Past Event

Revisiting the argument-adjunct distinction

Dr. Ida Toivonen
(Carleton University)

Abstract:

The goal of this project is to examine the distinction between arguments and adjuncts; a distinction fundamental to syntactic and semantic analysis. Consider example (1):

(1) Rhonda bought cookies on Monday.

An analysis of (1) should capture the intuition that “Rhonda” and “cookies” are more closely connected to “bought” than “on Monday” is: in (1), “Rhonda” and “cookies” are arguments, and “on Monday” is an adjunct. Although this underlying intuition is widely shared, a formal criterion for distinguishing arguments from adjuncts has yet to be successfully formulated. While several diagnostics have been proposed (we list eight of them below), they often give mutually inconsistent results. This study aims to take these inconsistencies seriously and make sense of them. Our leading hypothesis is that there is in fact no single criterion distinguishing arguments from adjuncts. Rather, there is a cluster of dissociable distinctions which tend to correlate.

Canonical arguments are specified within a verb’s core argument structure, they describe individual participants in an event, and they occur in certain phrase-structural positions. Canonical adjuncts are not specified in argument structure, do not describe event participants, and occupy different phrase-structural positions. Different argumenthood diagnostics probe different distinctions, which is why they sometimes give inconsistent results: the inconsistencies arise within a structured space of intermediate cases between canonical arguments and canonical adjuncts.

The novelty of this approach is that it reveals a middle ground between two common positions. The majority of current research, across different theoretical and experimental paradigms, presupposes that there is a clean argument–adjunct distinction, even if it is sometimes hard to classify a given phrase as an argument or adjunct. A minority, faced with the apparent failure to provide a single robust definition of the distinction, claims that there is no distinction, or that the distinction is gradient.

Neither approach is a good fit for the facts: the former is too coarse, but reflects the pervasiveness of the distinction in descriptions of syntactic and semantic phenomena. The latter reflects the slipperiness of the distinction, but obliterates a valuable tool in syntactic description. By postulating that arguments and adjuncts can be distinguished along multiple dissociable dimensions, we envisage a model in which the argument–adjunct distinction is real and not fully gradient, but with the flexibility to account for a range of categories with behaviour intermediate between arguments and adjuncts.

This event is sponsored by the School of Linguistics and Language Studies