Tuesday, November 28, 2017

The mystery of the DAF effect on stuttering

Hi and welcome!

My theory website on stuttering has been online for two years now. In that time I added many paragraphs and footnotes in order to include new empirical results, thus the text on the pages has become longer and longer, and if I continue in this way, then it may become confusing. Therefore I start this blog to discuss new papers on stuttering here. I will do this in the context of the theory published on my website: Are the assumptions in the theory consistent with new results of empirical research? Does the theory help to explain new empirical results? By that, I hope, I can a little bit contribute to the discussion about the causes of stuttering.

The first article I want to discuss is “Stuttering adults’ lack of pre-speech auditory modulation normalizes when speaking with delayed auditory feedback.” by Ayoub Daliri and Ludo Max. The study was recently published in the journal Cortex, see here (free full text).

The main findings, in my view, are:

(1) The function of pre-speech auditory modulation is probably not a suppression of auditory input during speech (as was assumed earlier), but rather an optimization of the processing of auditory feedback.

(2) Pre-speech auditory modulation was limited in adults who stutter compared with control participants. Figure 5 shows that the reduction of the N1 amplitude (indicating pre-speech auditory modulation) in all stuttering participants was smaller than the smallest reduction in the non-stuttering control group.

(3) Delayed auditory feedback (DAF) caused an increase of pre-speech auditory modulation in most of the stuttering participants, such that it was similar on average to that in the control group. The authors write that in adults who stutter, “DAF paradoxically tends to normalize their otherwise limited pre-speech auditory modulation.”

The results are very important because they first time shed a light on the mechanism how DAF works when reducing stuttering: It supports pre-speech auditory modulation and (probably) the processing of auditory feedback and its integration in speech control. The understanding in which way DAF reduces the disorder, should help us to better understand the disorder itself.

In Section 3.1 on my website, I assume that DAF and other kinds of altered auditory feedback work in the way that they draw the speaker’s attention to the auditory channel, at least as long as the feedback alteration is experienced as unaccustomed. Is that assumption consistent with the new findings?

I think it is. In the study (Experiment 2), pre-speech auditory modulation was measured as the attenuation of the amplitude of the N1 component of an auditory-evoked potential. The potential was elicited by a probe tone presented ca. 200ms prior to (a) silent reading and (b) reading aloud. With normal auditory feedback (NAF), the N1 amplitude in the control group was reduced prior to reading aloud versus silent reading, but there was no such difference in the stuttering group. To put it crudely, the stuttering participants (or their brains) behaved as if they did not expect anything to hear at speech onset.

Auditory N1 is weaker after standard stimuli, but stronger after a deviant stimulus. It is also weaker when a person controls the creation of auditory stimuli as it is the case at speech onset, but also when the person elicits a computer sound by pushing a button – the N1 is stronger when the sound is started by the computer (see Behroozmand &Larson, 2011, for an overview). This together strongly suggests that N1 has to do with expectation: It is stronger after an unexpected stimulus, but weaker after an expected stimulus.

And now, I think there is a relationship (an interaction) between expectation and attention: Someone expecting a sound will probably listen, and someone listening will probably expect to hear something. When a person’s attention is directed to the auditory channel, and he or she expects something to hear, then the N1 amplitude after a probe tone will be smaller than in a situation in which no attention is directed to the auditory channel such that the probe tone is more surprising.

Possibly, N1 reflects a brain mechanism which controls selective attention on the basis of sensory input in a way that attention is primarily drawn to new or unexpected things – a mechanism that certainly proved beneficial for survival in the course of the evolution.

If so, we can interpret pre-speech auditory modulation as a modulation of attention: In normal speakers, some attention is directed to the auditory channel prior to and during speech such that auditory feedback is properly processed (verbal input is poorly processed without attention to the auditory channel; see, e.g., Sabri et al., 2008).

In persons who stutter, this pre-speech modulation of attention is weaker or absent, but DAF draws their attention to the auditory channel because it sounds odd. Hence I think it is not paradoxical that DAF normalizes pre-speech auditory modulation in stutterers.

However, that was not the case in all of the stuttering participants in the study. Figure 4E shows that DAF increased the amount of N1 modulation only in 9 of 12 stuttering subjects – and even in only 4 of 12 control subjects. Is that consistent with the assumption DAF modulates attention?

One possible reason for the different response to DAF in the stuttering group may be that some participants experienced a delay of 100ms as unpleasant such that they strove to ignore it. In the SpeechEasy device, the manufacturer’s default setting is a delay of 60ms (Foundas et al., 2013), the two devices tested by Unger, Gl├╝ck,and Cholewa (2012) had a delay of 50ms as default setting. Antipovaet al. (2008) found delays of 50ms and 75ms to be effective. Hence, a delay of 100ms may have been too long for some individuals. In a future replication of Experiment 2, the effect of various delays should be tested.

But why did the DAF reduce the amount of pre-speech auditory modulation in the majority of the control participants? I think it was the same as with the three stutterers: They experienced the DAF as annoying and strove to ignore the auditory channel. Why did much more nonstutterers than stutterers behave in this way? A possible answer is: Since the natural auditory feedback of speech is optimally processed in normal speakers, they experience a feedback delay of 100ms more likely as a disturbance than stutterers do.

In sum, I think the exciting new results obtained by Daliri and Max are consistent with the assumption that DAF and other kinds of feedback alteration (like chorus reading, whispering, speaking in an assumed voice) reduce stuttering by drawing the speaker’s attention to the auditory channel.

The uneven response to frequency-altered auditory feedback in the stuttering group in Experiment 1 may again result from different subjective experience: Perhaps, some of the stuttering participants did not perceive the higher pitch as unpleasant, thus they did not compensate for (some stutterers report they perceive their own voice - with natural feedback - as unpleasant or droning).

The negative correlation between the response to frequency-altered feedback in Experiment 1 and the amount of pre-speech auditory modulation with natural auditory feedback in Experiment 2 (Figure 5) is really astonishing. It does not look like chance, but I have no idea what it could mean. No end of mystery...