So, Afroasiatic... is it really legit?

[dɮ the phoneme]

Afroasiatic is obviously widely accepted as a family, and I want to be clear that I'm not trying to peddle any crackpottery here. I'm a completely novice when it comes to the AA family, so I have no desire to butt against established scholarship. However, as a novice, there are a lot of things about the AA family that would be major red flags for me if they were claimed of any other, less established group of languages. The first is the time depth. I know that claims vary wildly, but the ones I've heard most dates it to before 10,000 BC, which is beyond the rule-of-thumb 10,000 limit on the efficacy of the comparative method. I understand that the earliest attestations of the family are uniquely ancient, but that still seems like a rather extraordinary claim. Beyond that (and maybe expectedly, if that kind of time depth is accurate), the reconstruction for PAA seems to be... seriously in need of work, at least if Wikipedia is to be believed.

There are two etymological dictionaries of Afroasiatic, one by Christopher Ehret, and one by Vladimir Orel and Olga Stolbova. The two dictionaries disagree on almost everything.[56] The following table contains the thirty roots or so (out of thousands) that represent a fragile consensus of present research:

[...]

The table contains only thirty vocabulary items, none of them attested in every brach of the family and most attested in less than half, and all with reconstructed proto-forms that look like this: *(ʔa-)bVr - "bull". Honestly, maybe I'm being too harsh, but that seems like something you'd see from an Altaicist.

The last red flag for me is that the family as we know it today was apparently first codified by Greenberg, whose track-record for the rest of the languages of Africa is bad, to say the least. Of his other two major proposals, Niger-Congo and Nilo-Saharan, the first is known to be dubious in scope, while the latter is generally regarded as more-or-less bullshit. Of course, proposals linking Amazigh with Semitic, and sometimes with Cushitic or Egyptian, do predate Greenberg's involvement, but I can't help but notice his prominent role in the current understanding of the family.

I've also heard mumblings at various points that the inclusion of Cushitic and, especially, Omotic is suspect, with the Amazigh-Egyptian-Semitic grouping being more confident, but I'm not sure what to make of these claims.

Anyway, I suppose my point boils down to this: the Afroasiatic proposal is an extraordinary claim, and I have yet to see extraordinary evidence. That doesn't mean it isn't out there! Again, I am very much a novice. So I'm curious, to anyone here who is more familiar with the AA family, what is the evidence, anyway? What should the non-expert do to convince themselves that the AA family has any more validity than other initially plausible looking macro-proposals like Altaic?

[Travis B.]

The argument for Afroasiatic is that while it is sorely lacking on reconstructable cognates, the morphological similarities amongst the languages within it are too striking to be mere coincidence. Even then, whether one can really treat all of the Greenbergian Afroasiatic, including Omotic, as a single family is highly questionable.

[dɮ the phoneme]

The argument for Afroasiatic is that while it is sorely lacking on reconstructable cognates, the morphological similarities amongst the languages within it are too striking to be mere coincidence.

Do you know of any good reference on these similarities? I'm assuming you're referring to the trilateral root structure, but my understanding was that Cushitic and Omotic are only really claimed to preserve vestiges of it, not the full system.

Even then, whether one can really treat all of the Greenbergian Afroasiatic, including Omotic, as a single family is highly questionable.

Which languages/groups are thought to be dubious, specifically?

[Travis B.]

Even then, whether one can really treat all of the Greenbergian Afroasiatic, including Omotic, as a single family is highly questionable.

Which languages/groups are thought to be dubious, specifically?

The inclusion of Omotic in particular is seen as questionable. If one takes more of a splitter approach, one could argue that only Semitic-Egyptian-Berber really holds up.

[Richard W]

The inclusion of Omotic in particular is seen as questionable. If one takes more of a splitter approach, one could argue that only Semitic-Egyptian-Berber really holds up.

But different authorities split the major branches differently, which suggests that unities between the branches were short-lived or non-existent. I believe there are serious doubts about the unity of Cushitic.

Omotic seems to show a geographical gradient of Afrasianity, which is troubling.

[Pabappa]

The argument for Afroasiatic is that while it is sorely lacking on reconstructable cognates, the morphological similarities amongst the languages within it are too striking to be mere coincidence.

Do you know of any good reference on these similarities? I'm assuming you're referring to the trilateral root structure, but my understanding was that Cushitic and Omotic are only really claimed to preserve vestiges of it, not the full system.

for me what stands out even more is the verbs. how could languages evolve the same y-/t-/n- prefixes just by chance? but as above, it doesnt stand on a single pillar .... there's the triliteral roots, the verb prefixes, the way the gender works (distinguished even on verbs), and many more things. Omotic is certainly the weakest link, but i would still call the family PAA even without Omotic, and my understanding is that Omotic is not considered the earliest branch to break off, but merely the one that has changed the most. thus the time depth would still be just as enormous without Omotic, and we won't really lose any of the majesty of our current reconstruction.

[Rounin Ryuuji]

On this point, I've wondered a few times if there might be a very distant (but so remote as to be untraceable) connection between these and Indo-European (Masculine-Feminine-Neuter systems seem fairly uncommon outside them), but this could also be an areal feature, so I've never put very much thought into the idea. I wonder now if it's ever occurred to anybody else here that some remote but ultimately unknowable connection might exist?

[bradrn]

On this point, I've wondered a few times if there might be a very distant (but so remote as to be untraceable) connection between these and Indo-European (Masculine-Feminine-Neuter systems seem fairly uncommon outside them), but this could also be an areal feature, so I've never put very much thought into the idea. I wonder now if it's ever occurred to anybody else here that some remote but ultimately unknowable connection might exist?

I personally suspect some sort of connection between Semitic and IE, mostly due to the number of times I’ve had my questions answered with ‘why yes, this feature is indeed attested outside IE, you can find it in Semitic’. (Gender is of course one feature; if I’m not mistaken, the other big one is a tripartite TA system.) But any connection would have to be areal — which is actually pretty plausible, if you assume a Semitic Urheimat in the Levant.

[Kuchigakatai]

I personally suspect some sort of connection between Semitic and IE, mostly due to the number of times I’ve had my questions answered with ‘why yes, this feature is indeed attested outside IE, you can find it in Semitic’. (Gender is of course one feature; if I’m not mistaken, the other big one is a tripartite TA system.) But any connection would have to be areal — which is actually pretty plausible, if you assume a Semitic Urheimat in the Levant.

Grammar distinctions and syntax seem to seep more easily across language families though, handled with unrelated morphemes. Like when early medieval Germanic and Romance figured it would be nice to develop definite articles from the distal demonstratives (even Swedish -en is related to English "yon", and Sardinian su/sa 'the' to Spanish "ese/esa" 'that', Italian esso/essa 'he/she'). I easily believe South Semitic is largely SOV due to living near Cushitic. And so the whole emphasis on cognate morphemes and sound correspondences...

In a way it's kind of sad comparative historical syntax can't be much of a thing though.

[WeepingElf]

Ah, Afroasiatic. As Travis B. has said, this family is accepted mainly on the basis of shared morphology; but that it is accepted while Indo-Uralic is not, is probably due to different standards between Africanists (who tend to lump things) and Eurasianists (who tend to split). Greenberg's four-family classification of African languages is IMHO not much better founded than his Eurasiatic and Amerind attempts, even if Afroasiatic and Niger-Congo (but not "Nilo-Saharan" or "Khoisan") are probably more or less actual families. The whole scheme is best understood as an improved version of an earlier fourfold classification distinguishing "Hamito-Semitic", "Sudanic", Bantu and "Khoisan" languages. What Greenberg did was essentially to break up "Sudanic", putting those languages that showed morphological resemblances to "Hamito-Semitic" in Afroasiatic, those that showed such resemblances to Bantu in Niger-Congo, while sticking the label "Nilo-Saharan" on the rest. In Eurasia, people demand more in order to accept a grouping, and Uralic languages such as Finnish or Hungarian have a long-standing tradition of being considered "utterly foreign" as non-IE languages, so why should those be related to IE? (I have also noticed that Uralicists tend to be more accepting of Indo-Uralic than Indo-Europeanists; this perhaps has to do with the fact that most Uralicists know some IE languages and have noticed the similarities, while most Indo-Europeanists know no Uralic languages and aren't aware just how similar those are.)

"Hamitic" is of course an obsolete label for those languages of Africa which showed a resemblance to Semitic: Egyptian, Berber and Cushitic (Chadic and Omotic were first not recognized as such). It turned out, though, that the "Hamitic" languages do not seem to from a valid node. What regards the relationship between IE and Semitic, it is IMHO vacuous, apart from some Neolithic Wanderwörter and a general notion of "both being inflectional"; but they inflect in utterly different ways. Some people even called the IE family "Japhet(it)ic" on the ground that it was assumed to be related to Semitic and "Hamitic", but this term has fallen completely out of use.

[Howl]

(I have also noticed that Uralicists tend to be more accepting of Indo-Uralic than Indo-Europeanists; this perhaps has to do with the fact that most Uralicists know some IE languages and have noticed the similarities, while most Indo-Europeanists know no Uralic languages and aren't aware just how similar those are.)

I have noticed the exact opposite. Prominent Uralicists like Aikio, Kallio, Janhunen etc. are dismissive of Indo-Uralic. Although Zhivlov (from the Moscow school) is supportive of Indo-Uralic as part of a wider Nostratic family. Prominent Indo-Europeanists like Kortlandt, Kloekhorst and Kümmel, etc. are accepting of Indo-Uralic. This has also been noted by Holopainen ("Fresh views on the early history of Indo-European and its relation to Uralic"):

It is good to remark here that the Indo-Uralic hypothesis is in general much more widely accepted among Indo-Europeanists (see, for example, Beekes 2011: 31–33; Rasmussen 2005) than on the Uralic side, and that most of the scholars who have worked on the topic in recent years are Indo-Europeanists.

Also, there are overlapping/competing proposals linking Uralic to Yukaghir and/or Eskimo-Aleut in North-East-Siberia. But these run into the problem that the similarities are strongest with the Eastern-most Samoyed branch of Uralic. And it is becoming more established within Uralics that the old theory of an early split into Finno-Ugric and Samoyed is no longer justifiable.

Nostraticists (Bomhard, Dogopolsky, Illich-Svitych) assume Afro-Asiatic is part of the Nostratic language family, together with Indo-European, Uralic, etc. The Moscow School now seems to exclude Afro-Asiatic from Nostratic, which makes their Nostratic very close to Greenberg's Eurasiatic. A problem for Nostratic+Afro-Asiatic is that there is no consensus reconstruction of Afro-Asiatic. The reconstructions of Ehret and Orel/Stolbova are very divergent. Bomhard basically uses his own Afro-Asiatic reconstruction that is intermediate between the two.

[Pabappa]

Indo-Uralic and Uralo-Eskimo-Aleut (usually called Uralo-Siberian, but i wanted to be specific) are definitely compatible .... the only question is which is the daughter of which. obviously Uralic cannot be the parent node since it is a fairly young family, and therefore we have to assume that there is a shared common ancestor for all three of these cold climate groups ... I believe in this theory, but I couldnt really offer a good guess as to which groups split off in which order. IE is certainly the odd one out in most respects, but the most changed member of a family is not always the first to split off ... plenty of counterexamples for that.

I am bullish on Afro-Asiatic being a very old family, perhaps as much as 18000 years old, ....far out of reach of our ability to reconstruct a language, so we can only go based on archeology, and figure out where people were living and when. If this is true, it is probably older than all of Nostratic, and obviously cannot be a child of Nostratic. I dont think it would be parent of Nostratic either .... if they're related at all, we would need to go still further back, to find the shared origin of the two macrocultures, one born in Africa and the other somewhere in Asia.

Of course, the question can be asked, if my idea of Nostratic excludes AA, can it really be called Nostratic? I dont really have an answer for that, because I dont know enough about the remaining languages in the group to know if knocking AA out would also knock out some other nodes that are considered central to the family.

[Richard W]

Some people even called the IE family "Japhet(it)ic" on the ground that it was assumed to be related to Semitic and "Hamitic", but this term has fallen completely out of use.

Well, Indo-European languages are human languages. Remember that only three sons are recorded for Noah.

[WeepingElf]

Thank you, Howl. So my impression was completely wrong.

[Nortaneous]

I personally suspect some sort of connection between Semitic and IE, mostly due to the number of times I’ve had my questions answered with ‘why yes, this feature is indeed attested outside IE, you can find it in Semitic’. (Gender is of course one feature; if I’m not mistaken, the other big one is a tripartite TA system.) But any connection would have to be areal — which is actually pretty plausible, if you assume a Semitic Urheimat in the Levant.

Contact between ~PIE and ~Proto-Semitic is known - IMO better demonstrated than contact between ~PIE and (pre-)Proto-Uralic. (Later IE contact with Uralic is of course also known - there are plenty of Iranian(?) loanwords, and Tocharian shows strong Samoyedic substrate influence - but the key word there is "later".)

[KathTheDragon]

(Gender is of course one feature; if I’m not mistaken, the other big one is a tripartite TA system.)

