"I did not realize Plautus wrote an entire monologue in Punic"

I am ashamed to admit I didn't know this either. That'll teach me to ignore the dramatists...

It is suggestive that the limits of where Arabic is spoken today are about the same as the limits of where Aramaic and Punic were spoken 2000 years ago.

Careful! Punic was spoken in southern Spain.

Actually, I believe it was Ernest Renan who claimed that the expansion of each Semitic language "paved the way" for the next expanding Semitic language: thus Akkadian expanded and eliminated (non-Semitic) Sumerian, Akkadian is then eliminated by Aramaic, meanwhile Punic expands in North Africa and is in its turn eliminated by Arabic (as Aramaic today soon will be, sadly). True, Punic was spoken in Southern Spain, but then so was Arabic for a long time: Punic was, however, also spoken in other places where it was replaced by non-Semitic languages (Sardinia, for example).

Nevertheless it is true that Arabic, today, is chiefly spoken in lands that were Semitic-speaking before they were arabicized. Hence the well-known "structural stability" of Semitic languages may have little to do with the linguistic structure PER SE and more to do with the fact that Semitic languages, for most of their history, were chiefly in contact with other Semitic languages.

Beyond just the horrible sacrifice of the Punic first-born, their introduction of the pun into Mediterranean culture has to make you question the worth of their civilization.

Probably they just gradually annoyed each other to death, and would have died out anyway without the Roman intervention.

the fact that Semitic languages, for most of their history, were chiefly in contact with other Semitic languages.
Um: Sumerian, Elamite, Old Persian/Pehlevi/Farsi, Greek, Coptic, Romance (from Latin to Sabir)... What do you mean, chiefly?
Also, consider those Semitic languages which underwent centuries of intensive contact with non-Semitic languages, like Maltese (Italian/Sicilian) or Amharic where the structure has remained recognizably Semitic (e.g. the perfect/imperfect oposition and the root system).

Bulbul: my point is that a large number of Arabic speakers (East of Egypt the vast majority) today descend from people who, before shifting to Arabic, spoke other Semitic languages as L1's: that neighboring languages (various stages of Persian, Greek...) influenced the Semitic languages spoken in the fertile crescent is indubitable, but there never was large-scale language shift from Greek or Persian to Arabic (unlike, say, the large-scale shift from Iranian to Turkic in Central Asia, or from Munda and Dravidian to Indo-Aryan in South Asia): thus, a country such as Lebanon has been predominantly Semitic-speaking for over three thousand years (with Phoenician, Aramaic and most recently Arabic having become the dominant form of Semitic): prestige languages such as (in the case of Lebanon) Greek, Latin, French and English have left their mark, of course, but there is a difference in nature between the influence of a prestigious elite language upon a language community as a whole, and the influence of large-scale shift to said language community: and large-scale language shift, in Lebanon as in many other Arabic-speaking countries, has as far as we know chiefly been a matter of shift between different Semitic languages.

I agree that Amharic and Maltese are still typologically Semitic, but less so, I would claim, than your typical Arabic dialect (Amharic has lost internal nominal plural-marking, for example, whereas as a rule Arabic dialects preserve it). And there is one modern Semitic language heavily influenced by contact with a non-Semitic language, Eastern neo-Aramaic (influenced by Kurdish), that is so deviant that a scholar recently wrote an article wondering whether it even deserves the label "Semitic" (from a typological perspective, of course). I may be able to dig up the exact reference should anyone be interested (my files are even messier than Ancient Middle Eastern history. Really!)

Etienne,

your point is well taken. But:
there never was large-scale language shift from Greek or Persian to Arabic
Are we really so sure? Do we even know enough about the population structure back then? Medieval Arab grammarians and language mavens often complained about the decline of Arabic (laḥn al-ʿāmma) and attributed it largely to what we would now call a language shift.
And then there's Egypt with Latin, Greek and Coptic spoken at the advent of the Islamic conquest. Again, one would assume a significant language shift took place there.

Eastern Neo-Aramaic and Neo-Mandaic are a good example of Semitic languages with heavy contact interference. The question is whether they should be regarded as a rule or an exception. Considering their sociological status, I'm leaning towards the latter.

This debate strongly reminds me of Owens' "paradigm stability". He essentially argues the opposite point to yours and to illustrate, he brings up two examples: the Akkadian preterite and the Nigerian Arabic imperfect. Over four millennia lie between the two and yet, so Owens, the paradigm has remained virtually unchanged. Owens also points out that peripheral varieties of Arabic like Nigerian Arabic and Uzbekistan Arabic where one would expect the most deviation from the "neo-Arabic" "standard" are in fact pretty conservative. I think he makes a compelling argument for the linguistic factor being responsible for the structural stability of (at least) Arabic.

Bulbul--

Owens makes a good point, although I believe the intrusion of Arabic into Nigeria took place comparatively recently (fourteenth-fifteenth century) and was due to Bedouin settlers (whose Arabic is notoriously conservative).

While Greek had penetrated deeply into Anatolian territory, further South in the Middle East it was found solely in settlements along the coast, and in Egypt in Alexandria only: elsewhere it was an L2 only (a very prestigious one, granted): from what little I know of Middle eastern epigraphy and the like it seems indubitable that most of the Arabic-speaking countries East of Egypt were predominantly Aramaic-speaking in the seventh century. So while there doubtless were non-semitic speakers who shifted to Arabic, a majority of the shifters were Semitic speakers (again, that is certainly true East of Egypt: whether to the West speakers of Punic, Romance or Berber were the more numerous language group in North Africa is probably impossible to determine).

Incidentally, there were few if any Latin-speaking communities in Egypt or the Middle East: indeed there is strong evidence that (originally Latin-speaking) veterans' colonies in Egypt (or the Eastern Mediterranean more generally) shifted to Greek rather quickly (which doubtless explains the paucity of Latin loans in Coptic or Aramaic when compared to the number of Greek loans in either language, and indeed many of these Latin loans seem to have been borrowed via Greek).

An example of Semitic languages being influenced by language contact is in the South Semitic Ethiopic languages, where the morphology remains Semitic but the syntax, especially the final position of the verb, is radically non-Semitic, presumably due to contact with Cushitic.

This is one of the more interesting discussions my comment sections have been graced with. I'm fascinated by this stuff.

It looks to me like Punic was quite restricted in distribution by the time the Arabs got to North Africa. Contrast the copious early Romance loanwords into Maghrebi and Andalusi Arabic documented by Corriente, and the several well-known Punic loanwords into Berber, with the paucity (nonexistence?) of Punic loanwords into Maghrebi Arabic, and contrast the Berber, Latin, and Greek etymologies given by al-Bakri with the absence of Punic ones (a few Punic placenames appear in his work, of course, but he is unaware of their etymology.) It seems probable that east of the Sinai Semitic languages were quite rare at the time of the Arab conquest - in which case, since West Asia accounts for only about a third of the Arab world by population (less by area), most Arabic speakers live in areas that were not Semitic-speaking.

Also, if the deciding factor were whether the previous language was Semitic or not, you would expect Egyptian Arabic to be the least conservative dialect - which is not obvious.

As to non-conservative Semitic languages, Central Asian Arabic too has become SOV, under the influence of Turkic and Persian. And of course there's Nubi, a fairly recently developed, surprisingly well-documented Arabic-based creole in Uganda and Kenya with radically simplified morphology and other interesting differences.

Lameen: you make a good point, except that if, at the time of the Arab invasions, Romance in North Africa was a predominantly urban and Punic a predominantly rural language, we would expect the former language to have been more "visible" to Arab invaders than the latter, doubly so if we consider that Romance/Latin would have been associated with Christianity (and hence enjoyed a certain degree of prestige) in a way Berber would not have been. Moreover, while there are a great many Romance words in Berber as well as in North African Arabic, a major problem lies in disentangling words that go back to indigenous North African Romance from Romance words borrowed from (early) Iberian or Sicilian Romance...

Hat: the fact that there are fascinating discussions here speaks volumes about the quality of your postings. Really, what can I say, except: keep it up!

And while Coptic's not actually Semitic, it has recognizable common factors with Semitic languages.

>copious early Romance loanwords into Maghrebi and Andalusi Arabic

These may have come into Andalusi Arabic first, though, and Spain was a richer region likely to export culture even aside from the impetus of forced emigration, so this may not be a measurement of the strength of Romance in early Islamic North Africa.

>well-known Punic loanwords into Berber, with the paucity (nonexistence?) of Punic loanwords into Maghrebi Arabic, and contrast the Berber, Latin, and Greek etymologies given by al-Bakri with the absence of Punic ones

But might a loan from Punic be invisible, if the cognate Arabic word was used? Or if Punic speakers quickly accommodated to Arabic, they might have left little trace. Are Aramaic loans into Arabic in the Mashriq distinguishable? Chinese people tend not to know that loans from Sino-Japanese words were originally coined in Japan.

And sometimes there is a lack of substrate influence even where we know a conquest took place, like the miniscule influence of British Celtic on English. Here my guess (which mysteriously I've never seen stated anywhere) is that both sides had enough familiarity with Late Latin to find it more convenient to communicate through that medium, though not enough to make it their own mother tongue.

>east of the Sinai Semitic languages were quite rare

I think you mean west...

There are enough phonetic differences between Punic and Arabic that one would expect many if not all loans to be recognisable - sh rather than th, uu rather than aa - and of course there are substantial semantic differences; enough, in fact, that Punic loanwords into Berber such as agadir "wall" or zalim "onion" are easily distinguished from Arabic ones. Then again, our records of Punic are sparse enough that a rare word might simply go unnoticed. In the Mashriq, a number of Aramaic loanwords have been spotted, although there the task is complicated by the fact that Classical Arabic itself already contained Aramaic borrowings, whereas it almost certainly does not contain Punic ones.

Incidentally, in an article a couple of years ago I actually suggested Maghrebi kaf "cliff" as a possible Punic loanword (cf. Hebrew kef); but the evidence is not particularly strong.

Caffeind: your idea about Late Latin being used for communication between Celtic and Germanic speakers in Early Britain is interesting, but unfortunately there do not exist any early Latin loans in Old English that are not to be found in its Continental relatives, which seems to preclude Latin having played a (non-elite!) role as a lingua franca.

We shouldn't make the mistake of believing there to be a correlation between the number of speakers of various substrate languages and the number of words borrowed from said substrate languages. I stand by what I wrote earlier: Latin was certainly a more prestigious language than Punic, and hence would be expected to influence North African Arabic and Berber more than Punic would. (By the way, in my most recent comment above, "in a way Berber would not have been", I meant of course "Punic"). Punic speakers' greater demographic weight (assuming them to have been more numerous than Latin speakers, for the sake of argument) might not have been enough to compensate for this difference in prestige.

Berber is a greater influence on North African Arabic than either Punic or early Latin; number of speakers (and ancestry of current Arabic speakers) seems a plausible enough explanation for that.

>phonetic differences between Punic and Arabic that one would expect many if not all loans to be recognisable - sh rather than th, uu rather than aa

Right, but what if Punic speakers dealing with Arabs (or perhaps the Arabs too) were familiar with the systematic differences, and simply transposed a Punic word to the corresponding Arabic one? That would hide a loan, or prevent it, depending on how you view this. While Arabs may have been new to North Africa, the idea of various Semitic dialects would not have been.

>unfortunately there do not exist any early Latin loans in Old English that are not to be found in its Continental relatives, which seems to preclude Latin having played a (non-elite!) role as a lingua franca

That is also compatible with Latin playing the same role on the Continent at the same time, as we in fact know that it did. The only difference is that we think Gaul was totally Romanized and Britain only superficially Romanized, an idea which in turn is based more on our knowledge of the outcome of French in France and English in England, than on actual records or archeology.

Lameen: the difficulty is that, Berber still being a living language, it is difficult to know how many Berber loans/features (assuming for the sake of argument that we can always clearly distinguish cases of Berber influence upon North African Arabic from North African Arabic innovations that thence made their way into Berber...) in North African Arabic are ancient (due to substratum) and how many are more recent (due to adstratum).

Caffeind: I half agree with you: I am inclined to think Britain was more heavily romanized than most people believe, but the absence in Old English of loans not found in other North Sea Germanic languages makes me suspect that Latin was no longer the dominant vernacular at the time of the Anglo-Saxon invasions.

As for the idea that Punic speakers might have been aware of Punic-Arabic sound correspondences: this is certainly something ordinary speakers can be aware of, but there is a sharp difference between synchronic and diachronic correspondences: Punic had a much simpler consonant system than Arabic, and thus there was no one-to-one equivalence between a Punic and an Arabic consonant. To believe that Punic speakers, in trying to make their words sound more Arabic (in and of itself not an unlikely scenario), always managed to hit upon the etymologically correct equivalents, strikes me as extraordinarily unlikely.

It is true that Punic speakers' initial approximations to Arabic would have been full of inaccuracies. But we are talking about the ultimate effect on Arabic. Effects on North African Punic would be visible only if it had survived either as a language or as an identifiable substratum. For the effects on Arabic, it only matters that an identifiably non-Arabic loanword is not being coined.

Punic's concentration in urban areas may have also decreased the need for loans from it. Rural languages contribute names for distinctive local rural features.

In even more basic terms, the volume of loanwords coined in a language encounter is highly dependent on the particular language contact situation and varies greatly.

As for Latin loans in Germanic, Germans served in the Roman army throughout the Empire, and much of their terminology for things in the Empire would have stabilized by the time of the invasions. How much new would they encounter in Roman Britain? And was there any German homeland unaffected by contact with Romans, or did they all exchange soldiers and traders?

We know the least about the initial Anglo-Saxon arrival and interaction with Romanized Britain, which may have been still somewhat intact and urbanized. This phase may have been the most likely to use Latin. In the mid-500s there were separate and hostile British and Anglo-Saxon kingdoms in the west and east, and little interaction. After the conquest was completed, there were pockets of British serfs remaining at the least in Wessex, and if some of these were Celtic-speaking, the explanation, if any, for the lack of loanwords has to be their social position. It certainly seems different from the wealth of Gaelic words recorded in Hiberno-English.

To caffeind: actually, there is regional differentiation in the Germanic-speaking area with regards to Latin loans: some such loans are only found in Anglo-Saxon and North Sea coast Germanic languages and dialects on the one hand, and some are only found in the High German area on the other: I believe (this is from memory) Old High German MILIZ "soldier", from Latin MILITEM, is a case of a loan not found outside the High German area. I'm not denying there was loanword diffusion between different Germanic groups, but it did not yield a unified "pan-Germanic" set of Latin loans.

And while there few Celtic loanwords in Anglo-Saxon, "few" is more than "none", and none is the number of Latin loanwords which are found in Anglo-Saxon and unknown elsewhere in Germanic. One might argue for diffusion from Anglo-Saxon to its continental brethren, but for this to have caused ALL Latin loans in Anglo-Saxons (and none of the Celtic loans? Explanations that involve speakers carrying etymological dictionaries around are not very credible) to have spread to Continental North Sea Germanic is improbable in the extreme.

"how many Berber loans/features... in North African Arabic are ancient (due to substratum) and how many are more recent (due to adstratum)."

Ancient or recent, I suspect they're mostly substratum. The slow process of language shift from Berber to Arabic has never yet stopped since the 1100s; there are many Arabic-speaking communities today whose grandfathers spoke Berber, and such recently shifted communities often have noticeably more Berber words than others. (Cases like this illustrate the folly of arguing, as I read once in The Languages of China, that substratum influence has to be confined to a single generation.)