Diedrich Hermann Westermann, a missionary and linguist, hesitated between assigning it to
Gur or considering it an isolate, and
Maurice Delafosse grouped it with
Mande. At present, Songhay is normally considered to be
Nilo-Saharan, following
Joseph Greenberg's 1963 reclassification of
African languages; Greenberg's argument is based on about 70 claimed
cognates, including
pronouns. This proposal has been developed further by, in particular,
Lionel Bender, who saw it as an independent subfamily of Nilo-Saharan.
Roger Blench notes that Songhay shares the defining
singulative–plurative morphology typical of Nilo-Saharan languages. As of 2011, he believes that Songhay is closest to the neighboring
Saharan languages and is not divergent. However, a Nilo-Saharan classification is controversial. Greenberg's argument was subjected to serious criticism by Lacroix, who deemed only about 30 of Greenberg's claimed cognates acceptable, and moreover argued that these held mainly between
Zarma and the
Saharan languages, thus leading one to suspect them of being
loanwords. Certain Songhay–
Mande similarities have long been observed (at least since Westermann), and Mukarovsky (1966), Denis Creissels (1981) and
Nicolaï (1977, 1984) investigated the possibility of a Mande relationship; Creissels made some 50 comparisons, including many body parts and
morphological suffixes (such as the
causative in
-endi), while Nicolaï claimed some 450 similar words as well as some conspicuous
typological traits. However, Nicolaï eventually concluded that this approach was not adequate, and in 1990 proposed a distinctly novel hypothesis: that Songhay is a
Berber-based
creole language, restructured under Mande influence. In support of this he proposed 412 similarities, ranging all the way from basic vocabulary (
tasa "
liver") to obvious borrowings (
anzad "
violin",
alkaadi "
qadi".) Others, such as Gerrit Dimmendaal, were not convinced, and Nicolaï (2003) appears to consider the question of Songhay's origins still open, while arguing against Bender's proposed etymologies. Greenberg's morphological similarities with Nilo-Saharan include the personal pronouns
ai (cf.
Zaghawa ai), 'I',
ni (cf.
Kanuri nyi), 'you (sg.)',
yer (e.g. Kanuri
-ye), 'we',
wor (cf. Kanuri
-wi), 'you (pl.)'; relative and adjective formants
-ma (e.g. Kanuri
-ma) and
-ko (cf.
Maba -ko), a plural suffix
-an (?), a hypothetical plural suffix
-r (cf.
Teso -r) which he takes to appear in the pronouns
yer and
wor, intransitive/passive
-a (cf.
Teso -o). The most striking of the Mande similarities listed by Creissels are the third person pronouns
a sg. (pan-Mande
a),
i pl. (pan-Mande
i or
e), the demonstratives
wo "this" (cf. Manding
o,
wo) and
no "there" (cf. Soninke
no, other Mande
na), the negative
na (found in a couple of Manding dialects) and negative perfect
mana (cf. Manding
má,
máŋ), the subjunctive
ma (cf. Manding
máa), the copula
ti (cf. Bisa
ti, Manding
de/le), the verbal connective
ka (cf. Manding
kà), the suffixes
-ri (resultative – cf. Mandinka
-ri, Bambara
-li process nouns),
-ncè (ethnonymic, cf. Soninke
-nke, Mandinka
-nka),
-anta (ordinal, cf. Soninke
-ndi, Mandinka
-njaŋ...),
-anta (resultative participle, cf. Soninke
-nte),
-endi (causative, cf. Soninke, Mandinka
-ndi), and the postposition
ra "in" (cf. Manding
lá, Soso
ra...) The Songhay languages are considered to be an independent family by Dimmendaal (2011), although he classifies
Saharan as part of Nilo-Saharan. ==Grammar==