Central Kenya Bantu The Syntax And Semantics Of Causative Affixes In

S.Afr.J.Afr.Lang., 2001, 4390(iv) Fourthly, the construction is used to expressthe meaning that an inanimate causerbrought about a certain change of state inthe causee.(lOa) Marua ni-ma-ken-i-a kaana(Gikuyu)Letter foc-SA-TNS-happy-CAUS-FV child'The letter has made the child happy'Downloaded by [Kenyatta University] at 00:03 08 March 2016(lOb) Valua ni-w-a-tan-i-a kana(Kikamba)Letter foc-SA-TNS-happy-CAUS-FV child'The letter has made the child happy''Mwangi had Nduku insulted by the child/had the child insult Nduku'We also noted that in Gikuyu and Kiembu the-itbi- form produces ambiguity in that theinterpretation can either be 'make somebody dosomething' or 'help somebody do something'.This happens when -itbi- is attached to dynamicverbs. Example (13) below could mean 'Mwangimade the child cook food' or 'Mwangi helped thechild cook food'.(13)(v) The fifth use is to express 'self-causation'.This is where one makes oneself do somethingor assume a certain state unnaturally. Theaction or state is simulated and not genuine,as in example (11) below:(lla) Mwangi ni-e-andik-itbi-a marua (Gikuyu)Mwangi foc-refl-TNS-wrote-CAUS-FVletter'Mwangi has made himself (pretended to)write a letter'(11 b) Mwangi ni-w-e-ken-itbi-a(Kiembu)Mwangi foc-SA-refl-be happy-CAUS-FV'Mwangi has made himself (pretended tobe) happy'(vi) Finally, the construction can be used toexpress 'indirect causation' where the causerinstigates or manipulates the causee to dosomething to somebody or to something. Inthese constructions, the causee can be leftout, as in example (12a) below. Comrie(1981:168) also notes this phenomenon ofindirect causation.(12a) Mwangi ni-a-rum-itbi-a Nduku (Gikuyu)Mwangi foe- TNS-insulted-CAUS-FVNduku'Mwangi had Nduku insulted'(12b) Mwangi ni-a-rum-itbi-a Nduku kaana(Gikuyu)M wangi foe- TNS-insulted-CA US- FVNduku childMwangi ni-a-rug-itbi-a kaana irio (Gikuyu)Mwangi foc-TNS-cook-CAUS-FV childfood'Mwangi has made/helped the child cookfood'In contrast, this ambiguity does not arise inKimeru and Kikamba since the only interpretation possible is that of 'make somebody dosomething'.Comrie (1981: 164) notes that a constructionlike (13) above could mean 'cause/help' or 'let'.We have, however, noted that the 'let' or'permissive' meaning is not possible with any ofthe four languages examined in this study.Syntactic features affecting thedistribution of the two affixesThe search for the syntactic features thatdetermine the distribution of -i- and -itbi- can beapproached from different perspectives. Therelevant perspectives have a direct bearing onthe classification of causative constructions.Firstly, we can concentrate on examining theeffects of the presence of either of the affixes onthe syntax - number of objects, type of objectsetc. (for this approach see Comrie (1981).Secondly, we can make use of the theoreticalapproaches that seek to establish whether aparticular morphological causative constructionis hi-clausal or mono-clausal (Marantz, 1984;Baker, 1988). Using this approach one caninvestigate whether the use of one of the affixesresults in a hi-clausal while the other results in

