Date of Award

Summer 7-17-2014

Document Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Rebecca Scarborough

Second Advisor

Esther Brown

Third Advisor

William Raymond


The role of perception in sound change is an open question. Perception may serve a functional purpose in sound change, for example, leading to enhanced distinctions, or it may affect sound change non-functionally as a result of misperception. This dissertation addresses this question by examining the stop/approximant alternation in Spanish. Traditional explanations of this alternation describe it as a lenition process motivated solely by articulatory factors. A widely accepted description of the core alternation is that the underlying stops /b,d,g/ are maintained after pauses and in post-nasal position, and through a lenition process these sounds lenite to [β,ð,Ɣ ] in intervocalic position and elsewhere. Detailed phonetic descriptions, however, have noted the appearance of approximants in non-leniting environments such as phrase-initial position, and cross-dialectal variation is common. This variation raises questions regarding how approximantization has spread to non-leniting environments, and suggests that a simple articulatory explanation is insufficient.

This dissertation explores how perceptual factors may be contributing to the spread of approximants across phonological environments and across dialects. To test whether enhancement or misperception has played a part in the spread of approximants across Spanish, four perception experiments and a production task were administered to speakers of two dialects of Spanish—one “conservative’” (Columbian) and the other “innovating” (Mexican). The production data confirmed that variation is the norm within and across dialects: Approximants appear in non-leniting environments with some regularity, and speakers of Colombian Spanish showed lower rates of approximantization than Mexican speakers. Results from the perception experiments showed that confusability between the two segment types is high; however, there was no clear evidence of a correlation between confusability and likelihood of approximantization. In the experiments that investigated perceptual enhancement, there was no indication that approximants were perceptually advantageous with respect to place of articulation or voicing. As a result, perceptual enhancement was also ruled out as a motivation for this sound change. Overall, the role of perception as a mechanism of sound change was found to be negligible, but results of the experiments point to promising directions for future research concerning the perception/production loop and the nature of stored representations.

Included in

Linguistics Commons