Al-Qantara, vol.

18, nº 1 (1997)

LENITION IN T H E M O Z A R A B I C DIALECTS:
A REAPPRAISAL

DAVID HANLON
Birkbeck College, University of London

Description of the varieties of Romance formerly spoken in al-An-

dalus is complicated by the documentation of the greater part of the ex-
tant textual evidence in an alphabet which had been contrived to repre-
sent the sound system and syllabic structure of an unrelated language,
and by the lack of any appreciable attempt on the part of the alphabet's
users to adapt it to this new task. Lenition provides a measure of the
difficulties this poses: whereas it has been possible to speculate on why
the intervocalic plosives of Latin voiced in other varieties of Hispano-
Romance, students of the Mozarabic dialects have sought to establish
whether such lenition took place at all. Opinion remains divided bet-
ween those, like Meyer-Lübke i, who believe that it did not occur, and
those, like Menéndez Pidal 2, who believe that it did.
Meyer-Lübke affirms that the voicing of the intervocalic plosives in
the Peninsula did not take place until after the Germanic invasions ^.
He concludes from an acritical examination of the material that conti-
nuations of Latin /-t-/ and /-k-/ remained voiceless in the Mozarabic
dialects at the time of the Islamic invasions '^, a view shared by Hall ^.
Griffin observes that Latin /-t-/ and /-k-/ are transcribed as (voiceless)
;â'and çâ/respectively in the Vocabulista in arábico. He argues that the
former was identified with Romance /-t-/, a dental plosive that had lost
some its articulatory force but remained voiceless 6, and that the latter
1 Meyer-Lübke, W., «La sonorización de las sordas intervocálicas latinas en espa-
ñol», Revista de Filología Española 11 (1924), 1-32.
2 Menéndez Pidal, R., Orígenes del español Qs/lsíáúá: Espasa-Calpe, 1968), 253-
259.
3 Meyer-Lübke, «La sonorización», 3-5.
4 Meyer-Lübke, «La sonorización», 32.
5 Hall, R., A., «La non-lenizione nella Romania occidentale». Romance Philology 28
(1974-75), 533.
6 Griffin, D., A., Los mozarabismos del «Vocabulista» atribuido a Ramón Martí(Ma-
drid: Offprint from Al-AndalusXXlll-XXV, 1961), 57.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 122 DAVID HANLON ^ Q , XVIII, 1997

was identified with Romance /-k-/, preferred to tó/which was felt to be

too palatal ^. In further support of his case he cites continuations of La-
tin /-t-/ and /-k-/ that remained voiceless in Hispanisms of fifteenth
and sixteenth century Granadan Arabic {ruta, rucata, xucr, etc.), and li-
kewise toponyms and Mozarabisms of modern Spanish (cornato, hor-
chata, marchito, Játiva, caroca, cernícalo. Mear, etc.) ^.
To counter the claims of Meyer-Lübke, Menéndez Pidal adduces
evidence that the first documented examples of voicing in the Peninsula
are from the south: IMUDAVIT < IMMUTAVIT of the second century AD and
LEBRA < LEPRA, lUBENTUDIS < lUVENTUTIS and EGLESIE < ECCLESIA of the Se-
venth century 9. He observes that from the eighth to the eleventh centu-
ries continuations of Latin /-t-/ and /-k-/ are consistently spelt with ta '
and qàfi and only occasionally by graphs for voiced phonemes such as
dàl and gayn. He concludes that voiced and voiceless pronunciations
coexisted in al-Andalus, but that the voiceles pronunciation, under the
guise of ia'and qâf, predominated because of Arab linguistic conserva-
tism. To support this assertion, he cites Qurtuba < CORDUBA and Sara-
qusta< CAESAR AUGUSTA, in which tó'and qâfsiïe transcriptions of ety-
mological / d / and /g/ respectively ^^. Conservatism of this kind, as a
result of the prestige of a literary standard, has been and remains a po-
werful force in Arabic-speaking communities. However, there is no evi-
dence of a strong trend towards voicing in Andalusian Arabic (there
are a handful of examples of /t/ > /d/ in final position, and one in in-
tervocalic position 11, as well as evidence of / t / > /d/ ^^), and therefore
the voiced pronunciation of intervocalic stops would not have been
considered a deviation from any received pronunciation among spea-
kers of Arabic. It is unlikely that Latin is the prestige variety in this ins-
tance, since it enjoyed little standing among speakers of Romance (wit-
ness the Indiculus luminosas of Alvarus), let alone speakers of Arabic.
Galmés de Fuentes overcomes this difficulty without making explicit
reference to it. He argues that the eighth-century grammarian Sîba-
wayhi classifies tá'and qàf as «voiced consonants» (hurüf majhüra),

'^ Griffin, Los mozarabismos, 60.

^ Griffin, Los mozarabismos, 60-61.
9 Menéndez Pidal, Orígenes, 254.
^0 Menéndez Pidal, Orígenes, 254-255.
1 ^ Corriente, F., A Grammatical Sketch of the Spanish-Arabic Dialect Bundle (Ma-
drid: Instituto Hispano-Arabe de Cultura, 1977), 37.
^2 Corriente, A Grammatical Sketch, 39-40.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 ^ Q , XVIII, 1 9 9 7 LENITION IN THE MOZARABIC DIALECTS 123

and that they were, therefore, originally graphs for voiced consonants
in Arabic i^. He observes that continuations of Latin /-t-/ and /-k-/
are habitually transcribed as fa ' and qàf, and occasionally as ta ' and
kàf in other positions; therefore they represent voiced dental and
velar plosives respectively i^. According to this view, Qurtuba < COR-
DUBA and Saraqusta < CAESAR AUGUSTA are no longer examples of
hypercorrection. Toponyms and Mozarabisms in later Granadan
Arabic and the lexis of modern Spanish that evince no voicing are at-
tributed to a trend towards devoicing in Andalusian Arabic ^^. This is
also the view adopted by Peñarroja i^, but one slightly modified by
Hilty in an independent survey of the material ^^. He concludes that
at the time of the Islamic invasions, there existed three groups of pho-
nemes resulting from the Latin intervocalic dentals and velars: /-ô-/
< /-d-/ (transcribed as dàl) and /-¥-/ < /-g-/ (transcribed as gayn)\
/-d-/ < /-t-/ (transcribed as ta ) and /-g-/ < /-k-/ (transcribed as qàf)\
/-t-/ < /-t:-/ (transcribed as td ) and /-k-/ < /-k:-/ (transcribed as
kdf). In a further development, continues Hilty, the results of Latin
I'd-/ and /-t-/ on the one hand, and Latin /-g-/ and /-k-/ on the ot-
her, merged into voiced dental and velar phonemes respectively with
alternative plosive and fricative allophones ^^.
The main argument of this paper is not to show that the Mozara-
bic dialects are characterized either by voicing of Latin /-t-/ and /-k-
/, or by the retention of a voiceless pronunciation, but to demonstrate
that the use of td ' and qdf cannot, by itself, be adduced as proof for
either possibility, and that previous arguments have assumed, mista-
kenly, that the Arabic used in al-Andalus was a stable linguistic va-
riety whose sound system may be fixed diachronically and synchroni-
cally.
If the Arabic alphabet was used to write down Romance, it must be
assumed that it followed the conventions of Arabic and not those of La-

1^ Galmés de Fuentes, A., Dialectología mozárabe (Madrid: Gredos, 1983), 61-62.

14 Galmés de Fuentes, Dialectología mozárabe, 92-96.
15 Galmés de Fuentes, Dialectología mozárabe, 92,177, 202.
16 Peñarroja Torrejón, L., El mozárabe de Valencia (Madrid: Gredos, 1990), 339-
348.
1^ Hilty, G., «Das Schicksal der lateinischen intervokalischen VerschBlaute -p-, -t-,
-k- im Mozarabischen», Festschrift Kurt Baldinger zum 60. Geburtstag, eds. Manfred
Hôfler, Herni Vernay and Lothar Wolf (Tubingen: Niemeyer, 1979) 1,145-160.
18 Hilty, «Das Schicksal», 158-160.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 124 DAVID HANLON ^ Q , XVIII, 1997

tin. Therefore, Galmés de Fuentes is right to seek to discover the pho-

nological value of ta 'and qàf. However, his study limits itself to the lite-
rary language but paradoxically appeals to evidence for the spoken lan-
guage to account for anomalous voiceless Mozarabisms and toponyms
that survived the extinction of the Mozarabic dialects. This procedure
relies on three assumptions that may be challenged: first, that an eighth-
century Persian grammarian is an authoritative source for the Arabic
used for written purposes in al-Andalus; second, that such a written
language was widely known in all epochs during which any loans are li-
kely to have been made; and third, that consequently it was the domina-
ting factor when the Arabic alphabet was used to write down vernacular
Romance.
To my knowledge, fa 'is universally realized in Classical Arabic as a
voiceless plosive. However, SIbawayhi, an early and respected autho-
rity, appears to prescribe a voiced pronunciation; he states that if were
not for the velar contour, /â 'would become dal (lawlà l-itbàqu la-sàrati
l'tà 'u dàlan) ^^. This is an unequivocal statement; the controversy over
whether the distinction mahmüs («whispered») and majhür («clearly
used»), to which Galmés de Fuentes appeals, corresponds to the voice
vs. voiceless distinction made by modern phonology need not detain
us 20. However, Sïbawayhi also documents a voiceless pronunciation
which he describes as «infrequent and unacceptable» ^i. It appears that
despite the prescription of Sïbawayhi, the literary standard throughout
the Arabic-speaking world favoured the voiceless pronunciation which
established itself rapidly 22. it would seem that al-Andalus was no ex-
ception to this norm on the evidence of the censorship of a voiced re-
alization of ta' in the lahn al-'àmma literature: qubbayd< qubbayp^;

