, where the break – the end of the clause chain – is marked by a finite verb. When a language innovates a new finite construction out of a Nominalized Clause Construction, the resulting sentence final particle takes over the functions of the older finite construction, especially that of marking the end of a clause chain. 3.2 Lolo-Burmese The first serious study of a final particle construction in Tibeto-Burman is Matisoff’s (1972) description of ve in the Loloish language Lahu. Matisoff (1985) identifies this, as well as the Jinghpaw final particle ai, and others, as a reflex of the same Proto-Tiberto-Burman copula *way, which we identified in the previous section as the source of the Classical Tibetan final particle. In fact, this particle is found throughout Lolo-Burmese. Bradley (1979:376-7) independently reconstructs a declarative verb particle *way3 for Proto-Loloish; this is obviously the same etymon. It is reflected in Burmish as well as Loloish, in the “mark of

10

assertion” / / which Vittrant (2002) shows can be internally reconstructed from the final particles of Burmese. Burmese has two “modal” final particles, realis te and irrealis me. There are also two clausal nominalizers, realis ta and irrealis hma, which mark many complement constructions, “cleft” constructions, and, most interestingly, exclamatory independent main clauses. These have exactly the same modal values as te and me. Nominalizers which encode aspectual or realis/irrealis distinctions are not uncommon in Tibeto-Burman (DeLancey 2005; Noonan 2008). So, if we interpret Vittrant’s / / as analogous (even if not cognate, although it would certainly appear to be) to Proto-Loloish *way3, we have here once again a synchronic system of finite main clause marking which derives directly from reanalysis of a nominalized clause construction, with nominalizers ta and hma plus a sentence-final copula.5 4. Summary Previous work on the topic of nominalized clause constructions in Tibeto-Burman has dealt substantially with synchronic issues surrounding the use of nominalized clauses in finite contexts. My concern here has been diachronic, and to some degree typological. We have examined a number of examples of finite constructions in various Tibeto-Burman languages which have finite verb forms or clause constructions which demonstrably developed from nominalized clause constructions. I have not given a survey of languages across the family, but anyone familiar with any Tibeto-Burman language can provide analogous examples. The Sunwar perfects have numerous parallels throughout Bodic, the best known being the Tibetan constructions exemplified by the Lhasa data discussed at the beginning of this paper. The Central Tibetan system exemplified there is no more than a few centuries old: in the vernacular 15th century biography of Milarepa we find the nominalized clause constructions which make up the modern system, but they do not appear to be fully grammaticalized, and have not yet completely replaced the older system (Saxena 1992, 1997). The older Classical system itself appears to be an old nominalized clause construction, so that the contemporary Tibetan languages are now on their second round of reinventing their finite clause structure. Parallel developments can also be found in the Western Himalayan languages (Saxena 1992, 2000), Tshangla (Andvik 1999), and elsewhere throughout the family. The innovation of new agreement prefixes from possessive proclitics on nominalized verbs, as we saw in Kuki-Chin, is not quite so common, but it has parallels in Kham (Watters 2002: 411ff.) and elsewhere. The final particle phenomenon which we saw in Kuki-Chin and Classical Tibetan is particularly important; if these are fossils of old nominalized clause constructions, then we have prima facie evidence for the ubiquity of this grammatical development throughout the family. We saw that Proto-Lolo-Burmese had a nominalized clause construction which already was used as a main clause, and which is the source of the primary finite main clause 5.

Simpson (2008) proposes a quite different history for these forms, in which the nominalizers -ta and -hma derive from the fusion of a construction in which the genitive-marked final particles te. and me. modified haa ‘one’ in a “headless” relative clause, i.e. represent an innovative nominalization construction. I find Vittrant’s account more convincing. In any case, Simpson agrees that the modern finite constructions in Burmese represent old nominalizations. 11

