Conceptual exploration

ABSTRACT

Conceptual engineering involves revising our concepts. It can be pursued as a specific philosophical methodology, but is also common in ordinary, non-philosophical, contexts. How does our capacity for conceptual engineering fit into human cognitive life more broadly? I hold that conceptual engineering is best understood alongside practices of conceptual exploration, examples of which include conceptual supposition (i.e. suppositional reasoning about alternative concepts), and conceptual comparison (i.e. comparisons between possible concept choices). Whereas in conceptual engineering we aim to change the concepts we use, in conceptual exploration, we reason about conceptual possibilities. I approach conceptual exploration via the linguistic tools we use to communicate about concepts, using metalinguistic negotiation, convention-shifting conditionals, and metalinguistic comparatives as my key examples. I present a linguistic framework incorporating conventions that can account for this communication in a unified way. Furthermore, I argue that conceptual exploration helps undermine skepticism about conceptual engineering itself.

KEYWORDS:

• Conceptual engineering
• metalinguistic negotiation
• counterconventional conditionals
• metalinguistic comparatives

Acknowledgments

For helpful feedback, I wish to thank audiences at the 2021 Eastern APA and the Auburn University Philosophical Society. Special thanks to Kelly Gaus, Arc Kocurek, and an anonymous reviewer for this journal.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 On metalinguistic negotiation and similar phenomena, see Plunkett and Sundell (2013); McConnell-Ginet (2006, 2008); Barker (2013); Sterken (2020); Stroud (2019); Hansen (2021); Kyburg and Morreau (2000); Barker (2002). Related work goes back to Gallie (1956); Carnap (1956).

2 In understanding matters this way, I am broadly following Plunkett (2015). It is also similar to Cappelen (2018), though he prefers to avoid talk of ‘concepts’ altogether, and simply focus on the meanings of terms. For an overview of recent approaches to conceptual engineering, see Cappelen and Plunkett (2020); Burgess, Cappelen, and Plunkett (2020).

3 This option also helps us see why Stalnaker's (1978) diagonalization technique will not help in our case. The diagonal proposition expressed by Ann's utterance of Pluto is a planet is that the referent of Pluto is in the extension of planet. This, on Stalnaker's picture, is still a traditional possible worlds proposition, a function from worlds to truth values. So this approach would presume that the world settles the meaning of planet. However, we not want to presume this in the present discussion. And more importantly, making such a presumption would misrepresent what Ann and Ben take to be at issue between them. They are not disputing what planet means as determined by some worldly facts, but rather what it should mean – this is what makes it a metalinguistic negotiation.

4 This is not so much intended as a criticism of the Stalnakerian framework, as much as a recognition that it was not yet designed to capture every aspect of communication. I am thus in agreement with Green (2017), when he remarks:

The [common ground]-context approach is […] highly abstract, so merely pointing out that it fails to account for an aspect of communication is an inconclusive criticism. Instead our question should be whether it can be extended or modified to account for such a phenomenon while preserving its spirit. (p. 1589)

Indeed, Stalnaker himself in more recent work has proposed extensions of his original framework that are in some ways similar to what I will propose here, except to account for communication about epistemic uncertainty rather than conceptual indecision. (See Stalnaker 2014, especially chapter 7.)

5 Compare Barker (2002), which uses sets of world-delineation pairs to account for the conversational dynamics of vague predications; also MacFarlane (2016).

6 Some discussions of metalinguistic negotiation, e.g. Plunkett and Sundell (2013), view it as entirely pragmatic, perhaps not requiring any revision to the view that assertive content is simply a set of worlds. On such an approach, what Ann and Ben are ‘really’ disagreeing about – whether the concept of planethood should apply to Pluto – is not part of the literal content of their assertions (see also Sundell 2011, 2017). Questions remain for such a view, most importantly about the mechanism by which this additional, non-literal normative content is conveyed by means of what is literally expressed. Answers may be given here (e.g. Mankowitz 2021), and it is beyond my present aims to argue that they cannot succeed. However, such a pragmatic approach will not easily extend to the embedded uses of metalinguistic material that I will discuss in Sections 3–4. (Similarly, Shan (2010) takes mixed quotation to operate at the level of semantics rather than pragmatics because of embedding constructions. Indeed, there are some commonalities between the present approach to metalinguistic constructions and Shan's account of quotation, though he hews closer to the traditional Stalnakerian line than we do; see Footnotes 3 and 8. Still, the relationship between quotation and metalinguistic uses merits further study.) The framework I develop here allows for a unified linguistic treatment of all kinds of communication about concepts, involving embedded and unembedded metalinguistic constructions. For approaches in a similar spirit, though not covering the same range of language-use, see Armstrong (2016); Muñoz (2019a, 2019b).

7 With contents as world-convention pairs, will we now overgenerate possible metalinguistic uses? An anonymous reviewer gives the following example:

I can't be understood as making a metalinguistic proposal about conventions concerning the word smartphone when I say My laptop is a smartphone (even though my laptop has most of the features one expects of a smartphone – phone (VoIP), email, social media, etc.) – so it arguably should be an accessible reading.

In a suitable context, though, I believe that utterance could be understood as a proposal to change the concept of smartphone so it applies to (at least some) laptops. Imagine it, perhaps, as a response to friends who are bugging you to finally get a smartphone. Granted, the speaker would probably be taken to mean this as a bit of a joke; but the reason for that, I think, is that no one in the conversation seriously thinks the meaning of smartphone is up for grabs to such an extent that it might be modified to apply to laptops. A metalinguistic use, after all, is a proposal to change the conventions adopted in the conversation. For it to succeed, the change must be accepted by the interlocutors; and for it to be understood in the way the speaker intends, the interlocutors must realize that the relevant conventions are up for discussion. For this reason, I am not fully on board with Sterken's (2020) claim that attempts to change language-use commonly cause miscommunication and confusion. While they no doubt occasionally do, it seems to me that, in usual cases, those in the conversation are aware, at least implicitly, that conceptual or linguistic matters are at issue. (This is still compatible with the linguistic interventions often being ‘disruptive’ and ‘transformative’, as she holds.) Overall, then, I think there are no limits in principle on which sentences can be used metalinguistically. It's just a question of how likely it is that the context will make such a use reasonable (cf., Kennedy and Willer 2016).

8 One might think that shifting the world is sufficient, since the world can determine linguistic conventions. That is, one might adopt a diagonalization approach inspired by Stalnaker (1978). While this will work for some cases, it won't capture the meaning of all counterconventionals. The reason is that we can shift to alternative conventions without shifting to a world at which those conventions are in effect, as shown in (i):

• (i) If Pluto were a planet, there would dozens more planets in the solar system, even if no life had ever evolved.

Here, the consequent is evaluated using shifted conventions for planet, but at a world where no life exists and so no linguistic conventions in effect. (Set aside the possibility of representational beings that are not alive.) See also Einheuser (2006) on ‘diagonal conventional conditionals’ and ‘counterconventional conditionals’.

9 For a formal implementation of this approach, see Kocurek, Jerzak, and Rudolph (2020, sec. 4).

10 See, e.g. Bresnan (1973); McCawley (1998); Huddleston and Pullum (2002); Embick (2007); Giannakidou and Stavrou (2009); Giannakidou and Yoon (2011); Morzycki (2011); Wellwood (2014, 2019).

11 She calls them ‘categorizing comparatives’, which is an apt label on my view; still, I stick with the more standard terminology of ‘metalinguistic’. Indeed, one might view the conventionalist framework as building into the semantics many things that have traditionally been viewed as ‘metalinguistic’. Note that she later moved away from the credence view in Wellwood (2019).

12 For a formal implementation as well as more discussion of how this approach compares to previous analyses of metalinguistic comparatives, see Rudolph and Kocurek (2020).

13 See, for example, Muñoz (2019a) and Kocurek, Jerzak, and Rudolph (2020) on attitude reports, and Armstrong (2013, chap. 3) on loose speech.

14 Note that Cappelen doesn't brand himself a full engineering skeptic. He holds that, even in the face of the challenges he identifies, ‘we will and should keep trying’ to engage in conceptual engineering (p. 72).

15 Note that even a view of conceptual engineering that takes it solely to operate on speaker meaning doesn't make such changes trivial or necessarily easy to accomplish, as Fischer (2020) argues drawing on cognitive linguistic evidence.

16 For an overview of possible responses to the implementation problem, see Jorem (2021).