Linguistic relativity: fact or wishful thinking?

One of the most fascinating questions in linguistics is whether and how much language influences thought. Do speakers of different languages think differently? There are some 6,000 languages in the world and they are all different from each other, sometimes in very striking ways. The question is whether these differences in language lead to any differences in thinking.

The idea of linguistic relativity is associated mainly with the writings of Benjamin Lee Whorf and Edward Sapir, two American linguists who postulated the hypothesis in the early 20th century. The “Sapir-Whorf Hypothesis” was motivated by the world’s vast linguistic diversity. Whorf himself investigated several exotic languages (exotic from a westerner's point of view, that is) and came to the conclusion that the make-up of a language strongly influences the way its speakers think about the world. Whorf’s most quoted example is Hopi, a native North-American language, which according to Whorf has “no words, grammatical forms, construction or expressions that refer directly to what we call time”. And sure enough, the notion of time plays no role in Hopi society, people live in the present and do not think about what happened in the past or will happen in the future.

When presented like this, the hypothesis seems like common sense. Surely, people who don’t think about time don’t talk about time either, they have no need for time-related expressions and therefore their language has no such expressions. This is uncontroversial. But the Sapir-Whorf Hypothesis claims that it works in the opposite direction as well: not only does thought determine language but also does language determine thought. So, the fact that their language has no time expressions effectively makes it impossible (or at least very difficult) for Hopi speakers to develop a notion of time in their minds. When a child is brought up as a monolingual Hopi speaker, she is pretty much predetermined to think about the world differently from the way she would if she were brought up with some other language.

It has to be said that Whorf’s observations about Hopi have later come to be questioned. It seems that Hopi does, after all, have expressions for talking about time and that poor Benjamin Whorf was so driven by a desire to prove his theory that he didn’t interpret the evidence correctly. There is a lesson there: one must always be cautious when a field linguist “discovers” bizarre features in some exotic language somewhere in the middle of a jungle which only he and the natives can speak – such as when Daniel Everett reported some time ago that the South-American language Pirahã doesn’t have recursion, thus contradicting a fundamental aspect of our understanding of what is and isn’t a language. The jury’s still out on whether that’s actually true.

But back to linguistic relativity. Probably nobody seriously believes any more that language influences thought in such breathtaking ways as Whorf believed. But once an idea like that appears on stage, it never goes away completely. Is it possible that language influences thought at least a little? And if so, can any evidence of that be found in more “mainstream” languages, including what Whorf himself called “Standard Average European” ones (a bunch of languages dismissed by Whorf as being all essentially variations on the same theme and possessing no interesting internal diversity)?

One of the most prominent researchers attacking this question today is Lera Boroditsky at Stanford University. Anybody with an interest in linguistic relativity will do well to read her accessibly written papers where she recounts evidence from languages such as English, Dutch, German, Spanish, French, Mandarin and Korean that suggest (if not prove) that linguistic relativity is real. Let’s take a look at some of those.

Languages differ in gender. As I’m sure you know, many languages divide nouns into feminine and masculine (and sometimes neuter) groups and the division is completely arbitrary: the same noun can be masculine in one language but feminine in another. The question now is, does the fact that a word is, say, feminine, make people think of it as being somehow more “female” than if it were masculine, and vice versa? There have been experiments that show that indeed it does. For example, native speakers of German and Spanish were asked to list adjectives that in their opinion describe concepts such as bridge and key which have opposite genders in the two languages. Consistently, speakers of German, where bridge is feminine, described bridges in terms that suggested femininity (slender, elegant, fragile) while speakers of Spanish, where bridge is masculine, used adjectives suggesting masculinity (big, strong, towering). And there have been numerous other experiments that show that people consistently attribute female-like and male-like qualities to concepts depending on whether those concepts are lexicalized as feminine or masculine in their own language. It is surprising that an arbitrary aspect of grammar such as gender assignment should have a discernible influence on thinking but apparently, it does.