type in the daughter languages. So we see again, as in Kuki-Chin, the finitization of a nominalized clause construction as the basis for the syntactic organization of an entire branch of the family. The same may well be true of the Naga and other less well-documented groups, but it is not possible to make definitive statements without a more secure subgrouping of these languages than we have at present. We can infer that Proto-Kuki-Chin, Proto-Lolo-Burmese, very likely Proto-Bodic, and, for that matter, no doubt Proto-Tibeto-Burman itself, were typical Tibeto-Burman languages, like the modern languages discussed by Bickel, Genetti, Noonan, and others, in which one or more nominalized clause constructions coexisted with an older finite system. A nominalized clause construction with the copula *way is very widely represented, and very likely existed in Proto-Tibeto-Burman – though at that level it presumably would have had the transparency of the modern Sunwar perfects (or, at least, of the slightly older Central Tibetan forms), and the *way would still have been a copula, which function has been largely lost in modern reflexes of the root such as Classical Tibetan ‘o, Jinghpaw ai, and Lolo-Burmese *way3. It is also possible that the proclitic construction in Kuki-Chin is ultimately rooted in a nominalization construction of Proto-Tibeto-Burman provenience (DeLancey 1988, Watters 2002: 406), although if so it must have undergone considerable alteration and regularization in Proto-Kuki-Chin. But we cannot simply explain the proclivities explored in this paper by reconstructing a particular source construction to Proto-Tibeto-Burman. My main point in this paper is not comparative, but typological – it is not simply the case that many Tibeto-Burman languages have undergone such a development, as documented in our examination of Kuki-Chin, Classical Tibetan, and Lolo-Burmese. The striking fact is that they continue to do so, over and over as we have seen in Sunwar and modern Tibetan. This is a consistent, repeated pattern across the family, synchronically and diachronically, and serves as another example of how nominalization is the primary engine driving Tibeto-Burman syntax and syntactic change. References Andvik, Erik. 1999. Tshangla Grammar. Ph.D. dissertation, University of Oregon. (To appear: Leiden: Brill) Bauman, James. 1975. Pronouns and pronominal morphology in Tibeto-Burman. Ph.D. dissertation, University of California, Berkeley. Bickel, Balthasar. 1999. Nominalization and focus in some Kiranti languages. In Studies in Nepalese Linguistics, Y. Yadava and W. Glover (eds.), 271-296. Kathmandu: Royal Nepal Academy. Borchers, Dörte. 2008. A Grammar of Sunwar. Leiden: Brill. Bradley, David. 1979. Proto-Loloish. (Scandinavian Institute of Asian Studies Monograph no. 39). London and Malmö: Curzon. DeLancey, Scott. 1986. Evidentiality and volitionality in Tibetan. In Evidentiality: The Linguistic Coding of Epistemology, W. Chafe and J. Nichols (eds.), 203-213. Norwood, NJ: Ablex. DeLancey, Scott. 1988. On the origins of the Kuki-Chin agreement prefixes. 21st International Conference on Sino-Tibetan Languages and Linguistics, Lund.

12

DeLancey, Scott. 1989. Verb agreement in Proto-Tibeto-Burman. Bulletin of the School of Oriental and African Studies 52: 315-33. DeLancey, Scott. 1990. Ergativity and the cognitive model of event structure in Lhasa Tibetan. Cognitive Linguistics 1: 289-321. DeLancey, Scott. 1991. The origin of verb serialization in modern Tibetan. Studies in Language 15: 1-23. DeLancey, Scott. 1992a. The historical status of the conjunct/disjunct pattern in Tibeto-Burman. Acta Linguistica Hafniensia 25: 39-62. DeLancey, Scott. 1992b. Sunwar copulas. Linguistics of the Tibeto-Burman Area 15(1):31-38. DeLancey, Scott. 1997. Mirativity: The grammatical marking of unexpected information. Linguistic Typology 1(1):33-52. DeLancey, Scott. 1999. Relativization in Tibetan. In Studies in Nepalese Linguistics, Y. Yadava and W. Glover (eds.), 231-249. Kathmandu: Royal Nepal Academy. DeLancey, Scott. 2001. The mirative and evidentiality. Journal of Pragmatics 33(3):371-384. DeLancey, Scott. 2005. Relativization and nominalization in Bodic. Tibeto-Burman Linguistics: Proceedings of the 28th Annual Meeting of the Berkeley Linguistics Society, 5-72. Berkeley: Berkeley Linguistics Society. DeLancey, Scott. 2008a. Towards a history of verb agreement in Tibeto-Burman. presented at the 14th Himalayan Languages Symposium, Gothenburg, August 21, 2008. DeLancey, Scott. 2008b. Kurtoep and Tibetan. In Chomolangma, Demawend und Kasbek: Festschrift für Roland Bielmeier zu seinem 65. Geburtstag. B. Huber, M. Volkart, and P. Widmer (eds.), 29-38. Halle: International Institute for Tibetan and Buddhist Studies GmbH. DeLancey, Scott. to appear. Nocte and Jinghpaw: Morphological correspondences. In Proceedings of the Third Meeting of the Northeast Indian Linguistics Society, S. Morey and M. Post (eds.). Denwood, Philip. 1999. Tibetan. (London Oriental and African Language Library 3). Amsterdam and Philadelphia: Benjamins. Driem, George van. 1987. A Grammar of Limbu. Berlin, New York, and Amsterdam: Mouton de Gruyter. Driem, George van. 1993. The Proto-Tibeto-Burman verbal agreement system. Bulletin of the School of Oriental and African Studies 16.2:292-334. Genetti, Carol. 1992a. Semantic and grammatical categories of relative clause morphology in the languages of Nepal. Studies in Language 16: 405-427. Genetti, Carol. 1992b. Segmental alternations in the Sunwari verb stem: A case for the feature [front]. Linguistics 30: 319-358. Genetti, Carol, E. Bartee, A. Coupe, K. Hildebrandt, and Y-J Lin. ms. The syntax of nominalization of five Tibeto-Burman Himalayan languages. Ms., Department of Linguistics, University of California at Santa Barbara. Hahn, Michael. 1974. Lehrbuch der klassischen tibetischen Schriftsprache. Bonn: Michael Hahn. Henderson, E.J.A. 1957. Colloquial Chin as a pronominalized language. BSOAS 20:323-7.1997.

