Folia Linguistica Historica Volume 28 Issue 1 2007 [Doi 10.1515_flih.2007.145] Gerner, Matthias -- The Lexicalization of Causative Verbs in the Yi Group

The lexicalization of causative verbs in the Yi group1

Matthias GernerCity University of Hong Kong

ABSTRACT. This paper explores a lexicalization process in an array of Yi(^ Tibeto-Burman) languages spoken in Southwest China. Almost every Yilanguage exhibits a set of minimal pairs, called simplex-complex pairs, of thefollowing sort. The simplex member is an adjective or verb starting with avoiced consonant (e.g. ve33 'sway'). The complex member is derived from theformer by devoicing the initial consonant and by augmenting its valency oradding causative meaning to it (e.g./e33 'wave = cause to sway*). Previousauthors have treated this kind of data as a blind process of sound change andwere undecided on the type of linguistic change, i.e. whether it represents akind of grammaticalization or lexicalization. With pan-Yi data and statisticalfigures it is demonstrated that the sound changes constitute a phoneticallyand semantically motivated process of lexicalization.

1. Introduction

The aim of this paper is to report and evaluate a remarkable process oflexicalization displayed in the lexicon of an array of Yi languages which areTibeto-Burman languages spoken in Southwest China (see appendix formore spatial-genetic information). For a limited set of mainly monosyllabicverbs - between 5 and 50 depending on the language - it is possible todevoice the initial consonant and derive a causative verb. In Weining Neasu(one of the Yi languages surveyed), there is for example the verb ve33meaning 'sway* being involved as the predicate of an intransitive clause (e.g.'the tree swayed by itself). Its devoiced counterpart fe33 'wave = cause tosway* appears as the predicate of a monotransitive clause (e.g. 'He waved the1 Part of this research was supported by a Strategic Research Grant (Project No.

7001921) of City University of Hong Kong.Folia Linguistica Historica 28/1-2 (2007), 145-185.ISSN 0165-4004, E-ISSN 1614-7308 Mouton de Gruyter - Societas Linguistica EuropaeaBrought to you by | University of Sussex

Authenticated | 172.16.1.226Download Date | 8/7/12 7:38 PM

146 Matthias Gerner

handkerchief). Throughout this paper, I will term the voiced component asthe simplex and the devoiced member as the complex of the pair.

The correlation devoicing-causation has been recognized before, mainlyin the Chinese literature, without, however, characterizing it as a motivatedprocess of lexicalization. The best-known western account is MatisofPs workon Lahu, a close genetic relative of the Yi languages (cf. Matisoff 1973: 32-34,1975, 1976: 414-419). Matisoff accounts for this phenomenon in aperspective of reconstructing PTB (Proto-Tibeto-Burman). He regards thedevoicing process as the vestiges of a reconstructed PTB prefix *s- thenfunctioning as transitivizer or causitivizer (see also Benedict 1972: 105-108).Synchronic data of two conspicuous Tibeto-Burman languages are adducedin this context, those of Written Tibetan2 and Jinghpaw/Kachin3 where verbsallow the prefixation of *s- (Jinghpaw: fa- or f a - ) to derive their causativependants. According to Matisoff, the PTB prefix *s- had successively evolvedinto a Proto-Loloish causative prefix *?-. (The Loloish or, more precisely, theBurmese-Lolo languages constitute one of the six components of the Tibeto-Burman language family.) The Proto-Loloish prefix *?- in turn was realizedin particular Loloish languages through various supra-segmental phenomenasuch as devoicing, aspiration and tone alterations. In another contribution,La Polla (2000) analyzed the various strategies by which Dulong/Rawangchanges the valency of clauses. Dulong/Rawang is a Central Tibeto-Burmanlanguage spoken on both sides of the China/Myanmar frontier. He reportsthe existence of a causative prefix siu21/tiu21 (tiu21 is used before voicelessfricative initial, while sui21 is used before all other initials), but not of anysuprasegmental phenomena as in the Yi languages. The Dulong/Rawangprefix may actually reflect the preservation of this prefix at an early PTB-stage. We will elaborate on this idea in section 5.2.

Besides a variety of individual Tibeto-Burman grammars, which mentionthe devoicing-causation correlation, focused accounts are mainly offered byChinese linguists. The articles of Chen Shilin et al. (1962) and Chen Kang(1990) report the devoicing-causation connection for Liangshan Nuosuwhich also belongs to the 11 Yi languages surveyed in this paper. Dai (1981)

2 Written (classical) Tibetan is classified in the Western or Bodish branch of TB

languages.3 Jingphaw/Kachin is a language spoken in northern Burma with minor extensions

into western Yunnan (China) and India. It belongs to the Sal branch of TBlanguages. See also the next footnote.

Brought to you by | University of SussexAuthenticated | 172.16.1.226

Download Date | 8/7/12 7:38 PM

The lexicalization of causative verbs in the Yi group 147

and Xu (1984) explore causative verbs in Jingphaw/Kachin. The article ofDai (1981) includes the information that in Zaiwa (a Burmese language4 ofthe Burmese-Lolo language group within the TB family), supra-segmentalprocesses such as aspiration and tone alteration are involved in thederivation of causative verbs. These processes seem to coexist with thecausitivizer prefix fa- mentioned above, albeit on different segments of thelexicon (see also Xu 1984). These data from the Burmese language Zaiwaseem to expand MatisofFs view which took suprasegmental causativeprocesses to hold only for Loloish languages (not necessarily for Burmishlanguages).

Finally there are works of Chinese linguists attempting to reconstruct andclassify the category of causative verbs for the whole range of Tibeto-Burmanlanguages. Du (1990) views PTB as exhibiting an agglutinative and laterinflectional morphology type. Causation belonged to the grammatical web ofthe verb at an equal footing with, for example, tense and aspect. Successively,when Old-TB shifted its morphological type to isolating, causation as agrammatical category was shed and became a mere lexical phenomenon.This vision is also supported by Dai (2001) who offers a reconstruction of thecausative in PTB. Similar to Matisoff, Dai establishes a proto prefix *s- forPTB, but in contrast with the former, Dai does not involve an intermediateglottal prefix *?- from which the suprasegmental processes would havestarted. Rather, Dai regards the direct merging of *s- with Stative verbs as thesource of the processes of devoicing, aspiration, etc. Curiously, Matisoff searlier reconstruction is not mentioned by Dai (2001). Sun's paper (Sun1999) is a synchronic classification of the morpho-syntactic strategies thatcertain, mainly Chinese, TB languages employ to express the category ofcausative verb. These strategies are prefixation, suffixation and supra-segmental processes, i.e. devoicing, aspiration and tone alteration.

4 Native speakers of Jinghpaw and Zaiwa share a common culture and residence

area. The superordinated ethnic group including both Jinghpaw and Zaiwa isreferred to as Kachin (in Myanmar) and Jinghpaw/Jingpo (in China). The majorityof these people live in Myanmar, where speakers of the Jinghpaw form the majority,while in China native Zaiwa speakers constitute the largest part of the Jingponationality and native Jinghpaw speakers form a tiny fraction. Nevertheless, becauseof historical factors not yet explained, native Zaiwa speakers consider themselves tobe Jinghpaw, even though both languages belong to different branches of the TBfamily (cf. Bradley 1997).



