2003 IFA Congress: Montreal, Canada

A Learning Theory Model of Stuttering

Volker Urban
Helmstedter Str. 14, D-38102 Braunschweig, Germany


Every stutterer has the ability to speak fluently. Why doesn’t a spontaneous learning process result in the stutterer using this ability all the time? The model presented here emerged from the following observation: When a stutterer interrupts a stuttering event, A he will experience fear and an urge to start speaking again immediately. This urge forms the basis of a new way of understanding stuttering. Various vicious cycles will result in the development of fear in the pause and the avoidance thereof rewards the stuttering.

  1. Introduction
The model presented in the following emerged out of some simple observations made during

the therapy of stuttering:
  1. When a stuttering person is afraid that he will start stuttering while saying a certain word he often finds it very hard to wait for a few seconds before trying to say the word. The reason for this behaviour is that while making such a pause the stutterer’s expectancy and fear of stuttering often increase to an astounding degree.
  2. This fear generates an intense urge to immediately start the attempt of speaking even though the stutterer is expecting further on that this attempt will lead into a stuttering event. He paradoxically has the feeling that this is the only way that will enable him to keep the intensity of the impending stuttering event within a limit.
  3. If he regardless of his urge to speak still decides to wait for a few seconds he often experiences the sudden feeling that he is now able to say the word fluently, and subsequently he really succeeds in doing so.

There is some evidence in the existing literature for the two former observations; as far as I know the third one is new. In my article “A new proposal for therapy: the Empty Pause” (Urban, this volume) I have described the observations in more detail and references are provided.
The observations made support Sheehan’s (1953, p. 36) hypothesis that “the occurrence of stuttering brings about a reduction of the fear that elicited it”. His “fear reduction hypothesis” can be specified as follows: By quickly starting the attempt of speaking the stutterer avoids the fear of stuttering which would increase if he postponed his attempt of speaking.

Provided this hypothesis is true, the anticipation of a stuttering event does not motivate the stutterer to approach his attempt of speaking slowly, even though one might expect that this would be the natural reaction. Instead he experiences a need to quickly start the utterance.
There is an indication that the person who stutters actually accelerates his efforts to speak during stuttering. Niermann et al. (1994) und Van Ark et al. (this congress) compared duration characteristics of single-syllable whole-word repetitions and part-word-repetitions in the speech of preschool children who stutter to those of control nonstuttering children. Both investigations show that children who stutter produce shorter silent intervals between the spoken segments, while the spoken segments are longer (Van Ark et al.) or similar (Niermann et al.).
Since the stutterer feels that his need to start speaking quickly is dominant and that he can hardly influence it with his willpower, the term “urge” is used. The phenomenon can be seen as comparable to the compulsive disorder. Those who are affected may check 20 times to make sure that they have really locked their door when leaving the house. The urge to do so can hardly be suppressed. If an affected person still tries to suppress this urge he experiences an increase in anxiety and fear.
It is plausible that the attempt to start speaking quickly, can lead to problems in the preparation and realisation of the articulation. Thus the urge to speak possibly becomes the cause of the stuttering event.
The following argumentation builds on the presumption that the urge to start speaking can not only be triggered by the conscious fear to start stuttering. Properties of the word or the situation that remind the stutterer of previous stuttering events, as well as a disfluency that has already begun are possible triggers, too. The latter assumption is supported by the clinical experience that an intentionally produced “pseudo-stuttering” often switches into a real stuttering event. Further reasons for this assumption are listed in my article about the “Empty Pause” (Urban, this volume).

lt f1

Figure 1

It is obvious that a person who has developed this pattern of reaction cannot get rid of it again easily. Every stuttering event would increase the risk that further stuttering events would be triggered. The mentioned article about the Empty Pause contains some suggestions how to empirically test this model.

The question which learning processes could maintain the pattern of reaction discussed here arises. Every stutterer pronounces a lot of words fluently. Seen from a behavioural point of View this means: The normal - not accelerated - preparation and realisation of the articulation will be rewarded every time by fluent speech, whereas attempts of speaking which are accelerated by the urge to speak will end in a stuttering event. What kind of learning process causes the stuttering events to still take place?

  1. A Theoretical Model
The development of the model presented in the following was stimulated by the theory of autopoietic systems which was originally developed by Maturana and. Varela (1980) in order to describe the functioning of biological ce'.ls. The model describes that the reaction pattern showed in Figure 1 (urge to quickly start speaking) can be seen as a small “self-reproducing system” that first generates the conditions which then form the prerequisites for its renewal.
  1. If the urge to speak is already existent as a learnt reaction some vicious circles arise. Their effect is that waiting before an attempt of speaking (the pause) causes a higher degree of fear than the attempt of speaking itself though this attempt leads into a stuttering event.
  1. Since the pause causes more fear than the attempt of speaking this attempt is reinforced by a reduction of fear which weighs more than the punishment by the beginning stuttering event. Thus the urge to quickly start the attempt of speaking is strengthened.

