BUCLD 36 Proceedings To be published in 2012 by Cascadilla Press Rights forms signed by all authors Clitic Production across Tasks in Young French-Speaking Children Mihaela Pirvulescu, Ana T. Pérez-Leroux, Yves Roberge and Nelleke Strik 1. Introduction1"/>* Previous research demonstrates that children acquiring French show optionality in the realization of object clitics – and that there are therefore object omissions – in contexts where clitic pronouns are obligatory in the target grammar (Müller, Crysmann & Kaiser, 1996; Jakubowicz, Müller, Riemer & Rigaut, 1997; Pérez-Leroux, Pirvulescu & Roberge, 2008; among others). Most previous studies attribute object omissions to issues related to the domain of verbal argument structure, that is, difficulties in realizing the clitic element or in representing the null element associated with the clitic. Our goal is to investigate clitic optionality in monolingual French-speaking children in order to determine whether there are conditions on clitic optionality in children’s production, and if there are, what the nature of these conditions is. We focus on the syntax of clitic optionality based on a minimalist analysis of the recoverability of silent arguments (Sigurðsson, 2011). To do so, we contrast the results in clitic elicitation obtained in three different tasks: indicative mood with third person subject, indicative mood with second person subject and imperative mood with second person subject. Our results show that clitic production varies significantly across different elicitation tasks. We propose an explanation that reinterprets as a condition on syntactic representations, the traditional view in acquisition according to which children allow argument drop because they over-rely on access to discourse (Allen, 2000). 2. Object Clitic Pronouns in French In French, clitic pronouns are normally used to replace a previously introduced element. For instance, the object clitic le in (1a) refers to the * Mihaela Pirvulescu, Ana T. Pérez-Leroux and Yves Roberge, University of Toronto, Nelleke Strik, Dalhousie University, ma.pirvulescu@utoronto.ca. We wish to thank the audiences at Linguistic Symposium on Romance Languages 41 and BUCLD 36, as well as H.A. Sigurðsson and the students from our Acquisition Project, Sophia Bello, Ailís Cournane, Mélanie Elliott, Anna Frolova and Joanne Markle LaMontagne. This research was partially supported by SSHRC grant 410-09-2026. antecedent ce livre. Following Roberge (1990) and Sportiche (1996, 1998) we assume that the clitic is coindexed with a null pronominal element pro, a null object, in the canonical object position. The English equivalent in (1b) shows that in this language a strong pronoun (it) is used in the canonical object position. (1) a. b. A: Tu veux ce livrei? you want this book B: Oh! Mais je li’ai déjà lu proi. oh but I it have already read A: Do you want this book? B: Oh! I’ve already read it. (Cummins & Roberge, 2005, p. 62) It is also possible to drop the clitic in French (Lambrecht & Lemoine, 1996) whereas in English a null object is not possible, as illustrated in (2). 2 (2) a. b. A: Tu veux ce livrei? you want this book B: Oh! Mais j’ai déjà lu proi. oh but I have already read A: Do you want this book? B: *Oh! I’ve already read Ø. (Cummins & Roberge, 2005, p. 62) Clearly, the null object in (2) has a definite anaphoric interpretation. However, null objects in French (and English) can also have an implicit interpretation, in which case we adopt Cummins and Roberge’s (2005) null cognate object analysis based on work by Hale and Keyser (2002). This null object is represented by a null Noun and an example in provided in (3). (3) Pendant mon congé sabbatique j’ai surtout l’ intention de lire N. During my leave sabbatical I have mainly the intent to read ‘During my sabbatical I mainly intend to read Ø.’ (Cummins & Roberge, 2005, p. 62) Although null objects are generally possible in adult French, certain constructions, including imperatives, favor object omission (cf. Cummins & Roberge, 2005 and references cited there). 2.1 Object Clitics in Child L1 French 2 How widespread clitic drop is in French still remains to be determined. In the classic object clitic elicitation task (4), where a short story with images precedes the prompt of the investigator (therefore introducing an antecedent), a response with a clitic pronoun is required. Children, however, have been observed to variably drop the clitic in this context. The alternation between the responses in (4a) and (4b) – produced by the same child in the same type of context – shows that clitics seem to be in free variation with null objects for some Francophone children. (4) a. b. Prompt: Dis à Kermit ce que l’ourson fait avec les framboises. tell to Kermit what the bear does with the raspberries ‘Tell Kermit what the bear is doing with the raspberries.’ Child: Il les mange. (object clitic) (M. 3;06) he them eats ‘He eats them.’ Prompt: Dis a Kermit ce que Sébastien fait avec le verre de jus. tell to Kermit what Sebastian does with the glass of juice ‘Tell Kermit what Sébastien is doing with the juice.’ Child: Il boit Ø. (omission) (M. 3;06) he drinks ‘He drinks.’ However, recent experimental studies indicate that clitic optionality might not be totally random. First, children do not overgenerate object pronouns in the absence of a discourse antecedent (Pérez-Leroux, Pirvulescu & Roberge 2008; Salomo, Lieven & Tomasello, 2008). Moreover, several studies have shown that children are sensitive to informativeness features in their patterns of argument omission, including omission of pronominals (Allen, 2000; Allen & Schröder, 2003; Serratrice, Serrace & Paoli, 2004). Research on clitic constructions in specific syntactic structures also shows that clitic production is drastically limited in topicalization structures in French (Pérez-Leroux, Pirvulescu & Roberge, 2011) or that it is boosted in island constructions in Italian (Brunetto, 2010). Finally, studying clitic production in a very controlled setting, Pirvulescu and Hill (2011) show that rates of clitic production vary drastically across elicitation contexts. Various analyses have been proposed to account for clitic omission in child language. Some proposed a deficiency specific to the clitic category (Jakubowicz, Müller, Kang, Riemer & Rigaut, 1996; Jakubowicz et al., 1997) or a problematic computation due to maturation (Wexler, Gavarró & Torrens, 2004) or to memory effects (Grüter, 2006). Schaeffer (1997, 2000) proposed a pragmatic deficit as the source of omissions. An alternative view, which does not consider omissions as the result of a deficit in either the morphosyntax of clitics or in children’s pragmatic abilities is given by Pérez-Leroux et al. (2008), which proposes a default null object in the early grammar. Most analyses share the view of the clitic construction as ‘static’, and therefore do not directly predict that clitic optionality should vary across different contexts. In contrast, the latter view predicts sensitivity to recoverability conditions. 2.2 Background and Hypotheses The goal of our study is to explore the effect that the syntactic and pragmatic contexts may have on the production of object clitics in child French. We use constructions with similar contextual linking, namely the clitic context, but which differ in some aspect of their morphosyntax. Some evidence can be found in previous studies on child language for the relevance of contextual features in early argument drop. It has been proposed that informativeness features control argument omission (Allen, 2000; Allen & Schröder 2003; Serratrice et al., 2004). Pirvulescu and Hill (2011) showed that manipulating the conversational context boosts clitic production to almost target-like responses. More specifically, these authors noticed a relation between the type of conversational setup and the rate of clitic omission: in the context of direct address (the child addresses the interlocutor, second person subject) the rate of object clitic omission is insignificant, while in the context of indirect address (the child talks about a character, third person subject) the rate of omission is considerably higher. Following Mavrogiorgios (2009), Cardinaletti and Starke (1999) and Déchaine and Wiltschko (2002), we analyze object clitics as phi-feature bundles which incorporate to their verbal host. Following Sigurðsson (2011), we link the referential silent argument (pro), associated with the incorporated clitic, to a constituent (CLn in (5)) in the edge of the clause (the local C-domain), which probes for a goal under Agree. Although this C/edge linking is decidedly syntactic in nature, it interacts with pragmatics through context scanning. (5) CP Sigurðsson (2011, p. 284) Force CLn context scanning . . . TP … [phon]/ø… C/edge linking C/edge-linkers are logophoric (speaker and hearer) features or topical features. Furthermore, we assume that referential null objects are available alongside with the clitic construction in child language (Pérez-Leroux et al., 2008) and that there is therefore a certain freedom in selecting items to Merge as direct objects in the course of the derivation: null object with referential properties, DP, pro, and overt pronouns (Castilla & Pérez-Leroux, 2010). Following Sigurðsson (2011), the distribution of any silent argument is constrained by surface factors such as lexical complementizers, agreement, etc., which can also