148 Matthias Gerner

The current paper differs somewhat from earlier contributions as itattempts to argue that the devoicing-causative correlation is not the result ofa blind process of sound change, as implied by previous authors,5 but rather amotivated process of lexicalization by which an entire lexical class came intoexistence. The structure of this paper is as follows. In section 2, I discusssome terminological issues, namely the notions of causation and lexicaliza-tion. A large part will be devoted to the description of the phenomenon fromthe phonetic and the semantic angle. In section 3, I produce a number ofcomprehensive tables where the causative verbs of the whole range of Yilanguages are represented and ordered with respect to the devoicedconsonant's articulation point and mode. In section 4,1 introduce the set ofWeining Neasu's causative verbs. With nearly 45 causative verb pairs,Weining Neasu is the TB language with the highest proven number of verbpairs. Weining Neasu's causative verbs are arranged according to thesemantic features of valency, control and situation type. In section 5,1 offersome statistics on phonetic and semantic properties of causative verbssuggesting that the sound changes represent a motivated process. In theappendix, spatial-genetic information of the Yi languages surveyed in thispaper is provided.

2. Terminology

In this section I want to lay some terminological foundations. In section 2.1,a number of semantic parameters transpiring from seminal works oncausation are introduced. I will moreover assemble some recent insights onthe issue of lexicalization (section 2.2).

2.1 Semantic parameters of causatives

There are many issues related to the concept of causation that have enrichedlinguistic theories and cognitive sciences in the last 40 years. Unfortunately,there is no way of doing justice to them in this introductory section. Below, Ibriefly discuss a few outstanding works that may be involved in support ofthe claims advanced in this paper, namely the hypothesis that thelexicalization of causative verbs in Tibeto-Burman is shaped by the causation

5 Some authors even suggest the term of grammaticalization, even though it is rather

a process in the opposite direction.




type of the simplex verb. Various causation types have been proposed byscholars depending on the semantic parameters taken into account. Morespecifically, four basic factors on which causation types are coined have beenidentified:

(a) a participant X, the causer,(b) a participant Y, the causee,(c) an event e called by some as "precipitating event" (Frawley 1992:

159), "causing event" (Shibatani 1976:1) or simply "cause" (Song1990: 10, 1996: 19-20),

(d) a situation s labelled as "result" (Frawley), "caused event" (Shibatani)or "effect" (Song 1990: 10, 1996: 19).

These parameters led to a widely agreed distinction between so-calledsingle-event and double-event causatives. Single-event causatives refer to aunique transitive event e(X, Y) in which the caused situation s is an integralpart of the event e: s(Y) ^ e(X, Y). Double-event causatives dissociate e and s:s(Y)

150 Matthias Gerner

(2) (a) 'John kicked the ice but nothing happened to it/(b) *7ohn melted the ice but nothing happened to it/

Both events constitute single-event causative events, but Shibataniclassified (2a) as uncausative and (2b) as causative. Yet, the acceptability of(2a) depends on how narrow one defines the notion of 'happen'. If itincludes the affectedness of the patient, then the state of 'being kicked' leadsto a contradiction in (2a). Consequently, transitive clauses that imply theaffectedness of the patient cannot be distinguished from single-eventcausatives. Nevertheless, there appears to be some ground for discriminatingbetween single-event causatives and transitive situations with no palpableimpact on participants (e.g. 'I know him').6

As we will see in the course of this paper, all causative verbs presented insections 3 and 4 are involved in single-event causatives and the distinctionsingle/double does not yield a sufficiently explanatory backbone for theprocess of lexicalization featured herein. A different finely graded list ofparameters is proposed by Dixon (2000: 62) who suggests nine semanticfeatures that causative constructions react to or show sensibility to:

6 Comrie (1975, 1976) takes a multi-stratal approach on causatives (in spite of his

typological mono-stratal inclinations) by posing a deep uncausative structure fromwhere causatives are derived through addition of a causer argument. The mapping ofthe grammatical relation of the causee is proposed to proceed from a case hierarchy:subject > direct object > indirect object > oblique. The case hierarchy predicts thegrammatical relation of the causee in the derived causative construction. The causeewill be mapped on the first grammatical relation of the case hierarchy that has noargument status for the underlying uncausative verb. The main problem withComrie's theory is the empirical lack of languages that follow this type of caseassignment (cf. Palmer 1994: 219). A typological work of a different type is Song'scross-linguistic endeavour based on 408 languages, cf. Song (1989, 1990, 1996). Hedistinguished three causation types: the COMPACT type, the AND type and thePURP type. The COMPACT type corresponds to single causatives and the ANDtype is comparable to double-event causatives. The PURP type refers to a third typeof causatives wherein the causing event is performed purposefully for obtaining thecaused situation.




(I) Relating toverb

(II) Relatingto causee

(III) Relating tocauser

(i) State/action.

(ii) Transitivity(intransitive, monotransitive, ditransitive).

(iii) Control over caused situation(lacking or having control),

(iv) Volition or willingness of causeto perform in the caused event,

(v) Affectedness of causeeby the causing event

(vi) Directedness: does the causer actdirectly or indirectly?

(vii) Intention of causer(accidental or intentional achievement of result),

(viii) Naturalness(caused event happens fairly natural or with efforts)

(ix) Involvement(causer involved or not involved in activity)

As it will become obvious in section 4, most causative verb pairs of the Yigroup that have undergone the process of lexicalization display for thecausee of the simplex verb a lack of control over the caused situation. Theseobservations will be made plain and exploited in section 5.

2.2 On the notion of lexicalization process

Lexicalization is acknowledged in the literature in a synchronic as welldiachronic perspective. Synchronically, it mainly refers to the encoding ofconceptual categories by words of the lexicon. This perspective is basicallypursued by Generative and more recently by Cognitive Semantics. Thecurrent trend of this line of research is cross-linguistic investigation onpossible conflation patterns of complex conceptual structures into one lexicalitem (e.g. McCawley 1968; Jackendoff 1990, 2002). Moreover, some scholarshave explored productive word formation such as compounding andderivation as cases of synchronic lexicalization (e.g. van der Auwera 2002:20). Diachronically, lexicalization is "the adoption into the lexicon of itemsthat have fallen outside the productive rules of grammar" (Brinton andTraugott 2005: 18; Anttila 1989 [1972]: 151). Diachronic lexicalization can



152 Matthias Gerner

follow two directions: (i) fusion of parts in a compound or complex word;7(ii) change from bound morpheme to simple lexeme8 (Brinton and Traugott2005: 61). As for fusional lexicalization, scholars have distinguished twomajor subtypes:

- Unproductive word formation processes: These are formerly productiveprocesses of word formation which have become frozen, irregular andunproductive (e.g. Bauer 1978: 6). The English derivational suffix-dom might be cited as an example of a formerly productive nowfrozen word formation process.

- Demorphologization: Hopper (1990: 154) defines demorphologizationas a process by which "a morpheme loses (most of) its grammatical-semantic contribution to the word and becomes an indistinguishablepart of the construction of the word, while retaining part of itsoriginal phonological substance." The English word alone (all + one)may be cited as example of demorphologization (Brinton andTraugott 2005: 52).

