Cognitive Linguistics: Basic Readings

≥
Cognitive Linguistics Research
34

Editors
Dirk Geeraerts
René Dirven
John R. Taylor
Honorary editor
Ronald W. Langacker

Mouton de Gruyter
Berlin · New York
Cognitive Linguistics:
Basic Readings

Edited by
Dirk Geeraerts

Mouton de Gruyter
Berlin · New York
 Mouton de Gruyter (formerly Mouton, The Hague)
is a Division of Walter de Gruyter GmbH & Co. KG, Berlin

앪
앝 Printed on acid-free paper
which falls within
the guidelines of the ANSI
to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data

Cognitive linguistics : basic readings / edited by Dirk Geeraerts.

p. cm. ⫺ (Cognitive linguistics research ; 34)
Includes bibliographical references.
ISBN-13: 978-3-11-019084-7 (hardcover : alk. paper)
ISBN-10: 3-11-019084-2 (hardcover : alk. paper)
ISBN-13: 978-3-11-019085-4 (pbk. : alk. paper)
ISBN-10: 3-11-019085-0 (pbk. : alk. paper)
1. Cognitive grammar. I. Geeraerts, Dirk, 1955⫺
P165.C6424 2006
415⫺dc22
2006023786

Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available in the Internet at [Link]

ISBN-13: 978-3-11-019084-7
ISBN-10: 3-11-019084-2
ISSN 1861-4132

쑔 Copyright 2006 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin
All rights reserved, including those of translation into foreign languages. No part of this book
may be reproduced or transmitted in any form or by any means, electronic or mechanical,
including photocopy, recording, or any information storage and retrieval system, without
permission in writing from the publisher.
Typesetting: Selignow Verlagsservice, Berlin
Printed in Germany
Contents

Introduction
A rough guide to Cognitive Linguistics
Dirk Geeraerts 1

Chapter 1
Cognitive Grammar
Introduction to Concept, Image, and Symbol
Ronald W. Langacker 29

Chapter 2
Grammatical construal
The relation of grammar to cognition
Leonard Talmy 69

Chapter 3
Radial network
Cognitive topology and lexical networks
Claudia Brugman and George Lakoff 109

Chapter 4
Prototype theory
Prospects and problems of prototype theory
Dirk Geeraerts 141

Chapter 5
Schematic network
Ambiguity, polysemy, and vagueness
David Tuggy 167

Chapter 6
Conceptual metaphor
The contemporary theory of metaphor
George Lakoff 185

Chapter 7
Image schema
The cognitive psychological reality of image schemas and their
transformations
Raymond W. Gibbs, Jr. and Herbert L. Colston 239
vi Contents

Chapter 8
Metonymy
The role of domains in the interpretation of metaphors and metonymies
William Croft 269

Chapter 9
Mental spaces
Conceptual integration networks
Gilles Fauconnier and Mark Turner 303

Chapter 10
Frame semantics
Frame semantics
Charles J. Fillmore 373

Chapter 11
Construction Grammar
The inherent semantics of argument structure: The case of the English
ditransitive construction
Adele E. Goldberg 401

Chapter 12
Usage-based linguistics
First steps toward a usage-based theory of language acquisition
Michael Tomasello 439

Epilogue
Trajectories for further reading
Dirk Geeraerts 459
Publication sources

Chapter 1. Cognitive Grammar

Ronald W. Langacker
1990 Chapter 1. Introduction in Concept, Image, and Symbol: The Cognitive Basis of
Grammar, Ronald W. Langacker, 1–32. Berlin/New York: Mouton de Gruyter.
(Reprint of An introduction to cognitive grammar. Cognitive Science 10(1):
1–40. 1986.)

Chapter 2. Grammatical construal

Leonard Talmy
1988 The relation of grammar to cognition. In Topics in Cognitive Linguistics,
Brygida Rudzka-Ostyn (ed.), 165–205. Amsterdam/Philadelphia: John
Benjamins.

Chapter 3. Radial network

Claudia Brugman and George Lakoff
1988 Cognitive topology and lexical networks. In Lexical Ambiguity Resolution:
Perspectives from Psycholinguistics, Neuropsychology, and Artificial Intel-
ligence, Steven L. Small, Garrison W. Cottrell, and Michael K. Tanenhaus
(eds.), 477–508. San Mateo, CA: Morgan Kaufmann.

Chapter 4. Prototype theory

Dirk Geeraerts
1989 Prospects and problems of prototype theory. Linguistics 27(4): 587–612.

Chapter 5. Schematic network

David Tuggy
1993 Ambiguity, polysemy, and vagueness. Cognitive Linguistics 4(3): 273–290.

Chapter 6. Conceptual metaphor

George Lakoff
1993 The contemporary theory of metaphor. In Metaphor and Thought, Andrew
Ortony (ed.), 202–251. Cambridge: Cambridge University Press.
viii Publication sources

Chapter 7. Image schema

Raymond W. Gibbs, Jr. and Herbert L. Colston
1995 The cognitive psychological reality of image schemas and their transforma-
tions. Cognitive Linguistics 6(4): 347–378.

Chapter 8. Metonymy
William Croft
1993 The role of domains in the interpretation of metaphors and metonymies. Cog-
nitive Linguistics 4(4): 335–370.

Chapter 9. Mental spaces

Gilles Fauconnier and Mark Turner
1998 Conceptual integration networks. Cognitive Science 22(2): 133–187.

Chapter 10. Frame semantics

Charles J. Fillmore
1982 Frame semantics. In Linguistics in the Morning Calm, Linguistic Society of
Korea (ed.), 111–137. Seoul: Hanshin Publishing Company.

Chapter 11. Construction Grammar

Adele E. Goldberg
1992 The inherent semantics of argument structure: The case of the English ditran-
sitive construction. Cognitive Linguistics 3(1): 37–74.

Chapter 12. Usage-based linguistics

Michael Tomasello
2000 First steps toward a usage-based theory of language acquisition. Cognitive
Linguistics 11(1/2): 61–82.

The papers are reprinted with permission. They appear in their original form, except for
the following changes: bibliographical entries, section numberings and other typographi-
cal elements have been added adjusted to the Mouton style, temporary and incomplete
references have been updated, and cross-references to the original volumes have been
deleted.
Introduction
A rough guide to Cognitive Linguistics

Dirk Geeraerts

So this is the first time you visit the field of Cognitive Linguistics, no? You may
need a guide then. Sure, when you move through the following chapters of this
volume, you get to see a top twelve of sights that you should not miss: a delightful
dozen of articles written by authorities in the field that each introduce one of the
conceptual cornerstones of the theoretical framework of Cognitive Linguistics.
Still, to give you a firm reference point for your tour, you may need some initia-
tion to what Cognitive Linguistics is about. That’s what the present chapter is for:
it provides you with a roadmap and a travel book to Cognitive Linguistics. It’s
only a rough guide, to be sure: it gives you the minimal amount of background
that you need to figure out the steps to be taken and to make sure that you are
not recognized as a total foreigner or a naïve apprentice, but it does not pretend
to supply more than that.
To understand what you may expect to find in this brief travel guide, we need
to introduce one of the characteristic ideas of Cognitive Linguistics first – the
idea, that is, that we should not just describe concepts and categories by means
of an abstract definition, but that we should also take into account the things that
the definition is about, if we are to achieve an adequate level of knowledge. Take
birds: you can define birds as a certain type of animal with certain characteristics
(like having wings, being able to fly, and being born from eggs), but if you want
to get a good cognitive grip on what birds are, you will want to have a look at
some typical birds like robins and sparrows and doves, and then maybe also at
some less typical ones, like chickens and ostriches.
It’s no different when you are dealing with linguistic theories. You have to
know about the scientific content of the theory, that is to say, the abstract defi-
nition of the approach: the topics it deals with, the specific perspective it takes,
and the observations it makes. But you also have to know about the sociology
of the theory: the people it involves, the conferences where they meet, the chan-
nels in which they publish. Introductions to linguistics tend to focus on the first
perspective only, but the present guide will take the second into account just as
much as the first.
2 Dirk Geeraerts

1. What is so special about this place?

Theories in linguistics tend to be fairly insular affairs: each theoretical frame-

work tends to constitute a conceptual and sociological entity in its own right, with
only a limited number of bridges, market places or even battlegrounds shared
with other approaches. Cognitive Linguistics, when considered in the light of
this metaphor, takes the form of an archipelago rather than an island. It is not
one clearly delimited large territory, but rather a conglomerate of more or less
extensive, more or less active centers of linguistic research that are closely knit
together by a shared perspective, but that are not (yet) brought together under
the common rule of a well-defined theory. The present volume contains an intro-
duction to twelve fundamental parts of that theoretical conglomerate – a tour of
twelve central islands, if you wish: Cognitive Grammar, grammatical construal,
radial network, prototype theory, schematic network, conceptual metaphor, image
schema, metonymy, mental spaces, frame semantics, construction grammar, and
usage-based linguistics.
We will define in a moment what links hold these concepts together and why
each of them separately is important, but at this point, the chief thing is to real-
ize that there is no single, uniform theoretical doctrine according to which these
research topics belong together: Cognitive Linguistics is a flexible framework rather
than a single theory of language. From the point of view of category structure (one
of the standard topics for analysis in Cognitive Linguistics), this recognition is
again one way in which Cognitive Linguistics illustrates its own concepts. As we
mentioned a moment ago, Cognitive Linguistics emphasizes the fact that defin-
ing a category may involve describing some of its principal members rather than
just giving an abstract definition. But it also stresses that the abstract definition
need not consist of a single set of defining characteristics that belong uniquely
and distinctively to that category. Think of birds again: when we describe the
features of birds, we soon notice that the features we would like to think of as
definitional for birds are not shared by all members of the species: we may even
find birds like the penguin or the kiwi, that have no wings to speak of, cannot
fly, and don’t have feathers but that are rather covered with some kind of fluff. In
such cases, we say that a category has a family resemblance structure: different
types of birds resemble each other like the members of a family would, but there
is no single set of attributes that necessarily shows up in all the members of the
family. Again, it is no different with a linguistic framework like Cognitive Lin-
guistics: it constitutes a cluster of many partially overlapping approaches rather
than a single well-defined theory that identifies in an all-or-none fashion whether
something belongs to Cognitive Linguistics or not.
Then again, the recognition that Cognitive Linguistics has not yet stabilized
into a single uniform theory should not prevent us from looking for fundamental
 Introduction: A rough guide to Cognitive Linguistics 3

common features and shared perspectives among the many forms of research that
come together under the label of Cognitive Linguistics. An obvious question to
start from relates to the ‘cognitive’ aspect of Cognitive Linguistics: in what sense
exactly is Cognitive Linguistics a cognitive approach to the study of language?
Terminologically speaking, we now need to make a distinction between Cog-
nitive Linguistics (the approach represented in this reader), and uncapitalized
cognitive linguistics – referring to all approaches in which natural language is
studied as a mental phenomenon. Cognitive Linguistics is but one form of cogni-
tive linguistics, to be distinguished from, for instance, generative grammar and
many other forms of linguistic research within the field of cognitive science.
What, then, determines the specificity of Cognitive Linguistics within cognitive
linguistics?
There are a number of characteristics that need to be mentioned: one basic
principle that is really, really foundational, and four tenets that spell out this
fundamental notion. The foundational point is simply that language is all about
meaning. As it says in the Editorial Statement of the very first issue of the jour-
nal Cognitive Linguistics, published in 1990, this approach sees language ‘as an
instrument for organizing, processing, and conveying information’ – as something
primarily semantic, in other words. Now, it may seem self-evident to you that
a ‘cognitive’ approach to language focuses on meaning, but if you are familiar
with generative grammar (i.e. Chomskyan linguistics), you will know that this
is a theory that thinks of language primarily in formal terms: as a collection of
formal, syntactic structures and rules (or constraints on such structures and rules).
And generative grammar is definitely also a ‘cognitive’ conception of language,
one that attributes a mental status to the language. So we have to be careful with
the term cognitive in Cognitive Linguistics. It does not only signal that language
is a psychologically real phenomenon (and that linguistics is part of the cogni-
tive sciences), but also that the processing and storage of information is a crucial
design feature of language. Linguistics is not just about knowledge of the language
(that’s the focus of generative grammar), but language itself is a form of knowl-
edge – and has to be analyzed accordingly, with a focus on meaning.
Conversely, Cognitive Linguistics is not the only linguistic approach focusing
on meaning: there are diverse forms of functional approaches to language that
go in the same direction. And further, formal semantics is clearly a semantically
oriented approach as well. It lies beyond the scope of this introduction to provide a
systematic comparison with these other semantic approaches, but you will certainly
be interested in what is particular about the way in which Cognitive Linguistics
deals with meaning. So that brings us to the four specific characteristics that we
announced earlier: each of them says something specific about the way Cognitive
Linguistics thinks about meaning. (By the way, the captions we use to introduce
the features may sound formidable, but don’t worry: an explanation follows.)
4 Dirk Geeraerts

LINGUISTIC¬MEANING¬IS¬PERSPECTIVAL
Meaning is not just an objective reflection of the outside world, it is a way of shap-
ing that world. You might say that it construes the world in a particular way, that it
embodies a perspective onto the world. The easiest way to understand the point is
to think of spatial perspectives showing up in linguistic expressions, and the way
in which the same objective situation can be construed linguistically in different
ways. Think of a situation in which you are standing in your back garden and you
want to express where you left your bicycle. You could then both say It’s behind
the house and It’s in front of the house. These would seem to be contradictory
statements, except that they embody different perspectives.
In the first expression, the perspective is determined by the way you look: the
object that is situated in the direction of your gaze is in front of you, but if there
is an obstacle along that direction, the thing is behind that obstacle. In this case,
you’re looking in the direction of your bicycle from the back garden, but the house
blocks the view, and so the bike is behind the house.
In the second expression, however, the point of view is that of the house: a
house has a canonical direction, with a front that is similar to the face of a per-
son. The way a house is facing, then, is determined by its front, and the second
expression takes the point of view of the house rather than the speaker, as if the
house were a person looking in a certain direction. Such multiple perspectiviza-
tions (and not just spatial ones!) are everywhere in the language, and Cognitive
Linguistics attempts to analyze them.

LINGUISTIC¬MEANING¬IS¬DYNAMIC¬AND¬mEXIBLE
Meanings change, and there is a good reason for that: meaning has to do with
shaping our world, but we have to deal with a changing world. New experiences
and changes in our environment require that we adapt our semantic categories
to transformations of the circumstances, and that we leave room for nuances and
slightly deviant cases. For a theory of language, this means that we cannot just
think of language as a more or less rigid and stable structure – a tendency that is
quite outspoken in twentieth century linguistics. If meaning is the hallmark of
linguistic structure, then we should think of those structures as flexible. Again,
we don’t have to look far for an example. Think back to what we said about birds:
there is no single, rigid set of defining features that applies to all and only birds,
but we have a flexible family resemblance structure that is able to deal with mar-
ginal cases.

LINGUISTIC¬MEANING¬IS¬ENCYCLOPEDIC¬AND¬NON AUTONOMOUS
If meaning has to do with the way in which we interact with the world, it is natu-
ral to assume that our whole person is involved. The meaning we construct in
 Introduction: A rough guide to Cognitive Linguistics 5

and through the language is not a separate and independent module of the mind,
but it reflects our overall experience as human beings. Linguistic meaning is not
separate from other forms of knowledge of the world that we have, and in that
sense it is encyclopedic and non-autonomous: it involves knowledge of the world
that is integrated with our other cognitive capacities. There are at least two main
aspects to this broader experiential grounding of linguistic meaning.
First, we are embodied beings, not pure minds. Our organic nature influences
our experience of the world, and this experience is reflected in the language we
use. The behind/in front of example again provides a clear and simple illustration:
the perspectives we use to conceptualize the scene derive from the fact that our
bodies and our gaze have a natural orientation, an orientation that defines what
is in front of us and that we can project onto other entities, like houses.
Second, however, we are not just biological entities: we also have a cultural
and social identity, and our language may reveal that identity, i.e. languages may
embody the historical and cultural experience of groups of speakers (and indi-
viduals). Again, think of birds. The encyclopedic nature of language implies that
we have to take into account the actual familiarity that people have with birds: it
is not just the general definition of bird that counts, but also what we know about
sparrows and penguins and ostriches etc. But these experiences will differ from
culture to culture: the typical, most familiar birds in one culture will be different
from those in another, and that will affect the knowledge people associate with
a category like ‘bird’.

LINGUISTIC¬MEANING¬IS¬BASED¬ON¬USAGE¬AND¬EXPERIENCE
The idea that linguistic meaning is non-autonomously integrated with the rest
of experience is sometimes formulated by saying that meaning is experientially
grounded – rooted in experience. The experiential nature of linguistic knowledge
can be specified in yet another way, by pointing to the importance of language
use for our knowledge of a language.
Note that there is a lot of abstract structure in a language: think for instance
of the pattern Subject – Verb – Direct Object – Indirect Object that you find in a
sentence like Mary sent Peter a message. In many languages, such structures are
not directly observable: what we do observe, i.e. what constitutes the experiential
basis for our knowledge of the language, is merely a succession of words (and even
that is not entirely without problems, but let’s pass over those). So the question
arises: how does this more concrete level of words relate to the abstract level where
you find functional categories like Subject and Direct Object? In more traditional
terms, the question reads: how does the lexicon relate to the syntax?
But if we think of grammatical patterns as having an experiential basis in
concrete, observable strings of words, there is yet another step we have to take:
the ‘observable strings of words’ do not exist in the abstract; they are always part
6 Dirk Geeraerts

of actual utterances and actual conversations. The experience of language is an

experience of actual language use, not of words like you would find them in a
dictionary or sentence patterns like you would find them in a grammar. That is
why we say that Cognitive Linguistics is a usage-based model of grammar: if
we take the experiential nature of grammar seriously, we will have to take the
actual experience of language seriously, and that is experience of actual language
use. Again, from the point of view of mainstream twentieth century linguistics,
that is a fairly revolutionary approach. An existing tradition tended to impose
a distinction between the level of language structure and the level of language
use – in the terms of Ferdinand de Saussure (generally known as the founder of
modern linguistics), between langue and parole. Generally (and specifically in
the tradition of generative grammar), parole would be relatively unimportant: the
structural level would be essential, the usage level epiphenomenal. In a usage-
based model that considers the knowledge of language to be experientially based
in actual speech, that hierarchy of values is obviously rejected.

2. What does the tour include?

You are right, of course: the first exploration of Cognitive Linguistics in the previ-
ous section remains somewhat superficial and abstract. You now have a general
idea of what type of scenery to expect in the Cognitive Linguistics archipelago,
but you would like to get acquainted with the specific islands, i.e. you now know
what the overall perspective of Cognitive Linguistics entails, but you hardly
know how it is put into practice. In this section, we will have a look at the twelve
basic concepts that are introduced by the dozen articles in this collection, and
we will show how these concepts relate to the overall picture that was drawn in
the previous pages.
As a preliminary step, let us observe that each of the characteristics that we
discussed earlier defines a number of specific questions for Cognitive Linguistics.
The perspectival nature of meaning raises questions about the specific mechanisms
of construal present in a language: what kinds of semantic construal, imagery,
conceptual perspectivization do languages implement? The dynamic nature of
meaning raises questions about the process of meaning extension: what are the
mechanisms of semantic flexibility, and how do the various readings of a linguistic
expression relate to each other? The encyclopedic nature of meaning raises ques-
tions about the interdisciplinary links of language to the other cognitive capacities:
to what extent are the cognitive mechanisms at work in natural language shared
by other cognitive systems? And the usage-based nature of meaning raises ques-
tions about the relationship between syntax and lexicon, and the acquisition of
language: what kind of experience do children need to learn a language?
 Introduction: A rough guide to Cognitive Linguistics 7

These questions are illustrated in various ways by the articles in the collection,
but before we can make that explicit, we need to introduce the articles separately,
and say something about the way in which they are grouped together. Roughly,
there are four groups of concepts and articles, corresponding to the four features
that we identified before. The following pages pay specific attention to the logic
behind the basic concepts that we introduce: why is it that these concepts are so
important to Cognitive Linguistics? What you should see, in particular, is how
they turn the fundamental features that we discussed in the previous section, into
a concrete practice of linguistic description. Reading through the following pages
will give you an initial idea of what you can expect in the volume, but of course,
until you get there yourself, you will never really know what it is about.

2.1. The perspectival nature of grammar

The first two concepts, COGNITIVE¬GRAMMAR and GRAMMATICAL¬CONSTRUAL,
illustrate the overall organization of a grammar that focuses on meaning. If con-
ceptual perspectivization is the central function of a grammar, the typical formal
categories of grammatical description (like word classes or inflection) will have to
be reinterpreted from a semantic point of view. In the context of Cognitive Linguis-
tics, two authorities in particular are systematically exploring these phenomena:
Ronald W. Langacker, and Len Talmy. The two initial papers in the volume will
introduce you to the thought of these two major thinkers – towering figures in
the context of Cognitive Linguistics, who have both provided the approach with
some of its basic vocabulary.

COGNITIVE¬GRAMMAR
Cognitive Grammar is the specific name that Langacker uses for his theory of
language. The paper included in the present volume originates from 1986, but is
here reprinted in the form in which it was published in 1990. It specifies a num-
ber of basic features of Cognitive Grammar that are still valid, and that form an
interesting backdrop for the rest of the articles in the present collection. Langacker
starts off with the very idea of a perspectival grammar – although he uses a slightly
different terminology: he talks about grammar as conceptualization and imagery.
He introduces a number of high level general features of grammatical ‘imagery’
(profiling, specificity, scope, salience) and then tackles the key question how to
build a descriptive framework for a grammar that starts from the assumption,
simplistically, that language is meaning and that meaning is conceptualization.
Central to his answer is the idea that a grammar is not built up out of gram-
matical rules on the one hand and a lexicon on the other (the idea that you tra-
ditionally find in generative grammar). Rather, a grammar consists of ‘symbolic
8 Dirk Geeraerts

units’, where a symbolical unit is a conventional pairing of a form and a meaning.

You can obviously think of lexical items here, but symbolic units can be more
abstract than that, like when nouns are claimed to instantiate the abstract notion
‘thing’, and verbs the abstract notion ‘process’. Given that you take the notion of
symbolic unit as the basis for a grammar, there are two questions that immediately
crop up, and Langacker does not fail to address them.
First, what could be the notional, conceptual characterization of abstract enti-
ties like word classes? What do we mean when we say that the meaning of nouns
is ‘thing’ and that of verb is ‘process’? On conceptual grounds, Langacker dis-
tinguishes between a number of basic classes of predications: entities and things
versus relations, and within the relational predicates, stative relations, complex
atemporal relations, and processes. The formal word classes of a language will
typically express a basic type of predication. For instance, while nouns express
the notion of ‘thing’ (a bounded entity in some domain), adjectives will typically
be stative relational predicates.
Second, if you have a grammar with no rules but only symbolic units, how
do you achieve compositionality, i.e. how do you ensure that different symbolic
units may be combined to build larger units, like phrases or sentences? Here,
the trick is to recognize that many predicates have open slots. If, for instance,
the meaning of above is defined in terms of a stative relationship between what
Langacker calls a ‘trajector’ and a ‘landmark’, the trajector and the landmark are
only included schematically, as an open slot, in the meaning of above. Filling
out the slots with other predicates then compositionally yields phrases like the
lamp above the table.
In the course of Langacker’s paper, you will come across a number of concepts
that will play a central role in some of the other chapters included in the present
collection: the schematic network idea (which will come to the fore in Chapter
5), the notion of domain matrix (which will play an essential role in Chapter 8),
and the concept of a construction and a continuum between lexicon and grammar
(which will constitute the focus of Chapter 11 and Chapter 12).

GRAMMATICAL¬CONSTRUAL
Talmy never suggested a specific label for his approach to grammatical description,
but the label grammatical construal captures very well what he is doing: what are
the forms and patterns of construal (in the sense of conceptual perspectivization
through language) that are realized by the grammatical structure of a language?
This adjective grammatical is quite important here: Talmy focuses on the spe-
cific types of conceptual construal that are expressed by those aspects of natural
language that have to do with syntax and morphology, rather than the lexicon. In
Langacker’s article, we noticed a specific interest in the relationship between the
 Introduction: A rough guide to Cognitive Linguistics 9

lexical dimension of the language, and the structural dimension – the syntax and
the morphology. Talmy notes that there are some forms of conceptual construal
that are hardly ever expressed by the grammatical structure (like color), whereas
others (like the distinction between singular and plural) are typically expressed by
syntax and morphology. The bulk of Talmy’s paper, then, provides an overview of
different types of conceptual construal systems that are typical for the structural,
grammatical rather than the lexical subsystem of natural languages.

2.2. The dynamic nature of grammar

If natural language signs are flexible, we will need a model to describe how the
different readings of the expressions relate to each other. Several such models
for the polysemic architecture of expressions have been proposed by Cognitive
Linguistics, and the three concepts in this group describe the most important of
them.

RADIAL¬NETWORK

The radial network model describes a category structure in which a central case
of the category radiates towards novel instances: less central category uses are
extended from the center. The paper featured in this collection, ‘Cognitive topol-
ogy and lexical networks’ by Claudia Brugman and George Lakoff is based on
Brugman’s seminal analysis of the preposition over. The study was seminal not
just in the sense that it popularized the radial network model, but also because
it spawned a whole literature on the analysis of prepositions (more on this in
the Further reading chapter). Brugman suggests the ‘above and across’ reading
of over (as in the plane flew over) as central, and then shows how less central
readings extend from the central case. These can be concrete extensions, as in a
‘coverage’ reading (the board is over the hole), but also metaphorical ones, as in
temporal uses (over a period of time).

PROTOTYPE¬THEORY

Radial categories constitute but one type of a broader set of models that fall under
the heading of prototype theory. For instance, the importance of specific birds
in the category structure of bird (this is a point we drew the attention to before)
belongs in the same set of phenomena as the radial set idea. The paper ‘Pros-
pects and problems of prototype theory’ by Dirk Geeraerts presents a systematic
overview of the different prototype-theoretical phenomena that are mentioned in
the literature. Specific attention is paid to the mutual relations that exist among
these phenomena: it is argued that prototype is itself a prototypically structured
10 Dirk Geeraerts

concept, i.e. that there is no single definition that captures all and only the diverse
forms of ‘prototypicality’ that linguists have been talking about.

SCHEMATIC¬NETWORK
Prototype theory as described in the previous article is a generalization over
the radial network model. But there is another generalization to introduce: the
schematic network model. What this adds to the radial network and prototype
models is the idea that the dynamism of meaning may also involve a shift along
a taxonomical dimension. This may need some explanation. Note that we can
think of birds at different levels of conceptual abstraction (or schematicity, as it
is also called). At one level, we have a prototypical idea of birds as living beings
that have feathers and wings and that can fly. If we stay on this level, we can move
from the central prototype cases (the ones that correspond to the central concept)
to peripheral cases, like birds without feathers and wings (we mentioned penguins
and kiwis before). But there are other levels at which you can think of birds: more
specific ones (as when you think about individual birds, like your great-uncle’s
parrot) and more general ones (like when you group bird species into categories
like ‘fowl’, ‘birds of prey’, ‘water birds’ etc.).
Moving from a more specific to a more general level is called ‘schematiza-
tion’, and the resulting model of readings for an expression is called a schematic
network. The idea of schematic networks is implicit in prototype theory, but it
has been made most explicit by Ronald W. Langacker. The concept plays an
important role in construction grammar (see below), but here, in David Tuggy’s
paper ‘Ambiguity, polysemy and vagueness’, it is applied to a crucial question
about meaning: the relationship between vagueness and polysemy: in a schematic
network, you accept that what is polysemy (different meanings) at one level is
vagueness (a less specified meaning potential) at an other, more schematic level.
In a very clear and graphical way, Tuggy shows how this shift between levels is a
contextual effect: in one situation we may use an expression rather more vaguely,
in another we use it at a more specific, polysemous level.
The consequences for our conception of semantic dynamism are tremendous.
The dynamism of meaning does not just imply that it is easy to add new meanings
to the semantic inventory of an expression, but also that we should not think of
this overall structure of meanings as stable. The semantics of lexical and construc-
tional units is not a bag of meanings, but is a (prototypically and schematically)
structured meaning potential that is sensitive to contextual effects.
 Introduction: A rough guide to Cognitive Linguistics 11

2.3. The non-autonomy of grammar

If meaning is non-autonomous and encyclopedic, it is important to investigate
the way in which different types of experience interact with each other. How
for instance does our bodily experience of a sensory or motor kind relate to our
more abstract thought, and are there any conceptual mechanisms that cut across
the sensorimotor and the abstract mode of human knowledge? The four papers
brought together in this group show how Cognitive Linguistics deals with the
encyclopedic entrenchment of linguistic meaning, from both perspectives: what
is the role of general cognitive mechanisms, and how do specific domains of
experience interact?
Two of the four concepts that will be introduced are not specific for Cognitive
Linguistics: metaphor and metonymy are traditional concepts in natural language
semantics. Cognitive Linguistics has however brought new perspectives to the
study of both metaphor and metonymy, and we will see in what sense. The other
two concepts were newly introduced by Cognitive Linguistics. Incidentally, the fact
that metaphor and metonymy are traditionally known as mechanisms of semantic
change, makes clear that there is a certain degree of overlap between the present
group of concepts and the previous one: some of the concepts mentioned here
may also be seen as illustrations of the dynamic nature of meaning.

CONCEPTUAL¬METAPHOR

Conceptual metaphor is probably the best known aspect of Cognitive Linguistics:

if you’ve heard only vaguely about Cognitive Linguistics, conceptual metaphor
is likely to be the notion that you’ve come across. You will have learnt by now
that there is much more to Cognitive Linguistics, but still, Conceptual Metaphor
Theory occupies a major place in the cognitive linguistic research program. Con-
ceptual Metaphor Theory rests on the recognition that a given metaphor need not
be restricted to a single lexical item, but may generalize over different expressions.
Such general patterns may then be summarized in an overall statement like LOVE¬
IS¬WAR, a pattern that ranges over expressions like the following:

He is known for his many rapid conquests. She fought for him, but his mistress
won out. He fled from her advances. She pursued him relentlessly. He is slowly
gaining ground with her. He won her hand in marriage. He overpowered her. She
is besieged by suitors. He has to fend them off. He enlisted the aid of her friends.
He made an ally of her mother. Theirs is a misalliance if I’ve ever seen one.

This way of thinking about metaphor was introduced in George Lakoff’s and
Mark Johnson’s book Metaphors we live by of 1980, a book that has achieved
something of a bestseller status. The article included here, Lakoff’s ‘The contem-
12 Dirk Geeraerts

porary theory of metaphors’ gives a systematic overview of the theoretical and

practical features of Conceptual Metaphor Theory.
You will notice how the two aspects of non-autonomy that were mentioned
earlier show up in conceptual metaphor theory. First, metaphor is treated as a
general cognitive mechanism, not as a specifically linguistic one that works on
the level of individual expressions. Second, metaphor involves the interaction
between different domains of experience: a source domain (in the example, war)
and a target domain (love). This notion of domain will turn out to be crucially
important for the other concepts in this group as well.

IMAGE¬SCHEMA
An image schema is a regular pattern that recurs as a source domain (or a structur-
ing part of a source domain) for different target domains. Typical image schema’s
include containment, path, scales, verticality, and center-periphery. The recur-
rence of image schemas may be illustrated by a closer look at the containment
schema. It occurs in conceptual metaphors in which containment is the source
domain for widely diverse target domains like the visual field (in sight, out of
sight, go out of view, inside someone’s field of vision), time (in two hours, he’s
into the first year of his retirement, do something in a short period), difficulties
(get yourself into difficulties, we’re in this together, how do we get out of this, to
be in a mess), obligations (what are you getting into, no way out, can he get out
of it), and the self as contained in the body (withdraw into oneself, a young man
in an old man’s body, there’s an insecure person inside).
Characteristically, image schemas involve some form of sensory or motor
experience, like a spatial configuration in the case of containment. In that sense,
the appearance of image schemas in metaphors is typical for the encyclopedic,
non-autonomous nature of meaning: prelinguistic domains, like the sensorimo-
tor or spatial ones are mapped onto more abstract domains, providing them with
structure. The notion of image schema, like the notion of conceptual metaphor,
was introduced into Cognitive Linguistics by George Lakoff in his collaboration
with the philosopher Mark Johnson. The paper included here, ‘The cognitive
psychological reality of image schemas and their transformations’ by Ray Gibbs
and Herb Colston, examines the psychological reality of image schemas: how can
you prove, by means of experimental methods, that image schemas do indeed
have a psychological reality?
In this sense, the paper is not just important as an illustration of the notion of
image schema, but also for methodological reasons. The particular theoretical
perspective of Cognitive Linguistics has a number of far-reaching methodologi-
cal consequences. For one thing, treating meaning as a mental phenomenon and
focusing on language as a cognitive tool implies that a rapprochement with the
methodology of psychological research is obvious: experimental methods should
 Introduction: A rough guide to Cognitive Linguistics 13

bear fruit in Cognitive Linguistics just like they do in psychology. For another,
the idea that grammar is usage-based implies that the analysis of actual usage
data (as in corpora, specifically) should play an important role in Cognitive Lin-
guistics. The paper by Gibbs and Colston is an example of such an experimental
approach. An example of a corpus-based methodology is presented in Tomasello’s
paper in this volume.
However, neither the use of an experimental method nor the use of corpus
data is as yet a dominant methodology in Cognitive Linguistics. They constitute
emerging tendencies that are likely to gain in importance in the course of the
following years, but a lot of the work done in Cognitive Linguistics is still based
on a more traditional analytic methodology.

METONYMY
There is yet another way in which thinking in terms of domains plays a role in
Cognitive Linguistics, viz. in the analysis of metonymy. In the tradition of lexical
semantics, metaphor and metonymy are distinguished on the basis of the type of
semantic association they involve. Metaphor is supposed to be based on similarity
(if love is war, it is like war in a number of respects), whereas metonymy is said
to be based on contiguity – a somewhat vague notion that could be clarified in
terms of ‘actual proximity or association’. For instance, when you fill up your car,
you don’t fill the entire vehicle with fuel, but only the gas tank. The name of the
whole comes to stand for the part, and part and whole are associated in reality.
Now, metonymy research in Cognitive Linguistics received an important impe-
tus from the recognition that metonymy could receive a definition that is nicely
complementary to that of metaphor. If metaphor is seen as a mapping from one
domain to the other, metonymy can be seen as a mapping within a single domain.
The shift from whole to part in car is a shift within the physical, spatial domain.
This view on the relationship between metaphor and metonymy was already made
in Lakoff and Johnson’s Metaphors we live by, but the article ‘The role of domains
in the interpretation of metaphors and metonymies’ by William Croft adds an
innovative perspective. The relevant shift, Croft argues, is not necessarily one
within a single domain, but it may be a shift within a domain matrix.
The domain matrix is a notion introduced by Ronald W. Langacker: it cap-
tures the idea that a concept may be simultaneously defined in various domains.
For instance, Shakespeare is not only defined as a physical person, but also in
the literary domain, as an author. So, when you say that you have read the whole
of Shakespeare, you metonymically mean the entirety of his literary production,
rather than the person. What Croft suggests, then, is to define metonymy overall
in terms of such a domain matrix.
14 Dirk Geeraerts

MENTAL¬SPACES
If metaphor is analyzed as a mapping from one domain to another, the ques-
tion arises how such mappings take place: how does the structure of the source
domain get mapped onto the target domain? The notion of conceptual integration
developed by Gilles Fauconnier, and represented here by the paper ‘Conceptual
integration networks’ by Gilles Fauconnier and Mark Turner, provides a descrip-
tive framework to answer that question. It distinguishes between four spaces: a
source input space, a target input space, a blend between both, and a so-called
generic space. For instance (to use an example first described by Seana Coulson),
you can think of ‘trashcan basketball’ as a game in which you throw crumpled
pieces of paper into a trashcan, as you might do in an office environment or in
a student dorm. The game of basketball is one input space, and the office or the
dorm situation the other. The mapping between the two spaces associates the ball
with the piece of paper, the basket with the trashcan, the players with the students
or the office people, and further elaborations are possible. This mapping creates a
blended space, and the relevant features of the blend are not just directly derived
from the original input spaces. On the contrary, you may find emergent structure
that is specific to the blended space: the fact that the trashcan would normally be
placed on the ground, in contrast with a basketball ring, would certainly influence
the way the game is played. The fourth type of space, the generic space, contains
the common structure of the input spaces; in this case, it would be the space of
someone throwing an object into a container.
The description of the four spaces may also explain some of the alternative
names that the conceptual integration approach is known by: the blending or
the mental spaces approach. As mentioned, the conceptual integration approach
clearly links up with the analysis of metaphor as mapping across domains: one
might say that the trashcan example elaborates the metaphor ‘a trashcan is a
basketball ring’. However, the blending analysis is more general than the study
of metaphor. Conceptual integration has proved to be useful in a wide variety of
phenomena, many of which are not even remotely associated with metaphorical
processes. Counterfactuals are a case in point. If Beethoven were alive, he would
use a synthesizer creates a blended space between the present-day musical situ-
ation and the historical space of Beethoven as an innovative composer, but you
cannot really say that the conceptual process is a metaphorical one.

2.4. The experiential grounding of grammar

You will remember that the experiential nature of language raised the question
of the relationship between lexicon and syntax. Cognitive Linguistics provides a
specific answer to that question that links up with what we said about prototype
 Introduction: A rough guide to Cognitive Linguistics 15

theory and schematic networks: we can think of a grammar as a schematic network

with abstract patterns at the schematic level, and the lexicalized instantiations
of those patterns (the words and strings of words that fill the patterns) at a more
specific level. From the point of view of mainstream twentieth century linguistics,
this is a bit of a strange idea. If you assume (as generative grammar so vehemently
stressed) that grammar is a set of rules, the lexical items instantiating those roles
are not all that important: you basically need an inventory of items, but the real
work is done by the rules.
However, Ronald W. Langacker has pointed out that there is a fallacy here: the
so-called rule/list fallacy, i.e. the idea that what can be handled by rules should
not be listed. If you start from that assumption, you get a strict separation between
lexicon and syntax. But if you think in terms of schematic networks, such a sepa-
ration is not necessary at all: you can both describe abstract patterns and their
concrete lexicalizations. In fact, you have to, because there are prototypicality
effects that you may need to describe. An indirect object construction of the type
Subject – Verb – Direct Object – Indirect Object is typically filled by verbs like
give or tell, and less typically so by verbs like envy. It is part of our grammatical
knowledge that we recognize those typicalities (just like we recognize blackbirds
as more typical birds than ostriches).
The specific form in which this idea is realized in Cognitive Linguistics, is
in the form of construction grammars. The plural is deliberate here: since there
are various forms of construction grammar, Construction Grammar is a family
of theories rather than a single well-defined approach. The three papers in this
group all involve Construction Grammar, but it is the paper by Goldberg (Chapter
11, the second one in the group) that introduces the approach most directly. The
first paper in the group presents frame semantics, which is one of the important
sources for Construction Grammar. The final paper show how the principles of
Construction Grammar can be applied in language acquisition research.

FRAME¬SEMANTICS

Frame semantics is the specific approach to natural language semantics devel-

oped by Charles Fillmore. The article included here, ‘Frame semantics’, sets out
the basics of the theory. One essential starting-point is the idea that one cannot
understand the meaning of a word (or a linguistic expression in general) without
access to all the encyclopedic knowledge that relates to that word. This obviously
ties in with the non-autonomous nature of natural language semantics: that mean-
ing in natural language is not separated from other forms of knowledge implies
that it is not very useful to maintain a strict separation between world knowledge
and knowledge of linguistic meaning. While this recognition would be shared
by all forms of semantics in Cognitive Linguistics, the individual identity of
16 Dirk Geeraerts

frame semantics involves the specific structures of encyclopedic knowledge that

it invokes. Basically, these ‘frames’ are things happening and occurring together
in reality. For example, in order to understand the word sell, you need to have
world knowledge about the situation of commercial transfer. This comprises, apart
from the act of selling, a person who sells, a person who buys, goods to be sold,
money or another form of payment, and so on.
A semantic frame of this type is a coherent structure of related concepts where
the relations have to do with the way the concepts co-occur in real world situ-
ations. Knowledge of the frame is necessary for an adequate knowledge of the
words referring to the concepts in the frame: a word activates the frame, highlights
individual concepts within the frame, and often determines a certain perspective
in which the frame is viewed. In the standard commercial transaction example,
for instance, sell construes the situation from the perspective of the seller and buy
from the perspective of the buyer.
Although frame semantics was originally applied predominantly to the seman-
tic description of words, there is a close relationship with Construction Grammar.
Words like sell come with their own set of constructions (like sell something to
someone or sell something for a certain price), and the different constructions
reflect different ways in which the frame can be highlighted. In this way, frame
semantics can be integrated with Construction Grammar as one way of specify-
ing the semantics of constructions. Overall, we can now see that frame semantics
occupies a transitional position in our grouping of concepts. On the one hand, if
we focus on the way in which it uses structured encyclopedic knowledge as the
background for the description of meanings in natural language, it belongs with
the previous group of papers: it describes one of the ways in which conceptual
knowledge of an encyclopedic (i.e. not specifically linguistic) nature is structured.
On the other hand, if we concentrate on the input that frame semantics provides
for the description of construction types, it links up with the present group of
articles.

CONSTRUCTION¬GRAMMAR

Simplistically, a grammatical construction is any string of words or morphemes

exhibiting a coherent pattern, whether it be an entire sentence or a clause or a
phrase (like a noun phrase or a verb phrase) or a complex lexeme (like a phrasal
verb). And of course, the abstract pattern itself may also be called a ‘construction’.
In classical contemporary grammars, an indirect object pattern of the type Sub-
ject – Predicate – Indirect Object – Direct Object would be considered a derived
structure, built up from the functional classes Subject, Predicate etc. That is to
say, the rules of the grammar would be defined in such a way that a grammatical
pattern of the form Subject – Predicate – Indirect Object – Direct Object could
 Introduction: A rough guide to Cognitive Linguistics 17

be assembled on the basis of the relevant functional building blocks. In Cognitive

Linguistics, a pattern of this type is considered to be non-derived, i.e. is taken to be
a sign of the language, or, if you wish, an (abstract) expression in its own right.
As you can read in Adele Goldberg’s paper ‘The inherent semantics of argument
structure: The case of the English ditransitive construction’, there are a number
of advantages to such an approach, two of which may be mentioned here, to give
you a first idea of why thinking in terms of constructions may be interesting.
First, if a construction is treated as an entity in its own right, it is possible that
the whole has characteristics that cannot be straightforwardly derived from the
constituent components. This is a property known as non-compositionality: the
meaning of the whole is not necessarily a compositional function of the meaning
of the parts. Second, if any construction is a distinct element of the inventory of
linguistic signs for a given language, you can treat constructions like any other
category – according to the usual practice of Cognitive Linguistics. That is to say,
it will then be quite normal to describe not just the meaning and the form of the
category, but also its salient members. In the case of constructions, we may then
think primarily of the lexical elements that can fill the slots of the construction
(like the verbs that occupy the predicate role in the indirect object construction).
Or in other words: describing grammar as a schematic network with lexical ele-
ments at the bottom and more abstract patterns higher up in the network is com-
pletely congenial to a construction grammar approach.

USAGE BASED¬LINGUISTICS

Defining Cognitive Linguistics as a usage-based model of language has a number

of consequences, like the straightforward methodological conclusion that cogni-
tive linguists will have to invest in the analysis of real language use if they are to
live up to their self-declared status. You would expect a lot of corpus research in
Cognitive Linguistics, then, but to be honest, this is a trend that is clearly emerg-
ing, but that has not yet gained as prominent a status as one would expect.
Using corpora of observed speech is natural in language acquisition research,
though: if you wish to study how children acquire their language, you will obvi-
ously want to observe and analyze their developing language. Also, language
acquisition is a domain par excellence to test a usage-based model of language
that believes that our experience of actual speech determines how we come by
more abstract patterns. In fact, if our more schematic knowledge is based on our
more concrete knowledge of lexical instantiations of such patterns, you should
be able to observe ‘lexical bootstrapping’ effects: the specific words in which we
begin to learn certain constructions, determine how we learn the construction.
That is exactly what Michael Tomasello investigates in his paper ‘First Steps
toward a Usage-based Theory of Language Acquisition’, which is part of a long-
18 Dirk Geeraerts

term research effort to develop a theory of language acquisition that ties in with
Cognitive Linguistics and Construction Grammar.
If you are familiar with the history of contemporary linguistics, you will
appreciate how important such an attempt is. The generative grammar idea that
language is a separate module of the mind is very much based on an argument
from language acquisition: if we do not assume that language is genetically wired
in, the argument goes, we couldn’t explain at all how the acquisition of language
proceeds so quickly as it actually does. This is particularly the case, the argu-
ment continues, because the input children get (the language they are exposed
to) is not sufficient to explain how they could learn all the intricacies of natural
language syntax. This is the ‘poverty of stimulus’ argument. Obviously, if Toma-
sello succeeds in his attempt to explain how children can learn language through
abstraction from the actual input they get, a central tenet of generative grammar
will be overturned.

2.5. A conceptual map

Let us summarize. What we have shown in this (somewhat detailed) introduction
to the tour of twelve papers is how they introduce, discuss, illustrate concepts
that follow logically from the central characteristics of Cognitive Linguistics
that we learned about in the previous section. It will not come as a surprise to
you that these twelve concepts are far from being the only relevant or interesting
ones that have been developed in the context of Cognitive Linguistics – but they
are certainly among the most basic ones. When you read the individual papers,
it may be wise to refer back once in a while to the characterizations that you find
in the foregoing pages: they will help you not to lose track and to interpret the
conspicuous features of the papers.
In Figure 1, the relations between the various papers are graphically rep-
resented. The fundamental features of Cognitive Linguistics (the perspectival,
dynamic, non-autonomous, experiential nature of natural language) are mentioned
in the corners of the figure. Intermediate between these cornerstones and the
twelve central concepts, you will find six fields of research that emerged in the
foregoing discussion as the logical link between the fundamental features and the
central concepts: the conceptual characterization of the grammar, the search for
models of polysemy, the analysis of mechanisms of polysemy, the importance of
thinking in terms of domains of experience, the relationship between grammar
and lexicon, and the relationship between structure and usage. The arrows in the
figure indicate how each more specific concept is motivated by a more general one
for instance: looking at mechanisms of categorial polysemy is one way of getting
a grip on the dynamic nature of linguistic meaning, and in a next step, conceptual
metaphor is one of the specific mechanisms of categorial polysemy.
 Introduction: A rough guide to Cognitive Linguistics 19

Figure 1. A conceptual map of Cognitive Linguistics

20 Dirk Geeraerts

Adding the intermediate level with the six fields of research brings in extra subtlety
in the overview. We can now make clear, for instance, that Langacker’s paper
with which the collection opens, not only illustrates the perspectival nature of
grammar, but also deals with the relationship between lexical structure and mor-
pho-syntax. In the same way, the figure illustrates how the experiential nature
of the grammar does not only involve questions about the acquisition and the
architecture of the grammar, but also links up with thinking in terms of domains
and their mutual relations.

3. Where do you go next?

Let us assume that, after roaming the present introductory volume, you really
like the look and feel of Cognitive Linguistics. It’s a safe assumption, in fact: you
are bound to be drawn in by an intellectual climate that is both hospitable and
inspiring, open-minded and exciting, wide-ranging and innovative. But where do
you go after the initial tour d’horizon that has won your heart? Let’s go over a
few trajectories that might cater to your personal interests.
A first thing to do would be to complete your initiation by reading the com-
panion volume to the present collection. In several ways, Cognitive Linguistics:
Current Applications and Future Perspectives (edited by Gitte Kristiansen, Michel
Achard, René Dirven and Francisco J. Ruiz de Mendoza) is deliberately comple-
mentary to the book that you are holding in your hands: it does not focus on Cog-
nitive Linguistics basics, but rather describes the contemporary state of affairs in
the main fields of application of Cognitive Linguistics. Also, it does not consist
of existing papers, but only contains newly written articles that have the explicit
purpose of presenting current discussions and domains. Taken together, the two
volumes will thoroughly familiarize you with Cognitive Linguistics, way beyond
the fragmented and uncertain knowledge that an incidental visit would impart.
From that point on, you may want to go beyond the level of introductions,
and you may, with the acquired confidence of the experienced traveler, do some
journeying on your own. However, in case your self-assurance has not reached an
optimal level and you would want to boost it with an additional round of intro-
ductory reading, there are some good book-length introductions to Cognitive
Linguistics that may help you. Here’s a list of commendable texts.

• Friedrich Ungerer and Hans-Jörg Schmid. 1996. An Introduction to Cogni-

tive Linguistics. London/New York: Longman.
• Violi, Patrizia. 2001. Meaning and Experience. Bloomington: Indiana Uni-
versity Press.
• John R. Taylor. 2003. Linguistic Categorization. Third Edition. Oxford:
Oxford University Press.
 Introduction: A rough guide to Cognitive Linguistics 21

• René Dirven and Marjolijn Verspoor. 2004. Cognitive Exploration of Lan-

guage and Linguistics. Second Revised Edition. Amsterdam, Philadelphia:
John Benjamins.
• William Croft and D. Alan Cruse. 2004. Cognitive Linguistics. Cambridge:
Cambridge University Press.
• Vyvian Evans and Melanie Green. 2006. Cognitive Linguistics. An Intro-
duction. Edinburgh: Edinburgh University Press.

Once you are ready to leave the introductory level, there are basically two things
to do: deepen your understanding of the existing body of work in Cognitive Lin-
guistics, and keep aware of current developments. With regard to the former, a
first set of routes to explore is provided by the present volume itself: accompany-
ing each of the chapters included in the collection, you find a broadly conceived
set of suggestions for further reading in the Epilogue. They will direct you to
elaborations and discussions of the basic concepts that are presented here – and
to some of the less basic (but no less important) concepts developed in the field
of Cognitive Linguistics. Alternatively (or in parallel), you could have a look at
the forthcoming Handbook of Cognitive Linguistics, edited by Dirk Geeraerts
and Hubert Cuyckens and published by Oxford University Press. It’s a collection
of some fifty commissioned articles that each offer an in-depth treatment of one
of the manifold aspects of Cognitive Linguistics.
To keep in touch with new work, you would certainly want to be aware of
new publications. So, what are the journals and the book series that you need to
keep an eye on? Note that a lot of studies in Cognitive Linguistics are now being
published in general publication channels and by a wide variety of international
publishers. Here, only journals and book series that are specifically dedicated to
Cognitive Linguistics are mentioned.
Two journals need to be cited: Cognitive Linguistics, and the Annual Review
of Cognitive Linguistics. The former, published by Mouton De Gruyter, is the
official journal of the International Cognitive Linguistics Association (more about
which in the next section). The journal was founded by Dirk Geeraerts in 1990.
Consecutive editors-in-chief so far were Arie Verhagen and Adele Goldberg.
The latter journal, published by the John Benjamins Publishing Company under
the auspices of the Spanish Cognitive Linguistics Association, first appeared in
2003. It is led by Francisco Ruiz de Mendoza. The journal Cognitive Linguistics
is not only the most reputable journal in the field, it also comes with a consider-
able bonus. A subscription to Cognitive Linguistics includes a copy of the digital
Bibliography of Cognitive Linguistics – and indispensable bibliographical tool
compiled through the relentless efforts René Dirven (undoubtedly the major
organizational force behind the entire Cognitive Linguistics enterprise). The
bibliography now covers 7000 publications (no, you won’t have to read them all
22 Dirk Geeraerts

to be recognized as an accomplished cognitive linguist), and it will be regularly

updated in the following years.
Five book series specifically dedicated to work in Cognitive Linguistics need
to be mentioned. Cognitive Linguistic Research or CLR is the oldest and most
complete series. Published by Mouton de Gruyter of Berlin, it was launched at the
same time as the journal Cognitive Linguistics, and now numbers over thirty vol-
umes. Recently, Cognitive Linguistic Research (with Dirk Geeraerts as managing
editor) has received a companion series in the form of Applications of Cognitive
Linguistics or ACL (managed by Gitte Kristiansen) – a series that will focus on
the descriptive applications of Cognitive Linguistics, while Cognitive Linguistic
Research harbors the more theoretically relevant studies. The other main publisher
for Cognitive Linguistics, the John Benjamins Publishing Company of Amster-
dam, features three specifically relevant (but as yet less extensive) series. While
Human Cognitive Processing (edited by Marcelo Dascal, Raymond W. Gibbs
and Jan Nuyts) has a broad cognitive orientation, the two other series are more
specific: Cognitive Linguistics in Practice (edited by Günter Radden) covers the
field from a textbook oriented perspective, and Constructional Approaches to
Language (edited by Mirjam Fried and Jan-Ola Östman) specifically deals with
Construction Grammar.
If you are interested in following what is going on in Cognitive Linguistics,
note that the book series are not just monograph series. A lot of what is being
published in the series consists of collective volumes with thematically intercon-
nected articles. In many cases, these constitute selections of papers presented at
Cognitive Linguistics conferences. Because quite a lot of relevant work is being
published in such collective volumes, you will profit from keeping an eye on them
once you’ve become a Cognitive Linguistics aficionado.
When you’ve reached this stage, you will be ready to take a step into the world
and take part in some real life Cognitive Linguistics activities. Where would you
go? All self-respecting cities and countries have their own festivals and fiesta,
and becoming part of the crowd involves participating in the celebrations. In
Cognitive Linguistics, the main community event is without any doubt the bi-
annual ICLC or Cognitive Linguistics Conference. The first ICLC took place in
1989 in Duisburg, Germany. It was one of the so-called LAUD symposia (where
LAUD stands for Linguistic Agency of the University of Duisburg) that had been
organized by René Dirven since 1977 and where some of the world’s most distin-
guished linguists were invited to present their work. The Duisburg conference was
of crucial importance for the institutionalization and the international expansion
of Cognitive Linguistics: it was there and then that the International Cognitive
Linguistics Association or ICLA was founded (the conference was accordingly
rebaptized as the First International Cognitive Linguistics Conference), that plans
were made to launch the journal Cognitive Linguistics, and that the monograph
 Introduction: A rough guide to Cognitive Linguistics 23

series Cognitive Linguistics Research was announced. Cognitive Linguistics as

an intellectual movement is too self-critical to recognize any historical sites or
places of pilgrimage, but if ever a commemorative plate were to be considered,
Duisburg would be a likely candidate.
The following ICLC conferences were consecutively held in Santa Cruz, US
(1991, organized by Gene Casad), Leuven, Belgium (1993, organized by Dirk
Geeraerts), Albuquerque, US (1995, organized by Sherman Wilcox), Amsterdam,
The Netherlands (1997, organized by Theo Jansen and Gisela Redeker), Stockholm,
Sweden (1999, organized by Erling Wande), Santa Barbara, US (2001, organized
by Ronald W. Langacker), Logroño, Spain (2003, organized by Francisco J. Ruiz
De Mendoza), and Seoul, South Korea (2005, organized by Hyon-Sook Shin).
The tenth ICLC in 2007, organized by Elzbieta Tabakowska, has its venue at the
Jagiellonian University of Krakow in Poland.
Next to the ICLC’s, there are a number of local events that you may consider
attending. The LAUD conferences are now being organized bi-annually in Landau,
Germany, in the year between the ICLC conferences. That holds for a number
of other regular meetings, like the US-based Cognitive Structure, Discourse and
Language conference, which reached its eighth edition in 2006; or the International
Conference on Construction Grammar, which had its fourth edition in 2006.
An important part in the organization of workshops and symposia in Cognitive
Linguistics is currently played by the ICLA affiliates. These are ICLA branches
defined by region or country (and occasionally by language). The first one to be
founded was the Spanish Cognitive Linguistics Association, whose affiliation
was formally approved at the 1999 ICLC. The year 2001 saw the affiliation of
the Finnish, the Polish, and the Slavic Cognitive Linguistics Associations. Fur-
ther affiliates include the Russian Association of Cognitive Linguists (2004), the
German Cognitive Linguistics Association (2005), the Discourse and Cognitive
Linguistics Association of Korea (2005), the Association Française de Linguis-
tique Cognitive (2005), the Japanese Cognitive Linguistics Association (2005),
the Conceptual Structure, Discourse and Language Association (2005), and the
UK Cognitive Linguistics Association (2006). Further local branches are likely
to emerge, and all of them are likely to organize regular meetings. If you are
interested in following what is on the agenda, you may want to consult the ICLA
website: it contains a calendar of Cognitive Linguistics events, and supplies links
to the individual websites of the ICLA affiliates. This is the ICLA address: http://
[Link]/
Meetings, lectures, workshops, symposia, and conferences are also announced
on Cogling, a mailing list for disseminating ICLA news, queries, and discus-
sions of interest to cognitive linguists. Again, details about subscribing may be
found on the ICLA website. Incidentally, the website also offers an overview of
24 Dirk Geeraerts

the courses or programs in Cognitive Linguistics that are offered at many places
around the globe.
Once you get to one of the conferences, who would you be likely and/or eager
to meet? Thinking in terms of people, the key figures of Cognitive Linguistics
are George Lakoff, Ronald W. Langacker, and Leonard Talmy. Round this core
of founding fathers, who originated Cognitive Linguistics in the late 1970s and
the early 1980s, two chronologically widening circles of cognitive linguists may
be discerned. (The lists that follow are obviously indicative only: they are not
meant to exclude anyone, but only to give you an idea of the different ‘genera-
tions’ of cognitive linguists.)
A first wave, coming to the fore in the second half of the 1980s and the beginning
of the 1990s, consists of the early collaborators and colleagues of the key figures,
together with a first generation of students. Names that come to mind include those
of Gilles Fauconnier, Eve Sweetser, Mark Johnson, Mark Turner, Raymond W.
Gibbs, William Croft, Adele Goldberg, Dave Tuggy, Gene Casad, Laura Janda,
Suzanne Kemmer, Sally Rice, Ricardo Maldonado, Karen Van Hoek, Geoff Nathan,
Margaret Winters, Sherman and Phyllis Wilcox, Margaret Freeman.
Simultaneously, a number of people in mostly Western and Central Europe took
up the ideas of Cognitive Linguistics and contributed to their international dissemi-
nation. Names include those of René Dirven (to repeat: his instrumental role in the
expansion of Cognitive Linguistics can hardly be overestimated), Brygida Rudzka-
Ostyn, John Taylor, Zoltan Kövecses, Chris Sinha, Brigitte Nerlich, Arie Verhagen,
Barbara Lewandowska-Tomaszczyk, Elzbieta Tabakowska, Peter Harder, Günter
Radden, Susanne Niemeier, Martin Pütz, Hans-Jörg Schmid, Hubert Cuyckens
and the author of the present introduction.
The mid 1990s and later witnessed a second wave of expansion, with second
generation students and a further geographical spread directed largely towards Asia
and the rest of Europe. Names include those of Alan Cienki, Michel Achard, Joe
Grady, Tim Rohrer, Seana Coulson, Todd Oakley, Gary Palmer, Jose M. Garcia-
Miguel, Antonio Barcelona, Francisco Ruiz de Mendoza, Carlos Inchaurralde,
Andrej Kibrik, Ekaterina Rakhilina, Michael Tomasello, Ted Sanders, Wilbert
Spooren, Gerard Steen, Stefan Grondelaers, Stefan Gries, Anatol Stefanowitsch,
Yo Matsumoto.
In addition, you might profit from the occasion to rub elbows with people
who would perhaps not describe themselves unreservedly as cognitive linguists
(coming as they do from other theoretical families or other disciplines, or simply
because they like their independence), but who would show up at the Cognitive
Linguistics conferences because they have relevant things to say: linguists like
Charles Fillmore, Joan Bybee, Elizabeth Traugott, Östen Dahl, Jan Nuyts, or
psychologists like Melissa Bowerman, Dedre Gentner, and Dan Slobin.
 Introduction: A rough guide to Cognitive Linguistics 25

