In more erudite, academic language, I have formulated some specific hypotheses based on research, clinical experience and intuition, and I invite all of you to respond.

I know of no one in the field who will argue that stuttering is a homogeneous disorder -- i.e. that it is a single entity with a single sufficient or necessary cause, or even that all stutterers suffer from the same underlying problem. Many authorities (including myself) have cited, acknowledged, and discussed evidence for heterogeneity in stuttering. I have found no paper challenging this evidence, or arguing the converse — that all stuttering is the same, and that all persons who stutter suffer from the same underlying problem.

However in spite of the general acceptance of "heterogeneity," most research makes little or NO provision for the explicit consequences of heterogeneity. Our research is seldom designed to address the problems inherent in studying a heterogeneous population. We continue to look for differences between persons who stutter (PWS) and those who do not, ignoring differences between PWS. We continue to evaluate stuttering therapies — pharmacological, behavioral, etc. — in terms of effectiveness for ALL PWS.

If stuttering arises from different etiologies, we cannot expect ALL PWS to differ from non-stutterers on the same variable(s). Nor can we reasonably expect ALL PWS to respond equally to the same therapy, whether that therapy is a drug or a behavioral technique.

Until our research specifically addresses heterogeneity in stuttering, we will fail to make significant progress toward understanding underlying causes or identifying the most appropriate and effective treatment(s) for each individual who stutters.

An epiphany in my understanding of stuttering occurred in 1973 when I heard Martin Adams expound on the fact that we do not have a basic understanding of the nature and measurement of "fluency." As he pointed out, all other speech/language disorders are defined by how they differ from normal. In the absence of knowledge of "normal performance" in fluency, we cannot approach fluency disorders as we do other communication disorders.

After a quarter of a century, our understanding of fluency has not progressed significantly. Therefore, it is not surprising that we still do not have a cogent, comprehensive, generally accepted definition of stuttering. Nor do we have any objective, comprehensive or generally accepted method of measuring fluency.

Martin Adams was correct then, and he is correct now. Until we understand fluent speech production, we will not understand disfluent speech production or stuttering. Until we can measure fluency, we will not be able to objectively and accurately measure disfluency.

This question (which is part of the problem of understanding normal fluency) has been explored in the writings of Gloria Borden and others. It is critical to the study and understanding of stuttering. From the time of Johnson, the concept of stuttering as an outgrowth or exacerbation of normal developmental disfluency — of a developmental continuum from normal fluency to stuttering — has been almost universally accepted and used as a basis for research and treatment. However, little empirical support can be found for this single continuum model. The recent developmental research of Ehud Yairii and his colleagues has shaken the most basic premises on which this model was originally based.

However, the single continuum model has not been subjected to explicit or rigorous investigation. If this model is false, then research based in our adherence to its premises is basically flawed. It is imperative that the premises upon which the single continuum model is based be made explicit, and these premises subjected to experimental study.

I believe that the biggest impediment to the design of research appropriate to studying stuttering as an heterogeneous disorder is the lack of explicit models or hypotheses regarding types or sub-groups. It is easy to say that ALL PWS are not the same, but difficult to determine the nature of potential sub-groups. If this paper has any value, it will be in the explicit identification of potential sub-groups..

Severity as a Grouping variable — The only variable to have received substantive attention as related to heterogeneity is severity. A number of studies have used severity to sub-group subjects, and/or have used severity in correlational analyses. By and large, results of such analyses have been mixed. There is some evidence that mild stutterers differ from severe stutters on physiological, neurophysiological, and behavioral measures.

However, severity, based on behavioral measures, is subject to a number of problems, including type, extent, and success of previous therapy. It appears that some etiological sub-groups may have more severe symptoms, may be more resistant to treatment and more prone to relapse. Thus, severity may be related to research findings, but this relationship could result from differences between etiological sub-groups.

Van Riper’s Four Tracks -- Perhaps the most intriguing sub-grouping of PWS can be traced to Van Riper (1971 pgs. 102-117). On the basis of his analyses of 300 cases of developmental stuttering he described FOUR TRACKS in the development of stuttering. In specifying these FOUR TRACKS, Van Riper departed from the "accepted wisdom" that ALL stuttering arose from normal disfluency and followed the same developmental course. This departure from the traditional view may be why this research by one of the most respected members of our profession has NOT been the basis for subsequent research. This is perhaps the single most significant neglected area for research in the field of stuttering. After almost 30 years, the hypotheses inherent in this sub-grouping remain uninvestigataed.

According to Van Riper, just under 50% of his subjects fell into TRACT I, and this sizable sub-group conforms very closely to the "traditional" developmental pattern for stuttering. Thus, the developmental pattern of the majority came to be viewed as THE DEVELOPMENTAL PATTERN OF STUTTERING. The remaining 50% of Van Riper’s subjects would be considered atypical or anomalous deviations from the norm. However it is entirely feasible that this 50% represent not deviations from established norms, but different sub-groups of PWS.

We will not review Van Riper’s Four Tracks, but would point out that they are based on some very important differences, including:

1. Developmental patterns of speech and language — age of first words, sentences, and the presence or absence of articulation errors.
2. Relationship of onset of stuttering to early speech/language — whether stuttering began with first speech or after the establishment of fluent speech.
3. Gradual or sudden onset of stuttering, including the presence or absence of tension, struggle, tremors, laryngeal blocking, etc. at or shortly after onset.
4. Awareness and development of fears and avoidance.
5. Remissions and Variability.
6. Severity (not all tracks progressed to severe, but rather remained mild or moderate).

While I have never investigated Van Riper’s tracks, I have consistently used these as a guide in differential diagnosis in clinical practice. My own hypotheses regarding the types or sub-groups of stuttering are predicated on Van Riper’s observations. However, before stating hypotheses, I would review the Model of Fluent Speech Production within which they are formulated.

The model of fluent speech production on which these hypotheses are based was briefly mentioned in 1994 (Watson, Freeman, Devous, Chapman, Finitzo, and Pool), described in detail in 1995 (Freeman), and discussed in 1997 (Watson and Freeman). Briefly summarized from this last publication, we state:

As Wingate (1988, pg. 267) reminded us, "Stuttering is not simply a problem of words per se, but of words as the pivotal elements in a system that can transduce ideas and thoughts into an audible code — speech."

As this model of fluency relates to models of stuttering, two comments are worthwhile. First, we agree with Kent that stuttering represents a failure in temporal processing — specifically a failure in the temporal integration of the basic processes of fluent speech generation. Second, we agree with Starkweather and others who have described stuttering in a capacities and demands model. Stuttering does represent an imbalance between the capacities of the fluency generating system and the demands of the situation. Within our systems model of fluency generation, a number of "types" of stuttering may be hypothesized as related to Cognitive, Linguistic or Speech Motor Processes.

At the level of Speech Motor Process, I believe there are at least two distinctly different etiological sub-groups. The first I call Dyspraxic Stuttering and the second I call Respiratory/Laryngeal Control Stuttering.

Hypothesis 1 — There is a sub-group of stutterers whose fluency problem is etiologically linked to developmental apraxia of speech.

Dyspraxic stuttering is characterized by the presence of developmental apraxia of speech, with co-occurring phonological and fluency problems. This type of stuttering is typically associated with delays in the appearance of intelligible words and sentences, with phonological and/or articulation problems, and with slow speech rates Other symptoms or expression of developmental apraxia, including oral and manual apraxia, may be present.

In older children and adults, problems in pronunciation of longer words, and inconsistent articulation errors, including consonant cluster reduction occur. Stuttering is most likely to occur on longer and/or unfamiliar words. In more severe problems of developmental dyspraxia, stuttering begins with the earliest words, and is present as soon as the child begins to string together longer sequences such as two and three syllable utterances. Many of these children fall into Van Riper’s Track II.

In its more severe form, this type is easy to spot, even in older children and adults. In cases of very mild dyspraxia, it may be more difficult to recognize. Since speech is the most complex sequential motor programming act in which humans regularly engage, stuttering may represent the predominant symptom of a deficit in temporal sequential motor programming. In such cases, the primary difficulty is not in producing accurate articulatory gestures, but in coordinating multiple systems, i.e. respiration, phonation, and articulation. These children must speak slowly in order to succeed. Any time pressure results in breakdown in temporal coordination.

