Multilingual transfer: L1 morphosyntax in L3 English

24
International Journal of Language Studies Volume 8, Number 2, April 2014, pp. 1-24 ISSN: 2157-4898; EISSN: 2157-4901 © 2014 IJLS; Printed in the USA by Lulu Press Inc. Multilingual transfer: L1 morphosyntax in L3 English Abdelkader HERMAS, Université du Québec À Montréal, Canada This study investigates the acquisition of two properties of the null subject parameter in L3 English: subject-verb inversion in declarative sentences and null expletive subjects. The participants are L1 Arabic- advanced L2 French adults learning L3 English in a formal foreign language context. The study considers the nature of the grammar in the L3 initial stages and answers the question “Which of the L1 or the L2 constitutes the source of morphosyntactic transfer?” The predictions of two L3 transfer scenarios are tested, the Typological Primacy Model and the L1 transfer scenario. The results of an acceptability judgement task and a preference task seem to show that L1 Arabic is the source of morphosyntactic transfer in the L3 initial stages, overriding the effect of language proximity, typology and psychotypology which would have instigated the transfer of L2 French. Keywords: L3 Initial Stages; L3 English; Null Subject Properties; L1 Transfer; Psychotypology; Language Proximity 1. Introduction Third language acquisition (L3A) is an emerging field in the generative approach to adult language acquisition. By the number of studies, it has stimulated much less interest in morphosyntactic properties compared to second language acquisition (L2A) literature (De Angelis, 2007; Rothman, Iverson, & Tiffany, 2011). This study considers the acquisition of the null subject parameter in L3 English in the minimalist framework (Chomsky, 2000, 2001a, 2001b). We examine the acquisition of two properties of the null subject parameter in the initial stages of L3 English, the verb-subject inversion in declarative sentences and null-lexical expletives. The learners are L1 Moroccan Arabic-L2 French bilinguals. We focus on the potential implication of the L1 or L2 in the construction of the L3 grammar, thereby testing the predictions of two L3 transfer hypotheses: (a) the Typological Primacy Model, and (b) the L1 transfer scenario. We start by presenting these two positions on multilingual transfer. Next, we describe the syntax of the target constructions in the three languages. The learning predictions and research objective follow. In the

Transcript of Multilingual transfer: L1 morphosyntax in L3 English

International Journal of Language Studies Volume 8, Number 2, April 2014, pp. 1-24

ISSN: 2157-4898; EISSN: 2157-4901 © 2014 IJLS; Printed in the USA by Lulu Press Inc.

Multilingual transfer: L1 morphosyntax in L3 English

Abdelkader HERMAS, Université du Québec À Montréal, Canada

This study investigates the acquisition of two properties of the null subject parameter in L3 English: subject-verb inversion in declarative sentences and null expletive subjects. The participants are L1 Arabic-advanced L2 French adults learning L3 English in a formal foreign language context. The study considers the nature of the grammar in the L3 initial stages and answers the question “Which of the L1 or the L2 constitutes the source of morphosyntactic transfer?” The predictions of two L3 transfer scenarios are tested, the Typological Primacy Model and the L1 transfer scenario. The results of an acceptability judgement task and a preference task seem to show that L1 Arabic is the source of morphosyntactic transfer in the L3 initial stages, overriding the effect of language proximity, typology and psychotypology which would have instigated the transfer of L2 French.

Keywords: L3 Initial Stages; L3 English; Null Subject Properties; L1 Transfer; Psychotypology; Language Proximity

1. Introduction

Third language acquisition (L3A) is an emerging field in the generative approach to adult language acquisition. By the number of studies, it has stimulated much less interest in morphosyntactic properties compared to second language acquisition (L2A) literature (De Angelis, 2007; Rothman, Iverson, & Tiffany, 2011). This study considers the acquisition of the null subject parameter in L3 English in the minimalist framework (Chomsky, 2000, 2001a, 2001b).

We examine the acquisition of two properties of the null subject parameter in the initial stages of L3 English, the verb-subject inversion in declarative sentences and null-lexical expletives. The learners are L1 Moroccan Arabic-L2 French bilinguals. We focus on the potential implication of the L1 or L2 in the construction of the L3 grammar, thereby testing the predictions of two L3 transfer hypotheses: (a) the Typological Primacy Model, and (b) the L1 transfer scenario. We start by presenting these two positions on multilingual transfer. Next, we describe the syntax of the target constructions in the three languages. The learning predictions and research objective follow. In the

2 A. Hermas

method section, we present the experimental groups, the tasks and the procedure. Finally, the results are analyzed and discussed and the research objective revisited. The study concludes that L1 Arabic, rather than L2 French, has a joint facilitative and non-facilitative influence on the acquisition of the two target null subject properties in L3 English. The L1 transfer scenario provides a better account of the data than the Typological Primacy Model.

2. Background

2.1. Transfer models in the L3 initial stages

A topic of inquiry in the generative L3A research concerns the source of transfer in the L3 initial stages. Here, we need to distinguish between two concepts in the L3 literature : (a) the initial state, and (b) the initial stages. Though used interchangeably in many L3 studies, they are not analogous. Initial state refers to the grammar at the outset of language acquisition. Garcia Mayo and Rothman (2012) state that the L3 initial state consists of the entire L1 system, the L2 end state and UG (especially those parameterized properties not instantiated in the L1/2). On the other hand, the initial stages are those stages of L3A just beyond what the third language acquirer/learner (L3er) brings to the learning task at the starting point (Rothman, in press). We also define transfer as a process that involves primarily the underlying mental representations of a linguistic system.

This study submits to empirical testing the scenarios of two positions on the source of morphosyntactic transfer in L3A: (1) the L1 transfer position, and (2) the Typological Primacy Model (TPM).1

1 Two other transfer models in the field of L3A are the Cumulative Enhancement

Model or CEM (Flynn, Foley, & Vinnitskaya, 2004) and the L2 status factor (Bardel & Falk, 2007; Falk & Bardel, 2011). The L2 status factor claims that the last system acquired chronologically (i.e., the L2) is a privileged source of morphosyntactic transfer. The L2 blocks access to the L1 because of the cognitive and sociolinguistic similarities between adult L2A and L3A, namely age of acquisition, outcome, formal learning, metalinguistic knowledge and learning strategies (Falk & Bardel, 2011). As for the CEM, Rothman & Cabrelli Amaro (2010, pp. 213-214) consider their TPM the ‘psychotypological version of the CEM’. Like the TPM, the CEM claims that transfer comes from either the L1 or the L2, but differs in that it supports that transfer is either facilitative or does not occur. It cannot be non-facilitative. We did not test the predictions of the three models altogether here since the language combination involved (L1≠L2=L3) would not allow us to distinguish between the first two models and the TPM in case transfer comes from French. In this case, French is the L2 (the L2 status factor), its impact would be facilitative (CEM), and it is (psycho)-typologically proximate to L3 English (TPM).

