Making a fish of a friend. Waris: The secret language of Arab Koranic school students in borno

Full text

MAKING A FISH OF A FRIEND. WARIS: THE SECRET LANGUAGE OF ARAB KORANIC
SCHOOL STUDENTS IN BORNO
JONATHAN OWENS and JIDDA HASSAN
1. The Koranic School in Borno
In the central and eastern Sudan, the tradition of Islamic learning is probably held
nowhere in such high esteem, and no area is so respected for this tradition, as in the cultural
area encompassed by the former Kanem-Borno empire, around Lake Chad. As far back as the
nineteenth century European explorers (e.g. Rohlfs 1874 (1984): 213, Nachtigal 1980: 161)
commented (not always flatteringly) on a flourishing system of Koranic learning in Borno.
Still today, Borno is noted within Nigerian Islam for its tradition of Islamic learning, its
historical depth and soundness of content. It is not uncommon for those seeking a solid
islamic education elsewhere in Nigeria to spend a number of years training with koranic and
islamic teachers in Borno. This esteem is based on a level of intellectual attainment which
reaches into the highest stages of Islamic scholarship. Arguably, however, it is the rather
simple basis of the system which provides, and has provided, the foundation on which the
entire ediface rests. This foundation is the memorization of the Koran, the first step in the
traditional islamic education of Borno. The Koran is the basis not only of Islam as a religion,
but of islamic culture in general, having a central role in the legal system and in many islamic
sciences, so a thorough knowledge of the Koran opens the door to other disciplines.
The Koran is a fairly long document, 114 verses, some 400-600 pages in a typical,
unannotated edition. Its memorization implies long years of study, and this in turn implies an
educational and social system to support this effort. The formal steps in the memorization of
the Koran in Borno have been described by Mustapha (1987).1 They culminate in a student
writing from memory the entire Koran (stage by stage, not all at one go), and qualify him for
continued Islamic study.
The social organization around which this task revolves is the sangá (or sangaaya), the
koranic school.2 The sangá is headed by a teacher (malam or gooní, see below). Virtually any
association of teacher and students constitutes a sangá, whether students are many or few,
whether or not teachers have reached the final stages of Islamic study. Some sangá can be
very large, headed by an established islamic scholar who will have hundreds of students and
tens of teachers working for him. At the other extreme a teacher, usually one who himself has
not finished his memorization studies, may have only 3 or 4 students under him. In all cases
the course of study is informal in that students work through the Koran at their own pace,
though there are daily, communal recitation exercises with the teacher, at least with younger
students. Having memorized a particular piece, students present themselves to the teacher, and
if their performance is adequate they proceed to the next part of the Koran.
1 Mustapha describes 5 formal stages students pass through in memorizing the Koran. However, it appears that
not all stages are equally important, and Mustapha’s description does not provide an adequately differentiated
picture about how the system functions in reality, as opposed to the theoretical ideal.
2 sangá is a Kanuri term. Nigerian Arabic uses this term (see “Loanwords...”) as well as makaranta, which is of
immediate Hausa origin (Hausa in turn has the term from Classical Arabic; see n. 6).

Waris
2
Students may begin study as early as 5 or 6 years of age. If established schools exist, as
they do in abundance in Maiduguri, young children will usually be sent to a local school. In
rural areas, however, there are relatively few villages with Koranic schools,3 and in this case a
father may choose to send a son (never a daughter) to a school. On one basis or another he
chooses a teacher, deposits his child, leaving him to the care of the teacher, and departs.
The choice of attending a koranic school as a full-time or part-time student is more than
one of convenience, however. Part-time students are usually sent to a koranic school largely
with a specific goal in mind, that of learning the Koran. The task of full time students is
broader than this. Their attendance at the school is seen as a challenge, a way of living that
serves to build their character. It is the full-time students (and their teachers) who form the
core of the traditional koranic school system. The term muhaajir (pl. muhaajiriin) (see below)
“koranic school student” refers only to this group, and as will be seen in section 6 below, the
different status of the students as full- or part-time is directly reflected in the knowledge of the
secret language.
The full-time student lives alone with the teacher and with the other students who were
enrolled in the school in the same way as himself, or are from the area where the school is
located, or sought out the teacher on their own. Even students who begin studying at their
neighborhood school will often, when they get older, begin studying itinerantly at schools in
outlying rural areas.4 However the students came together, they find that sangá life has a
character of its own, which can be likened to that of a family. The teacher, often married and
with his own children, supervises the progress of his students, while the students, besides
learning, provide labor, particularly farm labor in rural areas for their master, and beg daily for
food. The students may stay with a teacher for any length of time, weeks, months, years. Not
infrequently the teacher will move to a new location, taking his students with him. Changes of
scenery are said to promote Koranic memorization; the schools are itinerant, hence the name
of the students NA muhaajir-iin “those who emigrate”, = K majirí (<Ar haajara “emigrate”).
Not infrequently as well, students may seek out individually or together a new teacher, and
sometimes a teacher may abandon his students, leaving them to fend for themselves until they
find a new one.
Informal questioning – oral interviews with Kanuri teachers (Owens 1995) and taped
interviews for the NA corpus – indicates that the student body in Koranic schools, except
perhaps the very large ones, tend to be ethnically homogeneous. Kanuri speakers tend to
congregate with Kanuri speakers, NA speakers with other speakers of Nigerian Arabic. This is
true of the groups we have had contact with in the course of collecting data for this study,
composed to over 90% of native Arabic speakers (see section 2).
It is clear that the Koranic school students, particularly the full-time, itinerant ones, the
muhaajiriin, form a distinctive social sub-group. Their status is a recognized one throughout
Borno and further afield. In the Nigerian Arabic community it is often said that a student who
has studied in a koranic school is one who cannot go morally bad. Regardless of whether or
not a student succeeds in memorizing the entire Koran (most do not), the moral discipline
derived from living in a sangá is seen as an end in itself. The students’ regime dictates that
they come to form a particularly closed-knit and cohesive group. They eat, study, work, beg
3 Of the 19 villages used in Owens’ sociolinguistic study (see “Loanwords”), 6 reported having Koranic schools
in them at the time of the visits.
4 For a full-time student, residence in a small village is a standard part of the curriculum. There, away from
distractions, they can learn the Koran with facility.

Waris
3
for food and sleep together over a period of months or even years. One reflection of this close
social bond is in the secret language of the Koran schools, which will be described in the next
section.
Before proceeding to that, however, we should note that the focus of study is the secret
language of Arabic-speaking koranic school students. The system described above has been
generically applied to Borno as a whole. To what extent there are ethnically-specific
differences in the koranic school tradition in Borno is a matter requiring detailed study in
itself. It can be noted at this point, however, that the Arab koranic schools with which we have
contact follow, essentially, the system which has been described for the Kanuri, to the extent
that much of the terminology is taken from Kanuri. Various levels of attainment are referred
to with Kanuri names, for instance. A tegermí < K tegermí “one who excels in koranic
recitation” is a student who is in the process of memorizing the Koran, but has not yet done
so, while a kagnooma < K kagnoma “one who repeats Koran continuously” is a student who
has studied longer than a tegermí, though has not necessarily advanced further in terms of
memorization.5The highest level is termed gooní < K goní, a level attained when the koranic
scholar’s work has been scrutinized and approved by a panel of local koranic experts.
Attaining this title requires attendance at a school offering courses in various aspects of the
Islamic learning, called an ʔilm school (< Ar. ʕilm “knowledge”). There are also Arabic terms
for some of the stages of koranic learning. A beginner, for instance, is called an ambaddi < an
noun formative + bada “begin”, and informants suggested that a kagnooma could also be
termed a maalam. One who has memorized the Koran is a haafiz < Ar ħaafizṣ). The
formalized nomenclature reflecting level of scholarly attainment, however, is most
unambiguously expressed in terms of Kanuri origin. Furthermore, there is no native Nigerian
Arabic word for koranic school. The most common one (see n. 13, 40 in “Loanwords”) is
makaranta < H ma-karanta6 with sangá ~ sangáaya < K sangáya (ultimately < Latin
synagoga), being frequently used as well. Terms such as xalwa or kuttaab, names for
comparable institutions in Arabic countries were never heard. Historically, islamic learning in
Borno has been associated most closely with the Kanuri. This is not to say that there aren’t
(and weren’t) notable scholars from other ethnic groups (e.g. Arabs, Fulani). Nonetheless, it is
reasonable to suggest that Islamic learning in Borno, including the koranic school system, is
most closely bound up with the dominant cultural group of the Kanem-Borno cultural
tradition, Kanuri speaking peoples, and that their systems of learning have decisively defined
those of other ethnic groups in the area, including those of the Nigerian Arabs.
2. The Secret Language of Koranic School Students
In light of the preceding discussion it should come as no surprise that the social milieu of
the koranic school should be marked by defining forms of solidarity among its members. One
such artefact is a secret language, which will be described in the rest of this article.
5 The morphemic breakup is kag-no-ma “repeat-participle-agent”. NA tegermí is modern Kanuri teere-mí,
morphemically, “collect-son of”. Most Kanuri loans in NA with velars retain the velar (k, g), which are softened
post-vocalically in modern Kanuri (see “Loanwords: ). The Kanuri word teemerí refers to the practice of scholars
gathering and reciting the Koran, a teemerí being one who excels above others (we are thankful to Dr. Umaru
Bulakarima for this information).
We are indebted to our interviewees for their terminological interpretations. We have not compared the Arabic
interpretations of these terms with the way they are functionally embedded in the Kanuri system.
6 Morphologically the Hausa word is based on the root karantaa “read” + ma- “place prefix”. Ultimately the
Hausa root for “read” is taken from literary Arabic or Old Arabic, qaraʔa “read”. Nigerian Arabic, however,
takes makaranta from Hausa. That the borrowing did not go in the reverse direction is shown both by the fact
that the normal NA reflex of OA q is g, whereas in Hausa borrowings it most frequently appears as k (Greenberg
1947: 89), and the -nta suffix is inexplicable as a NA development.

