Evidence for Classifying Metathesis Patterns as Subsequential
Jane Chandlee, Angeliki Athanasopoulou, and Jeffrey Heinz
303-309 (complete paper
or proceedings contents
This paper provides a computational analysis of metathesis that shows that local metathesis (involving adjacent segments) is subsequential, while long distance metathesis (involving segments with one or more intervening segments) is only subsequential if the intervening string of segments is bounded by some length. If the metathesis occurs across arbitrarily long distances, it is neither subsequential nor even regular. The significance of this result is that (1) it provides a tighter computational bound on the set of possible phonological patterns, and (2) it has implications from a learnability perspective, since subsequential relations are identifiable in the limit from positive data (Oncina et al. 1993) and have been studied in the domain of phonology (Gildea and Jurafsky 1996). Johnson (1972) and Kaplan and Kay (1994) have shown that phonological patterns are regular. Using the formalism of finite state machines (FSMs), it is shown that local metathesis is not only regular, but subsequential. In addition, if the intervening string of a long-distance metathesis pattern is bounded, then it too can be described by a subsequential FSM. Furthermore, the unbounded, non-subsequential patterns appear to be restricted to the diachronic domain. Since subsequential relations are a proper subclass of regular relations, this result is evidence for a stronger hypothesis than that of Johnson (1972) and Kaplan and Kay (1994): namely, that synchronic phonological patterns are subsequential (Heinz 2007, 2009, 2010).
Proceedings of the 29th West Coast Conference on Formal Linguistics
edited by Jaehoon Choi, E. Alan Hogue, Jeffrey Punske, Deniz Tat, Jessamyn Schertz, and Alex Trueman
Table of contents