If you asked me what it means to call someone an “asshole” before I really thought about it, I probably would have suggested an “expressivist” analysis. The word, I might have elaborated, is just another term of abuse, a way of simply expressing one’s disapproval. Much as if one had said “Boo on you!”, one isn’t trying to say something that can be true or false, correct or incorrect. The job of foul language like “asshole” isn’t to describe the world, but simply to express one’s disapproving feelings, in an ejaculatory or cathartic burst facilitated by inherently emotive words.

I decided this was completely wrong one day in the summer of 2008, while surfing in a crowded line up. I was watching a guy brazenly break the rules of right of way and thought “Gosh, what an asshole”. That wasn’t a new thought, but I then noticed, for the first time, that this thought has what philosophers call “cognitive content”. I was trying to say that the guy in question was properly classified in a certain way. Other law-abiding surfers weren’t properly classified under that term, and so it could be true or false, correct or incorrect, to say that this guy was, in fact, an asshole.

That got me thinking about what it would be for someone to qualify as an asshole. Harry Frankfurt partly inspired this. I thought: Frankfurt put his finger on “bullshit”, and I am a philosopher, so I should define “asshole”. After considerable tinkering and with the help friends, I settled on this definition: the asshole is the guy who systematically allows himself special advantages in cooperative life out of an entrenched sense of entitlement that immunises him against the complaints of other people.

This definition is hopefully significant simply because it prompts one to think, “Hey, I’ve met that guy”. Maybe you encountered him this morning in the coffee shop. Maybe the guy wouldn’t quiet down on his mobile phone, despite obvious sneers. Maybe he drives as though he owns the road. He probably says, “Do you know who I am?” to the maître d’ at a restaurant when he’s not quickly seated. Although it of course matters how the definition’s details get worked out and applied, the main idea is that even those inclined to quibble in the small might agree that “asshole” doesn’t simply have expressive meaning. Its function is to classify a person, correctly or incorrectly, as having a particular kind of moral personality.

I soon discovered linguistic evidence for this “cognitivist” rather than “expressivist” treatment of the word. It makes perfect sense to say of someone, “Yes, he is my friend, and he’s fine to me personally, but I admit he’s an asshole”. You can also coherently say things like “General MacArthur was plainly an asshole, but in the end a force for good”. Now maybe those contexts don’t express all-out disapproval, because they still express disapproval in a muted form that is outweighed by other considerations. Yet even all-out approval seems perfectly coherent. It is coherent – and indeed commonplace – for an asshole to proudly own the name. He boasts “Yes I’m an asshole – deal with it”. He taunts his subjects with this pronouncement precisely because he seems to approve wholeheartedly.

An obvious worry here is that I have gone too far in a “cognitivist” direction. Surely the word “asshole” is often used to vent feelings with no concern for whether its target meets any set of necessary and sufficient conditions for its application. If nothing else, then, the emotional charge often associated with the term is something that must be explained.

An easy move here is to follow Stephen Finlay’s idea that ethical language “pragmatically indicates” attitudes of approval or disapproval within a normal conversation. If you call Trump an asshole while we are talking, it will normally make sense for me to interpret you as disapproving of Trump’s conduct. But that may only be a matter of what it is reasonable for me to infer about you given our conversational context and my general knowledge of how people feel when they speak. The very meaning of the word “asshole”, as set by the linguistic rules that govern its use, doesn’t itself imply that you disapprove. And so “Trump is an asshole” can count as true or false in a straightforward way. Expression is, in the standard parlance, a matter of “pragmatics” rather than “semantics”.

I take great comfort in Finlay’s idea, if nothing else as a last resort. It is also interesting that, if need be, a “cognitivist” account can go even further and simply admit expressive meaning as part of the very semantic meaning of “asshole”. As David Kaplan (and David Copp, in a different way) has explained, that needn’t undermine pretentions of objective truth.

