Whorfian Economics

« previous post | next post »

[This is a guest post by Keith Chen.]

Mark and Geoffrey were kind enough not only to write thoughtful columns on a recent working paper of mine here and here, but to invite me to write a guest post explaining the work. In the spirit of a non-linguist who’s pleased to be discovering this blog, I wanted to use Mark and Geoffrey’s insightful posts as a springboard to explain my work.

In a nutshell: I find a strong correlation between how a language treats future-time reference (FTR), and the choices that speakers of those languages make when thinking about the future. Specifically, in large data sets that survey families across hundreds of countries, I find a strong and robust negative correlation between the obligatory marking of FTR in the language a family speaks, and a whole host of forward-looking behaviors, like saving, exercising, and refraining from smoking. These correlations hold both across countries and within countries, even when comparing effectively identical families born and living in the same country. While the data I analyze don’t allow me to completely understand what role language plays in these relationships, they suggest that there is something really remarkable to be explained about the interaction of language and economic decision making. These correlations are so strong and survive such an aggressive set of controls, that the chances they arise by random lies somewhere between one in 10,000 and one in 10^32.

Starting with Mark’s post: Mark illustrates beautifully an idea that is really the central concern of all work done in modern econometrics: it can often be difficult to tell the difference between strong correlations produced by causal relationships, and correlations which arise through non-causal factors. Since questions of the connection between language and behavior have historically generated considerable controversy, it seems important to think hard about what exactly these correlations actually suggest. Towards this, I’ll discuss briefly why my analysis suggests that a non-causal story is unlikely, and that a language’s structure is causing its speakers to behave differently.

The basic idea is this: I find a strong negative correlation between a family’s savings rate and whether their language has grammatical obligatory marking of future time (call these strong-FTR languages). Obviously, the savings behavior of families can’t CAUSE their language’s structure: structure precedes behavior by large spans of time. What is possible however, is exactly what Mark’s post illustrates: the possibility of something like co-diffusion (correlated adoption) of both language and non-linguistic values, culture, or institutions. For example, might a value towards education have spread along with certain languages? More generally, might languages have co-diffused with attitudes towards work? With particular institutions or religions?

Most of my paper is spent checking for exactly these types of concerns, and what I find is that the data don’t suggest that they explain the correlations I find. How can we know? Well, if we thought languages co-diffused with proclivities towards education, this would suggest we should compare families with identical levels of education to see if the correlation persists. If we though the co-diffusion was mainly spatial (what Mark’s post is mainly concerned with), we might think to compare families born and living in the same country. Effectively, by trying to control for confounding factors that could be driving a spurious correlation between language and behavior, we can get a sense of what these patterns across families really suggest. This leads to a kind of statistical analysis epidemiologists call conditional-logistic regression. What this does, in effect, is drop families around the world into one of 1.4 billion buckets, where 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.

Now, at this level of detail, even the largest economic data sets only allow me to look at around twenty-five thousand families around the world. These families live in eight countries with enough native linguistic diversity to form sets of nearly identical families who also speak different languages. Those countries are Belgium, Burkina Faso, Estonia, Ethiopia, Malaysia, Nigeria, Singapore, and Switzerland. Surprisingly, what I find is that in every one of those countries, the strong vs. weak FTR language categorization I adopt from Östen Dahl (and the EUROTYP project) seems to have a large and consistent correlation with savings behavior. That is, the direction and the magnitude of these effects are statically identical in every country, despite spanning different regions of the world, and different language families.

These effects can be measured not only in savings behavior, but in many different future-regarding behaviors studied by economists. For example, exercising can be thought of as investing time and effort in exchange for future health. Smoking and overeating are basically the opposite: present pleasure at the expense of future (health) costs. I find exactly the same effect in all three of those behaviors, and only when comparing nearly-identical families. Most surprisingly: language FTR structure isn’t correlated with a self-reporting value of saving, even though both have strong effects on behavior. That is, families that report valuing saving (unsurprisingly) save more, as do families that speak weak-FTR languages. But these effects appear independent of each other: language appears to have the same quantitative effect on families whether or not they report value savings. If language simply co-diffused with a cultural value towards saving, we would not have expected to see this.

In short, the data simply don’t seem to suggest that co-diffusing factors play a significant role in explaining what is an extremely consistent set of correlations between the grammatical marking of future reference in languages and many different future-regarding behavior.

