HARVARD MIDDLE EASTERN MONOGRAPHS
x
COMMUNAL DIALECTS IN BAGHDAD
BY
HAIM BLANC ~
.• D ISTRIBUTED FOR THE
.CENTER F...
124 downloads
2949 Views
31MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
HARVARD MIDDLE EASTERN MONOGRAPHS
x
COMMUNAL DIALECTS IN BAGHDAD
BY
HAIM BLANC ~
.• D ISTRIBUTED FOR THE
.CENTER FOR MIDDLE EASTERN STUDIES OF HARVARD UNIVERSITY BY HARVARD UNIVERSITY PRESS CAMBRIDGE , MASSAC HUSETTS 1964
-
FOREWORD This study is the result of research carried on in Israel and in the United States between 1957 and 1962. I had the good fortune of being able to devote the years 1958- 1960 almost entirely to this project under a research fellowship of the Center for Middle Eastern Stud ies, and wish to express my indebtedness to Prof. H. A. R. G ibb, to D. W. Lockard and to the staff of the Center fo r the cord ial reception and generous help accorded me. I am also indebted to my chairma n, Prof. H. J. Polotsky, for releasing me from my duties at the Linguist ics Department of the H'ebrew Un ive rsity and for his inc isive comme nts on various parts of thi s work. Special thanks are due C.A. Ferguson, who was in strum en tal in getting me to the Center, follo wed thi s wo rk in its various stages, and gave the typesc ript a th orough go ing-over.
."
--
My colleague G. Baer was ki nd eno ugh to go over the historical and demographic sections, and the typesc ript was expertly a nd painstakingly prepared for publication by Mrs. M. Smith of the Center. The onerous typing job was done by Miss Carolyn Cross at the Center and Miss Esther Rosenthal in Ierusalem. Given the circumsta nces, research cou ld not be ca rried o ut in Iraq itself, and the material had to be gathered from speakers resid in g elsewhere. Speakers o f the A natol ian dia lects, consulted for comparative purposes, were also, though for different reaso ns, more easily accessible in the U.S. or Israe l. I am, of course , aware of the deficiencies of such a procedure, but I hope major mishaps have been avoided. More detailed data on the sources used are given in sect ion 1.2 and in the notes, though the names of the native speakers cons ulted
have been omitted. Those who have kept their Iraqi nationality must, I th ink , remain anonymous; these include all my Mus lim and mos t
of my Christian informants, without whose patient and friendly cooperation th is study could never have been comp'eted. Of the many others to whom I am indebted, I should li ke to mention Mrs. R. Gone nc, for much of my information on Christ ian Baghdadi, and
her husband, J. Go nenc, for my data on the dialect of Qarabiis; my
-
v
vi
COMMUNAL DIALECTS IN BAGHDAD
old friend and teacher A. J. Chaurize for several useful hints, including some on his native Siirt dialect ; Mrs. R. Shamash and her relatives (Carmac, Siverek), A. Zakay (Mosul) ; S. Khalil (Hit); A. Haskel (Basra); H. Kohen (Amara). Most of my information on Jewish Baghdadi comes from a good· friend , A. Loya, and from the persons whose acquaintance I made through him: his wife Gladys, his sisters Bunayya and Naomi, his friend s Reuben and Ruth Khazzam (the latter from Basra) and Miss E. Cohen. It is, in fact, to A. Loya's vast knowledge and keen insight that lowe some of my basic notions as to the structures and interrelations of the dialects studied. H.B. Jerusalem April 1963
,
c..:
u..
c:::;
CONTENTS I. INTRODUCTION Aim and Scope . Sources . Baghdad Other Dialects. . The Mesopotamian Dialect Area Overall view. The qeltu-gelet Split . Main Contrastive Features o f the qeltu-gefet Split. The Baghdad Situation . . Majority and Minorities: Some Figures Correlation between Community and Dialect Degree of Uniformity with in Each Dialect Communal Dialects in the Rest of Iraq . Affinities with Other Dialects
3
3 3 3 4 5 5
7 8 8 9
9 10 10
2. COMMUNAL DIALECTS IN THE ARAB WORLD Social Dialects. Socio-Economic Groupings Religious Groupings Communal Differentiation in Arabic Overall View Minor Differentiation Intermediate Differentiation Major Differentiation
12 12 12 13
3. PHONOLOGY Procedure and Notation Consonants . Overall View .' Reflexes of the OA Interdenta ls . Reflexes of OA/I l Reflexes of OA/rl Reflexes of OA/kl Reflexes of OA/ql Vowels . Overall View. . Reflexes of OAlil and lui Reflexes of OAlal . Reflexes of OA/i and luI Reflexes of OA final la'l The 'imlila. Reflexes of OA/ayl and lawl Stress .. Consonant Clusters and Anaptyxi s.
17 17 17 17 19 19
13 13 14
15 16
20 25 26 30 30 35
38 41 41 42
50 52 ~3
vii
" CONTENTS
COMMUNAL DIALECTS IN BAGHDAD
viii
Overall View
53
~O=
~
4. MORPHOLOGY
57
Procedure and Notation Basic Morphophonemics Overall View The Pronoun Subject Pronouns Object Pronouns . .
, . . Indirect and Double Object Pronouns The Femin ine Suffix Overall View. . . AJlomorphs in the Isolated Noun AlJomorphs in Sandhi Unit Nouns Some Ot her Suffixes The Sound Plural Suffixes Relational and Agent Suffixes Hypocoristic Suffixes The Noun Overall View Irregular Patterns Regular Patterns The Nu merals The Cardinal Numbers The Ordinals The Part iciple Fo rm I Participles . Participles of Forms II and III Other Participles The Verb Overall View Form I '" . . Forms lI , IIf , V. and vr, and Quadriconsonantal Verbs Form IV Form VII . Form VIII. Form IX Form X Verb Modifiers. Olher Morpheme Classes . Determination Markers. The Relative L' . . . ReOexes of OA/Ii / and /, ila/ . Some Prepositions
5. SOME SYNTACTIC FEATURES Introductory. . . The Postpositiona l Copula C vs. MJ Comparative Absence of Article
57 57 59 59 59 64 66 68 68 68 71
72 73 73
74 74 74 74
,
75 75
90 90 93 93 93 94
96 97 97 97 108 III III 113 114 115 115 11 8
11 8 120 120 121 124 124 124 124 125 125
Constructions with / miil/ Noun Plus Qua lifier Anticipatory Pronou n Suffix Plus L Object of Verb. Object of Preposit ion and Noun.
6. SOME LEXICAL FEA TU RES . Introductory. Basic Vocabulary. The Swadesh List The Ferguson-Said Lis t. Interrogatives and Demonstratives Overa II View Interrogatives The Demonstratives Elements of Non-Arabic Origin Turkish . Persian. Aramaic The Heb rew Element in J Selected Lexical Items
7. SUMMARY AND CONCLUS ION. Summary of Findings. Muslims vs. Non-Muslims Jews vs. Christians. Christians vs. Mosu l Musl im vs. the Countryside. Uniqueness of J Characterization of the Dialects Conservatism in JC Beduinization in M A Glance at Iraqi History
ix 125 126 128 128 131 133 133 133 134 135 136 136 136 138 140 140 140 145 160 160 160 162 164 165 166 166 166 167 168
ABBREV IATIONS USED IN REFERENCES ANO NOTES
172
REFERENCES CITED
173
NOTES.
18 1
MAP. The Mesopotamian Dialect Area
2
HARVARD MIDDLE EASTERN MONOGRAPHS
x
COMMUNAL DIALECTS IN BAGHDAD
BY
-
HAIM BLANC
I
I D ISTRIBUTED FOR THE
CENTER FOR MIDDLE EASTERN STUDIES OF HARVARD UN IVERSITY BY HARVARD UN IVERS ITY PRESS
CAMBRIDGE, MASSACHUSElTS
1964
z
-< ::;Ef-<
~f-<~~
:r:o......l~
INTROD UCTION
f-<~::;-<
00 V)
~
::;ElF
--<
.l ><~)
.!;
Ul
;;; ~
:;;
/ \
/
'.
/
/
w
m ~
/
~
~
••
,
~ ~
( ~
• :E «: • • ~
::>
~
)
r
><
'~
'~
.(J
•
•
LI AIM AND SCOPE. The present monograph attempts to describe the linguistic situation that obtains among the Arabic-speaking populations of Baghdad and other towns of Lower Iraq .1 Though detailed research was done for Baghdad o nly, the evidence indicates that a wholly a na logo us situation ex ists in the other communally mi xed urban centers of Lower Iraq. The basic feature of this situation is the unusually profound and sharply delineated dialecta l cleavage that divides these populations into three non regional dialect groups, corresponding to the three major rel igious communities, namely the Muslims, the Jews, and the Christians. Laconic statements on the existence of this trichotomy may lie fou nd here and th ere in the literature, beginning with the late nineteenth century,2 and some descriptive material on the several dialects is also available.' In the prese nt study, data collected at first hand from informants from the three communities will be set for th, collated, and compared. The remainder of this chapter lists the sources used and presents an overall view of the Meso potamian dialect area and of the Baghdad situation. Chapter 2 discusses socia-reli gious dialects in the Arab world a nd elsewhere, and Chapters 3- 6 study the main features of Muslim Bag hdad i (= M), Jewish Bag hdad i ( = J), and Christian Baghdadi (= C). Chapter 7 reeapitufates the main similarities and di fferen "es and probes the question of their ge nesis and preservati on. 1.2 SOURCES. 1.21 Baghdad. Data o n M and C ste m principally from persons residing or studying in the Un ited States, while data o n J was gathered chiefly from persons now residing in Israel. The half-dozen Muslim informants consist of four Sunnis and two Shiis, all nati ve Baghdadis, fhough in some cases one of the parents, and in most cases at least one gra ndparent, was not Baghdad i and even not Arabic-speaking. All the Muslim informants were stud ents in 3
4
-
COMM UNAL DI ALECTS IN BAGHDAD
l NTROD UCTION
their twenties, all males, and all from the middle o r upper strata of Baghdad society, hence had a good command of modern litera ry Arabi c. T he Christian info rmants consisted primarily of a woman of Jacobite extraction and two men of Chaldean origi n; all three were nati ve Baghdadis, but in each case one of the parents was from outside Baghdad. In the last two cases, o ne gra ndparent was no nArabic speakin g. Both of the men were university students, onc in his thirties and one barely twenty; the woman , in her thirt ies, was a housewife and d id clerica l wo rk ; all had a passable kno wledge of modern literary Arabic. The Jewish info rmants were all nati ve 8aghdadis of nati ve slock, and in most cases both grandparents were also nati ve Bag hdadis; th o ugh I was ex posed to this di alect from a score or more persons, I count here only those four men and three women, all in their thirties, whom I interviewed at length and repeatedly. All had a good knowledge of literary Arabic. In addition to these, [ made use of a number of tape recordings of Radio Bag hdad prog ra ms in Muslim Baghdadi, and of the available literafure on the dialects concerned. 1.22 Other dia lects. For purposes of pe rs pective and co mparison, it was found necessa ry to gather basic data on as many other Meso potamian dialects as possible under circumstances that made a fi eld trip to the area im practicable. I inter viewed at length in the United States two Muslim informants from Mosul , and more briefl y one Clujstian and (in Israel) one Jewish informant from that city. I obtained ' particularly detailed data on the 'Ana dialect fro m a single M uslim informa nt fro m that city. Fo r rural Lowe r Iraq, I interviewed a Columbia University student who was a native of Qa rya t Sayyid Razin, a hamlet of perhaps two hu~dred inhabitants in the al-l;Iilla province (n al,l iya of Sadd at al-Hindiya, qa<;la' of Musayyab); the in fo rmant, a Shiite in his middle twenties, seemed to have retained a good command of his nati ve dialect. For the rural dialects of the Amara region, I had no direct co ntact with info rmants, but have drawn on a number of tape reco rdings, one fro m Radi o Bag hdad (a play in the ru ra l di alect of th at regio n, wi th no specification of loca lity and an air of only partial authenticit y) and two kin dly forwa rded to me by Gra ham Leonard , who made them while in the Middle East. Both informants thus recorded were from th e AI-8u M ~a mm ad su b-d istrict. Eac h reco rded a brief auto biography. . a story or anecdote, and a number of wo rds and phrases
from a list provided by me. Mr. Leonard also sent me similar recordin gs of informants from Kuwayt. To Victor Ayoub [ am deepl y mdebted for a similar recording, made in [ran, of an infor ma nt from Ahwaz in Khuzistan. On the dialect of the Je ws of Amara I have had many communications from a frien d to whom th is d ialect is native and who has, in addition, permitted me to make use of his unpublished work o n the verb in th at di a lect ( Ko hen, n,d.) . For Basra, I have interviewed one Sunni Musli m in the United States and obtained detailed data from one man there and ~ne woman in !erusa.lem. ~he scanty information I have on Hit ste ms fro m a single 1Il1erVIew with a Karaite of that town now residing in Beersheba; for Tekrit I have only some indirect information from persons who made brief visits to the city. For the Anatolian dialects, I obtained rather detailed information fro m Jewish informants from Urfa and Swerek (Siverek) in Urfa province, a nd fro m CarmO c (also pronounced Carmiik, perhaps i. q. <";ermik in Diya rbe kir province), and from a Jacobite nati ve of Qara biis (perh aps i. q. Kara bahc;e in Diya rbeki r province), as well as some sketchier data fra in a Chaldea n native of Se'ert (Siirt, in the province of that name). All references to these dialects not attributed to a ~v ritte n source come from these informants.
5
1.3 THE MESOPOTA MIA N DiALECT AREA. 1.31 On the basis of the inco mplete data furni shed by the literature and by my informants, it seems possible to sketch a tentative outline of the dialect area that stretches from the Persian Gulf along 'and between the Tigris and the Euphrates nearly up to th e ve ry sources of th e two ri vers on the Anatolian plateau. In such a vast area, regional variation is bound to be great, the more so as the population is separated, in many points, by large stretches of desert with a nomadic population and, in addition, by large non-Arabic-spea king co ncentrations. One may, nevertheless, speak of a Mesopotamian dialect area . Within this area one ' must fi rs t distinguish between two large groups of dialects, each of which shares a great num ber of basic fea tures, and . each of which correlates roughly with a regional su bd ivisio n and' also (but even more ro ughl y) with a n ecological di visio n. T hese two gro ups may be called (fro m t he 1st pers. sing. of the perfect of the verb 'to say'), res pecti vely, qellu-dialects and gelel-dialects.4 The latter are spo ken by the M uslim p~p ul a ti o n (sedentary and non-
6
COMMUNA L DIALECTS IN BAGHDAD
I NTRODUCTION
sedentary) of Lower Iraq, and by the non-sedentaries in the rest of the area; the former are spoken by the non-Muslim' population of Lower Iraq and the sedentary population (Muslim and non-Muslim) of the rest of the area. This may be represented as follows:
to the verb and III - prefi xed to the object, e.g. Mlsefta laxuyal = Ise fet axuyal 'you saw my brother'; U) the use of laku/ ' there is', usually with its negative Imakul over
(no n-sed.) Lower Iraq Upper Iraq Anatolia
gelet gelet gelet
Muslim (sed.) gelet qellll qeltll
No n-Muslim
qeltll qellll qeltu
The qellll-dialects a re akin to the oasis dialects described by Cantineau· and to the sedentary dialects of the Aleppo region. The geleldialects are closely re lated with the Beduin dialects of the Samiya, on the one hand, and the dialects of Kuwayt, Khuzistan, and the Persian Gulf area on the other. 7 The relationship between the Iraqi dialects and Central Asian Arabic remains to be probed. 7a The main fea tures that c haracteri ze the Mesopotamian dialect area, in the
sense that they a re shared by most of the dialects irres pective of the qellu-gelet ~icho t o my, and are lackin g or not common in most other
7
most of the area, though apparently not in Anatolia; (k) many characteristic lexical items, e.g. Imall 'of, belonging to', IhassaC)1 'now', reflexes of OA /,albaril)al (not of /,al ba rib/) 'yesterday', Imezl ' table', Ibun!al 'wheat', Itemmanl 'rice', Ibazzun(a)1 'cat' (in Anatolia (j~an no r/), /j i gfua/ 'cigarette', /I axl and / lux/,another (r. and m. , respecti vely); and a large number of Turkish and Persian loanwords. 1.32 Of particular importance to the present stud y are the features characteri zi ng the qellu-gelel split. These are illustrated in the table below, with Mosul representing the qellu gro up and Muslim Baghdadi the gelet gro up. It must, of course, be kept in mind that not all of the differentiating features are eq ually general in the group they represent ; all arc, however, representati ve to some degree, in the sense that
none are peculiar only to the dialect mentioned. In the tab le, the items are chosen to illustrate the following features: (a) refl exes of OA Iq/; (b) reflexes of OA Ik/; (c) reflexes of OA Ir/; (d) reflexes of of OA lui ; (e) reflexes of OA/al before Iii ; (f) lreflexes of OA lal before Iii ; (g) reflexes of OA word final /a'l ; (h) reflexes of the nominal
dialect areas, may be summarized as follo ws :
feminine morpheme in several environments; (i) the pron. suff. of
(a) presence of a Ipi phoneme over most of the area; (b) preservation of the interdentals 10/, 15/, and II:)I with two exceptions .noted so far, both in the qeltu-dialects ; (c) presence of the a ffri cates lei and Ijl, though the individua l dia lects vary as to the historical correspondence of these ; (d) a bsence of a 1'1 reflex of OA8 Iql so common in the regions
the 3rd pers. masc. sing. and of the 3rd pers . fern . sing.; (j) the personal endings of the verb in the 1st pers. sing., the 3rd pers. fern. sing., and the 3rd pers. plur.; (k) the presence or absence of the two Form I verb paradigms, as in OA l kalabl vs. Isarib/; ( I) the terms for 'here', 't here', and 'thus',
Mosul
adjoining the Mediterranean, bo th western and eastern;
(e) the preservation of OA diphthongs laYI a nd lawl over a good part of the area; (f) the preservation of OA - Inl in the imperfect 2nd pers. fern. sing. and 2nd and 3rd pel's. plur., e.g. M/tketbinl 'you (r.s .) write', Itketbun/ 'you (pl.) write', /yketb un/ ' they write'; (g) the use of a sort of indefinite article, e.g. M/fadd / ; (h) plurals of the type qldla for adjectives of the type qalltill , e.g. M/ksalal sing. Ikaslanl 'lazy'; (i) the marking (with a frequency va rying from dialect to dialect) of the. definite direct object by a n anticipatory pron. suff. appe nded
(a) (b) (c) (d) (e) (f) (g) (h)
qal ka n yk un gal) kell jemc'
bazizin soda badli
Muslim Baghdadi gal
'he said'
can
'he was'
yku n
'he will be'
fal)
'he went' 'all' 'mosque'
kull jamc' bzazin soda badla
'cats' 'black (r.)' 'a suit of clothes'
8
(i)
(j)
(k)
(1)
COMMUNAL DIALECTS IN BAGHDAD
i)en!a badleti badletna badletu badleta abunu abuha katabtu katabet katabu katab segeb segbet segbu honi hnuka or honek hakeo
I) un!a badelti badlatna badelta badlatha aba abuha ktabet ketbat ketbaw ketab serab serbat
serhaw hna hnaka hici
'wheat'
'my suit' 'our suit' 'his suit' 'her suit' 'his father' 'her father '{ wrote' 'she wrote'
'they wrote' 'he wrote' 'he drank' 'she drank' 'they drank' ' here' 'there'
'thus'
1.4 THE BAGHDAD SITUATION . 1.41 Recent figures for the total population of Baghdad, with Kai)miya and other subur bs, give a total of abo ut 780,000' A breakdown of th is figure into members of the Muslim, Jewjsh, and Christian communi ties is hard to come by and has, moreover, undergone considerable changes in the past two or three decades. T his change results principaily from (a) the very large immigration of Muslims that has taken place from the countryside, (b) the smaller but steady infl ux of Christians from the north, (c) the wholesale emigration of Jews that took place about 1950. About 1920, the total population is estimated at around 200,000, with about 135,000 Muslims, 50,000 Jews, and 15,000 Ch ristians' 0 Judging from available ""stimates, I I it seems fair to state th at the ratio of Muslims to Jews to Christians for a century or so preceding )950 was not far removed from this, i.e. oscillated around a ratio of 7 to 2 to 1. There are, unfortunately, no estimates of how many non-Arabic speakers there were in each community: Kurds, Turks, and Persians among the Musli ms; Armenians, Assyrians, and Aramaicspeakin g Chald eans among the Christ ians; perhaps some Persianspeaking Jews. My impression is that the Jews were almost all Arabicspe'aking, while a good many Muslims (especially before World
INTRODUCTION
9
War J) and proportionately even more Christia ns spoke other languages. The only point that is essentia l for this prefatory synopsis of the Baghdadi Arabic dia lects is that there is a majority dialect (of the gelel type) and two minority dia lects (of the qellll type). 1.42 That three markedly different dia lects, each with its own phonology and (especially) morphology and each with its own syntactic and lexical peculiarities, are spoken by members of the three relig ions is beyond doubt. It seems nearly as certain, with all the reservations stemming from the fact that th e inves tigati on was not carried out in situ, that th is dialect cleavage correlates rully with communal affiliat ion. Provided th ey are Arabic-speaking native Baghdadis, all Muslims talk M, all Jews ta lk J, and all Christians ta lk C. A few qualifications which do not invalidate this statement are neverth eless in order. In the first place, J a nd C are spoken respectively by Jews and Christians largely at ho me and wit h coreligiollists, wh ile M, the domina nt dialect, is used in public and in intercommunal situations by the many Jews and Christians who have a a command of it. A good many non-Muslims seem to be nearly perfectly bi-dialectal. As a consequence of the special position of M as majority dia lect, non-Arabic speaking Christians (most notably the Armenians) who acqu ire A rabic as a second language may learn not C, but M ; I am personall y acquainted with a number of Baghdadi Armenians who, as a second language, speak o nly M. Another conseque nce of the special status of M is th e tendency among some nonMuslims, especially Ch ristians, to adopt it, or certain features of it, even when speaking with their coreligionists, s~ ' that there may conceivably be non-Muslim homes in which M or somethin g very much li ke M is spoken rat her than C or J. Questi oned as to this, my informants all indicate that th ey know of no such situation, but they do report a number of ihdividual M features replacing "old-fashioned" C or J features in the speech of the younger ge neration. 1.43 While J is a relatively unifo rm dialect, there are some mi nor differences among my C in formants and some variat ion within M as well.. The variati ons within C canno t, at the present stage, be cor-· related with the speaker's affiliatio n to a specific sect or ri te. The va riations within M do not seem to be in any way connected with the five Muslim dialects believed to exist in Baghdad by L. Massignon (see note 2); some are clearly due to the existence, in present-day Baghdad, of large groups of immigrants from the countryside. Some
10
COMMUNAL DIALECTS IN BAGHDAD
INTRODUCTION
of these va riants are id e ntified by my M informants as "provincial,"
though in some cases it might also be possible, given further resea rch, to correlate them witb social class, the lowe r strata being, to a considerable degree, made up of just such recent arriva ls from the
the dialect of the Muslims bears a stron g fa mil y resemblance to that of the Lower Iraq i co untryside and to the nomads of the area as a whole ; that the dia lect of the Ch ristia ns is very close to that of Mosul ; and that the dialect of the Jews, while no t as close as the other two to
vi lla ges. I have, so fa r, found no differences as between Sunnis and
any dialect so far described , is far more similar to the Christian and
Shiis (a nd inform ants do not seem to be aware of the existence of such differences) nor have I attempted to look for v'ariation as between the different quarters o f the city, though here informants are more inclined to believe that such differences exist, wh ile unable to cite insta nces (cf. note 97a). 1.44 A situation quite si milar to that described above seems to ex ist in the other to wns of Lower Iraq , though
J
II
Mosul type than it is to the Muslim dia lects of Lowe r Iraq. What Inferences may be drawn from th ese ali gnments wi ll be discussed in
Chapter 7.