4. Why would you want to come back?

So now you know your way around in Cognitive Linguistics. You can walk the
walk and talk the talk, and there’s no way that you’d be exposed as a novice. But
why would you be coming back? What would be a good reason to become a per-
manent resident? An obvious but relatively superficial motivation would be the
diversity of the panorama: there’s a lot to be found in the Cognitive Linguistics
archipelago, and the framework is not so strict as to stifle creativity. It’s a lively,
colorful, varied environment, and you’re likely to find some corner of special
significance to you, where you can do your thing and meet people with similar
interests. But beyond that? What would be the long-term importance of Cogni-
tive Linguistics?
Let us try to take a bird’s eye view of the history of linguistics, and see exactly
where Cognitive Linguistics fits in, and why it could be important for the future
of linguistics. Agreed, you can only achieve this sort of extreme synthesis if you
allow for massive simplification: let us try to keep that in mind as a proviso when
we do the exercise.
The development of linguistics in the twentieth century, then, is character-
ized by a succession of two dominant approaches: the structuralist one, and the
generativist one. Currently, in the first decade of the 21st century, the generativist
paradigm is no longer the principal framework, but there clearly is no new central
approach yet. If one looks at Cognitive Linguistics from this perspective, there are
indications that Cognitive Linguistics combines a number of tendencies that may
also be found in other contemporary developments in theoretical linguistics – viz.
in the broad range of functionalist approaches to linguistics. By combining these
tendencies, Cognitive Linguistics taps directly into the undercurrent of contem-
porary developments. Specifically, if we recognize that decontextualization is a
fundamental underlying characteristic of the development of grammatical theory
in twentieth century linguistics, Cognitive Linguistics strongly embodies the
recontextualizing tendency that is shared by most functionalist approaches.
The logic behind the decontextualization of twentieth century grammar may
be grasped if we take our starting-point in De Saussure, the founding father of the
structuralist approach. The Saussurean distinction between langue (the language
system) and parole (the use of the language system in actual usage) creates an
internally divided grammar, a conception of language with, so to speak, a hole
in the middle. On the one hand, langue is defined by De Saussure as a social
system, a set of collective conventions, a common code shared by a community.
On the other hand, parole is an individual, psychological activity that consists of
producing specific combinations from the elements that are present in the code.
When langue and parole are defined in this way, there is a gap between both: what
is the factor that bridges the distance between the social and the psychological,
26 Dirk Geeraerts

between the community and the individual, between the system and the applica-
tion of the system, between the code and the actual use of the code?
The father of generative grammar, Noam Chomsky, provided an answer by
introducing a distinction between competence and performance: the missing link
between social code and individual usage is the individual’s knowledge of the
code. Performance, in the Chomskyan sense, is basically equivalent with Sau-
ssurean parole, but competence interiorizes the notion of the linguistic system.
Competence is the internal grammar of the language user, the knowledge that
the language user has of the linguistic system and that he or she puts to use in
actual performance.
Remarkably, however, Chomsky introduces a new gap into the system. Rather
than the threefold classification that one might expect, he restricts his conception
of language to a new dichotomy: the social aspects of language are largely ignored.
In comparison with a ternary distinction distinguishing between langue, com-
petence, and parole/performance (between social system, individual knowledge
of the system, and individual use of the system), the binary distinction between
competence and performance creates a new empty slot, leaving the social aspects
of language largely out of sight.
This apparent lack of interest for language as a social sign system links up
logically with the Chomskyan emphasis on the genetic nature of natural language.
Where, in fact, does the individual knowledge of the language come from? If the
source of linguistic knowledge is not social, what else can it be than an innate and
universal endowment? If the language is not learned through acculturation in a
linguistic community, what other source could there be for linguistic knowledge
except genetics?
Further restrictions follow. Meanings constitute the variable, contextual, cul-
tural aspects of language par excellence. Because social interaction, the exchange
of ideas, changing conceptions of the world are primarily reflected in the meaning
of linguistic expressions, meanings are less interesting from a genetic point of
view. Also, if the lexicon is the main repository of linguistically encoded meaning,
studying the lexicon is of secondary importance: the focus will fall on the abstract
syntactic patterns. And finally, if linguistics focuses on formal rule systems, the
application of the rule systems in actual usage is relatively uninteresting. If the
rules define the grammar, it is hard to see what added value could be derived from
studying the way in which the rules are actually put to use. The study of perfor-
mance, in other words, is just as secondary as research into the lexicon.
In short, generative grammar led to a severe decontextualization of the gram-
mar, separating the autonomous grammatical module from different forms of con-
text: through the basic Chomskyan shift from langue to competence, linguistics is
separated from the social context of language as a social code; through the focus
on the genetic aspects of the language, linguistics is separated from the cognitive
 Introduction: A rough guide to Cognitive Linguistics 27

context of lived individual experience; through the focus on formal rule systems,
linguistics is separated from the situational context of actual language use.
Compared to this, Cognitive Linguistics is very much a recontextualizing
approach. First, it is an outspoken attempt to give meaning a central position in
the architecture of the grammar. Second, in contrast with formal semantics, the
conception of meaning that lies at the basis of this approach is not restricted to
a referential, truth-functional type of meaning – the type of meaning that you
could express in logical terms. Linguistic structures are thought to express con-
ceptualizations, and conceptualization goes further than mere reference. As we
have seen, it involves imagery in the broadest sense of the word: ways of mak-
ing sense, of imposing meaning. Also, the conceptualizations that are expressed
in natural language have an experiential basis, i.e., they link up with the way in
which human beings experience reality, both culturally and physiologically. In
this sense, Cognitive Linguistics embodies a fully contextualized conception of
meaning. Third, the link between linguistic performance and grammar is re-
established by the view that language is usage-based, i.e. that there is a dialectic
relationship between langue and parole.
We can observe, then, that the various characteristics of Cognitive Linguistics
that you learned about in the first section of this introductory chapter and that
were further spelled out in twelve crucial concepts, can be summarized under
one general denominator: the recontextualization of grammar. If we assume,
next, that this recontextualizing tendency is an underlying trend in contemporary
cognitive-functionalist linguistics, Cognitive Linguistics is probably one of the
most outspoken representatives of this tendency. Surely, it is not the only one,
and there are whole disciplines in linguistics that are devoted to the exploration
of specific forms of context, like sociolinguistics dealing with the social context,
or pragmatics and text linguistics dealing with the level of parole. But Cognitive
Linguistics is specific in the extent to which it tries to integrate these different
tendencies into an overall model of language.
All of this does not mean, however, that Cognitive Linguistics has reached its
goal yet. To begin with, because it is far from being alone in pursuing a recontextu-
alized line of linguistic research, one of the major tasks for its future development
will be to systematically confront similar approaches within the broad cognitive-
functionalist domain and see to what extent theoretical and empirical convergences
are possible. That is not going to be self-evident, given that Cognitive Linguistics
is not even a theoretically uniform framework on its own account.
Further, we cannot say that Cognitive Linguistics has already realized all the
consequences of the decontextualizing stance. For one thing, seeing language in the
context of the mind at large would seem to imply a lively interdisciplinary interac-
tion with the other sciences of the mind, but that is a trend that is only gradually
emerging in mainstream cognitive linguistic circles. For an other, recontextual-
28 Dirk Geeraerts

izing language in its social context implies an awareness of the variation that is
inherent in the social life of language. Here again, sociolinguistically oriented
studies inspired by Cognitive Linguistics are only beginning to come into view.
In short, there is still a quite a lot to be done. Ultimately, that may well be
the best reason for coming back: not what has already been achieved, but what
still has to be done – all the exciting, inviting paths for further exploration. So
we will meet there, right?
Chapter 1
Cognitive Grammar

Introduction to Concept, Image, and Symbol

Ronald W. Langacker

Despite the diversity of contemporary linguistic theory, certain fundamental

views are widely accepted without serious question. Points of widespread agree-
ment include the following: (i) language is a self-contained system amenable to
algorithmic characterization, with sufficient autonomy to be studied in essential
isolation from broader cognitive concerns; (ii) grammar (syntax in particular)
is an independent aspect of linguistic structure distinct from both lexicon and
semantics; and (iii) if meaning falls within the purview of linguistic analysis, it is
properly described by some type of formal logic based on truth conditions. Indi-
vidual theorists would doubtlessly qualify their assent in various ways, but (i)–(iii)
certainly come closer than their denials to representing majority opinion.
Since 1976, I have been developing a linguistic theory that departs quite radi-
cally from these assumptions. Called “cognitive grammar” (alias “space gram-
mar”), this model assumes that language is neither self-contained nor describable
without essential reference to cognitive processing (regardless of whether one
posits a special faculté de langage). Grammatical structures do not constitute an
autonomous formal system or level of representation: they are claimed instead to
be inherently symbolic, providing for the structuring and conventional symboliza-
tion of conceptual content. Lexicon, morphology, and syntax form a continuum of
symbolic units, divided only arbitrarily into separate components; it is ultimately
as pointless to analyze grammatical units without reference to their semantic value
as to write a dictionary which omits the meanings of its lexical items. Moreover,
a formal semantics based on truth conditions is deemed inadequate for describ-
ing the meaning of linguistic expressions. One reason is that semantic structures
are characterized relative to knowledge systems whose scope is essentially open-
ended. A second is that their value reflects not only the content of a conceived
situation, but also how this content is structured and construed.
Cognitive grammar is therefore quite distinct from any version of generative
theory. Moreover, it departs from most varieties of traditional and formal seman-
tics, as well as the newer “situation semantics” of Barwise and Perry (1983), by

Originally published in 1990 as Chapter 1. Introduction in Concept, Image, Symbol: The Cog-
nitive Basis of Grammar, Ronald W. Langacker, 1–32. Berlin/New York: Mouton de Gruyter.
(Reprinted with permission from An introduction to cognitive grammar. Cognitive Science 10(1):
1–40. 1986).
30 Ronald W. Langacker

equating meaning with conceptualization (or cognitive processing). It agrees in

this regard with the “procedural semantics” of Miller and Johnson-Laird (1976)
and Johnson-Laird (1983), and with the linguistic theories of Chafe (1970) and
Jackendoff (1983); however, it is very different from all of these in its concep-
tion of grammatical organization and its specific proposals concerning semantic
structure. Although cognitive grammar is not a direct outgrowth or a variant of
any other linguistic theory, I do consider it compatible with a variety of ongo-
ing research programs. Among these are Lakoff’s work on categorization (1982,
1987); Fauconnier’s study of mental spaces (1985); Haiman’s ideas on iconicity
and encyclopedic semantics (1980, 1983, 1985); Talmy’s research on spatial terms,
force dynamics, and the meanings of grammatical elements (1975, 1977, 1978,
1983, 1985a, 1985b, 1988a, 1988b); the proposals of Moore and Carling con-
cerning the nonautonomy of linguistic structure (1982); Fillmore’s conception of
frame semantics (1982); Wierzbicka’s insightful investigation into the semantics
of grammar (1988); the growing body of research on metaphor and image sche-
mas (Johnson 1987; Lakoff and Johnson 1980; Lakoff and Turner 1989; Sweetser
1984, 1987); recent studies of grammaticization (Bybee 1988; Kemmer 1988;
Sweetser 1988; Traugott 1982, 1986, 1988); and the rich, multifaceted work in a
“functional” vein by scholars too numerous to cite individually (though Givón
1979, 1984, 1989 must certainly be mentioned).
This chapter affords an overview of cognitive grammar as I myself conceive
it. The topics it briefly covers will all be taken up again in later chapters of Lang-
acker (1991a) and examined in greater detail. Readers interested in still further
discussion and illustration of the theory will find it in the following works: Casad
1982, 1988; Cook 1988, 1989; Hawkins 1984, 1988; Janda 1984, 1988, 1993; Lang-
acker 1981, 1982, 1985, 1987, 1988a, 1988b, 1991b; Lindner 1981, 1982; Poteet
1987; Rice 1987a, 1987b, 1988; Smith 1985a, 1985b, 1987, 1989; Tuggy 1980,
1981, 1986, 1988, 1989; Vandeloise 1984.

1. Linguistic semantics

Meaning is equated with conceptualization. Linguistic semantics must therefore

attempt the structural analysis and explicit description of abstract entities like
thoughts and concepts. The term conceptualization is interpreted quite broadly:
it encompasses novel conceptions as well as fixed concepts; sensory, kinesthetic,
and emotive experience; recognition of the immediate context (social, physical,
and linguistic); and so on. Because conceptualization resides in cognitive process-
ing, our ultimate objective must be to characterize the types of cognitive events
whose occurrence constitutes a given mental experience. The remoteness of this
goal is not a valid argument for denying the conceptual basis of meaning.
 Chapter 1: Cognitive Grammar 31

Most lexical items have a considerable array of interrelated senses, which

define the range of their conventionally sanctioned usage. These alternate senses
are conveniently represented in network form; Figure 1 depicts a fragment of the
network associated with the noun ring. Certain senses are “schematic” relative to
others, as indicated by the solid arrows. Some represent “extensions” from oth-
ers (i.e. there is some conflict in specifications), as indicated by the dashed-line
arrows. The nodes and categorizing relationships in such a network differ in their
degree of entrenchment and cognitive salience – for instance, the heavy-line box in
Figure 1 corresponds to the category prototype. The precise configuration of such
a network is less important than recognizing the inadequacy of any reductionist
description of lexical meaning. A speaker’s knowledge of the conventional value
of a lexical item cannot in general be reduced to a single structure, such as the
prototype or the highest-level schema. For one thing, not every lexical category
has a single, clearly determined prototype, nor can we invariably assume a high-
level schema fully compatible with the specifications of every node in the network
(none is shown in Figure 1). Even if such a structure is posited, moreover, there
is no way to predict precisely which array of extensions and elaborations – out of
all those that are conceivable and linguistically plausible – have in fact achieved
conventional status. The conventional meaning of a lexical item must be equaled
with the entire network, not with any single node.

Figure 1.

Because polysemy is not our central concern, we will nevertheless focus on indi-
vidual nodes. What is required to adequately characterize any particular sense of
a linguistic expression? Specifically rejected is the idea that a semantic structure
reduces to a bundle of features or semantic markers (cf. Katz and Fodor 1963).
Rejected as well is the notion that all meanings are described directly in terms of
semantic primitives. It is claimed instead that semantic structures (which I call
“predications”) are characterized relative to “cognitive domains”, where a domain
can be any sort of conceptualization: a perceptual experience, a concept, a con-
ceptual complex, an elaborate knowledge system, etc. The semantic description
of an expression therefore takes for its starting point an integrated conception
of arbitrary complexity and possibly encyclopedic scope. The basic observation
32 Ronald W. Langacker

supporting this position is that certain conceptions presuppose others for their
characterization. We can thus posit hierarchies of conceptual complexity, where
structures at a given level arise through cognitive operations (including simple
coordination) performed on the structures at lower levels. Crucially, the cogni-
tive domains required by linguistic predications can occur at any level in such
hierarchies.
Consider some examples. The notion hypotenuse is readily characterized given
the prior conception of a right triangle, but incoherent without it; right triangle
therefore functions as the cognitive domain for hypotenuse. Central to the value
of elbow is the position of the designated entity relative to the overall configura-
tion of the human arm (try explaining what an elbow is without referring in any
way to an arm!), so arm is a domain for elbow. Similarly, tip presupposes the
conception of an elongated object, and April, of the calendrical cycle devised to
plot the passage of a year. A meaningful description of shortstop or sacrifice
fly is possible only granted substantial knowledge of the rules and objectives of
baseball. The implications of this position are apparent: the full and definitive
characterization of a semantic structure must incorporate a comparable descrip-
tion of its domain, and ultimately of the entire hierarchy of more fundamental
conceptions on which it depends. Pushing things to their logical conclusion, we
must recognize that linguistic semantics is not an autonomous enterprise, and
that a complete analysis of meaning is tantamount to a complete account of
developmental cognition. This consequence is terribly inconvenient for linguistic
theorists imprinted on autonomous formal systems, but that is not a legitimate
argument against its validity.
What occupies the lowest level in conceptual hierarchies? I am neutral as to
the possible existence of innately specified conceptual primitives. It is however
necessary to posit a number of “basic domains”, i.e. cognitively irreducible repre-
sentational spaces or fields of conceptual potential. Among these basic domains
are the experience of time and our capacity for dealing with two- and three-
dimensional spatial configurations. There are basic domains associated with the
various senses: color space (an array of possible color sensations), coordinated
with the extension of the visual field; the pitch scale; a range of possible tem-
perature sensations (coordinated with positions on the body); and so on. Emotive
domains must also be assumed. It is possible that certain linguistic predications
are characterized solely in relation to one or more basic domains, e.g. time for
before, color space for red, or time and the pitch scale for beep. However most
expressions pertain to higher levels of conceptual organization and presuppose
nonbasic domains for their semantic characterization.
Most predications also require more than one domain for their full description,
in which case I refer to the set as a “complex matrix”, as illustrated for knife in
Figure 2. One dimension of its characterization is a shape specification (or a family
 Chapter 1: Cognitive Grammar 33

of such specifications). Another is the canonical role of a knife in the process of

cutting. Additional properties are its inclusion in a typical place setting with other
pieces of silverware; specifications of size, weight, and material; information about
the manufacture of knives; the existence of knife-throwing acts in circuses; and
so on indefinitely. Obviously these specifications are not all on a par. They differ
greatly in their degree of “centrality”, i.e. the likelihood of their activation on a
given occasion of the expression’s use. Moreover, some are probably incorporated
as components of others – for instance, Figure 2 plausibly suggests that a shape
specification is typically included in the conceptions constituting other domains
of the complex matrix. I do however adopt an “encyclopedic” view of semantics
(Haiman 1980). There is no sharp dividing line such that all specifications on one
side are linguistically relevant and all those on the other side clearly irrelevant.
Any facet of our knowledge of an entity is capable in principle of playing a role
in determining the linguistic behavior of an expression that designates it (e.g. in
semantic extension, or in its combination with other expressions).

Figure 2.

If we succeed in identifying and describing the domain or complex matrix invoked

by a linguistic predication, we have not yet finished its characterization. Equally
significant for semantic structure is the “conventional imagery” inherent in the
meaning of an expression. By imagery, I do not mean sensory images à la Shepard
(1978) or Kosslyn (1980), though sensory images – as one type of conceptualiza-
tion – are quite important for semantic analysis. I refer instead to our manifest
capacity to structure or construe the content of a domain in alternate ways. This
multifaceted ability is far too often neglected in semantic studies. Let us explore
its dimensions and briefly note their grammatical significance.
34 Ronald W. Langacker

2. Dimensions of imagery

The first dimension of imagery, observed in every linguistic predication, is the

imposition of a “profile” on a “base”. The base of a predication is its domain (or
each domain in a complex matrix). Its profile is a substructure elevated to a special
level of prominence within the base, namely that substructure which the expression
“designates”.1 Some examples are sketched in Figure 3, with the profile given in
heavy lines. The base (or domain) for the characterization of hypotenuse is the
conception of a right triangle; for tip, the base is the conception of an elongated
object; and for uncle, a set of individuals linked by kinship relations. The base
is obviously essential to the semantic value of each predication, but it does not
per se constitute that value: a hypotenuse is not a right triangle, a tip is not an
elongated object, and an uncle is not a kinship network. The meaning of hypot-
enuse, tip, and uncle is in each case given only by the selection of a particular
substructure within the base for the distinctive prominence characteristic of a
profile. An expression’s semantic value does not reside in either the base or the
profile individually, but rather in the relationship between the two.

Figure 3.

Some further examples will demonstrate both the descriptive utility and the gram-
matical import of these constructs. The predications in question represent specific
senses of go, away, and gone, namely those illustrated in (1):

(1) a. I think you should go now.

b. China is very far away.
c. When I arrived, he was already gone.

Consider first the particular sense of go that is diagramed in Figure 4(a). This is a
relational rather than a nominal predication, i.e. it profiles the “interconnections”
among conceived entities; these interconnections are indicated in Figure 4 by
 Chapter 1: Cognitive Grammar 35

the heavy dashed lines. The relevant domains are space and time. With the pas-
sage of time, one individual, referred to here as the “trajector” (tr), moves from
a position within the neighborhood of another individual, the “landmark” (lm),
to a final position outside that neighborhood. Only four states of the process are
shown explicitly, but they represent a continuous series. The dotted lines indicate
that the trajectors “correspond” from one state to the next (i.e. they are construed
as identical), as do the landmarks. Away profiles a relationship that is identical
to the final state of go: the trajector is situated outside the vicinity of the land-
mark. Observe now that the participle gone profiles this same relationship, but it
does so with respect to a different base. The base for away is simply the spatial
domain, but the base for gone is the process profiled by go – something cannot
be gone except by virtue of the process of going. The semantic contribution of
the past participial inflection is to restrict the profile of the stem, in this case go,
to its final state. Gone thus differs from go by virtue of its profile, and from away
by virtue of its base.
A second dimension of imagery is the “level of specificity” at which a situa-
tion is construed. For example, the same situation might be described by any of
the sentences in (2):

Figure 4.

(2) a. That player is tall.

b. That defensive player is over 6’ tall.
c. That linebacker is about 6’ 5" tall.
d. That middle linebacker is precisely 6’ 5" tall.

Each of these sentences can be regarded as schematic for the one that follows,
which elaborates its specifications and confines their possible values to a narrower
range. It is well known that lexical items form hierarchies with respect to level of
specificity, e.g. animal ĺ reptile ĺ snake ĺrattlesnake ĺsidewinder. Relation-
ships of schematicity are also important for grammatical structure. Consider the
36 Ronald W. Langacker

combination of break and the cup to form the composite expression break the cup.
As part of its internal structure, the predicate break makes schematic reference to
two central participants. The combination of break and the cup is effected through
a correspondence established between one of these participants (its landmark) and
the entity profiled by the cup, which is characterized with far greater specificity.
One of the component expressions thus elaborates a schematic substructure within
the other, as is typically the case in a grammatical construction.
A third dimension of imagery pertains to the “scale” and “scope of predica-
tion”. The scope of a predication is the extent of its coverage in relevant domains.
A predication’s scope is not always sharply delimited or explicitly indicated, but
the construct is nonetheless of considerable structural significance (cf. Langacker
1991a: Chapter 2). Consider the notion island with respect to the various scopes
indicated in Figure 5. The outer box, scope (a), is presumably sufficient to establish
the land mass as an island, but scope (b) is at best problematic. There is no precise
requirement on how extensive the body of water surrounding an island must be,
but the narrow strip of water included in (b) does not have the necessary expanse
(e.g. it could simply be a moat, and the land inside a moat is not thought of as an
island). Similarly, the finger of land projecting out into the water qualifies as a
peninsula given scope (c), but not (d); only from the former can we determine
that the overall land mass is quite large relative to the finger-like projection. We
can see that predications often imply a particular scale by noting the infelicity
of using island to designate a handful of mud lying in the middle of a puddle. In
my own speech, bay and cove are quite comparable in meaning except that bay
specifies the requisite configuration of land and water on a larger scale.

Figure 5.

Body-part terms illustrate the semantic and structural significance of these con-
structs. Essential to the characterization of terms like head, arm, and leg is the
position of the profiled entity relative to the body as a whole, whose conception
 Chapter 1: Cognitive Grammar 37

thus functions as their domain and immediate scope of predication. Each of

these designated entities functions in turn as immediate scope of predication for
other body-part terms defined on a smaller scale, e.g. hand, elbow, and forearm
in the case of arm. Hand then furnishes the immediate scope of predication for
palm, thumb, and finger, on a still smaller scale, and finger for knuckle, fingertip,
and fingernail. This hierarchical organization has structural consequences. For
example, sentences like those in (3), where have pertains to part-whole relation-
ships, are most felicitous (other things being equal) when the subject designates
the immediate scope of predication for the object (cf. Bever and Rosenbaum
1970; Cruse 1979).

(3) a. A finger has 3 knuckles and 1 nail.

b. ??An arm has 14 knuckles and 5 nails.
c. ???A body has 56 knuckles and 20 nails.

A similar restriction can be observed with noun compounds. We find numerous

terms like fingertip, fingernail, toenail, eyelash, and eyelid, where the first element
of the compound constitutes the immediate scope of predication for the second.2
Compare this to the nonexistence and oddity of expressions like *bodytip, *arm-
nail, *footnail, *facelash, and *headlid to designate the same entities.
In certain grammatical constructions the scope of predication plays a specific
structural role. A case in point is the “nested locative” construction exemplified
in (4).

(4) a. The quilt is upstairs in the bedroom in the closet on the top shelf
behind the boxes.
b. The rake is in the yard by the back fence near the gate.

Each locative expression confines the subject to a specific “search domain”, which
then constitutes the scope of predication for the locative that follows. Thus in
(4a) the locative upstairs confines the quilt to an upper story, and in the bedroom
is construed relative to this restricted region – only an upstairs bedroom need
be considered. The search domain imposed by this second locative functions in
turn as the scope of predication for in the closet, and so on. Formally, these rela-
tionships are handled by positing a correspondence between the search domain
of each locative and the scope of predication of its successor. Apart from the
abstractness of the entities concerned, this correspondence is just like that found
in any instance of grammatical combination (e.g. between the landmark of break
and the profile of the cup in break the cup).
The relative salience of a predication’s substructures constitutes a fourth dimen-
sion of imagery. Salience is of course a very general notion, so its descriptive
38 Ronald W. Langacker

significance depends on our ability to sort out the various contributing factors.
One factor is the special prominence associated with profiling (considered pre-
viously). A number of others can be discerned, but only two will be discussed:
the relative prominence of relational participants, and the enhanced salience of
elements that are explicitly mentioned.
Relational predications normally manifest an asymmetry in the portrayal
of the relational participants. This asymmetry is not strictly dependent on the
content of the predication, and is consequently observable even for expressions
designating symmetrical relationships, e.g. resemble. I maintain that X resembles
Y and Y resembles X are semantically distinct (even granting their truth-condi-
tional equivalence): the former characterizes X with reference to Y, and the latter
describes Y with reference to X. We can similarly employ either X is above Y or
Y is below X to describe precisely the same conceived situation, but they differ
in how they construe this situation; in the former, Y functions as a point of ref-
erence – a kind of landmark – for locating X, whereas the latter reverses these
roles. The subtlety of the contrast with predications like these hardly diminishes
its significance for linguistic semantics and grammatical structure. The asym-
metry is more apparent in cases like go, hit, enter, and approach, where one par-
ticipant moves in relation to another (which is stationary so far as the verb itself
is concerned), but its characterization must be abstract enough to accommodate
the full range of relational expressions.
I attribute this inherent asymmetry to figure/ground organization (for discus-
sion, see Langacker 1987: Chapter 6). A relational predication elevates one of its
participants to the status of figure. I refer to this participant as its “trajector”; other
salient participants are referred to as “landmarks”. This terminology is inspired
by prototypical action verbs, where the trajector is usually the initial or primary
mover, but the definitions make no specific reference to motion and are there-
fore applicable to any relational expression. The trajector/landmark asymmetry
underlies the subject/object distinction, but the former notions have considerably
broader application. In particular, a schematic trajector and landmark are imputed
to a relational predication’s internal structure, regardless of whether these entities
receive (or are capable of receiving) separate expression. The verb read conse-
quently has a trajector and a landmark in all the sentences of (5), despite the fact
that both are made explicit (by elaborative noun phrases) only in (5a):

(5) a. David read a new book.

b. David is reading.
c. The best way to learn is to read.

The terms subject and object are generally reserved for overt noun phrases that
elaborate a relational trajector and primary landmark at the clausal level. By con-
 Chapter 1: Cognitive Grammar 39

trast, trajector/landmark asymmetry is characteristic of relational predications at

any level of organization, even if left implicit.
The enhanced salience of explicitly mentioned elements can be illustrated by
the semantic contrast between pairs of expressions like the following: father vs.
male parent; pork vs. pig meat; oak vs. oak tree; triangle vs. three-sided polygon;
and sink vs. passively descend through a medium under the force of gravity. I am
not concerned here with differences in connotation or information content – for
sake of discussion, let us accept the members of each pair as equivalent in these
respects. My claim is that the paired expressions nevertheless contrast semantically
because the second expression in each case explicitly mentions certain semantic
components and thereby renders them more prominent than they would otherwise
be. Even for a speaker who knows perfectly well that pork comes from pigs, the
expression pig meat renders this provenience more salient than does pork, sim-
ply because the former incorporates a symbolic unit that specifically designates
this source. In similar fashion, the inclusion of the designated entity in a broader
class of geometrical figures is highlighted by three-sided polygon, but remains
latent in the case of triangle.
A linguistically appropriate characterization of meaning should accommodate
such differences. Cognitive grammar defines the meaning of a complex expression
as including not only the semantic structure that represents its composite sense,
but also its “compositional path”: the hierarchy of semantic structures reflecting
its progressive assembly from the meanings of component expressions. Let us
assume, for example, that the composite semantic values of pork and pig meat are
identical. As an unanalyzable morpheme, pork symbolizes this notion directly, so
its compositional path consists of the single semantic structure [PORK]. However
pig meat is “analyzable”, i.e. speakers recognize the semantic contribution of its
component morphemes. The meaning of pig meat therefore incorporates not only
the composite structure ;PORK], but also the individually symbolized components
[PIG] and [MEAT], together with the relationship that each of them bears to the
composite value. The two expressions arrive at the same composite value through
different compositional paths (a degenerate path in the case of pork), with the
consequence that they differ in meaning.
Besides accounting for the semantic contrast between simple and composite
expressions, this conception of meaning has the advantage of resolving a classic
problem of truth-conditional semantics. The problem is posed by semantically
anomalous expressions, e.g. * perspicacious neutrino and *truculent spoon,
which lack truth conditions and thus ought to be meaningless and semantically
equivalent. Not only is this counterintuitive, but it also predicts – quite incor-
rectly – the semantic anomaly of sentences like those in (6), which contain anoma-
lous constituents.
40 Ronald W. Langacker

(6) a. There is no such thing as a perspicacious neutrino.

b. It is meaningless to speak of a truculent spoon.

In the present framework, anomalous expressions are indeed both meaningful and
nonsynonymous. Though a coherent composite conceptualization fails to emerge
for *perspicacious neutrino, it has a semantic value, consisting of the meanings
of its components together with their specified mode of combination (as deter-
mined by the grammatical construction). The same is true for *truculent spoon,
and because its components are different from those of *perspicacious neutrino,
so is its semantic value. Lacking a coherent composite sense, these meanings are
defective, but they are meanings nonetheless. Sentences like (6) are semantically
well-formed precisely because they comment on the anomaly of a constituent.
I will mention two more dimensions of imagery only in passing, though each
is multifaceted and merits extended discussion. One is the construal of a situation
relative to different background assumptions and expectations. To take just one
example, either (7a) or (7b) might be used to describe the same state of affairs:

(7) a. He has a few friends in high places.

b. He has few friends in high places.
c. Few people have any friends in high places.
d. *A few people have any friends in high places.

Intuitively, the difference between few and a few is that the former is somehow
negative, and the latter more positive. This is corroborated by (7c) and (7d): any,
which requires a negative context (cf. Klima 1964), is compatible with few, but
not with a few. Analytically, I suggest that few construes the specified quantity as
being less than some implicit norm, whereas a few construes the quantity relative
to a baseline of zero. These respective predications therefore indicate departure
from an implicit reference point in a negative vs. a positive direction.
The final dimension of imagery is perspective, which subsumes a number of
more specific factors: orientation, assumed vantage point, directionality, and how
objectively an entity is construed. Orientation and vantage point are well known
from the ambiguity of sentences like (8a). The contrast between (8b) and (8c)
shows the importance of directionality, even for situations that appear to involve
no motion.
(8) a. Brian is sitting to the left of Sally.
b. The fall falls gently to the bank of the river.
c. The hill rises gently from the bank of the river.
d. The balloon rose swiftly.
 Chapter 1: Cognitive Grammar 41

I suggest, though, that (8b)–(8d) all involve motion in an abstract sense of the
term (see Langacker 1991a: Chapter 5). Described in (8d) is physical motion on
the part of a mover construed “objectively”, by which I mean that it is solely an
object of conceptualization, maximally differentiated from the conceptualizer
(i.e. the speaker and/or hearer). Motion along a similar trajectory is implied in
(8c), but in this case the movement is abstract and the mover is construed “sub-
jectively”: the mover is none other than the conceptualizer, in his role as the agent
(rather than the object) of conceptualization. Gradations between physical and
abstract motion on the one hand, and between objective and subjective construal
of conceived entities on the other, are important to the analysis of numerous lin-
guistic phenomena.3

3. Grammar as image

Lexicon and grammar form a continuum of symbolic elements. Like lexicon, gram-
mar provides for the structuring and symbolization of conceptual content, and is
thus imagic in character. When we use a particular construction or grammatical
morpheme, we thereby select a particular image to structure the conceived situa-
tion for communicative purposes. Because languages differ in their grammatical
structure, they differ in the imagery that speakers employ when conforming to
linguistic convention. This relativistic view does not per se imply that lexicogram-
matical structure imposes any significant constraints on our thought processes
– in fact I suspect its impact to be rather superficial (cf. Langacker 1976). The
symbolic resources of a language generally provide an array of alternative images
for describing a given scene, and we shift from one to another with great facility,
often within the confines of a single sentence. The conventional imagery invoked
for linguistic expression is a fleeting thing that neither defines nor constrains the
contents of our thoughts.
The most obvious contribution of grammar to the construal of a scene pertains
to designation. Grammatical constructions have the effect of imposing a particular
profile on their composite semantic value. When a head combines with a modifier,
for example, it is the profile of the head that prevails at the composite structure
level. Consider a simple situation in which a lamp is suspended over a table. Start-
ing from such simple expressions as the lamp, the table, above, and below, we can
combine them in alternate ways to form composite expressions that profile differ-
ent facets of the scene. The lamp above the table naturally designates the lamp.
By choosing the table for the head, and appropriately adjusting the prepositional
phrase modifier, we obtain instead the table below the lamp, which profiles the
table. Another option is to add the proper form of be to the prepositional phrase,
converting it into a process predication designating the extension of the locative
42 Ronald W. Langacker

relationship through a span of conceived time, e.g. is above the table. When a
subject is then supplied, the resulting sentence The lamp is above the table also
profiles the temporally extended locative relationship.
Let us further explore the sense in which grammar embodies conventional
imagery by considering the semantic contrast between (9a) and (9b).

(9) a. Bill sent a walrus to Joyce.

b. Bill sent Joyce a walrus.

The standard transformational analysis of these sentences treats them as synony-

mous and derives them from a common deep structure; depending on the particular
choice of deep structure, to is either deleted or inserted transformationally, and
the nonsubject nominals are permuted in the course of deriving the surface form
of either (9a) or (9b). Cognitive grammar does not posit abstract deep structures,
and neither sentence type is derived from the other – they are claimed instead to
represent alternate construals of the profiled event. Examples (9a) and (9b) differ
in meaning because they employ subtly different images to structure the same
conceived situation.

Figure 6.

The essentials of the analysis are sketched in Figure 6, where the small circles
represent Bill, Joyce, and the walrus; the large circles stand for the regions over
which Bill and Joyce exercise dominion; and heavy lines indicate a certain degree
of relative prominence. Up to a point the sentences are semantically equivalent.
Each symbolizes a conception in which a walrus originates in the domain under
Bill’s control and – at Bill’s instigation – follows a path that results in its eventual
location within the region under Joyce’s control. The semantic contrast resides in
the relative salience of certain facets of this complex scene. In (9a), the morpheme
to specifically designates the path followed by the walrus, thereby rendering this
aspect of the conceptualization more prominent than it would otherwise be, as
indicated in Figure 6(a). In (9b), on the other hand, to is absent, but the juxtaposi-
tion of two unmarked nominals (Joyce and a walrus) after the verb symbolizes a
possessive relationship between the first nominal and the second. Consequently
 Chapter 1: Cognitive Grammar 43

(9b) lends added prominence to the configuration that results when the walrus
completes its trajectory, namely that which finds it in Joyce’s possession, as indi-
cated in Figure 6(b).
All of the content present in one conception may be presumed to figure in the
other as well – what differs is the relative salience of substructures. This subtle
difference in imagery has an impact on the felicity of using to or the double-object
construction for certain types of situations.4 Consider the data in (10):

(10) a. I sent a walrus to Antarctica.

b. ?I sent Antarctica a walrus.
c. I sent the zoo a walrus.

Example (10a) is fully acceptable because to emphasizes the path traversed by

the walrus, and a continent can perfectly well be construed as the endpoint of a
path. However, it is harder to construe a continent as a possessor exercising control
over other entities, so (l0b), which specifically places Antarctica in a possessor
role, is felt to be marginal. The status of (l0c) depends on the construal of zoo. If
the zoo is simply construed as a place, it is difficult to view it as a possessor, and
(10c) is questionable for the same reason as (10b). But a zoo is also an institu-
tion, and it is conventional in English to treat institutions as being analogous to
people, which allows them to function linguistically as agents, possessors, and
so forth. Example (10c) is consequently well formed to the extent that this sec-
ond construal prevails. As viewed in the present framework, then, judgments of
well-formedness often hinge on the interplay and compatibility of images, and
are influenced by subtle shifts in context, intended meaning, or how a speaker
chooses to structure and interpret a situation.
The examples in (11)–(13) provide further illustration.

(11) a. I gave the fence a new coat of paint.

b. ?I gave a new coat of paint to the fence.

(12) a. I cleared the floor for Bill.

b. ?I cleared Bill the floor.
c. I cleared Bill a place to sleep on the floor.

(13) a. I baked her a cake.

b. ?I mowed her the lawn.

It is conventional in English to employ possessive locutions for part-whole rela-

tions, so construing a fence as the possessor of a new coat of paint, in the manner
of (11a), is quite natural. It is more difficult to envisage a coat of paint moving
44 Ronald W. Langacker

along a path to the fence; (11b) is thus a bit less natural, because to renders the
path more prominent than the eventual possessive relationship.5 The sentences
in (12)–(13) bring out another consequence of the analysis. Because the two
constructions are claimed to be parallel (i.e. neither is derived from the other)
and semantically distinct, it is to be expected that the double-object construction
– having no intrinsic connection with to – might serve as an alternative to other
prepositions also. It is well known from transformational studies (where the fact
has long been problematic) that the double-object construction alternates with
for as well as to. With for also the double-object construction is restricted to
instances where the first object is plausibly construed as winding up in posses-
sion of the second. In (12), for example, Bill does not come to possess the floor
just because I clear it for him, so (12b) is peculiar; (12c) is perfectly acceptable,
however, since the additional context provided by the second nominal (a place to
sleep on the floor) makes it apparent that the spot in question effectively comes
under Bill’s control and lies at his disposal by virtue of the act of clearing it. The
data in (13) is similarly explained. Baking someone a cake puts the cake at that
person’s disposal, but mowing a lawn can hardly have a comparable effect under
normal circumstances.

4. Grammatical organization

The ultimate goal of linguistic description is to characterize, in a cognitively

realistic fashion, those structures and abilities that constitute a speaker’s grasp
of linguistic convention. A speaker’s linguistic knowledge is procedural rather
than declarative, and the internalized grammar representing this knowledge is
simply a “structured inventory of conventional linguistic units”. The term “unit”
is employed in a technical sense to indicate a thoroughly mastered structure, i.e.
one that a speaker can activate as a preassembled whole without attending to the
specifics of its internal composition. A unit can therefore be regarded as a cogni-
tive routine. The inventory of conventional units is “structured” in the sense that
some units function as components of others (i.e. they constitute subroutines).
I speak of an “inventory” of conventional units to indicate that a grammar
is nongenerative and nonconstructive. That is, I reject the standard notion that a
grammar is properly conceived as an algorithmic device giving a well-defined
class of expressions (“all and only the grammatical sentences of a language”) as
output. This conception is viable only if one imposes arbitrary restrictions on the
scope of linguistic structure and makes gratuitous assumptions about its char-
acter. It is commonly assumed, for example, that judgments of grammaticality
are categorical rather than matters of degree; that figurative language is properly
excluded from the domain of linguistic description; and that a motivated distinc-
 Chapter 1: Cognitive Grammar 45

tion can be made between semantics and pragmatics. Although such assumptions
support the notion that language is self-contained and cognitively autonomous,
there is little factual basis for their adoption.
Instead, I conceive the grammar of a language as merely providing the speaker
with an inventory of symbolic resources, among them schematic templates
representing established patterns in the assembly of complex symbolic structures.
Speakers employ these symbolic units as standards of comparison in assessing the
conventionality of novel expressions and usages, whether of their own creation or
supplied by other speakers. The novel symbolic structures evaluated in this fashion
are not a well-defined set and cannot be algorithmically derived by the limited
mechanisms of an autonomous grammar. Rather their construction is attributed
to problem-solving activity on the part of the language user, who brings to bear
in this task not only his grasp of linguistic convention, but also his appreciation
of the context, his communicative objectives, his esthetic sensibilities, and any
aspect of his general knowledge that might prove relevant. The resulting symbolic
structures are generally more specific than anything computable from linguistic
units alone, and often conflict with conventional expectations (e.g. in metaphor
and semantic extension). Assessing their conventionality (or “well-formedness”)
is a matter of categorization: categorizing judgments either sanction them as
elaborations of schematic units or recognize them as departing from linguistic
convention as currently established.
Only three basic types of units are posited: semantic, phonological, and sym-
bolic. A symbolic unit is said to be “bipolar”, consisting of a semantic unit defining
one pole and a phonological unit defining the other: [[SEM]/[PHON]]. That lexical
units have this bipolar character is uncontroversial; pencil, for example, has the
form [[PENCIL]/[pencil]], where capital letters abbreviate a semantic structure
(of indefinite internal complexity), and a phonological structure is represented
orthographically. A pivotal claim of cognitive grammar is that grammatical units
are also intrinsically symbolic. I maintain, in other words, that grammatical mor-
phemes, categories, and constructions all take the form of symbolic units, and that
nothing else is required for the description of grammatical structure.
Symbolic units vary along the parameters of complexity and specificity. With
respect to the former, a unit is minimal (a “morpheme”) if it contains no other
symbolic units as components. For instance, despite its internal complexity at both
the semantic and the phonological poles, the morpheme sharp is minimal from
the symbolic standpoint, whereas sharpen, sharpener, and pencil sharpener are
progressively more complex. With respect to the second parameter, symbolic units
run the gamut from the highly specific to the maximally schematic. Each sense
of ring depicted in Figure 1, for example, combines with the phonological unit
[ring] to constitute a symbolic unit. Some of these senses are schematic relative
to others, so the symbolic units in question vary in their level of specificity at the
46 Ronald W. Langacker

semantic pole. Basic grammatical categories (e.g. noun, verb, adjective, adverb)
are represented in the grammar by symbolic units that are maximally schematic
at both the semantic and the phonological poles. A noun, for instance, is claimed
to instantiate the schema [[THING]/[X]], and a verb the schema [[PROCESS]/[Y]],
where [THING] and [PROCESS] are abstract notions to be described later, and [X]
and [Y] are highly schematic phonological structures (i.e. they specify little more
than the presence of “some phonological content”).
A grammatical rule or construction is represented in the grammar by a sym-
bolic unit that is both complex and schematic. For example, the morphological
rule illustrated by the deverbal nominalizations teacher, helper, hiker, thinker,
diver, etc. consists in a complex unit that incorporates as components the verb
schema [[PROCESS]/[Y]] and the grammatical morpheme [[ER]/[er]] (i.e. the suffix
-er, which is attributed substantial though schematic semantic content). This unit
further specifies how the component structures are integrated, conceptually and
phonologically, to form a composite symbolic structure. Using “-” to indicate this
integration (examined later), we can write the constructional schema as follows:
[[[PROCESS]/[Y]]-[[ER]/[er]]]. Its internal structure is exactly parallel to that of an
instantiating expression, e.g. [[[TEACH]/[teach]]-[[ER]/[er]]], except that in lieu of
a specific verb stem it contains the schema for the verb-stem category.
One constructional schema can be incorporated as a component of another.
In the top portion of Figure 7(a), the schema just described combines with the
noun schema [[THING]/[X]] to form a higher-order constructional schema, which
speakers presumably extract to represent the commonality of pencil sharpener,
lawn mower, mountain climber, back scratcher, taxi driver, and so on. The lower
portion of 7(a) represents the lexical unit pencil sharpener, which conforms to
the specifications of this schema but elaborates it greatly. The arrow labeled (a)
indicates that the upper structure as a whole is judged schematic for the overall
expression; this categorizing relationship is what specifies the membership of
the expression in the class that the schema characterizes. This global categoriz-
ing relationship is based on local categorizations between component structures:
relationship (b) identifies pencil as a member of the noun class; (c) categorizes
sharpener as a deverbal nominalization derived by -er; and (d) classes sharpen
as a verb.6 The full set of categorizing relationships of this sort constitutes the
expression’s “structural description”. Observe that pencil sharpener has a con-
ventional meaning which is considerably more specific than anything derivable
compositionally from the meanings of its parts – a pencil sharpener is not simply
’something that sharpens pencils’. Given the nonconstructive nature of the present
model, we can nevertheless accept the expression as a valid instantiation of the
construction in question, without relegating the unpredictable semantic specifica-
tions to the realm of extralinguistic knowledge. The constructional schema is not
responsible for assembling the expression, but only for its categorization.
 Chapter 1: Cognitive Grammar 47

Figure 7.

All of the structures and categorizing relationships in Figure 7(a) have the status
of units, which I indicate by enclosing them in boxes or square brackets. What
about a novel expression on the same model, for example chalk sharpener? Its
organization is sketched in Figure 7(b), where a closed curve (as opposed to a
box) indicates a structure that does not yet constitute a unit. The assembly of this
novel symbolic structure is largely prefigured by existing units, including the con-
structional schema, the components chalk and sharpener, and the categorization
of chalk as a noun. Taken as a whole, however, neither the full expression chalk
sharpener nor its categorization by the constructional schema (relationship (a))
has unit status. It does not matter for our purposes whether a speaker employs the
existing units to construct or simply to understand the novel expression – in either
case, all of the structures and relationships in 7(b) figure in its composition and
structural description, and in either case its contextual meaning may incorporate
specifications that are obvious from the situation being described (which functions
as the domain for the composite expression) but are not supplied by the conven-
tional meanings of its components. Despite this lack of full compositionality, the
expression may well recur with sufficient frequency to become established as a
conventional unit parallel to pencil sharpener, lawn mower, etc. If so, its contex-
tual meaning (in an appropriately schematized form) becomes the conventional
meaning of the new lexical unit. Full semantic compositionality is therefore not a
hallmark of either novel expressions as they are actually understood or the fixed
expressions which result from their conventionalization.
This conception of grammar makes it possible to impose the following restric-
tion on linguistic analyses: the only units permitted in the grammar of a language
are (i) semantic, phonological, and symbolic structures that occur overtly in
linguistic expressions; (ii) structures that are schematic for those in (i); and (iii)
categorizing relationships involving the structures in (i) and (ii). I call this the
“content requirement”, and consider it to be intrinsically more restrictive (at least
48 Ronald W. Langacker

in a certain, possibly nontechnical sense) than the constraints generally imposed

on algorithmic models. Essentially, it rules out all arbitrary descriptive devices,
i.e. those with no direct grounding in phonetic or semantic reality. Among the
devices excluded are contentless features or arbitrary diacritics; syntactic dum-
mies with neither semantic nor phonological content, introduced solely to drive
the formal machinery of autonomous syntax (cf. Perlmutter 1978); and the deri-
vation of overt structures from abstract, underlying structures of a substantially
different character (e.g. the derivation of passives from actives – see Langacker
(1991a: Chapter 4) for an alternative account).

5. Grammatical classes

The content requirement proscribes the use of diacritic features. How, then, does a
grammar indicate the behavior and class membership of conventional units? Some
classes are characterized on the basis of intrinsic semantic and/or phonological
content. In this event, a schematic unit is extracted to represent the shared content,
and class membership is indicated by categorizing units reflecting the judgment
that individual members instantiate the schema. The vowel [i], for example, is
classed as a high vowel by virtue of the categorizing unit [[HIGH¬VOWEL]ĺ[i]],
where [HIGH¬VOWEL] is a schematic phonological structure which neutralizes the
properties that distinguish one high vowel from another. Similarly, among the
categorizing units depicted in Figure 7(a), relationships (b) and (d) identify pencil
and sharpen as a noun and a verb respectively, whereas relationship (a) identifies
pencil sharpener as an instance of the grammatical construction characterized
by the overall schema. Only symbolic structures with actual semantic and pho-
nological content figure in these relationships.
Obviously, though, the membership of many grammatical classes is not fully
predictable on the basis of semantic or phonological properties, e.g. the class of
nouns that voice f to v in the plural (leaf/leaves, but reef/reefs), or the class of verbs
that conventionally occur in the double-object construction described earlier (cf.
Green 1974; Oehrle 1977). The fact that morphological and syntactic behavior is
often not fully predictable is generally taken as establishing the independence of
grammar as a distinct aspect of linguistic structure. However, this conclusion does
not actually follow from the observation – the tacit reasoning behind it confounds
two issues that are in principle distinct: (i) what KINDS¬of structures there are; and
(ii) the PREDICTABILITY of their behavior. The present framework accommodates
unpredictable behavior without positing arbitrary diacritics or rule features. To say
that leaf (but not reef) voices f to v in the plural is simply to say that the composite
symbolic structure leaves (but not reeves) is included among the conventional
units of the grammar. Similarly, to say that send participates in the double-object
 Chapter 1: Cognitive Grammar 49

construction amounts to positing the constructional schema [send NP NP], where

the verb is specific but the two noun phrases are characterized only schematically.
The nonoccurrence of transfer in this construction is reflected in the grammar by
the nonexistence of the parallel symbolic unit [transfer NP NP].7
Crucial to the claim that grammatical structure resides in symbolic units
alone is the possibility of providing a notional characterization of basic gram-
matical categories, nouns and verbs in particular. The impossibility of such a
characterization is a fundamental dogma of modern linguistics, but the standard
arguments that appear to support it are not immune to criticism. For one thing,
they presuppose an objectivist view of meaning, and thus fail to acknowledge
sufficiently our capacity to construe a conceived situation in alternate ways. Con-
sider the argument based on verb/noun pairs which refer to the same process,
e.g. extract and extraction. Such pairs demonstrate the impossibility of a notional
definition only if one assumes that they are semantically identical, yet this is not
a necessary assumption when meaning is treated as a subjective phenomenon.
It is perfectly coherent to suggest that the nominalization of extract involves a
conceptual reification of the designated process, i.e. the verb and noun construe
it by means of contrasting images. Another type of argument against a notional
characterization pivots on the confusion of prototypes with abstract schemas.
In the case of nouns, for instance, discussions of notional definitions generally
focus on physical objects (or perhaps “persons, places, and things”), which are
clearly prototypical; the existence of nouns like extraction, which do not conform
to this prototype, is then taken as demonstrating that nouns are not a semantic
class. Obviously, a schematic characterization of the class – one compatible with
the specifications of all class members – cannot be identified with the category
prototype representing typical instances. If a schematic characterization is pos-
sible at all, it must be quite abstract, accommodating both physical objects and
many other sorts of entities as special cases.
Cognitive grammar posits a number of basic classes that differ in the nature
of their profile (see Langacker 1991a: Chapter 3 for extensive discussion). As
previously indicated, a noun is a symbolic structure that designates a thing, where
“thing” is a technical term defined as a “region in some domain”; in the case of
count nouns, the profiled region is further specified as being “bounded”. Because
physical objects occupy bounded regions in three-dimensional space, expressions
which designate such objects qualify as count nouns, but the definition does not
specifically refer to them or to the spatial domain in particular. Examples of count
nouns characterized with respect to other domains include moment (a bounded
region in time), paragraph (a delimited portion of a written work), and B-flat
(a minimal, point-like region on the musical scale). Observe that the bounding
implied by a count noun need not be sharp or precise, and it may be imposed
50 Ronald W. Langacker

as a matter of construal when objective factors do not suggest any demarcation.

Where, for instance, does one’s midriff begin or end?
Contrasting with nouns are “relational” expressions, which profile the “inter-
connections” among conceived entities. The term “entity” is employed in a maxi-
mally general way, and subsumes anything we might have occasion to refer to
for analytic purposes: things, relations, boundaries, points on a scale, and so on.
Interconnections can be regarded as cognitive operations that assess the relative
position of entities within the scope of predication. It is speculated that only four
basic types of assessment are necessary, provided that cognitive domains have
been properly described: inclusion (INCL), coincidence (COINC), separation (SEP),
and proximity (PROX). Significantly, the interconnecting operations defining a
relational conception commonly associate entities other than the major relational
participants (trajector and primary landmark), or associate selected facets of these
participants rather than treating them as undifferentiated wholes.
By way of illustration, consider the predicate [ABOVE], sketched in Figure 8. Its
domain is space organized in terms of vertical and horizontal dimensions, including
an implicit reference point Ov (the vertical origin). The major relational partici-
pants are both things, characterized only schematically; one is further identified
as the trajector (relational figure).8 Among the entities invoked by specifications
of this predicate are the horizontal and vertical projections of the trajector (ht, vt)
and of the landmark (h1, v1). The expression above is optimally employed when
the horizontal projections of the trajector and landmark coincide, i.e. [ht COINC h1],
but is tolerated so long as they remain in proximity to one another: [ht PROX h1].
With respect to the vertical dimension, on the other hand, their projections must
not coincide – the specification [vt SEP v1] is obligatory. The pivotal specification
of [ABOVE] is provided by an operation interconnecting two entities that are still
more abstract. Let [Ov > vt] be the operation which registers the displacement
of the trajector from the vertical origin, and [Ov > vl] that of the landmark. The
specification in question resides in a higher-order operation assessing the relative
magnitudes of the component operations: [(Ov > vt) INCL (Ov > vl)].

Figure 8.
 Chapter 1: Cognitive Grammar 51

Interconnecting operations of roughly this sort must somehow figure in the cogni-
tive representation of a relational notion (though I take no position on the specifics
of their implementation). [ABOVE] is a “simple atemporal relation” (or “stative”
relation), in the sense that its specifications portray a single, internally consistent
configuration. We must also recognize “complex” atemporal relations, where such
is not the case. Consider the contrast between (14a) and (14b).

(14) a. There is a bridge across the river.

b. A hiker waded across the river.

Distinct senses of across are involved, diagramed in Figures 9(a) and (b). In
9(a), the trajector (in this case the bridge) simultaneously occupies all the points
on a path leading from one side of the primary landmark (the river) to the other.
In 9(b), on the other hand, the trajector still occupies all the points on the path
leading from one side of the landmark to the other, but does so only successively
through time. The profiled relationship involves indefinitely many distinct con-
figurations (or states), of which only a few are represented diagramatically. This
sense of across is consequently a complex atemporal relation.9

Figure 9.

Atemporal relations contrast with “processes”, which define the class of verbs.
The distinction between a process and a complex atemporal relation involves the
contrast between “sequential” and “summary scanning” (see Langacker 1991a:
Chapter 3). Sequential scanning is the mode of processing we employ when
watching a motion picture or observing a ball as it flies through the air. The
successive states of the conceived event are activated serially and more or less
instantaneously, so that the activation of one state begins to decline as that of its
successor is initiated; essentially, we follow along from one state to the next as
the event unfolds.10 On the other hand, summary scanning is what we employ in
mentally reconstructing the trajectory a ball has followed (e.g. in identifying a
52 Ronald W. Langacker

pitch as a curve, fastball, or slider and diagraming its degree of curvature). The
component states are activated successively but cumulatively (i.e. once activated
they remain active throughout), so that eventually they are all coactivated as a
simultaneously accessible whole. The difference between a complex atemporal
relation (like across) and the corresponding verb (cross) is therefore attributed
not to their intrinsic content, but rather to the mode of scanning employed in their
activation – a matter of conventional imagery.
Abbreviatory notations for the basic classes of predications are presented in
Figure 10. A circle is the natural choice to represent a thing. A simple atemporal
(or stative) relation profiles the interconnections between two or more conceived
entities, where an entity can be either a thing or another relation. (Dashed lines
represent these interconnections, and by convention the uppermost of the inter-
connected entities will be taken as the trajector unless otherwise indicated.) A
complex atemporal relation consists of a sequence of stative relations scanned in
summary fashion. A process is comparable to a complex atemporal relation in
profiling a sequence of relational configurations, but has certain other properties
as well: (i) the component states are conceived as being distributed through time;
(ii) these states are scanned in sequential fashion; and (iii) the trajector is always
a thing (never a relation). The arrow in Figure 10(e) stands for conceived time,
and the heavy-line bar along this arrow indicates that the component states are
scanned sequentially through processing time.

Figure 10.

6. Grammatical constructions

Grammar resides in patterns for the successive combination of symbolic struc-

tures to form more and more elaborate symbolic expressions. It is described by a
structured inventory of “grammatical constructions”, each of which specifies the
 Chapter 1: Cognitive Grammar 53

relation between two or more “component” structures and the “composite” struc-
ture resulting from their integration. The essential structures and relationships in
a grammatical construction are spelled out in Figure 11, where [SEM3 /PHON3] is
the composite structure formed by integrating the component expressions [SEM 1/¬
PHON 1] and [SEM 2 /¬PHON 2]. The two diagrams are notational variants: 11(b) is
an “exploded” version of 11(a); it shows the component and composite structures
separately at each pole.
Four symbolic relationships are indicated in Figure 11. The ones labeled s1 and
s2 hold between the semantic and the phonological poles of each component expres-
sion, whereas s3 indicates that the composite phonological structure symbolizes
the composite semantic structure. The fourth relationship, si, reveals an important
sense in which grammar is said to be inherently symbolic: the integration of com-
ponent structures at the phonological pole serves to symbolize the integration of
the corresponding component structures at the semantic pole. Consider the plural
noun walls. At the phonological pole, the component structures are integrated by
the suffixation of -s to wall, which involves the appropriate temporal sequencing,
syllabic organization, and minor phonetic adjustments. It is precisely the fact that
-s suffixes to wall (and not to some other noun stem) which symbolizes the fact
that the plurality it expresses is predicated of the notion wall in particular (rather
than the thing designated by some other noun in the sentence). Or to put it in
other terms, the symbolic association si, does not hold between a semantic and a
phonological structure per se – instead it associates the RELATIONSHIPS¬ between
two semantic and two phonological structures.

Figure 11.

Integration and composition work in essentially the same way at the phonologi-
cal pole and the semantic pole, but we will confine our attention to the latter. I
suggest that the integration of two component structures always involves “cor-
54 Ronald W. Langacker

respondences” being established between certain of their substructures. The

corresponding substructures provide points of overlap between the component
predications, which are necessary if a coherent composite conception is to emerge.
The composite structure is obtained by superimposing the specifications of cor-
responding substructures. In those instances where there is some conflict in their
specifications, a fully consistent composite notion cannot be formed, and the
result is what we perceive as semantic anomaly (or the violation of “selectional
restrictions”).
The semantic pole of a typical construction is sketched in Figure 12(a), which
diagrams the integration of above and the table to form the prepositional phrase
above the table (I will ignore the semantic contribution of the definite article).
[ABOVE] profiles a stative relation in oriented space between two things, each
characterized only schematically. [TABLE] profiles a thing characterized in far
greater detail with respect to numerous domains; purely for sake of diagramatic
convenience, it is represented by a mnemonic shape specification. The integra-
tion of these component predications is effected by a correspondence established
between the landmark of [ABOVE] and the profile of [TABLE] (correspondences are
represented by dotted lines). By superimposing the specifications of these corre-
sponding substructures, and adopting the relational profile of [ABOVE], we obtain
the composite predication (ABOVE -¬TABLE), which designates a stative relation
involving a schematic trajector and a specific landmark. Note that the compositional
process results in “vertical” correspondences between elements of the component
and composite structures, in addition to the “horizontal” correspondence(s) link-
ing the components.11

Figure 12.

Semantics is not fully compositional. When first assembled, an expression’s com-

posite structure may invoke a domain or incorporate specifications (e.g. the orien-
 Chapter 1: Cognitive Grammar 55

tation of the table) that are not predictable from the component structures or other
conventional units. Because such specifications are part of how the expression
is actually understood in context, and may well be included in its conventional
semantic value should the expression be established as a unit, it is arbitrary to
exclude them from the purview of semantic analysis. There are nevertheless con-
ventional patterns of composition that determine central aspects of a composite
structure’s organization. These are represented in the grammar by constructional
schemas, whose internal structure is parallel to that of the specific expressions
which instantiate them. For example, the grammar of English includes a schema for
the prepositional-phrase construction. Its phonological pole specifies the contiguity
and linear ordering of the preposition and its noun-phrase object; its semantic pole,
given in Figure 12(b), is precisely analogous to 12(a) except that the component
and composite structures are schematic rather than specific. The first compo-
nent is schematic for the class of prepositions. Basically, it is identified only as a
stative relation whose trajector and primary landmark are both things. The other
component is the noun-phrase schema: it profiles a thing, and implies additional
content (labeled X), but does not itself specify the nature of this content. As in
the specific structure 12(a), a correspondence holds between the landmark of P
and the profile of NP, and the composite structure is formed by superimposing the
specifications of these correspondents (and adopting the relational profile of P).
Speakers can employ this constructional schema in the computation and evalua-
tion of novel expressions. It serves as the structural description of any expression
which it categorizes when so employed.
This construction has various properties that can be regarded as prototypical.
There are just two component structures, one of them relational and the other
nominal. A correspondence holds between two highly prominent substructures:
the profile of the nominal predication, and the primary landmark (one facet of
the profile) of the relational predication. Moreover, there is a substantial asym-
metry in the degree of specificity at which the predications characterize the cor-
responding elements – the landmark of [ABOVE] is quite schematic, whereas by
comparison the profile of [TABLE] is specified in considerable detail. I have indi-
cated this diagramatically by an arrow (standing for a relationship of schematicity)
between [ABOVE]’s landmark and the other predication as a whole. Finally, it is
the relational predication which lends its profile to the composite structure (i.e.
above the table designates a stative relation, not a thing). I thus refer to [ABOVE]
in 12(a) as the construction’s “profile determinant”, and make this role explicit
by putting the box enclosing this predication in heavy lines.
None of the properties just cited is invariant except the existence of at least
one correspondence between substructures of the components. By recognizing
these properties as prototypical rather than imposing them as absolute require-
ments, we obtain the flexibility needed to accommodate the full range of attested
56 Ronald W. Langacker

construction types. It is probably necessary, for example, to allow more than just
two component structures at a particular level of constituency (e.g. for coordi-
nate expressions such as X, Y, and Z). It need not be the case that one component
structure is relational and the other nominal – in fact, there need be no relational
component at all. Appositional constructions involving two nominal predica-
tions, e.g. my good friend Ollie North, are straightforwardly accommodated in
this framework by means of a correspondence established between the nominal
profiles. In all the examples cited so far, the corresponding elements have been
things that either constitute or are included within the profile of the component
structure. Often, however, the correspondents are relational substructures, and
they need not be in profile. Consider once more the sense of gone diagramed in
Figure 4(c). The component structures are [GO], which designates a process, and
one particular semantic variant of the past-participial morpheme. This particular
predication profiles the final state of an otherwise unprofiled process that con-
stitutes its base. The participial morpheme itself characterizes this process quite
schematically; only in combination with a verb stem is the nature of the process
made specific. Their integration is effected by a correspondence between the
specific process profiled by [GO] and the schematic process functioning as the
base within the participial predication. By superimposing their specifications,
and adopting the profile contributed by the grammatical morpheme, we obtain a
composite structure that profiles just the final state of the process [GO].