Hopefully I’ve conveyed clearly that these correlations are surprisingly strong and survive a surprising amount of attempts to eliminate confounding factors. The working paper goes into much more detail and conducts many more tests than I could describe here, but all analyses I conduct are on publicly available data, and I’d love to talk more with readers interested in replicating, extending, or testing these results in ways I haven’t thought of. The paper also tries to explain why we might have thought that weak-FTR languages would lead to more savings, which comes down to a pretty simple intuition. Every theory of discounting studied in the behavioral sciences is strictly convex (that is, the value of a future reward is a strictly convex function of when you receive it). What I show is that given this, if languages with grammatical marking of the timing of events lead to more precise beliefs about timing of future rewards (an effect similar to many that have been demonstrated by psychologists), then the effect I find is what theory would suggest. Even assuming that language drives savings behavior, though, this analysis does not resolve the question of how it does this: especially because as Geoff points out in his extremely thoughtful and insightful post, it’s not entirely clear what the EUROTYP typology is measuring.

The main point I take away from Geoff’s post is that these types of typological classifications can be exceedingly difficult to interpret, and what they are measuring may not be entirely clear, even after quite a bit of thought. Geoff discusses the example of English, which is characterized in Dahl’s EUROTYP typology as strong-FTR (or more accurately, non weak-FTR). For those readers unfamiliar with this typological distinction, English is considered a strong-FTR language because of observation that unlike most Germanic languages, English generally requires speakers to grammatically mark future events, primarily with either the de-andative construction “be going to” or the de-volative construction “will”. It is this generally tendency toward obligatory grammatical marking of future time that characterizes the strong vs. weak FTR distinction.

What Geoff points out is that there are many exceptions to this, and that the exact nature of these constructions is complicated: neither is purely a tense marker. Indeed, there is substantial evidence that neither construction is a tense marker at all, but instead mark different temporal and modal properties which give rise to future reference in certain contexts (“going to” is prospective aspect, while “will” can be a modal auxiliary). Many linguists may wonder: if the precise function of the EUROTYP classification is not entirely clear, even for deeply studied languages like English, what are we to make of aggregate correlations between this classification and behaviors?

I share this concern, and think it’s worth thinking hard about both what it suggests careful work should look like, and how we might interpret results that survive careful analysis. Though I plan to take more steps in the direction of investigating more fine-grained linguistic distinctions in future work, I think it is important to note that despite these issues we can still learn a lot from correlations between even rough typological distinctions and behaviors.

To see this, note that while the obligatory grammatical marking of future time may collapse several important grammatical features, it is reasonably clear that it measures some distinction in how languages treat time, and that this type of distinction is stable enough to have been discussed by several different authors studying TMA systems. For example, the EUROTYP classification of English as a strong-FTR language and German as a weak-FTR language reflects similar distinctions in earlier work such as Comrie (1985), which goes so far as to use English and German as exemplars:

"Within languages that make a basic present-vs-past distinction, it is worth distinguishing two sub-types… which define end points on a continuum. The one would include languages where the present can always be used with future time reference, the only constraint on this use being avoidance of interpretations with present time reference… German and Finnish would fall into this category. At the opposite extreme would be languages where, although the present can be used with future time-reference, there are several constraints on the use of this form, constraints that are not explainable purely as strategies to avoid conversational implicatures. English would be an example of this category, since the present can be used with future time reference only under highly specific circumstances…"

Copley (2009) also notes this distinction in how English and German make reference to future events, noting a “plannability restriction” in English (on the use of present forms to refer to future events) that is not present in German. None of this is to claim that existing typological classifications are measured without considerable noise, or even that we have a good idea of what a complete typology of future-referring strategies would look like. What is important, though, is that the EUROTYP typology captures something about how future events can and cannot be spoken about across languages. That it also seems to capture an important tendency in future-directed behavior, is both surprising and bears explanation.

Note also, that while typological coarseness may complicate the interpretation of the correlations I find, if anything it strengthens the inference that language and economic behavior are interacting in powerful ways. Why? Well, since these typological distinctions are necessarily noisy, if anything, the correlations we see in data should be an UNDERESTIMATE of the correlations we would find if these characteristics of language could be measured more precisely. To see this, imagine that the EUROTYP FTR classification was effectively no better than a coin flip. Then, we would have expected to find no correlation at all between language and behavior out in the world. More generally, if we think that the classification is by its very nature coarse and messy, then we should also think that if anything, language structure and savings behavior are MORE closely tied than their correlation suggests.

In short, I believe the data suggest a strong and robust relationship between linguistic and economic data, a relationship that bears explaining. Where this leaves us is what I think is an exciting place: one where Economists have a lot to learn from Linguists.

(I gratefully acknowledge the helpful comments of Nicole Palffy-Muhoray. All errors are of course my own.)

