1 Introduction
Conceptual engineering concerns the assessment and improvement of our concepts.1 Its process typically involves four phases: (1) describing and (2) assessing our current concepts, and (3) designing and (4) implementing new concepts (Isaac, Koch, and Nefdt 2022, §2). While each phase has its own issues to address, the final phase, implementation, raises particularly difficult questions about its feasibility: What does it take to implement newly designed concepts? Is it feasible? Since Herman Cappelen’s pessimistic view that the factors relevant to implementation are both inscrutable and beyond our control (Cappelen 2018, 72–73), many theorists of conceptual engineering have addressed questions about the feasibility of implementation (Andow 2021; Deutsch 2020a; 2020b; Koch 2021c; 2021b; Machery 2021; Nimtz 2021; Pinder 2020; 2021; Schroeter and Schroeter 2020; Thomasson 2021).2
The difficulties of implementing newly engineered concepts have recently been formulated as a dilemma by Max Deutsch (2020a). Deutsch argues that the implementation of new concepts is either infeasible or trivial, depending on whether the goal is to change semantic meanings or speaker meanings. Several authors have already criticized Deutsch’s argument (Jorem 2021; Koch 2021c; Pinder 2020; Rudolph 2021), and Deutsch has also responded to some of these criticisms (Deutsch 2020b; 2021). In this debate, as we will see in more detail below, the goal of conceptual engineering tends to be set either too high or too low. As a result, implementation tends to be seen as either very unlikely to succeed or too easily achievable. The aim of this chapter is to offer a way out of this dilemma. I argue that the success conditions for implementation can be better understood if we distinguish between different stages in the implementation process. Implementation is a complex process involving several stages, each of which can be evaluated as a success or a failure. I argue that even if an implementation does not reach the final stage in which a new concept is widely used in society at large, it may not be a complete failure. Conceptual engineers may not even aim for a new concept to be widely used in society at large. And even if they do and a new concept only circulates in a smaller subgroup, this can still be a significant achievement. The upshot is that we should take more seriously the possibility that conceptual engineering can be implemented locally at the subgroup level.3
This chapter is structured as follows. In Section 2, I will examine Deutsch’s argument and some responses to it in terms of the success conditions for implementation, and suggest that we need a more detailed description of the implementation process. In Section 3, I will describe the implementation process as consisting of five stages and situate changing speaker-meaning and changing semantic meaning within it. In Section 4, I elaborate on the notions of dialects and local linguistic conventions, on which my account of local implementation at the subgroup level relies. In Section 5, I distinguish between two types of cases in which an implementation remains at the subgroup level and yet it is still a significant achievement. Finally, I conclude this chapter with some remarks about the implications that thinking about local implementation in a subgroup might have for thinking about implementation in society at large.
Before proceeding, three clarifications are in order. First, in this chapter I focus on linguistic approaches to conceptual engineering and assume that the target objects of conceptual engineering, what is to be engineered, include meanings that are somehow associated with words or their use, such as semantic meanings or speaker meanings (see Section 2 below for more on these). Note that this does not imply that meanings are the sole target of conceptual engineering and thus does not exclude other approaches such as those that include concepts in the psychological sense in the target objects (Machery 2017). Indeed, I see no good reason to think that meanings are the sole target of conceptual engineering, and I am inclined to think that linguistic and psychological approaches are not competing but rather complementary (Isaac, Koch, and Nefdt 2022, §3; Koch 2021a). However, it is beyond the scope of this chapter to explore the relationship between linguistic and other approaches.
Second, while there are two broad forms of conceptual engineering, conceptual re-engineering, which associates new concepts with existing terms, and de novo conceptual engineering, which introduces new terms and associates new concepts with them (Chalmers 2020), this chapter focuses on the former. This is because the problem of implementation is particularly important in the case of conceptual re-engineering (Deutsch 2020a; 2021; Koch 2021c). When I use the term ‘conceptual engineering’ in this chapter, I am referring to conceptual re-engineering.
Third, while it is often assumed that conceptual engineering will not be a worthwhile project unless the feasibility of implementation is shown (see, for example, Nimtz 2021, 2), I am not making this assumption.4 For even if the prospects for implementation are very poor, it can still be worthwhile to assess our current concepts and explore alternative concepts and their implications (Rudolph 2021). Of course, this is not to say that the problem of feasibility is unimportant. Investigating how new concepts might be implemented remains an important task for theorists of conceptual engineering, and indeed, this chapter is devoted to that task.
2 Deutsch’s Dilemma and the Success Conditions for Implementation
In this section, I reconstruct Deutsch’s dilemma and Koch’s (2021b) and Pinder’s (2020) responses to it in terms of the success conditions for implementation. I argue that these responses are inadequate and suggest that we need a more detailed description of the implementation process.
