A few years ago, I had a good talk with NPR's Gene Demby about why we have so many terms for people of color, and what linguists refer to as the euphemism treadmill. (He ended up writing this article). I've been thinking about this topic off and on since then. As with so much of language, for terms for other people the choice of word a speaker uses signals something to listeners about the person speaking.

What this means, is that we can think of the use of terms for groups of people as a signaling game, and we can situate it within the kind of discussion of strategic thinking that happens in Game Theory. What makes this instance particularly interesting is that it’s a coordination game on a massive scale. Basically, we’re all sometimes confronted with having a choice of words, and that choice of words may tell other people something about us (whether we want it to or not!). That, in turn, may affect how they react to us or treat us. Therefore, when confronted with such a choice, say between “colored” and “African American”, we have no choice but to strategize about what word we choose. (Technically, this is not strictly true, as we could just say whatever, but this is generally a losing strategy —- one that results in people thinking less of you and adjusting how they treat you. However, as with everything linguistic, it’s complicated; some people are willing to make an allowance for, say, the elderly, as when older relatives of mine asked after my groomsmen, and wanted to know how “the Canadian”, “the negro”, and “the oriental” were doing, and followed up with “send them [our] love.”)

It turns out these kinds of patterns look very similar to those from evolutionary models originally designed to capture gene flow, and predator-prey dynamics, with some minor tweaks. I won’t get into the math here, since it’s more complicated than I’m willing to dive into in a blog post, but the general idea is that there are a few factors that can all affect the euphemism treadmill. First is random drift — the creation of a new word or repurposing of an existing word can be thought of as analogous to mutation. Sometimes these forms disappear, and sometimes they completely take over; in the long run, in strict competition, it’s one or the other. Second is a predator-prey dynamic: we can think of these forms as being in competition for the same ecological niche, or we can think of the new form as literally preying on the old one(s). The second metaphor isn’t perfect, but it captures something about the ecology of word use.

On a very large scale, we can think of this as a coordination game. If “bad” people say “oriental” then when we hear “oriental” we can’t strictly determine that the speaker isn’t bad. So we go out of our way as speakers to signal that we’re “good” by picking a different word. However, in large populations, in the long run, these kinds of signaling games have interesting properties for a few reasons. First, we tend to be lazy, so if we can get away with saying “oriental” and it’s easy for us (say, it’s what we’ve always said), then we’ll do that. If we think there’s no cost, we go with the easy option. Second, sometimes bad people don’t want us to know they’re bad people. So if they know that everyone else assumes one is bad if one says “oriental” when referring to a person, they may hop on the new word bandwagon to avoid being “outed” as a bad person!

Eventually, two related things happen. First, people who use the new words just because it’s what everyone else expects of “good” people sometimes give themselves away, and the words they use may become associated with people who have their views. It doesn’t matter if I avoid saying “oriental” if I instead say “those damn Asian-Americans are ruining our cities” or some such nonsense. Second, things that people think of negatively are still associated with the new words to describe them, so eventually euphemisms become taboo themselves (see, for example, “toilet”). So for two related reasons, as a euphemism or a new term gains traction, especially if it becomes the main word people use, it leads to the need for another new word to separate out who means what. If I have negative beliefs about black people, but want to be thought of in a positive light by others, I might say “African American” … but it will be abundantly clear that I still mean something vaguely negative by it. If you want to signal virtue [sidenote: “virtue signalling” was a term from evolutionary game theory that has now been adopted by some regressive cranks, and is now slowly becoming something I had to think twice about using here…because of what it now signals by association with racist and sexist groups], you have to find a new way to differentiate yourself.

And so the wheel turns again.

For each of the above charts, we can generally think of social movements that relate to the terms used — for instance, the shift from “negro” to “black” coincides with the Black Power movement — but this doesn’t invalidate the game theory approach here. Rather, knowing about these social movements adds to our understanding just how this kind of massive signaling game plays out in society, and the kind of real world repercussions involved. Obviously, there’s a lot more involved here, and the above is a gross simplification that just scrapes the surface of the kinds of strategies involved, but I find it fascinating that the models developed to better understand gene flow and animal competition do a pretty good job of also capturing how words change in society in the long run. When I’m deciding whether to say “black” or “African American” based on what I think I know about the listener (or reader), millions of other people are making the same strategic calculations every day, and our individual decision-making (and the fallout from our decisions) in part drives massive social changes on a much larger time scale. It’s related to both how signaling strategy plays out in large groups in general, and to how other kinds of words change.

-----