Argument Omission in Signl2 Acquisition by Deaf Learners Back to the Inhibition

John Benjamins Publishing Company

This is a contribution from Three Streams of Generative Language Acquisition Research. Selected papers from the 7th Meeting of Generative Approaches to Language Acquisition – North America, University of Illinois at Urbana-Champaign. Edited by Tania Ionin and Matthew Rispoli. © 2019. John Benjamins Publishing Company This electronic file may not be altered in any way. The author(s) of this article is/are permitted to use this PDF file to generate printed copies to be used by way of offprints, for their personal use only. Permission is granted by the publishers to post this file on a closed server which is accessible to members (students and staff) only of the author’s/s’ institute, it is not permitted to post this PDF on the open internet. For any other use of this material prior written permission should be obtained from the publishers or through the Copyright Clearance Center (for USA: www.copyright.com). Please contact [email protected] or consult our website: www.benjamins.com Tables of Contents, abstracts and guidelines are available at www.benjamins.com Argument omission in SignL2 acquisition by deaf learners Back to the inhibition

Elena Koulidobrova Central Connecticut State University

We examine subject omission in American Sign Language (ASL) as a second sign language (SignL2) by deaf and hard-of-hearing users of Emirati Sign Language (EmiratiSL) with variable ages of exposure to SignL1. Results indicate that during a narrative, similarly to unimodal speech bilinguals, SignL2 learners of low ASL proficiency oversupply overt subject arguments – both NPs and pronouns (IX). In this, SignL2 learners behave differently from other ASL bilinguals described in the literature. We interpret the findings as offering support to the line of research arguing that certain atypicalities in multilingual production are best explained as resulting from the interaction between linguistic and general cognitive control mechanisms – i.e. that bilinguals oversupply arguments due to processing reasons. This phenomenon is modality independent.

Keywords: null subject, sign language, deaf, narrative

1. Introduction

Much of research on the nature of multilingual language acquisition has histori- cally been fueled by the following questions: How autonomous are the languages? How exactly do separate grammars interact in sequential and simultaneous multilinguals? In what ways do the languages influence one another during development? More recently, the field has also explicitly focused on asking whether any additional factors play a role in multilingual language development. For instance, if cross-language interaction is possible, can it be isolated and examined independently? While the starting point in much of the older literature is ‘yes,’ the question is not trivial – other characteristics of development in general and bilingual development in particular might interfere with such examination. Among them are the roles of general integration of linguistic experience in cognition and of

the linguistic modality. Concretely, because languages of a bilingual do not (co-) exist in a vacuum, we might expect various factors affecting multilingual linguistic experience to create a constellation of effects, only some of which may be visible/ immediately detectible. In this vein, in various works Sorace and colleagues (see Sorace, 2016 and references therein) have argued that anaphora resolution tasks reveal complexities of integration of linguistic and executive control forces in multilingual production. The argumentation has relied on the observation that at least from the point of view of phonology, only one of the languages of a multilingual can be overtly realized in production, and all the rest must be inhibited (Kroll, Bobb, & Wodniecka, 2006, i.a.). This creates an increased processing load, resulting in the multilinguals’ reliance on resumptive forms (Asudeh, 2004). As one type of evidence for this, researchers have used rates of overt argument suppliance. For instance, in the early days of research on argument omission in simultaneous bilingual production, the following observation emerged: bilinguals acquiring two languages, one of which allows/forces arguments to remain null (such as Italian, Spanish, Japanese, Inuktitut, and Turkish), were originally expected to also drop arguments in their non-null argument language (e.g. English) because the property was predicted to ‘transfer’ (cf. Hulk & Müller, 2000). Yet, the opposite appears to be true – bilinguals of all ages have been shown to oversupply non-target overt pronouns in the null-argument language instead. This finding has been well attested with various combinations of languages (Serratrice, Sorace, & Paoli, 2004 [Italian-English]; Juan-Garau & Pérez-Vidál, 2000 [Spanish-English]; Pinto, 2006 [Italian-Dutch]; Hacohen & Schaeffer, 2007 [French-English, Hebrew- English]; Schmitz, Patuto, & Müller, 2012 [Italian-German, French-German]; Mishina-Mori, 2007 [Japanese-English]; Blais, Oshima-Takane, Genesee, & Hirakawa, 2010 [Japanese-French]; Huang, 1999 [Chinese-English]; Haznedar, 2010 [Turkish-English]; Zwanziger, Allen, & Genesee, 2005 [Inuktitut-English]), leading to the suggestion that cross-linguistic interaction must be uni-directional for some reason – the issue to which we return below. Later works, however, have argued that morpho-syntactic influence cannot possibly account for the range of the observed effects. This is because studies involving pairs of languages both of which allow/force overt arguments to remain null (e.g. Spanish-Italian, Serratrice, Sorace, Filiaci, & Baldo, 2011) have yielded the same results: bilinguals, especially late learners, oversupply overt arguments in the manner that is non-target for either their native or additional languages – L1 and L2, respectively (Belletti, Bennati, & Sorace 2007; Sorace, 2003, 2005, 2006a, b, 2011; Sorace & Filiaci, 2006; Tsimpli & Sorace, 2006; Tsimpli, Sorace, Heycock, & Filiaci, 2004; Chamorro, Sorace, & Sturt, 2015). What bears a special note here is this: the effect is robust and covers both simultaneous (i.e. the effect is observed in both of L1s) and sequential (i.e. the effect is observed in the L2) bilinguals in various language families.

The question we ask here is whether this finding is modality dependent. The rationale for the question is this: the original sets of studies reported above involve language combinations within the same modality – all spoken. However, a number of studies on bi-modal language acquisition (e.g. English-American Sign Language (ASL), Brazilian Portuguese-Lingua de Signas Brasiliera (Libras)) have shown a few unattested linguistic patterns and behaviors (Lillo-Martin, Quadros, Koulidobrova, & Chen Pichler, 2009). For instance, research on bimodal sign- speech production has demonstrated that unlike, e.g., Spanish-English bilinguals, child and adult ASL-English bilinguals do not tend to code-switch but, rather, code-blend (Emmorey, Borinstein, Thompson, & Gollan, 2008a; Petroj, Guer- rera, & Davidson, 2014). Specifically, with respect to anaphora resolution tasks, relatively high rates of argument omissions as well as the types of null arguments unattested in the literature on monolingual and bilingual acquisition (Figure 1) have been recorded in the English of bilingual hearing children of deaf adults and deaf children with cochlear implants (Koulidobrova, 2013b, 2015, 2017a).

Unattested Null Subjects: English non-left edge with a modal subject of embedded clause subject of a question with Aux (%) multiple

TOM LEX Figure 1. Rates of subject omission in atypical environments (unattested in monolinguals and unimodal bilinguals) in the English of two ASL-English bilinguals (Koulidobrova, 2017a) By the same token, unlike other young bilingual children acquiring a null argument (NA) language and English simultaneously, bimodal bilingual children do not oversupply overt pronouns in their null argument language – ASL, at least in spontaneous production (Koulidobrova, 2014, 2017a), if not in an experimental condition, and not till they are older (Reynolds, 2016). Adult bimodal learners of ASL behave on par: Frederiksen & Mayberry (2015) demonstrate that the rate of argument omission in the elicited narratives of late learners of ASL whose L1 is English does not differ significantly from that of native signers in their sample. In other words, both simultaneous and sequential ASL-English bilinguals behave differently from spoken language bilinguals. Literature offers a few potential explanations for these observations. First, it is possible that cross linguistic transfer accounts have not been explored fully.

For example, languages may in fact interact with one another in the domain of argument omission, but there is just something about the previously examined language combinations that hides this transfer. The question of course is what that could be. We know, for example, that the problem cannot lie in the necessary ‘unidirectionality’ of cross-language interaction. Let us examine the reason closely (see Koulidobrova, 2012 for an extended discussion). Tsimpli (2011) suggests that overt lexical items (e.g. personal pronouns) have a higher chance of being observed in the language of a bilingual over covert ones (e.g. pro) because overt items are the default. Tsimpli’s view can be cast in the following way: a bilingual reaches into her lexicon for an item, and when two of them are in competition, the default wins. This account offers an explanation for the robust findings from the unimodal spoken language studies; however, it also predicts that bimodal bilinguals will not omit arguments because the default op- tion should win, irrespective of modality. In a similar vein, Serratrice et al. (2004) argue that the difference between overt pronouns in Italian and English lies in the constellation of features that make these pronouns. The use of pronouns in Italian is much more pragmatically restricted than the use of pronouns in English, and bilinguals use the lexical item that is the least pragmatically restricted between the two. If that were the case, however, we would not be seeing the difference between unimodal and bimodal bilinguals either: ASL omits its arguments, but this omission is tightly pragmatically constrained (Koulidobrova, 2017b). Thus, while the account predicts that ASL-English bilinguals may use English-like pronouns in ASL, it does not explain extra omissions in English. The third type of account we could rely on has been proposed by Sorace and colleagues (see Sorace, 2016 for an overview): that in tasks involving anaphora resolution (among others), linguistic processes interact with general cognition non-trivially. Concretely, because the two languages of a unimodal bilingual cannot be produced at the same time, and bilinguals must expend cognitive effort inhibiting one of them (Kroll et al., 2006, i.a.), overt arguments surface where they don’t belong for processing reasons, allowing the bilingual to track reference better. An explanation along these lines suggests that in a language combination where such a competition is either not present at all or simply relaxed in some sense (Emmorey, Luk, Pyers, & Bialystok, 2008b, i.a.), different patterns may arise.1 In terms of argument omission, this means that because of the reduced processing load, bimodal bilinguals will not oversupply overt arguments when unimodal bilinguals do – i.e. if there is any cross-linguistic interaction in the domain of argument omission at all, and the presence of the null-argument language in the

1. In such a scenario, cross language interaction, not obscured by processing difficulties, may be visible (Koulidobrova, 2012).

© 2019. John Benjamins Publishing Company All rights reserved Argument omission in SignL2 acquisition by deaf learners 301 linguistic repertoire of the bilingual affects her grammar in any way, the addition of another modality will offer a stage where this phenomenon can be observed. We will henceforth label this explanation an executive control-based account. There is, of course, another possible explanation for the apparent uniqueness of bimodal bilinguals: the fact that the other language is a sign language. In fact, atypical linguistic patterns of sign-speech bilinguals have been previously noted in the literature on linguistic development of hearing children of deaf adults (Todd, 1971; Schiff & Ventry, 1976; Sachs, Bard, & Johnson, 1981; Murphy & Slorach, 1983; Schiff-Myers, 1988; Johnson, Watkins, & Rice, 1992; Seal & Hammett, 1995) and deaf children with functional hearing restored via a cochlear implant (Koulidobrova, 2013b). While the field no longer asks whether sign languages are real languages, it remains clear that the contributions of gestural modality cannot be dismissed in theorizing about language acquisition, especially in terms of multilingualism (see Futrell et al., 2015 for an overview). Pushing this idea further results in a possibility that there is just something about the use of a sign language that affects the production of the spoken one. Let us call this type of explanation a visual language/language-on-the-hands-based account. In this paper, then, we attempt to disambiguate between the last two possibili- ties and offer support for the executive control-based account, relying directly on the data from acquisition of ASL as an additional sign language (SignL2).

2. Detour: Modality issues

Outside of perhaps the immediately obvious difference between languages in different modalities (different articulators and articulatory space, a possibility of bimodal production and non-concatenation, etc.), comparison between sign and spoken languages is additionally complicated in the following manner. The current literature on L2 acquisition of a sign language and of any language by signers is not easily comparable to the literature on L2 acquisition of a spoken language and of any languages by speakers. Consider this: when hearing users of spoken languages learn other spoken languages, this acquisition occurs in the same modality (M1L2). However, traditionally, claims regarding sequential acquisition of sign languages are made about hearing individuals acquiring a sign language (e.g. Spanish speakers learning Catalan Sign Language (LSC) in interpreter training programs, Bell, Or- tells, & Morgan, 2015). This is learning of an additional language across modalities (M2L2) and involves the use of articulators typically utilized for language-specific co-speech gesture. In some sense then, M2L2 learning may involve simultaneously less and more inhibition on the part of the learner – less because both M1L1 and M2L2 can be produced simultaneously (unlike M1L1 and M1L2), and more

because the L1 gesture actually must be inhibited during L2 production. These are the challenges of examining M2L2 acquisition by hearing individuals. Differently, claims regarding L2 acquisition by the deaf/hard-of-hearing tend to focus on learning the written version of the ambient spoken language (e.g. signers of ASL learning English literacy, Dostal & Wolbers, 2016). Such learning also occurs across modalities but cannot easily be labeled M2L2, since what is being learned directly is not a natural language per se but rather literacy representing a natural language, and often a privileged variety thereof (e.g. Standard American English). Further, research has shown that at least some deaf signers code-blend at the lexical level routinely, using lexical items adapted from the local spoken language. For instance, Bank, Crasborn, & Van Hout (2018) note that 12% of the Sign Language of the Netherlands (NGT) corpus produced by the deaf signers contains non-redundant mouthings from Dutch. If this finding generalizes, it suggests that such a signer is perhaps necessarily bimodal bilingual to a degree (M2L2), despite not having full access to the other language; this makes isolating effects of each of the languages more difficult. Finally, at least in the US, over 90% of deaf and hard of hearing children are born to hearing families, the vast majority of whom never learn how to sign. As a consequence, a sizable number of deaf children experience language delay which, in turn, has been shown to affect language learning as well as other types of learning, including literacy (Karchmer & Mitchell, 2003; Mayberry & Lock, 2003, Mayberry, 2007, i.a.). Therefore, comparing sequential acquisition of sign languages by the hearing- or of spoken languages by the deaf- to the acquisition of spoken languages by the hearing individuals is a problematic enterprise riddled with confounds (Chen Pichler & Koulidobrova, 2015; Koulidobrova & Palmer, 2016). For a true comparison then, what we need is a context in which users of some sign language (SignL1) acquire an additional sign language (SignL2) – i.e. also an M1L2 environ- ment.2 Although literature on this particular population remains sparse, research has shown that language interaction occurs, at least in the domains of phonology and phonotactics (Ortega & Morgan, 2010, Ortega-Delgado, 2013, Chen Pichler, 2011). We therefore expect other multilingualism-related effects to be visible as well.

