Tuesday, November 08, 2005

Curiosities of Semitic articles

As David Boxenhorn noted in his comment to the previous post, the definite articles of Hebrew and Arabic display two odd-seeming properties:
  • agreement: if a noun has a given article, so does any adjective modifying it. Thus "a short boy" in Arabic is walad-u-n qaSiir-u-n (where -n marks indefiniteness, and -u marks the nominative case), whereas "the short boy" is al-walad-u l-qaSiir-u. (The vowel of al- elides when preceded by another vowel.)
  • In direct compounds of two nouns (possessed-possessor, or more generally modifier-modified), the first noun cannot take any article. Thus in Arabic you can say yad-u l-walad-i "the boy's hand" or yad-u walad-i-n "a boy's hand" (-i marks the genitive case) but not *al-yad-u l-walad (intended to be "the hand of the boy") or *yadun al-walad (intended to be "a hand of the boy").

The second property isn't actually all that "exotic" - English does the same thing! You can say the man's hat, but never *the man's the hat or *the man's a hat; just as in Arabic or Hebrew, to make the full range of possible definiteness distinctions you have to resort to prepositions.

Definiteness agreement between nouns and adjectives is more unusual, but at least one Indo-European language has it: Norwegian. No question of substratum influence there, certainly... Anyone have another example?

However, in determining whether or not the article represents a shared innovation, the question is whether other relatives have it. I recall that Biblical/Imperial Aramaic did (later varieties lost it), but I'm not sure of the detailed behavior of its definite article. The Berber obligatory noun prefixes probably derive from an original article (see my post Beja and Beyond) but, though distinctly similar to the Beja definite article, don't seem directly comparable to the Arabic and Hebrew ones.


ACW said...

What is the situation in the Ethiopic group? An extremely varied subfamily, close to (or within?) Semitic; you'd thing there'd be good comparative data there.

In Hebrew, at least, in the possessed-possessor construction, the possessed noun surfaces in a special form, the "constructive"; I always think of it as an "ungenitive". Often the constructive is identical to the base form, but it varies often enough that it's clearly a separate "case". At any rate, the way the rule was expressed to me was that nouns in the constructive can't take the article: they are already determinated, one might say, by their marked connection to the possessor.

Example: ha'arets, the land, but 'erets yisra'el, the land of Israel. (In this case, the constructive shows an internal vowel change.)

Anonymous said...

I don't think it really counts as another example, but the structure is also found in Swedish, and I think in Danish, but my knowledge of the latter is minimal. It might be interesting to find out if there are any notable difference among the Scandinavian languages with regard to this.

Also, you could look at Bulgarian for another kind of articular (?) oddity. (I'm sure Bulgarian speakers don't find it odd, but I do.)

David Boxenhorn said...

My point about the curious Semitic articles is not that they are super-duper weird, but that they are unusual enough that it makes me suspect some kind of relationship - either areal or common descent.

But it seems perfectly clear that the set of demonstratives from which they descend was present in the proto-language! The only question is whether they were used as articles too.

Question: Do Norwegian articles derive from demonstratives? If so, what do they look like today? Are they applied to adjectives too?

ACW said...

One more slight weirdness about Hebrew — and I hope our host forgives me for wandering off topic — is that "second-generation" demonstratives like ze must agree in determination with the noun they modify.

Example: ha'aretz haze, this land, but literally "the land the this". How do demonstratives like "this" work in Arabic?

Lameen Souag said...

The Ethiopic group is within Semitic; however, the oldest member, Ge'ez, has no article, while Amharic has a postposed article which looks to derive from a different set of demonstratives (huwa "that" et al.) A special construct form of the noun is found, to varying degrees, in each of Arabic, Akkadian, Geez, Hebrew, and Aramaic; it must be at least partly proto-Semitic.

It has been argued that Hebrew and Arabic are actually more closely related to each other than either is to any other Semitic language, in which case the articles would surely be an innovation in their common ancestor; however, this position is controversial.

In Arabic, demonstratives, unlike adjectives, do not change according to definiteness, and normally precede the noun.

ACW said...

Ah, most intriguing. So the "the-this" construction is an innovation that postdates the divergence of Hebrew and Arabic.

The Ethiopic branch is of course much richer than just Ge'ez and Amharic, and perhaps more data from there might shed some light on how the Semitic determination-system evolved.

Of Aramaic, Hebrew, and Arabic, surely Arabic is the outlier? What about Phoenician? My sense of the branching order here is maddeningly vague.

Justin said...

Phoenician is VERY close to Hebrew.

The noun-adjective definiteness agreement thing does occur to some extent in Greek too. When an attributive adjective follows its noun, and the noun is definite, the adjective needs to be too. Otherwise the adjective is predicative. Thus:

ho kalos aner or ho aner ho kalos = "the handsome man"
but kalos ho aner or ho aner ho kalos = "the man is handsome"

And while you can't say "the man's the hat" in English you CAN say "the man's hat", which you can't in Hebrew (as opposed to "man's the hat," which is good Hebrew but bad English.)

Justin said...

Oops, slight but critical mistake. Please read:

kalos ho aner or ho aner kalos = "the man is handsome"

This is of course similar to Hebrew ha'ish hayafeh vs. ha'ish yafeh

Lameen Souag said...

"Of Aramaic, Hebrew, and Arabic, surely Arabic is the outlier? What about Phoenician?"

Phoenician and Hebrew are practically two dialects of the same language, Canaanite; Phoenician is perhaps the most divergent of the Canaanite varieties, but the similarities are nonetheless overwhelming. I too prefer the traditional position that Aramaic and Hebrew likely group together against Arabic, but Hetzron has made a strong if counterintuitive case for Hebrew and Arabic grouping together against Aramaic.

Ran said...

Regarding Hebrew's "ha-aretz ha-ze" ("the-land the-this"): that's the way people say it today, but Biblical Hebrew phrased it differently. I believe Biblical Hebrew used the construct form, saying "eretz ze" ("land-of this", "this's land") to convey the same idea.

So, it's not surprising that Arabic doesn't do the same thing; the change seems to me to have happened well after Hebrew was distinctly Hebrew.

John Cowan said...

The (mainland) Scandinavian articles do indeed descend from demonstratives; they aren't very different from the articles in the other Germanic language except for being postposed and therefore enclitic.

Indeed, I don't know of any language with definite articles aren't transparently derived from its demonstratives. It's also extremely common, though perhaps not universal, for indefinite articles to be derived from the numeral "one".

Rain_Drops said...

To ACW : The "the-this" construction is not a Hebrew innovation, it's present in most Arabic dialects, We say in Egypt : El ard di, and in other dialects : el ard hazi, el ard hay, el ard hadi, it's considered a way of affirmation :)