Let us consider the example of the Catalan word pesar, that has multiple functions and meanings:

(3)
a) Joan volia pesar el peix ell mateix / John wanted to weigh the fish himself.
b) La notícia li va causar un gran pesar / The news caused him great sadness.
c) A pesar de la notícia, va continuar el que estava fent / In spite of the news, she continued with what she was doing.

The verb pesar (‘to weigh’) is the original form, the literal sense of which is "to have a particular weight"'. The abstract noun (el) pesar (‘sadness’, ‘grief’) derives from the verb, and represents a change of category plus a modification of the meaning based on a metaphor: WHAT IS SAD "IS HEAVY", "IS DEPRESSING", "WEIGHS US DOWN". The noun is the basis for the preposition or conjunction a pesar de/a pesar que (‘in spite of’) where in it becomes concessive, the equivalent of though. This too can be explained metaphorically on the grounds that HEAVY THINGS ARE A BURDEN, A DIFFICULTY; thus the idea of 'difficulty overcome' is the origin of the concessive meaning developed by the collocation a pesar de/que.

These grammatical changes have to do necessarily with the expressive needs that are conventionalised in discourse: the repetition of new forms in similar or identical discourse contexts activate a conversational implicature than can become a conventionalised implicature and then is incorporated into the grammar. The same happens with a number of conjunctions, like mentre que (‘while’). This conjunction originally had a purely temporal meaning (simultaneity) which, by being combined with contexts in which there was a contrast between the linked elements, has taken on a contrastive meaning:

(4)
a) Mentre tu havies sortit, ha trucat algú (temporal = when) / While you were out someone phoned.
b) Jo no he sortit en tot el dia, mentre que tu encara no has entrat (contrastive = and, in contrast; however) / I haven't been out all day, while you haven't come in yet.

It is worth noticing that similar processes can be observed in other languages, as with the marker cependant in French or the English marker still, which derive from forms that indicate temporal simultaneity and have come to indicate contrast. This process implies changes from a more concrete notion (time) to a more abstract notion (contrast) which entail greater involvement of the speaker in the message being transmitted.

Grammaticalisation theory thus becomes a framework for the study of language variation, in "that part of the study of language that focusses on how grammatical forms and constructions arise, how they are used and how they shape the language" (Hopper & Traugott 1993: 1). In this process, pragmatic and discourse factors, and even extra-linguistic factors, play a major role. As Heine et al. (1991: 23-24) point out, "Grammaticalization is initiated by forces that are located outside language structure", somewhere between language per se and the outside or "real" world: the world "of experience". Thus, the concept of grammaticalisation takes on a cognitive and pragmatic dimension, and can be seen as a basic mechanism for understanding the nature of language and the way in which the variation within languages and between languages works.

In fact, if structuralism and generative grammar seeks to reveal the immutable principles of language, cognitive linguistics sets out to understand diversity and to show that there is more systematicity and order in the differences than had been thought until now. Hudson (1997) recalls the concept of inherent variability coined by Labov, in order to explain "the coexistence of alternative ways of saying the same thing within the speech of a single speaker who alternates between them in a statistically regular way" (1997: 73). Hudson observes that Sociolinguistics generally studies the diversity of usage, but it has not entered into the study of grammar and it has not influenced the (synchronic) theories of language. He adds that "it is hard to think of a single example (until very recently) where statistical data on inherent variability has been used as evidence in discussions of language structure" (Hudson 1997: 73). Cognitive linguistics aims at relating variants and invariants, system and use.

As can be seen, the common denominator in many of the examples that have been looked at is the importance given to the speaker (at the individual level and above all at the social level) in grammatical structure and the selection of forms. Cognitive linguistics introduces a dynamic element into the study of grammar associated with the speakers’ involvement in the messages which he or she emits. In the lines of Ungerer and Schmid (1996: xi-xii), the speaker’s involvement (subjectivisation) can be developed by three principles: the experiential perspective, the prominence perspective and the attention perspective of language.

a) According to the experiential perspective, the definition of words and other structures is not objective and is not only based on logical rules, since words and constructions involve associations of meaning and impressions that are part of the experience of the speaker.

b) The prominence perspective recognises the fact that not all associated elements have an equal value and that this difference in value can be translated into formal syntactic differences.

c) The attention perspective accounts for the fact that the parts of an event which attract our attention are more important cognitively and this is reflected by linguistic structure. This principle, then, complements and reformulates the previous one.

The example of the passives and the actives previously commented on exemplifies the way in which the attention perspective works. The following examples illustrate the prominence perspective.

(5)
a) La casa és davant del col·legi / The house is in front of the school.
b) El col·legi és darrere de la casa / The school is behind the house.

(6)
a) Juan s'assembla a Pere / John resembles Peter.
b) Pere s'assembla a Juan / Peter resembles John.

Prominence explains how information in a sentence is selected and organised, and justifies the differences between version (a) and (b) in the preceding pairs of examples.