Deutsch formulates a dilemma using the distinction between semantic meaning and speaker meaning. Semantic meaning is what words mean in a given language, while speaker meaning is what a speaker means in uttering certain words (Grice 1989; Kripke 1977). According to Deutsch, the goal of conceptual engineering is to change either semantic meanings or speaker meanings. However, if the goal is to change semantic meanings, then implementation is infeasible; and if the goal is to change speaker meanings, then implementation is trivial, if not infeasible. Thus, implementation faces a serious dilemma: it is either infeasible or trivial. The first horn of the dilemma is called the implementation challenge, and the second horn is called the trivialization challenge. In the remainder of this section, we will consider each of these challenges and some of the responses to them in turn.
Let us begin with the implementation challenge and Koch’s response. The first thing to note is that Deutsch is not denying that it is possible for the semantic meaning of a term to change. Rather, what he is denying is that it is feasible to intentionally change the semantic meaning of a term. According to Deutsch, we do not know what it takes to intentionally bring about a change in semantic meaning. For a mere stipulation is not sufficient, and we are ignorant of what else is necessary (Deutsch 2020a, 3940–41, 3953–54; see also Cappelen 2018, 72–74). Thus, he concludes that those who claim that it is feasible to intentionally bring about a change in semantic meaning “owe us an account of how they can actually succeed in doing so” (Deutsch 2020a, 3941).
Koch responds to the implementation challenge by sketching a metasemantic theory of how a change in semantic meaning can be brought about intentionally (Koch 2021b, 2284–87; see also 2021c). He argues that while semantic meaning is not within our individual control nor within our immediate control, it is within our collective long-term control. If a term is used in a new way by sufficiently many speakers over long periods of time, then its semantic meanings will eventually change (Koch 2021b, 2285; 2021c, 339, 343). Thus, an implementation is successful if conceptual engineers succeed in getting enough people to use the target term in accordance with their prescription, however long that may take.
While Koch’s response contains several insights, it is still inadequate in two respects. First, even if it shows how it is possible to intentionally change semantic meanings, the challenge is to show how this is feasible, and many are dissatisfied with Koch’s response in this respect (Deutsch 2021, 3665–66; Jorem 2021, 194; Nimtz 2021, 3). Second, it is not clear whether, in Koch’s view, the fact that many people change their use of the target term is not only a sufficient but also a necessary condition for a successful implementation. Koch is only explicit about the sufficient condition: that if the target term is used in a new way by sufficiently many speakers, then its semantic meaning will eventually change. Deutsch takes it that Koch’s view also implies a claim about the necessary condition: that a successful implementation requires that many speakers, and indeed, “huge numbers of speakers” (Deutsch 2021, 3667), change their use. Whether or not Koch’s view actually implies this, I will argue in the following sections that an implementation does not always require that many speakers change their use. If this is true, then any view that implies that a successful implementation always requires this sets the conditions too high. But before arguing this, we need to consider the other horn of the dilemma.
The second horn of the dilemma, the trivialization challenge, was that if the goal of conceptual engineering is to change speaker meanings, then implementation is trivial. To see why, we first need to clarify what it means to change speaker meanings. Since Mark Pinder is the only one who argues for the view that the goal of conceptual engineering is to change speaker meanings, let us consider his view. According to Pinder, a speaker S is engaged in conceptual engineering of a term t, which semantically means m, if S makes it explicit that they will speaker-mean m* by t, where m* ≠ m, S speaker-means m* by t, and S presents m* as a good thing to speaker-mean by t (Pinder 2021, 155). And S’s conceptual engineering of t is successful if S is engaged in conceptual engineering of t and the hearer interprets S accordingly (Pinder 2021, 157; 2020, 6).
Let us now return to the trivialization challenge. As Jorem (2021, 203) notes, Deutsch (2020a) does not explain in what sense he claims that implementation is trivial in the speaker meaning approach. But he later explains it in “Trivializing Conceptual Engineering” (Deutsch 2020b). As he makes clear there, his point is that in the speaker meaning approach, conceptual engineering would be too easy to be used “as a worthwhile philosophical method” (Deutsch 2020b, 1). However, I think this explanation does not work. For the question of the feasibility of implementation can be separated from the question of the effectiveness of conceptual engineering as a philosophical method. Conceptual engineering takes place everywhere, not only within philosophy but also outside of it (Cantalamessa 2021; Cappelen 2018, 27–33; Thomasson 2021). Thus, we can ask how it can be implemented, apart from whether it can provide a worthwhile philosophical method.
