“More research from the lunatic fringe”

A linguist send me an email with the above title and a link to a paper, “The Effect of Language on Economic Behavior: Evidence from Savings Rates, Health Behaviors, and Retirement Assets,” by M. Keith Chen, which begins:

Languages differ widely in the ways they encode time. I test the hypothesis that languages that grammatically associate the future and the present, foster future-oriented behavior. This prediction arises naturally when well-documented e§ects of language structure are merged with models of intertemporal choice. Empirically, I find that speakers of such languages: save more, retire with more wealth, smoke less, practice safer sex, and are less obese. This holds both across countries and within countries when comparing demographically similar native households. The evidence does not support the most obvious forms of common causation. I discuss implications for theories of intertemporal choice.

I ran this by another linguist who confirmed the “lunatic fringe” comment and pointed me to this post from Mark Liberman and this followup from Keith Chen. My friend also wrote:

I think it’d be well-nigh impossible to separate the effect of speaking West Greenlandic from living in West Greenland, or more reasonably, speaking Finnish from living in Finland. Who else speaks Finnish (maybe some Swedes?)

My reply:

B-b-but . . . the paper is scheduled to appear in the American Economic Review! Short of Science, Nature, and Psychological Science, that’s probably the most competitive and prestigious journal in the universe.

More seriously, this is an interesting case because I have no intuition about the substance of the matter (unlike various examples in psychology and political science). The theoretical microeconomic model in the paper seems ridiculous to me, that’s for sure, but I have no good way to think about the cross-country comparisons, one way or another.

18 thoughts on ““More research from the lunatic fringe”

  1. Maybe I’m totally missing something here.. it seems to me that Chen’s claims are quite conservative — he’s acknowledging that this is an area where, at best, we can make only predictive rather than causal claims. The underlying correlation seems interesting, and surprisingly robust. Obviously we don’t have enough to make the causal claim, but this seems interesting nonetheless, no? When I think lunacy, I think about people who can’t tell whether a causal parameter is identified, but this doesn’t seem to be Chen’s problem here.

    • I think the problem is where an author states in some places how they are careful to reach causal conclusions, but elsewhere, often in more public arenas which is arguably more damaging, they are not “quite conservative”. So, while his reply in the blog above is aimed at people who believe care is necessary he is more cautious, but his web page says:
      “Professor Chen’s most recent work focuses on how people’s economic choices are influenced by the structure of their language. His work has shown that how a person’s language encodes future events influences future-oriented behaviors as diverse as saving, smoking, and safe sex.”

  2. The prevailing belief in European development is that the rise of Protestantism (particularly the Calvinist influences) lead to wealth accumulation, as a sign of God’s good graces and a ticket to heaven as a member of the elect (that’s sort of the ultimate future-oriented behavior). Southern Europe, on the other hand, in the clutches of a hierarchical and static Catholic church that served the status quo and reaction, remained mired in the economic doldrums. I just mention this because romance languages are heavily represented in the south and germanic languages in the north. Not being a linguist, I have no idea of the degree of differentiation between intertemporal association in the types of languages, but this obviously points out the difficulty of separating language from other cultural references. I can’t judge the paper since I haven’t read it, but like so much of other grand theme historical determinism (“Guns, Germs and Steel”, anyone), there are suspicions about the validity of large claims.

    • Speaking of Europe, I should note that wealth accumulation in Roman times was essentially confined to the Empire (of which the two languages were Latin and Greek, both of which in some form are spoken today), and northern Europe was tribal and unorganized (with no wealth), and various forms of the Germanic (and Celtic) languages were spoken there. Two millenia later, the distribution of wealth accumulation has reversed. Have the languages also modified their temporal associations between present and future?

      • Work ethic and wealth accumulation are different things. The easiest way to accumulate wealth is to inherit it. The next easiest is to have the government hand it to you (tax laws, monopolies, etc). But really, most wealth accumulation has occurred because of the industrial revolution, and this is still a matter of intense debate as to why it occurred (first) in England (and then North America).

      • Popeye, as far as I know the question is still contested between Weber scholars, at least in Germany. Of course, most advocates are usually using more refined version of the argument.

  3. Ever read cultural anthropology books? They have this issue to the max. Even the famous ones – like The Fierce People – are full of bull.

    Problem is it’s – can’t find the word so I’ll give an example: take Raphael Patai’s The Arab Mind. He presents a neat discussion of cultural characteristics that many Arabs would agree with, such as the way boys are placed among women and then removed from their presence, but then he takes that to describe an infantilizing relationship. Maybe. And about language, he notes that Arabic has issues with tenses, something that Western readers might not grasp is relatively common in language, but then he extends that to say Arabs don’t get things done because the tenses confuse saying you’ll do x with actually doing x. Maybe but ….

