Semantic agreement and the dual model of language

Abstract

This paper will show that agreement cannot be considered a purely narrow-syntactic phenomenon, i.e., a phenomenon which cannot be given a functional-semantic explanation. It will show that person-number-gender (PNG-)affixes on verbs and adjectives are semantically interpretable even in languages such as English and Swedish, where they at first sight seem to be completely redundant. Several instances of so-called semantic agreement will reveal that these PNG-affixes convey important meanings not overtly marked by their co-referential DPs. In some cases, this even holds for the opposite case, i.e., that the PNG-affixes displaying grammatical agreement with their co-referential DPs also contribute with meanings not conveyed by the latter. Attempts to explain away semantic agreement by assuming there to be phonologically null determiners that give the DPs the same PNG-values as the agreement affixes will be refuted by data that show that the nouns often cannot be modified by such elements. Instead of being uninterpretable features that must be deleted before the derivation is sent to the semantic component, it will be shown that PNG-affixes on verbs and adjectives function like pronominal affixes, i.e., that they are arguments of the verbs and adjectives. They can either be anaphoric, in which case they can display semantic agreement, agreeing with the lexical semantics of the referent rather than with the PNG-values of the co-referential DP, or cataphoric, in which case they behave like formal-subject pronouns, signifying the PNG-value of the following logical DP subject.

Keywords

• grammatical agreement
• y-model
• anaphoric pronouns
• semantic component
• null pronouns
• lexical semantics