August 10, 2015

August 10, 2015

Written by Maximus Peperkamp, M.S. Verbal Engineer

Dear Reader, 

This writing is my tenth response to “Talker-specific learning in speech perception” by Nygaard and Pisoni (1998). "Serial recall of spoken word lists produced by multiple talkers was poorer than recall of lists produced by a single talker; but the result was found only in the primacy portion of the serial recall curve.” These results need to be analyzed in terms of whether the speaker produced Sound Verbal Behavior (SVB) or Noxious Verbal Behavior (NVB) and positively influenced the listener with an appetitive sounding voice or an negatively influenced the listener with an aversive sounding voice. 

“The primacy portion of the serial curve” is hypothesized to be absent in SVB and is believed to be mainly be a function of NVB. It was suggested that "variation in a talker’s voice from word to word in a list competes for processing resources in the recall task.” This interpretation doesn’t answer the question why this “competition for processing resources” occurs.  The SVB/NVB distinction, however, makes us realize that this “competition” occurs only due to NVB.

In SVB the verbal and nonverbal expressions of the speaker are aligned, but in NVB they are two different messages as they are disjointed. Moreover, in SVB the speaker is his or her own listener. This always positively effects the feelings of the listener, but in NVB the speaker is not listening to him or herself, which always negatively effects the listener. What is recalled by the listener from an aversive-sounding NVB speaker is mainly that he or she sounds aversive. 

Stated differently, because lexical information dominates and hinders the linguistic analysis of what the speaker says, the listener who listens to a NVB speaker is believed to remember less than the listener who listens to a SVB speaker. Whether it is possible or not, the listener who listens to a NVB speaker will try to move away from the aversive stimulation of the speaker and consequently remember less of the lexical information.

The “analysis of talker information during a memory task appears to be both time- and resource-demanding,” but only when the listener is dealing with a NVB speaker. Reasoned from the SVB/NVB distinction, we find that it is not the “talker variability” which “increases the capacity demands of the working memory system”, but it is SVB which increases this capacity and, by contrast, it is NVB, which decreases this working memory capacity. 

The researchers noted that recall is also affected by presentation rates. They  don’t mention that these, in turn, are determined by the kind of vocal verbal behavior of the speakers, that is, by their SVB or NVB. A SVB speaker's speech episode contains more instances of SVB than NVB, while a NVB speaker's presentation contains more NVB instances than SVB instances. 

The SVB presentation occurs at more relaxed pace and slower rate than the anxiety and stress provoking NVB presentation. Leaving out the influence of the talker’s voice on the listener, the researchers overlook what may be the most important independent variable, which is unspecified in the catch-all-phrase “talker variability.”

Authors unaware of the SVB/NVB distinction will maintain ‘mentalistic’ definitions, which are useless in any behavioral account. 'Talker variability' is a useless term if it doesn't address SVB and NVB. “This interaction between presentation rate and serial recall for the multiple- and single-talker word lists suggests that at fast presentation rates, when processing is constrained by time, talker variability affects both the perceptual encoding and the rehearsal of items in the serial recall task” (words underline by me). 

To consider the influence of the speaker’s sound, we should do away with constructs that represent verbal bias. Conclusions are drawn which prevent us from finding out what is happening. “At slower presentation rates, when listeners have more time and resources to encode and rehearse talker information, they are able to use that information to aid them in the encoding of item and order information.” With a SVB speaker the listener is at ease and better able to pay attention to what he or she is saying. The SVB/NVB distinction is a more parsimonious explanation than inferences about “encoding” and “rehearsing” of the “item and order information”.

Based on my knowledge about SVB and NVB, I object to the researcher’s conclusion. “These memory findings suggest that talker information may not be discarded in the process of spoken word recognition, but rather is retained in memory along with the more abstract, symbolic linguistic content of the utterance.” They seem to think that nebulous cognitive processes explain how “talker information” is “retained in memory.” 

What is left out by these authors is the fact that the listener’s neural behavior is altered by the sound of the speaker's voice, leading one listener to supposedly have better memory than the other. What actually happens is that the body of the listener who ‘remembers’ what the speaker has said was positively affected by the tone of the speaker’s voice. The stress that is produced by the NVB speaker always has an adverse effect on memory. 

If we don’t discard constructs as “information”, we continue to misrepresent classical and operant conditioning effects – in speakers and listeners –  of how the speaker sounds. Presumably “Talker-specific information is retained in memory along with lexical information" and "this information can facilitate listeners’ recognition memory.” SVB and NVB can be heard, but “talker-specific information” and “lexical information”cannot. 

Ironically, the researchers, who found that “Words repeated in the same voice were recognized better than words repeated in a different voice”, didn’t realize that fixation on words, a characteristic of NVB, distracts them from paying attention to how this “same voice” actually sounds.