La place de la valeur dans le langage naturel

This is a dissertation about words like good, better, bad, beautiful, fun or cruel. More abstractly, this dissertation is concerned with the question of how to accommodate evaluative expressions (evaluatives for short), which the former words exemplify, within a theory of natural language meaning. To do that, I recruit tools from both metaethics and formal linguistics. From the point of view of metaethics, this dissertation defends a form of NON-FACTUALISM, since it defends that evaluatives and the sentences in which they paradigmatically appear play a different role in communication from the standard role assigned to declarative sentences, namely, describing reality and imparting information about it. From the point of view of formal linguistics, this work relies on advances in the semantics of gradable adjectives, which evaluatives belong to, to gain insight into their semantic properties.

The intuition that evaluatives have special and intriguing features is, of course, not new: evaluatives have caught the attention of philosophers throughout history. But those features were not studied as pertaining to words like good or beautiful, but rather as properties of the very idea or concept of goodness or beauty. Indeed, some of the most recalcitrant problems of philosophy have to do with figuring out the place of value within certain world-views: the problem of evil, the nature of moral and aesthetic judgment, the characterization of the virtuous life or the definition of knowledge are all topics that in one way or other touch upon this issue.

With the turn of the XXth century, philosophers became preoccupied with language, and so a concern for value became a concern for the way in which value is represented in natural language, that is, a concern for evaluative expressions. Simultaneously, the development of formal linguistics and philosophy of language offered new and sophisticated tools to explore natural language meaning. As it turns out, evaluative expressions are interesting not only insofar as they vehicle traditional problems of philosophy, but are also intriguing in their own— linguistic—right. In this section, we first look at the distinction between fact and value, and then we turn to the difference between describing and evaluating .

Throughout the history of philosophy, one finds time and again the intuition that there exists an important difference between facts and values, and furthermore, that a gap separates the realms of the factual and the evaluative. Hume, Moore or Wittgenstein are famous defenders of this idea. Hume’s Enquiry (1739/1999) contains his notorious denial that OUGHT can be inferred from IS—what came to be known as ‘Hume’s law’, or the ‘is-ought problem’. In his Principia Ethica (1903/1993), Moore held the view that the property of being good cannot be reducible to any natural property, and to argue for that view he coined the famous ‘open question argument’: for any set of properties that one might offer as a definition of the word good, it seems to always remain an open question whether something that falls under that definition is indeed good. Finally, Wittgenstein’s Lecture on Ethics (1929/1965) contains a powerful statement of the apparently insurmountable gap between the factual and the evaluative or normative.

What makes these realms, the realm of the factual and the realm of the evaluative, so different? Here is where considerations about language have appeared to offer a different angle towards answering that question. If instead of considering the difference between the realms of facts and values we look at the difference between factual and evaluative language, there is a double hope: first, that attention to evaluative language will tell us something about evaluative concepts—that by studying the word good we might learn something about “the” good. And more specifically, following the first Wittgenstein, one might hope that attention to evaluative language might reveal some issues in ethics as pseudo-problems.

Regardless of whether the latter hope was in fact fulfilled, philosophers working in the analytic tradition started to distinguish, chiefly after Moore and following the development of ordinary language philosophy, the domains of NORMATIVE ETHICS and METAETHICS as two distinct sub-fields of philosophical enquiry (Darwall et al. 1992). The first remained preoccupied with the traditional questions of ethics, such as whether it is ever permissible to lie, while the second turned to the question about the meaning of ethical terms, such as permissible.  Questions about the meaning of moral terms are questions about moral terms, ontology, psychology and knowledge.

In recent decades, the scope of the distinction between normative ethics and metaethics has broadened to capture a similar contrast between enquiry into other evaluative domains versus enquiry into the language in which we conduct the former kind of enquiry. Thus, METAEPISTEMOLOGY has been distinguished from epistemology; and METAAESTHETICS from aesthetics (e.g. Alston 1978; Schellekens 2008). Authors like Chrisman (2016) speak more generally of METANORMATIVE THEORY. Moreover, by turning their attention to evaluative language, the metanormative theorist can (or rather must) take advantage of the developments of contemporary formal linguistics, which offers increasingly sophisticated tools to study the semantics of natural language expressions. What is really appealing is that evaluative language is not just of interest to the metanormative theorist. Much to the contrary, evaluatives are important case studies for central topics in formal semantics and pragmatics and the philosophy of language in general, such as the semantics of modal, gradable and vague expressions, the interaction between context and content or the performative aspects of language.