Speaker’s reference, stipulation, and a dilemma for conceptual engineers


Advocates of conceptual engineering as a method of philosophy face a dilemma: either they are ignorant of how conceptual engineering can be implemented, or else it is trivial to implement but of very little value, representing no new or especially fruitful method of philosophizing. Two key distinctions frame this dilemma and explain its two horns. First, the distinction between speaker’s meaning and reference and semantic meaning and reference reveals a severe implementation problem for one construal of conceptual engineering. Second, the distinction between stipulating meanings and conceptually analyzing allows us to see why, on another construal of what conceptual engineering involves, the practice is neither a new nor neglected philosophical methodology. The article also argues that semantic externalism is not the root of the implementation problem for conceptual engineering, and that the usual rationale for adopting the practice, one that ties its value to the amelioration of “conceptual defects”, is unsound.

This is a preview of subscription content, access via your institution.


  1. 1.

    These include Haslanger (2000), Cappelen (2018), Eklund (2014), Richard (2014), Nado (2019), Brun (2016), Chalmers (2011) and Burgess and Plunkett (2013a, b).

  2. 2.

    Most present-day advocates of conceptual engineering trace their lineage to Carnap (1950) and the method of “explication”. Most also draw a contrast between conceptual engineering/explication and conceptual analysis, and some are quite skeptical of the viability and fruitfulness of the latter. This is true of Nado (2019) and especially of Cappelen (2018), who has harsh things to say about philosophical work addressing the Gettier problem, for example, work that many regard as involving paradigmatic instances of conceptual analysis. (See Cappelen 2018, 26–27).

  3. 3.

    Later in the main text, I discuss Haslanger’s (2000) account of the concept ‘woman’ in particular. Still later, I discuss an alternative interpretation of this account, one given by Haslanger herself in Haslanger (2006) and again in Haslanger (2010) (though the alternative interpretation in Haslanger 2010 is applied primarily to her account of race concepts).

  4. 4.

    Cappelen (2018) ends up rejecting the view that the targets of conceptual engineering are concepts, proposing instead that these targets are the intensions and extensions of terms. This proposal is similar to the one I will soon make in the main text.

  5. 5.

    The distinction plays an important role in (a) Grice’s (1989) theory of implicature and the related demise of “ordinary language philosophy”, (b) the proper assessment of semantic contextualism, including epistemic contextualism, and the prospects for semantic minimalism (c) evaluating Radical Millianism about proper names, (d) determining whether the referential/attributive distinction, relative to uses of definite descriptions, is semantically significant, (e) understanding the significance of results in experimental philosophy, and (f) assessing theories of fictional discourse, metaphor, generic language, pejoratives, and slurs.

  6. 6.

    Haslanger (2000) is perhaps Nado’s inspiration here. Haslanger, in expressing advocacy for ameliorative analyses of gender and race concepts, writes that “[i]n the limit case of an analytical [ameliorative] approach the concept in question is introduced by stipulating the meaning of a new term, and its content is determined entirely by the role it plays in the theory” (Haslanger 2000, 224; my emphasis). Note, however, that Haslanger is here describing stipulating the semantics of a new term. I discuss this kind of stipulation in the next section of the main text.

  7. 7.

    Nado draws a distinction between semantic and functional conceptual engineering, one useful for understanding and assessing an objection to conceptual engineering due originally to Strawson (1963) to the effect that conceptual engineering changes the subject instead of shedding light on the subject it intends to address. Nado convincingly argues that this objection doesn’t apply to functional conceptual engineering, whose aim is to preserve the function(s) of a concept, not its semantics. Despite this, functional engineering, even on Nado’s account, proceeds via semantic change: the functional engineer tries to preserve function by changing meaning. So there is a sense in which functional engineering is no less semantic than semantic engineering. In both cases, the mechanism of conceptual change is a change in the semantics of a concept/term. This is why, despite Nado’s semantic/functional distinction, I regard her view as compatible with, and perhaps even committed to, the view that conceptual engineering involves changes in the semantic meaning and reference of a term.

  8. 8.

    On Nado’s account, conceptual engineers seek “efficient” concepts to replace the defective ones with which they start. This is a matter of the new concepts being better suited to serving various functions and purposes. The issue of efficacy is mostly irrelevant to the point I am making in the main text, which can be put as follows. There are two questions for any Nado-inspired conceptual engineer. First, can the new concept replace the old by stipulation alone? And, second, is the replacement a more efficient concept than the concept it replaces? The argument I am making in the main text is that the answer to the first question is often “no”. So, often, the question of whether an instance of attempted conceptual engineering is successful is answered in the negative, before the question of efficacy even arises.

  9. 9.

