Cognitive approaches to grammar
Radical Construction Grammar
Discussion
Radical Construction Grammar’s assumptions are very similar to those made in Cognitive Grammar. Indeed, they are radical especially if compared to Goldberg’s Construction Grammar, rather than Cognitive Grammar. This observation may lead us to view Radical Construction Grammar as a notational variant of Cognitive Grammar applied to typological research (since the main focus of Croft 2001 is typology). Nevertheless, Langacker (2005a) argues that there are some substantive differences between the two models.
The main difference between the two grammars (see also Broccias 2003) resides in the fact that grammatical categories are given a schematic characterization in Cognitive Grammar whereas they are claimed to be construction and language-specific in Radical Construction Grammar. Radical Construction Grammar, at most, offers a prototype description of parts of speech. By contrast, Cognitive Grammar claims that a schematic definition (hence, a definition valid for all cases) can be offered for both parts of speech (see Figure 1 in Word classes and grammatical functions in Cognitive Grammar) and (at least some) grammatical relations (i.e. in terms of trajector/landmark alignment). As was pointed out, the schematic characterization provided in Cognitive Grammar is not usage-driven but is advanced as a matter of conceptual necessity, i.e. it stems from our cognitive abilities. This amounts to saying that Cognitive Grammar, unlike Radical Construction Grammar, is, at the same time, a usage-based or bottom-up model and a universalist or top-down model. Like Radical Construction Grammar, Langacker views grammar as including constructions of varying degrees of specificity and complexity. But, as in non-cognitive models, universal categories are assumed to exist.
If the universalist stance adopted in Cognitive Grammar may be reminiscent of traditional and formal treatments in scope (but not in its implementation, since universal or schematic categories are based on cognitive processing in Cognitive Grammar), Radical Construction Grammar prototypical approach to words classes is not “immune” from “traditional” influences either. As was pointed out, the Radical Construction Grammar analysis rests on the interaction between propositional acts and semantic classes. Crucially, propositional acts seem to involve a rather objectivist view of semantics/pragmatics. For example, the act of predication is defined, quite traditionally, as the ascription of something to a referent (Croft 2001: 66) and the verb is taken to prototypically code such a role. However, the view that verbs prototypically code acts of predication may be debatable.1 It remains to be demonstrated quantitatively and cognitively that this is indeed the case. Lambrecht (1994), for example, reminds us that a sentence like The children went to school does not necessarily predicate the event of going to the school of the subject (i.e. this is what he calls a topic-comment sentence or predicate-focus structure) but has a variety of different uses. It can also identify who did the action of going to school (if it is used as an identificational sentence or argument-focus structure, i.e. to answer the question Who went to school?) and, more generally, what happened (if it is used as an event-reporting or argument-focus structure, i.e. to answer the question What happened?).
Langacker’s Cognitive Grammar dispenses with the traditional analysis of verbs as involving the act of predication and views verbs as relations (scanned sequentially). In this way, he can capture the intuition that the human ability to establish links between entities (i.e. what relations are about) is shared by verbs and (Croft’s unmarked) adjectives. The two only differ in their temporal profile: the former designate temporal relations; the latter profile atemporal relations. By contrast, (unmarked) adjectives are linked to a different propositional function in Croft’s theory. In fact, the very distinction between predication and modification (within a Croftian view) may be an intricate affair (see Broccias 2003).2 Nevertheless, it was pointed out earlier that Langacker’s analysis is also open to debate since it involves the controversial notion of scanning. Further, Langacker’s view that verbs always activate a relational content should also be the matter of empirical investigation (is for example an action verb like sleep (always) represented in our mind as a relation, akin, one may speculate, to “to be asleep”?).
Alongside grammatical categories, a further difference between Radical Construction Grammar and Cognitive Grammar involves the use of the term “syntactic representation”, which seems to be employed in Croft’s theory quite similarly to Construction Grammar. Croft subscribes to the Cognitive Grammar and Construction Grammar view of constructions as (partially arbitrary) symbolic units (or pairings of form and meaning). Rather controversially, however (see also Broccias 2003), he views form as including syntactic, morphological and phonological properties. This, as was pointed out for Construction Grammar, is potentially problematic because syntax in a cognitively oriented theory of language does not exist as an independent grammatical level. (Remember that Croft himself denies the existence of syntactic relations because they are taken to be independent of meaning.) Nevertheless, since the formal pole of constructions is said to include syntactic, morphological and phonological properties (and not only syntactic properties as in Construction Grammar), one may speculate that Croft’s use of the term “syntax” may just be a convenient label to refer to word order patterns in a construction.
All in all, a fundamental contribution of Radical Construction Grammar has not only been to have shifted the focus of scholarly research from lexemes to constructions (in the traditional sense) but, most of all, to have underlined the need for a truly usage-based or bottom-up model in language description.
1. Such a view derives from the Aristotelian analysis of copular sentences like Socrates is mortal as involving the ascription of a quality (e.g. that of being mortal) to the subject referent. This analysis is then extended to non-copular sentences (e.g. The children went to school). But observe (see Table 2 in Why Radical Construction Grammar is radical) that, paradoxically (with respect to the history of the notion of predication), non-copular sentences are not treated as the prototypical constructions coding the propositional act of predication in Croft’s theory.
2. My understanding of Croft’s analysis is that a clear-cut distinction between the three types of discourse functions and the three types of semantic classes is possible. Indeed, the distinction between predication and modification may seem intuitively straightforward. Consider, however, a sentence like John lives poorly: what is the function of poorly in this sentence? Is it predicating or modifying?