In the following this cycle will be described in detail. Then some suggestions for the empirical validation of the model will be presented. 
Figure 2 shows the possibilities the stutterer has in the moment that he fears to stutter or has already started stuttering. So we assume that the urge to quickly start the attempt of speaking has been triggered.

lt f2

Figure 2

Now the stutterer has the choice to start his attempt of speaking immediately (2) or to postpone it a little bit (1) (he finds himself in the conflict between speaking and not-speaking which is described e. g. by Sheehan, 1953). If he starts he gets into stuttering just because - as we assume - the urge to speak is already activated (5). Occasionally, however, he will hesitate, possibly because he hopes to find another word or because he simply wants to postpone the stuttering for another moment. The pause, however, impends to lead to an impatient response of the listener (3), the stutterer gets under time pressure. He feels helpless, his fear of stuttering increases (4).

Now a vicious circle arises. Given the hypothesis shown in fig. l is correct, fear of stuttering is one of the possible triggers of the ‘urge to start speaking. That is why this urge is increasing as well (7). It can be assumed that the stutterer knows from experience that his chance to get to a fluent utterance is sma 1 in particular if ae feels tne urge to start speaking quickly. So reactively his fear not to be able to avoid stuttering when saying the intended word increases (4).
Now the stutterer does not see any alternative but to start his attempt of speaking and to get into stuttering. At this moment, unfortunately, another vicious circle can occur. If the starting hypothesis is correct a stuttering event that has already begun triggers an intensification of the urge to speak as well (8). And if this urge increases the stuttering event is extended (5). This vicious circle can make the stuttering event more intense than the stutterer has expected it to be, even more intense than he has experienced it ever before with a similar word or with a similar situation.

Now though there is a way out. All continuing activities are slowed down due to a reactive inhibition after a while (6) (Hull, 1951). Presumably this is the reason why during stuttering the urge is alleviated so much that eventually the normal preparation of the articulation is possible and the utterance turns out to be successful (9).

During the pause, however, the reactive inhibition does not reduce the urge to speak because there is no action that is actually carried out. So, in the end the attempt of speaking brings a better result than the pause.

Up to this point the argumentation depends on the assumption that the urge to speak is already established as a learnt reaction which is triggered by the stimuli “stuttering” and “fear to stutter”. It was shown that on this condition, the pause causes a higher degree of fear than stuttering. In the following it is to be demonstrated that that very fact - the pause causes a higher degree of fear than stuttering - retroactively forms the basis for a learning process that reinforces the urge to quickly start speaking.

lt f3

Figure 3

Since the stutterer feels uncomfortable about his stuttering, he occasionally postpones the utterance by some seconds. He then experiences an increase of his fear to stutter.

lt f4

Figure 4

When the stutterer switches to an attempt of speaking after a pause this attempt will lead to a stuttering event (because - as we assume - the urge to speak causes an acceleration of this attempt). Since the stuttering event causes less fear than the pause, the attempt of speaking will be reinforced by a reduction of fear. Thus the stutterer establishes the attempt of speaking as an escape behaviour.

lt f5
Figure 5

After some experiences like that the pause will be left out and the attempt of speaking becomes the instant response on a stuttering event resp. the fear of stuttering. The escape behaviour has changed into an avoidance behaviour. This transformation leads to fatal consequences for the stutterer regarding two aspects:

  1. Avoidance behaviour has top priority for biological reasons. There is a tendency towards it following faster and faster on the triggering stimuli (Solomon et al., 1953); by generalization to preceding stimuli all characteristics of words and situations which remind the stutterer of previous stuttering become triggering stimuli of the urge to speak (cp. fig. 1). It is no longer necessary that the stutterer develops conscious anticipation and fear of the single stuttering event. The normal preparation and realisation of the articulation does not get a chance anymore as soon as a suitable trigger of the quick attempt of speaking exists, even if just in a slight extent.
  2. An avoidance behaviour is resistant to extinction in an extraordinary way (Seligman & Johnston, 1973). It is enough if the stutterer tries to make a pause once per every 500 stuttering events and again finds that he is experiencing an increase of fear (resistance to extinction with human beings: Birbaumer 1977).

