It feels very anglocentric to have a dedicated feature for just the English locale.

You either get a nice, small package if all you need is the rules for English, or you have to take on the entire dataset if you want to use it with any other language, even if it's just a single other one.