You are on page 1of 19

PARTIAL CLITIC FRONTING AND CLITIC REPETITION

IN KETAMA SENHAJA BERBER (NORTHWESTERN MOROCCO)

Evgeniya GUTOVA
Sorbonne Nouvelle – Paris 3, Lacito, CNRS

1. Introduction

This study investigates the behavior of verbal clitics in marked contexts (e.g. non-
realized, negation, etc.) in Ketama Berber (Senhaja, Western Rif, Morocco).1 The
present section gives an overview of the study and provides the background information
on the language. Section 2 presents the verbal clitics and some of their peculiarities: the
ventive clitic d (VENT) and the pronominal clitics (IO and DO series). Section 3 gives
an overview of clitic behavior in marked contexts. Such contexts usually trigger clitic
fronting. This phenomenon is known as attraction in Berber. In Ketama, not all clitics
are necessarily fronted. The major types of clitic behavior in Ketama are: no clitic
fronting (NCF),2 complete clitic fronting (CCF), partial clitic fronting (PCF), and clitic
repetition (CR), with ventive repetition and DO repetition as subtypes. There is a
hierarchy as to what is found depending on the number and type of the clitics and on the
dialect. PCF usually occurs when the IO clitic is fronted, while the DO clitic is left
postposed, although other constructions are also possible for the IO+DO clitic
combination (Section 4). Ventive repetition typically occurs when all the verbal clitics
are present (Section 5). In this case, the IO and VENT are usually fronted, while the DO
and VENT are postposed. The present study investigates different models of clitic
behavior with special attention to the deviant constructions. The study is descriptive in
nature. It also attempts to provide a possible explanation for the rise of new
constructions. The last section summarizes the study and draws some conclusions.
Ketama Berber is a variety of Senhaja Berber spoken by the Senhaja de Sraïr tribe
agglomeration in the western part of the Rif mountains.3 Senhaja Berber was
erroneously declared dead several times.4 All Senhaja Berber speakers are bilingual,
speaking both Berber and dialectal Moroccan Arabic. Arabic has exercised a great
influence on Senhaja. The main source of income for the local population is agriculture.

1
The author thanks Maarten Kossmann and Catherine Taine-Cheikh for their comments on the draft
version of this study. The symbols and abbreviations used in this paper are as follows: (-) indicates
morpheme boundary; (=) separates verbal particles and clitics from the verb (whether in a preverbal or in
a postverbal position), (^) indicates assimilation, elision, and sandhi; 1/2/3 = first/second/third person;
AOR = Aorist; DO = Direct Object; F = feminine; IMP = Imperative; IO = Indirect Object; IPF =
Imperfective; M = masculine; NEG = negation; P = plural (e.g. 1P = first person plural); PERF =
Perfective; S = singular (e.g. 1S = first person singular); VENT = ventive.
2
This construction is normally found in non-attraction contexts. However, in the Sahel dialect of
Ketama, clitics can remain postposed also in attraction contexts (especially in IPF.NEG and IMP.NEG).
3
Senhaja Berber is spoken among nine tribes: Ketama, Ayt Seddat, Taghzut, Ayt Bushibet (mostly
Arabized), Ayt Hmed, Ayt Bunsar, Ayt Khennus, Zerqet, and Ayt Bshir. Some sources also include Ayt
Mezduy. Each Senhaja tribe speaks its own Berber variety. One could speak of a dialect continuum from
Western to Eastern Senhaja. However, this dialect continuum is not always smooth.
4
E.g. Ethnologue 15 and Ethnologue 16. In the later editions (Ethnologue 17), the status of the language
was changed to living and the number of speakers was estimated at 50,000, which is not a reliable figure.
On the basis of Moroccan census data from 2014 (Haut Commissariat au Plan), the number of Senhaja
speakers can be roughly estimated at ca. 85,000, among which ca. 15,000 speak Ketama Berber.

1
The main publications on Senhaja Berber are: a descriptive grammar (Renisio 1932),
partly covering Senhaja; a dictionary (Ibáñez 1959); and a linguistic atlas (Lafkioui
2007). While Senhaja is geographically close to other varieties of the Rif (Tarifit), it
shows some unique linguistic features (Lafkioui 2007, Kossmann 2017a), distinguishing
it both from the varieties spoken to the East (Tarifit) and to the South (Central Atlas
Tamazight). It shares some features with Ghomara Berber spoken to the West (Mourigh
2015), but the languages also differ substantially.
This work describes the Ketama variety (westernmost Senhaja spoken in the
mountains of Ketama). The data originate from fieldwork conducted by the author
between 2013 and 2017 in Ketama, mostly in the villages of Beni Aisi (Berber: Ayṯ
ɛaḵsi), Beni Hmed (Ayṯ ḥmeḏ), Lmekhzen (Ayṯ Lmeḫzen), and Sahel (Ssahel).5 The
paradigms were elicited using Darija as the language of elicitation and checked with
multiple speakers (both men and women, from different age groups). It was carefully
verified that the elicited paradigms correlate with the data found in natural discourse
across different generations. Examples from texts are also provided.

2. Ketama Berber Verbal Clitics

There are three types of verbal clitics in Ketama Berber: the ventive clitic d (VENT)
and the pronominal clitics (the DO and the IO series). Only third person DO clitics can
appear together with IO clitics. In unmarked contexts, the clitics appear in the
postverbal position in the fixed order: IO+DO+VENT (with maximally three members,
and where any of the members can be absent). This corresponds to the “classical”
scheme of clitic ordering in Berber. Deviant constructions are restricted to the attraction
contexts and in particular, to the IO+DO and the IO+DO+VENT clitic combinations.

2.1. The Ventive

The ventive (often denoting direction toward the speaker, ‘hither’) is the only deictic
clitic in Ketama: its opposite, ientive (denoting direction away from the speaker) does
not occur. Besides the deictic meaning, ventive can add other senses to the verb.6
Ventive is especially common with verbs of movement. The ventive has different
allomorphs depending on the phonological and morphological context and on the
dialect. In IMP.SG, if the verb stem ends on a consonant, the VENT is id in Beni Aisi,
Beni Hmed, and Lmekhzen, and d in Sahel, e.g. ḵešm=id (Beni Aisi etc.) vs. kšem=d
(Sahel) ‘Enter [SG] (here)!’7 The same distribution (id vs. d) is found after other
suffixless verb forms (3MS, 3FS, and 1P). In the postverbal position, the ventive is iḏ in
Beni Aisi/Beni Hmed and i in Sahel/Lmekhzen after the 3MS:DO, and ed after the
3FS:DO clitic. When it is the only clitic, ventive is fronted in attraction contexts.

5
The village of Beni (Ayt) Hmed in the Ketama region should not be confused with the Senhaja tribe
Beni (Ayt) Hmed.
6
The semantics of the ventive has been addressed by multiple studies. See e.g. Galand 1959; Mettouchi
1998; El Mountassir 2000; Fleisch 2007 and 2012; Aoumer 2008 and 2011, Brugnatelli 2012; Taine-
Cheikh 2013 and 2017; Kossmann 2014, Belkadi 2015.
7
The fricativization of k is optional: ḵ is more common in Beni Aisi/Beni Hmed, and k in Sahel.

2
2.2. Pronominal Clitics (DO and IO Series)

2.2.1. Overview and some Peculiarities

The following table shows Ketama pronominal clitics (DO and IO series) in both
preverbal and postverbal positions. The forms with the ventive clitic are included in
parenthesis, when they have special allomorphs. The DO and the IO series are identical,
except for the third person. The role of the clitic in the sentence is usually apparent from
the context.

Table 1. Ketama DO and IO clitic pronouns

Preverbal DO Postverbal DO Preverbal IO Postverbal IO


1S y, yṯ, yt, t (y=d) (a)y [a]y, yṯ, yt, t ([a]y=d) (a)y
2MS ḵ (a)ḵ [a]ḵ (a)ḵ
2FS m (a)m [a]m (a)m
3MS ṯ (h=d) (i)ṯ (ṯ^iḏ, ih^iḏ) [a]s (a)s
3FS t (h=d) (i)t (ṯ^ed, ih^ed) [a]s (a)s
1P hen (a)na (naġ^d) [a]hen (a)na (naġ^d)
2P wen (a)wen [a]wen (a)wen
3P hen (i)hen, ṯen [a]sen (a)sen

Preverbal 1S (DO/IO) clitic. The preverbal 1S (DO=IO) clitic has different forms
depending on the context and on the dialect.8 It could be said that the form with the final
-t/ṯ is found only with 3MS and 3P verb forms, while the original y is found with the
remaining persons. Alternatively, it could be said that the form of the 1S:DO remains
the same, while the 2S/3FS/2P verb prefix is assimilated and realized as “zero” after the
clitic, e.g. š^a=y=t-ẓeṛ ~ š^a=yt^(Ø-)ẓeṛ ‘She will see me.’ The unusual form of the
fronted 1S clitic pronoun in Ketama probably originated as a reanalysis of the fronted
1S y followed by the “problematic PNG prefix” t (2S/3FS/2P) and/or by the 3S:DO
clitic ṯ/t. It is thus originally an amalgam of two morphemes. The same feature is found
in Ghomara Berber (Mourigh 2015).