Figure 13.

A factor we have not yet considered is “constituency”, which pertains to the order
in which symbolic structures are progressively assembled into larger and larger
 Chapter 1: Cognitive Grammar 57

composite expressions. Clearly, the composite structure resulting from the integra-
tion of component structures at one level of organization can itself be employed
as a component structure at the next higher level, and so on indefinitely. In Figure
13, for example, the composite structure (ABOVE-TABLE) from Figure 12(a) func-
tions as a component structure, combining with [LAMP] to derive the composite
semantic value of the noun phrase the lamp above the table. At this second level
of organization, it is the schematic trajector of the relational predication that is
put in correspondence with the profile of the nominal predication; moreover, it
is this latter which functions as the construction’s profile determinant. The com-
posite structure (LAMP-ABOVE-TABLE) consequently designates the lamp, not its
locative relationship vis-à-vis the table, though this relationship is included as a
prominent facet of its base.
Some grammatically significant observations can be made on the basis of these
examples. For one thing, we see that either a relational or a nominal predication
is capable of serving as the profile determinant in a construction. In Figure 12, it
is the relation [ABOVE] which contributes the profile of the composite expression,
whereas in Figure 13 it is the nominal [LAMP]. Moreover, the constructs now at
our disposal permit workable and revealing characterizations of certain funda-
mental grammatical notions that have long been problematic, namely “head”,
“modifier”, and “complement”. At a given level of organization, a construction’s
head can be identified with its profile determinant. Above is thus the head within
the prepositional phrase above the table, whereas lamp is the head within the
noun phrase the lamp above the table. In appositional expressions like my good
friend Ollie North there is no real basis for singling out either component noun
phrase as the head. But that is precisely what we expect: because their profiles
correspond, and each corresponds to the profile of the composite structure, it is
arbitrary to say that the latter inherits its profile from either one of the component
structures (as opposed to the other).
To the extent that one component structure, taken as a whole, serves to elaborate
a salient substructure within the other, I will speak of the elaborating component
as being “conceptually autonomous”, and the elaborated component as “concep-
tually dependent”. In Figure 12(a), then, [TABLE] is conceptually autonomous
with respect to [ABOVE] because it elaborates the latter’s schematic landmark. In
Figure 13, similarly, [LAMP] is autonomous by virtue of elaborating the schematic
trajector of the dependent predication (ABOVE-TABLE). The notions modifier and
complement can now be characterized explicitly in a way that reconstructs the
normal usage of these traditional terms: a “modifier” is a conceptually dependent
predication that combines with a head, whereas a “complement” is a conceptually
autonomous predication that combines with a head. The table is consequently a
complement (or “argument”) of above in above the table, and this entire prepo-
sitional phrase functions as a modifier of lamp in the lamp above the table. What
58 Ronald W. Langacker

about appositional constructions? Because there is no basis for recognizing either

component structure as the head (and often no autonomous/dependent asym-
metry), the definitions are correctly found to be inapplicable. In my good friend
Ollie North, neither my good friend nor Ollie North is considered a modifier or
a complement of the other.
This conception of grammatical structure has numerous descriptive advantages,
only a few of which will be noted at this juncture. One advantage is that it readily
accommodates variability of constituency, which is in fact quite common. The
present framework does not posit phrase trees of the sort familiar from generative
studies, nor does it rely on phrase structure configurations for the definition of
grammatical relations. Constituency is simply the sequence in which component
symbolic structures are progressively assembled into more and more elaborate
composite expressions. Though a specific order of assembly commonly becomes
conventionalized as the sole or default-case sequence, the choice is not inher-
ently critical in this model, because alternate constituencies often permit the
same composite structure to be derived. Moreover, because grammatical relations
are not defined in configurational terms, a unique constituency is not essential.
What identifies the table as the object of above in above the table, for example,
is the fact that the noun phrase elaborates the preposition’s landmark. Though
constituency happens to be invariant in this case, the critical factor in defining
the prepositional-object relation is the correspondence established between the
landmark of the preposition and the profile of the noun phrase.
We can better appreciate these points with regard to sentences like the ones
in (15).

(15) a. Alice likes liver.

b. Liver Alice likes.
c. Alice likes, but most people really hate, braised liver.

Sentence (15a) exhibits the normal, default-case NP + VP constituency of Eng-

lish clauses: liver elaborates the schematic landmark of likes at the first level of
constituency, yielding a processual predication with a specified landmark and
schematic trajector; Alice then elaborates the trajector of likes liver at the sec-
ond level to derive a process predication whose trajector and landmark are both
specific. It should be apparent, however, that the same composite structure will
result if the constituents combine in the opposite order, with Alice elaborating the
schematic trajector of likes, and then liver the schematic landmark of Alice likes.
This alternative constituency is available for exploitation, with no effect on gram-
matical relations, whenever special factors motivate departure from the default-
case arrangement. Two such factors are illustrated here. In (15b) we observe the
topicalization of the direct-object noun phrase, normally described as a movement
 Chapter 1: Cognitive Grammar 59

transformation. There is no need in this framework to derive this sentence type

by transformation – it can be assembled directly through the alternate compo-
sitional path. The second type of situation arises in conjoined structures when
two verbs have different subjects but share the same object, as in (15c). In lieu of
the transformational process of “Right Node Raising”, which supposedly derives
this type of sentence from conjoined clauses of normal NP + VP constituency,
we can once again assemble the overt structure directly. The two subject-verb
constituents are put together first and then combined in a coordinate structure.
A direct object NP is subsequently added, being integrated simultaneously with
each conjunct through a correspondence between its profile and the conjunct’s
relational landmark.
Also eliminable in this framework is the raising rule needed in certain trans-
formational accounts (e.g. Keyser and Postal 1976) to handle agreement between
a subject and an auxiliary verb, as in (16).

(16) The lamp is above the table.

The rationale for a raising rule goes something like this: (i) a verb is assumed to
agree with its own subject; (ii) the lamp is not the logical subject of be, which – if
anything – has a clause for its underlying subject; (iii) hence, to account for agree-
ment, some rule must raise the lamp from its position as subject of above and make
it the subject of be. However the need for such a rule is obviated given a proper
analysis of be and a suitably flexible conception of grammatical constructions.
The semantic pole of (16) is outlined in Figure 14.12 Pivotal to the analysis
is the semantic value attributed to be, of which three main features are relevant.
First, be is a true verb, i.e. a symbolic expression that profiles a process. Second,
all the component states of the designated process are construed as being identi-
cal; this is indicated by the dotted correspondence lines internal to [BE] that link
the three states which are explicitly represented (additional correspondence lines
specify that the trajector is the same from one state to the next, as is the landmark).
Third, apart from this specification of identity, the profiled process is maximally
schematic. Be is one of numerous verbs in English which designate a process con-
sisting of the extension through time of a stable situation (see Langacker 1991a:
Chapter 3) – others include have, resemble, like, know, contain, slope, exist, and
so on – but it abstracts away from the specific content that distinguishes these
predications from one another. In summary, [BE] follows through time, by means
of sequential scanning, the evolution of a situation that is construed as being stable
but not further specified (except for its relational character).
Any single component state of [BE] constitutes a schematic stative relation.
At the first level of constituency in Figure 14, the more specific stative relation
(ABOVE-TABLE) is put in correspondence with a representative state of [BE], the
60 Ronald W. Langacker

latter serving as profile determinant. The result is the composite predication

(BE-ABOVE-TABLE), which is like [BE] except that all the specifications inherited
from (ABOVE-TABLE) are attributed to the situation followed sequentially through
time. Observe that the landmark of (BE-ABOVE-TABLE) is now specific, whereas its
trajector remains schematic. At the second level of constituency, this schematic
trajector is elaborated by [LAMP] to derive the composite structure (LAMP-BE-
ABOVE-TABLE), which represents the composite meaning of the full sentence. It
profiles the extension through time of a stable situation in which the lamp and
the table participate in a particular locative relationship.

Figure 14.

Observe that the sentence is assembled directly, in accordance with its surface
constituency. In particular, there is no “raising” rule which derives it from a
hypothetical underlying structure by changing the grammatical relation of the
subject NP. But does the lamp function as the subject of be, as their agreement
 Chapter 1: Cognitive Grammar 61

presumably requires? It certainly does, given the way grammatical relations are
defined in this framework. A subject NP is one which elaborates the schematic
trajector of a relational predication by virtue of a correspondence established
between that trajector and its own profile. With respect to Figure 14, note first
that [BE] does in fact have a schematic trajector, characterized as both a thing (not
a clause) and a relational participant. Moreover, [BE]’s trajector does correspond
to the profile of the lamp, when both horizontal and vertical correspondences
are taken into account: the profile of [LAMP] corresponds to the trajector of (BE-
ABOVE-TABLE), which in turn corresponds vertically to the trajector of [BE]. It is
simply incorrect, in this analysis, to claim that be has no nonclausal subject, or
that the lamp is not its “logical” subject in (16). With no special apparatus, the
analysis establishes a relationship between the lamp and be which is perfectly
adequate as a basis for agreement.
Finally, the analysis permits a simple and natural account of sentences like
(17b), in which an auxiliary verb functions as a pro form:

(17) a. Q: What is above the table?

b. A: The lamp is.

As highly schematic process predications, auxiliary verbs are perfectly suited to

this role, and sentences of this type are derivable without any deletion operation.
Because constituency is potentially variable in this framework, we can derive
(17b) just by combining the lamp and be directly. A correspondence is established
between the profile of the former and the schematic trajector of the latter. Be is
the profile determinant, so the composite structure designates a process involving
the evolution of a stable situation through time. Apart from its trajector, identified
as the lamp, this situation is characterized only schematically.

7. Conclusion

This initial presentation of cognitive grammar has itself been quite schematic. I
do however hope to have shown that currently predominant linguistic theories do
not represent the only possible way of conceiving the nature of language structure
and linguistic investigation. By taking a radically different perspective on ques-
tions of meaning and grammar, it is possible to formulate a coherent descriptive
framework which promises to be both cognitively realistic and linguistically
well-motivated.
62 Ronald W. Langacker

Notes
1. Observe that designation, in my technical sense of the term, does not pertain to the
relation between a linguistic expression and the world – rather it is a relationship
holding between a cognitive domain as a whole and certain of its subparts. I do not
know whether profiling reduces to any independently established cognitive phenom-
enon. Possibly it constitutes one level of figure/ground organization, but not every
figure is a designatum.
2. In these expressions eye is evidently construed as the eye region, not the eyeball
itself.
3. The constructs needed to make this notion of subjectivity/objectivity precise are
introduced in Langacker 1985 and 1987, Chapters 3 and 7. For vantage point and
orientation, see Langacker (1991a: Chapter 2) and Vandeloise 1984.
4. Goldsmith 1980 presents a very similar analysis.
5. The importance of conventionality should be emphasized. Often a speaker is led to
employ a particular image simply because an alternative construction, which might
seem more appropriate, happens not to be conventionally established. For instance,
many verbs of transfer (e.g. transfer itself) are not employed in the double-object con-
struction; the to-construction represents the speaker’s only option with such verbs.
6. At this level of organization, we can ignore the fact that sharpen is morphematically
complex. The double-headed arrow labeled (e) in Figure 7 indicates identity of the
associated structures.
7. Fuller discussion is provided in Langacker (1991a: Chapter 10). (See also Langacker
1987: Chapter 11.)
8. By reversing the trajectory/landmark assignation, we obtain the predicate
[BELOW].
9. I omit the dashed line standing for the profiled interconnections, because the nature
of these interconnections is implicit in the position of the major participants within
the diagrams. Note that I regard these diagrams as heuristic in character, not as for-
mal objects. They are analogous to the sketch a biologist might draw to illustrate the
major components of a cell and their relative position within it.
10. Only for convenience do I speak of discrete states – a process is more accurately
viewed as continuous.
11. The component structures are enclosed in boxes, to indicate that above and the table
have the status of units. Closed curves surround the composite structure and the con-
struction as a whole on the presumption that above the table is a novel expression (in
the text, parentheses serve this purpose).
12. Omitted are the semantic contributions of the definite article and the verb inflection
on be. Note that our concern is not the nature of agreement (cf. Langacker 1991a:
Chapter 11), but rather the issue of whether the lamp can be considered the subject
of be in accordance with assumption (i).
 Chapter 1: Cognitive Grammar 63

References
Barwise, Jon and John Perry
1983 Situations and Attitudes. Cambridge, MA, and London: MIT
Press/Bradford.
Bever, Thomas G. and Peter S. Rosenbaum
1970 Some lexical structures and their empirical validity. In Readings in English
Transformational Grammar, Roderick A. Jacobs and Peter S. Rosenbaum
(eds.), 3–19. Waltham, MA: Ginn.
Bybee, Joan L.
1988 Semantic substance vs. contrast in the development of grammatical meaning.
In Proceedings of the Annual Meeting of the Berkeley Linguistics Society
14: 247–264.
Casad, Eugene H.
1982 Cora locationals and structured imagery. Unpublished Ph.D. dissertation,
University of California, San Diego.
1988 Conventionalization of Cora locationals. In Topics in Cognitive Linguistics, Bry-
gida Rudzka-Ostyn (ed.), 345–378. Amsterdam/Philadelphia: Benjamins.
Chafe, Wallace L.
1970 Meaning and the Structure of Language. Chicago: University of Chicago
Press.
Cook, Kenneth W.
1988 A cognitive analysis of grammatical relations, case, and transitivity in Samoan.
Unpublished Ph.D. dissertation, University of California, San Diego.
1989 A cognitive account of the Samoan lavea and galo verbs. Linguistic Agency
University of Duisburg.
Cruse, [Link]
1979 On the transitivity of the part-whole relation. Journal of Linguistics 15:
29–38.
Fauconnier, Gilles
1985 Mental Spaces: Aspects of Meaning Construction in Natural Language. Cam-
bridge, MA and London: MIT Press/Bradford.
Fillmore, Charles J.
1982 Frame Semantics. In Linguistics in the Morning Calm, Linguistic Society of
Korea (ed.), 111–137. Seoul: Hanshin.
Givón, Talmy
1979 The time-axis phenomenon. Language 49: 890–925.
1984 Syntax: A Functional-Typological Introduction. Volume 1. Amsterdam/Phila-
delphia: Benjamins.
1989 Mind, Code, and Context: Essays in Pragmatics. Hillsdale, NJ: Erlbaum.
Green, Georgia M.
1974 Semantics and Syntactic Regularity. Bloomington: Indiana University
Press.
Haiman, John
1980 Dictionaries and encyclopedias. Lingua 50: 329–357.
1983 Iconic and economic motivation. Language 59: 781–819.
64 Ronald W. Langacker

1985 Natural Syntax: Iconicity and Erosion. Cambridge: Cambridge University

Press.
Hawkins, Bruce W.
1984 The semantics of English spatial prepositions. Unpublished Ph.D. dissertation,
University of California, San Diego.
1988 The natural category MEDIUM: An alternative to selection restrictions and
similar constructs. In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn
(ed.), 231–270. Amsterdam/Philadelphia: Benjamins.
Jackendoff, Ray
1983 Semantics and Cognition. (Current Studies in Linguistics 8.) Cambridge, MA:
MIT Press.
Janda, Laura A.
1984 A semantic analysis of the Russian verbal prefixes za-, pere-, do-, and ot-.
Unpublished Ph.D. dissertation, University of California, Los Angeles.
1988 The mapping of elements of cognitive space onto grammatical relations:
An example from Russian verbal prefixation. In Topics in Cognitive Lin-
guistics, Brygida Rudzka-Ostyn (ed.), 327–343. Amsterdam/Philadelphia:
Benjamins.
1993 A Geography of Case Semantics: The Czech Dative and the Russian Instru-
mental. Berlin/New York: Mouton de Gruyter.
Johnson, Mark
1987 The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Rea-
son. Chicago/London: University of Chicago Press.
Johnson-Laird, Philip N.
1983 Mental Models. Cambridge, MA: Harvard University Press.
Katz, Jerrold J. and Jerry A. Fodor
1963 The structure of a semantic theory. Language 39: 170–210.
Kemmer, Suzanne E.
1988 The middle voice: A typological and diachronic study. Unpublished Ph.D.
dissertation, Stanford University.
Keyser, Samuel Jay and Paul M. Postal
1976 Beginning English Grammar. New York: Harper & Row.
Klima, Edward S.
1964 Negation in English. In The Structure of Language: Readings in the Philosophy
of Language, Jerry A. Fodor and Jerrold J. Katz (eds.), 246–323. Englewood
Cliffs, NJ: Prentice-Hall.
Kosslyn, Stephen Michael
1980 Image and Mind. Cambridge, MA: Harvard University Press.
Lakoff, George
1982 Categories: An essay in cognitive linguistics. In Linguistics in the Morning
Calm, Linguistic Society of Korea (ed.), 139–193. Seoul: Hanshin.
1987 Women, Fire, and Dangerous Things: What Categories Reveal about the
Mind. Chicago/London: University of Chicago Press.
Lakoff, George and Mark Johnson
1980 Metaphors We Live By. Chicago/London: University of Chicago Press.
 Chapter 1: Cognitive Grammar 65

Lakoff, George and Mark Turner

1989 More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago/London:
University of Chicago Press.
Langacker, Ronald W.
1976 Semantic representations and the linguistic relativity hypothesis. Foundations
of Language 14: 307–357.
1981 The integration of grammar and grammatical change. Indian Linguistics 42:
82–135.
1982 Remarks on English aspect. In Tense-Aspect: Between Semantics and Prag-
matics, Paul J. Hopper (ed.), 265–304. (Typological Studies in Language 1.)
Amsterdam/Philadelphia: Benjamins.
1985 Observations and speculations on subjectivity. In Iconicity in Syntax, John
Haiman (ed.), 109–150. (Typological Studies in Language 6.) Amsterdam/
Philadelphia: Benjamins.
1987 Foundations of Cognitive Grammar. Volume 1: Theoretical Prerequisites.
Stanford: Stanford University Press.
1988a An overview of cognitive grammar. In Topics in Cognitive Linguistics, Brygida
Rudzka-Ostyn (ed.), 3–48. Amsterdam/Philadelphia: Benjamins.
1988b A view of linguistic semantics. In Topics in Cognitive Linguistics, Brygida
Rudzka-Ostyn (ed.), 49–90. Amsterdam/Philadelphia: Benjamins.
1991a Concept, Image, and Symbol. The Cognitive Basis of Grammar. Berlin/New
York: Mouton de Gruyter.
1991b Foundations of Cognitive Grammar. Volume 2: Descriptive Application.
Stanford: Stanford University Press.
Lindner, Susan
1981 A lexico-semantic analysis of English verb-particle constructions with UP and
OUT. Unpublished Ph.D. dissertation, University of California, San Diego.
1982 What goes up doesn’t necessarily come down: The ins and outs of opposites.
In Papers from the Regional Meeting of the Chicago Linguistic Society 18:
305–323.
Miller, George A. and Philip N. Johnson-Laird
1976 Language and Perception. Cambridge, MA: Harvard/Belknap.
Moore, Terence and Christine Carling
1982 Language Understanding: Towards a Post-Chomskyan Linguistics. New York:
St. Martin’s Press.
Oehrle, Richard T.
1977 Review of Green 1974. Language 53: 198–208.
Perlmutter, David M.
1978 Impersonal passives and the unaccusative hypothesis. In Proceedings of the
Annual Meeting of the Berkeley Linguistics Society 4: 157–189.
Poteet, Stephen
1987 Paths through different domains: A cognitive grammar analysis of Mandarin
dào. In Proceedings of the Annual Meeting of the Berkeley Linguistics Society
13: 408–421.
Rice, Sally
1987a Towards a transitive prototype: Evidence from some atypical English passives.
66 Ronald W. Langacker

In Proceedings of the Annual Meeting of the Berkeley Linguistics Society

13: 422–434.
1987b Towards a cognitive model of transitivity. Unpublished Ph.D. dissertation,
University of California, San Diego.
1988 Unlikely lexical entries. In Proceedings of the Annual Meeting of the Berkeley
Linguistics Society 14: 202–212.
Shepard, Roger N.
1978 The mental image. American Psychologist 33: 125–137.
Smith, Michael B.
1985a An analysis of German dummy subject constructions. In Proceedings of the
Annual Meeting of the Pacific Linguistics Conference 1: 412–425.
1985b Event chains, grammatical relations, and the semantics of case in German.
In Papers from the Regional Meeting of the Chicago Linguistic Society 21:
388–407.
1987 The semantics of dative and accusative in German: An investigation in cogni-
tive grammar. Unpublished Ph.D. dissertation, University of California, San
Diego.
1989 Cases from conceptual categories: Evidence from German. Linguistic Agency
University of Duisburg.
Sweetser, Eve E.
1984 Semantic structure and semantic change: A cognitive linguistic study of modal-
ity, perception, speech acts, and logical relations. Unpublished Ph.D. disserta-
tion, University of California, Berkeley.
1987 Metaphorical models of thought and speech: A comparison of historical direc-
tions and metaphorical mappings in two domains. In Proceedings of the Annual
Meeting of the Berkeley Linguistics Society 13: 446–459.
1988 Grammaticalization and semantic bleaching. In Proceedings of the Annual
Meeting of the Berkeley Linguistics Society 14: 389–405.
Talmy, Leonard
1975 Sematic and syntax in motion. In Syntax and Semantics 4, John Kimball (ed.),
181–238. New York: Academic Press.
1977 Rubber-sheet cognition in language. In Papers from the Regional Meeting of
the Chicago Linguistic Society 13: 612–628.
1978 Figure and ground in complex sentences. In Universals of Human Language,
volume 4: Syntax, Joseph H. Greenberg (ed.), 625–649. Stanford: Stanford
University Press.
1983 How language structures space. In Spatial Orientation: Theory, Research,
and Application, Herbert Pick and Linda Acredolo (eds.), 225–282. New York:
Plenum Press.
1985a Force dynamics in language and thought. In Papers from the Parasession on
Causatives and Agentivity, William H. Eilfort et al. (eds.), 293–337. Chicago:
Chicago Linguistic Society.
1985b Lexicalization patterns: Semantic structure in lexical forms. In Language
Typology and Syntactic Description. Volume 3: Grammatical Categories and
the Lexicon, Timothy Shopen (ed.), 57–149. Cambridge: Cambridge University
Press.
 Chapter 1: Cognitive Grammar 67

1988a Force dynamics in language and cognition. Cognitive Science 12: 49–100.
1988b The relation of grammar to cognition. In Topics in Cognitive Linguistics, Bry-
gida Rudzka-Ostyn (ed.), 165–205. Amsterdam/Philadelphia: Benjamins.
Traugott, Elizabeth C.
1982 From propositional to textual and expressive meanings: Some semantic-prag-
matic aspects of grammaticalization. In Perspectives on Historical Linguistics,
Winfred P. Lehmann and Yakov Malkiel (eds.), 245–271. Amsterdam/Phila-
delphia: Benjamins.
1986 From polysemy to internal semantic reconstruction. In Proceedings of the
Annual Meeting of the Berkeley Linguistics Society 12: 539–550.
1988 Pragmatic strengthening and grammaticalization. In Proceedings of the Annual
Meeting of the Berkeley Linguistics Society 14: 406–416.
Tuggy, David
1980 ¡Ethical dative and possessor omission sí, possessor ascension no! In Work
Papers of the Summer Institute of Linguistics, University of North Dakota
24: 97–141.
1981 The transitivity-related morphology of Tetelcingo Nahuatl: An exploration in
space grammar. Unpublished Ph.D. dissertation, University of California, San
Diego.
1986 Noun incorporations in Nahuatl. In Proceedings of the Annual Meeting of the
Pacific Linguistics Conference 2: 455–469.
1988 Náhuatl causative/applicatives in cognitive grammar. In Topics in Cognitive
Linguistics, Brygida Rudzka-Ostyn (ed.), 587–618. Amsterdam/Philadelphia:
Benjamins.
1989 The affix-stem distinction in Orizaba Náhuatl. Linguistic Agency University
of Duisburg.
Vandeloise, Claude
1984 Description of space in French. Unpublished Ph.D. dissertation, University of
California, San Diego.
Wierzbicka, Anna
1988 The Semantics of Grammar. (Studies in Language Companion Series 18.)
Amsterdam/Philadelphia: Benjamins.
Chapter 2
Grammatical construal

The relation of grammar to cognition1

Leonard Talmy

0. Introduction

A fundamental design feature of language is that it has two subsystems which

can be designated as the grammatical and the lexical (as these are characterized
below). Why is there this universal bifurcation when, in principle, a language
could be conceived having only a single system, the lexical? The explanation in
this paper is that the two subsystems have distinct semantic functions, ones that
are indispensable and complementary. To develop this account further, we must
first note that we take a sentence (or other portion of discourse) to evoke in the
listener a particular kind of experiential complex, here to be termed a “cogni-
tive representation” or “CR”.2 Now, the grammatical and lexical subsystems in
a sentence seem generally to specify different portions of a CR. Together, the
grammatical elements of a sentence determine the majority of the structure of
the CR, while the lexical elements together contribute the majority of its content.
Lexical elements do incorporate some of the same structural indications that
grammatical elements express, but when the two are in association or in conflict
within a sentence, it is generally always the grammatical elements’ specifications
of structure that are determinative.3 The grammatical specifications in a sentence,
thus, provide a conceptual framework or, imagistically, a skeletal structure or
scaffolding, for the conceptual material that is lexically specified.
More generally, across the spectrum of languages, the grammatical elements
that are encountered, taken together, specify a crucial set of concepts. This set
is highly restricted: only certain concepts appear in it, and not others, as seen
below. The purport of the present paper is that this set of grammatically specified
notions collectively constitutes the fundamental conceptual structuring system of
language. That is, this cross-linguistically select set of grammatically specified
concepts provides the basic schematic framework for conceptual organization
within the cognitive domain of language.
Thus, grammar, broadly conceived, is the determinant of conceptual struc-
ture within one cognitive domain, language, and as such is the main object of

Originally published in 1988 in Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.),

165–205. Amsterdam/Philadelphia: John Benjamins. Reprinted with permission.
70 Leonard Talmy

this paper’s study. But such a study directly opens out into a broader investiga-
tion across other cognitive domains, such as visual perception and reasoning,
as discussed at the end of the paper. That is, the greater issue, toward which the
present study ultimately aims, is the general character of conceptual structure in
human cognition.
The present investigation into the semantics of grammar is of a scope that
follows in a progression from previous types of study. These have mostly been
either an in-depth semantic analysis of a selected grammatical element (or class
of elements) of particular interest within a language, e.g., the Turkish evidential
suffix -mis (Slobin and Aksu 1982); or an exposition of the meanings and func-
tions of all the grammatical elements of a single language, say, as in a grammar
of Dyirbal (Dixon 1972); or a cross-linguistic typology of the different kinds
of grammatical devices used for a single semantic function, say, to indicate the
interrogative (Ultan 1978). Some previous work has also treated broader issues
of grammatical meaning (Sapir 1921; Boas 1938; Whorf 1956, Jakobson 1971).
But the present study is perhaps the first to address grammatical expression in
language at the superordinate level, with the aim of determining the semantic
and cognitive properties and functions of this structural component of language
as a whole.4
The terms “grammatical” and “lexical” as employed here require some imme-
diate elaboration. The distinction between the two is made formally – i.e., with-
out reference to meaning – on the basis of the traditional linguistic distinction
between “open-class” and “closed-class”. A class of morphemes is considered
open if it is quite large and readily augmentable relative to other classes. A class
is considered closed if it is relatively small and fixed in membership. We can
identify the particular classes belonging to these two types. The open classes of
elements – i.e., the lexical classes – are the roots of nouns, verbs, and adjectives.5
Everything else is closed-class – and is here considered to be, quite generally,
“grammatical”. Among the overt elements of this type are such bound forms as
inflections and derivations, such free forms as determiners, prepositions, conjunc-
tions, and particles, and perhaps also such suprasegmental forms as intonation
patterns. Included among abstract, or implicit, closed-class forms are grammatical
categories and grammatical relations, word order, and perhaps also paradigms
and “zero” forms. Additionally here are regular combinations of simpler closed-
class forms, tending to have a unified or integrated semantic function – what are
below called “grammatical complexes”, including grammatical constructions and
syntactic structures.6
The issues presented in this introduction are treated below in three sections.
Section 1 examines the notions specified by a heuristic sampling of grammatical
elements, outlines the kinds of constraints on such notions, proposes a property
held in common by such notions but largely absent from excluded notions, and
 Chapter 2: Grammatical construal 71

contrasts such grammatically specified notions with ones that are lexically speci-
fied. Section 2 presents a number of categories in which grammatically specified
notions are seen to pattern, as well as broader conceptual systems in which these
categories in turn participate, ending with the identification of four compre-
hensive “imaging systems”. This section, further, examines the interactions of
grammatical specifications with lexical specifications within categories and the
nesting of such interactions across categories, as well as the cognitive processes
that accompany these interactions. And Section 3 presents an explanation of the
function of grammatical specification, as well as possibilities of its relations to
other cognitive systems.

1. The nature of grammatically specified notions

In this section we examine a small sampling of grammatical forms for the particular
component notions that they specify. The sample will give a heuristic indication
of the kinds of notions that get grammatically specified as well as of the kinds
of notions that possibly never do. By contrast, it will be seen that the excluded
kinds can be readily specified by lexical elements. A particular property will be
seen to run through most of the grammatical notions. To indicate this property at
the outset, it is preponderantly the case that grammatical specifications of struc-
ture are relativistic or topology-like, and exclude the absolute or the metrically
Euclidean. Finally, a systematic difference is shown between the characteristics
of grammatically specified notions and of lexically specified ones.
We begin with a simple demonstration that the concepts specified by gram-
matical forms are constrained in two ways: as to their categories and as to the
membership of these categories. Many languages have inflections on the noun
that specify the “number” of the object referred to by the noun, for example its
‘singularity’ or ‘plurality’, like the English -෬ and -s. By contrast, no languages
appear to have inflections that specify the “color” of the object referred to by
a noun, e.g., its ‘redness’ or ‘blueness’. Here, single quotes enclose “notions”,
while double quotes enclose categories of notions. The “number” category can be
specified grammatically and in that form is readily seen to play a structuring role
in a CR.7 The “color” category is perhaps never found specified by grammatical
elements, though it is readily found specified by lexical elements, e.g., English
red and blue. Further, though, even within a conceptual category acceptable for
grammatical expression, there are great constraints on the particular notions that
can be specified. Thus, “number” notions that are expressed grammatically include
little more than ‘singular’, ‘dual’, ‘trial’, ‘plural’, and ‘paucal’. They apparently
never include, say, ‘even’, ‘odd’, ‘dozen’, or ‘numerable’, whereas such notions,
again, can be specified lexically, as shown by the words just used.
72 Leonard Talmy

Given such constraints on grammatically specifiable notions, we can seek

properties that hold in common for included notions but need not apply to excluded
notions. In this regard, consider a deictic like the English this or that as in This/
That chair is broken. A closed-class element of this type specifies the location of
an indicated object as being, in effect, on the speaker-side or the non-speaker-side
of a conceptual partition drawn through space (or time or other qualitative dimen-
sion). This integral specification can be analyzed as containing the component
notions enclosed by quotes in (I):

(1) (a, b) a ‘partition’ that divides a space into ‘regions’/‘sides’

(c–e) the ‘locatedness’ (a particular relation) of a ‘point’ (or object ideal-
izable as a point) ‘within’ a region
(f, g) (a side that is the) ‘same as’ or ‘different from’
(h, i) a ‘currently indicated’ object and a ‘currently communicating’ entity

Notions that might at first be ascribed to such deictics, such as of distance or per-
haps size, prove not to apply, on the evidence of sentence-pairs like (2):

(2) a. This speck is smaller than that speck.

b. This planet is smaller than that planet.

The scenes referred to by (2a) and (b) differ greatly, involving tiny objects mil-
limeters apart or huge objects parsecs apart. But the sentences differ only lexi-
cally, not grammatically. Hence, the scenes’ differences as to the magnitude of
size or distance must arise from the lexical elements, they cannot be traced to
the deictics (or other grammatical elements) in the sentences. Thus, the notions
specified by a this or a that are abstracted away from any particularities of mag-
nitude and so, to this extent, are genuinely topological. Their specification of a
conceptual partition remains constant, but this partition’s distance can – by the
characterization of topology as “rubber-sheet geometry” – be “stretched” indefi-
nitely without challenge to any semantic constraints of the deictics. This find-
ing about deictics alerts us to noticing whether any grammatical elements make
specifications about magnitude. A spot check through English and various other
languages suggests that – while there are grammatical specifications for relative
magnitude8 – there are possibly never any for absolute or quantified magnitude,
whether of size, distance, or other parameters. We can provisionally conclude
that the referents of grammatical elements have the topological property of being
“magnitude-neutral”.
For another case, consider the type of adposition that specifies, for a mov-
ing object, certain characteristics of path and of reference-point or -frame. An
example of this type is English through as used, for instance, in I walked through
 Chapter 2: Grammatical construal 73

the woods. In this usage, through specifies, broadly, ‘motion along a line that is
within a medium’. The component notions contained here include those in (3):

(3) (a) ‘motion’

(b–e) which can be understood as ‘one-to-one correspondences’ between
‘adjacent’ points of ‘space’ and adjacent points of ‘time’
(f) motion that describes a ‘line’ (i.e., a ‘linear extent’)
(g) the locatedness of a line within a ‘medium’
(h, i) a medium, i.e., a region of three-dimensional space set apart by
the locatedness within it of ‘material’ that is in a ‘pattern of distribution’
with properties and a range of variation still to be determined

It can be first observed, from a sentence-pair like (4), that the concept specified
by through is indifferent to particulars of shape or contour in the linear path
described by the moving object. This is evident here because, as before, the two
sentences differ only lexically, not grammatically – they both use through while
referring to different path contours. Another cross-linguistic spot check of closed-
class elements suggests that they largely have this further topological property
of being “shape-neutral”.

(4) a. I zig-zagged through the woods.

b. I circled through the woods.

With a sentence pair like (5), it can be further determined that the ‘rate’ of motion
is not specified by through, a finding that also appears quite general among gram-
matical elements. And (6) shows that through, again like grammatical elements
generally, excludes specification of the ‘kind of material’ involved – here, com-
prising the “medium” – and of the ‘sensorimotor characteristics’ attendant on
executing the action involved – as, here, those attendant on wading in liquid vs.
weaving amidst obstacles. Thus, it can be further held that grammatical elements
are generally rate-neutral, material-neutral, and sense/motor-neutral.

(5) (a/b) I crept / dashed through the woods.

(6) (a/b) I walked through the water / woods.

In the aim of ascertaining any properties common to grammatically specified

notions, the notions examined above are gathered together in (7). For heuristic
purposes, the notions are provisionally divided into two groups on the basis of their
relation to topology. In group (a) are combined the notions that properly belong to
the specific mathematical system of topology and, with them, the intuitively com-
parable notions that might belong to a language-based system of topology – one
74 Leonard Talmy

that perhaps could serve as the model for the construction of a new topology-like
mathematical system.9 In group (b) are the notions that fall outside any usual
conception of topological properties. The number of notions in the first group is
fourteen, while the second has six – an indication of a preponderant propensity
for grammatical elements to specify quasi-topological notions. The ratio in this
direction is in fact improved if we consider that even several notions in group
(b) – the bottom three – resemble topological ones in the sense of involving rela-
tivistic relationships between quantities rather than absolutely fixed quantities.

(7) some notions found to be specified by grammatical elements

a. topological or topology-like b. non-topological
point singularity material
linear extent plurality space
locatedness same time
within different motion
region „adjacency“ of points medium
side one-to-one correspondence entity currently
partition pattern of distribution indicated/communicating

In the complementary aim of ascertaining any properties excluded from gram-

matical specification, the categories of notions found above not to be specified by
the elements investigated are listed in (8). Rather than topological, topology-like,
or relativistic, these notions involve Euclidean-geometric concepts – e.g., fixed
distance, size, contour, and angle – as well as quantified measure, and various
particularities of a quantity: in sum, characteristics that are absolute or fixed.

(8) some categories of notions seemingly rarely or never

specified by grammatical elements
absolute/quantified magnitude kind of material
(of distance, size, etc.) sensorimotor characteristics
shape/contour of line color
rate

The provisional conclusion to be drawn from these findings is that, if grammatical

specifications generally correspond to (linguistic-) cognitive structuring, then the
nature of that structure is largely relativistic or topological rather than absolute
or Euclidean.
This preponderant requirement for conceptual neutralities among closed-class
elements is in sharp contrast with the referential freedom of lexical items, which
can express not only structural abstractions but also wide-ranging specificities.
For example, specificity as to magnitude is seen in nouns like inch and mile; as
 Chapter 2: Grammatical construal 75

to shape, in nouns like circle, adjectives like square, and verbs like ricochet; as
to rate, in verbs like dawdle and hurry; in material, in a noun and verb like iron
and bleed; as to sensorimotor characteristics in watch and wade; and, of course,
as to color by such adjectives as red and blue.
To elaborate further the contrast between the grammatical and the lexical type
of specification, consider the full complement of both element-types in a single
whole sentence, viz., that selected in (9):

(9) A rustler lassoed the steers.

We first list the grammatical elements present in the sentence and the notions
that they specify in (10):

(10)
a. -ed ‘occurring at a time before that of the
present communication’
b. the ‘has ready identifiability for the
addressee’
c. a ‘not before in discussion or other-
wise readily identifiable for the
addressee’
d. -s ‘multiple instantiation of object’
e. a...-Ø ‘unitary instantiation of object’
f. the grammatical category ‘event character’
of “verb” for lasso
g/h. the grammatical category ‘entity character’
of “noun” for rustlerlsteer
i/j- the grammatical relations ‘agent’/’patient’ (among the pos-
of “subject’V’object” for sibilities)
rustlerlsteer
k. active voice ‘point-of-view at the agent’
1. intonation, word order, ‘the Speaker “knows” the Situation
pattern of auxiliaries to be true and asserts it’

The lexical items in the sentence have specifications that can be characterized
as in (11):
76 Leonard Talmy

(11) a complex of concepts involving:

a. rustler: a person, property ownership, illegality, mode of activity
b. steer: object of particular appearance, physical makeup, etc.
relation to animal kingdom
castration
institution of breeding for human consumption
c. lasso: certain objects (a body and a lasso) in particular configura-
tions
certain movement sequences
accompanying cognitive intending, directing, monitoring, etc.

In surveying the two lists, we can see these differences emerge: The grammati-
cal elements are more numerous, and their specifications seem more spare and
simpler, and more structural in function. Together, their specifications seem to
establish the main delineations of the scene organization and communicative
setting of the CR evoked by the sentence. The lexical elements are fewer in num-
ber, but their specifications are greater in quantity and complexity, and function
more to contribute content than structure. The lexical specifications are greater
in three ways: compared to a grammatical specification, each has a) more total
information, b) greater intricacy of information, and c) more different types of
information together. Taken together, their specifications comprise most of the
conceptual content of the CR scene that is evoked by the sentence.
These grammatical-lexical differences can be set into further relief by in
turn varying each element-type while keeping the other constant. Thus, varying
only the grammatical elements of (9), as is done in (12), seems to alter the scene
organization and discourse properties of the referent event but to leave its basic
contents intact:

(12) Will the rustlers lasso a steer?

By contrast, varying only (9)’s lexical elements, as in (13), shifts us to a new scene
altogether, and yet the basic breakup of the scene and of its communicative set-
ting seems to remain the same:

(13) A machine stamped the envelopes.

 Chapter 2: Grammatical construal 77

2. Categories of grammatically specified notions

The preceding sampling of grammatical elements has yielded a set of notions

helpful toward discovering common semantic properties. But the set has been
small and unstructured. With a broader and more systematic investigation, pat-
terns of organization among the notions become evident. Grammatically speci-
fied notions can be seen to pattern in categories, and the categories, in turn, in
integrated systems, as presented below. And within these notional patterns can be
seen certain regularities of function and process. These patterns and regularities
constitute principal features of conceptual organization in language.
Several such features are brought forward below. One feature is an extensive
homology between the representation of space and that of time. The first category,
“dimension”, includes this space-time homology, and largely crosscuts the remain-
ing categories. These categories will, in the majority, apply to both space and time,
and parallel examples from each dimension will be presented side by side.
Another feature is that, of the member notions of any category represented in
a language, often each notion will be incorporated in at least some lexical items.
Correlatively, the language will often contain grammatical forms that interact with
each lexicalization type in a way that yields the expression of another notion of
the category. Each such type of interaction can be regarded as a type of cognitive
operation that converts the indication of one notion to that of another within the
same category. A corollary feature is that a language with grammatical forms
for converting from notion A to notion B frequently has forms as well for con-
version in the reverse direction – that is, it can also trigger the reverse cognitive
operation.10
Some of the grammatical forms in a language function specifically to perform
a particular conversion operation. Others simply make structural specifications
that can come into conflict with the specification of a neighboring lexical item. In
such cases, the basic pattern is that the grammatical form’s specification always
takes precedence, and triggers a kind of operation, a “shift”, in the lexical item’s
referent that brings it into accord.”
As a note on methodology in what follows, efforts were made to determine
categories on the basis of particular grammatical meanings encountered, rather
than to posit the categories as part of an a priori schema which then sought cor-
roborative examples. In the research leading to this paper, grammatical forms
were sampled from a range of languages, but an effort has been made to take
most of the exemplification from English.
78 Leonard Talmy

2.1. Dimension
The category of “dimension” has two principal member notions, ‘space’ and ‘time’.
The kind of entity that exists in space is – in respectively continuous or discrete
form – ‘matter’ or ‘objects’. The kind of entity existing in time is, correspond-
ingly, ‘action’ or ‘events’ – terms here used neutrally as to whether the entity is
static or changing. These notions thus relate as in (14):12

(14) dimension continuous discrete

space: matter objects
time: action events

Homologies between the linguistic structuring of space and of time will be indi-
cated in the categories that follow. But here we can indicate operations of conver-
sion between these two main members of the dimension category. Thus, a verb
root that lexicalizes expression of an event or of action as a temporal quantity can
be associated with grammatical forms, including nominalizations, that signal a
cognitive operation of “reification”. By the semantic effect of this operation, the
referent becomes conceptualized as an object or a mass, one that can participate
in many of the same activities (such as being given or gotten) as a physical quan-
tity, as well as in many of the corresponding syntactic constructions (including
pluralization and modification: ...gave me two quick calls), as exemplified in (15).
(A way of representing the grammatical complexes involved here and in the next
operation is presented in connection with the following category.)

(15) an event: reified as an object:

John called me. John gave me a call.
I was called by John. I got a call from John.

action: reified as mass:

John helped me. John gave me some help.
I was helped by John. I got some help from John.

The reverse conversion also occurs. A noun referring to an object or mass can
be associated with grammatical forms, including verb-forming derivations, that
signal a cognitive operation of “actionalizing”. By this operation, the physical
referent is melded together with some of the activity in which it participates,
with the semantic effect that much of the referent’s tangible concrete character
is backgrounded, subordinated to a conceptualization in terms of a process of
occurrence, as illustrated in (16):
 Chapter 2: Grammatical construal 79

(16) object(s)lmass: actionalized as:

Hail(stones) came in through It hailed in through
the window the window.
Ice is forming over the windshield. It is icing up over the windshield.
I removed the pit from the cherry. I pitted the cherry.

2.2. Plexity

The category here to be termed “plexity” is a quantity’s state of articulation into

equivalent elements. Where the quantity consists of only one such element, it is
“uniplex”, and where it consists of more than one, it is “multiplex”. When the
quantity involved is matter, plexity is, of course, equivalent to the traditional lin-
guistic category of “number” with its component notions “singular” and “plural”.
But the present notions are intended to capture the generalization from matter
over to action, which the traditional notions do not do.13
Specifications as to plexity are made by both lexical items and grammatical
elements, and there is interplay between the two when they are both in association.
Example English lexical items that basically specify a uniplex referent are – for
matter and action, respectively – bird and (to) sigh. They can occur with gram-
matical elements that themselves specify a uniplexity, like those italicized in (17a)
(many languages have here a more regular, overt system of markers than English).
But they can also occur with grammatical elements that specify a multiplexity, as
in (17b). In this association, such elements can be thought to trigger a particular
cognitive operation, one of “multiplexing”. By this operation, an original solo
referent is, in effect, copied onto various points of space or time.

(17) matter action

a. uniplex A bird flew in. He sighed (once).
b. multiplex Birds flew in. He kept sighing.

The reverse of the preceding pattern is also found in language. First, there are
lexical items that intrinsically specify a multiplexity. English examples are furni-
ture or timber (i.e., ‘standing trees’) for matter and breathe for action, as used in
(18a). And, too, there are grammatical forms able to appear in association with
these, as in (18b), that signal an operation the reverse of multiplexing – one that
can be called “unit-excerpting”. By this operation, a single instance of the speci-
fied equivalent units is taken and set in the foreground of attention.
80 Leonard Talmy

(18) matter
action
a. multiplex Furniture overturned in the earthquake.
She breathed without pain.
b. uniplex A piece of furniture overturned in the earthquake.
She took a breath / breathed in without pain.

The English grammatical forms seen above that signaled multiplexing – -s and
keep -ing – consisted solely of explicit morphemes. The forms that signal unit-
excerpting differ in that they also include abstract elements: particular grammatical
categories that require the insertion of one out of a certain set of lexical items, as
represented in (19c,d). The forms can, moreover, contain two or more indepen-
dent elements. These forms are here considered to be “grammatical complexes”,
comparable to other grammatical constructions or indeed to lexical complexes
(collocations): they combine distinct elements within a structural whole serving a
single overall semantic function. Actually, by one analysis, all grammatical forms
are complexes, merely ranked along a dine of elaborateness. Under this analysis,
a grammatical form includes not only any explicit and generic elements, but also
the semantic and syntactic category memberships of its input and output forms,
as represented throughout (19). Thus, the English multiplexing forms, in (19a,b),
are merely at the simpler end of a continuum:

(19)

e.g., sigh: keep sighing

 Chapter 2: Grammatical construal 81

Support is lent to the thesis that a more elaborate grammatical complex can have
a semantic unity by the existence, within the same or another language, of a sim-
pler form with the same semantic function. As an example of just this circum-
stance, the English unit-excerpting complex for nouns, which is rather elaborate,
is paralleled in function by a simple suffix in Yiddish, either -l or -ele (otherwise
indicating diminutives), as illustrated in (20):

(20) zamd ‘sand’: zemdl ‘grain of sand’

groz ‘grass’: grezl ‘blade of grass’
shney ‘snow’: shneyele ‘snowflake’

2.3. State of boundedness

When a quantity is specified as “unbounded”, it is conceived as continuing on
indefinitely with no necessary characteristic of finiteness intrinsic to it. When a
quantity is specified as “bounded”, it is conceived to be demarcated as an indi-
viduated unit entity. In application to nouns, these notions largely correspond to
the traditional linguistic distinction between “mass” and “count”, and in appli-
82 Leonard Talmy

cation to verbs they can correspond to “imperfective” and “perfective”, among

other terms (the closeness of these correspondences varies with different usages
of the traditional terms). However, as with plexity, the concepts designated by the
new terms are intended to capture the commonality between the space and time
dimensions and to generalize over their usually separate analyses.
Among English examples of lexical items, water and (to) sleep basically specify
unbounded quantities, whereas sea and (to) dress basically specify bounded ones.
These specifications are demonstrated by the fact that these words are, respectively,
unacceptable and acceptable in construction with the grammatical complex “in
NPextent-of-time”, which itself specifies boundedness, as seen in (21):

(21) matter
action
a. unbounded *We flew over water in 1 hr.
*She slept in 8hrs.
b. bounded We flew over a sea in 1 hr.
She dressed in 8mins.

As with plexity, there exist grammatical elements that can, in construction with a
lexical item, shift its basic specification for state of boundedness to the opposite
value. Those acting in this way on an unbounded-type lexical item, in effect, trig-
ger a cognitive operation of “bounding”, or “portion-excerpting”. By this opera-
tion, a portion of the specified unbounded quantity is demarcated and placed in
the foreground of attention. Examples of such grammatical elements in English
are shown in (22).
The reverse of the preceding pattern also exists. The English nouns shrub and
panel each refer intrinsically to a bounded entity. But the grammatical elements
-ery and -ing can be added to them, yielding shrubbery and paneling, forms
which now refer to unbounded quantities. In effect, the grammatical elements
have triggered a cognitive operation of “debounding” whereby the quantity for-
merly within bounds is now conceptualized in a form with indefinite extension.
In English, however, such elements are not productive; they cannot, for example,
be used with sea to yield the meaning ‘pelagic water’, nor with (a) tear to yield
‘lachrymal fluid’.14
 Chapter 2: Grammatical construal 83

(22)

2.4. State of dividedness

The category of “state of dividedness” refers to a quantity’s internal segmentation.
A quantity is “discrete” (or “particulate”) if it is conceptualized as having breaks,
or interruptions, through its composition. Otherwise, the quantity is conceptual-
ized as “continuous”.15
Both lexical and grammatical elements are sensitive, in their specifications,
to the distinctions of this category. But there appear to be no grammatical ele-
ments that solely specify discreteness or continuity for a quantity, nor any that
signal an operation for reversing a quantity’s lexically specified state of divided-
ness. If forms of the latter type existed, we can describe how they would behave.
A grammatical form for a continuous-type lexical item would signal an opera-
tion of “discretizing”, whereby the originally continuous referent would become
conceptualized as a particulate aggregation. Conversely, a grammatical form for
a discrete-type lexical item would trigger an operation of “melding”, whereby
the separate elements of the original referent would be conceptualized as having
fused together into a continuum.
Although such grammatical forms seem lacking, there do exist certain indirect
or inexplicit mechanisms for these operations. Thus, the continuity specified by the
noun water can be reconceptualized as discrete with the locution particles of, as
in: Water / Particles of water filled the vessel. However, the grammatical complex
used here does not directly specify this shift but, like the complexes in Sections
2.5. and 2.13., comprises a several-stage sequence of other cognitive operations.
In the reverse direction, there appears to be a general conceptual tendency for a
basically discrete-type referent of a lexical root to undergo at least some degree
of spontaneous melding, without the addition of any explicit grammatical forms.
84 Leonard Talmy

Thus, foliage, timber, and furniture, as contrasted with leaves, trees, and pieces
of furniture, tend to evoke referents with a degree of blurring and fusion across
their component elements.
Because the category of dividedness has limited realization by itself, further
treatment of it will be deferred until the next section, where it can be seen in
interaction with the other categories.

2.5. The disposition of a quantity: A system of categories

The preceding four categories of attributes – dimension, plexity, boundedness,
and dividedness – all pertain to a quantity simultaneously and, taken together, can
be considered to constitute a system of attributes that may be termed a quantity’s
“disposition”. The intersections of these categories form an array that can be
schematized as in (23).

(23)

+ the distinction between matter and action, which crosscuts all of the above16

Each intersection of attributes indicated here is specified by various lexical items

(although one, a bounded multiplexity for action, is quite minimally represented
in English). An example or two (most seen earlier) is given for each intersection
in (24) :17
 Chapter 2: Grammatical construal 85

(24) A’: timber/furniture B’: water

(to) breathe (to) sleep
A: (a) family B: (a) sea/panel
(to) molt (to) empty
(The bird molted.) (The tank emptied.)
a: (a) bird
(to) sigh

Now if the particular contentful referent for which one chooses a lexical item
happens to be wedded, by that lexical item, to an unwanted set of structural speci-
fications, there generally are grammatical means available for converting this to
a desired set. Such means range in directness from specifying the single relevant
operation to involving a circuitous sequence of operations (cf. Section 2.13. on
“nesting”). A number of starting- and ending-points for such conversions, and
the means for accomplishing them, are indicated in (25):

(25) A’ ĺ A a stand of timber B’ ĺ B a body of water

breathe for an hour sleep for an hour
A’ ĺ a a piece of furniture ----
take a breath / breathe in
Aĺa a member of a family ----
?molt a single feather
A ĺ A’ members of a family B ĺ B’ paneling
(A ĺ a ĺ A’)
molt and molt empty and empty
a ĺ A’ trees ----
keep sighing
aĺA a stand of trees ----
(aĺ A’ ĺ A)
sigh for a while

2.6. Degree of extension

Implicit in the vertical dimension of the schematic arrangement in (23) is a further
category that can be called “degree of extension”. This category has three prin-
cipal member notions, terms for which are given in (26) together with schematic
representations of the notions for the linear case. Lexical items referring to either
matter or action may be taken to incorporate specifications as to their referent’s
basic degree of extension, and three examples of these for the linear spatial case
are also shown in (26) :18
86 Leonard Talmy

(26)

Now a lexical referent that is perhaps most basically conceived as of one particular
degree of extension can, by various grammatical specifications that induce a shift,
be reconceptualized as of some other degree of extension. For a first example,
consider the event referent of climb a ladder, which seems basically of bounded
linear extent in the temporal dimension, as is in fact manifested in (27) in con-
junction with the grammatical element “in + NPextent-of-time”:

(27) She climbed up the fire-ladder in 5 minutes.

With a different accompanying grammatical form, like the “at + NPpoint-of-time”

in (28), (as well as different contextual specifications), the event referent of the
preceding can be shifted toward a conceptual schematization as a point of time –
i.e., as being point-durational:

(28) Moving along on the training course,

she climbed the fire-ladder at exactly midday.

This shift in the cognized extension of the event can be thought to involve a cog-
nitive operation of “reduction” or, alternatively, “adoption of a long-range per-
spective”. This shift can also go in the other direction. The event referent can be
conceptually schematized as an unbounded extent by the effect of grammatical
forms like “keep -ing”, “-er and -er”, and “as + S”, as in (29):

(29) She kept climbing higher and higher up the fire-ladder as we watched.

Here there would seem to have taken place a cognitive operation of “magnifica-
tion”, or “adoption of a close-up perspective”. By this operation, a perspective
point is established from which the existence of any exterior bounds falls outside
of view and attention – or, at most, is asymptotically approachable.
The preceding event referent was continuous, but a discrete case can exhibit
the same shifts in extension. One such case, perhaps to be considered as most
basically of bounded extent, is shown with that degree of extension in (30a). But
the referent can also be idealized as a point, as in (30b) (clearly, the cows would
not all have died at the same moment, and yet the spread of their death times is
conceptually collapsed into such a single moment). Or, the referent can be sche-
matized as an unbounded extent, as in (30c):
 Chapter 2: Grammatical construal 87

(30) a. The cows all died in a month.

b. When the cows all died, we sold our farm.
c. The cows kept dying (and dying) until they were all gone.

The alternative schematizations of extension just seen as specifiable for an event

referent are generally also available for an object referent. Thus, e.g., the refer-
ent of (a) box can be specified for idealization as a point or as a bounded extent
(of area or volume). Some grammatical elements making such specifications are
illustrated in (31). Also set forth here are the homologies between these and the
event-specific elements:

(31) point The box is 20 feet away from the wall.

I read the book 20 years ago.
bounded extent The box is 2 feet across.
I read the book in 2 hours.
(point within) bounded extent The ball is in the box.
She left while I read the letter.

2.7. Pattern of distribution

The pattern of distribution of matter through space or of action through time
is a further category of notions that can be both grammatically and lexically
specified.19 For action through time – the only dimension we will be looking at
here – this category together with the preceding one largely constitute the tradi-
tional category of “aspect”.
Several of the main patterns of distribution for action through time are shown
schematically in (32) (the dots here, which represent situatedness in complementary
states, should really be adjacent, but they are sketched apart with a connecting
line to show the crossing of state-interfaces). Also shown are illustrative English
verbs, both non-agentive and agentive, that incorporate these patterns.

(32)
88 Leonard Talmy

One can determine that these verbs incorporate the specifications indicated by
noting the grammatical forms with which they can and cannot occur (or, to put
the latter case in our terms: ...grammatical forms toward whose specifications
they will not [readily] shift). A full demonstration is not in order here, but a few
examples show the principle: The resettable type of a one-way event is distin-
guished from the non-resettable type by its compatibility with iterative expres-
sions, as in: He fell 3 times; the non-resettable type cannot occur here: *He died
3 times. This same one-way form is distinguished from a full-cycle form by its
ability to appear in sentences like: He fell and then got up, which the latter can-
not do: *The beacon flashed and then went off. A gradient type can appear with
adverbs of augmentation, as in The river progressively widened, unlike a steady-
state type: * She progressively slept. And so on.
Grammatical elements can, of course, also specify differing patterns of tem-
poral distribution, and the present form of diagramming can readily reveal some
of their distinctions. Thus, the closed-class elements back and again, singly and
in combination, can indicate versions of full-cycle, sesqui-cycle, and double-cycle
patterns, as shown in (33):

(33)

Now consider the circumstance where a verb of one distribution type appears with
grammatical forms of another type. The resultant seems invariably to be that the
verb shifts its specifications into conformity with those of the grammatical forms.
For an example we again take die, whose basic specifications can be adjudged
as point-durational one-way non-resettable – schematizable, now more precisely,
as: ƔƔ. This verb is used with its basic specifications in a sentence like (34a). But
in a sentence like (34b), the grammatical form “be + -ing” induces a shift. In
effect, the infinitesimal interval between the two states involved for die – viz.,
‘aliveness’ and ‘deadness’ – is spread out, with the creation thereby of an extent-
durational gradient. This is the shift in the distribution pattern’s structural type.
But concomitantly, a shift in the basic contentful referent is engendered. Instead
of ‘dying’, the new gradient refers to ‘moribundity’. The distinction becomes clear
in noting that, as the conception is structured linguistically, one can have been
dying without having died, and, correlatively, one can have died without having
been dying.20
 Chapter 2: Grammatical construal 89

(34) a. He died as she looked on.

b. He was (slowly) dying as she looked on.

2.8. Axiality
The adjectives in a pair like well/sick behave contrarily when in association with
grammatical forms specifying degree like slightly and almost, as seen in (35a),
and they select for different readings of temporal forms like “in + NPextent-of-
time”, as seen in (35b). In this, perhaps surprisingly, they parallel the behavior
of certain kinds of expressions that specify spatial relations, e.g., at the border
I past the border:
(35)
a.
He’s slightly { sick / past the border.
*well / *at the border. }
He’s almost { well / at the border.
?sick / ?past the border. }
b. He got well / to the border in 5 days. – i.e., progressively in the course of
He got sick / past the border in 5 days. – i.e., first after the elapse of

This behavior can be accounted for by positing that such adjectives, in referring to
a more generic notional parameter, such as that of ‘health’, are not simply “oppo-
sites” but, rather, presuppose a schematic axis that is structured and directed in
a particular way. Each adjective, then, labels a different portion of that axis. The
adjectives here seem in particular to presuppose a directed line bounded at one
end; well refers to the end-point while sick refers to the remainder of the line,
correlating greater magnitude with greater distance along the line. These are the
“axial properties”, or “axiality”, of the lexical items, i.e., the specific relations
each has to a particular conceptual axis and to other lexical items with referents
along the same axis. It is the lexicalization of such axiality that can align adjec-
tives with expressions of spatial relation. Grammatical forms like the ones just
above also have axial properties, and these can function in consonance with
those of a lexical item, as in the acceptable cases of (35), now schematized as to
axiality in (36):
90 Leonard Talmy

In other cases, though, the axiality of a grammatical form can conflict with that
of a lexical item and, accordingly, can cause the latter to shift. Thus, sick in (37)
– now associated with grammatical forms that refer to an end-point – shifts from
its basic “directed shaft” type of axiality, and indeed from its reference to an
axis of ‘health’; it now specifies the end-point of an axis pertaining to ‘feeling
physically bad’.

(37) (After exposure to the virus, he felt worse and worse and)
he was almost sick at one point. / he finally got sick in 3 days.21

2.9. Perspectival mode

As seen earlier, a particular event, whether static or changing, can have a pattern
of distribution through time that is perhaps most basically associated with or
intrinsic to it in its own right. But, in addition, language has the means for speci-
fying an independent schema as to how one is to attend to the event. This schema
includes the location and deployment of the perspective point one adopts from
which to regard the event and the distribution of one’s attention over the event.
This category of specifications, here called the “perspectival mode”, can either
conform with or diverge from the event’s own basic pattern of distribution. Two
principal members of the category are characterized in (38):
 Chapter 2: Grammatical construal 91

(38) the assuming of:

a. a steady-state long-range perspective point with global scope of
attention
b. a moving close-up perspective point with local scope of attention

For illustration, consider first an example with a basically steady-state referent,

viz., objects in location. The (38a) type of perspectival mode – the one more con-
gruent with such a referent – is invoked in (39a), multiply specified there by the
set of grammatical forms shown underlined, namely, plural forms, an adverbial
expression of spatial dispersion, and the locative preposition in. But these can
be replaced by grammatical forms coding for the (38b) perspectival mode – as
in (39b) with singular forms, an adverbial expression of temporal dispersion,
and the motion preposition through. Thereby, the evoked CR is converted to one
where one’s perspective and attention or one’s own projected location shifts in
turn from object to object. In effect, a steady-state multiplexity of objects has
been converted to a sequential multiplexity of events consisting of conceptualized
encounters with the objects.

