See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/309386245

The Morphophonology of Tashlhit Agentive Nouns

Article · January 2012

CITATIONS READS

2 676

1 author:

Karim Bensoukas
Mohammed V University of Rabat- FLHS
8 PUBLICATIONS 36 CITATIONS

SEE PROFILE

All content following this page was uploaded by Karim Bensoukas on 23 October 2016.

The user has requested enhancement of the downloaded file.

Published in LANGUES ET LITTERATURES, VOL. 22, 2012, pp. 103-136.

THE MORPHOPHONOLOGY OF TASHLHIT

AGENTIVE NOUNS *

Karim BENSOUKAS
M5-Agdal University, Rabat.

1. Introduction

‘Un verbe étant donné aucune règle bien établie ne permet

de former son nom verbal.’ Such is Laoust’s (1936, 226) statement
concerning Amazigh morphology, and similar in spirit is
Aspinion’s (1953, 307): ‘il n’existe aucune règle pour construire
d’une manière certaine les noms verbaux d’un verbe donné.’ While
these two statements differ in their assertive force, they are good,
representative samples of early Berberists’ descriptions of Amazigh
morphology.
In this paper, we will show that although exhibiting
intricacies at different levels, Agentive Noun (AN) morphology is
not as ungoverned as implied in previous research. Two hypotheses
constitute the pillars of the analysis. First, the Unitary AN
Morpheme Hypothesis identifies the morpheme as a simple labial
prefix {m+}, while the initial vowel in ANs is treated as an
inflectional affix that nouns generally bear in Amazigh. The [m/n]
allomorphy affecting the AN, as in the examples amkraz and
anlmad (from the verbs krz ‘plow’ and lmd ‘learn’, respectively), is

*
This paper is a synthesis of the analysis of agentive nouns in my D.E.S. and
Doctorat d’Etat dissertations (Bensoukas 1994, 2001, respectively) as well as
the treatment of labial dissimilation in Bensoukas (1999, 2004). I have
benefited from discussions with and/or comments from A. Adnor, A.
Benhallam, A. Boudlal, A. Bouhlal, P. de Lacy, F. Dell, M. Elmedlaoui, R.
Faizi, E. Iazzi, S. Imouzaz, A. Jebbour, J. McCarthy, K. Moktadir, S. Parker, J.
Pater, J. Saib, E. Selkirk, A. Youssi and the members of the phonology-
morphology research group of FLHS, Rabat. Remaining errors are my entire
responsibility.

1
accounted for in terms of an independently justified constraint
interaction driving labial dissimilation. The [a/i] alternation
affecting the initial inflectional vowel will also be addressed for the
sake of providing a complete analysis. The second hypothesis we
defend is the Augmented Stem Hypothesis (ASH). The ASH
explains the presence or absence of the prefinal vowel [a] in
surface AN forms. ANs will accordingly be subdivided into two
major classes: those that require augmentation via a prefinal vowel
and those that do not. Both the UMH and the ASH are couched in
parallel Optimality Theory (OT) (Prince and Smolensky
1993/2004; McCarthy and Prince 1993a, 1995, 1999).
The remainder of this paper is organized as follows.
Describing AN classes and allomorphs, section 2 provides a factual
background for the analyses in the paper. Section 3 characterizes
the AN morpheme as a unitary form by (i) providing an analysis of
the initial, inflectional nominal vowel and its allomorphs and (ii)
explaining the labial dissimilation affecting the AN morpheme.
Finally, section 4 deals with the augmentation involved in the
majority of AN forms.

2. Tashlhit Agentive nouns

2.1 The class of agentive nouns

As a morphological category, the agentive noun is generally

defined as the word that refers to the doer of the action expressed
by the corresponding verb. In Tashlhit, the category of ANs is not
strictly defined in such a way. Our definition, inclusive as it were,
subsumes two noun types- the doer of the action (krz/amkraz,
‘plow’) and the undergoer of the action (ls/(t)imlsi(t), ‘wear’).
Both types of noun have the nominal inflectional prefix vowel as
well as the AN morpheme {+m}.The term AN, therefore, refers to
a class defined on the basis of inclusive morphological, rather than
strictly semantic, similarity.
Not covered in this analysis are two groups of nouns. First
are a few Tashlhit nouns that are clearly ANs but which do not
have the AN affix, such as inigi ‘a witness’, iniɣi ‘a murdered
person’, and imtti ‘a dead person’. Second are Moroccan Arabic

2
loanwords with an AN meaning but without the native AN
morpheme (e.g. axrraz ‘shoemaker’, agzzar ‘butcher’, and asjjax
‘goldsmith’). Although expressing the same meaning as Tashlhit
ANs, these borrowed items are discarded due to their
morphophonological difference.

2.2 The melodic make-up of Tashlhit roots

The Tashlhit consonantal system is very rich if compared to

the vocalic one. 1 The phonemic vowels of Tashlhit are i, u and a,
which makes the repertoire of Tashlhit vowels one of the simplest.
Quantity is not contrastive on Amazigh vowels in general. The
consonantal inventory of Tashlhit consists of the stops [b, t, d, ṭ, ḍ,
k, g, kw, gw, q, qw]; the fricatives [f, s, z, ṣ, ẓ, ʃ, ʒ, Š, Ž, x, ɣ, xw, ɣw, ħ, ʕ,
ɦ]; the sonorants [m, n, l, r, ḷ, ṛ]; and the glides [w, j].
The analysis to be provided for ANs is based on the
following characterization of radical material (for a full
presentation, see Bensoukas 2001a). First, we divide verb roots into
two classes only, consonant final verb roots and vowel final ones,
each of which contains two paradigms as in (1):

(1) Tashlhit verbal roots: Four paradigms

C-final roots V-final roots
/gn/ ‘sleep’ /afa/ ‘find’
Par. A

Par. C

/mḍl/ ‘bury’ /fka/ ‘give’

/krf/ ‘tie’ /rẓa/ ‘break’
/arw/ ‘give birth to’ /k nu/
w
‘bend’
Par. D
Par. B

/asj/ ‘take/lift’ /gwru/ ‘pick up’

/fsj/ ‘melt/loosen’ /rufu/ ‘be thirsty’

Second is the assumption that the verbal forms in paradigm

(1B) end in a consonantal glide that ‘vocalizes’ in nucleus position.
Third is the idea that the imperative in paradigm (1C) is
underlyingly vowel final. Accordingly, the C-final class of roots in

1
Conventional IPA symbols are used in this paper, except for emphatics
transcribed with a dot underneath. Š and Ž are the emphatic counterparts of ʃ and
ʒ, respectively. Gemination is indicated by doubling the consonant.

3
our conception contains the items in paradigms (1A-B), while the
V-final one contains the items in paradigms (1C-D) above. A
further distinction between the items that are V-final can be
proposed on the basis of whether the final vowel is underlyingly /a/
or /u/. We have been unable to find roots ending in a vowel [i];
therefore, we will assume as provisionally correct that these
constitute a lexical gap in the grammar of Tashlhit.
Most of the irregularity characterizing Tashlhit morphology
hinges on this bipartition. V-final verb roots and C-final verb roots
behave distinctly in the various morphological classes of the
language as the following items reveal:

(2) UR Passive Act. Noun

Par. A /frn/ ttufran afran ‘sort out’
Par. B /fsj/ ttufsaj afssaj ‘untie’
Par. C /ɣra/ ttuɣra tiɣri ‘read’
Par. D /gwru/ ttgwra tigri ‘pick up’

In passive verb and action noun morphology alike, C-final verb

roots (2a) appear with a prefinal vowel. This vowel is due to a
prosodically motivated epenthesis that has some generality in
Tashlhit, applying as well to ANs. C-final verb roots, including
glide-final ones, behave in the same way with respect to this
process of epenthesis.
V-final verb roots (2b), on the contrary, do not exhibit this
epenthetic prefinal vowel, but rather remain vowel final. In other
words, a general prosodic epenthesis process is blocked in the
bases having a final vowel. The fact emerges that Tashlhit roots
consistently remain either vowel final or consonant final in the
different morphological classes, and in this they faithfully
correspond to underlying structure. Moreover, just like the other
dialects of Amazigh, Tashlhit has ‘protected vowels’, i.e. vowels
that always surface as vowels (Guerssel 1986; Rosenthall 1994), as
is clearly demonstrated by vowel epenthesis facts.

2.3 Classes of agentive nouns

(3) below contains examples of the most productive class of

AN forms in our data. ANs belonging to this class are derived by

4
prefixing the derivational, AN morpheme {m+} to the verb root
and epenthesizing a prefinal vowel, generally [a] and sporadically
[u].2 All AN nouns bear an initial inflectional morpheme, generally
a prefix vowel {a+}. The AN morpheme shows allomorphy
between [m] and [n] (see 2.4 below). The initial vowel is also
subject to contextual variation [a]/[i] as in (4iv) below:

(3) ANs: Class A

Verb AN Gloss
nj amnaj ‘rider’
mgr anmgar ‘harvestman’
krz amkraz ‘plowman’
ṛmj anṛmaj ‘tired person’
aws amawas ‘helper’
azn amazan ‘messenger’
aḍn amaḍun ‘ill person’
ags amagus ‘injured person’

The second class of ANs is in fact an amalgamation of ANs

belonging to various subclasses. What these ANs have in common
is their being different from those in class A. (4i) contains verbal
forms which have an underlying prefinal vowel maintained in ANs.
(4ii) and (4iii) contain verb forms with more than three radical
elements, including one or two vowels. The AN stems of these
verbs do not require the prefinal vowel a. In (4iv), we give
examples of the ANs appearing, quite distinctly, with an identical
initial and final vowel i.
(4) ANs: Class B
i- aẓum anaẓum ‘sb. fasting’
lluẓ amlluẓ ‘hungry person’
ḍf(u)r anḍfur ‘follower/disciple’
irir amarir ‘singer’

2
Root material is considered here to contain vowel elements as well as
consonantal elements. Espoused by many scholars working within the generative
tradition (Jebbour 1988; Moktadir 1989; Bensoukas 1994), this position is a
major departure from the belief that Amazigh roots consist of consonantal
material only (Basset 1929; 1952).

5
 ii- tabʕ amtabʕ ‘follower/disciple’
ʕ(a)ʃr amʕaʃr/ amʕʃʃr ‘roommate’
iii- ddukkl amddakkwl ‘friend’
mmuddu anmmuddu ‘traveller’
iv- rẓa imrẓi ‘sb. with a fracture’
ṛʒu imṛʒi ‘hoper’
ħḍu imħḍi ‘hider’
ddullu imddilli ‘demeaned person’

Of particular interest is the class of ANs in (4iv). Unlike other ANs,

these start with a vowel [i]. Another alternation that ANs exhibit is
instantiated by the initial vowel alternating between the more
widespread [a] and this more restricted vowel [i]. We will claim
later that a process of vowel copying is at work.
To sum up, the AN forms in (3) and (4) all consist of the
prefixal nominal inflectional morpheme {a+} and the AN
morpheme {m+}. A prefinal epenthesized vowel a (or u) is attested
in (3) but not in (4), where it is not needed. The environment of the
epenthetic vowel is regular, always being the prefinal position.
With a few exceptions, this vowel is restricted to the class
containing the items listed above, which almost all consist of three
radical elements.

2.4 Allomorphs of the agentive noun morpheme

The allomorphy exhibited by the AN morpheme is quite

systematic: The AN morpheme {m+} changes to [n] whenever the
verb root contains one of the consonants [b, f, m]. This is
exemplified by the items in (5i), in which the [m] allomorph is
attested, and those in (5ii) with the [n] allomorph:

(5) i- ANs without labial dissimilation

azn amazan ‘messenger’
krz amkraz ‘plowman’
lluẓ amlluẓ ‘hungry person’
ddukkl amddakkwl ‘friend’
ddullu imddilli ‘demeaned person’
ħḍu imħḍi ‘hider’

6
 ii- ANs with labial dissimilation
bdd anbdad ‘stand’
ʕḍb anʕḍab ‘injure’
frn anfran ‘sort out’
ḍfur anḍfur ‘follow’
gwmr angwmar ‘hunt’
ag m
w
anagam ‘draw water’

The striking fact immediately emerges that {m+} changes to [n],

irrespective of whether it is adjacent to or distant from a co-
occurring labial consonant.

3. The unitary agentive noun morpheme hypothesis

Detailed as the previous descriptions of ANs in the

literature on Amazigh in general may be, no attempt has to our
knowledge been undertaken at determining a unified AN
morpheme, much less at explaining the related allomorphies it is
subject to. Rather, forms are simply classified as starting with
[am/an] or [im/in], together with a vowel infixed prefinally or
suffixed. 3
This section is devoted to explaining the allomorphy
exhibited by the AN derivational prefix. Departing from previous
literature, we take the strongest position possible and propose the
Unitary Agentive Noun Morpheme Hypothesis (henceforth UMH):

(6) Unitary Agentive Noun Morpheme Hypothesis (UMH)

The AN morpheme in Tashlhit is the bound, non-
discontinuous morpheme {m+} prefixed to verbal bases.

The UMH states what the AN morpheme is. It can be justifiably

valid only if explanation is provided for what the AN is not, i.e. (i)
the initial, inflectional prefix vowel and (ii) the prefinal vowel.
The aim of this section is elaborate on the UMH. In 3.1, we
deal with the initial nominal vowel and its allomorphs (the [a/i]

3
See for example the general works done within the colonialist tradition
(Aspinion 1953; Laoust 1936; Renisio 1932) and the more focused linguistic
approaches (Chaker 1978; Chami 1979; Elmoujahid 1981; Penchoen 1973).

7
alternation). In section 3.2, we provide an analysis of the AN
allomorphy (the [m/n] alternation). The analysis will be completed
when the provenance of the prefinal vowel, which we claim is not
part of the AN affix, is explained, the focus of section 4.

3.1 The initial nominal vowel

In this section, we will argue in favor of distinguishing the

initial vowel from the AN morpheme, the former being an
inflectional category and the latter a derivational one. Then we
analyze how the initial vowel behaves with respect to the AN
morpheme.
In (7), the ANs inflect for case and number in the same way
ordinary, non-derived nouns do. In the most unmarked case, a
masculine noun in CS starts with a vowel [u] (7a), and one in the
plural with a vowel [i] (7b): 4
(7)
AN Non-derived noun
a. CS: amkraz/ umkraz ‘plow’ argaz/urgaz ‘man’
amaḍun/umaḍun ‘be sick’ amadl/umadl ‘hill’
b. Pl.: amkraz/ imkrazn ‘plow’ argaz/irgazn ‘man’
amaḍun/imuḍan ‘be sick’ amadl/imudal ‘hill’

Theoretical considerations, especially the partitioning of

morphology into an inflectional sub-component and a derivational
one, militate for treating the AN morpheme and the inflectional
vowel as separate entities. The opposite standpoint, that the initial
vowel is part of the morpheme, in a sense radical, is vitiated by the
behavior of this initial vowel with regard to nominal inflectional
morphology, more specifically number and case.
Now that we have established the fact that the initial vowel
is not a particularity of ANs, but rather a characteristic of all nouns
in Amazigh, we provide an analysis of the prefixal nature of the
inflectional vowel. We start our analysis with the fact that (i) the

4
CS and plural formation facts in Amazigh are extremely complex (see Basset
1932; Chaker 1988; Elmoujahid 1982; Guerssel 1983; Jebbour 1988; Idrissi
2000; Saib 1982, 1986). For the sake of simplicity, we assume that the nouns in
(7) change their initial vowel [a] to [u] in CS and to [i] in the plural.

