Rules, constraints, and lexical phonology in Glenoe Scots

0 Introduction* Glenoe Scots, a variety of Ulster Scots spoken in the village of Glenoe about twenty miles north of Belfast (Gregg 1958, 1959, 1973, 1985), displays a peculiar interaction between phonology and inflection: a lexical pattern systematically fails to occur not only in irregular plurals but also in the present tense forms of a specific class of irregular verbs. I show that this interaction cannot be handled by rules, but instead requires an analysis with an associative and dynamic lexicon of the sort recently argued by Pinker and Prince (1992) to be necessary for the description of irregular inflection. I suggest how this notion may be expressed with the device of schemas (Bybee and Slobin 1982), which are relatively loose, nonderivational generalizations across forms that share a family resemblance in a particular lexicon. I further show that the constraints of Optimality Theory (Prince and Smolensky ms.), which are similar to schemas in being nonderivational and violable, face many of the same problems as rules in describing Glenoe Scots. However, if rather substantial portions of the associative analysis are carried over, in particular the reference to a specific lexicon, Optimality Theory does allow a more precise formalization of the insights into the lexicon made by Pinker and Prince (1992) and Bybee and Slobin (1982). I end the paper by considering the plausibility of developing a theory of lexical phonology founded on an associative lexicon.