The portion of the Sonagraph touching the abdomen is called the transducer, only in my instance I was applying the transducer to the side of the neck to study throat-movement patterns in cleft-palate patients slated for surgery. A consideration of these patterns was necessary to establish the donor site - the place in the throat from which a piece of tissue would be taken to close a hole in the roof of the mouth.

One of the patients stuttered, and with the Sonagraph it became clear that the throat constricted forcefully before every stutter. First would come the constriction, then the stutter. It was very regular, and I couldn't help but pause and consider this curious event.

Failing to find a ready explanation, I called a Speech Pathologist to ask if he knew anything of the relationship between throat constrictions and stuttering I had observed. He knew of none, he said, since Speech Pathologists had never been able to look at the throat during speech. My ultrasonic scan represented something new.

Excited by the prospect of a new discovery, we arranged to have several stutterers seen for ultrasonic examination. Within two weeks I had examined five - and all displayed the same pattern of constrictions.

Having confirmed the non-uniqueness of my initial observation, I began to systematically move the transducer along the side of the neck to see if the throat constrictions varied in intensity along the vertical dimension. The answer came quickly: the further down the throat, the more vigorous the constriction.

The throat rests upon the larynx, or voice box, which contains the vocal cords. The vocal cords are two small horizontal folds of tissue that lie within, one on either side of the box. The larynx rests on top of the trachea (or wind pipe) and its front cover is the Adam's apple. In order to speak, the vocal cords are brought together by several pairs of muscles so that they touch each other gently. The person then builds up an air pressure beneath them by expelling air from the lungs. When the air pressure becomes great enough, it blows the vocal cords apart, which sets them into vibration and makes sound. This sound is the raw material for speech production; it is converted into speech by moving the lips, tongue, jaw, teeth, palate and other articulators.

When the transducer was applied to the side of the stutterer's larynx, something happened which was not to be expected: just before every stutter, the vocal cords would rise slightly and then suddenly slam together in a constriction more violent still than any seen in the throat.

Here, I thought, was the center of the activity - the vocal cords - pressed together forcefully. For some reason the stutterers were tensing their vocal cords so powerfully that the air required for speaking couldn't pass. It reminded me of a phenomenon in medicine known as a laryngeal spasm, a forceful locking of the vocal cords which sometimes occurs after an operation. From my ultrasonic examination, it appeared that stutterers were exhibiting short-duration laryngeal spasms, but why, I didn't know.

I proposed an explanation which seemed outlandish at the time, but has since been substantiated repeatedly by investigators - namely, that the source of all stuttering is a locking of the vocal cords. But at that time I did not understand what the relationship was between the struggles I observed and the vocal cords.

I began to examine the struggles closely. For instance, some patients would suck air in briefly through their mouths just before speaking - a sort of inspiratory gasp. They reported that this facilitated fluency. It soon became clear why. The more rapidly one inhales, the wider the vocal cords open to allow a greater volume of air to pass. The patients were using this rapid inspiratory movement to widely open locked vocal cords so they could initiate speech before the cords locked again.

Similarly, other patients who spoke at the end of their breaths were unknowingly making use of another strategy to open locked cords. There are nerve endings in the lungs that detect air volume. When a person exhales most of his air before speaking these receptors detect what they interpret as the imminent collapse of the lungs, and signal the brain to initiate inspiration. A single pair of muscles at the back of the voice box begins to contract to force the vocal cords apart in preparation for the inflow of air. A person cannot stutter when a pair of muscles is being powerfully driven by the brain to open the vocal cords.

Other patients reported being able to speak after first swallowing. During the swallow reflex the vocal cords close tightly to prevent liquids or solid food from entering the lungs, and immediately after the swallow, as part of the same reflex, the cords are opened forcefully so that respiration may resume. The patients had unknowingly discovered that if they started speech at the instant a swallow was completed, they could speak without stuttering.

The non-speech behaviors of the stutterer now made sense. I was still at a loss, however, to understand precisely how a locking of the vocal cords could lead to the wide variety of speech struggles I observed.