8
initial vowel is obligatory in ANs and that (ii) it is a prefix. The
former characteristic suggests that the constraint Max-AffixNInfl
(8a) is high ranking. The latter issue ensues from the high-rank of
the constraint Align (AffixNInfl, L, Stem, L) (8b) (on Alignment, see
McCarthy and Prince 1993b, 2004a-b-c)

(8) a- Max-AffixNInfl: The input inflectional noun affix has a

correspondent in the output.
b- Align (AffixNInfl, L, Stem, L) (Align-NInfl): The
inflectional affix and the stem are left aligned.

Interaction between the inflectional nominal morpheme(s)

and derivational nominal morpheme(s) in Amazigh brings to the
fore interaction between different alignment constraints. Because
both affixes involved are prefixes, which affix is outermost
becomes an issue. A commonplace in morphology is the fact that
derivation is generally inner to inflection, a behavior obtained here
through constraint interaction.
Another alignment constraint is required, Align (AffAg.N, L,
Stem, L), the one governing the position of the derivational AN
morpheme {+m}:

(9) Align (AffAg.N, L, Stem, L) (Align-AffAg.N): The AN morpheme

and the stem are left aligned.

Interaction between these two alignment constraints accounts for

the order of the morphemes. As (10) shows, higher order Align-
NInfl ensures that the inflectional vowel is outer to the derivational
AN morpheme {m+}:
(10)
/a+m+krz/ Align-NInfl Align-AffAg.N
i. makraz *!
ii. amkraz *

(10i), which opts for the order derivation+inflection, fails at Align-

NInfl, a violation (10ii) spares. Other facts of nominal morphology,
especially those relating to case and number, show that Align-NInfl

9
is not completely undominated (see Bensoukas 2010 for details and
analysis.) 5
To rule out candidates in which the AN morpheme is away
from the left edge (*akmraz, *akrmaz...), we rely on the gradience
of violation which characterizes alignment constraints: the further
away from the left edge, the more alignment violations the AN
morpheme incurs. This positions the AN morpheme as closest as
possible to the left edge, specifically immediately after the initial
vowel. One argument we propose for this standpoint is that in the
absence of the initial vowel, the AN morpheme becomes word-
initial. A few putative frozen ANs exist that do not have the initial
vowel. In this case, Max-AffixNInfl must be dominated by another
constraint which bans the initial vowel in these items. Examples are
maf (>af ‘find’) in maf-aman ‘water-finder’, mslm (>slm ‘inhale’)
in mslm-aggwrn ‘moth’ (literally ‘flour-inhaler’), milluṭṭs/miṭṭus (>
ṭṭs ‘sleep’) and mixxi (>xxi ‘discharge excrement’). In the absence
of the intial vowel, the morpheme {m+} is aligned with the left
edge of the stem.
An additional issue with respect to the initial, nominal
vowel is its changing quality, as revealed by a group of derived and
non-derived nouns that appear with the vowel [i] instead of the
more regular [a]. Examples are given in (11):
(11)
Non-derived nouns Derived nouns
ifili ‘thread’ imddilli ‘demeaned person’
ifri ‘cave’ imkiri ‘hired person’
isli ‘bridegroom’ issgni ‘needle’

The account we provide is basically that in Bensoukas (2001a-b,

2002), in which the [a/i] allomorphy is conditioned by a general
vowel copy process in the language.
A conception of how vowel feature constraints interact and
a characterization of the spreading phenomenon are basic to the

5
Finally, the fact that syllables can be exceptionally onsetless at the beginning of
the word calls for another alignment constraint (McCarthy and Prince 1993a,
32), which dominates ONS (the requirement that syllables have onsets) namely
Align-L: [Stem = [PrWd.

10
analysis. Concerning vowel feature markedness constraints, we
assume Feature Class Theory (Padgett 1995), according to which
features are members of various classes, and constraints on well-
formedness may target individual features as well as classes of
features. (12) lists the feature classes along with their individual
members that are of relevance in the present context.

(12) Feature classes (Padgett 1995, 399): Vplace: {high, low, back,
round…}; Height: {high, low}; Color: {back, round}

Vowel markedness constraints are ranked on an individual

basis in (13a) and (13b). 6 A conflating of these individual rankings,
the one we assume, is given in (13c):

(13) a- Ranking based on place and color: *[+round], *[+back] »

*[-round], *[-back]
b- Ranking based on height: *[+high] » *[-high]
c- Ranking of vowel markedness constraints: *[+round],
*[+back], *[+high] » *[-round], *[-back], *[-high]

A convenient way of looking at vowel markedness is as a

scale of harmony (Prince and Smolensky 1993) as in (14). Thus, i
is less marked than [u], and [a] is less marked that both [i] and [u].
Using *u, *i, and *a mnemonically for the constraints that sum up
the individual markedness violations each vowel incurs, the scale in
(14a) can be converted into a constraint hierarchy as in (14b):

(14) a- Harmony scale for vowels: a > i > u

b- Vowel markedness hierarchy (mnemonically): *u » *i » *a

For ease of exposition, the mnemonic ranking in (14b) will be used

in the remainder of this section instead of the one in (13c).

6
In this conception, vowel markedness constraints are considered along two
dimensions: First, on the basis of labiality being universally more marked than
coronality (*Labial » *Coronal) (Prince and Smolensky 1993/2004), the vowel i
should emerge as less marked than u. Second, on the basis of a language-
particular preference of the low vowel over high vowels, a emerges as less
marked than the remaining two vowels.

11
 According to the precepts of Feature Class Theory, the
phenomenon of vowel copying is viewed as spreading the whole
vocalic constituent in an all-or-nothing fashion (see for example
Clements and Hume 1995; Hulst and Weijer 1995; Ní Chiosáin
1995). The result is a feature-sharing structure like the one in (15)
(b, h, and r stand for back, high and round respectively; the AN
imħḍi is derived from the verb ħḍu ‘hide’). As to the locality issue,
since vowels never occur contiguously in the language, copying is
distant. Note, however, that the vowel features associated with
glides or doubly articulated consonants never spread.

(15) Vowel copy: imħḍi

-b -h-r

Violation of constraints in vowel copying is assessed

gradiently (Beckman 1997; Ní Chiosáin and Padgett 1997; Walker
1998). In other words, a linked structure, where more than one
node is associated with one vowel feature, results in a single
violation of markedness. On the contrary, associating the vowels
involved in copying with similar or different feature specifications
in a one-to-one fashion results in multiple violations of
markedness.
Tolerating linked structures calls for a low ranking of the
constraint NOLINK (adapted from Itô, Mester and Padgett 1995),
which rules out linked structures:

(16) NOLINK: Features must not be multiply linked.

