NACAL 33, March 18th-20th 2005

The Left Margin of the Word in Berber Derivational Morphology Mohamed Lahrouchi CNRS-University Paris 8 [email protected]

In Berber varieties such as Tashlhiyt, Kabyle and Tamashek, the causative morpheme /s-/ and the reciprocal morpheme /m-/ appear either simple or geminated. Some examples from Tashlhiyt1 are given below in [1]: [1] a) Aorist causative mun « to be picked up » smun « to pick up » b´rb´r « to boil » sb´rb´r « to make boiled » knu « to lean » ss´knu « to make someone fall » xd´m « to work » ss´xd´m « to make someone work» b)

Aorist ħsad S´nn´q zri rwi

Aims -

« to be jealous » « to strangle » « to pass by » « to mix »

reciprocal mħisid mSinniq mm´zri mm´rwi

« to be jealous of each other » « to strangle each other » « to meet each other » « to mingle, mix with each other»

/s-/ and /m-/ quantity variation follows from the use of an empty site at the beginning of the word, in the sense of Lowenstamm (1999). The empty site is filled by means of two strategies: spreading and movement. Inchoative verb geminate their initial consonant by spreading into the initial site. Sibilant assimilation (/s/ becomes [S]) and [labial] dissimilation (/m/ becomes [n]) take place inside the domain of the verb.

I. Word-initial clusters Languages vary as to the kind of clusters they tolerate word-initially. Two types of languages. Two types of languages: i. Type I languages such as English and French which exclusively tolerate clusters of increasing sonority (#true, #blue). ii. Type II languages such as Moroccan Arabic and Berber where initial clusters are made sonority free, tolerating clusters of both increasing and decreasing sonority (glˆˆ÷ “remove” vs. lga “find”, d88rˆb “heat” vs. rd8a8 “accept” Moroccan Arabic; kru “rent” vs. rku “be dirty”, gnu “sew” vs. ngi “flow” Tashlhiyt). [2] Implicational Universal “If a language tolerates word-initial clusters of decreasing sonority, it tolerates word-initial clusters of increasing sonority, but not vice-versa.” 1

The use of epenthetic schwa in Tashlhiyt transcriptions is necessary though its phonological status is controversial; See Dell & Elmedlaoui (85, 88), Boukous (87), Coleman (96, 01), Puech & Louali (99), Louali & Puech (00) for more details.

1

NACAL 33, March 18th-20th 2005

[3] Word-initial Site hypothesis (Lowenstamm 1999) Every word of a major category has an empty initial CV site that replaces the traditional SPE word boundary symbol #: [4]

a. #kit#

b.

CV-CVCV | | | | ki t O This empty site is filled by means of phonological operations with respect to the licensing status (Kaye, Lowenstamm & Vergnaud 1990): In Type I languages, the initial site is always licensed while it is not always licensed in Type II languages. The initial site is licensed when it is properly governed by the following vowel. The initial site hosts proclitics: [5] French a.

CV+CVCVCV | | | | | | l´ ta pi s la le z

>>>>

CV+CVCVCV | | | | | | l ´ t a p i s la l e z

b.

CV+CVCVCVCV | | | | | | | | [p ø l] a t o l ´ l a l e z

>>>>

C V+ C V C V C V C V | | | | | | | | l ´ [p ø l] a t o la l ez

[6] Hebrew a.

b.

Licensed

ha [k ø l] a β i m [hakklaβim] | | / \ | | | | | | C V+ CV+C V C V C V C V | \ / | | | | | | | h a r ø q a ħ i m [ha:rqaħim]2 //

NOT licensed

II. CVCV Model (Lowenstamm 1996)–Government Phonology (Kaye, Lowenstamm & Vergnaud 1990) [7] Within the CVCV Model, the skeletal tier is made of a strict alternation of CV units. Complex constituents, geminates and long vowels are represented as in [8]: [8] a. onset b. geminate c. long vowel CVCV CVCV CVCV | | | | \ / | | \ / pOla b a b a [pla] [bba] [ba:] [9] Empty Categorial Principle (ECP) stipulates that for any V position to remain empty it must be properly governed by the following vowel. 2

/r/ does not undergo gemination in Hebrew. Instead, the preceding vowel lengthens.

2

NACAL 33, March 18th-20th 2005

[10] Proper Government Given V1 and V2, two V positions such that V2 is on the right of V1, V2 properly governs V1 if a. V2 is phonetically interpreted, and b. V1 and V2 are two adjacent V positions. [11] Proper Government and ECP constrain the distribution of empty vowel positions in Berber. e.g. a. xd´m « work- Imperative, 2 singular » b. x´dmat « work- Imperative, 2 plural » a.

b.

CV1CV2CV3 | | | | x d´m

CV1CV2CV3CV4 | | | | | | x ´d ma t

PG

PG

In [11a], V1 remains empty since it is governed by the following vowel. In [11b], V3 /a/ governs V2 and V1 because it is not governed by any adjacent vowel surfaces as a schwa. III. Causative and reciprocal data, observations, analysis [12] Causative a) Tashlhiyt (Lahrouchi 2001; Elmountassir 2003) Aorist (i) mun « to be picked up » faw « to be lit » gudi « to be in a pile » miyy´l « to bow » b´rb´r « to boil » (ii)

knu rku rmi xd´m

« to bow » « to be dirty » « to be tired » « to work »

b) Kabyle (Naït Zerrad 1994) Aorist (i) m´d « to be sufficient » g´n « to sleep » luƒ « to be cloudy » laq´b « to scoff at » (ii) fruri « to be crumbled » kc´m « to enter » nd´kwal « to get better » f´d8 « learn »

Causative smun sfaw sgudi smiyy´l sb´rb´r

« to pick up » « to light » « to put in a pile » « to tilt » « to make boiled »

ss´knu ss´rku ss´rmi ss´xd´m

« to tilt » « to soil » « to tire » « to make work »

Causative sm´d sg´n sluƒ sliq´b ss´fruri ss´kc´m ss´nd´kwal ss´f´d8

3

« to give enough » « to put to bed » « to make cloudy » « to ridicule » « to crumble » « to introduce, make someone enter » « to revive » « teach »

NACAL 33, March 18th-20th 2005

c) Tarifit (Cadi 1987) Aorist (i) b´dd qas ´yy´q qd8´÷ kkar qruZZ´÷ krumb´S

(ii)

Causative « to get up » « to taste » « to be displeased » « be sharp » « to stand up » « to tumble down » « be buckled, curl »

sb´dd sqas s´yy´q ss´qd8´÷ ss´kkar ss´qruZZ´÷ ss´krumb´S

« to get someone up » « to make taste » « to displease » « to sharpen » « to lift someone up » « to make tumble » « to buckle »

d) Tamashek (Prasse 1973, Heath 2004) Perfective Positive input gloss Causative Perf. Pos. (i) ´$kuf « be inflated » Q$skQf Q$bewQs « be wounded » Q$sbAwQs Q$dobQn « marry » Q$sdAbQn « rotate » QskQ$rAkQr Q$krAkQr (ii) ´$knu « show » ´$ss´kna ´$lmQd « inform » ´$ss´lmQd « be sick » ´$ss´rhQn ´$rhin ´$stik « be empty » ´$ss´stQk

(i) /s-/ is simple when prefixed to a verbal base that begins with a consonant plus a vowel (e.g. mun > smun in Tashlhiyt) (ii) while it is geminated when followed base by a string of two consonants or a geminate in the verbal (e.g. fruri > ss´fruri in Kabyle; kkar > ss´ ´kkar in Tamashek). [13] Reciprocal a) Tashlhiyt Aorist (i) Saw´r

(ii)

ħsad S´nn´q zri knu rwi

Reciprocal « to consult » « to be jealous » « to strangle » « to pass by » « to lean » « to mix»

mSiwir mħisid mSinniq mm´zri mm´knu mm´rwi

“go back” “wish sb ill” “see” “change”

mru mƒull mz8´r8 mb´ddal

«to consult each other » « to be jealous of each other » « to strangle each other » « to meet each other » « to lean together » « to mingle, to mix with each other»

b) Kabyle Aorist (i)

ru ƒull z8´r8 b´dd´l

(ii)

Aorist nt´g qb´l rnu

3

Reciprocal “go back with someone” “wish ill mutually” “see each other” “change each other”

Passive3 “agitate” “accept” “add”

mm´ntag nn´qbal nn´rni

Medio-passive in Tamashek (Heath 2004).

4

“be agitated” “ be accepted” “grow up”

NACAL 33, March 18th-20th 2005

[14] Hypothesis “The causative and the reciprocal morphemes are underlyingly simples. Their quantity variation depends on the way they use the empty initial site: (i) /s/ and /m/ move from their originate site into the initial site when it is licensed, i.e. properly governed by the following vowel. (ii) They geminate by spreading into the empty site when it is unlicensed. [15] “ The initial site cannot remain empty inside the domain of the verb.” [16] Tashlhiyt a. PG

s mun | | | | CV+vb[CVCVCV] b. CV+vb[CVCVCV] | | | | s k n u

s mun | | | | CV+vb[CV CVCV]

>>> >>>

CV+vb[CVCVCV] | | | | | s ´ k nu

[smun]

[ss´knu]

PG

[17] Kabyle a. PG

C V+vb[C V C V C V C V] | | | \ / s b ´ d

>>>

C V+vb[ C V C V C V C V] | | | \ / s b ´ d

[sb´dd]

C V+vb[C V C VC V C V] | | | | | | s ´ q d8 ´ ÷

[ss´qd8´÷]

b. _PG_

↓ | C V+vb[C V C VC V C V] | | | | | s q d8 ´ ÷

>>>

In [16a] and [17a], the prefix /s/ moves into the empty initial site since it is properly governed and thus licensed by the vowel /u/ in /mun/ and the schwa /´/ in /b´dd/. In [16b] and [17b], the prefix /s/ spreads into the initial site since it is unlicensed. [18] Causative-reciprocal combination Tashlhiyt Aorist Reciprocal Causative + Reciprocal a. « to put one opposite the other» gab´l ngibil s´ngibil « to cause a mutual embarrassment, to make ħ´SS´m nħiSSim s´nħiSSim people feel embarrassed »

5

NACAL 33, March 18th-20th 2005

b. knu rwi

mm´knu mm´rwi

s´mm´knu s´mm´rwi

« to make lean against each other » « to make mix with each other »

[19] a. PG

g a b´l | /\ | | | CVCVCVCVCV

>>>

n g ibi l | | | | | | CV+CVCVCVCV >>>

b.

CVCVCV | | | r wi

s ´ n gibi l | | | | | | | | CV+CV+CVCVCVCV

PG

>>>

CV+CVCVCV \ / | | | | m ´r wi

[20] Kabyle (Naït Zerrad 1994) Aorist “to learn” f´d8 “to get better” nd´kwal “to get dirty” am´s

>>>

Causative ss´f´d8 ss´nd´kwal ssim´s

CV+CV+CVCVCV | \ / | | | | s m ´r wi

Reciprocal+Causative ms´f´d8 ms´ndkwal msim´s

IV. Inchoative verbs Inchoative verbs denote a change in state or quality. They all show a geminated initial consonant which degeminates when preceded by a derivational morpheme such as the causative /s-/. [21] a. Tashlhiyt “to be a friend” “to be disgusted” “to be rotten” “to be dented”

Aorist ddukk´l mmuktu llugmu dduƒZu

Causative sdukk´l smuktu slugmu sduƒZu

b. Kabyle “to stammer” “to remember” “to be accustomed “to be invented”

ll´bl´b mm´kti nnum nnulfu

sl´bl´b sm´kti snum snulfu

6

NACAL 33, March 18th-20th 2005

[22] a. vb[CV-CVCVCVCV]

| | \/ | | du k ´ l b.

[ddukk´l]

PG

CV+[CVCVCVCVCV] | // | | \ / | | s d u k ´l

>>>

CV+[CVCVCVCVCV] | | | \/ | | s du k ´ l

[sdukk´l]

V. Sibilant harmony and [labial] dissimilation [23] Tashlhiyt a. Aorist ħ´SS´m f´rr´Z kS´m rSa

Causative Sħ´SS´m Sf´rr´Z SS´kS´m SS´rSa

*sħ´SS´m *sf´rr´Z *ss´kS´m *ss´rSa

« exchange »

Aorist gab´l sam´ s8´rr´f

Reciprocal ngibil nsimi ns8irrif

*mgibil *msimi *ms8irrif

« give »

fk

nn´fku

*mm´fku

« be timid » « amuse » « enter » « be decayed » b. « supervise » « forgive »

Whenever there is a sibilant [S] or [Z] in the verbal base, the causative /s/ becomes [S]. On the other hand, /m/ dissimilates to [n] when the verbal base contains a [labial] consonant /b, f, m/. [24] « be timid » [25] -

Aorist ħ´SS´m

Causative Sħ´SS´m

Caus.+Recip. s´nħiSSim

Recip.+Caus.+Recip. ms´nħiSSim

Why doesn’t the causative morpheme assimilate with /SS/ in s´nħiSSSim and why doesn’t the reciprocal morpheme in the leftmost of the form ms´nħiSSSim dissimilate? In other words, why are assimilation and dissimilation limited to morphemes that directly precede the verbal base?

[26] “The causative and the reciprocal morphemes undergo assimilation and dissimilation, respectively, whenever they are connected inside the domain of the verb.”

7

NACAL 33, March 18th-20th 2005

[27] a.

PG

CV+vb[CVCVCVCVCV] | | | \/ | | s ħ´ S ´m b.

>>>

CV+vb[CVCVCVCVCV] | | | \ / | | S ħ´ S ´m

[Sħ´S´m]

CV+vb[CVCVCVCV] | | | | | | S ´k S´m

[SS´kS´m]

CV+CV+vb[CVCVCVCVCV] | | | | | | \/ | | n ħi S im m s ´

[ms´nħiSSim]

//

CV+vb[CVCVCVCV] | | | | | s k S ´m

>>>

c. _PG__ ↓ | CV+CV+CV+vb[CVCVCVCVCV] >>>> | | | | | \/ | | m s m ħ i S im References Basset, A. 1929. La langue berbère. Morphologie. Le verbe : étude de thèmes. Paris Leroux. Boukous, A. 1987. Phonotactique et domaine prosodiques en berbère. Thèse de Doctorat d’Etat, Université Paris 8. Cortade, J-M. 1969. Essai de grammaire touareg (dialecte de l’Ahaggar). PhD. Université d’Alger. Dallet, J-M. 1982. Dictionnaire kabyle-français. Selaf. Paris. Dell F. & Elmedlaoui, M. 1985. Syllabic Consonants and Syllabification in Imdlawn Tashlhiyt Berber . Journal of African Languages and Linguistics 7, 105-130. Dell F. & Elmedlaoui M. 1988. Syllabic Consonants in Berber: Some New Evidence. Journal of African Languages and Linguistics 10, 1-17. Dell F. & Elmedlaoui M. 1991. Clitic ordering, morphology and phonology in the verbal complex of Imdlawn Tachlhiyt Berber. Part II, Langues Orientales Anciennes Philologie et Linguistique 3, 77-104. Iazzi, E. 1991. Morphologie du verbe en tamazight (parler des Aït Attab, Haut Atlas Central) : approche prosodique. Thèse de D.E.S., Université Mohamed V, Rabat. Guerssel M. (1992), “The Phonology of Berber Derivational Morphology by Affixation”, Linguistic Analysis, vol. 22, n° 1-2 . Jebbour, A. 1996. Morphologie et contraintes prosodiques en berbère (tachelhit de Tiznit) : analyse linguistique et traitement automatique. Thèse de Doctorat d’Etat, Université Mohamed V, Rabat. Kaye, J., 1990. “Coda” licencing. Phonology 7-2, 301-330. Kaye, J. Lowenstamm, J-R. Vergnaud, 1990. Constituent structure and government in phonology. Phonology 7-2, 193231. Lahrouchi, M. 2001. Aspects morpho-phonologiques de la dérivation verbale en berbère (Tachelhit d’Agadir). Thèse de Doctorat, Université Paris 7. Lahrouchi, M. 2003. Manifestations gabaritiques dans la morphologie verbale du berbère (Tachelhit d’Agadir). Recherches Linguistiques de Vincennes, n° 32 : 61-82. Paris. Louali, N. & Puech, G. 2000. Etude sur l’implémentation du schwa pour quatre locuteurs berbères de tachelhit. XXIII Journées d’Etude sur la Parole, Aussois. Lowenstamm, J. 1996. CV as the only syllable type. In. J. Durand & B. Laks (Eds.), Current Trends in Phonology : Models and Methods: 419-441. CNRS, ESRI, Paris X. Lowenstamm J. 1999. “The Beginning of the Word”, Phonologica, J. Rennison & K. Kühnhammer (eds), pp. 153-167, Holland Academic Graphics, The Hague. Prasse, K-G. 1973. Manuel de grammaire touarègue (tahaggart), Akademisk Forlag. Copenhague. Scheer, T. 1996. Une théorie de l’interaction directe entre consonnes : contribution au modèle syllabique CVCV. Thèse de Doctorat, Université Paris 7.

8