have direct evidence
for Basra and Amara o nly. Not only do Jewish informants from these cities speak a dialect which is, to all in tents and purposes, identical wit h Jewish Baghdadi, but the existence of the three-way split in their ho me cities, with the Muslim s spea king a gelet-dialect and the Jews an d Christian s a qellu-dialect is to them a matter of common kno wledge. In the case of Basra, thi s is confirmed by a Muslim info rmant, who speaks a dialect quite similar to Muslim
Baghdadi. In the case of Amara and of the Basra Christians, there is some ev id ence from th e literature l2 and fro m two of my Baghdad i Christian informants who , having visited these places, indicate that
the Christians there speak "just like we do." On the other hand, this socia-religio us clea vage does not seem to exist in Upper Iraq. From informants and from the Iitel'ature lJ it is clear that members of all co mmunities in Mos ul speak a qeltu-di~l ect, as do those of 'Ana and
(probably) th ose of Tekrit and Hit. There are unquesti onably subdialecta l differences within those citi es (my Jew ish informant from
Mosul happens to have a number of feat ures different from my Christia n a nd Muslim Mos ul info rm ants), but these may not correlate w:ith communal affiliation. Informants and literatu re (see note 13) attribute these differences to quarters within the c ity and to other variables (age, degree of educa ti on or modernization, ultimate provenience fro m some neighboring town). Though minor differences corre lating with communal affiliation may nevertheless exist, the
situation is at a ll events quite different from th at of Baghdad and (
,
L
•
:
Lower Iraq. 1.45 Underscoring once more wha t has already bee n sa id about the qeltll-gelet split, ihs worth pointing out that, in a city li ke Baghdad,
,• I
,
I I
,,I
COMMUNAL DIALECTS IN THE ARAB WORLD
2
COMMUNAL DIALECTS IN THE ARAB WORLD 2.l SOCIAL DIALECTS. 2.ll Dia lects correspo ndin g to group affi liation rather tb a n to geographical location have not bee n extensively studied. Descriptions of territorial dialects can be cou nted in the hundreds, a nd the various aspects of dialect geography have a prime place in ling ui stics a nd a chap ter or two in any treatise o n language. One looks in vai n for a nyth ing of the kind ha ving reference to socia l dialects or to Hdialect sociology," though rece nt decad es have brought forth, perhaps as a by-product of gene ra l develo pments j'n socia l science, so me attempts in thi s di rection. 14 In part thi s lag is, no doubt , due to the ract that geographicall y defined dialects are, a t least in Europe, more co mm o n, more striki ngly differentiated , and more eas ily pigeonholed. When o ne deal s, as t he dia lectol og ist usuaUy does, with a relativel y homoge neo us rural population, o ne arrives a t rough ly accurate results by fairly uncomp licated techniques requ iring a tnin-imum of extra-li nguistic knowledge a nd all boili ng down to a few common se nse field and mapping techniques, Co-terminous social g roupings are clearly not as a menable to delimitation, and the va riables wit h which a given ling ui stic feature is to be correlated may be considerably more proble matic than is th e variable of spatial location used for geog raph ical dialects, In addition, since socia l g roups and social "distance" do see m somehow essentially different from territoria l groups and spatial distance, there may well be a mo re intrinsic differe nce betwee n social dialects o n the o ne ha nd and geographical dialects on the other. On the evidence ava ila ble so far , it seems that differences a mo ng social d ia lects tend to be more s ubtle and mo re marginally ling uistic than diffe rences among reg io nal dialects. It is hard to find sociall y defi ned dia lects ex hi biting the so rt of all-embraci ng phono logical a nd morphological differences that are a com mo nplace in a rea dialectology, When this is the case, i,e. when two social groups show this so rt of structural differentiatio n, it see ms 12
l3
that a n explanation involving geography must inevitably be invoked: migratIOn, mfluence of another territorial d ia lect or of a different language. An exa mple fam iliar to Arabists is that of th e split between sedentary and non-sedenta ry dialects in Arabic. The Galilean villager and the Bed uin herdsman encamped nearby speak widely divergent dialects ex h l bltm~ major structural differences, but an explanat ion of the fact must Invoke the Bedu in's ultim a te provenie nce (perhaps ~any ge ~eratlO n s remo ved) from the A rabian desert , whence his ?Ialect migrated with him . The villager's dialect was, of co urse, also Imported from Arabia, but it was tran splanted at a different time and underwent a separate evolution . The origi n of the d i ffer~nce mu s; ~hus be ,sought in geography and history, a nd it is its prese rvat ion that IS due In large measu re to social dista nce. 2.1 2 Dialect differences among specifically reli gio us groupings ha ve attracted even less attention than th ose among socia-economic groups, and are in so me ways eve n more problematic. l s Some religious groups speak languages ex t raneou s to the area; others s pea k t,he local language with some diffe rences due to the influence of a n extraneQus .Iiturgical or sac red language. On th e basis o f th e fragment a,ry eVidence at hand, it appea rs that differe nces amo ng religious gro,upmgs are usually even more marginal than those among ot her SOCIal groupings; they tend , typically, to be few a nd non-structural in character: differences in the nam e of the Deity, different greetings and other formulas, different given names, different lang uages used as ~odel s for imitation or So urces of borrowing. In suc h cases, it is difficult to speak of " d ialects" at a ll, except in the broadest possible se nse of the word. A few insta nces of more substantial dialectal differenc~s among religiou s gro ups have, ho wever, been reported ; here agam, these seem to have originated in migrations and to have been preserved by social distance. These include those found here and there in Arab ic (see nex t sectio n) , in parts of the Serbo-Croa tian a rea and in parts of Persia,16 2.2 COMMUNAL DI FFERENTIATIO N IN ARABIC. 2.2 1 In matters of dialect differentiation a lon g li~es of religious affiliation the Ara~ic-speaking area presents a whole spectrum of situatio ns g'oino f rom complete or nearly compl ete absence of d ifferentiat ion ' to the0 sh arp cleavage of Lower Iraq, but aside from North Afri ca, the hterature ha s little to sayan the subject. Within this spec tru m one
14
COMMUNA L DIALECTS IN BAGHDAD
must distingui sh between the two ex tremes of "minor differentiation" and "major differentiation" with, of course, a number of intermediate cases. Cases of " minor" differentiation are characterized by two
features, namely (a) differentiation is primarily margillal to the
COMMU NAL DIALECTS IN TH E ARAB WORLD
IS
2.23 Intermediate differentiation. In some cities of North Africa (Algiers, Fez, Tlemcen, and some others) a differentiation which might be called " intermediate " has bee n reported. Here we find Jewish and Muslim dialects (there arc, so to spea k, no nati ve Arabic-
linguistic structure, and (b) there is fluctuation in usage and poor corre lati on of differences with religious affiliatio n. Cases of "major"
speaking Christians) wh ich prese nt, in addition to differences of the
differentiation exh ibit the opposite characteristics. There is considerab le evidence to suggest that minor and major differentiation are due to
correlate well, for the most part, with communa l affiliation. The follow ing table, abstracted from Cohen, 1912, lists all the differences reported there as between Muslim Algiers (= MA) and Jewish Algiers ( = JA):
entirely different sets of historical circumstances; some detai ls follow.
2.22 Minor differentiation . In the towns of Greater Syria, Egypt, the Arabian Penins ula, and Upper Iraq , either no differentiation alo ng religious lines has been reported , or the reports indicate
that differentiation is marginal, vague, and fluct uating. Jerusalem Arabic. for example, is communa ll y differentiated insofar as some
(not all) Christians and Jews deviate from majority usage wi th respect to intonation pattern s, voca l qualifiers, certain all ophones, the frequency of certain consonantal assimi lations, the proportion of Classical or Neo-Class ical vocabu lary item s, certain idioms, and the prop~rtion and phonetic trea tm ent of European loanword s. 17 In the case of the Jews , there are additional differences due to the presence of a Hebrew vocab ulary and , for some speakers, the influence of a number o r .non-loca l varieties of Arabic~ imported via immigration.
Some of the differences just listed are quite subtle, some are unstable, and few if a ny correla te perfectly with communal affiliation. In other words, it is quite difficult for anyone to say wit h certainty that a given utterance will be spoken in a given way by a Jew , in another way by a Muslim, and in a third way by Christian . A similar situation seems to obtain in Aleppo and Cairo , though to an even lesser degree. ls
a
In Yemen, the dialect spo ken by the San'a Jews and that spoken by the Muslims in that city differ principally (perhaps exClusively) by the Hebraic element in use among the former. 19 In the North Pales-
tinian villages studied by the writer (Blanc, 1953) differentiation among Muslims, C hristians, and Druzes is imperrectl y correlated
and limited to a handful of items, though some of them are structural (e.g. the reflexes of OA /q/ and the OA interdentals). The vague classification suggested by Feghali, 19.28, divid ing Mt. Lebanon into Christian and Dru ze areas seems to be of the same character ; more precise, if still preliminary, data on communal differentiation in
Lebanon may be found in Fleisch, 1959.
minor variety. a number or morc strictly structural differences. These
(a) OA/. / -+ JA/s/ vs. MA/. / OA/i/ -+ JA/z/ vs. MAIi/ (b) OA/r/ -+ lA/r/ vs. MA/R/ or /gj (c) OA/q/ -+ lA/,/ vs. MA/q/ (d) OAft! -+ lAft! vs. MA/e/ (lts/) (e) OA/ h/ -+ lA zero vs. MA/ h/ (f) OA/ iF'/ -+ lA/if 'a/ vs. MA/ iF'/ (g) fem . ending is lA [aj or [ej vs. MA [aj (h) in certain words, different treatment of OA/u l in closed syllables, e.g. / korsi/ vs. MAlkuRsi/ 'chair' (i) certain differences in intonation and stress, largely affective U) 3rd pers. masc. sing. pron. suff. after V: lA/ u/ vs. MA/a/ (k) differences in the phonemic shapes of so me five or six morphemes, viz. those for the particles 'of,' 'with ,' 'as,' 'fro m
here,' ' there', a nd (perhaps) 'yesterday' (I) some lexical differences, e.g. in kinship term inology and terms of the stone-cutting trade (a Jewish monopoly) (m) quan tity of Romance lexical loans: JA more, MA less (n) degree of phonetic assim ilation of loa nwords: lA less, MA morc (0) Hebrew loan words : JA many, MA none, exce pt one or two
loans from lA Features (a), (b), a nd (c) do not correlate perfectly, i.e. some speakers of lA show usage listed as MA and vice versa. Features (m) and (n) also imply relat ive frequency rather than absolute presence or absence
of the given usage. Ex ternal influences seem to be directly involved only in features (m), (n), and (0). A very simi lar picture emerges from Fez and. from less detailed accounts, al so . for Tlemcen and some towns of the Algiers region. 20 I call this differentiation interm ed iate
/
I
16
COMMUNAL DIALECTS IN DAGI-IOAD
because, though quite systematic and well-correlated, the differences are few and the dialects belong to the same class or type, i.e. to the urban or sedentary di alects of the region" 2.24 Major differentiation. For some other North African cities, notably Oran, and some smaller towns of the region of Algiers, reports indicate a differentiation between Jews and Muslims wherein the former speak a "sedentary type" dialect and the latter, like the Muslims of many North African towns, a " Beduin type" dialect. " This may be called "major" differentiation, in that (a) it permeates the whole phonology and grammar of the dialects, and (b) correlates fully with community membership. The fact that the Muslim townsmen speak a "Beduin type" dialect may be due to either migration or diffusion. The parallel with the Lower Iraq situation is striking in all respects, at least as regards the contrast between Muslims and nonMuslims; whether the contrast between Jews and Christians should be classified as major or merely as intermediate is of little consequence. The only other case of major differentiation that comes to mind is but partially ana logous: the Druzes of Mt. l;Ioran speak a dialect wholly" different in type from that of the Muslims of the Horan plain;23 these groups, howeve r, inhabit two adjoin ing but separate rural areas, and the Druze dialect was almost certai nl y imported ill tolo from Mt. Lebanon via migration. In this case, differentiation is therefore essentially historical and geograph ical, and only incidentally correlates with religious affiliation.
,
3
PHONOLOGY 3.1 PROCEDURE AND NOTATION. The present chapter surveys the phonologies of M, J, and C from a point of view that is essentially historical and comparative. Phonetic description of phonemes is held to a minimum ; on the other hand, a good deal of mo rphophonem ic detail is included, especially in the sections on conso nant clusters. In each of the sections, an overall view of the present situation is first given, followed by a more detailed disc ussion of the divergence in historical correspondences which constitutes the main differences among the three dialects. The symbols used are, it is hoped, largely self-explanatory and familiar; note, however, that since C is used for " Christian Baghdadi," the symbol K will be used for "any consonant" side by side with V for ".any vowel. " Word stress is marked only where its position cannot be determined by the well-known rule of thumb, "stress VKK or VK closest to the end of the word; if neither VKK nor VK are present, stress first V. " 3.2 CONSONANTS. 3.21 Overall view. (a) The inventory of consonant phonemes is nearly "the same" for M, J, and C. By this is meant that a single set of sym bols can be used for all three with, on the whole, very nearly the same allophonic content for each. One exception is the absence of interdentals in C , resulting in an additional phoneme NI in (} corresponding to MJ I¢f. There are, however, considerable differences in distribution, frequency, and historical correspondence. The following table lists the MJC consonants, with phonemes not shared by all three dialects in parentheses: b
p f
m
n
r
I;>
r
(<;I)
C
d z
s
(5
0)
I \
g
S w ~
y
(¢)
lJl 17
k
q g x
h
18
COMMUNAL DIALECTS IN BAGHDAD
PHONOLOGY
(b) The phoneme Ipl occurs in all three dialects: MJC/pOsij 'veil', I(opal 'ball', Ipardal 'curtain', Iplawl 'rice dish', Ipaspasl 'he wh ispered', I(appasl 'he threshed about'; occasionally J and C have Ipi for M/b/ : JC Iqapagl 'lid' for M Iqabagj. This Ipi is, of course, found largely in foreign words where it stands for an original Ip/, sometimes also for origina l Ib/ : Ipaysekel/ 'bicycle', Ipasl and Il}asl 'bus' . In the rural dialects, Ipl is less common, being often replaced by Ib/,24 (c) The phoneme Ihl is present in all three dialects, but while the M pronomina l suffixes-/ hal and-/huml preserve it in all positions, some
( :< Ibwahdak/); M/ franisl 'lamps' (J/ ffanisl and Ifwan is I but C/ fawe-
all omorphs of the corresponding J and C suffixes are with, and others
without, Ih/, which is also dropped in a few other JC morphemes. For details, see 4.2. (d) In the three dialects, /'I is phonemically stable in positions other than* _ V: MJC/t'axxarl 'he was late', M/se'all ' he asked', Itse'linl 'you (f.s.) ask', Jhill "veil-born'. In# _ Y, however, phonetic ['] varies freely with zero, the latter being the commoner, and since . ['] is ordinaril y also absent \\(hen prefixes are added , citati on forms C'
•
of such words will be written as beginning with a vowel: MJCjaswadl 'black',. labOnal 'our father', lesmakl 'your (m.s.) name', as also lIas wadi 'the black one', IwabOnal 'and our father' , MC Isesmak/, J/asesmakl 'what's your name?'. (e) Non-syllabic [i] is assigned to IY/, syllabic [i] to Iii ; similarly, non-syllabic [u] is assigned to Iwl and syllabic [u] to lui. Contrasts, if any, are few and unstable, but any other procedure wou ld be selfdefeating, in that it would 'needlessly obscure statements of history, comparison, . and morphophonernics. Tokens of such utterances as
M/iini umartakl " ordered you' and lani wmartakl " a nd your wife' usually sound exactly a like though they may sometimes be distinct; in general, transcriptions involving initial lyKI and IwKI should be understood as [iK] - [y;K] and' [uK] - [wOK], respectively. (j) On III vs. /! /,. see 3.23 belmv. As for I ~/, cf. MJCjjaHI 'it . creaked' vs. /iazzl ' he sheared'" and JI,erf/, M/ ~uruf/, C/ zeref 'envelope'. In the labia Is, n'ote M/ fakkl 'he opened' vs. Ifakk/ , more commonly Ifaeel 'jaw', though JCjfakkl 'he opened'; MJCjl}iil}al 'father' vs. Ibabal 'his door (M), her door (JC)'; the contrast Iml vs. Iml is somewhat moot: only ImaYY/, ImaYI 'water', but both M/maral
19
nisi), M/l}l}abij/, J/ l}l}abij/, but C/ bawebijl 'slippers'. On taftim in J words of Hebrew origin, e.g. lai:)elj (a general term of opprobrium) cf. 6.42. (g) Beyond these few remarks, the Baghdadi consonantal inventory ca ll s forth no specia l comme nt as to its gene ra l nature. It now remains for us to consider in detail th e historical correspondences which constitute the essential differences between M, J, and C. The para-
graphs below are accordingly devoted to reRexes of the OA interdenta ls, of OA/ I/, OA/r/, OA/k/ , and OA/q/, respectively. 3.22 Reflexes of the OA interdentals. (a) In C, the interdenta l, 19/, 10/, and II>I have been replaced by the corresponding stops It/, Id/, and /c)1, whereas they have been retained in M and J. Thus Cjtnenl 'two', Ihadal 'th is', lal}yarJI 'white' vs. MJ/Onen/, Ihaem/, jabyaQj. However, informants indi cate that th ere is fluctuation in this respect among C speakers, with a good many of the younger generation reintroducing interdentals as a result o f schoo l and majority influe nce; two of my three main C informants used inter-
dentals themselves (at least whe n speak ing to me) but on direct questioning readily volunteered variants with stops and characterized
them as "typically Christian"; the third used stops himself, often clea rl y distinguishing C from M forms by that very feat ure. 26 (b) This is one of the few features in which J and C do not go
together; in this respect C does not even go with Mosul, and the onl y other point in the Meso potamian area so far noted as having stops for interdenta ls is L:armt1 c (Urfa province, Jewish informants); Qa rabas and Mardin have int erdenta ls, while Siirl has the unusual
(for Arabic) replacement of the interdentals by the labiovelars IfI, lvi, and Ivj.2' All other dialects of the area have interdentals. In M and J, as well as in other dialects where interdentals are preserved, a few individual item s may nevertheless be found in wh ich they are replaced
by stops: JCjjgedil M/i redil 'rat'. Jawaliqi , writing in Baghdad in the twe lfth century, observed a tendency among the common people
and /mara/ 'woman'. A feature of the three dialects is the gem ination
to change 101 to Idl in three or four words (pp. 134, 155) including the ancestor of this word Ijredi/, namely Ijuradl for Ijurao/. 3.23 Reflexes of OA/ I/. A velarized /!I is to be found in M, J, and C, bu t whe reas all three have it in the name of God (except in jflm allaj
and tafxim of a labial resulting from the loss of a following Iw/ : MJCjffadl 'heart, entrails' ( < */fwad/), MJC/ l}l}ahdakl 'by yourself'
'goodbye', from jfi'amani- ll <1hj), it is mu ch more common in M than in J or C: M/x'''1 'mother's brother', JCjxa l/, M/xa!!1 'vinegar',
21
PHONOLOGY
20
COMMUNAL D IALECTS
I~
BAGHDAD
for which J has Ig/, notably larba'al 'four', Igarbell 'sieve'.
JCfxall/, M/~ u gul l 'work', JCfsegel/, M/gabu!! 'before', JC/qabel/ and many others." The J words Imxabbal/ 'crazy' and Ibolal 'whore' may be loans from M. The M words having /!I suggest that vela rization of OAIII occurred largely in roots where it was preceded by lxi, Ig/, or Iq/. This III is typical not onl y of M but of the geletdialects as a whole and of Beduin dialects in other areas as well ; the qeltll-dialects go along with J and C in this respect, as do most other sedentary dialects. 29 . 3.24 Reflexes of OA/r/. (a) One of the most striking features
Iqaral
' he read',
(b) Among literate and se mi-literate J and C speakers, th ere is a tendency to replace Igl by /rI wherever it corresponds to the Irl of literary Alabic and/or of the M dialect. Some do this only occasionall y,
as a styli stic device, while others seem to ca rry out the restitution more
consistently. Those who do this seem to be keenly aware of the IIlI reflex of OA/rl as a blemish or defect of their dialect, a nd carry out the restitut ion of / r/ even whe n they retain all other features of their
own dialects. ln Baghdad and the other towns of Lower [raq, jgl for Irl is a sort of hallmark of no n-Musli m identity, a nd see ms to be keenl y felt as such both by those who shun it and by those who
distinguishing J and C on the o ne hand fro m M o n the other is t~ e
treatment of OAlrl, for which M has Irl whcrcas J and C ha ve (In the majority of words) 1i\I. Thus JCfgasl ' head', Ibagl ' hot', l 'yaggal 'outside', Ifegbiinl 'glad', Iqayesgabl ' he's drin king' , Iqaygidl ' he wa nts' for M/ras/, II)ar/, l lia rra/, etc. However, the /rI vs. If!,/ contrast is retained in J and C, though in different wo rd s. In 1, there IS often Irl rat her than 1M in loanwords with etymological Ir/; this holds for Hebrew wo rds, e.g. Iseferl 'scroll , sacred book' , I~a'a rl 'pam', /ressana/ 'New Year's' ; and for such words of n ~.n-Arabic ~ri gin as Iqond ral 'shoes' (Turk. klilldura), Idondermal Ice crea m (":urk. dondlrllla) , Inafarl 'person' (Pers. lliijiir), learxl 'wheel'. (Pers. carx), leaderl 'tent' (Turk. f adlr), and for loans from modern hteraryArablc such as /ja ridaj 'newspaper', JsayyaraJ 'ca r'. However, th ere IS ~o~e degree of fluct uat ion jn such cases: some info rma nts ha ve /cadegj for 'tent', a ll seem to have Itayyagal for 'airplane'. Moreo ve r, so me words with Irl are not obviously loanwords: Isarql 'east', Igarbl 'west', Iqecr.yi/ ' Sabbath ' oil lamp', ./qerrabal .'fl ask' for whl~h , incidentally, only the plural /qegagibl is give n by IIlfo rmants; Igrabl
doggedl y reta in it , reflecting di ve rge nt attitudes toward the maj ority
community. As to fa lse restitution, i.e. Irl fo r etymo logical lilI, I have found very little of it, which may be due to the fact that my informants are all litera te , I have noted no instances amo ng J speakers, a nd
heard onl y one (and been told of another one) a mong C speakers, both in ~vo rds of non-Arabic origin whose etymo logica l pron unci-
ation a nd spelling are not likely to be known. One C speaker regularly says Iqasural 'spoon' for Iqasugal or Iqasollal (cf. Turk. ka~l k, Pers. qiissoq); anot her informant te lls me that though she herself has Igl in this word , she has heard it with Irl from other C speakers; she also volunteered the information that Iqapaill ' lid ' (cf. Turk. k apak) is sometimes pronounced Iqaparl by some C speakers. On M/gemarl for l Cfqemagj 'cream' (Turk. kaymak , Pers. qeymaq), see the lexicon. (C) A major co nsequence of this shift is the fact tlu~t, in J and C,
Igl
now occurs in positions strange to it in OA or M, notably in
contact wi th
'crow',
Iq/, lxi, and Ib/.These contacts have often resulted in arti-
culatory problems a nd further phoneti c c hange via to ta l o r pa rtia l
I n the last example a nd in Igarbl, the proxim it y of ety mological cannot be a suffic ie nt reason for the retentio n of Irl, cf. Ii,ellbill ' sieve', pI. Igegabil/ . Etymologica ll y related roots sometimes spht into doublets, the one reta ining Irl and the other showing Ig/ : j'aga bll 'Arabic', j' rubil 'a villager'; Ifaggl ' he poured , served' but Ifar~1 'he threw' ; Igayyagl 'he changed clothes' but Igayyarl 'he changed: I ~jagal 'stone' bUl /bajarj 'jewer. 30 There is an a nalogolls s lt~ atlO n In C but the distr ibution of Irl and Igj is not the same. Thus wh ile C has
assimilation. Thus M1Cflaxi 'another' has a feminine harking back to OAf' al'uxra/, which takes either the shortened form M/luxl JCjlex/, or the lo'nger form M/luxra/, l Cj1exxi/, with - Ixx/- fo r an expected - Ixill - ; similarly l Cfaxxasl 'dumb', fern. l lxegsa/, Cfxagsaj. There
Ii!,!