3. Study goals and predictions

Recall that previous studies on argument omission in general (Section 1) have shown that unimodal and bimodal bilinguals behave differently with respect to

2. Of course, even in this language combination, language delay affecting many signers remains a confound.

© 2019. John Benjamins Publishing Company All rights reserved Argument omission in SignL2 acquisition by deaf learners 303 argument suppliance in both of their languages in both spontaneous production and in elicited environments. The difference in recorded in Table 1 below.

Table 1. Apparent difference between unimodal and bimodal bilinguals M1 (e.g. Italian/Spanish/Inuktitut – English) M2 (e.g. ASL-English) Target-like in their non-NA language (do not Non-target-like in their NA language (omit) omit) Non-target-like in their NA language Target-like in their NA language (do not (oversupply) oversupply)

The target language under examination here is ASL – a null argument language which, in the relevant property, resembles Japanese (Koulidobrova, 2012, 2017b), Mandarin Chinese (Lillo-Martin, 1991), or Italian (Bahan, Kegl, Lee, MacLaugh- lin, & Neidle, 2000), depending on the analysis. The crucial point here is that ASL allows, and often prefers, its arguments to be omitted, as in (1).3 (1) a. A: COOKIE SEE? ‘Have you seen my cookie?’

B: (?IX1) EAT-UP (*IX3) ‘(I) ate (it)’ b. A: MARY HERE? ‘Is Mary here?’ B: NO OUT TODAY KID SICK ‘No (she) is out today, her kid is sick’ The overt pronominal element is IX, uttered in form identical to the pointing gesture used by the hearing community in the US. ASL differentiates between two forms of IX – 1st and non-1st (Meier, 1990; Lillo-Martin, 1995). Both forms productively participate in anaphoric relations (Sandler & Lillo-Martin, 2006; Koulidobrova & Lillo-Martin, 2016). Previous research on ASL narratives has shown that adults omit over 50% of arguments: Fredriksen & Mayberry (2015) cite ~ 55% of arguments being omitted by adult M2L2 learners and ~ 54% by L1 users.4 We therefore predict that if the reason for the previously recorded difference between unimodal (M1: speech)

3. Per Sign Language research conventions, the utterance is documented using SMALL CAPS for glosses, with ‘g[…]’ indicating a gesture rather than a sign and ‘#…’ indicating a finger- spelled word.

4. Frederiksen & Mayberry (2015) report much lower rates of argument omission than Wulf, Dudis, Bayley, & Lucas (2002) (the latter cites ~ 70%). However, this difference may be due to the use of markedly different methodologies: the former elicit using tightly controlled narratives using stimuli; the latter record ‘chit-chat’ about experiences/free discourse.

and bimodal (M2: sign-speech) bilinguals is related to the sign modality, then the SignM1L2 learners will pattern with bimodal bilingual learners (M2L2) (the right column of Table 1). If, however, the discrepancy in Table 1 results from the differences in executive control/processing, then unimodal bilingual ASL learners will pattern on par with unimodal learners of two spoken languages, robustly represented in the bilingual literature (the left column of Table 1), supporting the general view that unimodal bilinguals oversupply arguments not because they are default, for example, but because the forced choice into one set of articulators creates more of a processing load, resulting in a higher rate of overt arguments. In other words, unimodal bilingual ASL learners will behave differently from bimodal bilinguals. Further, considering the specific characteristics of the deaf population (see Section 2), one other prediction arises: if the phenomenon under the examination is in fact linguistic in nature, participants who were exposed to SignL1 early may be expected to behave differently from those who were exposed to language late (cf. Boudreault & Mayberry, 2006 and Mayberry, 2007).

4. Methodology

4.1 Participants

The other language of the participants (their SignL1) is Emirati Sign Language. To our knowledge, there is no literature on EmiratiSL; it appears to be related to other Arabic sign languages and, crucially, it resembles Jordanian Sign Language and Japanese when it comes to argument omission (Hendriks, 2008). In other words, like ASL, EmiratiSL is a null-argument language where something like (1) is not only grammatical but typically preferred. The participants of the study are twelve M1L2 learners of ASL in an immersion program in the USA. This program is a part of a residential school for the Deaf that is a sign-based (vis-à-vis oral). As part of the program, all participants receive direct ASL instruction as well as academic content instruction in ASL. Participants are male students from the United Arab Emirates (UAE) in the second year of the program. Additionally, knowledge of written Arabic could be verified for 2 out of 12 participants via an open-prompt writing sample.5 Some

5. The prompt asked students to produce anything they wished to write about in Arabic. Some topic options were (i) your family, (ii) your country, (iii) favorite holiday, (iv) personal introduction, (v) your favorite passage in the (Holy) Book, (vi) other. The rationale for the choices was as follows: with varying levels of linguistic proficiency and literacy, learners were expected to be able to produce texts on the given topics, concrete or abstract. The topic(v) was included because

© 2019. John Benjamins Publishing Company All rights reserved Argument omission in SignL2 acquisition by deaf learners 305 of the subjects report having deaf family members and other experiences with (home-)sign discourse; others were born to hearing parents, have previously at- tended oral-only institutions, and began acquiring their SignL1 concurrently with their SignL2. None of the subjects had any previous exposure to ASL. Details about the ages, hearing status, and age of exposure to L1 and L2 are found in Table 2. For the purposes of this study, we isolated EmiratiSL exposure as a variable (rather than exposure to gestural/visual communication in general, for instance, which we have investigated elsewhere, see Koulidobrova & Palmer, 20166). Figure 2 shows that about half of the participants report having acquired EmiratiSL early and the other half late – i.e. there are no native signers in the sample.7

Table 2. Participant information Age at testing Hearing Status Length of exposure Age of exposure to to ASL (SignL2) EmiratiSL (SignL1) 13,00–19,01 (mean: 10 deaf, 2 hard of 1–2 years (mean: 1.5 3–19 years old (mean: 15;03) hearing years) 7.4 years)

AoA of L

SignL early

SignL late

(%) Figure 2. Age of SignL1 acquisition (AoA)

in the UAE (as in other parts of the world where Arabic is the dominant language and Islam the dominant religion), religious texts is an access point to literacy (Carroll, Al Kahwaji, & Litz 2017).

