Sunday, August 29, 2010

Thoughtful and lucid on language and thought: Guy Deutscher ‘Through the Language Glass’

“Ever heard of es-tchay-den-frood?” asked the husband. “No, what is it?” un-answered the wife. “‘Relishing someone else’s misfortune’,” quoted back the husband. “It’s in this article that says how the language you speak affects how you think.” “Yeah, I’d have thought that,” thoughtfully responded the wife. I almost choked on my breakfast.

I doubt this blog has much of a readership, and I hardly expect to be influencing the august editors of International Herald Tribune (yet), but the timing was pretty vexing nonetheless: first Behavioural and Brain Sciences (article, response), then the Wall Street Journal (article, response), now this. The myth-peddling assault on sensible research into language rolls on unabated. So, I quickly left the breakfast room, retrieved our complimentary paper, and checked the article out. Sure enough, the title promised a parade of the usual errors: “Same planet, different worlds: Languages shaping reality”. And on the front page. The heart sinks, the brain revs.

The article (link), by Guy Deutscher (webpage), turns out to be a pleasant surprise, however (the title, perhaps, an editor’s addition): a deft and elegant style, facts simply not simplistically communicated, and, most importantly, avoidance of the array of errors that usual besmirch discussion of the link between language and thought. Deutscher identifies straightaway that the connection is one of tracks, not traps. That is, speaking Hopi doesn’t leave you trapped in a mindset that forever prevents you from learning to use the English word ‘time’. Rather, as speakers of English, or other languages, we get into the habit of thinking along certain tracks: we pay attention to some aspects of situations and fail to attend to others, because competent use of our day-to-day language has set us up with certain habits of expression and, hence, foci of attention, and, hence, patterns of thought. For the rest of the article, Deutscher illustrates this with two examples:
  • Speakers of, say, Spanish and German are inclined to describe bridges (amongst other things) in different ways, e.g., ‘strong’ versus ‘slender’. The choices they make may, plausibly, reflect differences in grammatical gender: for Spaniards, bridges are masculine, for Germans, feminine, and strength is a masculine property, slenderness a female one.
  • Speakers of Guugu Yimithirr—a language of northern Queensland, that eschews terms like ‘left’, ‘right’, ‘in front’, ‘behind’ for ‘to the north’, ‘to the south’, etc.—use and can recall details of cardinal directions in a way that the average city-bound English speaker can’t.
In both cases, it isn’t that Spanish speakers have thoughts about bridges unthinkable to Germans, or Guugu Yimithirr speakers have thoughts involving locations unthinkable to English speakers. Rather, it’s that there are certain thoughts that are more likely to occur to Spaniards than Germans, or Guugu Yimithirr than English speakers, in virtue of their respective languages.

My liking of the article notwithstanding, I’m unsure of what Deutscher thinks that these two examples are meant to show us. To me, they seem to show that society affects how people act, and, as action includes using language in an socially appropriate fashion, it affects how they speak and how they get ready to speak, that is, how they think. So, I don’t think either example shows us much about how language, alone, influences thought. Rather, they show how society influences thought—a conclusion that is surely unsurprising.

To make this clearer, consider first Spanish versus German. I question whether any association that speakers have between bridges and manliness/womanliness arises spontaneously, simply from using their native tongue, rather than from being put through an education system that tells them to think of nouns in terms of masculinity/femininity. The dichotomy between the genders is a socially central and salient one. Plausibly, it works its way, via grammatical terminology inculcated in the educational system, through to speakers’ conceptions of these objects. To be sure that it is language (a linguistic fact), rather than how society teaches people to think about their language (a social fact), that is at work in the ‘bridge effect’, we would want to look at Spanish and German speakers who are so far removed from formal education that they do not know anything about traditional grammatical gender terminology (such speakers might be found in rural parts of Latin America and, just possibly, amongst the Pennsylvania Dutch, or like communities). If they describe bridges in sex-biased terms, then we might have something to explain. If not, then we should conclude that how society tells people to think about their languages affects how people think about their languages down even to the level of the common noun—an interesting conclusion, but hardly compelling evidence of language influencing thought independent of ambient social facts.

And an incident from my time doing fieldwork makes me question Deutscher’s second example. When I’ve asked for directions from one place to the next, those directions have most often been in the form of cardinal directions: “Head south for two miles, then turn west, then ...”. I found this somewhat irritating at first (“Why can’t they just left-and-right me?”, I used to wonder). However, once the directions cease to involve roads but cross fields and follow rivers, it becomes pretty obvious. If you have to “turn left at the next road”, then it doesn’t matter whether than road is at 30 degrees to your current road, or at 60, or 113: following the road takes care of that. Try telling someone without a compass to turn left by 113 degrees and you soon appreciate that “head south for two miles, then turn west, then ...” is the superior system, once you’ve learned to use it. The crucial fact, however, is that I wasn’t off in the extremities of Australia, but smack in the middle of rural Oklahoma, and the people giving me directions were, more often than not, monolingual English speakers, sometimes just as white as me.