[Above is a guest post by Keith Chen.]


  1. Alex said,

    February 21, 2012 @ 5:59 am

    This is very interesting. I'm sure you are doing this already, but if you want to establish a causal relationship, why not do some experiments? For example, a cross-linguistic study using a delay discounting task.

  2. es said,

    February 21, 2012 @ 6:29 am

    Now that prof. Chen has makes the relevant linguistic distinction clearer, I'm even more puzzled by the assignments of languages to one class or the other. I don't know wether the assignment of Italian to the Strong FTR class is Chen's or is due to EUROTYP, but it is obviously untenable, at least if one takes the example of German seriously.

    Italian has future tense grammatical forms (synthetic forms, rather than periphrastic as in German): "andrò" ("I will go"). But the most common way to express future , in the spoken language at least, is by far the simple present: "Vado al seminario" ("I'll go the seminar"), "L'anno prossimo vado in Canada" ("Next year I'm going to Canada"). The planned/non-planned distinction which is relevant in English does not seem to play any obvious role, by the way.

  3. GeorgeW said,

    February 21, 2012 @ 6:34 am

    Do some societies speak about the future more than others? If so, would the speakers in the more future-oriented societies be more likely develop overt linguistic markers?

  4. Peter said,

    February 21, 2012 @ 6:47 am

    A very interesting and detailed post! One point that puzzles me: Chen controls against co-diffusion by looking at people living in the same country speaking different languages, and showing that the correlation remains.

    But this, surely, would be expected from the co-diffusion explanation too. In most cases that different languages co-exist, they do so in step with more or less different cultures. Similarly, when immigrant individuals/families/communities retain their original languages, they usually retain much from their original cultures, especially deeply-ingrained ideas like “the value of education”.

    What happens when one compares people speaking the same language in different countries — for instance, USAians and Brits, Iberians and various Latin Americans, Arabs from various countries… In these cases, the cultures have been separated long enough to diverge significantly in some ways, but the grammatical structure of the language hasn’t changed much.

  5. Claire Bowern said,

    February 21, 2012 @ 6:55 am

    If this is a general pattern, wouldn't we expect to find it in other domains apart from monetary savings behaviour? For example, how about hunter-gatherer food storage? Delayed-return trading strategies? Hunter-gatherer groups are known to vary considerably along these dimensions. If this is truly a robust relationship that's grounded in beliefs about the timing of future rewards, it would be extremely odd if it only surfaced in savings behaviour.

  6. Ben Hemmens said,

    February 21, 2012 @ 7:29 am

    English is considered a strong-FTR language because of observation that unlike most Germanic languages, English generally requires speakers to grammatically mark future events,

    Of the other Germanic languages, I only know German well, and it is much the same as English, minus the "going to" future. The future with "werden" is (I think) exactly analogous to that with "will" in English (I do wish the issue about whether to call it a future tense could be decided the same way for both languages – sigh) and the use of simple present tense with future meaning is also more or less the same in both languages. What do the other Germanic languages do that's so different?

  7. PeterL said,

    February 21, 2012 @ 7:45 am

    Japanese conflates present and future tense (it's often called the "non-past" tense), yet the Japanese are well-known to save a lot (except for the government, which borrows a lot). Smoking has decreased greatly from a few years ago.
    And if you've ever worked with a Japanese company, you'll realize that they are unsurpassed at building and executing detailed plans.

  8. John said,

    February 21, 2012 @ 7:48 am

    First, kudos to Prof. Chen, both for an interesting topic and for discussing it here.

    On the topic…

    I'm with es: Italian is incredibly weak in marking the future in practice, which presents a challenge to many English-speaking learners of the language.

    On this:

    Though I plan to take more steps in the direction of investigating more fine-grained linguistic distinctions in future work, I think it is important to note that despite these issues we can still learn a lot from correlations between even rough typological distinctions and behaviors.

    Uh, no. If the linguistic distinctions are bogus, they're bogus, no matter what correlations seem to exist in them.

  9. Ben Hemmens said,

    February 21, 2012 @ 7:53 am

    Ah, I see from the working paper that English has been classified as a strong-FTR but German as a weak-FTR language.

    That must be why I (Irish) still save less than my Swiss relatives ;-)

  10. Ben Hemmens said,

    February 21, 2012 @ 8:00 am

    the planned/non-planned distinction which is relevant in English

    I think a lot of this explanation of when we use different futures in English is made up. Of all the crappy explanations I was told to spout in my English teaching days, these were the least convincing. I think we choose the form, in speech, with a large amount of randomness.

  11. Östen Dahl said,

    February 21, 2012 @ 8:06 am

    It's a bit scary to see your own analyses invoked in this kind of discussion. But let me note here that the way Keith Chen describes them they sound much more categorical than they are in my own texts. In the paper from the EUROTYP volumes (Dahl 2000) that Chen refers to, I discuss the expression of future time reference in European languages from different points of view. One of the phenomena that I spend some time on is what I call the "futureless area" in Northern Europe, referring to a number of Germanic (excluding English) and Finno-Ugrian languages (also historically including Slavic) which lack inflectional futures and in which future time reference is less systematically marked grammatically than in the rest of Europe. I do not specify a binary classification of European languages (let alone the languages of the world) and I do not use the terms "strong-FTR language" and "weak-FTR language". (In the abstract of his working paper, Chen says "what linguists call strong-FTR languages" — what he should have said is "what I call strong-FTR languages"; Google Scholar yields no hits for the phrase "strong-FTR language" except Chen's own paper.) In my discussion of the "futureless" languages of northern Europe, I focus to some extent on predictive statements such as those found in weather forecasts; in the version of Chen's paper that I have seen, he also takes this as the criterion for distinguishing the two classes of languages he postulates, but in the blog post above, he seems to have forgotten about this and talks of "obligatory marking of FTR" in general. It is quite clear that FTR marking differs cross-linguistically on many parameters for which information is often lacking in grammars; precisely for this reason, a single criterion — inflectional marking — was focused upon in the chapter on future tenses in WALS (Dahl and Velupillai 2005).

    The working paper contains a table that purports to analyze countries in which languages are spoken which differ on the FTR parameter; I don't really know what kind of statistics is behind the numbers in this table; for instance, in Nigeria the speakers of the three languages English, Hausa, and Igbo are said to make up 70 percent of the population, but even counting generously the real percentage is no more than 30 (remember that Nigeria is home to around 500 languages).


    Dahl, Östen. 2000. The grammar of future time reference in European languages. In Tense and aspect in the languages of Europe, ed. by Östen Dahl, 309-328. Berlin: de Gruyter.
    Dahl, Östen and Velupillai, Viveka. 2005. The Future Tense. In World Atlas of Language Structures, ed. by Bernard Comrie, Matthew Dryer, David Gil and Martin Haspelmath, 270, 278-279. Oxford: Oxford University Press. http://www.livingreviews.org/wals/feature/67.

  12. Matthew Stephen Stuckwisch said,

    February 21, 2012 @ 8:14 am

    I think the issue on this board here is less with Chen's paper/analysis and more so with the classifications that he used. Admittedly, he didn't come up with them himself, he used the EUROTYP ones which seem pretty off to a lot of us.

    Perhaps one way to improve the analysis would be to grab a standard corpus, and see the percentage by which in spoken (vs written even) language, the present tense is used to express the future time frame versus using a strict future tensing. That way it's not a black/white classification, but rather graduated which could show a significantly stronger level of correlation (or weaker, of course).

    And for my two cents on EUROTYP and Spanish: the future tense isn't obligatory in many cases. In fact, in the subjunctive mood, with the exception of hyperformal/archaic texts, only the past and present is used. In the indicative, upon very casual reflection, for future perfective instances, Spanish seems to rarely if ever require it, but for future imperfective, it does. Though I'm sure I can find some solid counter examples to both. The synthetic future form is also used more probabilistically these days with the constructive (going to) form used more often especially in speech. Ultimately, I'm not really sure I would classify Spanish as a strong FTR and as es notes, it seems at least one other Romance language is equally questionably included.

  13. Ben Hemmens said,

    February 21, 2012 @ 8:14 am

    Many thanks to Östen Dahl. That's exactly what I was wondering, having browsed around a few of the EUROTYP documents and failed to find the binary classification.

  14. D Sky Onosson said,

    February 21, 2012 @ 10:12 am

    I, like others, am not convinced of the notion of being able to neatly categorize languages in terms of use of the future tense. There are many ways to incorporate future reference into language, tense being only one of them.

    But one thing that does strike me is that there is a significant difference between explicit reference to the future (through tense or the use of future-referring vocabulary like "tomorrow"), and implied reference that must be interpreted from context. While not a clean divide either, languages do tend to fall into two camps when it comes to explicit vs. implicit reference, and not just when it comes to the future. Japanese is a well-known example of a language that tends to rely heavily on context in a whole host of ways, from low frequency of use of pronouns, to not having a future tense, for example.

    In this regard, I wonder if it might be fruitful to explore the relationship between context-dependent and context-independent languages and/or language use, and the kinds of forward-looking behaviours that Dr. Chen is talking about.

