Consider these claims:
According to the sorts of thing one reads, these claims are the subject of a culture war between those who hold them and those who oppose them in favour of free speech, scientific progress, objective knowledge and human-independent reality. It’s a deep battle between on the one hand postmodernist inspired feminists, critical race theorists et al. and on the other the children and legacy bearers of the Enlightenment.
I doubt I’m the first person to suggest this but: couldn’t this all be a simple misunderstanding? Maybe there are different ways of reading these sentences according to one of which they are true while according to the other they are false. In order to investigate this, let’s consider what linguistics might have to say about those sentences. As we’ll see, it has rather a lot of suggestive things to say.
The above sentences are generalizations. As it happens, language provides us with a variety of means of generalizing: of talking about how things stand generally with a chunk of reality. Here are some:
Philosophers call expressions like ‘All ducks’ and ‘Every chip’ quantifiers, because they say how much or many of certain type of thing are a certain type of way. Quantifiers are pretty well understood by logicians, linguists, and philosophers of language (they present intriguing problems of their own, but that’s not for this post.)
But there is another way of generalizing without specifying exactly how much of a certain thing is the way you’re saying. Thus consider:
And:
These seem also to be generalizing, but they don’t even vaguely specify a quantity. Moreover, they differ in strengths: as we go down the list, the number or amount of the object or stuff with the property varies: thus all prime numbers are divisible by themselves and one, but only about 50% of ducks lay eggs (because only female ducks lay eggs), and very few mosquitoes actually carry the virus.
Similarly, while — let’s assume — almost all gold is shiny, a lot of water (sea water, for example) isn’t potable, and oftentimes laughter isn’t infectious (as you’ll know only too well if you’ve ever sat someone down to watch your favourite youtube video only to be met with an agonizing stony silence).
It seems that these sorts of expressions, which are (roughly) what linguists call generics, are weird. They bear different strengths, but the same linguistic form. That could get confusing, right?
Cats confusingly drinking milk — the cause of the culture wars?
Consider these two perspectives on the world. On the one, we look out at the world, survey how it is, and describe what we survey by saying something. On one, we hear somebody say something, and on the basis imagine what the world must be like.
You would expect these two perspectives to harmonize. We look out at the world, see there’s a dog on the floor, and describe that by saying ‘there’s a dog on the floor’. Or we hear someone say ‘there’s a dog on the floor’ and come to imagine the dog on the floor. In these cases, what we see in the first perspective is pretty much what we imagine in the second perspective.
With generics, a study suggests, things go badly wrong (the study is Cimpian, Brandone, and Gelman available here. I owe this reference, and the idea of this post, to Herman Cappelen and Josh Dever’s forthcoming book *Bad Language *(chapter 8), which discusses this study and more generally the possible socially deleterious effects of generics. For more on generics, see the Stanford Encyclopedia of Philosophy).
Remember the important fact that generics differ in their strength. It turns out this variable strength causes the two perspectives mentioned above to deharmonize. In particular, say we look out at the world and see 70% of K’s are G: we see 70% of cats drink milk, say. On that basis, we are happy to conclude ‘cats drink milk’.
If generics were harmonious, then from the perspective of hearing and imagining the world, we would expect that if we heard ‘cats drink milk’, we would imagine 70% of cats drink milk. We don’t do that. Instead, we conclude many more cats like milk — around 96%. Here’s the crucial point: we utter generics on much weaker evidence than we accept them.
Here’s a hypothesis: something like this explains the wars over sentences like the ones we started with. Here’s how it could go: a liberal witnesses some instances of speech being violent, science being biased, knowledge being perspectival, or reality being socially constructed.
On that basis, they conclude the generic in question. The opponent then hears them say it, and on that basis, they conclude that the liberal thinks pretty much all speech (science/etc.) is violence (/biased, etc.) The opponent disagrees, and the argument starts, and goes on and on and on.
On this understanding, though, neither is really doing anything wrong. It’s just this weird, tricky, and unexpected feature of generic claims that they have this asymmetry. Were the two parties to the debate to realize this, perhaps, the debate would either vanish, or at least could proceed on much more solid ground. The opponent could concede that some speech is violence, while the liberal could concede that not all speech is violence, and then the task could be to meet in the middle and work out exactly how much speech is violence, away from the potentially misleading effects of the generics in which these debates tend to be framed. While they’d probably still have much to disagree about, it seems that the debate could move on more profitably.