On subject use in English as a second language

ON SUBJECT USE IN ENGLISH AS A SECOND LANGUAGE Barbu Revencu* Abstract: The present article addresses the issue of syntactic transfer in child L2 acquisition, by presenting a two-part study in which Romanian monolinguals are compared to Romanian-Hebrew balanced bilinguals in two spontaneous production tasks. The main research question concerns the influence of Hebrew as a second L1 on the (re)setting of the Null Subject Parameter in English-L2, as Hebrew exhibits the same subject and verb morphology pattern as English for certain persons and tenses. The collected data provide evidence in favour of both access to UG and syntactic transfer, supporting the Full Access Full Transfer Hypothesis. Keywords: null subject parameter, agreement, full access full transfer hypothesis, L2 English, L1/2L1 Romanian, 2L1 Hebrew 1. Introduction Until very recently and with very few exceptions, linguists working within the generative framework regarded the process of second language acquisition as encompassing any acquisition of a non-native language, be it L2, L3 or Ln. It was only after 2004, when generativists began turning their attention to L3 acquisition as a way of gaining insight into the questions that have been the object of debate in L2 acquisition studies (Angelis 2007, Leung 2007, Rothman et al. 2010). One of the main arguments in favour of three-language acquisition studies concerns the issue of syntactic transfer, which has been at the core of the studies and analyses so far (Leung 2007: 107). Generativists studying L2 acquisition have been interested in the accessibility of Universal Grammar (UG) and in the influences of L1 on L2 (transfer). With respect to the latter, acquisition studies looking into simultaneous bilingualism might be more efficient, as previous linguistic knowledge is richer (more systems are available for transfer), and could thus shed light on the ongoing controversy in multilingual acquisition. Given the very low number of studies on bilinguals’ acquisition of English as L2 and the complete absence of Romanian in these studies, I undertook to investigate the acquisition of the null subject parameter in English by Hebrew-Romanian bilinguals. The main research question is in what way, if any, Hebrew as an L1/L2 might influence the (re)setting of the Null Subject Parameter in English-L2. The paper is organized as follows: Section 2 provides the syntactic background by describing the three languages involved in the study with respect to the null subject parameter, and the theoretical framework adopted in the present paper, Platzack’s (2004) Person Phrase Hypothesis. Section 3 presents the predictions for the study, driven by the Full Transfer Full Access hypothesis (Schwartz and Sprouse 1994, 1996). Sections 4 and 5 include the two studies and the discussion of the results with reference to the predictions made by the model proposed in Section 3. The main conclusions of the study are summarized in Section 6. * University of Bucharest, . 104 Barbu Revencu 2. The null subject parameter 2.1 In a nutshell The difference between languages which have an available null subject position (pro-drop languages) and those which do not has been addressed by generativists since the 1970s (Giorgi 2009: 1). Chomsky (1981) provided the first consistent account of the difference between the two types of languages, postulating a cluster of properties associated with the possibility of null subjects in pro-drop languages. Apart from missing subjects, he argued, post-verbal subjects, long wh-movement of subjects, empty pronouns in embedded clauses, and apparent violations of the that-trace filter are also possible in these languages. He linked this set of properties to the inflectional morphology of the languages in question, starting from the intuitive idea that Italian-like languages, which are allowed to drop their subjects, have rich inflection (strong agreement on the verb). Consequently, agreement has been proposed as the governor for null subjects. Conversely, null subjects are barred in languages like English, because the weak Agreement on the verb cannot properly govern them and the Empty Category Principle would thus be violated (Chomsky 1982 in Giorgi 2009: 3). Whether it is indeed richness of inflectional morphology that influences the parameter has been the object of a long-standing debate over the last decades. The existence of languages like Chinese, which lack morphology but allow null subjects, or like Icelandic, which have φ-features marked on the verb but do not drop their subjects, further complicates the issue. However, typological analyses point out that “the majority of languages do show a correlation between strong agreement and pro-drop” (Giorgi 2009: 21). This point will be of interest in the following sections, where the main subject properties of the three languages relevant to our study (Romanian, Hebrew, and English) are presented. 2.2 Romanian Romanian is a pro-drop VSO language, with the subject in post-verbal position, with the verb raising to Inflection (Dobrovie-Sorin 1994, Motapanyane 1995, Cornilescu 2000, a.m.o) and with rich inflectional morphology. The bundle of features associated to the parameter (Chomsky 1981) is also found in the syntax of Romanian subjects: (i) missing subjects (1) Am găsit cartea. have find-PERF book-DEF ‘I found the book.’ (ii) free inversion in simple sentences (i.e. post-verbal subjects) On subject use in English as a second language (2) 105 A mâncat Ion. has eat-PERF Ion ‘John ate.’ (iii) long wh-movement of subjects (3) Fata care cred că a plecat. girl-DEF that believe that has leave-PERF ‘The girl that I believe to have left.’ (iv) empty pronouns in embedded clauses (4) Aceasta este fata care crede că nu pleci. this is girl-DEF who think that not leave ‘This is the girl who thinks that you don’t leave.’ (v) apparent violations of the that-trace filter1 (5) Cine crezi că pleacă? who believe that leave ‘Who do you think that will leave?’ (vi) no expletive with weather verbs, impersonal and raising constructions (Pagurschi 2010:7) (6) a. b. Plouă. rain ‘It is raining.’ E adevărat că Vlad a obținut o bursă. is true that Vlad has get-PERF a scholarship ‘It is true that Vlad got a scholarship.’ Overt pre-verbal subjects are assumed to move to an A-bar position in the Cdomain (in TopicP, or FocusP, for instance), while post-verbal subjects occupy the specifier position of the VP, MoodP, or AgrP (see e.g. Cornilescu 2000). 2.3 Hebrew Hebrew, “an SVO language where the finite verb normally raises to INFL” (Vainikka and Levy 1999: 640), exhibits a mixed subject omission pattern, thus being neither a fully pro-drop nor a fully non-pro-drop language. Hebrew may drop its subjects for 1st and 2nd persons in past and future tenses, but disallows null subjects for 3 rd person 1 More recent studies claim, however, that the that-trace effect is not a property of the Null Subject Parameter (Pesetsky (...truncated)