You are on page 1of 8

Page 1 of 8

1,995 words (Limit 2,000)


John Newman
University of Alberta
Department of Linguistics, University of Alberta, Edmonton, AB T6G 2E7, Canada
(780) 492 5500
john.newman@ualberta.ca

COGNITIVE GRAMMAR

Cognitive Grammar (CG) refers to the theory of language articulated most

comprehensively in Langacker (1987, 1991), two mutually dependent volumes which are

best read together. Langacker (1988) provides a succinct chapter-length overview of his

theory, while Taylor (2002) and Evans and Green (2006, 553-640) are highly

recommended as student-oriented introductions to the theory. CG is wide-ranging in its

scope and provocative in its approach to understanding linguistic structure. It has played

a key role in the history of COGNITIVE LINGUISTICS.

Fundamental to CG is the idea that language is an integral part of human

cognition and cannot be properly understood without reference to cognitive abilities. A

pervasive feature of CG is the determination to reconcile accounts of linguistic structure

with what is known about cognitive processing in domains other than language. CG

contrasts in this respect with models which insist upon a discrete, autonomous “grammar

module” and the AUTONOMY OF SYNTAX. The cognitive orientation of CG is apparent

from a reliance on notions such as sensory imagery, perspective, mental scanning,

attention, and figure vs. ground asymmetry in accounting for linguistic phenomena. In

broad terms, grammatical structure is explained as conventional imagery, with alternate

structures reflecting alternate construals of the conceived situation. Not surprisingly, the
Page 2 of 8

cognitive notions assume a relatively abstract interpretation when applied to some aspects

of linguistic structure. For example, cognitive processes such as registration of contrast,

scanning of a field, and perception of a boundary are all deemed relevant to explicating

the notion of a count noun, understood as a “bounded region in some domain” in

Langacker (1987, 189-203). Such processes may be obvious factors in the

conceptualization of nouns with clear spatial boundaries (e.g., cup, pencil), but a more

abstract interpretation of these processes is clearly required in other domains. Body part

nouns (e.g., waist, shoulder, side), for example, must be explicated in terms of a “virtual

boundary” which does not correspond to any visible, objectively identifiable

demarcation. Likewise, the notions of figure and ground familiar from the study of

perception are seen as underpinning various relational asymmetries in language. These

notions have most obvious relevance in the case of words relating to the spatial domain,

such as the contrasting pair above and below, where there is a kind of figure-ground

reversal of the conceptual reference point. The terms “trajector” (an extension of the

notion of figure) and “landmark” (an extension of the notion of ground) are used to refer

to the specifically linguistic manifestation of the perceptual notions of figure and ground,

such that the book is the trajector and the table is the landmark in the book under the

table. Conversely, the table is the trajector and the book is the landmark in the table over

the book. More abstractly still, the traditional syntactic contrast between subject and

object is construed as a very specific instance of the trajector vs. landmark contrast.

At the heart of CG is the concept of a symbolic unit, consisting of semantic

structure standing in correspondence to a phonological structure. Consistent with the idea

that language is part of conceptual structure, semantic structure is understood as


Page 3 of 8

“conceptualization tailored to the specifications of linguistic convention” (Langacker

(1987, 99); see Talmy (2000, 4) for a similar view of semantic structure). CG takes the

notion of symbolic unit (similar to, but not to be equated simply with, the Saussurean

SIGN) as fundamental and applicable at all levels of representation, including lexical

items, grammatical classes and grammatical constructions. The lexical item tree, for

example, consists of a semantic unit [TREE] and a corresponding phonological unit [trɪ]

which combine to form the symbol for tree, [[TREE]/[trɪ]]. The same apparatus is

applicable to defining a word class such as a noun, [[THING]]/[…]], where both semantic

and phonological units (or “poles”) are schematic. A more complex lexical item such as

trees is represented as a composite structure integrating two symbolic units representing

the noun tree and the plural [-z]: [[[TREE]/[trɪ]]-[[PL]/[z]]]. Grammatical constructions are

in principle no different from a lexical item like trees in terms of the descriptive

apparatus required to capture all the relevant detail, with each of the component

structures of a construction represented by a symbolic unit. Grammatical morphemes

appearing in a construction, such as of, are treated as symbolic units in their own right,

with semantic structure (of, for example, specifies a part-whole relation). The integration

of any two symbolic units goes hand in hand with distinguishing the dependent and

autonomous parts of the composite structure. As far as semantic structure is concerned,