Kaplan suggests that, alongside “descriptive” terms whose meaning can be given by a definition (like “fortnight”), we can explain the meaning of “expressive” terms ( “damn”, “bastard”, “ouch”, “oops”, and “goodbye”) in terms of an idea of “expressive correctness”. So suppose someone sincerely uses an expressive term. This doesn’t simply report certain purported facts; “ouch” doesn’t just mean “I am in pain”. It does, however, purport to “display” things as being a certain way. I display the fact that I am in pain when I say “ouch”, and I display the fact that I despise someone if I say “that bastard”. Use of the term is expressively correct, says Kaplan, when the supposed fact holds: I say “ouch” and I am, in fact, in pain; I say “that bastard” and I do indeed despise the person. Use of the term is expressively incorrect when the supposed fact doesn’t hold – as when I’m not really in pain, but faking to get attention; or when I don’t despise the person at all.

But now consider “oops”, which, unlike “ouch”, has an element of objectivity. To say “oops”, Kaplan suggests, is (roughly) to purport to display the fact that “one has observed a minor mishap”. So saying “oops” will be (expressively) correct when one has just seen someone inadvertently break a wine glass, but incorrect when the mishap isn’t minor (for example, a whole building falls down, killing hundreds, in which case “oops” could at best be a macabre and vile joke). So whereas the correctness of “ouch” depends entirely on one’s state of mind, the correctness of “oops” also depends on the world, independently of one’s subjective attitudes.

Now turn to foul language. For a given foul term, we can ask whether its correctness conditions are fully subjective, as with “ouch”, or at least partly objective, as with “oops”. And, when you survey many foul terms – even the foulest of the foul – they easily seem to be objective expressives: their correctness can seem to depend, at least in part, on what is going on in out in the world. Here are several examples along with some tentative suggestions about what those “objective correctness conditions” might be.

“shit” – as when “shit!” is said after a fender bender (in contrast with something’s being “shitty”, or of poor quality). This implies that an unexpected event has occurred that frustrates the speaker’s aspirations (such as avoiding costly auto repairs, getting to work quickly). The aspirations are subjective, or up to the person, but the event in question has to actually occur. Shit happens, and has to happen, for it to be shit: shit has existence as its essence. Or, more plainly: if the fender bender or other untoward event hasn’t actually occurred, it won’t be expressively correct to say “shit!” (Except in fictional cases, when things are fictionally presented as actually happening – a pretty special context.)

“fucking”–as in “the fucking car wouldn’t start, right there in the middle of the road”. This implies major frustration of someone’s plans, though not necessarily the speaker’s. I could be hearing the story of a woman whose car had stalled in the middle of the road and say “really, and the fucking car just wouldn’t start?” Her plans would be majorly frustrated, but (unlike “oops”) I wouldn’t have had to observe the events myself or even myself have the frustrated plans (the event may have passed, in which case there is nothing, as regards that event, to plan for).

“fucker” – as directed at a toaster that burns one’s toast or an electrical outlet that shocks one. The object is personified as having malicious intent. The expressive correctness of the terms depends on whether the metaphor is apt, given the situation’s descriptive features. (We’d usually say this of more readily personified objects, such as toasters or computers, but perhaps not a stationary rock.)

“mother fucker” – a metaphorical way of saying someone can’t be trusted. You can’t trust him not to have sex with your own mother if he had the chance. The term is inapt if someone is completely trustworthy.

“bastard” – much as with “mother fucker”, implies treachery, that the person in question is a traitor, or would-be traitor. When we aren’t thinking of its non-derogatory meaning (“one without a father”), a person doesn’t count as a bastard, in the central, paradigmatic sense, unless they are treacherous or prone to betray others in relationships. It would be a mistake to call someone a bastard if he (or she?) were reliably faithful in what his relationships required.

Kaplan seems to disagree about “bastard”. (He treats “That bastard X” as akin to “That damn X”.) But no matter. Though “damned” or “damnable” does have an older and very rich set of meanings, especially religious ones, I think Kaplan is right about one use of “damn”, which seems to be a purely “subjective” expressive, without external correctness conditions. If I say “Damn you!” it seems pretty clear that I in some sense express my disapproval of you. The word “damn” itself carries that implication when it is sincerely used, by virtue of the linguistic conventions that set its meaning. I can’t sincerely (and properly) say “That damn Trump; but I really like him!”

