Boroditsky on Whorfian navigation and blame

« previous post | next post »

Several readers have sent me links to Lera Boroditsky's recent article in the Wall Street Journal, "Lost in Translation" (7/24/2010).  We've mentioned Prof. Boroditsky's work on LL several times, starting back in 2003, and so long-time readers won't be surprised to learn that I think this is an interesting popularization of solid work.  However, most LL readers will also know that there is probably no single linguistic idea that is more prone to exaggeration and mis-application than the "Sapir-Whorf hypothesis" about the relations between language and thought. And the WSJ editors' subhed for Boroditsky's article gives their readers a push down that road:

New cognitive research suggests that language profoundly influences the way people see the world; a different sense of blame in Japanese and Spanish.

That's all I have time for this morning — more later on the WSJ article, the research behind it, and the popular reaction to it. Meanwhile, the comments are open.

As usual, I strongly suggest that you read (at least some of) the research reports before sounding off. Boroditsky's preprints list is here, and much of the work discussed was done by Caitlin Fausey, whose publications would be a good place to start.



28 Comments

  1. Jason Merchant said,

    July 26, 2010 @ 12:34 pm

    Still, it's disappointing to see the caricature of the idea of Universal Grammar trotted out by someone who should know better (that UG means that languages don't differ from one another: from the article: "Dr. Chomsky proposed that there is a universal grammar for all human languages—essentially, that languages don't really differ from one another in significant ways. And because languages didn't differ from one another, …"). UG doesn't entail that the words of all languages should mean the same things or that grammatical formatives should have the same distribution across langauges (lexical differences being where most–all that I know of, in fact–of this research is concentrated, obviously).

  2. Geoffrey K. Pullum said,

    July 26, 2010 @ 12:45 pm

    Absolutely right, Jason. You could believe in an insanely strong Chomskyan universal grammar that controlled just about every aspect of the syntax of every language and held all but the tiniest details in a vice-like universal grip, and yet still believe that the individual words your language happened to have gave you your only access to concepts, so that if you weren't lucky enough to have a word for green and a word for blue you'd be green/blue colorblind. It is sad to see a goof as deep as this being published by a professional psycholinguist in a major newspaper feature. Sigh.

  3. bianca steele said,

    July 26, 2010 @ 12:52 pm

    I suspect a lot of the confusion over what Universal Grammar means comes from computer science types who misinterpret the "deep structure" as some kind of subjective mental representation in the brain of the individual.

  4. Jerry Friedman said,

    July 26, 2010 @ 1:03 pm

    Despite the problem with "Universal Grammar" (which I'm taking the word of you experts on), I thought the WSJ article and the two Fausey articles I read (this one and the Timberlake one) were fascinating. Unless you experts cast doubt on the results, this is just the sort of research that should inform the many discussions here of what Boroditsky and Fausey call "agentive" and "non-agentive" constructions. So thanks for the post, Mark.

    And should we have a pool about how soon someone commenting on Boroditsky's article changes nonagentive to passive.

  5. D said,

    July 26, 2010 @ 1:05 pm

    To be fair, the most surprising thing to me in that article was the claim that Justin Timberlake was the first to use the term "wardrobe malfunction", which indeed is "a wonderful nonagentive coinage". Maybe history should place him in the same category as Shakespeare rather than as R Kelly.

  6. Mr Fnortner said,

    July 26, 2010 @ 1:35 pm

    The American Dialect Society acknowledges the phrase as a 2004 phenomenon, following Timberlake's utterance. Whether he coined the expression or repeated an unpublished industry term is uncertain. Perhaps someone here can cull a reference in print before 2004. Here is a strong indication that it may not be found: http://www.phrases.org.uk/bulletin_board/28/messages/52.html

  7. John Lawler said,

    July 26, 2010 @ 3:06 pm

    To widen the discussion a bit, let me point to a lengthy discussion
    here last year on Anna Wierzbicka's work about meaning, language, and culture. This is linguistics, not cognitive psychology, but it's serious scholarship and seems quite consistent with this discussion.