Iri in
those of the words
is a similar total ass imila tio n next to
Iq/, so that in
J the ve rb 'to read'
has lilI in a few forms but mostl y Iqql < Iqg/, e.g. Iqqetul 'I've read; Iqadaqqal ' I' m reading', Iqqal 'read!' and so on ; the wo rd for ' near is Jiqqib/, comparative laqqa bj. InC, the verb 'to read' has Irl and the word 'near' has lal between Iql and IIl/, hence no contact, but
the loans from Turkish. Persian, a nd modern literary Arab i,c
mentioned above Uqandara/, learex/, Idonde rma/, and so rorth). It has Igl in fjogabi l 'sock' (J /j urabj). it has Irl in a good many roo ts
my C in form ants give unassim il atcd for ms such as laqgabl 'neare r',
\
22
.r,
COMMUNAL DIALECTS IN BAGHDAD
PHONOLOGY
j'aqgabil 'scorpion', perhaps because elicited. In the above J examples, assimilation is complete, i.e. Igl is not restored in more careful enunciation, whereas such words as J/maqqal 'sauce', J/waqqal 'piece of paper' do have more deliberate variants Imagqal and Iwagqa/ . On the other hand, no assimilation takes place in J/qgunl 'horns', J/begq'nl 'pitchers'. In the last example, and in such words as JCffeg~iinl ' glad', there is practically no audible velar constriction in the lill, which gives an auditory effect very close to that of [w). Instead of
speech peculiarity of individuals in the Syrian and Egyptian area, but as a stable phonetic feature of given dialects, it has been reported only for some North African areas and, less clea rly, for the Arabic dia lects of Central Asia. Ph. M'lf~ais mentions that many speakers of Djidjelli (Eastern Kaby lia, Alge ria) have a uvular spirant for Irl, and says that this is restricted to the town on ly (as distinct from the surrou nding countryside) and Occurs nowhere else in Eastern Kabylia. He indicates it is found in other old urban centers of the Maghreb
assimilation to the following consonant, there is thus a tendency
(Tunis, Constant ine. Algiers, Fez, Mek nes), considers it "une maladie des vi lles" though it also occurs "dans Ie Rif". 33 Cohen reports a uvular [R] for Muslim Algiers vs. an apical tr ill for Jewish Algiers
toward vocalization of lill, which can be noticed even more fully in other instances. In what ought to be phonemically J/lagqal ' frig ht' (cf. J/n!ailaql 'became frightened '), one speaker has II'qa/, i.e. the homophone of neo-classical I!iiqal 'energy', but the informant den ies homophony. On the other hand J/lbO~i/, Cfmbe~al 'yesterday' have lost all traces of Igl even in the speakers' reports] I (d) The same Irl > lill shift is to be found in Mosul (among all communities, but not among all speakers") and is reported for Tekrit. It does not occur in the other qellu-dialects thus far examined. As in J and C, Mosul has Irl in some loanwords, and in some Arabic words, e.g. Iqaral (but my Jewish Mosul informant has Iqagaf) and assimilations similar to those mentioned for J and C, e.g. Imbe!)al 'yesterday', j'aqqabil 'scorpion'. A Mosul peculiarity is the 101 of loba'al 'four', lob';nl 'forty', lob'iil 'Wednesday', no do ubt from the chronological sequence [arb] > [agb] > [awb] > [ob]. All my Mosul informants agree on these forms in 10/, but one of them uses only forms in ,ar/-and considers the 101 forms very "old-fash ioned." One is tempted to speculate that the presence of Irl rather than lill in these words in C may be due to a similar sort of resti tution, i.c. that C
originally had some such form as loba'al which, because of its unusual peculiarity in the Baghdadi milieu (into which it may have been in:tported from the north , as so many other C features seem to have
been) was changed back to lar/ ; this would explain the presence of Irl rather than If!,/ in such ,Yords. That some such explanation might be valid is indicated by another C fo rm with Irl rather than IP'!, namely Igarbell 'sieve', for which Mosul has lilubel/, with the vowel probably resulting from a chronological sequence [irb] > [,gb] > [,wb] > [lib], though the lal of the C form is puzzling (cf. also this word in the lexicon, Chap. 6). (e) The replacemcnt of Irl by Igl or the like is not unknown as a
23
(i.e. the reverse of the Baghdad situation) though he adds that there are individua ls in either community who ex hi bit the reverse phenomenon. 34 Colin li sts as one of the features of Moroccan urban
dialects the fact that Irl "often" sounds like the French [R]." The latter implies, and the first two state explicitly, that this [R] is not merged with [g] into a single phoneme. On the Irl of Central Asian Arabic, Tsereteli says the following: "Vibrant of back articulation: is sometimes pronounced as uvu lar r, with which it interchanges" ;36
Vinnikov is silent on the phonetic shape of Irl, a nd at any rate the vs. Igl contrast is, presumably, maintained . (j) The question that naturally comes to mi nd is whether this Irl > II;I sh ift is, as seems to be the case for other features of the qellu-dialects, a hallmark of Iraqi sedentary dialects. It is not present in any of the gelel-dialects, but only some of the qellu-dia lects do have it, namely Mosul, J, and C. The dialects of old urba n populations may
Ir/
be presumed to hark back to dialects that were spoken in the area before those of nomadic or recently sedentarized po pulations; exam ination of older sources may therefore be of interest. We find
the replacement of / fl by lill first mentioned in the ninth century by al-Jabi? of Basra (d. 864), though only as a luega (individual speech peculiarity). He says that of all the luegdl " it is the least ugly and the most commonly to be found among respected an d prominent personalities, as well as among individua ls of good speech and learn in g. "37 The phrasing suggests tha t, while not yet a dialect feature. thi s shift might at that lime have been on the way to becoming o ne. or co nsiderable interest, thou gh less explicit, is the description of the pronun-
ciation of the Hebrew Irl by Sa'adya Ga'on (d.940).38 Sa' adya was of Egyptian birth but spent the latter part of his life in Iraq; in commenting
I
24
COMMUNAL DIALECTS IN BAGHDAD
PHONOLOGY
on an obscure passage of the Sefer Yesira, he says that the Hebrew r belongs with b,g, eI, k,p, and 1, since it can be pronounced in two ways, one soft (rax) or weak (/Ialas) and the other hard (qase) or strong (gibbor) . What is mo re to the point, the r is pronounced in these two ways by Old Testament readers of the Tiberian school, whereas those of Iraq pronounce it in these two ways "in th eir speech , but not in their reading." He states that they refer to one variety of r by the term makrux and to the other by th e ter m gayr makn/x. It is tempting to believe (as do Gumpertz, Mansour, and Morag) th at this refers to a situation exactly parallel to that found in J today: they had two ways of saying r in their [Arabic] speech, namely one Irl and one 1M reflex for OA/r/, but only one, namely Ir/, in thei r reading of the Hebrew Scriptures, and the terms used to designate them (unknown to the Arabic dictionaries) so und as thou gh th ey might mean so mething like "uvular" and " non-u vul ar," res pectively (see note 37). It is surprising that if this had been the case, Sa'adya should say nothing of the similarity of one of these pronunciations with Arabic Ii!.! or the le nis variant of Hebre w g; the proposed interpretation seems plausible enough, but others a re not excluded. It must be added that the fact that Sa'adya speaks of the Jews onl y does not necessa ril y imply that thi s pronunciation was to be found among Jews to the exclus ion of Muslims and Christians; hi s focu s 0( att ention is Old Testament reading, not comparative dialecto logy. Incidenta ll y, it is not at all certain that the words " in their speech" refer to Arabic. A century later, Ibn Sinii (d . I 037) specifically includes a "/g/-like Ir/" among sounds Jlo t occurring in Arabic, and while he discusses other dialectal or ind ividua l peculiarities, is silent on Irl > /gj.39 Unlike Jiihi ~ and Sa'adya, however, Ibn Sin a did not reside in Iraq. Thus from the older sources that have thu s fa r come to my attention, the most th at can be sa id is that there may be traces of such a shift in the older sedentary dialects of Iraq. It certainly cannot be said to have been a very widespread shift , if one may judge from it s abse nce in some of the prese nt-day qelill-d ialects, notabl y the A natolian varieties. The evidence from the Central Asian dialects is moot and as yet hard to interpret. Another piece of negat ive ev idence is the treatment of Arabic loanwords in the Neo-Aramaic dialects of Upper Iraq and Kurdistan: a ll such words with etymological Irl have Ir/, implying Irl and not Igl in th e so urce dial ects' o For what it is worth, I note in pass ing that amo ng the Iraqis I have known, no less th an six (th ree
Muslims, two Jews, and one Christian, all of Baghdad) were unabl e to produce the ap ica l trill that is normal in their dialects, a nd rep laced it by 1£/ or a similar spirant. In other words, they all had a true lufJga for which, incidentally, they had no name, nor have I been able to find one in any Iraqi dialect, though in other dialects, where it may be less comm on, names for it do exist. The re lation ship between this luega, past or prese nt, and the so und shift Irl > 11;1- if any _ remains to be probed.'! 3.25 Reflexes o f OA/kf. (a) In J and C, the regular ren ex o f OA/kl is Ikl while in M, what must have been front varia nts and back variants of OA/kl have split into the two phonemes lei and Ik/. Thus JCf kiinl ' he was', M/ciin/, but MJCfykun/ ; JCfkbil;1 ' big', M/cebir/, JCfk biii;l, 'big (pl.)', M/ki?arl (probably from OA/kubiir/, cf. 3.3 below) ; J/kalb/, C/kalebl 'dog', M/ealeb/ , pI. J/klib/, C/kl eb/, M/eliib/ ; JCfke1i1 'all', M/ kullf. There is thus, in M, morphophonemic alternation of lei and Ikl in certain roots, e.g. those for ' to be' and 'big'; that th is alternat ion is not sub· phonemic is beyond doubt , cf. M/bikl 'in you (m.s.)', Ibiel 'in you (f.s.), Ibiieerl 'tomorrow' vs. Ibiikerl 'a virgin '. [n addition, this al ternation is used fo r stylisti c purposes, in that many word s have, in M, two poss ible fo rms, a "plain colloquial" o ne in lei and a "mildl y formal " or "sem i-li terary" one in Ik/, e.g. leiinl vs. Ikiinl 'he was', leene!1 vs. Ikenetl ·1 was, you were' lijiel 'that one (f.)' vs. lijik/ ; in so me cases this in vo lves other changes in the direction of Classical Arabic, e.g. leebirl vs. Ikabirl with the lal restored, or Ikalbl (esp. in the imprecation Ikalb ebn elkalbl 'dog son of a dog') wit h th e anaptyctic lei of lealebl re moved as wel1. 42 A phoneme lei also occurs in J and C, usually in loa nwo rds, which are common to MJC, e.g. IM!f 'desert, co untryside" (Pers. cII/), leiiyl 'tea' (Pe r~. cay), I~uel 'fau lt, guilt' (Tu rk. SIII'). As refle x of OA/kl it occurs in J and C in what must be loanwords from M or a similar dialect, e.g. leaHabl 'he clung', probably fro m k-I-b ; it also occurs in a few Arabic words where it is not a reflex of / k/ , e.g. leaqlabl 'tumbled ove r'. In cases where M has Ikl ra ther than an expected lei , i.e. where the OA form is likel y to have had a fro nt variant, borrowing from Classica l ca n usually be ass umed : Iktabj" ' book' (and throughout this root). . (b) The situation in Mis, ·in this respect, practicall y identica l with that obtaimng in th e rural gelel-dialects of Iraq, a nd with so me of the nomadic dia lects of the a rea, though so me Beduin dialects (t he
25
26
COMMUNA L DIALECTS IN BAGHDAD
large r, mo re full y nomadic tribes o f ca mel herders, a pparently of Ce nt ral Arabia n ori gin) have a dental affrica te /c/ as o ne of the re fl exes o f OA/ k/, the ot her being / k/ and the alternat ion simila r to t hat of M43 The I a nd C situati o n is exactl y p a ralleled by that of Mosul a nd th e Ana to lia n di a lects; the 'Ana d ia lect has p reserved / k/ by a nd large, but, t hough o f geltu type, does show so me words with /e/, a nd thu s exhibits in this respect a situa ti on th at is intermediate betwee n t he gelet and qe/tu types: / kbir/ ' big', /~ a ka/ /ye~ k i/ 'to spea k', but / baea/ / ye bei/ ' to wee p', /efm /, /ykun/ ' to be', l ealeb/ 'dog'. Broadenin g t he circle of co mp ari so n we note that a Sim il a r affricatio n o f OA/ k/ in certain environments is characteristic of the dialects furth er to the so uth (Kuwayt, Persia n G ulf) and t he rural populat io ns east of th e Jorda n, on the fri nges of th e Syri a n desert ; whereas non-affricat io n is characteristic of the sedentary populatIO n of t he rest of Greate r Syri a a nd, west of t he Jord a n , eve n of a good 44 many semi-nomad ic or rece ntl y sedentarized po pulalio ns. (c) Affricati on of Ikl does not see m to have been pa rticula rly co mmon in t he o lde r A ra bic dia lects." Abo ut all we fi nd are reports of kaskasa and kaikasa in certain tribes, i. e. the replacement o f k ci/ by si" or Sill a formulatio n that might possibly suggest some sort of affrication; at any rate, this replacement occurs o nly in the pron. suff. - Ikil df t he 2nd pers. fern. sing 4 6 Suggestions t hat it mi ght have 47 taken place in other cases as we ll are few, vague, and late. There is one early sta tement h inti ng a t possible affrication . Sibaway hi (d . 79 1) lists a mo ng co nso nants heard amo ng the Ara bs but lacki ng in the script "a ka/ that is between jim and k. iif,"48 but this is not explai ned further and, so fa r as [ know, isolated . To sum up, evide nce fro m present distribu tion as we ll as fro m older sources points. to. no~ affricat ion in o ld sedenta ry populations vs . co nd itio ned ~ffrIcatlOn tn popu l a~ions descended fro m, or in fluenced by, No rth A rabian no mads; th us J a nd C belo ng t o th e former, and M belo ngs to the latter, a fact wh ich I think th e ensuing sectio ns will te nd to co nfi rm. 3.26 Reflexes of OA/qj. (a) In I a nd C, t he reflex of OA/ql is, in practica lly eve ry insta nce, Iq/ ; in M, it is usuall y /g/, t ho ugh Iq/ is com m o n a nd there a re seve ra l insta nces of a Ikl reflex a nd a hi reflex. Mo~t hi gh-frequency roots a nd words ex hibit the typical co rres po ndence of JC/ql for M/g/ : JCfqal /, M/g.11 'to sa y'; JCfqam/, M/ga ml 'to rise'; J/qa lb/, C/qaleb/, M/ga!u ~1 ' heart', JC/q a ma£/, M/gumarl 'moon' ; JC/b'q/, M/bagl 'to steal'; 118qil/, Cftaqil/,
PHONOLOGY
27
M/8egil/ ' heavy ', etc. While IC/gl < OA/ql is relati ve ly ra re (JC / gam az/, M/ guma zl 'to jump') a nd presumably res tric ted to loa ns a I~/ reflex, in M is m o;e comm o n: / buqal 'to stay', /qubal/ ' to agree': Iqasmarl to in veigle, Iqa nn a'i ' to co nvince', /qeral ' to rea d ', etc. In M , th ere are three items wi th a Ik/ reflex: Iwa ketl ' time', / keta l/ ' to ki ll', I kufaxl ' to sla p'" a nd a numbe r o f insta nces of a Ijl reflex: IrnJ/ ' friend , lover', /jede rl 'pot' (fo r which J/geder/, cf. 6.5 below), II! a rijiyal :co nfl agra tio n' (also used in C, tho ugh I/hgiq if) ; so me form s In IJI ha ve equi va lents or doublets in Igl or Iq/ : Isarjil 'easte rn ' in Ihawa sa rjil 'east wi nd ', MJ C/ biib essa rj il 'the Eas t Gate' (name o f qua rter,) but o th erwise MIC I;a rq il; /jedd a ml an d Igedda ml ' befo re' (JCfqeddam / ; cf. a lso / mgaddi/-/ mj add ij 'begga r' in 6.5 below). In such cases, my M informants tend to co nsider the fo rm in Ijl as " pro vincia l" a nd to prefer e.g. Igeddiiml to Ijedd a m/ , Ige rib/ to IJeribl ' nea r', Iqaryal to Ijaryal 'village'so I,adiql fo r /,ad ijl in t heir own speech; they thu s use fo rm s in jj/ in a few words o nly. and speakers wi th add itio na l forms in Ijj wo uld, indeed, seem to show the i n fl~e nce of ru ra l dialects, wh e re jjj refl exes are comm on (see nex t sectIOn) . An alogously to the stylisti c va riati on descri bed above betwee n Ikl a nd lei , M spea kers have so me " m ildly fo rm al" fo rm s in Iql for " plain coll oquial" fo rm s in Igl o r Ik/ : Iqaribl for Igeribl 'near', Iwaqtl fo r Iwaketl 'time', Iqawil for Iguwi/ 's trong', etc. In all threedialects, Iql a nd Igl also Occ ur in loa nwo rds: MJ C / pur taqiil/ 'orange', MJCfg!ii,1 'glass'. (6) The qel/II-dialects as a who le go alo ng with J a nd C in having a Iql reflex of OA/q/. Here aga in, 'An a has /q/ in most cases but /gl in some : Iqall ' to say', 18qill ' heavy', /qaml ' to ri se', Iwaqetl ' t ime', Iqata l/ ' to. kill ', but Iga hwal 'co lfee', /grayyebl ' nea r', I bag ral 'cow'; a Simil a r situa tio n o btai ns in Hi t. In Carmue we have fo r OA/ql a / kl with a ll ophones [q] [k] and [k 'J that pa irs off wit h a shi ft of OA/kl to lei in a ll positions, pa ra lleling the situat ion in the sedenta ry dialec ts dubbed " par lers S," by Ca nt inea u, suc h as that of Sukh ne and Centra l Pal es tin e (Ca ntin ea u, 1939; 1956, p. 126). As for t he situa ti o n in M, it clo sely resembl es tha t fo und in th e other gelet-d ial ects , th o ugh there a re (as reg~rds re flexes of OA/q/) two d iffere nces between th e urban gelet-di alec ts (includ ing M) and th e rura l ones. The first is that som.c o f th e rural dialects have few refl exes in 1'1 /, so me a ppare nt ly haVin g no ne ; th e second is that all rll ra l gelet-d ialec ls examined
28
COM~'I UNAL D IALECTS IN B AG I~DAD
thus far 51 have a conditi oned affrica ti on of OA/q/ quite similar to that of OA/k/. What must have bee n back va riants of OA/ql have yielded Ig/, whereas fron t va ri ants yielded Iii· with concom ita nt morphophonem ic alterna ti ons: Irfii l 'compani on', pI. Irefgan / ; Isediil ' fr iend', pI. Isedga n/ ; j'aseil 'in love' , pI. /, ussag/ : IWed l 'sillin g', Ige'adl 'he sat'. My info rm ant from the Musayya b district indicated that upon returning to his village from his studies in the city, he was exposed to rid icule fo r pulling on 11)l:)eril (urban) airs because he used Iga'edl instead of IWed/. In his speech as we ll as in that of Kweres, this iii is ide ntica l with the reflex of OAlil, i. e. both are affr icates; in some areas th e two do not merge. as OAfj! yield s
[1.] (reco rdin g from Amara reg ion) or [y] (recordi ng from Khuzistan , Kuwayt), whi le the Iii < OA/ql is affr icated. In the few wo rds with iii OA/ql borrowed by M, the morp hopho nemic altern at ion does not take place: M/rfii/, pI. Irefian/. The affricati on of OA/ql is a hallmark of the co untrys ide ; it is al so character ist ic of the nomad dialects of the Iraq i and Syrian desert a reas ,52 with a dental affrica tion to [dz] differentiating, as in the case of Icl < Ik/, the full y no mad ic camel
-,
•
herde rs of Arabian ori gi n from the sheep-and- goat Bed uins who have, li ke rural Lower Iraq, Iii a nd Ie/. It is noteworthy that the sedentar y
a nd sedentari zed populations east of th e Jo rd a n who have lei < Ikl do fl ot have any affrication of OAjq/ . showing a similarity with M wh ich is probably not fortuitous. In those instances where M has preserved (o r re in troduced) Iq/, the rural dialects that lack Iql common ly have either th ei r nor ma l Ig/-/il (ige ral 'to read', /iese ml 'part') or, no doubt in an attempt to imitate the urban or Classical Iq/, they have 1M: Igubgiib/ ' wooden sli pper' , M/qubqab/ : Igas mar/ 'he inveigled', M/qasmar/ ; Iga nn a' i 'he convinced', M/qanna'l, etc. 53 To sum up, M has reflexes of OA/ql that are typical of sOI';e Beduin or Beduinized populations, tho ugh the partial preservati o n of Iql and the abse nce of affricatio n keeps M apart from th e Iraqi village and nomad di alects; whereas J and C go along with the typicall y urban Iql of the Upper Iraqi , A natolian, a nd Syrian areas and, for that matter, with the many sedentary dialects throughout the Arab world that have a/'l or a Ikl that is but o ne remove fro m IqJ. (c) The present-day distributio n of re flexes of OA/ql throughout th e A rabic-speaking wo rld presents a striking dichotomy : most sed e l~tary populations have a vo iceless reflex and all non-sedentary
PHONOLOGY
29
vo!ced :eflex. 54 111 considedng the historical development l ead l~ g to thiS Situat IOn, two question s must be kept distinct: ( I) the quest ion as to the sedentary vs. nomadic d iffere ntiation , and , (2) the question as to the quality of th e reflex of OA/ql in each gro upin g, i.e. w l~ether sedentanes always had a vo iceless and nomads always a vO iced reflex. From the eleventh century onward,55 Arab sources see m to agree as to. t~ e existence or. a sedentary vs. nomad differentiation a~ t.? pronunciatIOn of OA /q/. The earliest to me ntion it may be Ibn SIna (d. I037! \~!1O says that the "Arabs" (i.e. Bedu ins) have a kcil' like so u~d. fo r qaf In the fourteenth century this is made even more e~p hclt by Ibn Xal.dOn ,57 who expounds at length on the social sigmfi c~ nce of the difference (th.e Bed uin pronu nciation has prestige and ~ s used by ~a~y sedentanes in imitat ion of them), on its geographical spread (I t IS fo und everywhere, both in the east and in th e west or the Arab~c. area) and o n its phonetic quality (i ndicating a more rorw~ rd position for the Bed uin qnf). However, neither aut hor says anything that ca n be interpreted as a vo iced-voiceless distinction. In t~~ case or Ibn Sina, such an interpretation would seem to be expliCitl y precluded, si nce he un equi vocally describes a voiced [g] later .on 58 stating that it is found in Persian, that it is not ro und in ArabiC, and stands in the sa me relation to ka! as zny docs to sill. If we s u~ ~osc) as see ~ s likel ~, that the earliest description~ or OA/ q/ as ma)J~ur, e.g. by Slbawayl1l,59 mean that it was voiced, we may inrer ~hat thiS .o~der [G] underwent a number or changes. The earliest was Its de-VO ICIng to [q] and too k place among th e descendants of the first waves of Arab settlers, say in the eighth or ninth century; among some of these sa me populations it then underwent ru rther changes notably t o [']. Another change may have take n place around th e sa m~ time inSide Arabia, and was carried by later waves of migrants (whose descendants are often still no madic), namely its fronting to [g]. In stIil later waves, there was a further change of this [g] resulting in its .a ffr~cat~ o n to U] In certain positions, and in the latest wave we have affrIcatlOn to [dz]. No te th at this schema goes we ll with the developments noted for OA/ k/ : [0] de-voiced to [q], - no concom itant change In OA/ k/,. wh ich is already voiceless; [0] fro n:ed to [g] concomita nt fron tIng of [k], wi th a ne w phoneme lei in some dia lects; further frontII1 g and affri eatio n yielding iii for front allophones, always With concom itant fef, and simil ar.1 y ror dental affrication to [dz] and [cj. Th is may be a bstracted in the fo llowing table, with pop~I Jat i o ns tI.
30
PHONOLOGY
COMM UNAL DIALECTS I N BAGHDAD
"sedentary type" examples on the left , and nomadic type examples on the ri ght. OA [Ga'ad] [Gii' id] [kubr] [kabir] (I) de-vo icing: [qa'ad] [qii'ad] [kubr] [k a bir] (2) further changes: (a) ['a'ad] ['ii'ad] [kubr] [kabir] (b) [ka'ad] [kii'ad] [cubr] [cbir]
,.
