Arbitrary plural pronominals

Abstract

In this paper we have developed an analysis for certain pronominal elements which are interpreted as arbitrary in reference. The pronominals in question appear in tensed clauses with a third person plural subject. We have considered the behavior and distribution of such elements in English and Spanish. Several important conclusions can be drawn from our investigation. First, the existence of identical constructions in English and Spanish indicate that the properties of arbitrary plural pronominals are quite independent of the setting of the so-called null subject parameter. Arbitrary plural pronominals occur equally well in a language like Spanish which allows null subjects in tensed clauses as in a language like English which does not allow null subjects in tensed clauses. In English, an overt pronoun, they, functions as an arbitrary plural pronoun. In Spanish, only a null pronoun can be interpreted as arbitrary in reference. An overt pronoun in Spanish can only be interpreted as non-arbitrary. Our analysis accounts for these facts as follows. The class of pronominals is subdivided by the feature [+/−dependent]. Pronominals have arbitrary reference if they are [+dependent] elements which lack an antecedent, in the relevant sense. A [+dependent] pronoun fixes its reference by linking to an antecedent, in an extended sense of the ‘binding’ relation. The antecedent of a [+dependent] pronoun, then, can be either a coindexed Operator, a coindexed argument NP, or a coindexed head which in agreement features (like INFL in Spanish). In Spanish, where pronouns in certain syntactic positions are allowed to be either overt or null, we claimed that all overt elements are specified redundantly as [−dependent]. Hence, an arbitrary interpretation is not possible for these pronominals. In English, overt pronouns are ambiguous: they are either [+dependent] or [−dependent]. Thus, in English an overt pronoun is allowed to function as an arbitrary pronominal. To the extent that this analysis is well motivated, it establishes one of our central points: an arbitrary interpretation for a pronoun results when a referentially dependent pronominal element lacks an antecedent. Secondly, we also investigated the distribution of arbitrary plural pronominals. We discovered that they are syntactically quite severely restricted. They are not allowed to appear in passive, ergative, or middle sentences. They appear only in transitive, unergative intransitive, or raising sentences. The structures where they are allowed have the following property in common: the position occupied by the arbitrary plural pronoun is neither assigned objective Case nor a ϑ-role directly by the verb, nor is it in a chain which is assigned a ϑ-role by the verb. We postulated that this peculiar distribution follows from the interaction of the Projection Principle, rules of Case and ϑ-role assignment, and the indexing rule for arbitrary plural pronominals. The Projection Principle, and the rules of objective Case and ϑ-role assignment require the presence of a referential index at D- and S-structure. But arbitrary plural pronominals lack an index at these levels of representation. Hence, they are restricted to appear in positions which do not enter into crucial Case or thematic relations until LF. This is precisely [NP, S] position. This position need not be assigned a ϑ-role at D-structure. It may contain expletive elements, for example. Furthermore, it is assigned nominative Case independently of the index of the category occupying that position: even expletive elements are automatically assigned nominative Case. Thus, the fact that arbitrary plural pronominals lack an index at D- and S-structure is without consequence provided the pronominal occupies subject position which is assigned nominative Case. The distribution of arbitrary plural pronominals, then, is exactly what we would expect given the interaction of the indexing procedure for such elements, and central principles of GB theory such as the Projection Principle, Case assignment rules, ϑ-role assignment rules, and the Extended Projection Principle.