Experimental and Cross -Linguistic Support for the P-Map

Allan Schwade

INTRODUCTION : Phonological models based on Optimality Theory (OT) currently predict more variability than is displayed in natural language. As Donca Steriade has observed, a common phonotactic constraint prohibits voiced
obstruents from appearing word finally, yet only a devoicing repair strategy is ever employed in said circumstance. That is, speakers will change words like [tab] into [tap] but not [tan], [tabe], or [ta]. To redress the excess variability found in OT, Steriade proposed a grammatical mechanism called the p-map hypothesis, which ranks correspondence constraints based on psychoacoustic knowledge (Steriade, 2001). The logic behind her proposal is that speakers will only accept changes that are most similar to the input and that similarity judgments are based on psychoacoustic information (i.e. what they hear). In an attempt to gather evidence for this amendment, I replicated an online similarity judgment experiment conducted by Kawhara for both English and native Spanish speakers. Participants favored devoicing in almost every experimental condition, mirroring Kawahara’s findings and demonstrating positive evidence for Steriade’s proposal.

METHODOLOGY : An online multiple-choice questionnaire was created to evaluate the similarity preferences of 21 English and 21 native Spanish speakers. Participants were asked “Which of the following options sounds most similar to [x]?” where ‘x’ was one of six experimental conditions ([ab], [ad], [ag], [itab], [itad], itag]) or six controls. There were four response choices to select from, each reflecting a candidate in which the coda of ‘x’ was deleted, epenthesized, nasalized, or  devoiced. The order of the questions and answer choices was randomized.

RESULTS : As Figure I demonstrates, English speakers had a strong preference for devoicing in nearly all experimental conditions. The only exception was the [itad] stimulus, where speaker preference for epenthesis was about equal to their
predilection for devoicing. Spanish speakers on the other hand, illustrated a preference for epenthesis in the mono-syllabic
stimuli (see Figure II). However, this preference did decline the futhur back in the mouth the coda segment’s point of articulation was. In di-syllabic conditions, spanish speakers showed a strong preference for devoicing across the board.

DISCUSSION: Overall, English speakers appear to judge similarity based on psychoacoustic knowledge. The same can be said for Spanish speakers, even considering their responses in the mono-syllabic conditions. The reason being,
Spanish contains a minimal word constraint which requires prosodic words to be disyllabic (Colina, 2003). If said constraint dominated the correspondence constraints reflecting the projections of the p-map then we would expect speakers to prefer epenthesis.

CONCLUSION: The results of both questionnaires illustrate positive support for Steriade’s proposal. In the future, similarity studies using auditory stimuli should be used to test the p-map, so that results cannot be attributed to orthographic factors (such as the subject misinterpretation of the stimuli).


[1] Steriade, D. (2001). The phonology of perceptibility effect: The P‐map and its consequences for constraint
organization. Ms., UCLA.

[2] Colina, S. (2003) Diminutives in Spanish: A morphophonological account. The Southwest Journal of Linguistics

Make a Free Website with Yola.