PhDniX's blog

Proto-Semitic Case System #3

So today I'll discuss the dual somewhat. Many of the issues found in the Plural are also found in the Dual.

The most common reconstruction of the Dual is as follows:

Nom	*-ā(-ni)
Gen.-Acc.	*-ay(-ni)

The Canaanite languages only retain the Gen.-Acc form with an *m as the final consonant, just like the plural. Ugaritic actually retains both caseforms, and surprisingly the nunation suffix is found both with suffix *mi and *ma. This is probably due to analogy from the *ma found in the plural.
This means that Huenergard's hypothesis that *ni is a dissimilation after *ā from *na in Arabic can't be right, as Ugaritic seems to show and *i too.

It is interesting to note that, although Akkadian lost every indication of *-na in the plural, the dual nunation *-ni is retained. The Akkadian dual endings are Nom. -ān Gen-Acc. -īn. Yet just like The plural is invariable, as is the dual. So there has been an analogical leveling both in the plural and in the dual which made the function of Nunation get lost in the Plural and Dual of Akkadian. The plural generalised the form without nunation while the dual generalised the form with nunation.

The next part of this series will be a short discussion on the He Locale in Hebrew and how it relates to other Semitic languages.

Proto-Semitic Case System #2

Today I'll talk about another aspect of the case system, namely the plural.

There seems to be a lot of disagreement on the reconstruction of the masculine plural in Proto-Semitic.
There is conscensus on the case forms. But not on the mimation. Hebrew points to an *m in the plural while Arabic points to *n in the plural.

Nom.	*-ū(-ma)
Gen.-Acc.	*-ī(-ma)

Is Dolgopolsky's reconstruction which makes the *-na in arabic extremely difficult to account for. One could say that maybe the caseless variety of Arabic lost the final a on this plural marker (as is the case in modern dialectal Arabic where the plural marker is -īn which then had an *m > n shift at the end of the word. Later this n was reintroduced in the classical language.

This explanation is horrible convoluted, and is actually disproved quite easily by Ugaritic. While the majority of the plurals found in Ugaritic follow the form -ū/ī-ma we find several instances of -ū/ī-na instead. There is absolutely no basis to assume that *m in this word shifted to *n. More likely is that the na form is old and is shared with Arabic.

What though, gave rise to the *m in the plural in Hebrew and Ugaritic? Was it the *m in the mimation of the singular? I find this unlikely. *m and *na are not enough alike to make it likely that the speakers of the language thought that the *m and the *n in that suffix are supposed to be the same element and should have the same consonant. If *na had lost it's final vowel already as you see in Hebrew, the analogy would've been more likely, but cuneiform Ugaritic leaves little doubt that there was indeed a final vowel a.

Simply put, I am unsure where Hebrew got the m from.

Another issue with the plurals 'nunation' *-na is that it functions quite differently from the 'mimation' *-m that we find in the singular, and therefore maybe it shouldn't be equated to be the plural counterpart of the Definite marker *-m

If we look at Arabic this becomes quite clear:

	Singular	Plural
Definite	al-muslim-u	al-muslim-ūna
Indefinite	muslim-un	muslim-ūna
Construct	muslim-u	muslim-ū

As you can see, nunation in the plural appears even in the definite form, while in the singular it does not. In Hebrew we find the same distribution in the plural (no trace of nunation is left in the singular). Akkadian has no trace at all of the nunation in the plural while it keeps mimation in the singular. If they had the same function it is unlikely that they'd lose it asymmetrically.

So is nunation and mimation really the same element? Why does it behave differently in the singular and the plural? All these questions, very little answers.

Next installment I'll talk a little bit about the Dual, although it is in many ways a similar story to that of the plural.

Proto-Semitic case system #1

I'm currently writing a paper on the Proto-Semitic case system. While comparative linguists in the field of Semitic languages can be extremely precise when it comes to sub-classification I sometimes feel that they're ignoring rather vital things.

I see a lot of thing in the Semitic languages which simply cannot be unified with the accepted view of Semitic morphology. The problem is, whatever I see, I'm not quite sure what it means yet.

So I'll be writing a couple of blog posts on the subject of the Proto-Semitic case system, partially to organise my thoughts, and more so to hear ideas from readers.

Let's start of with the singular system.

Traditionally the Proto-Semitic singular case system is reconstructed as follows:

nom.	*-u(m)
gen.	*-i(m)
acc.	*-a(m)

The two most important languages for the reconstruction of the case vowels are Akkadian and Classical Arabic. These two languages are so far apart that it can't possible be due to chance that their case systems are so similar. So, they must be to some extent Proto-Semitic.

The mimation was probably used to mark definiteness in Proto-Semitic, unlike in Arabic where it's a marker of indefiniteness. How can this be?

Dolgopolsky gave a nice explanation to explain this descrepance in two problems of semitic historical linguistics. As *m is technically a marker of definiteness it cannot appear in the two environments where we find it in Arabic. If we asume that the modern article  (a)l- was originally a demonstrative pronoun, we find the lack of mination in two situations: in a demonstrative construction and in a genitive construction.

Similar to how in English you can't say 'this the book' or 'Peter's the book'; mimation can't be used in such constructions. When the demonstrative then later shifted to become a definite article the definite marker *m became a marker of indefiniteness instead.

But things are never so easy as the above system. Most importantly a rather important source of information is ignored and that is written Classical Arabic. It is no secret that the written language actually represents quite a different dialect than classical Arabic with all it's vocalism. For example the feminine marker -atun is always written <-h>. Without any stretch of the imagination can anyone uphold that writing a final <-h> was shorthand for a wordfinal -atun. Much more likely is that the realisation was [ah] or [a] which then directly coincides with the loss of word final -t in Hebrew and Aramaic nouns (which incidentally also write this feminine ending with an <-h>).

Nunation is never written in the consonantal script. Which is odd, nunation is a consonant after all. There's no reason not to do it. Consonantal Ugaritic for example does write its mimation, so why wouldn't Arabic? It seems clear that the dialect that is written is quite different from the Arabic dialect which is represented in the vocalisation.

The last part that we find in consonantal classical Arabic is the indefinite accusative singular which is marked with an <-A> (that is an alif) this points to an [-ā], which is quite different from the vocalisation which says [an].

How did this form come to be? The first option is that *an shifted to ā at some point. But is this shift likely? If loss of *n causes lengthening, wouldn't you expect that to have happened with *un and *in as well? They did lose the *n as well since you don't see it in consonantal script. This explanation therefore doesn't run that well.

So if Arabic has a unique feature which cannot easily be explained as an analogy or sound development, one would think it's old. So maybe *-ā is the original Accusative which was later fixed to look more like the other cases by analogy in Both Akkadian and Arabic.

I can imagine how *-ā is changed to *-am and how the indefinite is then formed from *-ā to *-a. So the Proto-Semitic form wouldn't have the defininite-indefinite contrast, similar to some other case forms like for example the directive particle *-aš found as the He locale in Hebrew and directive -iš in Akkadian.

Having a definitness-indeterminate accusative is attractive in other ways too. The plural form only has a u/i contrast where i takes the function of both the accusative and the genitive. If Accusative a undetermined marked that is found only in the singular as is the directive particle *-aš, then the 'case system' is actually a lot more symmetrical.

nom.	*-u(n)	*- ū(na)
gen.	*-i(n)	* -ī(na)

But symmetry and elegance of a system should never be the sole argumentation of a Proto-language system. After all plenty of languages lack an elegant system, and there's no reason to think that a Proto-Language had its ugly quirks as well.

For the system that I've stated above to work, I need to work out one development which I don't understand.

I imagine that the system was as follows:

*-ā >> split to *-a and *-am by analogy of the nom. and gen. case.

This could work for Classical Arabic and Akkadian, but not for Written Classical Arabic. There the development went like this:

*-ā >> split to *-a and *-ā.

For this explanation to work, I need some kind of basis of an analogy to explain this step. Right now I am absolutely clueless and very open to suggestions.

Next up is a discussion on the Dual and Plural system and especially the discussion on nunation versus mimation in these forms.

The Hittite Inherited Lexicon for all to see!

It does not happen every day that expensive academic books come online for free for everyone to download, so let's celebrate.

Alwin Kloekhorst's Etymological Dictionary of the Inherited Hittite Lexicon is free for download.

Enjoy. 

More on the emphatic Berber *ṃ

Last post I discussed the strange alternation between /m/ and /ṃ/ in the Zénaga Berber word for beard. I likened it to the common alternation in the word for dog that also had a non-emphatic consonant in the singular, and an emphatic consonant in the plural.

I was under the impression that there were more words like this, albeit not that many. But it turns out there are only very few, and none of them seem to give a consistent picture. Let's discuss them here.

One word is the word for 'ear' in some dialects is found as ameẓẓuγ pl. timezγin. But this is quite different. Not only does this word inexplicable have a gender switch in the plural, it also has the emphatic in the singular and the non-emphatic in the plural, this is the oppositie situation than that of dog.

According to Kossmann some dialects also show adeggʷal pl. iḍulan 'father-in-law'
The underlying root is D/ḌWL, /ww/ gives [ggʷ]. This word has the same change as the word for dog, but as you can see other processes are at work too (like gemination in the singular).

With the much too sparse and unclear data we cannot uphold the hypothesis that *ṃ is a Proto-Berber phoneme.

How Zénaga got this phoneme, is unclear it might be some kind of influence of the [u], but it's difficult to be sure.