Presupposition, implicature and context in text understanding1

This paper examines the roles which presupposition and implicature play with respect to what is asserted by a text and to its context, as a part of the process of text understanding. This process involves constructing and updating the representation of the context. Assertion, implicature and presupposition can be described as three different ways in which changes in the representation of the context are induced. On the basis of such a description, it is claimed that, contrary to most of the literature on the subject (in which presupposition and implicature seem not to be allowed to coexist without being identified with each other), there are reasons for considering presupposition and implicature as two distinct phenomenona. 1. Two perspectives on text understanding It is by now widely recognized that discourse understanding involves more than the understanding of what is explicitly said. In order to understand discourse, or as (I shall say here) in order to understand a text, we must understand more than what is encoded in the text itself, and draw inferences. This broader comprehension, which is closely connected with contextual knowledge, is often described as the comprehension of what is presupposed and/or implicated by the text. There are two main ways in which we can conceive of the role of this broader comprehension (which I will call comprehension of the implicitly conveyed meaning) with respect to the overall understanding of the text. (i) Sometimes it may seem that the understanding of what a text presupposes or implicates, as well as the contextual knowledge involved in such understanding, are necessary conditions for a full comprehension of the text. If we do not know the circumstances in which a text has been written, or if an utterance is reported to us without any information about the circumstances in which and the goals for which it has been uttered, we may not be able to make sense of it. If we do not share the speaker’s pragmatic presuppositions (the assumptions he or she takes for granted in speaking) (Stalnaker [18], [19]), we might misunderstand him or her. As to 1 Thanks to Christopher Gauker and to an anonymous referee for comments on earlier drafts. conversational implicatures, they depend on the assumption that the speaker is observing the Cooperative Principle (Grice [7]: p. 26) and therefore, in order to infer them, the hearer should already know whether, in the circumstances of the ongoing verbal exchange, the Cooperative Principle holds. According to relevance theory (Sperber and Wilson [17]), another pragmatic theory concerned with discourse understanding, in understanding a text we have to take into account the contextual premises which make the speaker’s contribution relevant. In all these ways, knowledge of, or at least beliefs about, what may in one word be called “the context” are represented as necessary to the comprehension of the text. This might lead us to conclude that we can, and must, acquire knowledge of or beliefs about the context prior to, and independently of, our understanding of a text. (ii) Suppose, however, that we find ourselves in a situation in which we have little independent access to the context, as it happens in reading, in certain phone calls, or in those cases of face to face interaction in which we know little about our interlocutor and his or her possible aims. Should we despair of making sense of the text we are faced with? In such cases, it might be convenient to exploit all the details of the text in order to project as much of its context as we can. After all, many presuppositions have linguistic markers or triggers, and this enables receivers to detect them even in absence of text-independent information. As to implicatures, it could be claimed that the speaker’s observance of the Cooperative Principle must not be known in advance, but can be assumed in absence of evidence contrary to it, so as to allow for the working out of as many implicatures as possible. Finally, relevance theory admits of the possibility of inferring missing contextual premises, when the assumptions which are already available to the hearer do not make the speaker’s contribution relevant. In this perspective, context (or more precisely, the representation of context which is associated with the understanding of the text) is not something which has to be given independently of the text, but something constructed in the very process of text understanding. Although I do not want to deny that perspective (i) has its merits, here I am going to adopt perspective (ii), because I would like to outline a description of the textcontext relationship which optimizes the chances of text understanding even in unfavorable conditions. In this framework, I would like to claim that presupposition and implicature play different roles with respect to text understanding and that therefore they should be considered as distinct phenomena. This runs contrary to most of the literature on the subject: presupposition and implicature belong, as it were, to two different conceptual frameworks and authors who use one of these notions do not use the other, so that only one of them does all of the work. Those authors who mention both notions have (since Karttunen and Peters [10]) identified presupposition with one kind of implicature, conventional implicature. I would like to claim that presupposition is different from both conversational and conventional implicature as to the role it plays with respect to what is asserted by a text and to its context. 