Conceptual engineers shouldn’t worry about semantic externalism

ABSTRACT

Conceptual engineers sometimes say they want to change what our words mean. If a certain kind of externalism is true, it might be nearly impossible to do that. For some of the external factors that determine meaning, like metaphysical naturalness or past usage, are not within our power to change. And if we can’t change what determines meaning, then we can’t change meaning. I argue that, if this sort of externalism is true, then conceptual engineers didn’t want to change what our words mean anyway. And if they did, they could always engineer externalism out of the language, or engineer a new sense of ‘meaning’ which could be changed. So the truth of externalism does not pose a threat to the possibility of conceptual engineering.

KEYWORDS:

• Conceptual engineering
• semantic externalism
• externalist challenge
• meaning

Disclosure statement

No potential conflict of interest was reported by the author.

ORCID

Jared Riggs http://orcid.org/0000-0002-6598-4990

Notes

1 See e.g. (Burgess and Plunkett 2013a, 2013b) or (Cappelen 2018) for overviews of conceptual engineering in philosophy.

2 For paradigm cases of the kind of externalism I have in mind, see e.g. (Putnam 1975) or (Kripke 1980).

3 I focus in this paper on semantic externalism, ignoring externalism about conceptual content. But I think the responses I give to this externalist worry work just as well with externalism about mental content.

4 (Burgess and Plunkett 2013a) mention this worry, and (Cappelen 2018) puts it forth at much more length, especially in chs. 6–7. And see (Koch forthcoming) for some more discussion of and a response to the challenge.

5 This suggestion appears in (Rorty 1989, 1991).

6 Cappelen (2018) places a lot of emphasis on the role of past usage and causal chains in fixing word meaning, as does (Kripke 1980). (Jackman 2005) and (Ball forthcoming) argue that features of future usage influence meaning. (Lewis 1983, 1984) advocates for something like reference magnetism (although see (Schwarz 2014) for an opposing interpretation), and (Sider 2011) and (Weatherson 2003) endorse the view.

7 This statement of the worry borrows a lot from (Cappelen 2018). But note that some of Cappelen’s worries are less about the metaphysical uncontrollability of meaning change and more about practical issues in organizing changes in people’s thought and talk (see e.g. pp. 200-1, where he compares conceptual engineering to reducing crime or getting speed bumps filled in). I won’t be concerned here with those more practical issues.

8 Burgess and Plunkett (2013a, 2013b) characterize things in this way, (Cappelen 2018) talks a great deal about changing meanings, and (Haslanger 2012) often frames her project in terms of meaning. Some conceptual engineers might also phrase things in terms of reference or content, but I take it that the same externalist worries apply to these notions.

9 In fact there are a few different ways of reading what exactly Haslanger is proposing in that work, but I take it that she is advocating that we change the meaning of those terms at least in some contexts, namely when we are engaged in a certain kind of politically-engaged theorizing. I’ll slide over further exegetical difficulties here.

10 See, for instance, (Thomasson 2017, 11): ‘how we use words matters, given their relations to other aspects of our conceptual scheme, and to our non-verbal behavior.’

11 Thanks for an anonymous referee for this journal for raising this worry.

12 Externalism still poses some threat on this proposal, since conceptual engineers patently don’t have complete control of the public. But this is a practical problem, not the sort of in-principle impossibility externalism seemed to entail.

13 The argument that follows borrows somewhat from (Fodor 2004) and (Schwarz 2014). (Ludlow 2014) also endorses a very strong version of the view.

14 David Chalmers suggested in conversation that some people might object that externalism is not a contingent truth, but is somehow embedded into the very nature of language, such that it must be true of any linguistic activity. Then we would not be able to engineer it away. This objection is worrisome only if there is nothing else, reasonably like a language, that externalism could be false of. Otherwise we could just reframe the proposal from one of engineering externalism out of our language to one of engineering ourselves out of using a language (of which externalism must be true) and into using a schmlanguage (of which externalism could be false). But no facts about the essence of language could establish this claim, which is an empirical one regarding how communicative systems must work.

15 (Ludlow 2014) represents a view from this end of the spectrum.

16 An anonymous referee for this journal suggests that this might not deliver what the title of the paper promises. Specifically, if all I’ve shown is that conceptual engineering is no harder to do per externalism than it is to do per internalism, but this is extremely hard in both cases, then it might not be right to say that conceptual engineers ‘shouldn’t worry’ about semantic externalism. That would be like saying that smokers ‘shouldn’t worry’ about Marlborough Reds, since even though there are some problems with those cigarettes (viz. giving you cancer), every other kind of cigarette has the same problems. I think the analogy does not quite transfer. The question here is just whether externalism poses an additional obstacle to bringing about what conceptual engineers want to bring about. If it doesn’t, then maybe they should be worried, but not about externalism. Compare: it’s very difficult to give up smoking. If chewing nicotine gum makes it even harder to quit, then people who want to quit should be worried about chewing nicotine gum. But if it doesn’t, they shouldn’t. Saying ‘people who want to quit smoking shouldn’t worry about chewing nicotine gum’ would be perfectly acceptable, I think.

17 E.g. (Yalcin 2014) and (Lewis 1980).

18 See (Grice 1991), especially ch. 18.

19 Although sometimes people do contest the way these distinctions are typically understood, e.g. (Davidson 1978) or (Recanati 2003).

20 See also (Cappelen forthcoming)’s ‘Master Argument’ for conceptual engineering, including the premise that ‘We have no good reason to think that the meaning that W ended up with the best meaning W could have,’ for any arbitrary W. On this line of thought, it would be a massive coincidence if ‘meaning’ turned out to be the best term to use for any particular project, including conceptual engineering.

21 To be clear, here I mean that we ‘can’ change the causal powers of tokenings of terms only in a limited sense, namely that completely coordinated social activities could bring that about. I don’t mean that it would be particularly easy for you or me to do so. But the former kind of capability is enough to undermine the claim that what conceptual engineers want to do is uncontrollable or impossible.

22 An anonymous referee for this journal suggests that conceptual engineers might want to change speaker meaning. While speaker meaning is a good candidate in part because externalist factors don’t much influence it, I’m not sure that changing it would suffice in all cases. For one thing, speaker meaning can only target believable contents, but some conceptual engineers want to change non-propositional features of utterances (consider people trying to reappropriate terms like ‘queer’ or ‘crip’). For another, speaker meaning doesn’t directly include what hearers actually come to believe (only what speakers intend for them to come to believe). If what speakers intended to convey changed, but what hearers came to believe did not, then speaker meaning may have changed, but many conceptual engineers’ goals would not be satisfied.

23 Once again eliding the details of how precisely to interpret her proposal.

24 See (Williamson 1994) for an influential defense of the view.

25 In fact, it might be that the feature of terms like ‘woman,’ ‘free will,’ and ‘justification’ that conceptual engineers want to change would be different from the feature of the term ‘meaning’ they want to change, so that there is no single thing, meaning_new, such that conceptual engineers want both to change the meaning_new of ‘woman’ etc. and to change the meaning_new of ‘meaning.’ But I don’t think this scenario poses any special new difficulties, and making room for it in the discussion would be terminologically unwieldly, so I’ll set it aside.

26 See pp. 73–74 for Cappelen’s development of this worry.

27 I am grateful to Andrew Sepielli, Nate Charlow, and a referee for this journal for their feedback on previous drafts of this paper. Thanks also to audiences at the University of Toronto Conceptual Engineering Workshop and the Foundations of Conceptual Engineering conference at New York University.