1 Introduction

There are no more important questions for philosophers of language than: what do words mean and how is that meaning fixed? The suggestion that the meaning of a word can be decided upon by a philosopher, shaped to best fit a particular conception of the world which that individual would like to promote, is, then, certain to represent a temptation. Not only does it imply that a meaning can be established by stipulation, it also allows philosophers to create better meanings, by which we mean meanings which enable them to accomplish their philosophical goals. Indeed, I shall maintain that this is a temptation which a great many philosophers throughout history have been unable to resist, and that the ‘engineering’ of concepts is a practice both common and venerable. I shall also suggest that it is, in a sense, dangerous and likely to cause more confusion and misunderstanding than it eradicates.

In Herman Capellen’s relatively recent book, ‘Fixing Language’, which has popularised the term and released a wave of publications on the issue, Conceptual Engineering is considered the best name for the ‘enterprise of assessing and improving our representational devices’ (2018, p. 3), despite the fact that in his view ‘the project isn’t about concepts and there isn’t really any engineering’ (2018, p. 4), which seems to have us striding rapidly in the direction of fallacy and confusion from the start. The term itself left aside, however, the description ‘assessing and improving our representational devices’ is helpful in understanding what is going on and what might be going wrong within Conceptual Engineering.

The first stage—assessing—seems harmless enough. We might want to discuss how we should go about making such assessments, but there is no apparent danger in the practice, whatever our methodology. It is the second phase, the process of improvement, with which we are concerned here and within which there is an obvious potential for disagreement and breakdown of communication. If our use of representational devices is to be improved, there must be some sense of better and worse meanings for our linguistic conceptualisations. That can only mean meanings which are better or worse for some end or object—the idea of an inherent goodness of meaning can, I think, be quickly dismissed—and since only individual people can have ends, that means meanings which are better for some to achieve what they wish to achieve (see Löhr 2024). It may be that there are cases where a change in meaning would be better for all concerned, but there is no reason to think that that will be the usual situation, or that it will be much of a consideration for budding engineers.

This work will not go further with the question of good and worse meanings, or, indeed, with the motivations of those who would advocate for the one over the other. Rather, it will suggest that many actions of Conceptual Engineering are too trivial to deserve such a grand title and then, for the rest, concentrate on the danger referred to above—that of leading users of engineered concepts into potentially fallacious argumentation.

In order to reach conclusions on these issues, there are two important distinctions which should be made at the outset. Firstly, there is a clear difference between suggesting that new concepts would be useful, then coming up with words for them, and the riskier step of advocating for existing words to be understood differently and taken to signify new, altered concepts. The creation of new words can, I think, be considered completely uncontroversial and unremarkable. The only test for coinages is whether or not they achieve sufficient popularity to enter the vocabulary, either of the population at large, or of the specialised community the word has been created to serve. In many cases, the rough reasoning that if such a word for such a concept had been needed it would have been invented long ago, will apply and the newly minted lexical item will never take hold. In others, particularly where new technologies are involved, the usefulness of the word will be apparent at once (to download, or to google, for instance) and society will soon forget that it ever got by without the term. While this practice may lead to some confusion on the part of those who do not know the new term, it is unlikely to lead to the kind of fallacies discussed below. The focus of this work, therefore, is on the adaptation of already existing terms, not the creation of new ones.

The second distinction, which is not much referred to in the literature, is that between the adjustment of concept/wordFootnote 1 pairings to assist speakers in better expressing what they really mean, that is, by bringing greater clarity to the usage of a word or separating out unhelpful fringe meanings, and the deliberate manipulation of meanings to fulfil the political or social goals of the engineer. Those of us who insist on the separation of the meanings of ‘uninterested’ and ‘disinterested’, for example, are often reminded that in the past the distinction did not exist, but that is not the point, we argue: it should exist because it is useful in achieving clarity of expression. There will, of course, be many borderline cases, perhaps a wide spectrum of increasingly less linguistic or philosophical, and more political instances: the example of John Stuart Mill’s Utilitarianism, discussed in Sect. 4, is a case in point, as he explicitly discusses the application of his philosophy, and thus his definitions, in a political system. The drawing of this distinction is not intended to imply that the latter are wrong or unacceptable, only that the two are not quite the same thing.