[TREE] is autonomous, while [PL] is dependent, requiring an elaboration by a noun to

complete the structure. In terms of phonological structure, [trɪ] is pronounceable as a

whole syllable and can be considered autonomous, while the single consonant [z] is

dependent.
Page 4 of 8

A striking feature of CG is the detail provided for in the integration of structures

into larger composite structures. The analysis of the English passive construction in

Langacker (2001, 101-147) illustrates the theoretical notions relevant to a detailed

grammatical description and is recommended as a prime example of a full-blown CG

account of a construction type. Briefly, and consistent with the foregoing remarks, each

morpheme in the passive (including by and the auxiliary verbs) has its own symbolic

representation, giving rise to the overall semantic structure, just as the active counterpart

has its own compositional structure and resulting semantic structure. Passive clauses do

not derive from active clauses in this view; nor do they derive from some abstract

structure underlying actives and passives. Rather, passive clauses exist in their own right

as instantiations of a construction type with its own distinctive way of integrating

symbolic units, reflecting a particular construal of the event.

While phonological structure can be fruitfully explored within CG (see Langacker

(1987, 328-348, 388-40) and Taylor (2002, 78-95)), it is semantic structure which has

received most attention and for which most theoretical apparatus has been developed.

Fundamental to semantic structure is the idea of a network which is employed to

represent polysemy relationships and to provide motivation for conventional and novel

extensions. Each node of the semantic network, together with the associated phonological

structure, represents a semantic variant of the lexical item. Two types of relationships

figure prominently in these networks: schematicity and extension. The word head, for

example, can be assigned a sense [PART OF A WHOLE WHICH CONTROLS THE BEHAVIOR OF

THE WHOLE] which is schematic relative to finer-grained elaborations such as [PART OF


Page 5 of 8

THE HUMAN BODY WHERE THINKING IS LOCATED ] and [PERSON WHO MANAGES AN

ADMINISTRATIVE UNIT]. In some cases, a highest-level node or “superschema” can be

proposed, encompassing all lower-level senses in the network, though such superschemas

are not feasible for every network. The extensive polysemy of head, for example, makes

one single superschema covering such diverse senses as ‘head of a lettuce’, ‘head of a

bed’, ‘head of a golf club’ etc. unlikely. Semantic extension holds between the more basic

sense of ‘human head’ and the sense of ‘head of an administrative unit’. The node which

is the source of the extension constitutes a “local” PROTOTYPE (with respect to the

extended sense); where one node is experienced as representative of the whole category,

as is likely in the case of the ‘human head’ sense of head, we speak of a “global”

prototype. There is clearly variation among speakers in their judgments about nodes and

relationships within the network, including their ability to identify relatedness of senses

and to extract schematic meanings. This variation poses challenges for description, but

does not negate the need to acknowledge the reality of such networks.

CG adopts a non-reductionist or “maximalist” stance in its analysis of linguistic

structure, contrasting with prevailing reductionist, minimalist approaches in

contemporary linguistics. The non-reductionist approach of CG explicitly provides for the

listing of highly specific patterns alongside the statement of more general patterns, rather

than recognizing only the most general rules and schemas. The existence of a general

pattern of plural formation in English with, say, a suffixed /z/, does not obviate the need

to recognize the more specific patterns of plural formation which hold between any

particular singular/plural pair, even when the plural formation is fully regular, such as

dog/dogs, cat/cats, horse/horses, etc. Acknowledging low-level, highly specific patterns


