Speaker: Edward Flemming (MIT) - Joint work with Giorgio Magri (University of Genéve)
Title: Strict domination in probabilistic phonology
Time: Monday, March 10th, 5pm - 6:30pm
Location: 32-D831

Abstract: Labov (1969) hypothesized that rates of application of variable phonological processes obey
an intriguing generalization that we dub the Strict Domination Generalizationː Where multiple
factors affect the rate of application of a process, the factors can always be ordered in a hierarchy
such that “each [factor] in the hierarchy outweighs the effects of all [factors] below it” (cf. strict
domination between constraints in Optimality Theory).

More specifically: suppose an optional consonant deletion process is favored by the presence
of each of three factors, F1, F2, F3 (e.g. properties like presence of a following consonant, or
occurrence in an unstressed syllable). The Strict Domination Generalization implies that these
factors have to be ranked, e.g. F1 >> F2 >> F3 (borrowing the OT notation for ranking). Given
this ranking, the rate of deletion must be higher in all contexts where F1 = yes than in all
contexts where F1 = no, regardless of the values of the lower ranked factors, and within each
value of F1, the rate of deletion must be higher where F2 = yes, regardless of the value of F3,
resulting in the order of deletion rates shown below. F1 outweighs all factors below it in the
sense that contexts where only F1 favors deletion have a higher rate of deletion than contexts
where both F2 and F3 favor deletion, but F1 does not.

I will review several test cases that are consistent with this generalization. We will then see
that the generalization does not follow from any current model of probabilistic grammar
(Stochastic OT, MaxEnt, Noisy Harmonic Grammar), so if it is correct, we need a new model.

On March 1st 2025, the 10th Workshop on Turkic and Languages in Contact with Turkic (Tu+ 10) was held at University of Southern California. Our current student Bergül Soykan (3rd year student) presented a poster entitled “The Underlying Structure of Correlatives and Unconditionals in Turkish”. You can read the abstract here, and a shorter version is provided below:

This study investigates the structure of correlatives and unconditionals in Turkish, focusing on their syntactic and semantic parallels and distinctions. I propose that correlatives align with standard if-conditionals, while unconditionals resemble antecedent-external “even if” da conditionals in Turkish, particularly through the interaction with the particle