The constraint against multiple linking of individual features being

dominated, a structure containing the feature-sharing configuration
is tolerated as long as it does not incur any fatal violations of
higher-ranked constraints.
In (17), we consider an example. We assume that the final
vowel [i] is part of the morphology of derived nouns like those in
the second column in (11). As such, a noun not containing this
vowel incurs a violation of M-Real, a constraint requiring the
realization of morphemic material.

12
(17)

/a+m+ħḍu+i/ M-Real *u *i *a NOLINK

i. a m ħ ḍ a
*! * *
-b -h -r
ii. i m ħ ḍ i
*!*
-b +h -r -b +h -r
iii. a m ħ ḍ i
* *!
-b +h -r -b -h -r
iv. i m ħ ḍ i
 * *
-b +h -r

Candidate (17i) fatally violates the constraint M-Real. (17ii)

violates the high-ranking markedness constraint banning the
features of the vowel [i] twice, which turns out to be fatal.
Candidate (17iii), which epenthesizes the least marked vowel of the
language, namely [a], violates the markedness constraint banning
the features of the vowel [i] and those banning the features of [a].
The latter violation, which is fatal in the case of (17iii), is spared by
(17iv). With the violation of NOLINK being minimal, the feature-
sharing candidate (17iv) is optimal, incurring only one violation of
markedness.
Tableau (18) below clarifies the Emergence of the
Unmarked (TETU) (McCarthy and Prince 1994/2004d) effect in
Amazigh vowel copying. The markedness constraint *a, which is
otherwise crucially dominated in situations where faithfulness
prevails, becomes active in epenthetic vowel copying situations:
(18)
/a+m+ħḍu/ *i *a
i. imħḍi * *!
 ii. amħḍi *

13
In tableau (18), the candidate (18i), with the initial vowel a, incurs
a violation of the markedness constraint *a, which (18ii), the
copying candidate, does not. Since the constraint that decides is a
markedness constraint that is otherwise inactive, this is a case of
TETU in vowel copying.
To recapitulate, the changing quality of the initial nominal
vowel is shown to be the effect of vowel copying, itself driven by
markedness reduction, the copying pattern being a case of TETU.

3.2 Labial dissimilation 7

In this section, we will show that the allomorphs of the AN

morpheme, [m] and [n], are to a large extent phonologically
explainable. In Tashlhit, there is a general labial dissimilation
process, of which AN morpheme allomorphy is just an example
(see Boukous 1987; Elmedlaoui 1992/1995; Lasri 1991; Selkirk
1993, 1995; Alderete 1997; Bensoukas 1999, 2004; Lahrouchi
2001; Alderete and Frisch 2007).
In avoiding the repetition of labial specifications in the
stem, Tashlhit can resort to more than one way: deleting one of the
features, fusing the two features into a linked structure specified for
only one labial feature, and deleting the whole segment. The
language resorts to neither segment deletion nor feature fusion. The
analysis in the present paper is compatible with feature deletion
(elaboration on why the other options are discarded will take us too
far afield; for details of the analysis, see Bensoukas 1999).
Our account of labial dissimilation relies on the self-
conjoined markedness constraint [*PL/Lab & *PL/Lab]Stem,
formulated as *PL/Lab2Stem (following Alderete 1997, 11): 8

7
This section draws heavily on the analyses in Bensoukas (1999) and (2004)
and references therein. We specifically adhere to the essence of the OCP » Faith
analysis (Selkirk 1995) and the self-conjunction analysis (Alderete 1997).
8
The essence of constraint conjunction is that two violations of the same
constraint are worse when they occur in the same domain. Smolensky (1993,
1995) defines local constraint conjunction as: “The local conjunction of C1 and
C2 in domain D, C1 and C2 is violated when there is some domain of type D in
which both C1 and C2 are violated.” Universally, the conjunction of C1 and C2
dominates the individual constraints conjoined, i.e. [C1 & C2] » C1, C2.

14
(19) *PL/Lab2Stem: Ban any stem containing two segments with
independent Place specifications [labial].

*PL/Lab2Stem dominates the faithfulness constraint Ident(Lab),

forcing labial dissimilation, as tableau (20) shows:
(20)
/a+m+lmd/ *PL/Lab2Stem IDENT(Lab)
 i. anlmad *
ii. amlmad *!

One further issue in this respect relates to the fact that, of

the two labial specifications, it is always the affix labial
specification that dissimilates, but never the root one. Our account
relies on Root Faithfulness (McCarthy and Prince 1994; Selkirk
1994, 1995; Beckman 1997, 1998).
The feature [labial] is equally attested in Tashlhit roots and
affixes, which argues for the ranking of both faithfulness
constraints governing the distribution of the labial feature in roots
and affixes (given in (21a) and (21b) respectively), above the
markedness constraint *Lab as in (22):

(21) a- Ident-Root(Lab): Input and output root labial

specifications are identical.
b- Ident(Lab): Input and output affix labial specifications
are identical.
(22) a- Labial root consonants b- Labiality in affixes
mRoot Ident-R(Lab) *Lab mAffix Ident(Lab) *Lab
m * m *
n *! n *!

The rankings established in (22a-b) predict that, by virtue of its

being outranked by both general Ident(Lab) and specific Ident-
R(Lab) faithfulness constraints, the markedness constraint *Lab
will not prevent any primary labial features specified underlyingly
in roots or affixes alike from surfacing, unless faithfulness is
sacrificed to satisfy some dominating markedness constraint.

15
 Second, if faithfulness is compelled to yield in satisfaction
of some higher ranked constraint, root faithfulness wins:
(23)
/a+m+lmd/ Ident-R(Lab) Ident(Lab)
 i. anlmad *
ii. amlnad *! *

As (23) shows, the loss of the labial place of the affix is less serious
than that of the root. This behavior reveals the dominance relation
holding between Ident-Root(Lab) and Ident(Lab).
The last decision to be made concerns the interaction of
*PL/Lab2Stem and Ident-Root. In the OCP » Faith analysis, featural
markedness outranks faithfulness in order to ban the tauto-
morphemic co-occurrence of labial consonants. Taking this
consideration into account, we give in tableau (24) the constraint
hierarchy accounting for Tashlhit labial dissimilation (Selkirk
1995, 7), in which Correspondence Theory constraints are
substituted for the original ones:
(24)
/a+m+mgr/ *PL/Lab2Stem Ident-Rt(Lab) Ident(Lab)
i. ammgar *!
ii. anmgar *
iii. amngar *! *

Although totally faithful by virtue of satisfying both general and

specific Ident constraints, candidate (24i) fatally violates
*PL/Lab2Stem. Candidate (24iii) dissimilates the root labial
consonant, a fatal violation of Ident-Root(Lab). Satisfying higher
ranked *PL/Lab2Stem, candidate (24ii) wins, even though it violates
the low ranked Ident(Lab). Accordingly, markedness satisfaction is
more important than general affix faithfulness. 9 The emerging

9
The surfacing of the coronal place instead of the input labial one ensues from
two more considerations. First, all consonants should have place specified in
satisfaction of the constraint HAVEPLACE (Padgett 1995; Fukazawa 1999).
Second, the choice of the coronal feature instead of any other feature emanates
from the ranking of the featural markedness constraints proposed in Prince and
Smolensky (1993/2004): *Dorsal, *Labial » *Coronal.

16
picture is, as highly predicted, one of identity avoidance observing
positional faithfulness (for a full development of root faithfulness
effects in Tashlhit, see Bensoukas 2004).
With no further stipulation, the same ranking predicts that if
there are no inimical labial specifications in a word, the labial
consonant [m] of the morpheme will surface unscathed:
(25)
/a+m+krz/ *PL/Lab2Stem Ident-R (Lab) Ident(Lab) *Lab
i. ankraz *!
ii. amkraz *

The prediction turns out to be quite borne out. The violation of

Ident(Lab) is gratuitous in (25i), which makes the candidate lose to
the faithful one, the optimal candidate (25ii). This is clear proof
that violation of faithfulness in the case of primary labial
dissimilation is forced only by the dominating markedness
constraint *PL/Lab2Stem. Thus, dissimilation in our analysis
systematically ensues from independently justified markedness and
positional faithfulness considerations.
To summarize, our central claim is that AN formation takes
place in two modules. First, the AN stem is constructed which
comprises the AN morpheme {m+} together with the melodic
constituents of the lexical entry of the verb, the base of the
derivation in our analysis. This first step corresponds to the
derivational, in contradistinction to the inflectional, morphology of
the language. The inflectional element is also affixed and is itself
subject to allomorphy affecting its featural makeup. Now, we deal
with the allomorphy affecting the AN base.

4. Prosodic augmentation in Tashlhit ANs

This section accounts for the prefinal vowel in AN bases.

The analysis now comes full circle, after we have established the
inflectional status of the initial vowel and accounted for the
allomorphy affecting the AN morpheme.
We start with the prosodic structure of AN bases. Following
Jebbour (1996, 54; 1999), we assume that the minimal syllable
consists of one mora (for example a ‘hey’, fi ‘suppurate’) and the

17
maximal one of two (for example ut ‘hit’, mun ‘accompany’). A
weight distinction between syllables is motivated in which the
status of the nucleus as vocalic or consonantal plays a decisive role
(Jebbour 1996, 55-57). 10 A heavy syllable consists of a nucleus
vowel that is followed by a singleton or a geminate consonant as in
the words ut ‘hit’, mun ‘accompany’, ass ‘day’, and bidd ‘stand’. A
light syllable, on the other hand, is either an open syllable with a
nucleus vowel or any syllable with a consonantal nucleus as in the
following examples, in which the consonantal nucleus is
underlined: a ‘hey’, fi ‘suppurate’, g ‘be’, gn ‘sleep’, ls ‘wear’, krz
‘harvest’, ʃʃ ‘eat’, qzz ‘look’.
In case a syllabic consonant is followed by a coda, both the
nucleus and the coda of the syllable in question are dominated by a
branching mora. A leftover consonant in the case of a syllable
headed by a vowel is provided a mora by the Weight-by-Position
convention (Hayes 1989), whereas a leftover consonant in the case
of a syllable headed by a consonant is attached directly to the
syllable node. Finally, following Dell and Elmedlaoui (1988),
Jebbour proposes that the first element of a geminate cannot be the
onset of a syllable, forcing initial geminates to be split between two
syllables. A final geminate consonant can be hosted by the same
syllable.
Consideration of the agentive nouns in Tashlhit yields the
following generalizations, and the output prosodic structure of the
ultimate syllable of agentive nouns is summarized in the table in
(26):
1. Bi/Triconsonantal roots and vowel initial trisegmental bases
have corresponding ANs derived by prefixation and prefinal
vowel epenthesis.
2. Verbs containing a prefinal or medial vowel undergo
affixation only.
3. Verbs containing a final vowel undergo prefixation and
appear with a final vowel [i].

10
In terms of its prosodic structure, Tashlhit is notable for the fact that any
segment in its inventory, even voiceless stops, can behave as a syllable peak
provided it is the most sonorous sound in the string to be syllabified (Boukous
1987, 2009; Clements 1997; Dell and Elmedlaoui 1985, 1988, 2002; Jebbour
1996, 1999; Prince and Smolensky 1993/2004; Ridouane 2008).

18
(26) Agentive noun stem prosody:
Augmented stems Unaugmented stems
Root Stem σ# Root Stem σ#
a. mgr anm[gar] a. σµµ a. bur an[bur] a. σµµ
agwm ana[gam] b. aẓum ana[ẓum] b. σµµ
awj ama[waj] c. ʒlu imʒ[li] c. σµ
b. gn amg[gan] b. σµµ kla imk[li] σµ
c. nj am[naj] c. σµµ ggalla imggil[li]
d. matr anmma[tr] σµ
ddukkwl amddak[kwl] σµ
sijħ amsi[jħ] σµ

The augmented nominal stems all exhibit a final heavy syllable

containing a vowel which does not belong to the root, whereas the
non-augmented ANs contain a heavy syllable because they have
enough root material to sustain it. The cell with the double outside
border contains items expected to show the prefinal vowel but
which fail to do so.
As a unitary account of the major bifurcation of AN stems
into augmented and unaugmented forms, we propose the Stem
Augmentation Hypothesis (henceforth SAH), stating roughly that
AN stems end in a final heavy syllable. We will argue in this
section that the prefinal vowel in most ANs is an epenthetic vowel
that has no lexical status, and as such cannot be considered part of
the AN morpheme. By epenthesis, we do not mean the exclusively
syllable-motivated epenthesis documented in Selkirk (1981) and Itô
(1986, 1989), for example. The vowel is epenthesized in this case
for the sake of augmenting a morphological stem (Morén’s 1999,
2003).

4.1 Augmented agentive noun stems

As the presentation in section 2.3 revealed, most C-final
roots end in a heavy syllable. For ease of exposition, a few
examples are given in (27):
(27) /nj/ amnaj ‘ride’
/sɣ/ amssaɣ ‘buy’
/awj/ (t)amawaj(t) ‘take away’

19
 /agwm/ anagam ‘draw water (from a well)’
/aws/ amawas ‘help’
/gwmr/ angwmar ‘hunt’
/lmd/ anlmad ‘learn’

We will claim that the grammar of Tashlhit has a strict requirement

that the stem final syllable be heavy, encoded in the constraint in
(28). We can think of this constraint as an Alignment constraint
requiring the coincidence of prosodic and morphological
constituents:

(28) Final Heavy Syllable (σµµ]): stems must contain a final heavy
syllable.

This constraint will rule out any stem that does not end in a heavy
syllable. On the other hand, given that the site of the epenthetic
vowel is consistently the prefinal position, this turns out to just
make the final syllable heavy, and the constraint σµµ] rules in
candidates which epenthesize a prefinal vowel to augment the AN
stem. In the spirit of OT, this constraint can be violated as long as it
is dominated.
The possibility of vowel epenthesis calls for a low ranked
faithfulness DEP-V constraint (29), which otherwise militates
against vowel epenthesis:

(29) DEP-V: A vowel in the output must have a correspondent in

the input (Vowel epenthesis is prohibited.)

The first ranking argument we will propose is the one determining

the interaction of the requirement to have final heavy syllables in
stems and the faithfulness constraint on vowel epenthesis, namely
σµµ] » DEP-V, illustrated with the form /mgr/ ‘harvest’ (in the
candidates in (30), the dots represent syllabic division):
(30)
/a+m+mgr/ σµµ] DEP-V
i. a.nm.gr. *!
 ii. a.nm.gar. *

20
Through this ranking alone, we obtain both the application of
vowel epenthesis and its prefinal site.
Kager (1999, 105) makes the following prediction about
epenthesis. It always applies minimally, that is precisely to the
extent that is necessary to improve structural markedness.
Moreover, epenthesis never results in an increase, but rather a
decrease, of structural markedness. By providing a vocalic nucleus
to a consonant cluster, Tashlhit vowel epenthesis results in the
decrease of structural markedness by reducing the number of
consonantal-peak syllables. Epenthesis also applies minimally, as it
never affects syllables with a vowel.

