An acoustical study of fluent speech in stutterers and nonstutterers

The research investigated coarticulatory timing and attainment of frequency targets in perceptually fluent utterances of adult, male stutterers and nonstutterers. The specific acoustic characteristics assessed were: voice onset time; transition, steady-state, and total vowel durations; velocity of movement through the transitions; and the amount of the shift in frequency across the consonant in selected V,CV sequences. These characteristics were assessed in two rate conditions: (1) the speaker's own judgment of a normal rate of speech, and (2) a fast rate of speech. A fast rate was operationally defined as a minimum of two seconds shorter than the time to complete five repetitions of the experimental utterance at a normal rate.

Recordings were made of each subject reading a constant carrier phrase into which a changing medial bisyllable was embedded ("Say the word please"). The bisyllabic stimuli consisted of a vowel- consonantvowel, in which the voiced plosives, /b,g/, and the vowels, /i, u/, were used. Four symmetric and four asymmetric stimuli were thus constructed.

Wideband, linear-expand spectrograms were made of the recorded stimulus items. Only fluent (medial) repetitions were analyzed for acoustic characteristics.

Temporal delays, as measured at the second formant, occurred within the speech intervals of voice onset time, vocalic transitions, and steady-state portions. The data concerning stutterers' velocity of movement, although equivocal, suggest adequate articulatory speed at iii

iv normal rates of speech. The specific conclusions, regarding group differences, drawm from the results of this study are:

1. Stutterers differ significantly from nonstutterers in voice onset time. The observed difference was approximately 4 msec at a normal rate of speech and 6 msec at a fast rate.

2. Stutterers, on the average, spent less time in vocalic transitions than did nonstutterers. At normal rates stutterers' transitions averaged 6 to 8 msec less than nonstutterers. At fast rates the two groups performed similarly, as stutterers exhibited up to 3 msec shorter transitions.

3. Stutterers averaged longer steady-state durations than nonstutterers. At normal rates stutterers' steady-state portions averaged 7 to 13 msec longer, while at fast rates differences were approximately 11 to 12 msec longer than those of nonstutterers.

4. The durations of the total vocalic elements were comparable between stutterers and nonstutterers. Greater similarities were observed at normal speaking rates than at fast rates.

5. Results regarding overall velocity of the transitional movement are equivocal. There was a tendancy for stutterers to exhibit smaller velocities and this difference was even greater at fast rates of speech.

6. Stutterers did not differ significantly from nonstutterers in the amount of frequency shift across the consonant in the V,CV sequences.

In as much as physiological movement can be inferred from changes in the acoustic signal, the observed results suggest that temporal disruptions exist in the perceptually fluent speech of stutterers. These disruptions in timing were observed in articulatory-phonatory and

steady-state patterns. Stutterers were apparently able to produce transitional shifts appropriate to the phonetic environment, but they appeared to delay the initiation of these coarticulatory movements. The results are discussed within the framework of two hypotheses. Aberrant speech timing may suggest an involvement in the planning or integration of articulatory/coarticulatory movements rather than difficulties with movement execution. An alternate hypothesis that aberrant speech timing is only one aspect of a Central Nervous System timing disruption is explored. Future research needs to investigate this relationship are put forth.