Northern Songhay


The best-known languages of the western and central Sahara have spread from its northern edges: Arabic and Berber.  However, a third family of sub-Saharan origin - Northern Songhay - has a scattered presence over a substantial part of this region.  Northern Songhay is a highly distinctive subfamily of Songhay.  It has been heavily influenced by Berber and to a lesser extent Arabic to the extent that its members have often been seen as mixed languages, posing a challenge for monogenetic models of language change and a test case for theories of language contact.  Its distribution and vocabulary in turn cast light on the history of the Sahara and Sahel.

Northern Songhay consists of the following languages (see also corresponding Northern Songhay KMZ file):
Distribution of the Songhay languages

  • Kwarandzyey (Korandjé), in Tabelbala (SW Algeria) - 3,000 speakers;

  • Tadaksahak, around Ménaka (NE Mali) - 30,000 speakers;
  • Tagdal/Tabarog, around Tahoua (N Niger) - 25,000 speakers;
  • Tasawaq, in In-Gall (N Niger) - 8,000 speakers;
  • Emghedesie, in Agades (N Niger) - extinct.
Kwarandzyey, Tasawaq and Emghedesie are the languages of their respective towns, each of which has depended historically on agriculture and the trans-Saharan trade; their speakers come from a variety of backgrounds.  Tadaksahak and Tagdal / Tabarog are spoken by nomads whose dominant class held the role of religious specialists within Tuareg society; their apparent integration was such that Basset (1934) treated Tadaksahak as a Tuareg dialect.

The Northern Songhay family has rather less than 100,000 speakers, making its status precarious, and Kwarandzyey at least appears to be endangered.

Phonological characteristics

All surviving Northern Songhay languages have adopted uvular and pharyngealised consonants (except in the latter case for some dialects of Tasawaq), and allow a wider range of consonant clusters than proto-Songhay did. Only Tasawaq retains phonemic tone, although phonemic stress in Tadaksahak reflects proto-Songhay tone patterns (Nicolaï 1981.)

Grammatical characteristics

All Northern Songhay languages use the clause order subject-verb-object, unlike most southern Songhay varieties.  Prepositional phrases with pronominal objects tend to be placed directly after the verb.  Complementizers are strictly clause-initial, and are extensively borrowed.

In Tadaksahak and Tagdal, plurality is always marked on the head noun as well as some modifiers, with Berber nouns taking Berber plurals and Songhay ones taking a suffix; in the other languages, plurality is marked on the head noun for many Arabic and Berber loans using the source language plurals, but is otherwise marked by a Songhay clitic whose placement within the noun phrase varies.

The basic noun phrase order is possessor n-noun-adjective-relative clause.  Possessors may also be placed after the noun, in which case they take the postposition *wane, also used to form absolute possessives (discussed in Kossmann 2009.)  Kwarandzyey and Tasawaq have retained the Songhay attributive adjective formant *-o(w), while Tadaksahak and Tagdal have largely replaced it with a borrowed Berber participial suffix.  The position of the numeral in the noun phrase presents interesting variation: 1-10 always follow the noun, but higher numbers often precede it.  All but a few of the lowest numerals are borrowed from Arabic or Berber.

All Northern Songhay languages have both prepositions and postpositions, the latter being used mainly for location and possession; the position of postpositions within the noun phrase depends on the language.  Prepositions are extensively borrowed.

The verb normally consists of the following elements, fused into a single word: subject marker - aspect/mood/negation marker - verb stem - directional marker - object pronoun.  The subject markers, optional in some languages, derive from the pan-Songhay subject pronouns, although the 2nd person plural is anomalous.  The aspect/mood/negation markers have been considerably reduced, and the positive imperfective ("present") marker *-b- is unique to Northern Songhay.  3rd person object pronouns often affect the phonological form of the verb stem, causing vowel lengthening. An inherited causative in *-nda is found in Kwarandzyey and Tasawaq, but is absent or unproductive in Tadaksahak and Tagdal, which instead use Tuareg loans (Christiansen 2002, Benítez-Torres 2009).  The directional markers are *-kate "hither", also found in some other Songhay languages, and *-nan "away", unique to Northern Songhay.  Borrowed Arabic or Berber triliteral verbs across the family tend to take a prefix i-/y-, corresponding to the 3ms. subject agreement prefix in the source language, while longer verbs are borrowed as bare stems.


All Northern Songhay languages have borrowed extensively from Arabic and Berber, to the point that such borrowing has become probably the principal means of lexical expansion.  The Songhay layer is limited to a few hundred words, usually basic terms not standing in a relationship of hyponymy or hypernymy to one another (for example, "camel" is retained whereas "male camel" and "animal" are borrowed); Nicolaï (1993) presents a brief overview of commonly retained lexemes.  Some of the Songhay vocabulary is absent from the major Songhay dialects of Gao and Niger; many forms are shared only with Western Songhay, for example.  Some show developments unique to Northern Songhay (for example, *kebar "show", elsewhere in Songhay *kerbu / *kebru / *kebe.)

Historical implications

The fact that a Northern Songhay language is spoken in Tabelbala, an oasis with no known historical connections to northern Niger, poses serious historical problems. Souag (fc) proposes two possible hypotheses: either Timbuktu originally spoke a Northern Songhay language, later transformed by influence from the dialect of the Songhay Empire's capital Gao, or Northern Songhay used to be spoken in the Touat region of Algeria.  Further data is needed to decide the issue.

The Berber influence on Northern Songhay has traditionally been identified with Tuareg, but the reality is more complicated: there is no documented Tuareg influence on Kwarandzyey, and both Kwarandzyey and Tadaksahak turn out to contain loans from a non-Tuareg Berber language closely related to Zenaga and Tetserrét (Souag 2010a).  This forces the conclusion that Western Berber was once rather more widespread.

Proto-Northern Songhay speakers were certainly familiar with herding (*kud), and a range of livestock including camels (*yoo), sheep (*feegi), goats (*hankini), and cows (*hawi).  However, the nomadic lifestyle of the speakers of Tadaksahak and Tagdal is probably not representative of proto-Northern Songhay.  Songhay words retained in proto-Northern Songhay include *fari "cultivated field", *lambu "garden", *kuumu "hoe", *hugu "house", *karkabu "key" (Souag fc), indicating a sedentary agricultural lifestyle.



Kwarandzyey (Korandjé) first came to linguists' attention following Cancel (1908), who provides a brief grammatical sketch and wordlist and notes its Songhay origin.  Champault (1969) is an ethnography of the oasis, including numerous Kwarandzyey words and phrases; the author also began an incomplete unpublished French-Kwarandzyey dictionary (Champault ms).  Tilmatine (1991, 1996) adds a few phrases of data to the previous, and discusses the language.  Souag (2010b) effectively contains a grammar of Kwarandzyey with a focus on contact effects.  A brief overview is given by Souag (2011.)

Nicolaï (1979b, c) discusses the phonology of Kwarandzyey based on the little data then available, to which Nicolaï (1981) adds a discussion of sound correspondences. Kossmann (2004b) discusses in passing r-loss in Kwarandzyey.  Kossmann (2004a) attempts to reconstruct the mood/aspect/negation system of Kwarandzyey from the little data then available, pointing out some areas of intense Berber influence.  Kossmann (2009) examines the genitive system across Northern Songhay, including Kwarandzyey.  Souag (2008) briefly summarises the numeral system.  Souag (2010a) argues for a stratum of Western Berber (Zenaga/Tetserrét) loans in Kwarandzyey, and discussing the principal sound changes affecting the language.  Souag (2014) discusses the Arabic-based orthography used in practice by Kwarandzyey speakers.  Souag (fc a) examines the synchronic syntax and historical development of the typologically unusual Kwarandzyey comitative, which takes external agreement.  Souag (fc b) re-examines the origin of Kwarandzyey, proposing a historical scenario for its emergence integrating linguistic, archeological, and historical data.


The only grammar of Tadaksahak so far is Christiansen-Bolli (2011), a slight expansion of her PhD thesis; it also includes a small Swadesh-based dictionary, a larger list of words of Songhay origin, and glossed texts.  Souag (2014) gives a detailed review of this work.

Probably the earliest source of lexical data on Tadaksahak is the Tuareg dialect maps in Basset (1934), providing eight terms (for "Daousak, Ihannakaten"): mu "oeil" [eye], tinšar "narines" [nostrils], miya "bouche" [mouth], išin "dent" [tooth], taməsandarat "canine" (ie *taɣməs n ḍarat "front tooth"), taɣməs "molaire" [molar], aḍaḍ "doigt" [finger], aurəž "talon" [heel]; Basset (1948) adds two more, bangu "tête" (head) and ašardən "âne" (donkey), commenting that "parmi les Touaregs, au Niger, dans l'énigmatique groupement des Idaousak, surgissent, ainsi pour le coq, l'âne ou les mulet, voire la tête, des variantes lexicographiques qui surprennent ici et ramènent immédiatement à des faits Zenaga, ksouriens ou maghrebin".  Chaventré (1983) adds a chart of kinship terms, in the context of a discussion of Idaksahak genealogy: baba beri "grandfather", nana beri "grandmother", aba "father", nana "mother", anber "elder sibling", antyena "younger sibling", nizey "child", nayan "grandchild / brother's son / maternal uncle's grandson", mushedi "paternal uncle" (glossed as "celui sur lequel on a l'oeil, l'exemple"), armat ma "maternal uncle", nana (tyena) "aunt", abubash "aunt's son", tabubash "aunt's daughter", n'tayaq "sister's son".  A more complete wordlist, along with texts, is to be found in Rueck & Christiansen (1999.)  Heath (np), formerly available from his website, is a brief but careful lexicon of Tadaksahak.  Lacroix (1981) presumably also provides a vocabulary.

Lacroix and Calame-Griaule (1970) provide a Tadaksahak text with translation.  Nicolaï (1979a, c) discusses Tadaksahak phonology, to which Nicolaï (1981) adds a discussion of sound correspondences.  Global Recordings Network (1999) is an audio recording in Tadaksahak gathered by missionaries summarising Christian beliefs; as the only publicly available Tadaksahak audio file, it may be of use for understanding the phonology.  Christiansen & Christiansen (2002, 2007) discusses one of the most cross-linguistically unusual aspects of Tadaksahak grammar: the systematic use of Tuareg loans to suppletively form the causatives and passives of words of Songhay origin.  Christiansen & Levinson (2003) discuss the relative clause.  Ag Almaki (2010) discusses some lexical semantic issues arising in Bible translation.
  • Ag Almaki, Mahamad. 2010. Étude du terme ḥèsèd et traduction en tadaksahak : Deuxième Partie. Le Sycomore 4.2, 2-9.
  • Basset, André. 1934. Note sur les parlers touaregs du Soudan. Bulletin du Comité d'études historiques et scientifiques de l'Afrique occidentale française 17 (3), pp. 496-509. Reprinted in: Basset, André. 1959. Articles de dialectologie berbère. Paris: Klinksieck.
  • Basset, André. 1948. La langue berbère au Sahara. Cahiers Charles de Foucauld 10, pp. 115-127. Reprinted in: Basset, André. 1959. Articles de dialectologie berbère. Paris: Klinksieck.
  • Chaventré, André. 1983. Evolution anthropo-biologique d'une population touarègue: Les Kel Kummer et leurs apparentés. Paris: Presses universitaires de France. (Google Books.)
  • Christiansen, Niels & Regula Christiansen. 2002. Some verb morphology features of
  • Tadaksahak, or Berber or Songhay, this is the question. SIL Electronic Working Papers (2002-005).
  • Christiansen, Regula and Stephen H. Levinson. 2003. Relative Clauses in Tadaksahak. SIL Electronic Working Papers (2003-003).
  • Christiansen, Niels and Regula. 2007. Tadaksahak Verb Morphology with Reference to Berber and Songhay Origins. In ed. Doris L. Payne, Mechthild Reh.  Advances in Nilo-Saharan Linguistics: Proceedings of the 8th Nilo-Saharan Linguistics Colloquium, August 22-25, 2001.  Köln: Rüdiger Köppe.
  • Christiansen, Regula. 2009. Une petite étude lexicologue du tadaksahak, langue songhay septentrional (Mali). In ed. Rainer Vossen, Dymitr Ibriszimow, and Harry Stroomer. Études berbères IV: Essais lexicologiques et lexicographiques et autres articles. Köln: Rüdiger Köppe.
  • Christiansen-Bolli, Regula. 2011. A Grammar of Tadaksahak: A Berberised Songhay Language (Mali). Köln: Rüdiger Köppe. (Roughly identical to her PhD thesis, Leiden 2011.)
  • Christiansen-Bolli, Regula. 2012. Quatre pattes et des cordes ou comment immobiliser des animaux domestiques au milieu nomade. In ed. Dymitr Ibriszimow, Rainer Vossen, and Harry Stroomer. Études berbères VI – Essais sur la syntaxe et autres articles. Köln: Rüdiger Köppe.
  • Heath, Jeffrey. np. Tadaksahak lexicon.
  • Global Recordings Network. 1999. Words of Life - Dausahaq.  (Audio.)
  • Lacroix, P. F. and Geneviève Calame-Griaule. 1970. "La Mère vendue. Essai d'analyse d'un thème de conte africain".  In ed. J. Pouillon and P. Maranda. Echanges et Communications, Hommages à Claude Lévi-Strauss, II, pp. 1356-1380. Paris: Mouton.
  • Lacroix, P. F. 1981. Les parlers du sous-ensemble songhay-zarma septentrional. Documents de terrain. Lexicologie tadaksahak. Microfiche A 77 023 289. Paris, Institut d'Ethnologie, "Archives et documents", micro-édition.
  • Souag, Lameen. 2014. Review of Regula Christiansen-Bolli's A Grammar of Tadaksahak. Journal of Language Contact 7.2, pp. 425-459.


No grammar of Tagdal/Tabarog has been written.  The closest available to a sketch is Benítez-Torres (2009, n.d.), discussing verbal morphology and focusing on the phenomenon of systematic suppletive verbal derivation discussed previously for Tadaksahak.  Nicolaï (1979a, c) discusses Tadaksahak phonology, to which Nicolaï (1981) adds a discussion of sound correspondences.  Wordlists for both Tagdal and Tabarog, along with texts, are to be found in Rueck & Christiansen (1999), which argues that they represent dialects of a single language.  Lacroix (1981) presumably provides a vocabulary.  A brief conversation in Tagdal can be heard in Al Jazeera (2014).
  • Al Jazeera English. 2014/01/09. Orphans of the Sahara I.  (Tagdal spoken, 10m13s to 10m52s)
  • Benítez-Torres, Carlos M. 2009. Inflectional vs. Derivational Morphology in Tagdal: A Mixed Language. In Masangu Matondo, Fiona McLaughlin, & Eric Potsdam (eds.), Selected Proceedings of the 38th Annual Conference on African Linguistics: Linguistic Theory and African Language Documentation, 69-83. Somerville, MA: Cascadilla Proceedings Project.
  • Benítez-Torres, Carlos. n.d. Tagdal of Niger. (PPT presentation).
  • Lacroix, P. F. 1981. Les parlers du sous-ensemble songhay-zarma septentrional. Documents de terrain. Lexicologie tihitit et tagdalt. Microfiche A 77 023 290. Paris, Institut d'Ethnologie, "Archives et documents", micro-édition.


The first grammar of Tasawaq is Alidou (1988), containing a sketch grammar and a vocabulary.  A more easily accessible sketch grammar is given in Wolff and Alidou (2001), focused on the description of Tasawaq as a mixed language.  Kossmann (2007a) similarly describes the grammar of Tasawaq with a contact focus, based on new fieldwork.

Bernus (1972), with a chapter by Lacroix, gives a short vocabulary and a text along with more extensive ethnographical and historical information.  Lacroix (1980) presumably contains a vocabulary of the language.  Rueck & Christiansen (1999) give an extensive vocabulary of Tasawaq (marred by lack of tone marking) and a text, along with some sociolinguistic discussion.  Bernus (1992) and Calame-Griaule (1992) are primarily focused on anthropology, but give some useful lexical data and a couple of phrases.  Anonymous (nd) gives two small vocabularies, including time, animals, colours, numerals, and a few greetings, in an impressionistic French transcription; ScriptSource (2011) adds the days of the week and the French months.

Nicolaï (1979b, c) discusses the phonology, to which Nicolaï (1981) adds a discussion of sound correspondences.  Kossmann (2007b) examines the assignment of tone to verbs borrowed from Tuareg, while Kossmann (2012) examines the variable phonemic status of pharyngealisation.  Sidibé (2010a) presents an overview of plural morphology, while Sidibé (2010b) discusses verbal derivation.

A useful site that can hardly be cited in the usual format is the Facebook group "J'APPRENDS LE TASSAWAQ, LE TAMASHEQ ET LE TAGDALT", with contributions from multiple authors.


The only first-hand description of Emghedesie we are ever likely to have is Barth (1851), who gives an extensive list of words and phrases along with a brief translated text.  It survived into the early 20th century; Abadie (1927:189) briefly (and somewhat inaccurately) notes that "On trouve encore à Agadez et à In Gall quelques vieillards qui peuvent parler un dialecte songhaï, mais ce dernier est tombé en complète désuétude."  Since then, however, the language appears to have become extinct; modern-day Agades speaks a dialect of Hausa (itself inadequately documented.)  Nicolaï (1981) occasionally discusses Emghedesie sound correspondences and phonology.  Lacroix (1981) reviews Barth's data from the perspective of a Songhay-speaking linguist.


Lacroix (1968) first coined the term "songhay septentrional" (Northern Songhay), drawing attention to these languages' differences from mainstream Songhay and to the heavy role of Berber influence in shaping them.  Nicolaï (1978, 1979a, b, c, 1981b) has studied their phonology in some detail, based mostly on first-hand data; Nicolaï (1981) is a general examination of the comparative phonology of Songhay, with an emphasis on contact issues and sound correspondences and a good deal of Northern Songhay data.  Nicolaï (1993) discusses the question of which Songhay vocabulary has been preserved in Northern Songhay, and why.  Rueck & Christiansen (1999) gives a very useful comparative vocabulary and discusses mutual comprehensibility.  Kossmann (2004b) discusses the meager evidence that a (possibly Northern) Songhay language was spoken in the Touat-Gourara region of Algeria.  Kossmann (2009) is a comparative overview of the genitive in Northern Songhay and its possible origins, while Kossmann (2011) examines adjective morphology across the family.  Souag (2012) confirms the cladistic validity of Northern Songhay based on shared innovations, argues that its closest relative within Songhay is Western Songhay (Timbuktu and Djenné), examines the implications of its vocabulary for its history, and proposes two scenarios for its development.