In order to better describe what he takes Conceptual Engineering to be, Capellen provides a large number of examples from philosophy and other fields of thinkers who have engaged more or less explicitly in the criticism of linguistic concepts and the task of improving language (2018, ch. 2). Some of these can be understood as attempts to address what are perceived as inadequacies and imperfections in the expressive ability of natural language, while others have other motivations and clear political agendas, but there is no explicit recognition in his discussion that they are doing very different things.

One of those works characterised as philosophy, but certainly at the activist end of the field, is the well-known paper on gender and race by Haslanger (2000), which is frequently held up as paradigm example of Conceptual Engineering, understood as semantic amelioration for the purposes of social advance (e.g. Koch 2021; Isaac 2023). This work is referred to in the sections below. One point of particular interest about it, however, is that while Haslanger is clearly in the business of changing the meanings of established—and commonly employed—words, she casually suggests that she would be happy to make up new ones, if people object. That this would completely alter the nature of her project, making it both uncontroversial and, I would suggest, invisible, is something she does not discuss.

With these distinctions drawn, I go on to discuss the triviality, for their theoretical rather than practical import, of many cases of Conceptual Engineering, in Sect. 2, below. I then enumerate a number of fallacies of language to which engineered concepts might make an arguer vulnerable in Sect. 3, and describe at length the ‘philosophical fallacy’ of Nelson (2016) in Sect. 4. I conclude, in Sect. 5, that the apparent ‘amelioration’ of meaning for social or political purposes may be little more than a rhetorical device with great potential to trip up the unwary.

2 The Charge of Triviality

Max Deutsch defines the goal of Conceptual Engineering as being ‘to replace semantic meanings and referents with new and better ones’ (Deutsch 2021, p. 3660). This can be referred to as ‘reengineering’ to distinguish it from ‘conceptual construction’ which is concerned with introducing the new, not replacing the old (see Koch 2021). Deutsch suggests that this practice is ultimately not particularly helpful or deserving of the attention being given it by the philosophical community. This is because he believes acts of Conceptual Engineering fall onto the horns of a dilemma: either we try to adjust speaker meaning, which is trivial, or we try to change semantic meaning, which is impractical. Koch refers to these as the ‘implementation challenge’ and the ‘trivialization challenge’ (2021, p. 2281). It is interesting to note that although both of these are essentially empirical rather than theoretical claims, Deutsch does not offer any evidence for them.

The stipulation of meanings within specialised contexts, is, obviously, nothing new; and philosophers have long indulged in the practice. Koch suggests that what makes discussion of Conceptual Engineering relevant now is ‘not that it is being done, but that it is being recognized and theorized’ (2021, p. 2284). Perhaps it is true that the practice has not been theorized in quite the same way previously, but there has been no shortage of discussion on the manipulation of meaning,Footnote 2 and there is a good deal of comment by philosophers on the use and abuse which they feel their colleagues have made of particular terms in changing or stipulating their meanings. Section 4 describes a theory, far from recent, of philosophical fallacy based upon the idea that philosophers have been guilty of unjustified stipulations and definitions. One of those criticised for doing so, Jeremy Bentham, himself noted that philosophers have shied away from calling pleasure by its name for fear of association with the gross: ‘to cleanse itself from the sordes of its impure original it was necessary it should change its name: the honourable, the glorious, the reputable, the becoming, the honestum, the decorum, it was to be called: in short, anything but pleasure’ (Bentham 1962, p. 42). Was Bentham not describing, and decrying, the efforts of philosophers of ethics to reengineer the concept of the good away from one based in pleasure and pain and towards something more refined and intellectual? From the utilitarian perspective, all these efforts at amelioration led only to an obfuscation and ultimately to error: a fair assessment or not, they certainly did not go unrecognised.