Preverbal 1P (DO/IO) clitic. The preverbal 1P (DO=IO) clitic is hen, while it is ġen in
other Senhaja varieties. It is originally a metathesized form (cf. the postverbal neġ found
in the rest of Senhaja). In Ketama, the final -eġ in the postverbal neġ became -a (when
not followed by some postverbal element), e.g. *i-ẓṛa=neġ > i-ẓṛa=na ‘He saw us’ (vs.
i-ẓṛa=naġ^d with the ventive); cf. also 1S PNG suffix: *ḵešm-eġ > ḵešm-a ‘I entered’
(vs. ḵešm-aġ^d with the ventive). It is possible that since ġ was lacking in the postverbal
1P clitic na, it also disappeared in the preverbal position (ġen > hen). The fronted 1P
hen merged with the fronted 3P:DO clitic (which originated from ten by means of
“extreme fricativization”: ten > ṯen > hen).

8
The 1S clitic yṯ is found most frequently in Beni Aisi; yt in Beni Hmed, Lmekhzen, and Sahel; t in Beni
Hmed (alongside yt). In Beni Hmed, when the preverbal 1S is realized as t, it becomes homophonous with
the 3FS:DO (and 3MS:DO in some contexts). When 1S clitic is realized as t, there is no opposition
between the variants with and without a fronted clitic when followed by the 2S/3FS/2P prefix t- in the
Aorist, e.g. š^a=tẓeṛ 1) ‘She will see me’ (= ‘She will see him/her’) = 2) ‘She will see’.

3
2.2.2. DO Clitics

Several DO clitic pronouns have different allomorphs. 3MS:DO is underlyingly ṯ. It is


realized is h in an intervocalic position, e.g. i-ẓṛa-h^iḏ ‘He saw him (from a distance)’.9
The 3MS:DO is optionally assimilated to the 1S PNG suffix -ġ that is devoiced when
followed by the clitic. When assimilated, the gender of the pronoun is neutralized
(M=F), e.g. *kki-ġ=ṯ > kki-ḫ^ṯ ‘I gave it (M)’ > kki-ḫ(^ḫ) ‘I gave it (M/F).’10
3FS:DO is underlyingly t.11 It is realized as ṯ before a vowel (in combination with
the ventive, which has the form ^ed following 3FS:DO) and as h in an intervocalic
position, e.g. ẓṛa-n=ṯ^ed ‘They saw her’; i-ẓṛa-h^ed ‘He saw her’.
3P:DO is underlyingly ṯen. It is realized as hen in an intervocalic position, and is
optionally assimilated to the 1S PNG suffix, e.g. i-ẓṛa=hen ‘He saw them’, kki-ḫ^ṯen ~
kki-ḫ^ḫen ‘I gave them.’

In the preverbal position, 3MS:DO is h when followed by a 1S, 3MS, 1P, and 3P verb
forms, e.g. š^a=h=i-k(k) ‘He will give it (M).’ When followed by a 2S, 3FS, or 2P verb
forms, 3MS:DO and 3FS:DO merge. In the Aorist, such verb forms with and without a
fronted 3S:DO clitic are homophonous, e.g. š^a=t^kke-ḏ/ š^a=t-kke-ḏ 1) ‘You (SG) will
give it (M/F)’, 2) ‘You (SG) will give.’12 The distinction between the fronted 3MS and
3FS:DO (and 1S:DO/IO in Beni Hmed) depends on the verb form (subject PNG prefix),
context (AOR vs. PERF/IPF.NEG), and the dialect (Beni Aisi has more distinctions
than Beni Hmed). In the paradigm presented in this study, 3MS verb forms are used,
where the fronted 3S:DO clitic is distinguished. This verb form also has the advantage
that it can be combined with all pronominal clitics.

2.2.3. IO Clitics

IO can be expressed by an IO clitic pronoun or by a prepositional phrase (usually the


Dative preposition i followed by a pronoun).13 The pronominal clitic can co-occur with
the Dative prepositional phrase, e.g. kki-ġ=as i netta ‘I gave (something) to him.’
Double expression of the object is common in Berber (especially with the verbs ‘to say’,
‘to give’).14 The IO can express that a certain person is affected by the event. The IO
clitics underlyingly start on the vowel a-, but a- is elided after the vowel of the
preceding particle/preverbal element (e.g. (š^)a Future/Irrealis, u Negator, and so on).15

9
Recall that following the 3MS:DO, the ventive is iḏ in Beni Aisi/Beni Hmed, and i in Sahel/Lmekhzen.
10
Gemination is difficult to hear word-finally.
11
When followed by a 3FS, 2S, or 2P verb forms in the context of AOR, the form with a fronted 3FS:DO
clitic is not distinguished from a variant with a fronted 3MS:DO or from a variant without a clitic. It is
thus distinguished in the Aorist only in combination with 1P, 3MS, 1P, and 3P verb forms.
12
In PERF.NEG and IPF.NEG, the 3FS/2S/2P verb forms with a fronted 3S:DO are distinguished from a
variant without a clitic, e.g. u=t^kka=š ‘She did not give it (M/F).’ vs. u=h-kka=š ‘She did not give.’
13
An alternative construction is a composite preposition a ġur- ‘to(wards)’ followed by the suffix
pronoun, e.g. i-ṣeṛṛḓ=as ~ i-ṣeṛṛeḓ a ġur-es ‘He sent to him/her.’
14
This phenomenon is also known as clitic doubling (e.g. Souag 2015). The term clitic doubling should
not be confused with clitic repetition (the phenomenon when the clitic occurs twice in the construction).
15
Alternatively, it could be said that the vowel of the preverbal particle is elided. Since a- is also absent
after the preverbal negator u, here we posit that it is elided from the clitic. This also makes the DO and the
IO clitics more uniform.

4
The pronominal clitics (IO and DO) in Ketama coincide both in their form (except for
the third person) and in their behavior – when a single clitic occurs. When a verb form
contains a single clitic, it is obligatorily fronted in the attraction contexts. The
combination of a pronominal clitic and the ventive is also normally fronted. Since this
study focuses on the deviant constructions, only the combinations of two pronominal
clitics (IO+DO) and three clitics (IO+DO+VENT) are discussed below.

3. Verbal Clitics in Marked Contexts

3.1. Marked Contexts

Marked contexts include irrealis, negation, subordination, relativization, clefts, and


content questions (cf. Kossmann 2012: 59). All these contexts can cause clitic fronting.
In Beni Aisi and Beni Hmed, clitic fronting occurs regardless of the verb aspect.16
Irrealis/Future. In the irrealis context, following the particle a (including the
future, following the particle (m)š^a), clitic fronting is obligatory.17 This context is used
in the paradigms below. The preverbal negator u does not co-occur with the particle
(š)a, and negation is achieved by the postverbal negator š/ši/šay alone.
Negation. Verbal negation is usually bipartite in Ketama, with the preverbal
particle u and the postverbal particle š/ši/šay.18 Sometimes, the preverbal NEG u is
absent. In Beni Aisi, the preverbal negator is still frequent, although it is often absent
with 3MS verb froms (starting on the i- prefix), e.g. i-ḵšem=šay ‘He didn’t enter’.
The IMP.NEG has a special form in Ketama. First of all, it has a different
preverbal negator i in Beni Aisi/Beni Hmed (vs. the usual preverbal negator u),19 e.g.
i=tak=ši ‘Do not give [SG]!’ Lmekhzen and Sahel dialects lack a “specialized”
prohibitive negator: Lmekhzen tends to use i in all contexts, while Sahel tends to use u
in all contexts. When the IMP.NEG form contains a clitic, there are two preverbal
negators, one on either side of the clitic. The scheme is: u + CLITIC + i + VERB + ši,
e.g. u=s=i=tak=ši ‘Do not give [SG] (something) to him/her!’20 Such constructions can
be described as ‘separation of the clitic from the verb’. This phenomenon (which is only
possible in IMP.NEG verb forms) is not discussed further in this study.
The preverbal negator ma is also sometimes used in Ketama prohibitives.21 The
subsequent verb form is in the Aorist and has the regular second person PNG affixes.
The preverbal negator ma causes clitic fronting in Ketama (unlike in Ghomara).

16
Sahel is different in that it allows for postverbal clitics also in IPF.NEG (NFC, “no clitic fronting”).
The IMP.NEG, which is based on the IPF stem of the verb, also allows for NCF in Sahel. NCF, which is
possible only in Sahel and only in specific contexts, is not discussed further in this study.
17
The future particle (m)š^a might be originally a combination of two morphemes, the future marker
(m)š, probably from the Arabic participle maši ‘going’, and the Berber irrealis particle a (<aḏ). The caret
hat indicates that the two morphemes merged into one (and the final -i of maši is elided). The same
particle is found in Ghomara (Mourigh 2015).
18
There is a tendency to use the postverbal negator ši in IMP.NEG and šay in PERF.NEG, although
postverbal negators š/ši/šay are normally interchangeable.
19
The “prohibitive” i found in Ketama might be etymologically related to the prohibitive il in Ayt
Warayn and wir in Tarifit. The author thanks Maarten Kossmann for this observation.
20
In Lmekhzen, the first negator can be i, e.g. i=s=i=tak=ši ‘Do not give [SG] (something) to him/her!’
21
Negator ma can occasionally be also used with the AOR to negate the future. However, it is not found
with PERF and IPF. The same restricted usage of ma is found in Taghzut.

5
3.1. Clitic Behavior in Marked Contexts

In those contexts that obligatorily cause attraction, a single clitic is always fronted in
Ketama, whether it is VENT, DO, or IO. The models are thus VENT-, DO-, and IO-,
where the hyphen indicates the position with regard to the verb. When the verb contains
more than one clitic, the behavior of the clitic complex differs depending on the type of
the clitics and on the dialect. In Beni Aisi/Beni Hmed, the combination of DO+VENT
and IO+VENT is usually fronted. When the verb includes two pronominal clitics
(IO+DO), there are different possibilities:
1) Partial clitic fronting (IO-DO), e.g. š^a=s=i-kk=it ‘He will give it (F) to
him/her’. This variant is especially frequent in Beni Aisi/Beni Hmed;
2) DO clitic repetition (IO+DO-DO), e.g. š^a=s=t=i-kk=it ‘He will give it (F) to
him/her.’22 This variant is especially frequent in Lmekhzen;
3) Complete clitic fronting (IO+DO-), e.g. š^a=s=t=i-k(k) ‘He will give it (M/F) to
him/her’. This variant is especially frequent in Sahel/Lmekhzen.

When the verb includes all the three clitics (IO+DO+VENT) in the context of
attraction, there are again different possibilities:
1) Ventive clitic repetition (IO+VENT-DO+VENT), e.g. š^a=s=d=i-kk=ih^iḏ ‘He
will give it (M) to him/her (here)’, especially frequent in Beni Aisi/Beni Hmed;
2) Partial clitic fronting with two sub-types: the IO-DO+VENT, e.g.
š^a=s=i-kk=ih^iḏ ‘He will give it (M) to him/her’, and the IO+VENT-DO, e.g.
š^a=s=d=i-kk=iṯ ‘id.’ Both models are possible in Beni Aisi/Beni Hmed;
3) Complete clitic fronting (IO+DO+VENT-), e.g. š^a=s=t^iḏ=i-k(k) ‘He will give
it (M) to him/her’. This model is especially frequent in Sahel.

To summarize, the major types of clitic behavior in attraction contexts are: complete
clitic fronting, partial clitic fronting, clitic repetition, and no clitic fronting (in Sahel).
There is a hierarchy as to what is found depending on the number and type of the clitics
and on the dialect:
1) The IO clitic does not appear postposed and is not repeated in marked contexts;
2) The DO clitic, when in combination with the IO clitic, can be preverbal
(complete clitic fronting: IO+DO- and IO+DO+VENT-), postverbal (partial
clitic fronting: IO-DO, IO-DO+VENT, and IO+VENT-DO), or repeated (DO
clitic repetition: IO+DO-DO);
3) The VENT, when in combination with the IO or the DO, is preverbal. With
IO+DO, the VENT can be preverbal (IO+DO+VENT- and IO+VENT-DO),
postverbal (IO-DO+VENT), or repeated (IO+VENT-DO+VENT).

Some models are more frequent than others.23 There are some complications when the
DO is in the plural, and there are some irregularities in the paradigm (see below). We
will now zoom in on the IO+DO and the IO+DO+VENT clitic combinations.

22
See Section 4.1 for the details and the alternative analysis of this construction.
23
All in all, there are 26 individual models possible in the attraction contexts, including the 7 models in
the “no clitic fronting” scenario (possible in Sahel).

6
4. The IO+DO Clitic Combination

4.1. Overview and Analysis

In Ketama (and in Senhaja, in general), the IO clitic can be followed by the DO clitic
only when the DO is represented by the third person.24 When the DO is the first or the
second person, the clitic combination is not allowed. Instead, the DO clitic pronoun is
used, followed by the IO expressed by the prepositional phrase + pronoun.25
When an IO and a DO clitic appear together, the order is IO+DO. After an IO
clitic, 3MS:DO clitic ṯ often becomes t, thus merging with the 3FS:DO. A special case
is the assimilation of DO clitics following the 1P:IO clitic na (underlyingly naġ,
realized as naḫ with the DO clitic; cf. the assimilation after the 1S PNG suffix -ġ treated
above). Here, the gender distinction in 3S:DO is usually lost, e.g. i-kka=(ya)naḫ(^ḫ)
‘He gave it (M=F) to us’. The distinction can be preserved in the non-assimilated forms,
e.g. i-kka=(ya)naḫ^ṯ ‘He gave it (M) to us’.
When IO+DO clitics appear together in marked contexts, different scenarios are
possible. In Beni Aisi/Beni Hmed, the IO+DO sequence is usually split. This results in
partial clitic fronting (IO-DO model), with clitics on both sides of the verb, e.g.26

(1) š^a=s=i-kk=it ‘He will give it (F) to him/her.’

The CCF (IO+DO-) is also possible. It is the only possibility in Sahel, and is frequent in
Lmekhzen. In Beni Aisi/Beni Hmed, CCF is rare for the IO+DO clitic combination.27
When the clitics appear together, just as in the postverbal position, the gender of the DO
pronoun is only rarely distinguished by means of the fricative ṯ for the 3MS:DO, e.g.28

(2) a. š^a=s=t=i-k(k) ‘He will give it (M=F) to him/her.’


b. š^a=s=ṯ=i-k(k) ‘He will give it (M) to him/her.’ (rare)

Finally, the DO clitic can be repeated in the IO+DO-DO model, e.g.

(3) š^a=s=t=i-kk=it ‘He will give it (F) to him/her.’

This variant is found in Beni Aisi/ Beni Hmed (alongside the PCF construction), and in
Lmekhzen, where it is the preferred scenario. It is agrammatical in the village of Sahel.

24
A situation similar to Ketama is found in (Western/Central) Tarifit, Ghomara (Mourigh 2015), and in
the Tunisian varieties described in Collins 1981-82.
25
With the third person DO clitic pronouns, two options exist: either the IO+DO clitic combination, or
the DO clitic followed by an IO expressed by a prepositional phrase. The combination of the two
(“double expression of the object”) is also possible.
26
The PCF is not common in Moroccan Berber, including the other Senhaja varieties, but is also found in
Ghomara (Mourigh 2015). It is also known in Tunisian Berber (Tamezret) (Collins 1982) and in Algeria
in Chaouia (Lafkioui & Merolla 2002). Within Ketama, there are differences: PCF with the singular form
of DO clitic is agrammatical in Lmekhzen and in Sahel.
27
This variant is especially avoided in Beni Aisi/Beni Hmed in cases of ambiguity such as when the
3S:DO clitic t is followed by verb forms starting on the PNG prefix t- (2S, 3FS and 2P persons).
28
In the discussion below, the gender of the DO pronoun is largely ignored, as our focus is on the
position of the clitic.

7
It must be noted that when the DO is plural, only the PCF scenario works without any
modifications, while the “pure” DO repetition (IO+DO-DO) and the “pure” CCF
(IO+DO-) scenarios do not work. They are possible only with the preverbal t. Thus,
there are new (sub)types: CR-t (DO clitic repetition by means of t), CR+t (DO clitic
repetition in addition to t), and CCF+t (complete clitic fronting in addition to t), e.g.29

(4) š^a=s=t=i-kk=ihen (CR-t: IO+DO(t)-DO) ‘He will give them to him/her.’