Is there anything else about the speaker meaning approach that might trivialize conceptual engineering? I think it is that the speaker-meaning approach sets the success conditions for implementation too low. As we have seen, in the speaker meaning approach, for a conceptual engineering of a term t by a speaker S to be successful, nothing more is required than that S is engaged in it and that a hearer H interprets it accordingly. On the part of H, it is not required that H accept m* as a good thing to mean by t, nor that H come to mean m* by t. On the part of S, it is also not required that S continue to mean m* by t, nor that S be expected to do so. In the speaker meaning approach, an implementation can be successful even if none of these conditions are satisfied.
Pinder would respond by appealing to the distinction between narrow-scope and wide-scope projects of the speaker meaning approach. Whether a project is wide or narrow in scope depends on how many people it involves. A narrow-scope project involves only a limited number of people and is easy to carry out successfully, while a wide-scope project involves a large number of people’s use and can be difficult. Thus, he argues, the speaker meaning approach does not always set the success conditions too low.
However, simply broadening the scope is not enough. Even in a wide-scope project, it is not required that the speaker or the hearers continue to mean, or be expected to continue to mean, m* by t. As a result, an implementation can be a success even if it has no lasting effect at all. I take this as suggesting that even a wide-scope project sets the success conditions too low.
If the above argument that both Koch’s and Pinder’s responses are inadequate is correct, do we have to accept Deutsch’s conclusion that the implementation of new concepts faces a serious dilemma? I do not think so. We can think of both changing speaker meanings and changing semantic meanings as part of a larger process of implementation, and there is no need to choose one or the other.5 What is needed, I think, is a more detailed description of the whole process of implementation. In the following sections, I will offer such a description and use it to examine the success conditions for implementation.
3 The Process of Implementation
In the previous section, I examined Deutsch’s dilemma and Koch’s and Pinder’s responses to it in terms of the success conditions for implementation, and suggested that we need a more detailed description of the implementation process. In this section, I offer such a description.
Implementation of new concepts is a complex process. In describing the implementation process, I think it is helpful to compare it to the process of slur reclamation, which has been the subject of several studies in recent years. Slur reclamation is a phenomenon in which a slur loses its negative valence as a result of the efforts of the members of the targeted group and their allies (Zyglewicz ms). As Lynne Tirrell notes, slur reclamation is “not achieved by one speaker by fiat in an instant; it is a community-wide achievement that takes time to occur” (Tirrell 1999, 61), and as Robin Jeshion notes, its process has a “diachronic structure” (Jeshion 2020, 108). In the literature on slur reclamation, it is common to distinguish between four to six stages in the process of a successful reclamation (Jeshion 2020, 108; Jusińska 2021, 701; Saka 2007, 109:146; Zyglewicz ms, 2).
We can think of the process of implementation of new concepts as having a similar structure.6 I propose to distinguish between five stages of a successful implementation:
Initial state: the target term t is governed in a whole linguistic community C by the conventions associated with the language L, in which t means m.
Initial acts of implementation, speaker-meaning m*: someone in C deviates from the conventions associated with the language L and starts to use or proposes to use t to speaker-mean m*, where m* ≠ m.
Secondary acts of implementation, imitation: other members of C imitate using t to speaker-mean m*.
Local conventionalization in C* and the emergence of a dialect L*: the new usage of t becomes locally conventional in a subgroup or subcommunity C* within C, and the new dialect L* emerges in which t means m*.
Conventionalization of L* in C: more and more people in C come to use L* so that the use of L* becomes conventional in C. Now t semantically means m*.
The process could be described more finely by drawing more distinctions in the stages, and each stage could be described in more detail. For example, one might distinguish between L* becoming conventional in C, on the one hand, and the semantic meaning of t changing from m to m*, on the other, as different stages. Or, even after L* has become conventional in C, L* may not have completely replaced L, and L* and L may be used for different purposes in different contexts.7 But for my purposes, the above description is sufficient. We can now ask the question: Is an implementation a failure if it does not reach Stage 5? In Section 5, I will answer this question in the negative, by presenting two types of cases in which an implementation has not reached Stage 5 and yet is not, at least, a complete failure. Before doing so, however, I will elaborate in the next section on the notions of local conventions and dialects, which appear in the description of Stage 4.
4 Dialects and Local Linguistic Conventions
In this section, I elaborate on the notions of local conventions and dialects, which appear in Stage 4 of the implementation process described in the previous section. Although local linguistic conventions and dialects have received little theoretical attention in the philosophy of language, their existence has been recognized in some of its classic articles. For example, David Lewis, in his “Languages and Language,” notes that when a population P is “linguistically inhomogeneous,” as when “P divides into two classes,” it is sometimes necessary to recognize that more than one language is used by P (Lewis 1975, 32).8 And Tyler Burge, in his “Individualism and the Mental,” makes some remarks about regional linguistic conventions and regional dialects in which terms have different meanings than in the language of the whole linguistic community (Burge 1979, 90, 114). More recently, as non-ideal theories of language have been developed,9 attention to linguistic subcommunities or subgroups has played a more important role in theorizing such linguistic phenomena as slur reclamation (Jeshion 2020), semantic contestation (Mühlebach 2021), linguistic hijacking (Anderson 2020), and hermeneutical injustice (Goetze 2018; Medina 2012; 2013; 2017). Of these, Burge’s remarks on regional linguistic conventions and regional dialects, though brief, are particularly important for our purposes. In what follows, I will consider and develop Burge’s remarks.
