The subclassification of Songhay and its historical implications

Just about two years after its acceptance, my article The subclassification of Songhay and its historical implications has finally appeared, in Journal of African Languages and Linguistics 33:2. (If you don't have access to this journal, you can contact me for offprints.) The linguistic history of Songhay is not well-understood; in particular, although much ink has been spilled in speculation about its possible distant relationships, very little has been done to understand its internal classification and what that tells us about its recent history. This paper is an attempt to get to grips with the latter. The abstract is as follows:

This paper seeks to establish the first cladistic subgrouping of Songhay explicitly based on shared arbitrary innovations, a prerequisite both for distinguishing recent loans from valid extra-Songhay comparanda and for determining how Songhay spread. The results indicate that the Northern Songhay languages of the Sahara form a valid subfamily, even though no known historical records link Tabelbala to the others, and that Northern Songhay and Western Songhay (spoken around Timbuktu and Djenné) together form a valid subfamily, Northwestern Songhay. The speakers of Proto-Northern Songhay practised cultivation and permanent architecture, but were unfamiliar with date palms. Proto-Northwestern Songhay was already in contact with Berber and probably (perhaps indirectly) with Arabic, and was spoken along the Niger River. Proto-Songhay itself appears likely to have been in contact with Gur languages, confirming its relatively southerly location. This result is compatible with two scenarios for the northerly spread of Songhay. On Hypothesis A, Northern Songhay spread out from an oasis north-east of Gao, probably Tadmakkat or Takedda, and Northwestern Songhay had been spoken in areas west of Gao which now speak Eastern Songhay. On Hypothesis B, Northern Songhay spread out from the Timbuktu region, and Western Songhay derives from heavy “de-creolising” influence by Eastern Songhay on an originally Northern Songhay language. To choose between these hypotheses, further fieldwork will be required.

Actually, since writing that I've put together another paper that suggests a more specific explanation for the presence of Northern Songhay in Tabelbala – but that's still under review...

Nightmares in the desert

Sleep paralysis, more colourfully known as "old hag's syndrome", is a phenomenon that seems to have left traces in the folklore of just about every culture around the world; but it also seems to be commoner in some areas than others. In particular, while rare to the point of unfamiliarity in Britain or in my Algerian hometown of Dellys, it seems to be familiar to almost everybody in Tabelbala (and, similarly, within the US, it seems to be commoner among African-Americans than whites). The following Kwarandzyey text explains how it is experienced there, and as such might be of some medical interest:

ah, tsaddərts ndza askundzan ləxla aɣudzi. ndza aggwạ niš nn axnuq ka, e! asbạ uɣ bsəlləkni kʷəll, ləxla aɣudzi. nəmgwạ ɣaṛ nəmtqaqa. bəɣ sabmmə̣w niši. nəbʕəyyəṭ, nbəẓẓəgga ndza nən žžəhd... wara affu asmmə̣w niši ni ɣar nn haya si. nən kudzi, amgạ ttsən. attsən amgwạbzda ɣaṛ ndza bəssyas... nəbʕəyyəṭ, nəbẓəgga; wara affu issabmmə̣w niši. uɣu ibtsas tsaddərt. abgwạ niš adaɣ ka nn axnuq ka, amgwạ niši nən kəmbi ka ndza an tsiyu, amkạnika mʕad - ʕad nəmmiħəmda ṛəb si, həlla ʕad nəmfạktsi. xəd nəffạktsi nəmdza ufff! itsa adri.

Oh, sleep paralysis, if it gets hold of you, it's a terror. If it sits on you at your throat, he! there's no one to save you, it's a terror. You start just squawking. Nobody hears you. You're shouting, you're screaming with all your might... no one hears you, you're just on your own. Your blood runs slow. It's slow, it starts moving only slowly... You shout, you scream; no one hears you. This, they call "taddert". It sits on you here, at your throat, it sits on your hands with its feet, it hits at you until - once you praise God, only then do you wake up. When you wake up, you go "Phew!" It's gone.