6. In Koulidobrova & Palmer (2016), we demonstrate that the main predictor of success on the phonological discrimination assessment for ASL as L2 is the overall length of exposure to sign modality, including home-sign systems developed within family units.

7. Due to the lack of literature on EmiratiSL, we could not assess participants’ SignL1 proficiency. Therefore, we relied on self-reports as well as descriptions of family structure: all of our subjects who classified themselves as early learners also reported having older deaf signing siblings and other older family members. These family members may not have been native users of EmiratiSL but they served as models in language acquisition (Carrigan & Coppola, 2012).

4.2 Tasks

We conducted three experiments, the goal of the first two of which was to assess the overall proficiency in ASL. We administered two instruments which have been argued in the literature to serve as valid assessments of ASL L2 proficiency for adults, although they have not (to our knowledge) been used to assess ASL of M1L2 signers: a comprehension test (ASL-CT, Hauser, Paludneviciene, Kurz, Emmorey, & Contreras, 2016) and a test of phonological discrimination (ASL-DT, Bochner, Christie, Hauser, & Searls, 2011). ASL-CT involves matching sign clips and pictures or short videos. In ASL-DT, participants are presented with a model and a set of two sentences contrasting (or not) along some phonological parameter (orientation, location, movement, and handshape). The participants are then required to determine whether the two utterances are the same or different. For example, (2) records HEADACHE and TOOTHACHE which are articulated with the same handshape, orientation, and movement but differ in location (jaw vs. temple); see Figure 3.

Figure 3. Proficiency measure procedure (2) Items: HEADACHE and TOOTHACHE Model: JOE a-IX TOOTHACHE ‘Joe has a toothache’ Signer (a): JOE a-IX TOOTHACHE ‘Joe has a toothache’ [Target: same] Signer (b): JOE a-IX HEADACHE ‘Joe has a headache’ [Target: different] The third experiment was elicitation of narratives. The data consist of 10–15 min free discourse (~ 100 participant utterances) elicited as part of a familiar school field-trip debriefing: recounting the sites that were visited during the trip, retell- ings of the events that had occurred during the time period, among others, thus ensuring a relative consistency in vocabulary. For all subjects, the interlocutor was

© 2019. John Benjamins Publishing Company All rights reserved Argument omission in SignL2 acquisition by deaf learners 307 the participants’ regular immersion program instructor, familiar with the subjects’ linguistic patterns and skilled in eliciting responses.

4.3 Transcription and coding

All of the experimenters were native ASL users (deaf for all participants across all experiments except for 5 subjects in ASL-DT. In the latter case, the experimenter was a hearing native signer who is also an expert interpreter). The data were annotated, coded, and analyzed in ELAN (Figure 4), using conventions of sign language transcription and coding in Chen Pichler et al. (2010). There was one primary transcriber, who is deaf. The transcriber, who also had minimal training in linguistic analysis, also served as one of the coders one year after the original transcription was completed. There were two other coders, both hearing, both trained linguists; one of the two is a native user of ASL. A subset of the data was subjected to the inter-rater variability analysis between two primary coders (deaf and hearing); cases resulting in disagreement (~ 4% of the data) were excluded from the analysis.

Figure 4. Narrative: ELAN window All utterances containing a verb8 in a finite context were analyzed for the presence/ absence of overt arguments (NPs and the pronominal IX) and their pragmatic appropriateness in a given context in adult ASL. Exact imitations as well as incom-

8. Because ASL does not have an overt copula, a number of utterances that could have been potentially relevant for the analysis were excluded.

prehensible/ unidentifiable strings (corresponding to the XXX, and YYY in Chen Pichler et al., 2010) were excluded from the analysis.

5. Results

In terms of the proficiency measures, scores varied (Figure 5), but for all of the participants they corresponded to the beginning level in Bochner et al. (2011) and Hauser et al. (2016), with the range of scores between 40–76%. Further, scores on ASL-CT and ASL-DT correlated (r = 98, p < .0001), as Figure 6 demonstrates.

ASL-DT

Proportion Correct ( ) Proportion Correct A A A A A A A A A A A A Figure 5. Performance (scores) on ASL-DT

Pro ciency measures ASL-CT ASL-DT

(%)

Figure 6. Discrimination vs. comprehension No correlation between the proficiency measures and age of acquisition, either L1 or L2 was observed (p>.05). Instead, the variable that predicted the success on ASL-DT was the total amount of exposure to what one might tentatively

© 2019. John Benjamins Publishing Company All rights reserved Argument omission in SignL2 acquisition by deaf learners 309 label communication in the visual modality: L1, L2, and homesigning (R = .44, p < .05).9 Results of the narrative task are shown in Figure 7, where the x-axis represents individual participants (A1–12) with the age of L1 acquisition coded (L1 early or late), and the y-axis represents subject suppliance in the utterances containing finite verb contexts (‘VU’). Overall, performance in terms of overt subject suppliance was variable; however, for all but one participants, the majority of subjects were overt. In terms of overt arguments, both NPs and pronominal expressions (IX(1) and IX(non-1) in Figure 7) were robustly represented.

Subjects in VUs (%) A (L A (L A (L A (L A (L A (L A (L A (L A (L A (L A (L A (L early) early) early) early) early) early) early) late) late) late) late) late) Null NP IX( ) IX(non- ) Figure 7. Subject suppliance in the narrative in utterances containing verbs (‘VU’). ‘IX’ is commonly analyzed as a pronoun referring to the 1st (as in IX(1)) or 2nd/3rd persons (as in IX(non-1)). ‘NP’ represents overt Noun Phrases, and ‘Null’ – omissions/zero anaphor. For each participant, 30–50% of overt subjects were pragmatically inappropriate, the majority of which (69–80%) were pronominal – IX. Sample utterances are recorded in (3)–(5). The italicized bolded items in both the utterances and the translations reflect inappropriate, from the point of view of a native deaf signer, use of overt subjects; parentheses in the translation indicate omitted subjects. (3) WALK AROUND TAP-ON-SHOULDER g[COME-ON] GO DOWN LINE IX(1) CURIOUS IX(1) IX(1) THINK FINE WATER BOTTLE g [HOLD-BOTTLE] TELL IX(1) NO WATER OUT MUST IX(1) LOOK g[CONFUSED-FACE] IX(1) THINK #OK g[OPEN-BOTTLE] DRINK FINISH g[THROW-AWAY] IX(1) SEE ‘(I) was walking around, (this other person) got my attention, ‘Come on, go on down the line’. I was curious, I thought ‘Fine’ (with) water bottle.

