Who knows?

« previous post | next post »

Over at the Brainstorm blog ("Psychology Today Editors Flood the Blog Zone"), Matthew Hutson asks "What does Caroline Kennedy know that we don't?" This is about Caroline Kennedy's filled pauses, of course, but what struck me first about Matt's post is the way that the blog format allows a journalist to take a more personal approach to the news:

Using the phrase is a pet peeve of my mom's. She'll interrupt my dad and say, "No, I don't know–you haven't told me yet." I suppose the peeve has latched onto me, as I'm more aware than most people are of its use.

But in fact, Matt's personal history with this expression has led him to some conclusions that are less subjective than most of the journalistic treatments of the subject, rather than more. He cites his family background as the reason he wonders "But what does saying 'you know' actually say, you know, about your personality?" And as a result, he looked into the question more than a year before Ms. Kennedy's candidacy even came up:

In 2007 I asked the U Penn linguist Mark Liberman if there might be differences between people who say "I mean" a lot and those who say "you know" a lot. He did some analysis and found that the "you know"/"I mean" ratio is greater in women and those who are older and those who are less well educated. Later he found that the uses of "you know" per conversational turn is highest in people 40-59, and is greater in men. But these data don't say much about personality.

About a year ago I asked UC Riverside psychologist Lisa Fast about you-know-ers and I-mean-ers. She'd published a paper in the Journal of Personality and Social Psychology titled "Personality as Manifest in Word Use: Correlations With Self-Report, Acquaintance Report, and Behavior," and she performed some additional analysis of the data for me.

Fast used a program called Linguistic Inquiry and Word Count and found that women use more nonfluencies (phrases such as "you know" and "I mean") and fillers (words including "um" and "uh") than men do. Both men and women who use more nonfluencies and fillers are considered by their peers to be more extroverted and more agreeable. Academic achievement is not correlated.

I'm glad to see the use of Jamie Pennebaker's LIWC program spreading among social psychologists. I'm also curious about why Lisa Fast's results seem to be somewhat different from mine (and some others' as well). Thus. Matt reports her as finding that "women use more … fillers (words including "um" and "uh") than men do", whereas in an experiment I reported in 2005, I found an interaction among filler, sex and age, so that the usage of "uh" increased with age in the material I looked at, with men on average higher than women in every age group, while the usage of "um" decreased with age, with women higher than men in every age group. And Bortfeld et al. 2001, cited in that 2005 blog post, found that "Men say uh and um [overall] more than women".

In order to sort this all  out, we'd need to start with more exact statements of the apparently divergent findings, compare the samples of speakers and contexts we looked at, the possible effects of the transcription processes involved, and lots of other factors. This kind of divergence and discussion is normal and productive, and I'm really happy to see some of my little Breakfast Experiments™ entering into the process — via a blog post by the editor of a popular magazine!


  1. jfruh said,

    January 7, 2009 @ 11:25 am

    Though it would presumably take decades to determine, it would be interesting to see if the age-related results are based on age or date of birth. In other words, if a 2009 experiment people over 50 are found to use a particular filler word more often than their younger peers, does that mean that (a) you tend to use that filler word when you're older, or (b) that you tend to use that filler word if you acquired language in before about 1960?

    [(myl) There's a substantial sociolinguistics literature on real vs. apparent time with respect to a variety of linguistic features in a variety of settings. Over-generalizing recklessly, I'd sum this literature up by predicting that the experiment would show a mixture of the two effects that you cite, mixed in with a third possibility as well: there may be a secular trend that older people participate in to some extent, so that the "apparent time" effect is smaller than the "real time" effect.

    There may be enough evidence in the (unfortunately unpublished) longitudinal data bases on which these studies are based in order to provide an answer with respect to (for example) rates of "uh" and "um" usage in English. But it may not be a very conclusive answer, since the number of speakers in these databases tends to be fairly small. ]

  2. Nathan Myers said,

    January 7, 2009 @ 3:56 pm

    Doesn't val-speak "like" fall into the same category as "uh/um" and "y'know/I mean"? Is it not measured because it was too regional until very recently?

  3. Mark P said,

    January 7, 2009 @ 4:32 pm

    I think "like" predates val-speak. I think it goes back at least to the beatnik days.

  4. Amy Vaughan said,

    January 7, 2009 @ 4:39 pm

    It's funny that despite seeing a correlation between nonfluencies/fillers and agreeableness, Caroline Kennedy has been getting skewered on TV as someone less fluent and eloquent. I guess being eloquent has nothing to do with being agreeable.
    Nevertheless, thinking of my own usage of such words, I find myself a bit perplexed by Caroline Kennedy's – doesn't being on national television qualify as a formal situation? If that means that we would expect less nonfluencies than usual, I'd be a little worried about Caroline Kennedy's ability to communicate in informal situations. But clearly she can communicate well enough to ask to be appointed Senator, so that leads me to speculate that she might be doing it on purpose. Hillary Clinton was panned during her run to the White House for being too cold and stiff (and faking those tears!) so perhaps Caroline Kennedy wants to appear to be warm and human and more pleasant?

  5. Philipp Angermeyer said,

    January 7, 2009 @ 5:10 pm

    I haven't investigated this at all, but off-hand it seems to me that an individual speaker's rate of filler use should not be expected to be an immutable characteristic, but rather should be expected to vary across contexts, by interlocutor, speech genre, etc. That is not to say that factors such as age and gender will not play a role, but the different findings reported above can likely be explained by contextual differences in the type of data (as suggested in the post's final paragraphs). For example, in research on courtroom talk, O'Barr and associates viewed fillers as characteristic of what they called "powerless language", and found that they were used especially frequently by witnesses who were unfamiliar with the courtroom setting — a "formal situation "(@Amy Vaughan). Perhaps Caroline Kennedy's example can be interpreted along these lines, namely that she is not (yet) experienced with giving TV interviews as a candidate for political office.

  6. James Chittleborough said,

    January 7, 2009 @ 8:17 pm

    I'd have thought that the biggest problem trying to analyse use of "um", "uh", and the like would be trying to consistently transcribe a decidedly inconsistent set of sounds.

  7. Thomas Allen said,

    January 7, 2009 @ 9:04 pm

    Why is disapprobation attached to a high count of uhs, ums and you knows? I don't believe the critics are making a non-moral judgment about her ability to communicate more information per word if she omitted the filled pauses, and paused. Instead the critics are focused on the high filled pause rate because they want to make judgments like liar, dumb, or unqualified. Simpler explanations for her quirks include nervousness from the high stakes, wanting to please, plus unfamiliarity or lack of practice speaking on camera. Having a low filled pause rate while speaking on tv is not a formal qualification for the Senate, and might disqualify more than a few.

  8. Ray Girvan said,

    January 8, 2009 @ 5:29 am

    Amy Vaughan > so that leads me to speculate that she might be doing it on purpose.

    Quite possibly. One angle so far not mentioned is that it could be a tactic to stop the interruption likely at long silent pauses. This vaguely relates to the 1982 Nature letter Why is Mrs Thatcher interrupted so often?, which concluded she was giving "completion of syntactic segment" cues in the wrong places.

  9. Wednesday Round Up #46 « Neuroanthropology said,

    January 14, 2009 @ 6:47 am

    […] Liberman, Who Knows? Linguistic Inquiry and Word Count – a methodology makes the news, in this case about Caroline […]

RSS feed for comments on this post