    Taken at face value, the proposal is actually the obviously false proposal that nothing is a dog unless it is a cat, and vice versa. Clearly, that’s not something one could stipulate to be so, nor does it concern concepts or the meanings of words. Note, though, that this is true of Nado’s hypothetical proposal about free will too: it says something about free will and H2O molecules, something no one could stipulate to be so, and something not about concepts or words. Despite this, in the main text, I am regarding the hypothetical proposal as a revisionary analysis of the concept ‘free will’, understanding concepts as (at least) the semantic meanings of terms. However, my view is that another problem with enthusiasm for revisionist analysis is that it is based on a deeply flawed picture of (non-revisionary) philosophical analysis, which is not a method of analyzing concepts at all. Philosophical analysts analyze knowledge, freedom, etc., not the concepts of these things.

  10. 10.

    This is to imagine something that Richard perhaps rejects, namely that ‘free action’ “determinately denotes” some property distinct from the property described in the revisionist’s stipulation. But this doesn’t affect the point I am making, which is just that if ‘free action’ determinately denotes, then mere stipulation is powerless to change this. In any case, even if ‘free action’ fails to determinately denote, stipulation is also powerless to change this fact about the semantics of the term, or so I would argue.

  11. 11.

    I have borrowed this abbreviated version of Haslanger’s analysis from Nado (2019). See Haslanger (2000) for the full version.

  12. 12.

    Again, I don’t deny that a term’s meaning and reference can change. The question is: What will turn a mere stipulation about the semantics of ‘woman’ into a genuine semantic shift? Haslanger can try to stipulate a new semantics for ‘woman’, and some of her readers can speaker-refer and speaker-mean in accordance with her stipulation. But without more details filled in, it is not at all clear that any of this activity will ever implement Haslanger’s desired semantic change.

  13. 13.

    I am open to the idea that, in some cases, perhaps especially in contexts of social/political speech, stipulative introductions do more than offer a convenient shorthand for lengthier descriptions. The introduction of the term “sexual harassment”, for example, might well have been a remedy for what Fricker (2007) describes as “hermeneutical injustice”. Still, I think these are special cases. Generally, stipulative introduction of new terminology is a remedy for inconvenience.

  14. 14.

    Although the example of ‘credence’ is not exactly stipulative introduction, but instead an example of a superficially different variety of stipulation that I describe in the next section as “stipulative addition”.

  15. 15.

    A similar motivation for conceptual engineering is described in Eklund (2015).

  16. 16.

    Some advocates of conceptual engineering suggest that some of what gets called “conceptual analysis” in philosophy is actually, in at least some cases, unwitting conceptual engineering. But, since stipulating the semantics of a technical philosophical term is so obviously a different sort of undertaking than specifying the conditions on the application of an existing term, it is very hard to take this suggestion seriously. Philosophers know when they are simply stipulating meanings and when, by contrast, they are proposing analyses.

  17. 17.

    I think insensitivity to the speaker’s/semantic meaning distinction also creates confusion about examples like ‘fish’ and ‘fruit’. Someone who says “don’t put that tomato in the fruit salad, it is not a fruit!” says something semantically false. She might nonetheless communicate—speaker mean—something true.

  18. 18.

    I have taken this example from Brun (2016), who uses it to make a somewhat different point.

  19. 19.

    Haslanger has long argued that gender is a “socially founded”, but fully objective, feature of people. But, as I read her, this is supposed to be a discovery about the nature of gender. It is not simply that she has redefined ‘woman’ so that it now refers to a socially founded feature; it is rather that it does, right now, so refer, and has all along. Viewing her analyses as revelatory nicely explains this connection between Haslanger’s social constructionism and the goal of her analyses. However, revelatory analysis can also potentially reveal that a feature one might regard as socially founded is not in fact socially founded. For example, such analysis might reveal that individual races are not socially founded features, and are, instead, something like breeding populations. For excellent critical discussion of both revelatory analysis (which he calls “revisionism”) and the idea that races are breeding populations, see Glasgow’s (2009) A Theory of Race.

  20. 20.

    In Saul 2006 Saul suggests that there’s a difference between an intuition that some perhaps merely hypothetical x is an F and a sincere application, by an ordinary speaker, of the termF’ to some actual x. Her complaint about Haslanger’s analyses, construed as descriptive analyses, is that they conflict with ordinary applications (she calls this the “linguistic counterintuitiveness worry”), and she grants that their conflict with intuitions proves little. But it is no more plausible that the criteria ordinary speakers employ in applying the term ‘woman’, say, are the conditions on its correct application than it is to think that correct analyses can’t conflict with widespread intuitions about hypothetical cases.

  21. 21.