Toward the end of his “Individualism and the Mental,” Burge makes some brief but important remarks about regional linguistic conventions and regional dialects (Burge 1979, 114). One of his main claims in this article is, as is well known, that even when a person deviates from the linguistic conventions of a whole linguistic community C, their words usually mean what they mean in the standard language L of C. However, it is less well known that Burge makes an important qualification to this claim. He admits the possibility that a person who deviates from the linguistic conventions of C may be “self-consciously opting out” of these conventions and instead following the regional linguistic conventions of a regional linguistic community C*.10 In such cases, Burge suggests, the person “speaks a regional dialect” (Burge 1979, 114), L*, in which some terms have different meanings than in L.
From Burge’s remarks, we can extract the following picture of dialects and local linguistic conventions:
The Burgean picture of dialects and local linguistic conventions: In addition to the language L, which is conventional in the whole linguistic community C, there is a dialect L*, which is locally conventional in a subgroup or subcommunity C* within C, where t means m in L, but m* in L*.
In the remainder of this section, I will elaborate on this picture by clarifying the notions of dialects and local linguistic conventions. Five clarifications are in order.
First, when Burge talks about dialects in his “Individualism and the Mental,” he uses the term “regional dialects” (Burge 1979, 114), but it is important to note that dialects need not be associated with particular geographic regions. In sociolinguistics, it is common to distinguish between regional dialects and social dialects. While regional dialects are varieties of language associated with geographic regions, including American English, British English, etc., social dialects are varieties of language associated with social groups, including English spoken by the upper class, English spoken by philosophers, etc. (Holmes and Wilson 2022, ch. 6; Wardhaugh and Fuller 2021, ch. 2).11
Second, dialects are sometimes defined in sociolinguistics as variations of a language that differ in pronunciation, grammar, or vocabulary (Holmes and Wilson 2022, 672), but what Burge calls dialects are variations of a language that differ in meaning, that is, those in which at least some terms have different meanings. Whether or not it is sociolinguistically appropriate to call such variations dialects, it should not be inappropriate to think of them at least as analogous to dialects. In what follows, I will use the term ‘dialects’ to refer to variations of a language that are associated with social groups and that differ in meaning.12
Third, since it is controversial where to draw the boundary between semantics and pragmatics (Szabó 2005), I want to remain neutral on whether the meaning of a term in a dialect belongs to semantics or pragmatics. I want to note, however, that the idea of local semantic meaning is at least not inconsistent. For example, Kripke calls the meaning of a term in one’s idiolect the semantic meaning of that term in that idiolect (Kripke 1977, 263–64). If we follow Kripke in this respect, then the meaning of a term in a dialect might also well be included in semantic meaning. Pinder also seems to suggest that semantic meaning can be local. He suggests that semantic meaning is relative to a linguistic community, and the relevant linguistic community can be a local linguistic subcommunity (Pinder 2021, 152).
Fourth, no special explanation is needed for the notion of local conventions because several accounts of conventions, such as those in Lewis (1969; 1975), Gilbert (1989), and Millikan (1998), are directly applicable to local ones. Of course, there are several important differences among these accounts,13 but both my description of the implementation process given in the previous section and my account of local implementation to be given in the next section are neutral with respect to these differences.
Fifth, and finally, members of a subgroup C* within C need not always speak L*, the dialect that is conventional in C*. They can switch, for example, between L* and L, the language that is conventional in the whole linguistic community C, depending on the context, for example, who they are speaking to. A member of C* may speak L* when speaking to other members of C* and speak L when speaking to outsiders. This is analogous to the fact that one can be multilingual, i.e., speak more than one language.
More could be said about dialects and local linguistic conventions, but this is enough for our purposes.14 In the next section, we will consider the success conditions for implementation.
