Saturday, March 29, 2014

Random Darja notes

I'm currently on a short break in Dellys, which is providing many incidental opportunities for linguistic observation. Here are a few, randomly chosen and not guaranteed to interest anyone but me:

- What kind of insufferable pedant "translates" merguez مرڤاز into Standard Arabic, on a butcher's signboard I spotted, as naqāniq نقانق "sausage"? And would they still do so if they were aware that the latter is a Greek loanword, deriving from loukanikos? Sometimes I feel that the problem with Modern Standard Arabic, for Algeria, is precisely that it's modern and standard: too extensively modernised to connect Algeria satisfactorily with its pre-colonial past, and too standardised for Algerians to feel comfortable tinkering with its vocabulary.

- There are very few Berber loanwords here that retain the nominal prefix, but I heard one or two new ones. The list so far: amalu أمالو "wet shady spot", axiṛ أخير "good morning", aqsil / lə-qsil أقسيل "grass sp.", tirẓəẓt تيرززت "small wasp", taɣənnant تاغنّانْت "stubbornness". None, unfortunately, seem to have plurals...

- Talking of which, I registered for the first time the handy "conjunction" məqqaṛ مقّار "at least", a concept I had previously had to express using French (au moins) or Standard Arabic (ʕala l'aqall) when speaking Darja. This conjunction is shared with Kabyle, but also with Andalusi Arabic (makkār مكّار) – Corriente derives it from Greek ō makarie "lucky you", but I'm not sure whether to accept that etymology.

- I belatedly realised that ṣəṛwəl صرْول, cypress, is actually from Arabic sarw سرو, with an unexplained extra letter. Another case in point: rəɣwən رغْون "to foam up" – cp. rəɣw-a رغوة "foam (n.)". Where extra letters like these come from is one of the great mysteries of Semitic, frequently discussed but never really explained.

- There's not much true code-switching into French going on here, at least not in my social circle, but I did overhear the following excellent sentence: ṛana en plein ṭyab رانا آن پلان طياب "we're in the middle of cooking". Note that en plein is selecting for a verbal noun: one could say ṛana mʕa ṭṭyab رانا معا الطياب "we are (busy with) cooking" with a preposition and a verbal noun, or ṛana nṭəyybu رانا نطيّبو "we are cooking" with a finite verb, but not *ṛana ṭṭyab.

- An interrogative relative clause with an unexpected nominal head: makaš drari mʕa-mən təlʕəb ماكاش دراري معامن تلعب "there are no kids for her to play with". The negative existential context is presumably what favours it.

Tuesday, March 18, 2014

Ghardaia: etymology, spelling, and politics

How is it possible that, in Algeria – a state normally obsessed with the need to combat divisive ideas – openly sectarian slogans like "Malikism, Malikism, national power" or "The Ibadi is the enemy of God" can be shouted by rioters and painted on the walls of burnt-out shops, in 2014? What's going on is far too localised to be explained in terms of "Arabs" and "Berbers" (contra AFP); at most, it's between Chaamba (Sh`ānba) and Mzabis (Mozabites). But the Mzabis speak Berber, practice Ibadi Islam (a small minority sect), are native to the town, and have a famously strong mercantile tradition; the Chaamba speak Arabic, practice Maliki Islam (like the rest of Algeria), used to be nomads with a strong martial tradition, and by and large are less well off. The potential economic and political causes for resentment should be obvious – and, indeed, the shops ransacked generally belonged to Mzabis, and the largely Arab local police stood aside. It is not surprising that both sides are using broader identities – Arab or Berber, Maliki or Ibadi – to appeal for help, and the Chaamba rioters are clearly demonising Ibadis, but it's housing shortages that reportedly initially triggered these riots.