(39) a. There are houses at various points in the valley.

b. There is a house every now and then through the valley.

For representing certain static spatial configurations, the moving-perspective mode,

though non-congruent in character, is greatly favored over the steady-state mode.
Thus, the ready colloquial formulation of (40b) for moving-perspective is matched
in the global steady-state mode of (40a) only by a stilted scientific style:

(40) a. The wells’ depths form a gradient that correlates with their locations
on the road.
b. The wells get deeper the further down the road they are.

The reverse of the preceding circumstances also exists. That is, a sequential
multiplexity of events, an example of which is represented in (41 a) with the
more congruent moving-perspective mode, can also become the object of a fixed
global viewing, as represented in (41b). Metaphorically, the effect here is as if
the vertical time line is tilted up into present-moment horizontality for integrated
or summational assessment.

(41) a. I took an aspirin time after time during / in the course of the last
hour.
b. I have taken a number of aspirins in the last hour.22
92 Leonard Talmy

2.10. Level of synthesis

The category to be considered now pertains to bounded quantities, like those
schematized in the A/B row in (23). One form of locution already seen to specify
such quantities is the particular type of “NP of NP” construction illustrated in
(42a). Here the second NP specifies the identity of the quantity involved, itself
conceptualized as without intrinsic bounds, while the first NP specifies the bound-
ing, or “portion-excerpting”, per se of the quantity. Moreover, in addition to such
a pure operation of bounding, the first NP can further specify the particular form
or configuration that the excerpted portion has, as in (42b):23

(42) a. a set of trees a body of water

b. a cluster of trees a puddle/drop of water

The two NPs here can be seen as coding for two different “levels of synthesis”.
Describing this for the internally discrete case, e.g., a cluster of trees, we can
say that the second NP specifies an unsynthesized multiplexity of independent
elements, while the first NP specifies a particular Gestalt synthesized out of that
multiplexity.
Furthermore, language can mark an additional cognitive distinction here.
Either level of synthesis can be placed in the foreground of attention while the
other level is placed in the background. One grammatical device for marking
this is the placement of the foregrounded NP at the head of the larger nominal
construction (in English, placing it first), as shown in (43a). With the use of this
device, moreover, predications can be made that pertain solely to one level of
synthesis or the other, as seen in (43b):

(43) a. the cluster of trees / the trees in the cluster

b. That cluster of trees is small.
The trees in that cluster are small.

There are certain forms, furthermore, whose referents are keyed to applying to
only one or the other level of synthesis. Thus, together (toward each other) tends
to correlate with multiple objects at large, while in upon -self tends to correlate
with a composite formed therefrom, as seen in (44):

(44) The bricks in the pyramid came crashing together / *in upon themselves.
The pyramid of bricks came crashing in upon itself / * together.

The preceding phenomena have involved the shift of attention from a multiplexity
to a Gestalt that it can constitute, a process that can be called “Gestalt forma-
tion”. Also encountered in language are means for specifying the reverse: shift-
 Chapter 2: Grammatical construal 93

ing attention from a Gestalt to components seen as constituting it, in a process

of “componentializing”. This procedure can take place when the starting lexical
item specifies an entity taken to be already at the more synthetic level, as is the
case with iceberg in (45a). By grammatical devices like those in (45b), such an
entity can be analytically converted from conception as a coherent whole to one
of component parts and their interrelations. Again we encounter a surface form
– in two – that correlates with only one level of synthesis and not the other:

(45) a. The iceberg broke in two.

b. The two halves of the iceberg broke apart (*in two).

The two levels of synthesis with the two directions of conceptual shift applicable
to them define four notional types, as indicated in (46). The term Figure is used
here as described in Talmy (1978b, 1983).

(46) example type operation

cluster of trees: “composite Figure”
“multiple Figure” ඍ
Gestalt-formation
iceberg:
two halves of iceberg:
“meta-Figure”
“component Figures” ඏ
componentialization

2.11. Level of exemplarity

A further cognitive distinction can be specified for a multiplexity of objects. This
distinction does not affect the basic reference to all the members of the multiplex-
ity, but addresses how attention is directed and distributed within that multiplexity.
Either the full complement of the multiplexity is in the foreground of attention,
with perhaps individual items here and there singled out in the background of
attention. Or a single exemplar out of the multiplexity is placed in the foreground
of attention, with the remaining items more dimly conceived in the background of
attention. This distinction as to “level of exemplarity” is specified by grammatical
devices in perhaps most languages. But English stands out in the extensiveness
of its specifications: there are different pairs of grammatical forms that mark the
distinction for a number of different types of multiplexity. A rather full list of
these pairs is indicated in (47), with examples showing first the full-complement
form and then the counterpart exemplar form:

(47) a. Oysters have siphons / a siphon.

An oyster has siphons / a siphon.24
94 Leonard Talmy

b. All oysters have siphons / a siphon.

Every oyster has siphons / a siphon
c. All the members raised their hand(s).
Each member raised his hand(s).25
d. Many members raised their hand(s).
Many a member raised his hand(s).
e. Some members here and there raised their hand(s).
A member here and there raised his hand(s).
f. Members one after another raised their hand(s).
One member after another raised his hand(s).
g. Hardly any members raised their hand(s).
Hardly a member raised his hand(s).
h. No members raised their hand(s).
No member (Not a member) raised his hand(s).
i. She held a gun in both hands.
She held a gun in either hand.26

2.12. Other categories and processes

A number of further notional categories and cognitive processes can be discerned
in language, but there is opportunity here to present briefly only two additional
examples:
Scene-division properties. A lexical item can have particular “scene-division
properties”, that is, a principal breakup of its referent into parts and participants.
For example, the referent of the English verb serve breaks up into an activity, an
item served, and a social dyad involving the two roles of ‘host’ and ‘guest’ – this
last being its particular “personation” type (Talmy 1985a) – as illustrated in
(48a). But grammatical forms can also have scene-division properties. Thus, a
subject-plus-reflexive-object complex has a single-role specification. When such
a grammatical form occurs with a dyadic verb like serve, it triggers an operation
of “monad-formation”: the verb’s referent shifts to one with monadic personation,
as in (48b). In this shifted state, its referent is equivalent to that of an intrinsically
monadic expression, like that in (48c):27

(48) a. The host served me some dessert from the kitchen.

b. I served myself some dessert from the kitchen.
c. I went and got some dessert from the kitchen.

Associated attributes. Lexical expressions like apartment and hotel room, in addi-
tion to their basic denotations, may be taken to have “associated attributes” – here,
respectively, those of ‘permanent residence’ and ‘temporary lodging.’ Such attri-
 Chapter 2: Grammatical construal 95

butes may mesh or conflict with the specifications of another element in the same
sentence. The attributes of the above two nominals mesh and conflict respectively,
e.g., with the closed-class directional adverb home, which specifies a permanent
residence. In the case of conflict, as in (49b), a cognitive process operates on the
lexical item to leave its essential characteristics intact but replace its associated
attributes with the closed-class element’s specifications:

(49) a. He drove home to his apartment.

b. He drove home to his hotel room.

2.13. Nesting: An interaction of categories

A number of what can be considered “meta-properties” govern the behavior of
categories of grammatical notions, in general and with respect to one another.
One of these, the capacity for nesting, already illustrated in Section 2.5., can be
readily presented here: The operations and shifts described throughout Section
2 need not take place singly. The output of one can serve as the input to another,
thereby building up hierarchical levels of embedding. While there are a number
of interesting examples of this for different types of matter and action, we will
go directly to illustrating one of the longer cases in (50):

(50) a. The beacon flashed (as I glanced over).

b. The beacon kept flashing.
c. The beacon flashed 5 times in a row.
d. The beacon kept flashing 5 times at a stretch.
e. The beacon flashed 5 times at a stretch for 3 hours.

In (50a), the lexical verb flash appears with its basic structural specification
as a point-durational full-cycle uniplex event. This undergoes the operation of
multiplexing, to yield the unbounded multiplexity in (50b). This then undergoes
bounding in (50c). This bounded multiplexity then first goes through the opera-
tion of reduction to become schematized as a new point-like uniplex quantity,
and this is in turn multiplexed, yielding (50d). This new unbounded multiplexity
is finally then bounded in (50e). The nesting of structural specifications in this
last stage can be represented schematically as in (51):

(51) [('"") - ('"")......("'") - ('"")]

Quite analogous to this temporal nesting, except for the lack of specific numerals,
is the spatial example in (52):
96 Leonard Talmy

(52) a. A duck landed on the pond.

b. Ducks landed on the pond
c. A flock of ducks landed on the pond.
d. Flocks of ducks landed on the pond.
e. A group of flocks of ducks landed on the pond.

2.14. Four imaging systems

Most of the preceding categories of grammatically specified notions, together
with categories not discussed here, group together under four much broader con-
ceptual systems, ones that can be understood as principal “imaging systems” of
language. These are great complexes in language that organize the structuring
and the “viewing” of conceptual material. The four systems outlined here (there
are additional ones) are relatively independent of each other in content, with each
adding a distinct conceptual parameter to those of the others, but their contribu-
tions can be coordinated and linked, at times by individual grammatical forms.
The first imaging system is “structural schematization”. This system comprises
all the forms of conceptual delineation that can be ascribed to a quantity, or to the
pattern in which two or more quantities are interrelated, whether in space or time or
some other conceptual dimension. A number of the categories of notions presented
above are part of this system. After “dimension”, all the categories pertaining to
the disposition of a quantity and its generalizations belong here, in particular, the
categories of plexity, state of boundedness, state of dividedness, degree of exten-
sion, pattern of distribution, and axiality. Belonging here, too, are the category
of scene-division properties and that of the “partitioning” of space or time that
is specified by such deictics as this and that, described in Section 1. A further
major component of this imaging system is the spatial or temporal “geometric”
schematization, including the topology-like kind, that is specified especially by
the adpositional systems of languages. This was only touched on here in Section
1’s discussion of the English preposition through, but it is an extensive domain,
one treated at length in such works as Bennett (1975), Gruber (1965), Jackendorf
(1977), Langacker (1986), Talmy (1975, 1982, 1983), Herskovits (1986).
The second imaging system is the “deployment of perspective”. Given a struc-
turally schematized scene, this system pertains to how one places one’s “mental
eyes” to look out upon that scene, including the location, the distance away, and
the movement pattern of this conceptual perspective point. Belonging to this
system from the discussion above is the category of perspectival mode, with its
options of a steady-state or a moving perspective point. Also here is the category
of degree of extension, when its alternatives are interpreted as “adopting a long-
range vs. a close-up perspective”.
 Chapter 2: Grammatical construal 97

The third imaging system is “distribution of attention”. Given a schematized

scene and a vantage from which to regard it, this system pertains to the alloca-
tion of attention that one can direct differentially over the aspects of the scene.
Belonging here from the discussion above are the categories of level of synthesis
and level of exemplarity, as well as the component of the perspectival mode cat-
egory that involves global vs. local scope of attention. In addition, a major cat-
egory, not treated here, comprises the obligatory “Figure/Ground” distinctions
that language imposes on a referent scene. Here, within a scene, there is ascribed
to one element the status of “Figure”, with its attentional primacy, and to another
element the status of “Ground”, with its function in the background of attention
as a reference object for the localizing of the Figure (Talmy 1978b, 1978c, 1983).
Additionally in this imaging system are such grammatically marked discourse
concepts as focus, topic, comment, given and new.
The fourth imaging system, not treated here at all, is “force dynamics”, which,
given a structured scene, involves the forces that the elements of the scene exert on
each other. Comprehended here are the notions of force exerted by one quantity
on another, as well as notions of resistance to such force, the overcoming of such
resistance, blockage to the exertion of force, and the removal of such blockage.
The system of force dynamics includes the traditional linguistic concepts of the
“causative”, but is a generalization over those concepts (Talmy 1976, 1985b).

3. Further cognitive connections

Grammatically specified structuring appears to correspond, in certain of its func-

tions and characteristics, to the structuring in other cognitive domains, such as
that of visual perception, compared below, or those of inference and memory. In
particular, perhaps the principal overarching function of the structuring common
across cognitive domains is that of providing conceptual coherence, that is, acting
as a means for integrating and unifying a body of otherwise disparate conceptual
material. In language and, as suggested below, in vision, this fundamental func-
tion has three main global forms of realization: coherence across a conceptual
inventory, coherence within a scene, and coherence through time.
Across the inventory of notions available for expression within any one lan-
guage, grammatical specifications bring coherence principally by constituting a
classification of the vast variety of conceived and perceived material. They gather
different portions of the material together into subdivisions distinct from each
other. By this, any particular currently cognized element is associated with its
implicit “subdivision-mates”. An illustrative case here are the twenty-plus motion-
related prepositions in English, such as through and into, which together subdivide
the conceptual domain of ‘paths considered with respect to reference-objects’.
98 Leonard Talmy

This domain covers a great and varied range, but any particular “path” gener-
ally falls within the purview of one or another preposition, associated there with
other “paths”. To a certain extent, such associations can be regarded as arbitrary
or idiosyncratic. Thus, as seen earlier, classed together by through are such dis-
similar cases as a straightforward liquid-parting course (walking through water)
and a zig-zag obstacle-avoiding course (walking through woods). The question
arises why such distinctions should be effaced by the grammatical system, while
they are observed by the lexical and other cognitive systems. Why are gram-
matical elements – say, such prepositions – not a large and open class marking
indefinitely many distinctions? One may speculate that the cognitive function of
such classification lies in unifying contentful material within a single conceptual
system and in rendering it manipulable – i.e., amenable to transmission, stor-
age, and processing – and that its absence would render content an intractable
agglomeration.
Providing coherence within a cognized scene was the function of grammati-
cal structuring that was originally indicated in the Introduction. There it was put
forward that the grammatical elements of any particular sentence together specify
the structure of the cognitive representation evoked by that sentence. Their speci-
fications act as a scaffolding or framework across which contentful material can,
in effect, be splayed or draped. It can be posited that such structuring is necessary
for a disparate quantity of contentful material to be able to cohere in any sensible
way and hence to be amenable to simultaneous cognizing as a Gestalt. That is,
without such structuring, not only does the inventory of concepts available for
expression in a language become less coherent, but also any selection of such
concepts concurrently juxtaposed by a sentence would tend to be only a collec-
tion of elements, rather than elements assembled so as to convey an integrated
idea or thought complex.
In the course of discourse, a great welter of notions are expressed in rapid suc-
cession, posing the potential problem of an unconnected sequence of ideational
elements. But grammatically specified structuring is a principal contributor to the
conceptual coherence through time that is requisite here. Through such structuring,
a cognitive continuity is maintained through this flux and a coherent Gestalt is
summated over time. A language can have a great stock of closed-class elements
participating in this function, for example, such English forms as “yes, but”,
moreover, nevertheless, besides, instead, also. Such forms direct the illocution-
ary flow, specify the “logical tissue” of the discourse, and limn out its rhetorical
framework. That is, these grammatical forms establish a structure that extends
over a span of time, and thus provides a conceptual level with temporal constancy
amidst more fleeting aspects of content.
The preceding three global forms of grammatically specified structuring
apply over the scope of any single language but, as indicated in the Introduction,
 Chapter 2: Grammatical construal 99

a fourth form must also be recognized that holds for language in general. While
each language has to some extent a different set of grammatical specifications,
there is great commonality across languages, so one can posit that each set is
drawn from an innate inventory of concepts available for serving a structuring
function in language.
Further, though, a qualifying property of this inventory can be adduced. It
can be observed that grammatically specified concepts range cross-linguisti-
cally from ones that are of extremely widespread (perhaps universal) occurrence
and of broad application within a language, down to ones appearing in a scant
few languages with minimal application. Thus, the innate inventory of available
structuring notions that is posited here seems to be graduated as to significance
for the language faculty (cf. the tabular listing of grammatical notions in Talmy
(1985a:126ff)). For example, the notions ‘entity’ and ‘occurrence’ as expressed
by the grammatical categories “noun” and “verb” are of great application and
probably universal distribution, the notional category “number” seems of roughly
middle standing in the ranking, while notions like ‘in the morning’ and ‘in the
evening’ are expressed inflectionally on the verb in just a few languages.
Notably, compared to spatio-temporal structuring, the notional category of
“affect” is rather low in the graduated inventory of concepts that language draws
on for structuring purposes, a fact that is significant considering its importance
in other cognitive domains (cf. the other cross-domain differences noted below).
The affect category does have scattered representation, for example ‘affection’
expressed by diminutive affixes, ‘scorn’ by pejoratives, ‘concern’ by a conjunc-
tion like lest, and ‘hurt’ by the “adversive” construction (as in the English: My
flowers all died on me.). But seemingly no language has a system of closed-class
forms marking major affect distinctions in the way that, say, the modal system in
English specifies distinctions of force opposition (Talmy 1985b). Such an affect
system can easily be imagined, however. Consider a parent addressing a child
in danger near an open window. Grammatical systems readily allow the parent
to refer to the spatial structure in this situational complex – Get away from the
window! – leaving the affectual component to be inferred. But there is no closed-
class form comparable to a modal, one that we could represent as “FEAR”, as in
FEAR the window!, that would allow the parent to refer to the affectual compo-
nent of the complex and leave the spatial component to be inferred. Comparably,
to a child near a freshly painted wall and about to harm it, a parent would likely
again express the spatial structure – Get away from the wall! – leaving the affect
to be inferred. There is no closed-class affect form for ‘like, be nice to’, which
we could represent as “FAVOR”, that the parent could use instead – FAVOR the
wall! – thereby leaving the spatial component for inference.
Parallels can now be drawn between the structuring system operating in lan-
guage and that in visual perception (cf. Jackendoff 1987).28 The principal func-
100 Leonard Talmy

tion of structure to provide coherence appears common across the two cognitive
domains, and the three global forms of such coherence just outlined for language
correspond to comparable forms in the operation of vision.
First, as proposed in cognitive psychology, the perception of any particular
object is mediated by its association with related objects in a schema for that object
type, and the set of such Schemas constitutes a classificatory system (Neisser
1967). This posited functioning of visual perception thus parallels the classifica-
tory function of linguistic structure across a language’s conceptual inventory.
Second, there is a parallel between the linguistic coherence within a referent
scene and the visual coherence within a perceptual scene. The welter of optical
sensations registered at any one moment from some whole visual scene is ren-
dered coherent by the perception of structural delineations running through it.
For example, one looking at, say, the interior of a restaurant from one corner of
the room does not see simply a pastiche of color daubs and curves but, rather,
perceives a structured whole that includes the framework of the room, the spatial
pattern of tables and people, and the individuated tables and people themselves.
And seeing a person in some posture involves perceiving a structural framework
in the human figure, as Marr (1982) describes this in his treatment of the “3-D
model” in visual perception. Children’s line drawings of scenes and stick-figure
sketches of people, animals, and objects (Kellogg 1970) demonstrate our early
capacity to abstract structure from visual scenes and scene parts.
Third, one can observe a parallel between the coherence through time in
linguistic discourse and that in visual perception. If the viewer in the illustrative
restaurant now walks through the room, the patterns in which visual stimuli and
the perception of structure change give rise in turn to the perception of a coher-
ent continuity of path and view occurring within an overall “scene-structure
constancy”.
It is reasonable to assume that, in addition to these language-vision parallels
in global structuring, a number of particular structuring devices match across the
two domains. Perhaps most of the grammatically specified conceptual categories
treated in this paper – including, for example, state of boundedness and level of
exemplarity – correspond to structuring factors in visual perception. Further,
the first three of the broader linguistic systems for conceptual organization, the
“imaging systems” outlined in Section 2.14., seem also to correspond to broader
systems of visual organization.
One can adduce still further parallels between language and vision as to the
properties of their structuring. The topology-like character of grammatical speci-
fications may have some parallel in the character of the perceived delineations of
a scene, or internal structure of a figure, or plan of a path to be followed through
obstacles. Such perceptions of structure seem in certain respects to abstract away
from Euclidean particularities of exact magnitude, shape, or angle, and more to
 Chapter 2: Grammatical construal 101

involve qualitative or approximate spatial relationships. As a further parallel,

the capacity of grammatical specifications to nest, one within another, and form
embedded structuring seems to correspond to embedded structuring within a
visual scene. The restaurant scene above was described in terms of an over-
all framework that embedded a spatial pattern, itself consisting of individuated
objects. Marr’s (1982) analysis of an object like the human figure then continues
the embedding, with perceived structurings of the body ranked from its overall
linearity, to its stick-figure-like limb structure, and further to its articulations of
these components.
Whereas the preceding has outlined a set of parallels between language and
vision, significantly, each of these two cognitive domains has prominent structuring
devices that play little or no role in the other domain. Thus, in visual perception,
three major parameters that structure (parts of) a scene are bilateral symmetry
(moving or static), rotation, and dilation (expansion or contraction) (Gibson 1966;
Palmer 1983) and, if color can be treated as structural, it is a fourth. In language,
by contrast, grammatical specification of symmetry is minimal, perhaps limited
entirely to the notion ‘reciprocal’. Closed-class indication of rotation is limited
in English to the prepositions and verb particles around and over, and is barely
augmented in other languages. Dilation is grammatically expressed in English by
the verb particles in and out when referring to radial motion (spread out / I shrink
in) and, again, such notions are not greatly more elaborated in other languages.
And color, of course, was this paper’s original example of a conceptual category
not grammatically specified.
In the other direction, there are several prominent linguistic categories of
seemingly little structural function in visual perception. Examples are “status
of reality”, as expressed, e.g., by inflections for mood, “status of knowledge”, as
expressed by evidentials, and “comparison of alternatives”, as expressed by a cat-
egory of particles that includes instead, only, and also. Further possible examples
are “relative temporal location”, as expressed by tense markings, “degree”, as
expressed by adjective inflections and modifiers (e.g., English -er, -est, almost,
too), and “force dynamics”, as expressed by modals (Talmy 1985b).
While language may not share these conceptual structuring categories with
visual perception, it may well do so with other cognitive domains. Thus, its closed-
class category “status of knowledge”, which distinguishes such notions as ‘known
as fact’, ‘inferred’, ‘deduced’, and ‘considered probable’ is very likely related to
a set of basic parameters in our reasoning faculty. And, significantly, certain
conceptual categories in language have a structuring apparently similar to that
of conceptual models that form part of our broader faculty for conceptualization,
in particular, our naive or folk models (“mental models” – cf. Gentner and Ste-
vens 1982; Lakoff 1987) as well as models in early science. For example, Talmy
(1985b) demonstrates that the way in which language structures its concepts of
102 Leonard Talmy

causation and force interactions greatly parallels the conceptual structuring of

naive physics and medieval physics (cf. diSessa 1988), while all three of these
forms differ conceptually in a similar way from modern physics.
Generalizing from all these findings, the possibility is that there is a funda-
mental core to conceptual structure that is common across cognitive domains – a
core that thus epitomizes the nature of conceptual structure for human cognition
– but that each domain has features of structuring, and perhaps also functions for
structuring, that are not shared by others. Determining the overall and particular
character of conceptual structure is the aim of the research advanced in the present
study, one requiring a cooperative venture among the cognitive disciplines.

Notes
1. This paper is a moderately revised and fully rewritten version of Talmy (1978a). Since
1978, the amount of additional material on the present subject, both descriptive and
theoretical, has grown to be quite extensive. The present version incorporates a certain
amount of this new material, as well as bibliographic updating, but the remainder
will be reserved for an entirely new paper at a later date.
2. The word “evoke” is used because the relationship is not direct. The CR is an emer-
gent, compounded by various cognitive processes out of the referential meanings of
the sentence elements, understanding of the present situation, general knowledge,
etc.
Our term “cognitive representation” is similar in purport to Fillmore’s (1975) “scene”
but is chosen over that more specifically visual term. The linguistically evoked com-
plex can include much from other sense modalities (notably somesthetic/kinesthetic
and auditory) as well as abstract conceptual aspects. More recently, Lakoff’s (1987)
notion of an “idealized cognitive model”, or ICM, points toward a comparable mental
entity.
3. For their part, grammatical elements are generally more unalloyed in their indication
of structure. They can express more contentful concepts, but this is largely limited.
An example of it is in English upon as used in We marched/rode/sailed/rushed upon
them (e.g., the enemy), where upon incorporates the notion of ‘attack’, seemingly
equivalent to the paraphrase ‘into attack upon’.
4. More recently, research on different aspects of this broader scope has included work
by Jackendoff (1983), Bybee (1985), Slobin (1985), Morrow (1986), and Langacker
(1987).
5. Not included are regular adverbs, which seem in all languages to be derived from
the three open classes just mentioned, rather than to comprise in their own right an
open class of specifically adverbial roots. Of possible inclusion as a type of open
class are the systems of ideophones, or “expressive forms” found, for example, in
a number of Asian and African languages. Also includable, at a level above that of
basic elements, are “lexical complexes” (collocations) like English kick the bucket
or have it in for.
 Chapter 2: Grammatical construal 103

6. More accurately, rather than a dichotomy between an open and a closed type of class,
there appears to be a cline. A class can range from having quite few members, like
that of number inflection in English, to having very many, like that of noun roots in
English, and the class’s properties may correspondingly range from relatively more
grammatical to more lexical. There exist some mid-sized classes – e.g., the several
score individual classifiers of Chinese, or the three dozen or so instrumental prefixes
in the polysynthetic verb of Atsugewi (Talmy 1972, 1985a) – that appear to have
properties part grammatical and part lexical in character.
7. One can note, for example, the effect on one’s cognitive representation in considering
first the sentence I looked at the dog and then I looked at the dogs. The addition of
the grammatical element -s has a major effect on the delineational breakup of – to
put it visually – the scene before the mind’s eye.
8. For example, augmentative and diminutive elements, insofar as they refer to actual
size, seem to specify size relatively greater or lesser than the norm for the particular
object indicated. And closed-class elements specifying distance – like English just
or way, as in just/way up there – specify notions of ‘near’ and ‘far’ that are relative
to the referent situation.
9. The properties of the specifically linguistic form of topology require determination.
In this regard, consider the English preposition in, which in one main usage speci-
fies a plane so curved as to define a volume of space. The referent of this morpheme,
as in mathematical topology, is magnitude-neutral: in the thimble / volcano; and it
is shape-neutral: in the well / trench. But in addition, its referent is closure-neutral,
i.e., indifferent to whether the curved plane leaves an opening or is wholly closed:
in the bowl / ball. And it is discontinuity-neutral, i.e., indifferent to whether the
curved plane is solid or gapped: in the bell jar / birdcage. These last two properties
would form a proper part of language’s topological system, whereas they are strictly
excluded from mathematical topology.
10. In many cases, a language favors only one such direction, having much lexicaliza-
tion with notion A and simple grammatical means for reaching notion B, but in the
reverse direction having only little lexicalization and complex grammatical forms.
Languages differ typologically in the directions they favor. This issue will not be
taken up here, but is treated at length in Talmy (1985a).
11. Shifts are actually one member of a set of “reconciliation processes” – including
blends, juxtapositions, schema-juggling, and blockage – that can be triggered by
the association of a grammatical and a lexical form with incompatible structural
specifications. In the non-shift processes, the grammatical specification does not
take precedence over the lexical one, but plays an equal role with it. Of all these
processes, this paper treats mostly shifts, but an additional number are discussed in
Talmy (1977).
12. In addition to space and time, language represents other conceptual dimensions that
also belong to the present category. For an example, recall from Section 1 that this
and that specify a partition drawn through space – and can do so through time as well
– and indicate that a referent entity is on the same or the other side of the partition
as the speaker. Now consider the English pronouns you and they in their indefinite
usage (akin to German man or French on). These also specify a partition, but one
drawn through “identificational space”, understood as a new conceptual dimension.
104 Leonard Talmy

They indicate, respectively, that ‘the average person’ is or is not identified with the
speaker in some relevant respect, i.e., is on the same or the other side of the identi-
ficational partition as the speaker. Thus, a person who smokes that is visiting a new
neighborhood can ask a passer-by about the purchase of cigarettes with you, but about
the sale of cigarettes with they:
(i) Where can you buy cigarettes around here?
Where do they sell cigarettes around here?
But a person looking for a location to open a tobacco shop would ask a business con-
sultant in the neighborhood about purchases and sales with the reverse assignment
of you and they:
(ii) Where can you sell cigarettes around here?
Where do they buy cigarettes around here?
13. It is true that there are the traditional terms “semelfactive” and “iterative” referring,
respectively, to one and more than one instantiation of an event. But there is no real
equivalent to number: “aspect” includes too much else about the temporal structure
of action. And in any case, none of the traditional terms refers generically to both
the dimensions.
14. The mechanism actually resorted to in many such cases, including that of tear, is the
use of the plural, as in:
(i) Tears flowed through that channel in Hades.
There seems to be a sequence of cognitive operations here in getting from a bounded
to an unbounded quantity. Speculatively, the bounded quantity is first treated as a
uniplex entity, it is then multiplexed, the resultant entities are conceived as spatially
juxtaposed, and their boundaries are lastly effaced, thereby creating an unbounded
continuum.
15. The present category may be prone to confusion with the preceding one. Contributory
here is the normal meaning range of continuous, which as easily covers ‘boundless-
ness’ as it does ‘internal seamlessness’. However, the two categories can vary indepen-
dently. Thus, in the preceding section, the lexical examples given for unboundedness,
water and sleep, happened also to be internally continuous; but the same demonstra-
tion of unboundedness could have been made with internally discrete examples like
timber and breathe.
In general, unbounded forms share many properties, whether continuous or discrete.
Thus, mass nouns and plural count nouns, both unbounded, share many syntactic
characteristics not shared by singular count nouns, e.g.:
(i) a / every – book / *ink / *books;
(ii) all / a lot of / more / some [unstressed] / 0 [generic] – ink / books / *book;
0 [‘progressively more’] (e.g., The machine consumed ink / books / *book
for an hour.)
16. For schematizing action along the one-dimensional time axis, an adaptation of the
two-dimensional A’, B’, A, and B diagrams would be necessary – and can be read-
ily visualized.
17. The lexical types for several of these intersections, it should be noted, do have tradi-
tional terms. Thus, nominal forms of the a, A or A’, and B’ types, respectively, have
been called count nouns, collective nouns, and mass nouns. And verbal forms of the a
 Chapter 2: Grammatical construal 105

and B’ types, respectively, have been called punctual and durative verbs. The matrix
presented here augments, systematizes, and generalizes the traditional notions.
18. This category can be considered a generalization over the earlier category of “state
of boundedness” by the inclusion of the “uniplexity” notion. It can in turn itself be
generalized – becoming the category “pattern of extension” – by the further inclu-
sion of such notions as a quantity bounded at one end but unbounded at the other
(see Talmy 1983).
19. This category clearly patterns with the preceding five within a single system of notions,
one that would be an expansion or generalization over “disposition of a quantity”.
20. Our main purpose here is to note the shift in structural distribution type. The shift in
content will doubtless prove part of a larger pattern as well, but this is not yet worked
out.
21. The category of axiality can be seen as an extension of the preceding category, pat-
tern of distribution. The two categories address temporal stasis or change, involving
both spatial relations (e.g., fall there, past the border here) and qualitative states (e.g.,
die/flash, awake/asleep there, sick/well here). But where the preceding category
focused on discrete states, the present category elaborates the notion of a scalar quan-
tity functioning in conjunction with a discrete state. Due to their structural character,
these two categories pattern together with all the categories after “dimension” as part
of a single broad conceptual system of “structural schematization”, described below
in Section 2.14., and are thereby distinguished from the categories described next,
which belong to different conceptual systems.
22. The use of the perfect here points to a principal function of perfect forms in general:
They can indicate the temporal counterpart of matter located within a bounded extent
of space, of the type seen in (i). That is, a sentence containing the perfect, as in (ii),
suggests a paraphrase like that in (iii), which is homologous with (i):
(i) There were 5 aspirins on the table.
(ii) I have taken 5 aspirins in the last hour.
(iii) There were 5 aspirin-takings in the last hour.
(In support of this interpretation, as pointed out to me by Peyton Todd, the perfect
seems always to involve a temporal span bounded at both ends.)
23. All three notions here – (a) identity of a quantity, (b) portion-excerpting from that
quantity, (c) configuration of that portion – can be respectively represented by three
distinct NPs together in a construction, as in:
(i) a clustering (c) of a set (b) of trees (a).
Many lexical items conflate the specification of two or all of these notions at once.
Thus, conflating (c) and (b) is a cluster ‘a clustering configuration of a set’ and a drop
‘a small globular form of an amount (of a liquid)’. A lexical item conflating all three
notions is a tear ‘drop of lachrymal fluid’. (See Talmy 1985a for a general treatment
of “conflation”.)
24. For the plural form oysters, the plural form siphons is ambiguous as to whether
there are one or more siphons per oyster. All the other combinations unambiguously
indicate the number of siphons per oyster. Thus, the exemplar form is always unam-
biguous in this regard – one of its advantages over the full-complement form. This
same arrangement holds through the list.
25. The difference between each and every arising in this analysis can now be added to
106 Leonard Talmy

those observed elsewhere (e.g., Vendler 1968). Each is the exemplar counterpart of
the full-complement expression all the, but not of all without the. Thus, *Each oys-
ter has a siphon cannot function as a generic assertion. Every is not as unilaterally
aligned in this way, but does serve more naturally as the counterpart of all without
the.
26. One more pair can be added to this list by adjoining two complementary unpaired
forms from two different languages. The English form some, as in some friends of
mine, is a full-complement form requiring the plural and has no exemplar counterpart
in the singular. The corresponding Italian form qualche, as in qualche amico mio,
requires the singular and lacks a plural counterpart.
27. Though the grammatical complex in (48b) is determinative in setting the role-number
as monadic, a trace of the verb’s original scene-division type does remain. In the CR,
the metaphoric suggestion of a dyad is blended in, as if both ‘host’ and ‘guest’ are
together present in the “I”, perhaps as separate subparts of the single person. For this
reason, (48b) is not the complete semantic equal of (48c). Such blending is, beside
shifting, another major process of reconciliation between incompatible specifications,
referred to in note 11.
28. Clearly, the language-related faculty of the brain evolved to its present character in
the presence of other already existing cognitive domains, including that of vision,
and no doubt developed in interaction with their mechanisms of functioning, perhaps
incorporating some of these.

References
Bennett, David C.
1975 Spatial and Temporal Uses of English Prepositions. An Essay in Stratifica-
tional Semantics. London: Longman.
Boas, Franz
1938 Language. In General Anthropology, Franz Boas (ed.), 124–145. Boston: Heath
& Co.
Bybee, Joan L.
1985 Morphology: A Study of the Relation between Meaning and Form. Amster-
dam/Philadelphia: John Benjamins.
diSessa, Andrea
1988 Knowledge in pieces. In Constructivism in the Computer Age, George Forman
and Peter Pufall (eds.), 49–70. Hillsdale, N.J.: Erlbaum.
Dixon, Robert M. W.
1972 The Dyirbal Language of North Queensland. London: Cambridge University
Press.
Fillmore, Charles J.
1975 An Alternative to Checklist Theories of Meaning. Proceedings of the First
Annual Meeting of the Berkeley Linguistics Society 123–131.
 Chapter 2: Grammatical construal 107

Gentner, Dedre, and Albert L. Stevens (eds.)

1982 Mental Models. Hillsdale, N.J.: Erlbaum.
Gibson, James J.
1966 The Senses Considered as Perceptual Systems. Boston: Houghton Mifflin.
Gruber, Jeffrey S.
1965 Studies in Lexical Relations. Doctoral dissertation, MIT. Reprinted as part
of Lexical Structures in Syntax and Semantics. Amsterdam: North-Holland,
1976.
Herskovits, Annette
1986 Language and Spatial Cognition: An Interdisciplinary Study of the Preposi-
tions in English. Cambridge: Cambridge University Press.
Jackendoff, Ray
1977 Toward a cognitively viable semantics. In Georgetown University Round Table
on Languages and Linguistics, Clea Rameh (ed.), 59–80. Washington, D.C.:
Georgetown Unversity Press.
1983 Semantics and Cognition. Cambridge, MA: MIT Press.
1987 Consciousness and the Computational Mind. Cambridge, MA: MIT Press.
Jakobson, Roman
1971 Boas’ view of grammatical meaning. Selected Writings of Roman Jakobson,
vol. 2: Word and Language, 489–496. The Hague: Mouton.
Kellogg, Rhoda
1970 Analyzing Children’s Art. Palo Alto: Mayfield.
Lakoff, George
1987 Women, Fire, and Dangerous Things: What Categories Reveal about the
Mind. Chicago/London: University of Chicago Press.
Langacker, Ronald W.
1986 An introduction to Cognitive Grammar. Cognitive Science 10: 1–40.
1987 Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. Stan-
ford, CA: Stanford University Press.
Marr, David
1982 Vision. San Francisco: Freeman & Co.
Morrow, Daniel
1986 Grammatical morphemes and conceptual structure in discourse processing.
Cognitive Science 10: 423–455.
Neisser, Ulrich
1967 Cognitive Psychology. New York: Meredith.
Palmer, Stephen
1983 The psychology of perceptual organization: A transformational approach. In
Human and Machine Vision, Jacob Beck, Barbara Hope, and Azriel Rosenfeld
(eds.), 269–339. New York: Academic Press.
Sapir, Edward
1921 Language. New York: Harcourt, Brace & Company.
Slobin, Dan I. and Ayhan A. Aksu
1982 Tense, aspect and modality in the use of the Turkish evidential. In Tense
– Aspect: Between Semantics and Pragmatics, Paul Hopper (ed.), 185–200.
Amsterdam/Philadelphia: John Benjamins.
108 Leonard Talmy

Slobin, Dan I.
1985 Crosslinguistic evidence for the language-making capacity. In The Crosslin-
guistic Study of Language Acquisition, vol. 2, Dan I. Slobin (ed.), 1157–1256.
Hillsdale, N.J.: Erlbaum.
Talmy, Leonard
1972 Semantic Structures in English and Atsugewi. Doctoral dissertation, University
of California, Berkeley.
1975 Semantics and syntax of motion. In Syntax and Semantics, vol. 4, John P.
Kimball (ed.), 181–238. New York: Academic Press.
1976 Semantic causative types. In The Grammar of Causative Constructions (Syntax
and Semantics 6.), Masayoshi Shibatani (ed.), 43–116. New York: Academic
Press.
1977 Rubber-sheet cognition in language. Papers from the Thirteenth Regional
Meeting of the Chicago Linguistic Society 612–628.
1978a The relation of grammar to cognition – a synopsis. Proceedings of TINLAP
2 (Theoretical Issues in Natural Language Processing), David Waltz (ed.),
14–24. New York: Association for Computing Machinery.
1978b Figure and ground in complex sentences. In Universals of Human Language,
vol. 4: Syntax, Joseph H. Greenberg (ed.), 625–649. Stanford, CA: Stanford
University Press.
1978c Relations between subordination and coordination. In Universals of Human
Language, vol. 4: Syntax, Joseph H. Greenberg (ed.), 487–513. Stanford, CA:
Stanford University Press.
1982 Borrowing semantic space: Yiddish verb prefixes between Germanic and
Slavic. Proceedings of the Eighth Annual Meeting of the Berkeley Linguistics
Society 231–250.
1983 How language structures space. In Spatial Orientation: Theory, Research,
and Application, Herbert Pick and Linda Acredolo (eds.), 225–282. New York:
Plenum Press.
1985a Lexicalization patterns: semantic structure in lexical forms. In Language
Typology and Syntactic Description, vol. 3: Grammatical Categories and the
Lexicon, Timothy Shopen (ed.), 57–149. Cambridge: Cambridge University
Press.
1985b Force dynamics in language and thought. Papers from the Parasession on
Causatives and Agentivity at the Twenty-First Regional Meeting of the Chi-
cago Linguistic Society 293–337.
Ultan, Russell
1978 Some general characteristics of interrogative systems. In Universals of Human
Language, vol. 4: Syntax, Joseph H. Greenberg (ed.), 211–248. Stanford, CA:
Stanford University Press.
Vendler, Zeno
1968 Each and Every, Any and All. In Linguistics and Philosophy, 70–96. Ithaca:
Cornell University Press.
Whorf, Benjamin L.
1956 Language, Thought, and Reality. John B. Carroll (ed.). Cambridge, MA: MIT
Press.
Chapter 3
Radial network

Cognitive topology and lexical networks*

Claudia Brugman and George Lakoff

This article is a rather long and detailed study of a polysemous lexical item – the
English word over – paying specific attention to the character of the relations
among its senses. Polysemy is a subtype of lexical ambiguity, contrasting with
homonymy, wherein a single lexical form is associated with more than one mean-
ing, and those meanings are unrelated. In the case of polysemy, one word is taken
as having senses which are related. The distinction is an important one for the
resolution of lexical ambiguity, for, as we will show in the bulk of this article,
the way semantic information is stored in a lexical entry may differ depending
on whether that lexical entry is taken as reflecting homonymy or polysemy. We
will show that the common practice of giving a list of meanings of ambiguous
items is neither the only way, nor, for polysemous words, the most efficient way,
of storing such semantic information. We will argue instead that a network-style
mode of storage is cognitively real, and that this allows for a maximum of shared,
and otherwise related, information between senses.
Network-style representation is common in many areas of AI. But we use it
here not by notational fiat but as part of a much more general conception of cat-
egorization, explored at length in Lakoff (1987). That work provides detailed
empirical evidence and theoretical argument against the classical view of categories
as collections of objects characterized by lists of necessary-and-sufficient con-
ditions, and in favor of an enriched view of categories. On that view, categories
may contain a great deal of internal structure – for instance, that one member
of a category should be more exemplary of that category than some other mem-
ber; that the boundaries of the category are not always clear-cut; that categories
may be characterized in part with respect to their contrast with other categories.
The category structure utilized here is called a “radial” structure, with a central
member and a network of links to other members. Each noncentral member of
the category is either a variant of the central member or is a variant on a variant
The theoretical claim being made is that a polysemous lexical item is a radial
category of senses.

Originally published in Lexical Ambiguity Resolution: Perspectives from Psycholinguistics, Neu-

ropsychology, and Artificial Intelligence, Steven L. Small, Garrison W. Cottrell, and Michael K.
Tanenhaus (eds.), 477–508. San Mateo, CA: Morgan Kaufmann. Reprinted with permission from
Elsevier.
110 Claudia Brugman and George Lakoff

What is important for our purposes is that the kind of network structure found
here is not made up ad hoc to characterize this set of facts. Instead, this is a com-
mon category structure that occurs in domains other than the lexicon.
There is an important consequence of using the general theory of radial cat-
egories to characterize polysemy. In the general theory, the links between mem-
bers of the network are not arbitrary. The theory of radial categories comes with
a characterization of possible link types. In the case of polysemy, the link types
are the types of relations linking the senses of the word. In general, some of the
links may involve shared information, some may involve the relation between a
general and a specific case, and some may be metaphoric. In the case under discus-
sion, most of the links are what we have called “image-schema transformations.”
But overall there is only a small number of types of relations between senses of
words, and this study is one of many that is being carried out in an attempt to
figure out what they are.
Such studies are significant in a number of respects. They show that the rela-
tions between senses are not arbitrary, but are rather principled, systematic, and
recurrent throughout the lexicon. Moreover, the relationships are natural, in the
sense that they are either relationships that arise naturally within the cognitive
system, or they are characterized by metaphors that have an independent exis-
tence in the conceptual system. From an explanatory point of view, the natural
and independently motivated character of the links allows us to explain why
polysemy should exist as a general phenomenon. From the point of view of lan-
guage processing studies, it suggests that the lexicon has a structure that is made
use of in processing.

1. Cognitive topology versus semantic features

The traditional mode of representing lexical information, whether in linguistics

or in cognitive science, has been the use of semantic features. Semantic features
arise in general within the symbol manipulation paradigm in cognitive science:
they are finitary symbols with no inherent meaning, but are to be made mean-
ingful by being connected to things in the world. The features have no inherent
structure, and any relationships among them are to be specified by a calculus
characterizing permissible operations on the symbols.
One reason that we have chosen the example of over is that it demonstrates
the inadequacy of feature analyses and shows the need for an oriented cognitive
topology, which characterizes structures oriented relative to the human body
that apply generally to spatial situations, structures like paths, bounded regions,
tops, etc. Structures in a cognitive topology differ from semantic features in a
number of ways: they are inherently meaningful (arising from sensory-motor
 Chapter 3: Radial network 111

operations), they have an inherent structure, they are analog rather than finitary,
and the relationships among them arise naturally via the operation of the human
sensory-motor system.
The evidence in this paper suggests that there are two respects in which cog-
nitive topology is superior to feature analysis for over.

The topological properties of the concepts are necessary to characterize “image-

schema transformations” in terms that are cognitively natural, rather than in terms
of an arbitrary calculus.

We will discuss this issue at the end of the paper. To facilitate the comparison
between topological and feature analyses, we will use both types of representa-
tion in this paper. The drawings indicate the topological representation, while the
names (such as [Link].C.E) indicate feature representations. We will demonstrate
that, while feature representations might be useful in computer simulations, only
topological representations characterize the cognitive reality of the meanings of
words like over:

Topological concepts are needed in order to account for how prepositions can be
used to characterize an infinity of visual scenes.

The semantics of even the most basic spatial senses of over is such that a feature
analysis simply will not do. Take, for example, the sense of over in The ball went
over the net. Given a scene with a ball moving with respect to a net, there is an
infinity of trajectories of the ball relative to the net that over will fit and another
infinity that it will not fit. To characterize those two infinities, one needs concepts
that generalize over possible trajectories and properties of the landscape: this sense
of over requires two bounded regions, and a path from one to the other that is
oriented vertically relative to the net. In short, what is needed is an oriented cog-
nitive topology with elementary structures (paths, bounded regions), orientations
(vertical), and means of fitting them together into an overall gestalt.
Our reason for going into this issue in such detail is as follows: It is some-
times claimed that the Symbol Manipulation Paradigm is necessary to account
adequately for natural language. In fact, the reverse is true. The Symbol Manipu-
lation Paradigm cannot account for natural language semantics. For a lengthy
discussion, see Langacker (1987).

2. Two levels of prototype structure

The lexical representation of over actually contains two levels of topological

structure. First, each sense of over is a complex topological structure. Second,
112 Claudia Brugman and George Lakoff

all the senses together form a radial category, which is itself a complex topologi-
cal structure. It is crucial to distinguish these two levels: The first is the level of
semantic content and the second is the level at which that content is structured
in the lexicon.
Correspondingly, there are two levels of prototype structure – one at each
level. At the second level, the level of lexical structure, the central sense in the
radial category is the prototypical sense of over. But at the first level, the level of
semantic content for each particular sense of over, the nature of prototypicality
is quite different. At this level, prototypicality concerns the degree of fit of some
real-world relation to an individual sense of the word. For example, consider The
plane flew over the mountain. The best fit is where the path goes right above the
center of the mountain. As the path of flight moves away from the center of the
mountain, the degree of fit lessens.
This introduction does not provide the whole story: we will not fully motivate
the independent existence of all principles that relate senses, and we cannot do
justice to the question of how contrasting lexical categories and the convention-
alization of boundaries figure in a full semantic description of this item. The
chief concern here is to provide a detailed example of how a lexical ambiguity of
a specific kind can be given a reasonably complete description and to show the
kinds of theoretical apparatus required for that description. To sum up: the criti-
cal features of this description are that feature-based descriptions are inadequate,
that a topological representation appears to be needed, and that the senses of a
polysemous item form a radially-structured lexical network.

3. The problem

To get some sense of the problem, let us consider a handful of the senses of
over:

The painting is over the mantel.

The plane is flying over the hill.
Sam is walking over the hill.
Sam lives over the hill.
The wall fell over.
Sam turned the page over.
Sam turned over.
She spread the tablecloth over the table.
The guards were posted all over the hill.
The play is over.
Do it over, but don’t overdo it.
 Chapter 3: Radial network 113

Look over my corrections, and don’t overlook any of them.

You made over a hundred errors.

Even this small number of examples shows enormous complexity. The problem
Brugman undertook was how to describe all these senses and the relations among
them. The analysis we will be presenting is a minor refinement of the semantic
aspect of the earlier analysis. Let us begin with what was found to be the central
sense.

4. The above-across sense

The central sense of over combines elements of both above and across (see Figure
18.1). In this example, the plane is understood as a trajector (TR) oriented relative
to a landmark (LM). TR and LM are generalizations of the concepts figure and
ground (Langacker 1983, 1987). In this case the landmark is unspecified. The
arrow in the figure represents the PATH that the TR is moving along. The LM is
what the plane is flying over. The PATH is above the LM. The dotted lines indicate
the extreme boundaries of the landmark. The PATH goes all the way across the
landmark from the boundary on one side to the boundary on the other. Although
the figure indicates noncontact between the TR and LM, the central sense is neutral
on the issue of contact. As we will see shortly, there are instances with contact
and instances without contact. In this respect the schema cannot be drawn with
complete accuracy. Any drawing would have to indicate contact or the lack of it.
The image-schema is neutral, and that is part of what makes it schematic. What
we have here is an abstract schema that cannot itself be imaged concretely, but
which structures images. We will return below to the question of what it means
for an image-schema to structure an image.

Figure 1. The plane flew over. Name: Schema 1

114 Claudia Brugman and George Lakoff

Let us now turn to some special cases of the schema in Figure 1. These are
instances of the schema that are arrived at by adding information, in particular,
by further specifying the nature of the landmark and by specifying whether or
not there is contact. We will consider four kinds of landmark specifications: (1)
LM is a point, that is, the landmark is viewed as a point with no internal struc-
ture. (2) LM is extended, that is, the landmark extends over a distance or area. (3)
LM is vertical. (4) LM is both extended and vertical. For each such case, we will
consider two further specifications: contact between TR and LM, and noncontact.
Each schema will be named using the following abbreviations: X: extended, V:
vertical, C: contact, NC: no contact. Thus, the schema name ‘[Link].C’ stands for
the special case of schema 1 in which the landmark is both vertical and extended
(VX) and there is contact (C) between the LM and TR.

Figure 2.7KHELUGÀHZRYHUWKH\DUG1DPH6FKHPD;1&

Figure 3.7KHSODQHÀHZRYHUWKHKLOO1DPH6FKHPD9;1&

Figure 4.7KHELUGÀHZRYHUWKHZDOO1DPH6FKHPD91&
 Chapter 3: Radial network 115

Figure 5.6DPGURYHRYHUWKHEULGJH1DPH6FKHPD;&

Figure 6.6DPZDONHGRYHUWKHKLOO1DPH6FKHPD9;&

Figure 7.6DPFOLPEHGRYHUWKHZDOO1DPH6FKHPD9&

The schemas in Figures 2–7 can be related by a diagram in Figure 8.

Figure 8. Links among schemas

These links indicate similarity. Thus, all the schemas are linked, as are all the
schemas that share noncontact. Moreover, each pair of schemas that share every-
thing except for the contact parameter are linked. In addition, they are all linked
to schema 1 (see Figure 1), since they are all instances of that schema.
116 Claudia Brugman and George Lakoff

Figure 9.

The schemas in Figures 2–7 can be viewed in two equivalent ways. Take, for
example, a sentence like Sam walked over the hill in Figure 6. We can think of
over in this sentence as being represented by the minimally-specified schema 1
of Figure 1, and we can think of the additional information as being added by
the object and the verb. Thus, a hill is vertical and extended (VX) and walk-
ing requires contact (C) with the ground. Let us refer to this as the minimal
specification interpretation. Equivalently, we can view the minimally-specified
over of Figure 1 as generating all the fully-specified schemas of Figure 2–7. On
this full specification interpretation, we can think of the over in Sam walked
over the hill as having the full specification of schema [Link].C in Figure 7. The
verb walk would then match the contact (C) specification, and the direct object
hill would match the vertical extended (VX) specification. The difference is
whether the verb and direct object add the VX and C information or whether
they match it.
These two interpretations make slightly different claims about the lexical rep-
resentation of over in these sentences. On the minimal specification interpretation,
only schema 1 exists in the lexicon; the other schemas all result from information
added by the verb and direct object. On the full specification interpretation, there
is a lexical representation for all these schemas; the more specific schemas are
generated by schema 1 plus the general parameters we have discussed: C–NC
and X–VX–V.
On the basis of what we have said so far, these two interpretations are com-
pletely equivalent; there is no empirical difference between them, and no a pri-
ori reason to choose between them. There is, however, additional evidence that
favors the full specification interpretation, and we will be citing it throughout
the remainder of this case study. We will be arguing that the senses of over form
a chain with schema 1 at the center. On the full specification interpretation, the
schemas in Figures 2–7 are part of that chain. Some of those schemas form links
to other senses. The existence of such links suggests that the full specification
 Chapter 3: Radial network 117

interpretation is correct. Consider the following case, where there is a focus on the
end-point of the path. We will use the abbreviation E in naming schemas where
there is end-point focus.
In Figure 10, there is an understood path that goes over the hill, and Sam lives
at the end of that path. The end-point focus is not added by anything in the sen-
tence, neither hill, nor lives, nor Sam. Over here has an additional sense which
is one step away from schema [Link].C – a sense in which end-point focus (E) is
added to yield schema [Link].C.E.

Figure 10. Sam lives over the hill. Name: Schema [Link].C.E

But end-point focus cannot be freely added to just any of the schemas in Figures
18.2–18.7. It can only be added to those with an extended landmark, as in Figure
18.11.

Figure 11. Sausalito is over the bridge. Name: Schema 1.X.C.E

In these cases, over has the sense of ‘on the other side of’ as a result of end-point
focus. However, over does not in general mean ‘on the other side of’. For example,
sentences like Sam lives over the wall and Sam is standing over the door, if they
occur at all, cannot mean that he lives, or is standing, on the other side of the door
and the wall. And a sentence like Sam is sitting over the spot, can only mean
that he is sitting on it, not that he is sitting on the other side of it. Thus there is no
end-point focus schema corresponding to schema l.V.C of Figure 6.
118 Claudia Brugman and George Lakoff

Assuming the full specification interpretation, we can extend the chain in Figure
8 to include the schemas in Figures 10 and 11, which is illustrated in Figure 12.

Figure 12. Links among schemas

So far, we have considered two types of links among schemas: instance links and
similarity links. Here are examples, where 'ĸ' indicates an instance link and 'ļ'
indicates a similarity link:

Instance link: 1 l l.V.C

Similarity links: [Link] ļ [Link].C

Thus, the link between schema 1 and schema l.V.C is an instance link, with l.V.C
being an instance of 1. And the link between schema [Link] and schema [Link].
C is a similarity link, where [Link] and C are shared.
So far, we have looked only at instances of the above-across sense. And we
have only looked at the least interesting links between schemas. Let us now turn
to other senses and more interesting kinds of links.

5. The above sense

Over has a stative sense, with no PATH. It is roughly equivalent in meaning to

above (see Figure 13).

Figure 13.+DQJWKHSDLQWLQJRYHUWKH¿UHSODFH1DPH6FKHPD
 Chapter 3: Radial network 119

Schema 2 has no particular constraints on either the TR or LM. It is linked to

schema 1 in that it has the TR above the LM. However, it differs from schema
1 in two respects: First, it has no PATH and no boundaries; in other words, the
across-sense is missing. Second, it does not permit contact between the TR and
LM. The no-contact requirement can be seen in examples like The helicopter
is hovering over the hill. If the helicopter lands, it is no longer over the hill, it is
on the hill.
From time to time, linguists have suggested that schema 2 is the core meaning
of the preposition over, that is, that schema 2 is present in all the uses of over as
a preposition. It should be clear from what we have seen so far that this is false.
Since schema 2 requires no contact, it cannot be present in those cases where
contact occurs, for example, in schema 1.X.C exemplified by Sam drove over the
bridge. Schema 2 also does not occur in the cases of end-point focus, such as
schema [Link].C.E, which is exemplified by Sam lives over the hill. In this case,
the TR is not above the LM.
One of the instances of schema 2 is the case where the TR is one-dimensional
(which we will abbreviate as 1DTR – see Figure 14).

Figure 14. The power line streches over the yard. Name: Schema 2.1DTR

This schema is a minimal variant of schema [Link], exemplified by The bird flew
over the yard, as shown in Figure 2. The extended path in Figure 2 corresponds
to the one-dimensional solid trajector in Figure 14. We will call this kind of link
between schemas a transformational link. This particular link between an extended
path (X.P) and a one-dimensional trajector (1DTR) will be represented as:

X.P ļ 1DTR

This relationship is not directly reflected in the naming system for schemas that
we have adopted. However, we can state the relationship more systematically if
we do a little renaming of a sort that reflects image-schema decompositions. Let
us use ABV for the ‘above’ subschema. And let us use PATH (P) for the ‘across’
subschema. Schema 1 would be renamed ABV.P, and schema [Link] of Figure
120 Claudia Brugman and George Lakoff

2 would be renamed [Link].X.P. This name would reflect the fact that in this
schema the TR is moving above (ABV) the LM, along a path (P), where the
landmark is extended (X) and there is no contact between TR and LM (NC).
Correspondingly, schema 2 would be renamed [Link], and schema 2.1DTR in
Figure 15 would be renamed [Link].1DTR.

Schema [Link] = [Link].X.P

Schema 2.1DTR = [Link].1DTR

This decomposition displays the relationship between the schemas directly. The
schemas are transforms of one another, given the transformational link X.P ļ
1DTR.
It is important to bear in mind the difference between similarity links and
transformational links. In the case of similarity links, the link is defined by shared
subschemas. In the relationship described above, there are, indeed, shared sub-
schemas: both schemas contain [Link]. But the transformational link is not a
matter of shared subschemas, but of related subschemas.
The links among the schemas that we have described so far can be represented
by the following diagram in Figure 15.

Figure 15. Links among schemas

6. The covering senses

There is a group of schemas for over that have to do with covering. This group
is linked to the grid of Figure 15 in two ways. The basic covering schema is a
variant of schema 2, where the TR is at least 2-dimensional and extends across
the boundaries of the LM.
 Chapter 3: Radial network 121

Figure 16. The board is over the hole. Name: Schema 3

There are two differences between schema 2 and schema 3. The first is that schema
2 is unspecified for the dimension of the trajector, while schema 3 must be at least
2-dimensional. But while schema 2 requires noncontact, schema 3 is neutral with
respect to contact, allowing either contact or lack of it.
There is a minimal variant of schema 3 in which the TR moves to the position
in schema 3. This schema is composed of schema 3 plus a path (P) indicating
motion to the final position.
Schema 3.P.E (see Figure 17) is linked to schema 1. It shares motion of the
TR above and across the LM. It also shares a lack of specification for contact.
Schema 3.P.E differs from schema 1 in two ways. It is specified for the dimension
of the trajector and it has end-point focus, which indicates that the final state is
that of schema 3.

Figure 17. The city clouded over. Name: Schema 3.P.E.

There are two covering schemas in which over is paired with a mass quantifier
that quantifies regions of the landmark, e.g., all, most, a lot of, entire, etc. The
quantifier all may combine with over in this sense to form the unit all over. The
first of these two schemas has a multiplex (MX) trajector, that is, a trajector made
up of many individuals.

He has freckles over most of his body.

There are specks of paint all over the rug.
There is sagebrush over the entire valley floor.
122 Claudia Brugman and George Lakoff

In these cases, the individuals – the individual hairs, specks of paint, and bushes
– don’t completely cover the part of the landmark quantified over. Rather, the
landmark has small regions which jointly cover its surface (or most of it), and
there is at least one trajector in each region.

Figure 18.7KHJXDUGVDUHSRVWHGDOORYHUWKHKLOO1DPH6FKHPD0;

The relationship between schema 3 and schema [Link] (see Figure 18) is the
relationship between a continuous region (or mass) and a multiplex entity. Such
relationships are very common in language. Compare cows (multiplex) and cattle
(mass). Quantifiers like all and most can occur with either masses (all gold, most
wine) or multiplex entities (all ducks, most trees). The relationship between mul-
tiplex entities and masses is a natural visual relationship. Imagine a large herd of
cows up close – close enough to pick out the individual cows. Now imagine your-
self moving back until you can no longer pick out the individual cows. What you
perceive is a mass. There is a point at which you cease making out the individuals
and start perceiving a mass. It is this perceptual experience upon which the rela-
tionship between multiplex entities and masses rests. The image transformation
that relates multiplex entities and masses characterizes the link between schema
3 and schema [Link]. We can characterize that transformational link as follows:

MX ļ MS

There is a second covering schema for over in which over is associated with a
mass quantifier. It is a minimal variant on schema [Link] in which the points
representing the multiplex entity of [Link] are joined to form a path (P) which
‘covers’ the landmark. Examples are:

I walked all over the hill.

We’ve hiked over most of the Sierras.
I’ve hitchhiked over the entire country.

We can represent this schema in Figure 19. This schema is linked to schema [Link]
by an image transformation that forms a path through a collection of points. We
will represent this transformational linkage as:

MX ļ MX.P
 Chapter 3: Radial network 123

Schema [Link].P is also minimally linked to schema 3.P. In schema 3.P, the land-
mark is gradually covered as the trajector moves along the path. This is also true
in schema [Link].P.

Figure 19.,ZDONHGDOORYHUWKHKLOO1DPH6FKHPD0;3

The covering schemas all have variants in which the TR need not be above (that
is, higher than) the LM. In all cases, however, there must be an understood view-
point from which the TR is blocking accessibility of vision to at least some part
of the landmark.

There was a veil over her face.

As the rain came down, it froze and ice spread all over the windshield.
There were flies all over the ceiling.
The spider had crawled all over the ceiling.

