Page 17 - Winter 2020

apple are fruits). If a deficit in phonemic hearing underlies comprehension impairments, then performance on the matching task should predict auditory comprehension; the poorer the performance on the matching task, the poorer the performance on an auditory comprehension task.
Recent research with a large number of subjects, careful clinical assessment, lesion mapping, and well-constructed test materials has shown strong support for the “phonemic hearing” hypothesis (Robson et al., 2012a). Nonetheless, it is also the case that lesions resulting in Wernicke’s aphasia may include the middle temporal gyrus (see Figure 1), an area implicated in semantic processing (Binder et al., 2009; Bonilha et al., 2017). Thus, it is possible that auditory comprehension deficits reflect both speech perception impairments and semantic impairments (Robson et al. 2012b). Regardless, it is clear that speech perception deficits have repercussions throughout the language system, affecting not only the perception of sounds but also access to words and their meanings (see also Blumstein, 2009; Dial and Martin, 2017).
Of importance, the overall patterns of speech perception impairments mirror those shown in speech production. First, similar to speech production, deficits emerge in both Broca’s and Wernicke’s aphasia despite the fact that the lesions producing these syndromes differ, indicating that the speech perception system is neurally distributed, encompassing more neural areas than just the temporal lobe. Second, the pattern of errors suggests that the sounds of language are composed of features organized in a network-like structure based on their articulatory and acoustic similarities. Third, brain injury does not affect the architecture of speech but introduces noise into the system, rendering similar sounds more likely to be misperceived than sounds further apart in the network. Fourth, despite similar patterns of errors, the number of errors distinguishes Broca’s and Wernicke’s aphasia, reflecting the neural areas specialized for perception on one hand and production on the other.
Perception of the Acoustic Properties Underlying Speech
The perception of speech requires an auditory analysis of the speech input and a mapping of this input onto the acoustic features associated with the sounds of language. A question is whether speech perception deficits in aphasia reflect failures to perceive these acoustic features
normally. For example, as discussed in On To Acoustics, VOT is a temporal cue that distinguishes voiced, [b d g], from voiceless stop consonants, [p t k]. As in production, there is a range of VOT values associated with voiced stops and voiceless stops (see Figures 3 and 4). Construction of a synthetic speech continuum with equal VOT steps (in milliseconds) between stimuli shows that listeners separate the stimuli into two distinct categories (see Figure 5). Importantly, despite equal acoustic steps, listeners can only discriminate stimuli that cross the phonetic boundary between voiced and voiceless stops, and they fail to discriminate stimuli that lie within either the voiced or voiceless phonetic category (Liberman et al., 1967). Thus, speech is perceived categorically, with
discrimination ability limited by category membership.
The perception of VOT is severely affected in aphasia regardless of clinical syndrome or lesion site (Carpenter and Rutherford, 1973; Basso et al., 1977). Discrimination performance is typically better than categorization (Blumstein et al., 1977b; Gow and Caplan, 1996). Importantly, discrimination functions are similar to those shown for non-brain-injured subjects, even for those aphasics who are unable to categorize the stimuli. This dissociation between categorization and
Figure 5. Idealized representation of categorization and discrimination responses to a continuum of synthetic [da]-[ta] stimuli ranging in VOT from −20 to +80 ms in 20-ms steps. Subjects failed to discriminate stimuli within either the [d] (blue line) or [t] (red line) category, but could discriminate stimuli (black line) perceived as members of two different (DIFF; blue vs. red) categories, i.e. [d] and [t].
Winter 2020 • Acoustics Today 17