I am inclined to agree with Deutsch on both counts as far as trivialisation is concerned: if we are concerned with only our own speaker meanings, then we are free to alter them as we wish, and if a group of us agree to the new or expanded meaning, we may use the concept in communication without problems; if we are interested in the philosophical practice of defining terms, then it is hard to see what there is to be excited about, as this is, and probably always has been, a perfectly normal and everyday part of philosophical theorising. In both instances, engineering and reengineering of concepts and definitions clearly does happen and is nothing out of the ordinary. What would be more noteworthy is the case of changes to broadly acknowledged semantic meaning brought about at the stipulation of an individual to promote a particular worldview, as Deutsch accepts. He does not, however, think this is a feasible project.

There are two reasons to question Deutsch’s view. Firstly, it is not clear just how much change needs to be brought about before an engineering project might be considered a success. Koch hints at this point when he writes: ‘It’s not that we can either aim to change semantic meanings, or else settle for individual-level speaker-meaning. There is room for a plausible, implementable and non-trivial project that lies between the two, namely, to convince other members of one’s group to use an existing term in a new way’ (2021, p. 2283). He supports this view with a somewhat convoluted discussion of metasemantics, of which Deutsch, understandably, is ‘deeply skeptical’ (2021, p. 3665). A clearer argument comes from Takaaki Matsui (2024), who discusses ‘local’ conceptual engineering and breaks up the implementation process into several distinct stages. In this way, engineers may be successful in the accomplishment of certain goals along the way to general societal use of an altered meaning even if they do not reach the end point of that process. Matsui reasons that these partial efforts may still have value, which can be recognised when the stages of the process are better understood.

Perhaps more importantly, there are some fairly obvious examples where semantic meaning change has been achieved in modern discourse. Capellen notes that the debates around the meanings of ‘marriage’ and ‘rape’ are not debates as to what those words do mean, but rather ‘are about what our words should mean, or what concepts those words should express’ (2018, p. 28). No-one, three decades ago, would have argued that the word ‘marriage’ is used to describe same-sex relationships, only that it ought to be, and now—with perhaps some hold-out traditionalists—it is. ‘Rape’ is a more complex case. Capellen focuses on the possibility of non-consensual sex within marriage being so classified, but there is also the question of whether any degree of violence or threat is required. In an article from 1995, Keith Burgess-Jackson discusses the dispute between two feminists (Christina Sommers and Marilyn Friedman) over, among other things, whether Scarlett O’Hara was raped in ‘Gone with the Wind’. Rather than accuse either party of being guilty of manipulating the meaning of the word, he suggests that ‘they have different theories of the nature of rape, that is, different accounts of what rape is and of what makes it morally objectionable […] different and competing theoretical definitions of the word “rape’’ (Burgess-Jackson 1995: p. 419). Burgess-Jackson goes on to describe how the battle between these theories played out, and describes three extant and competing theories of rape: the conservative, under which it is a form of property crime; the liberal, under which it is a form of assault; and the radical, under which it is a suppressive act of degradation. What is striking is how uncontroversial the ‘radical’ feminist definitions seem today. For instance, he cites a definition from Andra Medea and Kathleen Thompson, who ‘define rape in a very simple, clear-cut way. Rape is any sexual intimacy forced on one person by another. Rape differs from acceptable sexual relations in one, and only one, concrete way. One person is unwilling and is therefore forced’ (Medea & Thompson 1974, p. 12). Willingness is expressed through consent, and while high profile cases of sex apparently offered freely by those not considered legally capable of giving consent, being under-age or over-inebriated, for example, continue to stir mixed feelings (see Morris and Topping 2016), the general shift from ‘rape’ signifying a violent attack to ‘rape’ signifying sex which did not receive legally recognised consent from all parties does appear to have taken place. There can be little doubt that the conservative view has been left behind, in Western societies at least, and this change has taken place as a result of campaigning by certain identifiable individuals who have championed one theory of the concept of rape over others. So, of course, while it is difficult to bring about such change, it is not unfeasible. The important point, I would suggest, is that the engineer must be in tune with a wider movement in order to gain acceptance: the shift in the meaning of marriage has taken place as part of the widespread increase in public acceptance of homosexual relationships, and the changing definition of rape is connected to altered attitudes about the relationship of wives to their husbands and the more general realisation that women do not exist merely as objects of sexual gratification for men.

