Default Pronouns and Root Infinitives in Malagasy Acquisition

Dimitrios Ntelitheos UCLA Cecile Manorohanta Universit Nord, Antsiranana

In this paper we report results of a longitudinal study of the acquisition of pronouns in Malagasy. We show that there is a Root Infinitive (RI) stage in Malagasy, a language typologically distinct from other RI languages. Our assumption is supported by the fact that children use optionally the default 1st person pronominal form izaho in the prominent position of the sentence, an environment where adults use the marked form aho. This use of izaho co-occurs with verbal forms that lack prefixal tense and voice, while aho occurs mainly with fully inflected verbs. This empirical fact supports approaches that associate the use of default pronouns to the underspecification of some functional projection in the verbal domain and provides prima facia evidence that Malagasy has an RI stage. This is further corroborated by the distribution of null arguments in the language.

1. Introduction In this paper we report results of a longitudinal study of the acquisition of Malagasy, an Austronesian language spoken in Madagascar. We discuss issues associated with the acquisition of Malagasy pronominal forms including the acquisition of case morphology and its relation to the grammatical role that pronouns serve in the language, as well as differences in the use of default pronominal forms in child and adult language and environments that license these differences. In particular, we show that children overuse the strong 1st person singular form izaho, conforming to the crosslinguistic tendency of children to use default pronominal forms in early stages. This default form appears with bare verbal forms providing support to proposals that associate default case overgeneralization to the under-specification of verbal functional projections (Root Infinitives). The claim that child Malagasy has a Root Infinitive (RI) stage is further supported by the fact that children drop the sentence-prominent argument (trigger) more often with bare verb forms than with fully inflected ones, conforming to the pattern observed in other betterstudied RI languages (see Hyams et al 2004 and references therein). The paper is organized as follows: In section 2 we provide a brief overview of Malagasy voice morphology and the pronominal system, with special attention to the distribution of default pronominal forms. In Section 3 we present our predictions and the relevant data. We show that the nominative/default forms of pronouns are acquired earlier and appear more frequently than the genitive/bound forms, in contrast to the corresponding relative frequencies attested in adult language. Furthermore, when children make mistakes in the production of 1st person singular pronouns, these mistakes consist of substitutions of the default strong form izaho in place of both the

weak form aho and the genitive form ko and not the other way around. Finally, we show that these substitutions co-occur with bare verbal forms, supporting the claim that the latter are the equivalent of RIs in child Malagasy. Section 4 provides an additional argument for an RI stage from the domain of subject drop. Specifically, we show that null subjects occur most frequently with bare verbs while overt subjects occur most often with inflected verbs. This pattern conforms to the pattern of subject omission in other well studied RI languages. Finally, Section 5 presents our concluding remarks. 2. Overview of Malagasy Clause Structure and Pronominal System 2.1 Voice Malagasy is spoken by approximately 16 million people on the island of Madagascar, off the coast of East Africa. It is genetically a member of the Western Malayo-Polynesian branch of the Austronesian family. It is related closely to the languages of the Southeast Barito subgroup of southern Borneo and its closest relative is Ma'anyan of south Borneo (Kalimantan, Indonesia). It is also closely related to the Philippine languages such as Tagalog in that it is a verb-initial language (VOS) and has an elaborate voicing system. The voicing system has a distinctive morphology and it involves the promotion of an argument (actor, theme, instrument, etc.) to a referentially and syntactically prominent position, typically clause final position. Following traditional theoryneutral terminology we refer to this prominent DP as the trigger. The voice morphology on the verb identifies the grammatical function of the trigger, whether actor, theme, instrument, location, etc. Consider the following examples1: (1) a. Nametraka ny boky teo amboninny latabatra ny vehivavy PST.AT.put DET book Pst.there on.top DET table DET woman The woman put the books on the table. Hapetran ny vehivavy eo amboninny latabatra ny boky FUT.put.TTDET woman there on.top DET table DET book The books, the woman will put (them) on the table. Nametrahan ny vehivavy ny boky ny latabatra PST.CT.put.CTDET woman DET book DET The table, the woman put the books (on it). table

b.

c.