As we will see in the following sections, the devoicing-causation processof this paper is a formerly productive now unproductive word formationprocess generating during its course of fossilization some supra-segmentalalterations.

3. The causative verbs of the (other) Yi languages ordered by thepoint and mode of articulation

I surveyed the devoicing-causation correlation in 11 Yi languages representingthe whole range of Yi groupings. It turned out that the correlation in questionis attested to various degrees in the lexicon of these languages. As mentionedabove, Weining Neasu is the language with the highest number of causative

7 Brinton and Traugott feature this type of diachronic lexicalization as "increase of

dependency" (2005: 61). For example, 'cobweb' is a lexicalized noun that in OldEnglish consisted of two components: coppe 'spider' and web Veb' (cf. Brinton andTraugott 2005: 50).8 This type produces lexical items through an increase of autonomy of a formerly

bound morpheme. However, the examples that scholars have adduced to justify thiskind of lexicalization are controversial (cf. Brinton and Traugott 2005: 57-61 formore details).




pairs (nearly 45). On the other end of the spectrum lies Luoping Nase (seeappendix) with only one or two attested cases of devoiced causative pairs. Yet,in Luoping Nase a set of ambitransitive verbs can be used either as intransitiveor as transitive predicate. Some of these verbs are monosyllabic with voicedinitial consonant for the simplex as well as for complex forms. The other Yilanguages lie in between these extremes.

For all Yi languages, mono-syllabicity is a predominant lexical phenome-non, while two-syllabic words also make up a significant part of the lexicon.Multi-syllabicity, however, is much rarer. Yi languages and, more generally,Loloish languages are what Matisoff (1994: 115-116) calls omni-syllabictonal languages with an extremely simple core syllabic canon: (C) VT.

Devoicing is the basic phonetic process involved in the production ofcausative verbs. Additional suprasegmental changes such as aspiration, tonealteration, and sometimes even alteration of the point of articulation, oftenjoin in. In this section, the data are arranged according to the point ofarticulation of the initial consonant (bilabial, labio-dental, alveolar, retroflex,alvelo-palatal and velar) and secondarily by the mode of articulation (stop,nasal, fricative and lateral). It appears that the devoicing-causationcorrelation has effectively spread across the whole range of consonants.

3.1 Devoicing of bilabial stops

Example (3) displays the causative verb pair 'carry' / 'make carry' in GejiuNesu. The (a) version always incorporates a simplex verb, the (b) versionemploys a complex verb.

(3) Gejiu Nesu(a) k3>55 a55ko33 t

154 Matthias Gerner

cases, namely the verb pairs with identical simplex and complexcomponents. Besides Luoping Nase mentioned above, this concerns also onepair in Longlin Ngopho and two pairs in Mile Azhee.

'peel off'scrape off'carry on back''make carry''flow''cause to flow''collapse''make collapse''loose''loosen'Open' (intr.)Open' (tr.)'roll' (intr.)'roll' (tr.)'explode, crack' (intr.)'explode, crack' (intr.)'appear''reveal'

Nam

e of

Lang

uage

SimplexComplexSimplexComplexSimplexComplexSimplexComplexSimplexComplexSimplexComplexSimplexComplexSimplexComplexSimplexComplex

Li

angs

han

N

uosu

bu21

phu21

mbo33

po21

bi33

pi33

bi55

pi55

(2) W

eini

ng N

easu

be21

phe21

bu21

phu21

bu33

pu55

bi13

phi13

JGQu 4>

t 1! 't-3 3

_P93_3__pV3_pho33 pV3

pui2'pui21

bi21bi21

(5) Sh

izon

g K

opho

M

ile A

xi

(7) M

ile A

zhee

be13

2'

br21

bv21

bu33 po33

phu33 po33

=3.fl ,

g 3 *2 1 2:2 ff lo i l

ba33

pha33

ba55

pa55

br33

PY33

phui33 bi55

pui33 pi55

Figure 1: Devoicing of bilabial stops in the Yi group

3.2 Devoicing of labio-dental fricatives

The pair 'wear' / 'dress' has a devoicing process for almost every Yi language.In most languages of the sample it sounds like v/* / fi*. Only in LiangshanNuosu has the corresponding pair a velar onset: ga55 Vear' / ka55 'dress'(see subsection 3.6).




Name of Language Weining Neasu Longlin Ngopho Luoping Nase(5) Shizong Kopho(6) Mile Axi Mile Azhee Gejiu Nesu Wuding Aluphu

'wear'

Simplexvi13vei55vi13vi13vi33vi21vi21vi33

'dress'Complexfi13fei55VJ'3

fi13fi55fi21

fi21fi33

'sway'Simplexve33

'wave'

Complexfe33

Figure 2: Devoicing of labiodental fricatives in the Yi group

3.3 Devoicing of alveolar consonants

The group of alveolar consonants is well represented allowing the devoicingof both alveolar stops and laterals. Stops are the mode of articulation wherethe lexicalization process has spread and concentrated most of its output.

3.3.1 Devoicing of alveolar stops

Again, there is one pair in Longlin Ngopho, and one in Luoping Nase whereboth simplex and complex forms are of the same voice level. From the widespectre of causative verbs starting with alveolar stops, I have selected twoexamples, one illustrating the pair 'say* / 'tell' (Luoping Nase),9 the otherdemonstrating the couple 'study' / 'teach' (Weishan Lalo).10

9 There is no direct causative relation between 'say' and 'tell' in the sense that the

causee of 'tell' would be the agent of saying. Rather, 'tell' can be derived from 'say'through a valency increase of a different type. The new semantic role added to theframe of 'tell' is not the causer but rather a new role for recipient or beneficiary.10

Besides Weishan Lalo, the causative verb pair 'study' / 'teach' is elsewhere attestedin Lahu, a language related to the Yi languages spoken in parts of Yunnan Province(P.R. of China) and Thailand, see Matisoff (1975).



156 Matthias Gerner

S = Simplex

C = Complex

'drink''give todrink''eat'

'feed''say''tell''be leftover'

'leave over''stand''build'/'place''sit''make sit'/'build''lie down''place''break'(intr.)'break'(tr.)'split'(intr.)'split' (tr.)'collapse''makecollapse''study,learn''teach'

Nam

e of

Lang

uage

SC

SCSCS

CSC

Sc

scs

c

s

csc

s

c

2 -2 S 51 1 1 J a 1 -f !c Z ^ Z ^ > 3 >-> >-i ^ P c " o b ' o , o 3 - 5P ^rt ^ 7 ~ ^ D c 3

ndo33 ndo33 to33 ntfio21 do13 du33 da21 du55 do33 do33

to21 to21 to55 ntfio21 to33 tu33 te21 tu33 to33 to33

dzui33 dzu33 dz21 dzo21 dzo21 dze33 dza21 dzo33 dzo33

t$a33 tu33 tû55 tô33 t?o55 tse33 jtqa21 tô33 t?o33die55 di21 dzo33

thi2i th ie2i the2i

dzi33 dz21 dze33 dze33

tshi33 tsi13 tse33 tse33

dzu33 dzo33

tshu33 tso33

di33

ti33

da21

ta21

dui33

thui33

dz55

tsi55

d3>55 de33

thg-55 the33

dzu55

tsu33

Figure 3: Devoicing of alveolar stops in the Yi group




S = Simplex

C = Complex

'ascend''lift up''enter'

