Production of Sentence-Final Intonation Contours by Hearing-Impaired Children

Abstract

Studies of intonation in the hearing impaired (HI) are often concerned with either objective measures or listener perceptions. Less often has the focus been on how these two aspects of communication interrelate. This study examined the relationship between certain acoustic parameters and listeners’ perceptions of intonation contours produced by HI children. Six severe-to-profound HI children and 6 normal-hearing (NH) children, ages 7;9 to 14;7, were individually tape recorded while reading 10 declarative sentences and 10 phonemically matched interrogative sentences within the context of a script. Each sentence ended with a carefully chosen disyllabic (target) word. Twelve adult listeners, inexperienced with the speech of the HI, listened to a randomized audio tape presentation of all of these productions and categorized each one as a statement, question, or other. Fundamental frequency (FO) and duration measurements were obtained for the target (final) word of each sentence, and intensity measures were recorded for each entire sentence. Acoustic analysis showed that all 6 of the NH children and 4 of the 6 HI children produced acoustically different intonation contours for declarative versus interrogative sentences. The HI children’s productions were, in general, similar to the NH children, in that they used FO, duration, and intensity cues to mark the distinction. Their contrastive use of these acoustic cues, however, was less pronounced than for the NH children. Analysis of listener responses indicated that, although listeners were able to differentiate between some of the declarative and interrogative sentences produced by these 4 HI children, judgments corresponded with their intended type less often for the HI than for the NH children. (Judgments of NH children’s utterances were 100% correct.) Multiple logistic regression of listeners’ responses to the HI children’s utterances showed that 4 acoustic measures, all derived from the sentence-final word, were significantly predictive: (1) sentence-final FO, (2) slope between the target word’s initial and final FO, (3) duration of the target word, and (4) dB difference between the target word’s 1st and 2nd syllables. Results were similar for the NH children’s data, except that the ratio of the 2 syllables’ durations was significant, rather than total word duration. These findings differ in several important ways from previously published data for HI children’s intonation contours and suggest that many HI children have the ability to benefit substantially from training in the production of intonation.