LINGUIST List 5.623

Thu 02 Jun 1994

Review: _The Language Instinct: How the mind creates language_

Editor for this issue: <>


  1. claudia brugman, book review/discussion

Message 1: book review/discussion

Date: Wed, 1 Jun 1994 12:54:07 +book review/discussion
From: claudia brugman <>
Subject: book review/discussion

Initiation of discussion of _The Language Instinct: How the mind creates
language_, by Steven Pinker. 1994, William Morrow.

I learned about this book on the NPR show _Talk of the Nation_. As I
turned it on, the guest was being excoriated by a caller for promoting the
downfall of the English language. He responded both cogently and
graciously. I was intrigued that _Talk of the Nation_ had invited a guest
who would articulate a responsible anti-prescriptive attitude toward
language use, and even more so when I learned that that guest was Steven

Naturally, the book is not primarily about prescriptivism. Rather, it is
an argument for considering the language capacity as innately given, and
identifying as the innate faculties certain of the constructs of generative
grammar. In the service of this aim, P gives an overview of the chief
subdisciplines of linguistics as well as aspects of cognitive science which
support the conception of language as a biologically-based skill.

Because it is a trade book, I approached this review from three
perspectives: first, as a professional with my particular theoretical
predilections; second, as a teacher of students with no prior training or
interest in linguistics; and third, as a member of the general readership
myself. I evaluated the book with respect to two issues: how it presents
fact and theory, and how it represents the fields of cognitive science and

I'll give a general summary of the book's contents and then raise some
issues which I think are worth discussing by readers of LINGUIST.

The Table of Contents is as follows:

 1, "An Instinct to Acquire an Art" introduces the language capacity and
the endeavors of linguistics. Pinker here highlights Chomsky's
contributions to the "mental grammar" to the exclusion of non-Chomskian
 2, "Chatterboxes" shows the innate drive toward linguistic complexity:
that all languages and varieties are equally complex, and that children
acquire complexity with or without complex input.
 3, "Mentalese" discusses the relationship between language and thought,
and I will bring up this discussion below. P's general point is that
mental representation is not equivalent to the language spoken.
 4, "How Language Works [sic]" introduces the construct of a grammar as a
recursive device with rules distinct from interpretive rules. It also
introduces syntactic constituency and movement rules.
 5, "Words, Words, Words" is about morphology, and 6, "The Sounds of
Silence" is about phonetics and phonology. Topics include: inflection vs.
derivation, the constituency of the word, and P's analysis of "the Toronto
Maple Leafs". P also shows that mental representations of sound structure
involve much more than how to make or hear sounds. P also discusses
English spelling conventions here, and not terribly well.
 7, "Talking Heads" introduces the problems of sentence comprehension from
AI to human processing constraints. Here P introduces "branching" and
explains why processing constraints make some structures harder to parse
than others. He discusses structural and lexical ambiguity as parsing
problems, and then (in a somewhat boggling sweep from the Watergate
transcripts to indirect speech acts) shows how parsing is only the first
step in understanding language.
 8, "The Tower of Babel" goes from the particular to the universal. P
gives "parameter-setting" as evidence for evolution. There is also a
rather ambitious discussion of the evolution(s) of language (the
differentiation of systems, not of organisms) and a mention of the
Nostratic hypothesis.
 9, "Baby Born Talking--Describes Heaven" outlines the order and schedule
of first language acquisition. P also discusses the how errors reveal the
properties of the child's approximations to grammar.
10, "Language Organs and Grammar Genes" describes what is believed about
the anatomy of the "language cortex"; here P discusses the Gopnik's results
on Specific Language Impairment.
11, "The Big Bang" is devoted to the evolutionary question and discusses
critically the efforts to teach language to nonhuman primates.
12, "The Language Mavens" concerns prescriptivism. Here P rebuts the
arguments of the Safires by discussing the psychological and/or systemic
motivations behind common "grammatical errors".
13, "Mind Design" wraps up the nature-nurture debate from an evolutionary
perspective and especially criticizes models which emphasize
cross-linguistic or cross-cultural differences to the exclusion of
universals and which thereby de-emphasize whatever biological (including
cognitive) capacities are common to a species.

