Loanwords examined via Pozdniakov's Proto-Fula-Sereer

I recently finished Pozdniakov's Proto-Fula-Sereer, freely available through Language Science Press. This is obviously a very welcome and valuable contribution to West African historical linguistics, an area where much remains to be done. I have little experience of Atlantic languages as such, and therefore not much useful to say about most of the book (though it made me want to also read Merrill's work, with which much of it is in dialogue.) However, while proto-Fula-Sereer is dated by the author to 2000 years ago or more, some of the comparisons are relevant for studying contact with other regional families. Two forms are particularly interesting to me for exploring contact with Berber:

• *xiris "slay (vb)": Sereer xiris 'couper le cou, décapiter, égorger' (Merrill: 'slit the throat') ~ Fula hirsa 'égorger; sacrifier (un animal, pour en rendre licite la consommation)' [p. 63]
  Sereer x- : Fula h- is a very well represented regular correspondence; however, Fula -r- in -rC- would normally be lost in Sereer (p. 173), and no regular pattern of vowel elision is given in the book. The word also looks like Soninke xùrùsi "to kill by cutting the jugular vein", yet the vowel correspondence is difficult there as well. The explanation is to be found in their common source as a loanword from widespread (non-Zenaga!) Berber əɣrəs, with the same meaning. The religious importance of slaughtering an animal for meat in this precise manner is sufficient to motivate the borrowing, which would thus have spread with Islam - presumably through northern Saharan travellers rather than Zenaga scholars, given the form.
• *Guf "foam": Sereer kuf 'gonfler, écumer en bouillant', kuf a...al / kuf a... ak 'écume de la mer, à la marée montante' ~ Fula ngufo / (n)gufooji 'mousse, écume' (cf. Fula ƴufa 'mousser, écumer (trans.)', ƴufo 'mousse, écume) (Laala kuuɓ 'mousse', Nyun Gubaher gʊ-gʊfʊri 'mousse', Nyun Guñamolo tɪ-gʊf / tɪ-gʊf-ɔŋ 'écume, mousse', Joola Fonyi ka-gʊf 'bave, écume de mer, mousse du savon'). [p. 102]
  The correspondence of Sereer k to Fula ŋg (let alone ƴ) is completely irregular, with no other examples cited. A comparison to Berber forms such as Tamasheq tə-kuffe, Tamazight a-kuffi, Zenaga tu-ʔffukkaʔ-n "froth" is thus not ruled out, although the other Atlantic forms make it more likely that the resemblance is coincidental. Cp. also Zarma kùfú "écumer" and related forms in Songhay, which probably do derive from Berber.

Other forms are interesting to examine in the context of Songhay and Mande:

• *bon "bad (svb)": Sereer bon 'être mauvais, être méchant, être maigre', ponu l / ponu k "le mal [la chose mauvaise]' ~ Fula bona 'être mauvais, être mal; être méchant', mbonki / bonkiji 'méchanceté ; malfaisance ; perversité' (widespread root in Atlantic and Mel) [p. 86]
  Also widespread well beyond; looks originally Atlantic, but the suffixed vowel in Bambara bɔ̀nɛ and Zarma bòné betrays a borrowing path via Soninke rather than directly from Fula.
• *bul "blue (svb)": Sereer bule 'bleu' ~ Fula bula 'rincer au bleu (du linge blanc); passer au bleu de lessive; colorer en bleu pâle' (The root *bulu is common for Atlantic and Mel languages. It is not a European borrowing). [p. 86]
  If so, then this is also the source for Bambara búla and Zarma búlà "blue", and other forms across the region. But this is a widespread Wanderwort, and one wonders how a European source was ruled out.
• *mbedd "road, path": Sereer mbed o...ong/ped k 'petit chemin laissé entre deux champs à l'hivernage, ruelle, rue, allée" ~ Fula mbedda / mbeddaaji 'grand route' (Wolof mbedd 'rue', Jaad mbɛdɛ 'grand route'; Manjaku umbɛra 'chemin carrossable, route'). May be an ancient Soninke borrowing: < béddè 'rue principale, route'. [p. 87]
  Gao Songhay has albedda / mbedda, with an interesting prefix alternation; Heath very tentatively suggests a link to Arabic blṭ, but that probably doesn't work.
• *Birq (mb-/w-) "manure": Sereer mbiqi n 'fumier, tas de fumier' ~ Fula wirga 'labourer le sol en éparpillant la terre (en luttant au sol ou pour la mélanger ou encore pour brouiller des traces...); disperser du fumier (sur un champ)' [p. 88]
  The correspondence mb:w is not regular, arguably reflecting differences in consonant mutation; only four examples are found, although they look like plausible retentions. The loss of r in Sereer would be regular (p. 173). The correspondence of q to g does not appear regular either (p. 192), unless this is related to the preceding r; one would expect q:kk. It's just as well that the correspondence is irregular, since the Fula term is clearly at least in part a borrowing from Songhay, not vice versa: it reflects a merger of two tonally distinct verbs, found in Zarma as bírjí "fumer le sol; fumier" and bìrjí "mélanger, embrouiller", used in the expression laabu birji "mélanger la terre". Conceivably the "spread manure" sense could be original to Fula, with only the "mix" sense being borrowed; but it strains credulity to imagine Zarma borrowing the same verb but giving it two different tonal patterns depending on the intended meaning. Soninke boroko "manure" is suspiciously similar, but the vowels rule it out as an intermediary.
• *gaw "hunt (vb); throw (vb)": Sereer xaƴ 'lancer, envoyer un projectile, tirer une arme à feu; lancer un dard, pêcher au harpon', nGawlax n / qawlax k ~ nGaƴlax n / qaƴlax k 'la chasse [gibier]' ~ Fula gawoo 'chasser, être chasseur (professionel)'. [p. 111; poorly justified correspondences - 5 words for x:g]
  The Fula term is certainly the same root as (Songhay) Zarma găw "hunter", gáwáy "hunt (v.)". The term doesn't seem to be used in Mande, from a quick look. If the Sereer form is related to the Fula one, then the direction of borrowing must be Fula to Songhay. However, the correspondence looks rather poorly justified. For x-:g-, only 5 correspondances are given, including such eminently borrowable words as "indigo" and "okra". For -ƴ:-w, the expected regular correspondence is rather ƴ:ƴ (p. 192), cf. "limp" (p. 180), "lick" (p. 174). The question of borrowing direction thus remains open.

