Universal phonological restrictions and language-specific repairs: Evidence from Spanish

Berent, I. 1 , Rosselli, M. . 2 & Lennertz, T. 1

1 Northeastern University
2 Florida Atlantic University

Many studies show that speakers are sensitive to the sound-structure of their native language. Linguistic analyses, however, suggest that grammars also include broad, perhaps universal restrictions on language structure. To test this proposal, we investigate whether speakers posses phonological preferences concerning structures that are unattested in their language. Our case study concerns the restrictions on onset clusters (e.g., bl in block). Across languages, certain onset clusters are universally dispreferred (e.g., under-represented): Onsets like bn are preferred to bd, which, in turn, are preferred to lb. Previous research has further demonstrated that English speakers extend such preferences to onsets that are unattested in their language. Our question here is whether similar preferences might obtain for speakers of Spanish—a language whose repertoire of onset clusters is even smaller.
We examined this question in three experiments, eliciting syllable count (e.g., does lbif include one syllable or two?) and AX discrimination (e.g., is lbif=lebif?; is lbif=elbif?). We reasoned that ill-formed monosyllables should be systematically repaired as better-formed disyllables (e.g., lbif—>lebif or elbif), and consequently, these monosyllables should be misidentified as disyllabic (in a syllable count task) and harder to discriminate from those disyllables (in AX tasks). Results showed that, like English speakers, Spanish speakers tend to misperceive universally dispreferred onsets as disyllabic, suggesting that ill-formed onsets are repaired by inserting a vowel. Unlike English speakers, however, Spanish-English bilinguals often repair ill-formed onsets by prothesis (e.g., lbif—>elbif), rather than epenthesis (e.g., lbif—>lebif). Moreover, the choice of repair is modulated by language-dominance: Strongly-dominant Spanish speakers (acquiring English after the age of ten) favor prothetic repair, whereas their weaker-dominant counterpart opt for epenthesis, the repair favored by English participants. Results suggest that speakers might possess universal linguistic restrictions on onset structure, but their effect in perception is modulated by their linguistic experience.