August 9, 2015

August 9, 2015

Written by Maximus Peperkamp, M.S. Verbal Engineer

Dear Reader,

This writing is my ninth response to “Talker-specific learning in speech perception” by Nygaard and Pisoni (1998). It amazes me that the relation between “the indexical properties of the speech signal” and “the more abstract linguistic content of an utterance” needs to be pointed out. My reasoning is based on Sound Verbal Behavior (SVB) in which what we say is as important as how we say it. Reasoning which is based on what I call Noxious Verbal Behavior (NVB) creates a split between what we say and how we say it. The former is more important than the latter in NVB.

We must realize that “the problem” created by this split only occurs in NVB and never in SVB. There is no problem that in SVB these two are conveyed simultaneously. Surely, most researchers are unaware of the great difference between SVB and NVB. That is why they write that “The essence of the problem is that both types of information are conveyed simultaneously along the same acoustic dimensions within the speech signal."

Actually, they are unknowingly saying that NVB is problem. Only in NVB “the information about the talker must be disentangled from information about the linguistic content of the utterance.” What they call “perceptual normalization” I call SVB, as SVB includes “an account of the processing and representation of both the linguistic and the indexical information that are carried in parallel in the speech signal.” This is a sophisticated way of describing SVB. Moreover, while SVB normalizes our perception, NVB can be said to distort our perception. It is only a small step from “talker variability” to a different way of talking, that is to SVB and NVB.

“Several studies have shown that talker variability has a significant impact both on the perceptual processing of spoken utterances and on the memory representations constructed during the perception of spoken language.” The interpretation of such studies begins to make much more sense when we identify such impact as the positive or negative emotions in the listener.

The two subclasses of vocal verbal behavior, SVB and NVB, refer to how the listener’s affective experiences interact with “perceptual processing” and “memory representations constructed during the perception of spoken language.” A reinterpretation of the research makes clear that because of a certain way of talking we perceive reality as it is, as we embody that reality during our spoken language.

“Talker variability has been shown to affect both vowel perception." Also, it was found that "perceptual identification of words presented in noise was significantly poorer when the words were produced by multiple talkers than when they were produced by a single talker." Once we are familiar with the SVB/NVB distinction it is quite clear that only SVB can improve vowel perception and spoken word recognition, while NVB will always impair it.

Perceptual identification of words presented in noise will only be better if this single talker has SVB, but not if he or she has NVB. If among multiple talkers there would be a couple of SVB talkers and if the single talker would be a NVB talker, then perceptual identification of words uttered by multiple talkers is hypothesized to be higher than for the single talker. Also, the “difficulty ignoring irrelevant variation in the talker’s voice when asked to classify syllables by initial phoneme” is hypothesized to only occur with a NVB speaker, but not a SVB speaker. To the contrary, with a SVB speaker’s variability is believed to enhance perception.

“Aspects of the speech signal related to classifying talker identity seem to be integrally linked to attributes related to the processing of the linguistic content of the signal.” They are, but we can only acknowledge this during SVB, whereas during NVB we deny this. Thus, we understand each other better during SVB in which the speaker talks with, not at the listener. In SVB there is no need to classify talker’s identity as the listener is safe, but in NVB talker’s identity is important as the talker threatens the listener.