  15. William Ockham said,

    February 21, 2012 @ 10:17 am

    Obviously, the savings behavior of families can’t CAUSE their language’s structure: structure precedes behavior by large spans of time.

    This isn't obvious to me. If we assume that savings behavior is a culturally transmitted trait (which may or may not be true, but seems entirely plausible to me), the causal arrow could indeed point towards behavior causing a language's structure. The need to talk about the future in particular ways would influence the way a language evolves. This distinction could be important and perhaps even testable. Again, assuming the validity of the underlying correlation (something I have no opinion on because I don't have the expertise to evaluate it in any meaningful way), one could look at first generation language speakers whose parents spoke a differently typed language (in this typology). I would think immigrant communities that had roughly equivalent levels of new language adoption but different levels of cultural assimilation would be particularly interesting. Also, if the causal arrow really points from language to behavior, I would also expect to see measurable differences between bilingual speakers and strictly monolingual speakers.

  16. John Shutt said,

    February 21, 2012 @ 10:37 am

    Fwiw, I think a great deal of controversy surrounding "Whorfian" whatnot results from over-analysis: supposing a clean distinction between "language" and "culture" when really the two are so deeply entangled it's barely useful to distinguish them at all. Most, perhaps all, of language is part of culture, indeed a large part of culture, and indeed a very central part of culture as it serves as medium or catalyst for a great deal of cultural communication. Linguistic and non-linguistic facets of culture may tend to reinforce each other, but even that may be on the verge of over-distinguishing them. And here, it seems one may be trying to distinguish between a behavior being caused by culture versus by linguistic structure, when those "two" causes aren't separate from each other.

  17. Dierk said,

    February 21, 2012 @ 10:58 am

    This reminds me a lot of a hypothesis I once heard and like to use in jest myself:

    Africans and South-Americans tend to not think of the future and saving for it because nature has always something to eat easily pluckable all year round. Europeans and non-tropic Asians, on the other hand, have to think of the future and saving for it because only a small part of the year yields enough food. This clearly informs their languages …

    What I like to know, Mr Chen, is if your study categorised Belgium and Switzerland for their different language groups and found any economic variance between French, Italian, German, Rhaeto-Romance in Switzerland or French, Frisian, and German in Belgium?

  18. es said,

    February 21, 2012 @ 11:09 am

    @John Shutt
    I agree up to a certain point. I have never seen convincing evidence that people cannot share a language without sharing a culture and vice versa. Actually, one could think of many facts that suggest that such dissociations are possible. So I would not go as far as say that language and culture ARE objectively entagled. But I think that for a scientist to distinguish the two may be highly non-trivial at times. Think of some abstract nouns like "freedom", "love" and so on. In such cases, I don't know what difference it would make attaching different meanings to these words, or just having different concepts of freedom and love (this echoes something Donald Davidson argued for somewhere, but I'm not sure). Anyway, this is one of many reasons why when I read about far-reaching conclusions about cultural differences or language-thought connections I often reach for my gun…

  19. John Roth said,

    February 21, 2012 @ 12:34 pm


    I thought Prof. Chen specified a negative correlation, hence the Japanese example you cite is as expected, not a counter-example.

  20. Coby Lubliner said,

    February 21, 2012 @ 3:30 pm

    What Keith Chen's work best illustrates is the by now well-established tendency of economists to present themselves as experts on everything, typically using half-digested, poorly understood results from other disciplines such as mathematics, statistics, biology and, now, linguistics.

    One might have thought that, in view of their disastrous record in their own field in the last few years, practitioners of the dismal science would prefer to keep a low profile. But then one might have thought the same of the Catholic episcopate. It seems evident that in both instances, as with politicians, arrogance trumps discretion.

  21. Ben Hemmens said,

    February 21, 2012 @ 4:43 pm

    Looking at the figure in Chen's paper, it seems like we would need to take some people from Burkina Faso, give them a Swiss income and see whether they save like Malaysians. The inverse experiment might be even more revealing. It all reminds me of a certain wisecrack that has the form of a response to the phrase "Jesus saves!"

  22. peterv said,

    February 21, 2012 @ 5:53 pm

    " For example, exercising can be thought of as investing time and effort in exchange for future health. Smoking and overeating are basically the opposite: present pleasure at the expense of future (health) costs. "

    Except that these behaviours too are influenced by cultural factors (in some societies, fatter people are more highly admired or more commonly seen as beautiful than in others) AND are influenced by factors relating to body chemistry that differ across ethnic populations. People in some of the studied populations could more readily experience physical (ie, biochemical) addiction to nicotine, for instance, than people in others, just as some ethnic populations can more readily digest milk, or better tolerate alcohol. Nothing to do with views on the future, or an implicit discount rate, or anything except simple body biochemistry.

  23. Matthew Stephen Stuckwisch said,

    February 21, 2012 @ 7:27 pm

    Östen Dahl, thanks for the post. I was wondering why I, like Ben Hemmens, couldn't find some binary list from EUROTYP from a cursory search on-line, now I know why!

  24. Hayden said,

    February 21, 2012 @ 8:51 pm

    Interesting post, but my question, Keith Chen, is what led you to the idea that some grammatical feature of a language you speak could have anything at all to do with your economic activities?

    Surely culture is a big influence in your economic activities, but the language(s) you speak just incidental?

  25. Philip Spaelti said,

    February 22, 2012 @ 4:03 am

    As everybody knows, the locus in the brain of FTR marking is the crokus.

  26. Burns Cooper said,

    February 22, 2012 @ 8:08 am

    I'm curious about the following parenthetical comment in Prof. Chen's post, which may suggest where the idea for this correlation comes from:

    "if languages with grammatical marking of the timing of events lead to more precise beliefs about timing of future rewards (an effect similar to many that have been demonstrated by psychologists), then. . . "

    I'm not sure how 'similar to' is being used here. What have psychologists actually demonstrated? If they've demonstrated this grammar/ 'timing of future rewards' effect, then the economic correlation may be unsurprising. Psycholinguistics is not one of my strong points (to say the least), but I haven't heard of this.

  27. Chris said,

    February 22, 2012 @ 8:37 am

    I haven't had achance to read the working paper yet, but I've read the above post and LL's previous coverage and I'm a little skeptical. As a linguist, the most frustrating omission from Chen is the lack of historical linguistics. Languages change, do spending habits change with them? Is there any way to track the spending habit changes as languages change over time? If Chen is right, the simple prediction is yes, spending habits should follow FTR changes, right?

  28. kharris said,

    February 22, 2012 @ 9:07 am

    Some of the comments suggest the use of tense marker exclusively, rather than broader indications of future time. The key point here, since Chen chose EUROTYP typology, is what indications of future time EUROTYP employs.

    It's conceivable that there is just a huge spurious correlation, an amazing statistical artifact, in the comparison of non-linguistic behavior over time and EUROTYP typology, and nothing more. However, setting aside concerns over statistical methods, what Chen has shown is a strong negative correlation between EUROTYP categories and non-linguistic behavior that involves planning. There are, however, issues with exactly what EUROTYP measures. Chen's result (setting aside methodological concerns) shows that EUROTYP measures "something", even if that "something" is not well defined. The thing to do, then, is to figure out consider what EUROTYP measures. I'm not familiar enough with it to add to the debate, but I suspect other commenters are. Is there any consensus about EUROTYP measures other than what it's creators claim for it?

  29. peterv said,

    February 22, 2012 @ 12:18 pm


    "what Chen has shown is a strong negative correlation between EUROTYP categories and non-linguistic behavior that involves planning."

    Not so. Rather:

    what Chen has shown is a strong negative correlation between EUROTYP categories and non-linguistic behavior that someone ignoring cultural and biomedical factors has assumed to involve planning.

    I am contesting the extent to which the non-linguistic behaviors studied are volitional. If these behaviors are not freely chosen then the research falls in a heap.

  30. es said,

    February 22, 2012 @ 12:28 pm

    I don't know about obesity, but as for saving money I'm not sure biological factors can be plausibly invoked. Personally, I don't trust Whorfianisms of any sort, but biological determinism does not seem very promising as a general approach to culture either. Not to mention the fact that it sounds so terribly 'Nineteenth Century'… (just kidding).

  31. Assorted links — Marginal Revolution said,

    February 22, 2012 @ 12:36 pm

    […] 5. Whorfian economics. […]

  32. Östen Dahl said,

    February 22, 2012 @ 1:31 pm

    May I ask the discussants not to use expressions like "EUROTYP typology", "EUROTYP categories", and "EUROTYP measures", at least not without having read my paper referred to in my previous comment (and that of Rolf Thieroff in the same volume, also referred to by Chen)? In particular the word "measure" suggests a much more exact approach than we applied.
    What Thieroff and I did was to identify an area in northern Europe where languages lack a fully grammaticalized future tense (somewhat uncautiously labelled "futureless" by me). But we did not give criteria that could easily be applied in order to divide the languages of the world into two classes. As Chen makes clear in section 4.1. of his working paper, the extension to other languages is wholly his own.