The scope for successful engineering projects does seem rather limited though.Footnote 3 It is a separate question whether or not it is advisable: even in cases where language is brought into agreement with social mores, it is not clear that this is better done via the ‘amelioration’ of meanings, rather than through new coinages. For instance, everything written or said about marriage until quite recently referred to relationships between men and women. Is a relationship between two men or two women really the same thing, or would it be better to have new words? That is a question I shall not attempt to answer, but I do think it is worth raising. More importantly, for the purposes of this paper, where such cases lead to a successful, or partially successful, change in the language community’s use and understanding of a particular concept, I suggest that there is a real danger that it may also lead to instances of fallacious argumentation when the term is employed.

3 Fallacies of Language

The first known list of fallacies, Aristotle’s thirteen sophisms listed in the Sophistical Refutations, was divided into those relating to language and those not so-relating (Aristotle 1955, p. 13). While later authors have not generally made this the key distinction in their lists of fallacies, no such list is complete without mentioning at least equivocation and vagueness as potential weaknesses of an argument.

Of Aristotle’s six linguistic fallacies, three—accent, composition, and division—are essentially related to the written form of ancient Greek. The same names have been taken and applied to different phenomena more recently, but such uses should not be confused with the original conception. The others are homonymy, amphiboly, and form of expression. The first two lead to ambiguity, either at the semantic or syntactic level, while the third concerns the making of category mistakes under the influence of the linguistic representation used. This last is of particular interest from the philosophical perspective as it not only touches upon the relationship between thought and language, and the influence of the latter on the former, but also encompasses a whole range of conceptual errors into which philosophers may be thrown: not all nouns are things, not all verbs are actions, but their being nouns and verbs can seduce us into thinking they are so.Footnote 4

In the rest of this section, I shall discuss a number of errors or confusions which may come about because of language and to which, I suggest, any arguments involving engineered concepts are particularly vulnerable. I use the traditional fallacy names with the more systematic naming of the actual fault (Hinton 2021) given in parenthesis and include a number of reasoning errors which do not appear on Aristotle’s list and some which have not always been categorised as fallacies of language.

3.1 Ambiguity and Equivocation (Consistency Fallacy)

All language is, to a degree, ambiguous. Some words have multiple recognised meanings, others have subtler variances in use; many constructions may be understood and interpreted in different ways. This is an inescapable aspect of linguistic communication and we must make the best of it. Within argumentative discourse, this only becomes problematic when a word or structure is used inconsistently within one argument. For this to happen there must be an ‘active ambiguity’ (van Laar 2003, p. 35), that is, multiple meanings within the same context, and it must have an effect on the quality of the inference (cf. Tindale 2007). It is also important that the ambiguity must arise in the contributions of one of the arguers: if two discussants are using a word with a different meaning, we have a case of verbal dispute, not equivocation.

That being the case, the danger of an error in an argument involving an engineered concept appearing due to ambiguity lies within the competence of the speaker using that concept. When once we begin with the new meaning, we must stick to it and not be tempted back into a traditional use of the same word, through force of habit if nothing else. That danger does seem very real though when we are dealing with such common items of lexis as ‘woman’ and ‘man’. The word ‘women’s’ in ‘women’s clothes’ and ‘women’s athletics’ has two quite different meanings: in the former it refers to expectations of gender and its outward representation, in the latter to a biological status, subject to testing and enforcement. Much of the debate over the admittance of transgender women to participation in women’s sports events is brought about by a confusion over this difference: since women’s athletics is, explicitly and purposefully, designed for biological women, to the exclusion of biological men, it is not obvious that a person of female gender who is not a biological woman should be involved in it. To say that certain people should be allowed to compete in (biological) women’s sporting events because they are (gender self-identified) women, is to equivocate on the term. That is not to say that there are not other good reasons for allowing transgender women to compete, only to note that there is an error in any argument made to that effect from the definition of the event.