function as interveners between the null category and the C/edge-linkers. Within such an approach, we could expect that features of the syntactic and/or pragmatic context interact with clitic omission. With this background in mind, we formulate two different hypotheses: (6) a. b. H1: Clitics are optional regardless of syntactic contexts. H2: Patterns of clitic optionality can vary across syntactic contexts. As mentioned, previous analyses of clitic optionality in L1 French predict experimental results consistent with H1. For instance, according to the Decayed Features Hypothesis (Grüter, 2006), clitic drop reveals a deficit in the clitic construction itself, and predicts uniform patterns of omission of clitics across contexts with the same Agree distance between the clitic and the pro argument. 3. Study 3.1 Participants Participants in our study were 16 three year-old and 17 four year-old monolingual French-speaking children, as well as 14 adults who served as a control group. The children were recruited from daycares in the Montreal area. Most of the adults were daycare workers from the same daycares, while a small number of them were relatives of one of the testers. The age range, mean age and standard deviation (SD) for each of the groups are presented in Table 1. Table 1. Age ranges, mean ages and SDs for participants Age Range Mean Age SD 3 year-olds (n=16) 3;03 - 4;0 3;08 0;03 4 year-olds (n=17) 4;01 - 5;02 4;06 0;05 20 - 63 33 13 Adults (n=14) 3.2 Experimental task To test the object clitic production, we used an experimental method with three separate tasks. The context provided a definite referent, which was the topic of the sentence prompt. The first task was a classic elicitation task for object clitics using the indicative mood and a third person subject (cf. Jakubowicz et al., 1996, 1997; Schaeffer, 1997 and subsequent work). In this task, the child hears short stories accompanied by pictures. The puppet makes an erroneous statement and the investigator invites the child to correct the puppet (see (7)). (7) Task 1: third person subject - picture book and puppet Prompt: Dis à Kermit ce que la fille fait avec le chariot. tell to Kermit what the girl does with the carriage ‘Tell Kermit what the girl is doing with the carriage.’ Target clitic response: La fille le pousse. the girl it pushes ‘The girl is pushing it.’ The second task still involved the indicative mood, but with a second person subject (following Pirvulescu & Hill, 2011). The investigator is presenting various toys and props to the child and is explaining what she is doing with them. She then asks a question about it, as in (8): (8) Task 2: second person subject - toy manipulation Prompt: Dis-moi, qu’est-ce que je fais avec les voitures? Tell me what I do with the cars ‘Tell me, what am I doing with the cars?’ Target clitic response: Tu les pousses. You them push ‘You are pushing them.’ In the third task, the subject was a second person, but was not overtly expressed, due to the use of the imperative mood.3 In this task, the investigator is acting out scenes with a puppet, toys, and other props. Since the puppet is made to be somewhat distracted, the investigator invites the child to give orders to the puppet, so that it will understand what to do. The puppet asks the child what to do with each toy (see (9)). (9) Task 3: Imperative- toy manipulation and puppet Prompt: Dis-moi quoi faire avec le camion. Tell me what do with the truck ‘Tell me what to do with the truck.’ Target clitic response: Pousse-le! ‘Push it!’ There were 12 items per task. In each task, half of the items contained a singular referent and the other half a plural referent. Moreover the items were equally divided between masculine and feminine referents. To summarize, the distinction between Task 1 and Task 2 reduces to the contrast between third person subject (indirect address) and second person subject (direct address), whereas the difference between Task 2 and Task 3 is essentially that between indicative and imperative mood since in both cases the subject is a second person. To our knowledge, this is the first experimental study of clitic production utilizing imperatives. 3 3.3 Results We first consider the overall results for the three tasks and the three groups. In addition to target clitic responses, three other response types were produced: DPs, null objects and ‘other’ responses (see (10)). (10) a. Clitic: b. DP: c. Null: d. Other: Elle les coupe. (M. 4;03) She them cuts ‘She’s cutting them’ Tu manges la carotte. (R. 3;11) you eat the carrot ‘You’re eating the carrot’ Bois! (T. 4;03) ‘Drink!’ Responses not using the target verb or verb form, non-responses, non-relevant responses. Responses not using the target verb form are particularly frequent in Task 3. Instead of producing an imperative, children sometimes used infinitives or the construction il faut followed by an infinitive (see (11)).4 (11) a. b. Ouvrir les bouteilles. (I. 3;0) open the bottles ‘To open the bottles.’ Il faut couper. (J. 3;03) it must cut ‘You have to cut.’ Table 2 presents mean numbers and SDs in brackets. Recall that each task contained 12 test items. Hence, the maximum number for each response type is 12. The data in this table show that clitic production fluctuates across tasks and the rate of clitics increases between 3 year-old and 4 year-old children. At the same time, null objects and DP’s decrease. The adults produced particularly high rates of DPs, but clearly fewer null objects than the children.5 A mixed design ANOVA showed a highly significant effect of both age (F(2,44)=11.75, 4 Children who produced responses such as in (11) received a second (more explicit) prompt from the investigator, inviting them to use an imperative. This prompt indeed yielded more imperatives, but in order to be consistent and not to mix elicitation techniques, these responses have not been counted under imperatives in Table 2. 5 We speculate the high number of DPs is due to a task effect. In the experimental context, the adults had a tendency to be overly correct and to literally repeat the investigator in producing DPs instead of clitics. p<.000) and task (F(2,43)=26.92, p<.000), but not a significant interaction. Thus, both age and context are relevant factors in clitic production. Table 2. Mean numbers and SDs of response types per task per group Clitic DP Null Other Task 1 3rd p. Task 2 2nd p. Task 3 Imper. 3 year-olds 1.94 (2.9) 4.75 (3.2) 3.37 (3.0) 1.94 (1.3) 4 year-olds 5.06 (3.1) 3.0 (2.7) 2.53 (2.7) 1.41 (0.87) Adults 2.71 (2.8) 9.29 (2.8) 0 0 3 year-olds 4.06 (4.1) 3.62 (3.6) 2.88 (3.6) 1.44 (1.8) 4 year-olds 8.94 (3.6) 1.23 (2.1) 1.24 (1.8) 0.59 (0.8) Adults 3.5 (4.5) 8.14 (4.6) 0.36 (0.7) 0 3 year-olds 0.19 (0.5) 3.63 (3.2) 0.74 (1.6) 7.44 (3.7) 4 year-olds 2.35 (3.3) 2.82 (2.2) 1.18 (1.6) 5.65 (4.1) Adults 0.5 (0.5) 9.43 (2.8) 0.14 (0.4) 1.93 (2.7) We now take a closer look at clitic production in children across tasks considering only relevant responses (i.e., excluding ‘other’ responses). Figure 1 shows proportions of clitics among relevant responses (i.e., clitic, DP and null) for Task 1 and Task 2. Figure 1. Proportions of clitics among relevant responses per group: Task 1 versus Task 2 A mixed-design ANOVA showed a main effect of Task (F(1,31)=27.73, p<.000) and age (F(1,31)=71.45, p<.000), and no interaction. Task 1 (with a second person subject) yields many more clitics than Task 2 (with a third person subject). This replicates previous results obtained by Pirvulescu and Hill (2011) although the difference in our task, while significant, is not as polarized as in this previous study. In Figure 2 proportions of clitics among relevant responses for Task 2 and Task 3 are compared. Figure 2. Proportions of clitics among relevant responses per group: Task 2 versus Task 3 Figure 2 shows that mood is also a relevant factor in clitic production. A mixeddesign ANOVA showed a highly significant main effect of Task (F(1,31)=45.72, p<.000) and Age (F(1,31)=76.16, p<.000), and near significant interaction (F(1,31)=3.63, p=.066). The indicative mood context with a second person subject yields many more clitics than the imperative mood context with an (implicit) second person subject. For imperatives, most responses were of the DP type, but null objects were present as well. Crucially, development did not affect the proportion of clitics/nulls in the same way as in the indicative task: 4 year-olds give comparable rates of clitic and null responses; see Figure 3 below. Figure 3. Proportions of relevant responses per group: Task 3 The differences in clitic production between Task 1, 2, and 3 confirm Hypothesis 2, namely that patterns of clitic optionality can be affected by the syntactic context. 4. Discussion and conclusions Recall our theoretical assumptions: 1) the existence of a referential null object in child grammar alongside with the clitic option, and 2) licensing of the referential null object through the left periphery of the clause. While most previous studies have assumed that object omission is due to a problem within the verb-object area, we propose that the left periphery also interferes with this phenomenon (following, within a more articulated syntactic framework, previous proposals by Hulk & Müller, 2000 and Hill & Pirvulescu, 2011). We adopt Sigurðsson’s (2011) approach according to which referential null arguments are either phi-silent (Chinese argument-drop type) or phi-overt (Romance