So traditional emotivism gets one case right. Still, this seems an exceptional case. If so many foul terms seem to be “objective expressives”, then it is natural to see any expressive element in “asshole” in a similar way. My definition says when the term is descriptively correct. If we like, we can add that it will be expressively correct when, and only when, the speaker in fact disapproves, or finds disapproval appropriate, in virtue of the asshole’s acting upon entitlements that he does not in fact have.

This kind of reasoning has led me to be pretty sceptical about expressivist analyses of foul language generally. So much so that I’ve been flirting with the idea that there are four main categories of foul language and that expressivist treatments don’t apply to any of them straightforwardly.

The categories are these:

(1) vice terms, which purport to classify a kind of moral personality, correctly or incorrectly (e.g., “asshole”, jerk”, “bastard”, “motherfucker”, “schmuck”, “boor”, “cad”);

(2) pejorative terms, which assume false normative or moral claims about certain independently identifiable groups of people (e.g., “honkey”, “wop”, “kike”, “limey”, “chink”, “n—-r”);

(3) slurs, which invoke a metaphor (“four-eyes”, “pig”, “dickhead”, “cocksucker”) that is often literally false and yet may be “true”, metaphorically speaking.

(4) objective expressives (e.g., “shit” and “fucking”), which are expressively correct or incorrect depending on what is going on in the world.

My conjecture is that these categories cover a lot if not most of foul language, and that each requires an appropriate “cognitivist” treatment. That is only a conjecture at this point, but it is not hard to appreciate its potential significance. If my hunch pans out, then expressivism is wrong even about most foul language. That is, it is wrong about the area of language where it has the best, fighting chance of being right.

I might add that this is presumably a further count against expressivism about ethical language generally. The case for a cognitivist “constructivism” about ethics of the kind I favour gets even better.

To be sure, everything then depends on how the different categories of foul language get developed. So here are few thoughts about “asshole” in relation to pejorative terms and slurs, taking each in turn.

Consider racist pejorative terms. In contrast with “asshole”, I’m inclined to see these terms quite differently, as systematically wrong. Here I mean the class of pejorative terms that get directed toward a particular group of people (“Yankee” or “honkey” and whites; “wop” and Italians; “kike” and Jews; “chink” and Chinese people; “limey” and English people; “n—-r” and African-Americans). If we follow recent work by Chris Hom and Robert May, as I do, then the very meaning of these terms assumes certain normative beliefs or assumptions about the group in question. “Wop”, for instance, assumes that Italians are the appropriate object of contempt and discrimination, simply because they are Italian.

Now suppose the assumed judgements of appropriateness are radically mistaken. From a moral perspective, no one is the appropriate object of contempt or discrimination simply because they happen to belong to a racial group. In that case, the racist judgement that Berlusconi is a wop will lay claim to truth and yet fall into error by virtue of resting on a false presupposition. Since “wop” has a false presupposition, the claim that Berlusconi is a wop is neither true nor false.

We presumably won’t want to say that about “asshole”. We can surely be mistaken about what someone is or is not entitled to. Yet it is a pretty radical form of scepticism to hold that the moral presuppositions of an asshole judgement are systematically wrong. In that case, there are no kikes, no chinks, and certainly no “n—–rs”. And Berlusconi is not a wop – even if he is an asshole.

Now consider the different category of slurs, like “four eyes” and “dickhead”, understood as metaphors. “Asshole” probably initially got its meaning as a metaphorical slur.

Calling someone an asshole is of course literally false in the non-moral sense of “asshole” that refers to a physical body part, just as it is literally false to say of someone that he is his own left arm. To say either thing is at best some kind of metaphor. Yet, if I am right that “he’s an asshole” can be a literally true or false, from a moral perspective. This raises an interesting question: how could “asshole” come to have acquired a literal moral meaning? The real, full story is presumably complicated. Still, it is helpful, or at least interesting, to speculate with something like the following conjectural history.

In the beginning was the word, used as a mere metaphor. Geoffrey Nunberg tells us that “asshole” caught on in recent times among World War II soldiers. Imagine its first non-literal use: a solider called his superior officer – let’s call him Sargent Pug – “an asshole”, thereby inviting his fellow soldiers to imaginatively engage Pug in a certain unflattering light.