5 The Success Conditions for Implementation
In Section 3, I described the implementation process and used it to raise the question: Is an implementation a failure if it does not reach Stage 5, that is, if it remains at the subgroup level? I now answer this question in the negative. I argue that even if an implementation does not reach Stage 5 and thus remains local, it may not be a complete failure and may even be a success. In what follows I will present two types of cases.15
First, there are cases in which an implementation has not reached Stage 5 but the conceptual engineers may not have aimed for Stage 5 in the first place.16 Consider, for example, the botanical language, the dialect used in the botanist community. In the botanical language, the meaning of the term ‘fruit’ is not the same as in ordinary English. The extension of ‘fruit’ in the botanical language includes those objects that it does not in ordinary English, such as tomatoes, cucumbers, and pumpkins. Normally, most English speakers do not speak the botanical language. They would classify tomatoes, cucumbers, or pumpkins as vegetables, and even some “experts” on fruits such as culinary experts would do so (Martí and Ramírez-Ludeña 2021; Phillips 2014). However, even if non-botanists classify these things as vegetables, this does not make the implementation of the new concept of fruit by botanists a failure. For if the use of the botanical language becomes conventional in the botanist community, that would be sufficient for their purpose of scientifically classifying plants. And it is not clear that botanists have sought to change the meaning of ‘fruit’ in ordinary English at all.
It is worth noting here that the goal of a Carnapian explication, according to some commentators, is not to produce a new concept that is meant to replace a concept “in ordinary parlance, among laypeople” (Dutilh Novaes 2020, 19), but to produce a new concept “that will be in first instance adopted in the relevant professional dialects, i.e. the vocabulary of theorists and scientists” (Dutilh Novaes 2020, 19; see also Pearson 2017, 764). A new concept may eventually become more widely adopted, but, according to Dutilh Novaes, this is presumably not the initial goal of an explication (Dutilh Novaes 2020, 19), and perhaps in many cases, it is not even the ultimate goal. According to Dutilh Novaes, theorists and scientists are “the initial ‘consumers’ of the result of an explication” (Dutilh Novaes 2020, 20), and perhaps in many cases, they are also the main consumers.
Let us now turn to cases of the second type, in which the conceptual engineers aimed for Stage 5, but the implementation did not reach it, and yet the implementation was not a complete failure. Suppose that, in the community of English speakers, C, the term ‘disability’ initially referred to individual medical impairments (Stage 1).17 At this stage, members of C tended to assume that the primary cause of the restrictions of activities faced by people with disabilities was their own bodily conditions. Suppose, however, that someone with disabilities in C has proposed to use ‘disability’ to refer to a product of the interaction between individual medical impairments and the social environment (Stage 2), and that other members of C, especially people with disabilities and their allies, have started to imitate the new usage (Stage 3). Suppose also that the new usage has become conventionalized in a subgroup C* within C, and a new dialect L* emerges in which ‘disability’ refers to conditions that involve both individual and social factors (Stage 4). Now speakers of L* can better understand that the disadvantages experienced by people with disabilities are partly due to the way society is organized, and that accommodations ought to be provided as a matter of justice. Moreover, speakers of L* can also better communicate these experiences to each other. L* thus helps its speakers to promote solidarity. Now if speakers of L* further aim for Stage 5 and if they do not reach it, then the implementation would not be a complete success. But it would not be a complete failure. Rather, reaching Stage 4 would already be a significant achievement for speakers of L*. As I have just described, L* enables its speakers to do many things that they could not do without it.
Here it is worth noting a parallel between local implementation of the second type and the development of non-dominant hermeneutic resources in a marginalized group discussed by José Medina (2012; 2013, ch. 3; 2017). Medina’s discussion is intended to overcome the shortcomings of Miranda Fricker’s analysis of hermeneutical injustice. According to Fricker, a hermeneutical injustice occurs if (1) members of a group are prevented from participating in the practices by which collective hermeneutical resources are generated, (2) as a result of this unequal participation, there is a gap in collective hermeneutical resources, and (3) due to this gap in shared hermeneutical resources, members of that marginalized group are unable to properly understand and communicate their social experiences (Fricker 2007, ch. 7). Although Medina broadly agrees with Fricker’s analysis, he criticizes it for assuming a “monolithic” (Medina 2017, 43) conception of collective hermeneutical resources and ignoring their internal diversity and heterogeneity. According to Medina, Fricker’s analysis only pays attention to the dominant hermeneutical resources, but even if there is a gap in them, members of marginalized groups may have developed their own resources within their groups so that they can properly understand and communicate their own social experiences. Members of marginalized groups may not have tried at all to promote their own hermeneutical resources beyond their community. Or they may have tried but not succeeded. But in either case, they would still be much better off than if they had not developed their own resources. In thinking about the implementation of new concepts, we should also abandon the monolithic conception of language and linguistic conventions and pay more attention to their internal diversity and heterogeneity.
To summarize the discussion in this section, there are at least two types of cases in which an implementation has not reached Stage 5, and thus remains local, and yet is not a complete failure. The goal of a given conceptual engineering project may not be to implement a new concept in society at large; and even if it is, and even if the new concept has only circulated in a smaller subgroup and the implementation remains local, this can still be a significant achievement for members of that group.