( I ') fronting : [ga'ad] [gii'ad] [kubr] [k'abir] or [cabir] (2') fro ntin g: [ga'ad] [jii'ad] [kub r] [eabi r]
31
torical divergences in the treatment of OA vowels to be discussed below. In J, /iI and ful occur almost exc lusivel y in un stressed o pe n syllab les whe reas leI IS VIrtuall y rest ric ted to closed or stressed ope n syllab les; lal has greater freedo m of occurrence. Much th e sa me applies to C, except that IiI and luI are much rarer than in J, where they often result from a shortenin g of un stressed long vowe l th at does not take place in C. In M, o nly /ii, which is qu ite rare, is restricted to unstressed open sylla bles, whereas luI, leI, a nd lal are not strongly restricted. For some M speakers, the contrast iii vs . leI may not eXist, so that there may be speakers with on ly three short vowel
(3') fronting: [ga'ad] [dzii'ad] [kubr] [cabir]
3.3 VOWELS. 3. 31 Overal1 view. (a) As in the case of the consonants, the inventories o r vowe l phoncmes are very nearly the same . for M, J, and C. However, there are great differences in distribution and in historical correspondence. As in other Arabic dialects, it seems best to envisage the vowe ls as arranged into two sub-systems o f short and long vowe ls respectively, thus: Short Vowels
Long Vowels
u
u
e a
(b) T he sy mbol leI was chosen to represent a phoneme with allophones clustering abou t a mid, central [a] and is used for convenience only in the phonemic notation; a phonemic symbolj zatio n 101 would admittedly have been preferable. At any rate the symbol leI should in no way be taken to imply that t his phoneme is the " short equivalent" of leI. Neither leI nor 161 have co rrespondingly short phonemic parall els, though they each have shorter a nd longer phonetic varia nts; the final unst ressed [E] of M is assigned to lal, the (rare) final unstressed [0] of MJC is assigned to 161; for details and justification , see below. (e) The short vowels exhibit striking differences in frequency and distribution, most noticeably as between M on the one hand and JC on the other; most of these differences result directly from the his-
phonemes. In fact, in view of the distribution restricli o nsj ust outlined true minimal pairs contrasting Ii! and lei are hard to find even fa; those s ~eakers for whom the distinction is clear and consistent (and hence, 111 ou r terms, pho nemic): J/limal and J/lemal are sy non ymo us ('until ') b~t auditorily not interchan ged ; the morpheme I Ii/- as in e.g. MJC/lihassal ' until now', M/ligeddiim/, JC/liqeddiiml 'i n advance' IS one of the few instances (o ther than reflexes of long vowels in unst ressed syllables) in which /iI does not seem to result from C lassical ~ nfluence. In other instances such influence seems evident: it is fl agrant In ~he com~~n MJC/Ji 'an/ 'because', even more so in scmi-literary van ants (/kltabl for Iktiibl ' book' and the li ke), and is probably prese nt also in verbal nouns of Forms VII and VIlI, e.g. lextilMI 'dIfference ', lentiqiil/ ' transfer'. The contrast luI vs. leI is more sohdl y estabhshed for M, with many pairs such as I~ebbl 'water j a r' vs. I ~ubbl 'love', Igull al 'tel1 (m. s.) him' vs. Igellal 'dearth ', t hough there IS a good deal of neutra lization (cf. 3.32 a nd note 60); in C and J the contrast IS weaker, and / u/ occurs only in loans from M and Classical (in C) and in such loans as well as in forms where it reflects an older_long vowel in J: MJC/mudir/,director', MJC/mumken/' possible' , J/dulabl 'l1Iche' (cf. MC/duliib/). Unstressed final /iI and luI, quite common 111 MJC, ca n contrast neither with IiI o r /UI nor with leI, whIch do not occur in this position: MJc/jiiblil 'he brough t to me' Ijibul ' bring (imptv. p l.)'. ' (d) Al1 op honic distribution in the short vowels is, unl ike what has been said for the consonants, not exactl y the sa me for M, J and C. In M, even for those speakers who have a contrast leI-IiI, the leI has allophones clustering about a higher, more forwa rd position than the leI of JC, wh ic h tends towa rd s a lowe r more retrac ted position and towards greater rounding, especially ' near emphatics.
32
COMM UNAL DIALECTS IN BAGHDAD
PHONOLOGY
Thus to my ears (and, judging from informant reaction, to some of the speakers themselves) lC/el often so unds like M/ul and vice versa: M/ i;lU\ull and lCj I;>e\ell ' bottle' a re both phonetically [I;>~\~ I], and Ilbegdiidl 'Baghdad ' , for wh ich both M a nd C have Ibagdad/, sounds like Ibegdiidl to C speakers but like Ibugdiidl to M speakers and, frank ly, to me, though of course [ defer to the native 1 interpretation in my phonemicization. 6 0 Another major difference in allophone distribution is that obtaining fo r lal, where the differentiation is between 1 on the one hand and MC on the other. All three dialects have allophones spreading from mid front ttl through low front [a], low central [a] and low back [A]. In l , allophone distribution is quite similar to that found in most Arabic dialects, except perhaps for the more frequent occurrence of [tl. and is as follows: (i) near front consonants, ttl or [a] : [ji mt l], Ua mal] 'camei' (ii) [a] or [A] near back consonants, [ax], ' brother' [aku] or [~ku] 'there is' (but [a,iiku] 'what is there ?', because of the [s])
,.
,.
(iii) [A] near emphatics a nd Iq/ : [WAqqA] ' leaf', [A!!A] ' God'. The difference between this distribution and that found in M and C concerns the allophones of lal in final , unstressed syllables. In absolute final position, MC have ttl or [a] even if immediately preceded by an emphatic, though not if preceded by [h] or ['], in which case there is [a]; the same holds in final closed sylla bles unless the c10sirig consonant is an emphatic; thus :
\
\
1
C
M
[01.103] [A!!A] [i,lal.l!A]
[ol.1 nt] [A!!E]
[ahnt ] [A!!E]
[ 1)a~\E]
[h U~ \ E]
[I;>A~AI]
[I;>A~iil]
[ I;>u~a l]
[A 9YA(\] [sa'a]
[A I;>YAQ] [sii'a]
[A!;>YA(\] [sii'a]
In addi tion, the allophone of word-final lal preceded by Ihl depends, for M and C, on whether the Ihl is preceded by lal or not ; if preceded by lal, it may be either tt l or [aJ; if preceded by any other vowel, only ttl, thus:
1
C
[waddaha] or [waddiiha]
33
M
[wiiddaha] [wiiddiiht ]
[waddaha] or [wiiddiih t ]
[wiidd6ht]
[wiidd6ht]
It should be noted that this MC[t ] in word-final position is essentially a pausal phenomenon ; in close juncture with a following word , it behaves as in medial position, thus M[clnnaj 'we were', but [elll nahnilkt] 'we we re there'. While the behavior of a ll 1 and M informants is consistent with the allophonic distribution outlined
here, at least one of my C informants seems to fluctuate between the above C distribution and one more li ke 1. It may be that my C speakers showed M in fluence in this m a~er, and that more co nservative C speakers would have a pattern more like that of l , which is also that of Mosul , a dialect which C follows closely on many points. (e) The long vowels are less varied a llophonica lly, but a characteristic of 1 is that the vowels of unstressed sy ll ables are regu larly shorter than the corresponding vowels of MC. Phonemicall y, this results in the following: (i) Unstressed lei and 161 either remain "the sa me" (though phonetically shorter), or are replaced by Iii and lui : J/seme'l 'having heard', Iseme'a I 'having heard her', but Ilbetl ' house', Ibit.nl 'two houses' ; l/d6xanl ' he felt dizzy', Id6xa ntul 'I felt dizzy', but 1/1.16, 'courtyard', Ihusenl 'two courtyards'; for details, see 3.37b below. (ii) lla/, 11/, and lui are replaced by lal, Iii, and lui : I j'ayanl ' he saw', but j'ayantul 'I saw'; Iqqlbl 'near', pI. Iqqibln/; I/niimil 'lemons', nom. un . Inumiyyif. Thus MC/a/, III, and liil in unstressed syllables usually correspond to J/a/, /ii, and lui unless other historical changes have altered the corre~pondence even further: M/jiiriib/, Iljurabl 'sock' ; M/mlzan/, C Imlzenl, Ilmizlnl ' scale'; MCjcakiicl l leakiicl ' hammer'. Unstressed fin al vowels, except for the specia l case of lal vs. lal discussed in (j) below, show no qua ntity contrast in M. J, or C, and are represented by the short vowel sym bols, except . in the case of 16/, since our phonemic inventory includes no lof. This 161 is quite rare in Ml, e.g. MICjrady61 'radio' , somewhat less so in C, e. g. C/qar61 'they read', Iban61 'they built'; final unstressed lei does not occur.
COMMUNAL DIALECTS IN BAGHDAD
PHONOLOGY
(iii) Long [i] and [u] do, ho weve r, occur in unstressed syllables in J as a result of certain historical and mor phophonemic processes to be described below (3.33); these are equivalent to, a nd seem to vary freely with , the sequences [iy] and [uw] respectively: Isayyabl ' he abandoned', [siyyaba] or [siyaba] 'he aba ndoned her'; Imsayyebl 'having abandoned ', pI. [msibin], [msiybin]; /iawwazl ' he married', [juwwaza] or [juwaza] 'he married her'; Imjawwazl 'married', pI. [mjuzin], [mjuwzin]. In genera l, [i] and [iy] do not cont rast, nor do [u] and [uw] (cf. J/hi yyil 'she', Ihuwwil 'he', MC/ hiyya/, Ihuwwal, which could just as well be written with long vowels and a slDgle
One inf~ rm ant vol unteered the information that Ikallal 'sugarloaf' (I. e. [ka lle]) IS one thlllg whereas Ikallal 'certainly not' (in semiliterary or mock literary utterances), i.e. [k~ II Aj, is something else ~ga lfl ; In thiS case the distinction was made by a very clear lengthenIIlg of the final [Aj of the lalle r term as compared to the short front [e] of the former, bllt the quality co ntrast seems the more stable and quantity in this positio n does not seem to be distinctive pc;
34
se. Never theless, th e interpretatio n of the contrast as phonemically
lal
vs. lal in such cases seems the most reaso nable both fro m the point of view of notational simplicity and from the speakers' own interpretation. This would thus represent a c1carcut case of phonemic overlapping, with [A] assigned to / a/ in some environments and to lal in others: Fo r the speakers, the final vowel of [HimmAj, for example, IS qUIte explicitl y lhe "same" as lhat of [mal] a nd not the same. as ~hat of [QAt]. A striking confirmation of this native speaker
semi-vowel) and a finalized phonemic notati on might retain only
liyl and luwl to
the excl usion of Iii and lui ; at any rate this certainly seems indicated for the above verba l a nd participial forms; cf. also J/yhudl 'Jews', Ili yhud/, i.e. [Iihud], 'the Jews' ; J/wlanil 'first', Ilu wliin il i. e. [lulani] ' the fi rst'. On non-contrast wi th [ay] and [awl, see 3.37b. All three dialects have the typically Baghdadi mid allophones .[e] and [5] as norms for lei and lill and the norm for lal is, near front conso nants, more central than fro nt ; M speakers often have ~.
•
a back -variant in this position , whic h seems less true of J and C
speakers. All three a lso have the typical diphthongization of lei after fro nt consonants, i.e. a composite SQund with [i]-like on-glide that gives the effect now of a rising diphthong ri el, now of a falling diphthong [ie]: MJC [zie n] Izenl 'good'; there is no on-glide after back consonan ts or ernphatics. 6 1
(f) A phenomenon t hat is somewhat ma rginal but is of interest because of its absence in ·the Arabic. dialects hItherto descnbed is the contrast lal vs. liil in final unstressed position. This occurs principally in M, is present in C, perhaps as a result of M influence on my C informants, and seem,s abse nt fro m J. T his contrast arises from the fact that c1assicisms (some of them extremely common) are pronounced with a fina l (a] or ' [ft. ] irrespecti ve of environment, which contrasts with the final [e] which, in M and C, occurs in almost . any environment (cf. (d) above); the contrast seems to be essentially
one of quality rather tha n one of quantity, so that it is probably not maintained after [ill, ['J a nd [ah], where the central or back vanetles occur. Th us [iaA] or [Ia,, ] 'if', [JilmmA] or [Jii mmA] 'when', [ammA] or [a mm A) 'but, as for', in "se mi-literary" style, and such classicisms
as [hunA] ' here' [uxrA ] 'other (f.)', contrast clearly with such items as [Ije] 'he came', [ummel 'his mother', [h.ixre] 'another (f.); etc.
35
..
rea~tlo n IS to be fou nd in Ru~a fi's descri ption of the co ll oq uial, in whI ch he repeatedl y describes the so und at th e end of such word s as M/semaj 'sky' as an 'alif which is "written but not pronou nced" (/~~klab lVa-la /ulj{q) ,62 clearly reAecti ng the notion that if the 'alif wer~ pronou.nced, viz., if the vowe l were long, one would gel a phonemlca li y different pronunciation. Iraqi rad io announcers do so "pro n o u~ce the 'alif" when reading literary Arab ic, and informants e.ven deSignate the pronunciati on (an £J 'I' as wro ng for st rictl y clasSIcal usage because of "not pronouncing the 'a/if" the preferred version being [anA), [anA).63 ' (g) The vowel systems of MJC taken either separately o r in their interrelations are thus more comple x than the conso nant systems, and more co uld be said abo ut them if our main aim were a complete descripti on of the dialects. Havi ng elected to foc us attention on historical and comparative matters, however, we may end overall description here and proceed to examine the very divergent treatment of OA vowels that are to be found in our dialects. 3.32 Refl exes of OAlil and luI. (0) All three dialects have as do most other Arabic dialects, a zero reAex of the OA short high 'vowels where lhey were in unst ressed" open syllab le: OA/fula nl 'such-andsuch·, M1 Cf fl an/ ; OA/biladl 'cou nt ry', M1 Cfbiadf. There is, however, a considerable difference in the treatment of these vowels where they we re in closed sylla ble or in open stressed syll able, as between M on the one hand and .Ie on the ot her. TI~e i r treatment in JC (except
PHONOLOGY
36
for loanwords and special cases) is simple: both yield lei throughout a fact which explains in large measure the almost complete absence ~f a lui phoneme in lC discussed in the precedll1g secllon) . But tn M the situation is considerably more complIcated: M has both lui and lei in such cases, but these are not necessarily reflexes of OA/ul ~~~ Iii respectively. In some environments and tn some morpholog p;tterns, OA/ul is Mlel, in others it is M/u/ ; SIm ilarly, OAIII may yield either Mlel or Mlul depending o n a number of factors. Here are some examp les:
M
OA qubba qubab xubz bustan hamil) sidq
,
,
37
COMMUNAL DIALECTS IN BAGHDAD
'dome' 'domes' 'bread' 'gard en' 'sour' 'truthfulness'
gu!;>!;>a 'room' gu!;>a!;> 'rooms' xubuz bestan bamul) sudug
rub'
'quarter'
rubu'
aula ya'kul
'a third' 'he eats' '~ating'
OeleO yakul
*makil
makel
J
C
qebba qebab
qebbi qebab
xebz
xebez
bestan bamel) sedeq gebe' OelO yakel mokel
besUin I)amee;! sedeq rebe'
telet yakel mokel
(b) S'nce Mlel and Mlul are not only found as reflexes ofOAlil and OA/ u/ 1 the ensui ng discussion necessarily antic ipates some details , f OA/a/' and of OA consonant clusters. Wherever of the treat}llcnt a ' . ' eel a short historical or morphological conslderahons lead us to exp . 1I other than lall the choice between lei and lui IS usua Y VOW el ( ' I ntal environment redetermined by contextua l factors: t 1C co.~s~~a ., J I . " P . ""[ J coloTing or [u -co OTing. may be eit her "color-preservtng, C • . ' . r OAI I Mlel for OAIII (or, tn given The first will show M IuI or u, OA/'I OA/u/" s for OA/a/)· ·the seco nd will sho w Mlel for I or I' h' case, ' I I r OAI I /iI o r in given cases, la , t e the third will show M u . or U" , '11"1 Iy be lei ana t ctic vowel separating consonant clusters, WI Si mi ~r _ PulYaccOrd ing to whether the environment IS [eJ-coloTing or [uJ or I f II 'n 352 An exhausll ve colaTin but this is discussed more u Y I .. . g., f all oss ible envi ronments ca nn ot be given here,. ?ut ~~~C~~~I~~v~n; shouid give an adequate idea of the sa li en t re~ulartlt~esh' . . . t . re notably tho se tn w lI C (i) Color-pre servUlg envlronmen s (1 . , _ the vowe l is flanked by a velar o n one Side and a non-1Mck. non
emphatic, non-labial consonant on the other: I yakull vs. Imakell vs. Irekbat/ . If the other consonant is back, emphatic or labial, Ihe environment is no longer color-preserving but [uj-coloring (see below) which of course still preserves OAlul but not OA/e/. (ii) The [eJ-coloring environments are best defined negatively,
lyaxu51 vs. Imaxe5/, Irukbal
viz. as involving flanking consonants which are neither color-p re-
serving nor [uj-coloring; in such cases we have Mlel for OAlul and ~ friend s', Ibestanl 'garden', Iraje ll 'husband' , 10eleOI
OA/i/: Isedqanl
'third'. (iii) The preponderance o f [ul-coloring environments is one of the salient features of M in contrast to J a nd C and , indeed, to
most Arabic dialects. These fa ll into several categories. In the most general case, M will show lui for eit her OAlul or fiI or, in give n paradigms, fo r OA/a/, whenever the flankin g consonants a re a velar or emphatic on one side and a labial on the other : Ihamul)1 'sou r', Iwagufl 'standin g', l?a!;>u!1 "officer' (cf. Ibamell 'ca rrying' l lawnl 'necessary' etc.); Isudugl 'truth' (cf. Isejenl 'jail '); Imetxabbull 'going crazy', (but Imet'alleml 'educated'); Il;lTa! ul]11 'thick lips', Idanabugl 'drums' (but 11:l\vajebl 'eyebrows'). There may be Irl instead of the ve lar or emphatic, but th is is [uJ·coloring in a more restricted fashion.
There is fluctuation after lal, e.g. Isarebl and Isarubl 'drinking', I~arebl and I~a rub/' hitting', but there is only lui where Irl is preceded or followed by la/ : Imetgarrubl 'having gone abroad', and is most common in anaptyxis: Idarubl ' road', Igaburl 'grave" , Ibarubl 'war' (cf. Ibaredl 'cold', Isarejl 'saddle'). In addition to the environments "velar or emphatic plus labial" and '" /rl plus labial," the presence of an adjacent Iwl is usualry sufficient to cause [uj-coloring, except that if Iwl precedes the vowel there is fluctuation between lui and lei, Whereas if it follows lui seems stable : Iwu ledl and Iweledl 'children', Iwusall and Iwesal/ 'he arrived', but Isuwal 'together', Id uwa/ ·remedy'. . For two further cases of [uJ-coloring in anaptyxis, see 3.52 below. (e) The treatment of the short high OA vowels in J and C is largely analogous to their treatment in all the other qelru·dialects exam ined, i. c. Mosul, Mardin , 'Ana, and the Anatolian dia lec ts. However,
the M preservation of lui in some environments and the Iii> lui shift in others is not wholly ana logo us to the ge/ef-dialects of the countryside, which have more occurrences of lei for OAlul, i.e. fewer [u]·coloring
and
apparent ly
no
color-preserving
environments:
PHONOLOGY
COMMUNAL DIALECTS IN BAGHDAD
38
these dialects have e.g. Ikelll 'all', Iyakell 'he eats', unlik e M, which has lui in these forms but, like M Ibarput:)1 'so ur', Irul;lU' I 'quarter'.6' The nomadic dialects of tbe area seem to go along with the rural geleldialects in this respect. 66 3.33 Reflexes of OA/' /, (a) Whereas other different iating features tend to group two of the dialects as against the third (most commonly J and C as aga inst M), OAlal is treated in three different ways in the three dialects. Generally speaking, it will be seen that (i) in J, OAlal is preserved as lal in stressed syllables, and in closed poststress syllables, changed to lei in pre-stress closed syllables and zeroed-out in un stressed open syllables; (ii) in C, OAlal is preserved as lal in all positions; (iii) in M, the picture is far more complex: lal is preserved in closed syllables (stressed or un stressed), but in open sy llables the reflex may be lal, lei, lui, or ze ro , dependin g o n the
environment. The following are paradigmatic examples : M
OA 8aqiJ xaOf sawarib sak.ki n jamal jamalayn xabbaz
'heavy' 'light' 'mustaches' 'knives' 'ca mel'
8egiJ xafir swareb scacin jemal
'two came ls' jmalen xabbaz 'baker'
J
8qiJ xfir swigeb skakjn jamal jemlen xebb5z
C taqiJ xafif sawegeb sa kekin jamal jamalen xabbaz
(6) On reflexes of OA/al before Iwl and /y/, cf. 3.37 below. A number of specia l cases and deviations from the genera l scheme illustrated by the foregoing table will appear in the di sc ussion of specific morphological patterns or individual lexical items. Some additional facets of this threefo ld treatment of OA/al are the following. (i) In J, th e ze ro and lei reflexes appear in their expected places with the greatest regularity and few deviations. Where OA fal was preceded by initial /'/. it is usually reta in ed even if unstressed: /abu na / 'our father' , l axu nul 'his brother' l aka ltul ' I ate' , l aftahem l ' I understand', and so on thro ughou t ; howe ver, r~iI / 'well born' is regu lar. The J treatment of OA/a/ res ults in special morphonernic alternatio ns
39
of lal, lei, and zero that give J verb and noun paradigms their apparent intricacy, e.g. Ifatal)1 'he opened', Iftabul 'he opened it'; Ifatl)etl 'she opened', Ifetl)etul 'she opened it'; Iftabil 'open! (f.s.)', Ifetbihal 'open it!'; /jamalf 'camel', Ijmalul 'his camel', /jemlenl 'two ca mels', and many more. (ii) In C, the prese rvation of lal is nea rly as regular as its nonpreservation in 1. Deviations are less rationally classifiable than in J, and include some common wo rd s with Iii : Ikbigl 'big', fjd idl 'new', Iktil!.l 'much', where OA had lal between the first two co nsona nls ; informants give Ibmigl side by side with Ibamigl 'donkeys'. In such instances C goes along with 1, as it does also in adjectives which in OA had the pattern qatltill: Ikeslanl ' lazy', /, e!Sfml 'thirsty', etc., though in other patterns with a similar syll ab ic structure, lal is preserved : Ixabbazl ' baker', Imaftllbl 'open', etc. as expected. (iii) In M, lal is preserved in OA closed syllables; its trea tm ent in open syllables is complex and depends on several factors: whether stressed or unstressed, the quality of the following vowe l, and the quality of the adjacent consonants. The simplest case is th a t of laK I in the following syllable: the rellex is zero except if the initial was j'j . Thus Iswarebl ' mustaches', Irmadl 'ashes', Iksalal 'lazy (pl.)', bUI lawademl 'peo ple', lasamil 'names': Even this "simple" case has many exceptions, in part connected with morphology: many nouns of the qattil pattern prese rve the lal (fsalaml 'peace', Isa rabl 'wine', Ijamall 'beauty'), whereas most plural patterns regularly drop the la/. A second case of zeroing-o ut is that of OA/al preceded by a stressed syllable: lad mil 'a man' , Ibarbawl 'they fou ght', Isalmawl 'they greeted', Itwa nsawl 'they enjoyed themselves', Itfetbinl 'you (f. s.) open'. However, in the case of the last three verba l pa rad igms M also has Isallemawl, Itwannesawl and Iteftal)in/, cf. 4.82 and 4.83; my informants seem to use both forms interchangeably. A third case is that of OA/al followed by liKI o r lOKI in the next syllable; here the reflex is lal only if the preceding consonant was /,/, 11)/, /'I, Ixl or /i!,/ : j'atigl 'old', la~iJl 'well-born', IbaJibl 'milk', Ixafifl ' light', Igasill 'wash', j'ajuzj 'old woman' , instances of /a/ in qatil nouns in other environments seem to be due to borrowing: I~adigl 'friend'. Ifaqirl 'poor', Iqalill 'little', Iqaribl ' near' (cf. the more ordina ry Ige ribl ' near'). Whe n OAlal was not preceded by one of th e aforementioned consonants, it has been replaced by an lei which a lterna tes with ze ro (and .ma y be co nsidered anaptyctiS): loebirl 'big', f be'idl 'far',
40
COMMUNAL DIALECTS IN BAGHDAD
PHONOLOGY
'fat', Iyehudl 'Jews', /jenubl ' south', Ilewill 'long'. This lei is present when the wo rd in question is preceded by silence or by a word ending in a single consonant; if preceded by a closely joined word ending in a vowel or certai n consonant clusters, the lei drops out: Imudda lwilal 'a long period' , Ifadwebda jdidal 'a new one (f.)', leanna ebirl 'he seems big'. Thi s elision of lei is, however, optional,
'neck', etc.) and by the behavior of the vowels when pronom inal
Iseminl
as is its retention after a pause whenever the resulting consonant cluster
,.-
41
suffixes are added; details will be found in Chapter 4 68 (e) T~e treatment of OA/al in J is practically identical wit h that of the 'Ana dialect, whereas that of C is ve ry nearly the same as that of~os ul. The M treatment, while simi lar to that of the rura l geleldialects, differs from them in several important respects, notabl y in
the different dist ribut ion of the lei and lui reflexes and in the absence in M of the syllabic reshuffling that makes these rural dialects so
is permissible: in isolation one may hear both Iseminl and Isminl ' fat', leebirl and Icbirl 'big',/refijl and Irfijl 'friend', etc. F inally, a number of cases that seem historically related may be subsumed under a fou rth heading, namely "the case of two successive la/,s."
si milar to the no madic dialects of the area. Thus, from my informant fro m Mu sayya b district (s imilar examples may be fou nd in Meiss ner,
Thi s matter will re-appear in the morph ology , but deserves a separate
(M jxesba/),
paragrap h here. (iv) In M, OAlal in a stressed open syllable changes to lei or lui if it was immediately followed by anot her la/; if that second lal was in an open sy llable, it has dropped out; if it was in a closed syllable, it was retained . This phenomenon has thus had th e effect of consid-
Ighawal 'co ffee' (M/ga hwa/), /, rubi/ 'a vi ll ager' (M/,e rbij). 3.34 Reflexes of OA/il and lui. As a rul e, the OA high long vowe l ~ are represented by /i/ and lui, respecti vely, in all three dialects, though
°erably altering nouns on the pattern qalal, qa talT,67 and especially
add iti o n, a difference in their treatment th at is peculiar to C, namely
the perfect of Form I ve rbs with their basic qalal pattern. If the enviro nment is of the [ul-coloring type, the new vowel is lui, otherwise it is lei: Igu marl ' moon', l I:> u~a l l 'onions', Ikumasl 'he grasped', . IkuI:>arl 'he glew up', IUI" arl 'he ordered', contrasting with Ijemall 'camel', Ijeball ' mountain', Isemacl ' fi sh', Icefanl 'shro ud', Igela'i ' he cut', Ireka!)1 'he ran', l(je bakl 'he laughed'. The same applies to
their lowering respect ively to lei and 161 before Ib/, Iql and Ii!.! : Imale~1 'nice', l~a ~ e ~1 'who le', Ige~al 'smell', l~ t6 ~1 'roofs', lag61.11 'I go' (and so throughout the im pf. and impt v. of th is ve rb), Is6ql 'market', I ma~g6ql ' burnt', Imaql6ql 'worried', Isand6ql ' box', /~6ga/ 'picture'. However, this is not carried out wi th e ntire consistency : lasuql '[ convey', cf. Is6ql 'market', labuql 'I steal', a nd examples are not numerous eno ugh to a llow a conjecture as to the nature of the variati on. Here again, C is similar to Mosul; which shows /e/ a nd 161 in similar cases. As to the J shorten ing of OA/il and lui, (a phenomenon th at is com mon enough in the Egy pti an and Syrian areas but rare in Iraq) there is a pract ica ll y id e ntical shor tening of OA long vowels, includ ing la/, in 'Ana. 3.35 Reflexes of OA final la'i. In 3.36 the treatment of OA/al with respect to 'imala will be discussed in detail. In this ·paragraph attentio n is ca ll ed to anot her point of divergence among the three dialects as to treatment of OA/a/: in the feminines of adjectives denotin g colors and infirmities which had the OA pattern qat/a' , . the /'I has dropped in MJC as in the ot her Arabic dia lects, but in M the final vowe l is short and uns tressed , whe reas in JC it is lo ng and stressed. Thus M/s6da/ , J/suda /, C/s6ditl 'black·: M/ tarsa/ . J/tegsa/, C/lagSiI 'deaf'. a nd so on throug hout; On the other hand. M does
cases in wh ich the second / a/ is final , i.c. may once have been long but
was subsequently shortened: Isemal 'sky', Ibenal 'he built', Ige!)al ' he spent (time)', but Iduwal 'remedy', Isuwal 'together', Imu!)al 'he signed'~ Ibuqal 'he stayed'. With a second lal in open syllable eli minated , we find e.g. Isemcal 'a fis h', Ixesbal 'a piece .of wood', Isej ral ' a tiee', Irpu lratl 'it rained', Iqub(jawl 'they grasped', Isebhawl they swam', Ige'datl 'she sat', ·as well as /, erbil 'a vi llager' . (OAj'arabiy/), /,ejmil 'Persian'. Th is shift is most consistent in the verb, where the only exceptions are the verbs / aka l/ 'he ate' and /axao/' he took'; nouns are more problemat ic, with deviants includ in g,
aside from such loanwords as Isakarl 'sugar' and Iqanafal 'sofa', also such items as l ~ alaI:>1 'wood', Ixa!?arl 'piece of news', /,asall 'honey', Ihawal 'wind', /, a~al 'stick', and a good man y others. Fu rther complications are introduced · by fluctuation between lei and lui in certain . cases (jwe~all a nd I wu~al l 'he arrived', Irugbal and Iregbal
1903b; Weissbach,
1908, 1930)
I have
Ixsebal
'piece of wood'
Igu~aI:>1 ' reeds' (M/ge~aI:>/),jg~ u I:>al 'a reed' (M/ge~!?a/),
J stands apart from the ot her two in th at it regularl y shortens th e m
to
Iii
and
lui
in unstressed position (see 3.31e above). There is, in
42
COMMUNA L DiALECTS IN BAGHOAD
PHONOLOGY
share with J and C a similar retention of length and stress in M/tliiOli/ 'Tuesday ' (l/t/aSa/, C/tIata/), and larba'al ' Wednesday' (l/agb'a/, Cjarba'a/); and all t hree have a final stressed lal as a reflex of older lah/: M/sefnil 'we sa w him', MJCjmdaral 'ca re', Imesl)al 's hove l', etc. In all other instances, OA final lal and la'i have been reduced to lal, unless 'im{ila has taken place. Mos ul shows the same retenti on of length and stress in the feminines o f the (OA) qallti' pattern and in the words for T uesday and Wednesday as do J and C; ho wever 'Ana goes with M in this respect, as do the rura l gelel-dialects, which do not have lill in the wo rd s for Tuesday and Wednesday. 3.36 The 'imiila. (a) One o f th e most striking features differentiating J and C on the one hand from M on the other is the treatment of DAlai ; in M, we reg ularl y have lal (fal in final position), but in J and C there is o ften lei or /if. This phenomeno n is of co nsid erable co mparative and historical interest, an d it will be discussed in so me detai l. There are at least six cases that may be consid ered separately: (i) the case of DAlai near Iii or Iii in the noun ; (ii) the special case Qfthe ve rb ; (iii) the case of word final la/ ; (iv) the feminine endin g; (v) the special case of the numerals; (vi) so me isolated cases. This breakdow n poi nts up t he fact (already illustrated by several of the forego ing discussions of given sou nd shifts, notably in the short ·vowels) that different mo rphological classes in volve different diachronic treatment o f the sa me sound ; th is "morphological differentia tion" or "morphologica l conditioning" of sound shifts is remarkab ly extensive in J, seem's less commo n in C and is least frequent in M. (i) In the general case, when DAlai was in proximity to Iii or Iii in the affix less noun, it is represe nted by lei in C and by Iii in J: OA kilftb jami '
6ibban mizan
I!awajib
'dogs' 'mosque' 'flies' 'scale' 'eyebrows'
M
J
C
clab
klib
kleb
jame'
jime'
jeme'
6ebban
oe bbin
debben
mizan
mi zi n
mizen
I.l\vajeb
l)wijeb
ba wejeb
43
Ibasatinj 'gardens', M/ bsatin/, J/ bsatin/ , C/ basetin/. Where 'imtilaprone lal was Immediately fo llowed by Iyl, C has laYI (w hich in this pOSI tion does not co ntrast with the regu lar ly expected Ie/) and J has the reg ularly expected Ii/ whi ch in this pos ition varies free ly with Ily/ ; however, this does not ho ld for all J reflexes, nOla bl y of th e plu ral pattern qartiyil (qat ti'il), where J has la/ : /.1ikaya 'adiy is
'story'
i)caya
'brides'
'rayes
bkiyi 'gayes
bkayyi 'agayyes
There is, in bot h J, and .C~ at least one in stance of mo rpho log ica ll y defi na bl ~ pres~ rv~tlOn of / a/ where ';mdla might have been expec ted, namel y
In
adject ive plurals corresponding to the OA pattern qil (i1.
Thus JCj kbagl 'big', Imlal.ll ' nice', Is mflnl 'ra l', I!waif 'long', and so ? n throughout. In all lik elihood , th ese do not form real exceptions, In . th~t they are pr? bably based o n an OA pattern qUId! even though thiS IS unattes ted In Classical Arabic for thi s part icular word class (see p~r. (b) below),69 There are, howe ver, a goo d many unsys temati c rete~ tl o ns of / a( in give n words , which are not the sa me in J as they
a re In C: J/I.1Slbl but C/hsabl 'account', C/awedeml 'people' bu t J/awftd?m / , Cl ktebl ' book' bUI J/ktffb/, th o ugh Iktibil 'a (piece of) wr~tln g,; and !Cjlazeml 'necessary', IUIjerl ' merchant ', Is are'l 'street', Iwadl l va lley, Ibladl 'cou ntry', and o lhers. (ii) '1n th e ve rb, the JC treatment of DAlai is so mewhat dif-
ferent. ~n the active parliciple of Form I verbs, C has th e expected reftex lei (fayl before Iy/), and J in this case follows .lhe C pattern rather than the expected Iii :
,waqif nayim
OA
M
'standin g' 'sleeping'
naye m
waguf
J weqe f n,lyyem
C weqef nayyem
Thus while there is nothing unus ua l in C, there is in J a tre atme nt of ,pa,rti cip les that se ts th em apart fro m nouns o r adjec tives Corres-
~o ndlng to OA qiilil for ms, and yielding such doub lets as J/jeme'l
Ho weve r, J dev iates from this in the plural patte rn correspond in g
ha Vin g ga th ered' vs. /j im e'/ 'mosqu e'.70 In other ve rba l paradigms
to OA gartilil, in that it has lal rather th an the expected Iii : OA
where DAlai was nea r /ii, notably th e perfect, th e imperat ive, and
44
COMM UNA L DIALECTS IN BAGHDAD
PHONOLOGY
the active participle of Form III and the acti ve participle of Form VI, no 'imcila ta kes place at all in either J or C:
(iv) A discussion of the treatment of the feminine ending is So mewhat extraneous to the question of the reflexes of OA/a/, since the OA femmme endmg seems to ha ve been , in th e pause lahl or lal rather than i a/. It is mentioned here only beca use it w~s subject to a very specl~c. kind o,r 'h~llila resulting in an allomorphic split, so that the. fe mmme endmg m J and C is now lal in so me instances and /II m others. The fu ll de ta ils will ap pear in the mo rpho logy but It may be stated here that in a very ge nera l way, C has lal neal: emphatIcs and back consonants and Iii after other conso nants whereas J has lal or /if depending not on the preceding co nso na nt ' but on the phoneme preceding that consonant: IiI if that phonem~ was mOAN , III, or Iyl, lal if otherwise:
OA asami~
sam ib musamii)
M
'I forgive' 'forgive !' 'fo rgiving'
J
asameJ:! sa mel) msamel)
asameil samel) msa mcl)
x
C
asamcl)
sameD msa meJ:!
(iii) Some nouns and particles correspo nding to OA forms with final lal o r la'i have Iii in both J and C. These are either words which the Arabic script spells with 'alif maq$llra, or, in the case of final WI, words with Iii in the pre-final syllable. However, the prese nt JC /i/ does not appear in the reflexes of a ll such words; it does not appear in verbs (e.g. JC/ banal 'he built', Isammal 'he named') or in elatives of roots with KJ -y or '" (e.g. JCja' lal ' higher', laqwal 'stronger'); it does appear reg ularl y in plural adjectives o n th e OA
bayOa.
pattern qafdld, and its occu rrence or non-occurrence in other items
zayna
does not lend itself to classification:
kalba bazzuna
OA ka,saHi 'a' rna
bala sita' ~inna'
M
J
M
J
C
'egg' 'good (f.s.)' 'bitch'
boO. zena calba
'cat'
bazzuna
boOi zeni kalba bezzfma
bed a lcni kalbi
C
' lazy (pl.)' 'blind'
ksala
ksali
kasali
a'ma
a'mi
a'mi
'yes' 'wihter' 'henn a
bali
bali
bali
seta I)enna
se~i
seti I)enni
Denni
OA
. (v)
45
bazzuni
Both the cardinal and the ordinal numera ls are anoma lous
I~ severa l respects in sofar as the 'imdla is concerned. In J, the or-
?lIlal~ ha~e ,ro;ms w~th the expected Iii : ; Oinij 'seco nd ', IOHeOI thlfd '. Iglbe I fo~rth, IXimcs/ 'fifth', etc" but the corresponding
_C show_ la: (as ~n M and OA) instead of the expected lei : Irabe I, Ixa mes/, etc. The cardinal numerals o n the ~the; hand, are normal in C, bilt abnor mal in J. For OA/ wiil)id/ , one , we have the norm ally expected Cjwel)ed/ , but J also has Iwe iledl rather than an expected */ wiiled/ ; the feminine JC/ wei)di/, M/ wel)da/ , probabl~ hark back to OAj'iildaj with a Iwl due to contamination wlt,h l~ail ld (a)/, and is th erefore no rmal in M, J, a nd C with respecl te:ms
III
Itaml, Italetl,
In "the form j baJij the re is o ne of the rare in stan ces of 'imdla in M.
As to other instances of th e JC 'illlcil" o f final la/, J has l ilettil ' until , so that', but MCj i)atta/ ; for OA/mata/, 'w hen', J has lemtal, and M has jyemtaj, a form also give n by my C in forma nts. th oug h a
·form l e mtil seems to be attested fo r Basra C hristi ans." For OAr aliil ' o n', MJC have ( a laj. The fi nal Iii of JCj hollil ' here' may hark back to some such fo rm as Iha wnal < Ihii hunii/, but th e deictic words seem to mak e up a s pecial ca tegory a nd the hi story of their vocalic endi ng is problematic; a ll that ca n be said w ith certa int y
is that M has form s in lal whe re J and C ha ve fo rms in Iii : M/ hn ii kal 'there'.. J/ wnik i/. Cjh6niki/ ; M/ha56lal 'these . J/ ha06li /. Cj had 6li j.
to Imala. T~e num erals from 3 to 10 have, as in o ther dialects, each one form WIth and o ne form wit ho ut an ending that is e tymo logi-
cally th e sa me as th e femin ine ending; in C, th e fo rm s wit h th a t endlll g show the normal C alternation of Iii and la/; in J, howeve r, these form s show an /t/-/al alternation that follows the C rather than the J pattern (cf. par. (iv) above):
PHONOLOGY
COM MUNAL DIALECTS IN BA GHDAD
46
'three
xamsa sitta
'five' 'six'
'four
sab'a
'seven'
eamaniya
'eight'
lis'a
'nine' ' ten
'asara
JC/ megzibl 'gutter' and on Millarbil/. J/gegbil/, C/garbell 'sieve', for whic h late Classical and many dialects ha ve form s with lal in the second syllabl e, see the disc uss io n in 6.5 .
J
C
tlaO i agb'a
tlati arba'a
xamsl setti sab'a
xamsi setti sa b'a tmeni
(b) Except for 'Ana, and Hit where it is totally lackin g, the 'imdla is a characte ristic of the qellllMdia lects: Mosul , Tekrit , and th e Anatolian dialects all have lei (in Mosu l, also lill for OA/al
tes'a
holds for the Aleppo.
OA Oa laOa 'arba'a
47
9mini tes'a
'asga
'asga
in conditions quite similar to those indica ted for J and C. The same
sede ntary
dia lects
of
Northern Syria, notably
Beyond thi s, the Syrian area shows o nl y' imala of the feminin e e nd-
Taken individually, only the J numerals 'three' and 'five' deviate from the J norm , which should ha ve yielded *IWiOal and "/xamsa/, but the whole set taken together shows an altern~ti~n c,orresp.o~dlO~ to the C alternation. Moreover, the J numerals tlllTty and eighty
.. ~
Oamatlin
'thirty' 'eighty'
Here the J forms show ' inllila to
lal
but not of internal
Ift/, except
in so me Lebanese
dialects where it is large ly condi ti oned by the conso nanta l enviro n-
ment and is thus of a wholly different type. In Upper Egypt th ere is /i/ for fina l OA/al and for th e femin ine endin g und er cert ain conditions.72 The 'inlala found in C is closer to that of the Anatolian
are similarly abnormal: OalaOin
ing or of final
tliOin Gmin!n
/iI
(shortened fro m
tiWn tmenin
Iii in unstressed
sy lla ble) even though most nouns with a similar phonemic p_attern ,
. namely the plurals of the OA type qarali/, J has lal for OA/a/~ . (vi) A number of cases in which J, C or even M show lei, N or /iI where one would have expected lal o r lal m~st be dIScussed separately. No conditio ning from an adjacent OAN IS ~resent m C/nesl ' people' (OA and MJ/nas/), in J/keml 'how many (perhaps not a case of 'inlala, cf. note 157), in JC/l)eblel 'pregnant' (M/1)cbla / : OA/1)ubla/), which is taken up again in 6.5; in ?/'e~mel 'da:kness (OA/, atma/, M/, etma/ , C/,etmi/) or in J/i)elmel . darkness (OA Ii)alma/, M/i)elma/, c/cjalma/). As in many other dialects, meludmg some where 'ima/a is o ~herwise unknown, the names of the letters of the Arabic alphabet have lei in MJC, e.g. Ibe/, Itel, Il)el etc. A few M words ma y have preserved an 'imiila which was once not limited to the Jews and Christians of Baghdad, but chaqlle mol a son histoire and the examples noted are problem atic in v~nous ways: the history of M/hIcil 'thus' and M/anil ' I' is unc~rta!n; o~ MJC/balil 'yes', cf. 6.5; on an unusual M/jijl for M/dejajl he~s, for .which J/jijl and C/jejl are no rmal; see note 106; on M/meznb/,
d ialects than to that of Mosul in that Mosul has iii for OA/al in certain cases, whereas C has lei : Mosul Ibasitinl 'gard ens', C/baselin/ ; in addit ion both Mosul and some A natolian dia lects have what may be called a "productive" 'ill/ala that is lack ing in both C and J, i.e. lal in suffix less words changes to lei when certain suffixes with iii are added: Mosu l Ibestanl 'garden', IbestenCiI 'gardener'; Ibagdadl ' Baghdad', Ibagdedlil 'a Baghdadi'. The ' imala of J is unusual (among the dialects that have thus far come to my attention) in two respects: the prevalence and cond iti ons of occurrence of Iii for OA/a/, 73 and th e co nditi ons of the alternati o n in the feminine ending. In nonc of the dialects in which 'imalll of interna l OA/al
is conditioned by proximity of DA/i/ o r Iii do we find 'imdla in the plura ls correspo ndin g to OA qild!. Take n togeth er with certain data from other dialects (non-affrication of OA/kj or Iq/, e.g. M/ kbarl 'big (pl.), vs. loebirl 'big (s.)', rural Igriibl 'near (pl.)' vs. Ijeribl 'near (s.)' ; Egyptian forms s uch as Igud adl ' new (pl.)', Igum a ll 'beautiful (pl.)') this indicates that such fo rms may we ll be based on a non-attested OA pattern qllldl for adjectival plu. mls. " The following synoptic table is intended to summarize th e facts of 'imdla for th e Mesopotamian and North Syria n areas :
48
COMMUNAL DIALECTS IN 8AGHDAD
C
PHONOLOGY
J
Mosul
Anatolia
Aleppo
klib
kleb jeme' basitln a'mi ka lbi
kleb jeme' basetin
kleb jemc' basetin
a"ma
a"ma
kalbc;
kalb~
' blind' 'bitch'
be(\a
be<ja
'egg'
kleb jeme' basetin
jime'
a'mi
a'mi
kalbi
kalba
bsatin
'dogs' 'mosque' 'gardens'
ka l b~
be<ja
be(\i
bC(\a
(c) From abundant passages in the Arabic grammatical literature 75 it emerges quite clearly that OAlal (or, more acc urately, what traditional spelling rep resents by fatha plus 'alif or plus 'alif maq~lira) was rendered in Iraq, especiall y in Baghdad , wi th an "incl inati on"
('inlala) towards Iii . This inclination could be sligh t, medium, or strong, viz. presumably varying from [a] to [e] to [I], though it is rarely clear which degree is intended in every specific case cited.
49
any adjective plurals of the type qital which, as has been pointed out, show no 'imdla in prescnt-day dialects. Some nouns in which the conditions for 'il1l(i1a are present are said to reta in lal , including II.l imful 'ass' (cf. JCji)mag/) and others, in which the prox imity of a back co nsonant is sometimes thought to have prcvcntcd the
change,77 (ii) The 'imala prescribed by some for forms in which lal implies a radical y and alternates with IV is, as has been noted, unknown in our dialccts; this category included such forms as e.g.
{ab, 'be good' impf. ya{ib, ball a 'to build', impf. yabll;' Sibawayhi's statement that active participles of Form I ver bs havc no 'im ola (though other words of the same qatil pattern do) is contrad icted by (later) usage. but may have so mething to do with the qetel- qitel spli t in J. 7 8 No examples of 'imala in verbs of Forms III or VI have co me to my attention, and none occur in our dialects. (i ii) Final lal does not appear to be listed as a separate cate-
'inliila; l;Iariri (d. 11 22) reminds his readers that their pronouncing
and not M, which have preserved a feature that was more widespread
of hat/a with 'imdla goes against the foregoing ru le about particles
in Abbasid Baghdad. The gelet-dialects, as .well as those qeltl/-dialects that are wi thout 'imala, have probably been influenced by diaiects
and !;lim uses hiini 'here'; 79 cf. the forms Ih6ni/, Ibalil, Ii)etli/ , lemtil discussed above. There is an interesting parallel to present JC usage in an eighth century source quoted by Fiick and connected with the treatment of 'aqtal forms with KrY: a dis tinction seems to have been made between 'a'nla 'blind' a nd 'a'ma (mill) ' more blind (than)' in that the former was pronounced with 'illldia (cf. JCja'mil 'blind ') whereas the clative was not (cf. s uch clatives as JCja' lal 'higher').80 (iv) The 'imala of the feminine ending is briefly mentioned by
prescribed and collo'q uial, took place, It is noteworthy that words so pronounced tend to ar range themselves into categories not unlikc those listed in sub-paragraphs (i) to (I·i) above for the modern di-
alects, (i) In the noun, the proximity of a kasra is th e main cause of 'imdla ;76 the stock examples are words likc lisdn 'to ngue', njdl
'man', cf. Cjlsen/, Ig jel /, J/lsin/, liljil/; l;Iilli has, in his Baghdadi poems, forms likeflrid, lI'irid, mirid, birid (Hoenerbach, 1956, p. 135) I/rib (ibid., p. 154). None of the sources consulted seem to include
,
j '
gory, and the significance of the received spellings with yd' deserves in vestigation. There are many examples of 'inldla in this pos ition,
Moreover, 'imala was both a hallmark of Baghdadi speech and the refore a desideratum in vernacu lar poetry, alld prescribed , by some authorities at least, for correct dict ion and Qur'an reading; the two usages did not, however, e ntirely overlap, and it will be seen that it is on the whole the colloq uial rather than the prescribed 'imala which has come down to us in the qeltu-dialects and the related North Syrian varieties. In this as in other cases, it is J and C, then,
introduced into the arca. at latc r times, though of co urse they may contin ue o lder loca l" nOI1-'il11(ila dialects. I give here, abstracted from the sources consu lted, a cata logue under which 'imala, both
,
largely in words spelled with yo', some of which parallel presentday instances. Besides the verbs in K J-y already mentioned, and the proll. suff. -nd and -ha, which have no parallels in our dialects, we are told that particles do not undergo 'inuila. However, bald 'yes' and mota 'when' arc exceptiona l and are pronounced with
Sibawayhi; he merely states that he has heard such a pronunciation, which he interprets as "li ke ning the feminine ending to an 'aft/"SI In later times, it was actually prescribed by some authorities; some
forbade 'imala after back consona nts and emphatics, para lleling the usage of C, Mosul, Anatolia, and many Syrian dialects; while others accepted it after such consona nts as well, paralleling the
,.
