Re: Wilkins Neurolinguistic Preconditions

From: Stevan Harnad (harnad@coglit.ecs.soton.ac.uk)
Date: Wed Dec 08 1999 - 14:35:45 GMT


  ww> L]anguage is unlikely
  ww> to have evolved directly from communication-based precursors, nor
  ww> is it likely to have been based on those structures that subserve
  ww> communication.A communication-based account can
  ww> parsimoniously motivate neither the precise neural character nor
  ww> the apparent localization of human language cortex. Rather, as we
  ww> will show in detail,language came to utilize the processing
  ww> strategies available from newly evolving premotor cortex paired
  ww> with those aspects of
  ww> neural organization that allow for amodal concept formation and
  ww> yield structured abstract representations.

gj> Comment: I find it most interesting that they divorce language from
gj> communicational functions. This constitues, in my opinion, one of the
gj> novelties of this theory. In earlier hypotheses (bow- vow theory,
gj> social origin of language, language as a tool for working in community
gj> etc.) the need to communicate was the drive for language development.
gj> By separating communication from language, the authors do away with the
gj> telicity in previous theories, which is a welcome result in an
gj> evolutionary perspective (see for example the 'Evolutionary
gj> Perspectives' chapter in the Cognitive Neuroscience textbook, edited by
gj> Gazzaniga). This way, they can also free themselves from the heatly
gj> debated and very uncomfortable question of whether (the sophisticated
gj> structures of) syntax appeared all of a sudden, as if my chance or
gj> miracle or under selectional pressure. Their answer is neither, since
gj> language was a reappropriation.

True. But if language is NOT an adaptation of a communicative capcity,
what is it? It is certainly less a manipulative or motor behavior than
it is a communicative one. And although it certainly involves concepts
and the association between concepts, it is not clear how that is
manipulative rather than communicative.

  ww> in throwing, a two-handed release allows neither sufficient force
  ww> nor acceleration for an advantageous weapon trajectory. In flaking
  ww> stones, one
  ww> hand is used to steady the target (or core) stone while the skilled
  ww> movement is executed by the
  ww> other. With selection operating for behaviors associated with
  ww> unilateral performance,the stage
  ww> may be set for preferential handedness and cerebrallateralization.

gj> Comment: I feel a little dissatisfied with this argumentation. As the
gj> title of this subsection, and most importantly the main subject of the
gj> paper would suggest, the authors should show how language develops.
gj> What we get instead, however, is (I) the description of how the
gj> anatomical possibility of language evolved, and (ii) how cognitive
gj> processes in general (and conceptualization in particular) might have
gj> evolved. This however gives no account of language. If we accept that
gj> communicational needs constituted no driving force (as the authors have
gj> suggested earlier, they argued that communication could be done in ways
gj> other than language), then we have still no clue as to why and how
gj> language appeared.With the above given scenario, we might as well have
gj> ended up having a rich conceptual structure, some mimicing or gesturing
gj> for communication, but no language at all.

I agree with this criticism. In correctly noting that human language is
hardly just a variant of animal communication systems, the authors then
go too far, and divorce language origins entirely from communication,
hoping to find more powerful substrates in manipulation and multimodal
association. But although those no doubt played a role too, it still
leaves the basic origins question begged.

  ww> To relate the discussion of CS more directly to what we
  ww> have said regarding neuroanatomical structure, we suggest that CS
  ww> is the cognitive construct that is produced by the POT through its
  ww> interaction with Broca's area. By virtue of the POT,
  ww> human sensory input is highly processed in association cortex and
  ww> losesits modality-specific character; by virtue of Broca's area's
  ww> influence on the POT, the amodal representations are subject to
  ww> hierarchical structuring. Structured modality-neutral
  ww> representation, we suggest, is
  ww> the essence of CS. (Further,amodality and hierarchical structure
  ww> are necessary for feature abstraction, to which we return shortly.)
gj>
gj> Comment 1: Here again, it is quite visible ("CS is not part of the
gj> linguistic system per se.") that what the authors propose is not the
gj> development of language, but that of the conceptual system.

True, but there is nothing wrong with suggesting that there must be a
lot of cognitive structure in language capability. It is just that it
does not seem to get to what is peculiar to language, over and above
conceptual structure, and what its adaptive value and origin was.

gj> Comment 2: Jackendoff's model (in his The Architecture of the Language
gj> Faculty, 1997) is a complex architecture built up of the modules of
gj> many cognitive functions. Language itself consists of more modules.
gj> Conceptual structure is only a part of the whole model. The authors
gj> should indicate how modules other than CS are mapped into brain anatomy
gj> or brain evolution. However, they don't say much about it. They even
gj> ignore the purely linguistic modules. (Though elsewhere in the
gj> article, they make the (unmotivated) claim that syntax evolved later
gj> and gradually. )

Good point.

gj> Comment 3: They say that CS is hierarchical, and because it is in a way
gj> the representation of the world, there should be relations between the
gj> stored items. Some kind of "syntax" should exist even within CS (a
gj> language of thought), but where does it come from? In the authors'
gj> view, the syntax of language is only a development posterior to CS. And
gj> even if we somehow accept their arguments about CS, how is CS
gj> "syntax"/hierarchy mapped into the syntax of natural language. In
gj> linguistics, it is usually assumed that the link (one of the links)
gj> between syntax and semantics (that is CS) is assured by thematic roles
gj> (Agent, Patient, Goal, etc.). In the model described by the authors
gj> (later in the article), these roles and even some grammatical
gj> information (category, e.g. N, V etc. and subcategorisation) are
gj> included in CS (just like in Jackendoff), but no mention is made of how
gj> these categories and grammatical structures are created and put into
gj> CS.

These points are all valid. But W & W could still be right about these
cognitive components of language; it is that they do not give any
origins or adaptive scenario that makes the account miss the mark.

Let us say that the brain has nonlinguistic ways of learning categories,
and that it is true that language somehow involves bringing these
together in some way, and combining and recombining them. Is it not in
the service of communication that we do that? So we are back to a
communicative function after all, but not the rigid, species-specific
one of nonhuman species, but the much more powerful and general one that
is human language.

And yes, hand use and gesture and pantomime may well have been involved
in the origins of language, with the vocal modality taking over later,
because its advantages are obvious -- but only after the Adaptive
Advantages of language itself have made themselves felt: What are those
Adaptive Advantages of language itself (as opposed to the modality
advantages of vocalization over gestures)? And OVER WHAT are those
Adaptive Advantages advantages?

--------------------------------------------------------------------
Stevan Harnad harnad@cogsci.soton.ac.uk
Professor of Cognitive Science harnad@princeton.edu
Department of Electronics and phone: +44 23-80 592-582
Computer Science fax: +44 23-80 592-865
University of Southampton http://www.cogsci.soton.ac.uk/~harnad/
Highfield, Southampton http://www.princeton.edu/~harnad/
SO17 1BJ UNITED KINGDOM



This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:06 GMT