Waris
4
2.1 Groups and data
The present data was collected both from individuals who are studying in a koranic
school, or studied in one, and from collective groups of koranic school students still engaged
in the study of the Koran. The groups are based in the Maiduguri wards of Gwange, also the
home of one of the author’s of this article, Dikkeceri, Gamboru and Old Maiduguri. A total of
six koranic school teachers were interviewed at various times on various questions. Five of
these are full-time teachers, and have fairly large schools, up to 60 students. One is a part-time
koranic school teacher who takes only a small number of students. All the teachers are native
Arabic speakers, and something in the range of 90% of their students are Arabic speakers as
well. To cite one concrete statistic, one koranic teacher interviewed has about 55 students of
whom about 40 are full-time itinerant scholars, the rest from families living in the
neighborhood. Of these, two are Kanuri, two are ‘Hausa’, and the rest Arabs.
The secret language is known as waris, based on the Nigerian Arabic (NA) root meaning
“writing on koranic slate”.7 It is a productive root from which verbs (waras “write a text,
buurwas looha “he is writing his koranic slate”), nouns waaris “on who writes on slate”,
waris “koranic writing”, etc. are derived.
Before beginning a small amount needs to be said about the tradition of koranic reading.
There is no single standard way of pronouncing the Koran. Instead there are a number of
variants, seven are the usual recognized number, which differ in a myriad of very minor
(usually phonetic realizations) ways.8 Different variants have been/are prevalent at different
times and places in the Islamic world. In Borno the dominant tradition is based on the
Madinean koranic reader Nafiʕ (d. 785), as interpreted by Warsh, who lived in Egypt (d. 812,
Nöldeke 1981 III: 174). This is the version which became dominant in North Africa
(Neuwirth: 110). Certain aspects of the secret language can be understood only by reference to
this version (see (10) below and n. 18), so a more adequate study would give detailed
background about the actual form of the Warshian reading tradition in Borno. This is outside
the scope of the present piece, however.
Finally it should be noted that koranic editions may differ slightly as to their verse
numbering. We have followed that used in the Ali translation. The koranic citations are given
in standard orthographic transliteration, which does not necessarily reflect the actual
pronunciation in a given reading tradition.
2.2 Secret languages
In the terminology of one standard reference work (Domaschnev in Handbook of
Sociology: 313-4) the material described here may be classified as a “jargon” or “argot”, the
distinction between the two not being clear-cut. As opposed to specialized registers, the
vocabulary of a jargon or argot applies to common, every-day concepts for which words
already exist. The jargon or argot is thus used by, and may have been developed by, a discrete
social group for which it serves as one identity marker. The words used are deliberately
secretive, and often are affectively marked, masking pejorative, humorous, or ironic meanings.
7 The etymology is not entirely clear. The nearest phonological cognates are either the OA root *wrθ meaning
“inherit”, with the sound shift of θ → s (attested in Nigerian Arabic), or wrs meaning “become yellow with age
(as a cloth or leaf)” (Lisaan al-ʕArab 2: 199, 6: 254). The semantic associations are not immediately obvious
from these meanings, however. The term could derive from warsh, with depalatization of š, though this is not a
regular or common phonological change in Nigerian Arabic.
8 The seven variants were compiled by the tenth century scholar Ibn Mujahid.

Waris
5
Bagemihl (1988) comes perhaps the closest to providing a complete typology of secret
languages, and two of the three types which he identifies he handles in considerable detail. He
distinguishes three basic secret language types. ‘Surrogates’ involve the substitution of
alternative voice-producing mechanisms to the larynx (e.g. whistle languages, drumming
languages), ‘ludlings’ employ regular phonological and morphological manipulation to alter
the form but not the content of the message (e.g. Pig-Latin), while ‘argots’ have their own
vocabulary (1988: 10ff.).9
Bagemihl terms secret languages languages within languages, which aptly recognizes
their adventitious reliance upon a “normal” natural language. This parasite-like quality may
manifest itself in various linguistic ways, in the modification of the phonetics, phonology
(segments, syllable structure), morphology, lexicon, etc of the basis language.10 In order to
briefly underline the fact that we are describing a type of language, Conklin’s (1959) study of
“linguistic play” serves as an initial orientation point, because the Hanunóo of the Philippines
whom he describes have no less than 11, non-mutually exclusive types of “secret” codes.11
Five types, for instance, involve the syllabic rearrangement of normal Hanunóo words via
prefixation, suffixation, syllabic transposition etc (ludlings). Four types employ phonetic
modifications (whispering, falsetto, surrogates), and two operate at the lexemic level,
involving either the substitution of one sound for another (hiiraw for hiirak “jealousy”, w for
k) or the use of words with the opposite meaning for that intended (“very bad” when “very
good” is intended, argots). Except for the two lexemic modifications, the modifications can
operate on all words in a sentence. The lexemic modifications are irregular and apply only to a
few items (Conklin, 296).
The range of Hanunóo word-play types crosses various linguistic levels: phonetic,
phonological/morphological and lexical, each with various sub-types. In contrast to this, as
will be seen in the next two sections, the secret koranic school language is essentially a lexical
phenomenon, governed by a set of definable though semi-regular (i.e. derivational-like) rules.
In the functional domain Conklin suggests that the different types are context-sensitive, in
part specific to individual groups. Phonetic modifications, for example, are used in particular
by young men during courtship rituals, whereas the syllabic modifications are employed for
amusement, or to keep conversations secret from eavesdroppers. The syllabic modification
thus appears to be a society-wide ability, though it is used most intensively by younger
speakers, somewhat to the displeasure of older ones. Generally speaking, the Hanunóo secret
codes appear to be universal to Hanunóo society in that they are either used by all speakers, or
used during a universal phase of social activity, namely courtship.
9 Bagemihl does not discuss argots in detail, though it appears that the way in which they form their vocabulary is
very heterogeneous.
10 Bagemihl points out that secret languages can be characterized by rules which are absent from their ‘source’
language, and hence are independent systems. What he does not show, however, is whether there is a secret
language which exists entirely independently of a natural language. In the realm of argots the secret language is
always devised relative to a natural language (see 4.2 below).
11 Conklin sees his article as a contribution to “word play”, a cover term which may be used to include punning,
epigrams, riddles, as well as secret languages. In fact, the forms Conklin describes are in their linguistic and
functional characteristics closer to secret languages than say punning or riddles in that they are functionally
restricted, and defined by special linguistic rules which operate only within the secret language or code. By
contrast, Arnott (1957: 390) notes that the Fulfulde proverbs, riddles and epigrams which he describes diverge
little from normal Fulfulde grammar. The same applies to a number of other references to “word play” found in
Conklin’s article.

Waris
6
The waris, on the other hand, is largely restricted to one group, full-time koranic school
students (see section 6). The secret language is a group marker in the way the Hanunóo secret
codes are not.
3. The Formation of Secret Words
In the present section we shall be concerned to describe the way in which the secret
language is derived, although “obfuscation” is a better term than “derivation”. The latter
suggests an orderly process of forming one meaning from another, usually by the addition or
change of morphological material, whereas in the present case the emphasis is on encoding a
commonly known meaning in such a form that its contents can only be recognized to those
initiated in its rules. Although the rules themselves can, trivially, be represented as changing a
word or phrase of one class into a word of another class (usually a noun), the relationship
between the source term and target term (see 4.1.1 below) is so unsystematic and indirect that
the process may better be likened to the arbitrary coining of new words. Since so many of the
terms are taken from the Koran we may speak of a source (usually the Koran) and a target, the
language (usually Nigerian Arabic) in which the terms are embedded. The terms “source” and
“target” recall a process of borrowing, which is an intended innuendo as will be seen below
(4.2 and n. 20).
3.1 The sources
We have divided the sources into three broad categories, some of which can again be sub-
divided: Koranic citations, acronyms and non-koranic referential material. We will discuss
these in turn, taking a typical member of each class as illustration. Further examples are given
in the appendix. In all we have collected a non-exhaustive list of 113 items.
3.1.1 Koranic citations
Nearly half of the source material consists of koranic citations or direct reference thereto,
some 53 items. Within this class four subdivisions may be made. In each case a particular type
of associative feature can be distinguished. The associative feature may take various guises,
semantic, phonetic, orthographic, as described below, but its effect is always to establish a
link between a form and referent, a form-meaning dyad. This link is established by what we
term a process of displacement. (1) represents a basic schema illustrating this displacement.
(1) Displacement
source1 source2 = secret word (target)
Form F F Fs
Meaning M M Ms
In displacement, a link is established between two sources. Each source consists of a form and
a meaning, i.e. are normal words or phrases. The secretness of the secret word consists in the
fact that it contains a meaning from one source and a form from another. In (1) we have drawn
the link from the top right to the bottom left, though it could equally have been drawn along
the other diagonal. We term the process displacement because in the formation of the secret
word a form and a meaning are displaced. Again it is arbitary whether the displacement is
defined as a form from source 1 displacing the form from source 2, a form acquiring a new
meaning, or a meaning from source 1 displacing a meaning from source 2, a meaning
acquiring a new form.
In this section we make a fundamental distinction between the nature of the sources. On
the one hand both sources may be Koranic, that is, from the text of the Koran itself. This will
be detailed in 3.1.1.1. On the other hand, one of the sources may be Koranic, the other non-

Waris
7
Koranic. The non-Koranic source may be a form-meaning dyad from Nigerian Arabic, or from
some aspect of Koranic learning, as rescribed in 3.1.1.2.
In all examples summarized in this section one of the sources must have some sort of
associative link to the cultural environment of Nigerian Arabs. This link takes on various
guises. The most common is for an equivalence to be drawn between a word or phrase in the
Koranic text and a word in Nigerian Arabic. An initial example will make the discussion more
concrete. The Koranic verse ka-ṣaaħibi l-ħuut “...like the companion of the fish” (Koran 68:
48) has two referents, ṣaaħibi “companion” and ħuut “fish”. An association is drawn between
the Koranic word ħuut and the NA word huut, both meaning “fish”. The principle of
displacement dictates that the secret language use not a form which directly encodes the
meaning of “fish” (i.e. ħuut), but rather a different word or words (see 4.1 below). In this case
the meaning “fish” is substituted for the usual meaning of the adjacent prepositional phrase,
ka-ṣaaħibi, what we term “syntagmatic displacement” (see 3.1.1.2). The form ka-ṣaaħibi
substitutes the meaning of “fish” for its own meaning “like the companion” (alternatively, the
meaning “fish” acquires the new form ka ṣaaħibi). The following then becomes a normal
expression in the secret language, where an “x” marks the secret word in the glosses.
(2) akal-naa ka-ṣaahibi aloom
ate-we x today
“We ate fish today”
(see 4.2 for phonology)
In the following we summarize the different types of displacement under two general
categories, cases where there are two Koranic sources, and cases where there is one Koranic,
one extra-Koranic source.
3.1.1.1 Two Koranic sources
The example just cited introduced the format summarized in this section. A similar
example is
(3) wa ʕan ʔalqi ʕaṣaa-ka
“now do thou throw thy rod” (28:31)
Here the referent is a “rod”, which normally is encoded in the Koranic form ʕaṣaa, associated
with NA ʔaṣa “stick, rod, staff”. The two parts of the equation, the two sources, consist of wa
ʕan ʔalqi and ʕaṣaaka. In this case the referent is displaced to the entire phrase wa ʕan ʔalqi,
which now encodes the meaning of “rod”. The resultant meaning of the secret word waʕan
alqi is equivalent the NA word ʔaṣa, a staff used, for instance, to herd goats. In the formation
of the secret word no new referential material is added.
Note here that on a purely formal basis it could be said that the sequence wa ʕan ʔalqi =
conjunction + subordinating conjunction + verb is, by a derivational process, changed into a
noun, → N. The very arbitariness of the derivational relationship, however, underscores the
point that this is a process of obfuscation more than one of normal derivation.
In other cases of this type the associative feature is a meaning, but in the formation of the
secret word extra semantic material is added. Such a case is
(4) ka-ʔanna-hum xušub
like-that-they wood
“as pieces of timber” (63:4)

Waris
8
The meaning derived from this verse does not have the referent “timber” (associated with NA
xaďab “firewood”), but rather “uncivilized”. The rough quality of timber is singled out. This
meaning, “uncivilized”, is displaced then to ka-ʔannahum.
The displacement summarized thus far involves syntagmatic movements, which may be
sketched as in (5), which conventionally follows the schema in (1). The displaced term is
placed in brackets in (5), and an arrow is placed before the term which displaces the other.
The input to the process are two Koranic sources. One of these sources, ħuut, has an
association with the NA word huut. A secret word is formed by associating a phrase adjacent,
hence the term “syntagmatic displacement”, to ħuut with the meaning of this term, “fish”. The
secret word results from this form-meaning displacement. The meaning “like a friend” and the
form ħuut get shunted aside in the displacement.
(5) Syntagmatic displacement
sources = secret word
Koran Koran
F ka-ṣaahibi ħuut → ka-ṣaahibi (ħuut)
| | |
M “like a friend” “fish” → “fish” (“like a friend”)
Generally speaking it appears that the formation of secret words from two Koranic sources
always involve the principle of syntagmatic displacement.
3.1.1.2 Koranic source + another source
In other instances one of the sources is non-Koranic, though within this universe of
sources we have found three different types, summarized in the following three sub-sections.
3.1.1.2.1 Phonetic, Nigerian Arabic
In the first case a phonetic association between a part of the Koranic text and a NA word
is the basis of the secret word. In
(6) fa-ɣulib-uw hunaalika
so-defeated-pl there
“so the great ones were vanquished there and then” (7: 119)
the phonetic form qulib-uw is associated with the phonetic form qilb-at, which in the local
Nigerian Arabic dialect means “she became pregnant”. The form fa-qulibuw thereby acquires
the meaning of “pregnant”. This case can be sketched as (7), where the substitution may be
conceived as moving the koranic form qulibuw to the meaning “pregnant”, which derives
from Nigerian Arabic.
(7) Phonetic association
Koran Nigerian Arabic = secret word
F fa-qulib-uw qilbat → fa-qulib-uw (qilbat)
M “defeated” “became pregnant” →“became pregnant” (defeated)
Note that the association qulibuw “defeated”~ qilbat “become pregnant” is purely a phonetic
one in contrast to the examples discussed in the previous where an association exists between
a Koranic form and meaning and an NA form and meaning (e.g. ʔaṣaaka ~ ʔaṣa).
This particular example is revealing in another way, as it shows the perceived phonetic
identity between the voiceless uvular stop of Classical Arabic (q) and the phonetically

Waris
9
identical phoneme q of Nigerian Arabic (see “Loanwords” e.g. (8)). The interesting point is
that NA q derives etymologically from proto-Arabic ɣ (e.g. qalla “grain” < OA ɣalla).12 The
“original” sound is shown in (6). Obviously the association is made on the basis of spoken
Arabic, for orthographically the two sounds are represented differently (qaaf vs. ghain, ¶ vs.
®).
In the previous section (3.1.1.1), both parts of the form-meaning dyad derive from the
Koran, whereas in the present case one half is from the Koran and one half from elsewhere, in
this case, Nigerian Arabic.There also exist composite equations where dyads are formed using
both methods simultaneously. One such example uses the verse
(8) ʔa lam tara ʔilaa allađ-iyna baddal-uw niʕmata llaah
Q not you-see to who-pl change-pl favor God
“Hast thou not turned thy vision to those who have changed the favor of God”?
(14:28)
The koranic word baddal-uw is the same verbal stem as the Nigerian Arabic word baddal
which has a specialized meaning in the nomadic Nigerian Arabic society. It refers to sellers of
yoghurt, who customarily exchange yoghurt for grain. The substitute ʔa lam tara replaces the
form baddaluw here, but in the Nigerian Arabic meaning of “yoghurt seller”. It is the phonetic
identity between koranic baddaluw and the NA baddal (3mpl baddal-o) which allows the NA
meaning of baddal to insert itself in the verse.
(9) Composite formation
Koran Koran NA = secret word
ʔa lam tara baddaluw baddal →ʔa lam tara
| | |
“don’t you see” “change” “trade” →“traders, especially yoghurt sellers”
The process can be conceptualized (analytically at least) in two steps. The meaning of
baddaluw is transferred syntagmatically to ʔa lam tara (A), but at the same time baddaluw
forms a phonetic identity to NA baddal (B), which allows the NA meaning to substitute for
the koranic one (C). Hence ʔa lam tara comes to have the meaning “yoghurt trader”, with a
koranic form associated with a NA meaning.
3.1.1.2.1 Orthographic association
In two instances the extra-Koranic associative feature is orthographic terminology from
Koranic studies. Various terminology exists for the specialized problems of Arabic (and
koranic) orthography, some of it specific to the different reading traditions. Particular
attention is given to the way the glottal stop (hamza) is written, and the various forms of the
long aa (alif), since these may change form according to context, in some cases in
unpredictable (lexically determined) ways. There are, for example, different terms for the long
aa. That which is written ا (in some contexts known as ʔalif mamduuda) is termed the “tall
alif” ʔalif ďawiila (cf. NA ďawiil “tall”) since it is written long.12F
13 The meaning “tallness” is
seized upon in the following example.
12 And OA qaaf appears as g.
13 One informant said that the ʔalif ďawiila and ʔalif ħamraaʔ or “red alif”, its name due to the fact that it is
written in red to signify a palatalized (ʔimaala) pronunciation, are alternative names for the same symbol. Both
would stand in contrast, presumably, to the ʔalif maqṣuwra, a long -aa, orthographic mark which appears at the
A
B
C

Waris
10
(10) wa saraabiil (ﻞﯿﺑاﺮﺳ)
and coats of mail
“and coats of mail” 16:81
The long aa in saraabiil is the “tall alif”. By figurative extension, the tall alif may be taken to
represent a tall person, hence saraabiil comes to mean a tall person. As in the previous
section, the form in this case comes from the Koran, but the meaning from outside it, in this
instance, from orthographic pedagogy.
(11) Orthographic association
Koran orthography = secret word
saraabiil → ʔalif ṭawiila (in saraabiil) → saraabiil
| | |
“chainmail” “tall alif” → “tall person”
(containing tall alif) figuratively, “tall person”
3.1.1.2.3 Alphanumeric association
A final type of extra-Koranic associative feature ( 4 items in all) plays on the link
between letters and numeric values. The letters of the Arabic alphabet each have a numeric
value associated with them.14 In the Warsh version of koranic reading often used in Borno,
combinations of letters are inserted at the beginning of every quarter of a memorization unit
(ħizb) to mnemonically indicate how many pauses and elongations occur in that verse. For
example, the two letters y = 10, k = 20, if written together at the beginning of a quarter would
indicate that there are 10 pauses and 20 elongations in the quarter.15
In one chapter (12:77) yk appears above the verse which begins,
(12) qaal-uw ʔin yasriq
said-they if he steals
“They said, If he steals” (12:77)
The combination yk may therefore be associated with stealing (NA sirig), and thereby acquires
the meaning of “thief” or “theft”.
(13) Alphanumeric association
koranic reference system Koran = secret word
yk16
→
(qaaluw ʔin yasriq) → yakkin
| | |
20-10 at 12:77 “if he steals” = “thief, theft” → “thief, theft”
In this type of association an extra-koranic form (yk) acquires a meaning derived in part
from the Koran.17 It is closely analogous to the syntagmatic displacement described in section
end of certain words, and which has the same shape as the letter y. Fuller clarification requires an investigation of
the koranic reading tradition in the area.
14 The system used in Nigeria follows the ʔabjad version of the alphabet, which differs from the sequence used
today in the Arabic world in most functions (see Daniels 1997: 29). In Jewish Moroccan secret language the
letters assume a similar numerical value, though may be used to replace numbers in everyday speech (Chetrit
1994: 527). We have not found found this usage in waris, so that, whereas the letter waw (= w) with a value of
“6” in the abjad numerology may stand for “6” in Moroccan Jewish secret language, in waris we have waaw
associating phonetically with referents whose usual phonological signs begin with w- (see 4.1.2, and Berjaoui
(1994) for numerical encoding of lexemes among Koranic scholars in Morocco).
15 This at least is how the alphanumeric system in the Warsh version of the Koran was explained to us by one
group of koranic school students.
16 This is pronounced yakkin, with a case ending -in and doubled k to conform to Arabic phonotactic rules.

Waris
11
3.1.1.1 in that there is an interplay of forms located on the same page (though not, strictly
speaking, part of the same text).
3.2 Acronymic association
The associations in this and the following section are not derived from the Koran, though
the linguistic basis of the associative features is frequently the same.
We have collected three examples of acronymic associations. Here each letter in a
sequence of four letters (vowels do not count in Arabic orthograhy) is associated with a
meaning, the entire acronym representing the meaning of the individual letters.
(14) raďabaja (rďbj)
r = ragayge “thin”
ď = ďawiile “tall”
b = baaďile “thin”
j = jiʕaana “hungry”
“tall, thin, hungry one”
The reference happens to be to an ugly woman.
We can say little more about these at this time. Our informants volunteered only that they
had learned the acronyms from others, but could give no further information about why these
letters, these combinations of meanings were chosen.18
3.3 Non-Koranic associations
The final category utilizes associative means described in the previous sections, but not
drawing on a Koranic source. Instead, the resulting secret word is either a new invention, a
word from Nigerian Arabic, or from some other part of the cultural surroundings of northern
Nigeria. For example, “grain, millet” in Nigerian Arabic is qalla < OA ɣalla. Its secret
language equivalent is qaafiye. This is derived according to those we questioned from the
name for the phonetic symbol of the first sound of the word, q = al-qaaf “the qaaf”, with the
arbitrary addition of the suffix -iye.19
(15)
alphabet NA secret word
qaafiye → qalla → qaafiye
“qaaf” (?) “grain” → “grain”
As in 3.1.1.2.1, it is interesting to observe that the associations are established on the
basis of Nigerian Arabic (same etymological observation for OA ɣ = NA q). A further
example along these lines is adṣ-dṣaad which is the secret language word for dṣakar “male”.
NA dṣakar “male” derives from OA đakar, with the regular pan-Sudanic Arabic sound shift
17 We have found one instance where the same koranic verse yielding the same meaning is referred to in two
ways, one alphanumerically, and one via syntagmatic displacement. This is 5:25, where both kaddin-aay (<
alphanumeric kd = kaddin + -aay suffix) and qaala r-rajul-aani “the two men said”, refer to “leper”.
18 The arbitrary or partly arbitary association between letter and concept is perhaps as old as the study of Arabic
grammar itself. Ryding (1997) describes a treatise attributed (perhaps falsely) to the eighth century Arabic
lexicographer and grammarian Xalil in which each letter of the alphabet identifies a concept. The letter nuwn (n),
for instance, represents a fish.
19 This word brings to mind qaafiya, a poem whose rhyme letter is qaaf. No one suggested such a meaning,
however, and the study of poetry is not part of the koranic school students’ curriculum.

Waris
12
*đ
→
ḍ. The NA version of
ḍ
akar is then on a phonetic basis associated with the letter ḍaad
(ض) of the Arabic alphabet, which represents the meaning “male organ”.
In other cases the basis of association appears to be semantic. šajara “tree”20 comes to
represent “cassava” = galiisa. In one instance the association is social. The meaning “married,
with a wife” = be mart-a “with his wife” is represented by ali fato “Ali of the house”. Fáto is
the Kanuri word for “house”, as if Kanuri men are conceived of as being home bound.
We should note finally here that the koranic and non-koranic sources can cross-reference
each other. An example of this is
(16) yaa ʔuwliy l-ʔalbaab
“O ye men of understanding” (2: 179)
yaa ʔuwliy substitutes for ʔalbaab, not however, in its meaning of “intellect” (referring to
people, sg. lubb), but rather in a local NA sense as follows. The phonetic sequence -baab is
focused upon, a “baab” being an NA appelation for the Kanuri = “Borno”, hence yaa ʔuwliy =
“Kanuri”. The sequence of associations is doubly secret in this instance, however, since the
appelation baab = “Kanuri” itself is a part of the non-koranic secret language. baab =
“Kanuri” derives from the first letter = “b” of the NA designation of the Kanuri ethnic group,
borno.
4. Linguistic properties
We can now turn to a summary of the linguistic properties of the secret language. This
has two aspects, the linguistic properties of the source items, and those of the resulting secret
words. We begin with the source.
4.1 Sources
4.1.1 Koranic citations
There appear to be no morphological or syntactic constraints on which koranic sequences
can be used as substitutes. A great range of morphological categories are represented,
including the following.
(17) Morphological classes of koranic substitutes (standard meanings are given in glosses)
(a) verbs
perfect: qaal-uw “said-they” (12: 77)
maa kaana “was not” (33: 40)
imperfect: sa-yaquwlu “will-he say” (2: 142)
imperative: ađkur “remember”! (38: 41)
singular: sayaquwlu
dual: yaʔkul-aani “they 2 eat” (5: 75)
plural: qaal-uw
second person: kun-ta “were-you” (5: 120)
(b) nouns
dual: haađ-aani “those 2” (22: 37)
plural: al-ma
θ
ul-aat “the exemplary” (13: 6)
Furthermore, items of all morphological categories may serve as substitutes. Although
nouns and verbs predominate, one finds single particles, like the negative particle lan “not
future” (22:37), and conjunctions and particles in combination with nouns and verbs, as will
20 Note that the NA pronunciation of “tree” is šadara. Shajara is thus taken from Classical Arabic.

Waris
13
be exemplified presently. Bound pronouns occur in abundance, as in ka-ʔanna-hum “like-
complementizer-them” (61: 4).
Syntactically as well the substitutes are virtually unconstrained. An extreme, though not
untypical case is found in (9) above where ʔa lam tara “don’t you see” = “Q particle ʔa +
negative particle lam + second person msg imperfect verb, jussive form tara” join to form a
single formal whole. In fact, slightly over half of all substitutes are formed from combinations
of words. 8 substitutes are single verbs, 16 single nouns, and 28 composed of two or three
(never more) words.
Although the substitutes are formed of morphologically discrete words, phonetically they
are pronounced as single words, as marked, for example, by lack of juncture between the
koranic words, and a single stress on the entire unit. jiʔnaa bikum “we brought you pl” (17:
104), for instance, appears as [jínabíkum] “turban” (in this case following the stress pattern of
CVCVCVCV words, e.g. šíribáta “she drank it” (Owens 1993: 43)).
We do not have textual material to check the phonology. On the basis of elicited
responses we can say the following. A number of koranic sounds are not pronounced
according to standard CA rules. Some discrepancies are due to local adaptations, while others
probably stem from the reading tradition used in the koranic schools. To the latter belong the
rendering of the Classical Arabic sounds đ, θ,
ḍ
as z, s,ẓ. These contrast with the usual NA
correspondences of these sounds,
ḍ
(less frequently, d), t (or dialectally, s) and
ḍ
. Furthermore,
processes which might appear to be due to local interference may actually be due to the
reading tradition used. biʔsa “how bad” = “liar” is pronounced [biisa] in the secret language.
It might be tempting to explain the loss of ʔ as following the general loss of etymological ʔ in
Nigerian Arabic (e.g. bada < *badaʔ “begin”). It was explained, however, that in the practiced
Warshian reading tradition the ʔ is not pronounced in koranic recitation, whereas in other
traditions (one attributed to Hafṣ) the ʔ is used.20F
21
On the other hand, the depharyngealization of CA ʕ and ħ to ʔ/h, (e.g. ṣaahibi for ṣaaħibi
in (5)) and the rendition of ɣ as q (see (7) above) and ṭ as ď all would reflect normal NA
pronunciations (see “Loanwords” (8)). However, to the extent that these pronunciations are
used in the local versions of koranic recitation itself, something which we have not checked, it
may be that they carry over directly from koranic reading. There are clear instances where this
is the case. It is striking, for example, that the CA q is consistently pronounced q, in clear
contrast to the usual correspondence, CA q = NA g. Furthermore, a number of stress patterns
are not what one would expect in NA from the word structure. la-qaal-úw “sugar”, for
example, is consistently stressed on the last syllable, though NA norm would give la-qáaluw
with stress on the penultimate. Here as well one may expect the influence of the local reading
tradition.
The rather blatant disregard for the morphology and syntax of the koranic source stands in
contrast to the sometimes rather subtle associative features which explain the choice of a
particular koranic citation for a given meaning. Figurative associations are not uncommon, for
example (3.1.1.2), and chains of koranic and native Nigerian Arabic associations may be built
up to achieve a particular meaning (e.g. (9) above). One may therefore speculate that the
sometimes audacious choices of koranic forms (e.g. ʔa lam tara) are hardly dictated by an
21 This point is corroborated in Ibn Mujahid’s collection of the seven variant readings, where he reports (132)
that Warsh interpreted Nafiʕ as not pronouncing the glottal stop.

Waris
14
ignorance of koranic syntax and morphology, but rather by the opposite, by a desire to
deliberately choose forms too outlandish to be taken as proper language.
4.1.2 Non-koranic sources
The means by which non-koranic sources assume their waris form may be grouped under
four general categories. In all of them the target word takes over the meaning of the source
word directly. In the terminology used above, the new form is the substitute, the normal form
the displaced item.
The most frequent method (14 words) is via the re-formation of a common Nigerian
Arabic word, usually by means of shortening the original form, particularly by cutting off the
end of a word and adding the suffix -aay. Thus a baxiit “rich person” becomes a baxx-aay, a
pagan kaafir becomes kaaf-aay and a koranic teacher, a faqiir a faq-aay. In some cases a letter
may simply be deleted, as when sukkar “sugar” becomes sak, and muxxur “nose” becomes
maxi. In one case material is added to the beginning of a word, mux “brain” → dimaax, and in
three cases the noun formative prefix am- is added, kamfuuṣ “Hausa person” → an-kafaṣ-aay.
A second method consists of building a secret word on the basis of the first sound or
letter of the source word. In our data this consists of 18 items, and informants suggest that this
is the most productive means of forming secret words. Its relative frequency attests to the
prominence of the oral component in koranic learning, in particular the relation between the
written letter and its phonetic realization. In many cases the new word is based on the Arabic
name of the letter, so that weeke “soup” is termed waaw, after the name of the Arabic letter for
“w”, waaw, a nayra “one nayra (unit of Nigerian money) is a nuun < Arabic name for letter
“n”, salt is miim < Arabic name for letter “m” after milhe “salt”, etc. In other cases the
phonetic value of the source serves as the basis of the new word, as when gooro “kola nut”
becomes gaaʔ-iye.22 A phonetic observation is also said to be the origin of qaaqaad “person
from the Gwoza area”. The Gwoza area, some 150 kilometers SE of Maiduguri in southern
Borno is populated, inter alia, by ethnic groups which, according to the waris, speak
languages in which the sound q, as Nigerian Arabs customarily interpret ɣ, is common.23
Extra material may be added to the basic letter/phonetic form, a redoubled consonant plus
suffix -aay or -iye for example being common (as in gaaʔ-iye above), “man” = ra-r-aay <
Arabic name for letter “r” = raaʔ after raajil “man”, fa-f-aay “fulani person” < Arabic name
for letter “f” =faaʔ after fallaati “fulani person”. The addition of a suffix may have contrastive
value, as when to differentiate the word for “woman” (marʔa) from “salt” (cf. above), the
suffix -iye (f. suffix) is added, miim-iye “woman”. We might note here one productive way in
which the system is used. The word for “naira”, as noted above, is nuun. To specify how many
naira are involved one may qualify nuun with a secret numeral, derived from the sound of the
first letter of the respective numeral, thus nuun xaa “5 naira” (xaa < first syllable of xamsa
“5”), nuun si “6 naira” (sitta “6”).
A third method (11 words) involves the use of normal derivational processes to form
secret words. In one case a word is simply borrowed from Classical Arabic, maʔi “water” <
maaʔ (cf. Nigerian Arabic alme “water”). In two cases the suffix -iye is added to a word, as
when fift-iye “50 kobo” is derived via fifti + iye. Other cases make use of various
morphological derivational processes, mudangas “disorganized” = normal participle form for
22 Or it could be here that a letter gaaʔ is conceived of, which serves as the basis of gaaʔiye. No letter “g” exists
in the Arabic alphabet, though the sound g = OA *q is found in Nigerian Arabic.
23 Wolff and Ndaghra (1992: 8) note that ɣ is a common sound throughout the languages of the Mandara
mountains (Nigeria-Cameroon border), which include the languages of the Gwoza area.

Waris
15
class 1 derived verb < dungus “rubbish”, sammaaqi “dependent person” = normal nominal
form for agentive nouns < samqa “gum”. In some cases normal derivational processes are
applied to the secret words themselves, forming, as it were, derived secret vocabulary. For
example, an ʔali fato is a “Kanuri man” ( < ʔali “Ali” + fato “house” (< K fáto “house”)).
Ingeniously, the consonantal core of this word = ʔlft, is used to form an active participle
meaning “married”, again using the normal NA active participle pattern for roots with four
consonants, mi-ʔalfit “married”. From rufʔa “hunger” (see below for origin) are formed
adjectives, rafʔ-aan or muraffaʔ, both following standard NA models (cf. ʔaďš-aan “thirsty”
and participle pattern above). sajin “soldier” combines this and the previous type of
derivation. From the source sooja “soldier” is extrapolated the phonetic elements s...j, similar
to those found in sajjin “imprison”, which is converted to sajjin-aay “soldier” via suffixation.
The ironic double entendre is evident. We might note here that one of the two verbs in the
corpus falls into this sub-class. The verb is kaskaj “read, recite” being conjugated like other
quadriliteral verbs (e.g. imperfect bi-kaskij, cf. tarjam, bitarjim “translate”). It derives,
apparently, from the Kanuri kiská diya “woods outside”. The koranic students traditionally
search out quiet, isolated places, villages, where they can carry on their work undisturbed;
forested areas, rural areas, are such places.
Finally, there are four items formed from two words, what may be termed compounds. tur
rimeeli (also recorded as tur rimeedi), for instance comes from tur “must” + rimeeli
(probably) “diminutive of ramla “sand” = “food from corn and grits”.
4.2 Linguistic properties of the secret words
The question to be addressed in this section is how the secret words are integrated
linguistically into the language. Only 2 are verbs, 2 of the secret words function as adjectives,
while the remainder are nouns. The limited number of parts of speech in which the secret
words stand thus stands in sharp contrast to the variegated classes from which they are derived
in the koranic source. The heavy preponderance of nouns recalls a process of borrowing,
where very often (though not always, see “Loanwords”, this volume, 4.2.1) nouns
predominate.24
In Nigerian Arabic there are two main determinants of morphological variation, gender
(m or f) and number (sg or pl). There are two sub-classes of plurals, those formed by an
24 The formation of secret words resembles borrowing in other ways as well. Haugen (1950), for instance,
distinguishes between loanwords and loanshifts. Loanwords involve the complete importation of a new form +
meaning complex, in the way that ka ṣaahibi acquires the meaning of “fish” (5). Loanshifts, on the other hand,
may only involve changes of meaning in the native words. He cites the example of Portuguese grosseria “a rude
remark”, which in American Portuguese acquires the (additional) meaning of “grocery” (1950: 219). Here the
analogy is to be drawn with fa qulibuw (7), which acquires a new meaning due to the near homophony with NA
qilib.
Another type of analogy may be drawn between the process of secret word formation described here and the
transference of writing systems from one language to another. Akkadian, for instance, derived its writing system
from the unrelated Sumerian. One aspect of this derivation consisted of establishing phonetic correspondences on
the basis of translation. Thus, the Sumerian sign for “hand” = shu with the syllabic value “shu” became the
Akkadian “hand” = qat with the syllabic value “qat” (Coulmas 1990: 83). The Akkadian syllabic value displaced
the Sumerian phonetic value. In this case the displacement can be represented at three levels, sign, syllabic value
and meaning, with only the middle term changing value:
Sumerian Akkadian
shu qat
“hand” “hand”

Waris
16
internal change in the word (broken) and those formed by suffixing (sound). Secret words are
adopted to these classes. One class of secret words exhibits normal NA gender and number
(suffix only) marking, while one class is morphologically inflexible.
Gender is generally marked formally in Nigerian Arabic nouns, whereby feminine nouns
end in -a, masculine in other phonemes. In the secret language there are two tendencies. When
the secret word resembles in both form a meaning a regular NA word, it will behave
morphologically like it. madd-a “porridge”, for example, derives from madiid-e “porridge”.
madiid-e is a feminine noun, marked by the -e variant of the fsg suffix, and madd-a as well is
treated as f, marked by the -a variant of the fsg suffix (variants are phonologically
conditioned). This means that when a possessive suffix is added, in both cases the
morphological variant -it, which occurs before a possessive pronoun, will be substituted for
the -e/a suffix, madiid-it-na/madd-it-na “our porridge”. On the other hand, there are equally
many cases where the gender is simply determined by the gender of the item in the real
language. A jinabikum is a “turban”, = NA kadamool, which is m, hence jinabikum is also
masculine (see (19a i) below). tahtazzu and wa ʕan ʔalqi both = “rod, staff” represent ʔaṣá,
which in NA is treated as f, so tahtazzu and wa ʕan ʔalqi, though they do not have a typical
feminine formal ending, will also appear with fsg agreement (19a v). In respect of gender
assignment, the matrix language remains NA. Generally speaking when the referent is a
person (representing a large proportion of all referents in the sample) the gender is determined
naturally, so a sayaquulu “fool” will take m or f agreement (see (18, 19)) according to the
referent.
Regarding the morphological flexibility of the words, the general rule appears to be that
the secret words are morphologically inflexible (do not vary their morphological form) unless
a suffix, usually -aay (with the dialectal/sociolectal variant -aa) or -iye is added or unless the
words are derived according to a normal NA morphological pattern. Which items allow these
suffixes is a question of lexical specification, as described in section 4.1 above.
A number of words, particularly those of non-koranic origin (4.1.2) have such a suffix as
an inherent part of the word. A faf-aay is a fulani man. This then regularly varies for number
and gender, faf-aay-a “fulani-aay-f, fulani woman”, faf-aay-iin “fulani men”, faf-aay-aat
“fulani women”. Generally a mpl is formed by suffixing -iin (as just exemplified). In one
word, faqq-aay, pl faqq-aat “koranic teacher(s)” the plural -aat is used. -aat is normally a fpl
suffix, though in a limited number of words takes mpl agreement, notably in this case in
sayyid pl sayyid-aat “koranic school master”, which is probably the model for faqq-aat. The
words formed in a derivationally regular way also take normal suffixes, mudangas
“disorganized person”, mudangas-iin “disorganized people m”, where -iin is the usual mpl
suffix for passive participles. If the form allows (cf. above), countable nouns generally take
suffix plurals, fift-iye, fiftiy-aat “50 kobo pieces”. In addition, on what may be a lexically
irregular basis, some nouns also allow plural suffixes. The gender will either be determined
naturally (cf. above and (18) below), or if the referent is non-human the fpl suffix -aat will be
used (e.g. nuun-aat naira-pl), since in NA non-human plurals are uniformly feminine. The two
verbs in the corpus are fully inflectable.
Words with a single fixed form may still be said to be marked for number and gender, as
seen in their agreement patterns. Thus, the phrases,
(18) šiif kaʔannahum da/di/dool/deel
see uncivilized this m/this f/these m/these f
“Look at this/these uncivilized person/people (m or f)”.

Waris
17
are all possible, with the identity of the referent being made more specific by the choice of
demonstrative.
Among the words of koranic origin the vast majority are morphologically inflexible.
However, there are a few words, about 6 in all, which allow two possiblities. Either, as in
(18), they have a single fixed form, as qaalat = “Arabs” (m/f/sg/pl), or they may be made
morphologically flexible by substituting -aay for the final -V(C) of the koranic word, then
adding suffixes regularly. qaal-at, for example, becomes qaal-aay “Arab msg”, and from this
the other forms follow, qaal-aay-a “Arab fsg”, qaal-aay-iin “Arabs m”, qaal-aay-aat “Arabs
f”.25 In this context, one curious development is worth noting. qaaluw “they said”, is the
secret word for “thief”. This too may be morphologically inflected, but here a Kanuri agentive
suffix, -am, is inserted before -aay, qaal-am-aay “thief m”, qaal-am-aay-a etc.
The secret words seem to have a great deal of syntactic flexibility in that they appear in
the same range of functions which ‘normal’ nouns and adjectives do, as subject, objects,
objects of prepositions, modified by adjectives, demonstratives, other nouns, and so on. (19)
gives a range of elicited examples demonstrating this. (19a) exemplifies the secret word
modified in various ways (recall that the overwhelming majority are nouns), and (19b) the
words in a sentential context. All examples were elicited. The secret word is marked by “x”.
(19a)
(i) ʔaď-ni aj-jinabikum hanay-í aš-šaree-t-a ámis
give-me def-x of-sg/m-my which-bought-I-it yesterday
“Give me the turban which I bought yesterday”.
(ii) ʔaď-ni aj-jinabikum hiney-í aš-šaree-t-hin ámis
of-pl-my them-f
“Give me my turbans which I bought yesterday”.
(iii) aj-jolaqooki dool rufugaan-í
def-x these friends-my
“These koranic students are my friends”.
(iv) humma maa-hum hazaanii-na
they not-they x-ours
“They are not our enemies”.
(v) tahtazzu hiil-í
x of/sg/f-my
“my staff”
(19b)
(i) al-qaalamaay-iin ďoolaka ma binšaaf-o
def-x those not seen-pl
“Those thieves can’t be seen”.
(ii) ma šif-na al-qaalamaay-iin ďoolaka
not saw-we def-x-pl those
“We didn’t see those thieves”.
(iii) al-qaalamay-iin sirg-o nu xá min al-faqaay
def-x-mpl stole-pl x x from def-x
“The thieves stole five naira from the koranic teacher”.
Notes to examples:
25 In this particular example the final -VC sequence, -at, happens to be the fsg morpheme. In other cases,
however, the -aay suffix replaces non-discrete morphemes, as in kun-ta “you were” = “stiff” → kunt-aay, as in
mara kunt-aay-a “a stiff looking woman”. The replacement is generally conceived of as a phonological one.

Waris
18
All of the examples in (19a) show that the secret words occur with a full range of modifiers –
demonstratives, relative clauses, the possessive morpheme hana “having”. The only
construction where we found some hesitancy involved the suffixation of bound possessive
pronouns, as in (19 a iv). Some examples of this type had to be thought over carefully, but
most were accepted in the end. The suffixation follows standard NA phonological rules. The
secret word in (19 a iv) is hazaani. When a suffix is added, a final vowel (i here) is regularly
lengthened. (19a i/ii) demonstrate that agreement is ‘natural’, determined by the referent. Like
most secret words, jinabikum “turban”, has no morphological plural. When the reference is to
many turbans, however, the agreement will be fpl (see above).
In (19b), the first two examples show that standard active/passive rules can apply when secret
words are used, and (19 b iii), which was accepted with considerable laughter, shows that
virtually an entire sentence can be composed of secret words.
The secret language is closely related to Nigerian Arabic. Parts of it are closely integrated
into the language morphologically, and most parts (e.g. 3.1.2/3) derive their meanings from
specifically NA words. This does not mean that it is impervious to the multilingual
environment in which it is found. It happens that most of the koranic teachers and students
whom we talked to in Maiduguri are bilingual, or multilingual. Nearly all know either Hausa
or Kanuri or both. They relate that they may use their secret language in a Hausa or Kanuri
context as well as in an Arabic one. In the following two examples, the entire context is either
Kanuri or Hausa.
(20a) Kanuri
laqaalúw šayi-dǝ-ro fiine
x tea-the-in put
“Put sugar into the tea”.
(20b) Hausa
ka bá-ni nuun bíyú
you give-me x 2
“Give me two naira”.
In a sense then, the secret language is a property of the koranic school social group, rather
than of Nigerian Arabic as a whole. However, most of those asked did feel that the secret
language fits into Arabic better than it does into other languages, and as seen above, many of
its linguistic properties are understandable only in relation to NA.
4.3 Semantic and affective properties of the secret language
A rough classification of the secret language into semantic domains is given in (21).
(21) Semantic domains26
Words with negative implication: 38 words
Food: 18 words
Ethnic groups: 11 words
Cultural objects: 7 words
Money: 3 words
Qualities: 5 words
Koranic education: 4 words
26 This being a semantic classification, cross-classification is in principle possible. A faxaay “koranic teacher” is
a part of a koranic school, but more generally one of many social roles. We have used the practice of classifying
the words uniquely according to what, in our opinion, the most specific domain is. A faxaay is considered a part
of koranic education, for example.

Waris
19
Body parts: 3 words
Personal names: 3 words
Roles in the social milieu: 21 words
The meanings lexicalized in the secret language is revealing of what is important in the
koranic school milieu, and is largely indicative of an origin embedded in the cultural context
of the Lake Chad area. The 18 food items include “salad”, “water”, “sugar”, “meat”, “kola
nut”, “soup”, “salt” and “oil”. These will be commented on further below. The ethnic groups
include the largest in Borno and NE Nigeria: Kanuri, Arabs, Fulani, and Hausa, as well as
smaller ones, such as the Kotoko. The phonetic basis of one of the terms qaaqaad “Gwoza
person” was commented on in 4.1.2 above. The cultural objects include clothes items
(“turban”, “a type of expensive cloth”), and utensils, “plate”, “kettle”. The latter two objects
are among the standard accoutrements of koranic school students, a kettle for water and the
plate used for collecting alms. Money is clearly important, as the students are dependent to a
large degree on the financial generosity of benefactors. We have nothing to comment on
among the qualities (“fat person”, “thin person”, “pregnant” (perhaps a taboo term)). Koranic
education includes the verb for “reading” or “reciting”, obviously the central activity in the
koranic education, “writing portions of the Koran to earn money” and “alms”. The three
words classified as body parts are fairly neutral ones (“brain”, “nose”). There are others body
parts as well (e.g. “testes”), which have been included among the words with negative
implication. The personal names are all koranic ones, “Mohammad”, “Yunus” and
“Abraham”. The social roles include “wealthy person”. Koranic schools, particularly in cities,
are often subsidized by wealthy individuals. Two words for “guest” are indicative perhaps of
the need to differentiate those associated with the koranic school group from those who are
not. The importance of “pilgram” and “bookseller” for koranic students is obvious. Yoghurt
seller” is an interesting inclusion in this category. Yoghurt sellers (see (9) above) are much
more prominent in rural Borno than in Maiduguri, and its inclusion on a list gathered in
Maiduguri is indicative perhaps of continuing strong rural ties. These ties are reflected in
other cultural terms, two words, for example, for “rod used to herd goats or sheep”. Words for
“man”, “woman” and “child”, as well as “police” and “soldier” are also found in this category.
Over a third of the items, 38 words, are imbued with a negative implication. These
include “ugly person”, “blind person”, “lame person”, “cripple”, “thief” (4 different words),
“sorcerer” (2 different words), “uncivilized”, “parasite”, “fool”, “pompous”, “liar”,
“illiterate”, “cheater”, “feces”, “fart”, “enemy”, “non-Muslim”, “evil spirit” and perhaps
“beautiful woman”.
The words with negative implication include both those which often appear as taboo
words in languages, words relating to excrement and sexual parts, for instance, as well as a
great many more connected with the koranic school culture. Specifically, the large number of
words with a negative implication should perhaps be understood against the status and image
of the group who use the secret language. Attendance at a koranic school serves not only to
introduce students to koranic learning, but also to build character. Students practice
obedience, self reliance, concentration, perseverance and the value of loyalty and mutual help.
Koranic students are a respected segment of the population, embodying good moral qualities.
It is thus not becoming of them to use profane or derogatory language. Such language can be
avoided, superficially at least and in the eyes of outsiders, through the use of the secret words.
There are significant differences in the distribution of the semantic domains relative to the
source of the word. For example, among names of food, the four words represented by koranic
citations, “salad”, “fish”, “meat”, “sugar” all refer to more substantive, special types of food,

