Professional Documents
Culture Documents
Maarten Kossmann
Derivational gender in Moroccan Berber:
examples from Ayt Seghrushen
Abstract: Berber languages have a two-term gender opposition. Most nouns
are allowed to appear in both genders, expressing meaning diferences such as
natural gender and size. Nouns have a neutral gender value, which is lexically
determined, and a derivation which is marked by the opposite gender. In the
case of size diferences, this leads to a ternary opposition (diminutive – neutral –
augmentative), expressed by only two genders. The grammatical constraint on
the number of values is sometimes circumvened by using diferent allomorphs of
gender morphemes. With a number of sub-classes of nouns a ternary opposition
is made possible by a change in morphological sub-class.
Maarten Kossmann: African Languages and Cultures, PO Box 9515, NL-2300 RA Leiden,
The Netherlands. E-mail: m.g.kossmann@hum.leidenuniv.nl
1 Introduction1
All Berber languages have a gender system with two members, masculine and
feminine. These genders surface in agreement, but are in most nouns also overtly
marked on the noun. In most – if not all – Berber languages, this system is to a
large degree derivational, in the sense that a large proportion of the nouns occurs
in both genders. This article studies the way this derivation functions in diferent
sections of nominal morphology on the basis of one Moroccan Berber variety, the
Ayt Seghrushen (Aytt Sɣruššən) dialect as spoken between Tahala and Zrarda in
the province of Taza in the northeastern part of the Middle Atlas. Data were kindly
1 The facts of Berber nominal morphology and the basic meanings associated to gender are well-
known, and I have made no efort to provide full reference to the literature. For Ayt Seghrushen,
one may consult in particular Bentolila (1981), for Berber in general, among many others, Galand
(2010). The details of gender derivation, especially the change of sub-class, have hardly been
treated before. From discussion with participants at the Journée d’étude and my own experience
I have the impression that similar facts are to be found in other Berber varieties, but it is impos-
sible to say to what extent this is true of all the data presented here.
a. The irst system is the system with native Berber aixes. In this system, the
noun consists of maximally three parts: a preix, a lexical base, and a suix.
A typical example of nouns in this set is as follows:
(1) m f
el:sg a-funas ṯa-funas-ṯ
ea:sg u-funas ṯ-funas-ṯ
el:pl i-funas-ǝn ṯi-funas-in
ea:pl i-funas-ǝn ṯ-funas-in
‘male bovine’ ‘cow’
The preix encodes for state,2 number, and gender. The suix encodes for gender
and number. When the noun marks its plural by vocalic changes, plural suixes
are mostly absent, both in the masculine and in the feminine, e.g. a-ɣyul ‘donkey’,
pl i-ɣyal ‘donkeys’.
In one sub-set of nouns, the preix vowel is absent in the État Libre of the sin-
gular. All nouns of this type have a base starting in a single consonant followed
by a plain vowel a, i, or u. Example:
(2) m f
el:sg fus ṯ-fus-tt
ea:sg u-fus ṯ-fus-tt
el:pl i-fass-ǝn ṯi-fass-in
ea:pl i-fass-ǝn ṯ-fass-in
‘hand’ ‘baby hand’
The suix of the feminine singular is either -ṯ or -tt. In nouns with a basis ending
in i or a, it may be absent, e.g. ṯ-arya (f) ‘canal’.
The irst system is found both with nouns of Berber origin and with borrow-
ings from Arabic.
b. The second system uses Arabic-based aixes, although they are not all exact
copies from (dialectal) Arabic (see Kossmann 2010, 2013 for discussions).
The basic structure is as follows: An obligatory preix consisting of the Arabic
article, the lexical base, and a suix. The article does not express anything
except that it marks that the word is a noun belonging to this speciic mor-
phological sub-category. Diferent from Arabic, it does not have any link to
the marking of deiniteness. The suix is mostly a marker of feminine singu-
lar gender. Plural is formed as in the original Arabic word, in most cases by
stem changes, but sometimes by a suix. The feminine singular suix has
two variants: -(ə)ṯ and -a, e.g.
2 The term “state” refers to case-like forms; I will use the common French terminology for these
forms. In this variety, the État d’Annexion is used when the noun follows a preposition (with a
few exceptions) and when it is a non-topicalized subject of a verbal sentence. The État Libre is
used under all other circumstances.
c. The third system has no aixes at all. There is no overt marking for gender,
and number is either expressed by suppletion, or by a special pre-cliticized
plural marker iṯ. It is not a very large class, and mainly consists of kinship
terms, e.g.
The third system includes both nouns of Arabic and Berber origin.
The three systems function parallel to each other. Except in gender derivation
(see below), words are restricted to one system, and there are hardly any cases
where, for example, a singular would be in one sub-class, while the plural would
use another sub-class.
The three systems difer in the categories they express. The category of state
is only found in the system with Berber aixes, while the category of gender has
no overt expression in the sub-class without aixes.
There is a more subtle diference as regards the functioning of gender in the
system with Berber aixes and the system with Arabic aixes. In the system with
Berber aixes, the large majority of nouns allow for both genders, conveying dif-
ferent meanings. In the system with Arabic morphology, on the other hand, there
are hardly any words that allow for both a masculine and a feminine form ac-
cording to this system.4 Thus, while it is possible to have fus (m) – ṯfustt (f) ‘hand
– baby hand’, there is no way to do something similar within Arabic morphology.
Within Arabic-based aixal morphology, gender is a purely lexical feature that
cannot be manipulated using system-internal devices.
a. Natural gender with humans and higher animals (i.e., those animals for
which natural gender is considered relevant), e.g.
3 The voiceless lateral fricative is only attested in this word. Uɬma probably derives from *ulṯma,
as attested in other Ayt Seghrushen varieties (e.g., Pellat 1955: 153). Note however that the cluster
lṯ is maintained in other words, e.g. ṯaɣyulṯ (f) ‘female donkey’.
4 The only example I found is a loan from standard Arabic: lmuɛəllim (m) ‘male school teacher’,
lmuɛəllima (f) ‘female school teacher’.
b. Diference in size with other nouns; the masculine denotes something bigger
than the feminine, e.g.
c. Abstract nouns describing typical behavior are feminine. This is regular with
denominations of languages and occupations, but sometimes surfaces in
other forms as well, e.g.
Among these diferent categories, (a) and (b) cover a large part of the nouns in the
language. It should be noted, however, that most mass nouns allow for only one,
lexically determined, gender and number, e.g.
Within the relevant semantic categories, these ive meanings are highly produc-
tive. In the irst three meanings, one can speak of derivational processes uniquely
marked by gender change. The oppositions marked by diferent gender are, how-
ever, not equipollent. To the contrary, in most cases one can deine one lexically
determined gender that constitutes the neutral form, while the other gender is
derived from this.
Obviously, with natural gender, it is diicult to determine the neutral gender of
the pair. A priori, there is nothing that makes the female more marked than the male.
It is still possible to determine such neutral forms by looking at the gender when
The sentence ẓrix ifunasən can refer to a group of male bovines, to a mixed group,
or, most probably, to a group far away whose natural gender cannot be deter-
mined because of the distance. In this case, it appears that the neutral form of
funas- is masculine, and that the feminine is derived from this.
The derivation is much more easily detected in the case of size diferences.
In such words, there is one term that expresses a neutral statement as to the size
of the object, while the other term expresses an unusual size, either unusually
big (masculine, augmentative), or unusually small (feminine, diminutive). The
choice whether the size-neutral form is masculine or feminine is lexically deter-
mined; there are no clear correlations between semantics and gender here. Com-
pare for example:
(12) ṯamṣaṭṭ (f) ‘thigh’ ḍaṛ (m) ‘foot, part of the leg below the knee’
Gender change marks that the size is diferent from normal. This oten entails
connotations that may be positive (e.g., afection expressed by the diminutive) or
negative (esp. with the augmentative), e.g.:
(13) ḍaḍ (m) ‘inger’ ḍḍaṭṭ (f) ‘baby inger, sixth inger’
ṯiṭṭ (f) ‘eye’ aṭṭaw (m) ‘very big eye (negative expression)’
ṯamziyda (f) ‘mosque’ amziyda (m) ‘ridiculously big mosque’
There are also cases where the opposition has been lexicalized, and the two gen-
ders refer to diferent entities:
In such cases, it is not possible to refer to a collection of the two types of entities
by a single word, e.g.:
(16) lǝmžǝr n tɣǝnžayin d iɣnža ‘a drawer with spoons (f) and ladles (m)’
If one would use only ṯiɣənžayin, the drawer would only contain spoons, while
with iɣnža it would only contain ladles.
As stated above, with animate nouns, the gender opposition normally refers
to natural gender. If this is the case, a size interpretation is not possible, e.g.
Afunas (m) cannot refer to a ‘ridiculously big cow’, nor can ṯafunasṯ (f) be a ‘(cute)
little male bovine’.
Among animates whose gender is diicult to determine, or mostly irrelevant
to speakers, a size interpretation is possible, and oten constitutes the only likely
interpretation:
5 It is possible to have a derivation with natural gender for this item: ṯamkkarṯutt (f) ‘goat that has
not yet given birth with tender meat like a billy goat (which are mostly slaughtered at a younger
age)’.