Stutterers in this sub-group tend to experience an exacerbation of stuttering symptoms after consuming alcohol. Unlike some stutterers, whose stuttering gets worse when they consciously "try to speak more fluently," stutterers in this sub-group speak more fluently when they concentrate on trying to speak more fluently. Also, unlike other stutterers, they may report being less fluent when around good friends and family, and more fluent when talking with strangers or authority figures.

Hypothesis 2 — There is a sub-group of stutterers whose fluency problems are directly linked to difficulties in voluntary control of the muscles of respiration and/or phonation.

The respiratory and laryngeal systems are part of the respiratory system, which evolved to support a life-sustaining function, not speech. We are learning more about problems of vocal fold movement dysfunction through new studies of asthma and allergies. This research is currently impacting voice disorders.

For over 50 years research has indicated that children who stutter have more problems of asthma, allergies, and upper respiratory distress than non-stuttering children. I believe that a sub-group of stuttering is etiologically related to difficulties in voluntary control of the muscles of respiration and/or phonation.

Respiratory/Phonatory Control stuttering is characterized by blocks (i.e. complete arrest of speech) even in its earliest manifestations. It is most likely to occur in individuals or in families with a history of asthma and respiratory allergies. Momentary losses of voluntary control of respiratory/laryngeal function is the basic problem out of which stuttering develops.

Young children who demonstrate a rapid onset of symptoms with tension, effort, tremors, and struggle behaviors present at or near onset are likely to belong to this sub-group. Their disfluencies include complete blocks and/or unvoiced prolongations. The frequency, severity, and duration of respiratory/laryngeal control failures, as well as the child’s reactions/responses to these lapses, determines the severity of the associated stuttering.

While this sub-group may constitute only a small percentage of young children who stutter, they constitute a large percentage of adults who stutter severely. Their problem is resistant to treatment, and prone to relapse.

Just as asthmatic and allergic reactions are exacerbated by emotional reactions, respiratory/phonatory control problems appear to be highly related to affective arousal. The initial onset of stuttering in this sub-group is frequently associated with some traumatic or frightening event. In many respects, this sub-group appears to fit Van Riper’s Track III. This type of stutterer is frequently seen by clinicians who specialize in treating severe, complex, or treatment resistant stuttering.

It is possible that stuttering associated with Tourette’s Syndrome may be a special case of respiratory/phonatory control disruption.

I believe that for many stutterers, linguistic processing problems are the primary etiologic factor. Many of these children demonstrate deviant or delayed language development. Others fall within normal limits on measures of language development, but evidence specific "gaps" or wide discrepancies between their linguistic abilities and their cognitive or speech-motor abilities. Much work done at Northwestern University has focused on the role of linguistic complexity and word-retrieval problems in stuttering.

I have called this general category Linguistic Stuttering. (I was tempted to call it Developmental Aphasic Stuttering to parallel the Developmental Apraxic Stuttering, but decided this would both date me, and be politically incorrect.) I have further attempted to specify at least three sub-types of linguistic stuttering.

Hypothesis 3 — There is a sub-group in which the onset of stuttering is directly linked to developmental delays in aspects of linguistic processing.

This is probably the largest sub-group of young children who stutter. This group also has the largest number of individuals who "out grow" or spontaneously recover. They are represented in much smaller percentages among adults who stutter. As adults, they tend to cluster in the mild or moderate severity range. Only a very small number of this sub-group manifest severe stuttering as children or adults.

In most respects, this is the sub-group of stutterers described by Van Riper as Track 1. These are also the stutterers whose developmental pattern best fits Bloodstein’s developmental stuttering profile. These are the stutterers most frequently treated by clinicians who specialize in young children. Many early treatment programs are well suited to their needs. Many, but not all children in this group, demonstrate other linguistic problems, including reading, writing, and spelling difficulties.

Hypothesis 3A -- There is a sub-group of stutterers in which stuttering is directly linked to problems in word finding or word retrieval.

In non-fluent aphasic patients similar word-retrieval problems are labeled "anomia" (although nouns are not the only words affected). It is clear that inability to rapidly access the desired word will lead to a breakdown in the generation of fluent speech. Fluency breakdowns related to word finding problems are most typically marked by inappropriate pauses, fillers, interjections, repetitions of preceding words (delaying). Many so-called normal-type disfluencies, as well as phrase and word repetitions are characteristic of this type of stuttering.

