LINGUIST List 9.1209

Tue Sep 1 1998

Disc: Universal Grammar (Proof)

Editor for this issue: Martin Jacobsen <>


  1. Dr. John Skoyles, S. Pinker's Response-Universal Grammar (Proof)

Message 1: S. Pinker's Response-Universal Grammar (Proof)

Date: Tue, 01 Sep 1998 12:06:59
From: Dr. John Skoyles <>
Subject: S. Pinker's Response-Universal Grammar (Proof)

Steve Pinker sent me this interesting response.

Dear Dr. Skoyles,

Many thanks for forwarding that very interesting report. I doubt that
it will convince UG-skeptics, since body images may be innate at the
same time that UG is not, but it is a good corrective to the fad of
citing findings on the neural substrates of learning as if they
implied that everything is learned. Given that half of our 100,000
genes are expressed primarily in the brain, that species differ from
one another innately, that humans differ from one another innately on
every quantitative trait, and that human cognitive accomplishments are
solutions to remarkably difficult engineering problems, I myself don't
doubt that much of neural organization is innate. Of course that
leaves open the question of what aspects of language in particular are

I agree that UG has been poorly defended and documented in the
lingusitics literature. I don't think, however, that the evidence you
cited is relevant, because it is based on several urban legends. One
is that networks can extract grammars for human languages; what they
extract is very simple toy grammars. Another is that children receive
corrections for all grammatical errors ("negative evidence"); some
children receive some statistical feedback for some errors, which is a
different matter (see G. Marcus, "Negative evidence in language
acquisition," Cognition, 46, 53-85). A third is that arguments for UG
have depended on the non-existence of negative evidence, a popular
misconception. As I point out in the Foreword of the new edition of
Language Learnability and Language Development,

"There is a widespread misconception that the no-negative-evidence
assumption is central to nativist theories, but this is historically
incorrect. Chomsky (1965, p. 32) wrote "let us ... assume
tentatively that the primary linguistic data [i.e., input to the child
-- SP] consist of signals classified as sentences and nonsentences..."
-- that is, negative evidence. Indeed, it was a staunch anti-nativist,
Martin Braine (1971a), who first emphasized the lack of negative
evidence and its implications, with important amplification by another
empiricist, Melissa Bowerman (1983, 1988). The implications of the
lack of negative evidence are central to any thoughtful discussion of
learning, nativist or empiricist, for the simple logical reason that,
without negative evidence, if the child guesses too large a language,
the world can never tell him he's wrong."

Despite all these reservations, I agree with you that more attention
has to be given to the empirical status of UG, both cross-linguistic
studies and discussions of genetic influences on neural organization
more generally.

If you are posting replies and correspondence to your original
posting, please feel free to include this letter among them.

Steve Pinker

Dr. John R. Skoyles
6 Denning Rd,
Hampstead, NW3 1SU
London, UK

Check out my Golden House-Sparrow award winning homepage

Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue