What does a disfluency profile look like?

People vary widely in the manner in which disfluency appears in their speech. When it comes to filled pauses alone, there are people who use exclusively uh, others who use only um, and still others who seem to mix it up in some way. Some will use them only between utterances, while others (like me) will use them pretty much anywhere. There are some for whom the filled pauses are only long, drawn out articulations, and then there are others (again, like me) who use some filled pauses so quickly that the only way one would notice them is by listening to a recording afterward.

And that's just filled pauses only. If we mix in other sorts of disfluencies like lengthenings, repairs, false starts, and repeats, then we may start to see that every individual has their own unique disfluency profile. Now, this is no new claim: Many researchers have shown empirical evidence of individual variation in disfluency. What I have not seen, however, is any study (let alone evidence) of any sort of systematic variation in this variation. It may be possible that people just settle into a pattern of disfluency use which is largely randomly set. But this is an uninteresting hypothesis. So, if only for the fun of it, I'd like to entertain a different hypothesis: that the number of possible disfluency profiles is actually finite and predictable.

To confirm this hypothesis, we'd need evidence that certain disfluency behaviors correlate with (or against) each other. Let's consider here a few possibilities. First, though, let's limit this discussion to the case in which people use disfluency because of some kinds of speech production difficulty. In other words, let's ignore the possible functional and intentional uses of filled pauses as mediating or politeness devices. And let's also take it as a given that everyone experiences such production difficulties regularly.

One way that certain disfluency types might offset each other is in the degree that a speaker takes time to prepare what they are going to say. The person who says a long ummmmm before and between every utterance is presumably expending great cognitive energy to conceptualize and plan the following utterance carefully. Hence, it seems likely that when they articulate that utterance, it will come out smoothly with only rare need for repair. Perhaps in some individuals, they may still have a local need to recall a word from memory and thus use a silent or filled pause to regain that time. But on the whole, the long disfluency between utterances minimizes the short, local disfluencies.

Then, how about the opposite case: one who tends to start speaking immediately when it is their turn? Well, if they are speaking before they are fully prepared, then they will likely run into some local difficulty and end up displaying quite a lot of local disfluency as false starts, repairs, repeats, filled pauses and the like. Or they may learn to avoid many of these by using lengthening to buy the time needed to prepare their speech. Perhaps, if the lengthening is quite pervasive, they will simply use an overall slower speech rate.

I suppose it could also be the case that some people have quite over-active feedback loops that monitor their speech production too closely and raise flags perhaps more often than is really needed. These people would interrupt their speech (either overtly or covertly) quite often, leading to a high rate of repairs and possibly interregnum phrases (uh, no, I mean). These could even stack on top of each other and appear in rapid sequence.

A couple of further questions which could be considered here is learnability and malleability. So, how does one develop one's profile? Is it something that one picks up from others (say, family members) or does it fall out from the way one's linguistic development proceeds? Also, how changeable is one's profile? If one were to wish to change one's profile, how malleable is it? Can one switch from a rapid-fire repairer to a thoughtful ummer? I will suppose that some of this is just simple habits which are malleable, but how malleable are they? And what consequences could potentially follow after forcefully changing one's profile? Great questions for further consideration...