3.2 The Straw Man (Pertinence Fallacy)

Although the Straw Man has not traditionally been listed as a fallacy of language, recent scholarship has emphasised how it may come about as a misrepresentation through the change in the way a word is understood (Tindale 2007) or as a result of rephasing (Visser et al. 2018). A Straw Man argument may be strong or weak, but it is unsuccessful because it is not pertinent to the question at hand. By providing a misleading portrayal of an opponent’s position, the arguer committing this error is effectively arguing against the wrong position and, therefore, not achieving anything dialectically. At least not in a strict sense: in reality, the unfair misrepresentation of one’s adversary may have a number of rhetorical advantages, especially if the injustice goes unseen (see Schumann et al. 2021).

Whilst there is no reason to believe that conceptual engineers would be more likely than anyone else to deliberately misconstrue their opponents in debate, there is a danger that they will apply their own understanding of an engineered term when it is used by another and so misrepresent the position they are questioning. For example, if I were to say that not all races need help to overcome their oppression, and my interlocutor understands the term ‘race’ according to Haslanger’s engineered definition, which includes the characteristic of being oppressed, they may take my stance to be one denying aid to certain oppressed peoples, when, in fact, I meant to say that I do not think all peoples are oppressed. The first of those standpoints would be rather easier to discredit than the second.

This is likely to prove especially problematic when certain members of the language community are unaware of the degree to which a concept they wish to employ has been reengineered. Misunderstandings are a common part of communication, but that doesn’t mean that we should not seek to avoid them and refrain from introducing confusion into language purposefully. In his discussion of language and worldview, Manfred Kienpointner notes that ‘daily conflicts and misunderstandings in verbal confrontations between such subgroups are at least partially caused by the paradoxical fact that speakers of, say, English, do not speak the 'same' language’ (1996, p. 482), even though they are employing the same words. Knowingly using words in ways which others are unlikely to recognise or accept is, then, a dangerous game and one which may have a negative effect on one’s ability to communicate successfully with speakers of apparently the same language. Conceptual engineers, by definition, set out to employ terms quite purposefully in ways which do not conform to the general standard of a speech community and, therefore, have a special responsibility to take care that they do not misrepresent others or lead others to misunderstand them. A fallacy may occur through simple error rather than by design, but where our activities greatly increase the chances of such an error, we are to be held accountable, even if no sophistry was deliberately intended.

3.3 Begging the Question (Productivity Fallacy)

One major difference between formal and informal logic is that the latter demands that arguments be productive, that they add something to the debate. Aristotle, in his Prior Analytics, included this quality in his description of the syllogism (Striker 2009; Dutilh Novaes 2017) and while occasionally an apparent tautology may be used to make a point, it is generally accepted as a rule for arguers. When a question is begged, we find an assumption of the conclusion is contained within the premises which we must assume to be true in order to reach that conclusion. Again, if we employ Haslanger’s definitions, we cannot argue that women are oppressed, because her definition includes the characteristic of being oppressed as essential to being a woman. Our claim would effectively be that the oppressed are oppressed. Those occasions on which similar statements are meaningful actually rely on the ambiguity of the terms involved: the English saying ‘boys will be boys’, for instance, first applies the word ‘boy’ to mean a male child, and then employs the same term to refer to a stereotypical pattern of behaviour. Whilst it has popular currency, if only as an excuse for unruly males, any argument based upon its wisdom would be guilty of an equivocation.