The first position claims that the native language is a privileged source of transfer in the initial stages of adult

3 International Journal of Language Studies, 8(2) 1-24

L3A. Though this position is ‘logical’ and ‘certainly a possibility’, it has not been proposed yet as an initial stages model that can be called ‘the L1 factor’ (Rothman, Iverson & Tiffany, 2011) or ‘absolute L1 transfer’ (Garcia Mayo & Rothman, 2012). This position finds support in Lozano (2003) as well as Na Ranong and Leung (2009). To begin with the former, Lozano (2003) investigated the advanced L3 Spanish of L1 Greek-advanced L2 English adults. The study tested the strength feature of the head focus [Foc] in terms of two properties: overt/ null pronominal subjects, and SV/ VS word order. The distribution of pronominal subjects is constrained by a universal principle, the Overt Pronoun Constraint (OPC), and by a language-specific constraint, the Contrastive Focus Constraint (CFC). CFC is operative in Greek and Spanish but not in English (L1 ≠ L2, L1 = L3). Similarly, the distribution of SV/VS word order is constrained by two universal principles, the Unaccusative Hypothesis (UH) and the Uniformity of Theta Assignment Hypothesis (UTAH), and by a language-specific constraint, the Presentational Focus Constraint (PFC), active in Spanish but not in Greek and English (L1 = L2 ≠ L3).

Following an acceptability judgment test, the author claimed that the L3ers successfully acquired the universal principles (OPC, UH and UTAH). However, the acquisition of the parametric discursive constraints diverged from the native norm in case they were not previously activated in the L1. Specifically, the performance of the L1 Greek-L2 English adult learners of L3 Spanish converged with that of the Spanish controls on pronoun distribution (L1 ≠ L2, L1 = L3) but diverged from it on SV/VS word order (L1 = L2 ≠ L3) because the PFC regulating this property in L3 Spanish is not active in the L3ers’ native Greek. Lozano (2003, pp. 245-246) concluded that the L1 had an irreversible negative effect on parametric properties in the L3 end state if these were not previously operative in the L1 while the L2 was of no use even when it patterned with the L3.

Note that Lozano (2003) addressed the L3 end state but generalized his findings to the whole L3 process. Thus, what applies to the L3 near-natives applies to the L3 beginners. While the author excludes L2 transfer in the L3 end state, prior L1 or L2 transfer in the L3 initial stages needs to be excluded on empirical grounds. Herein, we empirically fill in this gap by investigating the initial stages of L3 English. However, while Lozano (2003) tested properties that belong to the syntax-discourse interface, the structures tested here are part of formal syntax.

The second study that found out L1 transfer is Na Ranong and Leung (2009). The study investigated the acquisition of null objects among L1 Thai natives who were L2 high-intermediate/advanced speakers of English and L3 beginners. While English rules out null objects, Thai and Mandarin allow both

4 A. Hermas

overt and null objects (Thai = Mandarin ≠ English). A written interpretation task was used to test the L1 Thai-L2 English learners of L3 Mandarin and a control group of L1 English-L2 Mandarin learners. The results showed that both groups performed similarly, with the L1 English group unexpectedly accepting null objects. However, the L1 Thai group was significantly more sensitive to the subtle distinction between null objects and overt objects than the L1 English group. The authors interpreted the data to indicate that the L1 was a privileged source of (non)-facilitative transfer in both the L2 and the L3 acquisition (ibid., p. 185).

Nevertheless, we remark that the factors of typology and L1 are confounded in this study because Thai is both typologically closer to Mandarin than English and it is also the L3ers’ native language. This means that the results of the study square with the predictions of the competing model TPM. Finally, Na Ranong and Leung (2009) warned against considering their finding as final due to different methodological shortcomings, the small scale nature of the study and the language combination involved. In this study, the language pairing involved (L1 Arabic ≠ L2 French = L3 English) avoids the above language overlap and therefore helps distinguish between the two transfer models.

The second position about transfer in L3 initial stages rejects the claim of a privileged transfer status for either the L1 or the L2. Specifically, abstract morphosyntactic knowledge can be transferred from either languages and the choice of one over the other is determined by their structural and psychotypological proximity with the L3. This is formalized by the Typological Primacy Model or TPM (e.g., Rothman & Cabrelli Amaro, 2010; Rothman, 2011, in press). For instance, Rothman and Cabrelli Amaro (2010) tested some properties of the null subject parameter for two groups of L1 English learners of L3 Italian and French who were advanced in L2 Spanish. The experiment consisted of a grammaticality judgment/correction task (null/overt expletive subjects and null/overt referential subjects) and a context/sentence matching task (Overt Pronoun Constraint). The results showed that both L3 groups transferred L2 Spanish grammar regardless of its (non)-facilitative influence on the L3. In particular, the L3ers of French did not transfer L1 English though this would have had a bootstrapping effect on the L3 development. The study concluded that (psycho)-typology was a deterministic variable and that the L1 was not a privileged source of transfer in L3A.

The TPM finds support in other studies investigating the acquisition of other structures using different language triplets such as Rothman (2010, 2011) and Montrul, Dias, and Santos (2011). In this respect, Rothman (2010) examined relative clause attachment preference in addition to SV/VS word

5 International Journal of Language Studies, 8(2) 1-24

order restrictions with unergative, transitive and unaccusative verbs in declaratives and interrogatives in the later part of the beginning stages of L3 Brazilian Portuguese. Two groups of learners were contrasted: L1 English-L2 Spanish advanced and L1 Spanish-L2 English advanced. Although Spanish and Brazilian Portuguese are typologically close, Brazilian Portuguese patterns much more like English than Spanish in the tested constructions. The results of a grammaticality judgment task with correction and a choice context-matching task indicated that both groups transferred syntactic knowledge from Spanish irrespective of whether it was the L1 or the L2 although English would have had a more facilitative effect.

From a methodological perspective, the L3ers in Rothman (2010) are not true beginners; they may be pre-intermediates. Their level in the L3 was not objectively determined by a proficiency test as their L2 English or Spanish was. The author made it clear that the L3ers were late beginners attending an intensive program in an immersion setting and he cautioned that the study may not tap the initial state given the significant amount of exposure to the L3 both in and out of the classroom, equal to or greater than a year of college level instruction.

Another study that supports the TPM is Rothman (2011) who considered adjective placement and its semantic interpretation. The experimental groups were L3ers of Spanish (L1 Italian/L2 English) and L3ers of Brazilian Portuguese (L1 English/L2 Spanish); both groups were at the low to intermediate proficiency level of L3. In these language combinations, L1 Italian and L2 Spanish are (psycho)-typologically similar to the L3. English lacks adjectival semantic nuances related to noun-raising while the other languages share them. The experiment included a semantic interpretation task and a context-based collocation task. Capitalizing on differences between the L3ers’ performance and that of L1 English learners at the same level of proficiency in L2 Spanish, Rothman (2011) maintained that L1 Italian was transferred into L3 Spanish, while L2 Spanish was transferred into L3 Brazilian Portuguese.