50
COMMUNA L DIALECTS IN BAGHDAD
treatment in J (where fro ntness or back ness of consonants is irrelevant for feminine endin g 'imiila), or the M treatment of all final /a/,s except after pharyngeals. This latter type, viz. 'imala of the fern . suff. after both front and back consonants, is even said to be typical of the 11I11lVol/a{hill of Kufa and Basra B ' (v) While the so urces have nothing to say that wou ld elucidate the special J treatment of the numerals they do mention a number of special cases in which 'imdla takes place "irregularly"; man y of these are parallel to present-day usage: the exceptional particles a lready mentioned, e.g. bala, cf. MJC/bali/ (and also Pers. bOle, Turk. beli), as well as the word lias (cf. C/nes/), the names of the letters of the alphabet (cf. the M1C treatment)" and the word !lUbla 'pregna nt' (cf. the lC/hebl(\j)B 4 .3.37 Reflexes of OA/ayl and law/. (a) On the whole, the treatment of the OA diphthongs in M1C is of a type familiar from many Arabic dialects : lei and 161, respectively, except before Iyl and Iwl, where we have layl and law/: M1C/betl 'house', Im6tl 'death', lawwalf 'first', Imayyetl 'dead'. Diphthongs are a lso preserved in certain morphological patterns: M1C/awsa'l ' broader', laybasl 'drier'. The contrasts lay/-/ el and law/-/61 are of very low functional yield, but are sufficiently well-established : M1C/jaysi 'army', Ibesl 'how much', fjebl 'pocket', Mllbetal 'his house (M), her house (1)" Imaytal 'dead (f.)'; C/gel).al 'smell', IgaYhal 'going (f.)'; M/zawjal 'wife', Im6jal ' wavele ngth', Ij6zal 'nut'; M/awsa'i ' broader', M /6~all 'I arri ve', M1C/aw/, 'or', /161 'or'. On the verba I. ending of the 3rd pers. pI., M/n."1 alternating \vith 10/, lC/ul alternating with 101, see 4.2 below." (b) A feature that sets 1 apa rt fro m M and C is the treatment of th ese OA diphthongs in unstressed syllable:" instead of lei and 16/, 1 has in such syllables Iii and lui and, in certain specific cases, liYI and luw/. T h us M/bU)a/, C/ber).i/, but llbil)il 'white (f.)'; M/s6da/, C/sodal llsudil 'black (f.)'; M1C/betl 'house', but MC/betenl vs. l l bitenl 'two houses'; M1C/y6m/, but MC/y6menl vs. J/yumenl 'two days'. T his al ternation between stressed lei or 161 and unstressed Iii or lui also ta kes place in such loanwords as Ikekl 'cake' , Ipatetal 'potatoes', which yield J/ kikayi/ 'a piece of cake' and l l patitayil 'a potato', In the J verb, it seems to have been extended to pronominal endings in which an etymological diphthong would be unexpected: Iqeltul ' I. said', but Iqelt61ul ' I said to him'; Iqalul 'th ey said', but
PHONOLOGY
51
Iqal6lu/ , 'they said to him' ; Iftah i/ 'o pen (f.)', but Ifetl)elul 'open for hIm , etc.; for ?ctalls and so me etymological conjectures, see 4.2 below. As for Ilyl and luwl, we find th em in cases where the prese nt-day st ressed alternants are still laYI and law/: llsayyabl 'he a bandoned ', IS,iyyabul ' he abandoned him' ; llsawwal ' he did' (suwwanul :he did it'; J/ mka yyefl 'glad', pI. I mkiyfin/ ; J/mxa wweri frr ghtenrng, pI. Imxuwfm/ . Howe ve r, luwl IS also found in passive partlclpJes of K , -IV verbs: Imuwjudl 'present', Imuwzunl 'weighed' . That these unstressed liYI an d luwl do not contrast with /i/ and lui, respectively, has alread y been mentioned (3 .3I e) ; whether or not they contrast wi th [oy] and [ow] is less clea r ; I doubt whether minimal pairs can be found , and speaker reaction leaves me uncertain as to whether [mowjud] is kept apa rt rrom [muj ud]. Morphophonemlca ll y, of co urse, such a spell ing as Imewjud/ , paralleling Ime ktub/, Imeftuh/, etc. wo uld be justifiable , as wo uld Isewwa nlll pa ralleling Ixell.nul (fxallal 'he put, left') or Imkey fin/, / mxewfin/, etc. for analogous ~easons . However, in s lIch cases it is prudent to "hug the phonetiC ground closel y," and citation forms for such words wi ll be written with fiy/ and juw/, which are to be interpreted as [IY] - [i] and [uw] - [U]B. T his who le 1 alternatio n is, o r co 1Irse, closely bound up with the red uctio n of OAlal to lei in unst ressed closed syllable discussed in 3.33 above: just as Ika/b/ 'dog' yields Ikelbenl 'two dogs', so */bayt/, when diphthongs were still diphthongs must have -yielded */beytaynl then */ bi ytenl and finally the present Ibitcn/ . Similarly for lawl > lewl > luwl > -Jul. No such alternation takes place in Form I acti ve participl es: jseme'/. 'hearing', pI. Isem'in/ , Igayyebl 'going', pI. Igaybin/; no r in verbs on the pa tterns qidal and qOla/ : Inesantul ' J aimed ', Id6xantul 'I rclt faint'. (e) In the Mesopotamian area, OA/ayl a nd lawl are preserved as phonetic diph thongs in so me places but not in others witho ut con-
nect io n to the qellll-gelel split. Mos ul has lei and /6/ like th e three Baghdad dialects and so, apparently , ha ve the lower Iraqi cities as a who le, as ha ve Urfa and Swerek- in Anatol ia ; however, Mardin , Qarabas, Carmuc, and 'Ana have preserved th e diphthongs, mostly as fey] and [ow] ; so have so me o f th e nomadic Kelel-d ial ects, as we ll us so me or th e rura l ones, including K huzistan and Kuwayt . where th e situation clo se ly pa rallel s that of M. In 'Ana, whe re the treatment of short / a/ in unstressed sy llabic c..::Joscly para llels th aL of J, there is an equall y close parallel to Iht: J red uction of OA diph-
'.
52
COMMUNAL DIALECTS IN BAGHDAD
thongs In unstressed syll ab les : [beyt] ' ho use', duo [beteyn]; [yowm] 'day', duo [yo meyn], etc. 3.4 STRESS. There is little or no difference in the place of word stress among the three dialects. As in many, perhaps all , other dialects, that place is usually determined by the syllabic structure of the isolated prefix less wo rd and, whe re this does not hold , by a number of morphological consideratio ns. In the majo rity of cases, stress is o n VKK or 'VK closest to the end of the word, and if neither VKK no r 'VK occur, on the first V: MJCjjebnal 'we brought', Ijeb nakl 'we brought yo u' Jjjbul 'brin g (imptv. pl.)' Iwaladl 'boy'. In the present notation , final lei, Iii, and lui are a lways stressed, final lal and 161 a lmost always, so that o ne could add this to the rule and refrain from marking stress on such fina l long vowe ls. e.g. Ml'ale/ .'o n him' , JCjl)eblel 'pregna nt', M/axul ' his brother', M/seftil 'you (f.s.) saw him', (but Ise ft iJ 'you (f.s.) saw/) M/lliiOal 'Tuesday' , .J/suda/, C/s6diil 'blac k (f.)', M/saf61 ' they saw him ' ; in the case of .fi nal lal and 16/, the few exceptions could easily be taken care of in a sma ll list. Turning now to cases w here morphological co nsiderati o ns alter this overa ll pattern , the follow ing types OCClIr:
(i) In J and C, certain nominal and ve rbal base types stress the syllab le immediately preceding an object pron. suff. regardless of the syll ab ic structure: JCjlebesl 'wearing', Ilebesul 'wearing it (m.)' ; JCjkammal1 ' he co mpleted ', J/ kemma la/, Cj kamma lal 'he completed it (f.)'; J/ hwijeb/, Cjl)awejebl 'eyebrows', J/ I)wijebi/, Cjl)awejeb ij 'my eyebrows"; this naturally results in some nearmin imal pairs, e.g. C/ madrasil 'schoo l', Cjmal)bas il ' my ring'; J/ mjannenal 'crazed, maddened (f.s.)', Imjennenal '(he) maddened her' . In M the patterns corresponding to the above JC exa mples stress acco rding to the genera l YKK-YK rute, but some pattern s do admit, optio nally, the shift of stress to the sy llab ic 'preceding the object su ffix: M/ ketbatal and Iketbatal 'she wrote it (m. )'; M/ xanjaral and Ixa njiual ' his dagge r' (vs. Iqililda ral 's hoes' with . only o ne poss ible stress) . . (ii) In C, a rather stable dist inction is created by anaptyxis in
Ib"tabetl 'she wrote' vs. j katabet j 'you (m.s.) wrote' and so throughout this pattern; with su ffixes, of co urse: / katabetu/ 'she wrote it', / katabtu/ 'you wrote it'. In other cases in MJ as we ll as in C, a vowe l that is historica ll y allaptyctic may, ir in stressable position accord ing to the YKK-'VK rule, be stressed o r not more o r less opt iona ll y: MJC
PHONOLOGY
53 Ib~netnal or Ibenetna/, M/rajelhal or Irajelhal 'her husband' M Igeletl ~1 or IgeJetla/, JCjqeletlul Or Iqeletlul 'yo u (m .s.) told h'il11 ' (III) In the Im perfect of Forms VII a nd VIII verbs M fl t . bet " I " ' uc uates ~veen reg u ar stress a nd stressing the first sy llable of the base' ,~/afteheml a nd, lafteheml ' [ understand ' la nxubu~1 and ja nxubusi get dIstracted ; J and C, on the other hand, always stress the fir~t syllable of the base: J/afuihem/, Cjaftehem/ ; J/anxilbes/, Cja nxebes/. A few other morphologically condItioned stress patterns will appea r In the morphology (Chapt. 4). A feat ure that seems equally cha racteristic of M, ~ , and C is the very rreq ucnt stressing or a nUlllber'or preposed. partIcles, such as Ili/- , Ibj-, Imen/-, /'al/- a nd es pecially the negatives Ima/- and Ila/- and the interrogative MC/s/- , J/a5/-; when the p~rt lcle does not itself have a vowe l, it draws the stress as ~ear tO , ltself as possible:. M/sdassawwil 'what are you doing ?' Isk~tabetl \~hat did you write', / setridl 'what do yo u want ', MJ Im.a,hel wa~ not pretty (f.s.)" MJCjlat naml 'don't sleep', MJC (bel aJal1 qU ickl y', M/, algii'l 'on the grou nd ', MJCjmenbet Jibetl fro m house, to house', M/ mnessugu!1 'from work', J/asakul 'what's the matter?, lasesmakl 'what's yo ur na me T.
3.5 CONSONANT CLUSTERS AND ANAPTYX IS. 8 7 3.5 1 Overall view (a) T he three dialects do not differ mar kedl y in the com binatio ns or co n~o ~ant s that Occur as initial or medial clusters; they do differ in a s tatl st l~a l sense, si nce such matters as the differential treatment of OA/al In open unstressed syll ab les alters the distribution and frequency of certain KK com bi nations as between M, J, and C. These matters have been illuslrated in 3.33 and 3.24 above.' In a ll three dIalects, a Consonant tends to be assi milated to the fo llowing consonant, un less that conso nant is a sonorant or /'/ or /'/ ; all three dialects show a m~rked. tendency toward assi milation or a stop to a fo llowing homorga nic spirant ; Ibl plus Ifara s/ is usually Iffarasl 'with a ho rse' It/ plus Isawwil is usually Issawwi! 'you do', etc. Simila rly, III tend; ~o ass l,?,late to a followin g In/: MJC/aka lj plus Inal is usua ll y laka nnal ~\fe ate. Wlt~ respect to voicing assimilation and emphasis assimilatIon, no detailed study was made, and the dialects do not see m in these. respects, to be either different from the Arab ic dialects ;s a whole or .differentiated from each other. There is, however. one feat ure whIch sets C apart from M and J, namely the grealer voic ing of consona nts preceding / bl- I nformants orten point out as "typ ica ll y
54
COMMUNAL DIALECTS IN BAGHDAD
Christian" such pronunciations as [zbu'), 'week' [ ~I;>A'tA'os)'seventeen ', for which M and J usually have initial [s) or [s). This is corro borated by my own (incomplete) observations, and though my partl y morphophonemic notation gives Ikbig /,big', pI. Ikbagl for both J and C, there is no doubt that the only form s I have heard for Care [gbig) and [gbag), whereas the J forms, as well as such words as M/ kl;>ar/, usually have initial [k). Initial clusters thus comprise many combinations of KK, but there are no occurrences of initial KKK with the exception of a few instances of initial Istgl in J (no others were noted): Istgaltul 'I worked', Istgetul 'I bought'. Medially, there are similarly many (perhaps all, taking account of assi milation) combinations of KK, and some of KKK: M/txarmsakl 'she'll scratch you', J/q6ndral 'shoes'. No medial KKK clusters were noted in C, which is not to say that no ne exist, but the historical differences in the treatment of older lal and certain morphological differences yield , in the C parallels to the above examples, Iqondaral 'shoes', Itxarmesakl .'she'll scratch you', without KKK clusters. . (b) There are in M, but not in C or J, two dIfferent sorts of initial clusters with respect to anaptyxis. In JC, all initial KK clusters are non-separable, and the extra vowel required by a preceding third consomint or, optionally, by a preceding pause, must come . before the cluster: JCfkbii!'/ 'big', in pause optionally lekbig/, and Igas ekbigl 'a big head'; similarly the plural Ikbag/, lekbag/, Igus ekbag/. In M, however, clusters in such words as Icbirl (resulting fro m loss of OA/a/) are "separable," so that in the pause there IS an alternative form Icebir/; and in sandhi one usually gets Iras rebir/, but sometimes Iras ecbir/; the cluster in Ikl;>arl and similar words is, however, non-separable, hence always Irus ekl)ar/· (c) Geminate clusters are reduced to a single consonant when, through historical or morphological processes, they are immediately followed by another consonant: MJCfsalmul 'greet! (pl.)'. Word . final geminates are, in the pause, similarly reduced, so that JCfke11 'eat!' and JCfkelll 'all' are homophonous; the distinction retained in the notation is morphophonemic, and reflects the fact that gemination is restored not only with vowel-initial suffi xes Ukellul 'all of it') but also when followed by a word with initial vowel, including anaptyctic vowels: JCf~el ental 'you eat', contrasting with Ikell ahhadl 'everyone'. But while final geminates are treated alike in MJC, other final clusters are handled in considerably
I
55
PHONO LOGY
different ways in each of the dialec ts, and it is to thi s that we now turn.
3.52 Fi nal clusters. (0) In ge nera l, OA fina l clusters are sepa rated by a vowel in M and C with very few exceptions, whereas in J anaptyxis takes place on ly in certai n con sonantal environments. Thus:
OA
M
uxt qalb
'sister'
uxut
'heart'
sa hr
'm onth '
mill) katabt
'sa lt' 'you (m. s.)
ga!u!;> sahar meleil ktabet
J ext qalb saheg
mel l) ktabt
C exe t
qaleb sa heg melel) kat.bet
r
wrote'
One does find in MC so me final clusters of which the first element is a sono ra nt or semi- vowel : Ikartl 'a card', Imartl' Ma rch', Imaysl 'May' Imardl ' manly', but these seem to occur in relatively rece nt loanwords ; compare Imardl (Pers. mard 'man') with OA/ward/, which is M/ wared/, J/wagdl and C/ wagedl ' flowers'. InC, the anaptyctic vowel is lei but in so me words where M inserts lal, C has lal also, appa rently as a result of borrowing: C/ laham/ 'meat', Ibaharl 'sea', but Isahegl ' month', Iba'edl 'after' with the normal lei though the M forms have la/. [n M, the anaptyctic vowel may be lei, lal, or jul depending on the environment : 88 (i) If the vowel preced ing the first co nsonant is lei (irrespective of what it may have been in OA), the anaplyctic vowel is lei : Ice5ebl 'lie', ImeSel1 '[ike', Iheleml 'dream', Ise'erl 'poetry', etc. (ii) [f the preceding vowel is lal, the vowel sepa rating the consonants of older clusters is likewise lei provided these consonants do not constitute [u)-coloring or [a)-coloring environ ments: Icalebl 'dog', Iwaketl 'time', Isabetl 'Saturday', Ibaredl'cold', Isarej/ 'saddle', Iktabetl 'you (m .s.) wrote' and all such ve rbal for ms, since the ItI is neve r part of a [u)-coloring or [a)-coloring enviro nment, etc. (iii) [f the environment is [a)-coloring, i.e. if the vowel precedi ng the fo rmer cluster is lal and the first of the two final consonants is a Ihl, f'/. Igl, or Ihl, the anaptyctic vowel is la/ : Ilahami 'meat', Isaharl 'month ', Ibaharl 'sea', Iba'adl 'after', Imaharl 'bride p rice', Ira'adl 'thunde r', I baga!! 'mule', etc. (iv) If the vowel preceding the former cluster is lal and the two
I
1
56
COMMUNAL DiALECTS IN BAGHDAD
final consonants are [ul-colori ng (cf. 3.32 above), the anaptyctic vowel is lui : laru!)1 'gro und ', Iga!ul;>l 'heart', l'Darufl 'letter of alp habet', Ixamurl 'wine' , ;'al;>url 'patience', Iratu!?1 wet', etc. (v) If the preceding vowel is lui in M (whatever it may have been in OA), the anaptyctic vowel is lui : Ixubuzl 'bread', I~udugl 'truth', Isugu!1 'work', luxutl 'sister', Isukurl 'tha nks', etc. (b) So far in this discuss ion of final clusters, we have taken the diachronic point of view and spoken of preservation or splitting up of older word final clusters, and regarded anaptyxis as a histo rical process. These word final clusters, however, also have a synchronic existence in the present-day dialects, since under certain condit ions the. anaptyctic vowel fall s out and the consonants are again in contact. This occurs in sandhi with a following word beginning with a vowel (or with two consonants, and hence preceded by
an anaptyctic vowel) and also when vowel-initial suffixes are added: M/ la'Dam/, but Ila'Dm ejmiill 'camel meat', /Ial!makl 'your meat', / Iai)mayaj 'small piece of meat'. The. anaptyctic vowel is retained if only a single consonant follows, whether in a suffix or a separate
word: Ilaham zenl 'good meat', Ilal!amha/ ' her meat'. The differential treatment of final clusters jn J as against M and C is thus a matter of pausal treatment only: the anaptyctic vowel appears in J as soon
as the word in question is fo llowed by a si ngle consonant: J/kalbl 'dog', but Ikaleb zenl 'a good dog', Ikalebkeml 'your (pl.) dog'. Thus despite the fact that Icalebl does not conta in a phonetic or a phonemic cluster whereas J/ kalbl does, both may be sa id to contain a "morphophonemic cluster"-lb-which
i~
actualized sometimes as
Ilbl and
sometimes as Ilebf under statable cond itions. Such morphophonemic clusters often correspond to OA phonemic clusters, but not always: M/rajelj'husband' < OA/rajulj, behaves precisely like /Caleb/: Irajlil 'my husband', Irajelhal 'her husband', l rajl ellaxl 'the other husband', IrajeJ laxl 'anot her husband'. We are thus justified in saying that Irajell contains a morphophonemic c1usterj1~and this will be one of th'e criteria whereby we will be able to dist in guish such .a noun as / lal)am/ 'meat' from such others as / xa'r)ari 'piece of news': the first conta ins a morphophonemic cluster(un- and the second has no such cluster, its second /a/ being stable. In the -morphology thi s will be put to use in such matters as classifying words by "bases" and iil simplifying statements regarding affixation.
4 MORPHOLOGY 4.1
PROCEDURE AND NOTAT ION.
4. 11 Basic morphophonemics.
(a) The terms "root" and "pattern " are here used in the traditional manner for the familiar discontinuous units of Semitic morphology. Thus MJCjbladl 'country' has a root bid, of which the first "radical" or K" is b, the second (K,) is I and the third (K,) is d; the same wo rd
has the pattern K , K,aK, or, using the traditional symbols for the rad icals~ the pattern qldl. In the classification of wo rd s by patterns, we shall refer less to the pattern of the actual phonemic word than
to the pattern of an underly ing fo rm or ' base'; thus J/kalb/, 'dog', Ik:llebnal 'our dog' , Ikelbenl 'two dogs' all contain a base kolb-, with or without modification of the base pattern in the uninflected
and inflected forms. (b) Many roots , patterns, and bases have more than one phonemic sha pe as they go through va ri ous derivational and inflectional processes. We therefore represent them by a morphophonemic notation , using symbols th at are , by and large , the same as phonemic symbols and that suggest th e phonemic shape of th e norm. Various rules then make it possible to arrive at the remaini ng phonemic shapes. The morphophonemic symbolization is in ita lics and unenclosed, the phonemic notation being enclosed in the customary slashes.
Thus Je/axxasl 'dumb', fern. J/xegsa/, Cjxagsal contain a root xgs or which K, is always Ixl and K, always lsi, but K, is a g wh ich is now Igl and now Ixl according to statable conditions. The elements of the patterns also show morphophonemic alternat ion: the masculine forms contain a pattern aqla/, of which a is always /a/. but Ii
is generally lal, but becomes Iii when the patlern is grafted on a root withK,- y:" JCja' mi/ ' bl ind', fern. J/,emytt/, Cj'a mya/. roo t '/I/y. As to the patterns conta in ed in the feminine forms, we could say tha t the J forms have a pattern qellii and the C form s a patter n qat/ci, but I have preferred wherever possible to keep the notation inter5)
•
58
COMMUNAL D IALECTS IN BAG HDAD
dialectal, as in the present case; we th erefore say that both the J and C feminines have a pattern qalla, tho ugh the pattern element a subsumes different alternants for C and for J. Finally, observe the more complex but highlY regular alternations th at take place if one of the radicals is IV o r y: fro m the roo t ·sIVd we get JC/aswadl ' black' ; the phonemic symbols are the same as the m o rphopho~e~lc ones, and we say there is "no modification ." Ho wever the femtnme~ are J/sud!i, Cjs6da/ ; we say that in J the pattern element a of 'lalla com bines wit h the rad ical \I' of slI'd in an unstressed syllable to Yield lui ; and that in C the same combinatio n yields 16/. (c) The phonem ic value of most of th ~ m o rphophoneme~ used in representing roots, patterns, and bases Will be specified pan passll wi th the discuss io n throughout ,his chapter. A number of the more common o nes are give n here . A radical' is 1'/ internall y, initially zero be fo re V and I'r before K. A radical y is Iyl when initial, when followed by another Iyl or when the pattern requires gemination, when followed by a vowel a nd, when final, if preceded by a long vowel: MJ Cjyabbasl ' he dried' (ybs), Isayyabl 'he abandoned' (syb), lazyadl 'more' (zyd). MCjI)effaYI 'barefoot (pl.)' (lify)· It IS zero If fina l and preceded by a pattern element a (JCjbanal 'he built', root bllY, pattern qalal) and Iii when final and preceded by any o th er pattern element (J CjI)aki/, ilky, pattern qall-). When preceded by a vowe l pattern element and followed by a consonan~al a ll:: It, ordinarily merges wi th the vowel into lei or Iii, sometimes lal (111 J also the equiva lent short vowels when unstressed), a~co rding to rules that are b·est deta iled when dealing with the indiVidual patterns. A rad ical IV merges into 16/, lui o r lal in the positio ns just mentioned; in other positions it behaves a nalogously to y, but it yields Iyl rather than Iwl in some patterns. As to symbols fo r pattern elements, no te especia ll y a, which in J nearly always yields lei in unstressed closed syllable, zero in unstressed open syllable, lal otherWise, except when merging with a following y or '" as o utlined. Note also e. wl11ch nearl y always yields zero in unstressed open syllable III MJC, othe r~ wise lei in JC , lei or lui in M depending o n the fl ankIn_g conso nants. M/kateb/ , Iwag ufl (pa ttern qalel), fern. Ikatba/, Iwagfal (pattern qatel + a); JCjketeb/, fe rn. Iketbi/, Iweqfi/. Pattern elements Written as lo ng vowe ls yield sho rt vowel pho nemes ill final unstressed position in M1C, in all unstressed positions in J. . . . (d) Other terms and sy mbols. Short morphemes In which It
MOR PHOLOGY
59
wo uld be useless to ,speak of roots and patterns a re, when necessa ry, designated by a simple formula, identica l with the phonemic symbolization of its most common a llo mo rph o r of its und erl yi ng form (i.e. th at from which the others can be prod uced by fa milia r rul es). Thus the pron. suff. -lla and -Ita a re represented morph opho nemically with lo ng vo wels,' · subsuming the stressed forms -/na/-and-/ha/as well as the unstressed -/nal and -/ha/, the quantity change with shift of stress being automatic. The symbo l T to subsume all the allomorphs of the fe rn. suff. is, on the ot her hand , pu rely ad Itoe, as are a number of others, e.g. L for the morpheme 'to' a nd L ' for the relative pronoun (q . v.). The use of the plus-sign in in flectional or deri vati o nal fo rmulas (cf. examples in precedi ng paragraph) is selfexpla natory. As for the grammatica l categori es and the cl assification of the parts of speech, I ha ve more or less stuck to the trad itional terminology , which seemed adequ ate fo r the purpose. 4.12 Overall vie w. There is little or no differentiatio n as between M, J, a nd C as to the overall structure of the respective morphologies . There is, however, considerable differentiatio n as to phonemic shape of mo rph emes, res ulting o nl y in par t fro m th e reg ula r phonetic corresponde nces d iscussed in the previous c hapte r; and there is also
considerable differenti atio n as to deg ree and kind of morph opho nemic alternation. The greatest differentiation is perhaps that fou nd in the inventory and behavio r of the prono minal affixes (4.2) and in the allo mor.phs of the feminine suffix (4.3) . The d ive rge nt morphopho nemics of the noun, particip le, numera ls, and ve rbs are, by and la rge, predictable from considerat io ns a lready outlined in Chapter 3, but a rather fu ller analys is wi ll be fou nd in the present chapter. Not unnatura ll y. a good many point s di scussed here are lexical
rat he r th an mor phological in any st rict sense. 4.2 THE PRONOUN. 4.21 Subject pronouns. (a) The free subject pronou ns, th o ugh lexical items with little or no role in the morphology, may first be . disc ussed here. Differentiation is, here again, more twofold (JC vs. M) than threefold: M
Is. 2m.s. 2f.s.