Waris
20
whereas the 12 from other sources, e.g. “corn”, “groundnuts”, “cassava”, “porridge”, and
“water” tend to designate more mundane staples.
A statistically larger difference pertains to the words with negative implication. The
words of non-koranic sources (including acronyms) have 9 of their 60 words in the negative
domain. Those from the Koran (including the 3 alphanumeric words here) have nearly half, 25
of 53, in the negative one. These figures can be looked at from two standpoints. Is the
negative domain underrepresented among the non-koranic sources; or is it overrepresented in
the koranic?
To answer this, one might attempt, for instance, to look for specific constraining local
factors. There are a few words which probably can only be derived from non-koranic sources.
One would have to be very inventive to find a term for “kola nut” or “groundnut” in the Koran
for example, and similarly for “naira” (the Nigerian currency), and for many ethnic terms.
However, such culturally specific constraints, all requiring non-negative words from the non-
koranic source, if operative at all, account for no more than 8 or 10 items. Moreover, the fact
that a number of domains have the same concept encoded in different words, from different
sources,27 indicates that there is no general determinism based on the nature of the referent.
Looking to a functional explanation, one can note that the secret language, by its very
nature, serves to define group solidarity. Many words of non-negative nature, those pertaining
to food or money, for example, may have no greater function than this. The negative ones,
however, have the further role of enabling the koranic school students to avoid unbecoming
language. In this sense these negatively-tinged words are more central to the koranic school
culture in that they directly reflect the social status of the group. These words may therefore be
thought of as the core vocabulary of the secret language. This core vocabulary is represented
overproportionally by words from the source which ultimately defines the group itself, the
Koran.
We believe, however, that there are too many open questions to merit further speculation
along these lines. Functional considerations are inherently circular, and we lack comparative
or historical perspectives which might help break this circularity. Moreover, the data base
itself has not been so thoroughly checked that we are completely sure that we are speculating
on the basis of the right facts. It is clear, however, that there are problems in the data, which
require more than mere description.
5. Choice of koranic verses, transmission and extent of the secret language
One question to which we cannot provide a very satisfactory answer is why one koranic
verse rather than another is chosen for the secret language. We have no historical perspective
on this issue, and functional answers are not particularly compelling. Only three of the verses
in our corpus from which secret words are derived, for instance, are from chapters said to be
efficacious in the making of talismans (chapters 12, 18, 72, said to be effective in making one
liked, preventing epidemics, and protection against evil spirits, respectively). One explanation
was considered along the following lines. It might be expected that citations from the second
half of the Koran would predominate, for two reasons. First, the length of the chapters
decreases from beginning to end, and it might be expected that the shorter chapters, being
more manageable than the long ones for memorization purposes, would have a higher degree
27 “Sugar”, for instance, is both laqaaluw, based on Koran 15:15, and sak, a shortened form of Nigerian Arabic
sukkar. A social parasite is both yaʔkulaani from Koran 5: 71, and a ṣammaaqi, from Nigerian Arabic ṣamqa
“glue”.

Waris
21
of prominence. Secondly, memorization is, in fact, begun from back to front, from end to
beginning. The chapters of the second half would therefore be memorized before the first half,
and for this reason might be expected to have a higher degree of prominence. Moreover,
relatively few students actually manage to complete the memorization of the entire Koran.
Many give up a quarter, half-way, three-quarters of the way through. Within the community of
koranic students as a whole then, it may be assumed that familiarity with the end chapters is
greater than that of the beginning ones, which are learned last.
In the light of these considerations, the actual division of citations 29 from the last half,
24 from the first, is not, however, at all striking.28
Preliminary considerations, then, may point in another direction, namely in the formal
transmission of the secret language, perhaps the koranic part of it in particular, from
generation to generation. The way in which ours interviewees described how they learned the
secret language would generally support this idea. Although they were never formally
instructed in the secret language, their teacher explained many of the meanings to them, as
asked. They claim further that there is little dialectal difference in the language in Borno.
While we have not been able to confirm this point in a detailed way, comparative data
presented in the next section does not contradict it. The claim does have plausibility,
moreover, in that both koranic students and teaches are so itinerant that they mix frequently
with different groups in different places. If this is indeed the case, it would indicate that
transmission through contact is a fundamental factor of stability.
Generational transmission implies conventionalized meanings. While the logic behind
most of the meanings is obvious to those with the relevant koranic and social knowledge,
there are words whose source was given variously by different students. The word rufʔa
“hunger”, for instance, was unanimously agreed by one group of students to derive from the
word rafʕ, which is the word for the short vowel sign29 -u, which resembles a “9” in
appearance. There was disagreement as to the basis of the association, however. Some said,
emphasizing that they were speculating, that it is because hunger makes one bend, and a bent
person resembles the sign of rafʕ, ُﺮ (the vowel sign is atop the letter “r”). Another suggested
that it comes from the spelling of the word “hunger”, the sign of rafʕ being written to
represent the vowel in juwʕ “hunger”. A degree of arbitariness has thus intruded into the logic
behind this word.29F
30
Comparison with the waris tradition among other ethnic groups will further elucidate this
question. In particular, the Kanuri have such a secret language as well (known as májir májir),
and a very cursory survey suggests that it functions basically in the same way as the Arabic
one, in terms of the mechanisms set out in section 3 above. To give two illustrations, laakin
“but” (e.g. in 7.143, wa-laakin unḍur “but look upon”) means a woman carrying a baby on her
back. This derives from the orthographic shape of the word laakin where the laak is shaped
roughly as , like a child on its mother’s back. The use of syntagmatic displacement + phonetic
similarity to the Kanuri language (analogous to (9)) is found in wa taraa, from the verse wa
taraa al-šamsa ʔiđaa ṭalaʕat tazaawar (18.17) “Thou wouldst have seen the sun when it
arose, declining...”. zówor is the Kanuri word for “harlot”, hence wa taraa becomes “harlot”.
28 The numbers are statistically insignificant by a chi-square distribution.
29 This assumes a Farraʔan vowel terminology; see Owens 1990: 159.
30 It will be seen in section 6, however, that the most arbitrary secret words, the acronyms, are also those with the
most restricted social/dialectal distribution. There thus appear to be bounds on the extent to which the words can
maintain a general currency independent of the manner in which they are formed.

Waris
22
These indices would therefore point to a widely established and basically inherited
system.31 We should note, however, that there is freedom to create new associations. That so
many meanings refer specifically to cultural phenomenon in NE Nigeria attests to this fact, as
does the reported license of students to create new words from koranic citations, if they can
establish an appropriate link. The proliferation of synonyms, four words for “thief” for
instance, may also be a reflection of this latter prerogative. The transmission therefore
involves not only the handing down of a fixed vocabulary, but also the rules to perpetuate,
adopt, and expand the language, as necessary.
One cognitive observation is in order here. It is striking that nearly half, some 23 of the
koranic citations, occur at the very beginning of a verse. Whatever the ultimate historical
origin of this secret language, it appears that verse-initial position commands a high degree of
prominence and a correspondingly high number of words are derived from this position.
6. Distribution of secret language
Establishing how widespread the secret language is serves to delimitate to what extent it
is a marker of a particular subgroup. Our initial impression is that it is indeed the property of
koranic school students. Two observations support this supposition. First, it appears that a
number of older, experienced koranic scholars, even if they know the secret language, are
hesitant to use it themselves. Some, indeed, are hostile to it, suggesting that it distorts the
Koran itself. One scholar we talked to admitted knowing of the language, but denied having
any knowledge of it himself, claiming he had forgotten it, and sent us on to a nearby group of
students to learn more about it. It thus appears that even within the circle of koranic
scholarship, the active use of the secret language is restricted to students in the schools.
Secondly, knowledge of the terms in the Arabic-speaking community at large does not
appear to be great. To get an external perspective on how widespread knowledge of the
language is, we chose 20 words from our corpus, 15 of them which seemed to require a more
rarefied koranic knowledge, 5 of non-koranic origin (like raraay “man”, fafaay “Fulani
person”) and asked 38 speakers (all Nigerian Arabs) whether or not they knew the meaning of
the terms. The following four groups were asked: 16 with no koranic education background
(11 males, 5 females), 12 itinerant koranic or full-time students 12-20 years old, 5 part-time
students, and 5 scholars who had achieved a higher level of koranic competence. Most of the
students were residing in Maiduguri, though we also visited one koranic school in the largely
Arab-populated village of Ambudha approximately 32 kilometers NW of Maiduguri where we
put the questions to 2 of the students we met there, plus one villager with no connection to the
school.32 The corpus itself is a composite assemblage from different sources, so there is no
expectation that any one individual would be familiar with the entire list.
The results unequivocally localize the domain of secret language knowledge within the
circle of koranic scholars. Table 1 summarizes the results, with the full-time students and
teachers included together in the first group.
Table 1 Familiarity with 20 waris words (correct/total)
full-time koranic scholars part-time no background
31 Similarly, Chetrit observes that the secret language of Moroccan Jews is largely conventionalized and
transmitted generationally (1995: 529).
32 The village has two koranic schools. When we visited it (Oct. 1995), at one of the schools we found no one,
students and teachers having gone to cultivate the teacher’s fields, at the other 5 students. They were living at the
teacher’s house, though they said that the teacher himself was residing in Maiduguri at the time. The students,
some 14 in all, were Nigerian Arabs from near the Cameroon border, quite far away from their homeland.

