Consonantal root extraction in two secret languages in Tashlhiyt Berber Mohamed Lahrouchi (CNRS-Université Paris 8)

&

Philippe Ségéral (Université Paris 7-CNRS)

(1) Root in Afroasiatic, a controversial object a. From linguists of the Middle Ages (Sibawayh) to McCarthy (1979, 1981), Frost et al. 2000, Prunet et al. 2000, Prunet 2006, Idrissi et al. 2008, etc.: The root is a grammatical morpheme entirely made of consonants Words are built on consonantal roots b. Ratcliffe 1987, Hammond 1988, Dell & Elmedlaoui 1992, Bat-El 1994, Ussishkin 1999, etc.: No need of such an abstract morpheme, words are derived from other words (2) Data to be studied: two secret languages used by women in Tashlhiyt Berber (TShl): a. Taqjmit (TQJM) (original data, Lahrouchi & Ségéral in press), Isouktane south-west Morocco b. Tagnawt (TGNW) (data from Douchaïna 1996, 1998), Tiznit south-west Morocco (3) Our claim: the root is a grammatical morpheme to which speakers have access. TQJM and TGNW formations show that speakers are able to a. extract the root-morpheme from any Tashlhiyt form b. build root-morphemes of a definite shape (triconsonantal) I. Extracting the root-morpheme = (3a) (4) Triconsonantal forms Disguised forms

TShl forms TQJM “be afraid” “be the first” “dream” “girl” “speak”

÷

TGNW ÷

iksud izwir wwarg tafruxt sawl

÷

ajkkasd÷wasd÷ ajzzawrwawr ajwwargwarg tiffarxwarx ajssawlwawl

tikkasd jusd tizzawrjuwr tiwwargjurg tiffarxjurx tissawljuwl

(5) Derivation TShl → TQJM, TGNW a. Morphological operations (survey): prefix tiajTQJM TGNW

1

geminate R1









-a-

infix -ju-

-wa-

(after R1)

(after R3)

(after R3)









b. Uniform pattern:

TQJM TGNW

ti R1R1a R2R3 ju R2R3 aj R1R1a R2R3 wa R2R3

c. Uniform vocalization1:

TQJM TGNW

i-a-u a

For an interpretation of vocalization in TQJM, see Lahrouchi & Ségéral (in press).

1

reduplicate R3 R2







d. Repetition:

each root-consonant is repeated twice two operations: gemination and reduplication

TQJM / TGNW i. ii.

R1 always never

gemination reduplication

R2 never always

R3 never always

(6) Template a. TQJM R1

R2

R3

t

R2

R3

I

C v C v

C

v C v C v

I

C

v

C

v C v C v

A

C

v

C

v

U

b. TGNW R1

R2

R3

I

C v C v C v

R2

R3

U

C

v C v C v

C

v

C

v C v C v

C

v

C

v

A The template is the same, encoding the various operations observed in the disguised forms as well as the principle of repetition (the additional Cv in TGNW hosts the vocalic material of the prefix). Note: in (6a-b), peripheral vowels are assumed to be underlyingly long, associated to two vocalic slots (cf. Lowentamm 1991 and Bendjaballah 2001, 2004). Evidence for this assumption will come at the close of the analysis of TGNW.

(7) Only root consonants are kept a. Any vocalic material of the input (TShl) is deleted in the disguised form. Whatever the original vocalism is, the disguised form vocalizes uniformly as in (5c). TShl tafruxt

agudi isliw argaz

TQJM au aui i a

TShl t-afrux-t au md≥uru u iz≥duj iu imz≥ij i

tiffarxjurx

tiggadjudi tissalwjulw tirragzjugz

iau

TGNW ajffarxwarx ajmmad≥rwad≥r ajzz≥adwaddi ajmmaz≥wazz≥i

2

a

“girl” “a lot” “be soft” “man”

