Temporal and spectral properties of voiceless fricatives in aphasia and apraxia of speech

Abstract

Background: This study represents an extension of a previous examination of fricative production in aphasic speakers with and without apraxia of speech (AOS; Haley, Ohde, & Wertz, 2000a). In our previous work, spectral overlap was found between alveolar and palatal voiceless fricatives produced by aphasic speakers with and without AOS, but not in normal speakers. Because all spectral measures were obtained at the midpoint of the fricative segment, categorical overlap may have been exaggerated by intra-segmental changes from one place of articulation to another. Moreover, the duration of the segments and the interaction between temporal and spectral properties were not examined. Aims: We asked three questions: (1) Are fricative segments longer than normal in aphasic speakers with and without AOS? (2) Are patterns of spectral overlap between places of articulation comparable at the beginning, middle, and end of the fricative segment? (3) What is the spectral evidence regarding intra-segmental changes in place of articulation? Methods & Procedures: The speech sample was the same as in our original investigation. Thus, 10 aphasic speakers with AOS, 10 aphasic speakers without AOS, and 10 normal control speakers each produced 48 words with an initial voiceless sibilant fricative. The duration of the fricative segment was measured and the Bark-transformed first spectral moment (FSM) time histories were computed for each utterance. Both quantitative and qualitative procedures were used to examine the data. Outcomes & Results: Fricative segment duration was significantly longer for speakers with coexisting aphasia and AOS than for the normal control group. Visual inspection of first spectral moment time histories revealed intra-segmental changes in spectral content for individual speakers in all three groups. The pattern of change across repeated productions was more variable for aphasic speakers with and without AOS than for normal speakers. Several individual patterns of spectral change were observed among aphasic and apraxic speakers. Spectral overlap between [s] and [] was noted in aphasic speakers with and without AOS at the beginning, middle, and end of the fricative segment, whereas normal speakers demonstrated limited overlap only at segment onset and offset. Conclusions: There was no evidence that intra-segmental modifications in place of articulation accounted for our previous observation of spectral overlap between fricative targets. The results indicate that abnormal phonetic properties are associated with, but not restricted to, a diagnosis of AOS.