Vowel duration as a cue to postvocalic stop voicing in aphasia and apraxia of speech

Abstract

Background: In American English, vowel duration is, on average, longer preceding postvocalic voiced stops than preceding postvocalic voiceless stops. Preliminary investigations have reported a preservation of this acoustic contrast for speakers with aphasia and apraxia of speech (AOS) on the basis of mean data. However, clinical interpretation of the available research is difficult due to lack of attention to the range of performance among and within speakers from both normal and disordered populations. Concurrent perceptual analysis is warranted to evaluate functional implications of acoustic variations, but standard approaches using a single listener and presentation may not be sufficiently sensitive to reveal subtle variations.Aims: (1) To determine whether aphasic and apraxic speakers produce a normal vowel duration differentiation between voiced and voiceless postvocalic stops. (2) To explore whether the produced vowel duration variations are associated with predicted perceptual effects.Methods & Procedures: Eight speakers with coexisting aphasia and AOS, eight with aphasia and no AOS, and eight normal control speakers produced 24 repetitions of the words “had” and “hat” in a short carrier phrase. For each utterance, the duration of the vowel was measured. Perceptual testing was conducted using three normal listeners and a forced‐choice perceptual identification paradigm.Outcomes & Results: As expected, all normal speakers, and most aphasic and apraxic speakers, displayed a mean vowel duration distinction between the voicing cognates. The magnitude of the distinction did not differ across groups. Instead, there was substantial inter‐speaker variability in the magnitude of duration contrast in all three groups. Some aphasic speakers with and without AOS did not distinguish in vowel duration between voicing cognates, and others displayed bimodal, but overlapping, distributions for /d/ and /t/. Results of the perceptual identification experiment indicated that there was good, but not perfect, agreement between variations in vowel duration and voicing perception and that several utterances produced by aphasic and apraxic speakers were perceptually ambiguous.Conclusions: (1) Although the mean duration for vowels preceding voiced and voiceless stops may be indistinguishable from normal, several abnormal acoustic patterns are found among individual aphasic speakers both with and without AOS. (2) The magnitude of acoustic distinction can vary considerably across normal speakers and this variation must be considered when evaluating disordered speech. (3) Perceptual identification testing facilitates the interpretation of acoustic data, particularly when the two levels of analysis are matched on an utterance‐by‐utterance basis. (4) Perceptual ambiguity can be demonstrated in disordered speech through perceptual identification testing.