6 Conclusion
In this chapter, I first examined Deutsch’s dilemma for implementing conceptual engineering and Koch’s and Pinder’s responses and suggested that we need a more detailed description of the implementation process. I then offered a description that distinguishes five stages in the process and used it to consider the success conditions for implementation. I argued that even if an implementation does not reach the final stage in which a new concept is widely used in society at large, it may not be a complete failure. In concluding this chapter, I want to note some of the implications and limitations of the above discussion.
If the above discussion is correct, then the implementation of new concepts is not always extremely difficult. The goal may not be to implement new concepts in society at large in the first place. And even if it is, local implementation in a smaller subgroup can still be a significant achievement for members of that group. Thus, before we become pessimistic about the feasibility of implementation, we need to consider, on a case-by-case basis, what is the goal of a given conceptual engineering project. We also need to think about the success conditions for implementation in a more fine-grained way. Implementation is not an all-or-nothing affair. Rather, it is a process that involves several stages, each of which can be evaluated as a success or a failure.
It is important to note that the above discussion does not imply that implementation is always easy. The point was that we should take more seriously the possibility of local implementation at the subgroup level, and nothing was said about the feasibility of implementation in society at large.
However, the above discussion may still offer some suggestions about how new concepts might be implemented in society at large. What is worth considering is the possibility that implementation may not always start at Stage 1. Even if an attempt to implement a new concept did not reach Stage 5 but only Stage 4, the next time the same attempt is made, it would not have to start over at Stage 1 but could start at the stage reached by the previous attempt, i.e., Stage 4. Then, it could be much easier to reach Stage 5 than if they had started at Stage 1. Implementation in society at large may be achieved as a result of such repeated efforts. My description of the implementation process helps us see such a possibility as well.
References
Anderson, D. 2020. “Linguistic Hijacking.” Feminist Philosophy Quarterly 6 (3).
Andow, J. 2021. “Conceptual Engineering Is Extremely Unlikely to Work. So What?” Inquiry 64 (1–2): 212–226.
Beaver, D., and J. Stanley. 2019. “Toward a Non-Ideal Philosophy of Language.” Graduate Faculty Philosophy Journal 39 (2): 501–545.
Brigandt, I., and E. Rosario. 2020. “Strategic Conceptual Engineering for Epistemic and Social Aims.” In Conceptual Engineering and Conceptual Ethics, edited by A. Burgess, H. Cappelen, and D. Plunkett, 100–124. Oxford University Press.
Burge, T. 1979. “Individualism and the Mental.” Midwest Studies in Philosophy 4: 73–121.
Burge, T. 2007. Foundations of Mind: Philosophical Essays, Volume 2. Oxford University Press.
Cantalamessa, E. 2021. “Disability Studies, Conceptual Engineering, and Conceptual Activism.” Inquiry 64 (1–2): 46–75.
Cappelen, H. 2018. Fixing Language: An Essay on Conceptual Engineering. Oxford University Press.
Chalmers, D. 2020. “What Is Conceptual Engineering and What Should It Be?” Inquiry, 1–18. https://doi.org/10.1080/0020174X.2020.1817141.
Deutsch, M. 2020a. “Speaker’s Reference, Stipulation, and a Dilemma for Conceptual Engineers.” Philosophical Studies 177 (12): 3935–3957.
Deutsch, M. 2020b. “Trivializing Conceptual Engineering.” Inquiry, 1–16. https://doi.org/10.1080/0020174X.2020.1853343.
Deutsch, M. 2021. “Still the Same Dilemma for Conceptual Engineers: Reply to Koch.” Philosophical Studies 178 (11): 3659–3670.
Dobler, T. 2022. “Pluralist Conceptual Engineering.” Inquiry, 1–27. https://doi.org/10.1080/0020174X.2022.2086171.
Dutilh Novaes, C. 2020. “Carnap Meets Foucault: Conceptual Engineering and Genealogical Investigations.” Inquiry, 1–27. https://doi.org/10.1080/0020174X.2020.1860122.
Fricker, M. 2007. Epistemic Injustice: Power and the Ethics of Knowing. Oxford University Press.
Gilbert, M. 1989. On Social Facts. Princeton University Press.
Goetze, T. 2018. “Hermeneutical Dissent and the Species of Hermeneutical Injustice.” Hypatia 33 (1): 73–90.
Grice, P. 1989. Studies in the Way of Words. Harvard University Press.
Holmes, J., and N. Wilson. 2022. An Introduction to Sociolinguistics. 6th ed. Routledge.
Isaac, M. G., S. Koch, and R. Nefdt. 2022. “Conceptual Engineering: A Road Map to Practice.” Philosophy Compass 17 (10).
Jeshion, R. 2020. “Pride and Prejudiced: On the Reclamation of Slurs.” Grazer Philosophische Studien 97 (1): 106–137.
Jorem, S. 2021. “Conceptual Engineering and the Implementation Problem.” Inquiry 64 (1–2): 186–211.
Jusińska, Z. 2021. “Slur Reclamation – Polysemy, Echo, or Both?” Organon F 28 (3): 689–707.
Keiser, J. 2023. Non-Ideal Foundations of Language. Routledge.
Koch, S. 2021a. “Engineering What? On Concepts in Conceptual Engineering.” Synthese 199 (1): 1955–1975.
Koch, S. 2021b. “The Externalist Challenge to Conceptual Engineering.” Synthese 198 (1): 327–348.
Koch, S. 2021c. “There Is No Dilemma for Conceptual Engineering. Reply to Max Deutsch.” Philosophical Studies 178 (7): 2279–2291.
Koch, S. 2024. “The Anti-Conceptual Engineering Argument and the Problem of Implementation.” American Philosophical Quarterly 61 (1): 73–85.
Koch, S., G. Löhr, and M. Pinder. 2023. “Recent Work in the Theory of Conceptual Engineering.” Analysis 83 (3): 589–603.
Kripke, S. 1977. “Speaker’s Reference and Semantic Reference.” Midwest Studies in Philosophy 2 (1): 255–276.
Lewis, D. 1969. Convention: A Philosophical Study. Harvard University Press.
Lewis, D. 1975. “Languages and Language.” Minnesota Studies in the Philosophy of Science 7: 3–35
Ludlow, P. 2014. Living Words: Meaning Underdetermination and the Dynamic Lexicon. Oxford University Press.
Machery, E. 2017. Philosophy within Its Proper Bounds. Oxford University Press.
Machery, E. 2021. “A New Challenge to Conceptual Engineering.” Inquiry, 1–24. https://doi.org/10.1080/0020174X.2021.1967190
Martí, G., and L. Ramírez-Ludeña. 2021. “Tolerance, Flexibility and the Application of Kind Terms.” Synthese 198 (12): 2973–2986.
Medina, J. 2012. “Hermeneutical Injustice and Polyphonic Contextualism: Social Silences and Shared Hermeneutical Responsibilities.” Social Epistemology 26 (2): 201–220.
Medina, J. 2013. The Epistemology of Resistance: Gender and Racial Oppression, Epistemic Injustice, and the Social Imagination. Oxford University Press.
Medina, J. 2017. “Varieties of Hermeneutical Injustice.” In The Routledge Handbook of Epistemic Injustice, edited by J. Kidd, J. Medina, and G. Pohlhaus Jr., 41–52. Routledge.
Millikan, Ruth. 1998. “Language Conventions Made Simple.” The Journal of Philosophy 95 (4): 161–180.
Mühlebach, D. 2021. “Semantic Contestations and the Meaning of Politically Significant Terms.” Inquiry 64 (8): 788–817.
Mühlebach, D. 2022. “Non-Ideal Philosophy of Language.” Inquiry, 1–23. https://doi.org/10.1080/0020174X.2022.2074884.
Nado, J. 2021. “Re-Engineering Knowledge: A Case Study in Pluralist Conceptual Engineering.” Inquiry, 1–26. https://doi.org/10.1080/0020174X.2021.1903987.
Nimtz, Ch. 2021. “Engineering Concepts by Engineering Social Norms: Solving the Implementation Challenge.” Inquiry, 1–28. https://doi.org/10.1080/0020174X.2021.1956368.
Pearson, J. 2017. “Carnap, Explication, and Social History.” Social Theory and Practice, 741–774.
Phillips, I. 2014. “Cetacean Semantics: A Reply to Sainsbury.” Analysis 74 (3): 379–382.
Pinder, M. 2020. “Conceptual Engineering, Speaker-Meaning and Philosophy.” Inquiry, 1–15. https://doi.org/10.1080/0020174X.2020.1853342.
Pinder, M. 2021. “Conceptual Engineering, Metasemantic Externalism and Speaker-Meaning.” Mind 130 (517): 141–163.
Pinder, M. 2022. “Is Haslanger’s Ameliorative Project a Successful Conceptual Engineering Project?” Synthese 200 (334): 1–22. https://doi.org/10.1007/s11229-022-03803-x.
Rudolph, R. 2021. “Conceptual Exploration.” Inquiry, 1–26. https://doi.org/10.1080/0020174X.2021.2002053.
Saka, P. 2007. How to Think about Meaning. Springer.
Schroeter, L., and F. Schroeter. 2020. “Inscrutability and Its Discontents.” Canadian Journal of Philosophy 50 (5): 566–579.
Szabó, Z. G. 2005. Semantics versus Pragmatics. Clarendon Press.