The above type is labelled CR-t (IO+DO(t)-DO) to indicate that the preverbal form of
the DO is t, which does not correspond to the postverbal plural form of the DO.
Apparently, the preverbal t (“3S:DO”) became petrified and can now be regarded as a
3rd person marker without the gender/number specification. More generally, it serves as
an indicator that the verb has two pronominal clitics (IO and DO). Hence, such
constructions could be regarded as a special category of the PCF with the (optional)
preverbal “dummy t” which plays the role of the DO “slot-filler”. In constructions with
the petrified t that has no true gender and number marking, the postverbal DO clitic can
be considered an “explicating double” that marks the gender and number of the DO.
Such constructions could have arisen due to the fact that, historically, in many
situations, third person DO pronouns led to ambiguity when in the preverbal position,
especially when followed by the 2S/3FS/2P verb prefix or by the VENT, and when the
DO has the plural form. As a result, the preverbal DO lost the gender and number
specification, became a 3:DO “slot-filler”, and the DO clitic was “repeated” in the
postverbal position to give more explicit information.30 In Sahel, the preverbal t
grammaticalized even further, e.g.

(5) a. š^a=s=en=t=i-kk=ihen (CR+t: IO+DO+t-DO) ~


b. š^a=s=en=t=i-k(k) (CCF+t: IO+DO+t-)
‘He will give them to him/her.’

Here, the t element (originally a 3S:DO clitic) follows a 3P:DO clitic, which developed
into (e)n as a result of “extreme fricativization”: ten>ṯen>hen>(e)n. In such examples,
the t element cannot be regarded as 3(S):DO marker any more, but has grammaticalized
into an indicator of a “two-object-construction”. In Sahel, unlike in Beni Aisi/Beni
Hmed, the grammaticalized t cannot be omitted. It plays the role of the “disambiguator”
as it signals that the preceding en has to be interpreted as a 3P:DO marker.31

29
The new labels are used to distinguish the constructions from the “pure” DO repetition and CCF types
found with 3S:DO. The CR+t and CCF+t constructions are illustrated in example (5).
30
Note, however, that in the PCF (I-D model), the DO is not repeated. Also, when the verb contains all
the three clitics (IO+DO+VENT), the DO is not repeated, while the VENT can be repeated. With the
ventive repetition, the postverbal ventive cannot be considered an “explicating double”. It is also possible
that originally, both DO and VENT were repeated, and then the preverbal DO became lost as it was fully
assimilated to the fronted VENT.
31
The form š^a=sn=i-k(k) exists, but it is understood as ‘He will give to them’, i.e. with a 3P:IO clitic
(a)sen rather than with a combination of 3S:IO+3P:DO clitics (a)s+en. Similarly, the form š^a=sn=i-
kk=ihen exists, but it is understood as ‘He will give them to them’ (the I-D model).

8
When both the DO and the IO are plural, the CR+t and the CCF+t scenarios are not
possible in Sahel. These scenarios require that the IO clitic, which in plural ends on -n
(1P hen, 2P wen, 3P sen), be followed by the 3P:DO clitic (e)n. This would mean that
the forms with the 3P:DO clitic and without it would have been neutralized (1P hen+n
> hen, 2P wen+n > wen, 3P sen+n > sen).
It would have been interesting to know if the “dummy t” could be found with first
or second person pronouns. However, since the IO clitic cannot be combined with first
and second person DO clitics (see introduction to Section 4.1), such examples are
impossible. Thus, clitic behavior depends on the person of the DO pronoun. The major
distinction is between third person vs. non-third person pronouns. Within the third
person pronouns, the distinction is between singular vs. plural forms. When the DO is
singular, there are three scenarios:
1) Partial clitic fronting (IO-DO), especially frequent in Beni Aisi/Beni Hmed;
2) DO repetition (IO+DO-DO), especially frequent in Lmekhzen; and
3) Complete clitic fronting (IO+DO-), especially frequent in Sahel.

When the DO has the plural form, there are four scenarios:
1) Partial clitic fronting (IO-DO), especially frequent in Beni Aisi/Beni Hmed;
2) DO clitic repetition (IO+DO(t)-DO), especially frequent in Lmekhzen;
3) DO clitic repetition+t (IO+DO+t-DO), especially frequent in Sahel;
4) Complete clitic fronting+t (IO+DO+t-), likewise frequent in Sahel.

The following table lists verb forms containing the IO+DO clitics in attraction contexts.
Column a) exemplifies the IO+3S:DO combination and illustrates the sentence ‘He will
give it (3S:DO) to [someone]’ (with various IO clitics). Column b) exemplifies the
IO+3P:DO combination and illustrates the sentence ‘He will give them (3P:DO) to
[someone]’. The table demonstrates that the clitic behavior differs depending on the
number of the DO clitic and (for the plural DO) also on the number of the IO. There are
some irregularities and some cases that require further explanation. Thus, there is no
CR+t (IO+DO+t-DO) scenario for the 2FS:IO clitic. The CCF scenario is possible in
Beni Aisi/Beni Hmed for the IO+3S:DO clitic combination (column a), but there is no
CCF+t type for the IO+3P:DO (column b). By contrast, the PCF scenario is possible in
Lmekhzen for the IO+3P:DO (column b), while it is impossible for the IO+3S:DO (a).
A note regarding 1S:IO must be made. Since 1S:IO can be realized as yṯ/yt and as
y depending on the context, there is a possible confusion between yt 1S:IO and y+t
1S:IO+3S:DO. Thus, for the 1S:IO+3S:DO clitic combination, the PCF (IO-DO)
construction š^a=yt=i-kk=it ‘He will give it to me’, is impossible to distinguish from
the CR (IO+DO-DO) š^a=y=t=i-kk=it ‘id.’. Similarly, for the 1S:IO+3P:DO, it is
impossible to distinguish the PCF (IO-DO) š^a=yt=i-kk=ihen ‘He will give them to me’
from the CR(-t) (IO+DO(t)-DO) š^a=y=t=i-kk=ihen ‘id.’.32 In spite of this, different
models are listed for the 1S:IO in the table. In the column ‘Dialect’, the following
abbreviations are used: BA/H = Beni Aisi and Beni Hmed, L = Lmekhzen, S = Sahel.
Parentheses indicate that this is not the preferred variant in the dialect in question.

32
Interestingly, the last construction (š^a=y(=)t=i-kk=ihen), which is usually agrammatical in Sahel, is
possible with the 1S:IO clitic, and even becomes the only possible variant in this dialect, while the usual
“Sahel models” (CR+t and CCF+t) are agrammatical with the 1S:IO clitic.

9
Table 2. The IO+DO with the verb ‘to give’ in 3MS:AOR

IO a) IO+3(F)S:DO b) IO+3P:DO Type Dialect


‘He’ll give it to...’ ‘He’ll give them to...’
1S š^a=yṯ=i-kk=it/ š^a=yṯ=i-kk=ihen/ PCF~ BA/H (b: L)
š^a=yt=i-kk=it = š^a=yt=i-kk=ihen =
š^a=y=t=i-kk=it š^a=y=t=i-kk=ihen CR(-t) L (BA/H) (b: S)
š^a=y=t=i-k – CCF S (L) (BA/H)
2MS š^a=ḵ=i-kk=it š^a=ḵ=i-kk=ihen PCF BA/H (b: L)
š^a=ḵ=t=i-kk=it š^a=ḵ=t=i-kk=ihen CR(-t) L (BA/H)
š^a=ḵ=en=t=i-kk=ihen CR+t S (L)
š^a=ḵ=t=i-k š^a=ḵ=en=t=ye-k CCF(+t) S (L) (a: BA/H)
2FS š^a=m=i-kk=it š^a=m=i-kk=ihen PCF BA/H (b: L)
š^a=m=t=i-kk=it š^a=m=t=i-kk=ihen CR(-t) L (BA/H)
š^a=m=t=i-k š^a=m=en=t=ye-k CCF(+t) S (L) (a: BA/H)
3S š^a=s=i-kk=it š^a=s=i-kk=ihen PCF BA/H (b: L)
š^a=s=t=i-kk=it š^a=s=t=i-kk=ihen CR(-t) L (BA/H)
š^a=s=en=t=i-kk=ihen CR+t S (L)
š^a=s=t=i-k š^a=s=en=t=ye-k CCF(+t) S (L) (a: BA/H)
1P š^a=hn=i-kk=it š^a=hn=i-kk=ihen PCF BA/H (b: L)
š^a=hen=t=i-kk=it š^a=hen=t=i-kk=ihen CR(-t) L (BA/H), S
š^a=hen=t=i-k – CCF S (L) (BA/H)
2P š^a=wn=i-kk=it š^a=wn=i-kk=ihen PCF BA/H (b: L)
š^a=wen=t=i-kk=it š^a=wen=t=i-kk=ihen CR(-t) L (BA/H), S
š^a=wen=t=i-k – CCF S (L) (BA/H)
3P š^a=sn=i-kk=it š^a=sn=i-kk=ihen PCF BA/H (b: L)
š^a=sen=t=i-kk=it š^a=sen=t=i-kk=ihen CR(-t) L (BA/H), S
š^a=sen=t=i-k – CCF S (L) (BA/H)

The IO + 3 Singular DO Clitic Combination


When the DO has the singular form, there are three models for each of the IO clitics
(with the exception of 1S:IO). Below follow additional examples involving IO+3S:DO.

(6) š^a=s=awi-ḫ(^ḫ) (PCF: IO-DO) ‘I will bring it (M/F) to him/her.’33


(7) š^a=s=y-awi=ṯ i ntaha (PCF: IO-DO) ‘He will bring it (M) to her.’
(8) š^a=s=t^e-kki-m=t (CR: IO+DO-DO) = š^a=s=te-kki-m=t (PCF: IO-DO)
‘You (PL) will give it (M/F) to him/her.’
(9) argaz a=y(=)t=i-kka-n (CCF: IO(+DO)-) ‘The man who gave (it) to me...’
(10) ška^(a)=k=t=i-nna-n? (CCF: IO-DO-) ~ ška^=k=t=i-nna-n=t? (CR:
IO+DO-DO) ‘Who told it (M/F) to you (MS)?’
(11) u=sen=kki-ḫ=t=šay (PCF: IO-DO) ‘I did not give it (F) to them’
(12) i-nna=s š^a=k=ṯ=zenz-aḫ(^ḫ) (CR: IO+DO-DO) žuž n lmlayen
‘He said (to him/her), I will sell it (M) to you (MS) for two million’.

33
Compare these four variants where the gender of the 3S:DO is optionally distinguished: PCF
š^a=s=awi-ḫ^ṯ (M) vs. š^a=s=awi-ḫ^t (F), and CCF š^a=s=ṯ=awi-ġ (M) vs. š^a=s=t=awi-ġ (F/M).

10
4.2. The IO + 3 Plural DO Clitic Combination

When there is a combination of an IO and a third person plural DO clitic, the behavior
of the clitics changes. Two of the three scenarios become agrammatical: the “pure” CCF
and the “pure” DO repetition. Both scenarios involve a fronted DO following an IO.
The usual fronted 3P:DO pronoun (not following an IO) in Ketama is hn, e.g.
š^a=hn=i-k(k) ‘He will give them (DO)’. There is no variant with a CCF: both a)
*š^a=s=hn=i-k(k) and b) *š^a=s=n=i-k(k) (intended: ‘He will give them to him/her’)
are agrammatical.34 The same is true for the DO repetition: *š^a=s=n=i-kk=ihen is
impossible in the meaning ‘He will give them to him/her’.35 The new models involve
the preverbal t (hence the labels CR-t, CR+t, and CCF+t). The preverbal t is clearly
grammaticalized, as it does not correspond to the plural form of the DO. There exist
three constructions with a “slot-filler t”:
1) CR-t (IO+DO(t)-DO), e.g. š^a=s=t=i-kk=ihen ‘He will give them to him/her’.
This model is frequent in Lmekhzen;
2) CR+t (IO+DO+t-DO), e.g. š^a=s=en=t=i-kk=ihen ‘id.’, frequent in Sahel;
3) CCF+t (IO+DO+t-), e.g. š^a=s=en=t=i-k(k) ‘id.’, likewise frequent in Sahel.

In all the constructions with the “slot-filler” t, the preverbal t is the same in origin, but it
is grammaticalized to a different degree. In the CR(-t) construction, the preverbal t can
be omitted, resulting in the PCF. This is not possible in the CR+t and CCF+t
constructions, where the t plays the role of a disambiguator. In the CR+t and CCF+t
constructions, the fronted 3P:DO clitic following the IO has the form (e)n and is
obligatorily followed by the t. As these constructions are frequent in Sahel, they can be
referred to as the “Sahel models”. Both constructions are ambiguous:

(13) a. š^a=s=en=t=i-kk=ihen (CR+t) ‘He will give them to him/her.’ =


b. š^a=sen=t=i-kk=ihen (CR) ‘He will give them to them.’
(14) a. š^a=s=en=t=i-k(k) (CCF+t) ‘He will give them to him/her.’ =
b. š^a=sen=t=i-k(k) (CCF) ‘He will give it to them’.

In Beni Aisi/Beni Hmed, only the b) interpretation (with 3P:IO) is possible. Thus, in
Sahel and Lmekhzen, the element (e)n can be interpreted either as part of the 3P:IO sn,
or as 3P:DO clitic n (when followed by the slot-filler t).36
It is clear that the “Sahel models” (CR+t and CCF+t) are ambiguous, as they are
open to different interpretations. Furthermore, these models do not work when both
pronominal clitics have the plural form, because they involve the 3P:DO clitic n after
the plural IO clitic that also ends on -(e)n. Thus, with the 1P:IO, both the CR+t and the
CCF+t constructions are agrammatical:

34
The b) variant, š^a=sn=i-k(k), is possible in the sense ‘He will give (to) them’. Thus, the fronted 3S:IO
s followed by the 3P:DO n would have been ambiguous as it could have been interpreted as a 3P:IO clitic
sen. Of course, this ambiguity is not there when 3P:DO follows other (single) IO clitics (1S, 2MS, 2FS).
But the ambiguity arises when the 3P:DO clitic follows a 3S:IO s or any plural IO clitic, which end on -n.
35
This utterance would be interpreted as š^a=sn=i-kk=ihen ‘He will give them to them’.
36
Of course, the 2MS:IO (a)k/ḵ does not exist in combination with n (there is no clitic *k(e)n), and hence
the “Sahel models” š^a=ḵ=en=t=i-kk=ihen (CR+t) and š^a=ḵ=en=t=i-k(k) (CCF+t) have only one
reading, ‘He will give them to you (MS)’.

11
(15) a. *š^a=hen=n=t=i-kk=ihen (CR+t) intended: ‘He will give them to us.’
b. *š^a=hen=n=t=i-k(k) (CCF+t) intended: ‘He will give them to us.’37

In this case, Sahel employs the “Lmekhzen” CR-t (IO+DO(t)-DO) construction. One
could also posit that when the IO is plural, the CR+t and the CR-t constructions merge,
e.g. *š^a=hen=n=t=i-kk=ihen (CR+t) > š^a=hen=t=i-kk=ihen (CR-t) ‘He will give
them to us.’ Note that when the IO clitic is singular, the CR-t model is not used in
Sahel, while with the plural IO clitic, this model becomes obligatory in this dialect.
When both pronominal clitics have the plural form, the PCF (IO-DO) construction
is possible without any modifications. It remains the preferred variant in Beni Aisi/Beni
Hmed. There are therefore only two models when both pronominal clitics are plural: the
PCF (IO-DO) used in Beni Aisi/Beni Hmed (and Lmekhzen), and the CR-t (IO+DO(t)-
DO) used in Lmekhzen and Sahel (also possible in Beni Aisi/Beni Hmed). The PCF
(IO-DO) construction is thus applicable in most cases, as it works with both singular
and plural DO clitics (including the combination of two plural pronominal clitics). Note
that when the DO is plural, the PCF scenario is possible in Lmekhzen (whereas it is not
attested with a singular DO in this dialect), alongside other variants. The PCF is still not
used in Sahel, which employs “new” models with the “slot-filler” t.

To sum up, when the DO has the plural form, there are differences depending on the IO
pronoun. The “Sahel models” are incompatible with all plural IO pronominal clitics as
well as with the 1S:IO. These are the encountered constructions with plural DO clitics:
a) when the IO is singular (except for 1S:DO):
1) PCF (IO-DO): Beni Aisi, Beni Hmed (rare in Lmekhzen);
2) CR-t (IO+DO(t)-DO): Lmekhzen (rare in Beni Aisi, Beni Hmed);
3) CR+t (IO+DO+t-DO) Sahel (rare in Lmekhzen) [except for 1S:IO and 2FS]
4) CCF+t (IO+DO+t-): Sahel (rare in Lmekhzen) [except for 1S:IO]
b) when the IO is plural:
1) PCF (IO-DO): Beni Aisi, Beni Hmed (rare in Lmekhzen);
2) CR-t (IO+DO(t)-DO): Lmekhzen (rare in Beni Aisi, Beni Hmed), Sahel.

