Search Results

It has been assumed that a pronominal agrees with its antecedent, but this book has shown that determining pronominal φ-feature values is actually a complex issue involving a number of different principles. The book has argued for the Pronominal Agreement Condition and has offered a syntactic account of pronominal alternations involving imposters and camouflage determiner phrases. It has also presented data, generalizations, and syntactic mechanisms that place important boundary conditions on any eventually successful semantic theory of pronominal anaphora. In addition, it has debunked the notion that the value of a pronoun is determined only by a contextually determined assignment function in the absence of an antecedent. This conclusion is supported by data that also involve pronominal agreement with the φ-features of ghosted antecedents.

This chapter presents the central assumptions about Agree and Distributed Morphology adopted in this study. These assumptions are sufficient to account for absence of Principle B effects, i.e., the observation that in many languages pronouns take over anaphoric duties when a dedicated anaphor is lacking. An account of this generalization is developed in terms of the Elsewhere principle. More specifically, it is shown that the framework of Distributed Morphology framework allows for a precise formal analysis of a wide array of data conforming to this generalization, including the diachrony of English, Haitian Creole, and the delay of Principle B effects in L1 acquisition.

This chapter discusses the syntax of simplex reflexives. It argues that simplex reflexives should be analysed on a par with possessive pronouns occurring in contexts of inalienable possession. Concretely, simplex reflexives are merged as the Possessum in a possessive constituent that also hosts its antecedent, the Possessor. Following Den Dikken (2006), the Possessum is merged in a position that is hierarchically higher than the Possessor. In this configuration, the reflexive Possessum is a probe c-commanding its goal, the Possessor-antecedent. The reflexive Possessum values its φ-features in an Agree relation with the Possessor, thus deriving Binding. Finally, it is shown that the constituent containing the Possessor and the Possessum is the internal argument of an unaccusative verb.

This chapter tackles the syntax of self-reflexives. Such reflexives are derived from pronouns by adjoining a self-part to them, which provides them with the syntax of floating quantifiers. This claim is developed in two steps: first, it is shown that self-reflexives share a number of properties with intensifiers (e.g., The headmaster has seen me himself). Second, it is argued that the syntax of such intensifiers closely matches that of floating quantifiers. Finally, the syntax of self-reflexives is shown to be reducible to the syntax of floating quantifiers. Floating quantifiers must c-command its antecedent. So do self-reflexives: they overtly or covertly raise to an adjoined position from which they c-command their antecedents. As probes, they value their φ-features via an Agree relation with the antecedent they c-command. An account is developed for the logophoric uses of self-reflexives.

The φ-feature values of a given non-expletive pronominal is expected to match the φ-feature values of some single determiner phrase with which it agrees. The matching would be with the values of a single source, but this assumed state of affairs, dubbed source uniqueness, is in no sense logically necessary. This chapter demonstrates that source uniqueness is factually untenable and how a single pronoun can be made to agree with multiple sources in different φ-features. It also discusses the pronominal agreement, camouflage and gender agreement, the concept of number attraction, the plurality of ultimate antecedents, and coordination involving pronominal agreement and determiner phrases.

This chapter provides novel empirical evidence from agreement in Czech copular clauses that the [+PERSON] feature is dependent on animacy (Adger and Harbour 2007, Nevins 2007, a.o.). The core evidence comes from agreement with φ‎-feature-deficient pronouns. It is argued that agreement with φ‎-feature-deficient pronouns yields a Multiple Agree configuration (Hiraiwa 2005). Strikingly, the interpretation of such a pronoun is restricted by all φ‎-features present in the Multiple Agree chain. Thus if a φ‎-feature-deficient pronoun with an unvalued PERSON feature enters a Multiple-Agree chain with valued φ‎-features, including a valued PERSON feature, then the PERSON feature restricts the interpretation of the antecedent of the pronoun. Crucially, this happens only if the pronoun gets valued as [+PERSON]. It is argued that this is because [+PERSON] pronouns have φ‎-feature valuation restricted by GENDER presuppositions associated with animate referents (Heim 2008, Sudo 2012, a.o.).