Talkative women, taciturn men

Yesterday's Arlo and Janis cartoon is a timely embodiment of the stereotypes that Mark Lieberman's been debunking pseudoscience about (e), over at Language Log lately:



The thing about the pseudoscience in question is, people think it's giving them objective confimation of something they are inclined to believe anyway, but which is mildly un-PC to discuss. This makes it a) very easy to believe and b) both interesting and safe to repeat, since any skeptical or disapproving raised eyebrows can be referred to the supposed(ly) scientific experimental results.

Maybe people experience women as being more talky than men not because they actually are that much more talky (Mark shows that they're not), but rather because people are less patient with women than with men. That is, a speech segment of length X might be perceived as being longer or shorter depending on the sex of the speaker. The idea would be that because women are generally lower-status community members than men, people are less inclined to invest a lot of time in listening to them. (Quiet and respectful attention is something accorded to higher-status people more than to lower-status people, after all.) Consequently, if a lower-status person talks the same amount as a higher-status person does, an interlocutor will perceive that person (negatively) as too talkative, that is, as talking more than their relative status 'should' warrant. If women are, on average, lower-status individuals than men, and if they, on average, talk more or less the same amount as men do, then they'll be perceived as 'talkative'

This would be very easy to test experimentally -- take the same speech segment, say 30 seconds worth; manipulate the pitch to create two copies of the recording identical in all respects except that one sounds like it's spoken by a female and one by a male, and play each to different randomly selected group of subjects. Ask the subjects to estimate the length of the recording (and perhaps for estimates of other factors too: e.g. is the rate of speech above or below average?) Analyze the estimates to discover if subjects tend to estimate that the 'female' speaker talked longer than the 'male' speaker, or vice versa, or neither.

That much would test whether people do in fact perceive women as more talkative than men. Assuming a positive result, the next step would be to test the conjecture that this experiential effect is correlated with social status, perhaps more than with gender. To do this, one would run the same experiment again with just a single recording (either male or female), but giving the subjects one of two short descriptions of the 'speaker' before they listen to it. Half the subjects would hear a description consistent with a high-status individual as the speaker, half would hear a description of a low-status individual. After listening to the stimulus, they'd be asked to estimate the talkiness of the speaker, exactly as in the previous experiment.

Perhaps such work has already been done? Anybody know of any?