Particularly if you speak Berber or Arabic, this text should pose an interesting challenge: how many words can you recognise? How much of it can you gloss?

Les déictiques en berbère oriental

Thanks to some filming at a recent conference, you can now listen to me butcher the French language while discussing the demonstrative systems of eastern Berber, in particular Siwi: Les déictiques en berbère oriental. It's not just Berber studies, although it has a good deal of Berber data. You see, it turns out that Siwi, like Qur'anic Arabic, has a typologically unusual feature called addressee agreement; so I attempt here to place this phenomenon within a wider typology of allocutivity, a phenomenon found in languages like Basque and Maithili too.

Gendered verbs: A thought experiment

Plenty of languages have gender for nouns. As far as I know, no language has gender for verbs. Obviously some languages have subject agreement in gender (Arabic, for one), but what I mean is a language in which some other word agrees with the verb it modifies or refers to. This is a bit odd, because - even if we assume that gender always starts out as a property of nouns, which is a pretty big assumption if you think about it - it's not too hard to imagine how verbs could develop gender.

Imagine a language that, like Japanese or Persian, expresses a lot of verbs using "do/make" plus a noun (English allows this in some contexts: "make a donation", "do a runner"). But, unlike Japanese or Persian, let's suppose this language has gender - like Kurmanji, say - and adjectives that agree with the noun in gender. In such a case, it would seem reasonable to have at least some adverbs expressed as adjectives agreeing with the verb's noun: "make frequent donation" for "donate frequently", say. Likewise, ellipsis could be handled with an appropriate pronoun: "I made a donation (masc.) to the library, and he made one (masc.) to a charity."

Now let the forces of phonetic erosion work on the verb phrase for a while, until the former support verb is reduced to a mere suffix attached to the stem (rather like what happened to Latin habere in the development of the future tense in Romance.) Unless the rest of the system has been reworked in the meantime for some reason, the result should be a language in which adverbs and pro-verbs agree in gender with verbs.

Is there any language that has done this? If not, why not?

Blida Atlas Tamazight

In the mountains above the town of Blida south of Algiers, there still survives an isolated variety of Kabyle, a remnant of the era when this region was Berber speaking. If you look through bibliographies for material on this, about the only thing that comes up is Laoust's (1912) Etude sur le dialecte berbère du Chenoua : comparé avec ceux des Beni-Menacer et des Beni-Salah, in which only the Beni-Salah material scattered here and there in the book comes from the region under discussion, while the rest deals with the varieties west of Algiers (almost as badly documented, but not closely related). But I recently came across a PDF that single-handedly changes this. Tamazight de l'Atlas blidéen (from Atlas de Blida) is a 57-page trilingual French-Berber-Arabic dictionary, followed by a sketch grammar, comments on toponymy, and a bibliography. The anonymous authors have done a fine job, and apparently are still working on it.

Leafing through it, I was struck by the extension of deg "in" to "from", replacing seg. While this is clearly not a Zenati variety, it seems to have picked up some Zenati characteristics, as you might expect from its westerly location: for instance, "one" is m. iǧ, f. ict. It's kept the word aryaḏ* "lion" (pl. iyraḏen); this explains the form recorded more than a thousand years earlier in Ibn Quraysh (with a slight copying error, 'ry'r), much better then the more widespread Berber form ar/ahar that Cohen (1972) suggested for it. There are some etymological puzzles to be examined - for example, why the prefixed q in iqic "horn", and why the shift d > l in laba "now"? The prefixed j in ijifer "wing" is presumably originally the indefinite article "one", but it occurs in the plural as well (ijufar). The double negative is ur...-k rather than ur... ara, a sort of non-Zenati version of the widespread Northern Zenati ur... -c (which I posted about a long time ago). No doubt plenty more remains to be seen. (* Typo corrected.)

