Colloquium with Jonah Katz, UCLA

Title:  Weakening, predictability, and representations

Abstract:  This talk introduces a type of intervocalic consonant weakening I've referred to as 'continuity lenition', using Campidanese Sardinian as an illustration. A number of recent acoustic studies, including my Campidanese work, have suggested that this type of sound pattern is mostly or entirely predictable from the duration and timing of the relevant consonants. I discuss here several types of models consistent with this informational dependency. A Strong Reductionist model claims that lenition-fortition patterns depend only on the language-specific gestural and auditory targets for a phonological segment/class, with changes in manner following by physical law from changes in timing. A Weak Reductionist model claims that changes in manner track changes in timing, but that this weakening trajectory is itself part of the acquired language-specific representation of a phonological class, rather than following from articulatory and acoustic laws. A Gestalt model claims that changes in manner track changes in timing because both are used to reinforce perceptual grouping principles in speech. Assessing and comparing such models will require an enormous amount of detailed phonetic data across languages. Here I present some preliminary data from ongoing studies of American English and Peninsular Spanish, discussing the implications of between-segment and between-language variability in lenition trajectories.