  8. michael ramscar said,

    July 26, 2010 @ 3:23 pm

    The confusion with "Universal Grammar" stems from its being ridiculously underspecified: the claim that people are born with a "language acquisition device," without more detail (and beyond the rhetoric, there is no more detail), is tautologous or trivially true, depending on the kind of morning you are having. What is clear is that it isn't falsifiable. (So I guess this means that I take issue with Mark's point about no single linguistic idea being more prone to exaggeration and mis-application than "Sapir-Whorf hypothesis…)

    The same point about falsifiability holds for Chomsky's rampant platonism ("However surprising the conclusion may be that nature has provided us with an innate stock of concepts, and that the child’s task is to discover their labels, the empirical facts appear to leave open few other possibilities." Chomsky, 2000, pp. 65–66) – Whorfian findings may increase the giggle factor, but they aren't going to falsify this kind of nonsense. (Note Chomsky's quite brilliant use of the word "facts", by the way — his misuse of "fact" and "facts" has no rival, save perhaps Erich Honecker's playful treatment of "democracy".)

    Which is the point, really, isn't it? Boroditsky's piece is theoretically vague, but so is the work it seeks to debate. Absent a clear picture of what "language," "thought," "think'" "syntax" (isn't everyone a lexicalist now anyway?) etc. are supposed to be, it is hard to know whether all this means that if you don't have a word for green and a word for blue you'd be green/blue colorblind, or whether it means something else entirely.

    You might see this, for better or worse, as a deft appropriation of the methods of the master (or write the whole thing off as two bald men fighting over a comb), but I'm not sure its entirely fair to criticize Boroditsky for caricaturing a caricature.

  9. ShadowFox said,

    July 26, 2010 @ 5:53 pm

    I am more puzzled by the claim of recency of the evidence than anything else. For example, Boroditsky's 2007 paper on Russian color discrimination is predated by 35 years by

    Heider, E. Rosch. (1972). Universals in color naming and memory. Journal of Experimental Psychology, 93, 10-20.

    Rosch, of course, went on to write a lot about "natural categories" and interplay between language and categorization over the next decade.

    One problem with Sapir/Whorf is the direction of influence. I suppose, the Russian light/dark-blue categorization is not exactly cultural–one could, I suppose, expect it to have originated as a language-based distinction. In Korean, for example, the direction is opposite–cultural formalisms lead to avoidance of 1st prs. sing. But, in general, the direction is not so clear-cut. And much of the philosophical argument (whether it spilled into linguistics is another matter) has been about the direction of influence, especially in categorization.

    Searle follows Merleau-Ponty and others in claiming both linguistic and natural categorization to arise from experience–in particular, from cultural experience. I am not sure that explains the Russian blue and similar cases. Whatever the case may be, pretending that the debate goes on solely or primarily in linguistics or cognitive neuropsychology or cultural psychology would be misguided. Therefore, the claim of "recency" is just wrong. Surely, some evidence is recent, but it's more of an accident than an artifact of the field. What is at issue has always been interpretation of evidence, not "evidence" per se.

  10. Nanani said,

    July 27, 2010 @ 12:33 am

    It seems like what's happening just that the need to mark tense or distinguish synonyms in one language means that a certain feature will be more salient to speakers of that language, while not as salient in languages that don't call for it.

    For example, take the sidebar finding: "Russian speakers, who have more words for light and dark blues, are better able to visually discriminate shades of blue."

    Having separate words for blue-A and blue-B means that the distinction between them is salient to a Russian speaker in a way that it isn't to an English speaker. This can be said to mean that "language shapes thought" but it's a far cry from what I have always understood the Sapir-Whorf hypothesis to be saying.

    As a trilingual, I find that I can often remember the information content of a convesation or document without remembering which language the information was originally in. I'll just recall it in the language I'm currently working in, instead.
    Sometimes, unless there is more than one possible language (like a conversation with another polyglot), I can figure it out by working backwards from the content to the likeliest language it was in.
    Seems that this shouldn't be possible if "language shapes thought".

  11. J. Goard said,

    July 27, 2010 @ 1:59 am

    @Jason & GKP:

    Well, if that's a "goof" by "someone who should know better" (fair enough), then what to make of the many popular introductory linguistics texts which describe Chomsky's 1959 review in Language as a thorough evisceration of behaviorist approaches to language acquisition, rather than an attack on a ridiculous straw man by someone who couldn't manage to summarize the book's major points with a reasonable degree of accuracy? When it comes to writing off broad theoretical approaches via caricature, nobody has done it better than early generativists. Live by the sword…

    [This isn't a very good analogy. Quite independently of whether one wants to take Chomsky's side in any particular dispute, there is an issue of whether the existence of an innate universal framework of limitations on syntax would entail that there couldn't be effects of lexical semantics on perception or cognition. As far as I can see, it can't possibly entail that, by any stretch of the imagination. That's just a point of logic. What you have raised is a very different matter indeed: whether a certain very lengthy book review by Chomsky correctly represents the content of the book he was reviewing. Readers who want a fascinating reading assignment in the psychology of language can judge for themselves by going to a university library and looking at (i) B. F. Skinner's book Verbal Behavior (1957), (ii) Chomsky's detailed critical review of it (Language 35.1 [1959], 26-58), and (iii) Kenneth MacCorquodale's retrospective response in "On Chomsky's review of Skinner's Verbal Behavior" (Journal of the Experimental Analysis of Behavior 13.1 [1970], 83–99). The judgment would have to be, to say the least, a matter of interpretation, and in no way a simple matter of a false entailment. Nobody says Chomsky was sympathetic and generous in his interpretation of what Skinner wrote; but nobody denies that he made some good points about why Skinnerian behavior analysis couldn't possibly be the basis for a full scientific understanding of language. In any case, you can hardly mean what your "live by the sword" remark seems to imply. Even if Chomsky wildly misrepresented Skinner, or if textbooks often misrepresent him as having definitively refuted behaviorism, or if scads of linguists in the late 1950s and the 1960s were misrepresenting their opponents left right and center, that doesn't mean we must assent to arbitrary falsehoods about other aspects of Chomsky's writings today, like what would be implied by an innate system of universal grammar, does it? —GKP]

  12. Ted said,

    July 27, 2010 @ 11:46 am

    If language (rather than culture) is the greatest influence on thought, people who learn a language from a textbook without learning about the culture of the target language ought to be able to communicate with and think like natives, absorbing the thought patterns right out of the language itself :-)

  13. Qov said,

    July 27, 2010 @ 11:53 am

    I first encountered the idea that language could shape thought when I read 1984 at about age 13. I was fascinated by the Newspeak proposition that vocabulary constrained thought, and I kept trying to find counterexamples. My teenage conclusion was that vocabulary makes it easier to manipulate ideas and build on them, the way having a single symbol to represent a complicated part of an equation makes it easier to manipulate the math, but that if the word was lacking and needed, one would be created and repurposed, the way a child wanting to comment on rain may point and say "juice!"

  14. dwmacg said,

    July 27, 2010 @ 1:09 pm

    @Nanani,

    You might be interested in this article by Dan Slobin on "thinking for speaking". Slobin's basic argument is that the language we speak forces us to attend to certain aspects of a situation when we're preparing to speak for it, but that that's not "a Whorfian straitjacket".