LACITO move; Siwi text online

First, a big update I should have posted a month ago: I'm working as a researcher at the CNRS now, in Paris; the lab is LACITO. It's a great research atmosphere, and a much easier commute!

Second: Every time I pass through Alexandria, I try to photocopy as much of Gen. Rif'at al-Jawhari's book "Garden of the Sahara" (1964) as the Library of Alexandria will let me - which is generally not much, since they take copyright very seriously. This important - and hard to obtain - resource for Siwi is now viewable free online in its entirety: جنة الصحراء سيوه أو واحة آمون (Garden of the Desert: Siwa, or the Oasis of Amon). It includes a rather long vocabulary at the end (starting on p. 131) along with a few poems (starting on p. 208). The main text itself is of some interest for its description of Siwi customs, although the author's attitudes are sometimes a bit patronising.

Sacred Phonology

The artistically fruitful intersection between geometry and mysticism has a fairly well-established label: "sacred geometry". Phonology and mysticism have historically intersected in a similar, but perhaps less familiar, way.

The notion of "place of articulation" predates modern linguistics by millennia, as is obvious from the order of the Indic alphabets. In the Arabic context, while it was first developed by early linguists such as al-Farahidi and Sibawayh, it is probably most commonly studied in the context of Qur'anic recitation, where it provides a cross-check on the pronunciation of consonants. The number of places of articulation used is rather larger than in the Western tradition, allowing a more linear ordering of the consonants, as follows: chest (long vowels ā ī ū), lower throat (glottal ʔ h), mid throat (pharyngeal/epiglottal ʕ ħ), high throat (uvular fricatives x ɣ), back tongue (uvular stop q), velar (k), mid-tongue (palatal/postalveolar j š y), back lateral (ḍ), front lateral (l), apico-alveolar (n), front tongue (r), apico-dental (ṭ d t), sibilants (ṣ s z), interdentals (ð̣ θ ð), labiodentals (f), bilabials (b m w). (I have used odd terminology for some positions in an attempt to reflect divisions not usually made by Western linguists.)

This ordering of consonants by place of articulation, familiar to any religious specialist of the period, gave Ibn Arabi a structure onto which he could map his vision of the cosmic order (see Appendix II of the link). The throat is the seat of the intellective world, ie universal underlying principles; the back of the mouth is the higher realm of imagination; the mouth proper is the celestial spheres, followed in front by the elemental globes; finally, the "progeny", or classes of beings, are at the gap between the teeth and lips. In short, the more contingent something is, the higher up the vocal tract - just as sounds originate at its bottom with air expelled from the lungs, are shaped as they pass through the vocal tract, to finally emerge from the mouth.

The metaphor is reasonably effective as it stands; but its one-dimensionality is somewhat unsatisfactory. Most consonants reflect combinations of articulatory gestures, rather than being elementary movements. For instance, the difference between d and n, for instance, lies not primarily (if at all) in the place of articulation, but rather in whether or not air is allowed to flow out of the nose, and the difference between t and d lies in how the vocal chords are held. Wouldn't it be nicer to have a symbolism for consonants that allowed for compositionality? What would Ibn Arabi have done with Element Theory, for instance?

Arabic /ē/ gets colloquial: the case of al-Kisā'ī

My description of Khalaf's reading in the previous post applies to all readings transmitted from Ḥamzah ibn Ḥabīb al-Taymī of Kūfa: Khalaf, Khallād, Idrīs al-Ḥaddād, and Isħāq al-Warrāq. There is one other set of readings with final -ē, however: those transmitted from ʕAlī ibn Ḥamza al-Kisā'ī of Kūfa, through his students Abū al-Ḥārith and al-Dūrī. Here's an example (sūrat al-Shams, al-Dūrī's reading):

This set shows two interesting differences for the words examined before:

• Verbs with medial /ē/ in the Ḥamzah tradition simply have [ā]; contrast Ḥamzah's xēba خاب "he lost" with al-Kisā'ī's xāba. In other words, medially *aya and *awa both become ā, just like in the standard Classical pronunciation.
• Verbs with final /ā/ in the Ḥamzah tradition have /ē/, just like the ones with /ē/; contrast Ḥamzah's talāhā تلاها "it followed it" with al-Kisā'ī's talēhā. In other words, final *aya and *awa both become ē, whereas original *ā remains ā.

The latter development is phonetically quite counterintuitive - why would *awa become ē, when it didn't even contain any front vowel? But it makes more sense when you look at it on a morphogical rather than phonological level. Arabic has a huge number of final-y verbs, and a much smaller number of final-w verbs. In the rather common 3rd-person perfect forms, they are indistinguishable. This makes it tempting to simplify the system by reducing the differences between the two classes, and in fact practically all modern Arabic dialects have taken this to its logical conclusion and simply conjugate all final-w verbs as if they were final-y: thus Algerian Arabic, for instance, has dʕa دعا, dʕit دعيت, yədʕi يدعي instead of daʕā دعا, daʕawtu دعوت, yadʕū يدعو. What al-Kisā'ī is doing looks like an early step along that road.

You may notice that another characteristic of this reading is also distinctly reminiscent of certain modern colloquials, in particular those of Syria: prepausal feminine -ah ة is pronounced -ih.

/ē/, Arabic's fourth long vowel

(Warning: This post assumes some knowledge of Arabic, although you should be able to follow the argument even without it.)

Everyone knows that Classical Arabic has three short vowels (a, i, u) and three long (ā, ī, ū). But is this true of all varieties of Classical Arabic? Listen to this recitation of sūrat al-'Aʕlā:

There are several distinct Qur'ān recitation traditions, thought to reflect (in part) early dialectal variation in pronunciation. The best known are Ḥafṣ (Asia and Egypt) and Warsh (mainly North and West Africa); the recitation above is in one of the more obscure ones, Khalaf (ʕan Ḥamzah). In it, you will notice that words like šē'a “he willed” شاء, tansē “you forget” تنسى, appear with ē where more common pronunciations of Classical Arabic would use ā. But not all cases of ā are pronounced ē: contrast for instance ġuθā'-an “chaff” غثاء, lā “not” لا. Let's try to figure out what's going on here.

Start with the verbs ending in ā. Verbs which end in ā in the 3rd person masculine singular (“he did”), such as hadā “he guided” هدى, ṣallā “he prayed” صلى, daʕā “he invited” دعا, sajā “it covered with darkness” سجا, divide into two classes in other forms, one ending in y, the other in w: haday-ta “you guided” هديت, ṣallay-ta “you prayed” صليت vs. daʕaw-ta “you invite” دعوت, sajaw-ta “you covered in darkness” سجوت. You have just heard that the former set become hadē, ṣallē. For the latter, we will have to examine different sūras: in the 10th verse of sūrat al-Qamar (1:20) we hear daʕā, and in the 2nd of al-Ḍuħā we hear sajā.

Now ordinary three-letter verbs have the same stem throughout: katab-a “he wrote” كتب vs. katab-ta “you wrote” كتبت. What if the same used to be true of these verbs: *haday-a “he guided” vs. *daʕaw-a “he invited”? (The asterisk means that these are just hypothetical forms.) As it happens, that idea is confirmed if you look at one of Arabic's closest relatives. In Ge'ez, the Semitic classical language of Ethiopia, the cognate verbs are pronounced precisely as reconstructed: with aya (eg ṣallaya “he prayed”) and awa (eg ṣalawa “he roasted”, Arabic ṣalā صلا, ṣalaw-ta صلوت). So if we assume those forms were original, then we can easily see what's going on: in ordinary Classical Arabic both original *aya and *awa end up as ā at the end of a word, but in the Khalaf reading they remain distinct: *aya becomes ē, but *awa becomes ā.

A similar division can be made among verbs with medial ā. Verbs with medial ā in the 3rd person masculine singular, such as zāda “he increased” زاد, ħāqa “he surrounded” حاق, kāna “he was” كان, qāla “he said” قال, divide similarly into two classes in their verbal nouns, one in y, the other in w: zayd زيد, ħayq حيق vs. kawn كون, qawl قول. So we might expect a similar original difference: *zayada, ħayaqa vs. *kawana, *qawala. Sure enough, the pronunciation is as expected. Listen to sūrat al-Baqarah, verses 10 and 11 (about 2:00) and sūrat al-'Anʕām, verse 10 (about 3:00): zēda, ħēqa vs. kāna, qāla. A near-minimal pair is provided by sā'a “he was bad” ساء (sūrat al-Munāfiqūn, v. 2, about 0:50) vs. šē'a “he willed” شاء (already heard in sūrat al-'Aʕlā.)

So – depending on how abstract you are willing to make your representations – this variety of classical Arabic seems to have four long vowel phonemes rather than three. It is also unambiguously more conservative in this respect than the mainstream pronunciation reflected both in the Ḥafṣ reading and in educated standard Arabic, which underscores the philological value of such reading traditions.

(Note: The Qur'ānic Arabic Corpus was useful in preparing this post.)

"Inability to read or write in your mother tongue was a prerequisite for upward mobility"

If Mohammed Hanif's account of growing up in Pakistan below doesn't ring any bells, then congratulations: you're from one of the minority of countries worldwide with relatively low levels of diglossia. If you're Arab, you know exactly what he's talking about: substitute Darja/3ammiyya for Punjabi, Fusha for Urdu, and French/English as appropriate for English.

"When I was growing up in Pakistan, the complete inability to read or write in your mother tongue was a prerequisite for upward mobility... In my rural version of the state education system, the first thing they did was to try and save me from my mother tongue. Everyone spoke Punjabi in my household and like every five-year-old I had a vocabulary. I could name a goat, a donkey, a chicken. But since the medium of instruction in my school was Urdu, I had to learn alien names for familiar things. I must have spent the next 10 years learning in a language that I would be considered pretentious for speaking in my own street. By the time I finished high school, I realised that there was no college physics in Urdu, forget mathematics, and if you were destined to study aviation, you might have had to wait for centuries while someone drew up navigation maps in Urdu. So I began to learn English and by the time I drifted into writing I had no idea what my own language was. I was more like, “How much are you paying?”"

You might also think it's odd that there's no college maths in Urdu, given that there is such a thing in, for instance, Polish, a language with about a third as many speakers...

Talking Dictionaries (Celtic, Tuvan, Athapaskan, ...)

This link needs posting: Talking Dictionaries, which does what it says on the tin. Included are comparative Celtic (Breton, Welsh, Cornish, Irish, Scottish, and Manx), which should interest some readers; Tuvan (a Turkic language of Siberia with which Richard Feynman was obsessed); a couple of Munda languages; three Native American languages; and an Oceanic language of New Guinea.

Inalienable possession and social change

Many languages make some sort of distinction between "alienable" and "inalienable" possession. The exact factors vary a lot from language to language, but typically, inalienable possession is when the possessor inherently can't end his relationship to the possessed at all - for example, your father, your brain, your birthplace - or at least not without unacceptably drastic measures - for example, your tongue or your kidney. Alienable possession is when the relationship is more easily changeable - for example, your book, your room, your beef tongue.

In Berber, the roughly corresponding distinction is in general limited to inalienable relations mediated through your parents, ie those that have an important role to play in determining your own social identity: thus father, mother, brother, uncle, aunt, etc. use a different possessive construction than other nouns. For these words, the bare form means "my _" - eg Siwi aṃṃa "my brother". To say "X's brother", you have to add a 3rd person possessive marker first - thus aṃṃa-s n X (brother-his of X) "X's brother". In contrast, for other nouns, including people outside this category - eg amdarrəs "teacher" - there's no 3rd person possessor: "X's teacher" would just be amdarrəs n X.

As widespread as it is, the notion of inalienability rests on an experience of the world that, although it would be immediately recognisable to most people anywhere any time, has been under increasing pressure in a modern context. High mobility - geographical and occupational - makes even parents a rather less meaningful determiner of your identity and position in society, let alone uncles or aunts or birthplaces. For a person who has spent life far from most relatives and with no very strong ties to them, the saying "friends are the real family" has a resonance to it which would seem bizarre to most people throughout history. The body and the brain remain more or less inalienable for life - organ transplants notwithstanding - but a few half-mad futurists like Moravec dream of changing even that. The point of inalienable possessions is not just that they happen to be inescapable but that they define your identity in a way you can't control. The spirit of this age resists such impositions on one's freedom - just ask Dr. Phil. But as Chomsky points out, blank slates don't get anywhere, and "creativity presupposes fixed structure".

Siwi semantics

I recently arrived in Siwa – a pleasant experience as always, quite unaffected by the political turbulence of Cairo – and since getting here, I've repeatedly been learning the meanings of words I thought I already knew. Siwi has turart and adrar, whose cognates throughout Berber mean “hill” and “mountain” respectively. But a turart can hardly be called a hill – some are rock outcrops not much taller than a man – and the flat-topped layered “mountains” of Siwa that they call adrar would in English usually be considered hills, though the term can be used for larger ones too. ləbħaṛ can obviously mean “sea”, as in Arabic; but in fact, in a Siwi context it primarily refers not to the sea but to the two large lakes of the oasis. lašqəṛ is familiar from Arabic – Ibn Hazm notes, for example, that the Umayyad dynasty of al-Andalus were all blond ('ašqar), thanks to their seemingly heritable marital preferences – but it doesn't actually mean “blond” in Siwi, though that's an associated symptom; it means “albino” (albinism is fairly common in Siwa for some reason; it must be hard having no melanin in a place which hardly ever sees a cloud, but they seem to manage.) iləm is “skin”, as elsewhere in Berber; but the thick skin or hide of a sheep, which Siwis cook on special occasions, is not iləm, it's the Arabic loan əjjəld. Semantic elicitation is trickier than it might seem! Another etymologically interesting item of vocabulary I've learned is agbez “cowrie shell”, used in decoration. The word must be related somehow to Kwarandzyey (Korandje) tsyagmʷəš, but the correspondences are fairly funky. I wonder what it's called in Libyan Berber...

Berber in Libya and Egypt

I am glad to announce a new collaborative blog, in which I will be participating along with Marijn van Putten, Adam Benkato, and possibly others: Oriental Berber, on the Berber languages of Libya and Egypt. Not much there yet, but keep an eye out... The subject seems timely, with Berber having started to be used in Libyan media.

In other news, my Dardja etymology blog now features posts on the origins of zṛudiyya / sfənnariyya / xizzu (carrot), čina (orange), njəm (a kind of grass), jṛana (frog), and ʕətrus (billy-goat.)

Some updates: Darja etymologies, sub-Saharan loans, Libyco-Berber

Back again :)

I've often talked about why it's not enough for developing countries to use English or French as a working language for research and leave the majority of their own citizens in the dark. So I'm putting my money where my mouth is (so to speak) and starting a blog in Arabic focused on dialect etymology, a subject rife with popular misconceptions: الأصول التاريخية للدارجة الجزائرية (Historical Origins of the Algerian Dialect). Some of this blog's readers may be interested.