391S.Afr.J.Afr.Lang., 2001, 4Downloaded by [Kenyatta University] at 00:03 08 March 2016a mono-clausal construction. A third approachwould be to explore the distribution of the affixesin terms of the types of verbs they attach to. Wewill follow the last approach and only make briefcomments on the other two approaches.We noted that the -itbi- form is moreproductive and can occur with all verbs, thoughsometimes with restricted interpretation. Indiscussing its distribution, we have employed thedistinction made between dynamic and stativeverbs (Quirk and Greenbaum, 1973:14-21).Dynamic verbs are those that denote an actionrather than a state.With dynamic transitive verbs only the -itbiaffix can be used. This can be seen in the wellformed (14a) and (14b). The ill-formedness in(14c) results from the use of the -i- affix with adynamic verb.(14a) Mwangi ni-end-itbi-a Wacuka marigu(Gikuyu)Mwangi foc-sold-CAUS-FV Wacukabananas'Mwangi has made Wacuka sell bananas.'(14b) Kioko ni-andik-itbi-a Nduku barua(Kimeru)Kioko foc-wrote-CAUS-FV Nduku letter'Kioko has made Nduku write a letter'(14c) *Kioko ni-andik-i-aNduku barua(Kimeru)With stative transitive verbs, only the -itbi- affixcan be used as is shown by the ill-formedness of(!Sc) below.(!Sa) Mwangi ni-w-a-vit-itby-a Nduku nzia(Kikamba)Mwangi foe-SA- TNS-miss-CA US-FVNduku way'Mwangi made Nduku miss the way'(!Sb) Mwangi ni-w-a-meny-itbi-a Nduku njira(Kiembu)Mwangi foe-SA- TNS-know-CA US-FVNduku way'Mwangi has made the way known toNduku'(!Sc) *Mwangi ni-w-a-vit-i-a Nduku nzia(Kikamba)Both affixes can be used with the dynamicintransitive verbs. Consider example (16a) and(16b).(16a) Mwangi ni-w-enam-itbi-a Wacuka(Kiembu)Mwangi foc-SA-bend-CAUS-FV Wacuka'Mwangi has made Wacuka bend'(16b) Mwangi ni-w-enam-i-a Wacuka (Kiembu)Mwangi foc-SA-bend-CAUS-FV Wacuka'Mwangi has made Wacuka bend'Stative intransitive verbs can also take either ofthe affixes as in example (17a) and (17b) below.(17a) Kioko ni-w-a-no-etby-a kana (Kikamba)Kioko foc-SA-TNS-tire-CAUS-FV child'Kioko made the child tired'(17b) Kioko ni-w-a-no-sy-a kana(Kikamba)Kioko foc-SA-TNS-tire-CAUS-FV child'Kioko made the child tired'From examples (16 and 17) above, we observethat the -i- affix, which can be realised as -sy- inKikamba as mentioned on page 5, is only usedwith intransitive verbs. This suggests that thisaffix has the function of changing an intransitiveverb into a transitive one, quite like that of theaffix -sas- in Japanese discussed in Shibatani(1973). The restricted use of the -i- affix maysuggest a lexical analysis in line with theproductivity criteria given in Shibatani (1976).However, this is ruled out by further evidenceshowing that -i- can be used with some transitiveverbs as in the examples below.(18a) Kioko ni-w-a-kund-y-a Nduku supu(Kikamba)Kioko foc-SA-TNS-drink-CAUS-FVNduku soup'Kioko made Nduku drink soup'

S.Afr.J.Mr.Lang., 2001, 4392(18b) Kioko ni-w-a-rum-i-a Nduku itunda(Kiembu)Kioko foc-SA-TNS-drink-CAUS-FVNduku soup'Kioko has made Nduku bite the fruit'Further, verbs of perception (verbs expressingsense experiences) whether transitive or intransitive can take the -i- affix.Downloaded by [Kenyatta University] at 00:03 08 March 2016The syntactic distribution of the affixes in termsof the verb types can be summarised as follows: Intransitive verbs take the two affixes -i- and-ithi Verbs of perception typically take the twoaffixes -i- and -ithi Transitive verbs mainly take the -ithi- affixSemantic features affecting thedistribution of the affixesIn discussing the semantic features affecting thedistribution of the two affixes, we examine theparameter of control either on the part of thecauser or the causee. Following the approach ofGivon (1975), we take control here to meandeliberateness or intentionality.The distribution of the two causative affixescan be predicted on the basis of the nature ofcontrol. The two main parameters are: 1) who hasthe control? Is it the causer, the causee or both? 2)When the causer has control, is it coercive or noncoercive? We note that with dynamic verbs,coercive control amounts to manipulation andnon-coercive control to a directive (Shibatani,1976).Where the causer has coercive control, the affixused in the four languages is -ithi- and -i- isunacceptable.(19a) Mwangi ni-end-ithi-a Wacuka marigu(Gikuyu)Mwangi foc-sold-CAUS-FV Wacukabananas'Mwangi has made Wacuka sell bananas'(19b) *Mwangi ni-end-i-a Wacuka marigu(Gikuyu)Mwangi foc-sold-CAUS-FV Wacukabananas'Mwangi has made Wacuka sell bananas'Here, Mwangi's action is deliberate andinvolves the use of force. The assumption is thatWacuka has no choice but to do what Mwangiwants. This agrees with Givon's assertion thatcontrol entails coercion (Givon, 1975:63).Although the manipulation in (19a) suggeststhat -ithi- is a lexical affix, we note that thecausee (Wacuka) is not non-agentive as is the casewith such affixes (Shibatani (1976).The four languages also express non-coercivecontrol. This obtains when the causer doessomething deliberately but without the use offorce. In such a case an animate causee also hascontrol since sfhe can choose to act or not to(Comrie, 1981: 166). The causative affix used tocapture this situation is the -i- as in (20) below.(20)Mwangi ni-w-ekal-y-a andu nthi(Kikamba)Mwangi foc-SA-sat-CAUS-FV peopledown'Mwangi influenced people to sit down'The interpretation of (20) is that Mwangi(causer) deliberately influenced or directed andu(people) to sit down but did not use force. Thepeople (causee) have an option of obeying ordisobeying. In fact we can add the clause Namalea (and they refused) to (20) above and thesentence will be acceptable. If coercion is involvedon the part of the causer and therefore giving thecausee no choice but to sit down, the causativeaffix used will be -ithi-. The non-coercive usewhere both the causer and the causee sharecontrol contradicts Givan's (1975) principle thatonly one participant in a causation chain hascontrol. Shared control is the only interpretationwe can posit for the phenomenon observed in thefour languages.Thirdly, in the four languages it is possible toexpress a situation in which the control lies withthe causee and not the causer. In such a situation