2. On text and context I choose here to use the word “text”, following the semiotic (Hjelmslevian in particular) rather than the philosophical tradition, in a sense akin to the one recently specified by M. Stubbs with reference to the practice of discourse analysis: “By text, I mean an instance of language in use, either spoken or written: a piece of language behaviour (...)” ([23]: p. 4). This definition of “text” leaves it open which size a text should have: a text (the relevant piece of language behaviour) could well coincide with the utterance of one sentence, but might also consist of the utterance of more or, for that matter, less than one. The utterance of a syntactically complete and isolated sentence is therefore one case falling under the more general idea of the production of a text. As to the problem of text delimitation, it should be remarked that, whenever what is focused upon as a text is in turn a part of a larger episode of language behaviour, we can (i) include relevant parts of this larger episode into the text focused upon, thus changing the delimitation of what is under consideration or (ii) consider the larger episode of language behaviour in which our text is embedded as a part of the context. Choice (i) turns linguistic context into text, while choice (ii) considers linguistic context as a part of the context. As to context, I believe that insofar as we are concerned with its capacity of being that against which a text is evaluated (as to appropriateness and/or truth), it must be conceived of as “objective” or mind-transcendent (Gauker [5]). It is only with respect to something external to speakers and independent of what is focused upon as the presently considered text, that it makes sense to evaluate, or attempt to evaluate, that text as a piece of linguistic behaviour. I will here conceive of objective contexts in an intuitive way, namely, as consisting of the set of facts which have to be taken into consideration by the participants if a given verbal exchange is to achieve its goals. One problem with this view it that evaluation may (or perhaps must) remain provisional, or defeasible. But this trouble is shared by all of our knowledge, which aims at objectivity, but is nevertheless persistingly defeasible. Here, however, we will not be concerned with the function of context in text evaluation, but in text comprehension. In particular, in conformity to perspective (ii) outlined above, I want to specify the ways in which the information contained in a text can tell or show us something about its context. So we will be concerned with that representation of the context, relative to a given text, which can be worked out on the basis of that text in the process of understanding it. This representation of the context is, of course, directed at the objective context, but should not be confounded with it, since it has to be worked out by the participants, while the objective context transcends their cognitive processes. 3. The dynamic relation between text and context A text entertains a dynamic relation with its context. During text production, the addition of new speech acts to those already performed can be described as having context-changing effects (Gazdar [6]), so that the context at time 2 is different from the context at time 1 as regards the addition or elimination of some contents. On the notion of text I am using here, a text T1 is correspondingly changed into a text T2 as soon as new parts are added to it. The updating of the context is always relative to a new delimitation of the text, so that one-to-one correspondence between texts and contexts is preserved. However, the distinction (outlined above) between the objective context and the representation of it has to be taken into account. It might be thought that the objective context, being independent of the text, cannot be changed by it. In fact, nonverbal actions, bearing on the circumstances relevant to the goal of the exchange within which the text is produced, change the objective context, but they do not belong to the text either. However, it has to be conceded that the very occurrence of linguistic behaviour changes the context, providing part of the context for subsequent text production. Moreover, it can be claimed that a text performs a context-changing action if the speech acts it contains have effects consisting of the bringing about of intersubjectively recognizable states of affairs (such as new obligations or rights or their cancelation). I have elsewhere called such effects “changes in the conventional context” (and following Gazdar [6], I have defined illocutionary acts in their terms: [14], [15]), where the “conventional context” may be construed as a specialized part of the objective context (insofar as we believe that human conventions too have their own kind of objectivity). Although the consideration of the ways in which the objective context is affected by changes can be an important issue, here we will be concerned with the changes which are produced in the representation of the context as a part of the process of text understanding. We will be concerned with (some) changes in the objective context only insofar as these play a role in inducing changes in the representation of the context. I shall try to describe three main ways in which the latter kind of change can be achieved: assertion, implicature, and presupposition. The proposed description will enable me to claim that presupposition cannot be identified with implicature.