One criticism of this study is that the low to intermediate proficiency level of L3ers is a stage of grammatical restructuring beyond the initial stage of relative stability. Given their experience in language acquisition, we suspect that the L3ers may have already gone beyond a preliminary stage of deciding which, between the L1 and the L2, was the most helpful to them. Specifically, for the L1 English-L2 Spanish-L3 Brazilian Portuguese group, nothing excludes the possibility that they tried their L1 initially. Since English lacks the relevant semantic contrast between pre- and post-nominal adjective position, the L3ers subsequently tried L2 Spanish.

6 A. Hermas

Comparable results are supported in Montrul, Dias and Santos (2011). They investigated clitics and object expression in self-estimated elementary and intermediate L3 Brazilian Portuguese. Two experimental groups of L1 English-L2 Spanish and L1 Spanish-L2 English adults completed an oral production task and a written acceptability judgment task. While English lacks object clitics, Spanish and Brazilian Portuguese have object clitic pronouns but differ in their use. The results generally revealed transfer effects from Spanish (as L1 and L2) in the L3 Brazilian Portuguese of the two experimental groups. Montrul et al. (2011) concluded that (psycho)-typology was deterministic for the source language of multilingual transfer. As a comment, we have reservations about the use of self-rating for proficiency measure, which is less reliable but justified by the lack of a standardized placement test. We also think the English native L3ers group resorted to transfer from L2 Spanish because their L1 lacks object clitics; there was nothing to transfer in the first place.

In summary, we presented two positions about morphosyntactic transfer in L3A. The L1 scenario supports the privileged status of the native language as a transfer source in the early stages of the L3. The second, formalized as the TPM, allows for either the L1 or the L2 to influence the L3 depending on the (psycho)-typological proximity between the three languages. The present study tests the predictions of the two positions in a rarely investigated language combination where Moroccan Arabic is the L1, French is the L2 and English is the L3.

Meanwhile, we need to describe the relationship among these languages and define the factors at play. We adopt Falk and Bardel’s (2010) distinction between the factors of language proximity, typology and psychotypology. Language proximity is based on genetic relatedness: French and English, two Indo-European languages, are proximate to each other compared to Moroccan Arabic, a Semitic language. Although English is a Germanic language, it presents several influences from Romance at the lexical level. By typology, we mean the similarity between specific structures. We see below that French and English, unlike Moroccan Arabic, are similar when it comes to the tested structures of the null subject parameter. Psychotypology is the learner’s perception of the proximity, whether real or imagined, between languages (Kellerman, 1983). According to the learners’ post-experiment interviews, French is similar to English because both use the Latin alphabet and share many lexical items and even grammatical properties like verb tenses. We test the weight of these three factors in a novel language triplet that is biased against the L1 transfer—L1 Arabic ≠ L2 French = L3 English. The learning scenario should advantage transfer from L2 French into L3 English as the TPM predicts. This would also have a facilitative effect on L3

7 International Journal of Language Studies, 8(2) 1-24

English. On the other hand, none of the three factors predicts transfer from L1 Moroccan Arabic.

2.2. Syntactic considerations

To investigate the acquisition of the null subject parameter, we describe two structures in the languages involved: verb-subject inversion in declaratives, and expletive subjects.

2.2.1. English

English rules out null subjects (Radford, 2004, p. 165). A minimalist account stresses the crucial role of the EPP feature. Consider this:

(1) John watches movies.

The DP John is first merged in the specifier of vP (SpecvP). As the head T is finite, it carries EPP and requires the presence of an item in the specifier position. The operation Move raises John to SpecTP to eliminate EPP. Here, we adopt Bobaljik’s (2002) proposal that the verb root watch contracts its inflection -es in PF. Here is a simplified representation:

(2) [TP Johni [T’ [T [EPP] -es [vP ti watch movies]]]]

In addition to Move, English eliminates EPP in raising constructions by the operation Merge of a lexical expletive in SpecTP (Radford, 2004, p. 243):

(3) It seems that John likes cartoons.

In the embedded clause, the verb likes is finite. The DP John moves to the lower SpecTP to eliminate EPP of the finite head T. Because the DP cannot eliminate EPP of the matrix T, observing the Phase Impenetrability Condition, the expletive it is introduced by Merge directly in SpecTP of the main clause to eliminate EPP of the higher T (Radford, 2004, p. 267). This computation also applies to sentences with raising adjectives such as likely or probable:

(4) [TP It [T [EPP] -s [vP seem [CP that [TP Johni [T [EPP] -s [vP ti like cartoons]]]]]

2.2.2. French

Some proposals like those of Roberge (1986) and Auger (1994) maintain that French is a null subject language and interpret lexical subject pronouns as verbal agreement. However, Auger (1994, p. 1) and Roberge and Vinet (1989, p. 45) limit this analysis to spoken French and do not claim it applies to the

8 A. Hermas

standard variety that the Moroccan participants here have been learning as L2. Additionally, De Cat (2007) rejects this analysis of spoken French and supports the standard argumental status of pronouns. We adopt the standard account where French requires a lexical subject following Arteaga and Herschensohn (2004), Belletti (1999), Laenzlinger (1998), and Rowlett (2007). The resulting language combination is then L1 ≠ L2 = L3.

In French, subjects need phonetic realization. Like English, only the sequence SVO is ruled in. For example:

(5) a. Jean regarde la télé. Jean watches the TV ‘Jean watches TV.’

The head T is specified for EPP. The DP Jean, merged in SpecvP, raises by Move into SpecTP to eliminate EPP:

(6) [TP Jeani [T [EPP] regardej [vP ti [tj la télé]]]]

Like English, raising predicates resort either to Merge of the expletive il ‘it’ or to Move of the embedded DP in order to eliminate the EPP of the matrix TP. Consider the first possibility:

(7) Il semble que Jean regarde la télé. it seems that Jean watches the TV ‘It seems that Jean watches TV.’

In (7), the verb regarde is finite, which requires the DP Jean to move into the lower SpecTP in order to eliminate the EPP of the embedded T. Observing the Phase Impenetrability Condition, the embedded DP cannot raise further. French resorts to Merge of the expletive il to eliminate EPP of the higher TP. This account is also valid for raising adjectives such as possible, difficile, etc.

2.2.3. Moroccan Arabic

According to Fassi Fehri (1993) and Aoun, Benmamoun, and Sportiche (1994), Arabic is a null subject language where the subject position can be filled with a null pro or a lexical subject. We describe the target structures starting with the Arabic equivalent of raising constructions with null expletives:

(8) χaşş Ali y-әmši. must Ali 3MS-leave ‘It is necessary that Ali leaves.’