Oddly enough, however, not only language but even etymology is being used as a tool of division. As I looked through page after depressing page on the events, I was surprised to notice that, while Mzabi pages, and neutral ones, spelled Ghardaia غرداية (Ghardāyah), Chaambi pages rather consistently spelled it غارداية (Ghārdāyah). The latter spelling turns out to be based on a folk etymology, deriving the name of "Ghardaia" from Arabic ghār "cave" plus Dāyah, the name of a woman – who some Chaamba claim was from the Arab tribe of Said Atba, proving that Arabs got there before the Mzabis did (قبائل الشعانبة… بنو سُليم الجزائر.) Mzabis have a version of the same etymology, in fact (chanson amazigh mozabit) – but according to them, Daya was a saintly Ibadi woman from Touat, proving that they were there first.

Either version is problematic, since the name is pronounced ɣərdāya (Berber taɣərdayt), not ɣārdāya. The Said Atba idea is especially implausible: in 1053, when Ghardaia was reportedly founded, Ibadi Berbers had been trading across the Sahara for centuries, whereas Arab nomads had barely begun to reach the area. Phonetically, the more obvious etymology is Mzabi Berber taɣərdayt "mouse" – but who'd name a town "Mouse"? Delheure suggested a derivation from tiɣərdin "shoulders", a term found in Ouargli Berber, based on its topography (followed eg here). Dabouz compares it to a Nafusi term reportedly meaning "land next to a wadi". No proposal seems entirely satisfactory, which is itself an indicator of the placename's antiquity.

Be that as it may, this pointed use of "cave of Dāyah" reinforces my impression that what's going on is a mapping of economic grievances onto ethnic/religious categories. Adding this one letter effectively says "Mzabis own this place, but by rights it should be ours" – a thoroughly wrong attitude. الله يهديهم ويهدينا!

Thursday, March 13, 2014

Book out: Berber and Arabic in Siwa (Egypt)

I am very happy to announce here (a couple of days late) that my book Berber and Arabic in Siwa (Egypt) has now come out. The oasis of Siwa in Egypt, already famous in Classical times for its Oracle of Ammon, is by far the easternmost place where Berber is spoken. As a result of its isolation from the rest of Berber, and of a history that includes significant immigration and language shift, the Berber variety spoken there is highly distinctive (and not mutually comprehensible with Moroccan or Algerian Berber). On the one hand, some of the most persistent quirks of Berber grammar have been substantially simplified; on the other, even highly irregular core Berber morphology has been retained, and massive influence from Arabic – including the borrowing of productive root-pattern morphology – has generated new complexities. Based on part of my doctoral thesis but significantly expanded, this book:
  • proposes a classification of Siwi within Berber, and a corresponding probable account of where this Berber variety originated;
  • describes the grammar of Siwi, in greater detail than any previous work;
  • establishes how, and how much, long-term contact with Arabic has affected its grammar;
  • examines the dialectal affiliations of Arabic loans in Siwi, providing further evidence that this contact involved very different varieties at different periods;
  • provides a number of fully glossed Siwi texts of different genres, illustrating Siwi grammar and casting light on Siwi culture.

Thanks once again to everyone who helped in this process, and especially my friends in Siwa. To all those who find this sort of thing interesting, I hope the book comes in handy!

Friday, March 07, 2014

Korandjé tale (Conte en korandjé - قصة بلبالية)

In the early 1950s, the French anthropologist Dominique Champault made a number of sound recordings in Tabelbala. Champault’s recordings have recently been made available online by the Centre de Recherche en Musicologie, through Cécile Funke’s archival work. Many are in Arabic or French, but the Korandjé ones are an irreplaceable resource for the study of the language; in her time, the language was under rather less pressure and verbal arts were in much better health.

One of the easiest recordings for me – the sound quality is good, and the language simple – is a short folk tale about a cat, narrated by Zohra Adda (70-01):

Like “the house that Jack built”, this cumulative tale helps children learn to understand recursive causation. There are a few dialectal or idiolectal differences from the Korandjé I’ve heard, minor but striking to my ears. Following Marijn van Putten’s example, I’ll put it up here – comments very welcome! Etymology is marked by colour: yellow for Arabic, blue for Berber, and unmarked for Songhay.

