Unusual borrowing patterns in Niger languages

Niger has about twelve indigenous languages, five major ones (Hausa, Songhay, Fulfulde, Tamajaq, and Kanuri) and seven minor ones (Arabic, Buduma, Gulmancema, Tadaksahak, Tagdal, Tasawaq, and Tubu). In Niger, Songhay is strongly differentiated into 3 major dialects, Kado Senni, Zarma Chiine, and Dandi. Tadaksahak, Tagdal, and Tasawaq are three mixed languages borrowing entire portions of their grammar and lexicon from Songhay and Berber languages (Benítez-Torres 2009, Sidibé 2010). Two important non-indigenous languages are French (the official language, inherited from colonization) and Standard Arabic.

My colleagues at the Department of History are always cautious about complaints heard from time to time that the colonial borders had enclosed people of various ethnic backgrounds. They would point out that in the history of the region, hardly any people lived in isolation from its neighbors and that typically, past empires and kingdoms (Songhay, Aïr, Borno, Sokoto, etc.), not unlike the modern states, comprised people of diverse ethnicity. Even when they were not sharing a common political fate, the various ethnic groups in the area always maintained strong commercial, cultural, scientific, and religious exchanges. This situation is well reflected in the relation between the languages.

The twelve indigenous languages represent three of the four Greenbergian African families: Afroasiatic (Arabic, Buduma, Hausa, Tamajaq), Niger Congo (Fulfulde, Gulmancema), and Nilo-Saharan (Kanuri, Songhay, Tubu). On this account, as one would expect, the languages can be very different. For example, some are strict basic SVO (example: Hausa), while others are basically SOVX (example: Zarma Chiine). Some of the languages have nominal classes (Fulfulde, Gulmancema), while some have a two-class gender system (Hausa, Tamajaq), and still others have no significant nominal classification (Kanuri, Songhay). Some languages have heavy nominal and verbal morphologies (Hausa, Tamajaq), while others have a simpler word structure (cf. Zarma Chiine, which is said to have just about 23 to 30 segmental affixes, depending on the author, see Issoufi Alzouma 1992).

Nonetheless, it is also a remarkable fact that the languages share a lot of features. As an illustration, let us look at the relationships between Hausa (the majority language, spoken by more than 80% of the country’s population) and Zarma Chiine (the second major language, spoken by 30% of the population; henceforth, ZC). A quick look at any ZC dictionary will turn out a substantial amount of lexical items shared with Hausa (excluding the obvious common Arabic loans). There are indications that some of these loans follow a rather non-trivial pattern. For example, Southern Songhay has a (partial, context-sensitive) /f/-/h/ alternation where ZC has the /f/ and all other dialects have /h/ (cf. ZC fûu vs. Kado hûu ‘house’). According to Nicolaï (1983), ZC would have retained the /f/ due to the reinforcing influence of Hausa. However, Hausa, too, has the /f/-/h/ alternation and an apparent problem for Nicolaï’s claim is that today ZC is in contact only with the Hausa western dialects, which have /h/ and no /f/ (or anything near it; the real alternation in Hausa is between the variants [f, p, ɸ] and [h]). Nicolaï’s claim nonetheless becomes plausible if one considers the fact that Hausa words implicated in the /f/-/h/ alternation were borrowed into ZC with the /f/ alternant (cf. bùfu ‘sack’, faadà ‘palace’, fansa ‘redeem’, fiili ‘open space’, fàtawcì ‘commerce’, fùulà ‘cap, responsibility, leadership’). One may first assume that ZC was borrowing Hausa words even before the differentiation of Hausa dialects through, amongst others, the /f/~/h/ alternation. However, this idea would not account for the fact that some of the borrowed /f/ words are new even in Hausa (cf. fàrantì/ pàrantì ‘plate’, fawdà ‘powder’, fànkàasu ‘pan cake’; the last two are clear borrowings from English). Another alternative account then would be that ZC speakers, somehow, go over the immediate Hausa /h/-dialects and borrow from the Hausa of Kano, which is an /f/-dialect but situated at about 800 km from Zarmaland. In this perspective, Standard (Kano) Hausa would exert its influence on ZC, just as it has been doing with other Hausa dialects even before colonization (Kano Hausa was the reference dialect well before it officially became the “standard” dialect during colonial times, see Abdoulaye 2009). It may be noted that Hausa words with a pandialectal non-alternating /h/ are borrowed into ZC with /h/ (cf. hàwka ‘violent spirit’ from Hausa hàukaa ‘folly’ and hayà from Hausa hayàa, itself from English hire).

Beside lexical items, the two languages also share some grammatical features, many of which surely are areal features. In both languages, negation marking is split. In Hausa, there is one marker for the subjunctive and another for the other TAMs. In ZC, the split is aspectual, with one marker for the perfective and another for the other TAMs. The two languages also have comparative constructions based on the verb for ‘surpass’, although in ZC this construction is in competition with a comparative based on the comitative phrase (so that ay gà koy habu ńdà Muusà means ‘I am going to the market with Musa’ or ‘I go to the market more than Musa’). In both languages, the comitative construction has also given rise to a causative construction where the verb and the erstwhile comitative particle are in a close-knit construction (cf. zaunar dà in Hausa and gorò ndà in ZC for ‘seat’, based on the verbs gòro and zaunàa ‘sit’). However, the construction is heavily restricted in most Songhay languages (implicating just three verbs) but it is being generalized in ZC, maybe under Hausa influence.

However, there is one shared feature between Hausa and ZC for which there is no obvious explanation. Parsons (1955) described in Hausa what he calls the “Abstract Nouns of Sensory Quality” (or ANSQ), which are semantically and formally characterizable in Hausa. Semantically, as the label indicates, the nouns refer to “…quality or attributes of people, animals or things that are perceptible by one or more of the senses”. Formally, the nouns are (a) disyllabic, (b) have a heavy initial syllable, (c) end with the long high vowel –ii, (d) have an all high tone pattern, and (e) are all masculine nouns. The phenomenon was considered so deeply characteristic of Hausa that when it was discovered in other Chadic languages it was taken as a West Chadic feature reflecting a genetic relationship (Haruna 1996). Some representative nouns in Hausa and Guruntum, a West Chadic language, are given in Table 1. Nonetheless, ZC also exhibits a very similar pattern, as can be seen in the table.

Table 1: Abstract Nouns of Sensory Quality in Chadic and ZC

Hausa Guruntum Zarma Chiine
coldness sanyii saanii yeeni ‘coldness’
toughness taurii sandi ‘toughness’
bitterness ɗwaacii ngyooŋi fotti ‘bitterness’
bad smell waarii waabii fumbi ‘bad smell’
sourness tsaamii Ɂaasii moori ‘sourness’
heat zaafii Ɂaanii konni ‘heat’
heaviness nauyii giisii beeri ‘bigness’
sharpness kaifii goobii gitti ‘acridness’
viscosity kaurii huulii doori ‘pain’
strength ƙarfii kwaamii jandi ‘length’
width faaɗii wulŋii kaari ‘glimmer’

The ZC forms, just like the Chadic forms, are not derived in any obvious way beside their semantic and formal similarity (i.e., there are no simpler bases from which to derive the forms). The short vowel in the ZC forms is probably due to the general tendency of ZC words to end in a short vowel (and Hausa borrowings are subjected to this tendency). It is clear that the ANSQs can no longer be taken as an exclusively (West) Chadic feature. As is clear from the table, it is not the individual words that may have been borrowed, but a lexical pattern, no matter how this may have happened.

The Sahel region, and in particular Niger, is linguistically diverse, with languages from different phyla. Nonetheless, they share many features in patterns of borrowing that attest to their high level of contact. This in turn attests to the shared history of their speakers.


