[xmca] Smotri, Sinochek!

From: David Kellogg (vaughndogblack@yahoo.com)
Date: Fri Mar 30 2007 - 15:20:06 PST

As Mike says, the relationship of Vygotsky to Freire is not at all straightforward (for one thing, you must pass through the Catholic church at some point). I think this is even more true of the relationship of Pinker to his peculiar brand of asocial politics (which I think is in a very curious way symmetrical to that of his mentor Chomsky to anarchism).
  First, a disclaimer. I didn't read "The Blank Slate", and the very title of "How the Mind Works" made my head spin. So I'm not terribly up to speed with Pinker's politics, and it is very convenient at this point for me to say they are in a separate module from his linguistics. It saves me money, reading time, and exasperation.
  But I do know a thing or two about his "semantic boostrapping hypothesis". It's an attempt to address the question that Cole and Cole ask with their beautiful watercolor (in the Development of Children, p.278) of a father holding a child at the window and saying "Smotri, sinochek, tam sidit ptitsa!".
  Cole and Cole want to know how the child is able to figure out what all the words in this utterance mean, and specifically that "ptitsa" refers to a bird, and not a helicopter, a cat, a man, a building or a tree, all of which are also visible outside the window.
  Pinker's initial answer was that ALL grammatical categories are "bootstrapped" from semantic ones. The child hears a stressed, utterance final sound "ptitsa" and sees an object, namely a bird, and assumes that the sound has a semantic relationship to the perceived object.
  Since the perceived object is an entity, the child assumes that the sound represents an abstract category which is NOT visible, a conceptual category of words that refer to entities, viz. a noun, a concept to which Pinker, as a loyal Chomskyan, assumes a genetic predisposition on the child's part.
  Verbs are handled the same way. The child sees a process, and assumes that the sounds accompanying the process refer semantically, and thus represent verbs. (It is absolutely unclear to me how this sits with the obvious fact that children acquire "be" verbs very early, and yet the processes that "be" verbs encode are anything but visible, or rather they are TOO visible to be salient.)
  But it's when Pinker tries to deal with argument structure that his argument (no pun intended) really falls apart. If, Pinker says, the child knows semantically that a phrase is playing the role of agent (say, "the boy" in "the boy threw rocks") then the child is able to infer that "the boy" represents the abstract category of "subject", for which the child also has a hard-wired predilection.
  You will, undoubtedly, have already noticed that this does not work for Cole and Cole's sentence, "Smotri, sinochek" ("Look, my son!") nor for the sequel "tam sidit ptitsa" ("There sits a bird") since in neither utterance is the grammatical subject the agent of a perceptible physical process. So did Pinker.
  He argued, at first, that parents deliberately avoided using sentences where the agents of physical actions are not subjects (like "Look! It's raining, isn't it?") and sentences where the patients of physical actions are not objects (like "Finish your milk. Your glass is still full"). As you can imagine, this didn't get him very far with actual data.
  So the next edition of the semantic bootstrapping hypothesis posited THREE modules (perception/cognition, semantics, and syntax), not just two (semantics and syntax). First, you've got perception and cognition from context (seen as the SAME process, much as Martin was assuming). This generates a conceptual structure. Then the child labels the resulting categories semantically (action, thing, agent, patient).
  But then the child does NOT use probabilistic rules to link these semantic categories to syntactic ones (verb, noun, subject, object). Not at all! The child uses "universal and exceptionless" principles (to which, of course, the child has a genetic predisposition).
  The problem is, of course, that the rules are NOT universal; they vary wildly from language to language. They are also NOT exceptionless; Tomasello presents evidence that only about 15% of what the child hears is canonical Subject-Verb-Object sentences.
  So Pinker has no option but to attribute more and more and more of his theory to genetic hard wiring on the one hand and sheer parental will power on the other. By now, even his writing style is starting to remind me of the attempts by neo-Ptolemaics to shore up the concept of epicycles:
  "In the simplified two-box version (...) the child assumes that the first argument of an ACT semantic structure or the first argument of a BE structure if there is no ACT structure, is a subject,and parents avoid sentences in which the child would not be likely to construe the meanign as involving an ACT or BE relation (thus they would be free to use passives in contexts in which the child was likely to realize that something was being predicated of the patient).Similarly, in the older view,the child would assume that patients of physical motion were objects and parents would have to avoid verbs like "fill" where that is not true. In the newer view, children would assume that entities serving as the second argument of ACT and the first argument of GO were objects, and parents would have to avoid verbs whose objects were not motional patients only if the child would fail to construe that situation as involving a change of state of the motional goal." (Learnability and
 Cognition, MIT Press, 1989: 364).
  Got it? Neither did I! But it's starting to look to me like the parental bootstrap is a lot bigger than the syntactic boot. And to the child, it all must mean something like "You are here. And what you want to know is on the other side of the universe."
  David Kellogg
  Seoul National University of Education

Now that's room service! Choose from over 150,000 hotels
in 45,000 destinations on Yahoo! Travel to find your fit.
xmca mailing list

This archive was generated by hypermail 2b29 : Sun Apr 01 2007 - 01:00:11 PST