SYMCO project

Project factsheet

Abstract

Coordination – as in “He saw [Jane or Mary]” or “[He came but she wasn't there]” – is one of the most frequent constructions in natural languages and, at the same time, one of the most controversial. One of the contentious issues is whether conjuncts must bear the same grammatical category; they do in the examples above: “Jane” and “Mary” are both nouns, “He came” and “she wasn't there” are both clauses (i.e. verbal constructions), etc. However, counterexamples are well known, as in “Pat is [a Republican and proud of it]” (a noun phrase and an adjectival phrase). Another controversial issue concerns the structure of coordination. In most kinds of phrases, one constituent is taken to be the syntactic head, e.g., the adjective “proud” in the adjectival phrase “proud of it" or the verb “came” in the verbal construction “He came”. But what is the head of “Jane or Mary”: is it “Jane”, “Mary", or “or”?

There are two competing types of theories of coordination: asymmetric and symmetric. According to asymmetric theories, the first conjunct is the head of coordination. For example, “Jane or Mary” is headed by “Jane”. According to symmetric theories, the status of all conjuncts is equal, i.e., neither is the sole head of coordination. One subtype of such theories has it that coordination is headed by the conjunction (e.g., “or” in “Jane or Mary”), while the other claims that coordination is multi-headed, i.e., headed by all conjuncts (e.g., by both “Jane” and “Mary” in “Jane or Mary”). Both symmetric and asymmetric approaches to coordination are pursued in dependency syntax (symmetric, e.g., in Hudson's Word Grammar, asymmetric, e.g., in Mel'čuk's Meaning–Text Theory) and in generative syntax (e.g., Neeleman et al. 2023 in Syntax vs. Zhang's 2009 CUP monograph).

The main argument given for asymmetric theories of coordination is based on examples such as (1), where only the first conjunct satisfies the requirements of the governing predicate ((2a) is fine), but the second violates them ((2b) is unacceptable).

The reasoning is that only the first conjunct matters for the distribution of the whole coordinate structure, so (only) the first conjunct is the head. However, examples such as (1) are very hard to find in real texts, and there are many other examples which show that actually both conjuncts decide about the distribution of coordination, e.g., (3a).

(3a) is unacceptable because the verb “become” does not combine with prepositional phrases such as “under suspicion” (*“Pat became under suspicion” is ungrammatical), so it also cannot combine with a coordination containing such a prepositional conjunct, even when it is the second conjunct. This should be contrasted with the verb “be”, which does combine with such prepositional phrases (“Pat is under suspicion” is grammatical), so it may also combine with a coordination containing such a phrase, as in (3b).

The aim of this project is to investigate and refute the argument for asymmetric theories of coordination based on data such as in (1). We will perform acceptability judgement experiments which will demonstrate that the perceived acceptability of sentences such as (1) is a processing effect: the longer the distance between the predicate (“depend on”) and the offending conjunct (“that he will be on time”), the more acceptable the sentence seems to be. Such “apparent grammaticality” caused by processing effects has been described in the literature, in the context of other phenomena.

The experiments will involve the initial (norming) stage, in which we will ascertain that certain verbs (such as “depend (on)”) cannot combine with clauses (such as “that he will be on time”) alone. At the subsequent main stage, we will attempt to demonstrate that the acceptability of such combinations improves with the distance between the verb and the clause. A number of constructions will be considered that make it possible to vary this distance: not only coordination (where the distance depends on the number and lengths of intervening conjuncts), but also fronting constructions, where the clause is topicalized, and possibly other constructions. Experimental results will be supported by corpus investigation.