Donca Steriade
MIT

Donca Steriade

MIT
United States

Phonology and paradigm uniformity

Most phonological systems display certain patterns of similarity between lexically related words. These similarities are unexpected, in that the normal functioning of the phonology would predict more dissimilar pairs; they are directional, in that one form appears to be basic and others are derived from it; they can be recursive; and they frequently involve non-distinctive phonological properties (Steriade 2000), suggesting that they are relations between surface structures.  Thus the stress of English pairs like dísciplin and dísciplining is identical, 100 and 100-0. The latter deviates from the stress pattern of English simple words, where 000 strings are impossible. Virtually every language we know contains such similar sets. Children appear to anticipate their existence, even when the evidence available to them contradicts this expectation (Do 2018).

The analysis of these similarities varies, depending on the syntactic relations between the forms. When a similar pair involves nested forms, as with dísciplin-dísciplining, they are analyzed using the tools of cyclic phonology (Chomsky, Halle and Lukoff 1956) or, in OT, Base-Derivative correspondence and related ideas (Benua 1997). When this is not the case, as with uniform inflectional paradigms (Albright 2010, McCarthy 2005) or morphomes (Aronoff  1994; Steriade 2016) or other forms of stem syncretism (Steriade 2008; Steriade and Yanovich 2015), different and incompatible devices are invoked in their analysis, or no analysis is offered.

The question, then, is how the shared properties of such similar morpho-lexical sets can be given a unified account. The tutorial will illustrate the shared properties, will motivate one idea that leads to a generalized analysis, and will provide some experimental support for it.