Expressivism, meaning, and all that

Abstract

It has recently been suggested that meta-normative expressivism is best seen as a meta-semantic, rather than a semantic view. One strong motivation for this is that expressivism becomes, thereby, compatible with truth-conditional semantics. While this approach is promising, however, many of its details are still unexplored. One issue that still needs to be explored in particular, is what accounts of propositional contents are open to meta-semantic expressivists. This paper makes progress on this issue by developing an expressivist-friendly deflationary account of such contents.

Keywords:

• Expressivism
• deflationism
• propositions
• meta-semantics
• meta-ethics

Acknowledgements

I owe special thanks to Matthew Chrisman, Guido Ehrhardt, and Mike Ridge for numerous conversations on the topic of this paper and for their more than generous support. For comments and discussion, I would also like to thank Vuko Andrić, Cameron Boult, Lars Dänzer, Christoph Fehige, Ben Ferguson, Simon Gaus, Susanne Mantel, Geoffrey Sayre-McCord, Robin McKenna, Stephan Padel, Joey Pollock, Huw Price, Peter Schulte, Christine Tiefensee, as well as an anonymous referee for the Canadian Journal of Philosophy, and audiences at work in progress events in Edinburgh and the Practical Philosophy Workshop in Saarbrücken. I owe the title of the paper to Robin McKenna.

Notes

1. Forms of expressivism so characterized are developed by Simon Blackburn (1998), Allan Gibbard (2003), Horgan and Timmons (2006), Michael Ridge (2014), and Mark Schroeder ([2008], though he does not accept it).

2. This way of arguing against expressivism and highlighting the cost of abandoning truth-conditional semantics is prominently championed by Mark Schroeder (2008, 2010).

3. Using Psychologized Semantics, Schroeder (2008) develops a view that ends up allowing expressivists to assign contents derivatively in a way that provides a unified semantics operating on those contents. He argues, though, that this requires too many problematic commitments, without being attractive enough to be a feasible alternative to truth-conditional semantics.

4. Of course, such independent linguistic evidence could then help semantic expressivism. Nate Charlow (2014, 2015), for example, argues that the semantics of imperatives already requires a different semantic approach, which helps semantic expressivism as well.

5. Of course, philosophers have worked hard to develop such alternatives (see e.g. [Blackburn 1993; Gibbard 2003; Horgan and Timmons 2006; Schroeder 2008; Baker & Woods 2015; Charlow 2015]). I will stay neutral on whether such an account works, as surely it is better if expressivism is not hostage to fortune to one.

6. These are e.g. Matthew Chrisman (2016), Michael Ridge (2014) and Alex Silk (2013).

7. In some sense, of course, semantics is also concerned with an enterprise like this. So, for example, when a semanticist spells out the notion of a truth-condition in terms of a set of possible world, she is trying to cash out a semantic notion in non-semantic terms. This is different, though, from the enterprise that concerns meta-semantics. After all, even if we concede that a truth-condition is a set of possible worlds, there are many different interpretations of what we could be saying when we concede this. If we are instrumentalists about semantics, for example, we might think that all we are doing is to introduce a useful, but strictly speaking false, fiction. On the other hand, we might be realists and hold that we are talking about what meaning really consists in. It is in this sense, in which meta-semantics is concerned with the named enterprise (compare: in one sense normative ethics tries to ‘cash out’ the moral notions in non-moral terms as far as possible. But, there are many different meta-ethical interpretations of what is going on when we do this).

8. I borrow this phrase from Paul Horwich (1998a: 5).

9. See e.g. (Sellars 1974; Davidson 1990; Field 1994, 2001; Shiffer 1996, 2003; Horwich 1998b; Williams 1999).

10. Ridge (2014) and Schroeder (2013) have explored views that allow expressivists to be realists about propositions (Ridge’s preferred account is a version of Scott Soames’ view [2010]). My aim is not to argue against these views, but to put another option on the table.