Obviously, Guugu Yimithirr culture and language have evolved in a world of without a grid of tarmacked roads leading you at just the right angle to get you where you want to go to. And, as a child, you don’t just learn the grammar of a language, or its lexicon, you learn what to talk about in day-to-day life and how to talk about it too. For objective reasons, cardinal directions play a prominent role in Guugu Yimithirr culture and, as a cultural fact, when you speak Guugu Yimithirr, you adorn your description of an event, it seems, with facts about these directions. In a similar vein, I remember the first time I read Warlpiri story, from a different part of Australia. It basically said that there was a group of people who headed in this direction, then that direction, then off in another, and ended, I think, with them being back where they started: no comment on why they left, what they found, what they did, or what it was like to get back again. (“Yep, that sounds about right,” said Ken Hale and Norvin Richards, who have, respectively, superabundant and abundant experience documenting Australian languages.) If Guugu Yimithirr culture is anything like the Warlpiris’ in this respect, then directions matter to the extent that recounting them is enough to create a story of interest.

As with Spanish and German, it seems to me that what makes Guugu Yimithirr different from English is not a linguistic fact (a case of language guiding thought), but a set of social facts that appear linguistic because they are bound up with language use:
  • It is a social fact that the world you live in and the life you lead there are better served by cardinal directions than by lefts and rights.
  • Whether your language has ‘left’ and ‘right’ is a linguistic fact, but how much your community expects you to mention directions when retelling events is a social fact. (Guugu Yimithirr could, one imagines, adopt ‘left’ and ‘right’ from English but use these only for humans, maintain a social preference for locating objects and events in terms of norths and souths.)
  • And whether sequences of directions alone suffice to make a story worth telling is certainly a social fact.
Deutscher wonders, with admirable tentativeness, “how geographic languages [like Guugu Yimithirr] affect areas of experience other than spatial orientation—whether they influence the speaker’s sense of identity, for instance, or bring about a less-egocentric outlook on life”. However, these questions strike me as premature: we still don’t have good reason for thinking that it is language, rather than social facts about language use, that are at play here.

It is an open, and probably prickly, question as to how one should tease apart purely linguistic influence on thought from social influence, mediated by patterns of language use, on thought. My suspicion, in part confirmed by Deutscher’s article, is that what is advertised as the former is in reality more often the latter. What the cases Deutscher considers probably show therefore is that society and socialisation influence how and what people think about (no surprise there) and that this influence infiltrates down to the level of quite subtle facts about language. In principle, the latter finding is not so surprising, but the precise details doubtless will be intriguing. I suspect that we are still far from knowing whether language itself, abstracted away from the society that speaks it, can influence thought and I doubt that the experiments devised to test the hypothesis are in principle capable of speaking to the issue. However, given how well Deutscher’s article frames the question and presents key research, I think his forthcoming book (link) will be an excellent way to confirm or disconfirm such suspicions.


  1. You could probably quite easily do the experiment to tease apart the actual effect of grammatical gender on common noun conceptualization, and the effect of using 'masculine' and 'feminine' to describe those genders (and thereby inheriting all the socio-cultural associations those terms have) with any of the millions of bilingual Spanish/English speakers in the US who speak Spanish as a first and dominant language, but have never had formal instruction in a language with a gender system. It could still be the case that its enough that 'la' is used with 'mujer' and 'madre' for 'la puente' to be associated with feminine characteristics, but at least you'd know it wasn't because speakers are taught that there is a 'feminine' noun class.

  2. I've been thinking about that since posting. The discourse literature on subjecthood and topichood often emphasises the dominant role that animates play in sentences: more likely to be grammatical subjects, to be ongoing topics, to have their point of view expressed, ... This is normally explained in functional terms, concerning what things humans are interested in and how we see other humans as affecting us and the world. If so, then the masculine/feminine divide between animates may be central enough to what and how we think to influence how we characterise the noun classes that contain other nouns that, e.g., take the article "la" versus "el", govern adjectival agreement in "a" versus "o", and so on. If so, then this explanation would predict (a) a bias towards conceiving of grammatically feminine (or masculine) nouns in more culturally female (or male) terms; and (b) independence of this effect from any formal education in use of grammatical terminology (though such education might strengthen the effect).

    The key driving force to any differential associations between inanimates and (fe)maleness would be the centrality of animates as subject-/topic-worthy entities in discourse and cognition—an extralinguistic fact. That this is embedded in a grammatical system that places males (plus some inanimates) and females (plus other inanimates) into different noun classes would then have knock on effects. In a weak sense, this could be taken as an effect of language on thought, but it would still have to be recognised that driving force is, as said, something extralinguistic. Still, it makes the gender example the better of the two that Deutscher presents.

    Either way, I think it would be interesting to look into the kind of experiment you outlined. How to check whether the speakers don't know about what "masculino" and "feminino" mean as grammatical terms...?

    (PS: It's "el puente". Were you having a Canadian moment, thinking of "la pont"? Like "le lait" ~ "la leche")

  3. People Francis has declared in a sermon that Christmas this year will be a charade because the whole world is at war. Speaking during mass at the casa Santa Maria, he said we are close to Christmas essay writer service.Any ways it is an interesting article is about christens and there culture.