Cognitive approaches to grammar
Conclusion
Although the present overview could not obviously do justice to its complex articulation, Langacker’s theory is by far the most comprehensive theory of grammar available in the cognitive linguistic camp. To some extent, other cognitive models such as Goldberg’s Construction Grammar, Radical Construction Grammar and Blending Theory (as applied to the investigation of linguistic structures) can be seen as notational variants of Cognitive Grammar. The similarities are numerous. They all share, among other things, the assumption that language must not be studied independently of general cognitive abilities. In fact, general cognitive abilities are taken to shape language. They also agree on the fact that language does not merely consist of a syntax plus a lexicon. Rather, language is viewed as a taxonomic hierarchy, where sharp distinctions should not be assumed; language is a “diffuse” network since much in language is a matter of degree. This means that linguistic units (or constructions) can be of any length and of any degree of complexity. Further, cognitive theories subscribe to the view that constructions are often ready-made pairings of form and meaning. We do not need to “generate” all their various parts every time we want to use them but they may be available to us automatically, as gestalts (i.e. we are often unaware of the contribution of their constitutive parts).
Still, important differences exist among the various theories. I have tried to point out that the nature of word classes and grammatical relations is still a matter of controversy. Similarly, the relation between constructional meaning and lexical meaning is still the topic of heated debate. Nevertheless, the theories briefly illustrated here alongside Cognitive Grammar serve a very important purpose if they are all given due consideration (especially) when specific phenomena are studied. They highlight certain facets of grammar (in the Langackerian sense) that may not have been worked out fully in the Cognitive Grammar model. This is most evident in the case of change constructions.
Blending Theory reminds us that when we examine complex expressions we should distinguish between two conceptual levels, the unblended (or unintegrated) level and the blended (or integrated) level. The latter can convey new meaning (with respect to the input spaces, cf. Langacker’s scaffolding metaphor mentioned in note 7), as is indeed the case with change constructions. An unintegrated, complex sequence of events is conceptualized as (or categorized by) a force-dynamic event. Thus, we construe the unintegrated causal sequence as implying instantaneous causation (cf. the discussion of I boiled the pan dry above).1
Goldberg’s analysis brings two important points to the fore. We should investigate the relation between the meaning of a construction and the meanings of its parts, e.g. the verb’s meaning in a change construction. In this respect, Langacker urges us to be cautious. The overlap between constructional and verbal meaning may be a matter of degree and, ultimately, we may add, the issue should be the object of psycholinguistic investigation. To be sure, in some cases, even if a verb is entrenched in a specific construction, this does not (at least intuitively) result in the verb’s acquiring a new sense.2 For example, if we consider the two constructions I laughed him out of patience and I laughed him into patience, we may find it intuitively odd to claim that laugh should be categorized as a verb of “removal” in the former case (since the object referent, metaphorically speaking, moved out of patience) and as a verb of “creation” in the latter case (since the object referent, metaphorically speaking, moved into a region of patience). Rather, we categorize laugh as a force-dynamic process by virtue of our knowledge that laughing can affect people surrounding us.
Goldberg’s other important point is that constructions can be polysemous. This is indeed the case with change constructions. Consider the examples in (3).
(3) a. She kissed the anxiety out of him.
b. The fans booed the team off the pitch at the interval.
c. The river froze solid.
d. She lifted the parcel up.
In (3a), kiss receives a force-construal. It is categorized as a force (because we know that kissing engenders some emotional response) causing anxiety to be removed from someone. Hence, (3a) is a prototypical example of a change construction.3 By contrast, (3b) shares its form with prototypical change constructions (cf. He was booed off the podium, which may have a causal interpretation) but only has a temporal interpretation, as the prepositional phrase at the end makes clear (see Broccias 2005b). The team would have abandoned the pitch even if the fans had not booed at them because it was half-time. The fans’ booing simply unfolded as the team left the pitch. Hence, (3b) shares with prototypical change constructions only unity of time, not (immediate) causation. (Remember that, at the blended level, change constructions construe the causing event and the caused event as obtaining one immediately after the other even if this is not the case in the “real world”.) (3c) can be analyzed as a change construction (i.e. the river became solid because it froze) but also shows that certain change phrases (e.g. solid) may be better analyzed as intensifiers. After all, if speakers understand freeze as “to become solid”, to say that “the river became solid because it froze (i.e. became solid)” is a tautology. Hence, solid may be interpreted as meaning something akin to “completely”, i.e. it intensifies the event denoted by the verb by specifying that the freezing affected all parts of the river. (3d) also illustrates redundancy of the change phrase (i.e. up) with respect to the verbal meaning. After all, lift already means to “move up”. But up does not obviously have an intensifying function. Rather, it seems to be used to make the spatial component implicit in lift explicit (since up explicitly refers to a spatial configuration, or at least this is taken to be the schematic meaning of up).
Croft’s approach is also of crucial importance. It stresses that grammar (hence constructions) should be studied using a bottom-up approach. In the specific case of change constructions, this means, for example, that it may be immaterial to try to distinguish sharply between caused motion constructions, on the one hand, and resultative constructions, on the other (hence the label “change construction” introduced by Broccias 2003). It may not be important to decide whether a construction like He opened the window wide is a caused motion construction (i.e. wide refers to a position) or a resultative construction (i.e. wide refers to a state). It may simply be an instantiation of either (cf. the idea of multiple inheritance, or multiple categorization). Indeed, it is a fallacy to assume that binary-categorization is always possible in the first place (or even useful, for that matter).
Further, Croft’s insistence on taxonomic hierarchies highlights the importance of specific (i.e. lexically-filled or lower-level) patterns for speakers. Nevertheless, I would like to add that the importance of higher-level schemas should not be discounted. They capture regularities across constructions (i.e. they allow us to orientate ourselves among specific linguistic instantiations) and are also important in language learning/acquisition (see also Tomasello 2003).
In sum, despite some (of course important) differences, the models illustrated in this overview shape a view of grammar that is largely unanimous. Grammar (in a broad sense) is best viewed as a structured inventory of conventional linguistic units. One of the next challenges for cognitive linguistics is to see how we can put this view into effect by relating it to the realm of applied linguistics.
1. The reader should observe that the present formulation is meant to be non-committal about the issue of whether the integrated sequence should be viewed as input 1 or just as a blend.
2. I am not suggesting that this is Langacker’s position (cf. especially Langacker 2005b). Rather, I am making a more general point concerning the usefulness of Goldberg’s view that a construction can impose a meaning (or construal) on a verb used in it (without the verb’s acquiring this meaning).
3. Incidentally, (3a) also shows that the “subcategorized” object of the verb (i.e. him, cf. Ske kissed him) is not always inherited as an object in the construction (although it shows up in the construction as the object of the complex preposition out of). The constructional object anxiety is not a possible object for the verb in isolation (cf. *She kissed the anxiety). This type of change construction is called an asymmetric construction by Broccias (2003).