5. The IO+DO+VENT Clitic Combination

In unmarked contexts, when all the three clitics appear together, they are found in the
fixed order IO+DO+VENT. In attraction contexts, there are several variants possible
depending on the number of the DO clitic and on the dialect. When the ventive clitic
occurs with the IO and the DO clitics, it can appear:
1) on both sides of the verb (i.e. it can be repeated): CR (IO+VENT-DO+VENT);
2) in the postverbal position: PCF (IO-DO+VENT);
3) in the preverbal position: PCF (IO+VENT-DO) and CCF (IO+DO+VENT-).

There is a difference depending on the number of the DO clitic. If the DO is singular,


the CCF (IO+DO+VENT-) scenario is possible. If it is plural, the CCF is not available.

37
The CCF+t construction would have been understood as š^a=hen=t=i-k(k) ‘He will give it to us’ (CCF:
1P:IO+3S:DO).

12
This is probably due to the fact that in this scenario, the form of the fronted DO would
have been reduced to (e)n, thus leading to ambiguity in a number of cases (such as
following 3S:IO and all plural IO clitics). Note that the PCF scenario involves two
models (IO+VENT-DO and IO-DO+VENT). DO repetition is not found with three
clitics and there are no traces of the “slot-filler” t. One could also posit that the
preverbal t is assimilated to the ventive d. This, however, does not happen in the CCF
scenario in Sahel. Note also that if the preverbal 3S:DO t could occur without the
ventive, it would not be distinguished with the 2S/3FS/2P verb forms (t- prefix).

There is a correlation between the constructions used with the IO+DO clitics and the
IO+DO+VENT clitics. Thus, in Beni Aisi/Beni Hmed, where with the IO+DO
combination, the PCF (IO-DO model) is frequent, the ventive clitic is usually repeated
when all the three clitics occur (IO+VENT-DO+VENT). The IO+VENT-DO+VENT
model is derived from the IO-DO model by placing the ventive clitic after each of the
pronominal clitics. Ventive repetition is optional, and either of the two ventives can be
omitted. This is how we acquire two PCF constructions (IO+VENT-DO and IO-
DO+VENT). Both models are possible in Beni Aisi/Beni Hmed, although they are less
frequent than the IO+VENT-DO+VENT model.
In Sahel, where with the IO+DO combination, the CCF (IO+DO-) is the only
attested model, the ventive clitic is also normally fronted when all the three clitics occur
(IO+DO+VENT-). The IO+DO+VENT- model is also attested in Beni Aisi/Beni Hmed,
albeit rarely. This is the model found in other Senhaja varieties, e.g. Taghzut
š^a=s=ṯ^iḏ=i-k(k) ‘He will give it (M) to him/her here.’ Below follow some examples
involving the three verbal clitics.38

(16) š^a=s=t^iḏ=awi-ġ (CCF: IO+DO+VENT-) ~


š^a=s=d=awi-ḫ^ḫ^iḏ (CR: IO+VENT-DO+VENT)
‘I will bring it (M) here for him/her.’
(17) š^a=ḵ=d^e-kk=ih^ed (CR: IO+VENT-DO+VENT) ~
š^a=ḵ=te-kk=ih^ed (PCF: IO-DO+VENT) ~
š^a=ḵ=d^e-kk=it (PCF: IO+VENT+DO)
‘She will give it (F) to you (MS) here.’
(18) š^a=s=d^e-nni-m=ṯ^ed (CR: IO+VENT-DO+VENT) ~
š^a=s=d^e-nni-m=t (PCF: IO+VENT-DO)
‘You (PL) will tell it (F) to him/her.’
(19) u=s=d=i-qqr=ih^ed=šay (CR: IO+VENT-DO+VENT) ~
u=s=d-i-qqr=it=šay (PCF: IO+VENT-DO)
‘He is not saying it (F) to him/her.’
(20) faywaḫ asen=d=i-wwi=h^iḏ... (CR: IO+VENT-DO+VENT)
‘When he brought him to them, ...’
(21) a=y=d^zenz-et^t^iḏ? (CR: IO+VENT-DO+VENT)
‘Will you (SG) sell it (M) to me?’
(22) u=hen=d=kka-n=ṯen=d=šay (CR: IO+VENT-DO+VENT)
‘They did not give them to us.’

38
Examples of IO+3P:DO+VENT in the texts are rare. The few found examples conform to Table 3.

13
The following table lists verb forms containing the IO+DO+VENT clitics in the
attraction context. Column a) exemplifies the IO+3S:DO+VENT clitic combination. It
illustrates the sentence ‘He will give it (3S:DO) to [someone] (with various IO clitics)
here/in this direction’. Column b) involves the IO+3P:DO+VENT clitic combination
and illustrates the sentence ‘He will give them (3P:DO) to [someone] here’.39

Table 3. IO+DO+VENT with the verb ‘to give’ in 3MS:AOR

IO a) IO+3S:DO+VENT b) IO+3P:DO+VENT Type Dialect


‘He will give it to...’ ‘He will give them to...’
1S š^a=y=d=i-kk=ih^iḏ š^a=y=d=i-kk=ihen=d CR BA/H
š^a=y=d=i-kk=iṯ š^a=y=d=i-kk=ihen PCF (BA/H)
š^a=(y)ṯ=i-kk=ih^iḏ/ š^a=yṯ=i-kk=ihen=d/ PCF (BA/H)
š^a=(y)t=i-kk=ih^iḏ š^a=yt=i-kk=ihen=d
š^a=y=t=iḏ=i-k(k) CCF S (BA/H)
2MS š^a=ḵ=d=i-kk=ih^iḏ š^a=ḵ=d=i-kk=ihen=d CR BA/H
š^a=ḵ=d=i-kk=iṯ š^a=ḵ=d=i-kk=ihen PCF (BA/H)
š^a=ḵ=i-kk=ih^iḏ š^a=ḵ=i-kk=ihen=d PCF (BA/H)
š^a=ḵ=t^iḏ=i-k(k) CCF S (BA/H)
2FS š^a=m=d=i-kk=ih^iḏ š^a=m=d=i-kk=ihen=d CR BA/H
š^a=m=d=i-kk=iṯ š^a=m=d=i-kk=ihen PCF (BA/H)
š^a=m=i-kk=ih^iḏ š^a=m=i-kk=ihen=d PCF (BA/H)
š^a=m=t^iḏ=i-k(k) CCF S (BA/H)
3S š^a=s=d=i-kk=ih^iḏ š^a=s=d=i-kk=ihen=d CR BA/H
š^a=s=d=i-kk=iṯ š^a=s=d=i-kk=ihen PCF (BA/H)
š^a=s=i-kk=ih^iḏ š^a=s=i-kk=ihen=d PCF (BA/H)
š^a=s=t^iḏ=i-k(k) CCF S (BA/H)
1P š^a=hen=d=i-kk=ih^iḏ š^a=hen=d=i-kk=ihen=d CR BA/H
š^a=hen=d=i-kk=iṯ š^a=hen=d=i-kk=ihen PCF (BA/H)
š^a=hn=i-kk=ih^iḏ š^a=hn=i-kk=ihen=d PCF (BA/H)
š^a=hen=t=iḏ=i-k(k) CCF S (BA/H)
2P š^a=wen=d=i-kk=ih^iḏ š^a=wen=d=i-kk=ihen=d CR BA/H
š^a=wen=d=i-kk=iṯ š^a=wen=d=i-kk=ihen PCF (BA/H)
š^a=wn=i-kk=ih^iḏ š^a=wn=i-kk=ihen=d PCF (BA/H)
š^a=wen=t=iḏ=i-k(k) CCF S (BA/H)
3P š^a=sen=d=i-kk=ih^iḏ š^a=sen=d=i-kk=ihen=d CR BA/H
š^a=sen=d=i-kk=iṯ š^a=sen=d=i-kk=ihen PCF (BA/H)
š^a=sn=i-kk=ih^iḏ š^a=sn=i-kk=ihen=d PCF (BA/H)
š^a=sen=t=iḏ=i-k(k) CCF S (BA/H)

39
The form of the ventive shows that the DO clitic is 3MS. No data from the Sahel dialect are available
for the IO+3P:DO+VENT clitic combination (column b). Abbreviations of the dialects are the same as in
Table 2 (BA/H = Beni Aisi/Beni Hmed, L = Lmekhzen, S = Sahel). Parentheses indicate that a given
construction is not the preferred variant in the dialect in question.

14
6. Summary and Conclusions

This study investigated the clitic behavior in the context of attraction in Ketama Berber.
Maximally, a combination of three clitics is allowed, IO+DO+VENT, provided the DO
clitic is a third person pronoun. Different models exist for various combinations of
clitics, depending on the context, the number and the nature of the clitics involved, and
the dialect. The major scenarios are:

1) No clitic fronting, usually found in non-attraction contexts, but also possible in


Sahel in some attraction contexts (especially in IPF.NEG and IMP.NEG forms).
The models in this scenario are: -VENT, -DO, -IO; -DO+VENT, -IO+VENT;
and -IO+DO+VENT;
2) Complete clitic fronting with the models VENT-, DO-, IO-; DO+VENT-,
IO+VENT-, IO+DO-; and IO+DO+VENT-. Fronted clitics appear in the same
order as in the postverbal position. The IO+DO- and the IO+DO+VENT-
models are mostly found in Sahel and Lmekhzen;
3) Partial clitic fronting with the models IO-DO; IO-DO+VENT, IO+VENT-DO.
The IO-DO model is frequent in Beni Aisi/Beni Hmed. IO-DO+VENT and
IO+VENT-DO are also possible in the same villages, but are less frequent;
4) Ventive repetition: IO+VENT-DO+VENT, frequent in Beni Aisi/Beni Hmed;
5) DO repetition: IO+DO-DO (and sub-models, see below).

The DO repetition scenario is possible to analyze differently. With a singular DO, the
DO is seemingly repeated, e.g. š^a=s=t=i-kk=it ‘He will give it to him/her’ (frequent in
Lmekhzen). However, the preverbal t is grammaticalized and is also found with a plural
DO (CR-t: IO+DO(t)-DO), e.g. š^a=s=t=i-kk=ihen ‘He will give them to him/her’ (also
frequent in Lmekhzen). Sahel dialect prefers the CR+t (IO+DO+t-DO), where the
3P:DO repetition is accompanied by the preverbal t, e.g. š^a=s=en=t=i-kk=ihen ‘id.’.
The postverbal DO is optional, and the CCF+t (IO+DO+t-) is also possible, e.g.
š^a=s=en=t=i-k ‘id.’

The major factors that influence the behavior of the clitics are as follows:
1) Syntactic context: unmarked/non-attraction vs. marked/attraction.
2) The number and the nature of the clitics involved:
- The IO+DO clitic combination is usually split in Beni Aisi/Beni Hmed, while
other combinations of two clitics (DO+VENT, IO+VENT) are not split;
- Ventive and DO can be repeated (in certain contexts), while IO clitic cannot.
3) The form of the clitic: person and number. For the IO+DO combination, there is
a major distinction between the singular and the plural forms of the DO:
- With plural DO, the “pure” CR and CCF constructions are impossible;
- With plural DO and IO, the CR+t and CCF+t constructions are impossible.

There are dialectal differences in the clitic behavior:


- the PCF (IO-DO) scenario is frequent in Beni Aisi and Beni Hmed;
- the CR(-t) (IO+DO(t)-DO) scenario is frequent in Lmekhzen;

15
- the CCF (IO+DO-) scenario is frequent in Sahel (whenever possible). The
modified models CCF+t (IO+DO+t-) and CR+t (IO+DO+t-DO) are used with
the plural DO, but are incompatible when IO is also plural. In the last case,
Sahel uses the CR-t model.

Thus, Beni Aisi and Beni Hmed have the tendency to split the IO+DO clitic complex,
Lmekhzen prefers a variant with DO repetition, and Sahel prefers the CCF scenario.
The divergent clitic behavior in Ketama may be in some cases linked to the wish to
avoid ambiguity. There are many contexts where the fronting of 3S:DO (M or F) leads
to ambiguity, mostly due to the fact that the fronted clitic is indistinguishable when in
combination with the 2S/3FS/2P t- prefix in the Aorist. The fronted 3S:DO clitic is also
indistinguishable after the 1S:IO, e.g. š^a=y=t^e-k(k) = š^a=y=te-k(k) ‘She will give it
to me’ = ‘She will give to me’. Keeping the clitic postposed or repeating it helps to
avoid ambiguity, e.g. š^a=yt^e-kk=it ‘She will give it to me.’ Thus, deviant clitic
behavior in Ketama might have a “functional” explanation. However, such clitic
behavior is not always restricted to the “problematic” contexts, and in many cases, there
is no ambiguity, e.g. š^a=s=te=nn-a (CCF) ~ š^a=s=enna-ḫ=t (PCF) ‘I will tell it to
him/her.’ Also, there are some examples which remain ambiguous (e.g. 3FS/2S/2P verb
forms with a single fronted 3S:DO), and there are no alternative models resolving their
ambiguity, e.g. š^a=t^ẓeṛ = š^a=t-ẓeṛ ‘She will see him/her’ = ‘She will see.’
Some innovations can be explained as the reanalysis of the original forms. Thus,
the fronted 1S:DO/IO clitic yt originated as the reanalysis of the 1S y + 3S:DO t and/or
the 3FS/2S/2P verb prefix t-. Similarly, some clitic models might have originated as a
reanalysis. The preverbal “slot-filler” t is a result of grammaticalization of the fronted
3S:DO, which gradually became devoid of its meaning, and hence a “real” DO appears
postverbally, e.g. š^a=s=t=i-kk=ihen ‘He will give them to him/her.’ Furthermore,
when fronted, 3P:DO (ten>ṯen>hen>n) might merge with the fronted IO and become
indistinguishable. In sum, there are different paths that could lead to the origin of a new
construction. The ambiguity in some constructions have led to the reanalysis.
Subsequently, the reanalysis led to the emergence of new constructions, which in many
cases help to avoid the ambiguity. Thus, the main drives behind the deviant clitic
behavior in Ketama seem to be: reanalysis, grammaticalization, and probably the wish
to avoid ambiguity wherever possible.

The described constructions have not been found in other Senhaja varieties.40 However,
Ketama is not the only Berber variety with the deviant clitic behavior. Various
deviations from the “classical” (CCF) scenario are found in other Berber languages.41
Given the geographical distance between the areas in which the innovations took place,
we assume that the innovations were not caused by contact between these varieties.

40
Partial clitic fronting and clitic repetition are possible in Taghzut, but in different contexts than in
Ketama. In Taghzut, deviant clitic behavior is limited to the IPF.NEG and IMP.NEG verb forms.
41
Among these varieties are: Ghomara (Mourigh 2015), Ayt Seghrushen (Kossmann 2017b), some
dialects of Kabyle such as Eastern Kabyle (Aoumer 2011: 454), Tasahlit, Chaouia (Lafkioui & Merolla
2002: 22-29; Reesink 1979), Aurès (Brugnatelli 1993: 236), Ngouça (Brugnatelli 1993: 236), Mzab
(Delheure 1989), Sud Oranais (Kossmann 2010: 93), Bissa (Reesink 1979: 374), some Tunisian Berber
varieties (Tamezret, Douiret) described by Collins (1981-1982), Zwara (Brugnatelli 1993: 235),
Ghadames (Kossmann 2013), and Mali Tuareg (Heath 2005: 597, 601).

16
Nor do the data suggest the innovations were caused by contact with the Arabic
language. Moreover, the precise patterns of change are significantly different in
different varieties. The question is then: what motivated the system-internal changes in
the different varieties? Were the innovations driven by functional parameters? By
pragmatic motivations? By different types of grammaticalization phenomena? There is
no one definite answer for all the attested types of deviant clitic behavior. However, it
can be observed that in a number of varieties, various reanalysis phenomena might be
responsible for the change in the clitic behavior. Those changes are neither
“complication” or “simplification” of the original system. They must have occurred
gradually and were not an ad hoc invention. The resulting new systems are not always
“symmetrical”, and a number of irregularities are attested. This points to the conclusion
that the deviant clitic behavior is most likely an innovation.

The variety that exhibits similar features to Ketama, is Ghomara (Mourigh 2015). Thus,
when there are two pronominal clitics, the IO is fronted, while the DO can be left
postposed (the IO-DO model). As Ghomara is also geographically close, the deviations
found in Ketama and Ghomara can be considered a shared innovation.42 Due to the lack
of old written sources, we cannot conclude how old this feature is. The fact that the
same feature is shared by Ketama and Ghomara, might be an indication that this feature
is not very recent (considering the fact that there is little contact between Ketama and
Ghomara at present). Combined with other isoglosses, this feature also points to the
common genetic origin of Ketama and Ghomara. We also cannot exclude the possibility
that Ghomara and Ketama speakers were once geographically closer than at present, and
their varieties might have influenced each other.