We will refer to these as rotated (RO) schemas, though with no suggestion that
there is actual mental rotation degree-by-degree involved. One might suggest
that instead of rotation from the vertical, there is simply a lack of specification
of orientation. If there were, we would expect that the contact restrictions would
be the same in all orientations. However, they are not. The rotated versions of the
MX schemas – [Link] and [Link].P – require contact, while the unrotated versions
do not. Here are some typical examples that illustrate the distinction:

Superman flew all over downtown Metropolis. (TR above LM,

noncontact)
*Superman flew all over the canyon walls. (TR not above LM,
noncontact)
Harry climbed all over the canyon walls. (TR not above LM, contact)

Thus, Superman’s flying alongside the canyon walls does not constitute flying
over them.
We will add ‘RO’ to the names of the unrotated covering schemas to yield
names for the corresponding covering schemas. The rotated covering schemas
124 Claudia Brugman and George Lakoff

have the following names: [Link], [Link], [Link], and [Link]. Figure 20
is a diagram indicating the links among the covering schemas and the links to
the other over schemas. And Figure 21 indicates the overall linkage among the
schemas discussed so far.

Figure 20. Links among covering schemas

Figure 21. Links among schemas discussed so far

 Chapter 3: Radial network 125

7. The reflexive schemas

Perhaps the most remarkable of the discoveries made by Lindner (1981, 1982) was
the discovery of reflexive trajectors. The concept can be illustrated most simply
using the example of out. The simplest use of out occurs in cases like Harry ran
out of the room. We can represent this by the schema in Figure 22.

Figure 22. Harry ran out of the room.

In this diagram the container (the room) is the landmark, and the trajector (Harry)
moves from the interior to the exterior of the room. But this schema won’t do for
cases of out like:

The syrup spread out.

The posse spread out.
They stretched out the taffy.
We rolled out the carpet.

Here the relevant trajectors are the syrup, the posse, the taffy, and the carpet. But
they are not moving out with respect to any other landmark. Take the case of the
syrup. Pour some syrup on a table. It will have a certain outer boundary at first;
but the boundary moves. Some of the syrup that was inside the initial boundary is
now outside that initial boundary. The syrup, or at least part of it, is moving ‘out’
relative to its own prior boundary. We can schematize this in Figure 23.

Figure 23. The syrup spread out.

126 Claudia Brugman and George Lakoff

In short, the syrup is its own landmark: TR = LM. Such a relation between a
landmark and a trajector is called reflexive. Since there is only one entity under
consideration, it is referred to as a reflexive trajector.
The '=' in ‘TR = LM’ is not strict identity; it is ‘identity’ of part of a bounded
mass relative to itself as it used to be bounded. As we will see below, there are
several ways in which ‘TR = LM’ can be realized. An important one is when parts
of a single entity act as TR and other parts of the same entity act as LM. This kind
of reflexive trajector occurs in the case of over. Consider examples like:

Roll the log over.

Here a major part (roughly half) of the log is moving above and across the rest.
That is, half the log is acting as landmark and the rest as trajector. The same is
true in a case like

Turn the paper over.

Both of these are variations on schema 1; they differ only in that LM = TR in

the sense just described.

Figure 24. Roll the log over. Name: Schema 4

We can represent the schema for these cases in Figure 24. Schema 4 can be viewed
as a transform of schema 1, with schema 4 adding the condition TR = LM. We
will represent such a transformational link as

NRF ļ RF

where ‘NRF’ means “nonreflexive” and ‘RF’ means “reflexive”. If we had cho-
sen to name schema 4 according to its status as a variant of schema 1, we would
have called it [Link].
The path of over in schema 4 traces a semi-circle above and across other
parts of the thing being moved. We will refer to this as a reflexive path. There
is a variant on schema 4 in which no part of the thing moving moves above or
across any other part; instead, the entity as a whole traces the reflexive path. This
occurs in cases like
 Chapter 3: Radial network 127

The fence fell over.

Sam knocked over the lamp.

These are cases where the TR is initially vertical and moves so as to follow the
last half of a reflexive path (RFP). The schema represented is as in Figure 25. The
relationship between schemas 4 and [Link] can be stated as follows: In schema 4,
half of the TR follows the whole reflexive path; in schema [Link], all of the TR
follows the last half of the reflexive path.

Figure 25. The fence fell over. Name: Schema [Link]

This schema is not only a variant of schema 4. It is also a minimal variant of one
of the most common instances of schema 1, the instance that characterizes over
in The dog jumped over the fence. In this case, there is a vertical landmark and
the path of the trajector both begins and ends on the ground (G). This results in
a semi-circular path (see Figure 26). If we take the reflexive transform of this
schema, letting TR = LM, we get the schema of Figure 25 – schema [Link]. Thus,
schema [Link] has close links to two other schemas.

TR

Figure 26.7KHGRJMXPSHGRYHUWKHIHQFH1DPH91&*

8. The excess schema

When over is used as a prefix, it can indicate excess, as in:

The bathtub overflowed.

I overate.
Don’t overextend yourself.
128 Claudia Brugman and George Lakoff

The excess schema for over is a variation on one of the instances of schema 1, in
particular, schema 1.X.C.E of Figure 11. In schema 1.X.C.E, there is an extended
landmark; the trajector has moved across the boundaries of the landmark, and
there is focus on the end-point, which is past the boundary. The excess schema has
all these characteristics. It differs from schema 1.X.C.E in three respects. First,
it is oriented vertically, with the end-point at the top. Second, it is understood as
indicating amount via the MORE¬IS¬UP metaphor (see Lakoff and Johnson 1980).
Third, the path is taken as indicating the course of an activity, via the ACTIVITY¬
IS¬A¬JOURNEY metaphor. Fourth, the boundary point is taken as the upper limit
of what is normal for that activity. Thus, being beyond that boundary point indi-
cates excess.
The excess schema is thus not merely an image-schema, but a complex image-
schema that makes use of an image-schema (schema 1.X.C.E), an orientation
transformation from horizontal to vertical, two metaphors (MORE¬IS¬UP and AN¬
ACTIVITY¬IS¬A¬JOURNEY), and propositional information about what is normal. In
link diagrams, we will refer to the excess schema as schema 5.

9. The repetition schema

One of the most common uses of over is to indicate repetition, as in

Do it over.

Over here is used an adverb. As in the case of the over of excess, the over of rep-
etition makes use of a complex schema built on an instance of schema 1, namely,
schema 1.X.C. This schema has an extended landmark, and indicates motion above
and across it (see Figure 5). The repetition schema uses schema 1.X.C, and adds
two metaphors to it. Again, the path is metaphorically understood as the course
of the activity. This is via the very general ACTIVITY¬IS¬A¬JOURNEY metaphor. The
landmark is understood metaphorically as an earlier completed performance of
the activity. This is a special-purpose metaphor, which is, to my knowledge, used
only in this complex schema. This is the part of the repetition schema for over
that is not motivated by an occurrence elsewhere in the conceptual system. It is
what makes this sense of over somewhat idiosyncratic.
At this point, we are in a position to give a link diagram that shows a good
deal of the complexity of over. In that diagram, we will refer to the repetition
schema as schema 6. Figure 27 displays all the links we have discussed so far. A
number of additional metaphorical links will be discussed below.
 Chapter 3: Radial network 129

Figure 27.

Figure 27 shows what is meant by a radial structure. schema 1 occupies a central

position; it and its instances are of primary importance in the system of links. The
links correspond to what Wittgenstein called ‘family resemblances.’ The links
are sometimes defined by shared properties. But frequently they are defined not
by shared properties, but by transforms or by metaphors.

10. Some metaphorical senses

It is extremely common for metaphors to apply to image-schemas. A great many

metaphorical models use a spatial domain as their source domain. Among the
most common source domains for metaphorical models are containers, orienta-
tions, journeys (with paths and goals), vertical impediments, etc. In this section,
we will give a number of cases where over has a metaphorical sense based on an
image-schema discussed above.
She has a strange power over me.
130 Claudia Brugman and George Lakoff

This is an instance of a very common metaphor: CONTROL¬IS¬UP¬LACK¬OF¬CONTROL¬

IS¬DOWN (see Lakoff and Johnson 1980). Over in this sentence is an extension of
schema 2 (Figure 14), where the trajector is simply above the landmark.

Sam was passed over for promotion.

Here we have an instance of schema 1 (Figure 1). Two metaphorical mappings

apply to it. The first is CONTROL¬IS¬UP¬LACK¬OF¬CONTROL¬IS¬DOWN. This entails
that the person who passed over Sam was in control of Sam’s status. The second
metaphor the applies to this schema is another common one: CHOOSING¬IS¬TOUCH
ING. This occurs in such sentences as He was tapped for service and The boss
handpicked his successor. Since the schema indicates that there is no contact, it
is entailed that Sam was not chosen.
We are now in a position to make sense of the difference between overlook
and oversee.

You’ve overlooked his accomplishments.

We need to find someone who can oversee this operation.

The over in overlook is based on schema 2.1 DTR (Figure 5). There are two
metaphors involved. The first is a metaphor for understanding vision: SEEING¬IS¬
TOUCHING. This occurs in examples like I couldn’t take my eyes off of her, Her
eyes picked out every detail of the pattern, He undressed her with his eyes, and
He fixed his gaze on the entrance. According to this metaphor, one’s gaze goes
from one’s eyes to what one sees. You see whatever your gaze touches. Under the
metaphorical mapping, the path in schema 2.1DTR is the gaze. Since there is no
contact in schema 2.1DTR, the metaphorical gaze doesn’t touch the landmark; thus
the subject of overlook is not looking at, and therefore does not see, the landmark.
The second metaphor is the general MIND AS BODY metaphor (see Sweetser 1984).
The relevant aspect of that metaphor is the part in which LOOKING¬AT¬SOMETHING¬
IS¬TAKING¬IT¬INTO¬CONSIDERATION. Accordingly, I‘ll take a look at it normally
entails I‘ll consider it. Therefore, to overlook someone’s accomplishments is not
to take them into consideration.
The over in oversee is based on schema 2 (Figure 15), in which the TR is
above the LM. There are a metaphor and a metonymy that are relevant to this
example. The metaphor is CONTROL¬IS¬UP. Thus, the one who does the oversee-
ing has control over the persons overseen. The metonymy is SEEING¬SOMETHING¬
DONE¬STANDS¬FOR¬MAKING¬SURE¬THAT¬IT¬IS¬DONE. This metonymy is based on
an idealized model in which making sure of something typically involves seeing
it. Because of this metonymic relation. See that he gets his money means Make
sure that he gets his money. Thus, to oversee means to be in control and make
sure that something is done.
 Chapter 3: Radial network 131

We can now compare overlook to look over.

Look over my corrections, but don’t overlook any of them.

The over in look over is based on schema [Link].P (Figure 18), and the SEEING¬IS¬
TOUCHING metaphor. The resulting complex schema is one in which the subject’s
gaze traces a path that ‘covers’ the direct object, corrections. In the resulting
schema, the gaze does make contact with the landmark. The MIND AS BODY meta-
phor again yields a sense of look in which looking at something involves taking
it into consideration. Thus, when one looks over X, one directs one’s attention to
a representative sampling that ‘covers’ X, and one takes into consideration each
subpart that one directs attention to.

11. Motivation

Before we go on, it is worth commenting on what is and what is not being explained
in these analyses. We are not explaining why oversee, overlook, and look over
mean what they mean. Their meanings cannot be predicted from the meanings
of over, look, and see. But their meanings are not completely arbitrary. Given
the range of spatial meanings of over, and given the metaphors present in the
conceptual system that English is based on, it makes sense for these words to
have these meanings. We are explaining just why it makes sense and what kind
of sense it makes.
In each of these cases, the metaphorical and metonymic models exist in the
conceptual system independently of the given expression. For example, we under-
stand seeing metaphorically in terms of a gaze that goes out of one’s eyes and
touches the object seen. This metaphorical understanding is present regardless of
whether any of the expressions just discussed have those meanings. Similarly, the
schemas for over exist for expressions in the spatial domain independently of the
existence of oversee, overlook, and look over. What one learns when one learns
these words is which of the independently existing components of their mean-
ing are actually utilized. Each of these expressions is a specialized ‘assembly’ of
independently existing parts. The only arbitrariness involved is the knowledge
that such an assembly exists.
The psychological claim being made here is that it is easier to learn, remember,
and use such assemblies which use existing patterns than it is to learn, remember,
and use words whose meaning is inconsistent with existing patterns. What is being
explained is not why those expressions mean what they mean, but why those are
natural meanings for them to have. Thus, if one is going to have a word that means
‘to fail to take into consideration,’ it is more natural to use overlook than to use
132 Claudia Brugman and George Lakoff

an existing unrelated word like sew, or a complex word whose components are in
conflict with the meaning, such as underplan, or taste at, or rekick. It is common
sense that such expressions would not be used with such a meaning, and we are
characterizing the nature of that ‘common sense.’
As we have mentioned before, such an explanation requires going beyond the
predictable/arbitrary dichotomy. It requires introducing the concept of motivation.
Thus, the meaning of overlook, though not predictable, is motivated – motivated
by one of the spatial schemas for over and by two metaphors in the conceptual sys-
tem. Similarly, all of the noncentral schemas for over in the chain given in Figure
27 are motivated – motivated by other senses and by principles of linking.

12. More metaphorical senses

There are some additional common metaphorical senses of over that are worth
discussing. Take get over in

Harry still hasn’t gotten over his divorce.

This use of over is based on schema [Link].C (Figure 6), and two metaphors. In the
first metaphor, obstacles are understood in terms of vertical landmarks – which
may be extended or not. This metaphorical model is the basis for expressions such
as There is nothing standing in your way. The second metaphorical model is one
that understands LIFE as a JOURNEY. This occurs in sentences like It’s time to get
on with your life. In the above use, the divorce in an obstacle (metaphorically, a
vertical extended landmark) on the path defined by life’s journey.

Pete Rose is over the hill.

Over the hill makes use of schema [Link].C.E (Figure 10) and a metaphor for
understanding a career in terms of a journey over a vertical extended landmark
like a hill. In this metaphorical model of a career, one starts at the bottom, may
go all the way to the top, and then goes downhill. Thus, over the hill means that
one has already reached and passed the peak, or “high point,” of one’s career and
will never have that high a stature again.

The rebels overthrew the government.

This is an instance of schema [Link] (Figure 25) and the CONTROL¬IS¬UP meta-
phor. Before the event takes place, the government is in control (metaphorically
upright), and afterwards it is not in control (metaphorically down).
He turned the question over in his mind.
This is an instance of schema 4 (Figure 24), plus an instance of the MIND AS
 Chapter 3: Radial network 133

BODY metaphor in which THINKING¬ABOUT¬SOMETHING¬IS¬EXAMINING¬IT. This

metaphorical model occurs in such sentences as Let us now examine the question
of factory chickens. In examining a physical object, one turns it over in order to
get a look at all sides of it. Questions are metaphorically understood as having
sides, and when one turns a question over in one’s mind, one is examining all
sides of it.

The play is over.

Here we have an instance of schema 1.X.C.E (Figure 11). In general, activities

with a prescribed structure are understood as extended landmarks, and performing
such an activity is understood metaphorically as traveling along a prescribed path
over that landmark. When one gets to the end, the activity is over. Thus, games,
plays, and political campaigns can be characterized at their end as being over.

13. Image-schemas as links between perception and reason

Two of our major sources of information are vision and language. We can gain
information through either perceiving something directly or being told it. And
we can reason about that information, no matter what its source. We can even
reason using information from both sources simultaneously, which suggests that
it is possible for us to encode information from both sources in a single format.
We would like to suggest that image-schemas provide such a format.
It is our guess that image-schemas play a central role in both perception and
reason. We believe that they structure our perceptions and that their structure is
made use of in reason. The analysis of over that we have just given is rich enough
for us to discuss such questions in some detail. Let us begin with the following
question. Are the image-schema transformations we have discussed natural, and
if so, what is the source of their ‘naturalness’?

14. The nature of image-schema transformations

There are certain very natural relationships among image-schemas, and these
motivate polysemy, not just in one or two cases, but in case after case throughout
the lexicon. Natural image-schema transformations play a central role in forming
radial categories of senses. Take, for example, the end-point-focus transformation.
It is common for words that have an image-schema with a path to also have the
corresponding image-schema with a focus on the end-point of the path. We saw
this in over in cases like
134 Claudia Brugman and George Lakoff

Sam walked over the hill. (path)

Sam lives over the hill. (end-of-path)

Pairs such as this are common.

Harry walked through that doorway. (path)

The passport office is through that doorway. (end-of-path)

Sam walked around the comer. (path)

Sam lives around the comer. (end-of-path)

Harriet walked across the street. (path)

Harriet lives across the street. (end-of-path)

Mary walked down the road. (path)

Mary lives down the road. (end-of-path)

Sam walked past the post office. (path)

Sam lives past the post office. (end-of-path)

It should be noted that although such pairs are common, they are not fully
productive.

Sam walked by the post office. (path)

Sam lives by the post office. (= near,  end-of-path)

Here, by has a path schema, but no corresponding end-point schema.

Sam ran from the house. (path)

Sam stood three feet from the house. (end-of-path)

Sam ran to the house. (path)

*Sam stood (three feet) to the house. ( end-of-path)

From allows both path and end-of-path schemas, but to only allows a path
schema.
Path schemas are so naturally related to end-point schemas that people some-
times have to think twice to notice the difference. The same is true of the schema
transformation that links multiplex and mass schemas. It is natural for words that
have a mass schema to also have a multiplex schema.
 Chapter 3: Radial network 135

All men are mortal. (MX)

All gold is yellow. (MS)

She bought a lot of earrings. (MX)

*She bought a lot of jewelry. (MS)

This schema transformation, of course, doesn’t hold for all quantifiers:

She bought two earrings. (MX)

*She bought two jewelry. (MS)

There are also verbs which have both schemas:

He poured the juice through the sieve. (MS)

The fans poured through the gates. (MX)

This will also work for other verbs of liquid movement, such as spill, flow, etc.

The wine spilled out over the table. (MS)

The fans spilled out over the field. (MX)

There is a special case of the multiplex-mass transformation in which the multi-

plex entity is a sequence of points and the mass is a one-dimensional trajector. A
variety of prepositions permit both schemas.

There are guards posted along the road. (MX)

There is a fence along the road. (1DTR)

He coughed throughout the concert. (MX)

He slept throughout the concert. (1DTR)

There were stains down his tie. (MX)

There were stripes down his tie. (1DTR)

There is a natural relationship not only between a one-dimensional trajector and a

sequence of points. There is also a natural relationship between a one-dimensional
trajector and a zero-dimensional moving trajector that traces a path.

Sam went to the top of the mountain. (0DMTR)

The road went to the top of the mountain. (1DTR)
136 Claudia Brugman and George Lakoff

Sam ran through the forest. (0DMTR)

There is a road through the forest. (1DTR)

Sam walked across the street. (0DMTR)

There was a rope stretched across the street. (1DTR)

Finally, there is a natural relationship between nonreflexive and reflexive trajec-

tors. Here are some examples:

He stood apart from the crowd. (NRF)

The book fell apart. (RF)

She walked up to me. (NRF)

Let’s cuddle up. (RF)

She poured the syrup out of the jar. (NRF)

The syrup spread out over the pancakes. (RF)

Let us consider for a moment what is natural about these image-schema

transformations.

Path-focus/end-point-focus: It is a common experience to follow the path

of a moving object until it comes to rest, and then to focus on where it is.
This corresponds to the path-focus / end-point-focus transformation.

Multiplex/mass: As one moves further away, a group of individuals: at a

certain point begins to be seen as a mass. Similarly, a sequence of points
is seen as a continuous line when viewed from a distance.

0DMTR/1DTR: When we perceive a continuously-moving object, we can

mentally trace the path it is following. This capacity is reflected in the
transformation linking zero-dimensional moving trajectors and a one-
dimensional trajector.

NRF/RF: Given a perceived relationship between a TR and a LM which

are two separate entities, it is possible to perceive the same relationship
between (a) different parts of the same entity or (2) earlier and later loca-
tions of the same entity, where one part or location is considered LM and
the other TR.

In short, these schema transformations are anything but arbitrary. They are direct
reflections of our visual experiences.
 Chapter 3: Radial network 137

The fact that image-schemas are a reflection of our sensory and general spatial
experience is hardly surprising, yet it plays a very important role in the theory of
image-schemas. Perhaps we can see that significance most easily by contrasting
the image-schema transformations we have described with the names we have
given to them. Take the transformation name ‘MX ļ MS.’ The names ‘MX’ and
‘MS’ are arbitrary relative to the character of what they name: a group of indi-
vidual entities and a mass. The transformation is a natural relationship, but the
name of the transformation is just a collection of arbitrary symbols.
This distinction is important because of certain versions of the Symbol Manipu-
lation Paradigm. On one theory of image-representation – the ‘prepositional theory’
– visual scenes are represented by arbitrary symbols which are linked together in
network structures. Arbitrary symbols such as X and Y are taken as standing for
some aspect of a scene, such as a point or an edge or a surface or an entire object.
Other symbols are used to express relations among these symbols, for example,
‘ABV(X,Y)’ and ‘C(X,Y)’ might represent relations which are supposed to cor-
respond to ‘X is above Y’ and ‘X is in contact with Y,’ but which, so far as the
computer is concerned, are just symbols. Such a symbolization describes how
various parts – points, edges, surfaces, etc. – are related to one another. Objects
in a scene are described using such symbolizations.
According to the Symbol Manipulation Paradigm as applied to visual infor-
mation and mental imagery (Pylyshyn 1981), only such propositional representa-
tions are mentally real, while images are not real. This view stems from taking
the Symbol Manipulation Paradigm very seriously. Since digital computers work
by the manipulation of such arbitrary symbols, the strong version of the Symbol
Manipulation Paradigm requires not only that visual perception and mental imag-
ery be characterizable in such a ‘propositional’ form, but also that such symbolic
representations, and only those, are mentally real.
The names that we have given to the image schemas, as well as to the image-
schema transformations, are in keeping with feature representations. They have
the properties that (1) they are arbitrary, in the sense that the internal structure
of symbols plays no role in how the symbols interact or what they mean; (2) they
are inherently meaningless, and have to be assigned meanings; and (3) they are
finitary in nature. Such feature-style names are in these respects opposite from
the corresponding image-schemas, which (1) are nonarbitrary, in the sense that
they have an internal structure that plays a crucial role in what they mean and
how they interact; (2) they are inherently meaningful; and (3) they are analog
in nature.
Suppose that, instead of merely using such symbols as convenient names, we
chose to take such a use of symbols seriously, as one would have to if one were
to adopt the Symbol Manipulation Paradigm for Cognitive Science. According
to that paradigm, topological representations such as image-schemas are not
available as cognitive representations; all that is available are symbolic repre-
138 Claudia Brugman and George Lakoff

sentations, which would look like the symbolic names we have given and would
have their properties. The Symbol Manipulation Paradigm would thus make the
implicit claim that the cognitively natural image-schema transformations of the
sort we described did not exist. In their place would be arbitrary transformations
relating the names we have given. Instead of a natural explanation of types of
polysemy, we would have no explanation at all, but only an arbitrary description.
We consider the lack of such explanatory force intolerable.
For instance, consider the relationship between the over of The bird flew over
the yard and that of The power line stretches over the yard. These are adjacent
senses in the network, linked by the natural relationship between a zero-dimen-
sional moving trajector and a corresponding one-dimensional stationary trajec-
tor. If there were no such natural relation between the senses, they would not be
adjacent in the network. Thus, the configuration of the network is anything but
arbitrary; it is determined by an account of what constitutes a “natural” link type.
Since it is the topological character of the representations that makes such rela-
tionships natural, it is that topological nature that makes the configurations of the
networks nonarbitrary. If symbolic feature representations are substituted for the
topological representations, then the naturalness of the relationships disappears
and, with it, the explanation for what is linked to what. In short, there are natural
reasons for the extensions of certain senses to other senses, and those reasons
must be given in a cognitively adequate account of polysemy.

16. Conclusion

Systematic polysemy of the sort we have just seen is a pervasive phenomenon in

language. It is so common and automatic that we often do not even notice it. Yet,
as we have seen, simple basic processes like image-schema transformations and
metaphors can interact to form rather large networks that characterize the natural
relationships among the senses of polysemous words.
Our primary conclusion is that lists will not do; networks of the kind we
have described are needed to characterize the relationships among the senses of
polysemous words.
Our secondary conclusion is that features will not do. Oriented topological
representations using image-schemas are necessary for two reasons: (1) to account
for the range of scenes that concepts like over can fit, and (2) to account for the
naturalness of image-schema transformations. This suggests that the Symbol
Manipulation Paradigm, which cannot tolerate the existence of such representa-
tions using a cognitive topology, is inadequate for natural language semantics.
 Chapter 3: Radial network 139

Note
* This paper is a shortened version of a discussion of polysemy and image-schemas
that appears as case study 2 in Lakoff 1987. It is published here with permission of the
University of Chicago Press. That study, in turn, was based on Brugman’s University of
California M.A. thesis (Brugman 1981). Portions of this work were supported by grant no.
BNS-8310445 of the National Science Foundation and a grant by the Sloan Foundation
to the Institute of Cognitive Studies at the University of California, Berkeley.

References
Brugman, Claudia
1981 Story of over. M.A. thesis, University of California, Berkeley.
Lakoff, George
1987 Women, Fire, and Dangerous Things: What Categories Reveal about the
Mind. Chicago: University of Chicago Press.
Lakoff, George, and Mark Johnson
1980 Metaphors We Live By. Chicago: University of Chicago Press.
Langacker, Ronald W.
1983 Remarks on English aspect. In Tense and Aspect: Between Semantics and
Pragmatics, P. Hopper (ed.), 265–304. Amsterdam: Benjamins.
1987 Foundations of Cognitive Grammar, Vol. 1. Stanford: Stanford University
Press.
Lindner, Susan
1981 A lexico-semantic analysis of verb-particle constructions with up and out.
Ph.D. dissertation, University of California, San Diego.
1982 What goes up doesn’t necessarily come down: The ins and outs of opposites.
In Papers from the Eighteenth Regional Meeting of the Chicago Linguistic
Society, 305–323.
Pylyshyn, Zenon
1981 The imagery debate: Analogue media versus tacit knowledge. Psychological
Review 87: 16–45.
Sweetser, Eve E.
1984 Semantic structure and semantic change. Ph.D. dissertation, University of
California, Berkeley.
Chapter 4
Prototype theory

Prospects and problems of prototype theory*

Dirk Geeraerts

1. Prototype theory within linguistics

The starting-point of the prototypical conception of categorial structure is sum-

marized in the statement that

when describing categories analytically, most traditions of thought have treated

category membership as a digital, all-or-none phenomenon. That is, much work
in philosophy, psychology, linguistics, and anthropology assumes that categories
are logical bounded entities, membership in which is defined by an item’s pos-
session of a simple set of criterial features, in which all instances possessing the
criterial attributes have a full and equal degree of membership. In contrast, it has
recently been argued ... that some natural categories are analog and must be rep-
resented logically in a manner which reflects their analog structure (Rosch and
Mervis 1975: 573–574).

As we shall see in section 2, the exact definition of the concept of prototypicality

as used in linguistics is not without problems. The major part of this introduc-
tion to the prototypicality-based studies collected here will, in fact, consist of an
attempt at clarification of some of the problematic aspects of the way in which
the notion of prototype has been used in linguistics. To begin with, however, we
shall be concerned with a brief overview of the state of the art in linguistic pro-
totype theory.1
The theory originated in the mid 1970s with Eleanor Rosch’s research into the
internal structure of categories. (Overviews may be found in Rosch 1978, 1988,
and Mervis and Rosch 1981; the basic research is reported on mainly in Heider
1972; Rosch 1973, 1975, 1977; Rosch and Mervis 1975; Rosch, Simpson and Miller
1976; Rosch et al. 1976.) From its psycholinguistic origins, prototype theory has
moved mainly2 in two directions. On the one hand, Rosch’s findings and proposals
were taken up by formal psycholexicology (and more generally, information-pro-
cessing psychology), which tries to devise formal models for human conceptual
memory and its operation, and which so, obviously, borders on Artificial Intel-

Originally published in 1989 in Linguistics 27(4): 587–612. A section of the original paper describ-
ing the various contributions to the thematic issue has been omitted from the present reprint.
142 Dirk Geeraerts

ligence. Excellent overviews of the representational and experimental issues at

stake here are Smith and Medin (1981), and Medin and Smith (1984); an inter-
esting sample of current research may be found in Neisser (1987). On the other
hand, prototype theory has had a steadily growing success in linguistics since
the early 1980s, as witnessed by a number of recent monographs and collective
volumes in which prototype theory and its cognitive extensions play a major
role (Wierzbicka 1985; Lakoff 1987; Langacker 1987; Craig 1986; Holland and
Quinn 1987; Rudzka-Ostyn 1988; Lehmann 1988a; Hüllen and Schulze 1988;
Tsohatzidis 1989; Taylor 1989). It is with the latter development that we shall be
concerned with here.
Against the background of the development of linguistic semantics, prototype
theory may be defined primarily in contrast with the componential model of
semantic analysis that was current in transformational grammar and that is stereo-
typically associated with Katz and Fodor’s analysis of bachelor (Katz and Fodor
1963); in an early defense of a prototypical approach, Fillmore (1975) called this
the ‘checklist theory’ of meaning. The prototypists’ reaction against this featural
approach had, however, the negative side-effect of creating the impression that
prototypical theories rejected any kind of componential analysis. This is a mis-
conception for the simple reason that there can be no semantic description without
some sort of decompositional analysis. As a heuristic tool for the description and
comparison of lexical meanings, a componential analysis retains its value (a value
that, incidentally, it did not acquire with the advent of componential analysis as
an explicit semantic theory, but which had been obvious to lexicographers from
time immemorial). Rather, the difficulties with the neostructuralist kind of feature
analysis that grew out of structuralist field theory lie elsewhere; it is not the use
of decomposition as a descriptive instrument that causes concern, but the status
attributed to the featural analysis. Two important points have to be mentioned.
In the first place, as suggested by the quotation at the beginning of this intro-
duction, featural definitions are classically thought of as criterial, i.e. as listing
attributes that are each indispensable for the definition of the concept in question,
and that taken together suffice to delimit that concept from all others. In contrast,
prototype theory claims that there need not be a single set of defining attributes
that conform to the necessity-cum-sufficiency requirement.3
In the second place, prototype theory is reluctant to accept the idea that there
is an autonomous semantic structure in natural languages which can be studied
in its own right, in isolation from the other cognitive capacities of man. In par-
ticular, meaning phenomena in natural languages cannot be studied in isolation
from the encyclopedic knowledge individuals possess; it is precisely the presup-
position that there exists a purely linguistic structure of semantic oppositions
that enables structuralist and neostructuralist semantics to posit the existence of
a distinction between semantic and encyclopedic knowledge. Prototype theory
 Chapter 4: Prototype theory 143

tends to minimize the distinction primarily for methodological reasons: because

linguistic categorization is a cognitive phenomenon just like the other cogni-
tive capacities of man, it is important to study it in its relationship to these other
capacities. More specific arguments have also been formulated to show that the
distinction between an encyclopedic and a semantic level of categorial structure
is untenable.4 For instance, given that the flexible extendibility of prototypical
concepts is a synchronic characteristic of linguistic structure, and given the fact
that these extensions may be based indiscriminately on allegedly encyclopedic or
on allegedly semantic features, the distinction between both kinds of information
loses its synchronic relevance. Take the case of metaphor: before lion acquires
the meaning ‘brave man’, the feature ‘brave’ is not structurally distinctive within
the semasiological structure of lion, and hence, it has to be considered encyclo-
pedic according to structuralist theories. But if it can be accepted (and this is
of course the crucial point) that the metaphorical extension of lion towards the
concept ‘brave man’ is not just a question of diachronic change, but is merely an
effect of the synchronic flexibility of lexical items, the feature clearly acquires
semantic status. If, furthermore, the argument can be repeated in the sense that
such synchronic metaphorical extensions may be based on any allegedly ency-
clopedic attribute, the distinction between semantic and encyclopedic concepts
as a whole falls.5
The matter need not, to be sure, be settled here. What is important for our
introductory purposes is rather to see what exactly prototype theory objects to
in componential theories of the Katzian type. First, the suggestion that lexical
concepts are criterial in the classical sense, and second, the suggestion that there
exists a purely linguistic level of conceptual structuring that is neatly separated
from other, ‘encyclopedic’ forms of conceptual information, and that may thus be
studied autonomously, in methodological isolation from other kinds of cognitive
research. As against these points of view, prototype theory defends a non-crite-
rial conception of categorial structure, and an interdisciplinary methodological
perspective that takes into account relevant research from the other cognitive
sciences. (The very transposition of the prototypical approach from experimental
psychology to linguistics derives from this attitude.)
But this historical positioning of prototype theory with regard to its imme-
diate predecessors within the field of lexical semantics clearly does not explain
why it has turned out to be such a successful alternative. Why did (and does)
the prototypical approach appeal to a sizeable part of the linguistic community?
On the one hand, the historical development of generative grammar had raised
a considerable amount of interest in semantic matters. It should not be forgotten,
in fact, that it was only after the incorporation of a semantic component into the
transformational framework that Chomskyanism became internationally popular;
the universal appeal of the generative Standard Theory was at least partly due to the
144 Dirk Geeraerts

promises held by its Katzian semantic component. On the other hand, the promises
were not fulfilled. Within the generative paradigm, Generative Semantics (which
most strongly embodied the semantic approach) withered in favor of Autonomous
Syntax, in which semantics hardly played a role worthy of note. Outside the gen-
erative approach, formal semantics of the Montagovian kind was too narrowly
restricted to sentential meaning to be able to hold the attention of those who were
interested primarily in the internal structure of natural language categories (and
not primarily in the way these categories combine into larger unities).6 In short,
as far as semantics was concerned, there was a gap in the linguistic market of the
early 1980s that was not filled by the major approaches of the day.7
But again, recognizing that there was an interest in the semantics of natural
language categories to which prototype theory could appeal does not tell the whole
story. Why didn’t people simply stick to the componential theory popularized by
Katz, or to the rival axiomatic method of representation – even if these gradually
moved out of the centre of the linguistic attention as Autonomous Syntax and
Formal Semantics took over? In general, there are a number of methodological
requirements people nowadays expect of linguistic theories: descriptive adequacy
(mainly in the form of a broad empirical scope), explanatory depth, productivity,
and formalization. Although prototype theory rates much lower on the formaliza-
tion scale than either the axiomatic or the featural approach, its assets with regard
to the other three points are considerable.
In the first place, it tackles a number of semantic phenomena that had been swept
under the rug by the more structurally minded approaches. The fuzzy boundar-
ies of lexical categories, the existence of typicality scales for the members of a
category, the flexible and dynamic nature of word meanings, the importance of
metaphor and metonymy as the basis of that flexibility – these are all intuitively
obvious elements of the subject matter of semantics that were largely neglected
by structural semantics. It is true that they were occasionally pointed at as an
indispensable aspect of any full-fledged semantic theory: think, for instance, of
Weinreich’s remark (1966: 471) that a semantic theory should be able to deal with
‘interpretable deviance’, or Uhlenbeck’s plea (1967) for a dynamic conception of
word meaning.8 These remarks did not, however, have much effect as far as theory
formation was concerned. In particular, it is only with the advent of prototype
theory that contemporary linguistics developed a valid model for the polysemy
of lexical items. This is perhaps the single most appealing characteristic of pro-
totype theory: here at last is a descriptive approach to lexical meaning in which
our pretheoretical intuitions about gradedness, fuzziness, flexibility, clustering
of senses etc. receive due attention.
In the second place, prototype theory appears to be a productive theory not
just in the sense that its insights into the structure of lexical categories can be
easily applied in various fields of the lexicon, but also in the sense that it may be
 Chapter 4: Prototype theory 145

extended towards other aspects of linguistics. Whereas prototype theory started

with being descriptively fruitful in lexical semantics, it soon became theoretically
fruitful in the sense that other areas of linguistics were taken into consideration.
A few recent examples of such extensions may suffice: phonology (Nathan 1986),
morphology (Bybee and Moder 1983; Post 1986), syntax (Van Oosten 1986; Ross
1987), historical linguistics (Winters 1987; Aijmer 1985), markedness theory (Van
Langendonck 1986), theoretical lexicography (Geeraerts 1985c). Through these
and similar extensions,9 prototype theory has become one of the cornerstones of
Cognitive Linguistics, which tries to account for the interaction between language
and cognition on all levels of linguistic structure: one need only have a look at the
prominent place attributed to a prototypical conception of categorial structure in
Langacker (1987) (one of the basic works of the Cognitive Linguistic approach)
to appreciate its importance.10 In this sense, the development of prototype theory
into Cognitive Linguistics contains exciting promises of a unified cognitive theory
of linguistic categorization.
In the third place, the explanatory depth of prototype theory resides partly in
its generalizable character, but also in its interdisciplinary nature. The importance
of its genetic link with psycholinguistics can only be fully appreciated against the
background of the Chomskyan requirements with regard to theories of grammar.
Chomsky’s methodology is, in fact, in the awkward position of declaring linguistics
a cognitive science, but refusing to deal directly with the findings of the other sci-
ences of the mind. Roughly stated, Chomskyan linguistics claims to reveal some-
thing about the mind, but imperviously prefers a strictly autonomist methodology
over the open dialogue with psychology that would seem to be implied by such a
claim. Prototype theory’s linguistic application of psycholinguistic findings, on
the other hand, takes the Chomskyan ideal of cognitive explanatory depth to its
natural consequences, viz. of giving up the methodological autonomy of linguis-
tics in favor of an interdisciplinary dialogue with the other cognitive sciences.11
Prototype theory takes the cognitive claims of Chomskyanism methodologically
seriously by its interdisciplinary openness. This is all the more important at a
moment when Cognitive Science is emerging as an interdisciplinary cluster of
psychology, neuroscience, Artificial Intelligence, and philosophy. It is probably
one of the reasons for the appeal of prototype theory that its interdisciplinary
connections hold the promise of linking linguistics to the most important devel-
opment that the human sciences are currently witnessing.
146 Dirk Geeraerts

2. Definitional problems, first series: ‘Prototype’ as a prototypical

notion

The appeal of prototype theory should not, however, obscure the fact that the exact
definition of prototypicality is not without problems. The purpose of this section
(and the following) is to analyze the sources of the confusion by making clear that
prototypicality is itself, in the words of Posner (1986), a prototypical concept. As a
first step, we shall have a look at four characteristics that are frequently mentioned
(in various combinations) as typical of prototypicality. In each case, a quotation
from early prototype studies is added to illustrate the point.
(i) Prototypical categories cannot be defined by means of a single set of crite-
rial (necessary and sufficient) attributes:

We have argued that many words ... have as their meanings not a list of necessary
and sufficient conditions that a thing or event must satisfy to count as a member
of the category denoted by the word, but rather a psychological object or process
which we have called a prototype (Coleman and Kay 1981: 43).

(ii) Prototypical categories exhibit a family resemblance structure, or more gen-

erally, their semantic structure takes the form of a radial set of clustered and
overlapping meanings:12

The purpose of the present research was to explore one of the major structural
principles which, we believe, may govern the formation of the prototype structure
of semantic categories. This principle was first suggested in philosophy; Wittgen-
stein (1953) argued that the referents of a word need not have common elements
to be understood and used in the normal functioning of language. He suggested
that, rather, a family resemblance might be what linked the various referents of a
word. A family resemblance relationship takes the form AB, BC, CD, DE. That
is, each item has at least one, and probably several, elements in common with
one or more items, but no, or few, elements are common to all items (Rosch and
Mervis 1975: 574–575).

(iii) Prototypical categories exhibit degrees of category membership; not every

member is equally representative for a category:

By prototypes of categories we have generally meant the clearest cases of category

membership defined operationally by people’s judgments of goodness of member-
ship in the category ... we can judge how clear a case something is and deal with
categories on the basis of clear cases in the total absence of information about
boundaries (Rosch 1978: 36).

(iv) Prototypical categories are blurred at the edges:

 Chapter 4: Prototype theory 147

New trends in categorization research have brought into investigation and debate
some of the major issues in conception and learning whose solution had been
unquestioned in earlier approaches. Empirical findings have established that ...
category boundaries are not necessarily definite (Mervis and Rosch 1981: 109).

As a first remark with regard to these characteristics, it should be noted that

they are not the only ones that may be used in attempts to define the prototypi-
cal conception of categorization. Two classes of such additional features should
be mentioned.
On the one hand, there are characteristics that do not pertain (as the four
mentioned above) to the structure of categories, but that rather pertain to the epis-
temological features of so-called non-Aristotelian categories.13 For instance, the
view that prototypical categories are not ‘objectivist’ but ‘experiential’ in nature
(Lakoff 1987) envisages the epistemological relationship between concepts and
the world rather than the structural characteristics of those concepts. In particu-
lar, it contrasts the allegedly classical view that ‘categories of mind ... are simply
reflections of categories that supposedly exist objectively in the world, indepen-
dent of all beings’, with the view that ‘both categories of mind and human reason
depend upon experiential aspects of human psychology’ (Lakoff 1982: 99). Such
an epistemological rather than structural characterization of natural concepts also
has a methodological aspect to it; it entails that prototypical categories should not
be studied in isolation from their experiential context. While such an epistemo-
logical or methodological conception of prototypical categorization is extremely
valuable, we shall take a structural point of view in the following pages; we shall
try to determine whether it is possible to give a coherent, structurally intrinsic
characterization of prototypical categories.
On the other hand, there are structural characteristics of prototypical con-
cepts that can be reduced to the four basic structural features mentioned above.
For instance, in my own work on prototypical categorization, I have repeatedly
stressed the flexibility of prototypical concepts (1983, 1985a), together with the
fact that a distinction between semantic and encyclopedic components of lexical
concepts cannot be maintained in the case of prototypical concepts (1985b). But
the flexibility of prototypical categories is linked in a straightforward manner with
the fourth characteristic: uncertainties with regard to the denotational boundaries
of a category imply that it need not be used in a rigidly fixed manner. Similarly,
the absence of a clear dividing line between encyclopedic and purely semantic
information follows from this very flexibility together with the first and second
characteristic. As illustrated in the previous section, the possibility of incorporat-
ing members into the category that do not correspond in every definitional respect
with the existing members entails that features that are encyclopedic (non-defini-
tional) with regard to a given set of category members may turn into definitional
148 Dirk Geeraerts

features with regard to a flexibly incorporated peripheral category member. The

resemblance between central and peripheral cases may be based on allegedly
encyclopedic just as well as on allegedly ‘semantic’ features. In short, features of
prototypicality that are not included among the ones mentioned in (i)–(iv) may
often be reduced to those four, and this in turn justifies a preliminary restriction
of the discussion to the latter.
A second remark with regard to the four characteristics is concerned with the
fact that they are systematically related along two dimensions. On the one hand,
the third and the fourth characteristic take into account the referential, exten-
sional structure of a category. In particular, they have a look at the members of a
category; they observe, respectively, that not all referents of a category are equal
in representativeness for that category, and that the denotational boundaries of
a category are not always determinate. On the other hand, these two aspects
(centrality and non-rigidity) recur on the intensional level, where the definitional
rather than the referential structure of a category is envisaged. For one thing,
non-rigidity shows up in the fact that there is no single necessary and sufficient
definition for a prototypical concept. For another, family resemblances imply
overlapping of the subsets of a category. To take up the formulation used in the
quotation under (ii) above, if there is no definition adequately describing A, B,
C, D, and E, each of the subsets AB, BC, CD, and DE can be defined separately,
but obviously, the ‘meanings’ that are so distinguished overlap. Consequently,
meanings exhibiting a greater degree of overlapping (in the example: the senses
corresponding with BC and CD) will have more structural weight than meanings
that cover peripheral members of the category only. In short, the clustering of
meanings that is typical of family resemblances implies that not every meaning is
structurally equally important (and a similar observation can be made with regard
to the components into which those meanings may be analyzed). The systematic
links between the characteristics mentioned at the beginning are schematically
summarized in Table 1.
As a third remark, it should be noted that the four characteristics are often
thought to be co-extensive, in spite of incidental but clear warnings such as Rosch
and Mervis’s remark that a family resemblance structure need not be the only
source of prototypicality (1975: 599). Admittedly, it is easy to consider them to
be equivalent; already in the quotations given above, partial reasons for their
mutual interdependence can be found. More systematically, the following links
between the four characteristics might be responsible for the idea that prototypi-
cality necessarily entails the joint presence of all four.
 Chapter 4: Prototype theory 149

Table 1. Characteristics of prototypicality

NON EQUALITY¬ NON RIGIDITY¬

differences in structural weight flexibility and vagueness

EXTENSIONALLY degrees of representativity absence of clear boundaries

INTENSIONALLY clusters of overlapping senses absence of classical
definition

First, linking the first to the second characteristic is the argument mentioned above:
if there is no single definition adequately describing the extension of an item as a
whole, different subsets may be defined, but since the members of a category can
usually be grouped together along different dimensions, these subsets are likely
to overlap, i.e., to form clusters of related meanings.
Second, linking the second to the third characteristic is the idea that members
of a category that are found in an area of overlapping between two senses carry
more structural weight than instances that are covered by only one meaning.
Representative members of a category (i.e., instances with a high degree of rep-
resentativity) are to be found in maximally overlapping areas of the extension of
a category. (In the example, A and E are less typical members that B, C, and D,
which each belong to two different subsets.)
Third, linking the third to the fourth characteristic is the idea that differences
in degree of membership may diminish to a point where it becomes unclear
whether something still belongs to the category or not. Categories have referentially
blurred edges because of the dubious categorial status of items with extremely
low membership degrees.
And fourth, linking the fourth to the first characteristic is the idea that the
flexibility that is inherent in the absence of clear boundaries prevents the formu-
lation of an essence that is common to all the members of the category. Because
peripheral members may not be identical with central cases but may only share
some characteristics with them, it is difficult to define a set of attributes that is
common to all members of a category and that is sufficient to distinguish that
category from all others.
These circular links between the four characteristics are, however, mislead-
ing. A closer look at some (familiar and less familiar) examples of prototypicality
reveals that they need not co-occur.

BIRD
The concept bird (one of Rosch’s original examples of prototypicality) shows
that natural categories may have clear-cut boundaries. At least with regard to
our own, real world, the denotation of bird is determinate; educated speakers of
150 Dirk Geeraerts

English know very well where birds end and non-birds begin. They know, for
instance, that a bat is not a bird but that a penguin is. Of course, the principled
indeterminacy described by Waismann (1952) as ‘open texture’ remains: when
confronted with an SF creature (a post-World War III mutant) that looks like a
bird but talks like a man, we would not be sure whether it should be called a bird
or not. A boundary problem that is typical for a prototypical organization of the
lexicon would then arise. As it functions now, however, in present-day English,
bird is denotationally clearly bounded, the archaeopteryx notwithstanding.14 As
has been remarked elsewhere (Lakoff 1987), the existence of prototypicality
effects in clearly bounded concepts such as bird implies that a strict distinction
has to be made between degree of membership and degree of representativity.
Membership in the category bird is discrete; something is or is not a bird. But
some birds may be birdier than others: the swallow does remain a more typical
bird than the ostrich.

RED
Color terms such as red constituted the starting-point for prototypical research;
drawing on the views developed in Berlin and Kay (1969), Rosch’s earliest work
is an experimental demonstration of the fact that the borderline between different
colors is fuzzy (there is no single line in the spectrum where red stops and orange
begins), and of the fact that each color term is psychologically represented by focal
colors (some hues are experienced as better reds than others) (Heider 1972; Heider
and Olivier 1972). These prototypical characteristics on the extensional level are
not matched on the definitional level. If red can be analytically defined at all (i.e.,
if it does not simply receive an ostensive definition consisting of an enumeration
of hues with their degree of focality), its definition might be ‘having a color that
is more like that of blood than like that of an unclouded sky, that of grass, that
of the sun, that of ... (etc., listing a typical exemplar for each of the other main
colors)’. Such a definition (cp. Wierzbicka 1985: 342) does not correspond with
either the first or the second characteristic mentioned above.

ODD¬NUMBER
Armstrong, Gleitman and Gleitman (1983) have shown experimentally that even
a mathematical concept such as odd number exhibits psychological representa-
tivity effects. This might seem remarkable, since odd number is a classical con-
cept in all other respects: it receives a clear definition, does not exhibit a family
resemblance structure or a radial set of clustered meanings, does not have blurred
edges. However, Lakoff (1982) has made clear that degrees of representativity
among odd numbers are not surprising if the experiential nature of concepts is
taken into account. For instance, because the even or uneven character of a large
 Chapter 4: Prototype theory 151

number can be determined easily by looking at the final number, it is no wonder

that uneven numbers below 10 carry more psychological weight: they are proce-
durally of primary importance.

VERS
As I have tried to show elsewhere (1988a), the first characteristic mentioned above
is not sufficient to distinguish prototypical from classical categories, since, within
the classical approach, the absence of a single definition characterized by neces-
sity-cum-sufficiency might simply be an indication of polysemy. This means that
it has to be shown on independent grounds that the allegedly prototypical concepts
are not polysemous, or rather, it means that prototypical lexical concepts will be
polysemous according to a definitional analysis in terms of necessary and suffi-
cient conditions (the classical definition of polysemy), but univocal according to
certain other criteria. These criteria may be found, for instance, in native speak-
ers’ intuitions about the lexical items involved, intuitions that may be revealed
by tests such as Quine’s (1960) or Zwicky and Sadock’s (1975). In this sense, the
first characteristic has to be restated: prototypical categories will exhibit intui-
tive univocality coupled with analytical (definitional) polysemy, and not just the
absence of a necessary-and-sufficient definition.

Once this revision of the first characteristic is accepted, it can be demonstrated

that the first and the second criterion need not co-occur. Lexical items that show
clustered overlapping of senses may either conform or not conform to the revised
first characteristic. An example of the first situation is the literal meaning of bird,
an example of the second situation the Dutch adjective vers, which corresponds
roughly with English fresh (except for the fact that the Dutch word does not
carry the meaning ‘cool’). Details of the comparison between both categories
may be found in the paper mentioned above; by way of summary, Figures 1 and
2 represent the definitional analysis of both items. The distinction in intuitive
status between vers and bird can be demonstrated by means of the Quinean test
(roughly, a lexical item is ambiguous if it can be simultaneously predicated and
negated of something in a particular context). Thus, taking an example based on
the corresponding ambiguity in the English counterpart of vers, it would be quite
normal to state that the news meant in the sentence there was no fresh news from
the fighting15 is fresh in one sense (‘recent, new’) but not in another (‘in optimal
condition’): it makes sense to say that the news is at the same time fresh and not
fresh. By contrast, it would be intuitively paradoxical to state that a penguin is
at the same time a bird and not a bird (disregarding figurative extensions of the
semantic range of bird). Nevertheless, the definitional analyses in Figures 1 and
2 make clear that both concepts exhibit prototypical clustering. In both cases, too,
the structural position of the instances just discussed (news, penguin) is not in the
152 Dirk Geeraerts

central area with maximal overlapping. In short, then, the revised version of the
first characteristic need not coincide with the second characteristic.

6 7

kiwi robin ostrich chicken

2
3
penguin

5
4

1 being able to fly 2 having feathers 3 being S-shaped

4 having wings 5 not domesticated 6 being born from eggs
7 having a beak or bill

Figure1.$GH¿QLWLRQDODQDO\VLVRIbird
Figure 1. A definitional analysis of bird

e.g. information

e.g. fruit and

other foodstuffs

e.g. air

1 new, novel, recent

2 in an optimal condition, pure, untainted

Figure 2.2.
Figure A Adefinitional
definitionalanalysis
analysisofofvers
vers
 Chapter 4: Prototype theory 153

The insight derived from a closer look at the four examples just described may be
summarized as in Table 2. It is now easy to see to what extent ‘prototypicality’
is itself a prototypical notion. There is no single set of attributes that is common
to all of the examples discussed here. Rather, they exhibit a family resemblance
structure based on partial similarities. In this sense, the set of prototypical concepts
characterized by clustering of senses overlaps with the subset characterized by
fuzzy boundaries (because of vers), and so on. At the same time, some concepts
are more typically prototypical than others. (Bird and vers are more prototypical
than red.) Notice, in particular, that the category fruit makes a good candidate for
prototypical prototypicality, in the sense that it seems to combine all four charac-
teristics. It shares the prototypical characteristics of bird, but in addition, things
such as coconuts and, perhaps, tomatoes, seem to point out that the denotational
boundary of fruit is less clear-cut than that of bird.
However, although the examples considered above do not have a set of attri-
butes in common, they do share a single feature, viz. degrees of membership
representativity. It is highly dubious, though, whether this feature alone suffices
to distinguish prototypical concepts from classical concepts. If the possibility of
a single necessary-and-sufficient definition is one of the features par excellence
with which the classical conception has been identified, it might justifiably be
claimed that degrees of representativity are entirely compatible with the classical
conception of categorization. It is, in fact, in that sense that Armstrong, Gleitman
and Gleitman (1983) deal with a category such as odd number. The experiments
used by Rosch to measure degrees of representativity are not, they claim, indica-
tive of prototypicality since they occur with classical, rigidly definable concepts
such as odd number. To say the least, representativity effects are only a peripheral
prototypical attribute according to Table 2 (cp. Lakoff 1986). But at the same time,
the debate over the status of odd number shows that the concept ‘prototypical con-
cept’ has no clear boundaries: it is not immediately clear whether a concept such
as odd number should be included in the set of prototypical concepts or not.

Table 2. The prototypicality of ‘prototypicality’

NUMBER
9(56
BIRD

ODD
RED

absence of classical definition + – – –

clusters of overlapping senses + + – –
degrees of representativity + + + +
absence of clear boundaries – + + –
154 Dirk Geeraerts

Of course, contrary to the situation in everyday speech, such a boundary conflict

should not be maintained in scientific speech. A discipline such as linguistics
should try to define its concepts as clearly as possible, and the purpose of this
section is precisely to show that what has intuitively been classified together as
instances of prototypical categories consists of distinct phenomena that have to
be kept theoretically apart. In line with prototype theory itself, however, such an
attempt at clear definition should not imply an attempt to define the ‘true nature’
or the ‘very essence’ of prototypicality. Determining an ‘only true kind’ of pro-
totypicality is infinitely less important than seeing what the phenomena are and
how they are related to each other by contrast or similarity.
Still, there might seem to be one way in which decent sense could be made of
the question what the true meaning of prototypicality would be. To begin with, let
us note that the prototypical character attributed to the concept of prototypicality
also shows up in the fact that the notion ‘prototype’ is an extremely flexible one.
This can be illustrated in two ways. First, the lexical item prototypical is spon-
taneously used to name a number of phenomena that are linked by metonymy,
next to the phenomena linked by similarity that are brought together in Table 2.
The lexical item does not only characterize structural features of concepts, and
the concepts exhibiting those features themselves, but sometimes even particu-
lar (viz., highly representative) instances of the categories in question (the robin
as a prototypical bird). Second, context may stress one feature of prototypical
organization rather than another (cp. the priming effects in Rosch 1975). The
general purpose of one’s investigations may lead one to devote more attention to
one aspect of the prototypical cluster than to another. To name a few examples:
degrees of representativity are important for language development studies (if
it is taken into account that most concepts in early language development are
acquired via their exemplars), while clustered overlapping of senses will come
to the fore in linguistic or lexicographical studies into the structure of polysemy.
And a cognitive interest into the epistemological principles underlying natural
language will attach more weight to the decoupling of intuitive univocality and
analytical, definitional polysemy.16
In this respect, the question with regard to the true nature of prototypicality
might be transformed into the question what might be the most interesting (or
perhaps even the most important) perspective for studying and defining prototypi-
cality. But here again, the ‘ultimate essence fallacy’ exposed by prototype theory
itself lurks round the corner: there will be different preferences for one perspec-
tive rather than another, but there will be no single ultimately and eternally most
important conception of prototypicality.
In short, the foregoing analysis corroborates Wierzbicka’s remark that there
are ‘many senses’ to the notion prototype, and that ‘the notion prototype has
been used in recent literature as a catch-all notion’ (1985: 343). However, a more
 Chapter 4: Prototype theory 155

systematic analysis than Wierzbicka’s reveals that this very multiplicity of usage
also supports Cognitive Semantics, in the sense that it shows that the same cat-
egorization principles may guide common sense and scientific thinking. This is,
then, a further indication of the metatheoretical relevance of a cognitive conception
of linguistic categorization, which I have explored at length elsewhere (1985b).
At the same time, it has become clear that one of the major tasks for the further
development of prototype theory is the closer investigation of the prototypically
clustered characteristics of prototypicality. A major reference in this respect is
Lakoff’s attempt (1987: chapter 4–8) to determine which different kinds of con-
ceptual models may lie at the basis of prototypicality effects.

3. Definitional problems, second series: ‘Prototype theory’ as a

prototypical notion

Whereas the previous section made clear that prototypicality as used in linguistic
semantics is a prototypically structured concept, it should now be noted that the
prototype-theoretical movement as well is a prototypically structured approach
to semantics. There are, in other words, central as well as more peripheral exam-
ples of prototypical theories. In particular, there exist a number of theories that
combine aspects of the classical approach to semantic structure with aspects of
the prototypical conception. In this section, two approaches will be considered
that are to some extent semi-classical as well as semi-prototypical; each of both
embodies a strategy for reinstating particular aspects of the classical view against
the background of an overall cognitive point of view.
To begin with, some of the clarity and neatness of the classical approach may
be recovered by concentrating on the prototypical centre of a category. If the
non-classical indeterminacy of lexical concepts stems primarily from the flex-
ible extendibility of concepts, discreteness may be reinstalled by avoiding the
problems of clustered polysemy, i.e., by restricting the definitional analysis to the
prototypical centre of the category. This approach is vigorously carried through
by Wierzbicka (1985), who explicitly defends the discreteness of semantics by
introspectively considering only the clear, salient centre of lexical categories. In
a discussion of Labov’s experimental investigation into the non-classical charac-
teristics of everyday concepts (1973), she notes:17

To state the meaning of a word, it is not sufficient to study its applicability to

things; what one must do above all is to study the structure of the concept which
underlies and explains that applicability. In the case of words describing natural
kinds or kinds of human artefacts, to understand the structure of the concept means
to describe fully and accurately the idea (not just the visual image) of a typical
representative of the kind: the prototype. And to describe it fully and accurately
156 Dirk Geeraerts

we have to discover the internal logic of the concept. This is best done not through
interviews, not through laboratory experiments, and not through reports of casual,
superficial impressions or intuitions ... but through methodical introspection and
thinking (1985: 19).