These differences cannot be attributed to semantic factors associated with truth conditions, since both versions are identical in this respect. The same holds if we look at their syntactic structure: both sentences have the same categorial configuration (noun phrase ser/assemblar-se preposition + noun phrase) and the same functional configuration (subject – verb – attribute/complement). The only apparent difference is the change of places between the noun phrases. However, each version reflects a different standpoint adopted by the speaker or writer, a difference in prominence (figure-ground relationship) of the entities referred to, similarly to the famous example of the bottle, containing the same quantity of liquid, which is said to be half empty or half full.

In some cases, the differences of prominence and experiential perspective may even justify the acceptability or non-acceptability of sentences that should be synonymous as in (7):

(7)
a) Maria s'assembla a la seva mare / Mary looks like her mother.
b) ? Marta s'assembla a la seva filla / ?Marta looks like her daughter.
c) ?? La mare de Maria s'assembla a la seva filla / ??Mary's mother looks like her daughter.

The fact that the second and third versions of sentence (7a) are odd, and in some contexts might not be acceptable, cannot be derived from their syntax or from differing truth conditions –understood as logical propositions they mean the same and, in fact, (7a) implies (7b). Rather, this effect derives from a difference of prominence and according to our knowledge-based experience: when looking for family resemblances –as it is often the case with newborn children– the perspective of the parent and not the child is generally adopted (the former is more prominent than the latter). Parents constitute a better reference point than children given the directionality of the relationship: the parents are the origin of their offsprings and not the reverse. This explains why (7a) is more acceptable than (7b) or (7c) in neutral context conditions.

Similarly, only our encyclopaedic knowledge of the world, linked to a culture and to prototypical or "idealised" social relations, can explain the asymmetries exhibited by (8) and (9), analysed by Hilferty and Valenzuela (2001):

(8)
a) Té marit / She has a husband.
b) Té marits / She has husbands.

(9)
a) No té fills / She doesn't have children.
b) No té fill / She doesn't have child.

The sentences in (8) and (9) are well-formed; nonetheless, the sentence Té marits ('She has husbands') is odd, as does No té fill ('She doesn't have child') or, rather, we are obliged to reinterpret them by creating a special context in which they are adequate. It is our knowledge of the world, cultural conceptions –that may not coincide with an external situation which is changing rapidly– that explains the asymmetry. In our culture, an idealised cognitive model operates whereby families consist of a man, a woman and more than one child (no matter how reality might show that in an increasing number of cases this is not so). Probably, as the model changes, the grammar will reflect this and allow for more structural possibilities. In point of fact, (9b), which would not be easily interpreted forty years ago, at least in Spain, is now interpretable in the context of someone who is divorced, referring to the fact that on that particular day he or she does not have to take care of the child and can go out in the evening.

In conclusion, grammar, pragmatics and cognition are interrelated in language use. Thus, from the cognitive point of view, grammar can be characterised as an entity in continuing evolution, "a set of cognitive routines, which are constituted, maintained and modified by language use" (Langacker 1987: 57). Many facts which are considered part of language use or performance can be integrated into system or competence, and at the same time the dividing line between diachrony and synchrony disappears.

Assuming this view, variation and the idiomatic or idiosyncratic aspects of languages are not seen as secondary, but, on the contrary, they constitute the real, central object of linguistics, the entering wedge for discovering the invariant, the system viewed as a living entity, an entity which takes shape and evolves through use, through the speakers as members of a group sharing a culture and a vision of the world.

CUENCA, M. J. "La gramaticalització com a teoria de la variació morfosintàctica". M. Lorente et al. (ed.). La gramàtica i la semàntica per a l'estudi de la variació. Barcelona: PPU-Secció de Lingüística Catalana de la Universitat de Barcelona. Col·lecció Lingüística Catalana, 5, p. 151-185, 2001.

CUENCA, M. J. & J. HILFERTY. Introducción a la lingüística cognitiva. Barcelona: Ariel, 1999.

GEERAERTS, D. "How does Cognitive Linguistics fit in the development of Linguistics". A. Soares da Silva (ed.). Linguagem, Cultura e Cognição: Estudos de Linguística Cognitiva. Coimbra, Almedina, [in press], 2004.

HEINE, B., CLAUDI, U. & F. HÜNNEMEYER. Grammaticalization: A Conceptual Framework. Chicago: University of Chicago Press, 2 vols., 1991.

HILFERTY, J. & J. VALENZUELA. "Maximality and idealized cognitive models: the complementation of Spanish tener". Language Sciences, 2001, no. 23, p. 629-63.

HOPPER, P. & E. C. TRAUGOTT. Grammaticalization. Cambridge: Cambridge University Press, 1993.

HUDSON, R. "Inherent variability and linguistic theory". Cognitive Linguistics. (1997), no. 8.1, p. 73-108.

LAKOFF, G. Women, Fire and Dangerous Things. What Categories Reveal about the Mind. Chicago: University of Chicago Press, 1987.

LANGACKER, R. Foundations of Cognitive Grammar. vol. I: Theoretical Prerequisites. Stanford (California): Stanford University Press, 1987.

UNGERER, F. & H.-J. SCHMID. An Introduction to Cognitive Linguistics. London/New York: Longman, 1996.

Maria Josep Cuenca  
Universitat de València
maria.j.cuenca@uv.es