11. Of course, the label ‘deflationism’ is used in a variety of ways in the philosophical debate and my use of this label might cut across the use of other people. What I have in mind are views committed to a particular way of deflating certain kinds of linguistic phrases that it is tempting to read in very theoretically committing ways. Specifically, these views do not deflate such notions via an representationalist order of explanation (which is also provides a possible way for ‘deflation,’ if one argues that the phrase is to be explained in terms of what it represents, but that theoretical requirements for the thing to be instantiated are minimal). Instead, they start with a non-representational explanation. Views that are deflationary in this sense are to be found e.g. in (Ramsey 1927; Grover, Camp, and Belnap 1975; Price 1988; Brandom 1994; Field 1994; Horwich 1998b). Note that such views can, but need not deny that, for example, truth is a property (or reference a relation, etc.). However, they will use their non-representational account to explain what it means to talk about that property (relation, etc.) in a way that doesn’t increase the metaphysical commitments of the account.

12. I explain how the account offered here also covers mental content in Köhler (2017). There I argue that the account covers the content of all propositional attitudes and not only allows expressivists to hold that normative judgements are beliefs in normative propositions, but also that there are other propositional attitudes with normative propositions as their contents, such as e.g. desires.

13. Blackburn (1998, 77–83) and Gibbard (2003, 180–196) have suggested sympathies towards deflationary accounts of propositions, but neither has provided the details of such an account or how it would fit with expressivism.

14. For reasons of simplicity I omit the time parameter here and in what follows.

15. Compare: a pro-sententialist theory of truth (e.g. [Grover, Camp, and Belnap 1975]) holds that there is a relation between the truth-predicate and some entity, but it does not hold that this entity is truth and so invokes a different relation to the one that would be employed by a representationalist about truth.

16. Note that this further step requires us to endorse a deflationary account of reference for the account not to collapse into a form of fictionalism. Note also, that this step is not required by the Sellarsian account. What I’ve said before only shows that on the Sellarsian account, that-clauses legitimately behave in a syntactic manner like referential terms. This view is compatible with holding that that-clauses do not in fact refer to anything – that the surface features of language are deceiving in this case. This would be a deflationary account of that-clauses, but not, strictly speaking, of propositional contents.

17. Some readers might wonder what kind of compositional semantics for ‘means that’ fits with the Sellarsian account. First of all I should highlight that the Sellarsian account itself should be understood as a meta-semantic account of that-clauses, as well as of ‘means.’ Even so, the Sellarsian account might itself have certain implications for semantics, depending on how we spell it out. For example, on Sellars’ original account of meaning-attributions, the that-clauses would be an indexical predicate and ‘means’ a specialized form of the copula. In this case that-clauses would have to be accounted for in semantics via extensions. However, if we follow some of my suggestions here as to how that-clauses function syntactically and how propositions might fit into the Sellarsian account, that-clauses would be singular terms and ‘means’ a relational term. In this case, that-clauses would require a referent. While the question which of these approaches we should endorse is important for further research, it is not a question that I will be concerned with here. It seems to me, though, that the availability of these options shows that the account could plausibly be compatible with different approaches to the compositionality of ‘means that’ depending on the further commitments one will want to endorse regarding that-clauses. Note, though, that in both cases, a deflationary account of that-clauses will require to deflate further notions (e.g. ‘extension’ and ‘reference’ in the cases above). I do not take this to be a problem in the context of my investigation, though, as an expressivist who wants his account to fit with truth-conditional semantics will need those anyways. I’d like to thank an anonymous referee for drawing me out on this issue.

18. Proponents of this view are e.g. Ned Block (1986), Hartry Field (1978), and Gilbert Harman (1999). It is related to inferentialism in that both emphasize role for content, but differs from inferentialism in that it emphasizes causal-functional, rather than inferential role (Brandom [1994] is the most well-known proponent of inferentialism). Note that the label ‘conceptual role semantics’ is misleading, given that the account explains in virtue of what mental states have their contents and so takes up the space in the philosophy of mind that meta-semantic accounts take up in the philosophy of language.

19. What follows is, of course, only a rough sketch of the account.

20. As already remarked in fn. 12, I fill this theoretical gap in (Köhler 2017).