Ah, yes, gender, well-known for being limited to Indo-European and Semitic. While I can't straightforwardly snarkily link to WALS maps for tense-aspect to prove the same point there, there's nothing special about a three-way tense-aspect system.

which is actually pretty plausible, if you assume a Semitic Urheimat in the Levant.

And if, as says my friend who's a speaker of Tigrinya and who I trust to have a better grasp of how Afrosemitic forces a reconsideration of all the "facts" about Semitic than anyone else I know, we assume a Semitic Urheimat in the Horn of Africa, what does that mean for the plausibility of this idea?

[bradrn]

(Gender is of course one feature; if I’m not mistaken, the other big one is a tripartite TA system.)

Ah, yes, gender, well-known for being limited to Indo-European and Semitic. While I can't straightforwardly snarkily link to WALS maps for tense-aspect to prove the same point there, there's nothing special about a three-way tense-aspect system.

Sorry, I was unclear. By ‘gender’, I was referring specifically to a tripartite masculine/feminine/inanimate Rounin Ryuuji’s comment:

On this point, I've wondered a few times if there might be a very distant (but so remote as to be untraceable) connection between these and Indo-European (Masculine-Feminine-Neuter systems seem fairly uncommon outside them) …

I did assume this comment was correct though; that WALS data indicates it may have been wrong (though it’s too low-resolution to say for sure).

But then there’s the issue of a tripartite TA system. More specifically, a tripartite system with present/past perfective/past imperfective seems limited to IE and Semitic — I’ve never seen it anywhere else. (Dahl observes that a specifically past imperfective category ‘seems to be restricted to an area around the Mediterranean, including languages from the Afro-Asiatic, Indo-European and Caucasian groups’. Earlier on, he calls such a tripartite system ‘the 'classical' Indo-European TMA system … the analysis proposed here makes the classical Indo-European system look very much less different from the 'Semitic' one’.)

which is actually pretty plausible, if you assume a Semitic Urheimat in the Levant.

And if, as says my friend who's a speaker of Tigrinya and who I trust to have a better grasp of how Afrosemitic forces a reconsideration of all the "facts" about Semitic than anyone else I know, we assume a Semitic Urheimat in the Horn of Africa, what does that mean for the plausibility of this idea?

In that case my theory would be wrong, or at least highly implausible. But given that Semitic is most diverse in the Levant, and least diverse in the Horn of Africa, I personally suspect its Urheimat is in the former area.

[KathTheDragon]

which is actually pretty plausible, if you assume a Semitic Urheimat in the Levant.

And if, as says my friend who's a speaker of Tigrinya and who I trust to have a better grasp of how Afrosemitic forces a reconsideration of all the "facts" about Semitic than anyone else I know, we assume a Semitic Urheimat in the Horn of Africa, what does that mean for the plausibility of this idea?

In that case my theory would be wrong, or at least highly implausible. But given that Semitic is most diverse in the Levant, and least diverse in the Horn of Africa, I personally suspect its Urheimat is in the former area.

But this is just simply untrue. Non-African Semitic is really not that diverse at all, while Afrosemitic is diverse to the point of raising doubts that it's even a unitary branch.

[bradrn]

And if, as says my friend who's a speaker of Tigrinya and who I trust to have a better grasp of how Afrosemitic forces a reconsideration of all the "facts" about Semitic than anyone else I know, we assume a Semitic Urheimat in the Horn of Africa, what does that mean for the plausibility of this idea?

In that case my theory would be wrong, or at least highly implausible. But given that Semitic is most diverse in the Levant, and least diverse in the Horn of Africa, I personally suspect its Urheimat is in the former area.

But this is just simply untrue. Non-African Semitic is really not that diverse at all, while Afrosemitic is diverse to the point of raising doubts that it's even a unitary branch.

…huh? Here’s a map:



The Levant region contains two primary branches of Semitic; Africa contains only one part of one group (which may not even be a primary branch). The north-western area of the map is clearly most diverse.

[zompist]

The Levant region contains two primary branches of Semitic; Africa contains only one part of one group (which may not even be a primary branch). The north-western area of the map is clearly most diverse.

That map is conflating five thousand years of history in about the busiest area of the world.

If you look at 3000 BCE, it's pretty simple: Eastern in Mesopotamia, Central in Canaan/Arabia, and Southern in S. Arabia/Africa.

And if you look at 2000 CE, it's also pretty simple, because there's no Eastern subfamily at all.

In between, the only real overlap is due to Eblaite, which isn't attested past the -23C BCE. So the overlap is just part of the long process of Central languages replacing Eastern ones.