Abdoulaye, Mahamane L. (2009). Mahzûratu-l-sâ’at, un manuscrit ajami en hausa: présentation et considérations d’ordre linguistique. In S. Moumouni & V. Pawlikova-Vilhanova (eds.). Le temps des Oulèmas: les manuscrits africains comme sources historiques, pp. 255-282. Niamey: AMU/IAU.
Benítez-Torres, Carlos M. (2009). Inflectional vs. Derivational Morphology in Tagdal: A Mixed Language. In M. Matondo, F. McLaughlin, & E. Potsdam (eds.). Selected Proceedings of the 38th Annual Conference on African Linguistics […], pp. 69-83. Somerville, MA: Cascadilla.
Haruna, Andrew (1996). Towards a comparative analysis of some lexical items and an unexpected form-meaning correspondence: A case of Hausa and some Southern Bauchi Area Languages. AAP 46: 119-136.
Issoufi Alzouma, Oumarou (1992). Etude lexico-sémantique du vocabulaire fondamental du zarma. PhD thesis, University of Montreal.
Nicolaï, Robert (1983). Position, structure and classification of Songay. In M.L. Bender (ed.). Nilo-Saharan Studies, pp. 11-41. Lansing, MI: Michigan State University Press.
Parsons, F.W. (1955). Abstract nouns of sensory quality and their derivatives in Hausa. In J. Lukas (ed.). Afrikanistische Studien, pp. 373-404. Berlin: Akademie-Verlag.
Sidibé, Alimata (2010). Procédés de pluralisation et marques du pluriel dans une langue dite “mixte”: le tasawaq d’Ingall. Nordic Journal of African Studies 19(2): 108-123.

2 thoughts on “Unusual borrowing patterns in Niger languages

  1. Thanks for posting this – good to see some discussion of the topic.

    A point of phrasing: Zarma did not “retain the /f/”, but rather innovated a change of *h > f before round vowels, as Nicolaï observes. All other Songhay varieties continue to contrast h and f before round vowels in inherited vocabulary (e.g. *hugu “house” vs. *futu “bad”).

    Etymologically, *fugula “hat” > fuula is reconstructible for proto-Songhay, on the basis of Korandje fugʷra, and the same word is found in Manding fugula (which, unlike Hausa, retains the original medial g), so it is not necessarily a Hausa loanword.

    The claim that “there are no simpler bases from which to derive the forms” is hardly obvious; Heath analyses the corresponding forms in related languages as reflecting an abstract noun suffix -i. The final vowel of the corresponding stative (adjectival) verbs usually seems predictable: -u after a round vowel, -∅ after n, -i otherwise – but the absence of assimilation in the nominal forms suggests a suffix, as do forms not fitting this template such as dúkùr “be indignant” > dúkùrì “indignation”. Whether derived or not, the consistently transparent alternation of these abstract nouns with related adjectives contrasts with their generally non-derived status in Hausa, although even in Hausa a few ANSQs are related to corresponding adjectives or nouns, e.g. ɗaci “bitterness” ~ ɗata “bitter”.

    As for the comitative to causative development, a complicating factor is that the productive causative suffix throughout all of Northern Songhay, including Korandje, reflects *-nda rather than *-ndi; probably the Zarma case represents an independent parallel development, as you imply, but the possibility of archaism needs to be ruled out. It would be interesting to learn more about the Zarma case.

    • Thanks Lameen Souag for your remarks. On your first point I guess Nicolaï held various views on the f/h alternation (I indeed cited an old 1983 article). Either way I guess is okay for the main point of the article.

      About the fuula-word, I now think that your remark is relevant, the situation may indeed be more complicated and the point of the contribution may have to rest on the other examples.

      As to the possible derived nature of the ZC abstract nouns, your remarks surely are relevant to the issue. But no matter the real situation in ZC, on the surface we see a word structure pattern that is shared between genetically unrelated languages and my question was how that came to be. I admitted to having no answer on that.

      Finally, on the “new” causative in ZC, it is very clear that contrary to the -andi causative suffix (which actually imposes a particular tone pattern on the verb root), the causative ndà is demonstrably a free particle, which can plausibly be related to the comitative ndà particle.

      Thanks again, hope the exchanges will go on on these matters.

Comments are closed.