I just noticed two things about the preposition per. It is usually followed by a bare singular noun phrase (i.e., one with no determiner aka specifier). This is odd because singular NPs almost always require a specifier unless the head noun is a proper noun. We say, take one pill per day, not *take one pill per a day or *per days.
But even more interesting is that sometimes it is followed by plural noun phrases, almost always specified by 100, 1,000, 100,00, etc. (e.g., The abortion rate had dropped from nearly 30 per 1,000 women of childbearing age, to less than 2.) And then when we speak this, we typically say per thousand women, instead of per one thousand women. So is thousand functioning as a specifier in thousand women?
By the way, you do find examples like a 0.75 ERA and an average of 15 strikeouts per seven innings, but these are rare.
Why do you suppose this is?
Two factors stand out to me:
1. Etymology, -slash- Latin influence. "Per" was generalized from Latin and quasi-Latin phrases like "per diem" and "per annum" and "per capita" and "per centum" and "per mil" and so on, which don't have indefinite articles (because Latin doesn't).
2. Semantics, -slash- influence from such quantifiers as "each" and "every". "We take one pill each day", "The abortion rate had dropped from nearly 30 in every 1,000 women of childbearing age, to less than 2" (typically read as "every thousand"), "a 0.75 ERA and an average of 15 strikeouts every seven innings."
1 sounds like good speculation. I think part of it is that these Latin expressions, often italicized until recently, were quite limited in their use, and small community of users may be more likely to deviate from established norms than is the everyman, especially when they feel they are being superior in doing so.
Not sure about 2. Both each and every are determinatives and they would fill the specifier function, so these are not bare NPs.
Post a Comment