WOC 2002

                                            dif
Workshop on Context 2002: Semantics vs. Pragmatics Genoa, October 25 & 26 2002

Abstracts



Kent Bach
(San Francisco State University)
Anne Bezuidenhout
(University of South Carolina)
Manuel Garcia Carpintero
(Universitat de Barcelona)
Robyn Carston
(University College, London)
Stephen Neale
Rutgers University
Stefano Predelli
(Universitetet i Oslo)
François Récanati
(Institut Jean Nicod, Paris)
Kenneth Taylor
(Stanford University)



Semantic Illusions

Kent Bach
San Francisco State University


Even though nowadays people see through the old Wittgensteinian slogan that the meaning is the use, there is still a tendency to let judgments about meaning be contaminated by considerations of use, hence for pragmatic phenomena to be confused with semantic ones. There are several reasons for this:
1. If the semantic-pragmatic distinction is not adequately appreciated, information conveyed by the act of uttering a sentence can be confused with the sentence's semantic content. In fact, figuring out what a speaker means in uttering a sentence generally requires taking into account the circumstances in which he is uttering it and possible alternatives to what he uttered, and clearly this goes beyond semantics.
2. It is often supposed that a standard use of an expression or construction is automatically a literal use. This ignores the effect of pragmatic regularities, of which there is a great variety.
3. It is is often thought that an account of the meaning of a sentence must explain our "intuitions" about its truth or falsity under various circumstances. However, these intuitions are semantic data, not semantic facts. They tend to be insensitive to the distinction between what is said and what is implicit in what is said (or, rather, in the speaker's act of saying what he says). These intuitions are, in effect, subject to 1. and 2. above. Semantics should not ignore intuitions, but with help from pragmatics, rather than explain them it can often explain them away.
The illicit intrusion of pragmatics into semantics is aided and abetted by a number of semantic illusions: I will describe and diagnose these illusions, and identify the theoretical and empirical benefits of seeing through them.

back to top



Procedural meaning and the semantics-pragmatics interface

Anne Bezuidenhout
University of South Carolina

Understanding and producing sentences of a natural language are, or at least involve, temporal-linear processes. (I will focus on spoken language production and comprehension and will sometimes lapse into an even narrower focus on spoken language comprehension). The process of articulation forces the speaker to utter a string of sounds in a certain linear order, and the hearer begins processing these in the same order. This is not to deny that at other stages of processing there may be operations that are performed in parallel. However, even parallel processes are temporal ones.
Given the dynamic nature of natural language understanding and production, one might think that it is promising to study natural language from a dynamic perspective, with an eye to describing those features of language that enable speakers and hearers to engage in these dynamic processes. And indeed, beginning with the work of (Kamp, 1981) on Discourse Representation Theory (DRT) and (Heim, 1982) on File Change Semantics (FCS), it has become increasingly popular to offer dynamic accounts of language. DRT and FCS have been lumped together as closely related variants of something called dynamic semantics, e.g., by (Kadmon, 2001). Furthermore, the work of (Groenendijk & Stokhof, 1991) and (Chierchia, 1992, 1995) has been seen as elaborating on and refining this dynamic approach to semantics.
However, (Geurts, 1995, 1999) has argued that it is a mistake to lump DRT together with FCS and other refinements on FCS that treat the meaning of an expression as its context change potential. Geurts argues that DRT is an idealized theory of language understanding, which incorporates a standard model-theoretic semantics. Geurts goes on to criticize dynamic semantics as incoherent. He rejects Chierchia's idea that 'certain aspects of language use enter directly into the compositional core of a semantic system' (Chierchia, 1995: xiii). Geurts thinks that a dynamic theory of understanding/interpretation along with a non-dynamic semantics can adequately deal with the phenomena that have motivated the dynamic approach (principally phenomena involving the interpretation of indefinites and pronouns anaphoric on them).
Similar claims that a static semantics together with a dynamic theory of interpretation are adequate to the task have been made by (Dekker, 2002a, 2002b, Forthcoming). Dekker sees himself as following out a project sketched in (Stalnaker, 1999). There Stalnaker, in a clearly Gricean spirit, says: "if we get clearer about the structure and purposes of discourse, we can better distinguish the idiosyncratic features of particular conventional devices from more general features of the practice that follow from assumptions about what people engaged in it are trying to do. ...And perhaps if we are clearer about the general structure of discourse this will help us defend simpler semantic analyses." (1999: 112).
In this paper I adopt the perspective of Geurts and use this as a framework for discussing the notion of procedural meaning that is a part of Relevance Theory (RT). (Blakemore, 1987, 1988, 1992) argues that certain lexical items (in particular, discourse particles such as 'but', 'however', 'moreover' and inferential 'so') have purely procedural meanings. These expressions encode instructions to process the propositions expressed by the utterances of which they are a part in certain kinds of context, namely ones giving rise to certain contextual effects. In this way the hearer is saved processing effort in the search for an optimally relevant interpretation of the speaker's utterance.
If there are such procedural constraints, they are constraints on processing and belong to a performance system, such as the language understanding and production system. It is not clear that knowledge of such constraints should be thought of as part of the speaker-hearer's semantic competence. This would seem to be building facts about language use into the core semantic system, something that Geurts argues against. On the other hand, clearly these procedural constraints are tied to the processing of specific lexical items, and thus the hopes of accounting for them in terms of general discourse principles may seem to be slim.
One possibility is that such procedural constraints belong to a special purpose linguistic-pragmatics module, which interfaces between the language system proper and the system that controls discourse-level processes. (Carston, 2000) in discussing the work of Ellen Prince seems to entertain the possible existence of such a subsystem. However, Carston concludes: "Nothing substantive hangs on whether we call this subcomponent of linguistic competence 'discourse competence', as Prince does, seeing it as that part of pragmatics which is properly linguistic, or 'procedural semantics', as relevance theorists do, seeing it as a component of 'linguistic semantics' (competence). On the latter approach, the term 'pragmatics' is reserved for that part of utterance meaning which is recovered by inferential processes dependent on guidance of a general principle of communication." (2000: pp.100-101). This paper will investigate whether indeed "nothing substantive" hangs on this choice of terminology, or whether there are good reasons against putting procedural constraints into the core semantic system of a language, as Geurts seems to suggest.

References:

Blakemore, D. (1987). Semantic Constraints on Relevance. Oxford: Blackwell.
Blakemore, D. (1988). "So" as a constraint on relevance. In R. M. Kempson (Ed.), Mental Representations: The Interface Between Language and Reality (pp. 183-195). Cambridge: Cambridge University Press.
Blakemore, D. (1992). Understanding Utterances. Oxford: Blackwell.
Carston, R. (2000). The relationship between generative grammar and (relevance-theoretic) pragmatics. Language and Communication, 20, 87-103.
Chierchia, G. (1992). Anaphora and dynamic binding. Linguistics and Philosophy, 15(2), 111-183.
Chierchia, G. (1995). Dynamics of Meaning: Anaphora, Presupposition, and the Theory of Grammar. Chicago: University of Chicago Press.
Dekker, P. (2002a). Meaning and use of indefinite expressions. Journal of Logic, Language and Information, 11(2), 141-194.
Dekker, P. (2002b). Pronouns in a pragmatic semantics. Journal of Pragmatics.
Dekker, P. (Forthcoming). Grounding Dynamic Semantics. In M. Reimer & A. Bezuidenhout (Eds.), The Semantics and Pragmatics of (In)definites. Oxford: Oxford University Press.
Geurts, B. (1995). Presupposing. Unpublished Doctoral dissertation, University of Stuttgart.
Geurts, B. (1999). Presuppositions and Pronouns (Vol. 3). Amsterdam: Elsevier.
Groenendijk, J., & Stokhof, M. (1991). Dynamic predicate logic. Linguistics and Philosophy, 14, 39-100.
Heim, I. (1982). The Semantics of Definite and Indefinite Noun Phrases. Unpublished Ph.D. dissertation, University of Massachusetts, Amherst.
Kadmon, N. (2001). Formal Pragmatics: Semantics, Pragmatics, Presupposition, and Focus. Oxford: Blackwell.
Kamp, H. (1981). A theory of truth and semantic representation. In J. Groenendijk & T. Janssen & M. Stokhof (Eds.), Formal Methods in the Study of Language (pp. 277-322). Amsterdam: Mathematical Center.
Stalnaker, R. C. (1999). On the representation of context, Context and Content: Essays on Intentionality in Speech and Thought (pp. 96-113). Oxford: Oxford University Press.