4.2 Unaugmented agentive noun stems

Exemplified in (31) is the class of ANs that differ from the

others by virtue of their verbal roots containing a final vowel.
These ANs quasi-uniformly have a final vowel i, whatever the
underlying specification of the root final vowel. We consider the
vowel i a requirement of nominal morphology. Since the verbs in
(31) always appear with the vowel i in their nominal forms, we
assume that the features of the vowel i make part of this
morphological component:

(31) Agentive noun roots containing a final vowel:

/ʒlu/ imʒli ‘lose’
/gwnu/ imgni ‘sew’
/ħḍu/ imħḍi ‘hide’
/ddullu/ imddilli ‘be demeaned’
/kla/ imkli ‘spend the day’
/rẓa/ imrẓi ‘break’
/lsa/ (t)imlsi(t) ‘wear’

While vowel epenthesis as an option to make heavy the

final syllable of stems is possible in C-final roots, consonant
epenthesis is not in the case of vowel-final roots, nor is vowel
lengthening. This ensues from the interaction of σµµ] with the
constraints in (32a) DEP-C (McCarthy and Prince 1995, 1999) and
No Long Vowels (Rosenthall 1994, 1997). The ranking that bans

21
consonant epenthesis and vowel lengthening is given in (33), where
NLV and DEP-C are not ranked with respect to each other.

(32) a- DEP-C: A consonant in the output must have a

correspondent in the input (Consonant epenthesis is
prohibited.)
b- NO LONG VOWELS (NLV): Two moras cannot link to a
single vocalic root node.
(33) NLV , DEP-C » σµµ] » DEP-V
/a+m+rẓa+i/ NLV DEP-C σµµ] DEP-V
i. i-m-r.ẓii. *! *
ii. i-m-r.ẓi. *!
iii. i-m-r.ẓiC. *!

When the input contains a final vowel, only consonant epenthesis

or vowel lengthening will derive the desired final heavy syllable
result. However, these two options only do so by fatally violating
DEP-C and NLV, respectively. Thus, vowel final roots show more
faithfulness to the input than consonant final ones.
Quite revealing of an epenthesis process at work, ANs with
an underlying prefinal vowel are not subject to vowel epenthesis.
Epenthesis is expected to apply only when necessary. In case the
root contains a prefinal vowel, the requirement for the stem to
contain a final heavy syllable is satisfied from the outset. The
following ANs provide illustration:
(34) /aẓum/ anaẓum ‘fast’
/lluẓ/ amlluẓ ‘be hungry’
/bur/ anbur ‘become a spinster’
/mjur/ anmjur ‘be accustomed’
Our ranking so far accounts for the facts in quite a straightforward
manner:
(35)
/a+m+aẓum/ NLV DEP-C σµµ] DEP-V
i. a.na.ẓu.ma *! *
ii. a.na.ẓum
iii. a.na.ẓu.maC *! *

22
Epenthesis not resulting in a final heavy syllable, candidate (35i)
incurs a gratuitous DEP-V violation in addition to a violation of
σµµ]. Though satisfying σµµ], candidate (35iii) violates both DEP-
C, fatally, and DEP-V. Only the faithful candidate (35ii) is optimal
in this case. So the ranking established so far shows how economy
of epenthesis is derived in the sense that epenthesis only applies
when needed.
So far, one possibility we have not considered via which the
final heavy syllable result may be achieved, but which should be
discarded immediately, is the deletion of some of the root material
in long roots. Examples are:
(36)
ddullu i-m-[d.dil.li] / *i-m-[d.dil] ‘be demeaned’
mmuddu a-n-[m.mud.du] / *a-n-[m.mud] ‘travel’

Tashlhit observes a requirement not to delete a vowel or a

consonant although the resulting structure may conform to the final
heavy syllable mold.
This we claim to be the result of the constraint MAX
(McCarthy and Prince 1999, 294) being ranked higher in the
constraint hierarchy than σµµ]:

(37) MAX: Every element of the input has a correspondent in the

output.
Simply put, every segment in the input has a corresponding one in
the output, thus banning consonant/vowel deletion.
Tableau (38) provides a ranking argument for the constraint
demanding final heavy syllables and the one militating against
deletion of root material:

(38) /a+m+ddullu+i/ MAX σµµ]

i. i.md.dil *!
ii. i.md.dil.li *

Candidate (38i) does not maximally parse root material with the
aim of achieving the final heavy syllable requirement; accordingly,
(38i) fails in the competition in front of the faithful candidate
(38ii).

23
 To sum up, the driving force of epenthesis is blocked in
vowel final roots by either the markedness constraint NLV or the
faithfulness constraint DEP-C. These have no effect on consonant
final stems. No provable ranking holds between the constraints
NLV and DEP-C and MAX. Therefore, these will appear at the top
of the hierarchy: NLV, DEP-C, MAX »σµµ] »DEP-V

4.3 Problematic AN forms

Problematic to the analysis are two groups of ANs. First,

certain bases having a medial geminate tend to show an
idiosyncratic behavior in that they do not call for the epenthesis of
a vowel although the motivation is not clear behind their resisting
the epenthetic vowel. This is the case of frrk/ anfrrk ‘boast’ and
qrrs/ amqrrs ‘be on the alert’, for example. The verb ɦḍr ‘play’
does not have the same input structure but has an identical output
AN as the two verbs above, i.e. amɦḍḍr.
Second, there is a problematic class containing ANs derived
from bases containing a medial vowel, in which case the base
undergoes the process of prefixation only. 11 Consider the
following examples:

(39) /ddukkwl/ amddakkwl ‘befriend’

/ggunzr/ amggunzr ‘have a nosebleed’
/matr/ anmmatr ‘guard’
/nakṛ/ amnnakṛ ‘deny’

The class of agentive nouns in (41) is challenging in the sense that

although its members are expected to display a prefinal vowel by
virtue of their corresponding roots being C-final, no such behavior
is displayed.
As things stand, we do not have a satisfactory analysis. If
verbal morphology is taken into consideration, the items in (39)
turn out to be quite intriguing by virtue of their containing a final

11
Note the exceptional form anḍalab derived from ḍalb (beg), a loan-word,
containing a medial vowel but nonetheless having a corresponding agentive noun
that does exhibit vowel epenthesis.

24
heavy syllable (ttdukkul, ttgunzur, ttmatar, and ttnakaṛ (see
Bensoukas 2001a for an account).

5. Conclusion

Instead of conceding with the early Berberists’

characterization of Amazigh nominal morphology as an
ungoverned component, we have shown how the intricacies of AN
formation in Tashlhit are amenable to a unitary account.
A proper characterization of the AN morpheme as a prefix
{m+} has been undertaken on the basis of explaining two different
aspects: (i) The initial vowel has been isolated as an inflectional
morpheme, and (ii) the prefinal vowel, we have argued, is the result
of an augmenetation operation affecting the stem so that it ends in a
heavy syllable. In so doing, we defended the UMH and the ASH.
Along the way, we have accounted for the allomorphies that are
exhibited by the AN morpheme itself and the inflectional nominal
vowel, covering thus a lot of ground in the morphophonology of
ANs in particular, and by the same token that of the language in
general.
The analysis provided herein can be extended mutatis
mutandis to other morphological categories. The morphology of
passive and intensive verb forms as well as that of action noun
forms seems to abide by the same principle accounting for AN
morphology (see Bensoukas 2001a). Another direction for future
research is to consider how AN formation works in other Amazigh
dialects, as a comparative approach may reveal aspects of the
morphology that are not readily visible on the basis of Tashlhit
facts alone.