9 International Journal of Language Studies, 8(2) 1-24

Benmamoun (2000, p. 41) maintains that the matrix clause in such examples has an expletive pro merged in SpecTP to eliminate the EPP. The same account applies to structures with raising adjectives. A representation of (8) would be:

(9) [TP pro [T [EPP] χaşşi [vP ti [TP Ali y-әmši]]]]

The lexical realization of pro in this sentence is optional in matrix (10) as well as embedded positions (11):

(10) (Huwa) χaşş Ali y-әmši (it) must Ali 3MS-leave ‘It is necessary that Ali leaves.’

(11) Smәʕ-t bәlli (huwa) χaşş Ali y-әmši. heard-1S that (il) must Ali 3MS-leave ‘I heard that it is necessary that Ali leaves.’

The second target structure is the verb-subject inversion in declaratives, a sequence banned in spoken French even under a null-subject analysis. Here is an example:

(12) Kәtb-u l-ulaad bra. wrote-3MP the-kids letter ‘The kids wrote a letter.’

The majority of accounts agree on the presence of an expletive pro in VSO, for example Aoun, Benmamoun and Sportiche (1994), Benmamoun (2000), Fassi Fehri (1993) and Ouhalla (1994). Benmamoun (1995, pp. 181-183) stipulates that an expletive pro is merged in SpecTP to eliminate the EPP whereas the DP argument remains in situ in SpecvP:

(13) [TP pro [T[EPP] kәtbi-u [vP l-ulaad [ti bra]]]]]

Finally, we consider the alternative licit SVO sequence:

(14) L-ulaad kәtb-u bra. the-kids wrote-3MP letter ‘The kids wrote a letter.’

The subject DP syntactically raises from SpecvP to SpecTP to eliminate the EPP. The verb also moves up in T to check off the verbal feature on T and contract its inflection:

10 A. Hermas

(15) [TP L-ulaadi [T [EPP] kәtbj-u [vP ti [tj bra]]]]]

In short, in VSO and the Arabic equivalent of raising constructions, the operation Merge of an expletive pro in SpecTP eliminates the EPP. In SVO, the operation Move of the subject DP from SpecvP to SpecTP does the checking. It follows that French, even if categorized as a null subject language, is different from Moroccan Arabic (MA) in two respects. First, it rules out VS in declaratives. Second, in constructions with expletives, it resorts to the verb morphology, instead of pro, to eliminate the EPP. The verb morphology has an argumental status in French, interpreted as a null subject language, but an agreement status in MA. This detail is important as it helps determine the source language of transfer.

3. L3 Predictions and research objective

From the syntactic description above, the language scenario we obtain as regards the properties of the null subject parameter is L1 MA ≠ L2 French = L3 English. MA shares with the other two languages the use of lexical expletives in raising constructions and the sequence SV. On the other hand, only MA sanctions null expletives and VS word order in declarative sentences.

The success of the L3 English beginners in acquiring the target structures would depend on the source language of transfer. If transfer comes from L1 MA, their performance will be defective since L1 ≠ L3. Specifically, for verb -subject inversion, the L3ers should accept both SV and VS sequences. The degree of acceptance may vary because the sequence VS is stylistically marked in the L1. Regarding expletive constructions, the L3 beginners should accept both null and lexical expletives since their phonetic realization is optional in the L1.

On the other hand, given the similarity between French and English, the L3ers are expected to accept the grammatical sentences and reject their ungrammatical counterparts if they transfer their L2 knowledge. Thus, VS and null expletives constructions should be ruled out in the L3 interlanguage. Taking account of the combined potential effects of the three factors at play—language proximity, typology and psychotypology—we support this second scenario.

The objective of this study is to consider the source of morphosyntactic transfer in the initial stages of L3 English. Thus, it seeks to determine which of the L1 scenario or the TPM provides a viable account of the acquisition of the two formal properties of the null subject parameter.

11 International Journal of Language Studies, 8(2) 1-24

4. Method

4.1. Participants

Three control groups of 65 native speakers and an experimental group of 14 L1-Arabic/L2-French adults learning L3 English in a formal setting took part in the experiment. The learners started formal learning of L2 French at the age of eight and L3 English around 16. We determined their proficiency in L2 French and L3 English using the Oxford Placement Tests.2

Group A: It has 14 trilingual learners with an average age of 17.7. They are advanced in L2 French but beginners in L3 English.

The four groups are:

3

ArNS: A control group of 15 native speakers of Moroccan Arabic (mean age 21.3).

FrNS: A control group of 25 French natives (mean age 25.5).

EngNS: A control group of 25 monolinguals of American English (mean age 22.5).

4.2. Experimental tasks

The experiment is an acceptability judgment test (AJT) and a preference test (PT) in Moroccan Arabic, French and English versions. While the control groups did their respective language tests, the L3ers completed the L2 and L3 tests, each on a separate day starting with the L2 tasks. We controlled for the L3 vocabulary using familiar words in the learners’ textbook or words that have similar meanings in French and English. The format of the tasks and scoring were adapted from Cowart (1997), Gass (1994) and Gass and Mackey (2007).

Each AJT has 12 grammatical targets (six exemplars per sentence type), 12 ungrammatical variants (six exemplars per sentence type) and 48 fillers (see the Appendix). There was a balance between the overall number of grammatical and ungrammatical items in the tasks. For illustration below, we use the English tokens. All of the pair sentences are grammatical in the L1 while those with an asterisk are not in the L2 and L3: 2 http://www.lang.ox.ac.uk/tests/ 3 The low number of L3 participants is a common problem to most L3 studies. We

could not get a larger L3 group because of the L2-L3 proficiency matching. Most L3 beginners we screened were L2 intermediates. With this level in French, their L2 knowledge may not be nativelike enough to transfer in L3 English. We concede this limitation.

12 A. Hermas

Verb-subject inversion in declaratives

(16) An Italian discovered America in 1492. *Enjoys a neighbour Egyptian movies.

Expletive construction

(17) It is certain that smoking causes cancer. *Is possible that the team wins the cup.

Four responses are possible: completely/maybe unacceptable and maybe/completely acceptable. For the scoring, a correct answer gets three points and a wrong one zero. An answer other than the correct one (i.e., maybe acceptable/unacceptable) gets two points if closer to the optimal answer and one point if closer to the wrong one.

The PT has 12 sentence pairs made of a grammatical and an ungrammatical sentence (six pairs per sentence type) and 54 filler pairs, balancing the total number of grammatical and ungrammatical items in the task (see the Appendix). Unlike the French and English tasks, both members of the target pairs were grammatical in the Arabic task. Three responses are possible: sentence A is better than B, sentence B is better than A, and A and B are equally acceptable (similar). The use of the PT is meant to draw the learners’ attention to the locus of the problem (the presence versus absence of the expletive or the (non)-reversed SV-VS word order), which excludes considering other structural or lexical elements as potential justification of the learners’ judgments.

Verb-subject inversion in declaratives

(18) a. *Relaxes Mom in a room upstairs. b. Mom relaxes in a room upstairs.

Expletive construction

(19) a. It appears that the boss is happy. b. *Appears that the boss is happy.

The scoring is as follows: in a pair containing a grammatical and an ungrammatical target, two points are granted for the choice of the grammatical target, one for the choice of both, and zero for the choice of the ungrammatical variant. Since both sentences are licit in MA, the scoring for the Arabic task is different: two points for the choice of both sentences, one

13 International Journal of Language Studies, 8(2) 1-24

for the choice of one of them, and zero for the choice of none. The dependent variable in both tests is accuracy calculated following the scoring above.

4.3. Procedure

Except for the Arabic experiment, the tests were carried out on a computer, using the online software Lea4

For French, the average reading times were 2700 msec for AJTL2 and 4500 msec for PTL2, and these were taken as the display times for the French controls and Group A, whose members are L2 advanced. For English, the display times for the natives were 2400 msec in AJTL3 and 5000 msec in PTL3. For the L3 beginners, seven non-experimental learners did the reading pre-tests. Their average reading times of the targets were 4500 msec in AJTL3 and 7500 msec in PTL3. We tested the French and English control groups in Quebec and the learners and Arabic controls in Morocco. The learners took the L2 and L3 tests on two separate days. Following the experiment, we had informal interviews (debriefing) in Arabic of the L3 beginners about what they thought of the relationship among the three languages. This was meant to inform us about the psycholinguistic proximity that they established among MA, French and English.

. The software randomizes item sequencing for every participant. The participants had to press the appropriate answer key as quickly as possible. To prevent them from re-reading the items and therefore reduce the learners’ recourse to metalinguistic knowledge, we controlled for the display time based on average reading times established by reading pre-tests involving seven native speakers; these pre-tests included only the target sentences.

5. Results

The results of the two tests were analyzed using repeated-measures ANOVAs. Bonferroni corrections were used for post hoc tests, keeping alpha at .05 for all analyses.

5.1. Acceptability judgment test

The results of the AJT are displayed in Table 1. The scores show the participants’ accuracy in correct answers, not error percentages. The maximal accuracy rate per condition is 18 points (6 targets × 3 points) or 100%.

4 www.microbe.ca; Since the software LÉA does not support the Arabic alphabet and

the sociolinguistic tradition has established Moroccan Arabic as a ‘dialect’ that is spoken but not written, we used a bimodal paper-and-pencil format for the AJT and PT tasks in Moroccan Arabic.

14 A. Hermas

Table 1 Accuracy Scores % by Structure in AJT Structure Group A L2 FrNS Group A L3 EngNS ArNS Inversion S-V sequence 88.88 94 84.50 94.88 93.33 *V-S sequence 95.22 95.55 76.16 98.44 95.94 Expletive Lexical 92.05 94.66 66.27 91.11 91.50 *Null 70.22 82.66 48.00 86.22 96.66

A repeated-measures ANOVA revealed a significant main effect for group [F (4, 88) = 25.59, p<.001], a significant main effect for structure [F (3, 264) = 23.42, p<.001], and a significant interaction effect for structure by group [F (12, 264) = 3.92, p<.01]. Bonferroni5

Between-group comparisons tested the learners’ performance for each structure against the native norm. Specifically, Bonferroni test indicated significant differences between the L3ers and the English controls for the grammatical SV constructions (p = .042), for the ungrammatical VS in declaratives (p<.001), for the grammatical constructions with lexical expletives (p<.001) and finally for their ungrammatical variants with null expletives (p<.001). In the discussion below, we add further comments on the scores obtained by the learners and the natives on some specific structures.

post-hoc test showed no significant difference between the L2ers and the French natives (p = .61), but the difference between the English natives and the L3 beginners was highly significant (p<.001).

5.2. Preference test

Table 2 presents the results of the PT. The maximal accuracy rate is 12 (6 targets × 2 points) or 100%.

Table 2 Accuracy Scores % by Structure in PT Structure Group A L2 FrNS Group A L3 EngNS ArNS Inversion 98.25 97.66 94.08 98.33 96.08 Expletive 92.83 94 70.83 89.33 90

The scores were better in PT than in AJT. This was expected given the comparative nature of the task. A repeated-measures ANOVA revealed that the main effect for group was significant [F (4, 88) = 13.42, p < .001]. A significant main effect for structure was also obtained [F (1, 88 = 36.27, p <

5 “Bonferroni is generally conservative, but if you want guaranteed control over the

Type I error rate [believing that there is a genuine effect when in fact there isn’t] then this is the test to use” (Field, 2009, p. 374).

15 International Journal of Language Studies, 8(2) 1-24

.001] together with a significant interaction effect of group by structure [F (4, 88 = 4.49, p < .01]. The post hoc Bonferroni test signalled no significant difference between the L2ers and the French natives (p = 1) but the difference between the L3ers and the English controls was highly significant (p < .001). Contrasting the scores for each structure of the L3ers and the English controls, a post hoc test indicated highly significant differences on lexical/ null expletive pairs (p = .01). However, the beginning L3ers and the English natives performed comparably on SV/VS pairs (p = .65).

6. Discussion

The results indicate that only L1 Arabic has a significant influence on the initial stages of L3 English. If L2 French were involved, we would expect a significant facilitative effect of French on the accuracy of the L3 beginners (L2 = L3), leading them to accept the grammatical sentences and reject the ungrammatical ones.

The analyses showed that the L2 knowledge of the L3 beginners did not have any bootstrapping effect on their L3 performance despite the (psycho)-typological proximity between the two languages and the learners’ nativelike L2 performance. While their performance was comparable to that of the French controls on all the conditions, this did not have any moderating effect on accepting non-target constructions in L3 English. On the contrary, it was the language that was more (psycho)-typologically distant that interfered. Remember that, for the grammatical conditions, the L1 allows SV and the phonetic realization of expletives is optional; therefore, the influence of the L1 and the L2 cannot be dissociated. We need to focus on the ungrammatical conditions, VS sequences and null expletive constructions. To show the extent of L1 Arabic negative influence, we use transfer rates or error percentages instead of accuracy percentages in the tables by deducing the L3ers’ scores on the ungrammatical sentences from the optimal 100% accuracy. The L3 beginners’ performance in the AJT showed 52% negative transfer on null expletive constructions and 23.84% on VS inversion in declaratives. Even the comparative nature of the PT and the nativelike L2 French knowledge did not help the L3ers minimize the error percentage of the lexical-null expletive pairs which reached 29.17%. Overall, the L3ers’ English interlanguage sanctions both the grammatical and ungrammatical constructions of the null subject parameter, reflecting facilitative and non-facilitative transfer from L1 Arabic. We contend then that L2 French did not help improve the accuracy of the L3 English beginners. We consider now whether L2 French had boosted the L3ers’ scores on the second structure, the verb-subject inversion in declaratives.

16 A. Hermas

The L3ers’ accuracy scores of 76.16 % on the ungrammatical VS declaratives in the AJT (significantly different from the English natives’ score) and 94.08% on SV/VS pairs in the PT (nonsignificant difference with the English natives) give the impression that it is the only property of the null subject parameter successfully acquired or probably transferred from L2 French into L3 English. We disagree with this interpretation for two reasons. First, the difference between the two scores may be task-based. Unlike the AJT, the PT has a comparative nature that sensitizes the learners to the structural difference in the SV/VS sequences. Second, and more important, both scores reflect the facilitative influence of L1 Arabic. In MA, VS is stylistically marked whereas SV is unmarked (Fassi Fehri, 2000, p. 129). The L3 beginners’ progress on declarative sentences with VS inversion, unlike the expletive construction, is accounted for by the psycholinguistic markedness hypothesis (Kellerman, 1983, pp. 117-118). The hypothesis claims that marked linguistic properties are not transferable and learners generally avoid them. It follows that we have here an instance of facilitative transfer from MA, whose native speakers avoid transferring the marked VS sequences into L3 English.6

In another respect, we need to comment on the scores of the French and English controls for null expletive constructions in the AJT. First, higher scores would not change the interpretation of the results, as this study investigates the L3 initial stages, not ultimate attainment, where the native speakers provide a baseline to which the performance of the (advanced) learners is compared. Second, the native speakers’ scores are not unexpected. Their acceptance of expletive subject omission is extremely common in the informal varieties. Some accounts describe subject omission in English as truncation (Radford, Atkinson, Britain, Clahsen, & Spencer, 1999) or diary drop (Haegeman & Ihsane, 2001; Liceras & Diaz, 1999)

In summary, unlike an account based on L2 French transfer, which is exclusively facilitative in nature, the alternative account of transfer from L1 Arabic has the advantage of explaining the facilitative and non-facilitative effects observed in the L3ers’ performance. This offers a unified account of the acquisition of the four grammatical and ungrammatical constructions in L3 English.

7

6 A third additional argument against claiming facilitative influence of L2 French on

SV/VS scores is that previous studies of L2A where the L1 is a null subject language showed that the SV/VS alternation was the first to be fixed by the L2ers (e.g., Liceras, 1989; Tsimpli & Rousseau, 1991; White, 1985, 1986).

.

7 Carroll (1991) states that null expletives are perfectly acceptable in informal English and claims that the expletive subject is deleted only at the surface level. There is no null pro underlyingly.

17 International Journal of Language Studies, 8(2) 1-24

Finally, we point out some observations pertaining to the individual data in L3 English. No learner scored consistently across the four constructions. The L3ers tend to score well on some conditions but they obtain only average or below average scores on others. The best accuracy scores were obtained by L3er 7, who scored above 83% on three conditions but got 33.33% on null expletives in the AJT and 75% on overt/ null expletive pairs in the PT. This was also the case of L3er 8, who obtained 83% on SV/VS pairs in the PT but only 44.44% on the grammatical expletive constructions in the AJT.

In summary, we conclude that the construction of grammar in the initial stages of L3 English is influenced exclusively by L1 Arabic while prior knowledge of L2 French did not play any facilitative role, at least for the SV-VS sequences and null-lexical expletive constructions of the null subject parameter tested here.

7. Contribution of the current study to the field

In light of the discussion, we review the research objective of the study. It considers the source language of morphosyntactic transfer in the initial stages of L3 English and seeks to determine which of the L1 transfer scenario or the TPM provides a viable account of the acquisition of the properties of the null subject parameter. The involved factors of language proximity, typology and psychotypology should have given advantage to L2 French.

For language proximity, French is holistically more proximate to English and both are typologically similar as non-null subject languages. Furthermore, the L3 beginners perceive French to be closer to English. Nevertheless, L2 French did not have the expected facilitative effect of rejecting the ungrammatical tokens (VS and null expletives). As argued above, in the L3 initial stages, it was L1 Arabic that was the source of facilitative effect on SV-VS sequences (in the PT) and non-facilitative effect on null-lexical expletive constructions. The predictions of the TPM (Rothman 2010, 2011) were not supported since Arabic is not (psycho)-typologically closer to English neither holistically nor on the parametric level, unlike L2 French. It follows that the L1 scenario accounts better than the TPM for the acquisition of the two formal properties of the null subject parameter.

This study provides additional empirical support for L1 (non)-facilitative transfer in L3A contra the expected deterministic influence of L2-L3 (psycho)-typological proximity. It also complements Lozano’s (2003) study of the L3 end state by empirically investigating the L3 initial stages and remedies Na Ranong and Leung’s (2009) variable language confound (their source of transfer L1 Thai is structurally and psychotypologically similar to L3 Mandarin). Both studies argued for L1 transfer in L3A. Finally, the present study empirically confirms Odlin’s (2003, p. 472) suggestion that is, in

18 A. Hermas

trilingualism, a different language (e.g., Arabic) may have an impact on the acquisition of another (e.g., English) especially if the first is the native language. A caveat is in order. We argue here for an L1 explanation of the L3 data, but we do not propose a new ‘L1 status factor’ model for multilingual transfer. To formalize such a model, we need a larger experimental sample size, more target linguistic properties and different language triplets.

8. Conclusion

This study examined the acquisition of two constructions of the null subject parameter, SV-VS in declarative sentences and null-lexical expletive constructions. The learners were L1 Arabic-advanced L2 French adults learning L3 English. The study supported the view that L1 Arabic was the source of morphosyntactic (non)-facilitative transfer in the L3 initial stages. Thus, the L1 transfer scenario provided a more viable account of the L3 beginners’ performance than the Typological Primacy Model that would predict L2 transfer. The L3 learners did not resort to transfer from L2 French despite the implication of three potentially deterministic factors, i.e., the language genetic proximity, structural similarity and psychotypological relatedness between L2 French and L3 English. Further L3 investigations of other linguistic properties would help clarify what conditions the primacy of the native language status over the influence of (psycho)-typological proximity.

The Author

Abdelkader Hermas, Université du Québec À Montréal, Canada Email: Address: C.P. 8888, succ. Centre-ville, Montréal (Québec), H3C 3P8, CANADA

[email protected]

References

Aoun, J., Benmamoun, E., & Sportiche, D. (1994). Agreement, word order, and conjunction in some varieties of Arabic. Linguistic Inquiry, 25 (2), 195-220.