Languages also differ in the expressions they use to describe spatial positions of objects relative to each other. In European languages such as English, we make a distinction between in and on: the apple is in the bowl, the magnet is on the fridge. It’s a distinction between being in a container and being on a surface. Korean, on the other hand, makes a distinction between tight and loose fit: putting a letter in an envelope and putting a magnet on a fridge are both expressed with the same preposition because they both involve a tight fit, while an apple in a bowl needs a different preposition because it’s a loose fit. In other words, the two languages cut the loaf in different places: English at the container/surface boundary, Korean at the tight fit/loose fit boundary. The question now is, is this difference just a difference of language, or do people actually think differently about the positions of objects? There have been experiments that suggest that the latter is the case. When asked to compare various spatial configurations of objects, Korean speakers spontaneously made distinctions along the tight fit/loose fit boundary while English speakers tended to distinguish along the container/surface boundary. In other words, which language a person speaks predicts what kinds of spatial relations the person considers important.

There have been experiments that look at other areas, such as time expressions, expressions that have to do with the shape and material of objects, and so on. And they all reveal what many will find shocking: which language you speak determines how you understand and interpret the world. The correspondence isn’t absolute, of course: people are able to liberate themselves from the prejudices imposed by their language if they try consciously or if they learn another language. Some domains are even completely immune to the effects of language, for example people’s ability to distinguish a colour seems unaffected by whether their language does or doesn’t have a word for the colour. But many other domains are in fact in alignment with the language you happen to be a speaker of.

To my mind, all these experiments give strong support to the linguistic relativity hypothesis, but we must be careful in how much credit we give them. They show that there is a correlation between linguistic structures and mental structures, but they don’t say which causes which. A correlation is not a causation, as you probably know if you’ve ever taken one of those universally hated college modules in research methods and statistics.

On the other hand, language is such an important and omnipresent part of our human existence that it would be strange if it didn’t have some influence on us. We humans are using language pretty much constantly, we are always talking to other people, listening to people, reading, writing. I recommend you to stand on a busy street corner for a while to observe people and pay attention to what they do. Chances are most of them are talking. A huge amount of our waking time is spent using language in some form. Realistically, something as pervasive as that is bound to leave at least some trace on the way we think.

I suspect that most linguists secretly wish the Sapir-Whorf Hypothesis to be true. I admit I certainly do. You see, we are people who love languages, we revel in the world’s linguistic diversity and we don’t want anybody to tell us that languages are unimportant because they’re nothing more than different sets of labels for the same underlying thoughts. We suspect that languages reach deeper in the human mind than that. Anybody who speaks more than one language habitually probably feels that switching from one to another is not just switching codes, it’s switching whole frames of mind. And those of us who believe that linguistic diversity is important and that minority languages in particular need to be saved, we want to be able to show to the world that linguistic diversity is really a diversity in ways of thinking about, and therefore ways of acting upon, the world. If this is true, then linguistic diversity has more than just sentimental value, it is an insurance against the extinction of human civilization, just like biological diversity is an insurance against the extinction of life.

Speaking of acting upon the world, some people don’t seem to need proof that linguistic relativity exists at all, they’re already using it themselves. Every time you hear a corrupt politician say “it is regrettable that mistakes happened” when he really should be saying “I’m sorry for the mistakes I’ve made”, you are witnessing a case of minds being manipulated by language – namely, in this case, your mind is being manipulated into not paying attention to the fact that it was “I” who made those mistakes. Why do corrupt politicians talk like that? Because they believe that linguistic relativity is real, that language does influence thought. And then there are people who want to change the way we think not by changing the way they talk to us but by making us change the way we talk ourselves – such as those who promote political correctness, who encourage us to use gender-neutral language (“he or she”, “chairperson”). They do this because they believe we will change the way we think if we change the way we talk. So it turns out that a lot of people already believe in linguistic relativity after all, regardless of how conclusive or inconclusive the scientific experiments are. And that, I guess, is all the proof I need.

Michal Měchura, 2009-08-16, CC BY-SA