Waris
23
Total Word type 314/340 100% types 17/100 types 65/301 types
non-koranic 84/85 4 10/25 1 30/75 0
koranic 211/221 9 7/65 0 31/196 0
acronym + 19/34 1 0/10 0 4/30 0
alphanumeric
Koranic scholars (students and teachers combined) have a 92% success rate overall. The
only word which gave them trouble was the acronym raďabaja “ugly woman”, known to only
2 of the 17. There is no significant difference between the koranic and non-koranic sectors,
nor is there a great difference between students and teachers, both having a 92% success rate
in the koranic part, for example. This group scored 100% on 14 of the 20 words (“types”
column). Besides raďabaja, two other words provided some difficulty, almusalaat “salad”
(12/17) and yawmʔiđin “dark” (13/17). The two rural informants from Ambudha faired
slightly worse than the others, 85% overall. The fact that both missed the same three words
may suggest dialectal differentiation, though the words turn out to be the same which provided
some difficulty for the Maiduguri speakers as well (raďabaja, almusalaat, yawmʔiđin).33 We
later queried the other three students present at the Ambudha interview, and they could not
provide the meanings of the three words either. In general, however, the results here confirm
the scholars’ own contention that there is little local differentiation in the waris language.
The non-koranic interviewees (third group) contrast sharply with the koranic scholars.
Their overall success rate of 22% is due largely to fairly good scores (40%) in the non-koranic
sector of the test. None knew the acronym, further confirming the very specialized nature of
this word formation. Boys and girls scored roughly equal, the girls doing slightly, though not
significantly better in the non-koranic part (44% vs. 38%), the boys in the koranic (15% vs.
16%). A perhaps unexpected result is the poor showing by the part-time students of the
koranic schools (second column), actually worse overall than those without a koranic school
background. Although these attend the same schools as the full-time students, the secret
language obviously does not rub off on them to any great extent. Clearly this finding
potentially tells us much about the social interaction, or lack thereof, between the different
groups learning the Koran.
Both of these groups tend to know the same words, if they know them at all. Kiskají
“reading the Koran”, for example, recorded a 100% level for the part-time students, 94%
among non-koranic students, and gaaʔiye “kola nut”, was recognized at the 75% and 60%
level respectively. Among the koranic vocabulary 3 words, qaalúw “thief”, muṣaaʔi “witch”
and zallí “madman” account for the vast majority of all successes, 83% for the two groups
together, and 100% for the part-time students. It therefore appears that certain secret words do
have some currency outside of the smaller circle of users.
This brief survey localizes the secret koranic language within a narrowly-defined sub-
group of Nigerian Arabs, namely the full-time koranic school students and their teachers.
Their uniformly good knowledge of the secret words contrasts with the rather modest
knowledge among both non-koranic and part-time koranic school youth.
7. Conclusion
The secret language of the waris is a testimony to the deep-seeded tradition of koranic
school learning in Borno. The language itself is not only taken from all of the Koran, but also
33 The only discrepancy between the two groups is wa zahaqa “thin person” missed by 4 from Maiduguri, none
from Ambudha.

Waris
24
from ancillary aspects of koranic learning, like orthography and alphanumeric references. At
the same time, at a number of points the koranic citations are juxtaposed with meanings
derived from the native Arabic of Nigeria. Classical Arabic and the local dialect mix freely in
this secret space.
The social functions of the secret language have properties common to all secret
languages. Its use signals a solidarity within the koranic school group, sets them off from
outsiders, and allows them to use language appropriate to their social status (see 4.3). More
than this, however, we would suggest that a secret language based to a large degree on the
Koran serves to bring the act of memorization of the Koran into everyday life. The work of
koranic memorization is surely a tedious, formal exercise whose rewards proceed in small,
incremental steps. Moreover, the focus is on memorization, rather than the language of the
Koran, so that students rarely are called upon to use what they learn in everyday life. The
secret language, however, brings koranic language into contact with the everyday experience
of the koranic students themselves in a number of ways. First, many of the meanings, even
those of koranic citations, derive from NA contexts. A direct link, often a facetious one, is
thereby established between koranic learning and the experiential background of the students.
Secondly, while as noted in section 5, the teacher explains the meanings of the koranic
citations to his students, he does not explain where in the Koran they are taken from. It
becomes a discovery exercise and discovery experience, therefore, when the students figure
out where they occur. Thirdly, the fact that students can, in principle, create new koranic
associations as they deem appropriate potentially imparts a creativity to their rote learning.
The secret language thereby imbues their task an immediacy which may be lacking in their
formal period of instruction.

Waris
25
Appendix
Koranic
The secret words are underlined.
qaalat al-ʔaʕraab ʔamannaa
“the desert Arabs say we believe” (49: 14)
qaalat straightforwardly substitutes syntagmatically for al-ʔaʕraab “the Arabs”, whose
meaning it represents, qaalat = “Arab”. Koranic form and meaning.
wa qaalat al-yahuwdu laysat al-naṣaaraa
“The Jews say, the Christians have naught...” (2: 112)
naṣaara is the NA term for European (or anyone with a light complexion, Arabs and Indians
can be naṣaara as well). By syntagmatic displacement, qaalat al-yahuwd comes to mean
“European”. Koranic form, but NA meanings are operative here.
An alternative, non-koranic term is naṣaay, derived directly from NA naṣaara.
wa ʔinna yuwnusa la-min al-mursiliyna
“So also was Jonah among those sent by us”. (37: 139)
wa ʔinna subsitutes syntagmatically for the neighboring proper name, Yunus.
wa bi-lħaqqi ʔanzalnaahu
“we sent down the Koran in truth” (17: 105)
Someone sent down is a guest, so wa bi-lħaqqi acquires, figuratively, the meaning of “guest”.
Acronym:
jalaqooki = full-time koran school students; j = jiʔaan “hungry”, since the students must
experience hunger; l = laammiin “joined together-pl”, representing the solidarity of the
students, q = qariibiin34 “close-pl”; solidarity; k = kariimiin “generous-pl”, an ideal quality of
the students.
34 The q derives from the phonetic value of the letter q. In NA this has the reflex g, gariib “near”.

Waris
26
References
Arnott, D. 1957. “Proverbial Lore and Word Play of the Fulani”. Africa 27: 379-97.
Bagemihl, B. 1988. Alternate Phonologies and Morphologies. PhD thesis, Univ. of British
Columbia.
Berjaoui, Nasser. 1994. “Preliminary Notes on a Secret Language of ṭǝlbaa in Morocco”. in
Caubet and Vanhove (eds.), 513-17.
Caubet Dominique and Martine Vanhove (eds.). 1994. Actes des Premières Journées
Internationales de Dialectologie Arabe de Paris. Paris: INALCO.
Chetrit, Joseph. 1994. “Formes et Structures du Mixage Linguistique dans les Langues
Secrètes Juives du Maroc”. In Caubet and Vanhove (eds.), 519-32.
Conklin, Harold. 1964 (1959). “Linguistic Play in its Cultural Context”. In Dell Hymes (ed.),
Language in Culture and Society. New York: Harper and Row, 295-300.
Coulmas, Florian. 1990. The Writing Systems of the World. Oxford: Blackwell.
Cyffer, Norbert and John Hutchison. 1990. Dictionary of the Kanuri Language. Dordrecht:
Foris.
Daniels, Peter. 1987. “Scripts of Semitic Languages”. In Robert Hetzron (ed.), The Semitic
Languages. London: Routledge.
Domaschnev, Anatoli. 1987. “Umgangssprache/Slang/Jargon”. In Ulrich Ammon, Norbert
Dittmar and Mattheier, Klaus (eds.). Sociolinguistics. Berlin: de Gruyter, 308-15.
Greenberg, Joseph. 1947. “Arabic Loanwords in Hausa”. Word 3: 85-97.
Haugen, Einar. 1950. “The Analysis of Linguistic Borrowing”. Language 20: 210-231.
Ibn Mujahid (d. 928). Kitaab al-Sabʕ fiy al-Qiraaʔaat. Ed. by Shawqi Dṣayf. Cairo: Dar al-
Maʕarif.
Koran. (tr. by Abdallah Yusuf Ali). Beirut: Dar al-Fikr.
Ibn Manđṣur (d. 1311). Lisaan al-ʕArab. Beirut: Dar Al-Fikr (n.d).
Mustapha, A. 1987. The Contribution of the Sayfawa ʕUlamaaʔ to the Study of Islam, c.
1086-1846 A. D. PhD thesis, Bayero University, Kano.
Nachtigal: Gustav. 1980 (1879). Sahara and Sudan, vol 2. (tr. by A. Fisher and H. Fisher).
London: Hurst.
Neuwirth, Angelika. 1987. Koran. In Helmut Gatje (ed.),. Grundriss der arabischen
Philologie. Wiesbaden: Reichert, 96-135.

Waris
27
Nöldeke, Theodor. 19812 (1909). Geschichte des Korans. Georg Olms: Hildesheim.
Owens, Jonathan. 1990. Early Arabic Grammatical Theory: Heterogeneity and
Standardization. Amsterdam: Benjamins.
--- 1993. A Grammar of Nigerian Arabic. Wiesbaden: Harrassowitz.
--- 1995. “Language in the Graphics Mode: Arabic among the Kanuri of Nigeria”. Language
Sciences 17: 181-99.
Rohlfs, G. 1984 (1874). Quer durch Afrika. Darmstadt: Wissentschaftliche Buchgesellschaft.
Ryding, Karen. 1997. “The Alchemy of Sound: Medieval Arabic Phonosymbolism”. In Asma
Afsaruddin and Mathias Zahniser (eds.), Humanism, Culture and Language in the Near
East. Winona Lake, Indiana: Eisenbrauns, 155-63.
Wolff, Ekkehard and Abdullahi Ndaghra. 1992. Ina Lama
ŋ
. Hamburg: Research and Progress
Verlag