lt f6

Figure 6

Nevertheless, it is a fact that the success of the normal preparation and realisation of articulation is more effective: It leads to fluent speech. Why doesn’t the better result cause the normal articulation to prevail over the accelerated attempt of speaking? The solution of this problem should be searched for in the following context: The positive results of the attempt to speak are more powerful than the ones of fluency because they -happen at moments of higher agitation. The normal articulation only comes about if word and situation are reminding the stutterer of a former stuttering event not at all or only in a slight extent. In cases like that the stutterer is likely to be rather calm. As soon as the urge to speak is triggered though, there is a possibility that the above-mentioned vicious circles drive up the agitation to a very high level. If the stutterer then tries a pause and re-closes it by an attempt of speaking, a powerful reinforcement for this attempt takes place.
Brutten and Shoemaker (1957) applied the two-factor-theory of Mowrer (1950) to stuttering. They hypothesise that autonomic arousal which arises in stress situations is classically conditioned to certain triggering stimuli, and is responsible for the disruptions in speech.

Instrumentally conditioned escape- and avoidance reactions would prevent extinction to take place. There are two problems this theory cannot resolve: 1. Stuttering is not connected to emotional arousal at all times. 2. The person who stutters is by far not able to avoid all stuttering events; therefore extinction processes should arise. - The model presented here offers the following solutions to these problems: 1. A stuttering event is not directly caused by emotional stress but by a learnt urge to accelerate the pronunciation. This acceleration can be triggered by adequate stimuli before an emotional arousal occurs. 2. The learnt avoidance reaction can undoubtedly be weakened by learning processes; nevertheless it will be re-strengthened by the vicious circles described above.

This course of events could be described as a “competition of reinforcements”. The normal articulation very frequently receives reinforcements but their intensity is low. The quick attempt of speaking is rewarded rarely but the intensity of these rewards is noticeably higher. The interplay of these reinforcing processes could be the reason for the fact that the intensity of stuttering can fluctuate considerably over time. In times of low strain successes of fluent speech prevail and the intensity of stuttering decreases. When characteristics of the situation cause a higher strain the escalating processes described above can raise the intensity of stuttering very quickly.

If fluent utterances are rewarded and stuttering events are punished the intensity of stuttering is lowered (Onslow 1996). Within the context of the model described above these findings can be interpreted as follows: The balance between the rewarding processes is shifted: As mentioned above the quick attempt of speaking gives rise to a reduction resp. avoidance of fear, which has a greater effect than the punishment caused by the emerging stuttering event. The therapeutic use of time-outs as a consequence to stuttering causes the aversive aspect of stuttering to weigh more than the reduction of fear; so it decreases the success of the quick attempts of speaking. Reinforcing the fluent utterances raises the emotional value of the successes of normal articulation.

It can be seen that the model as a whole describes a great vicious circle existing above the small ones described above that prevail during the pause and during stuttering. At the outset we assumed that the urge to speak already exists as an established response. Only if this is the case the small vicious circles that create the conditions of reinforcing the urge are able to arise. If there is no urge to speak - as it is the case with a non-stuttering person - no reinforcements for the quick attempt of speaking exist. So the model can only describe the maintenance of stuttering but not its origin.

However, as soon as the urge to quickly start speaking exists, the stuttering reproduces from generation to generation so to speak. Certain characteristics of a situation trigger the urge to speak; the resulting stuttering event retroactively connects the characteristics of the present situation with the experience of stuttering and subsequently these characteristics become possible triggers for future stuttering. 48 Theory, research and therapy in fluency disorders

  1. Empirical Validation
How can empirical proof for this model be found? One would have to split it up into its single components and look for proof for every single one of them. The following questions would have to be examined:
  1. Does the fear to stutter increase if the stutterer pauses instead of starting to speak?
  2. Does the urge to make an attempt to speak exist?
  3. Is the urge to speak triggered by the following stimuli: fear to stutter, a stuttering event that has already begun, and situation characteristics that make the stutterer remember former stuttering?
  4. Is there a vicious circle between the urge to speak and stuttering during stuttering? a) Does the urge to speak lead to stuttering? b) Does the intensification of a stuttering event lead to an intensification of the urge to speak?
  5. Is there a vicious circle between the urge to speak and the fear of stuttering during the pause? a) Does the fear to stutter trigger the urge to speak? b) Does an intensification of the urge to speak increase the stutterers fear to stutter?
  6. Do stutterers spontaneously make a pause instead of giving in to the urge to speak? The avoidance behaviour learning described above can not take place if such pauses do not occur.
In my article on the “Empty Pause” (Urban, this volume) some experiments that could give answers to the questions # l - 3, are described. Due to space constraints only a few proposals for experiments will be listed in the following, which could give answers to the questions # 4 - 6. This list will merely demonstrate that empirical test of the model is possible.
For question # 4: Is there a vicious circle during stuttering? General proof
  1. Search through imaging techniques for brain activation patterns which are typical for avoidance reactions (comparison group: compulsive disorder patients) and occur during stuttering.
  2. Persons who stutter get the instruction to try to pause for two seconds at certain moments. One could measure if pauses are shortened if current stuttering events have to be interrupted by a pause, and if the extend of the shortening correlates to the intensity of stuttering.