25
 REFERENCES

Alderete, John. 1997. “Dissimilation as Local Conjunction.” North

East Linguistic Society 27: 17-32.
Alderete, John D. and Stephan A. Frisch. 2007. “Dissimilation in
Grammar and the Lexicon.” In de Lacy, Paul ed., The
Cambridge Handbook of Phonology, 379-398. Cambridge:
Cambridge University Press.
Aspinion, Robert. 1953. Apprenons le Berbère: Initiation aux
Dialectes Chleuhs. Rabat: Félix Moncho.
Basset, André. 1929. La Langue Berbère. Morphologie. Le Verbe-
Etude de Thèmes. Paris: Librairie Ernest Leroux.
Basset, André. 1952. La Langue Berbère. First published for the
International African Institute; reprinted in 1969 by Dawsons of
Pall Mall, London.
Basset, André. 1932. “Note sur l’Etat d’Annexion en Berbère.”
Bulletin de la Société de Linguistique de Paris 33/2: 173-174.
Beckman, Jill. 1997. “Positional Faithfulness, Positional
Neutralization and Shona Vowel Harmony.” Phonology 14: 1-
46.
Beckman, Jill. 1998. Positional Faithfulness. Ph.D. diss.,
University of Massachusetts, Amherst.
Bensoukas, Karim. 1994. Tashlhit Agentive Nouns- An Optimality-
Theoretic Approach. D.E.S. diss., Mohammed V University,
Faculty of Letters, Rabat.
Bensoukas, Karim. 1999. “Labial Alternations in Tashlhit
Revisited: A Constraint-Based Analysis.” Ms. Mohammed V
University, Faculty of Letters, Rabat.
Bensoukas, Karim. 2001a. Stem Forms in the Nontemplatic
Morphology of Berber. Doctorat d’Etat diss., Mohammed V
University, Faculty of Letters, Rabat.
Bensoukas, Karim. 2001b. “Markedness and epenthetic quality in
Tashlhit imperfective verbs: An OT approach.” Linguistic
Research 6.1: 81-123.
Bensoukas, Karim. 2002. The Emergence of the Unmarked in
Berber Epenthetic Vowel Quality. Ms. Mohammed V
University, Faculty of Letters, Rabat.

26
Bensoukas, Karim. 2004. “Markedness, Faithfulness and
Consonant Place in Tashlhit Roots and Affixes.” Langues et
Littératures 18: 115-153.
Bensoukas, Karim. 2010. “Construct State Allomorphy in
Amazigh: What can Alignment Explain?” ms. UM5-Agdal,
Rabat.
Boukous, Ahmed. 1987. Phonotactique et Domaines Prosodiques
en Berbère. Doctorat d’Etat diss., Université Paris VIII-
Vincennes, Saint-Denis.
Boukous, Ahmed. 2009. Phonologie de l’Amazighe. Rabat:
IRCAM publications.
Chaker, Salem. 1978. Un Parler Berbère d’Algérie (Kabylie)-
Syntaxe. Doctorat d’Etat diss., Université Paris V.
Chaker, Salem. 1988. “L’Etat d’Annexion du Nom.” Encyclopédie
Berbère V: 686-695. Available at:
http://www.centrederechercheberbere.fr/tl_files/docpdf/annexion.pdf.
Chami, Mohamed. 1979. Un Parler Amazigh du Rif- Approche
Phonologique et Morphologique. Third Cycle diss., Université
Paris V.
Chomsky, Noam and Morris Halle. 1968. The Sound Pattern of
English. New York: Harper and Row.
Clements, Georges N. 1997. “Berber Syllabification: Derivations
or Constraints.” In Roca, Iggy ed., Constraints and Derivations
in Phonology, 289-330. Oxford: Clarendon Press.
Clements, Georges. N. and Elisabeth V. Hume. 1995. “The internal
Organization of Speech Sounds.” In Goldsmith, John. ed., The
Handbook of Phonological Theory, 245-306. Cambridge, MA:
Blackwell.
Dell, François and Mohamed Elmedlaoui. 1985. “Syllabic
Consonants and Syllabification in Imdlawn Tashlhiyt Berber.”
Journal of African Languages and Linguistics 7: 105-130.
Dell, François and Mohamed Elmedlaoui. 1988. “Syllabic
Consonants in Berber: Some New Evidence.” Journal of
African Languages and Linguistics 10: 1-17.
Dell, François and Mohamed Elmedlaoui. 2002. Syllables in
Tashlhiyt Berber and in Moroccan Arabic. Dordrecht: Kluwer.
Elmedlaoui, Mohamed. 1992/1995. Aspects des Représentations
Phonologiques dans Certaines Langues Chamito-sémitiques.

27
 Ph.D. diss., Mohammed V University, Faculty of Letters,
Rabat. [Published version Rabat: Faculty of Letters.]
Elmoujahid, El Houssain. 1981. La Classe du Nom dans un Parler
de la Langue Tamazight: Le Tachelhiyt d’Igherm. Third Cycle
diss., Université Paris V, René Descartes.
Elmoujahid, El Houssain. 1982. “Un Aspect Morphologique du
Nom en Tamazight: L’Etat d’Annexion.” Langues et
Littératures 2: 47-62.
Fukazawa, Haruka. 1999. Theoretical Implications of OCP Effects
on Features in Optimality Theory. Ph.D. diss., University of
Maryland at College Park.
Guerssel, Mohamed. 1983. “A Phonological Analysis of the
Construct State in Berber.” Linguistic Analysis 11: 309-330.
Guerssel, Mohamed. 1986. “Glides in Berber and Syllabicity.”
Linguistic Inquiry 17: 1-12.
Hayes, Bruce. 1989. Compensatory Lengthening in Moraic
Phonology. Linguistic Inquiry 20: 253-306.
Hulst, Harry van der and J. van de Weijer. 1995. “Vowel
Harmony.” In Goldsmith, John. (ed.). The Handbook of
Phonological Theory, 495-534. Cambridge, MA.: Blackwell.
Idrissi, Ali. 2000. “On Berber Plurals.” In Lecarme, J.,
Lowenstamm, J. and U. Shlonsky (eds.). Research in
Afroasiatic Grammar, 101-124. Amsterdam: John Benjamins.
Itô, Junko. 1986. Syllable Theory in Prosodic Phonology. Ph.D.
diss., University of Massachusetts, Amherst.
Itô, Junko. 1989. “A Prosodic Theory of Epenthesis.” Natural
Language and Linguistic Theory 7: 217-259.
Itô, Junko, Armin Mester and Jaye Padgett. 1995. “Licensing and
Underspecification in Optimality Theory.” Linguistic Inquiry
26: 571-613.
Jebbour, Abdelkrim. 1988. Processus de Formation du Pluriel
Nominal en Tamazight. Tachelhit de Tiznit- Approche Non-
Concaténative. D.E.S. diss., M5 University, FLSH, Rabat.
Jebbour, Abdelkrim. 1996. Morphologie et Contraintes
Prosodiques en Berbère Tachelhit de Tiznit- Analyse
Linguistique et Traitement Automatique. Doctorat d’Etat diss.,
M5 University, FLSH, Rabat.