Fifteen pages of notes cite scholarly works relevant to topics mentioned in
the text. The book also contains a glossary, with definitions, and often
examples, of the technical terms which Pinker uses liberally (but generally
comprehensibly) in the text. Some of these terms are: finite-state
device; dative; top-down. There are also an extensive bibliography and a
comprehensive index.

I enjoyed much about this book. Pinker's writing style comes as close as
one might hope to a series of public lectures. It contains a readable and
comprehensive overview of the major subdisciplines of linguistics as well
as the aspects of cognitive science and evolutionary biology that inform
our hypotheses about the innateness of the linguistic capacity. P
elucidates the issues which keep cognitive scientists fascinated, and
starts to give the reader a sense of the wonder of the language ability,
couching it in an evolutionary perspective. Perhaps this perspective will
make prescriptivism uninteresting to readers, and the cognitive sciences
important to them.

I appreciated the fact that Pinker has compiled results from acquisition,
from impairment, from neural-net modelling, and from comparisons with the
skills of other species. Together they offer compelling evidence that
something, indeed, is innate--something that plays out as language
behavior--and they make it in a way that will grab the general reader. I
take issue below with Pinker's conclusions from this evidence.

The topics covered would be pedagogically useful for introductory or
pop-linguistics courses. These include the notions of recursion and
grammaticality in the technical sense, and the point that languages are
comparably complex irrespective of the industrialization of their
associated societies. His outlines of the subfields of descriptive
linguistics are generally good for a class at the level of Linguistics for
Biologists, if the instructor doesn't object to the modularist bent
discussed below. It is also useful for professionals, since it provides a
quick-and-cleaned-up introduction to cognitive science topics which may
interest linguists but may not otherwise be accessible to them.

One unfortunatel property of P's easy style is a curled-lip disdain for the
vocabulary and modelling of linguistics, which finds no parallel in his
discussions of the other sciences. And Gazzaniga's back-cover blurb says
"He spares the reader the mumbo jumbo of linguistics . . .". I wonder why
linguist-bashing (which previous postings have noted is rampant in the
popular press) should find a haven here?

Also, the style sometimes allows for suspicious rhetoric. In several
places P is seeming to be objective while using tricks to undermine his
critics--e.g. referring to Gould and Lewontin's dissentions against certain
positions as "potshots" (p. 359). He sometimes decontextualizes or
inaccurately paraphrases others' positions, making them seem incoherent.
He uses suspicious argument forms throughout Chapter 3 to rebut
Whorfianism. And in Chapter 4 P says that "syntax is a Darwinian 'organ of
extreme perfection and complication'." (p. 124) He has certainly shown its
complexity, but has invoked, rather than elucidated, its evolutionary
significance by using the word "Darwinian". And again, the discussion in
Chapter 8 of the Nostratic hypothesis lends plausibility to the notion that
the language capacity evolved by showing possible parallels to the
evolution of the organism, rather than by giving any direct supporting

P wants to demonstrate that the language capacity is a product of
evolution, that it importantly distinguishes humans from all other
creatures, and that it is innate--a concept which he implicitly equates
with "instinctive". This equivalence is another marker of a pop-science
book, since "instinct" has no technical meaning within Darwinian
evolutionary biology. Here it goes undefined, though P admits that it's a
"quaint" term (p. 18). The closest he gets to characterizing it is to say
"people know how to talk in more or less the sense that spiders know how to
spin webs." (p. 18)