It should be noted immediately that Wierzbicka’s reinstatement of discreteness

does not imply that her definitions do in fact always consist of necessary-and-
sufficient conditions, and she acknowledges as much (1985: 60). In this respect,
Wierzbicka’s approach is only partly a departure from the hard core of proto-
type-theoretical studies: the absence of necessary-and-sufficient conditions for
the definition of certain core concepts is accepted, but the avoidance of the clus-
tered polysemy problem ‘tidies up’ the semantic description and reinstates some
of the classical neatness. Neither does Wierzbicka’s approach imply that lexical
items are always univocal; in her dictionary of English speech act verbs (1987),
several items receive multiple definitions. Each of the definitions does, however,
constitute a highly salient meaning, and again, by disregarding peripheral kinds
of usage, the clustered or radial structure of the polysemy of lexical items does
not enter the picture. The question to be asked, then, is whether Wierzbicka’s
restriction of the description to the salient meanings of a category is useful and
adequate from a cognitive point of view.
From a methodological point of view, the periphery of natural, non-uniquely
definable categories is as interesting as their salient centre(s), because it is pre-
cisely the relationship between both that typically characterizes natural categories.
Cognitive Linguistics is not only interested in what constitutes the centre of a cat-
egory, but also in how this centre can be extended towards peripheral cases, and
how far this extension can go. The mechanisms for incorporating marginal cases
into a category at the same time restrict the flexibility of that concept; it is only by
studying peripheral cases, for instance, that an answer may be found with regard
to the question how dissimilar things can be before they are no longer recognized
as basically the same. If, in other words, flexible polysemization is indeed one of
the major characteristics of natural language categories, a deliberate restriction
of the description to the salient meanings of a category is methodologically less
propitious, as it may lead to a neglect of this basic feature.18
A second strategy for salvaging aspects of the classical approach is to invoke
sociolinguistic mechanisms such as Putnam’s ‘division of linguistic labor’ (1975).
According to Putnam, ordinary language users possess no more than ‘stereotypi-
cal’ knowledge about natural kinds, that is to say, they are aware of a number of
salient characteristics, such as the fact that water is a transparent, thirst-quench-
ing, tasteless liquid. The technical definition of water as H2O, on the other hand,
is to be located primarily with scientific experts. It is the experts’ knowledge that
ultimately determines how natural kind terms are to be used. On the one hand,
 Chapter 4: Prototype theory 157

a ‘division of linguistic labor’ ensures that there are societal experts who know
that water is H2O, that there is a distinction between elms and beech, how to rec-
ognize gold from pyrites, and so on. On the other hand, laymen attune their own
linguistic usage to that of the expert scientists, technicians, etc.. The members
of the non-specialized group are not required to have expert knowledge, but if
they wish to be considered full-fledged members of the linguistic community,
they are supposed to know the ‘stereotype’ connected with a category. A ste-
reotype is, thus, a socially determined minimum set of data with regard to the
extension of a category. Given the similarity between Putnam’s stereotypes and
the prototypes of Cognitive Linguistics (both consist roughly of the most salient
information connected with a category), the division of linguistic labor might be
used to rescue the classical view of concepts.19 Expert definitions being classical
(they specify an essentialist ‘hidden structure’ for natural kinds), the stereotypi-
cal concepts of everyday language users might now be seen as hardly more than
a sloppy derivative of those classically defined expert categories. ‘True’ (expert)
definitions would be classical, and stereotypical/prototypical concepts might be
dismissed as sociolinguistically secondary phenomena.
It should be remarked immediately that such a reinstatement of the classi-
cal view is not as obvious for other words than the natural kind terms for which
Putnam’s theory is in fact intended (what is the expert definition of the prepo-
sition for?). Moreover, as a sociolinguistic theory about the social factors that
determine how lexical items may be used, the ‘division of linguistic labor’ theory
is incomplete to say the least. The primacy of expert definitions would seem to
imply that natural language follows the developments and discoveries of science
in a strict fashion. In actual fact, however, natural language categorization is not
only determined by the state of affairs in the sciences, but also by the communi-
cative and cognitive requirements of the linguistic community in its own right.
One of Putnam’s own examples may serve as an illustration. Although science
has discovered that jade refers to two kinds of materials (one with the ‘hidden
structure’ of a silicate of calcium and magnesium, the other being a silicate of
sodium and aluminium), ordinary usage continues to refer to both substances
indiscriminately as jade. That is to say, categorization in everyday language is
not entirely dependent upon scientific research, but seems to be determined at
least in part by independent criteria: if the classificatory exigencies of everyday
communicative interaction do not call for a distinction between the two kinds of
jade, the scientific splitting of the category is largely ignored. This implies that an
investigation into everyday language categorization as an independent cognitive
system is justified. More generally, if Putnam’s view is seen as a theory about the
sociolinguistic structure of semantic norms, his hierarchical model (with experts
at one end and laymen at the other) is only one among a number of alternatives,
some of which (such as the one described by Bartsch 1985) link up closely with
158 Dirk Geeraerts

a prototypical conception of categorial structure. At the same time, however, it

should be admitted that the relationship between classical scientific categorization
and prototypical common-sense categorization may be explored in more depth
than is yet the case.20
To summarize: the confusion associated with the notion of prototypicality is
further increased by the fact that more straightforwardly prototypical approaches
are surrounded by hybrid theories that contain particular strategies for combining
classical discreteness with typically prototypical phenomena. We have discussed
two such approaches (one in which the strategy in question is methodological, and
another one in which it is sociolinguistic), but this does not mean that these are
the only ones that might be mentioned.21 The two approaches mentioned here are,
however, particularly revealing, as they link up with two important currents in the
history of Western thought. The first one simplifyingly boils down to the view
that the mind is neat (if you look hard enough into it), but that the world is fuzzy:
the non-discreteness that Cognitive Linguistics is concerned with arises from the
fact that we have to apply clear-cut mental categories to an external reality that is
so to say less well organized. The conception that the world of mental entities is
somehow better organized than the outside world is obviously an idealistic one
(though it does not constitute the only possible kind of idealism); Wierzbicka her-
self stresses the Platonist character of her approach. On the other hand, Putnam’s
view that science is neat whereas everyday language is fuzzy, links up with the
empiricist objectivism of the Ideal Language branch of analytical philosophy:
the objective structure of reality is best described by the language of science,
and everyday language is at best a weak derivative of scientific categorization, at
worst a conceptual muddle teeming with philosophical pseudo-problems. As the
previous discussion suggests that hard-core Cognitive Linguistics steers clear of
both the idealist and the objectivist option, we have here one more indication22
for the necessity of a further investigation into the epistemological, philosophical
background of the prototypical conception of categorial structure.

Notes
1. The discussion in section 2 will make clear that the term prototype theory should
be used with care, since the theoretical uniformity that it suggests tends to obliter-
ate the actual distinctions between the diverse forms of prototypicality discussed in
the literature. The term is used here as a convenient reference mark only, to indicate
a number of related theoretical conceptions of categorial structure that share an
insistence on any or more of the various kinds of prototypicality effects discussed
in section 2.
2. Though not exclusively: see Rosch (1988: 386).
3. Notice that this claim applies just as well to the axiomatic, postulate-based form of
 Chapter 4: Prototype theory 159

description that developed as the major representational alternative for Katzian com-
ponential analysis. The notion of criteriality is just as much part and parcel of the
classical versions of the axiomatic alternative as it is of Katzian feature analysis.
4. See, among others, Haiman (1980) and Geeraerts (1985b).
5. The distinction between semantic and encyclopedic concepts against which Cognitive
Semantics reacts is often misconstrued. In particular, in the statement that there is
no principled distinction between semantic and encyclopedic information, the words
semantic and encyclopedic are not used (as implied by Lehmann 1988b) in the senses
‘as may be found in dictionaries’ and ‘as may be found in encyclopedias’, respectively.
Rather, the rejected distinction refers to an alleged distinction within an individual
language user’s conceptual memory; it involves the presupposition that there is an
independent level of semantic information that belongs to the language and that is
distinct from the individual’s world knowledge. The kind of information that is typi-
cally found in encyclopedias involves scientific information of the kind ‘ovulation
triggered by copulation’ for the item cat (the example is Lehmann’s); but while the
distinction between scientific and laymen’s knowledge is primarily a social one, this
kind of ‘encyclopedic’ information is only relevant for the psychological perspective
of Cognitive Semantics if the individual lexicon to be described is that of someone
with a certain amount of scientific knowledge of cats (or if, through sociolinguistic
idealization, the average language user’s lexicon may be supposed to contain that
piece of scientific information).
6. There are, of course, exceptions such as Dowty (1979) to confirm the rule. The his-
torical sketch of the advent of prototype theory given here is treated more thoroughly
in Geeraerts (1988b).
7. As the semantic interests of the former audience of Generative Semantics were so
to say no longer envisaged by the leading theories of the day, it does not come as
a total surprise, from this point of view, to find George Lakoff, one of the leading
Generative Semanticists, again as one of the leading cognitivists.
8. These antecedents are not the only ones that might be mentioned. I have elsewhere
(1988c) drawn the attention to the similarities between the prestructuralist, histori-
cal tradition of semantic research and present-day Cognitive Semantics, but there
are other (admittedly non-mainstream) traditions of semantic research with which
Cognitive Semantics is methodologically related: think, e.g., of the anthropological
research of Malinowski, Firth, and the London School in general. Even a structur-
alist such as Reichling has held views about the structure of polysemy that come
close to the point of view of prototype theory: his influential work on the word as
the fundamental unit of linguistics (1935) contains an analysis of the Dutch word
spel that is awkwardly similar to Wittgenstein’s remarks about the German equiva-
lent Spiel. The point to be stressed is this: as a theory about the (radial, clustered,
dynamically flexible) structure of polysemy, prototype theory is to a considerable
extent a rediscovery of views that were paramount during the prestructuralist era of
the development of lexical semantics, and that lingered on below the surface in the
structuralist and transformationalist periods.
9. Because of their large scope, the functionalist approach of Seiler (1986) and the
naturalist approach of Dressler (1985) are particularly interesting for the use of pro-
totypicality with regard to various aspects of the formal organization of language.
160 Dirk Geeraerts

10. A bibliography of work in Cognitive Linguistics is to be found in Dirven (1988). It

is worth mentioning that Cognitive Linguistics is currently in a stage of organiza-
tion: a first international conference of Cognitive Linguistics was held in Duisburg in
March 1989, and a new journal entitled Cognitive Linguistics, published by Mouton,
is scheduled to start appearing in the beginning of 1990.
11. Next to the link with psycholinguistics, there is a connection with Artificial Intel-
ligence research, through the correspondences between the notion of prototypicality
and that of frame; see Fillmore (1977). It needs to be stressed, though, that the link
is relatively weak; specifically, the correspondence just mentioned is to a certain
extent counterbalanced by Lakoff’s criticism (1987) of the objectivist assumptions
of mainstream Artificial Intelligence research (but then again, one of Lakoff’s cur-
rent research projects involves a connectionist approach to the formal modeling
of Cognitive Semantic notions such as metaphorical image schemata). In general,
sorting out the relationship between Cognitive Semantics and Artificial Intelligence-
oriented Cognitive Science will be one of the major tasks for the further development
of Cognitive Semantics.
12. See Lakoff (1987: chapter 6) for the notion of a radial set, and compare Givon (1986)
for a comparison between the views of Wittgenstein and those of prototype theory.
The stress Givon places on the distinctions between both is slightly exaggerated, as
it tends to obscure their mutual rejection of the so-called classical theory. See also
the next footnote.
13. The ‘so-called’ is added to stress, first, that the views of Aristotle also contain fea-
tures that correspond rather with a cognitive than with a ‘classical’ approach, and
second (more generally), that the philosophical position of prototype theory is in need
of further elucidation. The present situation is rather muddled: while the classical
Roschian position is to characterize prototype theory as non-Aristotelian and Witt-
gensteinian, Givon (1986) has argued that prototype theory is non-Wittgensteinian
(see the previous note), but whereas Givon also describes prototype theory as non-
Platonic, Wierzbicka (to whom we shall come back in section 3) precisely presents
an explicitly Platonic version of prototype theory. More generally, the philosophical
position of prototype theory has so far been discussed mainly against the background
of classical philosophy (Aristotle and Plato), and against the background of contem-
porary analytical philosophy (see Lakoff 1987). This means that a large part of the
history of Western philosophy passes unmentioned; this is to be regretted, as the
post-Cartesian period in the history of philosophy is concerned with epistemologi-
cal questions that are of immediate interest to Cognitive Semantics. In particular,
if it can be accepted that one of the major epistemological aspects of a prototypical
conception of categorial structure resides in the fact that categories are interpretive
schemata that are used flexibly and dynamically in our encounters with reality, a major
philosophical reference point for prototype theory will lie with those philosophical
theories that recognize the constitutive role of existing knowledge with regard to new
experiences. As I have argued elsewhere (1985b), the Husserlian phenomenological
movement (as represented, specifically, by Maurice Merleau-Ponty) provides a good
starting-point for a further confrontation with philosophy.
14. The archaeopteryx is probably regarded as a species separate from either bird or
reptile.
 Chapter 4: Prototype theory 161

15. The example is taken from the Longman Dictionary of Contemporary English.
16. Considered from this point of view, Lakoff’s radial sets as such are not particularly
unclassical: structured polysemy as such is entirely compatible with the classical
view. Kleiber (1988) offers an insightful discussion of the theoretical consequences
of the growing importance of the structure of polysemy in prototype-theoretical
research.
17. For a more extended discussion of Wierzbicka’s views, see Geeraerts (1988d).
18. Notice that the restriction to the prototypical centre of categories correlates with
Wierzbicka’s Platonic, introspective methodology: it seems probable that the appli-
cations of a category that can be accessed introspectively are only the more salient
ones; peripheral cases probably do not always pass the threshold of conscious atten-
tion. What is interesting from a cognitive point of view, however, is the way people
spontaneously categorize and classify things, not the way in which they introspectively
reflect upon their own conceptualizations. Any attempt to describe the peripheral
instances of a category together with its prototypical centre can therefore not be
restricted to an introspective methodology.
19. This is not say that Putnam actually intended his stereotypical theory as such an
attempted rescue: his problems lay with the notion of reference rather than with those
of polysemy and categorial structure. My remarks about Putnam are an investigation
into some of the possible consequences of the notion of division of linguistic labor,
not an attempt to give an account of Putnam’s view in its original setting. Further, it
has to be mentioned that some of Putnam’s later philosophical views open up entirely
different perspectives for a confrontation with Cognitive Semantics; in particular,
see Lakoff (1987) on Putnam and anti-objectivism.
20. An interesting contribution to such an exploration is found in Lakoff (1987: chapter
12), where it is claimed that scientific categories are far from being as classical as
is usually assumed.
21. Again, see Lakoff (1987: chapter 9) for some more examples; they are situated within
formal psycholexicology rather than within linguistics.
22. Next, that is, to the remarks made in footnote 13.

References
Aijmer, Karen
1985 The semantic development of will. In Historical Semantics – Historical Word-
Formation, Jacek Fisiak (ed.), 11–21. Berlin/New York: Mouton de Gruyter.
Armstrong, Sharon L., Lila R. Gleitman, and Henry Gleitman
1983 What some concepts might not be. Cognition 13: 263–308.
Bartsch, Renate
1985 Sprachnormen: Theorie und Praxis. Tübingen: Niemeyer Verlag.
Berlin, Brent and Paul Kay
1969 Basic Color Terms. Berkeley: University of California Press.
Bybee, Joan and Carol L. Moder
1983 Morphological classes as natural categories. Language 59: 251–270.
162 Dirk Geeraerts

Craig, Colette (ed.)

1986 Noun Classes and Categorization. Amsterdam/Philadelphia: John
Benjamins.
Dirven, René
1988 Bibliography of Cognitive Linguistics. Duisburg: Linguistic Agency of the
University of Duisburg.
Dowty, David
1979 Word Meaning and Montague Grammar. Dordrecht: Reidel.
Dressler, Wolfgang
1985 Morphophonology. Ann Arbor: Karoma.
Fillmore, Charles J.
1975 An alternative to checklist theories of meaning. In Proceedings of the First
Annual Meeting of the Berkeley Linguistics Society, Cathy Cogen, Henry
Thompson, and James Wright (eds.), 123–131. Berkeley: Berkeley Linguistics
Society.
1977 Scenes–and–frames semantics. In Linguistic Structures Processing, Antonio
Zampolli (ed.), 55–81. Amsterdam: North Holland Publishing Company.
Geeraerts, Dirk
1983 Prototype theory and diachronic semantics: A case study. Indogermanische
Forschungen 88: 1–32.
1985a Paradigm and Paradox. Explorations into a Paradigmatic Theory of Meaning
and its Epistemological Background. Leuven: Universitaire Pers.
1985b Les données stéréotypiques, prototypiques et encyclopédiques dans le diction-
naire. Cahiers de Lexicologie 46: 27–43.
1985c Cognitive restrictions on the structure of semantic change. In Historical Seman-
tics - Historical Word Formation, Jacek Fisiak (ed.), 127–153. Berlin/New
York: Mouton de Gruyter.
1988a On necessary and sufficient conditions. Journal of Semantics 5: 275–291.
1988b Katz revisited: aspects of the history of lexical semantics. In Understanding
the Lexicon, Werner Hüllen and Rainer Schulze (eds.), 23–35. Tübingen: Max
Niemeyer Verlag.
1988c Cognitive Grammar and the history of lexical semantics. In Topics in Cognitive
Linguistics, Brygida Rudzka–Ostyn (ed.), 647–677. Amsterdam/Philadelphia:
John Benjamins.
1988d Review of Anna Wierzbicka, Lexicography and Conceptual Analysis. Lan-
guage in Society 17: 449–455.
Givon, Talmy
1986 Prototypes: between Plato and Wittgenstein. In Noun Classes and Cat-
egorization, Colette Craig (ed.), 78–102. Amsterdam/Philadelphia: John
Benjamins.
John Haiman
1980 Dictionaries and encylopedias. Lingua 50: 329–357.
Heider, Eleanor R.
1972 Universals in color naming and memory. Journal of Experimental Psychology
93: 10–20.
 Chapter 4: Prototype theory 163

Heider, Eleanor R. and Olivier, D.C.

1972 The structure of color space in naming and memory for two languages. Cog-
nitive Psychology 3: 337–354.
Holland, Dorothy and Naomi Quinn (eds.)
1987 Cultural Models in Language and Thought. Cambridge: Cambridge University
Press.
Hüllen, Werner and Rainer Schulze (eds.)
1988 Understanding the Lexicon. Meaning, Sense, and World Knowledge in Lexi-
cal Semantics. Tübingen: Niemeyer.
Katz, Jerrold J. and Jerry A. Fodor
1963 The structure of a semantic theory. Language 39, 170–210.
Kleiber, Georges
1988 Prototype, stéréotype: un air de famille? DRLAV. Revue de Linguistique 38:
1–61.
Labov, William
1973 The boundaries of words and their meanings. In New Ways of Analysing Varia-
tion in English, Charles J. Bailey and Roger Shuy (eds.), 340–373. Washington:
Georgetown University Press.
Lakoff, George
1982 Categories and cognitive models. Trier: Linguistic Agency of the University
of Trier. Also as: Berkeley Cognitive Science Report no. 2. Berkeley: Institute
for Human Learning.
1986 Classifiers as a reflection of mind. In Noun Classes and Categorization, Colette
Craig (ed.), 13–51. Amsterdam/Philadelphia: John Benjamins.
1987 Women, Fire, and Dangerous Things. What Categories Reveal about the
Mind. Chicago: The University of Chicago Press.
Langacker Ronald W.
1987 Foundations of Cognitive Grammar I. Stanford: Stanford University Press.
Lehmann, Winfred P. (ed.)
1988a Prototypes in Language and Cognition. Ann Arbor: Karoma.
Lehmann Winfred P.
1988b Review of R. Langacker, Foundations of Cognitive Grammar. General Lin-
guistics 28: 122–30.
Medin, Doug L. and Edward E. Smith
1984 Concepts and concept formation. Annual Review of Psychology 35:
113–138.
Mervis, Carolyn B. and Eleanor Rosch
1981 Categorization of natural objects. Annual Review of Psychology 32: 89–115.
Nathan, Geoffrey S.
1986 Phonemes as mental categories. In Proceedings of the Twelfth Annual Meet-
ing of the Berkeley Linguistics Society, Vassiliki Nikiforidou, Mary VanClay,
Mary Niepokuj, Deborah Feder (eds.), 212–224. Berkeley: Berkeley Linguis-
tics Society.
Neisser, Ulrich
1987 Concepts and Conceptual Development. Ecological and Intellectual Factors
in Categorization. Cambridge: Cambridge University Press.
164 Dirk Geeraerts

Posner, Michael
1986 Empirical studies of prototypes. Noun classes and categorization, Colette
Craig (ed.), 53–61. Amsterdam/Philadelphia: John Benjamins.
Post, Michael
1986 A prototype approach to denominal adjectives. In Linguistics across Historical
and Geographical Boundaries, Dieter Kastovsky and Aleksander Szwedek
(eds.), II: 1003–1013. Berlin/New York: Mouton De Gruyter.
Putnam, Hilary
1975 The meaning of Meaning. In Mind, Language, and Reality. Philosophi-
cal Papers II, Hilary Putnam, 215–271. Cambridge: Cambridge University
Press.
Quine, Willard V.O.
1960 Word and Object. Cambridge, Mass.: MIT Press.
Reichling, Anton
1935 Het Woord. Zwolle: Tjeenk Willink.
Rosch, Eleanor
1973 On the internal structure of perceptual and semantic categories. In Cogni-
tive Development and the Acquisition of Language, Timothy E. Moore (ed.),
111–144. New York: Academic Press.
1975 Cognitive representations of semantic categories. Journal of Experimental
Psychology: General 104: 192–233.
1977 Human categorization. In Studies in Cross–cultural Psychology I, Warren,
Neil (ed.), 1–49. New York: Academic Press.
1978 Principles of categorization. In Cognition and Categorization, Eleanor Rosch
and Barbara B. Lloyd (eds.), 27–48. Hillsdale, NJ: Lawrence Erlbaum.
1988 Coherences and categorization: A historical view. In The Development of Lan-
guage and Language Researchers. Essays in Honor of Roger Brown, Frank
Kessel (ed.), 373–392. Hillsdale, NJ: Lawrence Erlbaum.
Rosch, Eleanor and Carolyn B. Mervis
1975 Family resemblances: Studies in the internal structure of categories. Cognitive
Psychology 7: 573–605.
Rosch, Eleanor, Carolyn B. Mervis, Wayne D. Gray, David Johnson, and Penny
Boyes–Braem
1976 Basic objects in natural categories. Cognitive Psychology 8: 382–439.
Rosch, Eleanor, Carol Simpson, and Scott R. Miller
1976 Structural bases of typicality effects. Journal of Experimental Psychology:
Human Perception and Peformance 2: 491–502.
Ross, John R.
1987 Islands and syntactic prototypes. In Papers from the 23rd Annual Meeting of
the Chicago Linguistic Society, Barbara Need, Eric Schiller, and Anna Bosch
(eds.), 309–320. Chicago: Chicago Linguistic Society.
Rudzka–Ostyn, Brygida
1988 Semantic extensions into the domain of verbal communication. In Topics in
Cognitive Linguistics, Brygida Rudzka–Ostyn (ed.), 507–553. Amsterdam/
Philadelphia: John Benjamins.
 Chapter 4: Prototype theory 165

Smith, Edward E. and Doug Medin

1981 Categories and Concepts. Cambridge, MA: Harvard University Press.
Taylor, John
1989 Linguistic Categorization. Prototypes in Linguistic Theory. Oxford: Clarendon
Press.
Tsohatzidis, Savas L. (ed.)
1989 Meanings and Prototypes. Studies on Linguistic Categorization. London:
Routledge and Kegan Paul.
Uhlenbeck, Eugenius M.
1967 The dynamic nature of word meaning. Actes du Xe Congrès International des
Linguistes II, 679–685. Bucarest: Editions de l’Academie.
Van Langendonck, Willy
1986 Markedness, prototypes, and language acquisition. Cahiers de l’Institut de
Linguistique de Louvain 12: 39–76.
Van Oosten, Jeanne
1986 The nature of subjects, topics, and agents: a cognitive explanation. Blooming-
ton: Indiana University Linguistics Club.
Waismann, Friedrich
1952 Verifiability. In Logic and Language (First Series), Anthony Flew (ed.), 117–
144. Oxford: Basil Blackwell.
Weinreich, Uriel
1966 Explorations in semantic theory. Current Trends in Linguistics III: 395–479.
Wierzbicka, Anna
1985 Lexicography and Conceptual Analysis. Ann Arbor: Karoma.
1987 English Speech Act Verbs: A Semantic Dictionary. Sydney: Academic
Press.
Winters, Margaret
1987 Syntactic and semantic space: the development of the French subjunctive. In
Papers from the Seventh International Conference on Historical Linguis-
tics, Anna G. Ramat, Onofrio Carruba, and Giulano Bernini (eds.), 607–618.
Amsterdam/Philadelphia: John Benjamins.
Wittgenstein, Ludwig
1953 Philosophische Untersuchungen. Philosophical Investigations. Oxford: Basil
Blackwell.
Zwicky, Arnold and Jerry Sadock
1975. Ambiguity tests and how to fail them. In Syntax and Semantics 4, John Kim-
ball (ed.), 1–36. New York: Academic Press.
Chapter 5
Schematic network

Ambiguity, polysemy, and vagueness

David Tuggy

1. Introduction: Tests for ambiguity vs. vagueness

The difference between ambiguity and vagueness is a matter of whether two or

more meanings associated with a given phonological form are distinct (ambigu-
ous), or united as non-distinguished subcases of a single, more general meaning
(vague). A standard example of ambiguity is bank “financial institution” vs. bank
“land at river edge”, where the meanings are intuitively quite separate; in aunt
“father’s sister” vs. aunt “mother’s sister”, however, the meanings are intuitively
united into one, “parent’s sister”. Thus ambiguity corresponds to separation, and
vagueness to unity, of different meanings.
Several kinds of tests have been proposed to render this difference measur-
able. The “logical” test of Quine (1960) will not centrally concern us here; it
involves seeing if a sentence “X and not X” can be true; the (supposed) accept-
ability of This is the bank [river edge] but it is not the bank [financial institution]
indicates that bank is ambiguous; the unacceptability of I have an aunt [father’s
sister] but I do not have an aunt [mother’s sister] indicates vagueness for aunt.
There is a “definitional” test (since Aristotle) which involves seeking a meaning
common to the two; if such is found, then it is taken as the correct definition,
and it is vague as to the differences between the original two meanings. Thus
since there is (supposedly) no meaning common to the two meanings of bank it
is ambiguous; since for aunt there is a common meaning (parent’s sister), it takes
precedence over the two more specific meanings and the word is ambiguous. The
“linguistic” tests (Lakoff 1970 and others) will concern us more; they involve
grammatical constructions which are taken as requiring semantic identity, such
as “X does/did Z and so does/did Y”. If Z as done by X and Z as done by Y can
be given “crossed” readings without the semantic oddness known as zeugma,
the meaning of Z is taken to be vague; if zeugma results, Z is ambiguous. Thus
the ordinariness of I have an aunt [father’s sister] and so does Bill [have an aunt
(mother’s sister)] indicates that the meanings are taken to be the same, and thus
aunt is vague; the unacceptability of the zeugmatic I went to the bank [financial

Originally published in 1993 in Cognitive Linguistics 4(3): 273–290.

168 David Tuggy

institution] and so did Bill [go to the bank (river edge)], which could only be
said facetiously, indicates that bank is ambiguous. A fourth distinction which is
sometimes mentioned (e.g. Kempson 1977: 182) is that puns can be constructed
with ambiguous but not vague structures; thus a pirate burying his gold at the
edge of the river could be said to be putting his money in the bank, which is an
(admittedly bad) pun; such a pun could not be readily made, if at all, on the two
meanings of aunt.
Geeraerts (1993) argues that (a) the three tests do not always agree with each
other, and (b) by changing the context, all three tests can be made to yield inde-
terminate results. He is particularly concerned to demonstrate that even for the
standard paradigm examples of ambiguity, if you try hard enough you can get
the tests to yield equivocal results. He concludes that the conception of ambiguity
and vagueness as classical categories with fixed boundaries and no gradations of
membership is, at the least, strongly called into question. The possibility that they
are categories formed on a prototype pattern thus merits investigation.

2. The fuzzy borderline between ambiguity and vagueness

2.1. Polysemy
Geeraerts’ conclusions are supported solidly by a different sort of data, not the
paradigm cases but rather borderline cases where there is clearly a meaning com-
mon to the sub-meanings in question (and thus by the definitional test there is only
one meaning), but nevertheless there are strong enough differences in meaning
to produce equivocal results or judgments of ambiguity from the linguistic tests.1
These cases are traditionally considered under the rubric of polysemy:2 traditional
definitions (e.g. Lyons 1977: 550; Zwicky and Sadock 1975: 2) have lexical ambi-
guity (or homonymy) as involving two lexemes, polysemy a single lexeme with
different distinct senses, and vagueness a lexeme with a single but nonspecific
meaning, i.e., the lexeme may subsume other meanings but those meanings are
not distinguished from each other or from the more inclusive “true” meaning.
Thus polysemy is a sort of halfway point between ambiguity and vagueness. As
Deane (1988: 327, 345) puts it, “In effect, the three types form a gradient between
total semantic identity and total semantic distinctness”, and “Polysemy seems
somehow to straddle the border between identity and distinctness.”
Any conscientious lexicographer will admit to at least occasional difficulty
in deciding whether two clearly related meanings associated with the same pho-
nological form are to be listed as a single entry in the dictionary or not (e.g.
Hartmann’s (1983: 6–7) discussion of senses of bank); most make frequent use
of sub-entries and sub-sub-entries to ease the difficulty. These cases constitute
 Chapter 5: Schematic network 169

massive evidence, from any and every language, that the categories of ambiguity
and vagueness are not absolute.

2.1.1. The verb paint

For an English example, consider the meaning(s) of the verb paint. A structure
such as she is painting (it) can conventionally be used to refer to an indefinitely
large number of situations, including the following (and many others): (i) painting
a portrait in oils on canvas, (ii) painting a landscape with watercolors on paper,
(iii) painting trompe l’oeil on the interior wall and floor of a house, (iv) painting
a mural on the exterior wall of a public building, (v) painting a decorative border
on an interior wall, (vi) painting the walls of a room with a single color of paint
for decorative purposes but also to preserve them, (vii) painting the exterior of a
house primarily to preserve it, (viii) painting furniture, (ix) painting a car with
an air gun, (x) painting stripes on a parking lot or roadway by driving a paint-
spraying machine, (xi) applying makeup to the face, or (xii) applying iodine or
some other colored disinfectant to the body after or prior to an incision, with a
swabbing motion. The concepts coded by paint in these different situations clearly
have something important in common: they all involve applying some colored
substance to a surface. This, by the Aristotelian definitional test, is the meaning of
the word, which is therefore presumably vague. However, the different situations
vary along a number of parameters such as the nature of the substance applied
(usually it is a liquid, properly called paint [N], which dries into a thin tough
layer on the surface, but not in the cases of (xi–xii)), the nature of the surface and
the extent to which it is covered as a whole, whether one color or a combination
of colors is applied, the instrument used for application (usually a brush or less
typically a roller, but not in (ix–xii)), the kind of physical activity involved in the
application, the purpose of the application (with protection and visual esthetic
enhancement being the two major types), the kinds of abilities and training needed
to do the application properly, etc. Many cases are obviously closer to each other
than others along these parameters; e.g. (i) and (ii) are much closer than (i) and
(x) or (i) and (xii).

2.1.2. Ambivalent results on the linguistic tests

The linguistic tests yield equivocal results in cases of this sort. If I have been
painting a watercolor landscape (ii) and Jane a portrait in oils (i), a sentence like
(1) I have been painting and so has Jane is perfectly normal, indicating vague-
ness rather than ambiguity for paint. If I have been painting stripes on the road
(x), however, while Jane painted a portrait, (1) feels zeugmatic: I do not believe I
could utter it except facetiously. This indicates that paint is ambiguous. Combi-
nations of (i) and (xi) or (i) and (xii) would be even more anomalous and clearly
170 David Tuggy

zeugmatic. If I was painting the exterior of a house (vii) and Jane a portrait, the
sentence would still strike me as odd, but not nearly as odd as clear cases of
zeugma do. I would not judge that this oddness amounts to a clear indication of
either ambiguity or vagueness, rather it might be taken to indicate something
in between.3 If I were painting a mural on a wall (iv), it would be better; if she
were painting a mural and I doing the more normal wall painting of (vi) or (vii)
I think it would be almost, but still not entirely, normal; if she were painting a
decorative border on a wall (v) and I a flat coat of paint (vi), the sentence would
be quite normal. The acceptability of sentence (1) in the different cases is not a
discrete, binary property, and it seems to correlate at least roughly to the close-
ness of the kinds of painting along the parameters mentioned above.

2.1.3. The influence of context

Equally importantly, the linguistic context can affect the acceptability of such test
sentences. For instance, when the stem paint is used with the nominalizing suffix
-er, the distinctions of purpose of painting and skills and training needed to paint
well become more important. Thus, although sentence (1) is almost normal in a
case of (iv) and (vi) or (vii) (mural vs. normal wall painting), it sounds facetious
to say, if Jane and I do those activities professionally, (2) I am a painter and so
is Jane. If I paint stripes on a road for a living and she paints oil portraits (i and
x), it sounds downright ludicrous. On the other hand, I could say, non-facetiously,
(3) When I’m painting I try to get the color on evenly, and so does Jane, with
virtually any combination of the kinds of painting involved.
If paint is used transitively further differences emerge. In cases (i–ii) and usu-
ally (iii–v) the direct object designates not the surface to which paint is applied,
but either the artistic figure created in paint on that surface, or something of which
that artistic figure is a representation. A similar construal is also normal for (x),
though roadway stripes are not precisely “artistic figures”, and the direct object
designates only the stripes themselves, not anything which they might represent.
Marginally in (iii–iv) and (x), more easily in (v) and (xii), and normally in (vi–ix)
and (xi) the object is the surface. Example (xii), however, would normally take the
disinfectant as object. This means that for some combinations (e.g. (v) and (xii)
a sentence like (4) I have been painting it and so has Jane is not possible since
there is no referent of it that can normally do service in both cases. In other cases
the same object can be construed, but its relationship to the process of painting is
different in the two cases, and thus the sentence still does not work: e.g., if Jane
is painting my house in sense (ii) (i.e., creating an artistic representation of it in
watercolors on paper) and I am painting it in sense (vi) or (vii) (as a housepainter),
to say (5) I am painting the house and so is Jane can only be facetious. Similarly
 Chapter 5: Schematic network 171

(6) Jane is painting her grandfather’s portrait – bright green sounds like a pun
on senses (i) and (vi) or (vii).

2.1.4. Summary
So, is paint ambiguous, or is it vague? Both, or perhaps neither. A full answer
would require elucidating more carefully than we have done here the relationships
among all these meanings and many others, including figurative meanings (paint
a gloomy picture of the prospects for peace), nominal meanings, and so forth.
These meanings are separable, yet they can also be correctly united. Webster’s
New Collegiate Dictionary makes separate entries for nominal and verbal mean-
ings, treating them exactly as it does clearly ambiguous forms,4 with multiple
subdivisions of both entries (the verb entry lists 4 transitive and 2 intransitive
meanings, plus submeanings for a total of 12 definitions; the noun entry has 3
main meanings and a total of 7 definitions). The American Heritage Dictionary
lists only one main entry, with nominal and verbal subsections (6 nominal and 10
verbal definitions); other dictionaries do it still differently. I do not think anyone
would want to say that one dictionary is right and the others wrong: rather we are
dealing with very complex groupings of meanings which can be rightly united
(treated as vague) for some purposes and distinguished (treated as ambiguous) for
others. The dictionary makers have made intelligent, even praiseworthy choices
for the purposes they have in mind, but the meanings they are representing do not
fall into classical categories, internally homogeneous and with absolute bound-
aries between them; thus no one way of splitting them up, or of uniting them, is
absolutely right.5
Thousands of other examples, from every language I know, could be given to
illustrate the same point: the same set of meanings can be both separable (ambigu-
ous-like) and unitable (vague-like). Any time a dictionary entry has sub-entries,
and especially sub-sub-entries, the dictionary makers have in effect said “these
meanings are united, but they are also usefully distinguished.” And surely our
native speaker intuitions confirm the same point. A chess set, a set in tennis, a
set of dishes, and a set in logic; or breaking a stick, a law, a horse, water, ranks,
a code, and a record; in each case the meanings are clearly rather different from
each other, but do they not have something in common as well? Our seeing the
differences between them does not cancel out our ability to see them as the same
thing (and thus call them by the same name). We see what is common between
getting food ready to eat and counselling a person about to be traumatized, and
that accounts for us calling both activities preparing, yet it feels like a pun when
you hear on the radio advice on how to prepare your turkey for Thanksgiving din-
ner – by breaking it to him gently (“Tom, are you religious?”). Wrenches, keys,
and faucets have a lot in common, and that doubtless is why they are all called
172 David Tuggy

llave in Spanish, but must we therefore suppose that Spanish speakers don’t distin-
guish them? Rather the evidence, from the linguistic tests, would show that they
can and often do do so. A Náhuatl speaker can certainly distinguish the rising
of the sun from leaving a house from a chicken hatching or a seed sprouting, but
he calls all these processes (and others) kisa ‘emerge’, and they clearly do have
important elements in common, some more than others.

2.2. Diachronic change

Another issue that is problematic for a model which posits classical categories for
ambiguity and vagueness, with no gradation between them, is diachronic change.
It is common for meanings that at one stage are united (vague) to become separate
over time (occasionally the opposite occurs as well), and the change seems to be
gradual rather than abrupt. For instance, Lewis (1967: 139), after a careful delinea-
tion of cumulative small changes in the meanings of sense, says of the juridical vs.
grammatical meanings of sentence: “This is an excellent example of the merely
homophonic status to which the different uses of a word are finally reduced. If
you said ‘Jeremy Taylor can boast the longest sentence of any English writer’ and
someone replied ‘Poor Wilde had a longer one’, this would be a pure pun.”

2.3. Summary
In sum, then, the synchronic existence of cases where the borderline between
ambiguity and vagueness is blurred, together with instances of gradual diachronic
change across that border, casts doubt on whether there is an absolute, hard and
fast line of demarcation between the two categories.

3. A cognitive model

If a model of ambiguity and vagueness (or ambiguity, polysemy, and vagueness)

as well-defined categories, internally homogenous and with clear boundaries,
does not work, what does? Categories based on prototypes have proved useful for
analyzing many other linguistic distinctions (e.g., Lakoff 1987; Langacker 1987);
perhaps they are appropriate for this one as well. I would suggest the following
model within Langacker’s Cognitive Grammar framework.6

3.1. Categorization in Cognitive Grammar

Following Langacker (1987, etc.), we call what two cognitive structures have in
common a “schema”, and represent its relationship to its elaborations (or subcases)
 Chapter 5: Schematic network 173

by arrows from the schema to each elaboration (Figure 1a); such relationships form
the basis for categorization.7 Both schemas and their elaborations can coexist in a
language; they exist to the degree that they are established (entrenched) in speak-
ers’ minds through repeated usage. Well-entrenched structures, ceteris paribus, are
more salient than less-entrenched structures, i.e., they occur more energetically.
Entrenchment can be viewed as a kind of enduring salience, i.e., salience apart
from relatively transitory effects such as directed attention or heightened activation
due to contextual factors. Degree of salience is represented by the thickness and
continuity of a box enclosing the structure in question (and secondarily by gray-
scaling the representation of the structure itself): thus Figure 1b shows a salient
structure A and a less salient structure B, both subsumed by marginally salient
schema C; if transitory salience effects are not factored in, the degree of salience
will be equivalent to the degree of entrenchment. Another parameter which we
should distinguish is elaborative distance, which we represent by the length of
the arrows from schema to elaboration: 8 a schema is distant from its elaborations
when relatively many specifications of the elaborations must be despecified to form
the schema, and close when relatively few must be despecified. Thus in Figure
1c, C is at a greater elaborative distance from A and B than in 1b. (One might,
for example, take A, B, and C in 1b to be CAT¬WEASEL¬and MAMMAL¬THAT¬EATS¬
MICE¬respectively, while in 1c they might be DOG¬STOOL¬and THING¬WITH¬LEGS¬
FOUND¬IN¬HUMAN¬DWELLINGS)

Figure 1. Schemas and elaborations

It will be readily seen that the Aristotelian definitional test amounts to looking to
see if there is a schema subsuming two meanings and making that the definition.
Many linguists seem to operate on the assumption that once such a schema is
found, the subcases it subsumes may be safely ignored, regardless of their degree
of entrenchment or salience. I am suggesting, however, that to the degree that they
are salient they must not be ignored.
174 David Tuggy

3.2. The characterization of ambiguity and vagueness

The prototypical case of ambiguity is where two semantic structures, associated
with the same phonological structure (which is called their phonological pole),
are both well-entrenched (and therefore salient), while there is no well-entrenched
and elaboratively close schema, also linked to the phonological pole, subsuming
them.9 This is represented in Figure 2a.10 Prototypical vagueness, on the other
hand, involves meanings which are not well-entrenched but which have a relatively
well-entrenched, elaboratively close schema subsuming them, as represented in
Figure 2e.11

Figure 2. The ambiguity–vagueness cline

Elaborateness (non-schematicity) of a semantic structure correlates with a smaller

extensionality; the extension of a schema includes those of its elaborations. Thus
Figure 2 corresponds to Figure 3, where Venn diagrams represent the extension of
the structures involved.12 Again thickness and continuity of the lines (with gray-
scaling) represents salience, and relative size represents schematic distance. 3a, like
2a, represents prototypical ambiguity, and 3e, like 2e, prototypical vagueness. It is
perhaps more intuitively obvious from this sort of diagram that in the ambiguous
case it is easy to separate the meanings and hard to unite them, whereas in the
vague case it is relatively hard to separate them and easy to unite them.
It is important to realize that Figure 3 is a representation of the extension of
the concepts (i.e., of the range of cases characterized by each concept’s cognitive
specifications), not a representation of the extent of the cognitive specifications
themselves (such as is implied by diagrams utilizing the “container” metaphor,
e.g., Langacker 1987: 75, Figure 2.4a, and many other places). More extensive
specifications (greater “elaboration”) corresponds to a lesser extension: thus a
diagram like Figure 4, representing the extent of semantic specifications, would
represent the same semantic structures as Figure 3 in quite opposite fashion (4a
= 3a, 4c= 3c, 4e= 3e). The diagrams in Figure 4 make it clear that the schema
is completely immanent to its elaborations (Langacker 1987: 180, 438–439): i.e.,
they contain all its specifications and more; the schema comprises those specifi-
 Chapter 5: Schematic network 175

cations which are common to the two subcases. Again relative size of a schema
vs. its subcase represents cognitive distance: when the schema is much smaller
than the subcase that means the subcase adds many specifications to those of the
schema and is thus quite distant from it.

Figure 3. The ambiguity–vagueness cline showing extensionality of semantic structures

Figure 4. The ambiguity–vagueness cline showing extent of semantic structures

For many purposes, speakers will “filter out” specifications below a certain level
of salience. If the threshold of salience is set higher than the level of salience
of any uniting schema in 2a/3a or of the elaborate structures (the subcases) in
2e/3e, 2a/3a will have two completely separate meanings attached to the same
phonological pole, while 2e/3e has only one meaning. (A rough visual analog to
imposing such a threshold can be achieved by squinting more or less tightly at
Figures 2–6.13) This corresponds exactly to the traditional characterizations of
the ambiguity/vagueness distinction.

3.3. The in-between cases: Polysemy

Since the differences between Figure 2a and 2e (3a and 3e) are gradual, we can
expect to find cases in the middle, where a schema exists but is not salient and/or
is distant (2b/3b), or where the subcases are somewhat salient but not so much
so as the schema (2d/3d), or even where both the schema and the subcases are
salient (2c/3c). These of course are the polysemous cases discussed in section
2.1, the border straddling cases, with meanings both clearly separable and clearly
united. But the important thing is that the differences among the categories are
gradual, not absolute.
Note that if a high enough threshold of salience is imposed (if a speaker
“squints” tightly enough), the schema in 2b (3b) is filtered out, leaving a con-
176 David Tuggy

figuration which is effectively the same as 2a (3a). Similarly, if the threshold of

salience is set above the level of salience of the sub-cases in 2d (3d), 2d (or 3d)
becomes effectively equivalent to 2e (3e). This, together with the influence of
contextual salience mentioned below (3.4.3), makes it even more impossible to
draw absolute boundary lines between the categories of ambiguity, polysemy,
and vagueness.

3.4. Fitting the data to the model

3.4.1. Straightforward cases

This model fits with the sorts of phenomena we find in languages. The case of
bank “financial institution” and “edge of a river” fits very nicely at the ambigu-
ous end of the spectrum: the two meanings are well established, and the nearest
schema subsuming the two would be something like “Thing”, which is quite distant
and is not linked to the phonological pole /baenk/ (5a, cf. 2a). At the vague end
(5e, cf. 2e) we can place aunt, with the non-salient subcases “father’s sister” vs.
“mother’s sister” subsumed at a minimal elaborative distance under the concept
“parent’s sister”, which is clearly better established and thus more salient than
they as a meaning of aunt. An example of 2c (5c) would be the notions of artis-
tic painting (including painting in oils and watercolors as in (i–ii) above) and of
typical utilitarian painting (subsuming (vi–viii) and others): both well-established
concepts, but clearly united under an also well-established schema not greatly
distant from the subcases.

3.4.2. Flexibility
Figure 2 of course oversimplifies things. Although elaborative distance between
a schema and its elaborations tends to correlate inversely to entrenchment of the
schema, the parameters are not absolutely parallel; Figure 2 does not represent
this. Also one typically is dealing not with two or three meanings but with many
more, arranged in multiply overlapping hierarchies – cf. Figure 6, of which 5c is
an abbreviation, with 5c=6a–c. (Figure 6 itself is of course far from exhaustive of
the meaning of paint.) Furthermore it is relatively rare for the subcases in ques-
tion to have the same degree of salience; they are more likely to differ along this
parameter, as in 1b–c (and, again, 6). These would all be valid criticisms if the
presentation in Figure 2 were intended as absolute, but they are in fact predicted
by the model, which has built into it the flexibility to accommodate them. Granted
that Figure 2 is incomplete in this way, the main point nevertheless remains
valid: To the degree that a group of meanings and any schema subsuming them
approaches the configuration given in 2a, we can call it a case of ambiguity, and
to the degree that it approaches 2e, it is vagueness.
 Chapter 5: Schematic network 177

Figure 5. Typical examples along the ambiguity–vagueness cline

Figure 6. Paint

3.4.3. The influence of context

Given this model, the influence of context comes for free. For salience is not a static
characteristic, but a dynamic one. Among the factors enhancing it is entrenchment,
as previously noted, but also degree of activation produced by context.14 When
a cognitive structure has itself been recently activated, whether in the linguistic
or in some other cognitive context, or when structures closely linked to it have
been activated, it retains a residual activation, and it will reactivate at a higher
level of energy.15 Thus when the word painter or even more especially the phrase
be a painter is used as in sentence (2) above, the specifications attached to -er
“one who does the verb professionally”, and to the phrase be a V-er, enhance the
salience of those portions of the meaning(s) of paint which have to do with innate
abilities and skills employed, purpose for which done, and so on. This moves
178 David Tuggy

the configuration of 5c towards one more like 2b, increasing the salience of the
elaborations at the expense of the schema. The fact that the same two readings of
paint feel more ambiguous in (2) than in (1) is thus to be expected. In sentence
(3), in contrast, the focus is on the even spreading of the color in painting, which
characterizes both subcases and therefore the schema as well; thus the differences
between the subcases are not especially activated and the schema (i.e., what is
common to the subcases) is rendered more salient. This moves the configuration
of 5c towards one more like 2d, which explains why all the senses of paint test as
vague by sentence (3).16 When paint is used intransitively, as in (1), the nature of
the object is less salient: When it is used transitively, as in (4) or (5), it is rendered
more salient, and must match in the two cases construed for identity-of-sense
anaphora to function normally.

3.4.4. Puns
Puns like (6) can also be explained: They are a special case of context enhanc-
ing the salience of meanings. In general puns involve a double context, in which
two meanings associated with a particular phonological pole are rendered salient.
Thus saying that a pirate burying his gold by the river is putting his money in the
bank involves the context of the river edge, which renders salient that meaning of
bank, but also the phrase put (your) money in the bank, which strongly activates
the financial institution meaning. The two meanings are overlaid on each other,
producing the semantic discomfort we call a pun. The situation in (6) is similar:
The construction paint X’s portrait strongly activates, and thus renders salient, a
meaning like (i) (Figure 6d), while paint Y color¬activates something like (vi–ix)
(6e). Both meanings are thus rendered salient by the context, yielding a configu-
ration like 2b, and overlaid on each other. In exactly the same way, the phrase
prepare your turkey for Thanksgiving dinner renders salient one meaning of pre-
pare, while the continuation by breaking it to him gently renders another meaning
salient, and the superimposition of the two salient meanings constitutes a pun.

3.4.5. Diachronic change

And, finally, the model clearly shows the distinction between ambiguity and
vagueness to be a gradual one, with synchronic variation of meanings along the
parameters of differentiation. This brings ease to the problem of diachronic change
from one category to the other. There is no hard and fast boundary that a form
needs to jump all at once: it can straddle the fence indefinitely, shifting its weight
back and forth, before gradually moving more to one side than the other.
 Chapter 5: Schematic network 179

4. Summary

I commend this model to you on the following grounds:

(1) The theoretical constructs involved are well motivated on independent
grounds (see extensive argumentation in Langacker 1987 and elsewhere).
(2) The traditional view of the ambiguity/vagueness distinction is seen as
essentially correct; its intuitive characterizations fit the model nicely. It was only
wrong if the categories were made absolute.
(3) The value of the traditional linguistic and definitional tests is affirmed:
they clearly and correctly distinguish prototypically vague from prototypically
ambiguous structures. The different behavior of ambiguous and vague structures
in puns also fits into the model.
(4) The model correctly allows for a range of in-between (polysemous) cases,
neither ambiguous nor vague in the prototypical sense (although in the cases we
examined they are predictably vague by the definitional test).
(5) The influence of context (including the possibility of puns between related
senses of a polysemous form) is accounted for automatically.
(6) The gradual nature of the ambiguity/vagueness distinction allows for grad-
ual rather than abrupt diachronic shifting from one category to the other.

Notes

1. Quine’s logical test is difficult for me to apply in these cases (as in most others).
Applying it, for instance, to examples mentioned below, I could say, of a case of
painting stripes in a parking lot (x) or even more of swabbing iodine (xii), “She’s
painting, but she’s not (really) painting”. By that I would mean that what she is
doing may be properly called painting, but it is not painting proper, i.e. it is not the
prototypical kind of painting. This might be taken as indicating ambiguity, but it is
not very satisfactory; it does not precisely yield the “painting in one sense but not
the other” reading that Quine was after. Also it does not work for more prototypical
cases – e.g. (i), or (vi) or (vii). In general, however, to say of a person in any of these
situations “She’s painting, but she’s not painting”, with no special intonation, seems
infelicitous if not anomalous. This rather indicates vagueness. On the other hand,
“This is the bank but it’s not the bank” is far from felicitous for me as well. Kempson
(1977: 129), for rather different reasons, also concludes that “the characterization of
ambiguity as the simultaneous assignment to a sentence of the values true and false
has not provided a criterion for deciding unclear cases; it merely accentuates the
point of disagreement.”
2. Traditional treatments of polysemy, however, tend to concentrate on cases in which
the senses involved are not as closely related as those I am interested in: e.g., they
would be more likely to discuss whether the nominal and verbal senses of paint can
180 David Tuggy

be considered the same lexeme than to consider the distinctions among verbal senses
I list in (i–xii).
3. Lyons (1977: 554) considers such in-between cases with the verb play: “Could we
delete the second occurrence of the form plays in [She plays chess better than she
plays the flute]? And what about [?He played scrum-half in the afternoon and Ham-
let in the evening]?” Lyons concludes that “It may well be that the whole notion of
discrete lexical senses is ill-founded.”
4. “1paint vb” vs. “2paint n”. The treatment is thus exactly parallel to that of “1painter
n ... one that paints” (with subcases for artists and for housepainters and such) vs.
“2painter n [... prob. fr. MF¬pendoir ...]: a line used for securing or towing a boat”
vs. “3painter n ... [alter. of panther]... COUGARv¬three meanings with little or noth-
ing in common beyond designating a thing of some sort. Cf. Lyons’ (1977: 21–22)
comment, “The fact that there are two separate entries means that the compilers or
editors of the dictionary have decided that two distinct lexemes are involved (and
not one lexeme with two meanings”).
5. Langacker’s discussion of the notion “lexical item” (1987: 388) is apropos: “A cat-
egory is coherent to the extent that its members are densely linked by well-entrenched
categorizing relationships of minimal distance. ... The coherence of a category is
naturally a matter of degree ... To the extent that a semantic network with common
symbolization [i.e. linked to a common phonological pole] approximates a coherent
category, we can reasonably speak of a lexical item. Despite its convenience, however,
this construct is more a descriptive fiction than a natural unit of linguistic organization.
Not only is coherence inherently a matter of degree, but also the definition allows a
single network to be divided into lexical items in multiple and mutually inconsistent
ways. I regard this as a realistic characterization of the phenomena in question.”
6. Most of this discussion is prefigured to some extent in Tuggy (1981: 56–59, 72), and
I have been using the essence of the diagram in Figure 2 at least since 1984. Never-
theless, and although Langacker (1988: 137–139) credits my work, the ideas involved
rest crucially on constructs of his theory, and in any case they came to me at least in
part through discussions with him, Sue Lindner, Mary Ellen Ryder, Jeff Burnham,
Barbara Levergood, and others. Thus I can claim no absolute credit for them.
7. For the discussion of this paper an essentially Aristotelian definition of schema is
adequate, in which the schema contains only (and all) material common to the sub-
cases, and any concept which includes that material is a subcase. This, while not the
only possible conceptualization, is doubtless the prototypical one, and is that described
repeatedly by Langacker (e.g. 1987: 68, 132–138). The main point of this paper with
respect to it is that (i) the existence of such a schema as part of the meaning of a
lexical item does not preclude the existence of its subcases: in fact the subcases may
well be more salient than the schema; and (ii) such variations in salience among a
schema and its subcases account for the non-absoluteness of the distinctions between
ambiguity, (polysemy), and vagueness.
A separate question is whether this conception of a schema is sufficient. (This is one
of the important issues addressed by Geeraerts 1993.) Specifically, can a schema
contain “either/or” specifications, or lists of alternative characteristics which do not
have anything in common, or can it consist of specifications which are not charac-
teristic (e.g. can one speak of a “schema” uniting checkmate and pawn, consisting of
 Chapter 5: Schematic network 181

the game of chess, which figures saliently in the meaning of both but characterizes
neither)? Although I do not argue it here, I believe such “schemas” (if the term is
appropriate to use for them) do indeed serve to unite concepts into categories and thus
are relevant to the concerns of this paper. Langacker (e.g. 1987: 69, 92–93) speaks
at length of a relationship of “partial schematicity” or “extension” in which there is
conflict between the specifications of the “schema” and the “elaboration”, using it to
describe the relationship between a prototype and non-prototypical members of the
same category, or between the literal and figurative senses of a metaphor, or a gram-
matical pattern and an ill-formed instance of that pattern; the prototypical kind of
schematicity is the limiting case along a parameter of compatibility, in which there is
no conflict of specifications. I would expect that there are other parameters involved
in the other cases of non-prototypical “schemas” (those involving alternatives or lists,
non-characteristic specifications, etc.), and that the prototypical schematicity will
again prove to be the limiting and most salient case. In any case, I believe that the
conclusions outlined above still hold: Both the uniting concept (of whatever type)
and the concepts which it unites may exist simultaneously, and differences in salience
among them will make the configurations we recognize as ambiguity and vagueness
non-absolute.
8. The limitations of 2-dimensional representation make this convention impracticable
for most purposes; it should not be taken too seriously in Figure 6, for instance. It
has not to my knowledge been used in other publications on Cognitive Grammar.
9. Ultimately there will always be some schema uniting any two cognitive structures,
even if it be only the superschema ENTITY¬(Langacker 1987: 198), which corresponds
to the notions “concept” or “cognitive structure”. Frequently there is even some
slightly more fully specified schema; e.g. THING¬which unites the two meanings
of bank (5a), or LARGE¬PHYSICAL¬THING¬which would unite the river edge sense
of bank with the physical building sub-sense of the financial institution sense. The
important things to notice, however, are that such a schema will be quite distant in
cases of clear ambiguity, and, more importantly, that it will not be linked directly to
the phonological pole of the lexical item in question. Thus THING¬is not a meaning
of /baenk/.
Another consideration, closely related to elaborative distance, is the possible existence
of schemas intermediate between the uniting schema and the elaborations, but not
linked to the same phonological pole. The uniting schema will not usually still be
linked to the phonological pole in such a case. One example, however, might be the
schema MOST¬IMPORTANT¬PART¬which is arguably a meaning of head (it is used pro-
ductively, especially with adjectival head). Although the cutting piece of a machine
designed for cutting is its most important part and in some cases it is called a head
(e.g., in a milling machine), in many other cases it is not (e.g., in a carpenter’s plane
or a pair of scissors).
10. In Figures 2–6 the connecting links from phonological structures are represented
as varying in salience directly with the salience of the semantic structures to which
they connect. Actually it is in some sense the salience of the link which is primary.
For assessing position along the ambiguity/vagueness continuum, what counts is
not salience of meanings as they might be measured objectively, from some outside
viewpoint, but rather their salience as measured from the phonological pole as point
182 David Tuggy

of access. E.g., the concept THING¬is, in and of itself, presumably quite a salient
one in English speakers’ minds, probably more so than RIVER¬EDGE¬or lNANCIAL¬
INSTITUTION¬but it is not directly accessed from the phonological structure /baenk/,
so from the perspective of that phonological structure it is relatively non-salient, as
represented in Figure 5a.
11. Figures 2a and e are essentially the same configurations as Langacker’s (1988: 138)
Figures 3b and 3a.
12. This diagram may be instructively compared with Geeraerts’ (1991: 202) representa-
tion of the meaning of vers; if Figure 6 were converted into this type of diagram the
result would be similarly complex. Note however that Geeraerts’ diagram does not
represent salience, except in the cross-hatching that marks the area which includes
the “most central meaning”.
13. This is somewhat misleading in that squinting requires effort; the default case of
looking is to have the eyes wide open. In cognition presumably the default case is
rather to have a fairly high threshold of salience (i.e., to attend only to structures
which are quite highly activated), and it will require effort to lower that threshold
and “see” more detail (cf. Kosslyn 1980: 53, 165–166; Langacker 1987: 381).
14. We are here dealing with what happens from the perspective of a hearer decod-
ing speech. Things are doubtless more complex from the perspective of a thinker,
for whom volition and other factors are more active in establishing what cognitive
structures are salient. It is here that Geeraerts’ “searchlight” metaphor (see Geeraerts
1993: 258–260, 263) is especially appropriate; the thinker can turn the searchlight of
salience (heightened cognitive activity) where he wants to (and has the mental energy
to), not just where the lights are already on most strongly. He then, if he wishes to
communicate as a speaker, can construct a context which will evoke parallel saliences
for the hearer; it is the hearer’s decoding of such a communication that we are dealing
with. Even here, of course, volition and other factors can enter the picture; hearers
are also active thinkers. But when these other factors are allowed to counteract the
effect of context, they generally foul up the communication process. In ideal com-
munication between a skilled speaker and a cooperative hearer, the pre-established
saliences of entrenchment and the influence of context are the most important factors
determining what is salient in the hearer’s mind.
15. This same mechanism is what accounts for hearers’ choosing of one possible meaning
instead of others because it fits the context. Thus if one says The current slammed the
boat against the bank or the current carried the boat into the reeds, the specifications
of current and boat, involving as they do the notions of water moving in a stream,
produce a secondary activation in concepts closely linked with them, such as bed (of
the stream) or the river edge sense of bank or the aquatic plant sense of reed. This
makes it quite improbable that hearers of the sentences will think of the financial
institution sense of bank or the musical instrument sense of reed at all, much less
to the exclusion of the river edge or aquatic plant sense. Of course a larger context
(e.g., of a flood overflowing a town, with perhaps an orchestra bravely carrying on
despite the water rising around them) might render those meanings salient instead
or as well. If both meanings are simultaneously salient, the sentences will feel like
puns (3.4.4).
16. In a context of overt comparison, any feature whatsoever which two homophonous
 Chapter 5: Schematic network 183

semantic structures have in common can function as a salient schema, rendering the
configuration vague and allowing identity-of-sense anaphora to function. This is the
significance of Deane’s sentence (1988: 345): Financial banks resemble those you
find by rivers: they control, respectively, the flow of money and of water. As Deane
says, “in effect, the context temporarily creates a single category”.
In the extreme case, where the only relevant feature two semantic structures have in
common is linkage to the same phonological pole, a context of naming or definition
can set up just that as the most salient feature. This is the point of Geeraerts’ sentence:
Daddy, what exactly do you call a bank: the place where we moor the boat or the
place where I bring my savings? – Well, son, the place where we moor the boat is a
bank, but so is the place where you bring your savings (Geeraerts 1993: 245).

References
The American Heritage Dictionary of the English Language
1978 New College Edition. Boston: Houghton Mimin.
Aristotle
1984 Post. Anal. [Link]. In The Complete Works of Aristotle: The Revised Oxford
Translation, Jonathan Barnes (ed.). Princeton: Princeton University Press.
Deane, Paul
1988 Polysemy and cognition. Lingua 75: 325–361.
Geeraerts, Dirk
1991 The lexicographical treatment of prototypical polysemy. In Tsohatzidis (ed.),
195–210.
1993 Vagueness’s puzzles, polysemy’s vagaries. Cognitive Linguistics 4(3):
223–272.
Hartmann, R. R. K.
1983 On theory and practice. In Lexicography: Principles and Practice, R. R. K.
Hartmann (ed.), 3–11. London: Academic Press.
Kempson, Ruth M.
1977 Semantic Theory. Cambridge: Cambridge University Press.
Kimball, John P. (ed.)
1975 Syntax and Semantics, vol. 4. New York: Academic Press.
Kosslyn, Stephen Michael
1980 Image and Mind. Cambridge, MA: Harvard University Press.
Lakoff, George
1970 A note on vagueness and ambiguity. Linguistic Inquiry 1: 357–359.
1987 Women, Fire and Dangerous Things. What Categories Reveal about the Mind.
Chicago: University of Chicago Press.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar. vol. 1: Theoretical Prerequisites. Stan-
ford, CA: Stanford University Press.
1988 A usage-based model. In Rudzka-Ostyn (ed.), 127–161.
184 David Tuggy

1991 Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin/New
York: Mouton de Gruyter.
Lewis, Clive S.
1967 Studies in Words. 2nd edition. New York: Cambridge University Press.
Lyons, John
1977 Semantics. 2 vols. Cambridge: Cambridge University Press.
Quine, Willard V. 0.
1960 Word and Object. Cambridge, MA: MIT Press.
Rudzka-Ostyn, Brygida (ed.)
1988 Topics in Cognitive Linguistics. Amsterdam/Philadelphia: John Benjamins.
Tuggy, David
1981 The transitivity-related morphology of Tetelcingo Nahuatl: An exploration in
Space grammar. Doctoral dissertation, University of California, San Diego.
Tsohatzidis, Savas L. (ed.)
1991 Meanings and Prototypes: Studies in Linguistic Categorization. London/New
York: Routledge.
Webster’s New Collegiate Dictionary
1977 Springfield, MA: Merriam.
Zwicky, Arnold M. and Jerrold M. Sadock
1975 Identity tests and how to fail them. In Kimball (ed.), 1–36.
Chapter 6
Conceptual metaphor

The contemporary theory of metaphor *

George Lakoff

Do not go gentle into that good night.