1 The following conventions in abbreviating labels in the examples will be used: DET, determiner; DEM, demonstrative; 1, 2, 3, person; SG, singular; PL, plural; AT, agent topic focus or actor trigger; TT, theme topic focus or theme trigger; CT, circumstantial topic focus or oblique trigger; ROOT, verbal forms with no overt voice/tense morphology; NOM, nominative; GEN, genitive; ACC, accusative; LOC, oblique/prepositional case usually manifested as prefix an-; PRS, present; PST, past; FUT, future; FOC, focus particle; TOP, topic particle dia.

In (1a) the agent argument of the verb is promoted as the external argument (underlined in the example) and the verb shows Actor Trigger (AT) morphology (boldface on the verb). In (1b) the theme argument occupies the rightmost prominent position and the verb exhibits Theme-Trigger (TT) morphology. Finally, in (1c) the Benefactor is promoted and the verb has Circumstantial Trigger (CT) morphology. For a detailed account of the properties of these voices (or focus structures, as they are known in traditional literature) see Keenan & Polinsky (1998); Pearson (2001); Paul (2000). 2.2 The Pronominal System Malagasy has a very impoverished system of inflection in the nominal paradigm. However, the pronominal system of the language exhibits morphological alterations that depend on the grammatical function that pronouns serve. There are three major classes of pronouns corresponding to external or topic noun phrases (c.f. 2), internal objects (c.f. 3), and internal agent phrases (c.f. 4)2. These forms are traditionally termed Nominative, Accusative, and Genitive respectively (c.f. Keenan & Polinsky, 1998; Paul, 1996): (2) Nahita ny ankizy izy PST.AT.see DET children 3.NOM He/She saw the children Nahita azy ny ankizy PST.AT.see 3.ACC DET children The children saw him/her Hita-ko ny ankizy see.TT.ROOT-1SG.GEN DET children The children, I saw (them)

(3)

(4)

Table 1 lists the three types of the Malagasy Pronominal Paradigm: Table 1: Malagasy Pronominal Paradigm Person Nominative Accusative Genitive SG. 1 aho, izaho ahy -ko / -o 2 ianao anao -nao / -ao 3 izy azy -ny / -y PL. 1 (inclusive) isika antsika -ntsika / -tsika 1 (exclusive) izahay anay -nay / -ay 2 ianareo anareo -nareo / -areo 3 izy (ireo) azy (ireo) -ny / izy ireo
2 Genitive case also marks possessors and objects of most prepositions.

2.3 Distribution of Default Pronouns Paul (1996), Zribi-Hertz & Mbolatianavalona (1999), and Pearson (2001) show that the nominative/free forms of Malagasy pronouns substitute for the genitive/bound forms in a number of environments, including pronominal augmentation used for pronoun-modification (5b) and co-ordination (6b): (5) a. Hita -ny see.TT.ROOT -3SG.GEN He/she saw Hita -n izy roa . see.TT.ROOT -GEN 3.NOM two The two (of them) saw Hita -ny t -any an -tokotany i Koto see.TT.ROOT -3.GEN PST-there LOC-garden DET Koto She/He saw Koto in the garden Hitan [izy sy ny zaza] t -any an -tokotany i Koto see.GEN3.NOM and DET child PST-there LOC-garden DET Koto She/He and the children saw Koto in the garden

b.

(6)

a.

b.

For the purposes of the present study we will assume a theory of default case as implemented in Schtze (2001:206), where it is assumed that default case forms are the forms that are used to spell out nominal expressions that are not associated with any syntactically determined case feature. Pronominal modification and co-ordination involve a blocking of case spreading from the head to the other elements in the constituent, resulting in the modified or coordinated pronouns emerging with default case (c.f. Schtze 2001:226). While the assumption that nominative forms of pronouns are defaults works satisfactorily for all other forms, it breaks down for 1st person singular pronouns. The 1st person singular in Malagasy allows for three different nonaccusative forms to surface: the nominative strong form izaho and weak form aho and the genitive (bound) form ko. In the cases in which the 1st singular is the trigger, it appears as the weak form aho (8a). However, in non-caselicensing positions, as in focalization, the strong form izaho appears (cf (8b)): (8) a. Nahita ny alika aho PST.AT.see DET dog 1SG.NOM I saw the dog Izaho no nahita ny alika 1SG.STR.NOM FOC PST.AT.see DET dog It was me who saw the dog.

b.

3. Theoretical Implications, Predictions and Data 3.1 Theoretical Implications Acquisition data from mainly European languages has shown that children tend to use default-case forms of pronouns quite early and in positions in which adults use other forms. For example, children acquiring English as a first language use accusative pronouns even in subject positions of clauses where adults use nominative forms (examples from Radford, 1990:175-176): (9) a. b. c. Me got bean. Me talk. Me look. Her go back in.

Similarly in French, children acquire the free forms of pronouns earlier than the clitic forms (Clarke, 1985:699). In fact children seem to use the free forms even in places where adults would use the clitic forms obligatorily, i.e. again as subjects of clauses. Importantly, French children use non-finite verb forms in these contexts in contrast to the finite forms used by adults, (examples from De Cat, 2002): (10) a. Je vais mettre a comme Pol. I will put that like Pol I will put it like Pol. Moi mettre a comme Pol. me put that like Pol It was me who saw the dog. (Adult)

b.

(Max, 2;3)

Similar patterns have been found in other languages (see for example Babyonyshev 1993 for Russian; Schtze 1995 for German; Powers 1995 for Dutch; Jonas 2002 for Faroese). As is obvious from the English and French examples, children tend to use these default forms with non-finite verbs that appear in root clauses, traditionally termed root infinitives (Rizzi, 1994; Wexler, 1994; Hyams, 1996; Schtze, 1997; Hoekstra and Hyams, 1998, among others). 3.2 Predictions Based on the patterns attested in the Indo-European languages mentioned in section 3.1, and the distribution of default forms in Malagasy illustrated in section 2.3, we predict the following: (1) Nominative forms of pronouns in Malagasy should be acquired before genitive forms. In particular, the strong form izaho should be acquired before the weak form aho and the bound form ko.

(2)

(3)

(4)

Child Malagasy is expected to have a greater number of izaho than aho and ko forms in comparison to the language of adults, and in general, nominative forms are expected to appear more frequently than genitive forms as compared to adult language. If children make mistakes in the production of 1st person singular pronouns, these mistakes should consist of substitutions of the strong form izaho in the place of both the weak form aho and the genitive form ko and not the other way around. If these substitutions take place in an environment that also includes reduced/bare verbal forms, then these forms can be argued to be the Malagasy equivalent of RIs.

3.2 The Data 3.2.1 Subjects and Data Collection The subjects of this study are 3 Malagasy-speaking children, Tsiorisoa, Sonnia, and Ninie. The children are from families that speak the Merina dialect spoken in and around the capital city, Antananarivo. Merina is also the basis for the standard written Malagasy and has been the focus of much of the linguistic research on Malagasy. The children are from middle class families and some of their parents are affiliated with the university. All of the children were taped 35 times monthly; Tsorisioa and Sonnia were taped over a 9-month period (from April to December 2000) and Ninie, over a 6-month period (from April to September 2000)3. Table 2 shows the number of files in the data, number of utterances, and childrens ages and corresponding MLUs for the first and last file for each child. Table 2. Age, MLU and number of utterances
Age TSIORISOA MLU Utter Age SONNIA MLU Utter Age NINIE MLU Utter

2;0 2;1 2;2 2;3 2;4 2;5 2;7 2;8 Total

1.68

4.5

24 200 31 35 41 58 85 38 512

1;6 1;7 1;8 1;9 1;10 1;11 2;0 2;1 2;2

2.84

3.46

61 122 27 50 81 90 31 29 107 598

1;10 1;11 2;0 2;1 2;3 2;4 2;5 2;6

3.09

4.09

88 156 42 14 33 33 68 74 508

3 Some of the sessions were rather short and thus we collapsed all recordings within a single month into a single file according to age. For example, Tsiorisoa was taped 4 times in April 2000. These 4 files are included in Tsiorisoa 2;0.

The children used 257 tokens of different types of pronominal forms. From these, 192 tokens (75%) are different 1st person singular forms. Because of the high frequency of 1st singular pronoun production and the idiosyncratic properties of the distribution of the 1st singular pronoun in Malagasy, we will restrict our discussion to the acquisition of this form, leaving the discussion of other forms to further research pending more data from later stages of acquisition. 3.2.2 Prediction 1 In almost all cases the free forms emerge earlier in the data than the bound forms. Bound forms functioning as Agent arguments in TT constructions (for example ko in (4)) appear later than possessive bound forms, which in turn appear later than free forms. Table 3 presents the order of appearance4 of the different forms. The emergence of the bound genitive form has been divided into two columns, one representing its function as a possessor (c.f. 4b) and the second as the Agent argument in TT constructions (c.f. 4c): Table 3: Comparison of Chronological Appearance of Bound versus Free Pronominal Forms
Person Free Form Form Age Weak Form Form Age Bound Form Form Poss. Agent

1SG 2SG 3SG 1PL (incl.) 1PL (excl.) 2PL 3PL

izaho ianao izy isika izahay ianareo izy (ireo)

1;8 1;11 1;9 1;10 2;4 2;6 ---

aho

1;10

-ko -nao -ny -tsika -nay -nareo -ny

1;8 1;11 1;11 1;11 2;5 2;7 ---

1;11 1;11 -----------

As we can see in the above table, in most cases the free form seems to emerge at least a month earlier than the bound form. This is the case for the 3rd singular, 1st plural exclusive and inclusive, and 2nd plural. The only apparent exceptions are the 1st person strong form izaho, which seems to appear at the same time as the bound form ko (1;8), and the 2nd person ianao, which appears at the same time as the bound form nao. In the first person the bound form ko emerges at 1;8 as a possessive. Its first appearance as an agent argument is three months later (1;11) and thus it confirms the prediction at least partially. In the case of the second person, the appearance of the bound form as an agent and a possessive is simultaneous (1;11). However, the number of tokens of these forms is very small (9 tokens for both types in total, compared to 159 tokens for the equivalent forms in 1st person singular) and so it is not clear
4 The order is calculated taking into account the first occurrence of the pronominal form. In all cases the first occurrence of a type is followed by sporadic use of the form in the immediately following and most subsequent files and thus is assumed to have been acquired by the children.

what to make of this finding. In any case, the data for second person singular is not representative, since in all other persons free forms appear before bound forms. 3.2.3 Prediction 2 The second prediction is that child language will have a greater number of izaho than aho and ko forms in comparison to the language of adults. More generally, nominative forms are expected to appear more frequently than genitive forms, as compared to adult language. The prediction is not easy to confirm. Child-directed adult language in the available recordings consists mainly of yes/no and wh-questions to the children with no occurrences of 1st person singular pronouns. Thus, the only numbers related to frequencies of case-marked DPs in Malagasy come from Keenan (1995); Keenan & Manorohanta (2004). In these studies a text count based on two newspaper articles and selections from three novels in Malagasy found that pronominal case is distributed as in (11): (11) Nominative/Free 33.6% Accusative 23% Genitive/Bound 43.4%

We see that in adult language the bound forms appear significantly more frequently than the free forms. A count of the pronominal forms in the child data, though, shows a different pattern (excluding proforms, indeterminate cases): (12) Nominative 130 (57.5%) Accusative 4 (1.7%) Genitive 92 (40.7%)

Thus, nominative/free forms are significantly more frequent than genitive/bound forms, in contrast to the adult data5. As far as the relative frequencies of izaho and aho are concerned, the second prediction seems to be confirmed. A word count of text from Malagasy romance novels6 reveals that in a sample of 23,241 words there are 124 free 1st person singular pronominal forms. (13) illustrates the relative frequencies of izaho versus aho in both adult and child counts: (13) Adult Count Child Count aho 120 (96.7%) 57 (61.3%) izaho 4 (3.3%) 36 (38.7%)

5 Keenan & Manorohanta (2004) provide a number of further reasons why nominative forms appear more often in child language. The first is that child speech consists mainly of short utterances, most of them headed by intransitive verbs with only one argument realized as the topic argument in the free pronominal form. The second reason is that prepositional elements like ami(na) which are quite common in adult speech and which take genitive complements are completely absent from child speech. To these two reasons we add the fact that the bound form ko is substituted by the strong form izaho in some cases. It is further substituted by izaho and aho in three cases when it functions as a possessor. 6 The texts used in the count are the same as in Keenan & Manorohanta (2001).

As we can see izaho appears a lot more frequently in the child data. This is an expected distribution if izaho is the default form. 3.2.4 Prediction 3 As far as the third prediction is concerned, the data again seem to confirm the hypothesis. The prediction is that any mistakes that the children make are expected to include substitutions of the bound form ko and the weak form aho by the strong form izaho. We found a limited number of substitutions in the data. These are illustrated in Table 47: Table 4: Correct vs. Incorrect Use of 1st SG Pronouns in Child Speech Function Environ. Correct Incorrect Topic DP (aho) Agent DP (-ko) Possessor (-ko) 76 35 33 55 (72%) 33 (94%) 30 (91%) 21 (28%) 2 (6%) 3 (9%) (aho)

As we can see from the table, the children make few mistakes in their production of pronouns. Most of the mistakes are with the weak form aho as the external argument/topic of active structures. As we have seen, these positions constitute environments in which aho appears obligatorily in adult language. Children in fact do use aho in these environments in most cases, as Table 4 indicates. Therefore, izaho-substitutions (14b) co-occur with correct usage of aho (14a): (14) a. Nitomany aho PST.AV.cry 1SG.NOM I cried Tomany za cry 1SG.NOM.STR I cried (Tsiorisoa 2;7)

b.

(Tsiorisoa 2;7)

3.2.5 Prediction 4 As is often the case in childrens grammars, verbal forms in the Malagasy acquisition data can appear reduced or bare with missing tense/voice morphology. To these we can add a number of verbal forms that are never affixed with voice or tense morphology in adult grammar and always appear as verbal roots that are inherently active or passive. We adopt as a working

7 The first row shows substitutions of the marked trigger form aho by the default form izaho. The second row exhibits substitutions of the bound form ko by izaho in predicate-internal positions of the agent. Finally, row three exhibits substitutions of ko by aho in when the former functions as a possessor in noun phrases.

hypothesis that these bare and (some of the) root forms are the equivalent of RIs in Indo-European languages8. If this is on the right track, the prediction is that izaho will emerge as a default case mainly with bare verbs in the childrens utterances, while aho will appear predominately with fully inflected forms. This seems to be true: Table 5: Distribution of izaho and aho with bare and inflected verbs Child Pronoun Inflected Verbs Bare Forms Tsiorosoa aho 10 1 1 izaho 0 1 Sonnia aho 1 8 izaho 5 29 13 Ninie aho 0 7 izaho Total aho 40 (73%) 15 (27%) izaho 5 (24%) 16 (76%)
2(1) = 15, p 0.0001

Furthermore, izaho appears in environments in which adults obligatorily use the bound form ko: (14) a. Lani -ko io finish.TT.ROOT -1SG.GEN DEM This was finished by me Io any any za DEM finish.TT.ROOT .RED 1SG.STR This was finished by me (Adult Form)

b.

(Sonnia 1;9)

If we add these cases to the percentage of izaho/aho substitutions we get an overall 88% percentage of izaho-substitutions with bare verbs. This is close to percentages of default-case substitutions in non-finite contexts observed in English, French, and German (c.f. Schtze, 1997). 4. Further Support for an RI Stage There are many syntactic properties that distinguish RIs from their finite counterparts, but one that is particularly relevant to the present discussion is that RIs typically occur with null subjects. Hyams et al (2004) show that in Malagasy bare verbs occur most frequently with null triggers while inflected verbs occur most often with overt triggers: Table 6: Rate of subject omission with finite and bare verbs
8 Similar patterns for example have been found in the acquisition of Swahili (Deen, 2003), Inuktitut (Swift & Allen, 2002), and Siswati (Kunene, 1979) (c.f. also the English examples in 9.a.-9.d.). Deen (2003) argues that the bare verb is an RI analogue in child Swahili. In English, as well, bare verbs (9.a.-9.d.) show properties of non-finite forms (Harris and Wexler 1996).

Bare Verb Finite Verb

Null Subject 251 (60%) 279 (46%)

Overt Subject 165 (40%) 325 (54%)

The relation of bare verbs to null triggers is marginally significant (p=.08) by a Friedman chi square9. This is predicted if we assume that the Malagasy bare verb is an RI analogue and a fortiori that Malagasy has an RI stage. No other analysis seems at present to be able to account for the empirical facts in a satisfactory way. 5. Conclusion We examined the acquisition of the pronominal system of Malagasy based on a longitudinal study of production data from three Malagasy children. We showed that the children overuse the strong 1st person singular form izaho. Furthermore, we showed that these substitutions occur predominately with bare verbal forms. This fact provides prima facia evidence that Malagasy has an RI stage. Finally, we provided further support for an RI stage in Malagasy from trigger omission patterns. We showed that bare verbs occur most often with null triggers, while inflected verbs occur most often with overt triggers. References
Babyonyshev, M. (1993). Acquisition of the Russian case system. In C. Phillips (ed.), Papers on Case & Agreement II, MIT Working Papers in Linguistics 19, Cambridge, Mass.: MIT Press, pp. 143. Clark, Eve V. (1985). The acquisition of Romance with special reference to French. In D.I. Slobin (ed.), The Crosslinguistic Study of Language Acquisition, Vol 1: The Data, Hillsdale, NJ: Lawrence Erlbaum Associates, pp. 687782. De Cat, Ccile (2002). Apparent non-nominative subjects in L1 French. In J. Paradis and P. Prvost (eds.), The Acquisition of French in Different Contexts: Focus on Functional Categories, Amsterdam: Benjamins, pp. 5388. Deen, Kamil Ud (2003). Underspecified verb forms and subject omission in Nairobi Swahili. In Barbara Beachley, Amanda Brown and Frabces Conlin (eds.), The Proceedings to the 27th Annual BUCLD, Somerville, MA: Cascadilla Press, pp. 220231. Hoekstra, Teun and Nina Hyams (1998). Aspects of root infinitives. Lingua 106: 81112. Hyams, Nina. (1996). The underspecification of functional categories in early grammar. In Harald Clahsen (ed.) Generative Perspective on Language Acquisition, Amsterdam, Philadelphia: Benjamins, pp. 91127. Hyams, Nina, Dimitris Ntelitheos and Cecile Manorohanta (2004). The acquisition of the Malagasy voice system. Ms., UCLA. Jonas, Dianne (2002). Case and clause structure in Faroese child language. Paper presented at the 19th Scandinavian Conference of Linguistics, 1012 of January 2002, Universitetet i Troms, Norway.

9 C.f. Hyams et al (2004) for discussion of factors that increase the strength of this association.

Keenan, Edward L. (1995). Predicate-argument structures in Malagasy. In C. Burgess, K. Dziwirek, and D. Gerdts (eds.), Grammatical Relations, Stanford, CA: CSLI, pp. 171217. Keenan, Edward L. and Maria Polinsky (1998). Malagasy (Austronesian). In A. Spencer and A. Zwicky (eds.), The Handbook of Morphology, Oxford: Oxford University Press, pp. 563623. Keenan, Edward L. and Cecile Manorohanta (2001). A quantative study of voice in Malagasy. Oceanic Linguistics 40: 6785. Keenan, Edward L. and Cecile Manorohanta (2004). Malagasy clause structure and language acquisition. Paper presented at AFLA XI, April 2004, Berlin. Kunene, Euphrasia (1979). The Acquisition of SiSwati as a First Language: A Morphological Study with Special Reference to Noun Prefixes, Noun Classes and some Agreement Markers. Doctoral Dissertation, UCLA. Paul, Ileana (1996). The Malagasy genitive. In M. Pearson and I. Paul (eds.), The Structure of Malagasy, vol. 1, UCLA Occasional Papers in Linguistics 17, Department of Linguistics, UCLA, pp. 7691. Paul, Ileana (2000). Malagasy Clause Structure. Doctoral Dissertation, McGill University. Powers, Susan M. (1995). The acquisition of pronouns in Dutch and English: The case for continuity. In Dawn MacLaughlin and Susan McEwen (eds.), Proceedings of BUCLD 19, Somerville, MA: Cascadilla Press, pp. 439450. Pearson, Matthew (2001). The Clause Structure of Malagasy: A Minimalist Approach. Doctoral dissertation, UCLA Dissertations in Linguistics 21, Department of Linguistics, UCLA. Radford, Andrew (1990). Syntactic Theory and the Acquisition of English Syntax. Cambridge, MA.: Basil Blackwell. Rizzi, Luigi (1994). Some notes on linguistic theory and language development: The case of root infinitives. Language Acquisition 3: 371393. Schtze, Carson T. (1995). Childrens subject case errors: Evidence for case-related functional projections. In L. Gabriele, D. Hardison, and R. Westmoreland (eds.), Proceedings of FLSM VI, Vol. 1, Bloomington, IN: IULC Publications, pp. 155166. Schtze, Carson T. (1997). INFL in Child and Adult Language: Agreement, Case, and Licensing. Doctoral Dissertation, MIT. Schtze, Carson T. (2001). On the nature of default case. Syntax 4: 205238. Swift, Mary and Shanley Allen (2002). Contexts of verbal inflection dropping in Inuktitut child speech. In Barbora Skarabela, Sarah Fish, and Anna H.-J. Do (eds.), Proceedings of BUCLD 26, Somerville MA: Cascadilla Press, pp. 689 700. Wexler, K. (1994). Optional infinitives, head movement, and the economy of derivations in child grammar. In D. Lightfoot and N. Hornstein (eds.), Verb Movement, Cambridge: Cambridge University Press, pp. 305350. Zribi-Hertz, A. and L. Mbolatianavalona (1999). Towards a modular theory of linguistic deficiency: Evidence from Malagasy personal pronouns. Natural Language and Linguistic Theory 17: 161218.

Dimitrios Ntelitheos Department of Linguistics, UCLA, 3125 Campbell Hall, Box 951543

Los Angeles, CA 90095-1543, USA [email protected] http://www.bol.ucla.edu/~dntelith/ Cecile Manorohanta Universit Nord, Antsiranana Madagascar, B.P. 0, UNM 201 Antsiranana, Madagascar, [email protected]