'bury= place in''wear'

(hat, shoe)'dress'(hat, shoe)'bright,shine''light,kindle''appear,exit''reveal''beperforated''perforate''burn'(intr.)'burn' (tr.)'peel off'scrape off'turbid,muddy''stir up'

Nam

e of

Lang

uage

SCSC

S

C

S

C

S

C

S

CS

CSCS

C

2 02 2 2z i o 1 c Z z0 * 3 S => 3If ! t s? 2 1 ! I I I t 3 > J J c ^ S ^ O > @

di33

ti33dy21

ty21

ndi55 di13 d321 dy21

ti55 ti13 t921 ty21

do13

to13

dr55do3 dou553

thY55 tu55

du21

thu21

ndu55 du21

tu55 tu21

dui21

thuj21de21

te21

Y

ongr

en Lo

lo

W

udin

g Alu

phu

dv55

tY55

Figure 3: Devoicing of alveolar stops in the Yi group (contd.)



158 Matthias Gerner

(4) Luoping Nase(a) tgra21 ?u21 tha55 mi55?u21 die55.

(5)(a)

3PSG 1PSG with word

'He is talking to me/speak

(b) ?u21 no551PSG affair

ta55 tar)NUM:1 CL

33 h 2 1thie2PSG tell

tell you something/

Weishan Lalo (Bj rverud 1998:139)de21 dzu55 li33.hit study IMP

'Learn to play!*

(b) de21 qa55 tsu33 li33hit 1PSG teach IMP

'Teach me how to play1/

3.3.2 Devoicing of alveolar lateral

There is only one pair in this category that constitutes a genuine case of ade voicing-causative correlation. The other two cases exhibit ambitransitiveverbs with intransitive as well as transitive uses (of the type S = O). TheWeining Neasu pair lr33 'move' (intr.) / tr33 'move' (tr.) is illustrated by thefollowing examples:

(6) Weining Neasu33

ga55 dze13 ko55 lY33.DEM:DIST CL DUR move

'That tree is moving [e.g. its branches and leaves by the wind]/33

-?a55 dze13 Qi21 le55 tr33.)EM:DIST CL 3P SG PASS move

'That tree is moved by him/her [e.g. by shaking the stem]/

(a)

(b)

sitree

SItree

Name of Language Weining Neasu Luoping NaseM eAxi

'move' 'make move'Simplex Complexhr33 4r33

'flow' 'make flow'Simplex Complex

lui33 lui33

'roll' 'make roll'Simplex Complex

122 122Figure 4: Devoicing of alveolar lateral in the Yi group




3.4 Devoicing of retroflex stops

The next group regroups causative verbs starting with a retroflex stop. Thiskind of stop belongs to the consonant canon of most Yi languages. Themajority of these stops are affricated. Only Weining Neasu (out of the set ofYi languages surveyed in this paper) has genuine unaffricated stops (i.e.

160 Matthias Gerner

(7) WeiningNeasu(a) se21vu33 mo21 tha21 mo13 se21vu33 khje55 c&e55.

peach tree fruit NUM:1 CL peach tree LOOon bear, beat

Teaches are on the peach tree/

(b) tol3b13 kje33


'be lacking''take off'left over''leave over''be afraid''frighten''go across''make go across''graze, pasture''turn out to grass''burn' (intr.)'burn' (tr.)'melt' (intr.)'melt' (tr.)

Nam

e of

Lang

uage

SimplexComplexSimplexComplexSimplexComplexSimplexComplexSimplexComplexSimplexComplexSimplexComplex

Li

angs

han N

uosu

dso33Jteo33

d?33jt(?933

dzi33Jtei33

(D W

eini

ng N

easu

d?e13jt

162 Matthias Gerner

3.5.3 Devoicing of alvelo-palatal fricatives

Within this devoicing category there is one uncorrupted example (that ofYongren Lolo) and one tainted example (that of Mile Axi whose complex isnot a fricative but an affricated stop). The following two sentences illustrateYongren Lolo's causative verb pair.

(10) Yongren Lolo(a)

(b)

be33tsi33 tS833lu33 thi21 kV3 si213P SG garment beautiful NUM: 1 CL wear

'She wears a beautiful garment.'

o21mo33 ^p21 thie21 be33tsi33 ge55mother 3P SG BEN garment PAT

'Mother dressed her with a garment.'

do33.DUR


'wear'

'dress''be afraid''frighten*'hear''tell''enclosed''enclose, arrest''return''bring back''curved, bowed, bent''curve, bend, bow''split, burst' (intr.)'split, burst' (tr.)'melt' (intr.)'melt' (tr.)

Nam

e of

Lang

uage

SimplexComplexSimplexComplexSimplexComplexSimplexCompkxSimplexComplexSimplexComplexSimplexComplexSimplexComplex

Li

angs

han

N

uosu

ga55ka55

gu33ku33gin33km33

gI33

ki33

W

eini

ng N

easu

gv33k*33

go'3

kho13gi33ki33

(9) W

eisha

n La

lo

gu33ku33

g355

ka55

Y

ongr

en Lo

logo21ko21

Figure 9: Devoicing of velar stops in the Yi group

4. Causative verbs in Weining Neasu ordered by semantic parameters

Weining Neasu occupies a remarkable position among the 11 Yi languagessurveyed in this paper. I counted a total of nearly 45 causative verb pairs infield data available to me. As mentioned above, the complex component ofthese verb pairs points to single-event causatives. Recalling Dixon's nineparameters of causatives mentioned in section 2.1 (cf. Dixon 2000:62), we willorganize the Neasu causative verb pairs around two semantic traits: the kind ofvalency extension, namely from intransitive to monotransitive (cf. section 4.1)or from monotransitive to ditransitive (cf. section 4.2), and the degree ofcontrol the causee has over the caused situation. Let us mark by S the soleargument role of the simplex verb in case that it is intransitive and by A themain argument of the simplex verb if and only if it is monotransitive. (For thesemantic macro-roles S and A, see Dixon 1979,1994).



164 Matthias Gerner

4.1 Causation as valency extension from one to two places

Causative verb pairs that exhibit a valency extension from one to two placesmake up the lion's share of all causative verb pairs with ca. 35 occurrences.SemanticaUy, the simplex verb may be involved in states, processes andactions with a unique participant S. We distinguish the case where S lackscontrol (section 4.1.1) and where it is in control of the situation (section4.1.2).

4. LI S of simplex verb lacks control

Intransitive simplex verbs that incorporate an S argument lacking controlover the situation may be involved in states (section 4.1.1.1) and processes(section 4.1.1.2). Both subcategories comprise the majority of all causativeverb pairs.

4.1.1.1 S of simplex verb lacks control of state

About seven simplex verbs in my sample are stative and describe apositional, structural or appearance state. The corresponding complex verbsadd a generally human causer who is directly and intentionally involved inthe activity. The S argument of the simplex verb equals the causee of thecomplex verb and both lack control in the respective situation they areengaged in.