I've written up a finding first posted here - Songhay words in El-Jadida, Morocco - as part of a recently submitted article on sub-Saharan loanwords into North African Arabic. (There aren't many, but more than you might think: one of them, شطة šaṭṭa "Cayenne pepper" from Hausa cìttā, has even made it into Modern Standard Arabic via Egyptian dialect, and another, كابوية kābūya "pumpkin" from Hausa kàbēwā̀, is quite widespread in Algeria.)

MNAMON have posted a video of my talk about Libyco-Berber at Pisa - if you can stand the poor delivery, the content may be interesting. Among other things, I discuss the question of where LB fits into the Berber family tree.

Genetic and linguistic perspectives on Afroasiatic

Via GNXP, I hear there's been a new study on North African genetics: Genomic Ancestry of North Africans Supports Back-to-Africa Migrations. It provides an interesting cross-check on linguistic hypotheses.

In brief, the story these geneticists propose is: the main ancestors of modern North Africans, in particular Berbers, migrated into North Africa at least 12,000 and perhaps as much as 40,000 years ago; this "Maghrebi" component is close to Western Eurasian populations, and is dominant in most of their Moroccan and Algerian samples (and prominent in Libya). Arabs migrated in more recently starting 1,400 years ago, and Near Eastern influence is prominent throughout, especially in Libya, and dominant in Egypt. The Sub-Saharan African component seems to have arrived even later (~1,200 years ago in southern Morocco) and thus probably reflects the trans-Saharan slave trade; in Morocco it looks West African, while in Egypt it appears more diverse. Some European admixture is visible in Algeria and northern Morocco as well, but its nature is not clear. The data set is a bit small: a better coverage of Sahelian populations would be highly desirable, as would more Near Eastern populations, and one wonders where the ancient Egyptians fit in. However, the overall picture seems reasonable.

The more recent stages fit trivially with the detailed linguistic and historical data available (see my earlier post on linguistic traces of sub-Saharan immigration into North Africa), but the genetic divergence between Maghrebis and western Eurasian populations takes us into a realm where both fields offer much less certainty. Linguistically, we know that Berber, Semitic, and Egyptian are all distantly related to one another (and to Chadic and Cushitic, though that doesn't show up in the genetic data here); but we don't know when they split apart. There is no generally agreed upon method for dating linguistic divergences, and Swadesh's original "radioactive decay" glottochronological formula has proved too poor an approximation to be relied upon. However, a much-modified glottochronological formula was more recently proposed by Sergei Starostin in an attempt to fit a curve of attested data points. As it happens, two of his followers, George Starostin and Alexander Militarev, have ventured to offer estimates for Afroasiatic; for the split between Semitic and Berber, they respectively estimate 9,700 or 11,000 years ago. This seems strikingly close to the lower limit of the geneticists' estimate here. But even if this estimate is rejected, if the divergence date is anywhere near what the genetics is suggesting, then we have to conclude that genetic relationships older than 10,000 years can be discerned, contrary to some claims in the literature.

There is a way around this: one could propose a pre-Phoenician immigration that changed the language but had relatively little impact on the gene pool. In fact, such an event may have to be postulated for Afroasiatic's history in at least some areas anyway: speakers of one Chadic language are represented in this paper - Hausa - and their genes look nothing like North Africans or Near Easterners. However, it hardly seems like a parsimonious hypothesis in this case, given the split dates suggested. So... is this a corroboration of Starostin's method, or just a lucky guess?

Do you speak Tashelhit?

If you speak Tashelhit (or, indeed, another Berber language) and feel like helping me test a hypothesis, I would really appreciate it if you can send me a translation of these two slightly inane short stories (preferably by email). I'll be happy to explain later...

Abdallah had two sons, Brahim and Cherif. Brahim told Cherif: “I can run faster than you can.” Cherif told Brahim: “Let's have a race.” Brahim said: “All right.” They raced 100 metres. Cherif won. Cherif said: “You thought you could run faster than I can?” Brahim got angry. He told his father Abdallah: “Cherif is bad, he went running instead of doing his homework.” His father laughed and told Brahim: “Who was running with him?”

