Search Results

This chapter presents an overview of central issues that must be confronted in developing a decent account of meaning, with various positions that might be taken with respect to them, and with some of the arguments that can be given for and against these positions. Topics addressed include forms of scepticism about meaning, the pros and cons of reductionism, the relationship between language and thought, the compositionality of sentence-meanings from word-meanings, the normative import of meaning, individualism, externalism, and the promise of a deflationary theory.

This chapter considers the phenomenon of meaning from the perspective of Chomsky’s ‘I-linguistics’ and his empirical postulation of the ‘language faculty’. After a sketch of that model, the question is raised as to how meaning should be incorporated within it. In accord with the use-theoretic perspective of this book, an answer is developed whereby the association of I-sounds with I-meanings is achieved by virtue of the conceptual roles of those I-sounds, i.e., their basic acceptance-properties. It is shown that this picture compares favourably with various alternatives, including those suggested by Fodor’s mentalese, Davidson’s view of compositionality, naturalizations of the reference relation, and by Chomsky-style explicitly-represented definitions.

The meaning of any sentence derives from the meanings of its words and from how those words are syntactically combined with one another. But what explains this ‘principle of compositionality’, and what is its import? The answer is due to Davidson: that since we know (from Tarski) how to deduce the truth conditions of sentences from the referents of their words, we should explain it by identifying (or replacing) sentence-meanings with truth conditions and word-meanings with referents. This chapter offers a deflationary alternative according to which the meaning of a sentence is trivially compositional, since it is constituted by the facts concerning its structure and the meanings of its words. An important implication of this idea is that contrary to the writings of Fodor and Lepore, compositionality can place no constraint whatsoever on how word-meanings are engendered.

In order to establish a typological picture of periphrastic negation, this chapter begins from the criteria for periphrasis established by Ackerman and Stump. These are feature intersection, non-compositionality, and distributed exponence. It is argued that while the first two work well for defining periphrasis, the third criterion is not sufficiently robust, and should therefore be substituted by the criterion of multiple exponence. Multiple exponence is a recurrent feature of morphology and therefore, when found in a syntactic construction, it signals its morphology-like status. The chapter analyses the applicability of the criteria by testing them on data from genetically and typologically diverse languages (such as Japanese, and languages within Oto-Manguean, Nilotic, Tungusic, Uralic, Nakh-Daghestanian, and Semitic languages). It shows the extent to which the existing criteria can be applied to languages of different types, and justifies the new criterion (multiple exponence) for identifying periphrasis.

Human languages are far more complex than any animal communication system. Furthermore, they are learned, rather than innate, a fact which partially accounts for their great diversity. Human languages are semantically compositional, generating new meaningful combinations as functions of the meanings of their elementary parts (words). This is unlike any known animal communication system (except the limited waggle dance of honeybees). Humans can use language to describe and refer to objects and events in the far distant past and the far distant future, another feature which distinguishes language from animal communication systems. The complexity of languages arises partly from self-organization through cultural transmission over many generations of users. The human willingness altruistically to impart information is also unique.

This chapter argues for the following theses. There are perfectly possible meanings (ones of a sort one would think are possessed by many vague predicates) which would necessitate a predicate's being gappy. Many arguments against the coherence of truth value gaps depend on a very narrow picture of saying, which ignores the possibility of such things as sui generis denial. Frege/Geach objections to things like sui generis denial dissolve once we observe that ‘not’ and other sentence compounding devices lead a double life, sometimes contributing to sense, sometimes to force. There is a simple compositional story about how (for instance) embedding a denial operator within a ‘force conditional’ makes if not A, then B fit to perform a sort of speech act which, when combined with B's denial, commits one to the aptness of asserting A. The trisection thesis — predicates trisect their domains into three groups, those they are true of, those they are false of, and the rest — is correct. The objection to the trisection thesis —that it is inconsistent with the idea that there are no sharp boundaries in a Sorites series — is not compelling: there is no conception of a ‘sharp boundary’ on which it's plausible, both that there are no sharp boundaries in a Sorites series, and that trisection involves the creation of sharp boundaries. Once we recognize that talk of indeterminacy is contrastive, we also recognize that higher order vagueness is not inconsistent with trisection. We also, once we think of indeterminacy as contrastive, come to see that indeterminacy itself is indeterminate: if it is indeterminate whether p, that indeterminacy itself is not something that is settled, but is itself indeterminate.

This chapter addresses the question how idioms are lexically represented, and the consequences that this has for a theory about the computational system. It argues that there is no straightforward way to use compositionality as a way to make a distinction between expressions that are idiomatic and expressions that are not idiomatic, and that only convention determines which phrases can be called an idiom.

This chapter presents the technical details of the proposed theory of conventional implicatures, which is based in the lambda calculus. Semantic types play an important role in organizing the logical space into the at-issue and conventional-implicature dimensions. The central rules of semantic composition are presented — those that are particular to this theory and those that are required to achieve a functioning compositional theory. The models for the theory are designed to accommodate basic discourse structures and relations. The chapter ends by discussing some of the major consequences of this general logical apparatus, connecting them to specific features of Grice’s original definition of conventional implicatures.

