The Weekly Newsletter of MIT Linguistics

Ling-Lunch 10/13 - Juliet Stanton (MIT)

Speaker: Juliet Stanton (MIT)
Title: Segmental blocking in dissimilation: an argument for co-occurrence constraints
Date: Thursday, October 13
Time: 12:30pm-1:50pm
Location: 32-D461

Most contemporary work assumes that dissimilation is motivated by featural co-occurrence (OCP) constraints (e.g. Alderete 1997, Suzuki 1998): a process that maps /X…X/ to [X…Y] (for example) would be explained by positing a ban on co-occurring [X]s.

I first show how this approach can be extended to analyze the typology of segmental blocking effects (name due to Bennett 2015), a term used to describe cases in which a dissimilatory process is blocked by some segments, but not others. For example, dissimilation might apply across some segment Z (/X…Z…X/ > [X…Z…Y]), but not some other segment Y (/X…Y…X/ > [X…Y…X]). This pattern can be explained in the following way (following Kenstowicz 1994, Steriade 1995): if a ban on co-occurring [Y]s (violated in the unattested /X…Y…X/ > *[X…Y…Y]) takes priority over the ban on co-occurring [X]s (violated in the attested /X…Y…X/ > [X…Y…X]), then dissimilation of /X…X/ to [X…Y] will fail if some [Y] is present elsewhere in the word.

I argue that all cases of attested segmental blocking should be analyzed as an interaction between two competing co-occurrence constraints (as above), and provide new evidence from lexical statistics in support of this conclusion. Time permitting, I will introduce an alternative correspondence-based analysis of blocking in dissimilation (Bennett 2015), and show that its predictions are less restrictive than those of the proposed analysis.