Learning Phonotactic Grammars from Surface Forms

This paper presents an unsupervised batch learning algorithm for phonotactic grammars without a priori Optimality-theoretic (OT) constraints (Prince and Smolensky 1993, 2004). The underlying premise is that linguistic patterns (such as phonotactic patterns) have properties which reflect properties of the learner. In particular, this paper explores one way in which a learner benefits from considerations of locality; i.e. “the well-established generalization that linguistic [here: phonological] rules do not count beyond two” (Kenstowicz 1994:597).1 The formalization of locality adopted here leads to a novel, nontrivial hypothesis: all phonotactic patterns are neighborhood-distinct (to be defined in §5).