28
Jebbour, Abdelkrim. 1999. Syllable Weight and Syllable Nuclei in
Tachelhit Berber of Tiznit. Cahiers de Grammaire 24
“Phonologie: théorie et variation”: 95-116.
Kager, René. 1999. Optimality Theory. Cambridge: Cambridge
University Press.
Lahrouchi, Mohamed. 2001. Aspects Morpho-Phonologiques de la
Dérivation Verbale en Berbère (Parler Chleuh d’Agadir).
Contribution à l’Etude de l’Architecture des Gabarits. Doctoral
diss., Université Paris VII.
Lasri, Ahmed. 1991. Aspects de la Phonologie Non-Linéaire du
Parler Berbère Chleuh de Tidli. Doctoral diss., Université de la
Sorbonne Nouvelle, Paris III.
Laoust, Emile. 1936. Cours de Berbère Marocain- Dialectes du
Sous, du Haut et de l’Anti Atlas. (2nd edn.). Paris: Société
d’Editions Géographiques, Maritimes et Coloniales.
McCarthy, John J. and Alan Prince. 1993a. Prosodic Morphology I:
Constraint Interaction and Satisfaction. Ms. University of
Massachusetts, Amherst and Rutgers University.
McCarthy, John J. and Alan Prince. 1993b. Generalized Alignment.
In Booij, G. And J. van Marle. eds., Yearbook of Morphology
1993, 79-153. Dordrecht: Kluwer.
McCarthy, John J. and Alan Prince. 1994. “The Emergence of the
Unmarked: Optimality in Prosodic Morphology.” North East
Linguistic Society 24: 333-379.
McCarthy, John J. and Alan Prince. 1995. Faithfulness and
Reduplicative Identity. UMOPL 18: Papers in Optimality
Theory. [Rutgers Optimality Archive#60].
McCarthy, John J. and Alan Prince. 1999. “Faithfulness and
Identity in Prosodic Morphology.” In Kager, R., van der Hulst,
H. and W. Zonneveld, eds. The Prosody-morphology Interface,
218-309. Cambridge: Cambridge University Press.
McCarthy, John J. and Alan Prince. 2004a. “Generalizaed
Alignment: Introduction to the Theory.” In McCarthy, John J.
ed., Optimality Theory in Phonology- A Reader, 72-76.
Malden, MA: Blackwell.
McCarthy, John J. and Alan Prince. 2004b. “Generalizaed
Alignment: Prosody.” In McCarthy, John J. ed., Optimality
Theory in Phonology- A Reader, 167-177. Malden, MA:
Blackwell.

29
McCarthy, John J. and Alan Prince. 2004c. “Generalizaed
Alignment: The Prosody-Morphology Interface.” In McCarthy,
John J. ed., Optimality Theory in Phonology- A Reader, 451-
463. Malden, MA: Blackwell.
McCarthy, John J. and Alan Prince. 2004d. “The Emergence of the
Unmarked.” In McCarthy, John J. ed., Optimality Theory in
Phonology- A Reader, 483-494.Malden, MA: Blackwell.
Moktadir, Khalid. 1989. The Passive Form in Tashlhiyt Berber- A
Prosodic Approach. D.E.S. diss., M5 University, FLSH, Rabat.
Morén, Bruce T. 1999. Distinctiveness, Coercion and Sonority: A
Unified Theory of Weight. Ph.D. diss., University of Maryland
at College Park.
Morén, Bruce T. 2003. “Weight Typology: An Optimality
Theoretic Approach.” The Linguistic Review 20: 281-304.
Ní Chiosáin, Maire. 1995. “Barra Gaelic Vowel Copy and (non-)
Constituent Spreading”. West Coast Conference on Formal
Linguistics 13: 3-15.
Ní Chiosáin, Maire. and Jaye Padgett. 1997. “Markedness,
Segment Realization and Locality in Spreading.” Report n°
LRC-97-01, Linguistics Research Center, UCSC, Santa Cruz,
CA.
Padgett, Jaye. 1995. “Feature Classes.” University of
Massachusetts Occasional Papers 18: Papers in Optimality
Theory: 385-420.
Penchoen, Thomas G. 1973. Tamazight of the Ayt Ndhir. Los
Angeles: Undena Publications.
Prince, Alan and Paul Smolensky. 1993/2004. Optimality Theory:
Constraint Interaction in Generative Grammar. Malden, MA,
and Oxford, UK: Blackwell. [Revision of 1993 technical report,
Rutgers University Center for Cognitive Science.]
Renisio, A. 1932. Etude sur les Dialectes Berbères des Beni
Iznassen, du Rif et des Senhaja de Sraïr. Paris: Leroux.
Ridouane, Rachid. 2008. “Syllables without Vowels: Phonetic and
Phonological Evidence from Tashlhiyt Berber.” Phonology 25:
321-359.
Rosenthall, Sam. 1994. Vowel/Glide Alternation in a Theory of
Constraint Interaction. Ph.D. diss., UMass, Amherst.
Rosenthall, Sam. 1997. “The Distribution of Prevocalic Vowels.”
Natural Language and Linguistic Theory 15: 139-180.

30
 Saib, Jilali. 1982. “Initial Vowel Syncope and Reduction in
Tamazight-Berber Nouns.” Langues et Littératures 2: 159-184.
Saib, Jilali. 1986. “Noun Pluralization in Berber: A Study of
Internal Reconstruction.” Langues et Littératures 5: 109-133.
Selkirk, Elizabeth. 1981. “Epenthesis and Degenerate Syllables in
Cairene Arabic.” MIT Working Papers in Linguistics 3:
Theoretical Issues in the Grammar of Semitic Languages: 209-
232.
Selkirk, Elizabeth. 1993. “[Labial] Relations.” Ms. University of
Massachusetts, Amherst.
Selkirk, Elizabeth. 1994. “Optimality Theory and Featural
Phenomena.” Lecture notes, LING 730, University of
Massachusetts, Amherst.
Selkirk, Elizabeth. 1995. “Language-Particular Violation of a
Universal Constraint: The OCP from the Perspective of
Optimality Theory.” Paper read at Table Ronde, Marrakech,
13-15 January 1995.
Smolensky, Paul. 1993. “Harmony, Markedness, and Phonological
Activity.” Handout from ROW-1, Revised. [Rutgers Optimality
Archive#87].
Smolensky, Paul. 1995. “On the Internal Structure of the Constraint
Component Con of UG.” Handout of talk, UCLA, April 7,
1995. [Rutgers Optimality Archive#86].
Suzuki, Keiishiro. 1998. A Typological Investigation of
Dissimilation. Ph.D. diss., The University of Arizona.
Walker, R. 1998. A Minimal Triplet in Altaic: Round Licensing,
Harmony, and Bisyllabic Triggers. Ms. University of
California, Santa Cruz. [Rutgers Optimality Archive #263].
Yip, Moira. 1988. “The Obligatory Contour Principle and
Phonological Rules: A Loss of Identity.” Linguistic Inquiry 19:
65-100.
Yip, Moira. 1995. “Repetition and Its Avoidance: The Case of
Javanese.” Ms. University of California, Irvine. [Rutgers
Optimality Archive#83]
Yip, Moira. 1998. “Identity Avoidance in Phonology and
Morphology.” In Lapointe, S., Brentari, D. and P. Farrell, eds.
Morphology and its Relation to Phonology and Syntax, 216-
246. Stanford: CSLI publications.

31

View publication stats