The Syllable in Domain Generalization: Evidence from Artificial Language Learning

Domain generalization refers to a type of diachronic account of certain domain-edge phenomena, in which the domain of application of a phonological pattern narrows from a larger level (e.g., the phrase/utterance) to a smaller one (e.g., the word). Building upon Myers & Padgett’s (2014) artificial language learning study, which showed that participants could learn an utterance-final pattern and generalize it to the word level, this paper investigates whether learners can generalize to the syllable domain. I consider differences between the word and the syllable as phonological units and present an original artificial learning experiment, in which participants were able to generalize to both the word and syllable levels. Results suggest that sub-word units like the syllable may be stored in memory or otherwise encoded in phonological representations in a way that makes them available for the types of calculations that lead to domain generalization. This also speaks to the nature of domain generalization, showing that generalization can go beyond the word level to smaller prosodic domains.