  33. andrew clearfield said,

    February 22, 2012 @ 1:53 pm

    very interesting, but to say that "because a correlation persists despite substantial noise, the correlation is probably even stronger than the results indicate" is just bad statistical analysis. You're trying to evaluate the relationship between savings and a specific aspect of language, FTR. But by not properly isolating FTR (and I'm not saying that you're method is actually failing to do this, but you admit that this might be the case) you have no way of knowing whether the correlation you see is actually between savings and FTR or between savings and some other aspect of language that has been mistakenly used to measure FTR (but actually measures something else entirely). In other words, all you can say (and I know this is obvious, but it's still worth stating explicitly) is that there is a clear correlation between savings and whatever formula is being used by Eurotyp to measure FTR, not between savings and FTR itself.

  34. Östen Dahl said,

    February 22, 2012 @ 2:45 pm

    EUROTYP has, or had, no formula to measure FTR. To measure FTR would not really make any sense. What could be measured is the degree to which marking of FTR is grammaticalized. But EUROTYP offers no formula for that either. EUROTYP is not a piece of software but was a huge research programme involving about a hundred linguists over five years (1990-94).

  35. es said,

    February 22, 2012 @ 3:13 pm

    I wouldn't like to be in prof. Dahl's shoes… I'm afraid he's going to have to repeat the same explanation again and again…

  36. peterv said,

    February 22, 2012 @ 3:32 pm


    I am not claiming biological factors impact saving behaviors (although they may do) but cultural factors certainly do, and in ways which limit individual volition. For example, ask what is the entity doing the saving — the individual, the nuclear family, the extended family, or the entire community or village? The answers differ from one population to another, and it may make no sense in some cultures to talk about individual decision-making powers when decisions are made by larger groups. This is common sense to any anthropologist although economists still seem to struggle with any notion of decision-making other than an individual-based one.

  37. es said,

    February 22, 2012 @ 4:02 pm

    I see. It sounds very reasonable.

  38. Karsten Kant said,

    February 22, 2012 @ 5:26 pm

    The attempt to account for confounding factors reduces each sample to minuscule size. I have trouble trusting the statistical analysis.

  39. Bill Fleming said,

    February 22, 2012 @ 9:24 pm

    Future tense is not used in Swiss German although there is a lot of subjunctive. (Ref.: Schwyzertu"u"tsch, Arthur Baur)

  40. andrew clearfield said,

    February 23, 2012 @ 12:06 am

    Re: Osten Dahl,
    you write: "What could be measured is the degree to which marking of FTR is grammaticalized. But EUROTYP offers no formula for that either. EUROTYP is not a piece of software but was a huge research programme involving about a hundred linguists over five years"

    Ok, but my point remains. You are trying to assert a correlation between savings and "the degree to which marking FTR is grammaticalized" (to use your words). Well, EUROTYP used a formula (call it what you want, but having an army of linguists measure the "degree to which FTR is grammaticalized" according to a series of rules and intuitions is equivalent to a "formula" for the purposes of my comment) to classify these languages. Then you say, but we admit this formula (again call it what you want) is noisy, meaning there is a lot of grey areas in the classifications of the degree to which FTR is grammaticalized. You then assert (and here's where you go wrong) that the fact that the correlation between savings and degree to which FTR is grammaticalized is strong in the face of so much noise means the correlation would probably be even stronger if the noise was somehow eliminated. But my point is that this conclusion ignores the possibility that something else (some of the noise) may be what's causing the correlation in the first place, and so it's possible that savings and degree to which FTR is grammaticalized are completely uncorrelated in your data set but you just haven't managed to isolate the variable.

  41. RF said,

    February 23, 2012 @ 4:44 am

    @andrew clearfield: You are replying to Östen Dahl as if he is the author of the post, or is involved in the work, or is defending it in some way. He is not. So when you say "you are trying to assert a correlation" etc., you are speaking to the wrong person.

    You are also mistaken about Dahl's work on EUROTYP. I think he was very clear about this so I will copy his comments (above) here:

    "[In my paper] I do not specify a binary classification of European languages (let alone the languages of the world) and I do not use the terms 'strong-FTR language' and 'weak-FTR language'."

    further expanded in a later comment:

    "What Thieroff and I did was to identify an area in northern Europe where languages lack a fully grammaticalized future tense (somewhat uncautiously labelled "futureless" by me). But we did not give criteria that could easily be applied in order to divide the languages of the world into two classes. As Chen makes clear in section 4.1. of his working paper, the extension to other languages is wholly his own."