back to top



A Gricean Alternative to Williamson's Theory of Assertion

Manuel Garcia Carpintero
University of Barcelona


Abstract: In his paper "Knowing and Asserting" (included as chapter 11 of his recent book, *Knowledge and Its Limits*, OUP 2001), Timothy Williamson defends an account of assertion, according to which it is constitutive of assertion to be governed by the following rule: One must assert that p only if one knows that p. This can be confusedly criticized on the basis that one can make an assertion, and people often do, without knowing the asserted proposition. Williamson's claim is not however that knowing that p is constitutive of asserting that p, but only that it is constitutive of asserting that p that the act is subject to the indicated norm. Now, Williamson's view appears to be at odds with broadly Gricean views of the nature of acts of meaning like assertion, according to which they are constituted by communicative intentions. These views are often criticized (for instance, by Searle, especially in his more recent writings) on the basis that one can make acts of meaning without having communicative intentions; but this criticism belies the confusion already illustrated. In my talk, I want to tentatively defend that assertion is constitutively governed by a stronger rule, which I find more congenial to a broadly Gricean view: roughly, that one must assert that p only if one transmits thereby the knowledge that p. I will argue that this view can claim the same virtues that Williamson invokes in favor of his. I will also appeal for its defense to a form of argument that Williamson uses to support his view. I will argue that the more Gricean view accounts properly for what is paradoxical in the examination paradox.

back to top



Semantics and Conversational Implicature

Robyn Carston
University College London

The linguistic semantic underdeterminacy thesis is widely recognised nowadays. According to this thesis, the encoded meaning of the linguistic expression-type employed by a speaker underdetermines the proposition explicitly expressed by the utterance. Acceptance of this view necessitates certain changes to the orthodox Gricean account of what is said and what is implicated, according to which any instance of maxim-guided pragmatic inference results in the communication of a conversational implicature. Different theorists attempt to accommodate the underdeterminacy facts in different ways. In this talk I compare the generalized conversational implicature approach of Stephen Levinson (1988, 2000) and that of relevance theory (Sperber & Wilson 1986/95, Carston 2002).
According to Levinson, underdeterminacy gives rise to an unacceptable circularity (which he calls 'Grice's circle'): 'what is said seems both to determine and to be determined by implicature' (2000, 186). Furthermore, given an equation of linguistic meaning with the truth-conditional content of an utterance, which Levinson assumes, 'the theory of linguistic meaning [i.e. semantics] is dependent on, not independent of, the theory of communication [pragmatics].' His solution to these circularity problems is to set up a notion of 'utterance type meaning', distinct from expression type meaning, on the one hand, and communicated content, on the other. It is a product of encoded linguistic meaning together with a special class of default pragmatic inferences which he calls generalized conversational implicatures. I shall argue that this move has unwelcome results: (a) it renders the concept of 'conversational implicature incoherent; (b) it forces an unnatural dichotomy between two types of pragmatic inference - the generalized and the particularized - each governed by quite distinct principles; (c) it makes false predictions since certain of the alleged default inferences simply do not arise even when they would be quite consistent with the context; (d) it loses the autonomy of linguistic semantics.
I claim that the solution to the interdependence of saying and implicating proposed within relevance theory is preferable. According to this approach, pragmatic inferences do not inevitably give rise to conversational implicatures but may contribute to the propositions explicitly communicated (explicatures). Explicatures and implicatures are derived by inferential processes of mutual parallel adjustment and are constrained by one and the same (relevance-based) communicative principle, without any recourse to default rules of inference. As a result, there is a single coherent concept of conversational implicatures which, as Grice maintained, are communicated assumptions which do not contribute to the truth-conditional semantics of the utterance, and linguistic semantics is both autonomous from pragmatics and distinct from truth-conditional semantics, which applies to propositional representations of all sorts, including both explicatures and implicatures.