Simplexc(255 'be at' (locational)dzo33 'stand' (for immobile objects)gr33 'enclosed'd^e13 'be lacking'(fr55 'loose'do13 'bright'clo55 'shine'

Complexthe55 'put at' (lit. cause to be located)tso33 'build'kr33 'enclose, arrest';fc

_ The lexicalization of causative verbs in the Yi group 165

verb tee13 'take off should be analysed as 'cause something to be missing'originating from the simplex verb (foe13 'be missing'. The causative verb pairin (16a+b) refers to a state of appearance.

(12)(a)lu21 dzD33 bo21 khV5 te13.city be built mountain LOCron placed

The house is built on a mountain.'

(b) 2155 hi21mo21 tha21 dzr33 tso33.IP PL house NUM:1 CL build

"We build a house.'

(13) (a) mu33 ko55 gv33 lo33.horse inside enclosed ADV

The horse is enclosed inside.'

(b)qo21 tgho55je21 tha21 jo21 kje33 kv33 13.1PSG friend NUM:1 CL COV arrest SEND

"I have arrested a friend.'

(14) (a) se33la13 na55 kV3

166 Matthias Gerner

4.1.1.2 S of simplex verb lacks control of process

Intransitive simplex verbs that refer to an uncontrolled process constitute themost productive group of verbs targeted by the causative lexicalizationprocess. Causative verbs pairs of this kind can be subsumed in turn under anumber of labels, namely verbs of motion, of disposition and verbs ofphysical affectedness. The first subgroup comprises simplex verbs that areverbs of motion.

Simplexbu33ve33

lY33da33

:21be21

'roll' (intr.)'sway''shake, move' (intr.)'collapse''collapse, come down''stretch out' (intr.)'fall down'

Complexpu55fe33ir33tha33fa'3thr33tH921phe21

'roll (tr.), incite''wave'

'shake, move' (intr.)'cause to collapse, destroy''cause to collapse''stretch out' (tr.)'throw down, cause to fall'

Figure 11: Pairs with uncontrolled simplex verb of motion (1st subgroup)

The motion that is expressed by the simplex verbs is basically preservedin the complex verbs with a few secondary meanings such as 'cause to roll' >'incite, provoke' (for pu55) in (17b) or 'cause to collapse' > 'destroy' (fortha33} in (19b). The complex verb appends a causer who directly andintentionally performs the activity in each case.

(17) (a) mo21ta13 ?o13mu33 bu33 lo33.ball oneself roll ADV

The ball rolls by itself.'(18) (a) si33thu33 tha21 33 ve33.

tree leave NUM: 1 CL sway

Tree leaves are swaying.'

(19) (a) hi21 tho33 (la33 lo33.house ART:PROX collapse ADV

The house collapsed.'

(b) go2l P SG snow

vu33 mo21ta13 pu55.ball roll

am rolling a snow ball.'(b)


(20) (a) se21vu33 si33 khje55 be21 za13 li21.peach tree LOC:on fall descend come

The peach has fallen from the tree/

(b) b33di33 tha21 kV3 ?i2ltha55 ci21 phe21 za13 ka55.bowl NUM:1 CL top of building 3P SG throw descend add

'He threw that bowl down from the top of the building/

A few simplex verbs refer to states of disposition a subject may be foundin. Two causative verb pairs in my data set belong to this subtype and arelisted in figure 12 below. The related complex verbs point to intentionalactivities performed by a causer and endured by a causee.

Simplexbu21dz21

Open' (intr.)left over'

Complexphu21tsi13(tu33)

'open' (tr.)leave over'

Figure 12: Pairs with uncontrolled simplex verb of disposition (2nd subgroup)One specific comment on the use of the complex verb tsin leave over' in

(22b) needs to be made. Even though it can occur as the sole predicate of theclause, it is often complemented by a beneficiary NP marked by the verb bi55'give1. The predicational frame of tsi13 is therefore basically monotransitive.

(21) (a) rjgo21 ?o13mu33 bu21 lo33. (b) Qi21 ggo21 kje33 phu21 33.door oneself open ADV 3P SG door COV open SEND

'The door opened itself/ 'He/she opened the door/

(22) (a) Jtea33 dz21 lo33. (b) na21 tea33 tsi13 tu33 rjo21 5521.food leftover ADV 2P SG food leave over FIX 1PSG give SEND

'Some food has been left over/ 'Keep some food for me.'

The main subgroup of uncontrolled simplex verbs is displayed in figure13 and refers to processes of physical affectedness and alteration. The voicedversion hints at an intransitive process of physical change, and the devoicedversion at a transitive activity. In English, by contrast, the causative verbs infigure 13 are often translated as ambitransitive verbs of the S = type(Dixon and Aikhenvald 2000: 4-5) where a noun phrase may occupy boththe S and roles of the ambitransitive verb as exemplified in 'the newspaperburnt' and 'he burnt the newspaper'.



168 Matthias Gerner

Simplexdu21dv13


33 21 33(25) (a) vu33 dUi21 lo3snow melt ADV

'The snow melted/

(b) vu33 kje33 tci13 dze33 ji21 13.snow COV melt then water get

'Melt the snow and get water/

(26) (a) do21 mo21 na55 go13.stick CL DEM:MED bent

'The stick is bent/

(b) go21 do21 mo21 na55 kV3 *i33to13.1PSG stick CL DEM:MED bend turn

have bent the stick/

(27)(a)ci2 1 do13 zo21.3P SG poisened endure

'He endured intoxication/

(b) Qi21 mo55 su21 le55 to13.3PSG COV:pointto others PASS poison

'He was poisoned by others/

(28) (a) mbu33 ?o13mu33 tse33 gi33 lo33.clothes oneself then torn ADV

'The clothes are torn by themselves/

(b) Qi21 mbu33 go21 le55 ki33 13.3PSG clothes IP SG PASS tear SEND

'His clothes were torn by me/The fourth and last subgroup of uncontrolled simplex verbs has one

member, the pair fao33 'be afraid' / ieo33 'frighten' which has anuncontrolled simplex verb with mental meaning. The following sentencemay serve as illustration. The simplex dzo33 'afraid' must be completed byfGi33 that occurs after the noun phrase that indicates the source of fear.

(29) (a) go21 *^33 pi21 ipi33. (b) go21

170 Matthias Gerner

4.1.2 S of simplex verb is in control of action

In the Neasu data set there is a small group of intransitive simplex verbswhose sole argument is in control of the action referred to. The simplexverbs in question are verbs of movement for which the moving entity exertscontrol over the movement. As for all causative pairs displayed in this paper,the corresponding complex verbs add a generally human causer who isdirectly and intentionally involved in the activity, though without movinghimself.

Simplexdr55do33iwio21dV5

'appear*'go across''come out*

ComplexthY55

tco13

tV5'reveal''cause someone to go across*'set free*

Figure 14: Pairs with controlled simplex verb of movement

Below are provided examples. The complex verb^o13 'cause someone togo across' in (32b) manifests a restriction of usage in that it must co-occurwith the polysemous verb ^V55 'set free, allow' in a serial verb construction.

