LINGUIST List 14.2954

Wed Oct 29 2003

Review: Cognitive Science, Semantics: Talmy (2003)

Editor for this issue: Naomi Ogasawara <>

What follows is a review or discussion note contributed to our Book Discussion Forum. We expect discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in. If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for review." Then contact Simin Karimi at


  1. Laura M. Wagner, Toward a Cognitive Semantics, 2 vols.

Message 1: Toward a Cognitive Semantics, 2 vols.

Date: Wed, 29 Oct 2003 17:51:47 +0000
From: Laura M. Wagner <>
Subject: Toward a Cognitive Semantics, 2 vols.

Talmy, Leonard (2003) Toward a Cognitive Semantics, Volume 1: Concept
Structuring Systems; Volume 2: Typology and Process in Concept
Structuring, MIT Press, Language, Speech, and Communication series.

Announced at

Laura Wagner, Wellesley College

This two-volume set of Talmy's collected works is now out in paperback
form. Volume 1 contains 8 papers which constitute the main pillars of
Talmy's program: The Relation of Grammar to Cognition, Fictive Motion
in Language and 'Ception', How Language Structures Space, The
Windowing of Attention in Language, Figure and Ground in Language,
Structures that Relate Events, Force Dynamics in Language, and
Cognition and The Semantics of Causation. Volume 2 contains another 8
papers including Talmy's important typological work (Lexicalization
Patterns, Surveying Lexicalization Patterns, A Typology of Event
Integration, and Borrowing Semantic Space: Diachronic Hybridization)
as well as some more speculative works (Semantic Conflict and
Resolution, Communicative Goals and Means: Their Cognitive
Interaction, The Cognitive Culture System, and A Cognitive Framework
for Narrative Structure). In addition, both volumes begin with a
brief introduction to Talmy's approach.

All of the papers in these volumes have been published previously.
However, Talmy has revised and expanded them, and has incorporated
some previously unpublished data and arguments (including some from
his dissertation). The intended audience for these volumes is
professionals and advanced students. The writing is dense and
technical. Talmy coins many of his own technical terms (e.g. agonist,
satellite) and although these terms are periodically used throughout
the volumes, they are only clearly defined in the original paper in
which they were coined.

As Talmy's general program is well-known and several of these papers
are old favorites, I will provide only a brief review of Talmy's
approach instead of summarizing each paper in turn. In a nutshell,
Talmy's Cognitive Semantics ''seeks to ascertain the global integrated
system of conceptual structuring in language'' (Vol. 1, p. 3). Talmy
takes basic conceptual categories such as space, time, events,
causation, perspective, agentivity, and intentions and delineates how
these concepts are structured in the linguistic system. In each
domain, language creates -- largely through the use of closed class
elements -- a schematic organizational structure into which more
particular knowledge -- largely in the form of open class elements --
is situated. The nature of the schematic organization is revealed
through a close analysis of the grammatical means for expressing these
basic concepts, the range of information within these concepts that
can (and in some cases, cannot) be expressed grammatically, and how
the pieces of information interact with each other in linguistic
expression. At the heart of the program is the idea that there are
intrinsic and inextricable links between the ways we conceptualize the
world and the way language structures those conceptualizations.

Evaluating these volumes is a little bit difficult because to a
certain extent, these papers have already been evaluated by the field
as a whole. Some of the ideas presented in these papers have become
thoroughly entrenched in linguistic and psycholinguistic research. In
particular, Talmy's typological analysis of motion events (with
languages being either predominantly verb or satellite-framed), and
the use of figure and ground as a way to understand argument relations
are simply part of the common currency of our field.

Although many of the ideas will be familiar to readers, there are
still many delights worth (re-)visiting in these volumes: the
arguments for maintaining the open/closed class distinction, the
wealth of cross-linguistic examples (particularly in the papers
originally written in the 1970's), the linguistic schematization of
space. There are lots of examples that make you think, and enough
data to mount a challenge to nearly anybody's semantic theory. Talmy
envisions his program extremely broadly, and one can't help but have a
certain affection for someone willing to put forward such a grand
theory. There's an appealing willingness to take on any problem; the
goodwill he generates with his successes is more than enough to carry
one through the somewhat more indulgent applications of the program.