Arteaga, D., & Herschensohn, J. (2004). Case, agreement, and expletives: A parametric difference in Old French and Modern French. In J. Auger, J. Clancy Clements & B. Vance (Eds.), Contemporary approaches to Romance linguistics, (pp. 1-16). Amsterdam: John Benjamins.

Auger, J. (1994). Pronominal clitics in Québec colloquial French: A morphological analysis. Unpublished doctoral dissertation, University of Pennsylvania, USA.

19 International Journal of Language Studies, 8(2) 1-24

Bardel, C., & Falk, Y. (2007). The role of the second language in third language acquisition: The case of Germanic syntax. Second Language Research, 23 (4), 459-484.

Belletti, A. (1999). Italian/Romance clitics: Structure and derivation. In H. van Riemsdijk (Ed.), Clitics in the languages of Europe, (pp. 543-580). Berlin: Mouton de Gruyter.

Benmamoun, E. (1995). The conditions on pro and the ECP. In R. Aranovich, W. Byrne, S. Preuss & M. Senturia (Eds.), The Proceedings of the Thirteen West Coast Conference on Formal Linguistics, (pp. 173-188). Stanford, CA: Center for the Study of Language & Information.

Benmamoun, E. (2000). The feature structure of functional categories: A comparative study of Arabic dialects. Oxford: Oxford University Press.

Bobaljik, J. D. (2002). A-chains at the PF-interface: Copies and ‘covert’ movement. Natural Language and Linguistic Theory, 20 (2), 197-267.

Carroll, S. E. (1991). Review of the book La variation dialectale en grammaire universelle, by Y. Roberge & M. T. Vinet, 1989. Revue Québécoise de Linguistique, 20, 303-308.

Chomsky, N. (2000). Minimalist inquiries. In R. Martin, D. Michaels & J. Urigareka (Eds.), Steps by steps: Essays on minimalist syntax in honor of Howard Lasnik, (pp. 89-155). Cambridge, MA: MIT Press.

Chomsky, N. (2001a). Beyond explanatory adequacy. Cambridge, MA: MIT Press.

Chomsky, N. (2001b). Derivation by phase. In M. Kenstowicz & K. Halle (Eds.), A life in language, (pp. 1-52). Cambridge, MA: MIT Press.

Cowart, W. (1997). Experimental syntax: Applying objective methods to sentence judgement. London: SAGE publications.

De Angelis, G. (2007). Third or additional language acquisition. Toronto: Multilingual Matters.

De Cat, C. (2007). French dislocation: Interpretation, syntax, acquisition. New York: Oxford University Press.

20 A. Hermas

Falk, Y., & Bardel, C. (2010). The study of the role of the background languages in third language acquisition. The state of the art. International Review of Applied Linguistics, 48 (2-3), 185-219.

Falk, Y., & Bardel, C. (2011). Object pronouns in German L3 syntax: Evidence for the L2 status factor. Second Language Research, 27 (1), 9-82.

Fassi Fehri, A. (1993). Issues in the structure of Arabic clauses and words. Dordrecht: Kluwer Academic.

Fassi Fehri, A. (2000). Al-Lisaniyat wa al-lugha al-Arabiya: Namadij tarkibiya wa dalaliya [Linguistics and the Arabic language: syntactic and semantic models] (Vol. 2). Casablanca: Toubqal Publishers.

Field, A. (2009). Discovering statistics using SPSS. London : Sage Publications.

Flynn, S., Foley, C., & Vinnitskaya, I. (2004). The Cumulative-enhancement model for language acquisition: Comparing adults’ and children’s patterns of development in first, second and third language acquisition of relative clauses. The International Journal of Multilingualism, 1(1), 3-16.

García Mayo, M., & Rothman, J. (2012). L3 morphosyntax in the generative tradition: From the initial state and beyond. In J. Cabrelli Amaro, S. Flynn & J. Rothman (Eds.), Third language acquisition in adulthood, (pp. 9-32). Amsterdam: John Benjamins.

Gass, S. (1994). The reliability of second-language grammaticality judgments. In E. Tarone, S. Gass & A. Cohen (Eds.), Research methodology in second language acquisition, (pp. 303-322). Hillsdale, NJ: Lawrence Erlbaum.

Gass, S. M., & Mackey, A. (2007). Data elicitation for second and foreign language research. Mahwah, NJ: Lawrence Erlbaum.

Haegeman, L., & Ihsane, T. (2001). Adult null subjects in the non-pro-drop languages: Two diary dialects. Language Acquisition, 9 (4), 329-346.

Kellerman, E. (1983). Now you see it, now you don’t. In S. M. Gass & L. Selinker (Eds.), Language transfer in second language learning, (pp. 112-134). Rowley, Mass.: Newbury House.

Laenzlinger, C. (1998). Comparative studies in word order variation. Amsterdam: John Benjamins.

21 International Journal of Language Studies, 8(2) 1-24

Leung, Y-K. I. (2002). Functional categories in second and third language acquisition: A cross-linguistic study of the acquisition of English and French by Chinese and Vietnamese speakers. Unpublished doctoral dissertation, McGill University, Canada.

Liceras, J. (1989). On some properties of the pro-drop parameter: Looking for missing subjects in non-native Spanish. In S. M. Gass & J. Schachter (Eds.), Linguistic perspectives on second language acquisition, (pp. 109-133). Cambridge, MA: Cambridge University Press.

Liceras, J. M., & Díaz, L. (1999). Topic-drop versus pro-drop: null subjects and pronominal subjects in the Spanish L2 of Chinese, English, French, German and Japanese speakers. Second Language Research, 15 (1), 1-40.

Lozano, C. (2003). Focus, pronouns and word order in the acquisition of L2 and L3 Spanish. Unpublished doctoral dissertation, University of Essex, UK.

Montrul, S., Dias, R., & Santos, H. (2011). Clitics and object expression in the L3 acquisition of Brazilian Portuguese: Structural similarity matters for transfer. Second Language Research, 27 (1), 21-58.

Na Ranong, S., & Leung, Y-k. I. (2009). Null objects in L1 Thai-L2 English-L3 Chinese: An empirical take on a theoretical problem. In Y-K. I. Leung (Eds.), Third Language Acquisition and Universal Grammar, (pp. 162-191). Bristol, UK: Multilingual Matters.

Odlin, T. (2003). Crosslinguistic influence. In C. Doughty & M. H. Long (Eds.), Handbook of second language acquisition, (pp. 436-486). Oxford: Blackwell.

Ouhalla, J. (1994). Verb movement and word order in Arabic. In D. Lightfoot & N. Hornstein (Eds.), Verb movement, (pp. 41-72). Cambridge, MA: Cambridge University Press.

Radford, A. (2004). Minimalist syntax: Exploring the structure of English. Cambridge: Cambridge University Press.