عيحاجانيس عسكاتنيسي: – حاجيتك ما جيتك
ʕa-yħaža=ni.s ʕa-s-kka-t=ni.si:
1Sg-tell=2Sg.to 1Sg-not-come-hither=2Sg.to
I have told you, I haven't come to you (ie “Once upon a time”):
Je t’ai raconté, je ne suis pas venu à toi (c’est à dire “Il était une fois”) :
Comments: This is an alarmingly literal translation of the widespread North African “Once upon a time” formula, ħajit-ək ma jit-ək. This fixed formula is barely interpretable in Arabic, but one grammatically possible parse corresponds to the Korandjé here. iħaža, of course, is a Maghrebi Arabic loan, from ħaji “tell a story” (possibly via Berber).

1. إيشنّ احّلّق موشفُكدّا. – خلق الله قطا صغيرا
išann a-ħħəlləq muš=fʷ kadda.
God 3Sg-create cat=one small
God created a little cat.
Dieu a créé un petit chat.
Comments: išannu, historically a compound “our master”, has fallen into disfavour in modern Tabelbala; most speakers now prefer the dialect Arabic equivalent mula-na. ħəlləq “create” is from Maghrebi Arabic xləq, with an irregular shift x > ħ, probably the result of place dissimilation, paralleled in the Arabic of the Touat region (cf. Bachir Bouhania). Songhay, Berber, and local Arabic all have more or less the same word for “cat”, so it’s difficult to say which source Korandjé got the word from; provisionally, I assume it’s inherited.

2. آدر آبينبش. – ذهب يخدش
a-a-dər a-ab-inbəš.
3Sg-PF-go 3Sg-PROG-scratch
He went scratching (in the ground).
Il est allé gratter.
Comments: inbəš “scratch” is Maghrebi Arabic nbəš.

3. افُّ ابساتاكا اتّاس: توغ نبابتلاّ؟ – مر به أحد فقال له: عمّا تبحث؟
a-ffʷ a-bbsa-t-a.ka a-tts=a.s tsuɣ n-bạb-tsə̣llạ?
Nom-one 3Sg-pass-hither=3Sg.at 3Sg-say=3Sg.to what 2Sg-PROG-seek
Someone passed by him and told him: What are you looking for?
Quelqu’un est passé à côté de lui et lui a dit : Qu’est-ce que tu cherches ?
Comments: I’ve never heard any modern speaker pronounce “seek” as emphatic; the pronunciation I always heard was [tsɛlla] / [tsɨlla]. For the etymology of this Berber loan, cf. Zenaga pf. yə-llāh, impf. yə-ttälla(a)h “chercher” (Taine-Cheikh 2010); unusually, it seems to derive from the imperfective rather than the perfective.

4. ايتا عابتلاّ (ذ) إدرامن. – ها أنا أبحث عن النقود
əytsa ʕ-ab-tsə̣llạ (ḏ) idṛạmən.
lo 1Sg-PROG-seek (?) money
I'm looking for... money.
Je cherche... de l’argent.
Comments: There’s a clearly audible before “money”, but I can’t figure out a plausible reason for it. əytsa “lo!” is probably Berber, cp. Kabyle aṯan. idṛạmən “money” is a formally plural noun taken from Berber, ultimately from Arabic dirham (itself from Greek drachma).

5. ما نبغ إدرامن؟ – لماذا تريد النقود؟
n-bə̣ɣ idṛạmən
why 2Sg-want money
Why do you want money?
Pourquoi veux-tu de l’argent ?
Comments: “Want” is one of two quasi-verbs in Korandjé – the other is “exist” – that does not take mood/aspect morphology. I’ve heard maɣạ and mạʕạ for “why” > Berber ma-ɣər, but never just ma/mə as here. bə̣ɣ “want” looks suspiciously like Arabic bɣa, but actually it has a regular Songhay etymology, *baga.

6. عمذينذي فركا. – لأشتري بها حمارا
ʕə-mm- ə dzay=ndz.i fə̣ṛka.
1Sg-IRR- uh buy=with.3Pl
So I can buy a donkey with it.
Pour que j’en achète un âne.
Comments: Modern speakers usually have a slightly more reduced vowel in “buy” – [dzɛi], or even just [dzɨi]. Note the 3Pl, agreeing with idṛạmən.