Hypothesis 3B — There is a sub-group of stutterers in which the onset of stuttering is directly linked to problems in generation of complex grammatical forms.

Onset of stuttering is temporally linked to the child’s attempts to generate more complex grammatical sentences. Stuttering will occur on complex sentences. Simple sentences will be relatively fluent.

Hypothesis 4 — There is a sub-group of stutterers in which stuttering is directly linked to problems in auditory processing and/or auditory sequential memory.

The ability to accurately reproduce complex sequential patterns is directly related to the ability to process and remember complex sequential patterns. Failures in auditory processing and/or in auditory memory, will lead to inadequately represented patterns. Inadequately represented patterns will lead to breakdowns in word retrieval and/or to inadequate information for generation of sequential patterns for speech production. Stuttering co-occurs with auditory processing and/or auditory sequential memory problems.

While disruptions or disorders of cognitive processing will lead to disfluent speech production, I do not believe that these problems are a primary cause of stuttering. The speech of those with cognitive disorders is frequently, even typically highly disfluent, but these disfluencies do not typically develop into stuttering. When stuttering does occur in individuals with cognitive disorders, the problem is more likely to lie in co-occurring linguistic and/or speech motor problems.

In other words, I believe that stuttering is most likely to develop when an individual’s cognitive abilities are superior to their linguistic and/or speech motor abilities. Limited cognitive ability in isolation will lead to disfluent speech, but not typically stuttering.

Obviously, the existence of a problem at one level in the fluent speech generating system does not preclude the existence of a problem at another level. Indeed, the existence of a problem at one level probably increases the potential for occurrence of problems at other levels. For example, developmental apraxia may occur in isolation, but is likely to occur in conjunction with developmental linguistic problems. Most pervasive cognitive disorders co-occur with linguistic and motor control disorders. This "real world" phenomenon complicates differential diagnosis for most communication disorders. The situation is no different in stuttering. However, the existence of co-occurring problems may impede, but should not preclude our search for etiological sub-groups.

While I have seen two individuals who exhibited full-blown anxiety/panic attacks in which stuttering "blocks" were associated with the inability to move. I do not believe these constitute a separate etiological sub-group. I believe the panic attacks post-dated rather than proceeded stuttering.

Anxiety certainly plays a role in stuttering, especially in older children and adults. Anxiety is apparently a critical modulating factor with respect to stuttering severity. Patients whose stuttering markedly improves after drinking a modest amount of alcohol, or after taking Valium, tend to demonstrate high anxiety levels. In spite of all this, I do not believe there is a significant sub-group of stutterers whose problem is etiologically related to anxiety.

I have no personal experience with stuttering associated with compulsive disorders. If stuttering can be an expression of a compulsive disorder, I suspect that it represents a very small number of individuals.

Similarly, when I have seen PWS with depression, the two disorders did not appear to be etiologically linked. In most cases, depression had its first expression in adult life while stuttering was developmental.

Like others, I have seen adult-onset stuttering in patients with stroke, closed head injury, and Parkinson’s Disease. I have also seen adult-onset stuttering in one patient after laryngectomy and in at least two subjects being treated with psychotropic medications.

In many respects, these hypotheses are not radical. All have been suggested by other authors. However, I believe (those scary words again) each is testable, and each is worthy of testing.

Freeman, Frances J. Fluency, Disfluency, Dysflulency, Nonfluency, Stuttering: Integrating Theories. Chapter 27 in Bell-Berti, Fredericka & Raphael, Lawrence J., Eds. Producing Speech: Contemporary Issues for Katherine Safford Harris. American Institute of Physics, N.Y. 1995.

Van Riper, Charles, The Nature of Stuttering, Prentice-Hall, Inc., Englewood Cliffs, N.J.,1971.

Watson, B.C., Freeman, F.J., Devous, M.D., Sr., Chapman, S.B., Finitzo, T., & Pool, K.D. Linguistic performance and regional cerebral blood flow in persons who stutter. Journal of Speech and Hearing Research. 37: 1221-1228. 1994.

Watson, Ben C. & Freeman, Frances J., Brain Imaging Contributions, Chaper 7 in Curlee, Richard F. & Siegel, Gerald M. Nature and Treatment of Stuttering: New Directions. Allyn and Bacon, 2nd Ed. 1997.