Does a stuttering event that has already begun, lead to an intensification of the urge to speak?

The way stutterer perceive their own stuttering when they stutter, is manipulated. Therefore, they get the task to read a list of words repeatedly; when a signal is given a word has to be said afterwards. In the course of the repeated cycles the intensity of stuttering is lowered by the adaptation effect. The utterances are fed back by headphones. Afterwards, the record of the first cycle (with more intense stuttering) is played to the stutterers. One might assume that this triggers more intense stuttering.

For question # 5: _Is there a vicious circle during the pause?
Persons who stutter are asked to rate the words of a text concerning the degree of their fear of stuttering when reading these words. When reading aloud afterwards they are to pause for two seconds before they try to say the critical words; adherence to this time allowed is controlled. Presumption: if the fear of stuttering is low, these pauses will lead to an improvement of the fluency; if the fear is high, the pauses will cause a worsening (compared to cycles: reading without pauses). This presumption is based on the hypothesis that a response prevention takes place which is successful resp. not successful (cp. my article on therapy, this congress). - A deterioration in case of a high level of fear to stutter would indicate that a vicious circle has taken place.

The author would like to thank the following people for the discussion of earlier versions of this work: Andreas Starke, Hamburg (Germany), Prof. Hans-Georg Bosshardt, Ruhr-Universitat Bochum (Germany), Prof. Dirk Vorberg und Dr. Jens Schwarzbach (both of them: Technische Universitéit Braunschweig, Germany), Dr. Albrecht Schumacher, Hohenroth (Germany), and Prof. Wolfgang Wendlandt, Alice-Salomon-Fachhochschule Berlin (Germany).

Birbaumer, N. (1977). Vermeidungslernen: Grundlage der Bestandigkeit von ,,Angstreaktionen”. In N. Birbaumer, Psychophysiologie der Angst. Miinchen: Urban & Schwarzenberg.

Brutten, G. J. & Shoemaker, D.J. (1957). A two-factor learning theory of stuttering. In L. E. Travis

(Ed.), Handbook of Speech Pathology (pp. 1035 - 1072). New York: Appleton-Century- Crafts.

Hull, C. L. (1951). Essentials of Behavior. New Haven: Yale University Press.

Maturana, H. E. & Varela, F. J. ( 1980). Autopoiesis and Cognition. Boston, MA: Reidel.

Mowrer, O. H. (1950). Learning theory and personality dynamics. New York: Ronald.

Niermann Throneburg, R. & Yairi, E. (1994). Temporal dynamics of repetitions during the early stage of childhood stuttering: An acoustic study. Journal of Speech and Hearing Research, 37, 1067 _ 1075. '

Onslow, M. (1996). Behavioral management of stuttering. San Diego, CA: Singular Publishing Group.

Seligman, M. E. P. & Johnston, J. C. (1973). A cognitive theory of avoidance learning. In F. J. McGuigan & D. B. Lumsden, (Eds.) Contemporary approaches to conditioning and learning (pp. 69 - 110). Washington, D.C.: Winston-Wiley

Sheehan, J. G. (1953). Theory and treatment of stuttering as an approach-avoidance conflict. The Journal of Psychology, 36, 27 - 49.

Solomon, R. L., Kamin, L. J. & Wynne, L. C. (1953). Traumatic avoidance learning: the outcomes of several extinction procedures with dogs. Journal of Abnormal and Social Psychology, 48, 291 - 302

Urban, V. (this volume). A new proposal for therapy: the “Empty Pause

Van Ark, M., Sandrieser, P., N atke, U., Pietrowsky, R. & Kalveram, K. T. (this volume). Characteristic features of single-syllable word repetitions in preschool children who stutter and controls.

join button


The IFA implemented Japanese translations of some pages on the site for the 2018 Joint World Congress. Choosing Japanese below to see these translations.

Not all pages are translated, but you can use Google translate to see a machine translation using the switch below

Google Translate


Journal of Fluency DisordersBrowse the current issue

The official journal of the International Fluency Association
IFA Members receive online access to JFD as a member benefit.

Read more: JFD

Twitter Feed @official_IFA

IFA on Facebook