Asked ''Who ate the cake?'', many people might answer ''Me and him.'', and look upon ''He and I.'' as pretentious.
These same people, however, would ridicule you for saying ''Me ate the cake.''.
Thus we have some evidence of a ''folk grammar'', if you will, with no small intolerance of infractions.
My question is, to what extent are such grammars similar across languages, geographical regions, and historical times? Has anyone ever described systematics in them? Has anyone ever theorized about their origin?
A very good question. And, yes, there is even a book out from Mouton by Nancy Niedzielski & Dennis Preston called _Folk linguistics_ from about 1998 or so, which treats of the subject. One topic especially is treated more frequently than others, and that is what we call 'folk etymology'. I did an article years ago called 'English stress as folk etymology', in which I claimed that English speakers take unknown long words and break them down into 2-, 3- or 4-syllable parts (usually) which are plausible (or real) English morphemes and apply stress to them accordingly, sometimes greatly mutilating the parts in the process. Thus, 'asparagus' becomes 'sparrow grass' dialectally. We don't even require that the parts (or the whole) make sense. Nobody cares that 'cran' doesn't seem to mean anything in 'cranberry' (actually, it comes from 'crane', but not even linguists usually care about that, or even notice it--we just accept that it is 'meaningless' without further thought), or that blackberries (not an obvious case) may be red, green or deep blue. All this shows that folk etymology has nothing to do with meaning -- we know what 'asparagus' means; we just feel that it is 'too long' to be a good English word.
Now, 'antidisestablishmentarianism', although only dictionary freaks like me have any idea what it means, is OK, since we can recognize all of its parts -- and notice that most literate people can reproduce the word, even though they don't know what it means.
Likewise 'pneumonoultramicroscopicsilicovolcanokoniosis', the *real* longest word in English (probably not true any longer), which just means 'black lung disease (common in miners)' and even the non-word 'supercalifragilisticexpialidocious' (a relative shortie, except in children's films), all of which we have no trouble lugging around as excess baggage in our heads.
There are also some more recent books and articles which touch on this subject. You might check the bibliography of Niedzielski & Preston's book for more references, or do a Google search on 'folk linguistics', 'folk etymology' and the like.
One point touched on by some recent writers in this regard is that folk linguistics does have a basis often in language structure, and therefore can give us valuable evidence for linguistic analysis, often unavailable or difficult to come by in other ways. We linguists very readily pooh-pooh what seem to us to be mistaken or silly ideas about language (especially when they are propagated by self-styled 'language gurus' and the like), and often this is merited. Still, many popular ideas will or may turn out to be right, so we should be careful about too readily dismissing an idea just because it wasn't a linguist (or a linguist who shares our fundamental assumptions) who had the idea.
Hope this helps.
James L. Fidelholtz
Posgrado en Ciencias del Lenguaje
Instituto de Ciencias Sociales y Humanidades
Benemérita Universidad Autónoma de Puebla, México