“girl” “feel better” “be heavy” “be small”

b. Any affixal material of the input is deleted in the disguised form (e.g. i- 3ms marker, t-…-t feminine marker, n- reciprocal marker, l- definite article and m- participial marker in Arabic). TShl t-afrux-t t-amƒar-t l-axbar m-bark

TQJM tiffarxjurx timmaƒrjuƒr tixxabrjubr tibbarkjurk

“girl” “woman” “news” Proper noun

TShl t-aknari-t t-afrux-t n-s≥br l-ħml

TGNW ajkkanrwanr ajffarxwarx ajss≥abrwabr ajħħamlwaml

“prickly pear” “girl” “we wait, endure” “load”

(8) The same holds for bi- and monoconsonantals as well. See examples below, part II. II. Building the root-morpheme = (3b) (9) Triliterality condition and its consequences: a. all items in TQJM and TGNW contain three radical consonants. The "lexicon" of TQJM and TGNW is entirely composed of triconsonantal roots. b. the root material inherited from Tshl is variable : Tshl roots are mono-, bi-, tri- and quadriconsonantal. Hence, repair strategies: • First case: TShl input has three radicals → all three are kept in TQJM and TGNW, see examples in (4) and (7) • Second case: TShl input has more than three radicals (4 R’s) → the number of consonants is reduced to three2 • Third case: TShl input has less than three radicals → TGNW: epenthesis provides the missing radicals → TQJM: vocalic and affixal I / U are redeployed as radical material (10) Quadriconsonantals: one consonant is dropped in the disguised form, always R2 or R4: TShl brahim kltum TShl gʒdr aglzim asrdun asngar

brhm kltm

TQJM tibbarhjurh tikkatmjutm

brhm kltm

proper noun id.

gʒdr glzm srdn sngr

TGNW ajggaʒdwaʒd ajggazmwazm ajssadnwadn ajssagrwagr

gʒdr glzm srdn sngr

“moan” “pickaxe” “mule” “corn”

2

In TGNW, not all TShl quadriconsonantal words loose one root consonant. Some keep all their radical consonants, but one of these consonants, namely R2, is not repeated, and the melody obtained in this case, and only in this case, contains ə after the geminated R1: e.g. ggrml / ajggərmlwaml “be crusty”, agnfur / ajggənfrwafr “face”, iħnbl / ajħħənblwabl “blanket”. The systematic appearance of ə instead of a in these forms is explained in Lahrouchi & Ségéral (submitted) as the result of the association of four radicals to a template that offers only three radical positions, and the necessary association of peripheral vowels to two V positions. Once the whole material is associated to the template, the vowel A preceding R2 does no longer surface as [a] since it has access to only one V position. Depending on phonotactic conditions, the remaining V position surfaces as [ə].

3

(11) Bi- and monoconsonantals bi-

mono-

bi-

mono-

TShl gn ƒr igr ils af asi ini immi

TQJM tigganjuni tiƒƒarjuri tiggarjuri tillasjusi tiffawiwi tissawiwi tinnawiwi timmawiwi

“sleep” “read” “field” “tongue” “find” “take” “say” “my mother”

TShl ad≥n i-fl sala kl i-zz≥a kk t-ʒʒi-t is

TGNW ajttad≥nwad≥n ajffalwalli ajssalwalli ajkkalwalli ajzz≥atwatti ajkkatwatti ajʒʒatwatti ajssatwatti

“be sick” “he let” “be involved” “spend the day” “he chases after” “pass by” “you recovered” interrogative “do...?”

a. TQJM: the affixal material I and U (which surface as ju in tri- and quadriconsonantals) compensate for the missing radicals: I compensates for the missing R3 in biconsonantal inputs. U and I compensate for the missing R2 and R3 in monoconsonantal inputs. b. TGNW: t or I compensate for the missing radical in biconsonantal inputs (t- replaces R1; I replaces R3) t and I compensate for the missing R2 and R3 in monoconsonantal inputs. (12)