    Some philosophers, including an anonymous referee for this journal, are not as comfortable as I am with the possibility of true but deeply counterintuitive philosophical analyses. I should therefore emphasize that the dilemma for conceptual engineers posed in earlier sections does not depend on the views I am here defending concerning what I am calling “revelatory analysis”. One can agree with me about the dilemma without agreeing with me about revelatory analysis, or, in particular, the interpretation of Haslanger as offering a revelatory analysis of womanhood.

  22. 22.

    For the same reason, we ought to take seriously the idea that the JTB theory of knowledge is the correct theory of knowledge, as Weatherson (2003) and Hetherington (1999) have argued, despite the existence of near universal intuitions to the contrary.

  23. 23.

    Cappelen 2018 (see especially Chapter 7) discusses the apparent incompatibility between externalism and conceptual engineering. This is also the topic of Koch (2018).

  24. 24.

    Cappelen (2018) claims that, on externalism, “[e]ven if we had all the information about the metasemantics of a term (about the use patterns, the histories, the sources of information, the interaction between the experts, etc.), it would appeal to factors that are in large part out of our control. For example, past facts play a role in determining the meaning of terms, but we can’t change the past” (74). This suggests that, if externalism is true, intentional semantic changes would require the ability to change the past. I don’t think Cappelen actually believes this suggestion—elsewhere, he says things that appear to contradict it. But the suggestion is there in the quoted passage.

  25. 25.

    I think of linguistic stipulations as simply intentions, usually made explicit in a stipulative act, to use a term in a particular way.

  26. 26.

    Koch (2018) would classify this as an externalist view, given that he defines internalism as the view that the semantic meaning of a speaker’s term is determined by that speaker’s mental states and nothing more. (See Koch 2018, 3.) I doubt there are very many internalists when the view is defined this way. In any case, it is worth noting that certain forms of metasemantic descriptivism (even if not internalist on Koch’s definition) are just as problematic for conceptual engineering as some anti-descriptivist views.

  27. 27.

    Take a metasemantic view that holds that a term’s semantic meaning and reference supervenes on actual use, where an actual use of a term is conceived as a speaker actually applying the term to this or that object or property. Since it’s perfectly possible for speakers to apply a term in a way that fails to comport with any given rule for its application, including a stipulated rule to which all speakers agree, it is also possible, on such metasemantic views, for there to be a mismatch between a term’s actual semantic meaning and reference and any stipulated semantics for it, even if every last speaker agrees to the stipulation and tries to abide by it.

  28. 28.

    Koch (2018) argues that if something like Evans’s (1973) view is correct, and our terms refer to the “dominant causal source” of the attitudes we express when using them, then we have what Koch calls “collective long-range control” over what our terms refer to. And we can exercise this control, Koch claims, simply by acting as if our terms have the referents we want them to have. Eventually, these acts can bring it about that the terms do refer to what we want them to refer to, by changing the causal source of the attitudes we express when using the terms. But the problem with Koch’s picture, even granting an Evans-style metasemantics, is that there is no telling, in advance, that a plan to get large numbers of speakers to use t as if it refers to x will actually turn x into the dominant causal source of the attitudes speakers express when using t. People can make widespread mistakes, for example, using t even in relation to things that are not x’s, despite the intention to use t only in relation to x’s.

  29. 29.

    Conceptual engineers might concede that it is unclear how to implement stipulative semantic revisions but deny that this matters. What matters, they might claim, is using the relevant terms in the stipulated way, to merely speaker-mean and speaker-refer in accord with the stipulation. This gives up on the idea that conceptual engineers seek semantic changes, but some of conceptual engineering’s defenders (e.g. Nado 2019) seem eager to give up this idea anyway. The deeper problem with the concession is that it amounts to recommending that we not aim to speak the literal truth when philosophizing using engineered terms. I hope to elaborate this problem in future work. For a problem that I take to be similar, see Cappelen’s (2018) discussion of reasons to avoid conceptual engineering merely for the sake of exploiting the “lexical effects” of a term (Cappelen 2018, 130–136).

  30. 30.

    The most puzzling aspect of Cappelen (2018) is its insistence on the inscrutability of metasemantics and meaning change, on the one hand, and its advocacy of conceptual engineering, conceived as stipulative revision, on the other. To his credit, Cappelen is aware that this will strike many as puzzling. See Koch (2018) and Deutsch (2019) for criticisms of Cappelen’s attempted resolution of this puzzle.

  31. 31.

    See Cappelen (2018, 34).