This study showed that the deviant clitic behavior in Ketama is most likely an
innovation, and that there are probably some reasons behind it. While the system might
appear asymmetrical and irregular at the first sight, there are some regularities within
this deviant system. Other Berber languages exhibit other deviations, the explanations
for which are still to be provided. Different modifications to the classical system of
clitic fronting in different Berber languages are most likely independent innovations.
However, there are some common tendencies. There are three most common deviations
from the classical system in Berber: no clitic fronting in attraction contexts, partial
clitic fronting, and clitic repetition, especially repetition of the VENT, more rarely of
the DO, and, most rarely, of the IO. It remains to be explained why some things tend to
innovate, and why certain innovations are preferred above others. Thus, the complete
story of clitic behavior in Berber still remains to be written.

42
At the same time, something structurally similar is found in Chaouia and Mzab in Algeria, and in
Tamezret in Tunisia. The IO+VENT-DO+VENT model is also shared by a number of Berber languages.

17
References

Aoumer, Fatsiha. 2008. Sémantique verbale, deixis et ‘orientation’ en berbère (parler


kabyle des Iâamranen). Thèse de doctorat. INALCO.
Aoumer, Fatsiha. 2011. ‘Une opposition perdue – la particule dite ‘d’approche’ ou la
deixis verbale dans un parler kabyle de Bejaïa’. In Amina Mettouchi (ed.).
« Parcours berbères », Mélanges offerts à Paulette Galand-Pernet et Lionel
Galand pour leur 90e anniversaire. Köln: Rüdiger Köppe Verlag. 453-468.
Belkadi, Aicha. 2015. ‘Deictic directionality and Space in Berber. A typological survey
of the semantics of =d and =nn’. In Sabrina Bendjaballah & Samir Ben Si Saïd
(eds.). Corpus 14. 189-233.
Brugnatelli, Vermondo. 1993. ‘Quelques particularités des pronoms en berbère du
Nord’. In: J. Drouin, A. Roth (eds.). A la croisée des études libyco-berbères.
Mélanges offerts à Paulette Galand-Pernet et Lionel Galand. Paris: Geuthner.
229-245.
Brugnatelli, Vermondo. 2012. ‘Syntaxe et figements en berbère’. In Dymitr Ibriszimow,
Rainer Vossen, Harry Stroomer (eds.). Études berbères VI. Essais sur la syntaxe
et autres articles Actes du 6 Bayreuth-Frankfurt-Leidener Kolloquium zur
Berberologie, Bayreuth, 19-21 juillet 2010. Köln: Rüdiger Köppe Verlag. 43-51.
Collins, Ridwan, 1981-82. ‘Un microcosme berbère : systeme verbal et satelites dans
trois parlers tunisiens’. Revue de l'Institut des Belles Lettres Arabes (IBLA). (I)
IBLA 148: 287-303. (II) IBLA 149: 113-129.
Delheure, Jean. 1989. ‘Etude sur le mozabite’. Etudes et Document Berbères 6: 120-
157.
El Hannouche, Jamal. 2008. Ghomara Berber. A Brief Grammatical Survey. MA
Thesis, Universiteit Leiden.
El Mountassir, Abdallah. 2000. ‘Langage et espace: les particules d’orientation -d / -nn
en berbère (tachelhit)’. In Mélanges offerts à Karl-G. Prasse : Etudes berbères
et chamitosémitiques. Peeters: Paris-Louvain. 125-153.
Fleisch, Axel. 2007. ‘Orientational clitics and the expression of path in Tashelhit Berber
(Shilha)’. Annual Publication in African Linguistics 5. 55-72.
Fleisch, Axel. 2012. ‘Directionality in Berber: Orientational clitics in Tashelhit and
related varieties’. In Mietzner & Claudi (eds.). Directionality in Grammar and
Discourse: Case Studies from Africa. Köln: Rüdiger Köppe Verlag. 127-146.
Galand, Lionel. 1959. ‘Une opposition perdue: note sur la particule d’approche dans un
parler kabyle des Bibans’. GLECS 8: 69-70.
Haut Commissariat au Plan (HCP). Royaume du Maroc. 2004. ‘Recensement général
de la population et de l'habitat’. http://www.hcp.ma/ (consulted 2017-10-01).
Heath, Jeffrey. 2005. A Grammar of Tamashek Tuareg. Berlin/New York: Mouton De
Gruyter.
Ibáñez, Esteban. 1959. Diccionario español-senhayi (dialecto bereber de Senhaya de
Serair). Madrid: Instituto de Estudios Africanos.
Kossmann, Maarten. 2010. ‘Grammatical notes on the Berber dialect of Igli (Sud
oranais, Algeria)’. In H. Stroomer, M. Kossmann, D. Ibriszimow, R. Vossen
(eds.). Etudes berbères V – Essais sur des variations dialectales et autres

18
articles. Actes du 5 Bayreuth-Frankfurt-Leidener Kolloquium zur Berberologie,
Leiden, 8-11 octobre 2008. Köln: Rüdiger Köppe Verlag. 69-120.
Kossmann, Maarten. 2011. A Grammar of Ayer Tuareg (Niger) [Berber Studies 30].
Köln: Rüdiger Köppe Verlag.
Kossmann, Maarten. 2012. ‘Berber’. In: Zygmunt Frajzyngier & Erin Shay (eds.). The
Afroasiatic Languages. Cambridge: Cambridge University Press. 18-101.
Kossmann, Maarten. 2013. A Grammatical Sketch of Ghadames Berber (Libya).
[Berber Studies 40]. Köln: Rüdiger Köppe Verlag.
Kossmann, Maarten. 2014. ‘The use of the ventive marker dd in Figuig Berber
narratives’. Nordic Journal of African Studies 23(4): 241-291.
Kossmann, Maarten. 2017a. ‘La place du parler des Senhaja de Sraïr dans la
dialectologie berbère’. In Á. Vicente, D. Caubet, A. Naciri-Azzouz (eds.). La
région du Nord-Ouest marocain. Parlers et pratiques sociales et culturelles 12.
Zaragoza: Prensas de la Universidad de Zaragoza. 93-105.
Kossmann, Maarten. 2017b. ‘Personal pronouns in the Ayt Seghrushen Berber variety
of the province of Taza’. In Abdelaziz Allati (ed.). "Auréoles berbères".
Mélanges offerts à Michael Peyron 47. Köln: Rüdiger Köppe Verlag. 253-265.
Lafkioui, Mena & Daniela Merolla. 2002. Contes berbères Chaouis de l'Aurès. Köln: Rüdiger
Köppe Verlag.
Lafkioui, Mena. 2007. Atlas linguistique des variétés berbères du Rif. Köln: Rüdiger Köppe
Verlag.
Lewis, M. Paul, Gary F. Simons, and Charles D. Fennig (eds.). 2013. Ethnologue:
Languages of the World, Seventeenth edition. Dallas, Texas: SIL International.
Online: http://www.ethnologue.com/.
Mettouchi, Amina. 1998. ‘La Particule D en berbère (kabyle) : transcatégorialité des
marqueurs énonciatifs’. In Bernard Caron (ed.). Proceedings of the 16th
International Congress of Linguists, Paris 20-25 juillet 1997. Oxford:
Pergamon. Paper n°0270 (CD-Rom).
Reesink, Pieter. 1979. Problèmes de détermination en Indo-européen et dans une
langue chamito-sémitique. These de 3eme cycle, Paris, EPHE.
Renisio, Amédée. 1932. Études sur les dialectes berbères des Beni Iznassen, du Rif et
des Senhaja de Sraïr. Grammaire, textes et lexique. Paris: Ernest Leroux.
Souag, Lameen. 2015. ‘The development of dative agreement in Berber: Beyond
nominal hierarchies’. Transactions of the Philological Society 113 (2): 232-248.
Taine-Cheikh, Catherine. 2013. ‘Les particules d'orientation en zénaga : du spatial au
temporel’. In H. Jarmouni & S. Moukrim (eds.). Études et recherches en
linguistique et littérature amazighes : la mesure du sens et le sens de la mesure,
Actes du Colloque international organisé en hommage au professeur Miloud
TAIFI. Saïs-Fès: Faculté des Lettres et des Sciences Humaines. 47-64.
Taine-Cheikh, Catherine. 2017. ‘Les particules d'orientation du berbère.
Fonctionnement, sémantisme et origine’. In A. Agostini & M. G. Amadasi
Guzzo (eds.). Quaderni di Vicino Oriente, XIII (Afroasiatica Romana.
“Proceedings of the 15th Meeting of Afro-Asiatic Linguistics” 17-19 September
2014, Roma). Rome. 247-257.

19

You might also like