References:
Carston, R. 2002. Thoughts and Utterances: The Pragmatics of Explicit Communication. Oxford: Blackwell.
Levinson, S. 1988. Generalized conversational implicatures and the semantics/ pragmatics interface. Ms. University of Cambridge.
Levinson, S. 2000. Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, Mass.: MIT Press.
Sperber, D. & Wilson, D. 1986/95. Relevance: Communication and Cognition. Oxford: Blackwell

back to top



Radical Contextualism means Radical Indeterminacy

Stephen Neale
Rutgers University


In the first part of this talk, I explain why we have no alternative but to view what is said i.e. what someone, U, said by uttering (or writing) a sentence, phrase, or word X on a particular occasion as typically indeterminate. Cases of determinacy are no more than limiting cases in which linguistic meaning is overbearing, and cannot be treated as "normal" cases from which those involving indeterminacy "deviate". In a sense, this should be unsurprising, for it is virtually dictated by something else that is now beginning to be appreciated thanks to the work of Sperber and Wilson, Récanati, Carston, and others: the role of pragmatic processes in fixing what U said by uttering X is far greater than has been assumed traditionally, going well beyond what is supplied by X's syntactic structure, the meanings of its constituents, and the anchoring of a few hackneyed parameters introduced by (e.g.) indexical, demonstrative and anaphoric pronouns. To the extent that there are aspects of what is said that are not directly traceable to particular semantic features of X, indeterminacy is going to be inevitable at least if what is meant by "indeterminacy" is that there are competing characterizations of what U said among which no principled choice can be made. To this extent, the notion of what U said is not that different from the notion of what U conversationally implicated, which as Grice himself stressed is typically indeterminate. The difference is that purely linguistic facts narrow down the indeterminacy of what U said considerably more than they narrow down the indeterminacy of what U conversationally implicated. Indeed, this is what the saying-implicating distinction ultimately amounts to. In the second part of the talk I turn to the matter of descriptive phrases: definite descriptions, possessive descriptions, and descriptive pronouns in particular. One consequence of the ubiquity of indeterminacy is that all versions of a type of argument used by Wettstein, Récanati, Reimer, Schiffer and others against traditional explicit (ellipsis or expansion-based) accounts of what U said by uttering X, where X contains a so-called incomplete description, such as "the book", are discredited: such arguments presuppose a notion of what is said that is quite simply unsustainable. In the final part of the talk, I turn to noun phrase incompleteness more generally, demonstrating that attempts to construct implicit (or domain-based) alternatives to the explicit approach, if successful at all, are no more pointlessly and perversely formal restatements of the explicit approach. This is so whether domain restrictions are introduced pragmatically (in the way Récanati suggests) or by way of elements in underlying syntax (LF) of a sentence that make no phonetic or graphic appearance in surface syntax (PF) (as Stanley and Szabó argue).
back to top



The Lean Mean Semantic Machine

Stefano Predelli
University of Oslo

According to widespread consensus, the analysis of certain utterances must allow for unarticulated constituents of semantic content, in violation of a principle of full articulation. The central argument in favor of this stance starts from premises reflecting our intuitions about the truth-conditional profile of certain examples, and concludes that any adequate systematic account of such intuitions must proceed along underarticulationist lines.
The main aim of this essay is that of challenging this argument, and of providing an independently plausible account of our intuitions, compatible with the principle of full articulation. I present what I call a 'lean' semantic apparatus, one in which all among the relevant component of semantic content are articulated. I then argue that such an approach is a 'mean' one, in the complimentary sense of the term: in particular, it is able to explain the intuitively correct semantic behavior for the utterances under analysis, and to account for the apparent differences in their truth-conditional profile.
The defense of the principle of full articulation is nevertheless not my main concern: whether underarticulatedness is at all admissible within semantic analysis, regardless of the unsoundness of the main argument in its favor, is a question I do not intend to address. Of more immediate relevance are certain results emerging from the response to the argument against full articulation, pertaining to the role played by context in the analysis of a variety of utterances.

