Yesterday's post, "Like Totally", cited Iyeiri et al.'s finding that men use discourse-particle like more than women in a range of professional settings. They also found the expected pattern of more use of discourse-particle like in settings that are more casual, less formal, less rehearsed, etc.. Several readers put these two things together and made the connection to the well-known tendency for men to use less formal speech styles, other things equal -- see for example the facts about g-dropping in English. Some readers went further, and speculated that professional settings might interact in a special way with sex differences in the use of discourse-particle like, since it's not only stigmatized but also female-associated. For example, Bryn LaFollette wrote:
I wonder if the disparity shown in this data might in fact be a result of the well-known stereotype being a negative influence on female speakers due to the public or group setting. That is to say, are female speakers in each of the settings described holding back from using like due to the worry, consciously or not, of fulfilling the stereotype and thereby being taken less seriously, whereas the male speakers are not similarly constrained?
This could well be true -- through the observed sex differences in like usage were not larger than reported differences in stigmatized forms that are not stereotypically sex-associated.
I thought I'd take a look at sex and age differences in use of discourse-particle like in a different sort of corpus, namely the collection of transcribed telephone conversations from the LDC, which includes 15,672 conversational sides where the speaker is female and 12,571 conversational sides where the speaker is male. The uses of like -- 184,184 for the women and 156,799 for the men -- are not tagged according to part of speech or function, and I don't have time this morning to select a sample and tag it, so I used some proxies.
I started by looking at the frequency of some strings like "like totally", "like really", "like wow". Here are the raw counts:
Normalizing by conversation, men used "like totally" once per 137 conversations (on average), while women used "like totally" once per 141 conversation; men used "like really" once per 22 conversations, women once per 21 conversations; men used "like wow" once per 59 conversations, women once per 45 conversations.
Normalizing by "like", men used "like totally" once per 1,704 overall instances of "like", while women used "like totally" once per 1659 instances of "like"; men used "like really" once per 274 uses of "like", women once per 244; men used "like wow" once per 704 uses of "like", women once per 531.
These results suggest rough parity, with perhaps very slightly more frequent use of (these cases of) discourse-particle like by women -- though men still used "like totally" slightly more often, at least on a per-conversation basis!
Looking quickly at the effects of age, it seems that discourse-particle like is associated with middle-aged as well as young people (in these conversations, recorded mostly around 2003). The numbers in the table below are overall counts for "like" for the specified group, divided by counts for each of the cited strings. Thus people aged 20-39 used "like totally" once per 1,648 uses of "like", and people aged 40-59 used "like totally" once per 1,619 uses of "like", whereas people aged 60 and over used "like totally" only once in 3,646 uses of all sorts of "like".
Finally, one weird statistic, for which I have no explanation. (This is no longer about discourse-particle like, but rather about the conjuunction-ish usage; but it's too strange to leave out, though it's not really relevant.)
In this collection, women used the expression "like I say" 545 times, and the expression "like I said" 2,264 times. But men used "like I say" 563 times, and "like I said", 1,302 times.
For women, the ratio of "like I said" to "like I say" was 4.51 to 1,, whereas for men, it was only 2.3 to 1. Why are women more likely to use the past tense in this expression -- or men more likely to use a timeless generic?Posted by Mark Liberman at August 18, 2007 06:47 AM