Session 2F

L1 transfer is a hallmark of bilingualism. Studies describing how L1 transfer affects late pubertal L2 grammatical systems abound in SLA research. This research has found L1 transfer to be linked to 3 key phenomena, divergence between L1 and L2 representations, interlanguages, and L2 fossilization (Vainikka & Young-Scholten 2011; Schwartz & Sprouse in Press). In contrast, far less is known of the role of L1 transfer when the two grammars come into contact earlier in life, otherwise dominant language transfer in heritage language grammars (Polinsky, 2018). Unlike adult L2 speakers, the dominant language of the heritage speaker potentially affects the weaker language already from childhood. The best evidence to date of dominant language transfer comes from studies focusing on L2 syntax exploiting language production tasks (Fenyvesi 2005; Cuza & Frank 2015; Cuza & Strik 2012). These studies, however, employed elicited imitation which has been criticized for failing to evaluate capacity to produce (rather than imitate) language (Vinther 2002) and bias monolingual populations when comparted to heritage speakers (Polinsky 2018). Another method which has been shown to tap into abstract representations fairly validly is structural priming (Jackson, 2018 for a state-of-the-art). Intrinsically linked to the use of priming and L1 transfer is the Basic Continuity Hypothesis (BCH) (Romano 2018) which maintains the sentence production mechanisms of monolingual (L1) and highly proficient L2Ss are similar enough for the latter to be able to integrate semantic and syntactic information in native-like manner despite any associated structures being absent in the L2Ss' L1 grammar. Thus, in the study we present, we extend the BCH to heritage speakers. The questions we address are as follows:
RQ1: To what degree does transfer affect L2 and heritage grammars?
RQ2: How similar are the language production mechanisms in L2 and HL to monolingual speakers, particularly when a structure requiring integration of semantic/syntactic information is absent from the L1/dominant language of the bilingual?

To address these questions, an oral structural priming task was employed to compare advanced Swedish speakers of Italian (n = 13), proficiency-matched adult heritage Italian speakers (n = 14) dominant in Swedish, and Italian monolinguals (n = 18). The production of 4 clitic structures requiring coordination of syntactic/semantic information, namely proclisis with lexical, modal, and causative verbs and si-passives which are only possible in Italian (1-4), in comparison to a structure shared by both Italian and Swedish, namely transitives (5) was measured. If L1 transfer (RQ1) occurs, it was predicted that the transitive structures would lead to higher priming rates than clitic structures in the L2 and HL groups as the bilinguals are eased by an overlap in L1/L2 transitive structures. Moreover, if the BCH applies to both L2S and heritage speakers(RQ2), priming rates will be significantly high (above 60%) and comparable to native speakers. Results showed that bilinguals are not primed any more on transitives than some clitic structures, rejecting transfer. Moreover, they are primed higher than 60% on all but clitics + modals structures with most contrasts by structure not being statistically significantly different from monolinguals, supporting the BCH. Finally, an analysis of divergent structures produced shows L2/HL passive structures to be remarkably compatible with those produced by monolinguals at earlier developmental stages reported in previous research (see L1 versus Dominant Language Transfer in L2 and Heritage Swedish Speakers of Italian.pdf)




References
Cuza, A, and N. Strik. 2012. 'Patterns of morphosyntactic convergence and child L1 attrition: Evidence from subject–verb inversion in Spanish–English bilingual children.' in Paper presented at the Linguistic Symposium on Romance Languages 42. Southern Utah University. 
Cuza, A., and J. Frank. 2015. 'On the role of experience and age–related effects: Evidence from the Spanish CP'. Second Language Research 31/1: 3–28. https://doi.org/10.1177/0267658314532939
Fenyvesi, A. 2005. 'Hungarian in the United States' in A. Fenyvesi (ed.): Hungarian Language Contact outside Hungary: Studies on Hungarian as a Minority Language. John Benjamins, pp. 265–318
Polinsky, M. 2018. Heritage Languages and Their Speakers. Cambridge University Press.
Romano, F. 2018. The Basis Continuity Hypothesis of L1 to L2 production. Second Language Research 3: 275-308. https://doi/abs/10.1177/0267658317729423. 
Schwartz, B. D., and R. A. Sprouse (in press). 'The role of Universal Grammar in nonnative language acquisition.' in I. Roberts (ed.): The Oxford handbook of Universal Grammar. Oxford University Press.
Vainikka, A., and M. Young-Scholten. 2011. The acquisition of German: introducing Organic Grammar. De Gruyter Mouton.

Background. Previous studies show that even though monolingual children find object relatives easier than subject relatives, their comprehension of object relatives can be facilitated by morphological cues, such as gender (Guasti et al., 2012). Given that in heritage contexts functional morphology is a vulnerable domain, a question that should be addressed is whether bilingual children, who are heritage speakers of their L1, will also be able to use morphological cues to comprehend complex syntax (Chondrogianni & Schwartz, 2020). To contribute to this line of research, we compared bilingual, first generation, Syrian Arabic-speaking children in Canada (N=108; Mean Age: 12;3; Mean AoA: 7;6) with monolingual, Syrian Arabic-speaking children in Syria (N=18; Mean Age: 11;6) with respect to their comprehension of Syrian Arabic relative clauses. Furthermore, we examined whether bilingual children's comprehension of relative clauses would be affected by their knowledge of gender while considering their Age of Acquisition of English (AoA).
Research Questions. We asked: A) How do bilingual, Syrian Arabic-speaking children in Canada compare with monolingual children in Syria in their comprehension of relative clauses?; B) Does AoA and knowledge of gender morphology affect bilingual children's comprehension of relative clauses in Syrian Arabic?
Methods. We used two offline sentence-picture matching tasks targeting relative clauses and gender (in object clitics). The task targeting gender included items with feminine and masculine object clitics (e.g., el–ʔarnab dˤam–o 'The rabbit hugged him'). The task targeting relative clauses included two conditions: Matching relatives (both Subject and Object matched in gender (1)); Mismatching relatives (Subject and Object differed in gender (2)).
el–ʔarnab yali yi–dˤum–o  el–ʔrd (2) el–ʔarnab yali t–dˤum–o  el–ʔitˤa 
the-rabbit.M that 3SG.M–hug-him the-monkey.M the-rabbit.M that 3SG.F–hug-him the-cat.F
'The rabbit that the monkey hugs.' 'The rabbit that the cat hugs.'
Predictions. We predicted that if gender morphology on clitics has a facilitative effect, then bilingual children, like monolingual children, will have a higher performance in mismatching relatives (where gender cues on clitics disambiguate reference) (A); and that there will be an association between children's knowledge of clitic morphology (gender) and children's comprehension of relatives (B).
Analysis and Results. We ran two mixed-effects logistic regression analyses. With respect to A, results revealed that monolingual children were more accurate than bilingual children, and that both groups were more accurate in subject relatives and in mismatching relatives. For question B, we focused on the comprehension of object relatives by the bilingual group. Results revealed that children's knowledge of gender morphology (on clitics) was a positive predictor of children's knowledge of relative clauses, even when controlling for AoA. 
Conclusion. Overall, our results support the conclusion that, like monolingual children, first generation, Arabic-speaking children living in Canada can use morphological cues to comprehend complex syntax in their L1. Furthermore, their comprehension of relative clauses is contingent on their comprehension of gender morphology on clitics.

In Japanese, the particles WA and GA are categorized into different usages according to Hasegawa (2014). WA has three different usages (Contrastive, Thematic, and Negative-scope marker) and GA into three (Object, Subordinate, and Indefinite). Previous second language acquisition studies have found that it is not necessarily the WA and GA that cause learners difficulties, but the difficulties are function-specific rather than particle-specific (Oshiro, 2021). However, to our knowledge, no-one has investigated the processing of particles in real time by bilinguals. 
The aim of this ongoing study is to compare heritage speakers and native Japanese in the acquisition of Japanese particles WA and GA. A forced-choice elicitation task (FCT) and a self-paced reading task (SPR) was conducted on the Japanese particles WA and GA. The FCT and SPR has thirty-six test items with WA and GA, and thirty-two fillers. Each test item was presented in a short independent conversational dialogue between two people. There were 10 participants, 5 heritage speakers and the 5 native speakers of Japanese. All heritage speakers (HS) speak their respective heritage languages with their families at home. For example, one participant is a HS of Spanish and Portuguese as her father is Brazilian and mother comes from Peru. They speak Spanish and Portuguese at home and Japanese outside of the home. 
We predict that given the HSs are early acquirers of Japanese they will likely have developed a high level of linguistic competence in Japanese, but less Japanese input in the home environment (Unsworth, 2016) may affect the degree of comprehension and real-time processing.
Figures 1-6 show the difference between heritage speakers and native Japanese speakers on the FCT. For the SPR task, the critical region for a Japanese particle is R2, and the regions before and after it are R1 (pre-critical region) to R4 (pre-wrap-up region) in the Figures 7-12. The grammatical RTs are subtracted from the ungrammatical RTs. A positive RT means that it took more time to read ungrammatical sentence for participants.
To find out if there were any particle-specific differences between different types of WA and GA on the SPR task, preliminary analyses were conducted using mixed-effects models with crossed random effects for subjects using the lme4 package. The analysis included four models, one for each condition: fixed effects for grammaticality of WA and GA; fixed effects for ungrammaticality of WA and GA. This included random intercepts for subjects - random slopes for groups, and subjects and their interaction on overall task accuracy. No significant differences were found between the HSs and the Japanese native speakers. 
References
Hasegawa, Y. (2014). Japanese: A linguistic introduction. Cambridge: Cambridge University Press.
Oshiro, D. (2021). The second language acquisition of Japanese particles WA and GA. Unpublished MA dissertation. Ball State University. 
Unsworth, S. (2016). Quantity and quality of language input in bilingual language development. In E. Nicoladis & S. Montanari (Eds.), Lifespan perspectives on bilingualism (pp. 136-196). Berlin: De Gruyter.