9. See Koulidobrova & Palmer (under review) for an extended set of findings and discussion.

(Another person) tells me there is no water. I look. (I) am confused. I think it’s ok, (I) open the bottle, try drinking, but there is nothing there, so (I) throw it away. I saw this …’ (Subject A1) (4) DIE IX(3) IX(1) LOOK TOUCH RESPECT g(QUIET) FEEL DV(HAND- ON-CHEST) GOOD IX(3) HELP IX(3) PEOPLE PRESIDENT IX(3) IX(1) LOOK STILL FIRE ‘(They) died. I looked around, touched with respect. (I) feel good. They helped, and they are regular people and president. I look around and see the fire still burning [=eternal flame].’ (Subject A6) (5) WOW HISTORY IX(1) LOVE WANT MORE WOW NEVER IX(1) SEE BEFORE HERE NOW IX(1) TIME WOW GUIDE GOOD NS(P) IX(2) HELP LOOK EXPERIENCE ‘This was quite a history. I loved this and wanted more. I have never seen anything like this before. I had quite a time. ‘P’ (=name sign) is a great guide. You help see and [you have?] experience.’ (Subject A12) Of note is the fact that in (3)–(5), a number of the inappropriate (according to the native signer) overt subjects are 1st person – something that, to our knowledge, remains unattested/not discussed in the literature but appears robustly present in the SignL2 of these learners (see Figure 7). Hierarchical regression revealed that the rates of overt 1st person pronoun (IX(1)) predicted the results on ASL-DT (R = .074, p = .02). No other effects were observed, including any of the effects of age of L1 exposure: as seen in Figure 7, subjects A1–7 report being exposed to EmiratiSL before the age of 5; A8–12 are late acquirers, and one of the subjects is acquiring L1 and L2 simultaneously. The age of language acquisition did not predict the difference in overall subject suppliance (t = 0.08, df = 9.56, p = .47), or suppliance of overt pronominal expressions IX (t = 0.14, df = 8.7, p = .45) in general or the 1st person (IX(1)) in particular (t = 0.54, df = 6.37, p = .33). To summarize the results: M1L2 learners of ASL, while performing on par with beginner M2L2 learners in terms of proficiency, nevertheless behaved differently from them with respect to argument omission, at least when it comes to the subject of the clause. For ease of comparison, see Table 3. Due to the difference in the methodologies, no statistical analyses could be applied here; however, the data themselves are telling: M1L2 signers, whose L1 also allows argument omission, supply overt subjects with much greater frequency than M2L2 and L1 users of ASL. In the narrative task, our subjects utilized all three types of reference cohesion strategies. That is, they omitted arguments, as required by the grammar of ASL (see (1)); they supplied overt NPs, like PEOPLE and PRESIDENT (see (4)), and they also supplied overt 1st and non-1st person pronominal forms IX (see (2)–(3)). Further, Frederiksen, and Mayberry (2015) report that in both their M2L2 and L1

Table 3. Subject omission results: means M1L2 (current M2L2 (Frederiksen & L1 (Frederiksen & L1 (Wulf et al., 2002) study) Mayberry, 2015) Mayberry, 2015) 30–43% ~ 55% ~ 53% ~ 70%

ASL data, pronominal expressions constitute a true minority of overt arguments (less than 1%). However, for M1L2 signers in this study, this number is much higher, ranging between 10% and 65%. In other words, despite the difference in methodologies, we can be reasonably sure that M1L2 ASL learner results differ considerably from those of L1 and M2L2 ASL signers.

6. Discussion

We hesitate to make a direct comparison with the previous research on reference cohesion since the methodologies differ. However, in terms of the qualitative analysis, everything about the SignL2 performance on the narrative task appears different from the L1 and M2L2 studies: rates, distribution across types of referential expressions, and their appropriateness from the point of view of the target grammar. In brief, L1 users of EmiratiSL who have been learning ASL as L2 are behaving in parallel with Spanish-Italian bilinguals in Sorace, Serratrice, Filiaci, & Baldo (2009) – both of their languages allow argument omission but the learners are oversupplying overt arguments their L2 – in their weaker language. Let us then return to the original set of questions we set out in Sections 1–2. A robust finding in the literature on bilingual (simultaneous and sequential) acquisition of spoken languages that allow their subjects to remain null is that bilinguals perform in a non-target manner here, often oversupplying overt forms where the null ones are more acceptable. However, albeit limited, the data from bimodal bilingual language production (both child and adult) have shown the opposite. In Section 3, and summarized here as (6), we articulated predictions for two types of accounts that might capture the difference in performance between unimodal spoken and bimodal bilinguals in terms of argument oversuppliance effects. (6) a. EmiratiSL-ASL bilinguals will not oversupply arguments in ASL = modality account b. EmiratiSL-ASL bilinguals will oversupply arguments in ASL = processing account

That is, if the reason for the apparent uniqueness of bimodal behavior lies in the sign modality itself, then the SignL2 users of ASL in this study should have behaved on par with other populations whose other language is ASL: the rates of argument omission ought to have been high, matching the previously reported numbers for L1 and M2L2 learners. This, however, is not what the data reveal: instead, SignL2 learners of ASL, whose SignL1 – EmiratiSL – matches the SignL2 in the relevant morpho-syntactic property, exhibit a lower rate of argument omission. We interpret this finding as suggesting that the account a la Sorace (2016) carries a better explanatory value than the alternative: because in the case of SignL2, one of the languages must be inhibited (as is the case in a unimodal spoken language combination but not the bimodal one), this creates more cognitive pressure and learners resort to overt forms for reference tracking. This conclusion directly implies that overt forms being observed in the languages of unimodal speech bilinguals may have nothing to do with L1-L2 interaction (or lack thereof); rather, it is a bilingual phenomenon, masking the potential language interaction in the domain of argument omission. This is precisely the path pursued in Kouli- dobrova (2017a) in explaining the increased rates of null arguments in the speech of bimodal bilinguals (see Figures 1). For this study, then, we can now say that the oversuppliance of IX and overt NPs in the SignL2 learners’ ASL may have nothing to do with cross-linguistic interaction either. Thus, while the distribution of IX in EmiratiSL should be examined independently, we cannot simply attribute its presence in the L2 ASL of the Emirati signers to transfer.10 There is another, albeit indirect, piece of evidence in favor of the view that the oversuppliance rates/patterns are not (cross-)linguistically-based. We know that age of acquisition of L1 affects linguistic competence and performance (Sec- tion 2): late L1 learners tend to perform differently in terms of bona fide linguistic tasks when compared to the native and early learners. However, no effect of the age of L1 acquisition was observed in this study. Certainly, numbers are small here, but overall, the Early Learner group (subjects A1–7) omitted more subjects, which is consistent with the grammar of ASL, than the Later Learner group; even though, the effect was not significant. Perhaps with a larger set we might see more here. No effect of L2 acquisition (such as age at arrival or length of study) was observed either.

10. The question of whether unimodal speech bilinguals oversupply 1st person pronouns remains, however.

7. Limitations and directions

The claims advanced here remain preliminary, and the data represent only the first step in the examination of the linguistic processes by SignL2 learners; yet, the results of the study shed light on a number of important issues which deserve attention. First, the fact that the results of the two proficiency measures strongly correlated offers some evidence that the two instruments, which, to our knowledge, were not normed on SignL2 population, actually appear to serve as valid measures of ASL as SignL2, at least with this small sample. Second, with respect to the phenomenon under examination here, for example, what has not yet been articulated is how the SignM1L2 learners compare to their speech M1L2 and M2L2 counter- parts in terms of the use of referential expressions as a cohesion tool in various discourse functions (introduced, maintained, re-introduced). More generally, the participant pool here was small with a fairly high degree of variability; it remains to be seen whether the generalization that has arisen here survives a larger N-size. Additionally, to date, very little is known about SignM1L2 processes in terms of either language interaction or extra-linguistic factors (Chen Pichler & Koulido- brova, 2015). To our knowledge, no research has been done on the morpho-syntax of EmiratiSL either. There are also, of course, other potential variables at play, such as language disabilities outside of audition that are typically not identified for the vast majority of the deaf in the world, including the participants of this study. The latter, for instance, adds an additional prediction stemming from the account advocated here: individuals with executive control-related atypicalities may be expected to present higher rates of argument suppliance. A direct finding in this study that also deserves careful examination with a larger sample is that while the participants’ scores varied widely, their performance was predicted by the amount practice with the visual modality – exposure to homesigning family members, for example. In other words, relative success on the proficiency measure in SignL2 depended not on the age of acquisition of this L2 (as is typically found in the L2 literature) or the age of acquisition of the L1 (as is typically found in the SignL1 literature) but the amount of practice in the visuo- gestural modality. This finding has thus far been unattested, to our knowledge, and can arguably be only be explored in multilingual signing populations. Finally, note that the participants in this study are also exposed to English in print, being explicitly taught it as part of the academic content. What the effects of the L3 acquisition here are, and what influences the development of foreign language learning for the deaf and hard of hearing (SignL2) vs. mastering literacy in representing the spoken language of the country where that SignL2 is used remains to be determined. These and other potential questions we leave for future research.

Acknowledgment

Warm thanks to students and faculty at the American School for the Deaf (particularly Patricia Pangborn and Dennis Palka), the Research Assistants Emily Crawford and Kurt Gagne and co-PI Jeff Palmer. Financial support is provided by the Connecticut State University AAUP Research Grants ##ARKOUG & ARKOUH.

References

Asudeh, A. (2004). Resumption as resource management (Unpublished PhD dissertation). Stanford University. Bahan, B., Kegl, J., Lee, R., MacLaughlin, D., & Neidle, C. (2000). The licensing of null arguments in ASL. Linguistic Inquiry, 31(1), 1–27. https://doi.org/10.1162/002438900554271 Bank, R., Crasborn, O., & Van Hout, R. (2018). Bimodal code-mixing: Dutch spoken language elements in NGT discourse. Bilingualism: Language and Cognition, 21(1), 104–120. https://doi.org/10.1017/S1366728916000936 Bell, A., Ortells, M., & Morgan, G. (2015). Reference control in the narratives of adult sign language learners. International Journal of Bilingualism, 19(5), 608–624. https://doi.org/10.1177/1367006914527186 Belletti, A., Bennati, E., & Sorace, A. (2007). Theoretical and developmental issues in the syntax of subjects: Evidence from near-native Italian. Natural Language and Linguistic Theory, 25, 657–689. Blais, M. J., Oshima-Takane, Y., Genesee, F., & Hirakawa, M. (2010). Crosslinguistic influence on argument realization in Japanese-French bilinguals. In K. Franich, K. M. Iserman, & L. L. Keil (Eds.), Proceedings of the 34th annual Boston University Conference on Language Development (pp. 34–45). Somerville, MA: Cascadilla Press. Bochner, J. H., Christie, K., Hauser, P. C., & Searls, J. M. (2011). When is a difference really different? Learners’ discrimination of linguistic contrasts in American Sign Language.Lan- guage Learning, 61(4), 1302–1327. https://doi.org/10.1111/j.1467-9922.2011.00671.x Boudreault, P., & Mayberry, R. I. (2006). Grammatical processing in American Sign Language: Age of first-language acquisition effects in relation to syntactic structure. Language and Cognitive Processes, 21, 608–635. https://doi.org/10.1080/01690960500139363 Carrigan, E., & Coppola, M. (2012). Mothers do not drive structure in adult homesign systems: Evidence from comprehension. In N. Miyake, D. Peebles, & R. P. Cooper (Eds.), Proceed- ings of the 34th annual Conference of the Cognitive Science Society (pp. 1398–1403). Austin, TX: Cognitive Science Society. Carroll, K. S., Al Kahwaji, B., & Litz, D. (2017). Triglossia and promoting Arabic literacy in the United Arab Emirates. Language, Culture and Curriculum, 30(3), 317–332. https://doi.org/10.1080/07908318.2017.1326496 Chamorro, G., Sorace, A., & Sturt, P. (2015). What is the source of L1 attrition? The effects of recent re-exposure on Spanish speakers under L1 attrition. Bilingualism: Language and Cognition, 19(3), 520–532. https://doi.org/10.1017/S1366728915000152