The second leg of this suspicious four-way equation is the familiar and
controversial one between "innate" and "universal". The third step
involves the implicit identification of (a modularist) UG with universal
properties of language, another issue which LINGUIST readers will recall
from recent discussions. So:

 instinctive = innate = universal = modular

Pinker presents this model as THE result from linguistics, and the only
interpretation of the facts. This raises an issue worth exploring: In a
trade book, the author should balance the need to be accessible to the
public against the need to portray accurately the field and its issues. So
is it acceptable to present a picture which in some way shows what we do,
and why it is interesting, but which does not represent the theoretical and
social diversity of the field? Does Pinker do a disservice to those whose
ideas are ignored or dismissed in this work, or does his service to the
field as a whole justify this one-sidedness? How would this work be
evaluated if P's assumptions reflected a non-dominant paradigm?

For a popular audience it would be virtually impossible to explicate the
theoretical Gordian knots which substitute for equals signs in the equation
above. Nevertheless, it is disturbing that the results from linguistics
and cognitive science aren't accorded higher status than interpretations of
those results. Pinker does not mention that other interpretations even
exist. The claim that the "language ability" is a product of evolution is
logically independent of the claim that the innate capacity looks like
"UG", but since the popular readership has no practical way to evaluate the
difference between empirical results and their theory-dependent
interpretation, P's argument amounts to propagandizing for the generative
paradigm. Because I am agnostic as to the autonomy of the "syntax module"
from other capacities, and sceptical of any claims that it's syntax in
particular that is innate, I was unhappy with the easy slide from one
conclusion--that something is innate--to the next--that such things as
grammar modules exist, and that the obviously innate parts of the language
capacity are the ability to distinguish phones and X-bar morphosyntax.

Pinker neglects to mention that the results of the cited studies, and the
conclusion that there is an innate capacity which is manifested in language
use, are consistent with a number of models which differ importantly from
his. We need not conclude the existence of a left-hemisphere cortex which
is specialized for language, and if we do accept that, we need not accept
that the cortex is modularized in the popularly-presented ways. (To P's
credit, he admits that we don't know what Broca's and Wernicke's areas are
for, and that MRI and PET scans are still crude indicators of cortical
activity; again, though, he doesn't spell out what our ignorance implies
about the plausibility of his model.) If we accept X-bar morphosyntax, we
might still want to look for an explanation (semantic, cognitive, and/or
functional) for why morphosyntax should have this geometry, rather than
ascribing it to evolution.

 A question important to professionals is implied by the evolutionary
perspective and the talk of "instinct": where does cognition fit in? Some
of P's discussion of the relationship between language and thought is
adequate, but the specifically-anti-Whorfian discussion is characterized by
sloppy argumentation, and he again doesn't make clear that "Whorfianism"
and "mentalese" don't exhaust the possible answers to the question.

P dispels the myth that "instinct" precludes learning, so there is no
danger of the book promulgating the position that (as a non-linguist
colleague put it) "the meaning of 'carburetor' is in our heads at birth".
Perception is similarly acknowledged as playing an important role in
language use. But despite the considerable ink spilled debunking
Whorfianism, P does not provide a positive model of how speakers relate
their language to the "mentalese" which P tells them to believe in. From
his Dawkinsian point of view, we're just bundles of very specialized
mechanisms, each as sophisticated as Turing machines (which P describes as
the "scientifically respectable" model of mental representation (p. 73 ff)
). The issue of how language expresses the infinity of human experiences
is addressed by appeal to recursive function theory, leaving imagination
and cognition (or whatever a Darwinian would identify as their analogues)
unmentioned. This is particularly unfortunate given Pinker's (and the
readers') acknowledged fascination with the relationship between language
and thought.

The question of the universality of the "language capacity" could have been
recast as the question of the universality of the cognitive capacity. P's
painstaking demonstration in the last chapter that there are important
universals of culture and conception should lead the thoughtful reader to
this question, except that Pinker has spent 400 pages precluding its


Dr. Claudia Brugman
English Department and
School of Languages
University of Otago
PO Box 56
Dunedin, New Zealand
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue