Ghost Languages?

Ghost words are words that have typically never existed in reality, but that get taken into account in reference works or scholarly research and come to have a life of their own until they are noticed and removed. Erroneous citation, spelling or printing mistakes, or misreadings can all be reasons that bring ghost words into existence. Ghost words have plagued linguistic research at various points—imagine publishing an analysis of a strange Latin word, only for a later scholar to point out that the word was originally misspelled in the text edition you were using. Something similar can happen when some source claims that a language is or was spoken in such-and-such place, and later sources adopt that claim as fact, but in reality there never was (or never plausibly was) such a language in such a place!

For example, I recently came across a Wikipedia page for the Zurg language, a Berber language apparently spoken in Kufra, Libya. (Note: this is probably not an inside joke referring to the Zerg). Clicking around the Wikipedia pages for Berber languages, one notices as well that a Tmessa language is classified as part of Eastern Berber, but clicking on it redirects one to the page for the language of Sokna, Libya (a real, though probably recently extinct, language). Strangely enough, most Berberologists have never heard of either Zurg or Tmessa.

These Wikipedia pages are not entirely to blame, as they derive their information from three sources: a 1999 work by Václav Blažek on the reconstruction of numerals (with a chapter on Berber numerals), a 2001 work by Roger Blench in which a subgrouping of Berber languages is proposed, and a 2006 work by Blench & Dendo repeating the 2001 subgrouping of Berber. *Note that all these subgroupings are out of date.*

The first of these, Blažek (1999:157-159) gives a subgrouping of Berber languages drawn largely from the Russian studies of Aikhenvald and Militarev (1987, 1991; which I have not been able to consult) as follows:

“Eastern Berber: Siwa (oasis Siwa in West Egypt), Zurg (oasis Kufra in East Libya), Fezzan (oases Tmessa and El Fodjaha [sic!] in South Libya), Augila (oasis Djalo in North-East Libya), Sokna (North Libya), Ghadames (oasis Ghadames in West Libya)”

It should be noted that the geography is a bit off, and Augila (Awjila) and Djalo (Jālu) are two separate oases in eastern Libya, only the former of which is Berber-speaking. “El Fodjaha”, also misspelled below, refers to El-Fogaha الفقها.

The articles by Blench (2001:86) and Blench & Dendo (2006:12) give the following subgrouping:

“Eastern Berber: Siwa, Awjila, Sokna, Ghadames, Zurg (=Kufra), and Fezzan which consists of Tmessa and Al-Fojaha [sic!]”

The Wikipedia article for “Eastern Berber” adds:

“The “Lingvarium Project” (2005) cites two additional languages: the extinct language of Jaghbub and the still-spoken Berber language of Tmessa, an oasis located in the north of the Murzuq District. [7] Blažek (1999) considers the language spoken in Tmessa as a dialect of Fezzan. [8]”

Note [7] of the Wikipedia article refers to an article in Russian which also names Zurg and Tmessa as Berber languages. Note [8] is the above-cited work by Blažek. As far as I know, there is no Berber language spoken in Jaghbub (an oasis near the Libyan-Egyptian border), except by Siwans who sometimes went there for work (and spoke Siwi, if they used Berber at all when there).

Again, I’ve never heard of either Zurg or Tmessa varieties of Berber. Of course, in the above sources none of the scholars can give any data for either language, because there isn’t any.

Tmessa (تمسّة) is a small town in west-central Libya, in the far east of the Murzuk-Sabha region. Since there are Touareg as well as Tebu in that region, it is possible that Touareg or Tebu speakers live in Tmessa, but such a claim or report has never appeared in a publication, as far as I know. There is no point in assuming so, and even assuming so would make Tmessa Touareg- or Tebu-speaking, which is quite different from it having its own unique variety of Berber. Though the above-mentioned sources do not cite it, the only reference to language in that town was made by the German traveler Gustav Nachtigal in the 1870s, in which he claims that it was Berber-speaking but gives no references or linguistic data (thanks to Maarten Kossmann for having noticed this a few years ago in a guest post on this blog):

“…haben in den vorher aufgeführten Berber-Kolonien Sôqna, Waddân, Temissa Berberdialecte, die dem von Ghadàmes nahe stehen, noch Bürgerrecht neben der arabischen Sprache: doch sind dies streng abgeschlossene Sprachinseln.

[…besides the Arabic language, Berber dialects, which are close to that of Ghadames, are still current in the Berber colonies of Sôqna, Waddân, and Temissa, but these are closed off language-islands] (Nachtigal 1879:195)”

So if “Tmessa” could be pulled from Nachtigal’s impressionistic report as a real Berber language, why not “Waddan”, which otherwise is never reported as having (or having had) a Berber language? Perhaps Nachtigal had good information that Berber was spoken in “Temissa”. But as probably both documented Berber languages of central Libya (in Sokna and el-Fogaha) have probably become extinct this past century, is it even likely that a Berber language of Tmessa would have survived until now? Doubtful.

Zurg is a locale within the oasis complex of Kufra, itself home to Tebu speakers (and perhaps various languages from Chad spoken by migrants) but not Touareg as far as I know. Interestingly, of all the Wikipedia entries for Kufra, only the English version mentions this locale (“ez-Zurgh”), along with a note pointing to the 1929 edition of the Guida d’Italia del Touring Club: Colonie e Possedimenti—however, no reference to the Zurg language is there. The Guida states that Zurg was inhabited by only a few hundred slaves—if accurate, this also makes it unlikely that a unique Berber language was spoken only in that specific locale within Kufra. Otherwise, I haven’t been able to track down other passages (e.g. in travelers’ accounts such as Rohlfs or Forbes) which may mention a connection between Zurg and Berber.

To see if anything else could be found about these mysterious languages, I turned to an obscure 1937 article by Francesco Beguinot, the most prominent Berberologist of the colonial period. In a chapter entitled “I Linguaggi” in the book Il Sahara Italiano: Fezzàn e Oasi di Gat, Beguinot gives a rough summary of the languages spoken in the Fezzan (I’ve uploaded a PDF for those interested). He concentrates mostly on Touareg and on ancient inscriptions, but has the following to say (pp. 502-503):

Attualmente nella parte orientale e centrale del Fezzán l’arabo domina quasi assolutamente. Parecchi anni fa un berbero di Augila che, come altri suoi confratelli, aveva molto viaggiato e conosceva vari paesi della Cirenaica e della Tripolitania, parlandomi di località ove il linguaggio berbero si era conservato, mi nominò, per il Fezzán, el-Fogaha, Trághen, Tméssa, et-Tuíla, Sébha, ez-Zíghen. Sarà necessario per accertare se almeno parte di tali osservazioni risponda a verità, fare una minuta indagine sui luoghi. Nel breve tempo che ebbi disponibile durante le missione geografico-linguistica dell’aprile-maggio 1934 constatai personalmente che a Sebha e a ez-Zighen non vi sono attualmente gruppi parlanti berbero, ma solo individui sporadici provenienti da altre località; quanto a Traghen, Tmessa e et-Tuila, le informazioni dataemi da indigeni di Sebha furono del tutto negative; mi risultò invece come berberofona el-Fogaha, che sarebbe quindi, stando a tali dati e notizie, la sola località del Fezzan orientale cha abbia conservato il berbero…trovai tuttavia, dopo molte ricerche a ez-Zighen una vecchia donna originaria di quella località, dalla qualie raccolsi un po’ di parole cha (salvo ulteriori accertamenti) mostrerebbero l’appartenenza di quel dialleto al gruppo della Zenatiya, mentre in qualche voce esso coinciderebbe in modo specifico con quello di Socna (per es. amâr uomo; miktâr asino).

[Currently Arabic dominates the eastern and central parts of the Fezzan almost entirely. Several years ago, a there was a Berber from Awjila who, like others from his community, had traveled a lot and was familiar with many locales in Cyrenaica and Tripolitania, and told me of some locations where the Berber language had been conserved. He named for me, in the Fezzan, el-Fogaha, Traghen, Tmessa, et-Tuila, Sebha, and ez-Zighen. It will be necessary to ascertain whether some of these observations correspond to reality and to make a detailed survey of the places. In the short time available to me during the geographic-linguistic mission of April-May 1934, I could personally confirm that there are no Berber-speaking groups in Sebha or ez-Zighen but only sporadically individuals from other places. As for Traghen, Tmessa, and et-Tuila the information given to me by natives of Sebha was entirely negative; the only remaining Berberophone (place) would be el-Fogaha, according to this information, the only location in the eastern Fezzan to have conserved Berber … I found, however, after much searching in ez-Zighen, an old woman originally from there, from whom I could gather a few words (unless further investigation) would show the membership of that dialect to the Zenatic group, indeed in some ways it would coincide with that of Sokna (e.g. amār man, miktār donkey). —Translation AB]

Regarding the woman in Zighen who Beguinot mentions, I think it would be a little much to assume an entire language on that basis—perhaps the words were Berber loans into the local Arabic dialect or perhaps the woman had learned them from visiting Berber speakers or was of Sokni descent. There are multiple possibilities. Beguinot also seems to shut down the possibility of Berber in Tmessa in the 1930s at least—nevertheless, if we don’t know anything about a language other than that it exists, how can we group it?

But anyways, I don’t think that most scholars have seen this piece, buried in an obscure and hard-to-find colonial volume as it is. In his other writings, Beguinot doesn’t seem to mention anything about these essentially random places where Berber may or may not have been or was spoken—but we have no data and no firsthand reports. It doesn’t look like Beguinot is the origin of Zurg or Tmessa. The problem may originate in the classifications of Aikhenvald which precede those of Blažek and Blench, but I haven’t been able to check those works. However, if Aikhenvald had any real data, it would have appeared in other studies of Berber languages. And it hasn’t.

To conclude, these handful of scholarly and internet sources seem to be the only places where the languages of Zurg and Tmessa existed. It is frustrating that some scholars simply accept them as being real without bothering to check where they have been mentioned or even if there is any data for them. So until further, firsthand, information can be found (which I think is unlikely), Zurg and Tmessa are simply ghost languages.


  • Aikhenvald, Aleksandra. 1987. Strukturno-tipologiceskaja klassifikacija berberskix jazykov. Sintaksis. Kratkaja istorija klassifikacij berberskix jazykov. Rezul’taty strukturno­tipologičeskoj klassifikacii berberskix jazykov. Moskva: Nauka.
  • Aikhenvald, Aleksandra & Militarev, Aleksandr. 1991. Livijsko-guančskie jazyki. In: Jazyki Azii i Afriki IV.2. Moscow, pp. 148-267.
  • Beguinot, Francesco. 1937. “I Linguaggi” In Sahara Italiano: Fezzan e Oasi di Gat. Roma: Societa Italiana Arti Grafiche Editrice.
  • Blažek, Václav. 1999. Numerals – Etymological Analyses of Numeral Systems and their Implications. Masarykova Univerzita v Brné.
  • Blench, Roger. 2001. “Types of language spread and their archeological correlates: the examples of Berber.” Origini 23, pp. 169–189.
  • Blench, Roger & Mallam Dendo. 2006. The Afro-Asiatic Languages. Unpublished Draft.
  • Nachtigal, Gustav. Sahara and Sudan, I Fezzan and Tibesti. Translated from the original German with and Introduction and Notes, by Allan G.B. Fisher and Humphrey J. Fisher. London, C. Hurst & Company (1974), p. 178. Original title Sahara und Sudan I, Berlin: Weidmannsche Buchhandlung (1879), p. 195.

كتاب المحاورة الإنسانية في اللغة العربية والسوكنية – Book of Human Dialogue in the Arabic and Sokni Language

في 1850 كثب علي بن الحاج عبد الطويل السوكني مخطوطة قصيرة بعنوان “كتاب المحاورة الإنسانية في اللغة العربية والسوكنية ” للرحال الإنجليزي جيمز ريتشاردسون. أرسله إلى إنجلترا ولا تزال في الإرشيف الوطني البريطاني.  وها هي.

James Richardson was an English traveller who made two trips to the Libyan desert in the mid-nineteenth century, gathering notes on half a dozen languages in the process which have ended up at the UK’s National Archives.  In 1850 he persuaded a man from Sokna named ʕAlī bən əl-Ḥājj ʕAbd al-Ṭawīl to write a short vocabulary of Sokni for him, rhymingly titled كتاب المحاورة الإنسانية في اللغة العربية والسوكنية Kitāb al-Muħāwarah al-‘Insāniyyah fī al-Luɣah al-ʕArabiyyah wa-s-Sawkniyyah “Book of Human Dialogue in the Arabic and Sokni Language”.  I’ve already posted on this source a few times (1, 2, 3); here follow the actual pages.  For a full analysis, look out for my forthcoming article:


Works on Ghadames

There are a few works on Ghadames in Arabic, published in Libya, that might turn up interesting leads on its history and language should anyone have the time to delve into them.

One is a grammar of the Berber language of Ghadames in Arabic titled السهل و المسير في تعلم اللغة الامازيغية بلهجة غدامس by Abu Bakr Hārūn. It’s available to order from a UK-based distributor here, though they’re currently sold out due to ‘overwhelming’ demand by Berberologists. There are 484 sayings and prayer formulas accompanying the grammatical remarks.


There are also two catalogues of manuscripts and documents held in Ghadames. The more recent is وثائق غدامس: وثائق تجارية، تاريخية، اجتماعية, published in 1995. It may be ordered online here. It seemingly follows on an older volume (with awesomer cover): فهرس مخطوطات غدامس, published in 1986. Both are by بشير قاسم يوشع. I had a brief flip through both volumes in the library and did not come across any mention of Berber, however, though that may be due to both having been published under the former regime. A more detailed reading might turn up clues.

ghadames-mss1 ghadames-mss2

Here is also a plug for Maarten Kossman’s recent grammatical sketch of Ghadames Berber, the first new monograph on that language in several decades. See the table of contents here.

A map of early 20th-century Awjila


Plan of Awjila from Scarin 1937 (insert between pp. 76 and 77).

Plan of Awjila from Scarin 1937 (insert between pp. 76 and 77).


This map of Awjila comes from a work by the colonial-era Italian “human geographer” Emilio Scarin on the oases of eastern Libya, focusing on Awjila, Jalu, Jaghbub, and Marada. The high density of mosques and tombs of marabouts that Awjila is famous for can clearly be seen.

The term “human geographer” is in quotes because Scarin was evidently a practitioner of the cranium– and lip-thickness–measuring type of fascist racial sciences (just check out his pamphlet “Le oasi del Fezzàn : ricerche ed osservazioni di geografia umana”). Not all of his works were of that type; thankfully the volume from which the above map is taken contains almost nothing of the sort and is full of detailed maps of the oases and plans of vernacular architecture.

Scarin, Emilio. 1937. Le oasi cirenaiche del 29 parallelo. Ricerche ed osservazioni di geografia umana. Firenze.

Recent publications

Two articles by Marijn van Putten, both historical linguistic studies of Libyan Berber varieties, are now published.

“Some notes on the historical consonantism of Awjila” (Folia Orientalia 51, 2014, 257–274) can be accessed at this link. The contents of the entire issue can be viewed here. Hooray for open access!

“Reflexes of the glottal stop in Nefusa and Ghadames” (Wiener Zeitschrift für die Kunde des Morgenlandes 105, 2015, 303–314), is not available online for a while but can be consulted in print format (or for an offprint contact Marijn directly or leave a comment below).

Ghadames story 1: The kitten and the rat

This text comes from Lanfry, J. (1968.) Ghadamès. Etude Linguistique et Ethnographique. Algérie: Fort-National. The texts have been transcribed by Maarten Kossmann, and all the corrections that Lanfry provided in a later publication have been included by him. For those who would like an idea of what Ghadames looks like:

Ghadames. Image by George Steinmetz, National Geographic.


Instead of a word-by-word gloss, this time we’ve only given the translation. A complete vocabulary can be found after the break.

yo̱t te̱žărt tǝlla takaṭṭuss d oḇǝǧǧan. ǧărrădăn, ǧărrădăn. aḇǝnnǝḇǝn rŏwwăḥăn. ikk asǝf sa.
One time there was a kitten and a rat. They played and played. In the evening they went home. Every day (it went) like that.

yo̱t te̱žărt, tǝnna-yas ma-yis n takaṭṭuss-e i yălle-s: – din twe̱das?
One time, the mother of this kitten said to her daughter: Where did you go?

tǝnna-yas: – gărrădǝɛ năšš d oḇǝǧǧan.
(The kitten) said: I was playing with rat.

tǝnna-yas: – iše wăl t-id-tăbbe̱t?
(The mother) said to her: Why have you not caught her?

tǝnna-yas: – azakka t-id-ăbbăɛ.
(The kitten) said: Tomorrow I will catch him.

oḇǝǧǧan-e tǝnna-yas ma̱-yis: – din twe̱das?
As for that rat, his mother said to him: Where did you go?

inna-ya̱s: – gărrădăɛ năšš ǝt takaṭṭuss.
(The rat) said to her: I was playing with kitten.

tǝnna-yas: – ak tǝkṣe̱dǝt šǝk-tǝšš?
(His mother) said to him: Are you not afraid that she’ll eat you?

inna-ya̱s: – azakka ak tiwiɛas [tiweɛas? JL]
(The rat) said to her: Tomorrow I won’t go to her.

takaṭṭus-e, eḇăḍ ǝnnăs imda ak tănădde̱m… ǝṣṣala, tǝkkăr-d zik, tǝzzăl i daž n oḇǝǧǧan-e, tǝnna-yas-ǝn: – oḇǝǧǧan, oḇǝǧǧan, wiǧǧǝz, ǝn_nǝǧrăd!
The cat, the whole night goes by and (lit. ‘her night ends’) she did not sleep… In the morning, she wakes up early and runs to the house of the rat. There she says to him: Rat, Rat! come down, so that we can play!

inna-yaz-d oḇǝǧǧan-e: – ke am-tăssǝlmăd ma-yim tăssǝlmăd-i-t imma!
And rat said to her: What your mother has taught you, my mother has taught me it!

yăbul-az-d esm-i yărwăl.
He pissed in her ear and fled.

tǝqqa tulless, wăl tǝqqe rrăḥmăt ǝn Răbbi!
The story is over, (but) the compassion of God is not!

Read more of this post

Intra-Berber borrowing in Yefren

At a recent conference, I met the Libyan activist Maziɣ Buzexxar, who was kind enough to answer some of my questions about Yefren Berber.  As we have already seen for “hand“, words that end up as CuC monosyllables in Yefren (and elsewhere in Zenati) regularly correspond to uCeC in Nefusi (and Ghadamsi).  One such word is “straw”: lum in Yefren (as in Siwa) vs. ulem in Jadu (and alim in Kabyle).  Thus, in Yefren, “a pile of straw” is aguday n lum.Milky_Way_Night_Sky_Black_Rock_Desert_Nevada

Throughout the Middle East and North Africa, the Milky Way is referred to as “the Road of Straw” or “Road of the Straw-bearers“, with the scattered stars conceived of as straws fallen from the load being carried, rather than, as in Greek tradition, as milk.  Yefren is no exception.  But in Yefren, according to Maziɣ, the Milky Way is called: abrid n ulem (which he understood, with no difficulty, as meaning “road of straw”).

If the facts are correct – obviously confirmation with more Yefrenis would be better – then the explanation is obvious: Yefren borrowed its term for the Milky Way as a whole from the Nefusi communities to its west, while keeping their original word for “straw”.  Intra-Berber borrowing – borrowing of words from one Berber language into another – must have been an important phenomenon historically, but it is often made invisible by the sound correspondences being trivial.  In this case, the Nefusi vowel shift makes it unusually conspicuous.  More such cases should be sought.

“Meat” and “drink” in Libyan Berber

The previously mentioned Facebook group has been continuing its activity – particular credit should go to Sami Halasa and Faycel Marir, for asking many dialectologically interesting questions.  Two recent ones seem interesting enough to share, and illustrate slightly different isogloss patterns – as usual, the Jebel Nefusa minus Yefren+Al-Qalaa and Wazzin stands out, but this time it shows significant internal divisions too, in a different place each time.

Meat“: aksum in dark green, aysum in green, usəm in yellowish-green, isan in white.


Aksum/aysum is pretty well universal in Berber along the Mediterranean – showing the usual Nefusi vowel shift in usəm – while isan is more associated with Tuareg.  It’s possible that isan is related to aksum/aysum/usəm: in Ghadames, it appears that isan is actually the irregularly formed plural of aksəm.  If so, however, I don’t understand the relationship well: was -um a singulative? or perhaps just -w, with -un-w > -um?

Also interesting is “he drinks“: isəss in green, itəssu/isəssu in yellowish green, yəssaw in white:


Of these, isəss, as the most irregular form, is presumably original (cf. Kossmann 2008); the other two have been partially regularised to fit better with more common conjugations.  The Nalut form looks exactly like Siwi, but that may be coincidence.  Sokna shows a combination of both innovations: itəssaw.


Umberto Paradisi is known best for the work he did on the several understudied dialects of Awjila and El-Foqaha. However, he has also published one text on Zwara Berber. I have rerecorded this story and discussed it with a native speaker, below you find the translation.

My informant told me that this ritual is still celebrated today.

tə́lt iyyám n uwə́ssu á ytəmm di-s əlmizán g užənná.

“During the three days of Awəssu, Libra will appear fully in the sky.”

  • tə́lt iyyám ‘three days’
  • n ‘of’, Paradisi records an m here as an assimlation of the n plus a following u. When I recorded this story, my informant would consistently pronounce it without the assimilation, despite the original text that I showed him having a translation. This seems to be an indication that for the speaker this rule is not active, or at least not in this context.
  • uwəssu ‘A summer ritual’, in the Etat d’Annexion (EL awəssu)
  • a Marker of the future
  • y-təmm aor.3sg.m. ‘to finish, complete’, in this context ‘appear fully’
  • di-s ‘in’ in the pre-prepositional form + 3sg. prepositional pronoun ending.
  • əlmizán ‘Libra’, a constellation that consists of three starts.
  • g ‘in’
  • užənná ‘sky’, in the État d’Annexion

íḍ aməzwár a yə́ffəɣ ítri, táni a yə́ffəɣ táni n itrán, əttálət a yə́ffəɣ ttálət n itrán. Baʕdén əlmizán.

“On the first night a star will come out, the second (night) the second of the stars will come out, on the third (night) the third of the stars will come out. After that: Libra.”

  • íḍ ‘night’
  • aməzwár ‘first’
  • y-ə́ffəɣ aor.3sg.m. ‘to come out’
  • ítri ‘star’
  • táni ‘second’
  • itrán ‘stars’, should be in the État d’Annexion, but is apparently not distinguished form the État Libre in Zwara.
  • (ə)ttálət ‘third’
  • baʕdén ‘then, after that’

Read more of this post

Towards a Libyan/Tunisian Berber dialect atlas

The polls on the previously mentioned Libyan Berber site are a wonderful data source, notwithstanding occasional difficulties caused by inexact transcription.  Here are a few more, selected from the minority of posts that include Wazzin:

Sun“: tufut in white, tfuyt/tfwit in green, təfuṯ in dark green:


Son“: təṛwa in white, məmmi in green:


Year“: suggəs/sukkəs in white, asəggas/asəggʷas in green, asəkkʷas in dark green, and (Arabic) əlʕam in yellow.


Note that, each time, the eastern and western extremes of the Berber-speaking area of Nefusa – Wazzin, Yefren, and Al-Qalaa – group outside of the core Nefusi area and with more Zenati-like dialects of the region.  One gets the same impression from other posts which sadly exclude Wazzin, eg:

Donkey” (حاشاكم:) ziṭ/aziṭ in white, aɣyul in green, aɣɣul in dark green:


Dog“: yudi in white, aydi / ayəddi in green, ayḏi in dark green, aɣərzul in blue:


That’s not to say, of course, that the central Nefusi dialect area is always homogeneous; there seems to be a certain number of isoglosses separating the Jadu area from the Nalut area.  Consider, for instance:

Figs“: iməṭkən in white, iməṭšan/iməṭšən in green, ifərgas in blue.  (I’ve ignored a few attestations of the Arabic loan kəṛmus, since they are inconsistent.  The form iməṭšən is attested only in the Nalut area, but some speakers from the same areas are writing iməṭšan.)


For all of these varieties, available grammatical descriptions are at best inadequate; practically nothing has been published on the dialects of Yefren, Wazzin, or even the Nalut region.