The following cases may be only coincidentally similar, but perhaps they reflect contact at a much earlier period in prehistory, related to the spread of the practice of milking:

• *Gang "chest": Sereer ngang n / kang k ~ Fula gannde / ganndeeje (Fula < gang-nde?) [p. 103; irregular initial correspondence with only two other examples found)
  Cp. Zarma gàndè "chest".
• *gand "nipple": Sereer hand 'être pleine (femelle), être en gestation, porter [femelle]', hand l / qand a...ak 'mamelle (des animaux), pis', and l / and a...ak 'mamelle (des animaux), pis, téton, tétine' (to note a variety of Sereer forms: h-,q-,Ø-) ~ Fula ʔenndu ~ ʔenɗi 'sein, mamelle; pis, trayon'
  Cp. Zarma gánì "udder".

The Fulani abstract noun formative -(aa)ku is analysed (p. 231) as an "extension suffix" -aa- plus a class suffix -ku explained as a taboo-motivated allomorph of -ngu, citing Koval 2000:230 (a source in Russian). This requires further investigation; it certainly cannot be unrelated to Soninke -aaxu with the same function, but what was the direction of borrowing?

Efforts to exclude Arabic loanwords were largely successful, but even so, one crept in: Fula waabiliire "pluie d'orage" is from Arabic waabil rather than proto-Fula-Sereer *(b)waam/b (p. 79). On the other hand, Sereer tuɓaaɓ and Fula tuubako "European, white man" are derived from nonexistent Arabic *tubaab (pp. 115-116), following a long if poorly evidenced tradition connecting this to the real Arabic word ṭabiib "doctor".

Being "upon the truth"

It's not too hard to think of words that are characteristically used in English almost exclusively by Muslims - salat, namaz, wudu, shahada, masjid... There are even a few such words that aren't borrowings from Arabic or Urdu: circumambulation comes to mind. It is much more difficult, at least for me, to think of characteristics of "Islamic English" that go beyond the lexicon.

I was recently struck, however, by the expression "upon the truth". Searching for "upon the truth" yields plenty of mainstream English examples like "hit upon the truth", "lay hold upon the truth", "an essay upon the truth of the Christian religion"... However, searching for "be upon the truth", "are upon the truth", "is upon the truth", etc. yields a very different picture. Suddenly almost every single search result is specifically Islamic:

• "a) Hindering from the path of Allah, b) and confusing the person into believing that he is upon the truth" (Anonymous, Mission Islam
• "Either all of them are upon the truth which is impossible since truth is not open to contradictory differences or one of them is upon the truth" (Shaykh Haytham Al-Haddad in Muslim Matters)
• "This is because ibn 'Arabee held that all pagans and idol-worshippers were upon the truth since Allah is in his view everything" (translated from Shaykh Muhammad Ibn Rabee Ibn Haadee Al-Madkhalee's The Reality of Sufism in Light of the Qur’aan and Sunnah)
• "who do I follow among all those who claim to be upon the truth?" (Zahra Anjum in the International Islamic Publishing House blog)

You get the idea. The rare exceptions, like "their ultimate dependence is upon the truth", reflect quite a different construction, as the inanimate subject shows. In English, referring to people or groups being "upon the truth" appears to be unique to Islamic discourse (perhaps even to some genres thereof; most of the hits seem to have a vaguely Salafi vibe).

While this construction uses only well-known English vocabulary, it literally translates the Arabic expression على الحق ʕalā l-ḥaqq "on the truth/right". Within Arabic, this expression has a bit of an archaic ring to it, but is familiar from a number of hadith, e.g:

فَجَاءَ عُمَرُ فَقَالَ أَلَسْنَا عَلَى الْحَقِّ وَهُمْ عَلَى الْبَاطِلِ
At that time `Umar came (to the Prophet) and said, "Aren't we on the right (path) and they (pagans) in the wrong?" (Bukhari 65.365)

Being "upon the truth" is thus a calque into Islamic English from Arabic. No doubt a wider investigation would reveal other such cases.

Two Bambara words in Gnawa songs of Meknes

Across North Africa, small groups dominated by descendants of slaves brought from the Sahel preserve musical traditions, with ritual and medical functions, usually called Gnawa in Morocco, Diwan in Algeria, and Stambeli in Tunisia. Aguadé's Die Lieder der Gnawa aus Meknes provides the lyrics of an extensive corpus of Gnawa songs from Meknes in northern Morocco. These songs are primarily in Arabic, but characteristically include a number of words with no plausible Arabic or Berber source, presumed to derive from languages of the Sahel. Their identification, however, is generally difficult, although Aguadé ventures a few suggestions drawn from Hausa. Anyone can comb dictionaries for sound-alikes, but similar forms may be found across unrelated languages of the Sahel with very different meanings. It would be much easier if the meanings were certain, but the singers do not necessarily know the meaning of such words, and the context often hardly narrows it down. Nevertheless, some cases can be identified more confidently than others.

Aguadé's song number 88, Lalla l-Batul "Lady Virgin" (pp. 128-129), is dedicated to a female genie whose song cycle corresponds to the colour yellow. Its refrain (accounting for 5 out of its 8 lines) is a lalla l-batul, saysay "Oh Lady Virgin, saysay". The word saysay has no meaning in Arabic or in Berber. In Bambara, however, sáyi means "yellow"; the refrain would then be "Oh Lady Virgin, yellow, yellow".

In his song number 90 (pp. 130-132), the refrain is fufu dənba ya sidi "fufu dənba, oh master" (repeated 14 times, including the opening line of the song). Bambara dénba means "mother". The first verse after the initial refrain is ma bɣatək kda ya sidi "she didn't want you like that, oh master"; no feminine singular subject to which this could refer appears anywhere in the Arabic text of the song, but the Bambara interpretation allows this line to be better understood. I'd like to relate the preceding fufu to Bambara fò "greet" and/or fɔ́ "say, speak" - "greet Mother" would seem contextually appropriate - but I can't quite see how the grammar would hang together.

Addendum: In song 5, Sidi Gangafu "Mr. Gangafu", almost every couplet ends in Bambaṛa or shortened ya Mbaṛa, so a Bambara etymology seems worth considering (although an allusion to Hausa is also found). As Aguadé notes, Ganga is simply a kind of drum used by the Gnawa, whose name is shared across most of the Sahel, so one would expect this name to mean something like "drum-player" or "drum-maker". In fact, Gangafu can readily be interpreted as Bambara gàngan-fɔ̀ "play the ganga-drum". "Drum-player" should properly be something like gàngan-fɔ̀-la, but it doesn't seem like much of a stretch to suppose that the Bambara used by slaves among themselves would have had some non-standard features, given that for many of them it would have been a second language to begin with.

The *Bugzu of Bagzan?

Mt. Băgzăn, at the heart of the predominantly Tuareg-speaking Air massif in Niger, bears a not very Tuareg-looking name. The only Berber meaning for the root BGZ found in Nait-Zerrad is a word used by the neighbouring Iwellemmedan, taken from Alojaly's dictionary: ebăgez, pl. ibəgzan "vessel for dogs or for rubbish"; this corresponds regularly to Tahaggart ebăǵăh, pl. ibəǵhan "crude vase or plate (used for giving dogs their food and for gathering rubbish)", with a feminine tebăǵăht, pl. tibəǵhin "flat, slightly concave instrument used as a dustpan" (Foucauld). Not a root one would want to reconstruct very far back in Berber, nor an obvious source for the name of a mountain.

Hausa provides a surely related form that may shed light on the term's history: the ethnonym būzu < *bugzu (by Klingenheben's Law, as shown by the pl. bugā̀jē) "serf of the Azben [Air] people" (Bargery). The term refers to ex-slaves, iklan, what in Mali would be called Bella. It presumably does not share an etymology with būzu pl. būzā̀yē "undressed skin mat, loin-cloth", with no *g, for which Skinner (1996) gathers plausible cognates elsewhere in Chadic.

Combining the two, we get what looks like a brief glimpse of morphology: the homeland of the *Bugzu is *Bagzan (perhaps their manufactures included crude plates). From a Tuareg perspective, -ăn looks like a masculine plural ending; but the specific vowel alternation would be hard to explain Tuareg-internally, though Tuareg has a-ablaut in other plural types. From a Chadic perspective, one is reminded of the -n plurals of Bade and Ngizim, e.g. Bade zawa-n pl. zawa-n-ən "stick" (Schuh ms), Ngizim gâzbə́r̃ pl. gázbàarín "tall" (Schuh ms, 1972); Ngizim even offers parallels for the vowel alternation, and a-ablaut plurals are widespread in Chadic more generally. The Bade-Ngizim subgroup includes geographically the closest Chadic varieties spoken to the Air besides Hausa, located almost due south of the Air, so it seems a promising point of comparison; could the *Bugzu have spoken a since lost West Chadic B.1 language? But of course, nothing guarantees that Bagzan should be an old plural; perhaps -ăn was a locative suffix, or something else entirely.

I wouldn't be surprised if some early 20th century work proposes this connection, but I haven't come across it in the literature so far; if you have, let me know.

Clitic doubling in Arabia: An update

Back in 2017, I published an article on "Clitic Doubling and Contact in Arabic" (ZAL 66, pp. 45-70), arguing that the various cases of clitic doubling reported across Arabic dialects in different regions - NW Africa, Malta, the Levant, Cyprus, Central Asia, Dhofar - differ in their behaviour, do not share a common origin, and in each case reflect substratum influence. The case of Dhofar turned out to be particularly tricky in that the only available evidence for clitic doubling in local Arabic and in its Modern South Arabian substratum came from the same speaker in each case - Mhammed bin Selim El-Kathiri, a bilingual speaker of Jibbali and Dhofari Arabic who worked with a team of Austrian linguists about a century ago. He used the same clitic doubling construction across both his languages (definite DO/IO/PrepO, no marker); but no such construction appears in more recent work on either language. I tentatively concluded that:

Only further data can determine whether this is a general feature of some particular Dhofari dialect (perhaps the second language dialect of Arabic spoken by Shihri speakers?) or just an unusual feature of El-Kathiri's idiolect. However, if this construction was not simply idiolectal, its origins seem more likely to lie in Jibbali than in Dhofari Arabic, since no parallels have been found in any Arabic dialect of the Arabian Peninsula.

A forthcoming article I recently came across, Pronominalization and Clitic Doubling in Syrian and Omani Arabic, changes the picture for this region. In a paper primarily focused on the generative syntax of clitic doubling rather than on its history, Peter Hallman and Rashid Al-Balushi demonstrate for the first time that the Arabic variety of al-Batinah in the north of Oman has productive clitic doubling, and that its distribution (definite/specific DO/IO/PrepO/Poss, no marker) largely matches El-Kathiri's usage a century earlier. Clitic doubling of this type thus a widespread Omani feature, not a Dhofar-specific one, and certainly not a merely idiolectal one.

Note that the dialect of Al-Batinah, like that of Dhofar, is a dialect with q for historic qāf, representing the earliest stratum of Arabic to reach the region. One hypothesis could be that clitic doubling of this type is a Modern South Arabian (MSA) substratum feature, calqued into the first Arabic varieties to reach Oman but never reaching the g-dialects that first come to mind when one thinks of Arabian dialects. On the other hand, no further evidence has yet come to light for clitic doubling in MSA; based purely on the available data, it seems equally or more plausible that this type of clitic doubling arose spontaneously in Omani Arabic and was calqued into Jibbali by bilinguals such as El-Kathiri. Much more dialectological data is needed to decide the question; available descriptions are evidently far from complete. In either case, independent origin appears far likelier than any kind of historic connection with the rather different types of clitic doubling observed in other parts of the Arabic-speaking world.

Mubi plurals from Arabic

Mubi, an East Chadic language spoken in the Guera Mountains of eastern Chad, stands out even in Chadic for the sheer complexity of its plural system, and all the more so for its extensive use of internal vowel changes. This seems likely to give it particular relevance for the reconstruction of Afroasiatic. However, Mubi is also profoundly influenced by Chadian Arabic, to a greater degree than even an Arabic-speaker might suppose at first sight. How much of Mubi's plural system reflects Arabic influence?

Looking through Jungraithmayr's (2013) La Langue Mubi^review, I find 14 plurals of the form BaCaaDi(i)F (e.g. àbàlány "patas monkey", pl. àbàalîny) and 5 of the form BaCaaDo/u (e.g. móngò "monkey sp." > mánáagò). Of these, 9 and 4 respectively are found in Jullien de Pommerol's (1999) Dictionnaire arabe tchadien-français (including, to my surprise, both the previous examples), and many of the remainder seem semantically likely to be features of some more localized Arabic variety (e.g. mánjàk "village chief" pl. mànáajìk, various ethnonyms). It seems rather clear that these two plural types are borrowed from Arabic; but there is no strong evidence that they have been extended to inherited vocabulary.

For the closely related plural form BaCaaDiFe, we find 3 examples, of which only one is definitely of Arabic origin: shàddáarì "shaman" (i.e. "herbalist", based on Chadian Arabic šadar < šajar), pl. shàdáadìrè. Suffixation of -e is not otherwise typical of Mubi plurals, and matches perfectly with the Arabic plural in BaCāDiF-ah; it thus seems reasonable to consider this plural type as a borrowing from Arabic as well. If so, it provides us with one good candidate for an extension to inherited vocabulary. Mubi érìny "scorpion", pl. àráarínyè is comparable to other East Chadic forms in its singular, eg Dangaleat ɛ́rîndílɛ̀ pl. ɛ́ríndílnà, Toram irindeeɗà pl. irindeɗ, Kajakse ʔàràari pl. ʔàràaràk (Fedry 1971, Alio 2004), but disagrees strikingly with them in its plural.

At first sight, one is tempted to go further and conclude that the plural types BuCooDuF and BiCeeDiF are also adaptations of the Arabic iambic plural. But the evidence in those cases is not so clearcut. BiCeeDiF is only attested for a single word with no Arabic counterpart that I've been able to find (dólgúm "a type of basket", pl. díléegìm.) BuCooDuF is far more frequent than BaCaaDi(i)F and seems to contain a much greater proportion of inherited vocabulary, although some Chadian Arabic loans are found as well (e.g. àngúmbùl "calabash", pl. àngùnóobùl, corresponding to Chadian Arabic amgunbul pl. amganâbil). Moreover, it can plausibly be unified with another plural schema with no possible Arabic counterpart, BuCoDFuG, e.g. áránjálà "kidney", pl. ùrònjúl. For the time being, it seems prudent to withhold judgement on the explanation for why these two plural types are so strikingly reminiscent of the Arabic iambic plural.

Other Arabic plurals borrowed only for the corresponding Arabic nouns include BuCuuD (e.g. tês "billy-goat", pl. túyúùs), BiCiDaan (e.g. jédì "dorcas fawn", pl. jídíyáàn), and the sound masculine plural suffix -iin (e.g. máanì "strong", pl. màanìʔíìn). In the case of the sound feminine plural suffix -a(a)t, only two of the four examples in Jungraithmayr are clearly of Arabic origin (àntàháarà "mantis", pl. àntàhàarât; ràbàʔíyè "young woman", pl. ràbàʔìyáàt.) The other two look suggestively Arabic, however (ìrèedíyè "small granary", pl. ìrèedìyât; ròomìyè "crushing-stone", pl. ròomìyáàt), and this plural type too probably consists entirely of Arabic loans.

So far, it looks like in Mubi, as in Berber, Arabic influence has had the effect of further complicating an already very complex plural system. But Mubi is spoken in a far more multilingual context than most Berber varieties; one wonders whether some of the complexity here might be due to contact with regional languages other than Arabic as well...

Kabyle-Arabic phraseological convergence

Kabyle Berber and (especially north-central) Algerian Arabic show the marks of massive convergence, often reflected in the use of phraseology that translates literally with identical meaning. A nice example I came across recently (p. 186, Poèmes kabyles anciens, Mammeri 1980/2009) is the following sentence:

Ur as zmireɣ, ur iffiɣ felli.
NEG 3SG.DAT be.able.NEG.PFV-1SG, NEG 3MSG-go.out.NEG.PFV on-1SG
Literally:
I can't handle it, it doesn't suit me.

This translates perfectly into (north-central) Algerian Arabic:

ما قدرتلو، ما خرج عليّا.
Ma qdertlu, ma xrej 3liyya.
NEG be.able.PFV-1SG-3SG.DAT, NEG 3MSG-go.out.PFV on-1SG
Literally: I can't to it, it doesn't go out on me.
I can't handle it, it doesn't suit me.

In languages further removed from the area, however, a literal translation would be comically nonsensical:

• EN: *I can't to it, it doesn't go out on me.
• FR: *Je ne lui peux, il ne sort pas sur moi.

Another case was highlighted on Twitter by Noureddine Chikh: the use in both languages of "where do I know?" for "how would I know?" The latter proved to have some near-parallels elsewhere (with "Whence do I know?"), but no perfect ones were reported. How about this one? Can you think of any other idiomatic phrases that translate literally across the two languages?

Getting lost in the NW Sahara

Two languages of the northwestern Sahara, spoken reasonably close to each other, have basic motion verbs derived from a word that originally meant GET LOST. Let's see if we can figure out how that happened.

For COME, practically all Berber languages consistently use reflexes of the proto-Berber word *asəʔ. In the largest Berber variety, however - Tashelhiyt, in southern Morocco - this root has been lost, and a quite different verb is used: ašk (ⴰⵛⴽ). The original meaning of this verb can still be seen in other Berber languages, such as Tamasheq: GET LOST (a meaning which in Tashelhiyt has been replaced by what's probably a borrowing from Arabic جلا.) Presumably, GET LOST came to mean WANDER, and WANDER (over) came to mean COME.

In Songhay, GET LOST is *dere(y), preserved as such in most varieties. In Korandje in western Algeria, however - uniquely within the family - this root's reflex has undergone a very similar shift in meaning: dri now means GO. (Songhay speakers might assume this comes from dira WALK, but this word, from Proto-Songhay *dida, rather corresponds to Korandje zda WALK.) Meanwhile, Berber *aškəʔ GET LOST has itself been borrowed - probably from Tamasheq - as wuška GET LOST (the vowels reflect the Berber perfective form.)

In summary:

	COME	GET LOST	GO
Tashelhiyt	ašk	žlu	ddu
Tamasheq	as	ašk	ăkk
Gao Songhay	kaa	dere	koy
Korandje	ka	wuška	dri

Both changes can be summarized as GET LOST > BASIC-MOTION-VERB. Lexically, Korandje shows heavy influence from southern Moroccan Berber, much of which seems to match Tashelhiyt better than it does the Southern Tamazight varieties currently spoken closest to Tabelbala. That makes it rather tempting to seek a contact explanation. But if Korandje was copying a Tashelhiyt pattern, why would it replace GO rather than COME?

To make sense of what happened, I think we have to envision an intermediate earlier stage where WANDER (from GET LOST) was getting used as a generic verb of motion irrespective of direction in some (perhaps expressive) contexts. Both Tashelhiyt and Korandje require direction towards (and sometimes away from) the speaker to be expressed with a directional morpheme outside the verb root proper, so no ambiguity would necessarily result. From this situation, WANDER could end up replacing either COME or GO, while still maintaining the existing (seemingly superfluous) lexical distinction between the two by keeping the other root.

Now I think about it, British English offers a possible parallel for the initial stages of such a development, with particles substituting for the directionals of Berber and Songhay. In phrases like "he wandered over" ("he came over"), "he wandered off" ("he went away"), the original implication of aimlessness has faded away in informal usage to the point of being virtually absent. Should we expect some peripheral English dialect to replace "come" or "go" with "wander" altogether? Check back in a few centuries to find out...

On reading Poplack 2018

It was a frustrating experience reading Poplack's Borrowing: Loanwords in the Speech Community and in the Grammar. On the one hand, it’s intelligent, well-written, and packed with a wealth of precious sociolinguistic data on borrowing and to a lesser extent code-switching; on the other hand, it appears to be largely dedicated to hammering home a definition of the former that appears to me to be fundamentally untenable. The author ably demonstrates that three criteria that one might expect to be closely correlated are not: conventionalization, morphosyntactic integration, and phonological integration are all independent of one another. Of these three, she chooses to define borrowing exclusively in terms of morphosyntactic integration. For her (enormous, but not very numerous) preferred corpora, this apparently works just fine. But...

