Abstract Summary
This study extends the Phonological Permeability Hypothesis to heritage phonological systems. L3 Portuguese learners’ Spanish global accent was evaluated at the onset and conclusion of a 16-week course. Accent ratings fell for heritage Spanish and L1 Spanish/L2 English speakers, indicating a primary role for age of acquisition in language stability.
Abstract :
The current study tests an extension of the Phonological Permeability Hypothesis (PPH, Cabrelli Amaro, 2013, 2017; Cabrelli Amaro & Rothman, 2010), which posits that adult-acquired phonological systems, even when evidencing native-like production and perception, are different from early-acquired systems with regards to the stability of the phonological system. While the PPH was conceived with late sequential bilinguals in mind and was first tested by comparing mirror image L1 Spanish/L2 English and L1 English/L2 Spanish speakers learning L3 Brazilian Portuguese (BP), the prediction that logically follows for Spanish heritage bilinguals learning L3 BP is that they should align with late L1 Spanish/late L2 English bilinguals. However, a number of variables that differentiate heritage speakers from late bilinguals could potentially impact the stability of their heritage Spanish system (e.g., dominance, input quantity).
The present study longitudinally examined L3 BP learners’ global accent at the onset and conclusion of a 16-week L3 BP course. We used accent ratings to determine the degree of influence from L3 BP on learner’s Spanish accent and whether the degree varied as a function of bilingual type (L1 Spanish/late L2 English, n = 11, heritage Spanish, n = 10). Fifteen-second error-free speech samples were extracted from oral interviews recorded at the two time points and evaluated by native Spanish raters familiar with English-accented Spanish and the US contact variety of Spanish. Raters assigned scores of 1 (very strong foreign accent) to 9 (native speaker) to 50 samples (42 experimental, 8 monolingual Spanish controls), which were then z-score transformed. Pilot results from 10 raters (data collection is ongoing) indicate a group-level trend of a minor reduction in accent rating for both groups, aligning with the prediction that the two groups pattern together by virtue of age of acquisition.
