"Like" youth and sex

« previous post | next post »

In confessing her like-aholism ("My Love Affair With 'Like'", Jezebel 6/26/2011), Erin Gloria Ryan framed the problem in terms of gender roles:

Any girl who's been teased for middle school nerdery has likely developed a long standing aversion for the feeling of being excluded for being too smart or opinionated. This is the way that socially acceptable people talk. This is the way that pretty people talk. Women are taught that it's more important to be pretty and socially accepted than it is to be smart. Ergo, like.

She's talking about the discourse-particle like, as in her example "so, like, my sentences, like, sound like this. And I, like, sound dumber than I actually am".  She reports a student evaluation that also noted the stereotypical association with youth: ""She says 'like' more often than a valley girl".

Are these stereotypes accurate? Is the discourse-particle like really characteristic of younger women? Today's Breakfast Experiment™ looks into the matter, and finds (in a limited and superficial survey of proxy measures) that one part of the stereotype is apparently valid, but the other is not.

As a source of data, I used the LDC's Fisher English collection, which comprises 11,699 telephone conversations recorded in 2003. The good news is that like occurs 317,509 times, and the demographic metadata for the speakers include age and sex. The bad news is that like is very ambiguous — we're not interested in like the verb

casual periodic dining out is fine but generally speaking i like eating at home

or like the complementizer

yeah it isn't as easy as it looks like it is on TV

or like the preposition

with a person like that you have no idea what they're gonna do

And I don't have time this morning to sort through even a large sample of the 317,509 cases. So as a proxy measure, I'm going to count hits for some simple patterns where nearly all of the hits will be the kinds of like that we're looking for.  For example, sequences of PREPOSITION like ARTICLE will mostly give us things like this:

… but you know just this feeling of you know like disorientation you know this feeling of wow where am I and where do I fit into this and where do we all fit into this you know trying to sort through like the recognition about what happened …

We want to look at specific prepositions and articles that occur together often enough for the comparison to be meaningful; and (because the data is far from orthogonal in the variables of interest) we'll look in each case at the proportion of like-insertion in a given lexical frame.

Here's "in (like) the":

in like the in the like per 10,000
Male 102 23,627 42.9
Female 95 27,919 33.9

Here's "on (like) the":

on like the on the like per 10,000
Male 62 11,483 53.7
Female 60 15,573 38.5

In both of those cases, it seems that men actually insert like more often than women do. This is not true for all of the lexical frames that I've tried, e.g. "for (like) the":

for like the for the like per 10,000
Male 92 6,081 149
Female 119 7,360 159

Two things are different in "for (like) the" compared to "in (like) the" and "on (like) the": the frequency of like-insertion is 3 to 5 times higher; and the women are doing it around 7% more often than the men, rather than about 20-30% less often. It might be interesting to look into these differences further, but it's time to pour another cup of coffee and move on.

So far, the femininity of like-itis is certainly not getting a ringing endorsement. How about the age dimension?  In order to break it down by age as well as sex, let's back off from the prepositions and look at sequences of the form ARTICLE like.  These mostly turn up things like

i saw that last on last night on fox news that some of the like suicide bombers are are trying to attack soldiers there
once you have a family you get married and so forth you know the like relationships change a little bit

you have to you have to register you know to have a like i don't know how to say certificate or something
i came across it on um a site on the internet in a like freebie type newsletter

If we again normalize the ARTICLE like counts by the ARTICLE counts, we get something like this:

[The "Age Group" categories are 20-39 ("yng"), 40-59 ("mid"), and 60-69 ("old").]

The first thing to note is that these data do confirm the youth hypothesis: for all combinations of sex and article, the frequency of like-insertion declines steeply with age.

But again, the relation to sex is complicated.

Overall, guys are doing a bit more like-insertion than gals — if we all up all the counts and normalize again, we find a rate of 20.0 per 10,000 for the males, versus 16.4 per 10,000 for the females.

The two articles come out differently with respect to sex.  In the case of "a", men consistently insert like more often than women — overall the male rate is 25.6 per 10,000 versus a female rate of 17.6, and the difference is maintained in every age group. However, in the case of "the", women generally insert like more frequently, especially at younger ages.

Where does this leave us? With more questions than answers, as usual; but there are a couple of clear conclusions.

First, there's definitely an association between non-traditional like and youth. As usual, whether this is a life-cycle thing or a change in progress is unclear; as usual, it's probably a bit of both; but it's plausible to assume that this is a change going forward, and that in a few decades phrases like "a really like lively community" will become part of the standard language.

Second, there's no evidence that women insert non-traditional like into their conversation more often than men do. There may be specific syntactic or pragmatic contexts where this is true; there may be effects in some registers and not others. But so far, I'm inclined to think that this is one of those cases where congruence with pre-existing stereotypes (here that women are less assertive) leads to post-hoc rationalization and confirmation bias.

[A set of links to previous LL discussions of like can be found here; also links to two relevant publications, Muffy Siegel, "Like: The Discourse Particle and Semantics", J. of Semantics 19(1), Feb. 2002, and Alexandra D'Arcy, "Like and Language Ideology: Disentangling Fact from Fiction", American Speech 82(4) 2007.]

Update — Thanks to a comment from Damon (below), I remembered a post from Language Log Classic ("Like totally presidential", 8/17/2007) which quotes  Yoko Iyeiri et al., "Gender and Style: the Discourse Particle like in the Corpus of Spoken Professional American English", English Corpus Studies 12, 2005:

We have investigated the Corpus of Spoken Professional American English and found that the discourse particle like is attested in the exploratory talk of the national meetings of mathematics tests and reading tests, both held in the 1990s, to a noticeable extent. By contrast, the expository talk of White House press conferences and faculty meetings of the University of North Carolina provides far fewer examples of the discourse particle like. As for gender differences, the same item is more frequently employed by male speakers. This result does not necessarily support the generally accepted view, which argues that it is a characteristic feature of young female speech.

They offer this graph:



  1. KevinM said,

    June 28, 2011 @ 11:09 am

    "As usual, whether this is a life-cycle thing or a change in progress is unclear" We can only hope the former. My son doesn't "like" much. My daughter, like my son dripping with Ivy League honors in the humanities, can't go more than five words without dropping the L-bomb. I choose to believe it will fade as she progresses through her twenties, but that may be just superstition on my part. Mel Brooks claims that, based on observations of his Brooklyn neighbors, he believes that speaking Yiddish is something that happens to you when you get old.

  2. Damon said,

    June 28, 2011 @ 11:36 am

    I wish I remembered some details but I swear I read a study in my undergrad years that like suggested that middle-aged men were the biggest users of discursive 'like'. At any rate, by this point I'm like more surprised when sociolinguistic research confirms people's intuitions about which groups most characteristically employ a given form. It is interesting that there's such a pervasive notion that discursive 'like' is particularly feminine.

  3. Nathan said,

    June 28, 2011 @ 12:06 pm

    No, like isn't ambiguous, just polysemous. Humans really have no problem identifying the difference; it's just hard to automate.

    [(myl) It appears that you're confused about what ambiguous means. From the OED:

    Of words or other significant indications: Admitting more than one interpretation, or explanation; of double meaning, or of several possible meanings; equivocal. (The commonest use.)

    This is the commonest use overall, and in discussions of linguistic analysis, it's essentially the only use. And it's correct in this case: like has several different syntactic functions, and several different meanings in each. It's not relevant to the question whether it's easy or hard for human or machines to disambiguate the alternative interpretations.]

  4. Matt McIrvin said,

    June 28, 2011 @ 1:36 pm

    I remember people complaining about this in much the same way when I was a teenager, 25-30 years ago (and "Valley Girls" had just entered the national consciousness). That suggests either that it's a life-stage effect rather than a cohort effect, or that we all use it so much now that frequencies that were once annoying are now considered normal. I suspect the former, but data would be interesting.

  5. Tanja said,

    June 28, 2011 @ 1:45 pm

    Sali Tagliamonte's study, which used conversational data collected in 2002-2003 from young Canadians between 10 and 19, found that the use of like as a discourse marker seemed to be age-graded: it grew until peaking at the age of 15-16, with a sharp drop at the next age group of 17-19 years. Females tended to use like more than males in every age group. This was based on only 18 speakers, though. She also looked at the frequency of like tokens in terms of the grammatical category that followed (NP was the most frequent one), but didn't separate the results by gender.


    [(myl) My data starts at age 20, so there's not necessarily any divergence.

    But maybe (that set of) Toronto teens were different -- note that in the 15-16 age group, there were just 2 males and 2 females, so that the "peak" in like-usage found at that age might well have been sampling error. Also note that Tagliamonte, like D'Arcy and me, found quite a bit of variation in the interaction of sex and syntax; and I found lots of interactions between sex and lexical choice. So depending on the mix of subcases in Tagliamonte's study, you could get lots of different things.

