Typically philosophy is conceived of as answering the big questions, and the big questions are questions about how reality is. This goes back to Plato: in his early dialogues he is concerned with what is called the ti esti question (ti=what, esti=is it, in Greek). Socrates in those dialogues asked what is knowledge, what is truth, what is justice, and so on (actually, he didn’t ask precisely these questions, but let’s pretend he did because I want to use them as examples below).
In recent years, some philosophers have started wondering whether the big questions of philosophy are questions about how reality is. Instead, they’ve suggested that instead of asking how reality is, the better question to ask is how reality should be. What should knowledge, truth, and so on, be?
Here’s one way to look at that question. Our word ‘knowledge’ picks out certain aspects of the world: certain states of mind, let’s say (that knowledge is a state of mind is controversial, but it doesn’t matter).
And that ‘knowledge’ (or ‘truth’, or ‘justice’, or whatever) picks out what it does pick out is important. If someone tells you they know something, that matters. If they tell you, for example, that they know the bank will be open on Saturday, so you can withdraw money then, that will affect your behaviour, perhaps substantially. If it’s Friday night and you’re wondering if you need to go out and get some cash before the weekend, and you hear that, you’ll relax, get into your pajamas, watch some Netflix.
This is not meant to be a sophisticated point: we rely on words massively to understand and to navigate the world, and to coordinate with each other. What words we have, and what they mean, is important.
We can see this by imagining languages which are, to use to technical term that linguistics and formal semanticists prefer, sucky. Thus imagine a language which, instead of our ‘hunger’ and ‘thirst’, had just the word ‘thunger’, which applied to somebody if they were either hungry or thirsty.
That would be really annoying. If your kid told you they were thungry, you wouldn’t know whether to bring it juice or a sandwich.
And we can see the opposite point by attending to languages that are — again, using the formal academic terminology — amazeballs. Thus imagine a language that contained, in addition to ‘thirsty’, ‘thirstsweet’, which applied to somebody if they were thirsty for in particular a sweet thing, and ‘thirstea’ which applied to someone if they were thirsty for tea.
That might — only might, this is just for the sake of a colourful example — lead to a better language. A language that would enable us to make better distinctions, to describe the world and interact with it better. This would never happen: you would never, on learning someone was thirsty and bringing them a glass of water, be told ‘nah, I wanted juice[/tea]’. That would save time.
Such reflections lead us to a version of what Herman Cappelen calls the master argument for conceptual engineering. Consider the language you speak. It’s not maximally amazeballs: it would be absurd to think that English or any other language makes all and only the right and most helpful distinctions that a language could make. English could be improved. If English could be improved, we should improve it.
But English contains words standing for philosophical concepts, and there’s no reason to think that those particular words are maximally amazeballs either. Kevin Scharp, for example, and maybe a bit roughly, argues that the concept of truth is kind of like the concept of thunger: it’s not very useful. This is manifest, in part, by its being subject to paradoxes such as the liar paradox (‘This sentence is not true’ — is it true or false?) Scharp thinks we should split the concept of truth up into two different concepts, kind of like we split ‘thirst’ up into ‘thirstweet’ and ‘thirstea’ (and of course we could introduce other words to cover the whole range of thirst) in our imagined amazeballs language.
And he does so, roughly, for the same reason: if philosophical English could be improved, we should improve it. This, I take it, is the conceptual engineering project. Find out which concepts or words are deficient and propose replacements for them.
Before considering some objections to this argument, I want to emphasize reasons for thinking that these sorts of considerations are important. What words we have matters immensely. Until very recently, the concept of sexual harassment didn’t exist. The story goes as so: in 1974 a teacher, a women named Lin Farley, was talking to the female students in her class. Per wikipedia:
she began to realize the extent of the problem was later termed “sexual harassment”. As the women in the class described their experiences in the workplace she noticed a pattern: every woman there had either quit or been fired from a job because they had been made so uncomfortable by the behavior of men. She discovered that this phenomenon of male harassment and intimidation of female workers had not been described in the literature and was not publicly recognized as a problem, although she continued to hear it described by women from all walks of life.
Philosopher Miranda Fricker explains what went on in this case by saying that the women lacked the appropriate concepts to make sense of a bit of their experience, and in particular that bit of their experience relating to how they were treated by men at work. Lacking the concept, they couldn’t properly make sense of their experience, and that prevented them from collectively coming together to speak out against it. They had no reason to think that their experience was instantiating a pattern that many other women’s experience also instantiated, and so didn’t say anything, until they were brought together to speak about it. (Google ‘hermeneutical injustice’ if you want more information about it.)
When you really think about cases like this, it can be quite overwhelming. Think about what distinctions we’re currently missing, what things we don’t have words for and which we consequently don’t understand our world in terms of, what things we suffer through alone because we don’t realize others are suffering in the same way. It makes your head spin, and an immediate thought is that we should try, where at all possible, to furnish our language with the right words. And so the conceptual engineering project has great possible social benefits, in addition to purely epistemic benefits(this is a point repeatedly made by Sally Haslanger, one of the big people in the field. She tends to speak of ‘ameliorative projects’ instead of ‘conceptual engineering, in case you want to google.)
But let’s return to the master argument. Take a word, any word. Almost inevitably, it could be better than it is. If so, we should change the meaning of the word so it means the better thing. It will be helpful to have an interesting example in mind. Andy Clark and David Chalmers famously argue (in their paper ‘The Extended Mind’) that we should count among the things known not only those which one can immediately access in one’s brain, but also, for example, things written down in notebooks or saved on one’s computer. Knowledge, they argue, is extended: it extends beyond the confines of our brains to comprehend the devices we use to store information. And, they suggest, we should use ‘knowledge’ so that it includes such things, even though they grant that that’s not how we currently do use it.
