Page 15 - Winter 2020

Acoustic Analysis of Speech Production in Aphasia
As Figure 3, right, shows, speakers do not have sufficient articulatory control to produce exactly the same VOT value every time they produce a voiced or voiceless stop consonant. Instead, there is a range of values for each category. Importantly, there is no overlap in VOT between voiced and voiceless stops.
The story is different in aphasia (Blumstein et al., 1980; Itoh et al., 1982; Verhaegen et al., 2020). As Figure 4 shows, those with Broca’s aphasia show a deficit in producing VOT. They fail to maintain the distinction between voiced and voiceless stops. Instead, there is an overlap in VOT between voiced and voiceless sounds, suggesting that they have an articulatory problem in laryngeal (vocal fold) timing. This is not the case in Wernicke’s aphasia where production of voicing is similar to the normal pattern.
Figure 3. Left: measurement of voice-onset time (VOT) for the syllables [di] and [ti]. VOT is the time (red arrows) between the release of the stop consonant closure (left vertical line) and the onset of voicing (right vertical line). Right: VOT distributions for [d] and [t] produced by English and French speakers. Note that English [d] and French [t] have similar distributions. Reproduced from Kessinger and Blumstein (1997).
Figure 4. The VOT distribution of alveolar stop consonants produced by a control subject (top), a Broca’s aphasic (center), and a Wernicke’s aphasic (bottom). Number is the number of productions at a particular VOT value. Black bars, target voiced stops, [d] (D); red bars, target voiceless stops, [t] (T).
Figure 3, right, shows that in English, there is a short delay in the onset of vocal cord vibration relative to the release of the stop consonant in the production of [d] and a long delay in the onset of voicing relative to the release of the stop consonant in the production of [t]. In contrast, in French, [d] is produced with voicing beginning before the release of the stop consonant, whereas [t] is produced with a short delay in the onset of voicing. Thus, the two languages differ in how they produce voicing in stop consonants. Indeed, English [d] and French [t] are both produced with a short delay in voicing.
The question then is whether a foreign accent syndrome patient who speaks English, now with a French sounding foreign accent, produces the voicing difference between [p-b], [t-d], and [k-g] as a French speaker or as an English speaker. Analyses showed that the production of voicing was the same as an English speaker, not a French speaker (Blumstein et al., 1987). Further examination showed that the foreign accent was not a foreign accent at all. Rather, there were global changes to a number of parameters of English, none of which could be associated with the sounds of any other language. In other words, acoustic analysis revealed that the foreign accent was in the “ears” of the listener, not in the “voice” of the speaker.
Winter 2020 • Acoustics Today 15