  1. Brun, G. (2016). Explication as a method of conceptual re-engineering. Erkenntnis, 81(6), 1211–1241.

    Article  Google Scholar 

  2. Burgess, A., & Plunkett, D. (2013a). Conceptual ethics I. Philosophy Compass, 8(12), 1091–1101.

    Article  Google Scholar 

  3. Burgess, A., & Plunkett, D. (2013b). Conceptual Ethics II. Philosophy Compass, 8(12), 1102–1110.

    Article  Google Scholar 

  4. Cappelen, H. (2018). Fixing language: An essay on conceptual engineering. Oxford: Oxford University Press.

    Google Scholar 

  5. Carnap, R. (1950). Logical foundations of probability. Chicago: University of Chicago Press.

    Google Scholar 

  6. Chalmers, D. (2011). Verbal disputes. Philosophical Review, 120(4), 515–566.

    Article  Google Scholar 

  7. Deutsch, M. (2019). Fixing language: An essay on conceptual engineering by Herman Cappelen. Analysis, 79(3), 574–578.

    Article  Google Scholar 

  8. Eklund, M. (2014). Replacing truth? In A. Burgess & B. Sherman (Eds.), Metasemantics: New essays on the foundations of meaning (pp. 293–310). Oxford: Oxford University Press.

    Google Scholar 

  9. Eklund, M. (2015). Intuitions, conceptual engineering, and conceptual fixed points. In C. Daly (Ed.), The Palgrave handbook of philosophical methods (pp. 363–385). London: Palgrave Macmillan.

    Google Scholar 

  10. Evans, G. (1973). The causal theory of names. Proceedings of the Aristotelian Society Supplementary, 47, 187–225.

    Article  Google Scholar 

  11. Fricker, M. (2007). Epistemic injustice. Oxford: Oxford University Press.

    Google Scholar 

  12. Glasgow, J. (2009). A theory of race. New York and London: Routledge.

    Google Scholar 

  13. Grice, H. P. (1989). Studies in the way of words. Cambridge, MA: Harvard University Press.

    Google Scholar 

  14. Haslanger, S. (2000). Gender and race: (What) are they? (What) do we want them to be? Noûs, 34(1), 31–55.

    Article  Google Scholar 

  15. Haslanger, S. (2006). What good are our intuitions? Philosophical analysis and social kinds. Proceedings of the Aristotelian Society Supplementary, 80(1), 89–118.

    Article  Google Scholar 

  16. Haslanger, S. (2010). Language, politics and ‘the folk’: Looking for ‘the meaning’ of ‘race’. The Monist, 93(2), 169–187.

    Article  Google Scholar 

  17. Hetherington, S. (1999). Knowing failably. Journal of Philosophy, 96(11), 565–587.

    Article  Google Scholar 

  18. Koch, S. (2018). The externalist challenge to conceptual engineering. Synthese. https://doi.org/10.1007/s11229-018-02007-6.

    Article  Google Scholar 

  19. Nado, J. (2019). Conceptual engineering, truth, and efficacy. Synthese. https://doi.org/10.1007/s11229-019-02096-x.

    Article  Google Scholar 

  20. Richard, M. (2014). Analysis, concepts, and intuitions. Analytic Philosophy, 55(4), 394–406.

    Article  Google Scholar 

  21. Saul, J. (2006). Gender and race. Proceedings of the Aristotelian Society Supplementary, 80(1), 119–143.

    Article  Google Scholar 

  22. Strawson, P. (1963). Carnap’s views on conceptual systems versus natural languages in analytic philosophy. In P. A. Schilpp (Ed.), The philosophy of Rudolph Carnap (pp. 503–518). Chicago: Open Court.

    Google Scholar 

  23. Weatherson, B. (2003). What good are counterexamples? Philosophical Studies, 115, 1–31.

    Article  Google Scholar 

Download references


Students in two undergraduate classes at HKU—Philosophy of Language (Fall 2018) and Metaphilosophy (Spring 2019)—were the first audiences for the ideas in this paper. I thank these students for their input, especially Lu Xiaoyi (Stephanie), Yang Qilin (Jaden), Au Siu Hong (Walker), To Ka Chun (Adrian), and Chan Sze Hoi (Steve). Thanks also to two visitors to these classes, Jenny Nado and Herman Cappelen, both of whom have had an obvious influence on my thinking about conceptual engineering. Jenny and Herman also provided me with written comments on an earlier draft, as did Steffen Koch, who gave me an especially detailed and useful set of written comments. Discussions with Amit Chaturvedi, Jamin Asay, Lam Ka Ho, Manuel Gustavo Isaac, Sigurd Jorem, and Anton Alexandrov had a significant influence on the final draft—for the better, I hope they will think. Lastly, thanks to an anonymous reviewer for Philosophical Studies for several suggestions for improvements.

Author information



Corresponding author

Correspondence to Max Deutsch.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Deutsch, M. Speaker’s reference, stipulation, and a dilemma for conceptual engineers. Philos Stud 177, 3935–3957 (2020). https://doi.org/10.1007/s11098-020-01416-z

Download citation


  • Conceptual engineering
  • Speaker’s reference
  • Stipulation
  • Semantic change
  • Conceptual analysis