Dylan Thomas
Death is the mother of beauty.
Wallace Stevens, “Sunday Morning”

1. Introduction

These famous lines by Thomas and Stevens are examples of what classical theo-
rists, at least since Aristotle, have referred to as metaphor: instances of novel
poetic language in which words like “mother,” “go,” and “night” are not used in
their normal everyday sense. In classical theories of language, metaphor was seen
as a matter of language, not thought. Metaphorical expressions were assumed to
be mutually exclusive with the realm of ordinary everyday language: everyday
language had no metaphor, and metaphor used mechanisms outside the realm of
everyday conventional language.
The classical theory was taken so much for granted over the centuries that
many people didn’t realize that it was just a theory. The theory was not merely
taken to be true, but came to be taken as definitional. The word “metaphor” was
defined as a novel or poetic linguistic expression where one or more words for
a concept are used outside of their normal conventional meaning to express a
“similar” concept.
But such issues are not matters for definitions; they are empirical questions. As
a cognitive scientist and a linguist, one asks: what are the generalizations govern-
ing the linguistic expressions referred to classically as “poetic metaphors”? When
this question is answered rigorously, the classical theory turns out to be false. The
generalizations governing poetic metaphorical expressions are not in language,
but in thought: they are general mappings across conceptual domains. Moreover,
these general principles which take the form of conceptual mappings, apply not
just to novel poetic expressions, but to much of ordinary everyday language.
In short, the locus of metaphor is not in language at all, but in the way we
conceptualize one mental domain in terms of another. The general theory of meta-

Originally published in 1993 in Metaphor and Thought, Andrew Ortony (ed.), 202–251. Cambridge:
Cambridge University Press. Reprinted with permission.
186 George Lakoff

phor is given by characterizing such cross-domain mappings. And in the process,

everyday abstract concepts like time, states, change, causation, and purpose also
turn out to be metaphorical.
The result is that metaphor (that is, cross-domain mapping) is absolutely central
to ordinary natural language semantics, and that the study of literary metaphor is
an extension of the study of everyday metaphor. Everyday metaphor is character-
ized by a huge system of thousands of cross-domain mappings, and this system
is made use of in novel metaphor.
Because of these empirical results, the word “metaphor” has come to be used
differently in contemporary metaphor research. It has come to mean “a cross-
domain mapping in the conceptual system.” The term “metaphorical expression”
refers to a linguistic expression (a word, phrase, or sentence) that is the surface
realization of such a cross-domain mapping (this is what the word “metaphor”
referred to in the old theory). I will adopt the contemporary usage throughout
this chapter.
Experimental results demonstrating the cognitive reality of the extensive sys-
tem of metaphorical mappings are discussed by Gibbs (1993). Mark Turner’s 1987
book Death Is the Mother of Beauty, whose title comes from Stevens’ great line,
demonstrates in detail how that line uses the ordinary system of everyday map-
pings. For further examples of how literary metaphor makes use of the ordinary
metaphor system, see More Than Cool Reason: A Field Guide to Poetic Meta-
phor by Lakoff and Turner (1989) and Reading Minds: The Study of English in
the Age of Cognitive Science by Turner (1991).
Since the everyday metaphor system is central to the understanding of poetic
metaphor, we will begin with the everyday system and then turn to poetic
examples.

1.1. Homage to Ready

The contemporary theory that metaphor is primarily conceptual, conventional,
and part of the ordinary system of thought and language can be traced to Michael
Reddy’s (1993) now classic essay, “The Conduit Metaphor,” which first appeared
in the first edition of Ortony (1993[1979]). Reddy did far more in that essay than
he modestly suggested. With a single, thoroughly analyzed example, he allowed
us to see, albeit in a restricted domain, that ordinary everyday English is largely
metaphorical, dispelling once and for all the traditional view that metaphor is
primarily in the realm of poetic or “figurative” language. Reddy showed, for a
single, very significant case, that the locus of metaphor is thought, not language,
that metaphor is a major and indispensable part of our ordinary, conventional way
of conceptualizing the world, and that our everyday behavior reflects our meta-
phorical understanding of experience. Though other theorists had noticed some
 Chapter 6: Conceptual metaphor 187

of these characteristics of metaphor, Reddy was the first to demonstrate them by

rigorous linguistic analysis, stating generalizations over voluminous examples.
Reddy’s chapter on how we conceptualize the concept of communication by
metaphor gave us a tiny glimpse of an enormous system of conceptual metaphor.
Since its appearance, an entire branch of linguistics and cognitive science has
developed to study systems of metaphorical thought that we use to reason and base
our actions on, and that underlie a great deal of the structure of language.
The bulk of the chapters in Ortony (1993[1979]), in which the present article
appeared originally, were written before the development of the contemporary
field of metaphor research. My chapter therefore contradicts much that appears in
the other chapters of Ortony (1993[1979]), many of which make certain assump-
tions that were widely taken for granted in 1977. A major assumption that is chal-
lenged by contemporary research is the traditional division between literal and
figurative language, with metaphor as a kind of figurative language. This entails,
by definition, that: what is literal is not metaphorical. In fact, the word “literal”
has traditionally been used with one or more of a set of assumptions that have
since proved to be false:

1.2. Traditional false assumptions

− All everyday conventional language is literal, and none is metaphorical.
− All subject matter can be comprehended literally, without metaphor.
− Only literal language can be contingently true or false.
− All definitions given in the lexicon of a language are literal, not
metaphorical.
– The concepts used in the grammar of a language are all literal; none are
metaphorical.

The big difference between the contemporary theory and views of metaphor prior
to Reddy’s work lies in this set of assumptions. The reason for the difference is
that, in the intervening years, a huge system of everyday, conventional, concep-
tual metaphors has been discovered. It is a system of metaphor that structures
our everyday conceptual system, including most abstract concepts, and that lies
behind much of everyday language. The discovery of this enormous metaphor
system has destroyed the traditional literal-figurative distinction, since the term
“literal,” as used in defining the traditional distinction, carries with it all those
false assumptions.
A major difference between the contemporary theory and the classical one is
based on the old literal-figurative distinction. Given that distinction, one might
think that one “arrives at” a metaphorical interpretation of a sentence by “start-
ing” with the literal meaning and applying some algorithmic process to it (see
188 George Lakoff

Searle 1993). Though there do exist cases where something like this happens,
this is not in general how metaphor works, as we shall see shortly.

1.3. What is not metaphorical

Although the old literal-metaphorical distinction was based on assumptions that
have proved to be false, one can make a different sort of literal-metaphorical
distinction: those concepts that are not comprehended via conceptual metaphor
might be called “literal.” Thus, although I will argue that a great many common
concepts like causation and purpose are metaphorical, there is nonetheless an
extensive range of non-metaphorical concepts. A sentence like the balloon went
up is not metaphorical, nor is the old philosopher’s favorite the cat is on the mat.
But as soon as one gets away from concrete physical experience and starts talking
about abstractions or emotions, metaphorical understanding is the norm.

2. The contemporary theory: Some examples

Let us now turn to some examples that are illustrative of contemporary meta-
phor research. They will mostly come from the domain of everyday conventional
metaphor, since that has been the main focus of the research. I will turn to the
discussion of poetic metaphor only after I have discussed the conventional system,
since knowledge of the conventional system is needed to make sense of most of
the poetic cases.
The evidence for the existence of a system of conventional conceptual meta-
phors is of five types:

− Generalizations governing polysemy, that is, the use of words with a num-
ber of related meanings
− Generalizations governing inference patterns, that is, cases where a pattern
of inferences from one conceptual domain is used in another domain
− Generalizations governing novel metaphorical language (see Lakoff and
Turner 1989)
− Generalizations governing patterns of semantic change (see Sweetser 1990)
– Psycholinguistic experiments (see Gibbs 1990)

We will be discussing primarily the first three of these sources of evidence, since
they are the most robust.
 Chapter 6: Conceptual metaphor 189

2.1. Conceptual metaphor

Imagine a love relationship described as follows:

Our relationship has hit a dead-end street.

Here love is being conceptualized as a journey, with the implication that the
relationship is stalled, that the lovers cannot keep going the way they’ve been
going, that they must turn back, or abandon the relationship altogether. This is
not an isolated case. English has many everyday expressions that are based on a
conceptualization of love as a journey, and they are used not just for talking about
love, but for reasoning about it as well. Some are necessarily about love; others
can be understood that way:

Look how far we’ve come.

It’s been a long, bumpy road.
We can’t turn back now.
We‘re at a crossroads.
We may have to go our separate ways.
The relationship isn’t going anywhere.
We‘re spinning our wheels.
Our relationship is off the track.
The marriage is on the rocks.
We may have to bail out of this relationship.

These are ordinary, everyday English expressions. They are not poetic, nor are
they necessarily used for special rhetorical effect. Those like look how far we’ve
come, which aren’t necessarily about love, can readily be understood as being
about love.
As a linguist and a cognitive scientist, I ask two commonplace questions:

− Is there a general principle governing how these linguistic expressions

about journeys are used to characterize love?
– Is there a general principle governing how our patterns of inference about
journeys are used to reason about love when expressions such as these are
used?

The answer to both is yes. Indeed, there is a single general principle that answers
both questions, but it is a general principle that is neither part of the grammar
of English, nor the English lexicon. Rather, it is part of the conceptual system
underlying English. It is a principle for understanding the domain of love in terms
of the domain of journeys.
190 George Lakoff

The principle can be stated informally as a metaphorical scenario:

The lovers are travelers on a journey together, with their common life goals seen
as destinations to be reached. The relationship is their vehicle, and it allows them
to pursue those common goals together. The relationship is seen as fulfilling its
purpose as long as it allows them to make progress toward their common goals.
The journey isn’t easy. There are impediments, and there are places (crossroads)
where a decision has to be made about which direction to go in and whether to
keep traveling together.

The metaphor involves understanding one domain of experience, love, in terms

of a very different domain of experience, journeys. More technically, the meta-
phor can be understood as a mapping (in the mathematical sense) from a source
domain (in this case, journeys) to a target domain (in this case, love). The map-
ping is tightly structured. There are ontological correspondences, according to
which entities in the domain of love (e.g., the lovers, their common goals, their
difficulties, the love relationship, etc.) correspond systematically to entities in the
domain of a journey (the travelers, the vehicle, destinations, etc.).
To make it easier to remember what mappings there are in the conceptual
system, Johnson and I (Lakoff and Johnson 1980) adopted a strategy for naming
such mappings, using mnemonics which suggest the mapping. Mnemonic names
typically (though not always) have the form: TARGET DOMAIN¬IS SOURCE DOMAIN¬
or alternatively, TARGET DOMAIN¬AS¬SOURCE DOMAIN¬In this case, the name of
the mapping is LOVE¬IS A¬JOURNEY¬When I speak of the LOVE¬IS¬A¬JOURNEY¬
metaphor, I am using a mnemonic for a set of ontological correspondences that
characterize a mapping, namely:

THE¬LOVE AS JOURNEY¬MAPPING
− The lovers correspond to travelers.
− The love relationship corresponds to the vehicle.
– The lovers’ common goals correspond to their common destinations on the
journey.

Difficulties in the relationship correspond to impediments to travel. It is a com-

mon mistake to confuse the name of the mapping, LOVE¬IS¬A¬JOURNEY¬for the
mapping itself. The mapping is the set of correspondences, Thus, whenever I refer
to a metaphor by a mnemonic like LOVE¬IS A¬JOURNEY¬I will be referring to such
a set of correspondences.
If mappings are confused with names of mappings, another misunderstanding
can arise. Names of mappings commonly have a propositional form, for example,
LOVE¬IS A¬JOURNEY¬But the mappings themselves are not propositions. If map-
pings are confused with names for mappings, one might mistakenly think that, in
 Chapter 6: Conceptual metaphor 191

this theory, metaphors are propositional. They are anything but that: metaphors
are mappings, that is, sets of conceptual correspondences.
The LOVE AS JOURNEY¬mapping is a set of ontological correspondences that
characterize epistemic correspondences by mapping knowledge about journeys
onto knowledge about love. Such correspondences permit us to reason about love
using the knowledge we use to reason about journeys. Let us take an example.
Consider the expression, we’re stuck, said by one lover to another about their
relationship. How is this expression about travel to be understood as being about
their relationship?
We’re stuck can be used of travel, and when it is, it evokes knowledge about
travel. The exact knowledge may vary from person to person, but here is a typical
example of the kind of knowledge evoked. The capitalized expressions represent
entities in the ontology of travel, that is, in the source domain of the LOVE IS A
JOURNEY¬mapping given above.

Two TRAVELERS¬are in a VEHICLE¬TRAVELING¬WITH¬COMMON¬DESTINA

TIONS¬The VEHICLE¬encounters some IMPEDIMENT¬and gets stuck, that is,
becomes nonfunctional. If the travelers do nothing, they will not REACH¬
THEIR¬DESTINATIONS¬There are a limited number of alternatives for
action:
They can try to get the vehicle moving again, either by fixing it or getting
it past the IMPEDIMENT¬that stopped it.
They can remain in the nonfunctional VEHICLE¬and give up on REACHING¬
THEIR¬DESTINATIONS.
They can abandon the VEHICLE
The alternative of remaining in the nonfunctional VEHICLE¬takes the least
effort, but does not satisfy the desire to REACH¬THEIR¬DESTINATIONS

The ontological correspondences that constitute the LOVE¬IS A¬JOURNEY¬metaphor

map the ontology of travel onto the ontology of love. In doing so, they map this
scenario about travel onto a corresponding love scenario in which the correspond-
ing alternatives for action are seen. Here is the corresponding love scenario that
results from applying the correspondences to this knowledge structure. The target
domain entities that are mapped by the correspondences are capitalized:

Two LOVERS¬are in a LOVE¬RELATIONSHIP¬PURSUING¬COMMON¬LIFE¬GOALS¬

The RELATIONSHIP¬encounters some DIFlCULTY¬which makes it nonfunc-
tional. If they do nothing, they will not be able to ACHIEVE¬THEIR¬LIFE¬
GOALS¬There are a limited number of alternatives for action:
They can try to get it moving again, either by fixing it or getting it past
the DIFlCULTY.
192 George Lakoff

They can remain in the nonfunctional RELATIONSHIP¬and give up on

ACHIEVING¬THEIR¬LIFE¬GOALS.
They can abandon the RELATIONSHIP
The alternative of remaining in the nonfunctional RELATIONSHIP¬takes the
least effort, but does not satisfy the desire to ACHIEVE¬LIFE¬GOALS

This is an example of an inference pattern that is mapped from one domain to

another. It is via such mappings that we apply knowledge about travel to love
relationships.

2.2. Metaphors are not mere words

What constitutes the LOVE¬AS¬JOURNEY¬metaphor is not any particular word or
expression. It is the ontological mapping across conceptual domains, from the
source domain of journeys to the target domain of love. The metaphor is not just
a matter of language, but of thought and reason. The language is secondary. The
mapping is primary, in that it sanctions the use of source domain language and
inference patterns for target domain concepts. The mapping is conventional, that
is, it is a fixed part of our conceptual system, one of our conventional ways of
conceptualizing love relationships.
This view of metaphor is thoroughly at odds with the view that metaphors
are just linguistic expressions. If metaphors were merely linguistic expressions,
we would expect different linguistic expressions to be different metaphors. Thus,
We’ve hit a dead-end street would constitute one metaphor. We can’t turn back
now would constitute another, entirely different metaphor. Their marriage is
on the rocks would involve still a different metaphor. And so on for dozens of
examples. Yet we don’t seem to have dozens of different metaphors here. We have
one metaphor, in which love is conceptualized as a journey. The mapping tells
us precisely how love is being conceptualized as a journey. And this unified way
of conceptualizing love metaphorically is realized in many different linguistic
expressions.
It should be noted that contemporary metaphor theorists commonly use the
term “metaphor” to refer to the conceptual mapping, and the term “metaphorical
expression” to refer to an individual linguistic expression (like dead-end street)
that is sanctioned by a mapping. We have adopted this terminology for the fol-
lowing reason: Metaphor, as a phenomenon, involves both conceptual mappings
and individual linguistic expressions. It is important to keep them distinct. Since
it is the mappings that are primary and that state the generalizations that are our
principal concern, we have reserved the term “metaphor” for the mappings, rather
than for the linguistic expressions.
In the literature of the field, small capitals like LOVE¬IS A¬JOURNEY¬are used
 Chapter 6: Conceptual metaphor 193

as mnemonics to name mappings. Thus, when we refer to the LOVE¬IS A¬JOURNEY¬

metaphor, we are referring to the set of correspondences discussed above. The
English sentence love is a journey, on the other hand, is a metaphorical expres-
sion that is understood via that set of correspondences.

2.3. Generalizations
The LOVE¬IS A¬JOURNEY¬metaphor is a conceptual mapping that characterizes a
generalization of two kinds:

− Polysemy generalization: a generalization over related senses of linguis-

tic expressions, for example, dead-end street, crossroads, stuck, spin-
ning one’s wheels, not going anywhere, and so on.
– Inferential generalization: a generalization over inferences across differ-
ent conceptual domains.

That is, the existence of the mapping provides a general answer to two
questions:

− Why are words for travel used to describe love relationships?

– Why are inference patterns used to reason about travel also used to rea-
son about love relationships?

Correspondingly, from the perspective of the linguistic analyst, the existence of

such cross-domain pairings of words and of inference patterns provides evidence
for the existence of such mappings.

2.4. Novel extensions of conventional metaphors

The fact that the LOVE IS A JOURNEY¬mapping is a fixed part of our conceptual
system explains why new and imaginative uses of the mapping can be understood
instantly, given the ontological correspondences and other knowledge about jour-
neys. Take the song lyric, We‘re driving in the fast lane on the freeway of love.
The traveling knowledge called upon is this: when you drive in the fast lane, you
go a long way in a short time and it can be exciting and dangerous. The general
metaphorical mapping maps this knowledge about driving into knowledge about
love relationships. The danger may be to the vehicle (the relationship may not last)
or the passengers (the lovers may be hurt emotionally). The excitement of the love
journey is sexual. Our understanding of the song lyric is a consequence of the
preexisting metaphorical correspondences of the LOVE¬IS A¬JOURNEY¬metaphor.
194 George Lakoff

The song lyric is instantly comprehensible to speakers of English because those

metaphorical correspondences are already part of our conceptual system.
The LOVE¬IS A¬JOURNEY¬metaphor and Reddy’s Conduit Metaphor were the
two examples that first convinced me that metaphor was not a figure of speech,
but a mode of thought, defined by a systematic mapping from a source to a target
domain. What convinced me were the three characteristics of metaphor that I
have just discussed:

1. The systematicity in the linguistic correspondences.

2. The use of metaphor to govern reasoning and behavior based on that
reasoning.
3. The possibility for understanding novel extensions in terms of the conven-
tional correspondences.

2.5. Motivation
Each conventional metaphor, that is, each mapping, is a fixed pattern of conceptual
correspondence across conceptual domains. As such, each mapping defines an
open-ended class of potential correspondences across inference patterns. When
activated, a mapping may apply to a novel source domain knowledge structure
and characterize a corresponding target domain knowledge structure.
Mappings should not be thought of as processes, or as algorithms that mechani-
cally take source domain inputs and produce target domain outputs. Each mapping
should be seen instead as a fixed pattern of ontological correspondences across
domains that may, or may not, be applied to a source domain knowledge structure
or a source domain lexical item. Thus, lexical items that are conventional in the
source domain are not always conventional in the target domain. Instead, each
source domain lexical item may or may not make use of the static mapping pat-
tern. If it does, it has an extended lexicalized sense in the target domain, where
that sense is characterized by the mapping. If not, the source domain lexical item
will not have a conventional sense in the target domain, but may still be actively
mapped in the case of novel metaphor. Thus, the words freeway and fast lane are
not conventionally used of love, but the knowledge structures associated with them
are mapped by the LOVE¬IS A¬JOURNEY¬metaphor in the case of We‘re driving in
the fast lane on the freeway of love.

2.6. Imageable idioms

Many of the metaphorical expressions discussed in the literature on conventional
metaphor are idioms. On classical views, idioms have arbitrary meanings, but
within cognitive linguistics, the possibility exists that they are not arbitrary, but
 Chapter 6: Conceptual metaphor 195

rather motivated. That is, they do not arise automatically by productive rules, but
they fit one or more patterns present in the conceptual system. Let us look a little
more closely at idioms.
An idiom like spinning one’s wheels comes with a conventional mental
image, that of the wheels of a car stuck in some substance – mud, sand, snow,
or on ice – so that the car cannot move when the motor is engaged and the
wheels turn. Part of our knowledge about that image is that a lot of energy is
being used up (in spinning the wheels) without any progress being made, that
the situation will not readily change of its own accord, that it will take a lot of
effort on the part of the occupants to get the vehicle moving again – and that
may not even be possible.
The LOVE¬IS¬A¬JOURNEY¬metaphor applies to this knowledge about the image.
It maps this knowledge onto knowledge about love relationships: a lot of energy
is being spent without any progress toward fulfilling common goals, the situa-
tion will not change of its own accord, it will take a lot of effort on the part of the
lovers to make more progress, and so on. In short, when idioms have associated
conventional images, it is common for an independently motivated conceptual
metaphor to map that knowledge from the source to the target domain. For a sur-
vey of experiments verifying the existence of such images and such mappings,
see Gibbs (1990).

2.7. Mappings are at the superordinate level

In the LOVE IS A JOURNEY¬mapping, a love relationship corresponds to a vehicle.
A vehicle is a superordinate category that includes such basic level categories as
car, train, boat, and plane. The examples of vehicles are typically drawn from
this range of basic level categories: car (long bumpy road, spinning our wheels),
train (off the track), boat (on the rocks, foundering), plane (just taking off, bail-
ing out). This is not an accident: in general, we have found that mappings are at
the superordinate rather than the basic level. Thus, we do not find fully general
submappings like A¬LOVE¬RELATIONSHIP¬IS A¬CAR¬when we find a love relationship
conceptualized as a car, we also tend to find it conceptualized as a boat, a train,
a plane, and so forth. It is the superordinate category VEHICLE¬not the basic level
category CAR¬that is in the general mapping.
It should be no surprise that the generalization is at the superordinate level,
while the special cases are at the basic level. After all, the basic level is the
level of rich mental images and rich knowledge structure. (For a discussion of
the properties of basic level categories, see Lakoff 1987: 31–50.) A mapping at
the superordinate level maximizes the possibilities for mapping rich conceptual
structures in the source domain onto the target domain, since it permits many
basic level instances, each of which is information rich.
196 George Lakoff

Thus, a prediction is made about conventional mappings: the categories mapped

will tend to be at the superordinate rather than the basic level. One tends not to
find mappings like A¬LOVE¬RELATIONSHIP¬IS A¬CAR¬or A¬LOVE¬RELATIONSHIP¬IS A¬
BOAT¬Instead, one tends to find both basic level cases (e.g., both cars and boats),
which indicates that the generalization is one level higher, at the superordinate
level of the vehicle. In the hundreds of cases of conventional mappings studied
so far, this prediction has been borne out: it is superordinate categories that are
used in mappings.

3. Basic semantic concepts that are metaphorical

Most people are not too surprised to discover that emotional concepts like love
and anger are understood metaphorically. What is more interesting, and I think
more exciting, is the realization that many of the most basic concepts in our con-
ceptual systems are also normally comprehended via metaphor – concepts like
time, quantity, state, change, action, cause, purpose, means, modality, and even
the concept of a category. These are concepts that enter normally into the gram-
mars of languages, and if they are indeed metaphorical in nature, then metaphor
becomes central to grammar.
I would like to suggest that the same kinds of considerations that lead to our
acceptance of the LOVE¬IS¬A¬JOURNEY¬metaphor lead inevitably to the conclusion that
such basic concepts are often, and perhaps always, understood via metaphor.

3.1. Categories
Classical categories are understood metaphorically in terms of bounded regions,
or “containers.” Thus, something can be in or out of a category, it can be put into
a category or removed from a category. The logic of classical categories is the
logic of containers (see Figure 1).

Figure 1.
 Chapter 6: Conceptual metaphor 197

If X is in container A and container A is in container B, then X is in container B.

This is true not by virtue of any logical deduction, but by virtue of the topologi-
cal properties of containers. Under the CLASSICAL¬CATEGORIES¬ARE¬CONTAINERS¬
metaphor, the logical properties of categories are inherited from the logical prop-
erties of containers. One of the principal logical properties of classical categories
is that the classical syllogism holds for them. The classical syllogism,

Socrates is a man.
All men are mortal.
Therefore, Socrates is mortal.

is of the form:

If X is in category A and category A is in category B, then X is in cat-

egory B.

Thus, the logical properties of classical categories can be seen as following from
the topological properties of containers plus the metaphorical mapping from
containers to categories. As long as the topological properties of containers are
preserved by the mapping, this result will be true.
In other words, there is a generalization to be stated here. The language of
containers applies to classical categories and the logic of containers is true of
classical categories. A single metaphorical mapping ought to characterize both
the linguistic and logical generalizations at once. This can be done provided that
the topological properties of containers are preserved in the mapping.
The joint linguistic-and-inferential relation between containers and classical
categories is not an isolated case. Let us take another example.

3.2. Quantity and linear scales

The concept of quantities involves at least two metaphors. The first is the well-
known MORE¬IS UP¬LESS¬IS DOWN¬metaphor as shown by a myriad of expressions
like prices rose, stocks skyrocketed, the market plummeted, and so on. A second
is that LINEAR¬SCALES¬ARE¬PATHS¬We can see this in expressions like:

John is far more intelligent than Bill.

John’s intelligence goes way beyond Bill’s.
John is way ahead of Bill in intelligence.

The metaphor maps the starting point of the path onto the bottom of the scale
and maps distance traveled onto quantity in general.
198 George Lakoff

Don’t Have
This Much

Been Here Not Have

Been This
Here Much

Figure 2.

What is particularly interesting is that the logic of paths maps onto the logic of
linear scale (see Figure 2).

Path inference: if you are going from A to C, and you are now at an inter-
mediate point B, then you have been at all points between A and B and
not at any points between B and C.

Example: If you are going from San Francisco to New York along Route 80, and
you are now at Chicago, then you have been to Denver but not to Pittsburgh.

Linear scale inference: if you have exactly $50 in your bank account, then
you have $40, $30, and so on, but not $60, $70, or any larger amount.

The form of these inferences is the same. The path inference is a consequence of
the cognitive topology of paths. It will be true of any path image-schema. Again,
there is a linguistic-and-inferential generalization to be stated. It would be stated
by the metaphor LINEAR¬SCALES¬ARE¬PATHS¬provided that metaphors in general
preserve the cognitive topology (that is, the image-schematic structure) of the
source domain.
Looking at the inferential structure alone, one might suggest a non-metaphorical
alternative in which both linear scales and paths are instances of a more general
abstract schema. But when both the inferential and lexical data are considered, it
becomes clear that a metaphorical solution is required. An expression like ahead
of is from the spatial domain, not the linear scale domain: ahead in its core sense
is defined with respect to one’s head – it refers to the direction in which one is
facing. To say that there is no metaphorical mapping from paths to scales is to
say that ahead of is not fundamentally spatial and characterized with respect to
heads; it is to claim rather that ahead is very abstract, neutral between space and
linear scales, and has nothing to do with heads. This would be a bizarre analysis.
 Chapter 6: Conceptual metaphor 199

Similarly, for sentences like John’s intelligence goes beyond Bill’s, the nonmeta-
phorical analysis would claim that go is not fundamentally a verb of motion at
all, but is somehow neutral between motion and a linear relation. This would also
be bizarre. In short, if one grants that ahead of and go are fundamentally spatial,
then the fact that they can also be used of linear scales suggests a metaphor solu-
tion. There could be no such neutral sense of go for these cases, since go beyond
in the spatial sense involves motion, while in the linear scale sense, there is no
motion or change, but just a point on a scale. Here the neutral case solution is
not even available.

3.3. The Invariance Principle

In the examples we have just considered, the image-schemas characterizing the
source domains (containers, paths) are mapped onto the target domains (catego-
ries, linear scales). This observation leads to the following hypothesis, called
“The Invariance Principle”:

Metaphorical mappings preserve the cognitive topology (that is, the

image-schema structure) of the source domain, in a way consistent with
the inherent structure of the target domain.

What the Invariance Principle does is guarantee that, for container schemas,
interiors will be mapped onto interiors, exteriors onto exteriors, and boundaries
onto boundaries; for path-schemas, sources will be mapped onto sources, goals
onto goals, trajectories onto trajectories, and so on.
To understand the Invariance Principle properly, it is important not to think of
mappings as algorithmic processes that “start” with source domain structure and
wind up with target domain structure. Such a mistaken understanding of mappings
would lead to a mistaken understanding of the Invariance Principle, namely, that
one first picks all the image-schematic structure of the source domain, then one
copies it onto the target domain unless the target domain interferes.
One should instead think of the Invariance Principle in terms of constraints on
fixed correspondences: if one looks at the existing correspondences, one will see
that the Invariance Principle holds: source domain interiors correspond to target
domain interiors; source domain exteriors correspond to target domain exteriors,
and so forth. As a consequence it will turn out that the image-schematic structure
of the target domain cannot be violated: One cannot find cases where a source
domain interior is mapped onto a target domain exterior, or where a source domain
exterior is mapped onto a target domain path. This simply does not happen.
200 George Lakoff

3.4. Target domain overrides

A corollary of the Invariance Principle is that image-schema structure inherent
in the target domain cannot be violated, and that inherent target domain structure
limits the possibilities for mappings automatically. This general principle explains
a large number of previously mysterious limitations on metaphorical mappings.
For example, it explains why you can give someone a kick, even if that person
doesn’t have it afterward, and why you can give someone information, even if
you don’t lose it. This is a consequence of the fact that inherent target domain
structure automatically limits what can be mapped. For example, consider that
part of your inherent knowledge of actions that says that actions do not continue
to exist after they occur. Now consider the ACTIONS¬ARE¬TRANSFERS¬metaphor, in
which actions are conceptualized as objects transferred from an agent to a patient,
as when one gives someone a kick or a punch. We know (as part of target domain
knowledge) that an action does not exist after it occurs. In the source domain,
where there is a giving, the recipient possesses the object given after the giving.
But this cannot be mapped onto the target domain since the inherent structure
of the target domain says that no such object exists after the action is over. The
target domain override in the Invariance Principle explains why you can give
someone a kick without his having it afterward.

3.5. Abstract inferences as metaphorical spatial inferences

Spatial inferences are characterized by the topological structure of image schemas.
We have seen cases such as CATEGORIES¬ARE¬CONTAINERS¬and LINEAR¬SCALES¬ARE¬
PATHS¬where image-schema structure is preserved by metaphor and where abstract
inferences about categories and linear scales are metaphorical versions of spatial
inferences about containers and paths. The Invariance Principle hypothesizes that
image-schema structure is always preserved by metaphor.
The Invariance Principle raises the possibility that a great many, if not all,
abstract inferences are actually metaphorical versions of spatial inferences that are
inherent in the topological structure of image-schemas. I will now turn to other
cases of basic, but abstract, concepts to see what evidence there is for the claim
that such concepts are fundamentally characterized by metaphor.

3.6. Time
It has often been noted that time in English is conceptualized in terms of space.
The details are rather interesting.
 Chapter 6: Conceptual metaphor 201

Ontology: Time is understood in terms of things (that is, entities and loca-
tions) and motion.
Background condition: The present time is at the same location as a
canonical observer.
Mapping:
Times are things.
The passing of time is motion.
Future times are in front of the observer; past times are behind the
observer.
One thing is moving, the other is stationary; the stationary entity is
the deictic center.
Entailment:
Since motion is continuous and one-dimensional, the passage of time
is continuous and one-dimensional.
Special case 1:
The observer is fixed; times are entities moving with respect to the
observer.
Times are oriented with their fronts in their direction of motion.

Entailments:
If time 2 follows time 1, then time 2 is in the future relative to time 1.
The time passing the observer is the present time.
Time has a velocity relative to the observer.
Special case 2:
Times are fixed locations; the observer is moving with respect to
time.
Entailment:
Time has extension, and can be measured.
An extended time, like a spatial area, may be conceived of as a
bounded region.

This metaphor, TIME¬PASSING¬IS MOTION¬with its two special cases, embodies
a generalization that accounts for a wide range of cases where a spatial expres-
sion can also be used for time. Special case 1, TIME¬PASSING¬IS MOTION¬OF¬AN¬
OBJECT¬accounts for both the linguistic form and the semantic entailments of
expressions like:

The time will come when . . . The time has long since gone when . . . The
time for action has arrived. That time is here. In the weeks following next
Tuesday . . . On the preceding day . . . I’m looking ahead to Christmas.
202 George Lakoff

Thanksgiving is coming up on us. Let’s put all that behind us. I can’t face
the future. Time is flying by. The time has passed when . . .
Thus, special case 1 characterizes the general principle behind the temporal use
of words like come, go, here, follow, precede, ahead, behind, fly, pass, account-
ing not only for why they are used for both space and time, but why they mean
what they mean.
Special case 2, TIME¬PASSING¬IS MOTION¬OVER¬A¬LANDSCAPE¬accounts for a
different range of cases, expressions like:

There’s going to be trouble down the road. He stayed there for ten years.
He stayed there a long time. His stay in Russia extended over many years.
He passed the time happily. He arrived on time. We‘re coming up on
Christmas. We‘re getting close to Christmas. He‘ll have his degree within
two years. I‘ll be there in a minute.

Special case 2 maps location expressions like down the road, for + location,
long, over, come, close to, within, in, pass, onto corresponding temporal expres-
sions with their corresponding meanings. Again, special case 2 states a general
principle relating spatial terms and inference patterns to temporal terms and
inference patterns.
The details of the two special cases are rather different; indeed, they are incon-
sistent with one another. The existence of such special cases has an especially
interesting theoretical consequence: words mapped by both special cases will have
inconsistent readings. Take, for example, the come of Christmas is coming (special
case 1) and We‘re coming up on Christmas (special case 2). Both instances of
come are temporal, but one takes a moving time as first argument and the other
takes a moving observer as first argument. The same is true of pass in The time
has passed (special case 1) and in He passed the time (special case 2).
These differences in the details of the mappings show that one cannot just say
blithely that spatial expressions can be used to speak of time, without specifying
details, as though there were only one correspondence between time and space.
When we are explicit about stating the mappings, we discover that there are two
different – and inconsistent – subcases.
The fact that time is understood metaphorically in terms of motion, entities,
and locations accords with our biological knowledge. In our visual systems, we
have detectors for motion and detectors for objects/locations. We do not have
detectors for time (whatever that could mean). Thus, it makes good biological
sense that time should be understood in terms of things and motion.
 Chapter 6: Conceptual metaphor 203

3.7. Duality
The two special cases (location and object) of the TIME¬PASSING¬IS MOTION¬meta-
phor are not merely an accidental feature of our understanding of time. As we
shall see below, there are other metaphors that come in such location/ object pairs.
Such pairs are called “duals,” and the general phenomenon in which metaphors
come in location/object pairs is referred to as “duality.”

3.8. Simultaneous mappings

It is important to recall that metaphorical mappings are fixed correspondences
that can be activated, rather than algorithmic processes that take inputs and give
outputs. Thus, it is not the case that sentences containing conventional metaphors
are the products of a real-time process of conversion from literal to metaphorical
readings. A sentence like The time for action has arrived is not understood by
first trying to give a literal reading to arrive, and then, on failing, trying to give
it a temporal reading. Instead, the metaphor TIME¬PASSING¬IS MOTION¬is a fixed
structure of existing correspondences between the space and time domains, and
arrive has a conventional extended meaning that makes use of that fixed structure
of correspondences.
Thus, it is possible for two different parts of a sentence to make use of two
distinct metaphorical mappings at once. Consider a phrase like, within the com-
ing weeks. Here, within makes use of the metaphor of time as a stationary land-
scape which has extension and bounded regions, whereas coming makes use of
the metaphor of times as moving objects. This is possible because the two meta-
phors for time pick out different aspects of the target domain. The coming weeks
conceptualizes those weeks as a whole, in motion relative to the observer. Within
looks inside that whole, conceptualizing it as a bounded region with an interior.
Each mapping is used partially. Thus, although the mappings – as wholes – are
inconsistent, there are cases where parts of the mappings may be consistently
superimposed. The Invariance Principle allows such parts of the mappings to
be picked out and used to characterize reasoning about different aspects of the
target domain.
Simultaneous mappings are very common in poetry. Take, for example, the
Dylan Thomas line “Do not go gentle into that good night.” Here go reflects
DEATH¬IS DEPARTURE¬gentle reflects LIFE¬IS A¬STRUGGLE¬with death as defeat.
Night reflects A¬LIFETIME¬IS A¬DAY¬with death as night. This one line has three
different metaphors for death, each mapped onto different parts of the sentence.
This is possible since mappings are fixed correspondences.
There is an important lesson to be learned from this example. In mathemat-
ics, mappings are static correspondences. In computer science, it is common to
204 George Lakoff

represent mathematical mappings by algorithmic processes that take place in real

time. Researchers in information processing psychology and cognitive science
also commonly represent mappings as real-time algorithmic procedures. Some
researchers from these fields have mistakenly supposed that the metaphorical
mappings we are discussing should also be represented as real-time, sequential
algorithmic procedures, where the input to each metaphor is a literal meaning. Any
attempt to do this will fail for the simultaneous mapping cases just discussed.

4. Event structure

I now want to turn to some research by myself and some of my students (espe-
cially Sharon Fischler, Karin Myhre, and Jane Espenson) on the metaphorical
understanding of event structure in English. What we have found is that vari-
ous aspects of event structure, including notions like states, changes, processes,
actions, causes, purposes, and means, are characterized cognitively via metaphor
in terms of space, motion, and force.
The general mapping we have found goes as follows:

The event structure metaphor

− States are locations (bounded regions in space).
− Changes are movements (into or out of bounded regions).
− Causes are forces.
− Actions are self-propelled movements.
− Purposes are destinations.
− Means are paths (to destinations).
− Difficulties are impediments to motion.
− Expected progress is a travel schedule; a schedule is a virtual traveler, who
reaches prearranged destinations at prearranged times.
− External events are large, moving objects.
– Long term, purposeful activities are journeys.

This mapping generalizes over an extremely wide range of expressions for one or
more aspects of event structure. For example, take states and changes. We speak
of being in or out of a state, of going into or out of it, of entering or leaving it, of
getting to a state or emerging from it.
This is a rich and complex metaphor whose parts interact in complex ways.
To get an idea of how it works, consider the submapping “Difficulties are impedi-
ments to motion.” In the metaphor, purposive action is self-propelled motion
toward a destination. A difficulty is something that impedes motion to such a
destination. Metaphorical difficulties of this sort come in five types: blockages;
 Chapter 6: Conceptual metaphor 205

features of the terrain; burdens; counter-forces; lack of an energy source. Here

are examples of each:

Blockages:
He got over his divorce. He’s trying to get around the regulations.
He went through the trial. We ran into a brick wall. We’ve got him
boxed into a corner.
Features of the terrain:
He’s between a rock and a hard place. It’s been uphill all the way.
We’ve been bogged down. We’ve been hacking our way through a
jungle of regulations.
Burdens:
He’s carrying quite a load. He’s weighed down by a lot of assignments.
He’s been trying to shoulder all the responsibility. Get off my back!
Counterforces:
Quit pushing me around. She’s leading him around by the nose. She’s
holding him back.
Lack of an energy source:
I’m out of gas. We‘re running out of steam.

To see just how rich the event structure metaphor is, consider some of its basic
entailments:

− Manner of action is manner of motion.

− A different means for achieving a purpose is a different path.
− Forces affecting action are forces affecting motion.
− The inability to act is the inability to move.
– Progress made is distance traveled or distance from goal.

We will consider examples of each of these one by one, including a number of

special cases.

Aids to action are aids to motion:

It is smooth sailing from here on in. It’s all downhill from here.
There’s nothing in our way.
A different means of achieving a result is a different path:
Do it this way. She did it the other way. Do it any way you can. How-
ever you want to go about it is fine with me.
Manner of action is manner of motion:
We are moving/running/skipping right along. We slogged through it.
He is flailing around. He is falling all over himself. We are leaping
over hurdles. He is out of step. He is in step.
206 George Lakoff

Careful action is careful motion:

I’m walking on eggshells. He is treading on thin ice. He is walking a
fine line.
Speed of action is speed of movement:
He flew through his work. He is running around. It is going swim-
mingly. Keep things moving at a good clip. Things have slowed to
a crawl. She is going by leaps and bounds. I am moving at a snail’s
pace.
Purposeful action is self-propelled motion to a destination; this has the
following special cases:
Making progress is forward movement:
We are moving ahead. Let’s forge ahead. Let’s keep moving
forward. We made lots of forward movement.
Amount of progress is distance moved:
We’ve come a long way. We’ve covered lots of ground. We’ve
made it this far.
Undoing progress is backward movement:
We are sliding backward. We are backsliding. We need to back-
track. It is time to turn around and retrace our steps.
Expected progress is a travel schedule; a schedule is a virtual trave-
ler, who reaches prearranged destinations at prearranged times:
We‘re behind schedule on the project. We got a head start on
the project. I’m trying to catch up. I finally got a little ahead.
Starting an action is starting out on a path:
We are just starting out. We have taken the first step.
Success is reaching the end of the path:
We’ve reached the end. We are seeing the light at the end of the
tunnel. We only have a short way to go. The end is in sight. The
end is a long way off.
Lack of purpose is lack of direction:
He is just floating around. He is drifting aimlessly. He needs
some direction.
Lack of progress is lack of movement:
We are at a standstill. We aren’t getting any place. We aren’t
going anywhere. We are going nowhere with this.
External events are large moving objects:
Special case 1: Things
How‘re things going? Things are going fine with me. Things
are going against me these days. Things took a turn for the
worse. Things are going my way.
Special case 2: Fluids
 Chapter 6: Conceptual metaphor 207

You gotta go with the flow. I’m just trying to keep my head
above water. The tide of events . . . The winds of change. . . .
The flow of history . . . I’m trying to get my bearings. He’s up a
creek without a paddle. We‘re all in the same boat.
Special case 3: Horses
Try to keep a tight rein on the situation. Keep a grip on the situ-
ation. Don’t let things get out of hand. Wild horses couldn’t
make me go. “Whoa!” (said when things start to get out of
hand).
Such examples provide overwhelming empirical support for the existence of the
event structure metaphor. And the existence of that metaphor shows that the most
common abstract concepts – TIME¬STATE¬CHANGE¬CAUSATION¬ACTION¬PURPOSE¬
and MEANS¬– are conceptualized via metaphor. Since such concepts are at the very
center of our conceptual systems, the fact that they are conceptualized metaphori-
cally shows that metaphor is central to ordinary abstract thought.

4.1. Inheritance hierarchies

Metaphorical mappings do not occur isolated from one another. They are some-
times organized in hierarchical structures, in which “lower” mappings in the
hierarchy inherit the structures of the “higher” mappings. Let us consider an
example of a hierarchy with three levels:

Level 1: The event structure metaphor

Level 2: A¬PURPOSEFUL¬LIFE¬IS A¬JOURNEY
Level 3: LOVE¬IS A¬JOURNEY¬A¬CAREER¬IS A¬JOURNEY

To refresh your memory, recall:

The event structure metaphor

− Target domain: Events Source domain: Space
− States are locations (bounded regions in space).
− Changes are movements (into or out of bounded regions).
− Causes are forces.
− Actions are self-propelled movements.
− Purposes are destinations.
− Means are paths to destinations.
− Difficulties are impediments to motion.
− Expected progress is a travel schedule; a schedule is a virtual traveler, who
reaches prearranged destinations at prearranged times.
− External events are large, moving objects.
208 George Lakoff

– Long-term, purposeful activities are journeys.

In our culture, life is assumed to be purposeful, that is, we are expected to have
goals in life. In the event structure metaphor, purposes are destinations and pur-
poseful action is self-propelled motion toward a destination. A purposeful life
is a long-term, purposeful activity, and hence a journey. Goals in life are des-
tinations on the journey. The actions one takes in life are self-propelled move-
ments, and the totality of one’s actions form a path one moves along. Choosing
a means to achieve a goal is choosing a path to a destination. Difficulties in life
are impediments to motion. External events are large moving objects that can
impede motion toward one’s life goals. One’s expected progress through life is
charted in terms of a life schedule, which is conceptualized as a virtual traveler
that one is expected to keep up with.
In short, the metaphor A¬PURPOSEFUL¬LIFE¬IS A¬JOURNEY¬makes use of all the
structure of the event structure metaphor, since events in a life conceptualized as
purposeful are subcases of events in general.

A¬PURPOSEFUL¬LIFE¬IS¬A¬JOURNEY
− Target domain: Life Source domain: Space
− The person leading a life is a traveler.
– Inherits event structure metaphor, with:
Events = significant life events
Purposes = life goals

Thus we have expressions like:

He got a head start in life. He’s without direction in his life. I’m where I
want to be in life. I’m at a crossroads in my life. He‘ll go places in life.
He’s never let anyone get in his way. He’s gone through a lot in life.

Just as significant life events are special cases of events, so events in a love rela-
tionship are special cases of life events. Thus, the LOVE¬IS A¬JOURNEY¬metaphor
inherits the structure of the LIFE¬IS A¬JOURNEY¬metaphor. What is special about
the LOVE¬IS¬A¬JOURNEY¬metaphor is that there are two lovers who are travelers and
that the love relationship is a vehicle. The rest of the mapping is a consequence
of inheriting the LIFE¬IS A¬JOURNEY¬metaphor. Because the lovers are in the same
vehicle, they have common destinations, that is, common life goals. Relationship
difficulties are impediments to travel.

LOVE¬IS¬A¬JOURNEY
− Target domain: Love Source domain: Space
− The lovers are travelers.
 Chapter 6: Conceptual metaphor 209

− The love relationship is a vehicle.

– Inherits the LIFE¬IS A¬JOURNEY¬metaphor.

A career is another aspect of life that can be conceptualized as a journey. Here,

because STATUS¬IS UP¬a career is actually a journey upward. Career goals are
special cases of life goals.

A¬CAREER¬IS¬A¬JOURNEY
− Target domain: Career Source domain: Space
− A careerist is a traveler.
− Status is up.
– Inherits LIFE¬IS A¬JOURNEY¬with life goals = career goals. Ideal: to go as
high, far, and fast as possible.

Examples include:

He clawed his way to the top. He’s over the hill. She’s on the fast track.
He’s climbing the corporate ladder. She’s moving up in the ranks quickly.

This inheritance hierarchy accounts for a range of generalizations. First, there are
generalizations about lexical items. Take the word crossroads. Its central meaning
is in the domain of space, but it can be used in a metaphorical sense to speak of
any extended activity, of one’s life, of a love relationship, or of a career.

I’m at a crossroads on this project. I’m at a crossroads in life. We’re at a

crossroads in our relationship. I’m at a crossroads in my career.

The hierarchy allows one to state a general principle: crossroads is extended

lexically via the submetaphor of the event structure metaphor that LONG TERM¬
PURPOSEFUL¬ACTIVITIES¬ARE¬JOURNEYS¬All its other uses are automatically gen-
erated via the inheritance hierarchy. Thus, separate senses for each level of the
hierarchy are not needed.
The second generalization is inferential in character. Thus the understanding
of difficulties as impediments to travel occurs not only in events in general, but
also in a purposeful life, in a love relationship, and in a career. The inheritance
hierarchy guarantees that this understanding of difficulties in life, love, and careers
is a consequence of such an understanding of difficulties in events in general.
The hierarchy also allows us to characterize lexical items whose meanings
are more restricted: Thus, climbing the ladder refers only to careers, not to love
relationships or to life in general.
Such hierarchical organization is a very prominent feature of the metaphor
system of English and other languages. So far we have found that the metaphors
higher up in the hierarchy tend to be more widespread than those mappings at
210 George Lakoff

lower levels. Thus, the event structure metaphor is very widespread (and may
even be universal), while the metaphors for life, love, and careers are much more
restricted culturally.

4.2. Duality in the event structure system

In our discussion of time metaphors, we noted the existence of an object/loca-
tion duality. There were two related time metaphors. In both, the passage of time
was understood in terms of relative motion between an observer and a time. In
the object-dual, the observer is fixed and times are moving objects. In the loca-
tion-dual, the opposite is true. The observer moves and times are fixed locations
in a landscape.
The event structure system that we have seen so far is based wholly on loca-
tion. But there is another event structure system that is the dual of the one we
have just discussed – a system based on objects rather than locations. In both
systems, CHANGE¬IS MOTION¬and CAUSES¬ARE¬FORCES¬that control motion. The
difference is this:

− In the location system, change is the motion of the thing-changing to a new

location or from an old one.
– In the object system, the thing-changing doesn’t necessarily move. Change
is instead the motion of an object to, or away from, the thing-changing.

In addition, the object in motion is conceptualized as a possession and the thing-

changing as a possessor. Change is thus seen as the acquisition or loss of an object.
Causation is seen as giving or taking. Here are some examples:

I have a headache. (The headache is a possession)

I got a headache. (Change is acquisition – motion to)
My headache went away. (Change is loss – motion from)
The noise gave me a headache. (Causation is giving – motion to)
The aspirin took away my headache. (Causation is taking – motion from)

We can see the duality somewhat more clearly with a word like “trouble”:

I’m in trouble. (Trouble is a location)

I have trouble. (Trouble is an object that is possessed)

In both cases, trouble is being attributed to me, and in both cases, trouble is meta-
phorically conceptualized as being in the same place as me (co-location) – in
one case, because I possess the trouble-object and in the other case, because I
am in the trouble-location. That is, attribution in both cases is conceptualized
 Chapter 6: Conceptual metaphor 211

metaphorically as co-location. In I’m in trouble, trouble is a state. A state is an

attribute conceptualized as a location. Attributes (or properties) are like states,
except that they are conceptualized as possessible objects.
Thus, STATES¬ARE¬LOCATIONS¬and ATTRIBUTES¬ARE¬POSSESSIONS¬are duals, since
possession and location are special cases of the same thing – co-location – and
since states and attributes are also special cases of the same thing – what can be
attributed to someone.
Given this, we can see that there is an object-version of the event structure
metaphor:

− Attributes are possessions.

− Changes are movements (of possessions, namely, acquisitions or losses).
– Causes are forces (controlling the movement of possessions, namely, giving
or taking away).

These are the duals of:

− States are locations.

− Changes are movements (to or from locations).
− Causes are forces (controlling movement to or from locations).

Similarly, ACTIONS¬ARE¬SELF PROPELLED¬MOVEMENTS¬(to or from locations) has

as its object-dual ACTIONS¬ARE¬SELF CONTROLLED¬ACQUISITIONS¬OR¬LOSSES¬Thus,
there is a reason why one can “take” certain actions – one can take a shower, or
take a shot at someone, or take a chance.
The submapping PURPOSES¬ARE¬DESTINATIONS¬also has a dual. Destinations
are desired locations and so the submapping can be rephrased as PURPOSES¬ARE¬
DESIRED¬LOCATIONS, and ACHIEVING¬A¬PURPOSE¬IS¬REACHING¬A¬DESIRED LOCA
TION¬Replacing “location” by “object,” we get the dual PURPOSES¬ARE¬DESIRED¬
OBJECTS, and ACHIEVING¬A¬PURPOSE¬IS¬ACQUIRING¬A¬DESIRED¬OBJECT (or ridding
oneself of an undesirable one).
Here are some examples:

ACHIEVING¬A¬PURPOSE¬IS¬ACQUIRING¬A¬DESIRED¬OBJECT
They just handed him the job. It’s within my grasp. It eluded me. Go for
it. It escaped me. It slipped through my hands. He is pursuing a goal.
Reach for/grab all the gusto you can get. Latch onto a good job. Seize the
opportunity. He found success.

There is also a hierarchical structure in the object-version of the event structure

metaphor. A special case of getting an object is getting an object to eat. Hence:
212 George Lakoff

ACHIEVING¬A¬PURPOSE¬IS¬GETTING¬SOMETHING¬TO¬EAT
He savored the victory. All the good jobs have been gobbled up. He’s hun-
gry for success. The opportunity has me drooling. This is a mouth-water-
ing opportunity.

Traditional methods of getting things to eat are hunting, fishing, and agriculture.
Each of these special cases can be used metaphorically to conceptualize achiev-
ing (or attempting to achieve) a purpose.

TRYING¬TO¬ACHIEVE¬A¬PURPOSE¬IS¬HUNTING
I’m hunting for a job. I bagged a promotion. The pennant is in the bag.

The typical way to hunt is to use projectiles (bullets, arrows, etc.)

I’m shooting for a promotion. I’m aiming for a career in the movies. I’m
afraid I missed my chance.

TRYING¬TO¬ACHIEVE¬A¬PURPOSE¬IS¬lSHING
He’s fishing for compliments. I landed a promotion. She netted a good
job. I’ve got a line out on a good used car. It’s time to fish or cut bait.

TRYING¬TO¬ACHIEVE¬A¬PURPOSE¬IS¬AGRICULTURE
It’s time I reaped some rewards. That job is a plum. Those are the fruits of
his labor. The contract is ripe for the picking.

I will not try to survey all the dualities in the English metaphor system, but it is
worth mentioning a few to see how subtle and persuasive dualities are. Take, for
example, the LIFE¬IS¬A¬JOURNEY¬metaphor, in which goals in life are destinations,
that is, desired locations to be reached. Since the dual of PURPOSES¬ARE¬DESTI
NATIONS is PURPOSES¬ARE¬DESIRED¬OBJECTS, the dual of LIFE¬IS A¬JOURNEY¬is a
metaphor in which life is an activity through which one acquires desired objects.
In this culture, the principal activity of this sort is business, and hence, LIFE¬IS A¬
BUSINESS¬is the dual of LIFE¬IS A¬JOURNEY.

A¬PURPOSEFUL¬LIFE¬IS¬A¬BUSINESS
He has a rich life. It’s an enriching experience. I want to get a lot out of
life. He’s going about the business of everyday life. It’s time to take stock
of my life.

Recall that LOVE¬IS A¬JOURNEY¬is an extension of A¬PURPOSEFUL¬LIFE¬IS¬A¬JOUR

NEY¬It happens that LOVE¬IS A¬JOURNEY¬has a dual that is an extension of the dual
of A¬PURPOSEFUL¬LIFE¬IS A¬JOURNEY¬which is A¬PURPOSEFUL¬LIFE¬IS A¬BUSINESS¬
The dual of LOVE¬IS A¬JOURNEY¬is LOVE¬IS A¬PARTNERSHIP¬that is, a two-person
 Chapter 6: Conceptual metaphor 213

business. Thus, we speak of lovers as “partners,” there are marriage contracts,

and in a long-term love relationship the partners are expected to do their jobs
and to share in both responsibilities (what they contribute to the relationship) and
benefits (what they get out of it). Long-term love relationships fail under the same
conditions as businesses fail – when what the partners get out of the relationship
is not worth what they put into it.
Duality is a newly discovered phenomenon. The person who first discovered it
in the event structure system was Jane Espenson, a graduate student at Berkeley
who stumbled upon it in the course of her research on causation metaphors. Since
Espenson’s discovery, other extensive dualities have been found in the English
metaphor system. It is not known at present, however, just how extensive dualities
are in English, or even whether they are all of the location/object type.
At this point, I will leave off discussing the metaphor system of English,
although hundreds of other mappings have been described to date. The major
point to take away from this discussion is that metaphor resides for the most
part in this huge, highly structured, fixed system, a system anything but “dead.”
Because it is conventional, it is used constantly and automatically, with neither
effort nor awareness. Novel metaphor uses this system, and builds on it, but
only rarely occurs independently of it. It is most interesting that this system of
metaphor seems to give rise to abstract reasoning, which appears to be based on
spatial reasoning.

4.3. Invariance again

The metaphors I have discussed primarily map three kinds of image schemas:
containers, paths, and force-images. Because of the complexity of the subcases
and interactions, the details are intricate, to say the least. However, the Invariance
Principle does make claims in each case as to what image-schemas get mapped
onto target domains. I will not go through most of the details here, but so far as I
can see, the claims made about inferential structure are reasonable ones.
For example, the logic of force-dynamics does seem to map, via the submap-
ping CAUSES¬ARE¬FORCES¬onto the logic of causation. The following are inferences
from the logic of forces inherent in force dynamics:

− A stationary object will move only when force is applied to it; without
force, it will not move.
− The application of force requires contact; thus, the applier of the force must
be in spatial contiguity with the thing it moves.
– The application of force temporarily precedes motion, since inertia must be
overcome before motion can take place.
214 George Lakoff

These are among the classic inferential conditions on causation: spatial contigu-
ity, temporal precedence, and that A caused B only if B wouldn’t have happened
without A.
At this point, I would like to take up the question of what else the Invariance
Principle would buy us. I will consider two cases that arose while Mark Turner
and I were writing More Than Cool Reason (Lakoff and Turner 1989). The
first concerns image-metaphors and the second, generic-level metaphors. But
before I move on to those topics, I should mention an important consequence of
invariance.
Johnson and I argued in Metaphors We Live By (Lakoff and Johnson 1980)
that a complex propositional structure could be mapped by metaphor onto another
domain. The main example we gave was ARGUMENT¬IS WAR¬Kövecses and I, in
our analysis of anger metaphors (Lakoff 1987: case study 1; Kövecses 1990),
also argued that metaphors could map complex propositional structures. The
Invariance Principle does not deny this, but it puts those claims in a very dif-
ferent light. Complex propositional structures involve concepts like time, states,
changes, causes, purposes, quantity scales, and categories. If all these abstract
concepts are characterized metaphorically, then the Invariance Principle claims
that what we had called propositional structure is really image-schematic struc-
ture. In other words:

− So-called propositional inferences arise from the inherent topological

structure of the image-schemas mapped by metaphor onto concepts like
time, states, changes, actions, causes, purposes, means, quantity, and
categories.

I have taken the trouble to discuss these abstract concepts to demonstrate this
consequence of the Invariance Principle: what have been seen in the past as propo-
sitional inferences are really image-based inferences. If the Invariance Principle
is correct, it has a remarkable consequence:

− Abstract reasoning is a special case of image-based reasoning.

Image-based reasoning is fundamental and abstract reasoning is image-based

reasoning under metaphorical projections to abstract domains.
To look for independent confirmation of the Invariance Principle, let us turn
to image metaphors.
 Chapter 6: Conceptual metaphor 215

5. Novel metaphors

5.1. Image metaphors

There are kinds of metaphors that function to map one conventional mental image
onto another. These contrast with the metaphors I have discussed so far, each
of which maps one conceptual domain onto another, often with many concepts
in the source domain mapped onto many corresponding concepts in the target
domain. Image metaphors, by contrast, are “one-shot” metaphors: they map only
one image onto one other image.
Consider, for example, this poem from the Indian tradition:

Now women-rivers
belted with silver fish
move unhurried as women in love
at dawn after a night with their lovers
(Merwin and Masson 1981: 71)

Here the image of the slow, sinuous walk of an Indian woman is mapped onto
the image of the slow, sinuous, shimmering flow of a river. The shimmering of a
school of fish is imagined as the shimmering of the belt.
Metaphoric image mappings work in the same way as all other metaphoric
mappings: by mapping the structure of one domain onto the structure of another.
But here, the domains are conventional mental images. Take, for example, this
line from Andre Breton:

My wife . . . whose waist is an hourglass.

This is a superimposition of the image of an hourglass onto the image of a woman’s

waist by virtue of their common shape. As before, the metaphor is conceptual; it
is not in the words themselves, but in the mental images. Here, we have a mental
image of an hourglass and of a woman, and we map the middle of the hourglass
onto the waist of the woman. Note that the words do not tell us which part of the
hourglass to map onto the waist, or even that only part of the hourglass shape
corresponds to the waist. The words are prompts for us to map from one conven-
tional image to another. Similarly, consider:

His toes were like the keyboard of a spinet.

(Rabelais The Descriptions of King Lent, trans. J. M. Cohen)

Here, too, the words do not tell us that an individual toe corresponds to an indi-
vidual key on the keyboard. The words are prompts for us to perform a conceptual
216 George Lakoff

mapping between conventional mental images. In particular, we map aspects of

the part-whole structure of one image onto aspects of the part-whole structure
of another. Just as individual keys are parts of the whole keyboard, so individual
toes are parts of the whole foot.
Image mapping can involve more than mapping physical part-whole relation-
ships. For example, the water line of a river may drop slowly and that slowness
is part of a dynamic image, which may be mapped onto the slow removal of
clothing:

Slowly slowly rivers in autumn show

sand banks
bashful in first love woman
showing thighs
(Merwin and Masson 1981: 69)

Other attributes are also mapped: the color of the sand bank onto the color of
flesh, the quality of light on a wet sand bank onto the reflectiveness of skin, the
light grazing of the water’s touch receding down the bank onto the light graz-
ing of the clothing along the skin. Notice that the words do not tell us that any
clothing is involved. We get that from a conventional mental image. Part-whole
structure is also mapped in this example. The water covers the hidden part of the
bank just as the clothing covers the hidden part of the body. The proliferation of
detail in the images limits image mappings to highly specific cases. That is what
makes them one-shot mappings.
Such mappings of one image onto another can lead us to map knowledge
about the first image onto knowledge about the second. Consider the following
example from the Navaho:

My horse with a mane made of short rainbows.