  15. J. Goard said,

    July 28, 2010 @ 3:09 am

    @GKP:

    I don't find any fault with your response, and of course I wouldn't assent to any such falsehoods. I was motivated to write by Jason's and your tone ("caricature", "a goof as deep as this" published, etc.), given what I view as the rampant caricature of pre-Chomskyan explanations for language acquisition and crosslinguistic variation in most of the introductory texts that mention them. Boroditsky, who would undoubtedly demonstrate a more nuanced understanding in other contexts, is writing a "gee whiz" piece for the general public here, not a serious academic salvo. So, given the history of the field, I thought personal criticism was a bit disproportionate in this case. Still, I accept your point. Thanks.

  16. David L Rattigan said,

    July 28, 2010 @ 7:13 am

    I found the article's examples pretty poor. For example, if the Indonesian language having "sit" for both present and past tense is significant, what of US English, which has "spit" for both present and past, where UK English has "spat"? Can we assume that reveals something important about how Americans and Brits think about the act of spitting?

    Then there's the Janet Jackson "wardrobe malfunction":

    The reports were identical except in the last sentence where one used the agentive phrase "ripped the costume" while the other said "the costume ripped." Even though everyone watched the same video and witnessed the ripping with their own eyes, language mattered. Not only did people who read "ripped the costume" blame Justin Timberlake more, they also levied a whopping 53% more in fines."

    Isn't this like saying, "Most people who read that the sea is blue were twice as likely to believe the sea is blue than those who read that the sea is green"? Aren't "the costume ripped" and "Justin Timberlake ripped the costume" fundamentally different pieces of information? Also, just out of interest, where does levying fines come into it? Was a bit baffled by that one, but maybe I was just being dense.

    My background is in biblical studies, a field in which some scholars have notoriously drawn all kinds of conclusions about the differences between Greek and Hebrew thought from supposedly significant linguistic differences. So I'm a bit skeptical of all that.

  17. Russell said,

    July 28, 2010 @ 11:33 am

    David:

    "The costume ripped" is part of the information conveyed by "He ripped the costume." I wouldn't say that they're "fundamentally" different. What's crucial is that these are two different descriptions of the same event, and that the way in which the event is described (transitive verb or not) affected the way some people thought about (certain aspects of) the event. Though the article doesn't make it clear, I'm pretty sure that this was supposed to be an attempt to show that, in certain circumstances, you can get English speakers to behave like Spanish and Japanese speakers (who prefer intransitive constructions for certain event types that English speakers would tend to describe with a transitive construction). She probably couldn't get many English speakers to say "the costume ripped," but she could get people to think about the event differently by describing it that way.

  18. Lane Greene said,

    July 28, 2010 @ 2:52 pm

    I've interviewed Boroditzky for my (forthcoming) book, and did my best with her underlying papers. While I disagree with some of her stronger interpretations of her finding, she certainly shouldn't be held responsible for the silly Whorfian claims that were being made before she was born and will no doubt go on being made after she and the rest of us have shuffled off this mortal coil. She sticks mainly to pretty careful statements about things she's tested. If I had to sum up in plain English my conclusion would be not "language shapes thought" (much less "language restricts thought"), but probably "language nudges thought" (in certain circumstances).

    What if silly Whorfian thinking were something we were innately prone to? Wouldn't that just blow her and Steven Pinker's minds at the same time?

  19. David L Rattigan said,

    July 28, 2010 @ 3:46 pm

    Okay, Russell, agreed that "He ripped the dress" implies or contains (or whatever the correct word is–I'm not a trained linguist, so I lack the technical terms) "the dress ripped," but the reverse is not true. So of course if you are told explicitly "Justin Timberlake ripped the dress," you would be more likely to believe that, well, that Justin Timberlake ripped the dress. "The dress ripped" doesn't say anything about the agent. So I'm still not sure why forming different conclusions after hearing the two is surprising, or how it reveals anything about underlying thought patterns. To me it just reveals that the one explicitly states (or reminds the reader) that Timberlake was the agent.