The danger, then, for the conceptual engineer is that by including certain ‘additional’ characteristics in a new definition, particularly if they are of an evaluative nature, the audience is being asked, in a covert fashion, to assume elements of the arguer’s worldview when discussing questions which may pertain to or reflect that worldview. If assumptions concerning the topic at issue are baked in to the language being used to discuss it, unbeknownst to one side of the debate, then the debate is not actually a live one at all. That does not mean, of course, that all arguments with engineered concepts will beg the question, but there is a clear vulnerability. There is also a similarity to the type of fallaciousness discussed in the next subsection. The famous ‘No true Scotsman’ fallacy is a form of persuasive definition, which, at the same time, leads an argument into circularity (Flew 1975; Anderson 2017). In Flew’s best known example, a Scot reads about the activities of a sex maniac in England and asserts that no Scotsman would behave in the same way. When confronted with an even more shocking case from Aberdeen, he insists that no true Scot would do such a thing, thus redefining what it is to be a Scotsman. As is so often the case, the traditional fallacy names trip over each other and confuse the issue by focussing on form of argument rather than types of error.

3.4 Persuasive Definition (Concept Fallacy)

The term ‘persuasive definition’ was introduced by Charles Stevenson (1944). It refers to the practice of adjusting the meaning of a word whilst maintaining that it remains the same word, often with the suggestion that the new meaning is the real or true one. This becomes a fallacy of argumentation when an inference relies upon the new meaning for its acceptability and would fail if the more usual meaning were understood. Andrew Aberdein has called this practice ‘gerrymandering a term’ (Aberdein 2006) and the phrase is apposite: as with electoral gerrymandering, it involves the unacknowledged shifting of boundaries for the accomplishment of one’s own purposes.

Within the Informal Argument Semantics (Hinton 2021) unacceptable moves concerned with definitions are considered concept fallacies, as they involve the shifting of concept/word relations. This raises the question of whether the whole project of Conceptual Engineering qualifies as one big exercise in persuasive definition. Conceptual engineers may not claim that their redefined concepts bear the ‘true’ definition, but they do assert that such concepts have a ‘better’ definition, or are simply better concepts. One might claim, therefore, that any argument featuring an engineered concept, where the engineered feature is pertinent to the quality of the inference, is guilty of a concept fallacy. This is an issue to be explored in greater depth elsewhere. The following section describes a particular variety of definitional fallacy, supposed to be of especial importance to philosophers in their work, and thus, of great significance to our theme.

4 Nelson’s Philosophical Fallacy

The English language publication ‘A Theory of Philosophical Fallacies’ (2016) is a translation from the original German of a series of lectures delivered by Leonard Nelson at the University of Göttingen in 1921. Across the twenty-two lectures, Nelson builds a view of philosophical practice from a Kantian perspective, and is harshly critical of contemporary work which he describes as relativist, artistic, or verbose. In the first lecture, he stresses that good philosophy is hard work, and it is his concentration on a careful, rigorous technique which brings him to consider the errors with which he finds philosophical writing to be riddled. Although he expresses a desire to ‘show specific types of errors of reasoning that are typically found in philosophy, i.e. that occur again and again, so that the multiple errors found in philosophy can be reduced to a definite and indeed quite small class of errors’ (2016, p. 33), he actually points to one particular core mistake, that of a definition which leads to an equivocation, which in turn leads to a false dichotomy. In this way, by attempting to more narrowly define concepts, philosophers reduce their field to a series of debates over competing ideas, which may very well both be wrong.

In his introduction to the translation, Fernando Leal, describes the move which leads philosophers into this error in a way which makes its connection to Conceptual Engineering very clear: it is ‘to replace the meaning of whatever piece of vocabulary we are using with a very special and different concept or idea, the consequence of that concept-swapping being that one can now prove practically anything’ (2016, p. 5). Whether or not we agree with the claimed consequence, there is no doubt that Nelson is concerned with condemning the very activity which conceptual engineers promote. The crux of the problem is that in reengineering a concept, we do not make the original version vanish; indeed, we may often need the original to be still active in the minds of our audience. Nelson himself puts it thus: ‘The fallacy consists in arbitrarily defining a concept that we are already acquainted with. When we do that, it often happens that the definition is circular in that the original concept is hidden in the definiens’ (2016, p. 157).