    To your larger point, that the "noisiness" may very well be causing, rather than weakening, the correlations found, I agree. The fact that Chen categorized languages himself (consulting available grammars to perform the classifications) certainly hints at the possibility of bias (however unintentional).

  42. Östen Dahl said,

    February 23, 2012 @ 7:58 am

    Thanks, RF, for defending me, that saves me some time. Andrew Clearfield merges Chen, EUROTYP, and me into some kind of neo-Whorfian Al Qaida. The picture of an army of linguists hired to apply some arcane formula to the world's languages in order to establish a connection between future tenses and savings is an amusing one, though. I should have been a bit clearer about what EUROTYP was. It was a research programme sponsored by the European Science Foundation with the aim of studying European languages from a typological point of view. It consisted of ten "thematic groups", each focusing on a particular area of grammar, phonology, and pragmatics. One of these was the group on "Tense and Aspect", led by me. Each group published a "final volume" of up to one thousand pages. The two papers referred to by Chen make up about one per cent of the total. The EUROTYP volumes have absolutely nothing to say about savings, smoking or health. I am personally no less skeptical towards Chen's claims than Andrew Clearfield.

  43. kharris said,

    February 23, 2012 @ 8:27 am


    "Not so. Rather:…"

    I believe you meant to say, or in any case were only justified in saying:

    "I object" or "I disagree". The fact that you hold a certain view in no way makes other views wrong. I realize that academic debate often stoops to the pretense that strongly held views are reality, but they are not.

    Now, I'm going to have one more try at this. There are two broad points regarding Chen's work that seem vulnerable to examination, aside from strongly expressed personal views about how the world works. There are that Chen either has his statistical analysis right or he doesn't, and that if his statistical analysis is right, then there is something going on, but objections raised about the interpretation of EUROTYP measures (or non-measures, claims one author of EUROTYP) means we don't really know what the "something" is that Chen has found. The rest seems clamor that largely ignores what we can know about Chen's work. Has he found anything, and if so, what has he found?

  44. peterv said,

    February 23, 2012 @ 1:59 pm


    You are incorrect in your presumption about what I meant to say. I said exactly what I meant.

    You did not express an opinion. You made a claim about what Professor Chen's work had shown. Your claim was incorrect, and I said so. I even explained why your claim was incorrect. You are free to tell me what precisely about my reasoning there was wrong.

    Re your second point: there is a third possibility: The statistical analysis in the paper may be right and yet the conclusion of "something going on" may still be invalid. For example, the variables between which a statistical correlation has been found may be meaningless, or may be subject to definitional contestation (as I demonstrated with the behavioral variables), or, most simply, a statistically-significant result may be a false positive. In other words, the null hypothesis of no correlation may be rejected when it is actually true. With a 95 percent significance level we would expect 1 false positive every 20 tests.

    This last reason is why statistical experiments are normally subject to replication and verification before the research community in question accepts their conclusions. But for replication to be sensible we first need agreement on the variables being tested for correlation. We ain't got that yet.

  45. Back to the Future « The Endless Knot said,

    February 24, 2012 @ 3:50 pm

    […] 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 […]

  46. Speaking of the future: Keith Chen at TEDGlobal 2012 | Krantenkoppen Tech said,

    June 28, 2012 @ 5:55 pm

    […] hash over the question, which is sparking ongoing debate and refinement (read a key blog post and Chen’s response); he’s working with the linguistics community at Yale to refine the theory. As he says: “This […]

  47. Speaking of the future: Keith Chen at TEDGlobal 2012 | Indoor Digital Billboards said,

    June 29, 2012 @ 10:03 pm

    […] hash over the question, which is sparking ongoing debate and refinement (read a key blog post and Chen's response); he's working with the linguistics community at Yale to refine the theory. As he says: […]

  48. Paper Talk: Language and Saving Behavior – REVISITED! « epicnomics said,

    November 23, 2012 @ 2:05 am

    […] Chen steps in to defend his paper! And he does a good job, especially on the correlation versus causation point. Of course, as […]

  49. Whorfian economics reconsidered: Why future tense? | Replicated Typo said,

    February 26, 2013 @ 9:27 am

    […] by several linguists, notably on language log (and a great model post by Mark Liberman), where Chen gave a response. The data has been criticised (e.g. English is marked as 'strong future tense marking', […]

  50. How Your Native Language Shapes Your Health and Spending Habits | InternetSuccess4You said,

    September 12, 2013 @ 3:19 pm

    […] The Effect of Language on Economic Behavior Whorfian Economics Basic Color Terms: Their Universality and […]

RSS feed for comments on this post