Without contradiction, linguistic meaning is seen as residing in conceptualization and as having a social-interactive basis. Conceptualization is fundamentally imagistic rather than propositional. Instead of there being a unique set of semantic primitives, there are different kinds of elemental conceptions, each basic in its own respect. Certain fundamental grammatical notions are semantically characterized both schematically, in terms of basic cognitive abilities, and prototypically, in terms of experientially grounded conceptual archetypes. Linguistic meanings do not reflect the world in any direct or straightforward manner, but rather embody particular ways of construing the situations described, often involving imagination and mental constructions. There is no specific boundary between linguistic and extralinguistic aspects of lexical meanings (which are better likened metaphorically to encyclopedia entries rather than dictionary entries), nor between semantics and pragmatics. Hence semantics is only partially (not fully) compositional. An expression derives its meaning by flexibly invoking an open-ended set of cognitive domains, i.e. concepts or conceptual complexes of any degree of complexity. These domains are connected in various ways, e.g. by overlap, inclusion, and metaphorical correspondences. There is no clear distinction between domains and mental spaces.

Grammar consists in patterns for assembling symbolically complex expressions. Such expressions are characterized as assemblies of symbolic structures, also called constructions. In large measure, symbolic assemblies are hierarchically arranged: at a given level of organization, component symbolic structures are integrated to form a composite symbolic structure, which can in turn function as component structure at a higher level, and so on. Component structures are integrated both semantically and phonologically, the phonological integration serving to symbolize the semantic integration. Although linguistic meanings are only partially compositional, compositional patterns are essential to the formation and understanding of novel expressions. These patterns are themselves symbolic assemblies, differing from expressions just by virtue of being schematic rather than specific; they are thus referred to as constructional schemas. Abstracted from occurring expressions, these schemas serve as templates for assembling and assessing new ones. A distinction is made between unipolar and bipolar organization, depending on whether the elements involved are delimited solely on semantic or phonological grounds or whether they are delimited by their participation in symbolic relationships. That is, unipolar organization is a matter of phonological or conceptual structure per se, considered independently of symbolic relationships, whereas bipolar organization pertains to the semantic and phonological structures which function in lexicon and grammar. Unipolar and bipolar organization do not have to match at either the semantic or the phonological pole.

Language is both cognitive and sociocultural, consisting in conventionally sanctioned patterns of communicative activity. These patterns take the form of schemas abstracted from usage events by the reinforcement of recurring commonalities. Conventional linguistic units are linked by relationships of composition and categorization (either elaboration or extension) and thus form intersecting networks of great complexity. Expressions are interpreted and assessed for well-formedness through categorization by linguistic units. Through a process of interactive activation, particular units are selected to categorize particular facets of an expression. The total set of categorizing relationships constitutes the expression's structural description, and whether the categorizations involve elaboration or extension determines its degree of conventionality. Despite the absence of explicit prohibitions, this model affords an account of distribution, restrictions, and judgments of ungrammaticality. One aspect of grammatical constructions is their characterization at different levels of specificity, including constructional subschemas incorporating specific lexical items. And since one aspect of lexical items is their occurrence in particular constructions, lexicon and grammar are overlapping rather than disjoint. The model accommodates degrees and kinds of regularity, which decomposes into generality, productivity, and compositionality. Regularities include higher-order generalizations, where sets of categorizations or lexical behaviors are themselves schematized to form productive patterns. Among the phenomena described in this manner are patterns of phonological extension (phonological rules), patterns of semantic extension (e.g. general metonymies), and patterns of morphological realization (like conjugation classes).

The semantic interpretation of complex words is governed by the general principle of compositionality. The semantic scope of morphosyntactic properties of words may be larger than the word on which they are marked. Polysemy is a pervasive phenomenon on the domain of word formation, and can be explained through general mechanisms of meaning extension such as metaphor and metonymy.

This chapter reports on research showing that it may be a universal structural property of human languages that they fall into a class of languages defined by mildly context-sensitive grammars. It also investigates the issue of whether there are properties of language that are needed to guarantee that it is learnable. It suggests that languages are learnable if they have a finite Vapnik-Chervonenkis (VC) dimension (where the VC dimension provides a combinatorial measure of complexity for a set of languages). Informally, a finite VC dimension requires that there be restrictions on the set of languages to be learned such that they do not differ from one another in arbitrary ways. These restrictions can be construed as universals that are required for language to be learnable (given formal language learnability theory). The chapter concludes by pointing out that formalizations of the semantic contribution (e.g., compositionality) to language learning might yield further insight into language universals.

This chapter defines word meaning in terms of speaker meaning and expression. The definition is recursive: in the basic case, individual words and idioms have meaning because of what they are conventionally used to mean; the recursion clause (the basis of compositionality) is provided by the fact that word structures are conventionally used to express certain idea structures. Different kinds of language are distinguished. What words mean today in living languages is dependent on the conventions that evolve from those of prior users of the language.

Quine’s naturalism is not well appreciated for either its force or its detail. It is a scientific view of what science, knowledge and objectivity amount to that takes seriously the fact that such a view is benignly circular: there is no point of view commanding knowledge or science that transcends science. After explaining how Quine himself accepted the view after his famous responses to Carnap, the chapter discusses naturalized epistemology and its implications for the study of language, the specter of indeterminacy, the fundamental importance for naturalism of the inscrutability of reference, and the place within naturalism of ontology. The chapter concludes with the implications for compositionality, truth, semantical holism, instrumentalism, realism, objectivity, the status of psychological characterization and the propositional attitudes.

This chapter introduces and motivates the compositionality requirement on natural language semantics. It argues that quantifying over meanings is neither necessary nor sufficient for meeting this requirement. It introduces Tarski's Convention T and explains how it gives rise to the proposal to use a truth theory to give a meaning theory. It illustrates the basic approach first with a simple context-insensitive language and then with a context-sensitive language. Finally, it gives an explicit statement of the form of a meaning theory, and draws out some of its consequences in responding to some objections.