(31) (a) rja33 ga55 kV3 jo13mu33


4.2 Causation as valency extension from two to three places

Besides intransitive verbs, monotransitive verbs were also targeted by thedevoicing process of lexicalization. Causative verb pairs that constitute avalency extension from two to three places occur with significantly lowerfrequency. I numbered less than ten occurrences in the field data available tome.11 We will again organize the causative verbs by whether or not the mainargument A exerts control over the situation referred to by the simplex verb.

4.2.1 A of simplex verb lacks control of state

There are two causative verb pairs where the main argument of the simplexverb has no control over the situation as provided in figure 15. Both aresimilar in meaning referring to the pair 'hear' / 'tell'. In both cases, thesimplex verb 'hear' is arguably a non-control verb. This kind of causativepair is elsewhere only attested in Liangshan Nuosu (refer to pair gut33 'hear'and kui33 'tell' in figure 9).

SimplexcUo33du55nu13

'hear''hear, understand'

ComplexiCO33

tu55'tell, cause to hear''tell'

Figure 15: Pairs of the 'hear' / 'tell' type

In Neasu, a difference appears for the complex verb of both causativepairs. The verb^o55 'tell' can make up the sole and independent predicateof a clause, while tu55 'teil' must co-occur with the verb hi55 'say' in a serialconstruction. Further, the pair du55nu13 'hear, understand' - \u55 'tell' issomewhat special. The voiced syllable du55 is a noun meaning 'word'. Incombination with the verb nu13 'notice' it means 'understand'. Thedevoicing takes place here between a noun and a verb.

11 There are also pairs that fail to fully qualify for a simplex-complex pair. They

devoice the initial consonant but do not display a causative meaning. These pairsinclude ,33 'estimate' /fact33 'calculate' and nde33 'equip' / te33 'put on'.



172 Matthias Gerner

(33) (a) du55 tho33 ^1 word ARTrPROX yesterday IP SG get

heard the word yesterday/55

i^o33.tell

'He will tell his aunt/

21 too33.hear

(b) Qi21 kr55 a3W3 du3P SG PUT aunt word

(34) (a) go21 a33ma33 du55nu13.1PSG mother hear

'I heard Mom's voice/

.21_13 hl55 QI21(b) go21 Io21pi

1PSG story say 3P SG tellCI told him a story/

55

4.2.2 A of simplex verb is in control

Two types of monotransitive simplex verbs exhibit the main argument A incontrol of the situation: stative verbs (section 4.2.2.1) and action verbs(section 4.2.2.2). For both types I counted more than six exponents.

4.2.2.1 A of simplex verb is in control of state

There are three stative causative verb pairs of the 'wear' / 'dress' type. Thepair vi13 / fi13 is the general term compatible with a wide range of vesture;the couple din I ti13 is restricted to the wearing of hats, gloves and shoes butexcludes textiles like trousers. Finally, gu551 ku55 refers to the state of beingcovered by cloth or clothing. All three simplex verbs refer to situationscontrolled by the person who is in the wearing position.

Simplexvi * 3 'wear' (for clothes)di13 'wear* (for hat, gloves, shoes)gu55 Covered' (with cloth or clothes)

Complexft l 3 'dress someone with clothes'ti13 'dress' (for hat, gloves, shoes)ku55 'cover' (with cloth or clothes)

Figure 16: Pairs of the 'wear' / 'dress' type

For a number of three-place predicational frames, the particle kjs33 canand should be employed to mark patient NPs. The order is fixed as

BV,12 as illustrated in (35a+b) for the pair vi13 / fi13. For the

12 As stated in part at the start of section 4, the labels A and are the macro-roles

inspired from Dixon (1979,1994). The abbreviation B stands for 'beneficiary'.




simplex verb gu55 'cover' in (37a) the covering entity is not marked with anyparticle or preposition.

(35) (a) Ci21 mbu33 ?a33?Y55 tha21 tgV1 vi13.3PSG garment beautiful NUM:1 CL wear

'He/she wears a beautiful garment/

(b) da13 ,2133 4?i55 a33ma33 mbu33 kV3 pi21 ft13.every morning all mother garment PAT 3 P SG dress'Every morning mother dresses him/

(36) (a) Qi21 mo55tsi33 di13.3P SG hat wear

'He wears a hat/

(b) th'55Je21 mo55tsi33 kje33 Qi21 ti13.friend hat PAT 3P SG dress

'The friend put a hat on him/

(37) (a)

174 Matthias Gerner

As in the previous section, the particle kje33 is generally employed tomark the patient with occasional exceptions such as in (39b). If the particle isnot involved in the sentence, the order in Neasu is fixed as ABOV. The use ofkjs33 achieves a topical foregrounding of the patient NP.

(38) (a) rjo21 d?a33 dzu33 go13. (b) Qi21


provided below represent near-complete numbers, I cannot exclude that forsome of the remaining Yi languages the actual number of causative pairscould be slightly higher if one pushed hard for finding more exponents.

c

Js"3

j

B abialLabio-dentalAlveolar

RetroflexAlvelo-palatal

Velar

G

3 2 or 2 > 3), whether the main argument exerts control over thesituation and finally the Aktionsart of the situation referred to by the simplexverb (state, process or action).

Valency Extension Control of S/A Situation Type Number of Pairsl ->2l->2l->22-^32->32-3

[- control][- control][+ control][- control][+ control][+ control]

StateProcessActionStateStateAction

7243233JL 7 J [ T V^UIILIUIJ -OX.UU11 J

Figure 19: Statistics of attested causative pairs across the semantic spectre inWeining Neasu

It is striking that intransitive verbs (34/42) have been the target of thelexicalization process much more often than monotransitive verbs (8/42).However, the more sensible parameter for the lexicalization process is [control]. In Weining Neasu, 33 simplex verbs refer to uncontrolled situationsand 9 to controlled situations. Especially uncontrolled intransitive processes(24/42) have been the target of the proposed lexicalization process. This basicstatistical distribution reflects a cross-linguistic tendency of (at least certainforms of) causatives not to apply to verbs in an iterative fashion (cf. Dixon2000: 59). As control verbs already express some kind of manipulation orcausation, it is not surprising that they had been less targeted by thedevoicing process. On the other hand, non-control verbs encode a lowdegree of causation and the devoicing process more likely seeks out suchverbs as it reacts less to their inherent semantics. I claim therefore that thesemantic nature of the simplex verb has also been a decisive factor for theselective process of devoicing these verbs have undergone.

5.2 Steps in the lexicalization process

It is thus proposed that the devoicing-causation correlation came historicallyinto existence in the Lolo-Burmese branch of TB languages through amotivated process of lexicalization. The proposed steps in this lexicalizationprocess are as follows:




Step 1 (morphological valency-increasing prefix *s- in PTB): Asnumerous authors remarked (see section 1), there seems to have been thetransitive prefix *s- at the stage of PTB (cf. Benedict 1972: 105-108; Dai 2001;Matisoff 1973: 32-34,1975,1976: 414-419). It appears that the language withwritten documentation in which the PTB-causative prefix *s- is preservedmost faithfully is the Central TB language Dulong/Rawang (La Polla 2000).The causative prefix is described in this language as a highly grammaticalbound morpheme without selectional restrictions. All intransitive andmonotransitive verbs may attach this valency-increasing productive prefix,while only marginal verb categories such as applicative verbs, ditransitiveverbs and verbalized nominals disallow its preposing (La Polla 2000: 296).

