Determination

The problem of restrictiveness of a specifier is intimately linked up with the problem of determination of a relative construction. We will here consider it for embedded relative clauses. The determination of adjoined ones works differently.

The first distinction to be made is between a concept and a referential expression. Lexemes, including nouns, designate concepts (or notions), not referents. A linguistic expression becomes referential in a speech situation and in a text. A referential expression refers to a referent. This is an entity in the universe of discourse. It is a particular entity or set of entities, delimited and fixed in its identity. In the subsequent discourse/text, more information can accrue on it by predication; but this does not change its identity.

A concept can be restricted and, thus, changed into a different concept by modification and anchoring; see the relevant section. A referent cannot.

The principal linguistic operation that delimits and fixes the reference of a nominal lexeme or syntagma is determination. However, there are different kinds of determination. Indefinite determination, as it is typically coded by indefinite pronouns, determiners and articles (cf. below), does not fix the reference of the nominal expression thus determined. Thus, it is compatible with modification that further restricts the concept designated. On the other hand, definite determination indicates that there is already a referent available for the expression thus determined. It is therefore incompatible with further restriction of the concept that may be the core of the referential expression.

Two other linguistic means to fix the reference of a nominal expression must be taken into account: Proper names are referential expressions. Consequently, they cannot be restricted; s. .a. If a proper name is accompanied by a restrictive relative clause, as in #b, this presupposes a category change from proper noun to common noun.

Nominal attributes, especially possessive pronouns, often have a determinative force that pins down the reference of their head, as in .a. This happens whenever it is assumed that the possessor possesses only one instance of the kind designated by the head. This presupposition does not always hold. Thus, the non-restrictive version #b is possible, though it might more naturally be phrased as the one of my daughters who ....

Given therefore a relative construction of the form [ [X]Det [Y]Nom [Z]S ]NP, there are two possibilities:

1. Z restrictively specifies Y; and the complex nominal thus formed is then determined by X.
2. X determines Y; and this noun phrase is then commented on by the non-restrictive relative clause Z.

The distinction in construction is generally mirrored by different intonation. Constructions reflecting the semantics in this way are compositional. Apart from the problematic cases mentioned in the next section, there is, however, a recurrent deviation from this in natural discourse: Not seldom a speaker first utters a determined NP which up to this moment he might think is sufficient to identify the referent he has in mind. However, he instantly realizes that more semantic information is necessary to identify this referent. Consequently, he supplies such information in a relative clause which is meant to be restrictive although from a normative point of view, this ought to fail.

In structural terms, a referential NP is here followed by a free relative clause, in its turn determined by the definite article. The combination is apposition. However, the communicative intention is to restrict ex post the concept designated by the initial head. The communicative strategy is apparently the following: The pronoun one resumes the head of the preceding NP by semantic anaphora. It is used as the head of the following light-headed relative clause, as if the previous head noun were repeated. Thus, the light-headed relative clause is meant to replace, in the communication, the first attempt at reference. This strategy may grammaticalize, with the effect that an erstwhile appositive combination of a definite NP with a definite light-headed relative clause becomes an attributive and even a restrictive relative construction. Such a construction may then, countericonically, involve a definite determiner preceding the relative clause. The grammaticalization of the construction, however, leads to the consequence that this determiner looses its referential force and gets a purely structural function in the formation of attributive relative clauses.

Restrictiveness

Given a core concept and a modifier or specifier applied to it, then the latter is restrictive if it narrows down the core concept to a more specific concept, i.e. if it enhances the intension and, consequently, reduces the extension of the core concept; otherwise, the modifier or specifier is non-restrictive. Non-restrictive modifiers are sometimes called appositive. However, until the form-function relations in this domain are clarified, it is safer to reserve the terms ‘apposition’ and ‘appositive’ to structural configurations rather than semantic relations.

In a polygamous situation, the attribute in would be restrictive, singling out the good one among the speaker's wives. In a monogamous situation, the attribute would be non-restrictive. Some languages – including Latin, but excluding English – can distinguish a restrictive from a non-restrictive adjective attribute by its postnominal vs. prenominal position. In , the distinction is made by prosody: in the restrictive interpretation, the attribute has primary stress.

The raison d'être of a relative clause is the specification of a term in order to narrow down a concept or to pin down a referent. Both is impossible if the referent in question is already identified. This is the case if it is designated by a definite noun phrase, including a definite pronoun, or by a proper name. Nevertheless some languages can combine a relative clause with such a term. Such a relative clause is then non-restrictive, in contrast with the default, which is the restrictive relative clause. is restrictive, is non-restrictive.

A restrictive adnominal relative clause is an attribute of its head nominal, while a non-restrictive relative clause combines with an entire noun phrase. Consequently, a sentence like has two analyses, one with a restrictive relative clause, as in #a, and another with a non-restrictive relative clause, as in #b.

The point to note is that with restrictive attribution, despite appearances, the determiner does not form a constituent with the head nominal; it forms one with the higher nominal. This makes no analytic problems in languages like English, where the determiner is on the opposite side of the relative clause. Nor does it constitute a problem – but would instead confirm the analysis – if both the determiner and the relative clause are on the same side of the head nominal, and in either case the relative clause is closer to the head than the determiner. This scope order of the determiner is found with the prenominal relative clause of Japanese and Tigre and with the postnominal relative clause in Indonesian (), Ewe and Yoruba.

Other orders do occur in restrictive constructions. Unless they have an independent explanation, one may hypothesize that in such cases, the determiner has grammaticalized on the head noun and therefore no longer has scope order. In other words, such constructions are countericonic.

Non-restrictive relative clauses generally show structural differences from restrictive ones. If they are postnominal, they are usually loose appositives, which implies an intonation break between head nominal and non-restrictive relative clause which does not exist in a restrictive relative construction. In English, non-restrictive relative clauses require a relative pronoun. Many more languages do not form non-restrictive relative clauses at all. Their function can be fulfilled by independent predications, including adjoined clauses. If a language has relative clauses, it has restrictive relative clauses, but not necessarily non-restrictive relative clauses.

The function of restrictive relative clauses varies between the poles of concept formation and entity selection. The relative clause in forms a concept; the relative clause in selects an entity.

This difference in function is related to the referential properties of the components of the situation depicted by the relative clause.