Vowel parameter shifts due to decreased stress for repeated words in passages

Ten experienced speakers read four paragraphs, each with ten repetitions of a high information CVC (consonant-vowel-consonant) word. Speakers were checked by judges to ensure that they spoke accurately. In three word repetitions (1st, 5th and 10th), which had identical syntactic contexts, vowels were analyzed spectrographically. For peak fundamental frequency, duration and peak amplitude, there were significant differences (p < 0.05) among repetitions for averaged values across three of the four CVC words. There was a significant (p < 0.05) decreasing linear trend across repetitions for all three parameters. When the raw data was rank ordered using difference limen criteria, results were the same. When the rank—order data was averaged across parameters, the decreasing trend was emphasized with significant differences between successive repetitions.

There were no differences among repetitions for F1 and F2 formant onset and target frequencies. As speakers spoke all target words accurately, all formant parameters remained within acoustic limits for vowel identity.

Correlations of the acoustic measures among the seven parameters were insignificant. The sample size (n = 10) may have been too small to detect any relationships between the parameters.

Trained listeners were not able to consistently rate isolated word repetitions for comparative stress levels. However, when entire passages were used, listeners were able to judge the stress levels of word repetitions. The 1st utterance was given the highest stress rating which agreed with the acoustical measures for peak fundamental frequency, duration and peak amplitude. Stress ratings for the 5th and 10 utterances were similar and also agreed with acoustical measures for the same parameters. It was concluded that perceptions of comparative stress required speech units greater than syllable length. Recommendations for future studies were presented.