wouldn'tbe very interesting. So consider: what if an index werenotcopied? This seems precisely the proper account of
predicate NPs, which do not pick out an independent individual. For instance, inEva became a doctor, there are not two
separate individuals, Eva and the doctor; there is only Eva, of who mdoctorhood is predicated. Accordingly, we could
set up this sentence as in (27).
The structural configurations of (26) and (27) are the same in syntax, and nearly the same in the descriptive tier. The
main structural difference is in the referential tier: there is no index corresponding to the predicate NPa doctor. Thus
the sentence asserts that one individual, describable as‘Eva,’has come to be describable also as‘a doctor.’This
captures thetraditional sense inwhichpredicateNPs are sort of likeadjectives:they contributeonly descriptionbut no
new individual.
On the other hand, suppose an independently referential NP such as a proper name is placed in predicative position,
forinstanceSuperman is Clark Kent. Inthiscase, thereferentialtierwillcontainan index fora secondindividual, and the
sentence will assert that the two individuals are identical—that is, it creates a“merger of indexicals”in the sense of
section 10.7. (Whether this requires a different sense ofbeis an interesting question; I have argued (Jackendoff 1983:
ch. 6) that it does not.)
The distinction between ordinary and predicate NPs leads to an important observation. Some NPs such as proper
names are inherentlyreferential. But other NPs, particularlyindefinites, are referential only by virtue of their role inthe
sentence as a whole. In most roles they are referential. But some roles, in particular the predicative role, do not carry
referential claims. The formalization of verb semantics in sections 11.7 and 11.8 allows us to localize the predicative
role precisely: it is the second argument of the function BEAscriptionOther conceptual functions that affect referentiality
will appear in the next section.
The separation between the descriptive and referential tiers leads to an interesting account of two different kinds of
anaphora. Standard definitepronouns such ashe, she, anditare understoodas coreferentialwiththeirantecedents. This
coreference can be notated in thereferentialtierby equating the pronoun's index withthat of its antecedent,as in (28).
(I abbreviate syntax/phonology (S/P) and the descriptive tier (DT) still further, eliminating self-explanatory brackets
and labels. The pronounitistaken to have the descriptive content‘3rd person singular neuter’.)