Step 2 (mixed segmental and supra-segmental valency-increasingstrategies in Proto-Burmese-Lolo (PEL)): The branch of South-EasternTibeto-Burman languages mainly consists of the Burmese-Lolo languagegroup (Bradley 1997: 38). In this branch of the TB family, large soundchanges seem to have occurred. In various BL languages, the originalcausative prefix *s- has undergone sound changes when attached to certainintransitive and monotransitive verbs. Matisoff (1973: 32) mentionsglottalization as a transition point between the segmental PTB-prefix *s- andsubsequent supra-segmental phenomena such as described in section 3 and 4of this paper. While no synchronic data with the glottal causative prefix *?-are available, there might be at least one Burmese-Lolo language which haspreserved this transitional hybrid picture: the Burmish language Zaiwaspoken in China near the Myanmar border (see section 1). Dai (1981)reports that in Zaiwa supra-segmental processes such as aspiration and tonealteration contribute to the production of causative verbs, whilst theexistence of a causative prefix fa- is also attested. Both causativizingstrategies are compatible with different segments of the lexicon. Theimportant feature of these supra-segmental alterations is that they exhibitselectional restrictions on the verbs that might undergo them. These supra-segmental alterations seem to have lost the productivity of the PTB prefix *s-which is typical for lexicalization. The split between verbs that derivecausative forms with a morphological prefix and those which form causativeversions through supra-segmental alterations has been arguably generated bymotivated lexicalization. At the onset, (intransitive and monotransitive)verbs with a voiced initial consonant were targeted. The selection of simplex-verbs with voiced initials has not been arbitrary, even though not everyvoiced-initial verb has been selected. In the following chart there are mono-



178 Matthias Gerner

syllabic verbs from Weining Neasu, which have not been sought out by theprocess of legalization: intransitive verbs with a voiceless initial andmono/ditransitive verbs with a voiced initial.

Voiceless initial-low transitivityt$hu55 'become bad' (intransitive)


which treated this brand of lexicalization as a blind process of linguisticchange. Furthermore, previous papers were ambiguous on the type oflinguistic change these verb derivations refer to, i.e. whether it isgrammaticalization or lexicalization. In this paper I identify the changes aslexicalization.

Appendix: Spatial-genetic information on the Yi group

In the linguistic classification literature, the Yi languages are included in thelarger Burmese-Lolo language group (inside the Tibeto-Burman languages).Some Yi languages, especially in the Chuxiong and Dali prefectures ofYunnan province, approach linguistically varieties of Lisu, which is anothernationality in China. In China there are 7.5 million ethnic Yi (2000 census)distributed over Yunnan, Sichuan, Guizhou and Guangxi provinces. Ifunintelligibility were a criterion to separate languages, then I would estimatethe total number of Yi languages to be 100-150. By unintelligibility I meansubstantial lack of comprehension that cannot be overcome by two monthsof language learning. The 11 Yi languages surveyed geographically representall branches of the Yi group. The lexical data presented in this paperoriginate from primary field research (conducted during 1996-2000), withthe exception of Weishan Lalo where I rely on a simplex-complex list inBjorvenuTs grammar (Bjrverud 1998: 66).



180 Matthias Gerner

Language

Liangshan Nuosu

@ Weining Neasu

Longlin Ngopho

Luoping Nase

Shizong Kopho

Me Axi

Me Azhee Gejiu Nesu

Weishan Lalo

Yongren Lolo Wuding Aluphu

Location

China/Sichuan/LiangshanChina/Yunnan/Lij iang/NinglangChina/Guizhou/Bijie/Weining, Hezhang,Nayong, ZhijinChina/Guizhou/Liupanshui/ShuichengChina/Guangxi/Baise/Longlin;China/G uizhou/Liupanshui/PanChina/Yunnan/Qujing/LuopingChina/Yunnan/Qujing/Luoping, Fuyuan,Shizong, Luliang, QujingChina/Guizhou/Liupanshui/PanChina/Yunnan/Qujing/Shizong, Luliang;China/Yunnan/Honghe/Luxi, MeChina/Yunnan/Kunming/ShilinChina/Yunnan/Honghe/Me, LuxiChina/Yunnan/Wenshan/QiubeiChina/Yunnan/Honghe/MeChina/Yunnan/Honghe/Gejiu, Jinping,Jianshui, LchunChina/Yunnan/Dali/Weishan,NanjianChina/Yunnan/Chuxiong/Yongren, DayaoChina/Yunnan/Chuxiong/Wuding, Luquan

Number ofSpeakers2.2 Million

300,000-600,000

10,000-40,000

100,000-300,000

20,000-40,000

ca. 82,000

ca. 60,000ca. 370,000

565,000

50,000-150,000ca. 41,000

References

Fu ([1950] 1997)Chen &Wu (1998)

Liotard(1909,1911)

Bjrverud(1998)

Figure 21: An overview of the Yi languages surveyed

The following map locates the Yi languages surveyed in this paper on amap of Southwest China.




Laos

Distribution of the Yi languages investigatedLlangshan Nuosu H Shizong Kopho $; Weishan LaloWeining Neasu ^=3. Mile Axl -.... Yongren LoloLonglln Ngopho = Mile Azhee ^JV Wuding Aluphu

Luoping Nase 44 Gejiu Nesu

Map 1: Geographical distribution of the Yi languages investigated



182 Matthias Gerner

List of abbreviations

IP PL1PSG2PSG3PSGADVARTARTiPROXBENCLcovCOV:point toDURDEMDEM:DISTDEM:MEDDEM:PROXDPFIXIMPLOGLOC:onNEGNUMNUM:9PASSPATPREPSEND:awaySTP

References

First person pluralFirst person singularThird person singularThird person singularAdversative modality particleArticleArticle: proximal distance to SpeakerBeneficiaryClassifierCoverbCoverb:with glossDurative AspectDemonstrativeDemonstrative: distal to SpeakerDemonstrative: medial to SpeakerDemonstrative: proximal to SpeakerDynamic Perfect particleFIX versatileImperativeLocation particleLocation particle with glossNegationNumeralNumeral with its valuePassive postpositionPatientPreparative phase particleSEND versatile with English glossStative Perfect

Anttila, Raimo (1989[1972]). Historical and Comparative Linguistics (Current Issuesin Linguistic Theory 6). Amsterdam: Benjamins.

Bauer, Laurie (1978). On lexicalization (neither a lexicalist nor a transformationalistbe). Archivum Linguisticum 9, 3-14.

Benedict, P.K. (1972). Sino-Tibetan: A conspectus (James Matisoff, contributingeditor). New York: Cambridge University Press.

Bjrverud, Susanna (1998). A grammar ofLalo. Ph.D. dissertation. Lund University.




Bradley, David (1997). Tibeto-Burman languages and classification. In Papers inSoutheast Asian linguistics No. 14: Tibeto-Burman languages of the Himalayas.Pacific Linguistics A-86, David Bradley (ed.) 1-72. Canberra: Australian NationalUniversity.

