2018 Sat Poster 6562

Saturday, November 3, 2018 | Poster Session II, Metcalf Small | 3:15pm

Initial phonological and morphosyntactic transfer in L3 Italian by early English/Spanish bilinguals
J. Cabrelli Amaro, C. Pichan, J. Rothman, L. Serratrice

Work on third language (L3) morphosyntax in recent years has focused on the dynamics of transfer source selection (L1, L2, or both). Available models cover the gamut of logical possibilities, ranging from L1 or L2 default transfer (Hermas, 2010; Bardel & Falk, 2007), to a cumulative effect of all previous systems (Flynn et al., 2004), to a deterministic role of underlying typological structural similarity between the target L3 and the L1/L2 (Rothman 2010, 2015; Westergaard et al. 2017). Virtually all available data examine adult L3 acquisition in late L2 learners. Thus, it is not clear if/how L3 models cover other types of bilinguals or extend to other modules, e.g. phonology. To address these issues simultaneously, we examine heritage speaker (HS) bilinguals (Spanish/English) acquiring an L3 (Italian) in adulthood and examine two modules of grammar, syntax and phonology, soon after initial L3 exposure.

There is some evidence that adult and child HSs follow a similar path of L3 transfer to adult L2ers in morphosyntax; these studies showed a typological primacy effect (e.g., Giancaspro et al., 2015 for adults; Hopp, 2018 for children). However, there are no true initial stages phonological data for HSs. Understanding transfer patterns across modules can help determine whether initial transfer is wholesale. Specifically, we examine copula choice (essere vs. stare), differential object marking (DOM), and intervocalic underlying voiced stop /bdg/ lenition. Crucially, all three phenomena pattern together in English and Italian—a single copula is used across the conditions we examine, there is no DOM or voiced stop lenition—and differently in Spanish and Italian, the more typologically similar language pair.

Fifteen English-dominant Spanish HSs in weeks 5-7 of a first-semester Italian class completed a delayed repetition task (DRT) to examine stop lenition, and an acceptability judgment task (AJT) to examine copula choice/DOM. Participants were tested in all three languages on separate days (L3 first to avoid priming effects). DRT stimuli were CV.CV nonce words in a carrier phrase (e.g., I say ‘faba’ for you), and consisted of 30 critical items with intervocalic /bdg/ and 15 fillers. Productions were categorized as [-cont] in the absence of energy, a sharp decline in the intensity contour, and, if present, a release burst, and [-cont] otherwise. A 4-point scalar AJT (1=odd, 4=natural) consisted of 32 DOM items (8 for each combination of [animate]/[specific], 32 copula items (all taking essere in Italian but corresponding, in equal numbers, to ser and estar in Spanish), and 20 fillers. To account for the substantial variation attested in L3 acquisition, we calculated individual-level between- language McNemar tests (for binary data) and paired t-tests (for continuous data). Outcomes indicate these HSs are more likely to produce and accept Spanish-like patterns in L3 Italian regardless of English dominance, proficiency in (non-dominant) Spanish, or the module tested. We take this as cross-module evidence—the first of its kind—of the role of global structural similarity in initial stages transfer.