Back to the Future

A quick update to my previous post about the correlation between the way languages mark future tim reference and future-oriented behaviour. Kieth Chen has written a guest post on Language Log explaining his working paper, and responding to the critiques by Language Loggers Geoffrey Pullum and Mark Liberman. He specifically addresses two concerns, that there is some linguistic imprecision in the classification of of strong FTR languages, and that the statistical correlation may be due to cultural co-diffusion, and I must say that I find his responses valid and persuasive, though clearly considerably more research needs to be done in this area.

The Ant and the Grasshopper

The Ant and the Grasshopper

In response to Chen’s response, another Language Logger, Julie Sedivy, has written an interesting post. Her main point, that additional linguistic experimentation needs to be conducted to really get at what is going on here, is an excellent suggestion, and I do hope that someone goes about this. However, I’m still troubled by some of the objections Sedivy raises. She again brings up this notion of cultural co-diffusion, that families in the same countries who choose to speak a particular language at home will also choose to hold to certain cultural values that are coincidentally associated with that language group, and that there is no causal relation between language and behaviour. This might explain away the behaviour in any one country, but if I understand correctly, Chen’s data is drawn from a wide range of countries which don’t necessarily have cultural connections. But again, perhaps I’m missing something here.

What really troubles me is the extent to which Chen’s data is being dismissed primarily on the grounds that the prevailing model shouldn’t allow. Surely the data should come first, even if it conflicts with the prevailing models. Certainly experimental evidence will be needed to corroborate Chen’s statistical correlation in order to establish a solid case for a causal relation. But Sedivy’s attitude is that such a result would be surprising (albeit interesting). There is, however, mounting evidence that such causal relationships exist, as demonstrated for instance by the research by Lera Boroditsky. Many linguists get uncomfortable when confronted by such evidence, since it can’t simply be dismissed, and yet it supports the notion of linguistic relativity, to which they do not hold. Perhaps we need Boroditsky to conduct such experimentation which Sedivy describes, and maybe then the notion of linguistic relativity will not be so cavalierly dismissed.

In any case, if you’re interested in the topic, have a look at these two new posts. I am glad that this topic is getting some attention, and I’m quite enjoying this ongoing discussion on Language Log. I will eventually post about my own research on the topic of linguistic relativity, and very soon I’ll post the rather long entry I’ve been working on about philology and cognitive linguistics.

About these ads
Categories: cognitive, linguistic relativity | 4 Comments

Post navigation

4 thoughts on “Back to the Future

  1. Jeff S.

    “[Julie Sedivy] brings up this notion of cultural co-diffusion, that families in the same countries who choose to speak a particular language at home will also choose to hold to certain cultural values that are coincidentally associated with that language group, and that there is no causal relation between language and behaviour. This might explain away the behaviour in any one country, but if I understand correctly, Chen’s data is drawn from a wide range of countries which don’t necessarily have cultural connections.”

    Yes, but Chen does not compare families in different countries *to each other*: he is trying to compare families identical in every relevant parameter except language, as he explains in his post: “…two families fall into the same bucket if and only if they are identical in COUNTRY OF BIRTH AND RESIDENCE, age, sex, income, family structure, number of children, and religion, where the religions of the world are broken up into 74 types. What I then compare, is families who fall into the same bucket, but who report speaking different languages at home.” [caps mine]

    Sedivy’s point is that cultural co-diffusion makes it impossible to conclude that grammar per se is responsible for differences in ‘forward-looking behaviors’ between families identical on parameters other than the language they (choose to) speak. She does not need to make this point about families in different countries, because they are already different and therefore cannot be usefully compared. In other words, your observation supports her argument.

    “…Chen’s data is being dismissed primarily on the grounds that the prevailing model shouldn’t allow. Surely the data should come first, even if it conflicts with the prevailing models.”

    That is neither a fair nor an accurate description of any of Language Log’s coverage of this topic. Chen’s data is not being dismissed at all, nor is his explanation of it:

    Pullum: “His work is not to be immediately dismissed — I am sure it deserves to be properly considered for publication.”
    Liberman: “My point is not that Prof. Chen’s analysis is wrong. There may very well be consequential connections of the kind that he explores. But the existence of statistically significant logistic regression coefficients among geographic distributions of cultural traits is not enough to convince me.”
    Sedivy: “But Chen has identified a not-crazy hypothesis, and some correlations that are worth thinking about, even worth experimentally investigating.”

    Chen’s study has important, perhaps fatal, flaws, as pointed out by the three LL’ers as well as Östen Dahl (in the comments on Chen’s guest post), who did the research on which the linguistic portion of Chen’s data is based! Those flaws, not conflict with the prevailing model, form the basis on which the linguists have been criticizing (not dismissing) his work.

    “Perhaps we need Boroditsky to conduct such experimentation which Sedivy describes…”

    I agree, that would be very interesting.

  2. Dr. Mark Sundaram

    Jeff: Thank you for your thoughtful comment. You are right — in general the Language Log treatment of the research has been excellent. Not only have Pullum and Liberman written thoughtful and engaging responses, but Chen has been allowed to contribute a guest post explaining his research. I was reacting mainly to the tone and subtext of Sedivy’s post. She characterises Chen’s hypothesis as “not-crazy” rather than simply as plausible, and humorously dismisses it, presupposing the results of further experimentation, and I don’t think that’s fair. It’s odd, given that her main point is dead right — it would be relatively easy to devise experiments to further test the hypothesis. Her subtext seems at variance with what she is on the surface saying.

    As for the statistical analysis, my point was that since he’s finding the same pattern in 8 different countries, largely unrelated linguistically and culturally, while it doesn’t rule out the cultural diffusion explanation, it makes it increasingly unlikely. But again, this kind of statistical analysis is not at all my field, so my understanding of it is largely informed by Chen’s own description. Of course, the only way to be sure would be to isolate the effect through experimentation, and on that point I think we both agree.

    In fact, I have my own misgivings about the research, though I haven’t had a chance to blog about them yet. In addition to the linguistic problems and imprecisions (which Dahl and others have raised), it all seems too tidy. While future-time marking may be one factor (as seems plausible to me), it surely can’t be the only one influencing what is highly complex human behaviour, and we have no idea how other factor, linguistic or not, interact. One would at least expect to see outliers. Plus, can you really categorise families in terms of one language? Would there not frequently be some level of bilingualism? I would imagine this kind of data might point towards areas of interest, nothing short of experimentation which could isolate the linguistic influence will give a clear answer. So I take the research in that light.

    Thanks again, Jeff, for the discussion!

  3. Pingback: Kalamazoo Time « The Endless Knot

  4. Pingback: Clip show: Links to links part 2 | The Endless Knot

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Blog at WordPress.com. The Adventure Journal Theme.

Follow

Get every new post delivered to your Inbox.

Join 742 other followers

%d bloggers like this: