Friday 18 May 2012

An Ecological Approach to Language

Language is often held up as an example against the possibility of the radical (non-representational) psychology we advocate for. You might be able to explain perception-action without representations, people say, but we can't see how you'll ever be able to explain 'real cognition, like language' without them. It's finally time for us to begin chipping away at this criticism. In the next few posts I'll lay out a first draft of an embodied, ecological analysis of language use.

Psychologists usually assume that catching a fly ball and talking about catching fly balls are two different kinds of thing. I reject this assumption (it is just an assumption) and I am going to treat language use as the same kind of thing as other examples of embodied cognition. Treating language as just another instance of embodied cognition allows me to import the lessons learned from perception-action type tasks and apply these to language tasks. This will lead to very different questions about language use than are typical in the literature. The next post will describe what I mean by this in some detail.

I will then sketch out an initial task analysis of language use. As promised, we are guided here by Bingham's task-specific device approach, which is a method for studying high-dimensional complex systems such as the human perception-action system. The idea is that you must study how systems like these behave one task at a time by 1) characterising the task in detail, 2) identifying what is required to solve the task, 3) looking for those resources across both the organism and the task environment (cognition is both embodied and extended), 4) identifying the information that might support an organism's access to those resources, and finally 5) experimentally test your hypotheses within the confines of your chosen model task and identify the smart solutions that are options. The task analysis post will not go through all these steps because these ideas aren't ready for it yet. But, I'll spend quite a bit of time talking about elements of steps 1, 3, and 4.

So, here's what's coming up in the next two posts, which I plan to roll out beginning this weekend so that people might actually have time to read and comment on them:

Language isn't magical (but it is special): I describe how to think about language as a form of embodied cognition with a particular emphasis on the similarities and differences between perceptual information and linguistic information.

Language: A task analysis (kind of): I characterise 6 types of tasks for which linguistic information is either necessary or useful. I also lay out the notion of perceptual-linguistic systems and explain why this account of language is explicitly non-representational.

If you're reading this and thinking "This can't possibly work! I want to yell about it - on the internet!" then please wait until you've read the next post so you have something specific to criticise. These posts are a first go at a difficult question, not a final, fully-fledged research programme. We hope to get some good discussions going in the comments to help hone these ideas further!

Bingham, G. (1988). Task-specific devices and the perceptual bottleneck. Human Movement Science, 7 (2-4), 225-264 DOI: 10.1016/0167-9457(88)90013-9  Download


  1. Cool. This is not yelling, but more chatting lightly while sipping on Scotch and awaiting the rest of the story.

    I'm still suspicious of why we would ever want to try to tackle this without thoroughly evaluating what the behaviorists have been up to for the past 40 years. But hardly anyone at ISEP or ICPA seems to care a wit about what the behaviorists have come up with. Sigh.

    I will also mention that one problem with Language is that it is not very amenable to full Gibsonian stories. A full Gibsonian story looks like this: A cup affords my picking it up, because of a match between my body and the object; and I can perceive the pick-up-able-ness because I am attuned to higher order patterns in ambient energy that specify the consequence my behavior will have (namely that the cup will be off the table and in my hand).

    In language this would look like: A man affords my saying 'good morning', because of a match between my body and the man; and I can perceive the good-morning-able-ness because I am attuned to higher order patterns in the ambient energy that specify the consequence my behavior will have (namely that he will reply 'good morning' to me as well and open the door I as I approach).

    It is not an impossible position to defend, but it is very awkward in comparison to the cup example. If you loosen the commitment to the full Gibsonian apparatus a bit, I think language can be dealt with more intuitively. On the other hand, if you loosen the commitment too much, then you are adrift. I am looking forward to seeing what you have come up with.

  2. I just came across this comment by eric, which I think highlights a problem with the ecological approach to language. It seems to break down as conversation gets more complicated. But I don't see that as a problem, just a requirement to bound the domain of applicability.

    First, a joke: a guy at an airport ticket counter makes an embarrassing malapropism with respect to the female agent. Later in the bar he ruefully recounts it to another patron, who responds "Don't worry! Everyone does that from time to time. Why just this morning I meant to ask my wife to pass the cream but instead blurted out 'You f*ing bitch, you ruined my life!'"

    The reason some find that funny is that it violates all kinds of rules, both social and logical. Point being that there is, of course, a lot of flexibility in language, but it's not unbounded. That presumably is one reason Wittgenstein used the game of chess as an example for his idea of "language games". You can certainly move chess pieces any way you like, but some moves don't count as moves in a chess game - there are rules that must be followed. Similarly with language games.

    My contention is that in much casual conversation - eg, eric's "Good morning" exchange - the environment (which I assume includes the subject's internal body state, in particular the social rules applicable to the current language game) is so determinative as to render consequent linguistic events essentially "lawful". In eric's scenario, one response to a casual "good morning" could be refutation via a detailed account of one's miserable life, but normal people follow rules that declare that an inadmissible move in that specific language game. OTOH, many times people don't follow rules, eg, innovative use of words as metaphors, creating neologisms, etc. And then whatever it is that makes language "special" kicks in.

    Returning to the tennis analogy, back in the '70s someone introduced the double strung racquet which put such weird spin on the ball that it was extremely hard even for pros to handle. Such stringing was a perception analog of a metaphor - it broke the ecological rules. Presumably, in time players would have learned the new rules, but the stringing was outlawed instead. Similarly, presumably the rules of strange language games (eg, eco psych!) can be learned to such an extent that they approach lawfulness, at least when dealing with familiar subject matter. Ie, although the domain of applicability of relatively lawful linguistic behavior is not unbounded, it may often encompass more complex language games than we think.