  20. Stephen Jones said,

    July 28, 2010 @ 11:36 pm

    While I disagree with some of her stronger interpretations of her finding, she certainly shouldn't be held responsible for the silly Whorfian claims that were being made before she was born …..

    We can certainly hold her responsible for the silly Whorfian claims in an article she's written.

    She sticks mainly to pretty careful statements about things she's tested.

    Not in this case she hasn't. She's deliberately setting up a false scenario (Whorf v Chomsky), and then pretending that Whorf has lately been justified because of new reports, even though the research on this has been mainstream since the 1970s at least.

    The fact that the woman can write the whole article without even letting any of the readers know there is such a field of study as cognitive linguistics, tells us exactly how warped the whole article is.

    And let's not talk at the way perfectly good information is turned into useless junk in the little blue box on the side.

  21. D said,

    July 30, 2010 @ 10:58 am

    The "Timberlake" experiment seems to me like it is almost exactly similar to Elisabeth Loftus' eye-witness studies, with the same results. Only, her point was that memory is unreliable and easily re-shaped, and theirs is a point about to what degree language might shape thoughts.

    It is interesting how such different conclusions can be drawn from what basically is exactly the same observation.

  22. UPDATE ON THE LINGUISTIC RELATIVITY WARS. | Pater Familias said,

    July 30, 2010 @ 1:05 pm

    […] and, some fifty years later, it is still debated. There is an excellent blog about linguistics, Language Log, which is a good place to follow this and other issues in […]

  23. Jerry Friedman said,

    July 31, 2010 @ 3:11 pm

    @David L. Rattigan: The participants in the experiment saw the video, so they have the information that Timberlake ripped Jackson's dress. And every American knows it. (Slight exaggeration.)

    The fining comes in because our Federal Communications Commission has the power to fine people and companies that violate its standards, including those on "decency". It did impose a record fine on the broadcaster, but the fine is still under appeal, according to the Wikipedia article.

    Spat for the past tense of spit is still used here by many right-thinking people who ignore pointless prescriptivists.

  24. David L Rattigan said,

    July 31, 2010 @ 3:36 pm

    Jerry Friedman:

    The participants in the experiment saw the video, so they have the information that Timberlake ripped Jackson's dress.

    Still I don't see the language-thought connection. All the agentive version did was remind the reader that it was Timberlake's fault. I don't think it's so much about the language used as the fact it hammered the point a bit.

    Thanks for the other info. Still don't quite get the "fines" thing, though. The way it's worded, it sounds like the study participants who read the report were the ones who were fined.

  25. HOW LINGUISTS THINK OF A “WARDROBE MALFUNCTION.” | Pater Familias said,

    July 31, 2010 @ 10:20 pm

    […] the other discussants. In Professor Boroditsky's article, and in the comments on Language Lab here, "wardrobe malfunction" is discussed as a "nonagentive coinage introduced into the […]

  26. Daniel Harbour said,

    August 13, 2010 @ 12:22 pm

    I agree with much of what has been said, beginning with Jason Merchant's surprise at the misportrayal of universal grammar's empirical commitments. My two cents' worth (on my blog): http://daniel-harbour.blogspot.com/2010/08/does-thought-influence-language-well-at.html.

  27. [Lesetipp] Kleine Auslese « [ʃplɔk] said,

    August 17, 2010 @ 9:05 am

    […] bei Language Log.) Von Duden-DNA und einer Außerirdischensprache von Stephan Matthiesen […]

  28. Pomplemoose said,

    September 1, 2010 @ 3:01 am

    @David: The connection is that certain languages are much more likely or even grammatically required to include the agent ("Justin Timberlake tore the dress") while other languages are much more likely to say simply "the dress tore". This distinction, in turn, affects people's propensity to assign blame or levy fines for that particular action. So a Japanese Timberlake may have found himself in more hot water than a Spanish Timberlake, simply by virtue of the grammar of the language in which he tore the dress.

RSS feed for comments on this post