English specificational and predicative clauses: A functional-cognitive account of their representational and textual organisation

Specificational clauses set up a pragmatically presupposed variable – i.e. that is presumed to be known or to be taken for granted by the hearer (Lambrecht 1994: 52) –, e.g. the best option in (1a), and assign a specific value to it, e.g. the Hotel Boomerang.

(1a) Where to stay: the Parador El Hierro … has doubles from £80 a night, B&B. Or in Valverde, the best option [variable] is *the Hotel Boomerang* [value]. (WB)

The role of such a clause is similar to a mathematical function, like “x=2”, with "x" being the variable and "2" the value specified for that variable. The variable in a specificational clause can be introduced either by a definite article (the), as in (1a), but also by an indefinite one (a(n)), e.g. (2a).

(2a) A major feature of functionalism [variable] is *the idea of societal needs* [value]. (WB)

Since presupposition and focus typically complement each other (Ward & Birner 2001: 121), the function of specification typically correlates with information focus (between asterisks) on the value, e.g. (1a) and (2a). A syntactic criterion of English specificational clauses is that they are reversible, i.e. they allow for a subject-complement switch (Huddleston 1984: 457; Higgins 1976; Declerck 1988; Halliday 1994; den Dikken 2006), e.g. (1b) and (2b).

(1b) *The hotel Boomerang* [value] is the best option [variable].

(2b) *The idea of societal needs* [value] is a major feature of functionalism [variable].

In the literature, the study of specificational clauses has typically focused on definite ones (cf. (1a-b)), while indefinite ones (2a-b) have either not been acknowledged or simply put on a par with their definite counterparts.

Picking up on Davidse’s (2010) suggestion to treat indefinite specificationals as a distinct subtype of copular clauses, Van Praet (2013, 2014) argued that indefinite specificational clauses cannot be thought of as having “identifying” semantics – i.e. they do not “express that the referents of two definite terms [...] coincide in the same entity” (Dik 1980: 32) – but are best accounted for in terms of conformity, or correspondence, between instance and schema (Langacker 1991). As pointed out in Davidse (1992), the gradability associated with predicational clauses is better accounted for by the relation of instantiation than by the traditional logical notion of class-membership: classes are “logical bounded entities membership in which is defined by an item's possession of a simple set of criterial (i.e. necessary and sufficient) features” (Rosch & Mervis 1975: 573-574); while a schema designates a “prototypical category, [which] exhibit[s] degrees of typicality – not every member is equally representative for a category” (Rosch & Mervis 1975: 574-575). Since the idea of more prototypical versus more peripheral instances of a schema translates directly into gradability, an instance-schema correspondence allows a more natural explanation of gradability.

In addition, the semantic model of a correspondence relation between instance and schema allows a natural account of the possibility to construe it either ascriptively (3) or specificationally (4) (Van Praet & Davidse 2016).

(3) Alice is *a thief*. (Langacker 1991: 67)

(4) An option is *Onstar’s personal calling service*. (WB)

The instance designated by an indefinite nominal like “a thief” in (3) is intrinsically semantically more general than a specific instance like Alice and can be thought of as designating “an arbitrary member of the thief category” (Langacker 1991: 67). Importantly, the fact that an indefinite NP with a common noun designates “an instance of the type T designated by the noun” helps explain why it can function either as an ascriptive complement or as a variable NP. While the semantics of (3) can be glossed as “Alice is an arbitrary member of the type or schema [thief]” or “Alice instantiates the schema [thief]”, a specificational clause like (4) specifies an instance (‘value’: Onstar’s personal calling service) as corresponding to the schema (‘variable’: an option). Indicating the instance-schema semantics that indefinite specificational clauses share with predicative ones allows making generalizations that are lost in the tradition that views them as identifying (e.g. Halliday 1994). By the same token, specificational clauses impose a different directionality on the correspondence relation between instance and schema than predicational ones. The predicational complement realizes type-attribution (Langacker 1991: 67), while in specificational clauses, instances (‘values’) are specified as meeting the criteria of the schema (‘variable’). Van Praet & Davidse (2016) captured the schematic overarching semantics of indefinite predicational and indefinite specificational clauses by referring to the two types as having ‘categorizing’ semantics (in the ‘prototype’ sense), yielding an ascriptive-categorizing and a specificational-categorizing type.

Since indefinite specificationals show that neither exhaustiveness nor identification are constitutive of the specificational function, the distinctive feature of specification must reside elsewhere, and a good candidate for this is the “superscriptional” character of the variable (Higgins 1976:203), which is always the lexicosemantically more general NP functioning like the heading of a list. This is manifested in the weakly referring status of the definite variable, e.g. “the best option” in (1a-b), (Declerck 1988:47, with reference to Donellan 1966) and the indefinite non-specific variable, e.g. “an option” in (4), both of which presuppose the existence of a referent (Declerck 1988:14ff), but do not enable identification of any specific referent.