One for Diogenes

« previous post | next post »

The philosopher Diogenes of Sinope was eccentric, to say the least — he begged for a living, slept in a large ceramic jar in the marketplace, and discarded the wooden bowl that was his only possession, deciding that it was excess baggage. He refuted the Platonic definition of human as "featherless biped" by exhibiting a plucked chicken. In response to the hypothesis that humans are rational animals, he wandered about in daylight with a lantern, explaining that he was looking for a rational individual — usually described in modern versions as looking for an honest man. Plato described him as "a Socrates gone mad".

But if Diogenes were still around, I'd put a tetradrachm or two in his hand, and urge him to go have a talk with Keith Chen.

Back in 2012, Keith published a working paper arguing for a Whorfian effect of tense morphology on cross-cultural variation in future-oriented behavior — a later version is available as "The Effect of Language on Economic Behavior: Evidence from Savings Rates, Health Behaviors, and Retirement Assets", American Economic Review, 2013).

Geoff Pullum expressed skepticism, based on concern about the nature and interpretation of the linguistic facts ("Keith Chen, Whorfian economist", 2/9/2012); and I expressed skepticism, based on concern about the nature and interpretation of the statistical model  ("Cultural diffusion and the Whorfian hypothesis", 2/12/2012).

Keith Chen responded to these concerns in a Language Log guest post ("Whorfian Economics", 2/21/2012), which impressed me with its serious and open-minded attempt to engage our criticisms.

Julie Sedivy added that "correlational studies can only get us so far when there are several viable competing explanations for the correlation", and suggested some experimental approaches that would shed further light on the question ("Thought experiments on language and thought", 2/22/2012).

A year later, Sean Roberts raised some more specific objections about the modeling ("Whorfian Economics Reconsidered", Replicated Typo 2/26/2013), which turned out not to be a strong as he thought ("Whorfian Economics Reconsidered: Residuals and Causal Graphs", 2/28/2013):

Yesterday I posted an analysis of some work by Prof. Keith Chen on the link between future tense marking and economic decisions.  Prof. Chen made some suggestions about changes to the analysis, some of which I’ve carried out here.  The new results below indicate that the link between future tense and the propensity to save is more robust than the previous post suggested, which is quite embarrassing, but I submit the findings here anyway.

But the interaction didn't stop there — Keith Chen has joined Sean Roberts and James Winters in a recently-published paper, "Future Tense and Economic Decisions: Controlling for Cultural Evolution", PLOS ONE 7/17/2015:

A previous study by Chen demonstrates a correlation between languages that grammatically mark future events and their speakers' propensity to save, even after controlling for numerous economic and demographic factors. The implication is that languages which grammatically distinguish the present and the future may bias their speakers to distinguish them psychologically, leading to less future-oriented decision making. However, Chen's original analysis assumed languages are independent. This neglects the fact that languages are related, causing correlations to appear stronger than is warranted (Galton's problem). In this paper, we test the robustness of Chen's correlations to corrections for the geographic and historical relatedness of languages. While the question seems simple, the answer is complex. In general, the statistical correlation between the two variables is weaker when controlling for relatedness. When applying the strictest tests for relatedness, and when data is not aggregated across individuals, the correlation is not significant. However, the correlation did remain reasonably robust under a number of tests. We argue that any claims of synchronic patterns between cultural variables should be tested for spurious correlations, with the kinds of approaches used in this paper. However, experiments or case-studies would be more fruitful avenues for future research on this specific topic, rather than further large-scale cross-cultural correlational studies.

Sean and James have explained what they did at greater length in a series of blog posts:

"Future Tense and Saving Money: No Correlation When Controlling for Cultural Evolution", Replicated Typo 7/20/2015
"Future Tense and Saving Money: Methods", Replicated Typo 7/25/2015
"Future Tense and Saving Money: Small Number Bias", Replicated Typo 728/2015

When a scientist comes up with an interesting result that gets a lot of press, but also provokes some skeptical responses, the usual behavior is to ignore the negative responses if it's possible to do so, and to mount a hostile defense if it's not.

What Keith has done is in sharp contrast to this norm. He's engaged criticism that he could easily have ignored, and has even joined some critics in a difficult re-evaluation of the work in question.