---

Khaled and Youcef were friends; they lived in a village. Khaled went on a long trip to the city with his son. A few days later Youcef went to the doctor, and the doctor told him: You need a medicine called such-and-such.” So Youcef called Khaled and asked him: “While you're in the city, can you get me this medicine?” Khaled was busy, so he told his son to find the medicine. His son went to the pharmacy; he said they didn't have it, but they might have it in another city nearby. So he went there and got the medicine. Then he came back and gave it to his father. When they got back his father gave it to Y.

Multilayered meanings in Dellys

In my hometown, the small port of Dellys in north-central Algeria (and probably elsewhere for all I know), older women traditionally throw water after a family member who is departing for a long voyage. When I asked my oldest aunt (who speaks no language but Arabic) about this recently, she said it was to bring them safety – Arabic 'amān – on the road. The Berber word for “water” is, precisely, aman. The action reveals itself as originally a pun rather than a mere superstition – but one that only makes sense in the light of both languages at once, not Arabic or Berber alone. A useful case to bear in mind in trying to understand North African culture...

When Emir Abdelkader came to Dellys in 1840 and inquired about its defences against the French (who would occupy it four years later), he was allegedly told that the town places its trust in its saints: Sidi Abdelkader by sea, Sidi Soussan by land. In the account of Bennaamane (2011:61, citing Daumas and Fabar 1847:197), Emir Abdelkader reacted in a very modern way: he got angry at their superstition and pointed out that Algiers had not been saved by its "patron saint" Sidi Abderrahmane. But somewhere along the transmission of this account, a bit of metonymy has been misunderstood. The tomb of the supposed saint Sidi Abdelkader was located at the tip of a 700-metre-long peninsula next to the town, from which you can see any incoming ship for at least 20 km (map). That of Sidi Soussan was located at the top of the hill on whose side Dellys stands, and was such an obvious location for defenses that the French turned it into a blockhaus soon after. The speaker was using religious language, but his trust was as much in the scouts posted there as in the saints buried there.

Dellys (dəlləs, medieval Tadallas), owes its name to a common plant used in net-making and thatching (Ampelodesmos mauretanicus), locally called dalis (better known in Algeria as dis.) The name is not attested before the 11th century, and does not resemble its earlier Latin name (Rusuccurium, from Phoenician rus “head, cape”.) However, Murcía (2011) points out that the plant name is a good deal more ancient: a 5th-century work, Ars sancti Augustini pro fratrum mediocritate breviata, states that non-Latin regional words are barbarous, ut si quis dicat in latino sermone dellas pro carice, quod utique punicum est (“like if someone says in Latin dellas, which is undoubtedly Punic, in place of carex (sedge)”). Murcía reasonably takes the word to be Berber rather than Punic in origin: as he points out, forms similar to adlis for this plant are found all across northern Berber. But as Bennaamane (2011:22) points out, there is a comparable classical Arabic form in Lisān al-`Arab – dalas “the remains of plants and vegetables; land that bears plants after having been barren; plant that leafs after late summer” – so this could be an old Semitic loan into Berber too; more extensive comparative work is called for.

References
بن نعمان، اسماعيل. 2011. مدينة دلس (تدلس) : دراسة تاريخية وأثرية خلال العهد الإسلامي. تيزي وزو: دار الأمل للطباغة والنشر والتوزيع.
Daumas, M. et M. Fabar. 1847. La Grande Kabylie : études historiques. Paris: Hachette.
Murcía, Carles. 2011. Que sait-on de la langue des Maures? Distribution géographique et situation sociolinguistique des langues en Afrique Proconsulaire. In C. Ruiz Darasse et E. R. Luján (éd.) Contacts linguistiques dans l’Occident méditerranéen antique. Madrid: Collection de la Casa de Velázquez (126), pp. 103-126.