For example: I don’t know what the last word of the last blog I wrote is. But according to Clark and Chalmers, I should count as knowing it, because that information is stored somewhere I can retrieve it easily, and that just makes for a better concept of knowledge.
Now, in an era when we outsource so much cognitive work onto google and, increasingly, digital assistants, this can seem attractive. But also, you might think, it’s kind of pointless. Clark and Chalmers want to change how we use ‘knowledge’. But who cares who we exactly use this one English word? It’s just, as the phrase goes, quibbling semantics, and that’s a waste of time. The question whether we should call googled-knowledge knowledge is a pointless one. It doesn’t matter — if we think googled-knowledge is interesting, we can just introduce a new term for it. Arguing about what words exactly mean is for lawyers (and their clients, and former presidents) and pedants.
I think this is wrong, and interestingly so, as I’ll now show. Quibbling semantics is a worthwhile thing to do, because words have power (roughly, what Cappelen calls ‘lexical effects’). Amazing power. Think about the word ‘knowledge’. You just read it, and doing so caused you to think about a bit of reality — the knowledgey bit. If I were famous and one million people were reading this, I would have caused one million people to think about that same bit of reality. That bit of reality gets a lot of attention by virtue of being meant by the word ‘knowledge’.
So words have the power to divert attention. And the attention thus diverted can be productive: it can make people think about the thing, learn about it, discover things about it. In the case of ‘sexual harassment’ it enabled people to legislate about it, and otherwise prevent people from harming others.
But — and this is the second important thing — not all words are created equal. ‘Knowledge’ is a very well known and popular word that enables hundreds of millions of people to coordinate their attention on the same bit of reality. ‘Process reliabilism’, a term from epistemology, is known by max a few thousand. It’s much less good at getting people to coordinate their attention on the same bit of reality.
This has important scientific and social consequences. Lawyers, economists, politicians, journalists and so on all care about and think about and argue about knowledge. Only philosophers argue about process reliabilism. Some philosophers think process reliabilism gives a good account of knowledge. If those philosophers want their findings to be uptaken and built upon, you might think, it’s imperative that they get the common word ‘knowledge’ to mean something involving process reliabilism. That is, they should engineer ‘knowledge’ because ‘knowledge’ is a big ticket word, an immensely powerful tool for coordination, while ‘process reliabilism’ is a relatively weak tool. The point generalizes: if you want the science of whatever you study to progress, then you need to get your — correct, you think — views about it out there. And attaching those views to the word is a great way to do it. But this leads nicely to the second objection.
Even if you’re on board with the above, and so think conceptual engineering might be a useful project, you might nevertheless think that it’s impossible. Philosophers sometimes speak of the ‘Humpty Dumpty’ theory of meaning after the following passage from Alice in Wonderland:
“When I use a word,” Humpty Dumpty said, in rather a scornful tone, “it means just what I choose it to mean — neither more nor less.” “The question is,” said Alice, “whether you can make words mean so many different things.”
Can Humpty Dumpty, or you, or a famous philosopher make a word mean what HD/you/FP choose it to mean?
You might think that nobody, especially academics like philosophers, has the power to change the meanings our words have. Even if it would be good, it’s not possible. And so conceptual engineering would be a kind of melancholy undertaking: philosophers would know (or think they know) exactly what words should mean, and think that this would lead to great social and scientific benefits … but they can do nothing about it!
I’m not as unoptimistic as some — including Cappelen, the master argument guy — are about this. Say you’re a professional philosopher who wants to spread the word about your concept for the social and epistemic benefit of the world. Here are some things you can do:
When organizing a conference about your concept, invite one less person, and use the money that would have been spent on their hotel, travel, and food to pay a youtube influencer to use the word as you want it to be used.
Befriend your local judge and try to convince them that the concept of ‘knowledge’ they use is no good, in the hope that they bear that in mind, even in a small degree, if they’re deciding a case that involves knowledge
Speak to non-philosophers. Do what people call ‘public philosophy’. Write on blogs and such and try to spread the word — evangelize about your discovery.
The first two might sound facetious but I’m actually somewhat in earnest. If you really believe that your concept should be the concept we all use, and in particular if you think so because the collective attention its use will lead to will in turn lead to breakthroughs in your field, then it doesn’t seem to me to outside of the remit of the philosopher to do such things. After all, the goal of the philosopher is to create knowledge. One way to do that is to write papers in journals. But another way is to make sure that people at large have the right conceptual toolkit, the better to enable them to create knowledge.
The point generalizes: philosophers who liked the master argument for conceptual engineering should think about ways to spread engineered words and concepts about. This is not a worthless or shallow undertaking — it could involve the close study of history, say the history surrounding sexual harassment, of the way slang gets propagated on social media, maybe even the — admittedly kind of debunked — theory of memes that Richard Dawkins liked back in the day.
In any event, I don’t think we should to too pessimistic about the possibility of changing meaning for the better. ‘Sexual harassment’ is a good example of a word that got introduced to great benefit; ‘marriage’ a good example of a word that got changed for socially-beneficial reasons (so as to include gay marriage); ‘mass’ a good example of a word that got changed for scientifically-beneficial purposes; and several pejorative words for people with developmental difficulties which I won’t use are good examples of word that most people have stopped using. Meaning change happens and so would-be conceptual engineers — and I’ve argued that could comprise many people — should study it.