J
ani
ana
cnla
cnla enti
en ti
C ana cnla enli
•I
•,
60
COMMUNAL DIALECTS IN BAGHDAD
M 3m.s. 3f.s. Ipl. 2pl. 3pl.
huwwa hiyya e~na
entu
humma
J
huwwi hiyyi nei)na entem hernmi
MORPHOLOGY C huwwa hiyya nei:lna
entem humma
The 3pl. C form is suspect because of the l ui , and may be a loan from M ; Mosul has the rather unusual form I hiyeml , and some such form may have existed in C a nd later been suppressed. One C informant cited (but does not use) the form Ihemm il, as in J. Somethi ng similar may be said of C/huwwa/, cf. the unusual Mosul form . Ihinu/ 9 0 (th ough Mosu l informants also use Ihuwwal). In the 2nd plur., JC go along with the qellu-dialects as a who le and M with the gelel-diale"ts, as is true also of the 1st sing., though some geleldialects have Ifmal, a form wh ich may be the precursor of the present lani I, with the fiI resulting from 'imdla or, more likely, from analogy w'th the pron.suff. A similar form is also found in the !;loran and a good many Beduin and other dialects; 'Ana, though havi ng a qeltu-dialect, "Beduinizes" in this instance as in many others and has lani/. For the 1st plur. JC also have a (rarer?) varia nt lehna/ ; form s with initial Inl are common throughout Ihe Arab world and are attested for Iraq as early as the twelfth century.9l As for the masculine-feminine distinction in the 2pl. and 3pl. which is a hallmark of Beduin dialects!' the rural gelel-dialects have forms like lentul a nd Ihummal for the masculine vs. such forms as lenta nl and Ihennal for the feminine ; my M informants have heard such feminines , but do not use them and characterize them as provincia1. 93 Finally, there are traces of shorter forms of the 3m.s. and 3r.s. pronouns, apparently on ly as enclitics attached to a few particles: M/ liihi .. walahi ..1 'she is neither ... nor .. .'; M/yiihu/ , fern. Iyahil pI. Iyahuml 'which ?' ; J/wen i/ 'where is she ?', though Iwenul can be interpreted as Iwenl plus the ordinary suffix, as in Iwe nakl 'where are you ?'. (b) The bound subject pronouns attached to the perfect base of verbs are listed below; the allomorphs appea rin g in the table are those that occur whe n no further suffixes are added: for alternants, see below.
M Is. 2m.s. 2f.s. 3m.s. 3r.s. I pI. 2pl. 3pl.
-ItI -ItI ':"-/til zero -'.i/atl
-/nal - /tul - /aw/"4
61
J
C
-/tul -ItI - /til zero - /etl - /nal - /teml
-lui
-/tu -ItI -Itil zero
-/etl -/nal -/teml -lui , - 101
These affixes call forth the following remarks: (i) The affixation of -ItI to most ver bal bases creates a morphophonemIC fina l cluster that behaves as outlined in 3.52: sal/am + I :-'. J/sellamt/, MC/sallametl 'you (m.s.) greeted'. Other consonant IIllttal suffixes have similarly predictable effects, largely as to posihan hof stress. . . . Vowel-initial . . suffi"es create open syllables , zeromgout t e precedlllg a (WIth vanous predictable side effects) in J, optionally III M, but not at all in C: sal/am + el ..... l lsalmetl C/sallamet!' sal/am+al ..... M/salmatl or Isall amatl 'she greeted'. ' , . (II) Bas~s 'end ing in - y undergo a number of further changes wIth the vanous suffixes, but these will be discussed in 4. 8 when de~lmg wIth the verb. In C, the 3 pI. suff. -lui has an alternant -/01lIl"such ve rbs : C/qarol 'they read' (root qry). (III) Suffixes ending in a vowel have stressed a lternants with a conco mitant change of quantity (in given cases also of quality) and the same holds for M -law/. These stressed a lternants occur when further suffixes are added, and have largely predictable effects on the base. The table shows the shape of all subject pronoun suffixes when add itional suffixes are add ed; items in parentheses are those III whIch the shape is the same as when no add itional suffixes are present; when two forms are given in 1, the second is that occurring before - lid, the first before all other suffixes . M Is. 2m .s. 2f.s.
UtI) Utf) l lil
•
J
C
Ito/, /tu/
llUl (ft/) l lil
(/t/)
Itil
62
M 3m.s. 3f.s. Ipl. 2pl. 3pl.
63
MORPHOLOGY COMMUNAL DIALECTS IN BAGHDAD
(zero) (fat!) /na/ /lU/
/0/
J (zero) (fet/) /na/ (ftem!) /o/ , /u/
C (ze ro) (fet/) / na/ (ftem/) / iJ/
The 3pl. alternant - /0/ in C (with K,-y verbs) remains unchanged. Some examples: JC/jebtu/ 'J brought', J/jebtonu/ C/jeblUnu/ 'J brought him', J/jeblUha/, C/jebtuwa/ 'I brought her' (on J and C allomorphs of - I/{i, see 4.22a below); MJc/jebti! 'you (f.s.) brought', /jebtlna/ 'you brought us', /jebtiln a/ 'you brought to us'; Mfjabaw/ 'they brought', M/jaboni/ 'they brought me'; Jc/jabu/ 'they brought', J/jaboni/, c/jabuni/ ' they brought me', J/jabUha/ , c/jabuwa/ ' they brought her'. (iv) The alternation /0/ ~ / u/ in J makes one wonder whether the stressed form is not older, having yielded /u/ when /0/ , perhaps harking back to /inv/ , was no longer admissible in unstressed syllable (cf. 3.37) . Thus '/jabawni/ > /jaboni /, and '/jabaw/ > ' /jab6/ > /jabu/ . Simi larly, the / 0/ ~ /u/ and the /e/ ~ fiI altern ation found in base fi nal syllables, e.g. in the imperative (g. v. in 4.8 below): in stressed syllables, as in K J-y verbs of Form I, we get /xbi/, /xbe/, /xM/ 'hide (m .s., f.s. , pl.)', /xbinu/, /xbenu/, /xbon u/ 'hide him'; in unstressed final syllable, e.g . K,-y verbs of Form II, we get /sawwi / ' do ·(m .s. and f.s.)' and /sawwu/ 'do (pl.)', but with suffixes (and hence stress) /suwwinu/ , /suwwenu/ , /suwwonu/ 'do it (m.s., f. s., and pl.)'. This suggests the possibility that whereas ihe masc. sing. form /sawwi/ may be from */sawwi/, the fern. sing. may be from */sawwe/ < */sawway/ , a nd so on. This leaves a good many questions unanswered, e.g. as to the /0/ ~ /u/ alternation in the 1st pers. . sing. suff. and as to the special alternants before - hii. (v) The 1st pers . sing. suff. - /tu/ is, of course, one of the distinguishing features of the geltu-dialects, not only in contrast to the gelel-dialects, but as agai nst the Arabic dialects as a whole, since all but t hese have lost the final /u/. It is retained in JC, Mosul, 'A na, and the Anatolian dialects, but not in Hit, where its loss may be one of many gelel-influences. That the gellu-dialects have, here again, preserved a form which was once more widespread
is attested to not only by the evidence of Classical Arabic, but also by data on older .v ernacu lar Arabic. In the fourteenth and fifteenth centuries, in fac! ,!. this final /u/ of the 1st pers. si ng. of the perfect was sti ll (as it must have been earlier) one of the hallmarks of Baghdadi speech and of Iraq i colloquial in general"' This retention of -/tu/ in the 1st sing. goes .together with a similar retention in -/tem/ for the 2nd plur., whereas its replacement by -ttl often goes together with - /tu/ in the 2nd plur. , as in M. Once more, M lacks the masc.- fem. distinction in the 2pl. and 3pl. found in the rural and Beduin dialects. The M 3pl. suff. -/aw/ is typical of the gelel-dialects as a whole and found in Beduin dia lects elsew here, e.g. that of the Negev seminomads. (e) The subject pronoun affixes attached to the imperfect base are listed in the following table. In the C col umn, where two forms are given, the second is that occurring with K J- y verbs, the first that occurring with all others. On the optional M alternant -/ha/ for -that , see 3.31 and note 99 below. M Is. 2m .s. 2f.s. 3m.s. 3f.s. I pI. 2pl. 3pl.
J
/a//t//t .. ·in/ /y//t//n//t .. ·un/ fy .. . un/ ,
.
C
/a//t/"/t .. ·en/ /y/-
/a//t//t ... in/ , / t .. ·en/
/t/-
/t/-
/ n/ /t .. 6n/ fy .. ·on/
/ n//t .. ·un/ , /t .. ·on/ /Y"'iJ n/, /y ... on/
IY/-
These affixes call "forth the following remarks: (i) The in itial morphophonemic cl usters created by the preposed consonant a~e resolved in conformity with rules outlined in 3.52, viz. KK clusters are unseparated, KKK clusters yield / KeKK/, whereas KKKK duste rs yield / KKeKK / in MJ but / KeKKeK/ in C: I+ndm .. MJC/tnam/ 'she sleeps'; I+kleb .. MJC/tekteb/ 'she writes' and since the e of the base is zeroed-alit when the post posed vowel changes the syllable structure (4. l le), we get I+kleb + iill, whence M/tketbiin/, J/tketbon/ , bllt C/te ktebunf. On initial /yK/ clusters, cf. 3.22e. The proposed /a/-, though a lways unstressed , is never altered in J.
64
COMMUNAL DIALEC TS IN BAG HDAD
65
MORPHOLOGY
(ii) The Inl of the post posed elements is elided when additional suffixes are added" a nd the rema ining lo ng vowels undergo, in J, modificatio ns before - I1a parallel to those outlined above: M/tketbiin/, J/tketbon/, Cftektebiinl 'you (pl.) write', but M/ tketbiila/ , J/tektbolu/, Cftektebu lul 'you (pl.) write to him '; M/tketbiiha/, J/tketbiiha/, Cftektebuwal 'you wri te it (f.)'; cf. also J/tketbenl but Itketbihal vs. Itketbelu/. (iii) T here is considerab ly less differentiation in this set of prono minal affixes than in th ose occurring with the perfect. All three dialects have the retention of fina l -/nl th at is typical of the Mesopotamian area and is found also in the Beduin, Peninsular, and Centra l Asian dialects. Jts retention in Mesopota mia , as opposed to its earl y disappearance in oth e" areas, is attested in "Middle Arabic" tex ts and recognized as a ha llm ark of the Iraqi vernacular by fourtee nth centu ry so urces 9 ' Like so me of the other qe/Ill-dialects, C has diffe rent postposed elements for K,-y verbs, viz. -/enl and -/onl vs. -/inl a nd -/iinl for all o th ers; both M a nd J ha ve a single set, but .I is the o nl y di alect so far no ted that seems to have generalized endings pr oper to K ,-y verbs;" a fact tha t is perhaps connected to the J a ltern atio n lei - Iii a nd 101 - lu/. Here again M lacks the masc.-fem. contrast in the 2pl. and 3pl. fou nd in the rura l a nd Beduin dialects, though some M speakers seem to have it. 4.22 Object pronouns. The pron. suffixes postposed to particles and no minal bases and to ve rba l bases enlarged wit h subject pronouns are listed in the following table. Where two forms appear, the first is the one occurrin g after base-fina l V, the second that occurring after base-final K ; however, o n the forms with or without Ihl in C, see (ii) below. M
J
C
- ya, - ) - k, - ak -e , - , -a - ha - na - kum - hum
-yl, - i
Here a re some details on these pro nouns: (i) In the I ~t · pers. sing. th e alternant --/nil occurs after ve rbal bases, but also, a~ in other dialects, after a small number of particle bases: MJCfjabnil ' he bro ught me', M/ba'adni/, J/ba"ad nif 'I still .. ·', J/lenil ' I have', J/llenil 'to me'. (ii) In C, the ,.3r.s. has the s hape - /hal a nd the 3pl. has -/heml o nl y if preceded by - /a/ : Cfjebn ahal 'we brought her', Ijebnaheml 'we brought them'" Ixall ahal 'he let her', Ixallaheml 'he let them'. Otherwise there is -/a/, -/em/, respectively, e.g. Ijabal 'he brought her', Ijabeml 'he brought them' ; this is true also after vowels other tha n lal, except tha,t in such cases the re is an a~tomatic glide Iyl o r Iwl depending on the base-final vowel : hi + "Ii ~ Ibiyal ' in her'; abu+hd ~ labiiwa/. ' her father '; similarly Ibiyeml, labiiwem/. In J, on the other hand , the Ih/-Iess forms occur o nl y after K , viz. JJjaba/, Ijabem/, but labuha/, labiihe m/ , Ibihal, Ibjhem/, etc. Note aga in J/jl before -Ita where lei occurs before other object suffixes: J/, Ienal 'on us', vs. /, Hha/<, o n her', cf. such forms as Itketbjhal and Ijabuhal (vs. Ijabona/) cited abo ve . (iii) In M a nd C, there is a single form for 2r.s. whethe r ·the base ends in V or K, viz. M/el a nd Cfkif, while J has Ikl for bases ending in K and Ikif for bases endi'ng in V, with predictable anaptyxis and/or loss o f gemination in morphophonemic clusters; using -k i as an interdialectical symbol for the 2f.s. morpheme, we get :
abiHki biil + k i 'amm +ki lIa/s+ki
M
J
C
abue betee 'ammee nafsee
abuk i betek
abuki betki
'ammek
'amki nafeski
nafsek
'your (f.s.) fathe r 'your (f.s.) house' 'your (r.s.) uncle' 'yoursel f (r.s.)'
Similar considerations obtain in bases terminati ng in the fe minine Is. 2m.s. 2f.s. 3m.s. 3r.s. Ipl. 2pl. 3pl.
-k
-yi, - k, - ak - ki
- nu, -u
- nu, - u
- ha, - a - na - kern - hem , - em'
(- ha), - a - na - kern - hem, - em
- k, - ak - ki,
- )
mo rpheme T, though for details, see 4.33 below. The behavior of M/kuml and JCfkeml is analogous to that of C/ki f. (iv) In the 3m.s., the alterna nt written 1'1 in the table for M simply means that the 3m.s. object suffix afler bases ending in V or Vy takes, in M , the shape of a zero that has the effect of a consona nt add ed to the base, so that the word ends in t he appropriate stressed long vowel.: using H as the generalized symbol fo r the 3m.s. object pronoun , we get:
~,
C'
66
COMMUNAL DIALECTS IN BAGHDAD
I,
M
J
C
abunu binu
abunu
b;+ H
abu bi
'omm + H
'am ma
'ammu
'ammu
abt'+H
binu
(v) It can be seen that J a nd C are, as rega rds object pronouns, much more similar to each other than either is to M. Apart from the treatment of Ih/-Iess alternants of the 3f.s. a nd 3pt., the only difference between 1 a nd C is the in varia nt - /kil of C as agamst Ikl - Ikil in 1. In this respec t, C goes with Mosu l, M ardin , and the Anatol ian dialects, as well as th~ Central ASla~8 dialects, suc.h an inva riant - j kij is altogether rare;: In other areas. The q~/(u-d l alect of 'Ana also has an in va ri ant 2f.s. suffix, but it is lei as m M: and is one of the several gelet feat ures o f this dialect; a,nather one IS the ze ro alte rnant o f the 3m.s. suffi x as in M , viz. labu!, though the alterna~t after K is lui as in JC a nd all o th er qelru-dialects.'" All th~ qeltu-dialects, including 'A na, also show Ih/-Iess a ltern ants of -"a, so me more like 1 a nd others more like C . Th,s IS true also of the Central Asia n di a lects; fo r the 3m.s. some o f the latter ha ve lui, othe rs la/. The m.S. alterna nt after V is !nul in Mosul, some Central Asian dialects, as in lC, but Ihul in the Anatolran dIalects. The geier-d ialects go along wi th M, exce pt that the rural and Bedum dialects · have a ' masc.-fem. distinc ti on in th e 2pt. and 3pt.; some Beduin dia lects have -l ui for the 3m.s ., but most geier-d Ialects have - /al as in M. The retention of Ihl in -lui is a hallmark of BeduIn and Beduinized dia lects. . 4.23 Indirect and doubl e object prono uns. (a) The three dIalects are not d ifferen tiated as to the marking of th e Ind,rect pronom inal object. All three postpose to th e verbal base the morp~eme L followed by object su ffi xes: MJCfjiiblil 'he bro ught to me. The allomo rp h of L in suc h cases is, in all three dial ects, always 11/, W~h anaptyx is whe re app rop nate and optrona l reph:e men~ of /II y . Inl before - lid: jab + L+ ltd yields M/jabel ha/ , JC/J ablal he brought to her'; jdb+L+ lld yield s MJCfjii belnal o r Ij ii benn al .'he brought to us' . The non-enclitic allom orphs of L are, however, dlfferen~Iated . MCfel/- vs. lll1e/- ; for detail s, see 4.93a below. The a ffixatIon of L ca uses no particular modificatio ns in verbal bases except In t~~t of th e verb 'to say', q. v. in 4.82/ below. J has a spec ial non-enclrttc
I (
I
MOR PHOLOGY
67
form denoting possession I le/- rat her than Ille/-, and we may be dealing with an entirely d ifferen t morpheme; see 4.93 below. (b) With a direct and an ind irect object pronoun, the d ialects
have two procedures', One is to affix the usual direct object pronoun to the verbal base, fQllowed by the non-enclitic a llo mo rph of L plus pron. suff.: M/jiiba eli/, C/jiibu eli/, J/jiibu lIen i/ 'he brough t it to
me'. This seems to ·be the more infrequent of the two construct ions, and may OCcur onl y when some special emphasis is placed on the indi rect object. The second procedure is 10 affix the enclitic allomo rph of L plus pron. suff. to the verba l base, fo ll owed by a reflex of OA /'iyyii/- plus the pro n. suff. referr ing to the di rect object. Some such co nstructi on is employed in all three dialects, yet they differ marked ly from one another in the detai ls. (i) In C, th e indirect o bject is affixed to the ver b in the usual fashion , a nd th e d irec t obj ect follows in th e gu ise of Iyiil plus pro n. s uff.: Cfjiibli yii nul ' he brough t it (m.) to me' ; Ijiiblak yahal 'he bro ught it (f.) to yo u (m.s.)" Ijiibl u yii hem/ 'he brought th em to him', etc. In this way C, unlike M and J, makes all the ordinary number, gender, and person distinctio ns in both objects. (ii) 1n M , L is attached as in C but th e pron. suff. a ttac hed to it exh ibit so me spet:ial a ll omorphs a nd a re fol lowed d irec tl y (i.e. witho ut any sembla nce of word boundary) by liya/- plus the pron. suff. referrin g to th e direct object : M/jiibl iyal 'he bro ught it (01.) to me', jjiibelkiyiihuml ' he brought them to you (m.s.)' wi th Ilkl instead of flak/ ; Ijiibel':iyiihal ' he brought it (f.) to you (f.s.)" cf. liiible':l and th e d ifferent position of th e a nap tyc ti c lei ; Ijiibelhiyii hal 'he brought it (f.) to ' him' o r 'to her', with I lh/ instead of Iial and in stead of Ilha/. TIA,; th ere is a pa rtial blurring of th e gender d istincti o n in the direct object because of this infixed -/lh/-: in the mascu line, th is fhl may be a remnant of an older pron. suff. - /ah/ of the 3m.s. , st ill to be hea rd in some dialects. (i ii) In J, there' is an unusua l res huffling that results in an infi xed -/yii/- preceded by L without pron. su ff. an d followed by the pron . suff. referrIng to the ind irect object, with no prall. 5uff. referrin g to th e direct object: J/jabelyii kl 'he brought him, her, or th em to you (m.s.)', II.l kitulYii nul ' I to ld it (th em) to him', Ihkitulyahal ' 1 told it (the m) to her', Istgitulyiiheml ' I bo ugh t it (the m) for them'. Th lls th e indirect object is specified in the normal way as to ge nder, person , n umber, but the d irect object completely un-
.. I
, i
;:
68
COMMUNAL DI ALECTS IN BAGH DAD
MORPHOLOGY
specified. Side by side with these very com mo nly used forms, I have more rarel y heard a form reminiscent of the M constructi o n (albeit in persons who do not otherwise show M influence), viz. / kemmeliyll/ 'he fu lfilled it for him' /ake mmelki yi/ ' I shall fu lfill it fo r yo u (m .s.)', instead of / kemmelyii nu/ and /akem melyak/ . Note the normal vowel changes wroug ht by the stressed -/ya/- upon the ver bal form s invo lved, the respective suffi xless for ms being Ija b/, / \:1ketu/ , /stgetu/, / kammal/, /aka mmelf. (e) My data on pa ra llel constructions in the other Mesopolamian dialec ts are insufficient for a fu ll-fledged compa riso n. The C construct ion is co mmon throughou t the Syrian area and seems to be
found in 'Ana; Mosul , however, has so me special features of its ow n: /ta'Wuw",a/ 'you (m.s.) gave it (m.s.) to him (her)" /ta'etIi yya/ 'you (m.s.) gave it (f) to him (her)'. The J construction is, so far as I know. unique ; the sa me may be true of so me features of M though one would expect the gelet-dialec ts to show so me si mila ri ties to it. 4.3 THE FEM ININE SUFFIX. 4.31 Overall view. One of the features that most clearly distingu ishes the dialects from each other is the treat ment of the mor pheme T, viz. the refl ex of OA - aCt), usua lly ca lled the feminine suffix. The three dia lects diffe r but little in the use made of th is suffix in inflection a nd derivation (though cr. 4.34); they do differ co nsiderably in two ways: in the phonemic shape of the a llomorph of
T
in the isolated word and in the phonem ic
shape of the a ll omorph occu rrin g in sandhi . On the special case of the nume ra ls, see 4.6 below.
