Differences between Tasawaq on the one hand and Tagdal and Tadaksahak on the other extend to the composition of the lexicon, which is always heavily mixed in mixed languages. Although there are no publicly available dictionaries for these languages, it may be judicious here to add some observations at least on the basic lexicon.
An examination of the lexical proportions of elements of various origins in Tadaksahak (Christiansen-Bolli 2010) shows that 62% of the Swadesh 207-item list entries such as are provided in Rueck and Christiansen (1999) are or can be expressed by forms of Songhay origin. In addition, Christiansen-Bolli (2010) listed over 280 forms of Songhay origin for Tadaksahak in her grammar (including two or three forms found only in Northern Songhay languages which may constitute exclusively shared innovations, setting aside a few forms of Arabic or French origin which are common to all NS varieties), and this list of inherited Songhay forms is, as far as we know, complete. Of these 111 are nouns, 138 verbs and 37 belong to other form classes.
We also compared a table of 325 basic lexical items, taken from Rueck and Christiansen (1999), for Tadaksahak, Tagdal and Tasawaq. In Tadaksahak 161 of these 325 forms are of Berber origin (not necessarily all Tuareg), with 157 forms from Songhay and 7 are of other origin (mostly Arabic numerals from ‘five’ onwards). In Tagdal the Berber forms numbered 154 against the 158 of Songhay origin and the 13 of other origins, while in Tasawaq the 110 forms of Berber origin are swamped by the 197 forms of Songhay origin and the 18 forms taken from other sources – again mostly Arabic, which has also donated heavily to both Tamajaq (Berber) and Songhay. Regarding borrowing of basic lexicon, Tasawaq’s degree of borrowing is unusually high among the world’s languages; Tadaksahak and Tagdal show even higher degrees of borrowing. The most common personal pronouns, interrogatives, adverbs and so on (not all of which are on the Rueck and Christiansen list) nonetheless derive from Songhay, with less frequently-used forms in these categories deriving from Berber.
A recent examination of comparable data for Korandje (Souag 2014) shows that only 108 items on the list (there are three gaps in the Korandje data for items which are not lexicalized in the language) are shared by these three languages plus Korandje; the list for the latter contains 183 words of Songhay origin, and the number of items of non-Arabic origin (from Songhay or a Berber language) in the various languages which are shared with Korandje on this list ranges from 135 (with Tadaksahak), 136 (Tagdal), 138 (Tabarog) to 142 (with Tasawaq).
It is by no means certain that the majority of Berber forms on the Korandje list come specifically from Tuareg languages rather than from (say) Tetserrét, while the extensive overlay of Arabic forms in Korandje may conceal or replace some former Northern Songhay shared innovations, Western Berber forms, later loans from Zenati Berber and other Northern Berber varieties, and other elements which could have been historically informative. Of the forms on the list which Korandje shows, 136 are shared with Songhay and only 11 of the forms which it shares with Tasawaq are not originally from Songhay. In all Korandje shares 16 non-Songhay plus non-Arabic forms on the list with at least one of Tagdal, Tadaksahak or Tasawaq but not with Songhay. 35 Korandje forms are only shared with Songhay.
What we can also do with the data available is examine the contents of the other Northern Songhay wordlists in the collection assembled by Rueck and Christiansen. It is manifest first of all that lexically Tadaksahak and the two Tagdal varieties – Tagdal and Tabarog – presented are extremely similar, though not identical, and that unsurprisingly the Tagdal varieties show the greatest similarity with one another. Assuredly the Tagdal varieties and Tadaksahak go back to a common ancestor. We may also compare these forms with the Tasawaq data provided. Of the Tasawaq forms on the list 220 are certainly shared with Tagdal varieties and with Tadaksahak (and 2 more may be), 4 with Tadaksahak alone and 12 with one or both forms of Tagdal (especially Tagdal proper) but not Tadaksahak. 28 forms are common to Tadaksahak and Tagdal (either Tagdal proper, Tabarog or both) but are not found in other forms of Songhay.
