ABSTRACT The truth-conditions of utterances are often underdetermined by the meaning of the sentence uttered, as suggested by the observation that the same sentence has different intuitive truth-values in different contexts. The intuitive difference is usually explained by assigning different truth-conditions to different utterances. This paper poses a problem for explanations of this kind: These truth-conditions, if they exist, are epistemically inaccessible. I suggest instead that truth-conditional underdetermination is ineliminable and these utterances have no truth-conditions. Intuitive truth-values are explained by (...) the effect that all the most reasonable interpretations have on the common ground: An utterance is intuitively true when it is true on all interpretations that answer the question under discussion. (shrink)

People often speak loosely, uttering sentences that are plainly false on their most strict interpretation. In understanding such speakers, we face a problem of underdetermination: there is often no unique interpretation that captures what they meant. Focusing on the case of incomplete definite descriptions, this paper suggests that speakers often mean bundles of propositions. When a speaker means a bundle, their audience can know what they mean by deriving any one of its members. Rather than posing a problem for the (...) interpretation of loose talk, the underdetermination of a uniquely correct interpretation allows for various ways in which the audience can grasp the speaker’s meaning. (shrink)

Generic generalisations like ‘Opioids are highly addictive’ are very useful in scientific communication, but they can often be interpreted in many different ways. Although this is not a problem when all interpretations provide the same answer to the question under discussion, a problem arises when a generic generalisation is used to answer a question other than that originally intended. In such cases, some interpretations of the generalisation might answer the question in a way that the original speaker would not endorse. (...) Rather than excising generic generalisations from scientific communication, I recommend that scientific communicators carefully consider the kinds of questions their words might be taken to answer and try to avoid phrasing that might be taken to provide unintended answers. (shrink)

It seems true to say that Sherlock Holmes is a detective, despite there being no Sherlock Holmes. When asked to explain this fact, philosophers of language often opt for some version of Lewis’s view that sentences like ‘Sherlock Holmes is a detective’ may be taken as abbreviations for sentences prefixed with ‘In the Sherlock Holmes stories …’. I present two problems for this view. First, I provide reason to deny that these sentences are abbreviations. In short, these sentences have aesthetic (...) properties that we should not expect of abbreviations. Second, I argue that the apparent truth of these sentences would not be explained even if they were abbreviations. An alternative is presented that avoids these problems. Following Walton, talk about fiction is viewed as a game of make-believe; following Lewis, interpretations of fiction are modelled using possible worlds. (shrink)

This paper presents Elmar Unnsteinsson’s novel theory of Edenic Intentionalism, on which a speaker cannot refer to an object when the speaker is relevantly confused about its identity. A challenge to the theory is presented and several possible responses considered. The challenge is this: According to Edenic Intentionalism, reference often fails even when speakers seem to refer successfully. Elmar therefore supplements Edenic Intentionalism with an explanation of how communication can succeed without reference. If such an explanation is available, it isn’t (...) clear what need there is for Edenic Intentionalism. (shrink)

The debate over the meanings of indexical expressions has relied heavily on the method of counterexamples. This paper challenges that method by showing that purported counterexamples can often be explained away by appeal to perspective shifts. For these counterexamples to establish anything about indexical reference, we must identify the conditions under which theorists can legitimately appeal to perspective shifts. Some tests for semantic content are considered and it is argued that none of them can tell us when appeal to perspective (...) shift is admissible. The paper then considers how we should proceed if we become convinced that there is no way to identify the content of indexical expressions, offering reasons in favour of a nihilist conception of character over an epistemicist or pluralist conception. (shrink)

It is often possible to know what a speaker intends to communicate without knowing what they intend to say. In such cases, speakers need not intend to say anything at all. Stanley and Szabó's influential survey of possible analysis of quantifier domain restriction is, therefore, incomplete and the arguments made by Clapp and Buchanan against Truth Conditional Compositionality and propositional speaker-meaning are flawed. Two theories should not always be viewed as incompatible when they associate the same utterance with different propositions, (...) as there may be many ways to interpret speakers that are compatible with their intentions. (shrink)