Radford, A., Atkinson, M., Britain, D., Clahsen, H., & Spencer, A. (1999). Linguistics: An introduction. Cambridge, MA: Cambridge University Press.

22 A. Hermas

Roberge, Y. (1986). The syntactic recoverability of null arguments. Unpublished doctoral dissertation, University of British Columbia, Canada.

Roberge, Y., & Vinet, M.-T. (1989). La variation dialectale en grammaire universelle. Montréal: Presses de l’Université de Montréal.

Rothman, J. (2010). On the typological economy of syntactic transfer: Word order and relative clause attachment preference in L3 Brazilian Portuguese. International Review of Applied Linguistics, 48 (2-3), 245-273.

Rothman, J. (2011). L3 syntactic transfer selectivity and typological determinacy: The typological primacy model. Second Language Research, 27 (1), 107-127.

Rothman, J. (in press). Cognitive economy, non-redundancy and typological primacy in L3 acquisition: Evidence from initial stages of L3 Romance. In S. Baauw, F. Dirjkoningen & M. Pinto (Eds.), Romance Languages and Linguistic Theory 2011. Amsterdam: John Benjamins.

Rothman, J., & Cabrelli Amaro, J. (2010). What variables condition syntactic transfer? A look at the L3 initial state. Second Language Research, 26 (2), 189-218.

Rothman, J., Iverson, M., & Tiffany, J. (2011). Introduction: Some notes on the generative study of L3 acquisition. Second Language Research, 27 (1), 5-19.

Rowlett, P. (2007). The syntax of French. Cambridge : Cambridge University Press.

Tsimpli, I., & Roussou, A. (1991). Parameter resetting in L2? UCL Working Papers in Linguistics, 3, 149-70

White, L. (1985). The pro-drop parameter in adult second language acquisition. Language Learning, 35 (1), 47-62.

White, L. (1986). Implications of parametric variation for adult second language acquisition: An investigation of the Pro-drop parameter. In V. J. Cook (Ed.), Experimental approaches to second language acquisition, (pp. 57-72). Oxford: Pergamon Press.

23 International Journal of Language Studies, 8(2) 1-24

APPENDIX

1. L2 French Tests

1.1. Acceptability Judgment Test

A. Verb-subject inversion

1. Une amie fait le ménage chez-moi. 7. *Prévoit la météo une belle journée. 2. Un ami fume beaucoup de cigarettes. 8. *Ferme le cinéma ses portes à minuit. 3. Une secrétaire préfère le poulet rôti. 9. *Cherchait une dame son sac à main. 4. Une fille apprend la musique berbère. 10. *Arrêtait un policier le voleur en fuite. 5. Une dame souffre de maux de tête. 11. *Affecte la malnutrition notre santé. 6. Un bébé souffre de problèmes de santé. 12. *Inspirent les chanteurs nos jeunes.

B. Expletive construction 1. Il est interdit de fumer dans les salles. 7. *Est interdit de nuire à l’ordre public. 2. Il est probable que des amis viennent. 8. *Est possible qu’un soldat quitte l’armée. 3. Il est certain que tu réussiras le test. 9. *Est nécessaire que le travail commence. 4. Il faut que le président soit honnête. 10. *Semble que les employés sont satisfaits. 5. Il semble que la crise va s’aggraver. 11. *S’est révélé que le pétrole est polluant. 6. Il paraît que le monde va vers le chaos. 12. *Importe que tout citoyen ait ses droits.

1.2. Preference test

A. Verb-subject inversion 1a. *Détruisait un incendie le magasin. b. Un incendie détruisait le magasin. 2a. Un patron supprimait dix postes. b. *Supprimait un patron dix postes. 3a. *Aidait un passant Laura à se relever. b. Un passant aidait Laura à se relever. 4a. *Suit un soldat les ordres à la lettre. b. Un soldat suit les ordres à la lettre. 5a. Une dame travaille dans la librairie. b. *Travaille une dame dans la librairie. 6a. Un policier arrêtait deux casseurs. b. *Arrêtait un policier deux casseurs.

B. Expletive construction

1a. *Semble que la météo va s’améliorer. b. Il Semble que la météo va s’améliorer. 4a. Il est possible que le froid soit mortel. b. *Est possible que le froid soit mortel. 2a. *Semble que Nick gagnera la course. b. Il semble que Nick gagnera la course. 5a. Il est difficile que Léo obtienne justice. b. *Est difficile que Léo obtienne justice. 3a. Il est certain que le sida est mortel. b. *Est certain que le sida est mortel. 6a. *S’est révélé que la rumeur était vraie. b. Il s’est révélé que la rumeur était vraie.

2) L3 English Tests

2.1. Acceptability Judgment Tests

A. Verb-subject inversion 1. Some people stay at work until night. 7. *Needs a man some money to travel. 2. An Italian discovered America in 1492. 8. *Repaired a mechanic the bus easily. 3. Few pupils learn Spanish at school. 9. *Comes a teacher to school on time. 4. Many people prefer private hospitals. 10. *Finds a friend geography very easy. 5. Pollution spreads illnesses and death. 11. *Enjoys a neighbour Egyptian movies. 6. Some guests came to the party late. 12. *Invented Bell the telephone in 1876.

24 A. Hermas

B. Expletive construction

1. It is certain that smoking causes cancer. 7. *Is necessary that offices get computers. 2. It is possible for the Iraqis to live in peace. 8. *Is possible that the team wins the cup. 3. It seems that health depends on nutrition. 9. *Appears that a man is waiting outside. 4. It appears that my son is a bad child. 10. *Are some exciting films on TV tonight. 5. There remains no hope for the victims. 11. *Seems that money is everything today. 6. There were very few clouds in the sky. 12. *Remain many problems to understand.

2.2. Preference test

A. Verb-subject inversion 1a. *Relaxes Mom in a room upstairs. b. Mom relaxes in a room upstairs. 2a. Anne plays boxing with a friend. b. *Plays Anne boxing with a friend. 3a. *Revises Sami history after dinner. b. Sami revises history after dinner. 4a. *Needs Paula a syrup for her cold. b. Paula needs a syrup for her cold. 5a. Children play chess in the garden. b. *Play children chess in the garden. 6a. Lucy prepares delicious dishes. b. *Prepares Lucy delicious dishes.

B. Expletive construction

1a. *Surprises me that Hind is absent. b. It surprises me that Hind is absent. 2a. *Were many people in the cinema. b. There were many people in the cinema. 3a. It is certain the US has the best athletes. b. *Is certain the US has the best athletes. 4a. It appears that the boss is happy. b. *Appears that the boss is happy. 5a. It is possible that James arrives late. b. *Is possible that James arrives late. 6a. *Seems that Sandra has a problem. b. It seems that Sandra has a problem.