null subjects type). All arguments have to be C/edge linked. However, in the case of phi-overt arguments this linking is made through the overt phi-features, which has consequences for the recoverability of the null argument. Specifically, a phi-overt null argument can match CLn features in the C domain across lexical categories. For the case of French, we propose that, as in the case of Romance null subjects, the clitic construction involves a phi-overt null argument, with the pro as the null argument and the clitic as the overt manifestation of the argument’s phi-features. For the case of the referential null object, we consider that this might be a case of a genuinely phi-silent null argument. If this is the case, then following Sigurðsson (2011), this type of null argument is subject to intervention effects: “Radically phi-silent arguments differ from phi-overt arguments (including Romance Ø-Tphi) in that their C/edge linking is invisible, hence uninterpretable across a spelled-out intervener in the C-system” (p. 270). We propose that the following constructions are available in the child grammar for the direct object argument: (12) a.[CP.. X ... [TP ... Ø ... ]] = object omission b.[CP..X … [TP ... Cl ... Ø... ]] = clitic construction c.[CP..X … [TP ... DP... ]] = Overt DP Construction (12a) is specific to the child grammar while constructions (12b) and (12c) exist in both the child and adult grammars. Assuming that the default construction in child grammar is a., this will be used unless something prevents it. According to our results, the context of second person subject seems to disfavor the use of the null object construction, favoring clitics instead. The second person subject can therefore be considered a potential intervener between the C/edge linker and the null object as in (13) and (14) below: (13) [CP.. X ..2Pers .. [TP ... cl-V Ø... ]] Null object linking OK Potential intervener (2nd person subject) (14) [CP..X …2Pers .. [TP ... V Ø... ]] impossible linking of the null object Intervener (2nd person subject) Therefore, according to our analysis, the second person subject blocks the C/edge linking of the null referential argument, this context resulting in a higher use of the clitic construction. If this is the case, the null object in French early grammar appears to be a true phi-silent null argument of the Chinese type: that is, it has to match the C/edge features under distant agreement, as opposed to the German type where the null object has the possibility of moving into the Cdomain. The C/edge blocking by the second person subject seems to conform to the situation observed in Sigurðsson (2011) for Swedish, where a more specific (subject) clitic intervenes between the silent logophoric C-features and the clause internal null object (i.e., the Relative Specificity Constraint). Interestingly, while our imperative task yielded few null and clitic responses, favoring DP-responses, both clitic and null responses seem available, in comparable proportions for 4 year-olds. Null objects are not blocked in this context. We speculate that this is because in the imperative mood, no syntactically (and phonologically) active second person subject features are present (cf. Mavrogiorgios, 2009). Therefore, no intervention effects occur. To conclude, we have shown in this paper that clitic omission in child French is not uniform across tasks within the same modality, an elicited production context. Some tasks generate significantly higher rates of object clitics and have a steeper developmental curve. This shows that we cannot consider omission as a developmental deficiency in the syntactic computation of the clitic. We propose that omission is the result of several options available in the child grammar, including the null referential object and the syntactic constraints regulating its distribution. Therefore, omission is made possible within an appropriate discourse situation, but is regulated by the syntactic context. References Allen, Shanley (2000). A discourse-pragmatic explanation for argument representation in child Inuktitut. Linguistics, 38, 483–521. Allen, Shanley & Schröder, Heike (2003). Preferred argument structure in early Inuktitut spontaneous speech data. In John W. Du Bois, Lorraine E. Kumpf & William J. Ashby (Eds.), Preferred argument structure: Grammar as architecture for function (pp. 301-338). Amsterdam: John Benjamins. Brunetto, Valentina (2010). Syntactic diagnostics for referentiality marking in early null objects: evidence from Italian. In Barry Heselwood & Leendert Plug (Eds.), Leeds Working Papers in Linguistics and Phonetics 15 (pp. 1-13). Leeds: Department of Linguistics and Phonetics, University of Leeds. Cardinaletti, Anna & Starke, Michal (1999). The typology of structural deficiency: A case study of the three classes of pronouns. In Henk van Riemsdijk (Ed.), Clitics in the Languages of Europe (pp. 145-233). Berlin: Mouton de Gruyter. Castilla, Anny P. & Pérez-Leroux, Ana T. (2010). Omissions and substitutions in Spanish object clitics: Developmental optionality as a property of the representational system. Language Acquisition, 17(1/2), 2-25. Cummins, Sarah & Roberge, Yves (2005). A modular account of null objects in French. Syntax, 8, 44–64. Déchaine, Rose-Marie & Wiltschko, Martina (2002). Decomposing Pronouns. Linguistic Inquiry, 33, 409-442. Grüter, Teres (2006). Object clitics and null objects in the acquisition of French. Unpublished doctoral dissertation, McGill University. Hale, Kenneth & Keyser, Samuel J. (2002). Prolegomenon to a Theory of Argument Structure. Cambridge (MA): MIT Press. Hill, Virginia & Pirvulescu, Mihaela (to appear) French object clitics, L1 and the left peryphery, In Sandrine Ferré, Philippe Prévost, Laurie Tuller, & Rasha Zebib (Eds.), Selected Proceedings of the Romance Turn IV Workshop on the Acquisition of Romance Languages. Cambridge: Cambridge Scholars Publishing Hulk, Aafke & Muller, Natasha (2000). Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism: Language and Cognition, 3(3), 227–244. Jakubowicz, Celia, Müller, Natasha, Kang, Ok-Kyung, Riemer, Beate & Rigaut, Catherine (1996). On the acquisition of the pronominal system in French and German. In Andy Springfellow, Dalia Cahana-Amitay, Elizabeth Hughes & Andrea Zukowski (Eds.), Proceedings of the 20th annual Boston University Conference on Language Development (pp. 374–385). Somerville (MA): Cascadilla Press. Jakubowicz, Celia, Müller, Natasha, Riemer, Beate & Rigaut, Catherine (1997). The case of subject and object omissions in French and German. In Elizabeth Hughes, Mary Hughes & Annabel Greenhill (Eds.), Proceedings of the 21st annual Boston University Conference on Language Development (pp. 331–342). Somerville (MA): Cascadilla Press. Lambrecht, Knut & Lemoine, Kevin (1996). Vers une grammaire des compléments zéro en français parlé. In Jean Chuquet & Marc Frid (Eds.), Absence de marques et représentation de l’absence 1 (pp. 279-309). Rennes: Presses universitaires de Rennes. Mavrogiorgios, Marios (2009). Proclisis and Enclisis in Greek. Unpublished doctoral dissertation, University of Cambridge. Müller, Natasha, Crysmann, Berthold & Kaiser, Georg A. (1996). Interactions between the acquisition of French object drop and the development of the C-system. Language Acquisition, 5, 35–63. Pérez-Leroux, Ana T., Pirvulescu, Mihaela & Roberge, Yves (2008). Null objects in child language: Syntax and the Lexicon. Lingua, 118(3), 370-398. Pérez-Leroux, Ana T., Pirvulescu, Mihaela & Roberge, Yves (2011). Topicalization and object drop in child language. First Language, 31 (2), 1-20. Pirvulescu, Mihaela (2006). Theoretical implications of clitic omission in early French: spontaneous vs. elicited production. Catalan Journal of Linguistics, 5, 221-236. Mihaela Pirvulescu & Hill, Virginia (2011). Object Clitic Omission in French-Speaking Children: Effects of the Elicitation Task. Language Acquisition, 19:1, 73-81. Roberge, Yves (1990). The Syntactic Recoverability of Null Arguments. Montreal: McGill-Queen’s University Press. Salomo, Dorothé, Lieven, Elena, & Tomasello, Michael (2010). Young children’s sensitivity to new and given information when answering predicate-focus questions. Applied Psycholinguistics, 31, 101–115. Schaeffer, Jeannette (1997). Direct Object Scrambling and Clitic Placemen in Dutch and Italian Child Language. Unpublished doctoral dissertation, University of California, Los Angeles. Schaeffer, Jeannette (2000). The Acquisition of Direct Object Scrambling and Clitic Placement. Amsterdam: John Benjamins. Serratrice, Ludovica, Sorace, Antonella & Paoli, Sandra (2004). Crosslinguistic influence at the syntax–pragmatics interface: subjects and objects in English–Italian bilingual and monolingual acquisition. Bilingualism: Language and Cognition, 7, 183–205. Sigurðsson, Halldór A. (2011). Conditions on argument drop. Linguistic Inquiry, 42(2), 267-304. Sportiche, Dominique (1996). Clitic constructions. In Johan Rooryck & Laurie Zaring (Eds.), Phrase Structure and the Lexicon (pp. 213-276). Dordrecht: Kluwer. Sportiche, Dominique (1998). Partitions and Atoms of Clause Structure. London: Routledge. Wexler, Kenneth, Gavarró, Ana & Torrens, Vicenç (2004). Feature checking and object clitic omission in child Catalan. In Reineke Bok-bennema, Bart Hollebrandse, Brigitte Kampers-Mahne & Petra Sleeman (Eds.), Romance Languages and Linguistic Theory 2002 (pp. 253–268). Amsterdam: John Benjamins.