6 Diferent from other varieties of Berber, aməṭṭuḍ does not necessarily imply masculine
behavior.
It is marginally possible to use ṯaryazṯ (f) in the sense of ‘little man’, but it is nor-
mally an abstract noun meaning ‘courage’.
As a result, the system does in principle not allow for expressing an augmentative
and a diminutive with a single word. Only the derived meaning that belongs to
the gender opposed to the gender of the neutral term can be lexically expressed.
A feminine neutral term can have an augmentative, but not a diminutive, while a
masculine neutral term can have a diminutive, but not an augmentative. This will
be called the ‘binary system’. Examples:
(21) ṯažllabṯ (f) ‘gown’ ažllab (m) ‘type of very densely woven gown’
[no diminutive]
aslham (m) ‘mantle’ ṯaslhamṯ (f) ‘mantle for children’
[no augmentative]
ṯaddarṯ (f) ‘house’ addar (m) ‘gigantic house’
[no diminutive]
ṯafuyṯ (f) ‘sun’ afuy (m) ‘big sun’
[no diminutive]
The second group consists of nouns that have the zero form of the masculine sin-
gular preix in the État Libre. Some of these nouns allow for the adjunction of the
regular nominal preix a- in order to express the augmentative; the opposition is
not possible in the État d’Annexion, which has the preix u- anyhow, e.g.
(23) baṛiq (m) ‘slap in the face’ abaṛiq (m) ‘very hard slap’ (ea: ubaṛiq)
baw (m) ‘faba bean’ abaw (m) ‘very big faba bean’ (ea: ubaw)
This is far from systematic, and many nouns which belong to this sub-group do
not allow for an augmentative derivation. Thus it is impossible to have an aug-
mentative from fus (m) ‘hand’ or ḍaḍ (m) ‘inger’, and **afus, **aḍaḍ are not
accepted.
As stated above, nouns with Arabic-based aixal morphology do not allow for
gender derivation using Arabic devices. There is no way, for example, of making
a feminine **lkamyuna ‘the little truck’ on the basis of lkamyun ‘truck’. However,
this does not mean that these nouns cannot have diminutives or augmentatives.
In order to achieve this, the noun changes its morphological sub-class – as men-
tioned above, these classes are formally diferent:
Words in this class have a ternary system, i.e., they can express both the diminu-
tive and the augmentative, as the neutral term is entirely diferent:
7 Note that in ṭṭunubil the initial ṭṭ represents the assimilated form of the Arabic article l-, while
in ṭṭunubilṯ it represents the assimilation of preixal ṯ to a following ṭ. In this noun, the vowelless
allomorph of the preix is used (see section 2) in the diminutive, while the augmentative has a
preixal vowel.
In such words, the feminine singular suix -(ǝ)ṯ of the Arabic-based forms is lost,
something which is easiest shown looking at the augmentative:
On the other hand, the allomorph -a of the feminine singular of the Arabic-based
forms is maintained:
Nouns with three forms are well-attested, but the ternary opposition is by no
means found with all Arabic-based nouns. In many cases, only the diminutive or
the augmentative are accepted, e.g.:
A similar situation occurs with nouns of the third morphological sub-class, the
nouns which have no aixes at all. Most of these nouns refer to humans; however,
diferences in natural gender are always expressed by suppletion (type: ‘my son’
– ‘my daughter’). In order to make diminutives and augmentatives, the nouns
are introduced into the morphological class of nouns with Berber aixes. Kinship
terms of the class without aixes are inherently irst person possessed; only by
means of pronominal suixes is a diferent interpretation possible. The diminu-
tives and augmentatives, however, are not inherently possessed, and in order to
express ‘my X’, it is necessary to use the genitival pronoun inu ‘my’, e.g.
The basic terms in both examples belong to the third sub-class. It is impossible to
say whether this is coincidence or not.
Change of sub-class is not restricted to gender derivation concerning size.
It is obligatory in the derivational system of collectives and unity nouns. Like a
number of other Berber varieties (e.g., Beni Iznasen, cf. Kossmann 2009), in the
Ayt Seghrushen variety studied here all nouns that oppose collectives to unity
nouns express this by means of a change in sub-class. Collectives invariably
belong to the system with Arabic aixes, while unity nouns belong to the system
with Berber aixes. Neutral forms of unity nouns are always feminine. As a result,
it is not possible to make diminutives from such nouns, but augmentatives are
allowed. Examples:
(31) lḥumṣ (m) ‘chick peas’ ṯaḥumṣtt (f) ‘one single chick pea’
aḥumṣ (m) ‘very big individual chick pea’
[no diminutive]
8 Muka is a feminine noun without a preix, cf. išṯ n muka ‘an owl’.
5 Conclusions
In Berber the gender opposition can take many diferent meanings: natural
gender, size, individuation, and abstract behavior. In principle, these meanings
are not mutually exclusive – one can easily imagine a diminutive of a naturally
gendered animate, or a small individualized fruit. Still, size diferences and the
other meanings are mutually exclusive. The size interpretation is only regularly
available when the other interpretations are impossible. This is especially clear in
the case of animates that have suppletive expression of gender, such as ṯaməṭṭuṭṭ
‘woman’ (as opposed to aryaz ‘man’). In such nouns, a size derivation is available
and indeed represented (aməṭṭuḍ ‘very big woman’).
