
Generalization Beyond Similarity: Support for Abstract Phonology
Author(s) -
Sara Finley
Publication year - 2016
Publication title -
proceedings of the annual meetings on phonology
Language(s) - English
Resource type - Journals
ISSN - 2377-3324
DOI - 10.3765/amp.v3i0.3647
Subject(s) - grammaticality , phonology , markedness , linguistics , similarity (geometry) , grammar , generalization , phonotactics , vowel harmony , learnability , natural language processing , computer science , psychology , task (project management) , artificial intelligence , mathematics , philosophy , mathematical analysis , image (mathematics) , management , economics
The present paper provides evidence from an artificial grammar learning task that supports abstract representations for phonology. Learners exposed to a novel back/round vowel harmony pattern were able to discriminate between grammatical and ungrammatical items that contained novel affixes, even though participants rated these novel items as highly dissimilar to exposure items. These results suggest that learners do not use exemplar-based notions of similarity as a metric for acceptability, but use abstract notions of grammaticality and markedness.