Ikalanga, a virtually unstudied Bantu language spoken in Botswana has optionally overt preverbal subjects and obligatory subject markers (SAs). This paper presents an analysis of Ikalanga syntax that has important consequences for Bantu morphosyntactic variation and for the formal characterization of null subject languages. It shows detailed evidence that a) Ikalanga preverbal subjects are not in the IP domain; b) Ikalanga is a null subject language (NSL) and SAs represent standard subject verb agreement, contrary to what has been proposed for different Bantu languages.
Considering (a), syntactic and semantic evidence indicates that overt preverbal subjects in Ikalanga are base generated in an A’ position (see Alexiadou and Anagnostopoulou 1998 (A&A) re- Greek and Spanish). One piece of evidence comes from the position of the subject of the subordinate clause introduced by the Co if. Thus, in Ikalanga, the subject of a subordinate clause, Nchidzi, precedes the complementizer a in (1), something impossible in English (2). Quantifier interpretive tests also provide evidence that preverbal subjects in Ikalanga occupy an A' position (3). The only reading of (3a) is one where the quantified subject NP ngwana nngwe (some child) has wide scope over the universally quantified NP thathobo ingwe ne ingwe (every examination) suggesting that the subject NP ngwana nngwe (some child) does not reconstruct, precisely because it did not undergo from a VP internal position. (3b) provides evidence that the wide scope reading in (3a) is not due to lexical semantic properties of the quantifier nngwe since this quantifier can have a narrow reading if in an A position such as that occupied by the object NP koloi ingwe (some car). Indefinite sentences (4) also provide evidence that subjects are in an A' position. In (4) only the specific reading is available suggesting that the NP ngwana (child) did not move out of VP, the domain of existentiality (see Diesing 1992). (4b), on the other hand, in which the subject NP is VP internal, has an existential reading. Scope of negation facts also indicate that subjects are not and cannot be interpreted in an A position at any point in the derivation. Example (5a) for instance, means "the children did not read the book". It cannot mean "there are no children who did not read the book" (5b). I argue that this is because there is no reconstruction of the subject bana (children) into an A position under the scope of negation. WH constructions such as Ikalanga (6) provide further evidence that preverbal subjects occupy an A’ position. In (6) the wh-agreement marker, which I argue is a Co, intervenes between the subject NP Neo and the subject agreement marker a, suggesting that the subject is not in an A position such as spec AgrS or spec TP. A similar phenomenon is observed in Luganda relative clause constructions (7) (Ashton, Mulira, Ndawula and Tucker 1954) where the subject NP abasajja (men) precedes the relative marker bye, suggesting that this analysis may extend to other Bantu languages.
Turning to (b) above, I argue first that Ikalanga is a null subject language (NSL) as supported by (8) and then propose that a null NP, pro, is merged in spec VP and later moves to a spec AgrS or spec TP. This argument is supported by the analysis above and comports with the non-Bantu analysis proposed by A&A (1998) and others that subjects in NSLs occupy an A’ position (see also Bresnan and Mchombo (B&M) 1987)). I argue that SAs are not incorporated pronouns but simply subject verb agreement morphology, (contra B&M (1987), Keach 1995 and Omar 1990) who argue that SAs in Chichewa and Swahili are sometimes incorporated pronouns. Agreement between the subject and the verb is obligatory in Bantu finite clauses (Givόn 1976) as B&M (1987) themselves point out for Chichewa. Thus, if the SA were an incorporated pronoun, we would expect two SAs, one used as an incorporated pronoun, the other simply used as an SA (9), but this is unattested in Ikalanga and possibly other Bantu languages. Secondly, if the SA was an incorporated pronoun, we might expect other lexical DPs to occur in its position, but this is not possible (10). In addition, the SA changes its form depending on the tense of the clause (11) strongly supporting the view that it actually corresponds to standard agreement morphology. I thus conclude that the SA in Ikalanga is an agreement morpheme. It agrees in phi features with a pro (either in spec AgrS or spec TP) which is coreferent with a preverbal overt subject NP in an A’ position in case an overt preverbal subject occurs. The latter is consistent with the argument that overt preverbal subjects in NSLs may actually be in a clitic left dislocated (CLLD) position (see A&A 1998, Cinque 1990). By using islands diagnostics, I also evaluate an alternative analysis in which preverbal subjects are the result of topicalization, in which case pro would not occur. The proposal made here has the advantage of providing a unified account of the properties of subjects and SAs in Ikalanga. This account, which arguably extends to other Bantu languages, is also relevant for the evaluation of analyses of argument licensing across languages.
(1) Nekuti Nchidzi a e zha, Ndiko u noobe e enda
Because Nchidzi if SA come Ndiko SA future SA go
Because if Nchidzi comes, Ndiko will leave.
(2) *Because John if he comes, Mary will leave. (* if no parentheticals)
(3) a. Ngwana nngwe we ikwele wa ka kwala thathobo ingwe ne ingwe.
Child some of school SA past write exam every
Some student wrote every examination.
b. Nlume nngwe ne nngwe u no kgweetsa koloi ingwe a e enda ku nshingo.
1.Man every SA pres drives car some when SA goto work
Every man drives a car when he goes to work. (1,1a,2…= noun classes)
(4) a. Ngwana wa ka bala buka inoyi Denjebuya.
1.Child SA past read book called Denjebuya
The child read a book called Denjebuya.
b. Kuna ngwana kuzhe.
Agr-exist child outside
There is a child outside.
(5) a. Bana a ba zo bala buka.
2.child NEG 2.SA past read book
The children did not read the book.
(6) Ndiani Neo wa a ka bona? (Foc = focus marker)
Foc-who 1.aNeo Foc-agr SA past see
Who did Neo see?
(7) Ebikajjo abasajja bye batema si byaffe.
The sugar canes men rel SA-cut not ours (rel. = relative)
The sugar canes which the men are cutting down are not ours. (Ashton, Mulira, Ndawula and Tucker (1954)
(8) pro u no bika nyama.
pro SA pres cook meat
pro S/he cooks meat.
(9) *U-u no bika nyama.
Pronoun-SA pres cook meat
She cooks meat.
(10) *Neo no bika nyama.
1a.Neo pres cook meat
Neo cooks meat.
(11) a. Neo u no bika nyama.
1a.Neo SA pres cook meat
Neo cooks meat.
b. Neo wa ka bika nyama.
1a.Neo SA past cook meat
Neo cooked meat.
Alexiadou, Artemis and Anagnastopoulou, Elena. 1998. Parameterizing Agr: Word Order, V-Movement and EPP Checking. Natural Language and Linguistic Theory 16: 491-539.
Ashton, E.O., Mulira E. M. K., Ndawula, E. G. M. and Tucker, A. N. 1954. A Luganda Grammar. London: Longmans, Green and Co.
Bresnan, Joan and Mchombo, Sam. 1987. Topic, Pronoun and Agreement in Chichewa. Language 63: 741-782.
Diesing, Molly. 1992. Indefinites. Cambridge, Mass: MIT Press.
Cinque,Gugliemo. 1990. Types of Dependencies. Cambridge, Mass: The MIT Press.
1. Rose Letsholo
2. Subjects and Agreement in Ikalanga
4. University of Michigan
6. Department of Linguistics
1076 Frieze Building, 105 South State ST.
Ann Arbor, MI 48109-1285