This may in the end be an effective strategy of self-promotion — but if so, it's one that remarkably few others have discovered. I prefer to think that Keith is someone who would make Diogenes put his lantern down, at least for a while. And I'm certain that society in general, and the activity of rational inquiry in particular, would be better off if more people were to take up Keith's ethos.




  1. Guy said,

    August 14, 2015 @ 1:31 pm

    A great deal of time is spent discussing the statistical methods, but little is spent on the classification of languages into "strong FTR" and "weak FTR" types. It appears to be based on the Östen Dahl paper, but skimming through it, it's not immediately obvious to me where the distinction is made and based on what criteria. This is interesting to me less because it concerns the validity of the paper's conclusion and more because making such a classification seems a daunting and complicated task. Presumably one would have to identify constructions that have a primary temporal meaning, distinguished from modal and aspectual meaning, and then measure how frequently they are used, when they are obligatory and when they are forbidden, and then compress that into a single bit of data. There are also questions about how grammatically central a construction must be before it can be considered the type of "marking" we are concerned with. For example, if a hypothetical language has no commonly used inflectional or auxiliary construction marking future time reference, but future time reference usually requires an explicit temporal adjunct in a wide range of cases – perhaps even wider than where typical languages with an inflectional future tense would typically require it, should we regard that as a "strong FTR" language?

    [(myl) Take a look at Geoff Pullum's 2012 post, and Keith Chen's response.]

  2. Xmun said,

    August 14, 2015 @ 2:54 pm

    Doesn't everybody know that Jews are good at saving money although Hebrew has no future tense? (Allow me to share this thought although I know it's pretty silly.)

    [(myl) Actually, that's the prediction. Lack of obligatory future tense makes the future seem more present and thus more real.]

  3. Guy said,

    August 14, 2015 @ 4:09 pm


    Chen's hypothesis is that the presence of obligatory future time reference marking makes people less likely to plan for the future, not more. So that anecdotal argument would be consistent with, not contrary to, the hypothesis.

    I didn't notice the comment by Östen Dahl in the thread by Chen before, it would seem to confirm my suspicion that Chen engaged in some degree of synthesis in converting the "raw data" into a stron/weak FTR classification. In that post, Chen seems to indicate that his criteria is the obligatory marking (it sounds like only inflectional and grammaticized periphrastic marking is considered to "count") of future time reference under conditions that can't be explained as the need to avoid present time reference. This is a little odd as it seems to me that these are the situations where it is most likely a modal or aspectual meaning is the core reason for the marking. This doesn't necessarily undermine Chen's reasoning in coming up with the hypothesis (the hypothesis could be rephrased as regarding "languages where future time reference is more routinely modalized as being uncertain or mutable") and if a statistically significant correlation is found, that would show a correlation with whatever he is measuring, even if it isn't necessarily future time reference, but I'm not sure it's a satisfactory test of marking of future time reference if that's specifically what we are looking for.

    One interaction with tense and aspect that might be interesting to investigate is that some languages (like English) require explicit aspectual marking of stative verbs to show what I would consider to be "core" present tense meeting. "I'm watching Keeping Up with the Kardashians" typically refers to a situation that is occurring now (though it can also be a specific future occurrence), whereas "I watch Keeping Up with the Kardashians" usually refers to an enduring habitual state that includes the present, but doesn't to me feel any more present than "two plus two equals four", which is usually considered to not have specific time reference.

    This really raises the question of when a marking is considered to be a future time reference marking in the first place. The "futurative progressive" can occur more easily in a wider range of circumstances than the "simple present futurative": compare ?"I heard it rains tomorrow" and "I heard it's raining tomorrow". Presumably we wouldn't classify the progressive as "future time marking", since that isn't its primary function – and even here it is an aspectual difference that explains the difference in acceptability – but since it makes future time reference more available, there is a sense in which it is marking future time reference.

  4. Guy said,

    August 14, 2015 @ 4:15 pm

    That should be "require explicit aspectual marking of non-stative verbs".

  5. Catsidhe said,

    August 14, 2015 @ 4:42 pm

    Xmun, that is (AIUI) precisely the effect Chen predicts.

  6. Daniel de França said,

    August 14, 2015 @ 6:19 pm

    It might be that things are correlated, but the order of causation is inverted. The lack of future mark is due not saving, not the other way around.

  7. TR said,

    August 14, 2015 @ 8:35 pm

    Modern Hebrew has a future tense, though Biblical Hebrew doesn't. So the prediction would be that Jews have gotten better at saving since the days of Eliezer Ben-Yehuda?

  8. Xmun said,

    August 14, 2015 @ 10:55 pm

    Thanks to Catsidhe and Guy for explaining the point. The moral, no doubt, is that one shouldn't comment on topics without first reading the posts. But I doubt if I am the first LL reader to do this.

  9. phspaelti said,

    August 15, 2015 @ 12:46 am

    you are repeating the same mistake that Xmun made. The relationship is the other way around. If the older form of a language did not have obligatory future marking and the modern form does, then presumably the propensity to save should drop. And the reverse should be true as well.
    In fact historical cases of this kind would be far the best evidence that there is anything to this theory.

  10. Iskander said,

    August 15, 2015 @ 12:59 am

    @Xmun: Silly indeed. First, because it treats "the Jews" as a single entity with well-defined behavioural patterns (are Ethiopian Jews known as good savers? Yemeni ones? Malayali or Georgian Jews?). Second, because it assumes that what "what everyone knows" about the Jews is true (any evidence that poor Hassidic Jews in nineteenth-century Ukraine were better at saving than their neighbours?). Third, because it equates "the Jews" with Hebrew speakers which is absurd: by the time the stereotype of the Jews as especially skilled at commerce and finance arose (due to the legal environment in the Western Christian world, not in other areas of Jewish settlement), Hebrew had long been a dead language.

  11. Harold said,

    August 15, 2015 @ 1:20 am

    @Xmun "Jews are good as saving money"

    Jesus saves Moses invests.

  12. Keith said,

    August 15, 2015 @ 3:03 am

    @Xmun and others

    It's my understanding that up until the renaissance of Hebrew through the efforts of Eliezer Ben-Yehuda the Jews would speak the local language of their place of residence.

    So by studying the effects of the future tense systems of various local languages on the economic behaviour of a variety of Jewish communities we could not only indulge in some Whorfianism, but also in a good deal of stereotyping.

  13. anon said,

    August 15, 2015 @ 6:11 am

    "Jesus saves; Moses invests."

    Onan withdraws.

  14. Victor Mair said,

    August 15, 2015 @ 8:01 am

    A couple of years ago, I presented a great deal of empirical evidence against Chen's claims in the following comments and the many articles to which they are linked: here, here, here, and here.

    Considering the economic woes currently besetting China (stock market collapse, mountains of bad debt, rapid devaluation of the yuan — I could provide any number of current articles describing the dire circumstances), the empirical evidence against Chen's thesis about the economic impact of grammatical futurity is even stronger now than it was two years ago. But Chen's thesis isn't just about economics, it is also concerned with gambling, alcohol consumption, unhealthy habits, and so forth. In all of the areas that I examined, his predictions did not comport with reality.

    This incompatibility with empirical data ties in with the confusion among the commenters to this post about what Chen is actually claiming, which also has to do with its hypertheoretical and counterintuitive nature.

  15. Stitch said,

    August 15, 2015 @ 11:03 am

    I'm so old…I can remember when English speakers were considered the rock-ribbed Calvinist work-ethical inventors of capitalism. Now it's apparently necessary to come up with an explanation of why we're so shiftless and feckless compared to everybody else. Has the language (or the behavior) changed a lot without me noticing it?

  16. Harold said,

    August 15, 2015 @ 7:56 pm

    I don't understand how God could have told Abraham: :"I will multiply your seed as the stars of heaven", if there were no future tense, not to mention the fire next time, beating swords into plough shares and countless other prophesies.

    On the other hand since Hebrew was not used by Jews except as a ritual language until quite recent times, whether it had a future tense is not relevant to their behavior during the diaspora or even earlier, as Keith points out above.

  17. Guy said,

    August 16, 2015 @ 1:58 pm


    Many languages don't have future tenses, that doesn't mean that they don't have ways of referring to the future. English is usually considered by most modern linguists not to have a future tense ("will" is a modal auxiliary), and even if we considered "will" to be the English future tense marker there are still many other ways to speak of the future that are uncontroversially present tense. "Your seed is to be spread", "I'm going to find out", "my license expires next year" etc.

RSS feed for comments on this post