(War God’s Horse Song I, words by Tall Kia ahni, interpreted by Louis
Watchman)

The structure of a rainbow, its band of curved lines for example, is mapped onto
an arc of curved hair, and many rainbows onto many such arcs on the horse’s
mane. Such image mapping allows us to map our evaluation of the source domain
onto the target. We know that rainbows are beautiful, special, inspiring, larger
than life, almost mystic, and that seeing them makes us happy and inspires us
with awe. This knowledge is mapped onto what we know of the horse: it too is
awe-inspiring, beautiful, larger than life, almost mystic. This line comes from a
poem containing a series of such image mappings:
 Chapter 6: Conceptual metaphor 217

My horse with a hoof like a striped agate,

with his fetlock like a fine eagle plume:
my horse whose legs are like quick lightning
whose body is an eagle-plumed arrow:
my horse whose tail is like a trailing black cloud.

Image metaphors raise two major issues for the general theory of metaphor:

− How do they work? What constrains the mappings? What kinds of internal
structures do mental images have that permit some mappings to work read-
ily, others only with effort, and others not at all?
– What is the general theory of metaphor that unifies image metaphors with
all the conventional metaphors that map the propositional structure of one
domain onto the propositional structure of another domain?

Turner and I (Lakoff and Turner 1989) have suggested that the Invariance Prin-
ciple could be an answer to both questions. We suggest that conventional mental
images are structured by image-schemas and that image metaphors preserve
image-schematic structure, mapping parts onto parts and wholes onto wholes,
containers onto containers, paths onto paths, and so on. The generalization would
be that all metaphors are invariant with respect to their cognitive topology, that
is, each metaphorical mapping preserves image-schema structure.

5.2. Generic-level metaphors

When Turner and I were writing More Than Cool Reason, we hypothesized the
existence of what we called “generic-level metaphors” to deal with two problems
we faced – first, the problem of personification and second, the problem of prov-
erbs, which requires an understanding of analogy. I shall discuss each in turn.

5.2.1. Personification
In studying a wide variety of poems about death in English, we found that, in
poem after poem, death was personified in a relatively small number of ways:
drivers, coachmen, footmen; reapers, devourers and destroyers, or opponents in
a struggle or game (say, a knight or a chess opponent). The question we asked
was: why these? Why isn’t death personified as a teacher or a carpenter or an
ice cream salesman? Somehow, the ones that occur repeatedly seem appropri-
ate. Why?
In studying personifications in general, we found that the overwhelming num-
ber seem to fit a single pattern: events (like death) are understood in terms of
218 George Lakoff

actions by some agent (like reaping). It is that agent that is personified. We thus
hypothesized a very general metaphor, EVENTS¬ARE¬ACTIONS¬which combines with
other, independently existing metaphors for life and death. Consider, for example,
the DEATH¬IS DEPARTURE¬metaphor. Departure is an event. If we understand this
event as an action on the part of some causal agent – someone who brings about,
or helps to bring about, departure – then we can account for figures like drivers,
coachmen, footmen, and so forth. Take the PEOPLE¬ARE¬PLANTS¬metaphor. In the
natural course of things, plants wither and die. If we see that event as a causal
action on the part of some agent, that agent is a reaper. So far, so good. But why
destroyers and devourers? And what about the impossible cases?
Destroying and devouring are actions in which an entity ceases to exist. The
same is true of death. The overall shape of the event of death is similar in this
respect to the overall shapes of the events of destroying and devouring. More-
over, there is a causal aspect to death: the passage of time will eventually result
in death. Thus, the overall shape of the event of death has an entity that over time
ceases to exist as the result of some cause. Devouring and destroying have the
same overall event shape. That is, it is the same with respect to causal structure
and the persistence of entities over time.
Turner (1987) had noticed a similar case in Death Is the Mother of Beauty,
his classic work on kinship metaphor. In expressions like necessity is the mother
of invention, or Edward Teller was the father of the H-bomb, causation is under-
stood in terms of giving birth or fathering, what Turner called the CAUSATION¬
IS PROGENERATION¬metaphor. But, as he observed (Turner 1987: 145–148), this
metaphor could not be used for just any instance of causation. It could only be
used for cases that had the overall event shape of progeneration: something must
be created out of nothing, and the thing created must persist for a long time (as
if it had a life).
Thus, for example, we can speak of Saussure as the father of modern syn-
chronic linguistics, or of New Orleans as giving birth to jazz. But we cannot use
this metaphor for a single causal action with a short-lived effect. We could not
speak of Jose Canseco as the father of the home run he just hit, or of that home
run as giving birth to the Oakland As’ victory in the game. We could, however,
speak of Babe Ruth as the father of modern home-run hitting, and of home runs
giving birth to the era of baseball players as superstars. The overall event shape
of the target domain limits the applicability of the metaphor.
Recalling Turner’s observation about CAUSATION¬IS PROGENERATION¬we there-
fore hypothesized that EVENTS¬ARE¬ACTIONS¬is constrained in the following way:
the action must have the same overall event shape as the event. What is preserved
across the mapping is the causal structure, the aspectual structure, and the per-
sistence of entities. We referred to this as “generic-level structure.”
 Chapter 6: Conceptual metaphor 219

The preservation of generic-level structure explained why death is not meta-

phorized in terms of teaching, or filling the bathtub, or sitting on the sofa. These
actions do not have the same causal and overall event structure, they do not share
“generic-level structure.”

5.2.2. Proverbs
In Asian figures – proverbs in the form of short poems – the question arises as to
what the limitations are on the interpretation of a proverb. Some interpretations
are natural; others seem impossible. Why?
Consider the following example from Asian Figures, translated by William
Merwin.

Blind
blames the ditch

To get some sense of the possible range of interpretations, consider the following
application of the proverb:

Suppose a presidential candidate knowingly commits some personal impro-

priety (though not illegal and not related to political issues) and his candidacy
is destroyed by the press’s reporting of the impropriety. He blames the press
for reporting it, rather than himself for committing it. We think he should have
recognized the realities of political press coverage when he chose to commit
the impropriety. We express our judgment by saying, “Blind / blames the ditch.”

Turner and I (1989) observed that the knowledge structure used in comprehend-
ing the case of the candidate’s impropriety shared certain things with knowledge
structure used in comprehending the literal interpretation of “Blind / blames the
ditch.” That knowledge structure is the following:

− There is a person with an incapacity, namely, blindness.

− He encounters a situation, namely a ditch, in which his incapacity, namely
his inability to see the ditch, results in a negative consequence, namely, his
falling into the ditch.
− He blames the situation, rather than his own incapacity.
– He should have held himself responsible, not the situation.

This specific knowledge schema about the blind man and the ditch is an instance
of a general knowledge schema, in which specific information about the blindness
and ditch are absent. Let us refer to it as the “generic-level schema” that structures
our knowledge of the proverb. That generic level knowledge schema is:
220 George Lakoff

− There is a person with an incapacity.

− He encounters a situation in which his incapacity results in a negative
consequence.
− He blames the situation rather than his own incapacity.
– He should have held himself responsible, not the situation.

This is a very general schema characterizing an open-ended category of situations.

We can think of it as a variable template that can be filled in many ways. As it
happened, Turner and I were studying this at the time of the Gary Hart scandal.
Hart, a presidential candidate, committed certain sexual improprieties during a
campaign, had his candidacy dashed, and then blamed the press for his downfall.
“Blind / blames the ditch” fits this situation. Here’s how:

− The person is the presidential candidate.

− His incapacity is his inability to understand the consequences of his per-
sonal improprieties.
− The context he encounters is his knowingly committing an impropriety and
the press’s reporting it.
− The consequence is having his candidacy dashed.
− He blames the press.
– We judge him as being foolish for blaming the press instead of himself.

If we view the generic-level schema as mediating between the proverb “Blind /

blames the ditch” and the story of the candidate’s impropriety, we get the follow-
ing correspondence:

− The blind person corresponds to the presidential candidate.

− His blindness corresponds to his inability to understand the consequences
of his personal improprieties.
− Falling into the ditch corresponds to his committing the impropriety and
having it reported.
− Being in the ditch corresponds to being out of the running as a candidate.
− Blaming the ditch corresponds to blaming the press coverage.
– Judging the blind man as foolish for blaming the ditch corresponds to judg-
ing the candidate as foolish for blaming the press coverage.

This correspondence defines the metaphorical interpretation of the proverb as

applied to the candidate’s impropriety. Moreover, the class of possible ways of
filling in the generic-level schema of the proverb corresponds to the class of pos-
sible interpretations of the proverb. Thus, we can explain why “Blind / blames
the ditch” does not mean I took a bath or My aunt is sitting on the sofa or any of
the myriad things the proverb cannot mean.
 Chapter 6: Conceptual metaphor 221

All the proverbs that Turner and I studied turned out to involve this sort of
generic-level schema, and the kinds of things that turned up in such schemata
seemed to be pretty much the same in case after case. They include:

− Causal structure
− Temporal structure
− Event shape; that is, instantaneous or repeated, completed or open-ended,
single or repeating, having fixed stages or not, preserving the existence of
entities or not, and so on
− Purpose structure
− Modal structure
– Linear scales

This is not an exhaustive list, but it includes most of the major elements of generic-
level structure we discovered. What is striking to us about this list is that everything
on it is, under the Invariance Principle, an aspect of image-schematic structure.
In short, if the Invariance Principle is correct, the way to arrive at a generic-
level schema for some knowledge structure is to extract its image-schematic
structure.
The metaphoric interpretation of such discourse forms as proverbs, fables, alle-
gories, and so on seems to depend on our ability to extract generic-level structure.
Turner and I have called the relation between a specific knowledge structure and
its generic-level structure the GENERIC¬IS SPECIlC¬metaphor. It is an extremely
common mechanism for comprehending the general in terms of the specific.
If the Invariance Principle is correct, then the GENERIC¬IS SPECIlC¬metaphor
is a minimal metaphor that maps what the Invariance Principle requires it to and
nothing more. Should it turn out that generic-level structure is exactly image-sche-
matic structure, then the Invariance Principle would have enormous explanatory
value. It would obviate the need for a separate characterization of generic-level
structure. Instead, it would itself characterize generic-level structure, explaining
possible personifications and the possible interpretations for proverbs.

5.3. Analogy
The GENERIC¬IS SPECIlC¬metaphor is used for more than just the interpretation of
proverbs. Turner (1991) has suggested that it is also the general mechanism at work
in analogic reasoning and that the Invariance Principle characterizes the class of
possible analogies. We can see how this works with the Gary Hart example cited
above. We can convert that example into an analogy with the following sentence:
Gary Hart was like a blind man who fell into a ditch and blamed the ditch. The
mechanism for understanding this analogy makes use of:
222 George Lakoff

− A knowledge schema for the blind man and the ditch

− A knowledge schema concerning Gary Hart
– The GENERIC¬IS SPECIlC¬metaphor

The GENERIC¬IS SPECIlC¬metaphor maps the knowledge schema for the blind man
and the ditch into its generic-level schema. The generic-level schema defines an
open-ended category of knowledge schemata. The Gary Hart schema is a member
of that category, since it fits the generic-level schema given the correspondences
stated above.
It appears at present that such analogies use this metaphorical mechanism.
But it is common for analogies to use other metaphorical mechanisms as well, for
instance, the Great Chain Metaphor and the full range of conventional mappings
in the conceptual system. Sentences like John is a wolf or Harry is a pig use the
Great Chain metaphor (see Lakoff and Turner 1989: Chapter 4).
A good example of how the rest of the metaphor system interacts with GENERIC¬
IS SPECIlC¬is the well-known example of Glucksberg and Keysar (1993), my job
is a jail. First, the knowledge schema for a jail includes the knowledge that a jail
imposes extreme physical constraints on a prisoner’s movements. The GENERIC¬
IS SPECIlC¬metaphor preserves the image-schematic structure of the knowledge
schema, factoring out the specific details of the prisoner and the jail: X imposes
extreme physical constraints on Y’s movements. But now two additional conven-
tional metaphors apply to this generic-level schema: The event structure metaphor,
with the submetaphor ACTIONS¬ARE¬SELF PROPELLED¬MOVEMENTS¬and PSYCHO
LOGICAL¬FORCE¬IS PHYSICAL¬FORCE¬These metaphors map “X imposes extreme
physical constraints on Y’s movements” into “X imposes extreme psychological
constraints on Y’s actions.” The statement my job is a jail imposes an interpreta-
tion in which X = my job and Y = me, and hence yields the knowledge that “my
job imposes extreme psychological constraints on my actions.” Thus, the mecha-
nism for understanding my job is a jail uses very common, independently existing
metaphors: GENERIC¬IS SPECIlC¬PSYCHOLOGICAL¬FORCE¬IS PHYSICAL¬FORCE¬and
the Event Structure Metaphor.

5.4. The Glucksberg-Keysar Claim

I mention this example because of the claim by Glucksberg and Keysar (1993)
that metaphor is simply a matter of categorization. In personal correspondence,
however, Glucksberg has written, “We assume that people can judge and can also
infer that certain basic level entities, such as ‘jails,’ typify or are emblematic of a
metaphoric attributive category such as ‘situations that are confining, unpleasant,
etc.’ ” Glucksberg and Keysar give no theory of how it is possible to have such a
“metaphoric attributive category” – that is, how it is possible for one kind of thing
 Chapter 6: Conceptual metaphor 223

(a general situation) to be metaphorically categorized in terms of a fundamen-

tally spatial notion like “confining.” Since Glucksberg is not in the business of
describing the nature of conceptual systems, he does not see it as his job to give
such an account. I have argued in this essay that the general principle govern-
ing such cases is the Event Structure Metaphor. If such a metaphor exists in our
conceptual system, then the Glucksberg-Keysar “jail” example is accounted for
automatically and their categorization theory is not needed. Indeed, the category he
needs – “situations that are confining, unpleasant, etc.” – is a “metaphoric attribu-
tive category.” That is, to get the appropriate categories in their categorization
theory of metaphor he needs an account of metaphor. But given such an account
of metaphor, the metaphor-as-categorization theory becomes unnecessary.
Even worse for the Glucksberg-Keysar theory, it cannot account for either
everyday conceptual metaphor of the sort we have been discussing or for really
rich poetic metaphor, such as one finds in the works of, say, Dylan Thomas, or
for image metaphor of the sort common in the examples cited above from the
Sanskrit, Navaho, and surrealist traditions. Since it does not even attempt to deal
with most of the data covered by the contemporary theory of metaphor, it cannot
account for “how metaphor works.”

5.5. More on novel metaphor

At the time most of the chapters in Ortony (1993[1979]) were written (the late
1970s), “metaphor” was taken to mean “novel metaphor,” since the huge system
of conventional metaphor had barely been noticed. The authors therefore never
took up the question of how the system of conventional metaphor functions in
the interpretation of novel metaphor. We have just seen one such example. Let
us consider some others.
As common as novel metaphor is, its occurrence is rare by comparison with
conventional metaphor, which occurs in most of the sentences we utter. Our every-
day metaphor system, which we use to understand concepts as commonplace as
TIME¬STATE¬CHANGE¬CAUSATION¬PURPOSE¬and so forth is constantly active,
and is used maximally in interpreting novel metaphorical uses of language. The
problem with all the older research on novel metaphor is that it completely missed
the major contribution played by the conventional system.
As Turner and I discussed in detail (Lakoff and Turner 1989), there are three
basic mechanisms for interpreting linguistic expressions as novel metaphors: exten-
sions of conventional metaphors, generic-level metaphors, and image metaphors.
Most interesting poetic metaphor uses all these superimposed on one another. Let
us begin with examples of extensions of conventional metaphors. Dante begins
the Divine Comedy:
224 George Lakoff

In the middle of life’s road

I found myself in a dark wood.

“Life’s road” evokes the domain of life and the domain of travel, and hence the
conventional LIFE¬IS A¬JOURNEY¬metaphor that links them. “I found myself in a
dark wood” evokes the knowledge that if it’s dark you cannot see which way to
go. This evokes the domain of seeing, and thus the conventional metaphor that
KNOWING¬IS SEEING¬as in I see what you’re getting at, his claims aren’t clear,
the passage is opaque, and so forth. This entails that the speaker doesn’t know
which way to go. Since the LIFE¬IS A¬JOURNEY¬metaphor specifies destinations
are life goals, the speaker must not know what life goals to pursue, that is, he is
without direction in his life. All this uses nothing but the system of conventional
metaphor, ordinary knowledge structure evoked by the conventional meaning of
the sentence, and metaphorical inferences based on that knowledge structure.
Another equally simple case of the use of the conventional system is Robert
Frost’s

Two roads diverged in a wood, and I –

I took the one less traveled by,
And that has made all the difference.

Since Frost’s language often does not overtly signal that the poem is to be taken
metaphorically, incompetent English teachers occasionally teach Frost as if he
were a nature poet, simply describing scenes. (I have actually had students whose
high school teachers taught them that!) Thus, this passage could be read non-
metaphorically as being just about a trip on which one encounters a crossroads.
There is nothing in the sentence itself that forces one to a metaphorical inter-
pretation. But, since it is about travel and encountering crossroads, it evokes a
knowledge of journeys. This activates the system of conventional metaphor we
have just discussed, in which long-term, purposeful activities are understood as
journeys, and further, how life and careers can also be understood as one-person
journeys (love relationships, involving two travelers, are ruled out here). The
poem is typically taken as being about life and a choice of life goals, though it
might also be interpreted as being about careers and career paths, or about some
long-term, purposeful activity. All that is needed to get the requisite range of
interpretations is the structure of conventional metaphors discussed above, and
the knowledge structure evoked by the poem. The conventional mapping will
apply to the knowledge structure yielding the appropriate inferences. No special
mechanisms are needed.
 Chapter 6: Conceptual metaphor 225

5.6. Searle’s theory

I will not pursue discussion of other more complex poetic examples, since they
require lengthy treatment which can be found in Lakoff and Turner (1989), Turner
(1987), and Turner (1991). Instead, I will confine myself to discussing three exam-
ples from John Searle (1993). Consider first Disraeli’s remark, “I have climbed
to the top of the greasy pole.”
This could be taken nonmetaphorically, but its most likely metaphorical inter-
pretation is via the CAREER¬IS A¬JOURNEY¬metaphor. This metaphor is evoked
jointly by source domain knowledge about pole climbing, which is effortful, self-
propelled, destination-oriented motion upward, and knowledge that the metaphor
involves effortful, self-propelled, destination-oriented motion upward. Part of the
knowledge evoked is that the speaker is as high as he can get on that particular
pole, that the pole was difficult to climb, that the climb probably involved back-
ward motion, that it is difficult for someone to stay at the top of a greasy pole, and
that he will most likely slide down again. The CAREER¬IS A¬JOURNEY¬metaphor
maps this knowledge onto corresponding knowledge about the speaker’s career:
he has as much status as he can get in that particular career, it was difficult to
get to that point in the career, it probably involved some temporary loss of status
along the way, it will be difficult to maintain this position, and he will probably
lose status before long. All this follows with nothing more than the conventional
CAREER AS JOURNEY¬mapping, which we all share as part of our metaphorical
systems, plus knowledge about climbing greasy poles.
The second example of Searle’s I will consider is Sally is a block of ice. Here
there is a conventional metaphor that AFFECTION¬IS WARMTH¬as in ordinary sen-
tences like she’s a warm person, he was cool to me, and so forth. A block of ice
evokes the domain of temperature and, since it is predicated of a person, it also
evokes knowledge of what a person can be. Jointly, both kinds of knowledge
activate AFFECTION¬IS WARMTH¬Since a block of ice is something very cold and
not warmed quickly or easily, this knowledge is mapped onto Sally as being very
unaffectionate and not able to become affectionate quickly or easily. Again, com-
mon knowledge and a conventional metaphor we all have is all that is needed.
Finally, Searle discusses the hours crept by as we waited for the plane. Here
we have a verb of motion predicated of a time expression; the former activates the
knowledge about motion through space and the latter activates the time domain.
Jointly, they activate the TIME AS MOVING OBJECT mapping. Again the meaning
of the sentence follows only from everyday knowledge and the everyday system
of metaphorical mappings.
Searle accounts for such cases by his Principle 4, which says that “we just
do perceive a connection” which is the basis of the interpretation. This is vague
and doesn’t say what the perceived connection is or why we “just do” perceive it.
226 George Lakoff

When we spell out the details of all such “perceived connections,” they turn out
to be the system of conceptual metaphors I have been describing. But given that
system, Searle’s theory and his principles become unnecessary.
In addition, Searle’s account of literal meaning makes most of the usual false
assumptions that accompany that term. Searle assumes that all everyday, con-
ventional language is literal and not metaphorical. He would thus rule out every
example of conventional metaphor described not only in this chapter, but in the
whole literature of the field.
The study of the metaphorical subsystem of our conceptual system is a cen-
tral part of synchronic linguistics because much of our semantic system, that
is, our system of concepts, is metaphorical, as we saw above. Because this huge
system went unnoticed prior to 1980, authors like Searle, Sadock, and Morgan
could claim, incorrectly as it turns out, that metaphor was outside of synchronic
linguistics and in the domain of principles of language use.

6. The experiential basis of metaphor

The conceptual system underlying a language contains thousands of conceptual

metaphors – conventional mappings from one domain to another, such as the
Event Structure Metaphor. The novel metaphors of a language are, except for
image metaphors, extensions of this large conventional system.
Perhaps the deepest question that any theory of metaphor must answer is this:
why do we have the conventional metaphors that we have? Or alternatively: is there
any reason why conceptual systems contain one set of metaphorical mappings
rather than another? There do appear to be answers to these questions for many
of the mappings found so far, though they are in the realm of plausible accounts,
rather than in the realm of scientific results.
Take a simple case: the MORE¬IS¬UP¬metaphor, as seen in expressions like prices
rose; his income went down; unemployment is up; exports are down; the number
of homeless people is very high.
There are other languages in which MORE¬IS¬UP¬and LESS¬IS DOWN¬but none
in which the reverse is true, where MORE¬IS DOWN¬and LESS¬IS UP¬Why not?
Contemporary theory postulates that the MORE¬IS UP¬metaphor is grounded in
experience – in the common experiences of pouring more fluid into a container
and seeing the level go up, or adding more things to a pile and seeing the pile get
higher. These are thoroughly pervasive experiences; we encounter them every
day of our lives. They have structure – a correspondence between the conceptual
domain of quantity and the conceptual domain of verticality: MORE¬corresponds
in such experiences to up and LESS¬corresponds to DOWN¬These correspondences
in real experience form the basis for the correspondences in the metaphorical
 Chapter 6: Conceptual metaphor 227

cases, which go beyond real experience: in prices rose there is no correspondence

in real experience between quantity and verticality, but understanding quantity
in terms of verticality makes sense because of a regular correspondence in so
many other cases.
Consider another case. What is the basis of the widespread KNOWING¬IS SEEING¬
metaphor, as in expressions like I see what you’re saying; his answer was clear;
this paragraph is murky; he was so blinded by ambition that he never noticed his
limitations? The experiential basis in this case is the fact that most of what we
know comes through vision, and in the overwhelming majority of cases, if we
see something, then we know it is true.
Consider still another case. Why, in the Event Structure Metaphor, is achiev-
ing a purpose understood as reaching a destination (in the location subsystem)
and as acquiring a desired object (in the object subsystem)? The answer again
seems to be correspondences in everyday experience. To achieve most of our
everyday purposes, we either have to move to some destination or acquire some
object. If you want a drink of water, you’ve got to go to the water fountain. If you
want to be in the sunshine, you have to move to where the sunshine is. And if you
want to write down a note, you have to get a pen or pencil. The correspondences
between achieving purposes and either reaching destinations or acquiring objects
is so utterly common in our everyday existence, that the resulting metaphor is
completely natural.
But what about the experiential basis of A¬PURPOSEFUL¬LIFE¬IS A¬JOURNEY¬
Recall that the mapping is in an inheritance hierarchy, where life goals are special
cases of purposes, which are destinations in the event structure metaphor. Thus,
A¬PURPOSEFUL¬LIFE¬IS A¬JOURNEY¬inherits the experiential basis of PURPOSES¬ARE¬
DESTINATIONS¬Thus, inheritance hierarchies provide indirect experiential bases,
in that a metaphorical mapping lower in a hierarchy can inherit its experiential
basis indirectly from a mapping higher in the hierarchy.
Experiential bases motivate metaphors, they do not predict them. Thus, not
every language has a MORE¬IS UP¬metaphor, though all human beings experience
a correspondence between MORE¬and UP¬What this experiential basis does predict
is that no language will have the opposite metaphor LESS¬IS UP¬It also predicts
that a speaker of a language without that metaphor will be able to learn it much
more easily than its reverse.

6.1. Realizations of metaphor

Consider objects like thermometers and stock market graphs, where increases in
temperature and prices are represented as being up and decreases as being down.
These are objects created by humans to accord with the MORE¬IS¬UP¬metaphor.
They exhibit a correlation between MORE¬and UP¬and are much easier to read and
228 George Lakoff

understand than if they contradicted the metaphor, if, say, increases were repre-
sented as down and decreases as up.
Such objects are ways in which metaphors impose a structure on real life,
through the creation of new correspondences in experience. And once created in
one generation, they serve as an experiential basis for that metaphor in the next
generation.
There are a great many ways in which conventional metaphors can be made
real. They can be realized in obvious imaginative products such as cartoons, liter-
ary works, dreams, visions, and myths, but they can be made real in less obvious
ways as well, in physical symptoms, social institutions, social practices, laws, and
even foreign policy and forms of discourse and history.
Let us consider some examples.

6.1.1. Cartoons
Conventional metaphors are made real in cartoons. A common example is the
realization of the ANGER¬IS A¬HOT¬mUID¬IN¬A¬CONTAINER¬metaphor, in which one
can be boiling mad or letting off steam. In cartoons, anger is commonly depicted
by steam coming out of the character’s ears. Social clumsiness is indicated by
having a cartoon character fall on his face.

6.1.2. Literary works

It is common for the plot of a novel to be a realization of the PURPOSEFUL¬LIFE¬
IS A¬JOURNEY¬metaphor, where the course of a life takes the form of an actual
journey. Pilgrim’s Progress is a classic example.

6.1.3. Rituals
Consider the cultural ritual in which a newborn baby is carried upstairs to ensure
his or her success. The metaphor realized in this ritual is STATUS¬IS UP¬as in: he
clawed his way to the top; he climbed the ladder of success; you’ll rise in the
world.

6.1.4. Dream interpretation

Conceptual metaphors constitute the vocabulary of dream interpretation. The
collection of our everyday conceptual metaphors makes dream interpretation pos-
sible. Consider one of the most celebrated of all examples, Joseph’s interpretation
of Pharaoh’s dream from Genesis. In Pharaoh’s dream, he is standing on the river
bank when seven fat cows come out of the river, followed by seven lean cows that
eat the seven fat ones and still remain lean. Pharaoh dreams again. This time he
sees seven “full and good” ears of corn growing and then seven withered ears
 Chapter 6: Conceptual metaphor 229

growing after them. The withered ears devour the good ears. Joseph interprets the
two dreams as a single dream. The seven fat cows and full ears are good years
and the seven lean cows and withered ears are famine years that follow the good
years. The famine years devour what the good years produce. This interpreta-
tion makes sense to us because of a collection of conceptual metaphors in our
conceptual system – metaphors that have been with us since biblical times. The
first metaphor is TIMES¬ARE¬MOVING¬ENTITIES¬A river is a common metaphor for
the flow of time; the cows are individual entities (years) emerging from the flow
of time and moving past the observer; the ears of corn are also entities that come
into the scene. The second metaphor is ACHIEVING¬A¬PURPOSE¬IS EATING¬where
being fat indicates success, being lean indicates failure. This metaphor is com-
bined with the most common of metonymies, A¬PART¬STANDS¬FOR¬THE¬WHOLE¬
Since cows and corn were typical of meat and grain eaten, each single cow stands
for all the cows raised in a year and each ear of corn for all the corn grown in a
year. The final metaphor is RESOURCES¬ARE¬FOOD¬where using up resources is
eating food. The devouring of the good years by the famine years is interpreted
as indicating that all the surplus resources of the good years will be used up by
the famine years. The interpretation of the whole dream is thus a composition of
three conventional metaphors and one metonymy. The metaphoric and metonymic
sources are combined to form the reality of the dream.

6.1.5. Myths

In the event structure metaphor, there is a submapping EXTERNAL¬EVENTS¬ARE¬

LARGE¬MOVING¬OBJECTS¬that can exert a force on you and thereby affect whether
you achieve your goals. In English the special cases of such objects are “things,”
fluids, and horses. Pamela Morgan (in unpublished work) has observed that in
Greek mythology, Poseidon is the god of the sea, earthquakes, horses, and bulls.
The list might seem arbitrary, but Morgan observes that these are all large mov-
ing objects that can exert a force on you. Poseidon, she surmises, should really
be seen as the god of external events.

6.1.6. Physical symptoms

The unconscious mind makes use of our unconscious system of conventional meta-
phor, sometimes to express psychological states in terms of physical symptoms.
For example, in the event structure metaphor, there is a submapping DIFlCULTIES¬
ARE¬IMPEDIMENTS¬TO¬MOTION¬which has, as a special case, DIFlCULTIES¬ARE¬
BURDENS¬It is fairly common for someone encountering difficulties to walk with
his shoulders stooped, as if carrying a heavy weight that is burdening him.
230 George Lakoff

6.1.7. Social institutions

We have a TIME¬IS MONEY¬metaphor, shown by expressions like he’s wasting time;

I have to budget my time; this will save you time; I’ve invested a lot of time in that;
he doesn’t use his time profitably. This metaphor came into English use about the
time of the industrial revolution, when people started to be paid for work by the
amount of time they put in. Thus, the factory led to the institutional pairing of
periods of time with amounts of money, which formed the experiential basis of
this metaphor. Since then, the metaphor has been realized in many other ways.
The budgeting of time has spread throughout American culture.

6.1.8. Social practices

There is a conceptual metaphor that SEEING¬IS TOUCHING¬where the eyes are limbs
and vision is achieved when the object seen is “touched.” Examples are: my eyes
picked out every detail of the pattern; he ran his eyes over the walls; he couldn’t
take his eyes off of her; their eyes met; his eyes are glued to the TV. The metaphor
is made real in the social practice of avoiding eye “contact” on the street, and in
the social prohibition against “undressing someone with your eyes.”

6.1.9. Laws

Law is a major area where metaphor is made real. For example, CORPORATIONS¬
ARE¬PERSONS¬is a tenet of American law, which not only enables corporations to
be harmed or assigned responsibility so they can be sued when liable, but also
gives them certain First Amendment rights.

6.1.10. Foreign policy

A¬STATE¬IS A¬PERSON¬is one of the major metaphors underlying foreign policy

concepts. Thus, there are friendly states, hostile states, and so forth. Health for a
state is economic health and strength is military strength. A threat to economic
health can be seen as a death threat, as when Iraq was seen to have a strangle-
hold on the economic lifeline of the United States. Strong states are seen as male
and weak states as female, so that an attack by a strong state on a weak one can
be seen as a rape, as in the rape of Kuwait by Iraq. A just war is conceptualized
as a fairy tale with villain, victim, and hero, where the villain attacks the victim
and the hero rescues the victim. Thus, the United States and allies in the Gulf
War were portrayed as having rescued Kuwait. As President Bush said in his
address to Congress, “The issues couldn’t have been clearer: Iraq was the villain
and Kuwait, the victim.”
 Chapter 6: Conceptual metaphor 231

6.1.11. Forms of discourse

Common metaphors are often made real in discourse forms. Consider three
common academic discourse forms: the guided tour, the heroic battle, and the
heroic quest. The guided tour is based on the metaphor that THOUGHT¬IS MOTION¬
where ideas are locations and one reasons step-by-step, reaches conclusions, or
fails to reach a conclusion if engaged in circular reasoning. Communication in
this metaphor is giving someone a guided tour of some rational argument or of
some intellectual terrain. This essay is an example of such a guided tour, where
I, the author, am the tour guide who is assumed to be thoroughly familiar with
the terrain and the terrain surveyed is taken as objectively real. The discourse
form of the heroic battle is based on the metaphor that ARGUMENT¬IS WAR¬The
author’s theory is the hero, the opposing theory is the villain, and words are
weapons. The battle is in the form of an argument defending the hero’s position
and demolishing that of the villain. The heroic quest discourse form is based
on the metaphor that knowledge is a valuable but elusive object that can be dis-
covered if one perseveres. The scientist is the hero on a quest for knowledge,
and the discourse form is an account of his difficult journey of discovery. What
is discovered is a real entity.
What makes all these cases realizations of metaphors is that in each case
something real is structured by conventional metaphor, and thereby made com-
prehensible, or even natural. What is real differs in each case: an object like a
thermometer or graph, an experience like a dream, an action like a ritual, a form
of discourse, and so forth. These examples reveal that much of what is real in a
society or in the experience of an individual is structured and made sense of via
conventional metaphor.
Experiential bases and realizations of metaphors are two sides of the same
coin: they are both correlations in real experience that have the same structure
as the correlations in metaphors. The difference is that experiential bases pre-
cede, ground, and make sense of conventional metaphorical mappings, whereas
realizations follow, and are made sense of, via the conventional metaphors. And
as we noted above, one generation’s realizations of a metaphor can become part
of the next generation’s experiential basis for that metaphor.

7. Summary of results

As we have seen, the contemporary theory of metaphor is revolutionary in many

respects. To give you some idea of how revolutionary, here is a list of the basic
results that differ from most previous accounts.
232 George Lakoff

7.1. The nature of metaphor

− Metaphor is the main mechanism through which we comprehend abstract
concepts and perform abstract reasoning.
− Much subject matter, from the most mundane to the most abstruse scien-
tific theories, can only be comprehended via metaphor. Metaphor is funda-
mentally conceptual, not linguistic, in nature.
− Metaphorical language is a surface manifestation of conceptual metaphor.
− Though much of our conceptual system is metaphorical, a significant part
of it is nonmetaphorical. Metaphorical understanding is grounded in non-
metaphorical understanding.
− Metaphor allows us to understand a relatively abstract or inherently
unstructured subject matter in terms of a more concrete, or at least more
highly structured subject matter.

7.2. The structure of metaphor

− Metaphors are mappings across conceptual domains.
− Such mappings are asymmetric and partial.
− Each mapping is a fixed set of ontological correspondences between enti-
ties in a source domain and entities in a target domain.
− When those fixed correspondences are activated, mappings can project
source domain inference patterns onto target domain inference patterns.
− Metaphorical mappings obey the Invariance Principle: The image-schema
structure of the source domain is projected onto the target domain in a way
that is consistent with inherent target domain structure.
− Mappings are not arbitrary, but grounded in the body and in everyday
experience and knowledge.
− A conceptual system contains thousands of conventional metaphorical
mappings which form a highly structured subsystem of the conceptual
system.
− There are two types of mappings: conceptual mappings and image map-
pings; both obey the Invariance Principle.

7.3. Some aspects of metaphor

− The system of conventional conceptual metaphor is mostly unconscious,
automatic, and used with no noticeable effort, just like our linguistic sys-
tem and the rest of our conceptual system.
− Our system of conventional metaphor is “alive” in the same sense that our
 Chapter 6: Conceptual metaphor 233

system of grammatical and phonological rules is alive; namely, it is con-

stantly in use, automatically, and below the level of consciousness.
− Our metaphor system is central to our understanding of experience and to
the way we act on that understanding.
− Conventional mappings are static correspondences, and are not, in them-
selves, algorithmic in nature. However, this by no means rules out the pos-
sibility that such static correspondences might be used in language process-
ing that involves sequential steps.
− Metaphor is mostly based on correspondences in our experiences, rather
than on similarity.
− The metaphor system plays a major role in both the grammar and lexicon
of a language.
− Metaphorical mappings vary in universality; some seem to be universal,
others are widespread, and some seem to be culture specific.
– Poetic metaphor is, for the most part, an extension of our everyday, conven-
tional system of metaphorical thought.

These are the conclusions that best fit the empirical studies of metaphor conducted
over the past decade or so. Though many of them are inconsistent with traditional
views, they are by no means all new, and some ideas – for example, that abstract
concepts are comprehended in terms of concrete concepts – have a long history.

8. Concluding remarks

The evidence supporting the contemporary theory of metaphor is voluminous

and grows larger each year as research in the field continues. The evidence, as
we saw above, comes from five domains:

− Generalizations over polysemy

− Generalizations over inference patterns
− Generalizations over extensions to poetic cases
− Generalizations over semantic change
– Psycholinguistic experiments

I have discussed only a handful of examples of the first three of these, enough, I
hope, to make the reader curious about the field.
Evidence is convincing, however, only if it can count as evidence. When does
evidence fail to be evidence? Unfortunately, all too often. It is commonly the
case that certain fields of inquiry are defined by assumptions that rule out the
possibility of counterevidence. When a defining assumption of a field comes up
against evidence, the evidence usually loses: the practitioners of the field must
234 George Lakoff

ignore the evidence if they want to keep the assumptions that define the field they
are committed to.
Part of what makes the contemporary theory of metaphor so interesting is
that the evidence for it contradicts the defining assumptions of so many academic
disciplines. In my opinion, this should make one doubt the defining assumptions
of all those disciplines. The reason is this: the defining assumptions of the con-
temporary theory of metaphor are minimal. There are only two.

1. The generalization commitment: To seek generalizations in all areas of

language, including polysemy, patterns of inference, novel metaphor, and
semantic change.
2. The cognitive commitment: To take experimental evidence seriously.

But these are nothing more than commitments to the scientific study of language
and the mind. No initial commitment is made as to the form of an answer to the
question of what is metaphor.
The defining assumptions of other fields do, however, often entail a commitment
about the form of an answer to that question. It is useful, in an interdisciplinary
volume of this sort, to spell out exactly what those defining assumptions are, since
they will often explain why different authors reach such different conclusions
about the nature of metaphor.

8.1. Literal meaning commitments

I started this chapter with a list of the false assumptions about literal meaning
that are commonly made. These assumptions are “false” only relative to the kinds
of evidence that support the contemporary theory of metaphor. If one ignores all
such evidence, the assumptions can be maintained without contradiction.
Assumptions about literality are the locus of many of the contradictions between
the contemporary theory of metaphor and various academic disciplines. Let us
review those assumptions. In the discussion of literal meaning given above, I
observed that it is taken as definitional that what is literal is not metaphorical.
The “false assumptions and conclusions” that usually accompany the word “lit-
eral” are:
− All everyday conventional language is literal, and none is metaphorical.
− All subject matter can be comprehended literally, without metaphor.
− Only literal language can be contingently true or false.
− All definitions given in the lexicon of a language are literal, not
metaphorical.
– The concepts used in the grammar of a language are all literal; none is
metaphorical.
 Chapter 6: Conceptual metaphor 235

We will begin with the philosophy of language. The generalization commitment

and the cognitive commitment are not definitional to the philosophy of language.
Most philosophers of language would feel no need to abide by them, for a very
good reason. The philosophy of language is typically not seen as an empirical dis-
cipline, constrained by empirical results, such as those that arise from the applica-
tion of the generalization and cognitive commitments. Instead, the philosophy of
language is usually seen as an a priori discipline, which can be pursued using the
tools of philosophical analysis alone, rather than the tools of empirical research.
Therefore, all the evidence that has been brought forth for the contemporary theory
of metaphor simply will not matter for most philosophers of language.
In addition, the philosophy of language comes with its own set of defining
assumptions, which entail many of the false assumptions usually associated with
the word “literal.” Most practitioners of the philosophy of language usually make
one or more of the following assumptions.

− The correspondence theory of truth.

− Meaning is defined in terms of reference and truth.
– Natural language semantics is characterized by the mechanisms of math-
ematical logic, including model theory.

The very field of philosophy of language thus comes with defining assumptions
that contradict the main conclusions of the contemporary theory of metaphor.
Consequently, we can see why most philosophers of language have the range of
views on metaphor that they have: they accept the traditional literal-figurative
distinction. They may, like M. Johnson (1981), say that there is no metaphori-
cal meaning, and that most metaphorical utterances are either trivially true or
trivially false. Or, like Grice (1989: 34) and Searle (1993), they will assume that
metaphor is in the realm of pragmatics, that is, that a metaphorical meaning is
no more than the literal meaning of some other sentence which can be arrived
at by some pragmatic principle. This is required, since the only real meaning for
them is literal meaning, and pragmatic principles are those principles that allow
one to say one thing (with a literal meaning) and mean something else (with a
different, but nonetheless literal, meaning).
Much of generative linguistics accepts one or more of these assumptions from
the philosophy of language. The field of formal semantics accepts them all, and
thus formal semantics, by its defining assumptions, is at odds with the contem-
porary theory of metaphor. Formal semantics simply does not see it as its job to
account for the generalizations discussed in this chapter. From the perspective
of formal semantics, the phenomena that the contemporary theory of metaphor
is concerned with are either nonexistent or uninteresting, since they lie outside
the purview of the discipline. Thus Jerrold Sadock (1993) claims that metaphor
236 George Lakoff

lies outside of synchronic linguistics. Since he accepts mathematical logic as the

correct approach to natural language semantics, Sadock must see metaphor as
being outside of semantics proper. He must, therefore, also reject the enterprise of
the contemporary theory of metaphor. And Morgan (1993), also accepting those
defining assumptions of the philosophy of language, agrees with Grice and Searle
that metaphor is a matter of pragmatics.
Chomsky’s (1981) theory of government and binding also accepts crucial
assumptions from the philosophy of language that are inconsistent with the con-
temporary theory of metaphor. Government and binding, following my early theory
of generative semantics, assumes that semantics is to be represented in terms of
logical form. Government and binding, like generative semantics, thus rules out
the very possibility that metaphor might be part of natural language semantics as
it enters into grammar. Because of this defining assumption, I would not expect
government and binding theorists to become concerned with the phenomena
covered by the contemporary theory of metaphor.
It is interesting that much of continental philosophy and deconstructionism is
also characterized by defining assumptions at odds with the contemporary theory
of metaphor. Nietzsche (see Johnson 1981) held that all language is metaphori-
cal, a theory at odds with those results indicating that a significant amount of
everyday language is not metaphorical (see subsection, “What is not metaphori-
cal”). Much of continental philosophy, observing that conceptual systems change
through time, assumes that conceptual systems are purely historically contingent,
that there are no conceptual universals. Though conceptual systems do change
through time, there do, however, appear to be universal, or at least very widespread,
conceptual metaphors. The event structure metaphor is my present candidate for
a metaphorical universal.
Continental philosophy also comes with a distinction between the study of the
physical world, which can be scientific, and the study of human beings, which it
says cannot be scientific. This is very much at odds with the conceptual theory
of metaphor, which is very much a scientific enterprise.
Finally, the contemporary theory of metaphor is at odds with certain tradi-
tions in symbolic artificial intelligence and information processing psychology.
Those fields assume that thought is a matter of algorithmic symbol manipulation,
of the sort done by a traditional computer program. This defining assumption is
inconsistent with the contemporary theory of metaphor in two respects.
First, the contemporary theory has an image-schematic basis. The Invariance
Principle both applies to image metaphors and characterizes constraints on novel
metaphor. Since symbol manipulation systems cannot handle image-schemas,
they cannot deal with image metaphors or imageable idioms.
Second, those traditions must characterize metaphorical mapping as an algo-
rithmic process, which typically takes literal meanings as input and gives a meta-
 Chapter 6: Conceptual metaphor 237

phorical reading as output. This runs counter to cases where there are multiple,
overlapping metaphors in a single sentence, and which require the simultaneous
activation of a number of metaphorical mappings.
The contemporary theory of metaphor is thus not only interesting for its own
sake. It is especially interesting for the challenge it presents to other disciplines.
If the results of the contemporary theory are accepted, the defining assumptions
of whole disciplines are brought into question.

Note
* This research was supported in part by grants from the Sloan Foundation and the
National Science Foundation (IRI-8703202) to the University of California at Berkeley.
The following colleagues and students helped with this essay in a variety of ways, from
useful comments to allowing me to cite their research: Ken Baldwin, Claudia Brugman,
Jane Espenson, Sharon Fischler, Ray Gibbs, Adele Goldberg, Mark Johnson, Karin Myhre,
Eve Sweetser, and Mark Turner.

References
Chomsky, Noam
1981 Lectures on Government and Binding. Dordrecht: Foris Publications.
Gibbs, Raymond W., Jr.
1990 Psycholinguistic studies on the conceptual basis of idiomaticity. Cognitive
Linguistics 1: 417–462.
1993 Process and products in making sense of tropes. In Metaphor and Thought,
Andrew Ortony (ed.), 252–276. Cambridge: Cambridge University Press.
Glucksberg, Sam and Boaz Keysar
1993 How metaphors work. In Metaphor and Thought, Andrew Ortony (ed.), 401–
424. Cambridge: Cambridge University Press.
Grice, Paul
1989 Studies in the Way of Words. Cambridge, MA: Harvard University Press.
Johnson, M.
1981 Philosophical Perspectives on Metaphor. Minneapolis: University of Min-
nesota Press.
Kövecses, Zoltan
1990 Emotion Concepts. New York: Springer-Verlag.
Lakoff, George
1987 Women, Fire, and Dangerous Things: What Categories Reveal about the
Mind. Chicago: University of Chicago Press.
Lakoff, George and Mark Johnson
1980 Metaphors We Live By. Chicago: University of Chicago Press.
238 George Lakoff

Lakoff, George and Mark Turner

1989 More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago: Uni-
versity of Chicago Press.
Merwin, William S.
1973 Asian Figures. New York: Atheneum.
Merwin, William S. and Masson, J. Moussaieff (trans.)
1981 The Peacock’s Egg. San Francisco: North Point Press.
Morgan, Pamela
1993 In Metaphor and Thought, Andrew Ortony (ed.). Cambridge: Cambridge
University Press.
Ortony, Andrew (ed.)
1993 Metaphor and Thought. Cambridge: Cambridge University Press. (First edi-
tion published in 1979.)
Reddy, Michael
1993 The conduit metaphor. A case of frame conflict in our language about lan-
guage. In Metaphor and Thought, Andrew Ortony (ed.), 164–201. Cambridge:
Cambridge University Press.
Sadock, Jerrold
1993 Figurative speech and linguistics. In Metaphor and Thought, Andrew Ortony
(ed.), 42–57. Cambridge: Cambridge University Press.
Searle, John R.
1993 Metaphor. In Metaphor and Thought, Andrew Ortony (ed.), 83–111. Cam-
bridge: Cambridge University Press.
Sweetser, Eve
1990 From Etymology to Pragmatics: The Mind-as-Body Metaphor in Semantic
Structure and Semantic Change. Cambridge: Cambridge University Press.
Turner, Mark
1987 Death Is the Mother of Beauty: Mind, Metaphor, Criticism. Chicago: Univer-
sity of Chicago Press.
1991 Reading Minds: The Study of English in the Age of Cognitive Science. Prin-
ceton: Princeton University Press.
Chapter 7
Image schema

The cognitive psychological reality of image schemas

and their transformations
Raymond W. Gibbs, Jr. and Herbert L. Colston

One of the important claims of cognitive semantics is that much of our knowledge
is not static, propositional and sentential, but is grounded in and structured by vari-
ous patterns of our perceptual interactions, bodily actions, and manipulations of
objects (Johnson 1987, 1993; Lakoff 1987, 1990; Talmy 1988). These patterns are
experiential gestalts, called image schemas, that emerge throughout sensorimotor
activity as we manipulate objects, orient ourselves spatially and temporally, and
direct our perceptual focus for various purposes (Johnson 1991).
Studies in cognitive linguistics suggest that over two dozen different image
schemas and several image schema transformations appear regularly in people’s
everyday thinking, reasoning, and imagination (Johnson 1987; Lakoff 1987).
Among these are the schematic structures of CONTAINER¬BALANCE¬SOURCE
PATH GOAL¬PATH¬CYCLE¬ATTRACTION, CENTERPERIPHERY¬and LINK¬These
image schemas cover a wide range of experiential structures that are pervasive
in experience, have internal structure, and can be metaphorically elaborated to
provide for our understanding of more abstract domains. For example, cognitive
linguistic research has examined how image schemas are used to create gram-
matical forms (Langacker 1987, 1991), to represent the underlying meaning that
relates the seemingly disparate senses of prepositions (Brugman and Lakoff 1988;
Vandeloise 1993), to motivate verb-particle constructions, such as those focusing
on up and out (Lindner 1983), adverbs, such as very (Brugman 1984), certain
verbs, such as take (Norvig and Lakoff 1987), as well as to explain the many kinds
of cognitive relationships that can form the basis of the extension of a category
such as Japanese hon (Lakoff 1987). More recent investigations from linguistics
and philosophy examined the role that image schemas have in motivating abstract
metaphorical concepts, such as causation, death, and morality (Johnson 1993;
Lakoff 1990; Lakoff and Turner 1989; Turner 1991).
Although these studies provide important evidence on image schemas in every-
day thought and linguistic understanding, the question remains as to whether
there exists independent empirical evidence on the psychological reality of image

Originally published in 1995 in Cognitive Linguistics 6(4): 347–378.

240 Raymond W. Gibbs, Jr. and Herbert L. Colston

schemas. Our aim in this paper is to describe some of the findings from psycho-
linguistics, cognitive psychology, and developmental psychology that, in our
view, support the claims of cognitive semantics about image schemas and their
transformations.
There are two important reasons for considering this psychological evidence.
First, cognitive linguists, following the cognitive commitment to construct theories
that are consistent with what is known about the mind and brain (Lakoff 1990,
1993), should be aware of the experimental findings from neighboring disciplines,
especially data that bear on the possible connections between perception, thought,
and language. Second, psychologists are sometimes skeptical about theoretical
notions from linguistics that are primarily based on an individual analyst’s intu-
itions about linguistic structure and behavior. One of the main reasons for con-
ducting experiments with large groups of people is to minimize the uncertainty in
making inferences about thought and behavior in whole populations of people.
We do not entirely agree with the skepticism of psychologists about the theo-
retical claims of cognitive linguists (e.g., Kennedy and Vervaeke 1993). Yet we
think there exist different kinds of empirical evidence from psychology that both
psychologists and cognitive linguists should be aware of regarding the importance
of image schemas in ordinary cognitive functioning. This paper describes some
of this evidence. We begin by first elaborating the notion of image schemas and
how they are transformed. We then review work from psycholinguistics that has
explicitly examined how image schemas motivate people’s understanding of word
meaning. The next section of the paper describes work from cognitive psychology
that seems quite consistent with claims for the importance of image schemas in
everyday cognition. We then review work from developmental psychology that
also supports the cognitive reality of image schemas. The final section discusses
the significance of the different work from psychology for future studies in cog-
nitive linguistics.

1. Image schemas and their transformations

Image schemas can generally be defined as dynamic analog representations of

spatial relations and movements in space. Even though image schemas are derived
from perceptual and motor processes, they are not themselves sensorimotor pro-
cesses. Instead, image schemas are “primary means by which we construct or
constitute order and are not mere passive receptacles into which experience is
poured” (Johnson 1987: 30). In this way, image schemas are different from the
notion of schemata traditionally used in cognitive science, which are abstract
conceptual and propositional event structures (see Rumelhart 1980). By contrast,
image schemas are imaginative and nonpropositional in nature and operate as
 Chapter 7: Image schema 241

organizing structures of experience at the level of bodily perception and move-

ment. Image schemas exist across all perceptual modalities, something that must
hold for there to be any sensorimotor coordination in our experience. As such,
image schemas are at once visual, auditory, kinesthetic, and tactile.
We can illustrate what is meant by the notion of image schema, and how its
internal structure is projected onto new domain via metaphor, by considering the
BALANCE¬schema (Johnson 1987). The idea of balance is something that is learned
“with our bodies and not by grasping a set of rules” (Johnson, 1987: 74). Balanc-
ing is such a pervasive part of our bodily experience that we are seldom aware of
its presence in everyday life. We come to know the meaning of balance through
the closely related experiences of bodily equilibrium or loss of equilibrium. For
example, a baby stands, wobbles, and drops to the floor. It tries again and again, as
it learns how to maintain a balanced erect posture. A young boy struggles to stay
up on a two-wheeled bicycle as he learns to keep his balance while riding down
the street. Each of us has experienced occasions when we have too much acid in
our stomachs, when our hands get cold, our heads feel too hot, our bladders feel
distended, our sinuses become swollen, and our mouths feel dry. In these and
numerous other ways we learn the meanings of lack of balance or equilibrium. We
respond to imbalance and disequilibrium by warming our hands, giving moisture
to our mouths, draining our bladders, and so forth until we feel balanced once
again. Our BALANCE¬image schema emerges, then, through our experiences of
bodily equilibriums and disequilibriums and of maintaining our bodily systems
and functions in states of equilibrium. We refer to these recurring bodily experi-
ences as image schemas to emphasize means of structuring particular experiences
schematically so that we can give order and connectedness to our perceptions and
conceptions (Johnson 1987).
One of the most interesting things about image schemas is that they motivate
important aspects of how we think, reason, and imagine. The same image schema
can be instantiated in many different kinds of domains because the internal
structure of a single schema can be metaphorically understood. Our BALANCE¬
image schema, to continue with this example, is metaphorically elaborated in a
large number of abstract domains of experience (e.g., psychological states, legal
relationships, formal systems) (Johnson 1991). In the cases of bodily and visual
balance, there seems to be one basic scheme consisting of a point or axis around
which forces and weights must be distributed so that they counteract or balance
off one another. Our experience of bodily balance and the perception of balance
is connected to our understanding of balanced personalities, balanced views,
balanced systems, balanced equilibrium, the balance of power, the balance of
justice, and so on. In each of these examples, the mental or the abstract concept
of balance is understood and experienced in terms of our physical understand-
ing of balance. Image schemas have internal logic or structure that determine
242 Raymond W. Gibbs, Jr. and Herbert L. Colston

the roles these schemas can play in structuring various concepts and in patterns
of reasoning. It is not the case that a large number of unrelated concepts (for the
systematic, psychological, moral, legal, and mathematical domains) all just hap-
pen to make use of the same word balance and related terms (Johnson 1991).
Rather, we use the same word for all these domains because they are structurally
related by the same sort of underlying image schemas, and are metaphorically
elaborated from them.
Image schemas do not simply exist as single entities, but are often linked
together to form very natural relationships through different image schema trans-
formations. Image schema transformations have been shown to play a special
role in linking perception and reason. Among the most important image schema
transformations are (Lakoff 1987: 443):

(a) Path-focus to end-point focus: Follow, in imagination, the path of

a moving object, and then focus on the point where it comes to rest, or
where it will come to rest.
(b) Multiplex to mass: Imagine a group of several objects. Move away (in
your mind) from the group until the cluster of individuals start to become
a single homogeneous mass. Now move back down to the point where the
mass turns once again into a cluster.
(c) Following a trajectory: As we perceive a continuously moving object,
we can mentally trace the path it has traversed or the trajectory it is about
to traverse.
(d) Superimposition: Imagine a large sphere and a small cube. Increase
the size of the cube until the sphere can fit inside it. Now reduce the size
of the cube and put it within the sphere.

Each image schema transformation reflects important aspects of our visual, audi-
tory, or kinesthetic bodily experience. To illustrate, consider how these transfor-
mations might apply to our earlier example of the image schema for balance or
equilibrium. A situation where several of these transformations interact with the
balance image schema is that of handling a group of animals. In order to success-
fully control and navigate a large number of animals, cattle or sheep perhaps, one
needs to maintain the cohesiveness of the group. If a portion of the herd begins to
drift apart from the whole, an instance of the Multiplex to mass transformation,
equilibrium has been lost and action must be taken to restore it. Such a corrective
action requires that the path of the drifters be ascertained, following a trajectory,
and that their destination be determined and “headed off”, path-focus to end-point
focus. There are many examples like this that illustrate the role of image sche-
mas and different transformations in structuring our understanding of real-world
phenomena. We will consider other instances of image schema transformations
 Chapter 7: Image schema 243

as demonstrated in several studies from cognitive and developmental psychology.

But we will first consider some of the experimental evidence on the role of image
schemas in motivating people’s understanding of word meaning.

2. Psycholinguistics and image schemas

Consider the word stand in the following sentences: Please stand at attention.
He wouldn’t stand for such treatment. The clock stands on the mantle. The law
still stands. He stands six-foot five. The part stands for the whole and She had
a one-night stand with a stranger. These sentences represent just a few of the
many senses of stand that are common in everyday speech and writing. Some of
these senses refer to the physical act of standing (e.g., Please stand at attention,
The clock stands on the mantle, He stands six-foot five), while others have non-
physical, perhaps figurative, interpretations (e.g., We stood accused of the crime,
The part stands for the whole, He wouldn’t stand for such treatment). What are
the principles that relate the meanings of polysemous words? For instance, what
relates the different physical and nonphysical senses of stand in the examples
noted above?
Some linguists in recent years have argued that many polysemous words
resist being defined by a general, abstract, core sense (Brugman and Lakoff
1988; Fillmore 1982; Geeraerts 1993; Sweetser 1986). Cognitive linguists have
suggested that the meanings of polysemous words can be characterized by meta-
phor, metonymy, and different kinds of image schemas (Lakoff 1987; Johnson
1987; Sweetser 1990). Under this view, the lexical organization of polysemous
words is not a repository of random, idiosyncratic information, but is structured
by general cognitive principles that are systematic and recurrent throughout the
lexicon. Most important, perhaps, is the claim that these principles arise from our
phenomenological, embodied experience. One possibility is that bodily experi-
ence partly motivates people’s intuitions as to why different senses of stand have
the meanings they do.
Gibbs et al. (1994) attempted to experimentally show that the different senses
of the polysemous word stand are motivated by different image schemas that arise
from our bodily experience of standing. Their general aim was to empirically
demonstrate that the meanings of the polysemous word stand are not arbitrary
for native speakers, but are motivated by people’s recurring bodily experiences
in the real world.
As a first step toward understanding how image schemas partly motivate the
meanings of the polysemous word stand, a preliminary experiment sought to
determine which image schemas best reflect people’s recurring bodily experiences
of standing. A group of participants were guided through a brief set of bodily
244 Raymond W. Gibbs, Jr. and Herbert L. Colston

exercises to get them to consciously think about their own physical experience
of standing. For instance, participants were asked to stand up, to move around,
bend over, to crunch, and to stretch out on their tip-toes. Having people actually
engage in these bodily experiences facilitates participants’ intuitive understand-
ings of how their experience of standing related to many different possible image
schemas. After this brief standing exercise, participants then read brief descrip-
tions of 12 different image schemas that might possibly have some relationship
to the experience of physical standing (e.g., VERTICALITY¬BALANCE¬RESISTANCE¬
ENABLEMENT¬CENTERPERIPHERY LINKAGE ¬Finally, the participants rated the
degree of relatedness of each image schema to their own embodied experience
of standing. The results of this first study showed that five image schemas are
primary to people’s bodily experiences of standing (i.e., BALANCE¬VERTICALITY¬
CENTERPERIPHERY¬RESISTANCE, and LINKAGE).
A second experiment investigated people’s judgements of similarity for dif-
ferent senses of stand. The participants sorted 35 different senses of stand into
five groups based on their similarity of meaning. An analysis of these groups
revealed that participants did not sort physical senses of stand separately from
the nonphysical or figurative senses. For example, the physical idea of standing
in to stand at attention was often grouped with the metaphorical senses of stand
in let the issue stand and to stand the test of time.
The third experiment in this series examined the relationship between the five
image schemas for the physical experience of standing and the various senses of
stand studied in Experiment 2. Once again, participants were first asked to stand
up and focus on different aspects of their bodily experience of standing. As they
did this, the participants were presented with verbal descriptions of the five image
schemas BALANCE¬VERTICALITY¬CENTERPERIPHERY¬RESISTANCE¬and LINKAGE¬
Afterwards, the participants were given a list of 32 senses of stand and asked to
rate the degree of relatedness between each sense and the five image schemas.
The rating data from this third study allowed Gibbs et al. (1994) to construct an
image schema profile for each of the 32 uses of stand. Several interesting similari-
ties emerged in the image schema profiles for some of the 32 senses of stand. For
example, it stands to reason and as the matter now stands both have the same image
schema profile (in their rank-order of importance) of LINKAGEnBALANCEnCENTER
PERIPHERYnRESISTANCEnVERTICALITY¬The expressions don’t stand for such treat-
ment and to stand against great odds are both characterized by the image schema
profile RESISTANCEnCENTERPERIPHERYnLINKAGEnBALANCEnVERTICALITY.
The primary goal of this study, though, was to assess whether the senses of
stand seen as being similar in meaning in the second experiment were reliably
predictable from the image schema profiles obtained in this study. Statistical
analyses showed that knowing the image schema profiles for different senses of
stand allowed us to predict 79% of all the groupings of stand in Experiment 2.
 Chapter 7: Image schema 245

These data provide very strong support for the hypothesis that people’s under-
standings of the meanings of stand are partly motivated by image schemas that
arise from their bodily experiences of standing. A fourth study showed that par-
ticipants’ sortings of stand in different groups cannot be explained simply in
terms of their understanding of the contexts in which these words appeared. Thus,
people did not sort phrases, such as don’t stand for such treatment and to stand
against great odds, because these phrases refer to the same types of situations.
Instead, it appears that people’s similarity judgments are best attributed to their
tacit understanding of how different patterns of image schemas motivate different
uses of the polysemous word stand.
This psycholinguistic research has demonstrated that people make sense of
different uses of stand because of their tacit understanding of several image
schemas that arise partly from the ordinary bodily experience of standing. These
image schemas, the most important of which are BALANCE¬VERTICALITY¬CENTER
PERIPHERY¬RESISTANCE and LINKAGE, not only produce the grounding for many
physical senses of stand (e.g., he stands six-foot nine, stand in the