Thomasson, A. 2021. “Conceptual Engineering: When Do We Need It? How Can We Do It?” Inquiry, 1–26. https://doi.org/10.1080/0020174X.2021.2000118.
Tirrell, L. 1999. “Derogatory Terms.” Language and Liberation: Feminism, Philosophy and Language, 41–79.
Wardhaugh, R., and J. M. Fuller. 2021. An Introduction to Sociolinguistics. John Wiley & Sons.
Zyglewicz, T. ms. “Slur Reclamation and the Polysemy/Homonymy Distinction.”
Versions of this chapter were presented at the Eighth International Conference on Philosophy of Language and Linguistics (PhiLang 2023) at the University of Lodz, at the ELSI Philosophy Forum at Kyoto University, and at the Eighth Conference of Online Political Theory Forum. I would like to thank the audiences at these events for their helpful feedback. I would also like to thank Shimpei Endo, Kotaro Hara, Céline Henne, Koshi Hamamoto, Guido Löhr, Arata Matsuda, Hironori Mori, Mark Pinder, Yuki Suda, Amie L. Thomasson, and Tomasz Zyglewicz for valuable comments or discussions. This work was supported by JSPS KAKENHI Grant Numbers JP21J00061 and 23K25257. It was also supported by crowdfunding through the platform Academist. I would like to thank Yuji Tatemata, Shin Ichihara, Takuya Hitomi, Shunji Furukuma, and Shinya Abe for their generous support.
See Koch, Löhr, and Pinder (2023) for a useful overview of Cappelen’s view and some responses to it.
Isaac, Koch, and Nefdt (2022, 6) also note that implementation can be done at the subgroup level, but they do not develop this point further. I will distinguish between two types of local implementation at the subgroup in Section 5.
The term ‘local conceptual engineering’ is sometimes used differently. For example, Cappelen uses it to refer to conceptual engineering done for a particular purpose in a particular context (Cappelen 2018, 36).
Koch (2024) examines and rejects this assumption.
Rudolph (2021, 19) makes the same point. Koch (2021b, 2283) also notes that Deutsch’s dilemma rests on a false dichotomy but his reason is different from mine.
See Tomasz (ms) for a more detailed discussion on the relationship between conceptual engineering and slur reclamation. I thank Tomasz Zyglewicz for allowing me to read this unpublished manuscript, from which I learned a great deal.
See Brigandt and Rosario (2020); Dobler (2022); Nado (2021) for such pluralist conceptual engineering.
However, no sooner does Lewis point out this complicating factor than he abstracts it away. See Keiser (2023, ch. 2) for Lewis’s idealizing assumptions about language.
See Beaver and Stanley (2019); Keiser (2023); Mühlebach (2022) for the recent development of non-ideal theories of language.
Burge makes a similar point in the introduction to his collection of essays, Foundations of the Mind (Burge 2007, 14), which includes “Individualism and the Mental.”
In fact, Burge also mentions “dialects […] shared by social groups” in his “Wherein is Language Social?” (Burge 2007, 275).
It is important to emphasize a difference between what I call dialects and what Peter Ludlow (2014) calls microlanguages: dialects are governed by linguistic conventions, no matter how local they are, whereas microlanguages are “one-off fleeting things” that “we build on a conversation-by-conversation basis” (Ludlow 2014, 3). Thanks to Céline Henne for raising this issue.
Three differences are worth noting here. First, whereas in Lewis’s account, it is necessary for a convention to exist among a population P that (almost) everyone in P conforms to it, and that it is common knowledge in P that (almost) everyone conforms to it, neither general conformity nor common knowledge is required in Gilbert’s and Millikan’s accounts. Second, whereas normativity is essential to conventions in Gilbert’s account, it is not essential in Lewis’s and Millikan’s accounts. Third, whereas in Lewis’s and Millikan’s accounts, an explicit agreement is not required for conventions to emerge or change, in Gilbert’s account, something analogous to an agreement, which she calls “the joint acceptance of a principle,” is required. As for the third, it is worth noting that Deutsch assumes that “intentional changes in semantic meaning require huge numbers of speakers to first agree that a term’s semantic meaning should be changed” (Deutsch 2021, 3667; emphasis added). I am inclined to think that an explicit agreement is not required, but I will not argue for it here.
One important question I cannot address here is how to demarcate the boundaries of linguistic communities. See Anderson (2020, 8–15).
In his recent paper, Pinder (2022, 16) also notes that “different conceptual engineering processes have different success conditions.” Thanks to Arata Matsuda and Mark Pinder for bringing this point to my attention.
Isaac, Koch, and Nefdt (2022, 6) note this type of case, but they do not note the second type, which I will present below.
Cantalamessa (2021) argues that those working in disability studies and disability rights activism are engineering the concept of disability.