Recently, Caroline Kennedy has been facing a lot of heat for her disfluent speech during an interview with the New York Times. In his Psychology Today blog, Matthew Hutson summarizes the event and discusses some linguistic research regarding the use of such disfluencies (um, uh, you know, like). He shares that the frequency of the disfluencies 'you know' and 'I mean' are, "...greater in women and those who are older and those who are less well educated..." This is surprising to me given that the research I have read suggested they are used more by men, perhaps because men tend to hold the floor longer during conversations (which contradicts the myth of the talkative woman...but that's another story).
In any event, what seems to be lost in this hullabaloo is that everyone uses disfluent speech...perhaps less so during well-practiced speeches, but even then, yes, there is a tendency for us to be disfluent. The trick is that we seem to filter out these disfluencies when we perceive speech...during the perception of our own speech and often during the perception of the speech of others. However, record a 15 minute segment of relaxed conversation among you and your friends (or a speech from an academic conference!) and you will hear the grammatical breakdowns and disfluencies that go altogether unnoticed as we talk and listen to others talk. This is the natural outcome of a production system that is imperfect, but highly effective in communicating the gist of our thoughts & ideas. Disfluencies in speech are normal and, research suggests, can often serve functions that facilitate spoken language comprehension.
Still, sometimes the perceptual filter breaks down and then we begin to notice the 'ums,' 'uhs', and 'you knows' of other people. What is interesting about the Kennedy case is that it may illustrate a bias in when this "um filter" breaks down. Those who are perceived as high status or well-educated are often forgiven their linguistic glitches because they are obviously having a hard time selecting among the myriad of thoughts, words, and ideas they want to share. Those who are perceived as low status or poorly educated, on the other hand, are not afforded this tolerance. Instead, we perceive these people as stumbling, uncertain, and clearly without many thoughts, words, and ideas to share.
What has the potentially biased breakdown of our um filter have to do with the Kennedy case? Well, since this is speculation, I'm not sure. However, the apparent discrepancy in the literature regarding sex differences in the rates of disfluency seems significant. Perhaps a confound exists between sex and status that creates the stereotypes of a disfluent woman versus an absent-minded, but intelligent, man. In any event, the perceptions and misperceptions of disfluent speech, and how we interpret such speech, is important to consider. A biased application of an um filter could have serious consequences for speakers in the workplace, in the classroom, and at academic conferences! In other words, two speakers may use similar rates of disfluencies and while the ums of one these speakers goes unnoticed, the ums of the other speaker could cost her a job, a raise, or a position of leadership. The situation is even more complicated if we consider that different types of disfluencies may serve different purposes and may be perceived differently (and more/less frequently) by listeners as well.
For a more detailed consideration of the Kennedy case, visit the Language Log site where Mark Liberman scrutinizes the transcript of the Times interview and reviews some of the relevant research in his posts More (dis)fluency and (in)coherence and Who Knows?.
And, for an eye-opening discussion of the history, prevalence, and functions of disfluencies in our speech read Um. . .: Slips, Stumbles, and Verbal Blunders, and What They Mean by Michael Erard.