Arguments for context-sensitivity are often based on judgments about the truth values of sentences: a sentence seems true in one context and false in another, so it is argued that the truth conditions of the sentence shift between these contexts. Such arguments rely on the assumption that our judgments reflect the actual truth values of sentences in context. Here, I present a non-semantic explanation of these judgments. In short, our judgments about the truth values of sentences are driven by heuristics (...) that are only fallible reflections of actual truth values. These heuristics can lead to different truth-value judgments in different contexts, even when the sentence at issue is not semantically context-sensitive. As a case study, I consider Sterken’s (Philos. Imprint, 15, 2015a) argument for the context-sensitivity of generic generalisations. I provide a non-semantic explanation of Sterken’s truth-value judgments, which builds on Leslie’s (Philos Perspect 21(1):375–403, 2007; Philos Rev 117(1):1–47, 2008) theory of default generalisation. (shrink)

We outline a novel theory of natural language meaning, Rich Situated Semantics [RSS], on which the content of sentential utterances is semantically rich and informationally situated. In virtue of its situatedness, an utterance’s rich situated content varies with the informational situation of the cognitive agent interpreting the utterance. In virtue of its richness, this content contains information beyond the utterance’s lexically encoded information. The agent-dependence of rich situated content solves a number of problems in semantics and the philosophy of language (...) (cf. [14, 20, 25]). In particular, since RSS varies the granularity of utterance contents with the interpreting agent’s informational situation, it solves the problem of finding suitably fine- or coarse-grained objects for the content of propositional attitudes. In virtue of this variation, a layman will reason with more propositions than an expert. (shrink)

This paper summarises the contributions to our Topical Collection on indeterminacy and underdetermination. The collection includes papers in ethics, metaethics, logic, metaphysics, epistemology, philosophy of science, philosophy of language and philosophy of computation.

This paper introduces a notion of cognitive context-sensitivity, in contrast with linguistic context-sensitivity. When a sentence is cognitively contextsensitive, the truth-value assigned to the sentence can vary with context, without any corresponding shift in the interpretation of the terms or structure of the sentence. The notion will be deployed to explain the context-sensitivity of generic generalizations.

From Aristotle’s puzzle about the indeterminacy of future contingents to Duhem and Quine’s observations about the underdetermination of theory by evidence, the concepts of indeterminacy and underdetermination have been a recurrent theme in philosophy. As well as a continued interest in classic problems, recent years have seen new applications of these notions in various research contexts. This Topical Collection showcases recent work on indeterminacy and underdetermination from diverse branches of philosophy, including philosophy of language, logic, philosophy of mathematics, metaphysics, ethics, (...) metaethics, epistemology, philosophy of science, and philosophy of computation. The collection aims to highlight the similarities and differences between the many manifestations of these phenomena so that insights and conceptual tools developed in each philosophical sub-discipline can stimulate research in the others. (shrink)

According to the Encoding Model, speakers communicate by encoding the propositions they want to communicate into sentences, in accordance with the conventions of a language L. By uttering a sentence that encodes p, the speaker says that p. Communication is successful only if the audience identifies the proposition that the speaker intends to communicate, which is achieved by decoding the uttered sentence in accordance with the conventions of L. A consequence of the Encoding Model has been the proliferation of underdetermination (...) arguments, each of which concludes against some linguistic theory T, on the grounds that, were T true, audiences would be unable to know what was said by utterances of some particular linguistic form, and therefore unable to know what speakers intended to communicate by these utterance. The result, if we accept the conclusion of these arguments, is radical restriction of the domain of viable linguistic theory. This Thesis defends an alternative model according to which there need be nothing encoded in an uttered sentence – nothing that is said by its utterance – for the audience to retrieve. Rather, there are indefinitely many ways to interpret uttered sentences – indefinitely many routes to the propositions that speaker intend to communicate – which proceed through different interpretations of what is said. (shrink)

The lottery puzzle can elicit strong intuitions in favour of skepticism, according to which we ordinary language-users speak falsely about knowledge with shocking regularity. Various contextualist and invariantist responses to the puzzle attempt to avoid this unwelcome result and preserve the competence of ordinary speakers. I will argue that these solutions can be successful only if they respect intuitions of a certain kind, and proceed to judge competing solutions by this criterion.