Chen Pichler, D., Hochgesang, J., Lillo-Martin, D., & Müller de Quadros, R. (2010). Conventions for sign and speech transcription in child bimodal bilingual corpora. Language, Interaction and Acquisition/Language, Interaction et Acquisition, 1(1), 11–40. https://doi.org/10.1075/lia.1.1.03che Chen Pichler, D., & Koulidobrova, E. (2015). Acquisition of sign language as a second language (L2). In M. Marchark, & P. Spencer (Eds.), The Oxford handbook of Deaf studies, language, and education (pp. 118–230). Oxford: OUP. Chen Pichler, D. (2011). Sources of handshape error in first-time signers of ASL. In G. Mathur, & D. J. Napoli (Eds.), Deaf around the world: The impact of language (pp. 96–121). Oxford: OUP. Dostal, H., & Wolbers, K. (2016). Examining student writing proficiencies across genres: Results of an intervention study. Deafness & Education International, 18(3), 159–169. https://doi.org/10.1080/14643154.2016.1230415 Emmorey, K., Borinstein, H. B., Thompson, R., & Gollan, T. H. (2008a). Bimodal bilingualism. Bilingualism: Language and Cognition, 11(1), 43–61. https://doi.org/10.1017/S1366728907003203 Emmorey, K., Luk, G., Pyers, J. E., & Bialystok, E. (2008b). The source of enhanced cognitive control in bilinguals. Psychological Science, 19(12), 1201–1206. https://doi.org/10.1111/j.1467-9280.2008.02224.x Frederiksen, A. T., & Mayberry, R. I. (2015). Tracking reference in space: How La2 learners use ASL referring expressions. In E. Grillo & K. Jepson (Eds.), Proceedings of the 39th annual Boston University Conference on Child Language Development (pp. 165–177). Somerville, MA: Cascadilla Press. Futrell, R., Hickey, T., Lee, A., Lim, E., Luchkina, E., & Gibson, E. (2015). Cross-linguistic ges- tures reflect typological universals: A subject-initial, verb-final bias in speakers of diverse languages. Cognition, 126, 215–221. https://doi.org/10.1016/j.cognition.2014.11.022 Hacohen, A., & Schaeffer, J. (2007). Subject realization in early Hebrew/English bilingual acquisition: The role of crosslinguistic influence. Bilingualism: Language and Cognition 10(3), 333–344. https://doi.org/10.1017/S1366728907003100 Hauser, P. C., Paludneviciene, R., Kurz, K. B., Emmorey, K., & Contreras, J. (2016.). American Sign Language Comprehension Test: A tool for sign language researchers. Journal of Deaf Studies and Deaf Education, 21(1), 64–69. https://doi.org/10.1093/deafed/env051 Haznedar, B. (2010). Transfer at the syntax-pragmatics interface: Pronominal subjects in bilingual Turkish. Second Language Research 26(3), 355–378. https://doi.org/10.1177/0267658310365780 Hendriks, B. (2008). Jordanian Sign Language: Aspects of grammar from a cross-linguistic perspective (PhD dissertation). Utrecht: LOT Dissertations in Linguistics. Huang, P.-Y. (1999). The development of null arguments in a Cantonese-English bilingual child (Unpublished MA thesis). Chinese University of Hong Kong. Hulk, A., & Müller, N. (2000). Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism: Language and Cognition, 3(3), 227–244. https://doi.org/10.1017/S1366728900000353 Johnson, J., Watkins, R., & Rice, M. L. (1992). Bimodal bilingual language development in a hearing child of deaf parents. Applied Psycholinguistics, 13(1), 31–52. https://doi.org/10.1017/S0142716400005415

Juan-Garau, M., & Perez-Vidal, C. (2000). Subject realization in the syntactic development of a bilingual child. Bilingualism: Language and Cognition, 3(3), 173–191. https://doi.org/10.1017/S1366728900000328 Karchmer M., & Mitchell, R. (2003). Demographic and achievement characteristics of deaf and hard-of-hearing students. In M. Marchark & P. Spencer (Eds.), Oxford handbook of deaf studies, language, and education (pp. 21–37). Oxford: OUP. Koulidobrova, E. (2012). When the quiet surfaces: ‘Transfer’ of argument omission in the speech of ASL-English bilinguals (Unpublished PhD dissertation). University of Connecticut. Koulidobrova, E. (2013a). Influence uninhibited: Null subjects in the speech of ASL-English bilinguals. In S. Stavrakaki, M. Lalioti, & P. Konstantinopolou (Eds.), Advances in language acquisition (pp. 346–355). Newcastle upon Tyne: Cambridge Scholars. Koulidobrova, E. (2013b). She said __ was ok: Embedded subject omission in the English of ASL-English bilinguals. Paper presented at the LSA 2013 (Winter). Boston, MA, January 2013. Koulidobrova, E. (2014). Null arguments in bimodal bilingualism: Code-blending (and the lack of) effects in American Sign Language. In W. Orman & M. J. Valleau (Eds.), Proceedings of the 38th annual Boston University Conference on Language Development (pp. 253–265), Somerville, MA: Cascadilla Press. Koulidobrova, E. (2017a). Argument omission in ASL-English bilinguals: Language interaction effects in bimodal bilingualism. Linguistic Approaches to Bilingualism, 7(5), 583–613. https://doi.org/10.1075/lab.13047.kou Koulidobrova, E. (2017b). Elide me bare: Null arguments in ASL. Natural Language and Linguis- tic Theory, 35, 397–446. https://doi.org/10.1007/s11049-016-9349-5 Koulidobrova, E., & Lillo-Martin, D. (2016). Point of inquiry: The case of the (non) pronominal IX in ASL. In P. Grodsz & P. Patel (Eds.), Impact of pronominal form on interpretation (pp. 221–250). Berlin: Mouton de Gruyter. Koulidobrova, E., & Palmer, J. (2016). Phonological discrimination in Sign L2: Within, vs. across, modality effects. Theoretical Issues in Sign Linguistics (TISLR), 12, Melbourne, January 2016. Kroll, J. F., Bobb, S. C., & Wodniecka. Z. (2006). Language selectivity is the exception, not the rule: Arguments against a fixed locus of language selection in bilingual speech. Bilingual- ism: Language and Cognition, 9(2), 119–135. https://doi.org/10.1017/S1366728906002483 Lillo-Martin, D. (1991). Universal Grammar and American Sign Language: Setting the null argument parameters. Dordrecht: Kluwer. https://doi.org/10.1007/978-94-011-3468-2 Lillo-Martin, D. (1995). The Point of view predicate in American Sign Language. In K. Emmo- rey & J. Reilly (Eds.), Language, gesture, and space (pp. 155–170). Hillsdale, NJ: Lawrence Erlbaum Associates. Lillo-Martin, D., Quadros, R., Koulidobrova, E., & Chen Pichler, D. (2009). Bimodal bilingual cross-language influence in unexpected domains. In J. Costa, A. Castro, M. Lobo, & F. Pratas (Eds.), Language acquisition and development: Proceedings of Generative Approaches to Language Acquisition 2009 (pp. 269–275). Newcastle upon Tyne: Cambridge Scholars. Mayberry, R. I. (2007). When timing is everything: Age of first-language acquisition effects on second-language learning. Applied Psycholinguistics, 28(3), 537–549. https://doi.org/10.1017/S0142716407070294 Mayberry R. I., & Lock E. (2003). Age constraints on first versus second language acquisition: Evidence for linguistic plasticity and epigenesis. Brain and Language, 87(3), 369–384. https://doi.org/10.1016/S0093-934X(03)00137-8

