Restrictive vs. non-restrictive relative clauses
In (35) the head of the relative clause (woman) is a common noun which could refer to any one of a few billion individuals. The function of the modifying clause is to identify (uniquely, one would hope) which particular woman the speaker is referring to. This is a typical example of a RESTRICTIVE relative clause. In this construction, the reference of the NP as a whole is determined in two stages: the head noun designates a class which the referent must belong to; and the modifying clause restricts (or narrows) the identity of the referent to a specific member of that class.
Not all relative clauses work this way. A NON-RESTRICTIVE relative clause is one in which the referent of the head noun can be identified independently, and the clausal modifier simply presents additional information about that referent. The difference between restrictive and non-restrictive relative clauses is illustrated in (36). In(36a) the head is a common noun which identifies the class of men in general. The modifying clause serves to identify which particular man the speaker is talking about. In (36b), on the other hand, the head is a proper name; this indicates that the speaker assumes that the hearer already knows who Al Capone is. The clausal modifier serves only to provide additional background information about this individual.
(36) a RESTRICTIVE:
The police are looking for [the man who escaped from prison yesterday].
b NON-RESTRICTIVE:
The police are looking for [Al Capone, who escaped from prison yesterday].
Thus, in restrictive relative clauses, like (36a), the modifying clause contains old or presupposed information while the identity of the referent is new information. But in non-restrictive relative clauses, like (36b), the identity of the referent is old information while the modifying clause contains new information.
In many languages there is little or no difference in grammatical structure between restrictive and non-restrictive relative clauses, but this is not always the case. In English the most obvious difference is into national: nonrestrictive relative clauses are always set off by pauses, usually indicated by commas as in (36b), whereas restrictive relative clauses are not. Another difference is illustrated in (37) and (38). In a restrictive relative clause like (37), the modifying clause may be introduced by that, a Wh- word, or Ø.
But in a non-restrictive relative clause like (38), the modifying clause must be introduced by a Wh-word; the use of that or Ø makes the construction ungrammatical.1
(37) a The woman whom I love is moving to Argentina.
b The woman that I love is moving to Argentina.
c The woman I love is moving to Argentina.
(38) a Mary Martin, whom I dearly love, is moving to Argentina.
b *Mary Martin, that I dearly love, is moving to Argentina.
c *Mary Martin, I dearly love, is moving to Argentina.
As we have seen, proper names occur quite naturally as heads of nonrestrictive relative clauses, but they are quite unnatural in restrictive relative clauses. Conversely, phrases involving quantifier words such as any, every, or some are only acceptable as heads of restrictive relative clauses, and do not work in non-restrictive relative clauses:
(39) a Every student who failed the exam will be asked to repeat the course next year.
b *Every student, who failed the exam, will be asked to repeat the course next year.
(40) a Any student who brings cigarettes to school will be sent home.
b *Any student, who brings cigarettes to school, will be sent home.
(41) a No one who saw the new James Bond movie liked it.
b*No one, who saw the new James Bond movie, liked it.
1. These differences are also discussed in Bickford (1998:332). For a more detailed discussion of these and several other grammatical differences between restrictive and non-restrictive relative clauses in English, see McCawley (1988:417 432).