TQJM a. biconsonantal

gn → tigganjuni

g

n

n

t

I

C v C v

I

C

v C v C v

C

A

v

C

I

v C v C v

U

4

C

v

C

v

b. monoconsonantal g → tiggawiwi

g t

U

C v C v

C

I

v C v C v

C

I

v

C

I

v C v C v

C

v

C

v

A

(13) TGNW a. biconsonantal

adɭn → ajttadɭnwadɭn

t

n

dɭ

I

C v C v C v

U

n

dɭ

U

C

v C v C v

C

v

C

v C v C v

C

v

C

v

A b. monoconsonantal is → ajssatwatti s

t

t

I

C v C v C v

I

C

v C v C v

C

v

C

I

U

v C v C v

C

A Note: in (13b), the epenthetic I does surface in the position normally identified by the first instance of R3. The same happens in triconsonantal forms such as ajzz≥adwaddi ← iz≥duj “be heavy” and ajmmaz≥wazz≥i ← imz≥ij “be small”, where the final I surfacing as [j] is lexical. is does not lead to *ajssatjwati just as iz≥duj does not lead to *ajzz≥adjwadi. However, in both cases the copy of R2 is geminated. This is analysed as a compensation of the non-appearance of I in the first instance of R3.

(14) How do word-based models can account for these phenomena?

5

v

C

v

References Bat-El, O. 1994. Stem modification and cluster transfer in Modern Hebrew. Natural Language and Linguistic Theory 12: 571-596. Dell, F. & Elmedlaoui, M. 1992. Quantitative transfer in the nonconcatenative morphology of Imdlawn Tashlhiyt Berber. Journal of African Languages and Linguistics 3: 89-125. Douchaïna, R. 1996. Tagnawt, un parler secret des femmes berbères de Tiznit (sud-ouest marocain). PhD, Paris: INALCO.

Douchaïna, R. 1998. La morphologie du verbe en tagnawt. Etudes et Documents Berbères 15/16: 197-209. Frost, R., Deutsch, A., & Forster, K. 2000. Decomposing morphologically complex words in nonlinear morphology. Journal of Experimental Psychology: Learning, Memory, and Cognition 26: 751-765. Idrissi, A., Prunet, J-F. & Béland, R. 2008. On the mental representation of Arabic roots. Linguistic Inquiry 39/2: 221-259. Hammond, M. 1988. Templatic transfer in Arabic broken plurals. Natural Language and Linguistic Theory 6: 247-270. Lahrouchi, M & P. Ségéral. In press. Morphologie gabaritique et apophonie dans un langage secret féminin (Taqjmit) en berbère tachelhit. Revue canadienne de Linguistique / Canadian Journal of Linguistics 54.2 (37 pp.). Lahrouchi, M & P. Ségéral. Submitted. Peripheral vowels in Tashlhiyt Berber are phonologically long: evidence from Tagnawt, a secret language used by women. Brill’s Annual of Afroasiatic Languages and Linguistics. McCarthy, J. 1979. Formal Problems in Semitic Phonology and Morphology. PhD, MIT. McCarthy, J. 1981. A prosodic Theory of Nonconcatenative Morphology. Linguistic Inquiry 12: 373-418. Prunet, J-F. 2006. External evidence and the Semitic root. Morphology 16: 41-67. Prunet, J-F., Béland, R. & Idrissi, A. 2000. The mental representation of Semitic words. Linguistic Inquiry 31/4: 609-648. Ratcliffe, R. 1987. Prosodic templates in a word-based morphological analysis of Arabic. In. M. Eid & R. Ratcliffe (eds) Perspectives on Arabic Linguistics X, Current Issues in Linguistic Theory 153: 147-171, Amsterdam / Philadephia: John Benjamins. Ussishkin, A. 1999. The inadequacy of the consonantal root: Modern Hebrew denominal verbs and output-output correspondence. Phonology 16: 401-442.

6