7. ما نبغ فركا؟ – لماذا تريد حمارا؟
ma n-bə̣ɣ fə̣ṛka
why 2Sg-want donkey
why 2Sg-want donkey?
Why do you want a donkey?
Pourquoi veux-tu un âne ?

8. عمنڤّر لابو. – لأنقل الطين
ʕa-m-dza- ʕa-mm- ə- nəggə̣ṛ lạbu.
1Sg-IRR-bu- 1Sg-IRR- uh- transport clay
So I can bu- so I can- uh- transport clay.
Pour que j’ach- pour que je- euh - transporte de l’argile.
Comments: I’ve never heard “clay” with an emphatic vowel either; modern speakers usually say [læ:bu]. nəggə̣ṛ is presumably from Arabic naqala “transport”, but I need to double-check its meaning is also normally not emphatic now.

9. ما نبغ لابو؟ – لماذا تريد الطين؟
ma n-bə̣ɣ lạbu?
why 2Sg-want clay
Why do you want clay?
Pourquoi veux-tu de l’argile ?

10. عمكا آضّب. – لأصنع الطوب
ʕa-m-kạ ạḍḍə̣b.
1Sg-IRR-hit brick
So I can make bricks.
Pour que je fasse des briques.
Comments: kạ ạḍḍə̣b “make (lit. hit) bricks” is a fixed expression; kạ can be used in some other contexts to mean “work”. ạḍḍə̣b “brick” (pl. iḍḍụbən) is a Berber-style adaptation of Arabic al-ṭūb, which, transmitted via Spanish, also gives us English adobe.

11. ما نبغ آضّب؟ – لماذا تريد الطوب؟
ma n-bə̣ɣ ạddəb?
why 2Sg-want brick
Why do you want bricks?
Pourquoi veux-tu des briques ?

12. عمكيكي ڤا. – لأبني بيتا
ʕa-m-kikəy gạ.
1Sg-IRR-build house
So I can build a house.
Pour que je construise une maison.

13. ما نبغ ڤا؟ – لماذا تريد بيتا؟
n-bə̣ɣ gạ?
why 2Sg-want house
Why do you want a house?
Pourquoi veux-tu une maison ?

14. عمّيكنا محمّد نذا فاطنة إمّيـ – أصنعها، محمد وفاطمة يـ
ʕa-mm-ikn-a muħəmməd ndza fạṭna, i-mm-i-
1Sg-IRR-make-3Sg Muhammad and Fatima, 3Pl-IRR-h-
I’ll make it, and Muhammad and Fatima, they’ll- (Je la construirai, et Mohamed et Fatma, ils –)
Comments: Note that fạṭna “Fatima” shows place dissimilation, a regular process in many Atlas Berber varieties, suggesting that this version of the name reached Korandjé via Moroccan Berber rather than directly via Arabic.

15. عمذاكا محمد نذا فاطنة. – لأضع فيها محمدا وفاطمة
ʕa-m-dza=a.ka muħəmməd ndza fạṭna.
1Sg-IRR-put=3Sg.at Muhammad and Fatima
So I can put Muhammad and Fatima into it.
Pour que j’y mette Mohamed et Fatma.

16. ما نبغ محمد نذا فاطنة؟ – لماذا تريد محمدا وفاطمة؟
ma n-bə̣ɣ muħəmməd ndza fạṭna?
why 2Sg-want Muhammad and Fatima
Why do you want Muhammad and Fatima?
Pourquoi veux-tu Mohamed et Fatma ?

17. عمكيكيغيس تا- إمّيسرحغيس تاوالا. – لأبني لي قـ- ليسرحوا لي قطيعا
ʕa-m-kikəy=ɣəy.s ta- i-mm-isrəħ=ɣəy.s tsawala.
1Sg-IRR-build=1Sg.to fl- 3Pl-IRR-herd=1Sg.to flock
So I can build myself a fl- so they can herd for me a flock.
Pour que je me construise un tr– pour qu’ils me paissent un troupeau.
Comments: isrəħ “herd, graze” is Arabic srəħ. tsawala “flock (cared for by turns)” is Moroccan Berber, and is probably a later re-borrowing of the same word that yields Korandjé tsara “(a) time”.