Although it is literally false that Pug is a body part, likening him to a foul and hidden part of his own body called attention to his arrogant disposition and repugnant personality. As with any good metaphor, at least as Richard Moran explains it, the point was to see or experience Pug as an asshole, to interpret him and his actions from an imaginative frame of reference that treats a man as the physical embodiment of a body part that behaves much as that body part might. They were to feel and almost believe that Pug is as problematic and as foul and yet shamefully exposing himself in public.

The more literal minded of soldiers might have objected: “Look, it’s not strictly speaking, literally true that Pug is an asshole any more than it can be literally true that Pug is his own left arm”. In reply, the soldiers would have laughed off the objection as beside the point. Sure, they’ll say, the statement “Pug is an asshole” is literally false; “asshole” is just a metaphor. The point is that it is especially apt. Or as we might elaborate the idea, the use of the word by the soldiers says things about Pug that could not have been literally said of him up to this point. Perhaps some of those things can be put as literal truth-claims, such as the claim that Pug isn’t worthy of respect, or that he abuses his office, or that he is contemptible. But those truths aren’t the whole point of the metaphor, which is mainly to see Pug in an unflattering light that cannot be fully expressed in so many truths.

Because the metaphor was apt, it quickly caught on as a way of speaking. Use of the term became popular among the other soldiers on the base, in the wider army, and then in further reaches of society. The term was especially useful as communication. People found calling someone an asshole an especially handy way of conveying and perhaps even venting their contempt for abusive authority figures who are not, for them, worthy of respect. Those who heard the metaphor invoked found it especially easy to grasp what the speaker meant: they meant not simply to vent feelings of contempt, but to invite an interpretation of their target that would make those feelings of contempt fully appropriate. That’s because, when someone calls someone an asshole, you could easily tell that he or she has a certain moral view of things: the view that person called an asshole is not worthy of respect, because of how he treats those around him.

Soon enough, when someone used the word, you could readily land upon this interpretation of the user’s meaning, without knowing very much about his or her context of utterance. The metaphor thus came into a different kind of meaning: calling someone an asshole moved away from mere metaphorical communication and became a literal, more routine claim to truth, a claim to the truth of a moral judgement: that the person in question is not worthy of respect because of the way he treats others.

Even now few were especially careful or aware about exactly what kind of unsavoury people they were calling an asshole. Still, there was a rough but remarkable convergence. The invited perspective would be appropriate for certain kinds of people and not for others. Knowing or not, people began to grasp the rules of normal usage, which called for one type of person to be called an asshole and left other types for better names.

Eventually, the rules settled. They became well enough established that a competent speaker of the language could entertain asshole judgements without meaning to express contemptuous attitudes. The curious user of the language would wonder whether this person is the right kind of person to qualify as unworthy of respect, even without feeling at all exercised about this, let alone speaking out about it. The curious person would ask questions that don’t express negative attitudes, such as “Is Trump an asshole?” And people could reason with hypotheticals, such as, “If Trump is an asshole, I probably shouldn’t watch his show; Trump does seem to be an asshole: so I probably shouldn’t watch his show”.

Beyond private reflections, all of this could be discussed in a mode of cool-headed argument among friends. Over coffee, friends might agree that, yes, Trump indeed qualifies as an asshole, literally speaking. They might conclude on that basis that they probably shouldn’t go out of their way to listen to him or watch his show, having reasoned together and reached agreement upon what they all regard as an objective matter of moral fact: Trump is, in fact, an asshole.

Now when a less agreeable fellow in the coffee shop says it isn’t literally true that Trump is an asshole, the friends don’t say what the soldiers said to their literal-minded fellow solider. They don’t agree and say that this is beside the point. They beg to differ. They reply: “no, that’s wrong; Trump definitely qualifies; he’s literally an asshole; you’ve made a mistake”. At that point, the journey of “asshole” from metaphorical slur to vice term was complete.

Aaron James is Associate Professor Of Philosophy at The University Of California, Irvine. He is the author of Fairness In Practice: A Social Contract For A Global Economy and Assholes: A Theory.