Quasi-Indexicals and Knowledge Reports

We present a computational analysis of de re, de dicto, and de se belief and knowledge reports. Our analysis solves a problem first observed by Castaneda, namely, that the simple rule • (A knows that p) implies P , apparently does not hold if P contains a quasi -indicator. We present a single rule, in the context of an AI representation and reasoning system, that holds for all propositions P, including quasi-indexical ones. In so doing, we demonstrate the importance of representing proper names expli citly, and we provide support for the necessity of considering sentences in the context of extended text (e.g., discourse or narrative) in order to fully capture certain features of their semantics.

1984

This thesis is a study in “knowledge ” representation, specifically, how to represent beliefs expressed by sentences containing quasi-indicators. An indicator is a personal or demonstrative pronoun or adverb used to make a strictly demonstrative reference. A quasi-indicator is an expression that occurs within an intentional context and that represents a use of an indicator by another speaker. E.g., if John says, “I am rich”, then if we say, “John believes that he himself is rich”, our use of ‘he himself ’ is quasiindexical. Quasi-indicators pose problems for natural-language question-answering systems, since they cannot be replaced by any co-referential noun phrases without changing the meaning of the embedding sentence. Therefore, the referent of the quasi-indicator must be represented in such a way that no invalid co-referential claims are entailed. I discuss the origin of the problem of quasi-indicators in philosophy of language, and the lack of recognition of its importance by r...

1998

The Review of Symbolic Logic, 2016

This paper presents a proof system for discourse representation theoretic reasoning and dynamic predicate logical inference. It gives a sound and complete characterization of the dynamic declaration of discourse referents and the essentially indexical means to refer back to them. The indexical outlook upon discourse reference is argued to further our understanding of some issues deemed relevant both theoretically (philo-logically) and practically (computationally).

Since Kaplan(1978) first provided a logic for context-sensitive expressions, it has been thought that the only way to construct a logic for indexicals is to restrict it to arguments which take place in a single context -- that is, instantaneous arguments, uttered by a single speaker, in a single place, etc. In this paper, I propose a logic which does away with these restrictions, and thus places arguments where they belong, in real world conversations. The central innovation is that validity depends not just on the sentences in the argument, but also on certain abstract relations between contexts. This enrichment of the notion of logical form leads to some seemingly counter-intuitive results: a sequence of sentences may make up a valid argument in one sequence of contexts, and an invalid one in another such sequence. I argue that this is an unavoidable result of context sensitivity in general, and of the nature of indexicals in particular, and that reflection on such examples will lead us to a better understanding of the idea of applying logic to context sensitive expressions, and thus to natural language in general.

Aristotelian Society Supplementary Volume

When we report a belief about a particular object, we often implicitly indicate the way the believer thinks of that object. This is why two different beliefs can be reported by 'John believes that Cicero is an excellent writer' and 'John believes that Tully is an excellent writer'. One report can be true and the other false, even though 'Cicero' and 'Tully' refer to the same person. To account for this fact, it is natural to assume that the names 'Cicero' and 'Tully' are somehow associated with different 'modes of presentation' of the individual they both refer to. What is the nature of the link between the referring expression in the 'that'-clause (in my examples, the proper name) and the mode of presentation under which the believer is understood to think of the reference? Is it semantic or pragmatic? Arguably, it is the speaker's choice of a particular referring expression (rather than the expression itself) which, in some contexts, conveys the suggestion that the believer thinks of the referent under this or that mode of presentation. To that extent the link between the proper name and the mode of presentation is 'pragmatic'. Many philosophers take this conclusion to be inconsistent with the thesis of Opacity, according to which the mode of presentation in question affects the truth-conditions of the beliefreport. But this is incorrect. It is possible for something which is pragmatically suggested by the use of some expression to affect the truth-conditional content of the utterance in which the expression occurs. Thus the order of the clauses in 'He took off his clothes and went to bed' (or 'They got married and had many children') pragmatically suggests that one event antedates the other, but this indication, though pragmatic, affects the truth-conditions of the utterance. That this is so is shown,

Journal of Pragmatics 39. 2007. 934-959

Journal of Pragmatics, 2008

In this paper, I explore Bach's idea that null appositives, intended as expanded qua-clauses, can resolve the puzzles of belief reports. These puzzles are crucial in understanding the semantics and pragmatics of belief reports and are presented in a section. I propose that Bach's strategy is not only a way of dealing with puzzles, but also an ideal way of dealing with belief reports. I argue that even simple unproblematic cases of belief reports are cases of pragmatic intrusion, involving null appositives, or to use the words of Bach, 'qua-clauses'. The main difference between my pragmatic approach and the one by Salmon (1986) is that this author uses the notion of conversational implicature, whereas I use the notion of pragmatic intrusion and explicature. From my point of view, statements such as ''John believes that Cicero is clever'' and ''John believes that Tully is clever'' have got distinct truth-values. In other words, I claim that belief reports in the default case illuminate the hearer on the mental life of the believer, that includes specific modes of presentation of the referents talked about. Furthermore, while in the other pragmatic approaches, it is mysterious how a mode of presentation is assumed to be the main filter of the believer's mental life, here I provide an explanatory account in terms of relevance, cognitive effects, and processing efforts. The most important part of the paper is devoted to showing that null appositives are required, in the case of belief reports, to explain certain anaphoric effects, which would otherwise be mysterious. My examples show that null appositives are not necessitated at logical form, but only at the level of the explicature, in line with the standard assumptions by Carston and Recanati on pragmatic intrusion. I develop a potentially useful analysis of belief reports by exploiting syntactic and semantic considerations on presuppositional clitics in Romance. #

Proceedings of the 10th European Conference on Artificial Intelligence, 1992

Feature structures are partially speci ed, record-like structures which are employed in many recent grammar formalisms to represent linguistic objects of various kinds. Building on previous approaches to the logical representation of linguistic knowledge, this paper presents a logical language which is su ciently expressive to allow for the encoding of recursive constraints on feature structures. A particular concern of this paper is to show how formulas of the logic can be used to capture the denotation of a grammar considered as a recursive de nition of a class of linguistic objects. However, the logic may be of interest to researchers working in the area of general knowledge representation.