This work investigates the question of how phonological systems might arise from phonetic cues, focusing on the contribution of the language learner. Specifically, I present a series of experiments on the learning of morphologically conditioned epenthesis patterns. A theoretical account that situates synchronic phonological systems as the product of natural sound changes (such as Evolutionary Phonology (Blevins 2004) ) would attribute the emergence of an epenthetic consonant to the listener's

more »

... sperception of the natural transition between two adjacent vowels: ratu+əәk, pronounced as ratuʷəәk, re-analyzed as ratuwəәk. The proposed origins of such a pattern guarantee that such epenthesis would be, at least at first, contextually conditioned (what I will call Type 1). A distinction has been made, however, between such systems and those in which a unique (unmarked) segment is epenthesized regardless of context (Type 2) (see, e.g., Lombardi (2002), de Lacy (2006) ). Within a phonetically-based learning model, such a system would require an additional stage of regularization or generalization on the part of the listener.