December 14, 2020

Why does stuttering disappear when one’s own voice is not heard?

Some days ago, I put a paper with the above title online (see here). Previously I submitted it for publication in JSLHR and JFD; in both cases, it was rejected by the editors without regular anonymous peer review. By JSLHR, the paper was rejected without rationale; JFD told me that it “is very speculative, not supported by direct evidence.”

I think it is important to understand the purpose of the paper: to explain why a fact that has been regarded as evidence against feedback theories of stuttering is no evidence against such theories. To this end, it is sufficient to propose a valid explanation, and ‘valid’ means that this explanation is (a) coherent and consistent in itself and (b) consistent with the available data.

I do neither claim that my explanation is true, nor that it is the only possible one. However, as far as I know, it is the only explanation of this problem that has been proposed until now. Therefore, and as long as no better one is proposed, my explanation should be accepted as a candidate for the true explanation. Previously, there was no explanation for some facts that seemed to be paradoxical and enigmatic. Now, there is a plausible explanation that allows us to understand those facts such that they no longer appear paradoxical and enigmatic. I think that makes a difference.

In the table below, I have listed all arguments my explanation consists of and the corresponding empirical evidence or logical relation. All references in the table are also included in the paper. I think that’s a good way to review a theory paper. If you still find something that is speculative, i.e., neither empirically nor logically reasoned, please let me know.

Tablee: Structure of the arguments contained in the theory paper “Why does stuttering disappear when one’s own voiec is not heard? External and internal auditory feedback and their impact on stuttering.”
 Assumptions and assertions 
 contained in the paper
 Empirical evidence, logical  
 derivation, or reference
 Auditory feedback, i.e.,
 feedback of one’s own speech
 in the auditory modality
 is necessary for the control
 of fluent speech. 
 This is evident by the Lee
 effect: Delayed auditory
 feedback causes prolonged and
 disfluent speech in normally
 fluent individuals.
 (e.g., Lee, 1950)*
 Developmental stuttering is
 caused by insufficient inte-
 gration of auditory feedback
 in the control of speech, or,
 in other words, by a dissocia-
 tion of speech production
 from speech perception.

 Suggested by weaker or de-
 layed responses to manipula-
 tions of auditory feedback,
 (e.g., Cai et al., 2012, 2014;
 Daliri et al., 2018; Loucks
 et al., 2012; Tourville et al.,
 2013) and by deactivation
 of secondary auditory cortex
 areas during speech, (e.g.,
 Brown et al., 2005; Budde et
 al., 2014; Fox et al., 1996;
 Ingham et a., 2003; Stager et
 al., 2003; Toyomura et al., 2011) 
 Auditory feedback
 can be provided in two ways:
 via an external and via an
 internal feedback loop. The
 external loop is what is usually
 called auditory feedback. 
 Perceptual Loop Model
 (Levelt, 1989) which is part
 of the standard model of
 speech production.

 During silent mouthing and
 during speaking under com-
 plete auditory masking, one’s
 own voice is heard via
  the internal feedback loop
 just as during inner speech
 (silent reading, verbal thinking).
 This has been assumed and
 was presupposed in experi-
 ments by several experts (e.g.,
 Brocklehurst & Corley, 2011;
 Lackner & Tuller, 1979;
 Oppenheim & Dell, 2010,
 2008; Postma & Kolk, 1993)

 The internal feedback loop
 closely connects speech
 production with speech
 perception in the brain.
 e.g., Smith, Wilson, & Reisberg 
 (1995), Smith, Reisberg, &
 Wilson (1992), Tian & Poeppel
 (2010, 2012)

 PWS are fluent during
 mouthing and under auditory
 masking not because they do
 not hear themselves speak.
 According to (2), they hear
 themselves speak internally
 via the internal feedback loop.

 PWS are fluent during mouth-
 ing and under complete audi-
 tory masking because of
 the close connection between
 speech production and speech
 perception when they use the
  internal feedback loop.
 This is only the inversion of the
 hypothesis (2): Sufficient inte-
 gration of auditory feedback
  in spech control saves from

 The internal feedback loop
 (in humans generally) works
 only if external auditory feed-
 back is not provided, i.e.,
 when one’s own voice is
 not heard externally.

 Direct evidence: Smith,
 Reisberg, & Wilson (1992)
 Indirect evidence: We would
 hear ourselves doubly
 because internal and external
 feedback take different time
 (Lackner & Tuller, 1979;
 Vigliocco & Hartsuiker, 2002).
 Detection of speech errors
 prior to articulation is no
 evidence that both feedback
 loops work concurrently.

 No monitoring takes place via
 the internal loop during overt
 speech(Huettig & Hartsuiker,
 2010). Pre-articuiatory error-
 detection seems to be part of a
 general (not speech-specific)
 production-based conflict
 monitoring mechanism
 (Nozari, Dell, & Schwartz,
 2011;Ganushchak & Schiller,
 2008; Ries et al., 2011;
 Trewartha & Philips, 2013)
 The State Feedback Control
 model (Hickok, Houde, &
 Rong, 2011; Hickok, 2012)
 which implies pre-articulatory
 monitoring of internal auditory
 feedback during overt speech
  is incorrect.
 The model is inconsistent
 with the Lee effect: Delayed
 external auditory feedback
 would not affect speech
 fluency if external auditory
 feedback was irrelevant for
 speech control (as the model
 The internal feedback loop
 cannot compensate for a
 possible deficit in the external
 feedback loop in normal condi-
 tions, i.e., when one’s own
 voice is externally heard.
 This follows from (8):
 Both the feedback loops
 do not work concurrently,
 thus, one cannot compensate
 for a deficit in the other one.

 The fact that stuttering dis-
 appears when no external
 auditory feedback is available
 does not contradict the hypo-
 thesis that poor involvement
 of external auditory feedback
 in speech control causes
 This follows from (3) and (8).

*) Further evidence comes from audio-phonatory coupling, i.e., the fact that the duration of long syllables is controlled by the auditory feedback of the syllable start (Kalveram & Jäncke, 1989). I have not mentioned this in the paper because the Lee effect (which is more famous) is a manifestation of audio-phonatory coupling.

to the top

previous page