18. ما نبغ تاوالا؟ – لماذا تريد قطيعا؟
n-bə̣ɣ tawala?
why 2Sg-want flock
Why do you want a flock?
Pourquoi veux-tu un troupeau ?

19. عمكواكا هوّا. – لأستخرج منه الحليب
ʕa-m-kaw=a.ka huwwa.
1Sg-IRR-remove=3Sg.at milk
So I can get milk from it.
Pour que j’en obtienne du lait.
Comments: As with “buy”, modern speakers usually have a rather more reduced vowel in “remove” – more like [kçəu].

20. ما نبغ هوّا؟ – لماذا تريد الحليب؟
ma n-bə̣ɣ huwwa?
why 2Sg-want milk
Why do you want milk?
Pourquoi veux-tu du lait ?

21. عمكواكا ڤي. – لأستخرج منها السمن
ʕa-m-kaw=a.ka gi.
1Sg-IRR-remove=3Sg.at ghee.
Pour que j’en obtienne du s’men.
Comments: By an amusing coincidence of sound and meaning, gi means more or less the same as English “ghee”.

22. ما نبغ ڤي؟ – لماذا تريد السمن؟
ma n-bə̣ɣ gi?
why 2Sg-want ghee
Why do you want ghee?
Pourquoi veux-tu du s’men ?

23. عمْيننذا رسول الله ن تالبّسْت(؟). – لأدهن به ؟؟ رسول الله
ʕa-m-yən=ndz.a ṛạsuləḷḷạh-n tsagʷḍḍə̣st[?].
1Sg-IRR-anoint=with.3Sg Messenger_of_God GEN lock
So that I can anoint with it the Messenger of God’s hair-lock.
Pour que j’en oigne le ?? de l’Envoyé du Dieu.
Comments: I can’t seem to make out that last word – the speaker tails off – but it seems to have the Berber feminine circumfix. ṛạsuləḷḷah is an Arabic compound, rasūl “messenger” and Allāh “God”. By the way, despite appearances, n is not Berber – given the associated word order, it can more plausibly be derived from an irregular shortening of Songhay wane (see Kossmann).

Korandjé is generally thought of as a contact-intensive language – so how mixed is this sample? Well, there are two ways to count (excluding, in any case, bound morphemes and incomplete words), depending on what we do with words that occur more than once. If we count by token, then we count the same word each time it appears; if we count by type, then we count the same word only once.

By token, we have 84 words: 52 Songhay, 12 Arabic, 20 Berber. So 62% of the text is Songhay, 14% Arabic, and 24% Berber.

By type, we have:
  • 24 Songhay words: ka “come”, išannu “God”, -fu “one”, kadda “small”, dri “go”, bsa “pass”, tsi “say”, tsuɣu “what”, bəɣ “want”, dzay “buy”, fəṛka “donkey”, lạbu “clay”, kạ “hit, work”, dza “put, do”, ndza “and, with”, kikəy “build”, kaw “remove”, huwwa “milk”, gi “ghee”, yən “anoint”, aɣəy “I”, ni “you”, ana “he/she/it”, ?muš “cat”
  • 8 Arabic loans: iħaža “tell (a story)”, ħəlləq “create”, yinbəš “scratch”, yisrəħ “herd”, ṛasuləḷḷạh “the Messenger of God”, muħəmməd “Muhammad”, fạṭna “Fatima”, nəggə̣ṛ “transport”
  • 8 Berber loans: idṛạmən “money”, ạḍḍə̣b “brick”, ikna “make”, tsawala “flock”, əytsa “lo”, tsə̣llạ “seek”, ma “why”, tsagʷḍḍə̣st “hair-lock”
So this rather repetitive text has a total vocabulary of only 40 words; 60% of its vocabulary is Songhay, 20% Arabic, and 20% Berber.