Monday, June 29, 2015

Anomalous gender agreement in Algerian Arabic

In Algerian Arabic (here, Dellys dialect), the feminine singular form of an adjective is formed just by adding a suffix -a, with almost no exceptions. In two of the exceptions, a full look at the paradigm suggests that it's really the masculine form rather than the feminine which is irregular (though the situation is less clear-cut in other dialects - in traditional Algiers, for example, the plural of "beautiful" is شبّان šəbban):
m. sg.f. sg.pl.
beautifulشباب šbabشابّة šabbaشابّين šabbin
otherآخُر axŭṛأُخرى ŭxṛaأُخرين ŭxṛin

A third case is rather different. "Such-and-such (a person), so-and-so" is expressed by the noun m. sg. فلان flan, f. sg. فلانة flana, with no known plural. (This originally Arabic form is rather widely borrowed; you may be familiar with it from Spanish fulano). From this we can derive an adjective "such-and-such a" by adding a nisba suffix -i: m. sg. فلاني flani, but f. sg. فلانتية flantiyya. To make matters worse, we suddenly find ourselves with a gender distinction in the plural, something otherwise absent from adjectival agreement in this dialect: m. pl. فلانيين flaniyyin, f. pl. فلانتيين flantiyyin.

What's going on, though anomalous, is pretty clear (recall that feminine -a regularly becomes -t in the construct state): this adjective is displaying double agreement, gender agreement alone on the nominal root flan, and normal gender+number agreement on the adjectival derivational suffix -i. Can you think of any comparable cases elsewhere?

7 comments:

David Marjanović said...

Well. The German adjective declensions are such a mess, because there are very, very few endings for a reasonably large number of gender/number/case/definiteness combinations, that such a thing has probably arisen by chance. (Almost everything is -e or -en, scattered at random, and most of the rest is -er.) I'm too tired now to burrow through the grammar of my own native language right now, though. :-)

George Gibbard said...

I can think of diminutives in Zenaga, which must originally have meant 'son(s) of …'/'daughter(s) of...'. So in a feminine diminutive you get the feminine marker *t not twice as usual, but three times (at the beginning of '*daughter of' and at the beginning and end of the base). I don't have examples to hand unfortunately.

George Gibbard said...

I may be misremembering about the feminine marking, but Zenaga diminutives certainly have plural marking twice, on '*son(s)/*daughter(s)' and on the base.

Lameen Souag الأمين سواق said...

Going back to Catherine's paper on Zenaga diminutives, it seems that some feminine diminutives - not all - do have triple marking as you described, eg tyāḏä'Märt "little beard". But that's agreement of a prefix with the noun to which it's attached, not of both with a third party...

George Gibbard said...

Maybe what you're looking for is this?
http://people.umass.edu/acharris/Resources/web%20research%20current%20EE.pdf

Lameen Souag الأمين سواق said...

That is rather relevant, thanks. I read some of her stuff a while back - need to refresh my memory...

George Gibbard said...

You might also be interested in Somali. A small number of irregular verbs preserve the Proto-Afro-Asiatic pattern of subject agreement with some prefixes and some suffixes. The open class derives from a construction with an uninflected base followed by an inflected auxiliary, so that the Proto-Afro-Asiatic verbal agreement prefixes appear as suffixes. But another small set of irregular verbs has the Proto-Afro-Asiatic prefixes twice, once as prefixes and once as suffixes. So they have both the irregular (prefixal) affixes plus the regular (suffixal) affixes.