Downloaded by [Kenyatta University] at 00:03 08 March 2016S.Afr.J.Afr.Lang., 2001, 4393the -i- affix is used. Such constructions mainlyinvolve verbs of perception, as in example (21)below.(Except with verbs of perception in K.ikamba asin (22a) where -ithy- may be used).(21)The lexical and non-lexical distinctionMwangi ni-a-thur-i-a Wacuka (Kimeru)Mwangi foc-SA-annoyed-CAUS-FVWacuka'Mwangi has annoyed Wacuka'Here we note that whatever Mwangi (causer)does, whether accidentally or deliberately, he hasno control over the reaction of Wacuka (causee).Thus the causee retains some control over herreaction. If Mwangi's (causer) actions are deliberate, then he has control over his own actions.This is a situation that seems to allow both thecauser and the causee to have control. With verbsof perception, the four languages differ in the waythey use the two causative affixes to express thismeaning. Gikuyu, K.imeru and K.iembu use -iwhether the causer acts deliberately or not, whileK.ikamba makes a distinction by using -itbi- whenthe causer acts deliberately and -i- when thecauser acts accidentally (see example (22) below).(22a) Eka kuthat-ithy-a Wacuka(K.ikamba)Stop annoying-CAUS-FV Wacuka'Stop annoying Wacuka'(2.::,) Eka kuthat-i-a Wacuka(K.ikamba)Stop annoying-CAUS-FV Wacuka'Stop annoying Wacuka'(22a) carries the meaning that the causer isacting deliberately and therefore has control overhis/her actions while in (22b) the causer may noteven be aware that his/her actions are causinganger.The above discussion has shown that thedistribution of the two causative affixes in theCentral Kenya Bantu languages can be predictedby using distinctions related to the parameter ofcontrol. When the causer has coercive control, theaffix used is -ithi-. When the causer has noncoercive control or has no control, the affix usedis -i-Now we look at the relevance of the semantic andsyntactic distribution of the two causative affixesto the distinction drawn between lexical and nonlexical causatives. The analysis above has shownthat the two affixes differ on two parameters:productivity and control. On the parameter ofproductivity, the -itbi- affix is the more productiveof the two. Thus, if we were to recognise a lexicaland a non-lexical causative, the -itbi- would besaid to be non-lexical and the -i- the lexical on thebasis of productivity. On the parameter ofcontrol, data from the four languages showsthat the -itbi- is used with manipulated/coercivecontrol while -i- is used with directives and noncoercive control. This would suggest a lexicalanalysis for the -itbi- affix and a non-lexicalanalysis for the -i- affix. Thus, this criterion givesus just the opposite of what the productivitycriterion yields. Given these mixed results, it isdifficult to call one affix lexical and the other nonlexical. To validate this point, we review each ofthe criteria given on the first 2 pages of this articleand show that while some contradict each other inthe face of our data, others are not applicable.Firstly, the criterion of productivity wouldmake the semi-productive -i- lexical while thehighly productive -ithi- would appear to be nonlexical.Secondly, though ambiguities resulting fromadverb scope may be observed in these languages,we have not found it significant in determiningthe distribution of the two affixes. Upon examining Bemba data, Givon (1976: 50) drew theconclusion that the 'adverbial interpretation donot converge sufficiently to differentiate lexicalfrom non-lexical causativization [and therefore] . a clear-cut distinction between thebehaviour of lexical and non-lexical causatives israther difficult to draw'. After analysing datafrom the four languages, we concur with Givon 'sconclusions on this aspect.