A growing interest in exploring converging models that systematically triangulate experimental evidence and naturalistic corpus data is making significant contributions to long-standing acquisition debates in SLA (Granger, 2021; Lozano, 2021; Lozano & Mendikoetxea, 2018). A grammatical area at the core of the discrepancies is variability in the acquisitional outcomes of L1English-L2Spanish accusative pronominalization with third-person clitics–le/lo(s), la(s)–. Clitics are a highly vulnerable domain for bilinguals as they require integration of discursive, morphosyntactic and lexical information at once. Non-economical full DP overuse, inflectional deficits, and clitic omission have been attested in varying degrees across task modalities and proficiency levels in experimental research (Malovrh & Lee, 2013; McCarthy, 2008). 
However, the discursive nature of clitics has been widely overlooked and very few studies have adopted a methodologically triangulated approach to L2 Spanish object anaphora (Sánchez & Al-Kasey, 1999; Zyzik, 2008). Crucially, difficulties with the gender feature in ultimate attainment and recently uncovered factors (i.e., the animacy feature and the salience of referents) have been identified as potential triggers for acquisitional variability (Shimanskaya & Slabakova, 2017; Zyzik, 2008). In this study, we explore the distribution of L1 English-L2 Spanish accusative clitics and full DPs in terms of persistent morphosyntactic, but also discursive challenges in natural and controlled production. Particularly, we investigate a) the mechanisms used by learners as anaphoric forms (e.g., clitics, full DPs, omission) in contexts where clitics are expected in natives, b) the factors that determine the use of such forms –i.e., gender and animacy features (Lardiere, 2009), but also discursive factors (e.g., antecedent distance and ambiguity)– and finally, c) whether such factors converge in controlled and natural production.
A multifactorial analysis of four written learner subcorpora (30 upper-advanced, 30 lower-advanced, 30 upper-intermediate, 30 lower-intermediate) and a native subcorpus extracted from CEDEL2 (Corpus Escrito del Español L2, Lozano, 2021) was carried out with the software UAMCorpusTool (O'Donnell, 2009). 773 accusative objects were submitted to a fine-grained linguistically motivated tagset and statistical analyses (χ2). A further methodological triangulation with preliminary experimental data from a controlled elicited production question & answer oral task (Arche & Domínguez, 2011) testing clitic accuracy (15 L1 Spanish, 10 L2 Advanced, and 10 Intermediate learners) exhibited convergence in the distribution of accusative anaphora with a significantly higher rate of full DPs in learners (Figures 1 and 2). Natives use full DPs for disambiguation or distant antecedents, whereas learners use them non-economically (Figure 3). Feminine and inanimate antecedents trigger full DP overuse in corpora (see Figure 4), reduced clitic use in controlled production (Figure 5), and non-target gender agreement. In sum, results exhibit the well-documented effect of gender, but also a significant effect for animacy (Olsen & Juffs, 2020) in both natural and controlled production. This suggests (1) a developmental pattern in L2 Spanish clitic feature reassembly with avoidance as a strategy to supply for feature underspecification (Lardiere, 2009) and (2) methodological triangulation as a useful model to reveal undetected factors in L2 discourse and morphosyntax.