The prototypical grammatical construction
Of course, the four factors we have just discussed (correspondence, profile determinacy, autonomy versus dependence and constituency) are not independent properties of constructions. In particular, correspondence is closely linked to the properties of autonomy and dependence, which in turn give rise to constituency. These factors should therefore be viewed as interrelated aspects of what it means to define a grammatical relationship in terms of valence. Furthermore, Langacker argues that the four factors discussed here are not of equal importance to valence relations. He argues that correspondence is a central factor because it participates in every kind of valence relation. In contrast, not all composite grammatical structures have a readily identifiable profile determinant (consider the compound noun puppy dog, for example), and it is equally difficult to establish an autonomous and a dependent component in such constructions. For this reason, Langacker suggests that it is only meaningful to refer to a profile determinant in cases where there is a clear asymmetry between component structures: because puppy and dog each profile a THING and together profile a THING, it may not be useful to identify one or the other as profile determinant. Langacker (1987: 185) describes this type of valence relation in terms of apposition, which means that both component parts of a construction designate the same entity (see Taylor 2002: 235–8). Langacker also argues that constituency is not fundamental to valence because a given complex construction might be arrived at via various routes. In other words, the ‘order’ in which constructions are ‘built’ is not important in this model because of its usage-based nature. As we have seen, in Cognitive Grammar many complex constructions are stored as units, which in turn give rise to schemas. These schemas do not contain step-by-step ‘instructions’ for the composition of novel instances (for example, ‘build head-complement structures before adding modifiers’), because the instances give rise to the schema and not vice versa.
According to Langacker, the prototypical grammatical construction involves two component structures, an idea that reflects the assumption that the combination of component structures into composite structures is binary in nature. In other words, regardless of the ‘order’ in which constructions are composed, their internal constituency tends to reflect ‘layers’ that can be described in terms of binary relations. For example, in the NP that slipper under the bed, the PP ‘layer’ under the bed involves a relation between P under and NP the bed, while the larger NP ‘layer’ that slipper under the bed involves a relation between N slipper and PP under the bed. In the prototypical grammatical construction, one of the component structures is a RELATION and the other a THING. The RELATION is dependent and is the profile determinant. The THING is autonomous and serves to elaborate the schematic aspect of the dependent unit’s structure. This prototype represents the head complement structure, which corresponds to the preposition phrase (PP) structure and, as we will see below, it also corresponds to the structure built around a verb and its arguments which gives rise to the clause (section 17.2). Observe that the head-modifier structure departs from the prototype despite the fact that it is a frequently attested structure. The head-modifier structure departs from the prototype because it involves a RELATION (the modifier) that is not the profile determinant. Langacker (1987: 326) accords the head modifier relation the status of a ‘secondary prototype’. The valence relation apposition (which we discussed above in relation to the expression puppy dog) represents a more extreme departure from the prototype, since it relates two autonomous THINGS, does not contain a RELATION and lacks a profile determinant.