Brinton, Laurel and Elisabeth Gloss Traugott (2005). Lexicalization and LanguageChange. Cambridge: Cambridge University Press.

Chen Kang (1990). Yiyu zidongci yu shidongci de xingtai biaozhi jiqi youlai [Onmorphological marking of simplex and causative verbs in the Yi language].Minzu Yuwen (2), 40-45.

Chen, Kang and Wu Da (1998). Yiyu yufa [Yi grammar]. Beijing: Zhongyang MinzuDaxue Chubanshe [Central University of Nationalities Press].

Chen Shilin, Bian Shiming, Li Xiuqing and Luohong Waku (1962). Liangshan Yiyude shidong fanchou [The category of causative in Liangshan Yi]. ZhongguoYuwen (8-9), 412-426.

Comrie, Bernard (1975). Causatives and universal grammar. Transaction of thePhilological Society (1974), 1-32.

Comrie, Bernard (1976). The syntax of causative constructions: Cross-languagesimilarities and divergencies, In Syntax and Semantics Volume 6: The grammar ofcausative constructions, Masayoshi Shibatani (ed.), 261-312. New York:Academic Press.

Dai Qingxia (1981). Zaiwayu shidong fanchou de xingtai bianhua [On themorphology of the causative category in the Zaiwa language]. Minzu Yuwen (4),36-41.

Dai Qingxia (2001). Zangmian yuzu yuyan shidong fanchou de lishi yanbian [Thehistorical development of the causative category in Tibeto-Burmese languages].Journal of Chinese Linguistics 29(1), 1-10.

Dixon, Robert M.W. (1979). Ergativity. Language 55(1), 59-138.Dixon, Robert M.W. (1994). Ergativity. Cambridge studies in linguistics 69.

Cambridge: Cambridge University Press.Dixon, Robert M.W. (2000). A typology of causatives: Form, syntax and meaning. In

Changing Valency: Case studies in transitivity, R.M.W. Dixon and AlexandraAikhenvald (eds.), 30-83. Cambridge: Cambridge University Press.

Dixon, Robert M.W. and Alexandra Aikhenvald (2000). Introduction. In ChangingValency: Case studies in transitivity, R.M.W. Dixon and Alexandra Aikhenvald(eds.), 1-29. Cambridge: Cambridge University Press.

Du Ruoming (1990). Zangmianyu dongci shidong fanchou de lishi yanbian [On theevolution of the causative category of Tibeto-Burman languages]. Yuyan Yanjiu18(1), 133-148.

Fodor, J. A. (1970). Three reasons for not deriving 'kill' from 'cause to die'. LinguisticInquiry 1,429-438.

Frawley, W. (1992). Linguistic semantics. Hillsday, N.J.: Lawrence Erlbaum.



184 Matthias Gerner

Fu, Maoji (1997 [1950]). A descriptive grammar of Lolo. (Ph. D. Thesis August 1950,Cambridge University.) Linguistics of the Tibeto-Burman Area, 20(1), 1-227.

Hopper, Paul (1990). Where do words come from? In Studies in Typology andDiachrony: Papers Presented to Jospeh H. Greenberg on his 75th Birthday(Typological Studies in Language 20), William Croft, Keith Denning andSuzanne Kemmer (eds.), 151-160. Amsterdam: Benjamins.

JackendofF, Ray (1997). The Architecture of the Language Faculty (Linguistic InquiryMonograph 28). Cambridge, MA: The MIT Press.

Jackendoff, Ray (2002). Foundation of Language: Brain, Meaning, Grammar,Evolution. Oxford: Oxford University Press.

La Polla, Randy (2000). Valency-changing derivations in Dulong/Rawang. InChanging Valency: Case studies in transitivity, R.M.W. Dixon and AlexandraAikhenvald (eds.), 282-311. Cambridge: Cambridge University Press.

Lietard, A. (1909). Notions de grammaire Lolo (dialecte A-hi). Bulletin de VEcoleFrancaise d'Extreme Orient 9(2), 285-314.

Lietard, A. (1911). Notions de grammaire Lolo, dialecte A-hi. ToungPao 13,1-42.Matisoff, James (1969). Verb concatenation in Lahu: The syntax and semantics of

'simple' juxtaposition. Acta Linguistica Hafniensia (International Journal ofStructural Linguistics K0benhavn) 12(1), 69-120.

Matisoff, James (1973). The grammar of Lahu. Publications in Linguistics 75.Berkeley: University of California.

Matisoff, James (1975). A new Lahu simplex/causative pair: "study/train". Linguisticsof the Tibeto-Burman Area 2(1), 151-154.

Matisoff, James (1976). Lahu causative constructions: Case hierarchies and themorphology/syntax cycle in a Tibeto-Burman perspective, In Syntax andSemantics Volume 6: The grammar of causative constructions, MasayoshiShibatani (ed.), 413-442. New York: Academic Press.

Matisoff, James (1994). Tone, intonation, and sound symbolism in Lahu: loading thesyllable canon, In Sound symbolism, Leanne Hinton, Johanna Nichols and John J.Ohala (eds.), 115-129. Cambridge: Cambridge University Press.

McCawley, James (1968). Lexical insertion in a transformational grammar withoutdeep structure. In Papers from the 4th Regional Meeting, Bill J. Darden, Charles-James N. Bailey and Alice Davison (eds.), 71-80. Chicago: Chicago LinguisticSociety.

Palmer, F. (1994). Grammatical roles and relations. Cambridge University Press,Cambridge.

Shibatani, Masayoshi (1976). The grammar of causative constructions: A conspectus,In Syntax and Semantics Volume 6: The grammar of causative constructions,Masayoshi Shibatani (ed.), 1-40. New York: Academic Press.

Song, Jae Jung (1989). A universal typology of causative constructions: Functionaland diachronic perspectives. Ph.D. dissertation. Melbourne: Monash University.




Song, Jae Jung (1990). On the rise of causative affixes: a universal-typologicalperspective. Lingua 82,151-200.

Song, Jae Jung (1996). Causatives and causation. London: Longman.Sun Hongkai (1999). On the category of causative verbs in Tibeto-Burman

languages. Linguistics of the Tibeto-Burman Area 22(1), 183-199.van der Auwera, Johan (2002). More thoughts on degrammaticalization. In New

Reflections on Grammaticalization - Proceedings from the InternationalSymposium on Grammaticalization 17-19 June 1999, Potsdam Germany(Typological Studies in Language 49), Ilse Wischer and Gabriele Diewald (eds.),19-29. Amsterdam: Benjamins.

Xu Xijian (1984). Jingpoyu de shidong fanchou [The causative category of theJingphaw language]. Minzu Yuwen (1), 40-45.

Author's addressMatthias GernerDepartment of Chinese, Translation & LinguisticsCity University of Hong Kong83 Tat Chee AvenueKowloon, Hong Konge-mail: [email protected]



Folia Linguistica Historica Volume 28 Issue 1 2007 [Doi 10.1515_flih.2007.145] Gerner, Matthias -- The Lexicalization of Causative Verbs in the Yi Group

Documents

array of yi languages

lexicalization process

aretibetoburman languages

panyi data

theburmeselolo languages

simplex member

complex member

thedevoicing process