With a collection this broad and this rich, it seems small- minded to
identify the occasional argument that falls flat or example that isn't
convincing. That said, there are a few elements that recur throughout
the volumes that do bear criticizing. The first criticism has to do
with the revised nature of the papers. Talmy could have opted to
simply publish his collected papers in their original form but he
chose instead to update and revise them. Given this choice, it is a
little disappointing how little reference there is to more current
work in this area. There are a few brief asides about the work of
Goldberg and Fauconnier, for example, but virtually no attempt is made
to connect to the wealth of work in Cognitive Linguistics and
psycholinguistics that directly develops some of the ideas that Talmy

The second criticism concerns the way Talmy undersells the cognitive
dimension of his cognitive semantics. In several places in these
volumes, Talmy makes explicit claims about how the mind itself works
(e.g. Vol 2-ch.1's claim that backgrounded elements yield lower
cognitive costs , and Vol 1-ch.2's discussion of apparent motion, the
lasting effects of infant locomotion, and perhaps most astonishingly,
identification of 13 parameters of cognitive functioning). In his
introductory remarks, Talmy quite clearly articulates that his program
begins with language but is aimed to extend towards (hence the
''towards'' in the title of these volumes) cognition more generally.
However, the casual assumption that the linguistic analysis has
already told us about cognition seems to be suggesting that the
program has already hit its mark. This is simply not the case.
Implicit in Talmy's program is a strong hypothesis, namely, that the
linguistic structuring of concepts is deeply revealing about how those
concepts are structured in the mind more generally.

Testing this hypothesis requires two steps: first, we need to have a
detailed idea about how language structures the concepts and second,
we need to investigate the extent to which these structures actually
correspond to our non- linguistic mental structures. Talmy's work is
focused exclusively -- perhaps brilliantly, but exclusively
nonetheless -- on the first step in this testing procedure. There are
plenty of researchers out there who have taken up the second step
seriously (I count myself among them) but Talmy isn't one of them --
certainly he isn't in these volumes. It is perhaps instructive to
compare Talmy's treatment of Psychology to his treatment of Atsugewi.
Talmy provides many linguistic examples from Atsugewi and these
examples strengthen his arguments and his program in general because
his knowledge is both specific and deep. His knowledge of Psychology
- at least as illustrated in these volumes -- is neither. Talmy is
certainly not obligated to discuss the cognition end of his program
(he has certainly done his share of the work already), but he needn't
undermine it by discussing it in such a careless fashion.

My final criticism has to do with the sheer overwhelming force of the
number of lists presented in the volumes. There are no fewer than 20
features identified as relevant for structuring spatial configurations
using closed-class elements and another 20 elements define the
force-dynamic system; there are also 6 sub-categories within the
perspective system, 9 principles governing semantic borrowing, 8
associated characteristics of figures and grounds, 15 communicative
goals, 7 sub-types of the degree- of-differentiation parameter within
narrative structure, 35 semantic categories expressed (or not) by
verb-complex elements, and so on and on. What's more, Talmy
periodically notes that other elements besides those listed are
relevant, leaving one with the depressing feeling that however
overwhelming these lists are, they aren't even exhaustive. In some
cases -- e.g. the typological analysis of motion verbs and possibly
the analysis of force dynamics -- the lists are part of a systematic
exploration of a few well articulated semantic parameters. But more
often, the lists seem simply to be descriptions of a set of phenomena
without necessarily stemming from or leading to any greater
theoretical organization. This criticism may in fact be deeply
unfair: nobody ever criticizes a chemist for identifying yet one more
element to add to the long list of elements already discovered.
Language is a complicated creature and it may simply be the case that
it takes this many features to adequately describe it; it's not
Talmy's fault that language is what it is. Nevertheless, one does
start longing for some general principles which, while they might not
be as descriptively accurate, would at least give the illusion of a
coherent explanation.

Criticisms notwithstanding, these volumes present an extremely
important body of work that has had, and no doubt will continue to
have, extensive influence in linguistics, psycholinguistics, and
cognitive science more generally. Bringing all of Talmy's papers
together into one place is quite helpful and these volumes will
doubtless grace the shelves of researchers for years to come.


Laura Wagner is currently at Wellesley College. Her research centers
on the acquisition of tense and aspect and the non-linguistic
representation of event concepts in development.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue