Berber Negation in Diachrony Brugnatelli

Embed Size (px)

Citation preview

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    1/17

    Berber negation in diachrony*

    Vermondo BrugnatelliUniversity of Milan-Bicocca

    e morphosyntax of negation in Berber is rich and complex, and appears

    to be the outcome of multiple processes that have taken place over different

    time-periods from prehistory to the present day. e most noteworthy issueis the tendency towards a redundant marking of negation, not only by means

    of discontinuous morphemes (circumfixes) but also through the use of special

    “negative verb stems” – a feature that is attested in nearly all of the Berber-

    speaking area, regardless of the type of negative affixes in use. In this paper,

    I attempt to single out the main processes that have led to the present stage,

    taking into account the etymologies of prefixal and suffixal negative morphemes,

    the origin of negative stems and the role of the so-called Jespersen cycle in the

    evolution of Berber negation.

    e Berber languages are a linguistic family 1 that is scattered right across Northern

    Africa, from the western oases of Egypt to the Atlantic Ocean and from theMediterranean coast to the southern borders of the Sahara. ey make up a branch

    of the Hamito-Semitic (Afroasiatic) macro-family and are still spoken by some

    forty million people, although dialects of the Arabic language are nowadays also

    spoken in many regions that were previously only Berber-speaking.

    * In this paper, Arabic is written according to the standard transcription, while Berber is

    written according to the standard orthography of Kabyle, whenever appropriate. e differ-

    ences between both systems are as follows: Berber 〈 c 〉 stays for Arabic 〈 š  〉, 〈   〉 for 〈  ġ  〉, 〈 x 〉 

    for 〈 h    〉, 〈  〉 for 〈   〉. Moreover, the interdental fricatives [θ] and [ð] are represented by 〈 t     〉 and

    〈 d     〉, schwa [ə] is represented by

    〈 e

    〉 and Tuareg [e] is written

    〈é〉. In Medieval Berber, whatI transcribe 〈 g〉 may represent both [g] and [g]. Abbreviations in the examples are:  =

    aorist, aorist particle;  = auxiliary verb;  = negator, negative particle;  = negative

    perfective;  = orientation particle;  = participle, mark of participle;  = perfective;

     = plural;  = predicative particle. e sign = marks an amalgam, a dash separates clitics,

    while underscore marks the existence of sandhi phenomena.

    . From a linguistic point of view, Berber is more a linguistic family than a language.

    However, many people, especially in countries where it has little or no recognition, still regard

    it as one language, for ideological and political (extra-linguistic) reasons involving prestige.

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    2/17

      Vermondo Brugnatelli

    With regard to negation, the Berber languages display several clear-cut com-

    mon features but also many innovations that are peculiar to isolated dialects or

    shared by a small number of languages. A comprehensive description of such a

     vast and manifold subject has not yet been written, although many partial sketches

    and studies are already available.2 As far as comparative linguistics is concerned,

    an overall outline of Berber negation within a Hamito-Semitic framework is given

    in Brugnatelli (2006).

    From a diachronic perspective, there are many interesting phenomena to be

    observed in the domain of negation in relation to both historic and pre-historic

    times. Given that it would be both over-lengthy and superfluous to deal with all of

    these in the current paper, I have chosen to focus on the key aspects that appear tooffer the most potential for cross-linguistic comparison, namely the origin of the

    negative particles, the negative forms of the verb and the question of “discontinu-

    ous negation” and how it relates to the issue of Arab-Berber interference.

    . Origin of the negative particles

    e Berber languages possess many different negative particles. Some of these are

    widely attested and quite likely to have derived from a common element, while

    others are restricted to smaller areas and may nearly always be considered innova-

    tions. e creation of new negative elements took place above all in the domain of verbless sentences, possibly as a consequence of contact with Arabic, a language

    that tends to use different negative strategies for nominal and verbal sentences

    (Brugnatelli 2006).3

    e negation used with verbs oen consists of circumfixes:4 one particle (1)

    precedes the verb and another (2) follows it. However, 2 is not used in the

    . For a general picture of negation in the Berber languages, letting aside the specific issues

    that will be dealt with below, see the contributions in Chaker & Caubet (1996) and the papers

    of Galand (1994), Brugnatelli (2006), Lucas (2007), Mettouchi (2009), Taine-Cheikh (2011).

    . Some of these new negative tools are dealt with in Kahlouche (2000), Brugnatelli (2010),and Laioui (1996, 2007: 234–236; 2011: 62–69). For the sake of convenience, the present

    paper will deal mostly with verbal negation.

    . Circumfixes are frequent in the Berber grammar: in addition to their use in negation,

    they are normally used in nouns (for which gender and number are expressed by prefixes +

    suffixes) and in the conjugation of verbs (see Footnote 12 below). e sporadic “doubling”

    of clitics before and aer the verb could also be viewed as the development of circumfixes;

    it occurs in Libya (Zuara), Algeria (Mzab, Aurès, Ngouça): Brugnatelli (1993: 234–237) and

    Morocco (Tarifit: Laioui 2007: 128).

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    3/17

      Berber negation in diachrony

    southern area of the Berber domain: Zenaga (Mauritania), Tashelhit (Southern

    Morocco) and Tuareg, nor in the Saharan varieties of Mzab, Ouargla, Ghadames,

    Sokna, El-Fogaha and Siwa, nor in Yefren (Tripolitania).5 On the other hand, some

    languages in Tunisia and Libya omit 1 and use 2 as the only negator: Sened

    (Provotelle 1911: 26) and Awjila (van Putten 2013: 83–85, 263). e optional omis-

    sion of 1 is also attested in Zuara (Tunisia; Mitchell 2009: 100–103) and in

    Western Tarifit (Morocco; Laioui 2007: 234–235).

    In addition, many verbs take on peculiar stems in negative sentences. Nearly

    all the Berber languages possess such negative forms in the perfective, while a

    smaller, though appreciable, number also have a negative imperfective (Kossmann

    1989; Brugnatelli 2002; Laioui 2007: 175–176).6 is redundant way of markingnegative sentences through verbal morphology is also attested in most of the lan-

    guages that do not use post-verbal negative particles.

    . e preverbal negative particle

    e negator that is placed before the verb – in some dialects also before nominal

    predicates – takes many forms, which may easily be taken as cognates:wәr /ur , wәl /

    ul , wә/u, wәd /ud , etc.7 Other preverbal negators, which do not require 2 aer

    the verb, are attested in some Libyan dialects: Sokna (Fezzan) ingî  (mostly), but

    also yul , lā, abû; El Fogaha (Fezzan) (e)nk and bak; Ghadames (near the Tunisian

    and Algerian borders) ak; Yefren (Tripolitania) mi.8

    . Some of these languages have innovated the negator. See below, § 1.1.

    . As a general rule, most of the Berber languages have a threefold verbal system based on

    the opposition between a perfective and an imperfective stem, while a third stem, called aorist,

    is unmarked as far as aspect is concerned and has distinctive, mostly modal, functions.

    . An interesting restriction in the use of this particle is the fact that in most languages it

    cannot be used with the aorist, but only with perfective or imperfective forms. To examine

    the origin of this state of affairs would be a lengthy process and outside the scope of this

    paper, given that it would also require a wider investigation of the genesis of the modern

    tripartite TAM system (aorist, perfective, imperfective), which is not that originally in use but

    appears to be the result of long-term processes that reshaped the entire verbal system (Galand2010: 202–207).

    . Most of these particles have not yet been investigated. With regard to mi in Yefren, see

    Brugnatelli (2014: 130). e origin of the preverbal negator ak used in Ghadames as an alter-

    native to wel  is puzzling and difficult to account for. Mettouchi (1996:191) considers it to be

    a cognate of the most widespread words used for 2 (discussed below): “Il peut arriver que

    l’élément généralement réalisé après le verbe prenne la place de ur . C’est le cas pour le parler

    de Ghadames” (“it may happen that the element usually uttered aer the verb replaces ur . is

    is the case with the variety of Ghadames”). A misinterpretation of this statement leads Lucas

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    4/17

      Vermondo Brugnatelli

    e most widespread form is wәr /ur , which appears to be similar to another

    particle, war-, used before nouns as a prefix denoting the lack of something, for

    example:

      (1) Jerba (Tunisia)

      war abekkad   su

      lacking sin

      ‘sinless’, i.e. ‘young child’.

    is similarity has led some scholars to posit a link between the two particles

    (Loubignac, 1924: 177; Basset 1940: 221; Chaker 1996: 12), but two facts challenge

    this view: firstly, the privative particle has a feminine form tar- implying that theinitial wa- sounds should rather be regarded as the reflex of a masculine pronomi-

    nal marker (wa, which has a feminine counterpart ta); secondly, the presence of

    the prefix war- in languages that use wel  and not wer  as a negator, points to an

    independent origin. See, for instance, an example from an unpublished old Ibād  site

    text9 containing both particles:

      (2)  d iwalen  g g  wel yelli war elh  seqq

      word= in 3=be= lacking truth

      ‘they are utterances in which there is nothing untruthful’ (f. 82b, l. 8)

    In any case, Lionel Galand has repeatedly stressed the fact that, given the multiple

    forms of the negator, it seems preferable to consider only the beginning soundwe-/u- as  the “basis of the negative particle”, because it is the common element

    shared by all the languages (Galand 1994: 171; 2010: 279–280). is implies that -r ,

    -l  and -d  were added during a later phase, and this could account for the different

    forms attested.

    to erroneously list the language of Ghadames among those that only possess a post-verbal

    negator (2013: 412). More convincingly, Galand (1994: 172) points to another pan-Berber

    particle, akw/akkw  ‘all’, which is oen used to strengthen other particles. In the Moroccandialect of the Ayt Youssi of Enjil it precedes the preverbal negator ur  when the meaning is ‘not

    at all’. More details in Galand (2011: 2nd part 66). An example:

      ccifur   s-addx akkw  ur issin zznaqi  driver-this at all 1  knew=  streets  ‘this driver does not even know the streets’ (Galand 2011: 1st part 26).

    . e old Berber text referred to in this article is the manuscript Ms.Or. 2550 of the Na-tional Library of Tunis. It contains a medieval commentary to Abū Ġānim’s Mudawwanah, animportant legal work of the Ibād  site branch of Islam.

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    5/17

      Berber negation in diachrony

    Galand’s hypothesis was recently corroborated by a discovery (Brugnatelli

    2011) regarding a little-studied dialect (Chenoua region, Algeria). In this Berber

     variety, a negative construction meaning ‘not yet’ is formed using a kind of auxil-

    iary verb; this is also the case in many other Berber languages,10 but here the para-

    digm also demands a form of conjugation of the negator ur . More precisely, the

    conjugated element is the second part, r , while the first part, u, remains invariable:

      (3) 1.s. u r t   uci   ad     2.s. u her t   ucid     ad     3.m.s.  u yer t   uci ad   

      3.f.s. u her t   uci ad   

      1.p.  u ner t   

    uci ad     2.p. u her t   ucim ad   

      3.m.p. u r t   ucin ad   

      3.f.p. u r t   ucinţ ad   

    Such a state of affairs may be viewed as the incipient grammaticalisation of a pair

    of verbs that have lost the personal indices occurring between them: according to

    this scheme, the second verb (uci, here with the imperfective stem t   uci) displays

    only the suffixed, and the first (*r ) only the prefixed, indices.11 e previous phase

    was probably as follows:

      (4) 1.s.  *u r (e )  t   uci   ad     2.s. *u her (d   )  (h)t   ucid     ad   

      3.m.s. *u yer (i)t   

    uci ad   

      etc.

    is discovery, first presented and more thoroughly explained and discussed in

    Brugnatelli (2011), reopens, on a new basis, a long-debated issue in Berber his-

    torical studies, namely the proposed verbal origin of the negative particle. is

    hypothesis was put forward by Loubignac in the work cited earlier (1924: 177), in

    which he suggested bringing together under one and the same root R the negative

    particle wer , the privative prefix war  and the verb ar  ‘be void, desert’ of Central

    Morocco; later on, A. Basset took up the debate in a detailed study concluding that:

    . For instance, in Nafūsī wel yuc we dd-yusu-c ‘he has not yet come’, lit. 1 3= 

    1 -3=come=-2. Other verbs are used in other regions. Elsewhere, for instance

    in Jerba, this verb is lexicalised and has become an invariable particle wecci. More details in

    Brugnatelli (2011: 521–524).

    . e Berber personal indices may be: prefixes (3  y-, 3  t-, 1  n-), suffixes (1  - ,

    3-n, 3 -nt ) or circumfixes (2 t-…-d , 2 t-…-m, 2 t-…-mt ). In Chenoua dialect,

    morphological t  usually becomes h.

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    6/17

      Vermondo Brugnatelli

    “We would be strongly tempted to recognize in war  an old 3rd pers. masc. sg.

    preterite of a quality verb whose characteristics are already fully determinable. In

    this case, the root would be biliteral, WR; tar  would be due to reshaping (…) and

    wər , whose vocalism remains obscure, would probably be a frozen masc. sg.” 12

    What he considered important was the fact that most nouns following the priva-

    tive war  are in the free state instead of the annexed state required by most prepo-

    sitions.13 e free state is typical of the direct object and could be easily explained

    if the original construction was a verb phrase and not a prepositional phrase (see

    also Prasse 1972: 244).

    On the other hand, the argument put forward by Prasse (1972: 244; 1973: 12) –

    quite similar to the point of view of Marcy (1936, 1940/41) – took into consider-ation the behaviour of negation in the relative and wh-interrogative clauses, which

    require a special form of the verb, the so-called participle. In these constructions,

    the morphemes of the participle are not suffixed as in positive sentences, but

    placed before the verb and aer the negator. As a consequence of this shi, these

    morphemes may be seen as the participial mark placed aer the negator, and the

    method adopted to transcribe a negative expression in relative or interrogative

    clauses depends on whether this theory is accepted or rejected:

      (5) a. uren  yufil 

      = 3=be tanned=

      b. our en ioufil 

      3=be tanned=

      ‘(which is) not tanned’ (newer and older spelling of the same phrase in

    Drouin 1996: 244)

    e case of Chenoua negation examined above suggests an earlier scenario some-

    where in between the reconstructions of those positing an ancient verb meaning

    ‘not being’ and those who have not accepted this theory, considering the negator

    to have been an invariable particle from the earliest stages of Berber. Actually, it

    seems that only the last part of the negator may be traced back to a verb, while

    . “Nous serions fort tentés de retrouver en war  une ancienne 3e pers. masc. sg. de prétéritd’un verbe de qualité aux caractéristiques dès maintenant parfaitement déterminables. En ce

    cas, la racine serait bilitère, WR ; tar  serait dû à une réfection (…) et wər , dont le vocalisme

    reste obscur, serait sans doute un masc. sg. figé” (1940: 222). Chaker (1996: 12–14) accepts this

    theory and assumes that the different vocalism of wər and war  points to a contrast between

    aorist and “preterite” (i.e. perfective).

    . e opposition of two “states” in nouns (“free” vs. “annexed”) is a well-known feature of

    Berber grammar and does not require in-depth presentation here. e best updated descrip-

    tion of the phenomenon, with a critical bibliography on the topic, is offered by Galand (2010),

    Chapters 4.1–4.3 and 5.3.

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    7/17

      Berber negation in diachrony

    the initial part u/we was a basic uninflected negator, as previously suggested by

    Galand.

    According to Galand (2010: 284), the conjugated element of the final part of

    the negator derives from a monoconsonantal verb meaning ‘will, want’. is verb

    still exists as iri in Moroccan Berber (Tashelhit in the South as well as dialects of

    central Morocco), and as әr  or äru in Tuareg. is is fully consistent with the form

    of the negative particle in all those dialects where it appears as ur /wer . However,

    some dialects display a lateral [l ] in the negator, which appears as ul /wel  (Ouargla,

    Ghadamès, Jerba (partially), Mzab, Nefousi, Central Morocco: Zemmour, Iziyan).14 

    In this respect, it is worth noting that the verb itself may have a variant with [l ]. As a

    matter of fact, in a Medieval Berber text in which the preverbal particle is wel , I wasable to single out a verb il  meaning ‘will, want’: in some instances, the word yil  is

    translated by an Arabic gloss yurīdu ‘ he wants’. For instance:

      (6) Arab text:  yurīdu iqtit   sā a māli-ka

      Berber translation:  yil [Gloss: yurīdu] s we res n wed   rim-ik

      ‘he wants to take possession of your money’ (f. 286b, l. 2)15

    In any case, this phonetic issue requires further investigation, given that the Berber

     varieties in which the preverbal negator is ul /wel  usually also display l- variants

    of other morphological elements, such as the preposition  el /al  ‘towards’ (vs. the

    more frequent  er /ar ) or the particle ala (vs. ara) used before relatives.16

    . e post-verbal particle

    As has already been pointed out, in most Berber languages the verbal negation con-

    sists of two elements, one placed before the verb ( 1) and one aer it (2).

    e second element is not the same in all varieties, and the differences among the

    particles attested in different regions appear to be greater than those existing in the

    prefixed negator. e situation is complicated by interference, given that most of the

    Arab Maghribian dialects also display a suffixed negative particle that is sometimes

    similar to the Berber equivalent. When neighbouring Berber and Arabic dialects

    possess particles that resemble each other, the oldest descriptions of Berber tended

    . Concerning ud , at the beginning the consonant d   might have been an element added

    in order to avoid a hiatus. For instance, in Jerba 1 is usually u/we (seldom wel ), but a free

    choice exists between w-  and wed   (d   )- before a vowel: w-ud   ife -c  / wed   d   -ud   ife -c ‘I did not

    come in’.

    . e spelling is sure, since the word is fully vocalised: the consonant y  bears the sign of the

    i vowel, and the final l  has a sukūn, which means lack of a following vowel.

    . For a more detailed discussion of this issue concerning the preposition, see Brugnatelli

    (2006: 69). As for the other particles, see below (§ 1.2.2.).

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    8/17

      Vermondo Brugnatelli

    to consider 2 a borrowing from Arabic, but this is not always the case (Brug-

    natelli 1987a). In reality, the most widely used particles for 2  may be traced

    back to one and the same lexical item, “the proto-Berber forms *k y ăra ~ (h)ăra(t )

    ‘thing’ ” (Kossmann 2013: 332). e original forms have undergone divergent devel-

    opments in different areas, and 2 may appear as: kra/k  ra/cra, ara, ka/k  a/ca or k/ 

    k   /c.17 us, the palatalisation of velars occurring in certain Berber varieties (above

    all in the so-called Zanata dialects), has led to a form of 2 comparable to the

    corresponding Arabic morphemes (šay , ši or š  from the classic word šay   ‘thing’).18

    .. “Jespersen cycle” 

    e frequent usage of circumfixes in Berber negation suggests that its historicaldevelopment may be described in terms of the so-called “Jespersen cycle”, a pro-

    cess through which many world languages renew their negative structures by add-

    ing redundant elements that over time become part of the standard negation.

    is process is customarily represented as comprising three main stages:

    Stage I = only preverbal negator (1 …);

    Stage II = twofold negator, preverbal and post-verbal (1 … 2);

    Stage III = only post-verbal negator (… 2)

    As pointed out in the most recent studies, this tripartite classification is somewhat

    sketchy and would benefit from the inclusion of subtypes based on the degree of

    coexistence in a given language of the structures of each of the three main stages(van der Auwera 2009, 2010: 75–85).

    e similarity between the negative structures of North-African modern

    Arabic dialects and those of Berber has prompted scholars to explore the ques-

    tion of the origin of discontinuous morphemes within the framework of linguistic

    interference:

    “e fact that those varieties of Arabic and Berber which have reached stage II or

    III of JC are spoken in largely the same geographical area raises the question of

    whether the stage II construction was spread from one language to the other via

    contact, and, if so, which was the source and which the target language as far as

    this structure is concerned.” (Lucas 2007: 401)

    . Other particles used for 2 will not be examined here. Kossmann (2013: 332) suggests

    that ani, used in Eastern Kabylia (Algeria), is derived from ani ‘where’. Laioui (2013a and

    2013b) examines the Riffian 2 bu, which has also spread to the neighbouring Arabic dia-

    lects. Another particle attested in Eastern Kabylia is ula (Rabhi 1992 and 1996).

    . Kossmann (2013: 333) set up a table summarizing the forms of 2  in the principal

    Berber varieties and regrouping them into three columns according to what he considers

    surely Berber, surely Arabic and ambiguous.

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    9/17

      Berber negation in diachrony

    It is difficult to identify a clear-cut solution, given the lack of material from the

    earliest stages of spoken Arabic and Old Berber. For this reason, Lucas’ suggestion

    that, in Berber, Stage II “developed under the influence of Arabic” (2013: 402) is

    not conclusive (see also Laioui 2013a for a critical discussion of Lucas’ hypoth-

    esis). e main reason put forward concerns the areal distribution, “consistent

    with a gradual spread westwards and southwards of the cycle in the local con-

    tact varieties of Arabic” (Lucas 2013: 413). But the areal data may also be read in

    the opposite way: in terms of the loss of a redundant feature in peripheral areas.

    For instance, both the easternmost language, Siwa, and the westernmost, Zenaga,

    no longer possess state opposition in nouns, but this alone does not justify the

    straightforward assumption that this is an innovation they never shared with theother Berber languages (Brugnatelli 1987b). Furthermore, the sporadic presence

    of 2 in Ancient Ibād  sī Berber, in both a more archaic form (-cra) and a phoneti-

    cally reduced one (-c) seems consistent with viewing it as an ancient construction

    that is losing ground, rather than as a newly acquired innovation. e influence

    of Arabic might be seen, rather, as a stimulus to preserve 2  in the Berber

     varieties in which it had become similar to the Arabic equivalent, while most of

    the so-called kem dialects, where 2 did not undergo a palatalisation, have lost

    it (Brugnatelli 1987a: 58).

    A decisive argument in favour of a very early stage characterised by a twofold

    negator across the whole Berber area derives, in my opinion, from the wide diffu-

    sion of negative stems in all the verbal systems. For a more in-depth look at thissubject, see § 2 below.

    .. Kabyle 2 ara

    e Kabyle particle ara  has attracted the attention of researchers because it is

    homophonous with another particle ara, which is the aorist particle used in rela-

    tives and wh-interrogatives. Mettouchi (2001) strongly asserted that both derive

    from one and the same lexical unit, which in the course of time came to take

    on such differing functions via processes of reanalysis (whereby 2 in negative

    sentences preceding a relative would have been reinterpreted as the antecedent

    of the relative itself and subsequently as an aorist particle).19 However, the series

    of steps implied by this theory does not match any known process in other lan-guages; furthermore, the paper does not examine comparative data from other

    Berber varieties.

