We investigate the extent to which L1 versus adult L2 phonological systems resist influence from an L3. We test the Phonological Permeability Hypothesis (Cabrelli Amaro & Rothman, 2010), which states that adult L2 phonological systems are different from L1 systems with regards to instability.

To isolate the variable of age of acquisition, we examined the acquisition of L3 Brazilian Portuguese (BP) by two types of sequential bilinguals: L1 English/L2 Spanish, L1 Spanish/L2 English. We tested perception via a forced-choice goodness task and production via a delayed repetition task. First, we assessed acquisition of the phonological property in BP (in this case, word-final vowel reduction, and excluded learners’ data that was not target-like in BP. We then tested the learners’ Spanish to determine the level of BP influence.

Perception data were analyzed for accuracy and reaction time. Production data were analyzed acoustically for formant structure, duration, and intensity. We compared L1 English/L2 Spanish data (n=15) with L1 Spanish/L2 English data (n=8), and with Spanish (n=11) and BP controls (n=14).

While data from the preference task do not signal instability of perception for early or late acquirers of Spanish, L2 Spanish production data for vowel height measured differs from the L1 Spanish and Spanish control data. We take this as preliminary support for our hypothesis.

By comparing L1 and L2 vulnerability to L3 influence, this study takes a novel approach to the debate over the constitution of phonological systems acquired in childhood versus in adulthood.

The novel methodology implemented, together with these empirical findings, will afford further development of a research program dedicated to L3 bidirectional influence and the study of what L3 acquisition can tell us about language acquisition more generally.