Imagine there was a society in which blue eyed people are referred to with blee/bler pronouns, and green eyed people are referred to with glee/gler pronouns, and one day someone from that society saw a brown eyed person and had no idea whether to categorise them as blee/bler or glee/gler
I don’t think it’s all that arbitrary. A bit, sure, but not fully. If i’d have to guess, it probably has to do with the cultural, social and biological relevance of sexual preference in bonding and maybe reproduction and all that.
Language tends to evolve to convey culturally relevant information. You know, like tu/vous or status/relationship distinctions in japanese or the amount of words there are for friendship in arabic.
If that was the case it would be a much more common feature. There are about 7000 languages in the world and only about a hundred of them have that feature.
But, cultures speaking languages with that feature tend to be much more likely to subjugate other cultures than other cultures are.
So, while if you take a random culture on this planet, it most likely won’t have the he/she distinction in their language. But if you take a random human on this planet, they most likely do have it.
Yeah, we have all kinds of labels that are not arbitrary, but also also not 100% accurate in all cases. They only tend to be a problem when people try to treat them as absolutes.
There are plenty of examples of languages with two, three, or no grammatical genders (which don’t always correlate with biological genders—e.g., some languages see ”animate” and “inanimate” as genders.)
If it isn’t arbitrary, why is there so much variation?
Edit: I’m not saying that adding information to nouns by splitting them into rough categories isn’t useful, just that the specific dimension on which we make the split (sex vs animacy vs eye color) is arbitrary.
I think it is, just not fully. You’d be hard pressed to find a language with 6 genders, for example.