Emma Borg Saying What You Mean: Unarticulated Constituents and Communication

In this paper I want to explore the arguments for so-called ‘unarticulated constituents’ (UCs). Unarticulated constituents are supposed to be propositional elements, not presented in the surface form of a sentence, nor explicitly represented at the level of its logical form, yet which must be interpreted in order to grasp the (proper) meaning of that sentence or expression. Thus, for example, we might think that a sentence like ‘It is raining’ must contain a UC picking out the place at which the speaker of the sentence asserts it to be raining. In §1 I will explore the nature of UCs a little further, and, in §2, suggest that we can recognise two different forms of argument for them in the literature. I will argue that ultimately neither is convincing, and they will be rejected in §3 and §4 respectively. The claim will be that, though the need for an appeal to such things as time and speaker are undoubtedly necessary in order to specify what a speaker said in a given context, advocates of the semantic relevance of UCs have failed to hold apart crucially different aspects of our understanding: first, the difference between knowledge of truth-conditions and the knowledge that truth-conditions are satisfied; second, the difference between knowledge of meaning and the understanding of communicative acts. Instead of ceding contextual information the kind of semantic role envisaged by advocates of UCs, we should, I will argue, see it as part of a theory of speech acts.1 I will suggest that what we need to recognise here is the proper division of cognitive labour, for once this division is in place, we can recognise the role and function of the information attributed to UCs, and its crucial relevance to communication, without ceding it semantic value. Sketching a model of our cognitive architecture which can underpin this stance, and showing why it might be thought independently attractive, will be the task of §5. Clearly, then, although the main focus of this paper rests with UCs, there are some big issues hovering in the wings here, and perhaps before we turn our attention squarely on the main target it would be in order to say why I think discussion of UCs cannot be had in isolation from these bigger issues. The reason, as I see it, is that arguments for UCs are part and parcel of a particular perspective on semantic theorising, one which is over-ambitious about the aims of a semantic theory. Consider the tasks we might expect an adequate semantic theory to fulfil: on the one hand, we might be concerned that such a theory explain quite ‘low level’ linguistic data, such as the meaning possessed by basic lexical items and how, given this base, our language displays properties like systematicity and productivity, which have been made so much of in recent linguistics and cognitive science. On the other hand, however, we might think that an adequate theory should do this and more,