Page 6 of 8

runs counter to deeply entrenched practices in contemporary linguistics which has been

preoccupied with higher-level generalizations and the principle of economy in

description. Langacker has repeatedly emphasized the desirability of both general and

particular statements in linguistic description, referring to the assumption that a

phenomenon is to be accounted for in a mutually exclusive way as either a rule or a list as

the “rule/list fallacy” (Langacker 1997, 40-42). Grammar, in CG terms, amounts to “a

structured inventory of conventional linguistic units” (Langacker 1987, 73). The units, so

conceived, may be semantic or phonological; they range from the symbolic units

consisting of a single morpheme to larger composite symbolic units at the clause level;

they include highly specific, as well as highly schematic units. This conception of

grammar makes CG comparable to CONSTRUCTION GRAMMARS, which are also

“inventory-based” (cf. Evans and Green 2006, 475-483), particularly Radical

Construction Grammar (Croft 2001).

By including quite specific patterns within a grammatical description, CG is able

to comfortably accommodate phenomena which have been largely neglected in linguistic

theorizing, e.g., the collocational patterning of great idea, absolutely fabulous etc.

involving combinations of particular words. The greater emphasis on specific patterning

makes CG highly compatible with the methodology of CORPUS LINGUISTICS and

other approaches which focus on LANGUAGE IN USE whereby actual usage, including

frequency of occurrence and patterns of co-occurrence, can be observed and used as a

basis for extracting patterns of varying generality. Fully general, exceptionless rules are

seen as atypical and, while it is valid to seek out such rules, it would be misguided in this

approach to attend only to the most general patterns.


Page 7 of 8

Finally, a word on notation employed in CG. There is an array of notational

devices used by Langacker who employs a distinctive and highly original, geometric

style of representation (in his earlier publications, Langacker used the term “Space

Grammar” to refer to his approach). To some extent, the notation is intuitive: a circle is

used to denote a [THING] entity; thicker, darker lines represent the “profile”, i.e., the

designated thing or relation in the semantic structure of a morpheme. A full appreciation

of the notation, however, requires careful study. Of course, not all the detail needs to be

represented all the time and CG ideas can be effectively incorporated into linguistic

analyses simply in prose or with a minimum of notation (as in Taylor 2002).

--John Newman

Works Cited and Suggestions for Further Reading

Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological

Perspective. Oxford: Oxford University Press.

Evans, Vyvyan and Melanie Green. 2006. Cognitive Linguistics: An Introduction.

Edinburgh: Edinburgh University Press.

Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. I: Theoretical

Prerequisites. Stanford: Stanford University Press.

Langacker, Ronald W. 1988. “An Overview of Cognitive Grammar.” In Topics in

Cognitive Linguistics, ed. Brygida Rudzka-Ostyn, 3-48. Amsterdam and Philadelphia:

John Benjamins.
Page 8 of 8

Langacker, Ronald W. 2001. [1st ed. 1991]. Concept, Image, and Symbol: The Cognitive

Basis of Grammar. Berlin and New York: Mouton de Gruyter. The chapters in this

volume cover key areas of grammar (grammatical valence, case, passive etc.) and can

be read more or less independently of one another - a good balance between CG

theory and application to data.

Langacker, Ronald W. 1991. Foundations of Cognitive Grammar. Vol. II: Descriptive

Application. Stanford: Stanford University Press.

Newman, John. 2004. “The Quiet Revolution: Ron Langacker’s Fall Quarter 1977

Lectures.” In Imagery in Language: Festschrift in Honour of Professor Ronald W.

Langacker, ed. Barbara Lewandowska-Tomaszczyk and Alina Kwiatkowska, 43-60.

This chapter gives a first-hand account of an early presentation of the material which

eventually became Langacker (1987).

Talmy, Leonard. 2000. Toward a Cognitive Semantics. Vol. 1: Concept Structuring

Systems. Cambridge, Mass.: MIT Press.

Taylor, John. 2002. Cognitive Grammar. Oxford: Oxford University Press.

You might also like