Futured Money: A Review of Chen's Tense and Savings Theory

By Patrick Welsh, February 2014

Several months ago, an economist and professor at UCLA, M. Keith Chen, published a paper in the American Economic Review, which argued that countries with languages that are future tense dependent (where they necessarily must speak about the future using tense) save less money. Or in simpler words, if you speak a language that talks about the future and the present in the same tense, you're more likely to save your money. English is a language that tends to invoke a future tense. You can say I will go to the bank later but not I go to the bank later.

If you want to learn more about the theory to prepare yourself for my review, here is an article in The Atlantic about this; here is Chen's article (pdf warning); and here is a TED talk by Chen himself.

Chen's study involved 39 languages, only nine of which were categorized as non-future tense dependent. Many of the nine have interconnected economies. More frustratingly, Chen categorizes languages based on popular consensus, not linguistic consensus. Some have Danish and Swedish are closely inter-related in an East Scandinavian language clade. Basque was categorized as one language when there are definitely two language divisions between east and west, and in reality probably many more Basque languages. Why is "East Scandinavian" divided but Basque not? The idea that the Basque language consists of simply different dialects is not a reflection of the current state of Basque today; two speakers of the most divergent Basque dialects (read: languages) will have a much harder time communicating than a Danish and Swedish speaker.

But back to the problem of interconnected economies. Danish, Swedish, and Finnish utilize the Scandinavian economic and fiscal models while Danish and Swedish are highly interconnected languages. This artificially inflates the N-sample and makes the report look statistically more significant than it should be.

Chen's characterization of language was awfully simplistic. English can say I am going to the store (later) in which case the sentence is free from a present-future distinction. Chen's paper, however, acknowledges these nuances in language but makes no attempt to distinguish them. Such a feat would be Herculean.

Some have pointed to Chen's paper and said, "Well this was true even within countries! A French speaker in Switzerland saves less than a Swiss German." While Chen did find that was true, his findings were not statistically significant, so the point is moot.

Critics of Chen's articles have not been entirely on point. I was fairly disappointed with Jason Merchant's comment on Language Log. Merchant says,

Because Chen did not control for cultural factors though, it remains at best a supposition that language, and not the cultures of the people using them, are responsible for the savings and other behavioral differences found.

Did he read Chen's article? Control variables included legal inheritance and Family Values survey findings. You may say that the control variables were poorly chosen (as in 'what is the legal origin going to do with the savings rate?') but you can't say that he wasn't controlling for cultural factors. Part of the reason stems from the dearth of reliable cultural control variables. Cultural controls are notoriously tricky to quantify, and useful data are difficult to find and utilize. 

It's not surprising that Chen's argument has created such a gut-level rejection among linguists. The linguistics community has been plagued for years from the outside with Sapir-Worf advocates. The Sapir-Worf theory says that the way we see the world is shaped by the language we speak, and that people who speak different languages will see the world differently.

There is some evidence this is the case, such as the famous 1969 color test of Berlin & Kay which showed that people who speak languages with limited color vocabulary seem to distinguish fewer colors as a whole. Their findings have been the subject of much debate over the years, but the gist is that there is some evidence that Sapir-Worf, in its weakest form, holds true. The problem is when we extend these small evidences of Sapir-Worf to entire modes of thinking with little justification. Do people without a conditional mood view the world in black and white? Sapir-Worf. Are African political quagmires and violence the inevitable result of Niger-Congo philosophies imbued into their languages? Sapir-Worf. Do people whose language merges the future and present tenses save less/more? Sapir-Worf. For years, linguists have had to deal with non-linguists constantly using Sapir-Worf for the most grandiose of conclusions, and it's gotten old. Real old.

So you can see why the linguistic community may be hesitant to get behind Chen's thesis. They're sick and tired of the same old conclusions warranted by the same old theories. You can also see why the non-linguistic community is much more eager to get behind him. They haven't had to endure the years of defending scholarly consensus against wishful thinking that wears an academic out.

Chen's argument is mildly interesting and mildly persuasive. It will take more studies over time to convince the older linguists behind their tenured pulpits, but the groundwork has been lain. Chen brings new evidence that supports a conclusion, an albeit unpopular one. I look forward to what else Chen has to say in the future and the present.