    . A different opinion on the origin of this particle is expressed by Chaker (1983: 121): “asa matter of fact, ara is the amalgam of ay  ‘this’ + ad  ‘non-real’ ” (“ara est en fait l’amalgame deay  ‘ce’ + ad  ‘non-réel’ ”) (see also ibid. 159).

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    10/17

      Vermondo Brugnatelli

    As a matter of fact, a specific aorist particle that is used in relatives and wh-

    interrogatives is also found outside of Kabylia. In the language of Figuig this

    particle is ala (7). It has also been documented elsewhere, for instance in the

    old language of Jerba (in a religious text composed at the beginning of the 19th

    century) (8) and even in Medieval Berber from the Ibād  site regions (Tunisia/

    Libya) (9).

      (7) Figuig: wi stt ala nawey? 

      who her take==

      ‘who will marry her?’ (Kossmann 1997: 268)

      (8) Jerba w’ ala s nuc  who to him give==

      ‘who will give him’ (Brugnatelli 2011: 529)

      (9) Ibād  site w’ ala nz   sem lxemr 

      who press== wine

      ‘whoever will produce wine’ (f. 398a, l. 1–2)

    e similarity of forms and functions makes it probable that ala and Kabyle ara 

    are cognate. e only slight phonetic difference lies in the quality of the liquid,

    much in the same way as the negative particles ur  and ul  or the prepositions ar  

    and al  mentioned above.20 e Medieval Berber evidence is important not only

    because it shows that this particle is old (while Mettouchi’s theory frames it as a

    recent development), but also because it makes clear that it has nothing to do with

    2, attested as c or cra.21

    Most probably, another cognate is the particle la of Zuara (Libya), which is a

    connective (“relateur”: Galand 2005: 192–193) introducing a relative that may be

    either verbal or verbless:

      (10) wuh la yemmut 

      that one 3=die=

      ‘the one who is dead’

      knečč la d_tamet   st   sut 

      me _woman

      ‘I, who am a woman’

    . Kabyle ara has several phonetic variants: a a,  a, aa, aa, ara. Other possible cognates

    (such as Gourara  a or Mzab a a) are discussed in Brugnatelli (2011: 529–530).

    . In the Medieval text, cra is also used as an indefinite pronoun in non-negative expres-

    sions, such as cra n ‘some of’, ‘a little of’, in the same way as the Kabyle kra.

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    11/17

      Berber negation in diachrony

    In Zuara,  la may be used as a subordinating connective in other cases too (11),

    similarly to ala in Figuig (12) and Old Ibād  sī Berber (13):

      (11) bed la kemməlnet ccáhi

      aer 3=finish= tea

      ‘aer they had finished tea’ (Mitchell 2009: 274)

      (12) qbel ala idjiwen

      before 3=be sated=

      ‘before he is sated’ (Kossmann 1997: 330)22

      (13) qabbel ala yeg     ad   in

      before 3=do= this  ‘before he does this’ (f. 310a, l.11)23

    At this early stage of the language, the particle was also used in cle sentences, as

    in the following example:

      (14) d elqimet     en yelemlemen g     essuq ala yuc

      price of  y.  in market 3=give=

      ‘it is the market price of y.-s [unknown word] that he will give’ (f. 403b, l. 8)

    e fact that all these particles are strongly connected with relatives raises the

    question of a possible link with the Tuareg particle éré , which is not a connective

    but a pronoun – ‘anyone who’ – that is solely used as an antecedent of a relative

    clause. Prasse (1972: 189) proposed the following etymology of the last-mentionedterm: “i-irâ, i-irê ‘un qu’il (on) veut’ ”, i.e. a pronoun i (“pronom d’appui”) followed

    by the verb ăr  ‘will, want’. Interestingly, this same construction is widely accepted

    as the source of a “future” particle in Tashelhit (Southern Marocco), namely ra, rad  

    /ara, arad .24 An in-depth analysis of the usages and functions of all these gram-

    matical devices falls outside the scope of this paper, but this brief comparative

    outline is sufficient to support the claim that the Kabyle ara used as 2 should

    be regarded as independent of the aorist particle ara.

    . Such cases of use with the perfective prevent us from considering ala a simple “variant

    of the prospective preverb under ‘attraction’ [i.e. in relatives and wh-interrogatives: V.B.]”

    (“variante du préverbe prospectif en attraction”, Kossmann 1997: 442).

    . Interestingly, the same construction bə d la ‘après que’ “aer (that)” appears in Takrouna

    Arabic (Tunisia): Galand (2005: 193). Maybe in this case a borrowing from Berber is more

    probable than the other way round. See bə d la with the same meaning.

    . A possible link between the Kabyle ara and the Tashelhit rad  has already been proposed

    by Taine-Cheikh (2009: 98).

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    12/17

      Vermondo Brugnatelli

    . e negative verbal forms

    Another distinctive feature of the Berber languages is the use of special forms of

    the perfective and imperfective stems in negative constructions. Negative stems

    are seldom used without the presence of other negators, yet their negative value is

    indisputable and in some – admittedly few – cases they may be the only means of

    expressing negation. For instance, in Kabyle (Dallet 1982: 530):

      (15) a. mazal yet   st   ses

      still 3=sleep=

      “he is still sleeping”

      b. mazal yet   st   sis  still 3=sleep=

      “he is not yet sleeping”

    e most widespread form is the negative perfect, which appears in nearly all

    the varieties, while the negative imperfective is less generalized but nonetheless

    scattered across the whole area and should probably also be considered a com-

    mon form. is is further confirmed by the fact that the negative imperfective is

    attested in ancient texts even in areas in which it is not currently in use.

    Both perfective and imperfective stems undergo similar modifications in the

    negative form. A survey reported in Brugnatelli (2002: 168) indicated that these

    changes may be summarised as follows:

    1. vowel fronting (a > é /i and ä > e/é )

    2. shortening of the first vowel

    3. lengthening of the last vowel

    In general, shortening and lengthening of the vowels may be detected in Tuareg

    only, given that the other Berber languages do not usually distinguish between

    short and long vowels.25 On the contrary, vowel fronting is a general rule, affecting

    the negative stems in all varieties.

    Some examples from different areas:

      (16) a. – Tuareg

      perfect neg. perfect imperfect neg. imperfect

      ikräs ikrés ikârräs ikerres ‘knot’

      ilsa ilsé ilâss iless  ‘wear’

      ibberäġ = itâbärâġ iteberiġ ‘show off’

    . For the changes in vocalism of the perfective in the negative form as well as in the so-

    called resultative, see Brugnatelli (2005: 376–378).

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    13/17

      Berber negation in diachrony

      b. – Jerba

      yez   swa   yez   swi iz   sugga yez   suggi ‘go down’

       yebbes yebbis yetbessa yetbessi ‘be switched off’

       yewet     = yeččat     yeččit     ‘strike’

    From a diachronic point of view, these forms are easily explained as the result of

    phonotactic changes involving the final part of the stem, possibly under the influ-

    ence of a suffixed negative particle. is would account for the shortening of the

    beginning vowels and the lengthening of the final ones (via a le-to-right stress

    shi) as well as for the fronting of (final) vowels (as a consequence of Umlaut,

    assuming that the original particle contained front vowels).

    An interesting parallel comes from the Arabic dialects of Egypt (DakhlaOasis), in which negative verbal forms have arisen from positive ones, displaying

    a vocalic difference most likely provoked by “consonant clustering and heavy syl-

    lable formation”, due to the affixation of 2 (Woidich 1995–97: 14–15):

      (17) Western dialects: i  gom > ma-ti  gam-š  ‘do not speak Cairene!’

      Central dialects: si an > ma-si in-š /ma-si en- š «he did not ask».

    Likewise, in a 19th century religious poem from Jerba I detected an opposition

    between forms such as  er-s (‘by-him’ = ‘he has’) and we  r-is-c (‘1 by-him-

    2’ = ‘he has not’), with the development of a full vowel i under the influence of

    an affix added at the end of the complex.

    In spite of the strong evidence in favour of a purely phonetic origin of the phe-nomenon, some scholars still share the opinion put forward by Picard (1957) that

    the negative perfective represents a sort of “intensive” form of perfective (“prétérit

    intensif”). Chaker (1996: 18) affirmed that it was “a former intensive form which

    must have been used in environments strongly characterised by modality: negative

    statements (prohibition), wishes, unreal hypotheses, etc.”26 As already pointed out

    by Brugnatelli (2002: 171), the negative perfect is absent when modality is most

    heavily involved, such as in wishes (optative) and oaths, in which Berber uses,

    respectively, a wer/wel  + aorist and ma (or equivalents) + positive perfect, without

    2. As for the counterfactual conditional – the only other instance apart from

    negative sentences in which the negative perfect may occur – it should be noted

    that this implies a negation, and that some connectives introducing it are formedby an amalgam containing the negative particle ur/wer  (e.g. the Tashelhit mur , and

    possibly also the Kabyle lemmer ).

    . “Une ancienne forme intensive qui devait être employe dans des environnements à

    forte modalisation: noncs ngatifs (interdiction), de souhait, d’hypothse irrelle, etc.”

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    14/17

      Vermondo Brugnatelli

    e existence of negative verb stems in all Berber languages could be viewed

    in itself as a pan-Berber strategy of double-marking the negation (Schmitt-Brandt

    1979: 235; Brugnatelli 1987a: 59; Laioui 2013a: 87). e fact that such forms

    probably derived from elements placed towards the right end of the verbal com-

    plex strengthens the hypothesis that Berber achieved Stage II of the Jespersen cycle

    in very ancient times, earlier than any contact with Arabic. e actual shape of the

    added element is difficult to reconstruct. It is even possible that the attested forms

    of 2 are innovated forms which replaced or were added to earlier morphemes,

    given that the most important phonetic change is palatalisation, which in principle

    entails the presence of a front vowel. We nd similar phenomena in many other

    world languages, such as the Old Irish genitive maicc (from macc ‘son’), in whicha nal -i, preserved in Ogam maqqi, had completely disappeared leaving only a

    phonetic vestige in the palatalisation of the nal consonant, or the well-known

    phenomenon of Umlaut in German, in which nal vowels undergo fronting under

    the influence of i-endings that have now disappeared.

    . Conclusion

    e diachronic evolution of Berber negation is associated with the so called

    Jespersen Cycle in a number of ways and may contribute to a deeper understanding

    of the linguistic phenomena determining the cycle’s multiple outcomes.In particular, a rst key observation is that Berber, with its widespread use of

    two concatenative negators (1, 2) combined with a third, non-concatenative

    one (namely the use of a negative apophonic stem), can be considered one of the few

    languages to possess a “triple negation” (Laioui 2013a: 87), a feature rst pointed

    out and studied in Lewo, a Malayo-Polynesian language spoken in Vanuatu, and

    later also in some Brabantian and Italian dialects (van der Auwera et al. 2013). As a

    consequence, in relation to the origin of the negative stems, the phonetic modica-

    tions triggered by a post-verbal negator should be viewed as a sixth source of new

    (non axal) negators, in addition to those already known, which are:

    1. a word expressing minimal value: French pas ‘(not even a) step’;

    2. a negative word: English not , which originally meant ‘nothing’;3. an emphatic element: such as French du tout  or English at all ;

    4. a particle of negative answer: Brazilian Portuguese naõ;

    5. repetition of the rst negator: Brabantian nie.

