![Sample our Humanities journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5939/TOC-Subject-Sample-2021-Humanities.jpg"})
ABSTRACT
A fundamental principle of all truth-conditional approaches to semantics is that the meanings of sentences of natural language can be compositionally specified in terms of truth conditions, where the meanings of the sentences’ parts (words/lexical items) are specified in terms of the contribution they make to such conditions their host sentences possess. Thus, meanings of words fit the meanings of sentences at least to the extent that the stability of what a sentence might mean as specified in a theory is in accord with the stability of what a word might mean as similarly specified. In this paper, I shall be concerned with Ludlow’s (2014) idea that, in fact, there need be no such sympathy between words and sentences. He proposes that we can square what he calls a dynamic lexicon, where word meaning is not stable at all, with a traditional truth-conditional approach of the kind indicated, where sentence meaning is delivered via ‘absolute truth conditions’. I share Ludlow’s aspiration to accommodate dynamic features of word meaning with a truth conditional approach, but not his belief that the marriage is an easy deal. Thus, I shall present a problem for Ludlow’s position and show how resolving this problem leads to an alternative picture of how the meaning of a sentence may be truth-conditionally specified with all relevant dynamic features of the lexicon retained.
Notes
1 Think of LF as a syntactic level that realises certain structural or logical features of meaning, such as argument structure, scope assignment, variable assignment, and so on. See May (Citation1977, Citation1985) and Chomsky (Citation1981) for seminal discussion. It doesn’t matter to the following whether LF is viewed as an independently constituted syntactic level or as simply a cipher for whatever syntactic properties fix the relevant features of semantics.
2 The basic thought here is that a semantic theory does not indulge in general conceptual analysis (bracket the issue of meaning postulates), but simply maps metalanguage expressions onto object-language ones with an understood preservation of extension. Of course, a semantic theory should and does do more insofar as it captures algebraic aspects of argument structure, functional properties (tense, modality, etc.), and relations of scope and quantification. Still, every proposal in the truth-conditional traditional has what we might think of as a primitive component that simply maps unitary concepts onto unitary lexical items. See Glanzberg (Citation2014) for an insightful discussion of this issue.
3 Ludlow dwells on kind nominals (artefact and natural), but examples can be readily developed for verbs and adjectives, and all kinds of other nominals. Closed class items appear insusceptible, presumably because they express functional/structural relations, which fall outside of negotiations over what is to count as an instance. See note 2.
4 For broad discussion of underdetermination, see Recanati (Citation2004) and Carston (Citation2002). The basic insight, however, is long-standing; see, for example, Austin (Citation1962) and Ziff (Citation1972).
5 Again, see Asher (Citation2011).
6 Perhaps too swayed. There is a growing literature, initiated by Borer (Citation2005a, Citation2005b), that rejects lexical complexity in favour of lexical roots as unstructured atoms. My present point, however, simply presupposes that all parties accept complex lexical items.
7 Here, I treat a name as if it simply picks out an object. A standard Montagovian treatment takes a name n to pick out a set of properties possessed by some entity (λP.P(n)), such that predicating a property R of n is true iff R is one of the properties P possessed by n. There are other ways of proceeding.
8 Lahav, for one, pitches underdetermination as a problem for compositionality. Such a putative quandary has attracted a number of solutions via the semantic accommodation of context (cp., Reimer Citation2002; and Szabó Citation2001). Similarly, Travis has been taken to raise issues for compositionality and been dealt with in a similar manner (cp., Predelli Citation2004; Kennedy and McNally Citation2010; Lasersohn Citation2012). Regardless of how successful or otherwise justified such responses might be, my current appeal to the style of reasoning Lahav and Travis popularised pertains just to the identification of underdetermination phenomena. For what it is worth, I think it is clear that some notion of compositionality remains unmolested by underdetermination, which only putatively shows that type-level meaning doesn’t fix a context-invariant proposition. Independent work must show whether or not the relevant contextual factors are genuinely extra-linguistic or somehow under a linguistic license. For present purposes, I am simply assuming the latter.
9 Ludlow thanks David Braun for raising the issue. Lepore and Ludwig (Citation2005, 138–42) address essentially the same concern in review of Davidson’s ‘mystery’-transferring stratagem whereby lexical properties of the object-language are simply inherited by the metalanguage.
10 My thanks go to Herman Cappelen for the kind invitation to contribute to this symposium and to Peter Ludlow, for various discussions of these issues over the years.