|
|
 E-mail this message to a friend
|
|
Title:
|
An Autonomous System for Quantifying the Perceptual Use of Acoustic Speech Cues: Voicing in intervocalic /t~d/ in French
|
|
Author:
|
Thomas Sawallis
|
|
Email:
|
click here to access email
|
|
Degree Awarded:
|
University of Florida
, Program in Linguistics
|
|
Degree Date:
|
1996
|
|
Linguistic Subfield(s):
|
Phonetics
|
|
Subject Language(s):
|
French
|
|
Director(s):
|
Caroline Wiltshire
William Sullivan
|
|
|
Abstract:
|
|
Research has shown that phonemic decisions in speech perception are influenced by the information in multiple speech cues, and that those cues are produced with different settings across various different contexts, including different languages. However, there has never been an accepted method for quantifying the perceptual use or importance (often termed 'weight') of cues in a way that would allow cross language comparison, perhaps the most basic of linguistic study types. The existing assessment methods are either limited or designed in ways that make them incapable of rendering such a comparison.
This research proposes a quantitative, independent, abstract, and autonomous measurement system for a cue's perceptual importance. The system begins with a survey of the cue's acoustic measurements in natural speech, since a memory of that distribution is the presumed basis for the listener's expectations of that cue. Then perceptual stimuli are designed according to a pattern based on the standard deviation of the cue's surveyed distribution, which has the effect of normalizing the step size in the stimulus series across cues in all possible environments, including language. Finally, the stimuli are played to listeners and the responses are compared using mathematical techniques from Signal Detection Theory, specifically, the d' sensitivity measure. This has the advantage of quantifying the cue's importance independently of the overall performance accuracy, which can vary for reasons extraneous to the perceptual system, and the quantification is done on an abstract scale suitable for any cue in any context in any language.
The measurement system is tested on four cues for the voicing contrast of intervocalic /t~d/ in French, with native and nonnative listeners, and the results, though limited, are promising. For instance, native French speakers are shown to use hold duration (as operationalized) more in perceiving /t/ than in perceiving /d/. Also, certain weaknesses are discovered in the experiment design used to implement the measurement system, and remedies are proposed to allow more and stronger conclusions as the measurement system is refined and developed.
|
|
|
|
|
|
Page Updated: 26-Nov-2009
Please report any bad links or misclassified data
LINGUIST Homepage | Read
LINGUIST | Contact us
While the LINGUIST List makes every effort to ensure the linguistic relevance of sites listed
on its pages, it cannot vouch for their contents.
|
|