    Across the vast area in which Berber is spoken, all possible stages of the “cycle”

    may be found (1 V; 1 V 2; V 2), although the relative chronology of

    the changes has not yet been rmly established.

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    15/17

      Berber negation in diachrony

    Further in-depth comparative study of the different developments taking

    place in different regions will contribute to a more advanced understanding of the

    multiple factors involved in the evolution of negation.

    References

    Basset, André. 1940. Quatre études de linguistique berbère: 1. gar ‘mauvais’; 2. Touareg әrk‘mauvais’; 3. war; 4. bla en berbère. Journal Asiatique 232: 161–291.

    Boumalk, Abdallah. 1996. La négation en berbère marocain. In Chaker & Caubet: 35–48.

    Brugnatelli, Vermondo. 1987a. La negazione discontinua in berbero e in arabo-magrebino. In Atti della 4. Giornata di Studi Camito-semitici e Indeuropei, Bergamo 28 November 1985,Giuliano Bernini & Vermondo Brugnatelli (eds), 53–62. Milano: Unicopli.

    Brugnatelli, Vermondo. 1987b. Deux notes sur l’état d’annexion en berbère. In Proceed-ings of the Fourth International Hamito-Semitic Congress  [Current Issues in Linguisticeory 44], Hermann Jungraithmayr & Walter W. Mueller (eds), 349–359. Amsterdam:John Benjamins.

    Brugnatelli, Vermondo. 1993. Quelques particularités des pronoms en berbère du Nord. In  Àla  croisée des études libyco-berbères. Mélanges offerts à Paulette Galand-Pernet et LionelGaland , Jeanine Drouin & Arlette Roth (eds), 229–245. Paris: Geuthner.

    Brugnatelli, Vermondo. 2002. Les thèmes verbaux négatifs du berbère: Quelques réflexions. In Articles de linguistique berbère. Mémorial Werner Vycichl , K. Naït-Zerrad (ed.), 165–180.Paris: L’Harmattan.

    Brugnatelli, Vermondo. 2005. Voyelles et accents dans l’histoire du berbère. In Proceedings ofthe 10th  Meeting of Hamito-Semitic ( Afroasiatic) Linguistics, Florence, 18–20 April 2001.[Quaderni di Semitistica 25], Pelio Fronzaroli & Paolo Marrassini (eds), 371–380. Firenze:Dip. di Linguistica-Università.

    Brugnatelli, Vermondo. 2006. La négation berbère dans le contexte chamito-sémitique. In Leslangues chamito-sémitiques (afro-asiatiques), Vol. 2 [Faits de Langue 27], Atoine Lonnet &Amina Mettouchi (eds), 65–72. Paris: Ophrys.

    Brugnatelli, Vermondo. 2010. Problème de la négation en berbère: À propos de l’origine d’ulac,ula, ula d. In Camsemud 2007. Proceedings of the 13th Italian Meeting of Afroasiatic Linguis-tics held in Udine, May 21st–24th, 2007 , Fredrick Mario Fales & Giulia Francesca Grassi(eds), 401–405. Padova: S.A.R.G.O.N.

    Brugnatelli, Vermondo. 2011. Négations, participes et figement en berbère: Nouvelles hypothèses.In Parcours berbères. Mélanges offerts à Paulette Galand-Pernet et Lionel Galand pour leur

    90e anniversaire, Amina Mettouchi (ed.), 521‒532. Cologne: Rüdiger Köppe.Brugnatelli, Vermondo. 2014. Typology of eastern medieval Berber. STUF – Language Typology

    and Universals 67(1): 127–142.Chaker, Salem. 1983. Un parler berbère d’Algérie (Kabylie). Syntaxe. Aix-en-Provence: Publica-

    tions Université de Provence.Chaker, Salem. 1996. Quelques remarques préliminaires sur la négation en berbère. In Chaker &

    Caubet (eds), 9–22.Chaker, Salem & Caubet, Dominique (eds). 1996. La négation en berbère et en arabe maghrébin.

    Paris: L’Harmattan.

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    16/17

      Vermondo Brugnatelli

    Dallet, Jean-Marie. 1982. Dictionnaire kabyle-français. Parler des At Mangellat. Paris: SELAF.

    Drouin, Jeannine. 1996. Les formes participiales en berbère. Essai de dialectologie comparée.Littérature Orale Arabo-Berbère 24: 233–260.

    Galand, Lionel. 1994. La négation en berbère. Matériaux Arabes et Sudarabiques 6: 169‒81.Galand, Lionel. 2005. Quelques traits du parler berbère de Zouara (Libye). Studi Maġrebini 3:

    187–195.Galand, Lionel. 2010. Regards sur le berbère. Milano: Centro Studi Camito-Semitici.Galand, Lionel. 2011. Deux mille phrases dans un parler berbère du Maroc. Application et évalu-

    ation de la méthode d’enquête linguistique d’Henri Frei. Rabat: Institut Royal de la CultureAmazighe.

    Kahlouche, Rabah. 2000. Le présentatif négatif ulac ‘il n’y a pas’, est-il de souche berbère ouun emprunt à l’arabe? In Études berbères et chamito-sémitiques. Mélanges offerts à Karl-G.

    Prasse, Salem Chaker & Andrzej Zaborski (eds), 233–236. Louvain: Peeters.Kossmann, Maarten. 1989. L’inaccompli négatif en berbère. Études et Documents Berbères  6:

    19–29.Kossmann, Maarten. 1997. Grammaire du parler berbère de Figuig ( Maroc oriental ). Louvain:

    Peeters.Kossmann, Maarten. 2013. e Arabic Influence on Northern Berber . Leiden-Boston: Brill. DOI:

    10.1163/9789004253094Laioui, Mena. 1996. La négation en tarifit . In Chaker & Caubet, 49–77.Laioui, Mena. 2007. Atlas linguistique des variétés berbères du Rif . Köln: Rüdiger Köppe.Laioui, Mena. 2011. Études de la variation et de la structuration linguistiques et sociolinguis-

    tiques en berbère. Köln: Rüdiger Köppe.Laioui, Mena. 2013a. Reinventing Negation Patterns in Moroccan Arabic. In  African Arabic:

     Approaches to Dialectology   [Trends in Linguistics. Studies and Monographs 25], Mena

    Laioui (ed.), 51–94. Berlin: De Gruyter Mouton.Laioui, Mena. 2013b. Negation, grammaticalization and language change in North Africa:

    e case of the negator NEG___bu. In Tilelli. Scritti in onore di Vermondo Brugnatelli,Giorgio Francesco Arcodia, Federica Da Milano, Gabriele Iannàccaro & Paolo Zublena(eds), 113‒130. Cesena: Caissa.

    Loubignac, Victorien. 1924. Étude sur le dialecte berbère des Zaïan et Aït Sgougou. Paris: Leroux.Lucas, Christopher. 2007. Jespersen’s cycle in Arabic and Berber. Transactions of the Philological

    Society  105(3): 398–431. DOI: 10.1111/j.1467-968X.2007.00189.xLucas, Christopher. 2013. Negation in the history of Arabic and Afro-Asiatic. In e History

    of Negation in the Languages of Europe and the Mediterranean, Vol. I: Case Studies, DavidWillis, Christopher Lucas & Anne Breitbarth, 399–452. Oxford: OUP.

    Marcy, Georges. 1936. Note sur le pronom relatif-sujet et le pseudo-participe dans les parlersberbères. Bulletin de la Société de Linguistique de Paris 37: 45–57.

    Marcy, Georges. 1940–41. Observations sur le rélatif futur en touareg Ahaggar. Bulletin de laSociété de Linguistique de Paris 41: 129‒133.

    Mettouchi, Amina. 1996. La négation dans les langues du Maghreb: Synthèse. In Chaker &Caubet (eds), 177–195.

    Mettouchi, Amina. 2001. La grammaticalisation de ara en kabyle, négation et subordination rela-tive. In Travaux du CerLiCO 14, Gilles Col & Daniel Roulland (eds), 215–235. Rennes: P.U.

    Mettouchi, Amina. 2009. e system of negation in Berber. InPatterns of Negation in West AfricanLanguages  [Typological Studies in Language 87], Norbert Cyffer, Erwin Ebermann &Georg Ziegelmeyer (eds), 287–306. Amsterdam: John Benjamins.

  • 8/17/2019 Berber Negation in Diachrony Brugnatelli

    17/17

      Berber negation in diachrony

    Mitchell, Terence F. 2009. Zuaran Berber (Libya). Grammar and Texts. Köln: Rüdiger Köppe.Picard, André. 1957. Du prétérit intensif en berbère. In  M émorial André Basset (1895–1956 ),

    107–120. Paris: Adrien Maisonneuve.Prasse, Karl-G. 1972.  Manuel de grammaire touarègue (tăhăggart ) , I–III: Phonétique Écriture-

    Pronom. Copenhagen: Éditions de l’Université.Prasse, Karl-G. 1973. Manuel de grammaire touarègue (tăhăggart), VI–VII: Verbe. Copenhagen:

    Éditions de l’Université.Provotelle, Paul. 1911. Étude sur la tamazir’t, ou zénatia de Qalaât es-Sened . Paris: Ernest Leroux. van Putten, Marijn. 2013. A Grammar of Awjila Berber (Libya). Based on Umberto Paradisi’s

    Material. Ph.D. dissertation, Leiden.Rabhi, Allaoua. 1992. Les particules de négation dans la Kabylie de l’Est. Études  et Documents

    Berbères 9: 139–146.

    Rabhi, Allaoua. 1996. De la négation en berbère: les données algériennes. In Chaker & Caubet(eds), 23–48.

    Schmitt-Brandt, Robert. 1979. Berberische Adstrateinflüsse im maghrebinischen Arabisch.Folia Linguistica 13: 229–235. DOI: 10.1515/flin.1979.13.3-4.229

    Taine-Cheikh, Catherine. 2009. Les morphèmes de futur en arabe et en berbère. Réflexions pourune typologie. Faits de Langues 33: 91–102.

    Taine-Cheikh, Catherine. 2011. L ’énoncé négatif en berbère zénaga. In Parcours berbères. Mélanges offerts à Paulette Galand-Pernet et Lionel Galand pour leur 90e anniversaire,Amina Mettouchi (ed.), 533–551. Köln: Rüdiger Köppe.

     van der Auwera, Johan. 2009. e Jespersen cycles. In Cyclical Change  [Linguistik Aktuell/Linguistics Today 146], Elly van Gelderen (ed.), 35–71. Amsterdam: John Benjamins.

     van der Auwera, Johan. 2010. On the diachrony of negation. In e expression of Negation,Laurence R. Horn (ed.), 73–109. Berlin: De Gruyter Mouton.

     van der Auwera, Johan, Vossen, Frens & Devos, Maud. 2013. Le cycle de Jespersen à trois ouquatre négations. In La linguistique de la contradiction, Jacques François, Pierre Larrivée,Dominique Legallois & Franck Neveu (ed.), 19–30. Bern: Peter Lang.

    Woidich, Manfred. 1995–97. Negation in the Egyptian Arabic dialect of the Dakhla-Oasis: Acase of rule morphologization. Mediterranean Language Review 9: 13–28.