On Mar 19, 2010, at 9:56 AM, Alex Fink wrote:

> On Fri, 19 Mar 2010 13:28:40 +0100, Andreas Johansson <[log in to unmask]>
> wrote:
>
>> It just occured to me that if that language were English, the most
>> straightforward adaption to Meghean phonetics of English "saint" would
>> be _seint_ [Seint], which happens to look like the genitive plural of
>> a possible Meghean word _seth_ [SeT], so if there were to be a Meghean
>> word for "saint", I'd be sorely tempted to make it just that. I seem
>> to recall hearing of similar cases of backformation of loans in
>> natlangs, but I can't offhand think of any examples. Does anyone?
>
> _Cherry_ is a simliar case in English: it's had the apparent plural stripped
> out from Norman French _cherise_. Of course in this case similar things
> were happening to words which had been around longer, e.g. _pea_ < _pease_
> (into OE from Latin). And this maybe isn't such a satisfactory example,
> anyhow, since these are the sort of words we might expect vacillation
> between count and mass for, independently.
>
> Oh, I've got a better one. Swahili, as I've heard it, has taken several
> loans in forms that look plural and singularised them by swapping native
> plural for singular class prefix, at violence to the original root: so
> 'video' is sg. _kideo_ pl. _video_, 'bartender' is sg. _mutenda_ pl.
> _batenda_, 'mudguard' is sg. _digadi_ pl. _madigadi_.
>
> Alex

This is one of the things that fascinates me about language change. But I wonder how it happens when the words in question have already had a long history in a language, where they fill specific slots -- I would think that the speakers would have a good grasp of which slots they fill, and not go changing them around too much. With loanwords, at least, there is no previous history of patterning, so I can understand that.

For example, in Vulgar Latin we find several neuter plural nouns in -a being reanalyzed as feminine singular (which is also signified by -a); e.g. gaudium "joy", lignum "stick (of wood)" > *gaudia, *ligna. Taken out of context, this change seems completely likely to me, due to the phonological equivalence; but semantically and grammatically, I don't understand it -- those nouns had always fallen into slots where only a plural would be appropriate (e.g. they triggered plural agreement). How did people lose sight of that? (As you say above, perhaps confusion between mass nouns and plural count nouns played a role, especially in these two examples, but I'm sure there are others where that isn't a big issue.)

Another thing I wonder is why the French definite article becomes part of the noun in French-derived creoles, but only with words beginning with a vowel. (AIUI, sometimes the singular definite article is used, and sometimes the plural.)