Saving for the Future

The Ant and the Grasshopper

Another quick post today relating to a recently reported story about a working paper by M. Keith Chen, an economics prof at Yale, which claims a correlation between tense marking and behaviour.1 I was originally going to hold off for a longer post on the topic of linguistic relativity, but instead I decided to post right away as this topic has been making waves on the web due to Chen's paper.2 The Ant and the Grasshopper

Essentially, Chen points out a statistical correlation between languages that do not strongly mark a distinction between present and future time, and people who provision for the future, for instance by saving money or engaging in behaviour that is beneficial to long term health like diet and exercise. The crux of it seems to be that if your language makes a strong distinction between present and future time, you don't take note of future consequences as much as someone who speaks a language which lumps together present and future --- when the future is the present, the future is really a present concern, but if you linguistically isolate the future you can put it out of mind more easily, so the theory goes. Thus on average, speakers of languages without a distinct future tense save more money and live healthier lives than those who speak languages with separate future tenses. The abstract from Chen's paper summarises his findings and reasoning:

Languages differ dramatically in how much they require their speakers to mark the timing of events when speaking. In this paper I test the hypothesis that being required to speak differently about future events (what linguists call strongly grammaticalized future-time reference) leads speakers to treat the future as more distant, and to take fewer future-oriented actions. Consistent with this hypothesis I find that in every major region of the world, speakers of strong-FTR languages save less per year, hold less retirement wealth, smoke more, are more likely to be obese, and suffer from worse long-run health. This holds true even after extensive controls that compare only demographically similar individuals born and living in the same country. While not dispositive, the evidence does not seem to support the most obvious forms of common causation. Implications of these findings for theories of intertemporal choice are discussed.

I first heard about this research a while ago, and it's being widely reported on the web in the past few days (see here and here) as the discussion paper (not yet published in a peer-reviewed journal) has recently been made available online. For a rather more negative appraisal of this research see this post, and for a critique of the statistical analysis see this post,5 both at Language Log.

What lies behind this sort of argument is the concept of linguistic relativity, which you can read about on Wikipedia or even better in this easily approachable and very well explained entry by Lera Boroditsky in the Encyclopedia of Cognitive Science, available along with much other research on the topic on her excellent website.3 Indeed, Boroditsky is doing some of the most compelling and convincing work on the issue of linguistic relativity. Essentially the claim is that language can affect the way you think in very fundamental ways; language differences can lead to differences in cognition. The idea originally came out of the research of Edward Sapir and Benjamin Lee Whorf and is thus also known as the Sapir-Whorf hypothesis.4 The original strong version of this theory, linguistic determinism, claimed that the language you spoke determined the way you were able to think about the world. The more recent versions of this idea claim that language can have an effect on the way you think but is not an absolute determiner of it.6

I first encountered the concept of linguistic relativity when working on my dissertation, which examined the development of future tense constructions, such as “I shall” or “you will”, in Old English texts. English did not originally have a future tense in the same way that it had a present and a past tense (“I walk” and “I walked”). You could (and still can in some circumstances) use the present tense with an adverb (“I walk to the doctor’s office tomorrow”) to indicate future time, or you can use some sort of auxiliary with the verb (“I will walk” or “I am going to walk”). Latin, on the other hand, does have a series of simple ending you can tack onto the end of verbs to indicate future time, just like English adds “-ed” to verbs to show past time. I was curious how Old English translators of Latin texts handled these Latin future tense forms when English at that time did not have a regularised system for marking future time. It was a straightforward philological topic. But from there I branched out into all kinds of conceptual areas like anthropology and philosophy, and literary areas, such as the construction of narrative. For instance, particularly relevant is the effect Christianization had on the Anglo-Saxons, Christianity having a well developed notion of the future and Latin, the language that Christianity was transmitted in, having a future tense. It was in the conceptual area that I first encountered cognitive science, how we conceptualise future time. I briefly discussed these issues and the issue of linguistic relativity, and then filed them away for a later date as an interesting topic that might be worth another look, and am now coming back to the issue of linguistic relativity in a conference paper I'm preparing for the International Congress on Medieval Studies in Kalamazoo in May. Chen's reseach has obvious parallels to all of this. I'll post a fuller account of these implications later as I continue to work on my conference paper.

Full disclosure: my position on all of this is that I support the notion of linguistic relativity, at least in its weaker formulation. I am convinced by the evidence of Boroditsky and others that language does have a profound and deep influence on the way you think. It is not, however, the only factor, and the more complex the human behaviour you're looking at, the harder it is to ascribe it to one simple linguistic feature. Other linguistic and non-linguistic factors influencing cognition come into play. That being said, Chen's methodology would, at first glance, seem to deal with these limitations, as he controls for these variables, and given that he is showing large-scale statistical influences, individual exceptions aren't a big problem for his conclusions. Future time marking may well be a powerful factor influencing your behaviour, though not the only factor. Other personal and large-scale cultural factors all have a role to play in this complex behaviour, but in the aggregate, your language may well have a profound effect on your long-tem planning. I'll write more about this research when I've had a chance to go over it in more detail, as well as about linguistic relativity, which underlies much of the thinking behind the ideas I'm writing about on this blog.

If you do not think about the future, you cannot have one. —John Galsworthy, Swan Song


1 Well, it was going to be a brief post, but it ended up as a rather long one. The topic of linguistic relativity is very much on my mind at the moment...[back] 2 I will soon get back to the post I was writing to explain cognitive linguistics and philology, I promise.[back] 3 For a good introduction to her research have a look at the popular press essays, such as the Scientific American article, and the Economist debate with Language Logger Mark Liberman on the topic of linguistic relativity. Also a good introduction is this public lecture given by Boroditsky, which is also available through iTunes.[back] 4 See for instance E. Sapir, Language (1921), Time Perspective in Aboriginal American Culture, a Study in Method (1926), Culture, Language and Personality (1957); Benjamin Lee Whorf, Language, Thought & Reality (1956).[back] 5 I'm a little confused by Mark Liberman's complaint here. From his comments, he seems to be implying that due to cultural diffusion, any number of factors will be exchanged by geographically adjacent cultures, and so having a linguistic feature and a cultural feature in common doesn't imply a connection between the two. Fair enough, but isn't that the point? Germanic and Romance languages are geographically very close together, and yet linguistically different, and strong-FTR languages around the world may happen to have that one linguistic trait in common, in spite of the languages being completely unrelated, and yet Chen is reporting a statistical correlation between this linguistic trait and the economic one. But perhaps I'm missing something here. The Language Loggers frequently complain about the lack of basic statistical knowledge in the general public, and they are no doubt correct; however, given the complexity of the statistical analysis under discussion, one would wish that he could explain it more clearly for someone who doesn't have that level of statistical knowledge.[back] 6 For counter arguments to the Sapir-Whorf hypothesis, see for instance the very approachable and well written book The Language Instinct (1994) by Steven Pinker, and various Language Log posts on the topic such as this one.[back]