Meier, R. (1990). Person deixis in American Sign Language. In S. Fischer, & P. Siple (Eds.), Theo- retical issues in sign language research, 1: Linguistics (pp. 175–190). Chicago, IL: University of Chicago Press. Mishina-Mori, S. (2007). Argument representation in Japanese/English simultaneous bilinguals: Is there a crosslingusitic influence? In H. Count-Nulton, S. Kulatiklake, & I.-H. Woo (Eds.), Proceedings of the 31st annual Boston University Conference on Language Development (pp. 441–450). Somerville, MA: Cascadilla Press. Murphy, J., & Slorach, N. (1983). The language development of pre-preschool hearing children of deaf parents. British Journal of Disorders of Communication, 18(2), 118–126. https://doi.org/10.3109/13682828309019829 Ortega, G., & Morgan, G. (2010). Comparing child and adult development of a visual phonological system. Language, Interaction and Acquisition, 1, 67–81. https://doi.org/10.1075/lia.1.1.05ort Ortega-Delgado, G. (2013). Acquisition of a signed phonological system by hearing adults: The role of sign structure and iconicity (Unpublished PhD dissertation). University College London. Petroj, V., Guerrera, K., & Davidson, K. (2014). ASL-dominant code-blending in the whispering of bimodal bilingual children. In W. Orman & M. J. Valleau (Eds.), Proceedings of 38th annual Boston University Conference on Language Development (pp. 319–330). Somerville, MA: Cascadilla Press. Pinto, M. (2006). Subject acquisition in bilinguals: Interference or maturation. In L. Escobar & V. Torrens (Eds.), The acquisition of syntax in Romance Languages (Language Acquisition and Language Disorders 41) (pp. 331–352). Amsterdam: John Benjamins. https://doi.org/10.1075/lald.41.16pin Reynolds, W. (2016). Early bimodal bilingual development of ASL narrative referent cohesion: Using a heritage language framework. PhD dissertation, Gallaudet University. Sachs, J., Bard, B., & Johnson, M. (1981). Language learning with restricted input: Case studies of two hearing children of deaf parents. Applied Psycholinguistics, 2(1), 33–54. https://doi.org/10.1017/S0142716400000643 Sandler, W., & Lillo-Martin, D. (2006). Sign Languages and Linguistic Universals. Cambridge: CUP. https://doi.org/10.1017/CBO9781139163910 Schiff, N., & Ventry, I. (1976). Communication problems in hearing children of deaf parents. Journal of Speech and Hearing Disorders, 41(3), 348–358. https://doi.org/10.1044/jshd.4103.348 Schiff-Myers, N. (1988). Hearing children of deaf parents. In D. Bishop, & K. Mogford (Eds.), Language Development in Exceptional Circumstances (pp. 47–61). Edinburgh: Churchill Livingstone. Schmitz, K., Patuto, M., & Müller, N. (2012). The null-subject parameter at the interface between syntax and pragmatics: Evidence from bilingual German-Italian, German-French and Italian-French children. First Language, 32(1–2), 205–238. https://doi.org/10.1177/0142723711403880 Seal, B., & Hammett, L. (1995). Language intervention with a hearing child whose parents are deaf. American Journal of Speech-Language Pathology, 4(4), 15–21. https://doi.org/10.1044/1058-0360.0404.15 Serratrice, L., Sorace, A., Filiaci, F. & Baldo, M. (2011). Pronominal objects in English-Italian and Spanish-Italian bilingual children. Applied Psycholinguistics, 33(4), 1–27.

Serratrice, L., Sorace, A., & Paoli, S. (2004). Crosslinguistic influence at the syntax-pragmatics interface: Subjects and objects in English-Italian bilingual and monolingual acquisition. Bilingualism: Language and Cognition, 7(3),183–205. https://doi.org/10.1017/S1366728904001610 Sorace, A. (2003). Near-nativeness. In M. Long & C. Doughty (Eds.), Handbook of second language acquisition (pp. 130–152). Oxford: Blackwell. https://doi.org/10.1002/9780470756492.ch6 Sorace, A. (2005). Selective optionality in language development. In L. Cornips & K. P. Corrigan (Eds.), Syntax and variation. Reconciling the biological and the social (pp. 55–80). Amster- dam: John Benjamins. https://doi.org/10.1075/cilt.265.04sor Sorace, A. (2006a). Possible manifestations of shallow processing in advanced second language speakers. Applied Psycholinguistics 27, 88–91. Sorace, A. (2006b). Gradience and optionality in mature and developing grammars. In G. Fanselow, C. Fery, M. Schlesewsky, & R. Vogel (Eds.), Gradience in grammars: Generative perspectives. Oxford: OUP. Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguistic Approaches to Bilingualism, 1(1), 1–33. https://doi.org/10.1075/lab.1.1.01sor Sorace, A. (2016). Referring expressions and executive functions in bilingualism. Linguistic Ap- proaches to Bilingualism, 6(5), 669–684. https://doi.org/10.1075/lab.15055.sor Sorace, A., & Filiaci, F. (2006). Anaphora resolution in near-native speakers of Italian. Second Language Research, 22, 339–368. https://doi.org/10.1191/0267658306sr271oa Sorace, A., Serratrice, L. Filiaci, F., & Baldo, M. (2009). Discourse conditions on subject pronoun realization: Testing the linguistic intuitions of older bilingual children. Lingua, 119, 460–477. https://doi.org/10.1016/j.lingua.2008.09.008 Todd, P. (1971). A case of structural interference across sensory modalities in second-language learning. Word, 27(1–3), 102–118. https://doi.org/10.1080/00437956.1971.11435616 Tsimpli, I. (2011). External interfaces and the notion of ‘default.’ Linguistic Approaches to Bilin- gualism, 1(1), 101–103. https://doi.org/10.1075/lab.1.1.17tsi Tsimpli, I., & Sorace, A. (2006). Differentiating interfaces: L2 performance in syntax-semantics and syntax-discourse phenomena. In D. Bamman, T. Magnitskaia, & C. Zaller (Eds.), Proceedings of the 30th annual Boston University Conference on Language Development (pp. 653–664). Somerville, MA: Cascadilla Press. Tsimpli, I., Sorace, A., Heycock, C. & Filiaci, F. (2004). First language attrition and syntactic subjects: a study of Greek and Italian near-native speakers of English. International Journal of Bilingualism, 8, 257–277. https://doi.org/10.1177/13670069040080030601 Wulf, A., Dudis, P., Bayley, R., & Lucas, C. (2002). Variable subject presence in ASL narratives. Sign Language Studies, 3(1), 54–76. https://doi.org/10.1353/sls.2002.0027 Zwanziger, E. E., Allen, S. E. M., & Genesee, F. (2005). Cross-linguistic influence in bilingual acquisition: Subject omission in learners of Inuktitut and English. Journal of Child Language, 32(4), 893–909. https://doi.org/10.1017/S0305000905007129