While reading An Introduction to Information Theory by John R. Pierce, I was distracted by a linguistic artifact (on page 251 of the second edition):
We can tell our friends apart, […] but we find much less that is distinctive in strangers. We can, of course, tell a Chinese from our Caucasian friends, but this does not enable us to enjoy variety among Chinese.
Referring to an individual as
a Chinese is no longer grammatically correct. When the book was written (1961) and revised (1980), the grammar apparently was as proper as
I met a Brazilian. might be today (2019). Now it sounds at least as wrong as
My friend is a French. — if not also offensive. (The explicit attention drawn to race in the passage certainly makes the language stand out. The author was illustrating a larger point: "To be distinguishable, sounds must be to a degree familiar").
I found a number of discussions regarding the current grammaticality of this phenomenon:
Several of the answers note a pattern in the word ending, e.g. "In English there is a distinction between nationalities that end in 'ian' like Canadian or Italian and those that end in 'ese' like Japanese or Burmese."
So there is a rule via which we might derive that "Ask a Canadian" is grammatical but "Ask a Burmese" is not. But this rule seems to have changed, and quite recently! A whole set of demonyms (those ending in
-ese) shifted from use as both nouns and adjectives to proper use only as adjectives, within the lifetime of many speakers alive today.
I am curious as to why this shift happened. Was it a conscious change? Has this particular case been formally studied from a linguistics perspective? Are there other examples where a production rule related to morphemic inflection has changed for reasons [which I suspect are] unrelated to pronunciation?