|Title:||Perception of gestural overlap and self-organizing phonological contrasts|
|Abstract:||Languages that maintain distinctive secondary articulation contrasts tend to avoid multiple vowel contrasts, particularly rounding contrasts in front and back vowels. At the same time, languages with complex vowel inventories very rarely show distinctions in secondary consonant articulations, for example, in palatalization or labialization. These observations are based both on an analysis of the UPSID Database (Maddieson & Precoda 1990) and on an examination of inventories of a number of languages of Europe that exhibit at least one of the above mentioned contrasts.
In this paper I provide an explanatory account of these co-occurrence restrictions on seemingly unrelated segments and derive the two mutually exclusive patterns through a learning simulation – an acquisition of a hypothetical language with excessively marked segment inventory (4 high vowels differentiated by backness and/or rounding and 4 consonants with secondary articulations corresponding to the vowels). I demonstrate that the observed markedness effects emerge naturally from low-level interactions between a speaker and a listener/learner as a result of limits on what can be successfully transmitted through the speech communication channel. The key factor in the process is the failure on the part of the listener to correctly process overlapped gestures (Browman & Goldstein 1989) that happen to share the same articulator.
The results suggest that no “innate” restrictions against having both types of contrasts in inventories need to be assumed (cf., de Boer 2000): A language having these contrasts will inevitably “self-organize” by shifting to a more stable pattern: with either rounding contrasts in the vowels, or secondary articulation contrasts in the consonants, or none of these marked contrasts.