The meaning “size” is diferent from the other meanings. Other meanings can
overlap, and nouns can be ambiguous as to their meaning because of this. Thus,
both the unity noun and the fruit tree are marked by feminine gender, and such
words have two interpretations:
Similarly, feminine gender can both refer to a female animate and to an abstract
noun of common behavior with the same basis, e.g.
The Berber system of gender derivation is highly interesting for a number of rea-
sons. In the irst place, it represents a not that common case of highly productive
gender derivation. In the second place, there is a fundamental mismatch between
the binary system of the gender opposition, and a ternary system in one of the
major meaning of the opposition, size. This mismatch is to some degree accepted
in the language, in that certain oppositions are simply not expressed with certain
words, but sometimes morphological allomorphies are used to resolve this. In the
third place, Berber has several diferent sub-classes of nouns. Gender derivation
is only possible within one of these sub-classes. In order to express the derived
meanings of gender, nouns belonging to the other sub-classes have to change
their class. This interplay of etymologically deined sub-classes provides a fas-
cinating insight in the dynamics of borrowed morphology once it has entered a
language.
Abbreviations
ea État d’Annexion m masculine
el État Libre pl plural
f feminine sg singular
Acknowledgments: The research for this article has been undertaken in the frame-
work of the projects Arabisch im Mittleren Atlas at the universities of Graz and
Vienna, inanced by the Austrian Science Fund (FWF) and How Arabic Inluenced
Berber at Leiden University, inanced by the Netherlands Organization for Scientiic
Research NWO. I wish to thank Abderrahmane Assini for his invaluable help. I also
wish to thank Utz Maas for many inspiring discussions about Berber and Arabic
grammar, and for involving me in the project and making my stays in Graz possible.
I also wish to thank Maarten Mous for his comments on an earlier version of this
paper. I greatly proited from the discussion at the Journée d’étude internationale :
Le berbère dans une perspective typologique (Paris, October 26, 2012). All mistakes
and wrong interpretations are of course to be blamed on me.
References
Abdel-Massih, Ernest T. (1971): A reference grammar of Tamazight (Middle Atlas Berber). Ann
Arbor: Center for Near Eastern and North African Studies, University of Michigan.
Bentolila, Fernand (1981): Grammaire fonctionnelle d’un parler berbère. Aït Seghrouchen
d’Oum Jeniba (Maroc). Paris: SELAF.
Destaing, Edmond (1920): Étude sur le dialecte berbère des Aït Seghrouchen (Moyen Atlas
marocain). Paris: Leroux.
Galand, Lionel (2010): Regards sur le berbère. Milano: Centro di studi Semitici.
Kossmann, Maarten (2009): The collective in Berber and language contact, in: Lakioui,
Mena & Brugnatelli, Vermondo (eds.), Berber in contact. Linguistic and sociolinguistic
perspectives. Cologne: Rüdiger Köppe, 53–62.
Kossmann, Maarten (2010): Parallel system borrowing: parallel morphological systems due to
the borrowing of paradigms, in: Diachronica, 27(3), 459–487.
Kossmann, Maarten (2012a): Berber-Arabic code-switching in Imouzzar du Kandar (Morocco),
in: Maas, Utz & Procházka, Stephan (eds.), Moroccan Arabic in typological perspective.
STUF – Language Typology and Universals, 65(4), 369–382.
Kossmann, Maarten (2012b): Berber, in: Frajzyngier, Zygmunt & Shay, Erin (eds.), The
Afroasiatic languages. Cambridge etc.: Cambridge University Press, 18–101.
Kossmann, Maarten (2013): The Arabic influence on Northern Berber. Leiden: E.J. Brill.
Kossmann, Maarten (forthcoming): Personal pronouns in the Ayt Seghrushen Berber variety of
the province of Taza, in: Allati, Abdelaziz (ed.), Mélanges Michael Peyron.
Pellat, Charles (1955): Textes berbères dans le parler des Aït Seghrouchen de la Moulouya.
Paris: Larose.