LingLunch : Alex Martin

Alex Martin (Groningen University)
Feature-based generalization in sound pattern learning depends on phonetic motivation

Vowel harmony is a typologically common, phonetically-motivated pattern, while vowel disharmony, though formally equally complex, is barely attested and clearly lacking in phonetic motivation. Previous work using the artificial language learning paradigm has shown asymmetrical generalization of vowel harmony during learning, when comparing to vowel disharmony, lending support to the hypothesis that learning of phonological patterns is biased by phonetic knowledge. In the present talk, I will review some previous studies and report results from two new experiments with both French and English speakers which extend previous work. Both samples show clear patterns of generalization of vowel harmony to new morphological contexts involving previously unseen vowels. No generalization of vowel disharmony is observed for either group. These results support previous findings and lend strong support to the substantively-biased phonological learning hypothesis.