    Unfortunately, Tagliamonte's corpus is not published, so we can't look into any of these issues further. I strongly prefer studies based on published data sets, since that makes it possible to go back and check things that were not originally taken into consideration.]

  6. Sili said,

    June 28, 2011 @ 3:07 pm

    Its syntax and meaning are nearly always clear in context, admitting only a single interpretation.

    But when searching, you can rarely include context – save by proxy as done here. The point, as I understand it, is that just searching for "like" in the corpus will not yield any answers, because "like" without context is ambiguous exactly due its polysemy.

    [(myl) Yes, that's part of the point.

    In fact, though, some of the distinctions in question remain ambiguous in many contexts. See "Divine ambiguity", 1/4/2004, for an extended discussion of one example. For another example, D'Arcy (following others) distinguishes between a "discourse marker" use of like, which functions somewhat like well or so in connecting sentences or clauses, and a "discourse particle" like, which can occur nearly anywhere and modifies sub-clausal constituents. Examples might be "Like I was already sick of it yesterday" vs. "I was already sick of it like yesterday". But if someone says "Like yesterday I was already sick of it", the analysis could either be "like [yesterday I was already sick of it" or "[like yesterday] [I was already sick of it]".

    For this morning's experiment, I considered tagging a random sample of the hits. But in practice, I found that this was as hard to do as such things nearly always are, and so I decided to explore the results using reliable contextual proxies instead. ]

  7. Elizabeth Braun said,

    June 29, 2011 @ 3:17 am

    Ugh! That first quote showing the 'cool' (or should I say 'sick' these days??) use of 'like' has THREE instances of the filler 'you know' in! I remember listening to someone's comment a few years ago and counting that they used 'you know' about 12 times! It was so distracting that I couldn't tell you what they were talking about….

    I think a lot of use of 'fillers', into which category this use of 'like' might well be said to fall as it has no real function, but does carry a tiny bit of meaning, (i.e. something like…), could mostly be down to nerves and being unsure. People add fillers ever more when they are under pressure of speaking into a microphone or something like that!

  8. Keith M Ellis said,

    June 29, 2011 @ 9:41 am

    I think it's amusing (and disappointing) that even on LL (at least in comments) there seems to be widely shared assumption that "fillers" truly have no function and are "bad".

    I guess if you scratch hard enough, you'll find language peeving even here.

  9. This Week’s Language Blog Roundup | Wordnik ~ all the words said,

    July 1, 2011 @ 9:49 am

    [...] love affair with peppering her speech with "like" while Mark Liberman at Language Log questioned Ryan’s proposal that women may use “like” more often than men, and jokingly devised a possible solution, the [...]

  10. martin said,

    July 1, 2011 @ 10:50 am

    All very, like, interesting stuff. And some good linguistics, but no reference to the wierdness of how the word 'like' started out. I have tried in vain to find sources to confirm what I was taught by my Anglo-Saxon tutors that the word originally meant 'body'. The log entry squabble over the ambiguity/polysemy tag aside, the word (and suffix) 'like' has to be the most versatile piece of kit in the English language – checkout http://dictionary.cambridge.org/search/british/?q=-like. Finally, Suzanne Romaine and Deborah Lange produced an important paper way back in 1991 pointing to the new use of 'like' under review in this blog and which Prof Pullum (a self-declared non-semanticist) overlooks – it's a quoting device in the making – http://users.ox.ac.uk/~romaine/romainelange.pdf (you need to have JSTOR clearance for this)

  11. Librarian M. said,

    July 14, 2011 @ 6:52 pm

    But, like, you missed, like, the part where people are all like, "like what's happening?"

  12. Filler words: The enemy of eloquence? « thesociallinguist said,

    February 21, 2012 @ 3:16 pm

    [...] was this gem from the Star Tribune about fillers (Language Log have talked about fillers here and here if you're interested). The article opens with the cracking line 'If Martin Luther King [...]

  13. “The most annoying word of 2012″: Whatever. | Linguistics @ Canterbury said,

    January 2, 2013 @ 3:43 am

    [...] Log 2007 'Like totally presidential'  Language Log 2011 'Like, youth and sex' Share this:TwitterFacebookLike this:LikeBe the first to like [...]

RSS feed for comments on this post · TrackBack URI

Leave a Comment