13

Henderson, E.J.A. 1964. Tiddim Chin: A descriptive analysis of two texts. London: Oxford University Press. Konow, Sten. 1908. Tibeto-Burman Family, Vol. III of George Grierson, Linguistic Survey of India. Calcutta, Office of the superintendent of government printing. (repr. Delhi: Motilal Banarsidass, 1967-8). Konow, Sten. 1911. Tibeto-Burman languages. Encyclopedia Britannica, 11th edition. Matisoff, James. 1969. Verb concatenation in Lahu: The syntax and semantics of 'simple' juxtaposition. Acta Linguistica Hafniensa 12(2):169-120. Matisoff, James. 1972. Lahu nominalization, relativization, and genitivization. In Syntax and Semantics I, J. Kimball (ed.), 237-57. New York: Seminar Press. Matisoff, James. 1985. God and the Sino-Tibetan copula, with some good news concerning selected Tibeto-Burman rhymes. Journal of Asian and African Studies (Tokyo Foreign Languages University) 29:1-81. Matisoff, James. 2003. Handbook of Proto-Tibeto-Burman. (University of California Publications in Linguistics 135). Berkeley and London: University of California Press. Mazaudon, Martine. 1978. La formation des propositions relatives en tibétain. Bulletin de la Société de Linguistique de Paris 73: 401-414. Noonan, Michael. 1997. Versatile nominalizations. In Essays on Language Function and Language Type. In Honor of T. Givón, J. Bybee, J. Haiman & S. A. Thompson (eds), 373-94. Amsterdam/Philadelphia: John Benjamins. Noonan, Michael. 2008. Nominalizations in Bodic languages. In Rethinking grammaticalization: New perspectives, M. J. López-Couso and E. Seoane (eds), 219-37. Amsterdam/Philadelphia: John Benjamins. Przyluski, Jean, and Marcelle Lalou. 1933. Le da-drag tibetain. BSOS 7: 87-9. Saxena, Anju. 1992. Finite Verb Morphology in Tibeto-Kinnauri. Ph.D. dissertation, University of Oregon. Saxena, Anju. 1997. Aspect and evidential morphology in Standard Lhasa Tibetan: A diachronic study. Cahiers de Linguistique, Asie Orientale 26(2):281-306. Saxena, Anju. 2000. Diverging sources of new aspect morphology in Tibeto-Kinnauri: External motivation or internal development. In Historical linguistics 1995. Volume 1: General issues and non-Germanic languages, J. Smith and D. Bentley (eds.), 361-375. Amsterdam/Philadelphia: John Benjamins. Simpson, Andrew. 2008. The grammaticalization of clausal nominalizers in Burmese. In Rethinking grammaticalization: New perspectives, M. J. López-Couso and E. Seoane (eds), 265-88. Amsterdam/Philadelphia: John Benjamins. Stern, Theodore. 1963. A provisional sketch of Sizang (Siyin) Chin. Asia Major 10: 222-78. Stern, Theodore. 1984. Sizang (Siyin) Chin texts. Linguistics of the Tibeto-Burman Area 8(1):43-58. Vittrant, Alice. 2002. /Ta/ et /Ma/, deux nominalisateurs particuliers du birman. In Aspects de la prédication. S. Leroy and A. Nowakoska (eds.), 335-51. Montpellier: Université Paul Valéry. Watters, David. 2002. A Grammar of Kham. Cambridge: Cambridge University Press. Zoller, Claus. 1983. Die Sprache der Rang pas von Garwhal. Wiesbaden: Harrassowitz.

14

Suggest Documents