The notion of “borrowing” emerged from diachronic studies of the vocabulary used in monolingual discourse. As such, whatever necessary criteria we choose to use to delineate marginal cases, conventionalisation must remain a sufficient criterion for borrowing: if the whole speech community uses the form irrespective of individuals’ level of competence in its source language, it must be a borrowing, not a code-switch. Poplack rejects the criterion of conventionalization as essentially extra-linguistic, preferring the criterion of morphosyntactic integration; yet the latter invokes community conventions just as much as the former, the only difference being the type of conventions invoked (grammatical vs. lexical.) Finding that single words of foreign origin overwhelmingly display morphosyntactic integration and are thus by her definition nonce borrowings, she concludes (p. 213) that “loanwords do not originate as code-switches… the very first mention of a nonce form already features the full complement of morphosyntactic integration into [the recipient language]”. But this makes some problematic predictions.

First of all, if this is true, borrowings should never retain source morphosyntax. This is clearly not tenable. Borrowings retain source morphology all the time: Berber nouns in several Arabic dialects, and Arabic nouns throughout Berber, keep their plurals; Latin nouns in German keep their case markers; in a tiny scattering of languages around the Mediterranean, such as Ghomara Berber, borrowed verbs even keep their conjugation. Some categories of borrowings retain their syntax as well: larger borrowed numerals precede or follow the noun according to the rules of the source language, not of the recipient, in Korandje; borrowed primary adpositions and complementizers rather consistently place their complement as in the source language wherever they are found (cf. Moravcsik 1978). Poplack attempts to dispose of the latter with a short footnote (p. 50): “More wide-ranging proposals for borrowability hierarchies […] including prepositions, determiners, pronouns, clitics, and complementizers may be characteristics of certain extreme borrowing situations, such as pidginization or creolization, or, alternatively, the result of confounding code-switches […] and borrowing. The latter is so heavily restricted to content words that this is practically a defining characteristic.” But this really will not do. Turkish (which has borrowed the complementizer ki from Persian along with the associated word order) is hardly anyone’s idea of a pidgin or creole!

Second, such a claim (along with the book as a whole) seems to presuppose that borrowings are necessarily single lexical items. This is manifestly not the case. In English, borrowings that consist of multiple source language words (quid pro quo, per cent, hors d’oeuvres…) are sufficiently unanalysable to be considered as single lexical items in the recipient language; these need not pose a problem for Poplack. But in quite a few languages, including many Berber varieties, at least two classes of multi-word borrowings remain clearly analysable as multiple words, and productive, even for monolingual speakers: numerals, and numeral+measure noun combinations. Such borrowings must necessarily start out as code-switches in Poplack’s terms.

From these facts, I conclude that the process of conventionalization is even more independent of morphosyntactic integration than Poplack assumes. Morphosyntactic integration, as Myers-Scotton implies, is far stricter for structure than for semantics, and is strictly obligatory in neither case. And for function words, at least, syntactic integration only concerns relations up the tree, not down it. It follows that neither morphosyntactic nor phonological integration can be considered necessary or sufficient criteria for borrowing.

Ga3 c'est que ga3!

Among the many responses to recent events in Algeria circulating on Facebook, a particularly linguistically interesting one caught my eye:

His Excellency the People
orders:
"Throw 'em all out!"
That means all!
Million Man March of Friday 29 March
Peaceful, Popular, National, Civilised

The first two lines and the last two are in Fusha (Standard Arabic); the two middle lines are very much in Darja ("dialectal" Algerian Arabic). The clash of registers produces an amusing effect. But even more striking is the first word of the fourth line: سيك sik. You'll search for it in vain in Arabic dictionaries, or even in Algerian Arabic dictionaries (they do exist) printed before, oh, 2000 or so; it's a word from French - well, in French it's three words - c'est que, literally "it is that..." In French, this structure is used to mark sentence focus. You can find examples of it being used that way in Algerian Arabic too, eg in Lotfi DK's "Aar alikoum": "المشكلة فهذ البشر سيك في راسهم كاين لحجر" [The problem with these people is that there are rocks in their heads]. But that's obviously not quite what's going on here, though. I haven't come across this construction often enough to have a good sense of all its uses, but it seems to be gaining new functions as it becomes integrated into the wider system of information structure marking in Darja.

Kabyle-Arabic code-switching

One of the great understudied subjects in North African linguistics is Berber-Arabic codeswitching (unlike French-Arabic and French-Berber codeswitching, which are massively overstudied); among the few references available are Kossmann 2014, for a central Moroccan community, and Hamza 2007, for Tunisia. A video from "Nass Bejaia" that's been circulating on social media might seem to provide a promising data source for this: in an eloquent plea against efforts to divide and conquer by setting ethnic groups against each other, the speaker alternates fluent Darja [Algerian Arabic] and Kabyle [Berber] to convey his message to both groups. What better context for code-switching? But, surprisingly enough, there is almost no insertional code-switching, almost no embedded language islands. About the only example is (Kabyle in bold, Arabic underlined):

makanš la d aqbayli, la d aɛeṛbi, la d amẓabi, la d annayli, la d attargi, ula ma... yji waħəd məlkamrun ysəggəm ldzayər nqululu mərħba bik.

There is no Kabyle, no Arab, no Mozabite, no Naili, no Tuareg, even if... there comes someone from Cameroon who would fix Algeria, we'll tell him welcome!

Even here, the second shift comes after an audible pause, and it's probably no coincidence that all the Kabyle elements of this sentence except ula ma are immediately comprehensible to Arabic speakers; even copular d is widely used in Jijel and Bejaia, though unfamiliar elsewhere (and la is ambiguous, used in both languages, which probably facilitates the first shift). Otherwise, the language shifts are rather consistently at phrase boundaries, as in the sentence that follows the previous:

yji waħəd mənnižir, waħəd məḷḷalmạn, lɛaslama

Someone comes from Niger, someone from Germany, welcome.

or as in this later sentence (French in red):

Anda ara aɣ terrem? Wac, ad ɛawdeɣ? Lukan par exemple - Ya xawti, ya xawti, had əlmisaž muhimm židdan məbjaya, makanš əljihawiyya.

Where are you taking us? What, should I say it again? If for example - Brothers, brothers, this message is extremely important from Bejaia: No regionalism!

Sometimes the same content is repeated in both languages successively, sometimes it's left in only one language, but in general, any one phrase should be perfectly comprehensible to a monolingual. It remains to be studied whether this is typical of Kabyle-Arabic code-switching, or just a fact about this short clip.

We're all related: a calque from Kabyle into Darja

Algerian Arabic (or at least Dellys Arabic) has a verb for "be related to" (as family): kul كول, taking the dative, as in waš y-kul-lek? واش يكوللك "what relation is he to you?" In the reciprocal form, this yields tkawel تكاول "be related to each other"; "we're related to each other" is ne-tkawl-u نتكاولوا. These only seem to be used in the (present) imperfective; I've never heard anyone say *kal كال.

This verb clearly derives from an Arabic word still used in its own right in Algerian Arabic: kun كون "be", with regular assimilation of n+l to ll and reinterpretation of the root. waš y-kul-lek واش يكوللك "what relation is he to you?" was originally waš y-kun-lek واش يكونلك "what is he to you?" But that construction seems rather odd and unidiomatic from a Classical Arabic perspective. You don't normally use an equational verb "to be" in the indicative present tense like that, in Classical Arabic or even in Algerian Arabic; you would rather expect something with a pronoun, like *wašen huwwa lik واشن هو ليك (which you don't hear). What's going on here?

Flipping through Dallet's (1982) enormous dictionary of Kabyle as spoken by the Ait Menguellet, I came across the answer. The Kabyle verb ili "to be" (imperfective ttili) matches Arabic kun كون fairly well in its usage. In the imperfective, with the dative, it means "be related to" (his gloss: "être parent avec, avoir relation de parenté à"): d acu i-m tettili? "what relation is she to you?") It likewise has a reciprocal myili (imperfective ttemyili) "have in common; be related to each other", which in the latter sense only seems to show up in the imperfective: nettemyili "we're related to each other".

It seems clear that the Algerian Arabic verb derives from an excessively literal translation - a calque - of the Kabyle expression, probably by people whose first language was Kabyle. But since then it's taken its own path; whereas in Kabyle the meaning "be related to" remains a context-specific sense of the verb "be", in Algerian Arabic the change of n to l has allowed it to become an independent lexeme in its own right with no one-to-one Kabyle translation equivalent. Contact catalyses change, but the resulting change follows its own path.

Bilingual suppletion comes from selection conflicts: Supporting evidence from Pichi

Azeb Amha recently directed my attention to a very interesting passage in Kofi Yakpo's grammar of Pichi, the English-based creole spoken in Equatorial Guinea. In this language, English-based numerals are used up to seven (and known in theory by some speakers up to ten), while Spanish numerals are familiar to all speakers and are used consistently above seven. The English numerals are followed by singulars (plural marking in Pichi is handled with postposed dɛ̀n, and occasionally the suffix -s):

So	à	dɔn	gɛt	tri	nacionalidad	nà	dis	wɔl.
so	1SG.SBJ	PRF	get	three	nationality	LOC	this	world
‘So I have three nationalities in this world.’ [fr03ft 102]

When Spanish numerals are used, however (p. 545), we get "bilingual suppletion" (Matras 2012) - i.e., a grammatical rule of one language that seems to require switching into another one:

The attributive use of Spanish numerals goes along with the insertion of Spanish head nouns – there is no instance of a mixed combination of a Spanish numeral and a Pichi noun:

Lɛf=àn	mek	è	rich	a	los	quinze	años.
leave=3SG.OBJ	SBJV	3SG.SBJ	reach	to	the.PL	fifteen	years
‘Leave her, let her reach [the age of] fifteen years.’ [ab03ay 138]

In Spanish, of course, numerals other than 1 select for plural nouns.

Now I would prefer to see a wider range of examples before reaching any firm conclusions, because counters like "years" are inherently more likely to cause borrowing of numeral+noun units. But, as described, this language precisely fits the explanation proposed for bilingual suppletion in Souag & Kherbache (2016), based on Myers-Scotton's Embedded Language Island Hypothesis:

[W]here bilingual suppletion in numeral+noun combinations emerges, it will occur only following borrowed numerals whose noun selectional requirements in the source language differ from those in the recipient language.

I was, of course, unaware that Pichi displayed bilingual suppletion when I proposed this generalization, so I take this as corroborating evidence. I would be interested to hear of any further examples.

fatta: a loan from Chadic into Songhay?

The Proto-Chadic word for "go out" was reconstructed by Newman and Ma (1966) as *p-t-, with attested reflexes in all primary subgroups of the family; the best known of these is of course (West Chadic A.1) Hausa fìtā.  The vowels vary across languages, and there is often no final vowel.  Only one subgroup, as far as I can see on a quick check, shows the consistent vocalisation *patā: the Bole languages (West Chadic A.2), spoken in Nigeria's Yobe State along the boundary between Hausa and Kanuri.  Thus Bole pàtā, Ngamo hàtâ, Karekare fàtā.

Most Songhay varieties have reflexes of two near-synonyms for "go out": *hùnú and *fáttá.  Usually, the distinction seems to be roughly "leave (a place or event)" vs. "go out of (an enclosed or concealed space)".  In Northern Songhay - the subgroup most isolated from the rest for longest, spoken in the Sahara - only reflexes of *hùnú seem to be attested, covering both senses (eg Korandje hnu).  This could be interpreted as reflecting Northern Songhay's general tendency to reduce its inherited vocabulary by widening the usage of generic terms.  In light of the Chadic data, however, it is tempting to interpret it the other way around: did Northern Songhay preserve the original situation, while a West Chadic borrowing spread throughout the rest of the family via the Niger River?

Drawing water in Songhay and Zenaga

Almost every attested Songhay variety (Tasawaq is perhaps the only exception) has a reflex of the proto-Songhay word *gúrú "draw water" (from the river, from a pond, from a well, etc.)  To express this concept, most Berber varieties (including Tashelhiyt, Kabyle, Tumzabt, Ghadames, Awjila, Tamajeq...) use reflexes of a verb *āgum "draw water", which is thus equally securely reconstructible for proto-Berber.  Zenaga, however, has a rather different verb: ägur "puiser l'eau d'un puits, remonter le delou, tirer la corde du seau; faire parvenir qqc (à qqn)" and "se lever (astre)", with an irregular corresponding noun tgäʔrih "eau tirée du puits".  It seems to be distinct from äggur "pull".

The only Berber cognates Taine-Cheikh suggests for ägur are reflexes of a verb that may be reconstructed as *agir "throw; rise (of sun)" (eg Tashelhiyt gr, Kabyle gər, Chaoui gər).  Presumably the semantic shift of "throw" to "draw water" would be explained via the idea of throwing the bucket down the well.  If the comparison is accepted, then the verb shows an innovative semantic shift specific to Zenaga.  (It would be interesting to see if Tetserrét shares this, but unfortunately the relevant term doesn't seem to have been recorded.)

If the Zenaga word is indeed cognate to the suggested Berber forms, then it seems reasonable to draw the conclusion that proto-Songhay borrowed *gúrú "draw water" from an early relative of Zenaga.  This would fit well with the evidence for a Western Berber language having played an important role in the history of at least northern Mali.  If not, then it would become tempting to draw a conclusion much harder to fit with what is known of the region's history: that Zenaga borrowed the word from proto-Songhay.

Berber feminine nouns in Dellys Arabic: an update

In Dellys, Berber nouns borrowed into Arabic are not very common, and ones that preserve the Berber nominal affixes are even rarer, so I'm always on the lookout for them. A few days ago, listening to my eldest aunt, I heard one that was completely new to me, in an old idiom:

xəlləṭ tazalt u bəḷḷuṭ
خلّط تازالْت وبلّوط
mix up tazalt and oak/acorns (ie mix good with bad)

Tazalt was described as a vine with white flowers; probably the reference is to Cistus (rockrose), whose Kabyle name is tuzzalt, "little iron". Why that would be particularly easy to confuse with an oak tree is beyond me. There are a few other plant and animal names retaining the Berber feminine circumfix t(a)-...-(t), including tirẓəẓt تيرززت (a kind of small wasp), tubrint توبرينْت (a kind of seaweed), taɣanim تاغانيم (a variety of fig, from Berber taɣanimt "small reed"), and originally plural timəlwin تيملْوين (another variety of fig). Otherwise, this circumfix seems to be almost exclusively reserved for abstract nouns referring to negatively judged character traits (see previous posts): eg taɣənnant تاغنّانْت "stubbornness", taklufit تاكلوفيت "meddling", tayhudit تايهوديت "malice", tastutit تاستوتيت "malicious trickiness". An amusing variant on this theme came up recently: taṭnuhist تاطنوهيست "open-mouthed stupidity", presumably a blend of unrecorded *taṭnuhit تاطنوهيست and French -iste. (This in turn derives from ṭnəh "mooring-post", as in "dumb as a post".)

More Darja notes: oath complementisers, free choice indefinites, kids' morphology, finger rhymes

Oath complementisers

In North Africa, the oath wəḷḷah والله, literally "by God", is used so frequently to emphasize statements - religious scruples notwithstanding - that a more appropriate synchronic translation might be "seriously". (It can even be used with imperatives, which can hardly be read as committing the speaker to the truth of any given statement.) Perhaps as a result of their high frequency, constructions with wəḷḷah have a number of unique morphosyntactic characteristics. Negation after wəḷḷah uses ma ما alone, whereas in most other contexts negation is bipartite ma... š(i) ما... شي. Positive sentences after wəḷḷah are introduced by what seems to be a complementiser, ɣir غير or la لا, which in other contexts mean "just, only". What struck me this time is that in certain syntactic contexts this complementiser systematically shows up twice, once right after the oath and once at the start of the main clause proper; I've come across this in topics:

wəḷḷah la lyum la sxana والله لا اليوم لا سخانة
by.God just today just heat
By God, today, it's hot.

wəḷḷah ɣir anaya ɣir dərt-ha والله غير أنايا غير درتها
by.God just I.EMPH just did.1sgPf-3FSgAcc
By God, me, I did it.

and in conditionals with the condition preposed:

wəḷḷah ɣir lukan t-dir-ha ɣir nə-ʕṭi-k ṭṛayħa والله غير لوكان تديرها غير نعطيك طرايحة
by.God just if 2Sg-do-3FSgAcc just 1Sg-give-2SgAcc beating
By God, if you do that I'll give you a beating.

In generative grammar, it is generally supposed that sentences are complementiser phrases. The complementiser is unpronounced in normal declarative sentences here, as in many languages, but is pronounced overtly in specific circumstances such as, here, oaths. A popular hypothesis in the cartographic approach to generative grammar proposes that the complementizer phrase needs to be split into a more fine-grained set of projections: Force > Topic > Focus > Topic > Finiteness, following Rizzi 1997. Prima facie, this complementiser-doubling data suggests otherwise: it looks very much as though right-adjunction of both topics and conditions is being handled by embedding the CP within another CP.

---

Free choice indefinites

In traditional Algerian Arabic, it seems pretty clear that the function of free choice indefinites ("anyone could do that", "take anything (you want)") isn't very strongly grammaticalised. In French, however, it's expressed using a relatively frequent, dedicated series of forms based on "no matter" plus the interrogative pronouns: n'importe qui/quoi/quel "anything, anyone, any..." Younger speakers of Algerian Arabic have borrowed the morpheme n'importe, but not the construction as a whole; instead, they simply prefix n'importe to existing indefinite nominals, in which interrogative pronouns play no role. Thus the phrase I heard today:

fə-z-zit wəlla f næ̃mpoṛt ħaja في الزيت ولا في نامبورت حاجة
in-the-oil or in any thing
in oil or in any thing

---

More children's morphology

Algerian Arabic has very few native bisyllabic words ending in the vowel u, but in loanwords it's not so unusual; for instance, it uses French triku تريكو (ie tricot) for "t-shirt". The first person singular possessive has two allomorphs: -i after consonants, -ya after vowels. I caught the younger of the two kids mentioned in the last post saying trikuww-i تريكوّي "my T-shirt" and trikuww-ək تريكوّك "your shirt"; his father (and everyone else, as far as I've noticed) says triku-ya تريكويَ and triku-k تريكوك. So it would seem that this kid has reanalysed the word as phonologically /trikuw/. Further inquiries are called for.

---

This little piggy...

I've encountered two finger rhymes in Algerian Arabic around Dellys; compare them to a Kabyle version below from Hamid Oubagha:

Dellys A	Dellys B	Kabyle
hađa ʕaẓẓi məskin هاذا عزّي مسكين This one is a robin, poor thing	hađa sɣiṛ u ʕaqəl هاذا سغير وعاقل This one is small and gentle	Wa meẓẓiy, meẓẓiy meskin ! This one is small, poor thing!
u hađa ṣbəʕ əssəkkin وهاذا صبع السكّين And this one is the knife-finger	u hađa ləbbas əlxwatəm وهاذا لبّاس الخواتم And this one is the ring-wearer	Wa d Ɛebḍella bu sekkin ! This one is Abdallah of the Knife!
u hađa ṭwil bla xəsla وهاذا طويل بلا خسلة And this one is long without function	u hađa ṭwil u məhbul وهاذا طويل ومهبول And this one is tall and crazy	Wa meqqer, meqqer bezzaf ! This one is big, very big!
u hađa ləħħas əlgəṣʕa وهاذا لحّاس القصعة And this one is the dish-licker	u hađa ləħħas ləqdur وهاذا لحّاس القدور And this is one is the licker of pots	Wa d ameccaḥ n teṛbut ! This one is the dish-licker!
u hađa dəbbuz əlgəmla وهاذا دبّوز القملة And this one is the louse-club	u hađa dəbbuz ənnəmla وهاذا دبّوز النملة And this one is the ant-club	Wa d adebbuz n telkin ! And this one is the lice-club
u yəmma tqul: mʕizati, mʕizati, mʕizati! ويمّا تقول: معيزاتي، معيزاتي، معيزاتي And mother says: my little goats, my little goats, my little goats!	dəbb əđđib, dəbb ənnəmla, dəbb əđđib, dəbb ənnəmla... دبّ الذّيب، دبّ النملة، دبّ الذّيب، دبّ النملة... Debb the wolf, Debb the ant, Debb the wolf, Debb the ant...	(n/a?)

All three clearly share a common background. Obviously, Dellys B has been deliberately made more posh - ants substituted for lice, pots (with urban q) for dishes (with villagers' g), ring-finger for knife-finger... Dellys A remains defiantly unrefined, but shows at least one sign suggesting an original in Kabyle: ʕaẓẓi məskin "a robin, poor thing" makes a lot less sense for referring to the little finger than meẓẓi meskin "small, poor thing", but sounds almost the same. On the other hand, Dellys A shows a near-rhyme between verses 3, 4, and 5 which doesn't work at all in the attested Kabyle version. It would be interesting to compare more versions in both languages

Sara, sara

With only 30,000-odd inhabitants, and fairly poor road connections, Dellys is a reasonably small and out-of-the-way place. In summer it briefly fills up with the unfamiliar faces of other Algerians looking for a quiet beach holiday, but I've never seen, for instance, a Chinese person here, even though there are plenty in Algiers. Nevertheless, the problems of the Sahel have made themselves felt even here: this year, for the first time, a couple of families from Niger seem to have made it to Dellys. As I was browsing in a little bookshop, a little girl came in, holding up a bowl and saying "Sara, sara". She said the same word to each of us in turn, then left to proceed along her route. Shortly after she left, I belatedly realised what she was saying. In Zarma (the main language of western Niger), historic intervocalic d became r, and intervocalic velars were lost. Arabic ṣadaqah "alms" (Hausa sadaka) is thus reduced to sara. She can't have been here long, or surely she would have found a more effective expression to use; I imagine everyone else was assuming that she was simply repeating her own name.

As a town, Dellys is not particularly fond of strangers, though it leaves them alone; coincidentally, the owner of the bookshop had just been complaining to me about how all the post-independence immigrants into town - from villages a few kilometres away - had made a mess of the place. Absorbing Nigerien immigrants may take some work. But I expect more will arrive; right now, Niger has the fastest growing population in the world, with a birthrate last seen in Algeria in the 1970s, and in the industrialised world during the 19th century. Many Algerian young people dream of escaping the country's sclerotic economy, sometimes illegally by boat from Dellys - there used to be a graffiti near the lighthouse alluding to the early Muslims' flight to Abyssinia: "I shall go to Spain, for it is ruled by a king who does not oppress anyone." But compared to Niger, Algeria might as well be the US.

Language Contact in the Sahara: An overview

I am very happy to announce the publication of my freely accessible overview of Language Contact in the Sahara, written for the Oxford Research Encyclopedia of Linguistics. Apart from being the first introduction to this topic to cover both sides of the Sahara, it encapsulates a good deal of my research program over the past few years, and gives some idea of what remains to be done in this domain. Here's the abstract; if it sounds interesting, go read it!

As might be expected from the difficulty of traversing it, the Sahara Desert has been a fairly effective barrier to direct contact between its two edges; trans-Saharan language contact is limited to the borrowing of non-core vocabulary, minimal from south to north and mostly mediated by education from north to south. Its own inhabitants, however, are necessarily accustomed to travelling desert spaces, and contact between languages within the Sahara has often accordingly had a much greater impact. Several peripheral Arabic varieties of the Sahara retain morphology as well as vocabulary from the languages spoken by their speakers’ ancestors, in particular Berber in the southwest and Beja in the southeast; the same is true of at least one Saharan Hausa variety. The Berber languages of the northern Sahara have in turn been deeply affected by centuries of bilingualism in Arabic, borrowing core vocabulary and some aspects of morphology and syntax. The Northern Songhay languages of the central Sahara have been even more profoundly affected by a history of multilingualism and language shift involving Tuareg, Songhay, Arabic, and other Berber languages, much of which remains to be unraveled. These languages have borrowed so extensively that they retain barely a few hundred core words of Songhay vocabulary; those loans have not only introduced new morphology but in some cases replaced old morphology entirely. In the southeast, the spread of Arabic westward from the Nile Valley has created a spectrum of varieties with varying degrees of local influence; the Saharan ones remain almost entirely undescribed. Much work remains to be done throughout the region, not only on identifying and analyzing contact effects but even simply on describing the languages its inhabitants speak.

From codeswitching to borrowing in une génération?

It's not unusual to hear sentences like the following from middle-class Algerian adults, especially women:

عندنا ان تيليفيزيون كبير
ʕəndna æ̃ tilivizyõ kbir
"we have a big TV"

شرينا لو تيلي
šrina lœ tele
"we bought the TV"

If I had in fact heard these from an adult, I would unhesitatingly classify them as code-switching, with a French noun phrase inserted into an otherwise Arabic sentence. That goes especially for the former - monolingual adults simply don't use the French indefinite article [æ̃ ] (un). In fact, however, I heard them from a monolingual 5-year old, born and bred in Algeria, who only took her first French class this term. Unless she knows more French than she or her parents are letting on, that necessarily makes them monolingual sentences. And that means that, for this young lady, [æ̃ ] (un) has become a borrowing into Algerian Arabic - an indefinite article used with words that take the definite article le.

Earlier, I noted that children don't typically initiate effective language change; and, in terms of output, this isn't a change at all. She's simply learned to produce the kind of sentences she hears all the time directly, without going through all the effort of learning French first. In terms of the underlying system, however, it's a significant change. Instead of having one indefinite article used with all nouns, she now has two: one with Arabic nouns, and one with French nouns. (Rather like the nouns borrowed from Berber that we looked at earlier, which can't take the Arabic definite article.) In Saussurean terms, one generation's parole (the relatively free Arabic-French codeswitching practiced by her parents) has become the next generation's langue. And that sort of change is by its nature something children, and only children, are extremely likely to lead.

(Note, by the way, that in French télévision is feminine; I'm not sure why she gives it masculine agreement, but probably this reflects the influence of the earlier borrowed form tilivizyun, which regular Arabic phonological gender assignment rules make masculine.)