394Downloaded by [Kenyatta University] at 00:03 08 March 2016Thirdly, in all the preceding examples, thecauser nominal is realised as the grammaticalsubject and the other nominals are ranked forobjecthood according to the hierarchies operatingin Bantu languages (Kioko, 1994; 2000). None ofthe object nominals in a causative construction inthese languages seems to have logical subjectfeatures. We therefore cannot apply the reflexivization test of Shibatani (1973) to establishwhether the two affixes can be differentiated onthe basis of whether the resulting sentence ismono-clausal or hi-clausal.Fourthly, lexical causatives are said to expressmanipulative causation in which the causee is anon-agentive entity. Though this kind of causation mainly attracts the -ithi- form as opposed tothe non-lexical -i- which expresses directivecausation, our data shows that the main factordetermining the distribution of the two affixes isthe parameter of control expounded in Givon(1975) as is evident in examples 19-22.Fifthly, the four languages have basically noobject marking. Therefore, we cannot use Cooper's (1976) object marking criterion to distinguishthe two affixes. In all the examples we have givenabove, the participants increase by one regardlessof the causative affix used. That is, a basicallyone-place verb becomes two-place as in (4a) and(4b) while a two-place verb becomes a three-placeverb (see example (9).ConclusionOur data has shown that the distribution of thetwo causative affixes can be predicted on the basisof verb types and the semantic parameter ofcontrol. However, none of the syntactic criteriaproposed for distinguishing lexical from nonlexical causatives has independent motivation forthe four languages examined. We have found thatsyntactically, the constructions resulting from theaffixation of -i- are similar to those resulting fromthe affixation of -ithi-. Thus, if there were reasonsfor analysing one as non-lexical and therefore hiclausal, the same reasons would suffice to accordthe other the same hi-clausal status. The criteriaS.Afr.J.Mr.Lang., 2001, 4of productivity and that of directive versusmanipulative causation, though useful in explaining the distribution of the two affixes, haveyielded contradictory results in as far as thedistinction between lexical and non-lexical causative is concerned. While -ithi- would be classifiedas non-lexical on the basis of the productivitycriterion, it turns out to be lexical on the directive/manipulative criterion. Consequently, it is difficult to classify either causative affix as lexical ornon-lexical using the semantic criteriaFinally, the data has shown that the semanticparameter of control/deliberateness is the mainfactor determining the distribution of the twocausative affixes in the Central Kenya Bantulanguages. That is, -ithi- is used in situations thatentail control while -i- is used in REFLSATNSapplicativeCausative affixFocusVerb final vowelReciprocal affixReflexiveSubject agreementTenseReferencesBaker, M. 1988. Incorporation: A Theory ofGrammatical Function Changing. Chicago:The University of Chicago Press.Comrie, B. 1976. 'The Syntax of CausativeConstructions: Cross-Language similaritiesand divergence'. In Shibatani (ed.) Syntaxand Semantics 6. 261-311.Comrie, B. 1981. Language Universals andLinguistic Typology. Oxford: Basil Blackwell.Cooper, Robin. 1976. 'Lexical and non-lexicalCausatives in Bantu'. In Shibatani, (ed.)Syntax and Semantics 6. 313-325.Gathenji, H.W. 1981. 'The Morphology of theVerbal Extensions in Kikuyu'. UnpublishedM.A thesis, University of Nairobi.

Downloaded by [Kenyatta University] at 00:03 08 March 2016S.Afr.J.Afr.Lang., 2001, 4Givon, T. 1975. 'Cause and Control: On theSemantics of Interpersonal Manipulation'. InJ. Kimball (ed.) Syntax and Semantics 4. 5989. New York: Academic Press.Givon, T. 1976. 'Some Constraints on BantuCausativization'. In Shibatani, (ed.) Syntaxand Semantics 6. 325-351.Katamba, F. 1993. Morphology. London: Macmillan Press.Kioko A.N. 1994. Issues in the syntax ofKikamba: A Bantu Language. UnpublishedPh.D. Thesis, Monash University.Kioko, A.N. 2000. 'On Determining the PrimaryObject in Bantu.' South African Journal ofAfrican Languages. Vol. 20 No. 3. 225-233.Leaky, L. 1959. First Lessons in Kikuyu. Nairobi:The Eagle Press.Marantz, A. 1984. On the Notion of GrammaticalFunctions. Cambridge Massachusetts: MITPress.395Mwangi, P. 2001. 'Verb Morphology in Gikuyuin the light of morpho-syntactic theories'.Unpublished Ph.D. thesis, Kenyatta University.Mwangi, PandA. Kioko. 1998. 'The Syntax andSemantics of Morphological Causatives inCentral Kenya Bantu'. A Paper Presented atthe Association of African Languages ofsouthern Africa. June-July 1998.Quirk and Greenbaum. 1973. A Concise Grammarof Contemporary English. New York: HBJ.Shibatani, M. 1973. 'Semantics of JapaneseCausativization'. Foundations of Language, 9.327-373.Shibatani, M. 1976. 'The Grammar of CausativeConstructions: A Conspectus'. In Shibatani,(ed.) Syntax and Semantics 6. 1-39. London:Academic Press.

causative constructions found in languages viz. non-lexical and lexical. The non-lexical causative, . The non-lexical causative shows ambiguity when used with adverbs Downloaded by [Kenyatta University] at 00:03 08 March 2016 . 388 but the lexical causative does not have this ambiguity (Cooper, 1976:323). To illustrate,