2018 Friday Poster 6427

Friday, November 2, 2018 | Poster Session I, Metcalf Small | 3pm

Low Proficiency Does Not Mean Ab Initio: Exposure Matters for L3 Transfer Studies
E. Puig-Mayenco, J. Rothman

In the present study, we contribute to better understanding the role actual exposure plays— above and beyond beginner proficiency—for testing competing theories of morphosyntactic transfer in L3/Ln acquisition at so-called initial stages (see e.g., Rothman, Iverson and Judy, 2011; Rothman and Halloran, 2013, González Alonso, et al. 2017, for relevant models). We present data from a Picture-Sentence Matching (PSM) task (see examples on 2nd page) tapping into the semantic interpretation of L3 English Negative Polarity Items (NPIs) ‘anything/anybody’ and Negative Quantifiers (NQs) ‘nothing/nobody’ by Catalan-Spanish bilinguals. Catalan and Spanish only have one corresponding lexical item, labeled a Negative Concord Item (NCI), covering the English NPIs and NQs (nada ‘Sp’, res ‘Cat’: anything/nothing; Nadie ‘Sp’, Ningú ‘Cat’: anybody/nobody). Data from 4 conditions are presented involving either the NPI or NQ in conditional sentences in object position of a conditional structure (1a-b); or in pre-verbal position with the co-occurrence of sentential negation (2a-b). Sentences in (3a-b) and (4a-b) show the respective counterparts in Spanish and Catalan. As shown in table 1, the interpretation of these lexical items in the three languages differ, allowing us to explore morphosyntactic transfer selection.

We tested 60 Catalan-Spanish bilinguals—35 L1-Catalan-L2-Spanish and 25 L1-Spanish- L2-Catalan learners—after an 8-week program of English that we designed. The course was offered free of charge in Osona, a rural community in central Catalonia where bilinguals with L1 Spanish and L1 Catalan (the other language being the L2) could be found with extremely high proficiency in both languages and no previous exposure to English was in principle possible. Designing our own curriculum meant we could attract enough participants—while providing them a service in kind—and control the type and amount of L3 English input each learner would receive in the case of true ab initio learners—the group we are (and theories of transfer should be) most interested in for the purpose of determining transfer selectivity.

The statistical analysis consisted of a generalized linear mixed effects logistic regression with, among others, amount of exposure as a continuous variable. The results favor Catalan as a transfer source, yet show some variability for one condition (1b) where Spanish also seems to exercise some influence (Figure 1). The model, however, showed only one, yet crucial significant predictor accounting for the variability: amount of exposure to the L3 (p .< 001). Based on the results of the analysis, we ran a secondary analysis in which we differentiated between true ab initio versus low proficiency beginner learners with significant exposure to the target L3. The analysis shows that the results are more uniform when isolating ab intio learners, Catalan-like interpretations are consistent across all four conditions (Figure 2).

We provide empirical evidence to back of the claim that actual exposure matters beyond proficiency for L3 initial stages transfer testing (González Alonso & Rothman, 2017); that is, even when proficiency is held constant at very low levels (< A1 level of the EU framework), low proficiency L3 learners who have had significant exposure to an L3 in the past pattern differently from truly ab initio L3-learners. And so, what seems to be hybridity in transfer selection might be—as was the case in this snapshot—reflective of previous language exposure obscuring initial transfer selection. We discuss how this reality complicates isolating L3-transfer from effects of L3-development/acquisition and its multiple implications.