The present paper is the result of a long struggle to understand how the notion of compositionality can be used to motivate the structure of a sentence. While everyone seems to have intuitions about which proposals are compositional and which ones are not, these intuitions generally have no formal basis. What is needed to make such arguments work is a proper understanding of what meanings are and how they can be manipulated. In particular, we need a definition of meaning that (...) bans all mentioning of syntactic structure; it is not the task of semantics to state in which way things are put together in syntax. The present paper presents such a theory of meaning. This, in tandem with some minimal assumptions on the syntactic process (that there can be no deletion) yield surprisingly deep insights into natural language. First, it rehabilitates a lot of linguistic work as necessary on semantic grounds and defends it against potential claims of redundancy. For example, θ-roles and linking are an integral part of semantics, and not syntax. To assume the latter is to put the cart before the horse. Second, as a particular example we shall show that Dutch is not strongly context free even if weakly context free. To our knowledge, this is the first formal proof of this fact. (shrink)

Within natural language semantics, pronouns are often thought to correspond to variables whose values are contributed by contextual assignment functions. This paper concerns the application of this idea to cases where the antecedent of a pronoun is a plural quantifiers. The paper discusses the modelling of accessibility patterns of quantifier antecedents in a dynamic theory of interpretation. The goal is to reach a semantics of quantificational dependency which yields a fully semantic notion of pronominal accessibility. I argue that certain dependency (...) phenomena that arise in quantificationally created contexts require a representation of context wherein the labelling of antecedents is not rigid but rather dynamic itself. I propose a stack-based alternative to classic assignment functions, along the lines of Vermeulen (1993) and van Eijck (2001), and give a dynamic semantics of quantification which correctly accommodates the problematic anaphoric phenomena. (shrink)

Relational Grammar (RG) was introduced in the 1970s as a theory of grammatical relations and relation change, for example, passivization, dative shift, and raising. Furthermore, the idea behind RG was that transformations as originally designed in generative grammar were unable to capture the common kernel of, e.g., passivization across languages. The researchconducted within RG has uncovered a wealth of phenomena for which it could produce a satisfactory analysis. Although the theory of Government and Binding has answered some of the complaints, (...) still it left many phenomena unaccounted for. Referent Systems (RSs) have been introduced by Vermeulen (1995) to overcome certain weaknesses of Dynamic Semantics. Their usefulness has not yet been realized in semantical theory. We shall show here that their significance cannot be overestimated. Namely, we will show in this paper that there exists a fundamental affinity to RG. Both handle the relation between an argument and a functor by means of a shared relational sign, which is unique for each argument. This assignment can be changed. What is interesting is that the notion of a chômeur, which is central to RG, finds a natural treatment in RSs. This coincidence is in our view not accidental but reveals some fundamental properties of the human language faculty. (shrink)

This paper presents a new take on how argument dependencies in natural language are established and constrained. The paper starts with a rather standard view that (quantificational) argument dependencies are operator-variable dependencies. The interesting twist the paper offers is to eliminate the need for syntax that serves to enforce what the operator-variable dependencies are. Instead the role of ensuring grammatical and generally unambiguous forms is taken up by semantics imposing what are dependency requirements for any interpretation to go through at (...) evaluation time. With this viewpoint there emerges an essential need for localities as the hiding of (embedded) expression material, together with a (limited) form of garbage collection to release grammatical resources used in the current locality for reuse in subsequent localities. (shrink)

In this paper we describe a framework for the construction of entities that can serve as interpretations of arbitrary contiguous chunks of text. An important part of the paper is devoted to describing stacking cells, or the proposed meanings for bracket-structures.