Uh, more on the mysterious case of 'uh' and 'um'

A recent summary on this blog (Er, what about this?) discussed the intriguing finding that while male speakers of British English used um and uh (or erm and er, in British English) more often than female speakers, females preferred um over uh. Now recent research in the US has revealed that female speakers of American English behave in the same way – at least in the two sets of data that Eric K Acton analysed.

Acton analysed two distinct corpora of spoken American English. One was a collection of 992 audio recordings from three speed-dating sessions held for graduate students in 2005. He found small but statistically significant differences between the overall rates of um and uh in men’s and women’s speech, with men using them more frequently, just as in Britain. A far more dramatic difference, though, was in the proportion of um to uh in men and women’s speech. This was more than three times higher in the women’s speech than in the men’s speech. 

The second corpus was the Switchboard Corpus, a database of more than 2400 telephone conversations between people across the USA, recorded in 1990. Again Acton found that the proportion of um to uh was higher for women than for men. In this case, the proportion of um to uh in female speech was two and half times as high as in male speech – not as high, then, as in the Speed Dating corpus, but sizeable nonetheless. The Switchboard Corpus includes conversations from different regions of the country: although the degree of women’s preference for um over uh varied across the country (it was highest in New England and lowest in the South), the gender differences persisted across the different regions. Acton considered the possibility that the gender of the listener might affect the use of um rather than uh: however, while both men talking to men and women talking to women used higher proportions of um relative to uh than when talking to the other gender, the proportion of um to uh for men talking to men was less than half that of when women were talking to women. As in the research on British English, younger speakers in the Switchboard Corpus used more um than uh compared to older speakers, suggesting that a language change may be occurring towards the use of um rather than uh. 

Like previous researchers who have analysed um and uh, Acton is unable to find an explanation for the dramatic gender differences in his data. He notes that he now intends to investigate whether um and uh may differ in what they communicate. He does not expect there to be an explanation as direct as “um means ‘female’” and “uh means ‘male’”. This would run counter to other research on the relation between social categories such as gender and social meaning, and would not account for the different frequencies he found among speakers of different ages (nor amongst speakers of a different social status, as found in the research on British English). Perhaps there is something about the meanings of um and uh and women’s relation to society (in both the US and Britain) that can explain why women seem to leading a change towards the increased use of um rather than uh. Even so, Acton would not expect all women or all men to behave in the same way, and exceptions to the general rule may turn out to be just as informative as the original generalization.

Acton concludes that um and uh, both in their ubiquity in spoken English and the degree to which their use is socially stratified, provide a rich site for understanding the dynamics of language use and social meaning. Certainly they present an intriguing puzzle for researchers.

Acton, Eric K. 2011. On gender differences in the distribution of um and uh. University of Pennsylvania Working Papers in Linguistics 17/2.

This summary was written by Jenny Cheshire