    I think if we looked in ourselves we’d find this is how we construct our working mental models of the world: telling ourselves a story, extending it until it reaches too far from an objective standpoint while we’re still seeing it works from our perspective to explain what we see.

    Consider Jews. How would you explain the success of Jews? So many academic prizes. So much success in many fields. Look at the Supreme Court. We can easily see that because anti-abortion activists can’t nominate Evangelicals to the Court, there is rationally a link to nominating Catholics. But why are all the non-Catholic Justices Jewish?

    People construct models to explain this. Many are objectively ridiculous but very powerful: the vast Jewish conspiracy crap that somehow manages to exist in places where there are no Jews (or the only Jews are old and poor). We see versions of this every day in the weird allegations – and books by real academics – about how Israel controls US policy … which genuinely need to go to the lunatic extreme that the US is somehow blinded by Jews into acting against its own best interests. People look for explanations and find them in the weirdest places.

    This story-telling or model making is also revealing about what it can’t accept. Given the importance of Christianity in the West and Islam in other places, they can’t open their minds to the idea that maybe Judaism itself is the reason, that the Jewish method of questioning, which is taught through Talmud and is deeply embedded in the nature of Jewish concepts – to the point where you can find ancient lists of Jewish questioning methods (that seem to me to be highly influenced by the Greeks), actually matters in real world performance. That might also be nuts but it makes more sense than the ridiculous notion that, as I daily read, Jews are responsible for Sisi and/or the Muslim Brotherhood and Jews are responsible for the war in Syria or, my most recent favorite, an American “scholar” on Iranian TV saying that Israel was behind the Malaysian airplane disappearance and that there’s a duplicate airplane in a hangar in Tel Aviv. (Have to say I’m not sure what that detail actually means but it sounds ominous.)

    I go through this to note that sometimes oddity ends up being correct, as in Cantor’s exploration of infinity. But that many of these explorations are never going to be provable and that they express our inherent need to order the universe through developing lines of thought – story lines – that make sense of things.

    And of course what you do connects directly to this because you work with story lines that relate what you believe might be true to what happens and so on. Sorry for the length of this comment but I’m having trouble getting my butt out the door today.

  4. the follow up from Keith Chen suggests the work is accessible to being reproduced and replicated and that he has thought through many of the challenges of learning from non-randomised data.

    Perhaps he should send it to that new journal for observational studies – something the editors there and their students likely would find a worthy challenge.

    The strong Worfian hypothesis was sent aside due to low prior and low data support but a weak one of subtle effects had some prior support – now the question is whether data support can come from the observation data now available. That mostly a challenge of causal inference – no?

  5. The Sapir-Whorf hypothesis is pretty plausible in its weak form: cultures will tend to have more words and more precise one for things they care more about: e.g., snowboarders have lots of words for different types of snow (that Flying Tomato fellow in the Winter Olympics could bring you up to date on them).

    Note that Whorf himself worked for a big fire insurance company (he did his scholarly work on vacations) and a big issue in fire safety was that imprecise and confusing terminology can get people killed. I don’t know if Whorf himself was involved in the big and successful push by the fire safety community to change the label on tanker trucks from the confusing “Inflammable” to the better “Flammable,” but that’s representative of the issues he dealt with on his day job:


    • The empty gas cans lesson really hit home with me and also that Worf apparently like Mozart, did not make drafts – the original drafts went straight to publication.

      But in this post, I really do think that _the_ issue is a challenge of causal inference – no? not that the prior for the hypothesis (weak form) is so low to be dismissed outright as it _seems_ in the comments if not also post.

    • Manoel:

      Conditional logistic regression might be better explained as conditioning from a larger to a smaller probability model so that nuisance parameters (e.g. the average response of a pair being contrasted) disappear (aren’t in that smaller probability model at all). The _magic_ is there is very little loss of information about the interest (language effect) parameter.

      Additionally, Hierarchical Bayes _can be_ exactly the same if you use Ken Rice’s priors to get the same marginal likelihoods for interest parameters. Rice; Equivalence Between Conditional and Random-Effects Likelihoods for Pair-Matched Case-Control Studies, JASA 2008

      I too am guessing that the challenge of causal inference is going to be very hard if not too hard but that is the question here (I think) and that will require some careful deliberation.

      As ezra point out below, languages are extremely flexible, but I don’t think that rules out common habits in language use having some effects.

  6. imo, this paper is why econ is not going anywhere: the theory math/data ratio is way to high
    Look at figure one, and then eqn 11 at the top of page 20
    we all know that you can make something significant if you try; just looking at Figure one, would you bother ?
    sort of what reinhart rogoff should have done

    and, I don’t think he can write: isn’t that comma in the second sentance of the abstract un needed ?

    The famous linguistics professor gives a talk on how in every language except english, there is a phrase, made up of two positive words, that has a negative meaning
    at the end, a voice from the back calls out, yeah , right.

Leave a Reply

Your email address will not be published. Required fields are marked *