19 Quoted from Matar, 'A., Lahn al-'àmma fî daw' al-diràsàt al-lugawiyya al-hadîtha
(Cairo: Al-dár al-qawmiyya li-1-tibà'a wa-1-nashr, 1966), 224.
20 See Blanc, H., «The "Sonorous" vs. "Mufled" Distinction in Old Arabic Phono-
logy», To Honour Roman Jakobson (Mouton: The Hague, 1967) I, 295-308; Garbell, I.,
«Remarks on the Historical Phonology of an Eastern Mediterranean Arabic Dialect»,
Word 14 (1958), 303-337; al-Nassir, A. A., Sibawayh the Phonologist (London and New
York: Kegan Paul International, 1993), 35-38.
21 Al-Nassir, Sibawayh, 19.
22 Cantinean, J., Etudes de linguistique arabe (Paris: Klincksieck, 1960), 32,171.
23 Al-Zubaydî, Lahn al-'àmma, éd. 'Abd al-'Aziz Matar (Cairo: Dar al-ma'àrif,
1981), 112.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 v4Q, XVIII, 1997 LENITION IN THE MOZARABIC DIALECTS 125

qubbayd< qubbayt^^\ muliddun< mulittun^^; dàbid< dàbit^^. In the

case of qáf, the pedagogic tradition, in particular that of Quranic reci-
tation, universally favours the voiceless pronunciation, again despite
the prescription of Sibawayhi. Cantinean suggests that this is because
of the predominance of the voiceles pronunciation in urban commu-
nities where schoolmen were educated ^7. it is difficult to know whet-
her the written Arabic of al-Andalus was characterized in the same
way since there is no explicit censure of a voiced pronunciation in the
lahn aWàmma literature. However, grammarians do comment on the
confusion of gâ/and kàf, which appear to have suffered a phonemic
merger, at least partially, since they were distinguished only by a na-
rrow difference in point of articulation 2^: hakk< huqq^^; tarka-
wa < tarquwa ^O; maqqàs < makkàs ^i; qub < küb ^2; istaktala < istaq-
tala^^; Washka< Washqa^^. Educated speakers were particularly
careful when reciting the Quran to give qàfan uvular pronunciation,
but would disregard the uvular/velar distinction in speech, on the evi-
dence of the following remark by al-Safadï concerning the gramma-
rian Abu Hajryán (1256-1344): «His pronunciation is correct fin] the
speech of al-Andalus, he articulates the qáf close to káf although in [re-
citation] of the Quran his pronunciation [of qàj\ is correct» ^^. This
indicates a voiceless pronunciation in the written language and upper
registers of the spoken languages.
It would seem that, despite his authority as a codifier of the written
language, Sîbawayhi's prescriptions regarding the realization of ta ' and
(jâ/were largely ignored by many educated speakers of Arabic in al-An-
dalus.

24 Ibn Hishám al-Lakhmï, Al-madkhal ilà taqwîm al-lisàn wa-ta'lîm al-bayân, éd. José
Pérez Lázaro (Madrid: Consejo Superior de Investigaciones Científicas, 1990) 2, 56.
25 I b n H i s h á m , ^/-mflúfA:/zfl/2,190.
26 Ibn Hishám, Al-madkhal 2, 260.
2"^ Cantinean, Études, 67.
28 Corriente, A Grammatical Sketch, 54; for later periods see Barceló Torres, M. C,
Minorías islámicas en el País Valenciano. Historia y dialecto (Valencia: Instituto Hispano-
Arabe de Cultura, 1984), 175.
29 Al-Zubaydi, Lahn al-'ámma, 80.
30 Al-Zubaydï, Lahn al-'ámma, 122.
31 Al-Zubaydï, Lahn al-'ámma, 143.
32 Al-Zubaydï, Lahn al-'ámma, 152.
33 Al-Zubaydï, Lahn al-'àmma, 200.
34 Ibn Hishám, Al-madkhal 2, 287.
35 Quoted from al-Tayyár, R. 'A., Al-dirását al-lugawiyya fí l-Andalus (Baghdad: Dar
al-Rashid li-1-nashr, 1980), 223.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 126 DAVID HANLON AQ, XVIII, 1997

A good deal of the evidence for the Mozarabic dialects consists of

Romance items borrowed by Arabic. Loans of this kind are typically
made during a period of cultural clash; the invading language is forced
to borrow since it does not have the lexical resources to denote the re-
ferents (flora, fauna, etc.) of its new environment. In the context of al-
Andalus, this period is likely to comprise the three centuries between
the invasion of the Peninsula and the political zenith of the Umayyad
caliphate. Assuming for a moment that the prescriptions of Sîbawayhi
are relevant, was the written language, for which he is regarded an aut-
hority, well known throughout this period to the extent that it could in-
fluence the orthography of Romance borrowings?
During the period immediately following the invasion, written Ara-
bic is unlikely to have been well known. The conquerors of the Penin-
sula where a mixture of Berbers and Arabs, of which the former appear
to have been the larger group. Linguistically little is known about
them 36. Presumably they spoke Berber, but they had recently been sub-
dued by Arabic-speaking forces and converted to Islam. We would as-
sume that as a speech group they were characterized by an incipient
Berber/Arabic bilingualism whose exact nature is impossible to ascer-
tain. The Arabs comprised Yemeni tribesmen who entered the Penin-
sula under the command of Mùsà ^^, and Syrian forces commanded by
Baly b. Bisr who arrived in 741 under caliphal orders to calm Berber
unrest ^^. The vast majority of this Arabic-speaking body of men is unli-
kely to have known the literary dialect well. It is likely that the speakers
who were responsible for borrowing Romance items in this period
comprised the invading troops and their immediate descendants, who
were possibly a poorly educated and, in some cases, illiterate group.
Even if ta ' and qàfwcxt voiced consonants in written Arabic, speakers
would probably have given them the values of their native phonemic
systems if they could write. Since the Arabic spoken in the Peninsula
during this period was necessarily characterized by imported dialectal
divergence, the geographical back-ground of the invaders and early
settlers determined its major constituents, that is to say, Yemeni and

36 Wasserstein, D. J., «The Language Situation in al-Andalus», Studies on the Mu-

wassah and the Kharja, eds. Richard Hitchcock and Alan Jones (Reading: Ithaca Press,
1991X4,12-13.
3"^ Táhá, 'A. D., The Muslim Conquest and Settlement of North Africa and Spain (Lon-
don: Routledge, 1989), 94,118-128.
3^ Táhá, The Muslim Conquest, 132.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 AQ, XVIII, 1 9 9 7 LENITION IN THE MOZARABIC DIALECTS 127

Syrian tribesmen and their clients who dominated the political stage of
the nascent Andalusian state and whose dialects would have enjoyed
widest currency. Orthographic vagaries would have occurred because
of this dialectal divergence: a modern example would be English /g/
transcribed as jim in Egypt and qáf in the Gulf region. Residual Yeme-
ni features have been detected in Andalusian Arabic ^9. Indeed, Yeme-
ni linguistic features have clearly influenced the orthography of Ro-
mance loans in Peninsular Arabic, for example the plosive jîm in Tàjuh
< TAGus and Jallîqiyya < GALLAECIA ^O. Another Yemeni feature is the
voiced pronunciation of tà'^^. Therefore, during this period, for spea-
kers of Arabic in the Peninsula ta ' was likely to have been a graph for
voiced and voiceless dental stops. The voiced pronunciation of qàfïs
widely distributed among the rural varieties of spoken Arabic: the north
of the Arabian Peninsula, the Yemen, Oman, Iraq, Upper Egypt, the
Hawrán, Jordan, Algeria, Morocco and southern Tunisia ^2. The voice-
less pronunciation is characteristic of the settled urban areas of Syria,
Iraq, North Africa, Tunisia, Morocco, Algeria and Oran ^3. Whether
the same rural/urban distribution was characteristic of al-Andalus im-
mediately following the Islamic invasions is not known, but it is almost
certainly true that both pronunciations existed. Therefore, the written
language is unlikely to have influenced the orthography of Romance
loans made during this period, and furthermore it is impossible to iden-
tify with any degree of accuracy what values ta 'and çâ/may have had.
There are two valuable sources for determining the extent of the
knowledge of the written language in al-Andalus from the post-inva-
sionary period to the fourth century AH. The first is the Tàrîkh 'ulamà '
al-Andalus whose author, Ibn al-Faradi, was born in Córdoba in 351/
962 44. He achieved renown as a jurist and /za^ii/scholar, was appointed
qàdî of the great mosque of Valencia, and died in the Berber sack of

39 Corriente, F., «South Arabian Features in Andalusi Arabic», Studia Lingüistica et

Orientalia Memoriae Haim Blanc dedicata, eds. Paul Wexler, Alexander Borg and Sasson
Somekh (Wiesbaden: Otto Harrassowitz, 1989), 94-103.
40 Corriente, «South Arabian Features», 98.
^'^ Cantinean, Études, 32.
"^2 Cantinean, Études, 69-70; al-Nassir, Sibawayhi, 37.
^^ Cantinean, Études, 68-69; al-Nasssir, Sibawayhi, 37.
44 Ibn al-Faradî, Tàrîkh 'ulamà'al-Andalus, éd. Ibrahim al-Abyàrï, 2 vols. (Cairo and
Beirut: Dàr al-kutub al-islâmiyya, 1983), henceforth known as IF for the sake of conve-
nience.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 128 DAVID HANLON AQ, XVlll, 1997

Cordoba led by al-Musta'ïn in 403/1013 ^5. The work is among the

older and more complete of the Andalusian biographical dictiona-
ries 46^ whose chronological scope ranges from the date of the inva-
sion to the end of the fourth century AH. The second is the Tabaqàt
al-nahwiyyîn wa-l-lugawiyyîn, whose author, Abu Bakr al-Zubaydï,
was born in Seville in 316/982 ^7. He studied in Córdoba under Abu
'All l-Qàlï and became judge of the great mosque and sahib al-shurta in
Seville, where he died in 379/989 ^s. The chronological scope of the
Tabaqàt ranges from Abu 1-Aswad al-Du'ali (d. 69/688), to al-Zu-
baydi's teacher Muhammad b. Yahyà al-Rabàhï (d. 358/968). The
two sources yield sufficient data to construct a sample of one hun-
dred and fifty-one individuals said to be grammarians or interested in
grammar and lexicography who lived during the period circumscri-
bed by the Islamic invasions and the zenith of the Umayyad Calipha-
te. The sources mention a further thirty-nine individuals whose date
of birth/and or death are not given either independently of or relative
to contemporary political events. I have omitted twenty-nine from the
final sample but have included ten by supplementing the data of the
sources with extensive biographical research carried out by Marin ^^\
it should be noted, however, that the quality of the sample has been
compromised which qualifies the conclusions reached in this study

45 Pons Poigues, F., Ensayo bio-bibliográfico sobre los historiadores y geógrafos arábi-
go-españoles (Madrid: Establecimiento Tipográfico de San Francisco de Sales, 1898),
105-108.
46 Ávila, M. L. and Marín, M., «Le Ta'ríh Vlamá' al-Andalus d'Ibn al-Faradí: étude
et informatisation», Cahiers d'Onomastique Arabe 1985-87, (Paris: Editions du Centre
National de la Recherche Scientifique, 1989), 41-42.
47 A l - Z u b a y d ï , Tabaqàt al-nahwiyyîn wa-l-lugawiyyîn, éd. M u h a m m a d A b u 1-Fadl
Ibrahim (Cairo: Muhammad Sàmî Amïn al-Khanji, 1954), henceforth know as A Z for
the sake of convenience.
48 IF no. 1355, pp. 786-789; Pons Boigues, Ensayo, 90-93.
49 Marín, M., «Nómina d e sabios d e al-Andalus (93-350/711-961)», Estudios
onomástico-biográficos de al-Andalus I, ed. M. Marín (Madrid: Consejo Superior d e
Investigaciones Científicas, 1988), 23-182. The twenty-nine scholars omitted are (no-
te that in cases where I a m unsure of the proper vocalization of names, I have ommit-
ted the short vowels): Ibrahim b. Wahb (IF no. 4 4 , p. 52); Bakr b. Khatib al-Murádí
1-Makfüf al-Nahwî (IF no. 290, p. 174; A Z , p. 297); Sulaymán b. Yûsuf al-Qaysï (IF
no. 562, p. 329); Sá'id al-Muqri' (IF no. 612, p. 356); Diyà' b. A b i l-Daw' (IF no. 616,
p. 357; A Z , p. 3 Í 7 ) ; ' A b d al-Rahmàn b. Hassan al-Khawlânï (IF no. 792, p . 447);
'Uthmàn b. Shann (IF no. 893, p. 515); Abu 1-Fath Sa'dà (AZ, p. 308); Ahmad b. 'Abd
al-Karïm (AZ, p. 313); Muhammad b. Asbag b. Násih al-Murádí 1-Mujaddir (Az, p.
313); Abu l-'Abbás Bajwam"(AZ, p. 314); Ibn Waqqás al-Qurasï (AZ, p. 315); Mud-
haj al-Mu'addib (AZ, p. 315); Muhammad b. Ganim al-Udaym ( A Z , p. 315); ' A b d

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 AQ, XVIII, 1 9 9 7 LENITION IN THE MOZARABIC DIALECTS 129

(the blame for this lack of detail is usually to be laid at the door of al-
Zubaydi whose standards of scholarship fall short of those set by Ibn
al-Faradï).
Towards the end of the first century of Islamic administration, and
the beginning of the second, we see the first signs of an attempt to esta-
blish Arabo-Islamic cultural supremacy, and to foster a knowledge of
the literary koine of the Islamic world. Only two scholars are mentioned
who are reputed to have known the literary dialect well: the Cordobán
al-Gàzï b. Qays (d. 199/814), who was a student of Malik b. Anas and
also met al-Asma'i and other linguists (IF no. 1013, p. 578; AZ, pp.
276-278); and Jùdï b. 'Uthmàn (d. 198/813) from Morón, but who later
settled in Córdoba. He met al-Kisá'í, al-Farrà' and others, was said to
have introduced a work by al-Kisà'ï to al-Andalus {wa-huwa awwalu
man adkhala hitaba l-Kisà 'í), and to have written a work on grammar
(AZ, pp. 278-279).
Towards the end of the second century AH and throughout the
third, a knowledge of written Arabic spread further through the popula-
tion. Thirty-one scholars are mentioned by Ibn al-Faradï and al-Zubay-
dï who achieved fame for their knowledge of the written standard and
who died in the third century AH. Although the majority (seventeen)
are from Córdoba, there is evidence of a wider geographical spread of
expertise, with scholars from Saragossa, Jaén, Carmona, Niebla, Elvira,
Ecija, Tàkurunna, Morón and Algeciras. There are several figures of
importance. The polymath and author of a universal history (Kitáb al-
tàrîkh), 'Abd al-Malik b. Habïb (d. 238/853), was also considered a
grammarian by his contemporaries (IF no. 814, p. 459; AZ, pp. 282-
283) 50. 'Uthmàn b. al-Muthannà (d. 273/886) from Córdoba trave-
lled to the East, met various scholars of grammar and studied under
Abu Tammàm whose dlwàn he brought to al-Andalus (IF no. 889, p.

Allah b. Mu'min b. 'Udáfir al-Tujîbî 1-Marwakî (AZ, p. 316); 'Isa b. Abî Jurthümat
al-Khawlànï (AZ, p. 316); Bahlül al-Ja'thamï (AZ, p. 317); Tâhir (AZ, p. 317); 'Abd
al-Samad (AZ, p. 317); 'Uthmân b. 'Amr al-Mawmrï (AZ, p. 318); Yùsuf b. Sulaymàn
al-kátib (AZ, p. 322); Abu 'Abd Allah al-Mklfkhï (AZ, p. 329); Ahmad b. Madá' Ibn
al-Hisár (AZ, p. 331); Idrîs b. Mytm (AZ, p. 332); Faraj b. Fzlmân (AZ, p. 334);
'Uthmân b. Ibràhïm al-Brshqïrï (AZ, p. 334); Ishàq b. Ibrahim b. Muhammad (AZ,
p. 334); 'Àfî b. Sa'ïd al-Makfûf (AZ, p. 334); Muhammad b. Fath (IF no. 1296, p.
738).
50 See also Pons Boigues, Ensayo, 29-38; Aguadé, J. in Ibn Habib, 'Abd al-Malik,
Kitàb al-ta'rîj, éd. Jorge Aguadé (Madrid: 1991), 15-56.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 130 DAVID HANLON ^ Q , XVIII, 1997

513; AZ, p. 288) ^i. The Cordobán traditionist Muhammad b. 'Abd

al-Sallàm al-Khushanï (d. 286/899) travelled to the East where he met
grammarians of the school of al-Asma'î, and introduced works of
grammar to al-Andalus (IF no. 1132, p. 648; AZ, p. 290) 52. 'Abd al-
Rahman b. Mùsà from Ecija travelled to the East and studied under
Malik b. Anas and al-Asma'I (IF no. 776, p. 439; AZ, pp. 275-276).
'Abd al-Wáhid b. Sallám (d. 209/824), was a famed teacher of gram-
mar and author of a work on grammar (IF no. 856, p. 494; AZ, p.
279). 'Abbas b. Firnás (d. 274/887), a polymath and inventor who
pioneered the production of rock crystal and made a half-successful
attempt at flight ^3, was also famed for his knowledge of the literary
standard (AZ, pp. 291-292) 54. 'Abbas b. Násih al-Thaqafí of Algeci-
ras, famed for his panegyrics of al-Hakam b. Hishám (180/796-206/
822), travelled to the East with his father and met Abu Nuwàs and al-
Asma'i (IF no. 879, pp. 504-505; AZ, p. 284). Other authors of
works on grammar include Mufarrij b. Malik al-Nahwî from Córdoba
(IF no. 1446, p. 843; AZ, p. 297), and Yahyà b. 'Abd al-Rahmán (d.
263/876) of Saragossa (IF no. 1559, p. 902).
In this as in other periods of Andalusian history, state intervention
in education was non-existent 55. For this reason, mosques assumed im-
portance as centres of education; they did duty as public libraries and
were the only exposure to learning and the literary standard for those
who wished to better themselves but did not have the means to provide
for private tuition. Scholars and clerics would establish a circle (halqd)
and give lectures on matters religious, linguistic and literary. For the
common believer, mosques brought familiarity with the received pro-
nunciation used in common prayer and liturgy. In the major provincial
centres great mosques were founded throughout the ninth century: Se-
ville (214/829); Jaén (218/833); and Lérida (289/901).
Sponsorship of the literary standard reached its fruition in the reign
of 'Abd al-Rahmán III (300/912-350/961), which saw the political
unification and apogee of the Andalusan state. The indigenous popula-

51 See also Ibn Sa'ïd al-Magribï, Al-mugrib fi hulà l-magrib, éd. Shawqï Dayf (Cairo:
Dáral-ma'árif, 1953)1,112.
52 See also Pons Boigues, Ensayo, 48.
53 Ibn Sa'ïd, Al-mugrib 1, 333; Vernet, Juan, La cultura hispanoárabe en Oriente y
Occidente {B2ivœ\on2L\ Ariel, 1978), 28.
54 Date of death given in Ibn Sa'ïd, Al-mugrib 1, 333.
55 Ribera y Tarrago, J., La enseñanza entre los musulmanes españoles (Córdoba: Pu-
blicaciones de la Real Academia de Córdoba, 1925), 15-25.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 AQ, XVIII, 1 9 9 7 LENITION IN THE MOZARABIC DIALECTS 131

tion began to participate in Arabic and Islamic culture in large num-

bers; they assumed administrative posts and conversion became com-
monplace. Accordingly, the figures given by Ibn al-Faradï and al-Zu-
baydi increase dramatically. There are one hundred and tv^enty-eight
scholars in total reputed to know written Arabic well who died in the
fourth century AH. Again, nearly half are from Córdoba (fifty-eight),
but the geographical distribution is much greater with representatives
from Beja, Seville, Sidonia, Elvira, Ecija, Jaén, Saragossa, Fahs al-Ba-
llüt, Carmona, Algeciras, Osuna, Tortosa, Badajoz, Morón, Calatayud,
Toledo, Pechina, Guadalajara and Malaga. This period witnessed one
of the most important events in the history of grammatical studies in al-
Andalus: the arrival of Abu 'All Ismà'ïl b. al-Qasim al-Qàlï (288/900-
356/966) from Baghdad in 330/941 to dictate his al-Amàlî (IF no.
221, p. 138) ^6. He exerted considerable influence because of his tea-
ching activities. Those who studied under him include al-Zubaydï him-
self (IF no. 1355, p. 768), Ibrahim b. 'Abd al-Rahmàn al-Tanasï (d.
387/997) of Córdoba (IF no. 47, p. 53), 'Abd Allah b. Asbag (d. 373/
983) of Córdoba (IF no. 726, p. 407), 'Abd Allah b. Shu'ayb b. AM
Shu'ayb (d. 389/998) of Osuna (IF no. 753, p. 421), Muhammad b.
Ibrahim b. 'Abd al-Rahmàn b. Mu'áwiya b. al-Mundir (d. 373/983) of
Córdoba (IF no. 1340, p. 761), Muhammad b. Aflah (d. 385/995) of
Pechina (IF no. 1369, p. 781), and Hishàm b. Hubaysh of Toledo (IF
no. 1540, p. 891). Another outstanding figure is Ibn al-Qütiyya (d. 367/
977), remembered to posterity principally for his historical writings,
but who was also famous to contemporaries as a grammarian and aut-
hor of such works as Kitàb tasàrlf al-afàl and Kitàb al-maqsür wa-l-
mamdüd(lF no. 1316, p. 747). Makkí has documented the spectacular
rise in the importation of eastern grammatical and lexicographical
works in this period, which further stimulated the cultivation of home-
grown scholarship ^'^. It is unsurprising, therefore, to note that many
scholars from the sample are reported to have made trips to learn di-
rectly from scholars in the heartlands of Islam. A number of Andalu-

56 See RasMd, B. Y., «Abu 'Alî al-Qâlï: Vida y obra. Estudio de sus transmisiones lin-
güístico-literarias», Miscelánea de Estudios Arabes y Hebraicos XXXI (1982), 17-45;
XXXII (1983-1984), 37-54; XXXIV-XXXV (1985-1986), 271-286; XXXVII (1988),
211-225.
5"^ Makkï, M. 'A., Ensayo sobre las aportaciones orientales en la España musulmana
(Madrid: Instituto Egipcio de Estudios Islámicos, 1968), 256-268.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 132 DAVID HANLON AQ, XVIII, 1997

sians studied under the Egyptian Abu Ja'far Ahmad b. Muhammad b.

Ismá'Il al-Nahhàs (d. 338/950), for example: Khattáb b. Maslama (d.
372/982) of Carmona (IF no. 402, pp. 243-244); 'Abd al-Salàm b. al-
Samh of Morón (IF no. 855, p. 492); Muhammad b. Ishaq b. Mundir (d.
367/977) of Córdoba (IF no. 1317, p. 749); Husayn b. Fath of Seville
(IF no. 352, p. 207); and Muhammad b. Mufarrij b. 'Abd Allah (d. 371/
981) of Córdoba, who is reported to be the first individual to bring to
al-Andalus the Kitàb Vràb al-Qufàn, Kitàb ma'ánil-Qur'àn and Kitàb al-
nàsikh wa-l-mansükh of al-Nahhàs (IF no. 1329, p. 756). Two further
students of Abu Ja'far al-Nahhás also studied under the Egyptian scho-
lar Ibn Wallád (d. 332/943): Muhammad b. Yahyà b. 'Abd al-Salám al-
Rabàhî (d. 358/968), who imported a copy of Sîbawayhi's Kitàb into
the Peninsula (IF no. 1290, p. 736; AZ, p. 336); and Mundir b. Sa'id (d.
355/965) of Córdoba (IF no. 1452, p. 845; AZ, pp. 319-320). One
Musa b. Asbag al-Muràdî of Córdoba also studied under the Basran
scholar Ibn Durayd (d. 321/933) (IF no. 1462, p. 853). Lastly, Qàsim
b. Thàbit b. Hazm (d. 302/914) of Saragossa travelled to the East with
his father and brought to the Peninsula a copy of the Kitàb al-'ayn (IF
no. 1060, pp. 605-606; AZ, p. 309).
In sum, the knowledge of written Arabic was limited to a small élite
towards the end of the first century of Islamic administration. It was not
to become widespread until the end of the ninth and beginning of the
tenth centuries when it was possible for it to exert a levelling influence
on the spoken languages and the orthography of borrowings from Ro-
mance. For long periods, therefore, it remains impossible to predict the
phonological values of ta'and qàf when used to write down vernacular
Romance. In later periods, it is possible that the prestige of the written
language, on the evidence of the lahn al-'àmma literature, favoured voi-
celess pronunciations that were inevitably used in upper registers of the
spoken language. It is doubtful, however, whether the lower registers of
the spoken language can be ignored, which brings me to my third point.
Was the written language the dominating influence when the Arabic
alphabet was used to write down vernacular Romance? The dichotomy
established between H(igh) and L(ow) varieties of Arabic by Fergu-
son ^8 has been questioned by Kaye ^^. A linguist who has spent any

58 Ferguson, Ch. A., «Diglossia», Word\5 (1959), 325-340.

59 Kaye, Alan S., «Remarks on Diglossia in Arabic: Well-defined vs. Ill-defined»,
Linguistics 18 (1972), 32-48.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 A Q , XVIII, 1 9 9 7 LENITION IN THE MOZARABIC DIALECTS 133

length of time in an Arabic-speaking country cannot have failed to noti-

ce the myriad gradations that exist along the social and stylistic axes of
Arabic, which depend on the education, background and sex of the
speaker, as well as the topic of conversation and the interlocuter. Any
decision as to where H begins and L ends will inevitably seem arbitrary;
this would suggest a continuum rather than a discrete H/L distinc-
tion 60. For example, in Egypt thà ' may be realized as /0/, / s / or /t/.
Therefore, we must take into acount the values of ia'and qàf in lower
registers of Arabic.
An urban spoken standard seems to have emerged in the tenth cen-
tury, the result of the merger of the native dialecte of the early settlers,
the influence of the Romance substrate, and the levelling influence of
the written superstrate ^i. In this and subsequent periods, m'had two
possible realizations: voiceless and voiced. The former was the prestige
norm and the latter suppressed as substandard ^2. The real reason for
this may be particular to the language attitudes of al-Andalus (and, the-
refore, beyond our discovery), but one obvious possibility is that the re-
alization of ta ' in written Arabic was voiceless. In the ninth and tenth
centuries, qdf wdis consistently unvoiced and merged in many areas with
/-k-/, but a voiced pronunciation remained which was substandard and
possibly rural ^^. Since before this period spoken Arabic was characte-
rized by a dialectal divergence that has already been commented on, at
no time from the invasion to the establishment of the caliphate can qàf
and ta ' in spoken Arabic be identified with fixed phonological values.
Throughout the entire period under review, the phonological values
of ta ' and (ja/when used to transcribe reflexes of Latin /-t-/ and /-k-/
cannot be predicted, for different reasons over different periods of ti-
me: from the invasion to the eighth/ninth centuries because of the exis-
tence of imported dialectal divergence, and the negligible to uneven re-
gulating influence of a literary superstrate; and from the eighth/ninth
centuries onwards because, despite the existence of a widely-known li-

60 Cf. Blanc, H., «Style Variations in Spoken Arabic: A Sample of Interdialectal

Educated Conversation», Contributions to Arabic Linguistics, ed. Charles A. Ferguson
(Cambridge, Massachusetts: Harvard Middle Eastern Monograph Series, 1960), 81-
156; El-Hassan, S. A., «Educated Spoken Arabic in Egypt and the Levant: A Critical Re-
view of Diglossia and Related Concepts», Archivum Linguisticum 8, no. 1 (1977), 112-
132.
61 Corriente, A Grammatical Sketch, 7.
62 Corriente, A Grammatical Sketch, 39-40.
63 Corriente, A Grammatical Sketch, 53-54.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 134 DAVID HANLON AQ, XYlîl, 1997

terary superstrate, voiced and voiceless realizations of these sounds

coexisted in the spoken language, and the social and stylistic variables
that determined the use of one rather than the other are possibly be-
yond our investigation. Corriente may be right that the voiced realiza-
tions characterized those communities who prided themselves on their
Romance heritage or were stigmatized because of it ^'^. But the asser-
tion, on the one hand by an Arabist, that ta 'was voiced because it was a
graph for Romance /d/ continuing Latin /-t-/ 6^, coupled with the as-
sertion, on the other hand by a Romance scholar, that Latin /-t-/ voi-
ced because ta 'was voiced in Arabic ^^, produces a circular argument.
The problem is compounded by the commonly-held assumption
that the Mozarabic dialects are a single linguistic variety. Given the lin-
guistic diversity that existed in the area north of the Andalusian border,
which for most of the period under review existed within narrower geo-
graphical confines, we are probably dealing with several typologically
distinguishable linguistic varieties at least as diverse as those to the
north (the adoption of a working hypothesis of diversity has been the
great contribution to this field of study by Galmés de Fuentes). If this is
the case, and given the evidence of voicing of /-t-/ and /-k-/ in non-li-
turgical Latin texts written in al-Andalus in the ninth and tenth centu-
ries 6'^, it is very likely that one or more of the Mozarabic dialects were
characterized by voicing of /-t-/ and /-k-/. The task facing those who
wish to identify successfully such examples of lenition is to adopt a
structuralist approach similar to that used by Hilty ^^, and match refle-
xes of /-t-/ and /-k-/ transcribed by ta'and qâf with examples of frica-
tion or deletion of /-d-/ and/or degemination of /-t:-/ from the same ti-
me and place. Because their phonological values cannot be predicted,
the use of the graphs fâ 'and qáf is not sufficient evidence of such leni-
tion 69.

64 Corriente, A Grammatical Sketch, 40.

65 Corriente, A Grammatical Sketch, 40-
66 Galmés d e Fuentes, Dialectología mozárabe, 6 1 - 6 2 et passim.
67 F o r example, didissimas < DITISSIMAS, repedat < RÉPÉTÂT, Hludouicus < G e r m a n i c
Hlutowigch, instripidu < INSTRIPITU, eglesia < ECCLESIA, Gregi < GRECI, digumur < DICUNTUR
in Vespertino Rodríguez, A., «La sonorización de las consonantes sordas intervocálicas
en el latín de los mozárabes». Homenaje a Alvaro Galmés de Fuemes (Madrid: Gredos,
1985)1,348-349.
68 Hilty, «Das Schicksal».
69 I wish to record my gratitude to my colleague Professor Roger Walker, and Dr
Richard Hitchcock and Dr Lynn Williams of the University of Exeter, who generously
gave their time and expertise to comment on earlier drafts of this paper.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)
 v4Q, XVIII, 1 9 9 7 LENITION IN THE MOZARABIC DIALECTS 135

RESUMEN

El propósito de este estudio no es defender ni negar la sonorización de las

I-i-1 y /-k-/ latinas en los dialectos mozárabes, sino proponer que, siendo difí-
cil fijar el valor fonológico de las grafías ta 'y qáfen los varios registros del ára-
be andalusí, su uso no prueba la sonorización de dichas consonantes ni su con-
servación como sordas.

ABSTRACT

The main argument of this paper is not to show that the Mozarabic dialects
were characterized either by voicing of Latin /-t-/ and /-k-/, or by the reten-
tion of a voiceless pronunciation, but to demonstrate that, because their phono-
logical values in the various registers of Andalusian Arabic cannot be fixed, the
use of tá 'and ^a/cannot be adduced as proof for either possibility.

(c) Consejo Superior de Investigaciones Científicas http://al-qantara.revistas.csic.es

Licencia Creative Commons
Reconocimiento 4.0 Internacional (CC BY 4.0)