References
Bach, K. 1994. Conversational Impliciture. Mind and Language 9: 124-162.
Borg, E. 2002. Saying what you mean: unarticulated constituents and communication. Manuscript
Carston, R. 1988. Implicature, Explicature, and Truth-Theoretic Semantics. In R. Kempson (ed.) Mental Representations. Cambridge University Press.
Crimmins, M. 1992. Talk About Beliefs. Cambridge, Massachusetts: The MIT Press.
Perry, J. 1986. Thought Without Representation. Proceedings of the Aristotelian Society, Supplementary Volume 60 (1986): 137-152.
Recanati, F. 1993. Direct Reference: From Language to Thought. Blackwell Publishers.
Recanati, F. 2002. Unarticulated Constituents. Linguistics and Philosophy 25: 299-345.
Stanley, J. 2000. Context and Logical Form. Linguistics and Philosophy 23:391-434.
Stanley, J. 2002. Making It Articulated. Mind and Language 17
Taylor, K. A. 2001. Sex. Breakfast, and Descriptus-Interruptus. Synthese 128: 45-61.
back to top



"What is said" and the semantics/pragmatics distinction

François Récanati
Institut Jean Nicod, Paris


I will attempt to show that there are five basic positions concerning the role of context in the determination of truth-conditions, and I will provide arguments against most of them. The five positions can be ordered on a scale from Literalism to Contextualism:
Literalism (in its modern form) holds that the truth-conditions of a sentence are fixed by the rules of the language independent of speaker's meaning. This position must be rejected from the outset because the well-documented phenomenon of semantic underdetermination makes it unavoidable to appeal to speaker's meaning in determining truth-conditions. According to the next two positions, we need to appeal to speaker's meaning in determining truth-conditions, but we do so only when the sentence itself demands it. In other words, 'optional' pragmatic processes are not allowed to affect truth-conditional content. Indexicalism holds that " all truth-conditional context-dependence results from fixing the values of contextually sensitive elements in the real structure of natural language sentences ". The Syncretic View acknowledges the fact that the intuitive truth-conditions may be affected by primary pragmatic processes of the optional variety; but it draws a distinction between what is said in the intuitive sense, and what is strictly and literally said - between the 'apparent' truth-conditions and the 'real' (or minimal) truth-conditions. The last two positions I will consider fully acknowledge the role of pragmatic processes, including pragmatic processes of the optional variety, in the determination of truth-conditional content. They differ from each other in their respective attitudes towards the minimal proposition posited by the Syncretic View. One position, which I call Quasi-Contextualism, simply considers the minimal proposition as a theoretically useless entity which plays no role in communication. Contextualism goes much further and denies that the notion even makes sense.

back to top



The Syntax and Pragmatics of The Naming Relation

Kenneth Taylor
Stanford University


Philosophers of language have lavished attention on names and other singular referring expressions. But they have focused primarily on what might be called lexical-semantic character of names and have largely ignored both what I call the lexical-syntactic character of names and also what I call the pragmatic significance of the naming relation. Partly as a consequence, explanatory burdens have been heaped upon semantics that are better discharged elsewhere. This essay takes some steps toward correcting these twin lacunae. I argue that when we properly distinguish that which belongs to the lexical-syntactic character of names, from that which belongs to the lexical semantic character of names, from that which rests on the pragmatics of the naming relation we expose many misbegotten claims about names and their presumed semantic behavior. I argue, for example, that the solution to Frege's puzzle about the possibility of informative identity statements turns almost entirely on facts about the lexical-syntactic character of names and shows almost nothing about the semantic character of names. Moreover, I show that apparent failure of names to be intersubstitutable in propositional attitude contexts is entirely traceable to largely pragmatic factors and shows nothing in particular about the semantic behavior of names within such contexts. Once we see what explanatory burdens semantic theory does and does not carry many phenomena which have been wrongly thought to be obstacles to a referentialist semantics for names turn out to be nothing of the sort.