What Happens When We Stutter?

A new technique, real time magnetic resonance imaging, enables us to observe the vocal tract during the moment of stuttering. I recently took part in such a study in Professor Shri Narayanan‘s Speech Production and Articulation Knowledge group at the University of Southern California‘s Dynamic Imaging Centre. This was part of the doctoral dissertation project of researcher Yijing Lu.

Here I am stuttering in the scanner.

Articulation requires the coordination of many muscles shaping air flow through the vocal tract. Stuttering occurs when there is excess tension in one or more of those muscles. Such excess tension can be clonic or tonic, having a corresponding spoken manifestation as repetitions or prolongations. For example, in clonic tension muscles are contracted and relaxed rhythmically. Thus clonic tension underlies the repeated sounds, typically consonant sounds, which are a hallmark of s-s-stuttering.

Much of the stuttering I am doing in the video is clonic. This was not on purpose. During the recording session, I deliberately did nothing when I anticipated a moment of stuttering, and clonic stuttering was the result. It was interesting to stutter in this way. I am not sure that I have ever historically stuttered in exactly this manner – prior to therapy, much of my stuttering was via secondary behaviours such as head movements and jaw jerks. I didn’t like those behaviours, and am only able to stutter without them now because I am so desensitised to stuttering. The type of stuttering I am modelling in the scanner is often referred to as “primary stuttering”, since it does not involve secondary behaviours. One of the aspects I found interesting about deliberately adopting primary stuttering was that I have no idea why the stuttering eventually resolved. I wasn’t doing anything on purpose to resolve the stuttering and allow me to continue to the next part of my planned utterance. But somehow, after several repetitions of a vocal gesture, I proceeded to the next syllable anyway. I felt like a passenger to the stuttering experience.

Tension in stuttering can also be tonic. In tonic stuttering, muscular tension is maintained for longer than intended. Tonic tension is heard as prrrrrolonged syllables. Or, if tonic tension occurs in the larynx, there can be a cessation of air flow, experienced by the speaker as a silent block. This type of stuttering can be alarming, because it affects breathing patterns. Anything which alters breathing can feel threatening, as if one is suffocating or drowning. Stuttering which affects breathing can be very unpleasant, and can understandably lead to avoidance of speaking situations.

Several instances of tonic tension can be seen in the first video. We also recorded a second video, in which I read the same passage but deliberately changed any anticipated instances of stuttering into a short prolongations:

There is a lot of tonic tension, but these instances of stuttering resolve quickly compared to the stuttering in the first video. I learnt to stutter in this way by following Van Riper’s recommendations about stuttering modification. It is now my natural mode of speaking. Speaking in this mode is automatic for me – my speech is spontaneous, and requires no thought. The stuttering which remains is not troublesome to me. What is more, when I follow this speaking strategy my delivery takes about a third as long as the primary stuttering in the first video. You can see this in the timings of the videos. There still noticeable stuttering in the second video, but the stuttering is minimally time consuming. Of course, it should not be an issue to anyone if people who stutter need longer to communicate. However, people who stutter may themselves not wish to take appreciably longer to deliver a message than people who do not stutter, or than they would take themselves were it not for their stuttering. Such a circumstance is referred to in disability studies as an impairment effect – something which is intrinsic to a particular type of impairment (such as stuttering) but is nothing to do with listener reaction or social attitudes to stuttering. Impairment effects in stuttering are real, and can often be the most wearing and traumatising part of the stuttering experience. Treatments such as stuttering modification aim to reduce impairment effects.

During a moment of stuttering, stutterers will sometimes switch between tonic and clonic tensions. This can happen even from an early age (e.g. the typical age of onset of stuttering at 3–5 years old). The explanation for stutterers changing between tonic and clonic tension is that they have found changes to articulatory posture can help to resolve a moment of stuttering. This situation can become substantially complicated with the introduction of secondary stuttering behaviours (e.g. grimaces, eye closures or jaw jerks). It is unclear why such behaviours should ever resolve stuttering, just as it is unclear why stuttering occurs in the first place. Vocal tract imaging has the potential to map out the course of a moment of stuttering, and thereby to transform our understanding. There are at least three way I can think of in which it can do so.

The first is to answer a question of whether stuttering occurs on the sound currently being repeated (e.g. in the word “stutter”, the /st/ of /stʌtɚ/) or is on the immediately following sound (e.g. the /ʌ/, or “u”, which follows the /st/ of /stʌtɚ/). This has already been assessed, with initial indication that it is restraint to the release of a consonant sound, more than a lack of preparedness for the issue of the following vowel sound, which is the locus of the repetitions. However, this description applies to an articulatory gesture, and may not fully reflect phonetic planning preceding that gesture. There is moreover scope for more detailed analyses. For example, in an instance in which stuttering occurs through repetition of the beginning of the word “stutter”, is there excess tension in articulatory positions other than those necessary for the /st/ and the /ʌ/? Another question might be what causes an instance of stuttering to resolve – e.g. why would a certain articulation of the /st/ in “stutter” lead to articulation of the subsequent /ʌ/, when prior articulations of /st/ in “stutter” were followed by repetitions of /st/? What differences are seen in the vocal tract when stuttering resolves?

A second way vocal tract imaging could be helpful is in speaking tasks involving auditory perturbation. In these, ongoing speech is altered through manipulation of sounds heard over insert earphones. For example, delays and frequency shifts to ongoing speech have repeatedly been found to reduce the amount of stuttering. There are at present scant data describing changes in vocal tract gestures which accompany auditory perturbations. Obtaining and analysing such data, perhaps in conjunction with spectographic analysis of speech, would seem an obvious step towards understanding how alterations in speech perception can affect speech production.

A third use for vocal tract imaging could be comparing different modes of stuttering. For example, the two videos of myself speaking in this post compare primary stuttering to stuttering modified by the introduction of deliberate short prolongations when stuttering is anticipated. Other comparisons might be to smooth speech, in which every syllable is deliberately prolonged, or to costal breathing, in which the creation of large amounts of pressure at the diaphragm overpowers tension in vocal tract articulators.

Vocal tract imaging offers the potential for an extensive programme of research which could lead to better treatments for stuttering, as well as a greater understanding of how linguistic planning overlaps with and influences speech-motor production. A limitation is the cost of time in high resolution scanners, as well as researcher time. However, the study at the University of Southern California remains ongoing. If you are interested in this type of research, and can travel to the Los Angeles downtown area, full details are available here.