This leads to the two questions: does it, and, if so, how? Nelson answers these with examples from the history of philosophy. Suppose we follow Bentham and state that pleasure is what has VALUE, we are then asked what gives us pleasure: that which we value, of course. As Nelson puts it: ‘all pleasure, all liking, all wishing, striving and desiring is just a specific form of the act of valuing’ (2016, p. 129). Things get worse when Mill attempts to rescue Bentham’s theory from charges of being nothing more than a base Epicureanism. Mill introduces the idea of ‘higher pleasures’ and distinguishes them from others which are ‘less valuable’ (Mill 1962, p. 260–261), a move which, as Nelson points out, requires ‘a valuation of pleasure according to an independent standard’ (Nelson 2016, p. 129). Nelson skewers Mill’s sophistry well: there is much talk of what is ‘noble’, of what an ‘intelligent human being’ would want, and the needs of a ‘being of higher faculties’ (Mill 1962, p. 259), but it is clear that this simply boils down to Mill’s personal tastes—my pleasures are high, their pleasures are base. In fact, Mill’s theory is that things which give Mill pleasure are more valuable than things which give less exalted mortals pleasure. Unsurprisingly, his argument for this position is paper-thin. Of interest to us, however, and Nelson maintains this is the case with all value-based ethical theories, is that the particular notion of VALUE can only be defended by appeal to the previously held concept of value.

Something similar happens with theories of TRUTH. Nelson discusses Poincaré as an example of a pragmatist who believes that truth can be equated with convenience. He notes that any affirmative statement that we make carries with it the implied claim that our statement is true. Thus, when one says ‘the truth of a statement lies in its convenience’, one is actually saying ‘it is true that the truth of a statement lies in its convenience’. It then becomes clear that in making the original claim about truth one was presupposing a concept of truth. Nelson’s comment rings out as a warning to would-be engineers of concepts:

A pragmatist can naturally escape the objection by saying: ‘I want to use the word truth in the sense of CONVENIENCE.’ That is of course allowed. But in that case he should not forget that by so changing the meaning of words they have not eliminated the concept that was previously associated with the word ‘truth’, a concept that other people still associate with the word, indeed a concept that he himself still uses in the very act of redefining it, for in doing so he is of course expressing a judgment. (2016: 124)

In all these cases, the concepts at issue, truth, value, good, are well understood even if they are hard to define. Any attempt at redefinition, or precisification relies on the original, vaguer concept for acceptance, which means we have gone round in a circle. Further, the very desire to redefine comes from the familiarity and importance of the original concept—there is little appetite to interfere with obscure concepts which do not have much of a role in our thinking or discourse.

It may appear that many engineered concepts would not be vulnerable to the type of problem Nelson describes in the way in which judgements about truth and value are, but the example below illustrates how attempts to employ an altered concept are always, or at least often, likely to require an understanding of the original concept if they are to communicate their intended meaning. The following comes from a newspaper article which is not concerned primarily with gender issues, but is a review of a comedy show, performed by Dave Chappelle. The review is unremittingly critical and takes issue with the comedian’s jokes which are seen as insulting towards transgender people. In response to an attempt by Chappelle to be more conciliatory by likening trans men to himself as fellow ‘dreamers’, the author asserts: ‘these statements are incorrect—trans men don’t dream of being men, they are men’ (Bramesco 2024).Footnote 5

This can be fleshed out into an argument thus:

Believing oneself to be a trans man does not make one a dreamer, because being a trans man is a reality, not a dream.

The argument would rely on a very simple definitional warrant premise that for a state to make one a dreamer, it must involve dreaming, not reality.