A selection of high-frequency lexical items in NS languages, with data on their origins, can be found in Table 24. Forms which are not etymologized are derived or continued from Songhay. Tagdal data are taken from Benitez-Torres (n.c.), Tadaksahak data from Christiansen-Bolli (2010), Tabarog from Rueck and Christiansen (1999), Tasawaq data from Kossmann (n.d.), Emghedeshie data, unphonemicized and thus in plain type, from Barth (1851) and Korandje data are from Souag (n.d.). Words in bold are Amazigh (Berber); underlined words are from Arabic.
Table 24 Lexical similarity and variation in Northern Songhay languages
In this section data from Tagdal, Tadaksahak and Tasawaq are compared with data from several stable mixed languages in an attempt to see the parallels between the sources of elements in each language. Table 25 gives a general conspectus of language mixing (including NM and VM – nominal morphology and verbal morphology) and Table 26 below presents a picture of the origins of subsets of structural elements in various mixed languages.47
[INSERT TABLE 25 HERE]
Table 25. Sources of various elements in four stable mixed languages, plus Berbice Dutch, Saramaccan and Northern Songhay lgs.
Mixed: Bantu (incl. Swahili), West Rift, Eastern Cushitic, Maasai
85-90% Spanish; rest Ecuadorian Quechua
Predomi-nantly Attuan Aleut; some Russian
French, Plains Cree, many English loans
Dutch, Eastern Ijo, Arawak and Guyanese
English, Portuguese, Dutch, Kikongo, Gbe
75% Berber, depend-ing on domain
75-80% Berber, depending on domain
West Rift Cushitic predominates; Bantu and Maasai elements
Almost all Aleut
52% Cree; 47% French, 1% English
60% Dutch, 35% Ijoid
50% English, 35% Portuguese. 15% Gbe, Loango Bantu, Dutch, other
47% Berber, 48% Songhay
49.5% Berber, 48% Songhay
61% Song, 34% Berber
Tanzanian Bantu with three ‘exotic’ phonemes /x ł ʔ/
10a time adverbs Unk Ru;Al Sp Cr(Fr) EnPo* Du;EI B B B
10b phasal adverbs Unk Al Sp Cr En* Du Ø Ø Ø
11a 'abstract' adpositions Ban Ru Q Cr,Fr Po EI;DU S/B S/B S/B
11b place adpositions Ban Al Q(Sp) Cr,Fr EnPo* Du S S S
11c time adpositions Ban? ? Q Cr,Fr EnPo* Du S S S
12 dative word ‘to’ Ban? Ru Q Cr,Fr En Du S S S
13a instrumental word ’with’ Ban Ru Q Fr En Du S S S
13b 'and' linking NPs Ban Ru Sp;Q Fr En* Du S S S
14a 'and' coordinating Ban Ru Sp;Q Cr,Fr En Du S S S
14b 'or' Cush? Ru ? Fr En Du S S S
14c 'but' Cush Ru ? Cr Du Du B B S
14d other coordinators Ban Ru Sp;Q Cr,Fr En Du S/B S/B S/B
15a most complementisers Ban? Ru Sp Cr En* EI Ø Ø Ø
15b 'if' Cush Ru;Al Sp Cr En Du S S S
15c 'because' Ban Ru Sp;Q Fr En Du B B B
15d 'in order to' Swa Ru ? Cr En* Du;EI Ø Ø Ø
15e 'when' (temporal) Swa Ru ? Cr En* Du S S S
15f other subordinators Swa Ru Sp Cr,Fr Du? Du S/B S/B S
16a copulas Ban Al Q Cr,Fr En* Du S S S
16b 'to have' Cu-Ban Al Sp Cr En Du S S S
17a INFL morphemes Ban Ru Q Cr Ø Du;EI S S S
17b modal verbs Ban? Ru ? Cr,Fr Du Du;EI Ø Ø Ø
17c causative Ban Al Qu Cr En EI B B S
18 negative particles Ban Ru;Al Sp;Q Cr,Fr En Du;EI S S S
19a definite articles Ø Ø Ø Fr En Du Ø Ø Ø
19b indefinite article Ø Ø Sp Fr En Du Ø Ø Ø
19c demonstratives Cush Al(Ru) Sp Cr En Du S S S
19d other determiners Cush Al Sp Cr,Fr En Du Ø Ø Ø
20 plural NP markers Ban Al Q Cr,Fr Ø EI B B S/B
21 question particle Ø Ru Ø Cr(
22 deictic adverbs Unk Ru Sp Cr Po EI+Du S S S
23 manner adverbs Unk Ru Sp Cr En EI B B ?
24 focus particle ? Ru Q ? Gb ? S S S
* If by this you mean the POSS marker n, this is probably Berber. If it means the order of the NP constituents (Owner n owned), this is Songhay.
Table 26 Some structural features in stable mixed languages and mixed-lexifier creoles.
‘Quantifier’ nouns are items such as a lot of. ‘Abstract’ adposition include items such as at. INFL morphemes indicate tense, aspect, person and number on finite verbs. The focus particle is used in some languages to emphasis one argument (usually a nominal) over others.
Of the languages listed in Tables 25 and 26, Ma’á (combining Pare Bantu and various Cushitic languages; Tanzania; Mous 2003), Media Lengua (Quechua and Spanish; Ecuador; Muysken 1997, Gómez Rendón 2008), Mednyj Aleut (Western Aleut and Russian: Commander Islands, Siberia; Vakhtin and Golovko 1990) and Michif (Plains Cree and Canadian French; Manitoba and North Dakota; Bakker 1997, Laverdure and Allard 1984) are stable mixed languages. Meanwhile Berbice Dutch has its inflectional morphology and much of its basic lexicon from Eastern Ijo while most of the rest of the basic vocabulary is from Zeeuws Dutch (Kouwenberg 1994), and Saramaccan is a mixed English-Portuguese–Gbe creole of Suriname (Good 2009).
In each case the language which has provided the bulk of the morphosyntax is listed first, followed by the language(s) which provided the bulk of the basic lexicon. Tagdal (TG), Tadaksahak (TD) and Tasawaq (TS) appear in the final three columns, and data are provided for as many fields as we have data for. S stands for Songhay, while B and A stand for Berber and Arabic respectively. NM and VM are Nominal and Verbal Morphology respectively.
As we have more historical information for Saramaccan than for some other creoles, we have provided information for the Saramaccan column as to the extent to which the forms used to express concepts on the list below indicate the restructuring or redeployment of elements (from whichever source they may come) in Saramaccan structure. Thus forms that are asterisked, such as En*, derive from English but that because of processes of grammaticalization and morphosyntactic/semantic change the form is not used in Saramaccan in the sense in which it is used in English. Cases of grammaticalization by reason of morphosymtactic change of what were previously only lexical items or of semantic change in function words are much less prominent in the other languages surveyed.
According to our findings, as supported by the data displayed in Tables 25 and 26, both the nomadic and sedentary varieties of Northern Songhay are essentially Songhay languages in respect of both basic lexicon (nominal and verbal) and basic nominal and verbal structure, with varying but always very large quantities of Berber features. These include forms such as ‘because’ and ‘but’, while the morphosyntax of possession and plurality in NS nouns seems to resemble Berber as much as it does Songhay. Although both the plural affix in sedentary Tasawaq and the possessive marker can be traced to Songhay, most nouns in NS languages come from or through Berber and bear the marks of Berber morphology. Similar uniformity of origin of elements in a sentence cannot be claimed for the four stable mixed languages cited. It would be difficult to create a similarly etymologically consistent sentence in the case of Berbice Dutch.
The profiles of Northern Songhay languages – especially the nomadic Tadaksahak and Tagdal – are seen to be different from those of other ‘mixed’ languages. They seem to follow a pattern which we will call core-periphery languages, in which the ‘core’ of the languages is Songhay but the peripheral features (including anything from 50%-75% of the vocabulary) is Berber or of other origins. (In short, this reflects the usual profile of patterns of borrowing in non-mixed languages as well, including English, Berber and Songhay languages, because in each of these the most used morphemes – including the bulk of the items on the Swadesh lists - tend to come from the genealogical ancestor of the language, while the less-used morphemes are very often absorbed or copied from other languages.) What is remarkable about NS languages is the thinness of the genealogical core – a few hundred morphemes – when compared with the profundity of the penetration of these languages’ lexica by non-core and thus peripheral elements from other linguistic traditions.
In terms of the distribution of sources of elements in these languages, one possible close parallel to this set of Core-Periphery mixed languages is with Berbice Dutch, though NS languages, unlike Berbice Dutch, do contain a couple of numerals from the language which furnished the material of their core inflectional morphological systems.
Another possible parallel to the patterns of Berber-Songhay mixing in Tadaksahak and Tagdal can be found in the variety of Montagnais or Innu-aimun (a Cree-complex variety belonging to the Algonquian family), spoken in Betsiamites, Quebec. It has been strongly affected by local forms of French, especially in the speech of members of the younger generations. Lynn Drapeau has studied this language extensively and describes the situation brought about by Montagnais-French code-mixing in Drapeau (1994, 1995). Montagnais had previously borrowed heavily from French, integrating all borrowed nouns with /le-/ which derives from the French definite article, but which serves to mark off loan nouns in Montagnais and which does not carry any sense of definiteness. French verbs are borrowed in their infinitive forms and integrated with Montagnais tuutam ‘to do’. Other forms – pronouns, demonstratives, et cetera – are inherited from Montagnais; Betsiamites Montagnais is less ‘mixed’ than Michif.
The need to express concepts for which Montagnais previously did not have labels is great, with the result that many items of French origin (and indeed the first layers of French loans) in Montagnais are acculturational borrowings. But the amount of borrowing in people’s speech progresses from older to younger age groups (as do the number of French noun phrases and prepositional phrases in everyday use). For such speakers a Montagnais-French code-mixing lect is now the default Montagnais lect and the only one of which younger people have productive command (though they do not appear to mix Montagnais words into their French, and any French verbs borrowed into Montagnais are still incorporated by virtue of the use of an infinitive with an inflected Montagnais verb ‘to make’). Drapeau (1994) reports that when they were shown 100 pictures of objects, adults named the terms in Montagnais 92% of the time, high school students 72%, middle school students 62% and 4th grade children only 48%. Of the time; children simply did not know the original Montagnais words for things such as ‘waterfall’ or ‘neck’. This suggests that lexical erosion has proceeded apace in Betsiamites Montagnais and that the Montagnais language of speakers born after say 1985 is a lect which is full of code-mixing and with a depleted reservoir of Montagnais lexicon, even if Montagnais structure prevails; already according to Drapeau’s work the bulk of nouns which were recorded in a set of radio broadcast corpora in Montagnais from the 1980s were items from French, even though many of the nouns had Montagnais equivalents.
Code-mixing alone will not create a mixed language, creole or not, but stable and stereotyped patterns of code-mixing plus a high degree of lexical erosion (especially erosion of the noun lexicon), coupled with a tendency to borrow firstly acculturational and later on other lexicon, may create a mixed language which becomes the primary lect for some speakers. This may (as is the case with Montagnais and French in Betsiamites) coexist with the source of its replacive lexicon. What is happening in Betsiamites in the 20th and 21st centuries may well have happened in the Sahel some generations ago, before linguists were around to observe the fact.
Thomason (2003) examines the mechanisms of contact-induced linguistic change, finding seven. They are code-switching (intersentential and intrasentential), code alternation, passive familiarity, ‘negotiation’ (in which speakers make linguistic changes which attempt to replicate what they think happens in a second language), second-language acquisition strategies, bilingual first-language acquisition and change by deliberate decision. On the basis of the evidence that we have, code-switching, second-language acquisition strategies, and also bilingual first-language acquisition (certainly in previous generations), have had parts to play in shaping these languages. The combination of mechanisms which gave rise to the Northern Songhay languages may not necessarily also account for the genesis of other mixed languages.
Some other remarks on the broad topic of multilingualism are in order. Speakers of these languages are often multilingual, many speaking a Tuareg variety (in addition to one or more of Hausa, Fulfulde, Arabic and French). However, with the exception of Tadaksahak speakers, who reside in a region where Songhay is the LWC, most speakers of NS languages do not speak mainstream Songhay languages. Furthermore, the culture shared by speakers of these languages, Islamic in orientation, includes a number of realia and concepts which typically have Arabic-derived names in all the languages of the area, and nowadays there is also a topsoil of loans from French (e.g. Tadaksahak lakol ‘school’). There is also the use of Berber languages, including Tuareg, as sources for loans in the languages of this area, so that the presence of Berber loans in Songhay should not surprise us. Finally, in these languages (as demonstrated in Christiansen-Bolli 2010), there is a small stratum of words of Songhay origin which are the names of certain artifacts (such as fendí “winnowing fan”) which are typical of the material culture of the area, and which are the kinds of words which Tamajaq and other Tuareg languages might borrow from Songhay as loanwords.
One further note on mixed Berber input. It would be easy but unwise for us to assume that the Berber component in Northern Songhay languages is all from Tamajaq or other Tuareg varieties. In fact, lexical evidence suggests the opposite. As pointed out in section 1, Souag (2014) in his review of Christiansen-Bolli (2010) shows that many of the Berber but non-Songhay and non-Tuareg elements in Tadaksahak originate in Western Berber, a subgroup comprising Zenaga of Mauritania and Tetserrét of Niger. This is the most divergent branch of Berber, and there may have been other members of this branch which have since been absorbed by other languages. Stray elements such as the few Dogon words found in Korandje may even have entered Northern Songhay at this time as part of a loan stratum within Western Berber, and Berber elements occurring further south in Southern Songhay languages need not always be from Tuareg varieties. The availability of fuller resources for Northern Songhay languages would make these questions simpler for us to answer – for all we know they may have entered Korandje or other languages as part of the Songhay component.
Of the Western Berber languages, Tetserrét is the likelier source than distant Zenaga, but our lexical sources on it are not as extensive as we may wish, so we cannot be certain. The same is true for other Northern Songhay languages. Nonetheless it is likely that the earliest Berber component in Northern Songhay languages was from Western Berber, but that Tuareg elements, which have been acquired later, outnumber these.
Review of the data and conclusion
Table 27 demonstrates the key similarities and differences between the Northern Songhay languages in question, as well as between Northern Songhay, mainstream Songhay (Zarma and Gao) and Berber (Tǝwǝllǝmmǝt and Tayart).48 [INSERT TABLE 27 HERE]
Table 27: summary of basic structures in NS, mainstream Songhay and Berber languages
Default inflectional subsystem
Default derivational subsystem
All three Northern Songhay languages have similar inflectional subsystems of Songhay origin. However, the nomadic varieties, Tadaksahak and Tagdal, both have default Berber derivational subsystems. These include a series of affixes – the causative s-, the reflexive nǝm- or the passive tǝw- – which attach exclusively to verb roots of Berber origin. Neither Tagdal nor Tadaksahak have any known productive Songhay derivational affixes.52 Tasawaq only has the causative suffix -nda, which affixes onto Songhay roots. The few recorded cases of Berber causatives in Tasawaq seem to have been borrowed ‘as-is’ and are not representative of productive processes, with the possibility, as yet unattested, of underived y- verbs possibly taking Berber causatives. Tasawaq lacks passive or reflexive affixes.
All three languages have certain syntactic structures in common, with similar functions to those of the passive and reflexive prefixes. These include the functional-inverse construction, and the construction using ʃarayen, tʃaːren or siraːyen.53
With these data in mind, as well available historical data (see section 1.1), we can begin to come to a few tentative conclusions about the origins of the three Northern Songhay languages in question. First, given the ethnic and cultural differences between the peoples and the languages in question, it seems that Robert Nicolai’s (1979) division of Northern Songhay into sedentary and nomadic branches seems reasonable from a grammatical standpoint.54
Second, it seems likely that the structures of all three languages in view were originally the result of bilingualism. Both the Songhay and Berber portions are too well-formed and recognisable as coming from either Songhay or Berber, to be the result of any kind of imperfect learning.55 In any case, with these two observations in view, we can begin to guess at some scenarios that may have possibly led to the development of Northern Songhay languages.56 We see two possible scenarios, presented below in Figures 2 and 3.
[INSERT FIGURE 2 HERE]
Figure 2, Scenario 1: NS languages developed gradually, via normal parent-to-child transmission, from Proto-NS
Proto NS (Azawagh region)
Tadaksahak Tagdal Tasawaq Emghedeshie (extinct)
In Scenario 1, all Northern Songhay languages originated gradually, via normal parent-to-child transmission. In this scenario, all three Northern Songhay languages in question are descended from an ancient vehicular Songhay language that existed in the region, which over time would have acquired Berber features and developed into Proto-Northern Songhay (Souag 2009), for lack of a better term.57 If so, then Proto-Northern Songhay most likely at one time possessed both Berber and Songhay derivational forms. Subsequently, NS languages developed into the present nomadic and sedentary subdivisions. Nomadic Tagdal and Tadaksahak would have lost Songhay derivational processes, leaving only kanda ‘cause to fall’ in modern-day Tagdal, along with some syntactic constructions, only some of which exist today in Tadaksahak. Sedentary Tasawaq would have lost most of its Berber derivations, leaving a default Songhay system with some essentially fossilised, Berber forms. All of this hinges on whether the derived and underived Berber verbs (see Table 15 and discussion above) are cases of historical Tasawaq or recent innovations by young bilinguals.
[INSERT TABLE 28 HERE]
Table 28: scenario 1, gradual development of NS via normal parent-to-child transmission, from Proto-NS
Songhay derivational subsystem
Proto NS (LWC)
Proto-Tasawaq (sedentary, vernacular)
(gradually lost, only few “fixed” cases of causative left)
Proto-Tagdal/Tadaksahak (nomadic, vernacular)
(gradually lost, only one case of Songhay causative left in Tagdal)
This contrasts with the second possible scenario.
[INSERT FIGURE 3 HERE]
Ancient Songhay LWC
\\Figure 3. Scenario 2: nomadic varieties developed in a short period of time, sedentary variety actually a heavily berberised mainstream Songhay language.
Tadaksahak Tagdal Tasawaq Emghedeshie (extinct)
In scenario 2, the nomadic NS forms originated abruptly, in the course of one generation, not via parent-to-child transmission. Songhay was the lingua franca of the region of modern-day Mali and northern Niger. Nomadic Tagdal and Tadaksahak would have originally developed over a short period, perhaps single generation, from the code switching patterns of relatively bilingual Tuareg-Berbers. Unfortunately, we cannot know with exactitude what would have motivated the generation in question to adopt their speech patterns into a new language. However, it probably would have had to do with the establishment of a separate identity, similar to the social function of mixed languages today (Thomason 2001).
If true, then nomadic Tagdal and Tadaksahak at one time most likely would have had both Berber and Songhay derivations, as well as some Songhay syntactic constructions, some of which remain to this day. Over time, Songhay derivational affixes would have fallen out of use until all that is left today is kanda ‘cause to fall’ in Tagdal, as well as some relevant syntactic constructions which may no longer exist in Tadaksahak.
Further, if this were true, then sedentary Tasawaq might essentially be a mainstream Songhay language, as Lacroix (1968, 1975, 1981) suggested – a vernacularised, admittedly heavily-berberised, variety descended from the ancient lingua franca of the region. If so, then the Berber derivational forms in Tasawaq in section 2 above would be instances of more recent borrowing of foreign vocabulary.
[INSERT TABLE 29 HERE]
Table 29 Scenario 2, abrupt genesis
Songhay derivational subsystem
Ancient Songhay LWC
Tasawaq (sedentary, vernacular)
(instances of Berber causatives only recent innovations by bilinguals)
Proto-nomadic NS (vernacular)
(gradually lost, only one case of Songhay causative left in Tagdal)
Finally, the question of the development of Korandje, the “outlier” (Souag: fc) among the Northern Songhay languages, still remains. How different is it from mainstream Songhay languages, as well as from the various NS languages? The evidence given here, especially the lexical evidence, suggests that the differences are considerable and may be enhanced by the large-scale absorption of Arabic words in Korandje. Further comparisons of vocabulary and grammatical structures between Korandje and NS languages of the Azawagh valley could resolve many of these questions; the ongoing work on Korandje by Lameen Souag will be invaluable in this regard. A common origin for all these Core-Periphery mixed languages cannot and should not be ruled out.
Adamou, Aboubacar. 1979. Agadez et sa région. [Études nigériennes 44]. Niamey, Niger: Institut de recherches en sciences humaines.
Alidou, Ousseina. 1988. Tasawaq d’In-gall: esquisse linguistique d’une langue dite «mixte». MA thesis, Univ. of Niamey.
Bakker, P., and M. Mous, eds. 1994. Mixed languages: 15 case studies in language intertwining. Amsterdam: IFOTT.
Bakker, Peter. 1997. A Language of our Own: The Genesis of Michif - the Mixed Cree-French language of the Canadian Métis.New York, Oxford: Oxford University Press. Oxford Studies in Anthropological Linguistics, 10.
Barth, Heinrich. 1851. Progress of the African Mission, Consisting of Messrs. Richardson, Barth, and Overweg, to Central Africa. Journal of the Royal Geographical Society of London 21. 130-221. Bender, M. Lionel.1996.The Nilo-Saharan Languages: A Comparative Essay. München: LINCOM-Europa. Benítez-Torres, Carlos M. 2005. Review of Nicolaï 2005. Journal of West African Languages. 32.1.: 188-191.
Benítez-Torres, Carlos M. 2009. Inflectional vs. Derivational Morphology in Tagdal (Northern Songhay): a Case of Language Mixing. In Selected Proceedings of the 38th Conference on African Linguistics. ed. Fiona McLaughlin, Matondo Masungu, and Eric Potsdam, 69-83. Somerville, Ma.: Cascadilla Proceedings Project.
Benítez-Torres, Carlos M. n.d. Tagdal lexicon.
Bernus, Suzanne and Edmond. 1972. Du sel et des dattes: Introduction à l’étude de la communauté d’Ingall et de Tegidda-n-tesemt. Niamey, Niger : Centre Nigérien de Recherches en Sciences Humaines.
Cancel, Lt. 1908. Etude sur le dialecte de Tabelbala. Revue Africaine, Nº 270-271, 302-347.
Champault, Dominique. 1969. Une oasis du Sahara nord-occidental, Tabelbala. Paris: CNRS.
Christiansen, Niels and Christiansen , Regula. 2002. Some verb morphology features of Tadaksahak: Berber or Songhay, this is the question. SIL Electronic Working Papers 2002-005. Dallas, Tex: SIL International. Christiansen-Bolli, Regula. 2010. A grammar of Tadaksahak: a Northern Songhay language of Mali. Doctoral dissertation, University of Leiden.
Dimmendaal, Gerrit. 1992. Review of Nicolaï (Parentés linguistiques). Bulletin of the School of Oriental and African Studies 55: 73-5.
Dimmendaal, Gerrit. 1995. Do some languages have multi-genetic or non-genetic origin? An exercise in taxonomy. Nilo-Saharan Studies, ed. by Franz Rottland and Robert Nicolaï, 357-372. [Nilo-Saharan 10]. Cologne : Rüdiger Köppe.
Drapeau. Lynn. 1994. Bilinguisme et érosion lexicale dans une communauté montagnaise, Langues et sociétés en contact, edited by Pierre Martel and Jacques Maurais, 363-376. Tübingen: Niemeyer.
Drapeau, Lynn. 1995. Code-switching in caretaker speech and bilingual competence in a native village of Northern Quebec. International Journal of the Sociology of Language 113: 157-164.
Ehret, Christopher. 2001. A comparative reconstruction of Proto-Nilo-Saharan. Cologne: Rüdiger Köppe.
Gómez-Rendón, J. A. 2008. Mestizaje lingüístico en los Andes: génesis y estructura de una lengua mixta. Quito, Ecuador: Abya-Yala.
Good, Jeff. 2009. Loanwords in Saramaccan. Loanwords in the world’s languages: a comparative handbook, edited by Martin Haspelmath and Uri Tadmor, 918-943. Berlin: De Gruyter.
Grant, Anthony P. 2002. ‘Fabric, Pattern, Shift and Diffusion: What Change in Oregon Penutian Languages Can Tell Historical Linguists.’ Proceedings of the Meeting of the Hokan-Penutian Workshop, June 17-18, 2000, U. of California at Berkeley. Report 11, Survey of California and Other Indian Languages, 33-56, edited by Laura Buszard-Welcher. Department of Linguistics, University of California at Berkeley.