Now, clearly for the author, the word ‘men’, as used in the clause ‘they are men’, is a term referring to gender and may include some people who are not biologically male, and that must also be so in the phrase ‘trans men’. That being the case, there is a sense in which the argument is a tautology saying simply: all types of men are men. If it isn’t, then the two uses of ‘men’ must somehow differ and the author is guilty of an equivocation. Suppose that the goal of the author is simply to remind us that all types of men, whatever we or anyone else think of them, are men. The motivation for doing so depends on our acknowledging the relationship between the traditional and engineered versions, and thus relies on the audience’s having the traditional concept, which would suggest that the category of ‘trans men’ is not a subset of the category of ‘men’, in mind, even if they are fully accepting of the more recent one. This is important because it suggests that the force of engineered concepts may be reliant on their not actually replacing older concepts, but existing alongside them as ‘special’ uses. The purpose of the engineered concept, in that case, would be to stand in contrast to the original, and this presupposes that the original has not, in fact, been supplanted. Many examples of this effect are discussed by Perelman and Olbrechts-Tyteca (1958) in Chapter IV of the New Rhetoric dealing with what they term ‘dissociation’, where a Term I and a Term II version of one word coexist and are contrasted with each other for rhetorical effect [see Schiappa (1985) and Frank (2020) for a full discussion].Footnote 6 Indeed, they make it clear in section 90 that Term II can only be understood with reference to Term I, and suggest in section 94, on dissociative definitions, that the use of Term II may actually make the notion more confused.

The way in which such situations can lead to false dichotomies is obvious. If someone wishes to deny that the ‘real’ meaning of ‘man’ is connected to gender stereotypes, they will have to come up with an alternative, presumably linked to biology. This can lead to a debate over whether ‘man’ is a biological or social concept, where it should be perfectly obvious that it is neither, or rather that it is both. The argument is pointless, yet also totemic for some on both sides of the debate.

5 Conclusion

This paper has suggested a new dilemma for Conceptual Engineering. It repeats the charge of triviality made by Deutsch, and adds an alternative of extreme vulnerability to fallaciousness. That vulnerability, of course, does not necessarily undermine the entire project, but it does call into question the overall impact of engineered concepts. Whatever advantages may be gained through the spread of an ameliorated concept might well be outweighed by the negative consequences in the form of error and confusion in communication generally and argumentation in particular. The onus is very much on the engineer to ensure that their use of words which have been repurposed to some extent does not put others at a disadvantage or, in the worst case, serve to manipulate their audience into accepting certain positions which they would reject if they fully understood the speaker’s intended meaning. At the same time, the engineer risks being misunderstood and thought to hold positions which, in fact, they do not. There may well be cases, of course, where the engineer considers this risk to be one worth taking, given the possible advantageous social or epistemic outcomes they believe would ensue. It is important that they do so advisedly.

There is a telling passage in Haslanger’s famous paper (2000, p. 52) where she accepts that if others are particularly attached to the conventional meanings of words, such as ‘woman’ and ‘race’, then she is prepared to find alternatives. She is reluctant to do so, as she admits that the reengineering of an old word has a rhetorical effect which would not be matched by the coining of a new one. This I find troubling. If the engineer is aware that they are attempting to fit a new conceptual meaning onto a commonly used word which other speakers may be keen to preserve, but they find it rhetorically advantageous to ignore the possible confusion and so press on, the suspicion begins to arise that the intention is one of either provoking controversy and discord or employing a rhetorical device to gain the agreement of the unsuspecting, rather than of improving understanding or expression. Where this is the case, they can be held responsible for any fallacies which arise, even if their intentions are not to confuse or manipulate. This is clearly something different from the activities of philosophers, however forlorn, who have sought to explicate or precisify philosophically important concepts such as truth and knowledge. Although I remain highly sceptical as to the meaningfulness of the very idea of ‘better’ meanings for words, one conclusion that can be drawn from the foregoing is that Conceptual Engineering understood as a practice by which philosophers try to solve philosophical puzzles through the discussion of the meaning of words and possible suggestions for changes in use, is a very different thing to the politically inspired promotion of new meanings to achieve certain defined social ends. There is something a touch disingenuous about grouping the two under the same heading of Conceptual Engineering simply because both involve a reassessment of the meanings of words and concepts.

Finally, to end on a more positive note, there is clearly wide scope for the development of a literature comparing the Conceptual Engineering enterprise and its conceptions of the mutability and malleability of meanings with existing work on definitions, argument, and rhetoric. Would-be engineers have everything to gain from a better acquaintance with that body of work.