4.32 Allomo rphs in the iso lated noun. (a) In the isolated feminine nou n ending in T, M words always have _/aj99 whereas J and C have either -/i/ or -/a/, dependi ng o n the structure of the base, th o ugh the conditioni ng factors are very ditferent in the two dialects . Here are some illustrative examples, to be fo llowed by a detailed discussion:
'bitch' 'egg' , 'trifl ing' (r.s.) 'big (f.s.)' 'snake' 'brolhers
M
J
C
calba bei,la i;la,ila cebira I)ayya exwa
kalba bei,l i i;la,ili kbigi I)ayyi exwa
kalbi beQa 9a~ita
kbigi ~ayy i
ex W I
69
Note that besides t,hese voca lic a llomorphs, all three dialects have a zero allomorph that acts lI ke a consonant, i.e. preserves the len th ~nd stress of the base vowels: MJC/meshll/ 'shovel' MC ~;\ prayer', J/$!fi/ 'synagogue'; MCfmda ril/ , J/";darAI 'care.'. and ~':' I others; that these w~rds end in T is decided by the fact th~t the of the zero allomorph IS taken by -/t/- in sandhi, J/ , Ifi t lekbigi/ 'th Great Synagogue' . "On / bibi/ 'grandmother', /'abal 'ma ntle' se: 4.33b ~elow. As for the vocalic a llomorphs, the a lternation la/-/i/ IS, In ,controlled ' by the following factors: (I) A base final emphatic or back consonant Ukl and any thin g forward of It are front consona nts) except lei a lways require -/a/ Iwesxal 'dIrty' Ida I ' , - - . . .' , ," ~qa, narrow, jmalei)aj 'l1lee', jga fi 'aj 'thin', l I!ab_ha fron t , Ihaya/ orga niza ti o n', /qe"a/ 'story' / henta/ 'wheat' , '. , / agl(;Jaj 'broad '. --:
Pi;';:
b f (ii) Base final /r/ , th ough rare, seems to req uire -Iii if preceded y ront vowels, -/a/ If not : I mudiri/ 'directo r (f.)' vs. Isayyaral 'car" slmIiarly Ig'l th t . fl· f ' . , a IS a re ex 0 OA Irl requires -/il o r -/al according to fro~t vs. non-front enviro nment: Ikbigil 'big', Izgaygi/ 'smal/' /ebgl/ need le'. but Imagal 'woman', 1,6gal 'picture'; the few exam ple; of b~se_ fin,~1 /g/ that IS not a re fl ex of Irl aU have -/a/ : lrargal 'empty' /qa50gal spoon'. · , (iii) . With no?-back, non-emphatic base fi na l consonants there IS ~/I/ : Iqebbll roo m', 15effil 'lip', Idsut i/ 'pots', Iba'idi/ 'distant,' 1"_aJsll Impure', Izelzil 'naughty', Idesdfisil ' loose overgarment,' !JeJII ~h:n ', Isamakil 'fish', Isallil ' basket', Isani/ 'year', I kelmir 'word,' Il;elwll pretty', Imiyyil ' hund red'. ' (iv) In a good ma ny wo rds whe re o ne wou ld expect -IiI, there -/a/ (but appa~ently never the reverse); some cases are clearly d ue to _borr~wm~ : Ilopa! '~al/', Iferca/ 'brush', I kleca/ ' ki~d of pastry', Ihay5al cow, /Iagwal trouble'; there is -/al also in so me ki nShip terms where -N wo uld be expected: Ixalal 'maternal aunt' /' amma/ 'paternal aun t', Ijed?al 'grandl]'lOther'IOO ' (v) Fmally, there is hesitation after base final 1\:1 / preceded by front v~wels: one Ihform ant says /ge\:1i/ 'smell' where the others have /ge\:1a/ , but one of the latter reports his mother says /geh i/, SImIlarl y, Igayl;l/ arid /gayl)al 'going (r.s.)'; the fo rms with ~jij' perhaps the older o.n~", are closer to the J pattern. As for the facto r; condltlonmg the /I7-/al altern ati on in J, they are as follows: ( I) After av y base fina l consonant except Iw/, there is -IiI is
70
COMMUNA L DI ALECTS IN BAG HDAD
if the base final syllable contai ns IY/, Iii or Ie/: Imiyyil ' hundred', Inaymil 'sleeping', Igayl)iI 'going', i¢ayqil ' narrow', Icerpayi/ 'bedstead', Izgaygil 'small', Igil) i! 'smell', Iwsi'il ' broad', Ikbig il 'big', /'i!,i¢il 'wide', I!wilil 'long', Ifa qitil ' poor', Ijijil ' hen', Ijimdil 'frozen', Imigq il 'rotten (egg)', Isminil 'fat', Izenil ' nice', Ibe¢i/ 'egg', Imedil 'table set for the Sabbath', Ilebsil 'wearing', Ise m'il 'hearing'. (2) The same applies if the base final syllable contains e that may be zeroed-out through suffixation (this holds fo r some of the above exa mpJes as weJl): / zalziJ 'naughty' (base zalez-), Iwe~x i/ 'd irty' (base lI'e~ex-) and especially all the forms of the acti ve participle: not only Ilebsi/, Isem'i/, etc. as a bove, but a lso Imkamlil ' having completed' (base mkCllllmel-), Imtal'il ' having take n out' (base lI1!olle'-), Imsflml)il ' having forgi ven' (base II1sam eb-), and so throughout (see 4.7 below); other cases of e in the base final syllable must be viewed diachronically: leb!!,il 'needle' but Isejgal 'tree' where o'A/, ibral vs. Isaja ral explain the different treatmen!.lol (3) A slight ly specia l case that is actually in keeping with the stateme nt in (i) a bove is that of nouns on the patterns qiitel a nd qitel with K,-y; thi: corresp onding feminin es still end in - Iii as ex pected, but slo not differ phonemically fro m the masculine form s: Imesil 'walking (m.s. and f.s .)" Ibeni/ ' building' , Il)ifil ' bare-foot', /'i lil ' high' for both the m.s. and the f.s. (4) In a ll other cases, i.e. when the base final syllable is other than outlined a bove, the allo morph is -/a/ : Iqebbal ' roo m', Igekbal 'k nee', Imotal 'a death', /iem'al 'Friday', Iwaqqal 'leaf', Ids"tal 'pots', Isanal 'year', Isall al ' basket', Ihasal 'cow', Ileqmal ' morsel', II)agal 'hot' , Iweccal 'page' a nd of co urse all passive participles except those with K ,-y : Imkamlal 'completed' (base II1kommol-) , Imta!'al 'taken out' (base m!alla'-land so on ; base fina l ", seems to require the -/al allomorph no matte r what the preceding vowel: Ilabwal ' lioness/', I balwal 'genius', Ilagwal 'fuss', I I)elwal 'pretty', but also l illewal 'pretty', lexwal 'brothers' , Imiwal 'frui t', Idewal 'fema le demon'. (5) There a re a few insta nces with -/al whe re one might have expected -Iii (t hough, as in C, apparently no examples of the reverse): Ibental 'wheat' « ' /l)unta/ ?), /, aylal ' famil y',1 sarikal 'company' (probably loans), Iswayyal 'a little' and Imlil)a/, m. Imlil)1 'nice'; I 02 cf. a lso plu ra ls on the pattern of Ibgaddal 'Baghdad is', despite the /iI of Cl. Ar. Ibagad ida/ , and the discussion in 4.53 below.
,.
71
MORPHOLOGY
(b) The inva ria nt fem . suff. -/al found in M is also fo und in all the other gelel-dialects that have so far come to my attention, as well as in 'Ana and ,Hit, the only qellu-dialects that have no 'imala whatever, and in the Centra l Asian dialects. As already mentioned in 3.36, all other ,qf/lu-dia lects as well as most Syrian sedentary dialects, have a n alternation resembl ing that of C. The Mosul dialect differs from C in at )east one respect, namely in th at there seems to be always -/al aften Ig/, even when it is a refiex of OAfr/ ; thus Ikabigal ' big', Ikeigal 'abundant', vs. C/ kbigi/, Iktigif. The J alternation, which depends essentiall y on the vocal ic co loring of the base final syllable regard less of the quality of the base fi nal co nso nan t, seems unique" It may not be irrelevant to point out that in ot her dialects which haye this allomorphic split there is often a J-Iike conditioning in bases ending in r (now phonemically often split into Irl and /r/). Thus North Pa lestinian Ikbiril ' big', lfatril 'lukewarm', vs. Ibyaral 'well s', Imaksural ' broken'. This suggests that the present-day J a lternatio n may owe its origin to the Irl > Igi shift. Assuming the alternation was once conditioned, as in most dialects, by the quality of the fi nal consonant except in the case ·of Ir/, a problem arises when e.g. Ikbirl beco mes I kbii!,/ and so ends in a velar, like say, /,tlq / , which presumably had " fem in ine ·/'tiqaf. One solution is for this new h i to behave like all other velars, and this was ada pted \>y Mosul (fkabiga/, like /,atlqa/); a second solution is for Ii!.! to ·go on as formerl y, viz. to behave like non-velars, and this is what occurred in C (fkbigi/, like l~awHi/ but unlike /'atiqa/); a third solution is that surmised for J, viz. Igl goes on behaving li ke a non-velar, and other vela rs follow suit (fkhigi/, so now also /, tiqi). Th is explanation per se clearly cannot account fo r all the fact s of the J alternation, and will eventuall y have to be either supplemented or discarded . 4.33 Allomorphs in sandhi. (a) The three dialects differ as to the aJ\omorphs of T that occur in external a nd internal sandhi, i.e. in annexation and suffixation but this time with J and C practically identical while M remains apart: M
'room' 'my room" 'our room"
gul?l?a gubti gul?l?atna ,
,
J qebba qebbeti qebbetna
C
qebbi qebbeti qebbetn"
72
COMMUNAL DIALECTS IN BAGHDAD
M 'two rooms'
'a bedroom' 'the bedroom'
gubte n gu\>\>at nom gubt ennom
l . qebten qebbet nom qebt ennom
MORPHOLOGY
C qebten qebbet nom qebt en nom
In lC there is -/et/- and in M -/at/- with the respective vowels yielding phonemic zero when morphophoncmically in unstressed open syllable, and with the concomitant anaptyxis and/or reduction of gemination as usual; the main ditTerence is that the -/et/- of JC is always stressed with pronomina l suffixes. The zeroing-out of unstressed open syllab le takes place even across word boundaries but in such cases a looser juncture opti o nally restitutes Ihe vowel: Igu\>\>at ennom/ , Iqebbet ennom /. In cases where this zeroing-out leads to word-internal clusters, anaptyx is takes place as outlined in 3.52: rukb + al + i yields M/nikubtil 'my knee'; this cannot occur in JC, cf. JCfgekbeti/ . With base-final y or IV, the phonetic result is unchanged in lC (exlV + el + i --+ lCfexwetil 'my brothers'), but in M we get exlV + al + i --+ lexiJtil ' my brothers', and leby + al + i --+ Ilei)itil 'my beard', with stress fluctuating between the first and second syllable. Similarly, with the dual suffix : M/ karwal 'fare', IkariJtenl 'two fares', Ilei)itenl 'two beards', whereas lC have such forms as llmaswal . 'way of walking', j meswetenj 'two ways of walking' for which C has Imasyetenj or Imasiten/. The shift of stress away from the base has the usua l etTect on J base vowels: Jj kalbal 'bitch' but Ikelbetil 'my bitch' ; l jxalal ' maternal aunt' but j xaletij 'my maternal a unt'; J/doga 'turn', but jdugetnaj 'our turn', etc. (b) A small number of words exh ibit slightly deviant behavior. While M/maraj, Cfmagaj ' wife' are norma l, J/magal has a base mgal- wit h suffixes: j mgatij ' my wife', Imgatul 'his wife', etc. There a re regu la r sandhi a llomo rph s with irregular isolated forms in M/bibil 'grandmother', i.e. Ibibitij 'my grandmother' Ibibiyathuml 'their gra ndmother' . Both M/, abaj and /, abaya/ 'mantle' give /' abatecl 'your (f. s.) mantle', /,abataj 'his mantle', etc . (c) The qellu area as a who le has the same -/etj as JC which, without tbe same stress pattern, is common throughout the Arab world. The -jatj of M is found in the other gelel-dia lects and in the peninsular area, and is rarc elsewhere. 4.34 Unit nouns. (a) One of the functions of the feminine morpheme
73
is the formation ' of unit nouns out of nouns denoting aggregates: M/semac/, l CfsafQakl 'fish', M jsemcaj, Jjsamk~j, Cfsamakil 'a fish'. But while this procedure is the rule for M , it is the exception in JC, where unit nouns are ordinarily formed by affixing -/ayij i.e. a sort of enlarged fern. s~tT. ay + T; thus:
M 'piece of bre~d' 'piece of meat' , '3 fly' 'a hailstone,', ~
"
'a star'
xubza lai)ma 5ebbiina i)aliJba naJma
1 xebzayi, lei)mayi 5ebbinayi i)alubqy.i nejma~ i
C xebzayi lai)milyi debbenayi i)alUbiiyi nej mayi
The last instance is formed on the aggregate MJCjnajem/, but in J there is also jnjumayil formed on the plura l jnjOmj. In the last example but one, the aggregate is MCfi)iilUbj, l li)a lObl 'hail'. A suffix - jayaj is also founo in M, either with diminutive con notation (informants say jxubzayal and jlai)milyal are smaller pieces than jxubzaj and / lai)ma/) or when the aggregate nou n already has the feminine ending, e.g. j kubbaj, a dish consisting of meatballs each of which is a Ikubbayaj ; both uses of this suffix are illustrated in the proverb: Ii)abbaya ysawwOh'a kubbayal ' making a mountain out of a molehill' (lit. 'a Ikubbaj ball out of a small grain'). Nole also M1Cfhefj 'hot wind', J/ hifiiyij, . Cfhefayi j 'gust of hot wind' (M equivalent not noted). .. (b) Here again' J and C go along with Mosul and the Anatolian dialects, as well as with some of the Syrian sedentary dialects; even Cairo makes frequen t use of this - /ilyaj suffix. On the other hand the qeltu-dialects of 'Ana and Hit, as well as the gelel-dialects examined , are c1o~er to M usage.
,
4.4 SOME OTHER·SOFFIXES. 4.4 1 The sound plural suffixes. All three dialects have the sound plural suffixes -/inl and -/iltl and do not ditTer from ea~h other or from the Arabic dialects as a whole as to their use, degre~ of productivity, and so forth . T hey share a morphological feature which is not 'shared by most d ialects, namely the dropping of th~ jnj of -jinj before suffixes : M jsam'inj, lCfsem'in 'having 'heard (pl.)', M/sam'iha/ , J/sem'ihaj, Cfsem'iyal 'having heard
1
,."I
I
., 74
COMMUNAL DIALECTS IN BAGHDAD
it (f.)'; M/jayik/ , Ic/jayyikl '(we are) coming to yo u', M/~ayl)ilal '(they) have called him', etc. Cf. the dropping of the Inl of the imperfect subject pron . affixes. 4.42 Relational a nd agent suffixes. All three dialects make use of the lIisba suffix -Iii, sometimes - / awi/ , ' in a manner that differs little from fa miliar Arabic practice: /, anil 'from 'Ana' , M C/J1la~!awi/, III11e~ ! awil 'fro m Mosul'. All three occasionally use the suffix -/li/, MJC/kerkOklil 'from Kirkuk'. A very common agent suffix is - /eil for all sorts of professions, h abitual activities and the like: Ikababeil 'meat ve ndor', Iballameil ' boat man' (Jbliiml 'flat river boa t') M/J1l!ayyereij, IC/J1l!ayyegeil 'bird raiser', I~axlaeil ' wily individual'. These suffi xes in -fil can be considered mo rphophonemically as terminating in - Iy to simplify statements of morphological processes : 'araballclY+T -> M/,arabaneiya/ , JC/'a rabaneiyil 'coachmen'; a similar notation is indicated for many other nouns ending in - Iii, cf. , for example, nouns on the pattern qatll, in 4.53 below. 4.43 H ypocoristic suffixes. The three dialects have occasional instances of a diminuti ve - /Onl suffix, e.g. M/zgayrOnl from Izgayyerl 'little' and I q ~ay ronl from Iq~ayyerl 'short'; cf. also M/darbOna/, J/dagbOnal 'alley' a nd note the unusual retention of lal in J. 103 In certain terms of address referring to kinship, there is, aga in in
all th ree dialects, a hypoco ristic - /u/ :MIC/,a mmu/, M/xa!ul IC/xalu/ ; many 'Syrian dia lects have a similar lui or 10/. 4.5 THE NOUN. 4.5 1 Overall view. All three dialects use roughly the same no minal base patterns , though in a few cases phonemic differentiation has resulted in some degree of morphological diverge nce, cf. e.g. the pattern qtll- in 1 as opposed to its equivalents in M and C. The greatest differentiation is in the realm of morphophonemics. In what foll ows no attempt is made at exhaustiveness, though most of the nomina l base patterns will be listed. The category " noun" includes all words susceptible of taking the a~ticle, and this in turn subsumes at least two sub-categories, substantives
and adjectives. Participl es will be dealt wi th separately (4.7), as will numerals (4.6) . It may be worth pointing out again that in the discussion of roots and patterns, such terms as "biconsonantal" and "trico nsonantal" refer not to phonemes, but to morphophonemes
represented by consona ntal symbols.
MORPHOLOGY
75
4.52 Irregular patterns. As in other areas, many OA biconsonantal nouns have becume trico nsonantal in MIC a nd fit more or less smoothly into the regular nominal patterns: MI C/damml ' blood', lidl 'hand' (pattern qell- wi th K I -' and K, -y, (see 4.54a ~low!, laxxl ' brother" Vabbl 'father' ; this is true alsQ of Iisen/, Isenij thIDg, less so of M/si/, C/se/. The three dialects show greater irregularity in patterning and fluctuation betwee n biconsonantalism and triconsonantalism in other cases: M/uxut/, 1/ext/, C/exetl 'sister', plur._ MC/xaway,. Ilxwat/ ; MJC/eseml 'name', plur. M/asiim i/, C/ase ml/ , MIl lsa ml/ ; MIC/ebenl 'son' is largely confined to use in sandhi lebnil 'my so n', (Jebn eHanl 'so-and-so's son'), plur. Ilbninl (MC pI. not noted). As for MC/benet/ , Ilbentl 'girl , daughter', plur. MC/ banat/ , J/bniit/, there is in M a more regular (and preferred) form Ibnayya/, for wh ich J has Ibnetij; I 04 Mlisanal 'year' and l C/magal 'woman' are truly biconsonantal, as is M/mara/, thou gh the latter has a preferred form Imrayyaf. Note also MIC/laxl 'another', fern . M/lux/, IC/lex/, also with longer forms M/luxra/, lC/lexxi/. 4.53 Regular patterns. (a) Reflexes of OA patterns q VII, q Viii, q Villi. From the discussion of short vowels and a naptyxis in Chapter 3, the di.fferences in the treatments of such pattern~ should already be famllt ar; the essen!ial facts may be summed up as follows: (i) With sound roots, M has a base pattern "ell- where the e is lei or lui depending al most entirely on the consonantal environment, JC have p actica lly only leI, a nd the pattern serves mostly for singular substanti ves; hOlVever, M/weledl or Iwuledl 'children', M/kutub/, l CJketebl 'books', a nd the plurals of color adjectives, e.g. M/~umur/, IC/~emegl ' red'. The difference in the treatment of final c1us!;rs in 1 as opposed to MC res ults in a different classification of base patterns : MC/wesexl 'd irty', C/najesl 'unclean', are on the same pattern as Isekell or /~a!el/, i.e. "ell- and qall-, respectlvely ; whereas the homonymous forms in 1 belong to bases of t he pattern qelel- a nd qalel-, respecti vely. Cf. J/ malek! ' king' (pattern ' qalel-) vs. J/ ~ a lq l 'mouth' (paitern "all-), vs. M/malek/, I ~aleg/, Cjmalek/, I ~ a l eqj ; the leI of MC is only anaptyctic, the lei of 1 is not. Furthermore, the -e- in such J bases as those of Inajes/ , Iza lezl explains the feminine in - Iii (Jnajsil, Izalzi/, IlVesxij) whereas qallnouns have feminines in - /aj : l l kalba/ , ' bitch', Igamzal 'hint', Igafsal 'kick'. Some examples of "ell- qall- a nd in 1 also qhe!- and "alelnouns (last two of list):
76
COMMUNAL DIALECTS IN BAGHDAD
MORPHOLOGY
M
J
C
rejel quful galub
gejel qefel qaleb
bared
gejel qefel qalb ,aheg bagd
' king'
malek
malek
malek
'impure'
nages
najes
najes
'foot'
'lock' 'heart' 'month' 'cold'
~ahar
~aheg
baged
(ii) Nouns like MICjtiitl 'mulberries', IsOdl 'black (pl.)' and ltinl 'figs', Ibi51 ' white (pl.)', can easily be viewed as containing qel/~,
KrY or w, respectively, and a pattern
where e merges with y to IiI and with IV to 101; similarly, words like MJC/betl 'house' and Ijozl 'nuts' can .be viewed as having K,-y and IV and a pattern qall-, where a merges with y into Ie/ and with IV into 101; the existence of a very few "unmerged" words such as Ijay~1 'army', the proper noun Iqays/, M/zawjl 'husband' and Izawjal 'wife' need not deter us and may simply be listed as special cases. Words like MJC/jaml 'glass' . are slightly ·more problematic, but if we wish to consider them as having a pattern qall-, we may consider K2 in such cases to be a ' with which the pattern element a merges into la/. At any rate, qetl- and qat/-words with K,-y or IV show nothing special, except of course lack of anaptyxis in MC and the normal behavior of long vowels in 1. However, in the "unmerged" forms with phonemic Iyl and Iw/, note that /jaysl undergoes no change when suffixes are added, whereas Izawjl and Izawjal behave as with any other K, : lzawejhal 'her husband', Izawejtil 'my wife'. (iii) With K,-y or IV, the suffix less word ends in /i/ or luI, respectively, but has Iyl or Iwl with suffixes: from the bases MJC /lellV- and M/lacy-, IC /wky- , we get MIC Ihelu/, M/haci/, IC/hakil and: M
' pretty (f. s.), ' his talk'
helwa I)acya
J helwa hakyu
, C
helwi hakyu
(iv) There are, of course, a good many qetfT- and qal/T-nouns (i.e. with fern. suff.) in all three dialects; indeed, their number has
77
been increased b~cause most OA qalafT-nouns have dropped the second lal and are now represented by qatlT- and qetlT-nouns so that M/ rukba/ 'knee' (OA/rukbaf) and M/rugbal 'neck' (OA/raqaba/) n.ow show the Same pattern, as do the JC equivalents I/gekbal and Igeqba/, Cjgekbll and Igeqbi/ ; see par. (b) below. (b) Reflexes of .OA pattern qatal. There are a good many such reflexes wIthout, tllough very few with, the femin ine ending. In M most are on a ~attern qelal-, in
Ie
on a pattern qatal- ; the fate of
the first vowel In M has been dealt with in 3.33; in some cases M has qatal- as well: ,
'onions' 'camel' 'boy'
.,M
J
C
I:>u~al
I:>a~al
jemal walad
I:>a~al
jamal walad
jamal walad
In some, cas~s, M has qetl- where JC have qatal-; M/ henec/ , JCjhanakl chm . As In all bases ending in VK, JC stress the syllable Immedla.tely ~recedl~g pron. suff., M does so optionally; M/jemal~1 or / lemalal h,s camel', Iljmalu/,. C/jamalu/. Feminine nouns for whIch OA (at least as attested in CI. Ar.) had the pattern qatala(t) have, m the three dialects, generally not a pattern qatalr but qatfT or qetlT; the zeroing-out of the lal in unstressed open syllable is normal for J and to f ame extent for M, but not for C; and the change of the first vowel is normal for M but not for IC, yet in this case the three dialects have somehow converged: M 'ten' 'tree'
'neck' 'fish'
J
C
'a§ra
'asga
sejra r,ugba S,e mca
sejga geqba samka
'asga sejga geqbi samaki
Nouns corresponding to OA pattern qataiT- are particularly prone to all sorts of reshufflmgs m other dialects, cf. Mosul Igaqqabil and the many Syrian dialects that have e.g. Iraqabal but Iraqbatl wIth suffixes. The M forms are more or less what one would expect (I.e. conform to a more general M treatment of) a/, cf. 3.33), the \ I I
78
COMMUNAL DIALECTS IN BAGHDAD
J forms are not always what we would expect (whence the lei of Isejga/, Igeqba/ ?; however, note that an expected */gaqbal would have yielded Igeqbet·1 in suffixation) and the C forms apparently reftect, as do other C features, a mixture of J·like features and Mosul· like ones. There are a few qatal· and qetal·nouns with medial lV, though they hark back to different OA patterns. T have noted on ly the fol· lowing, which also have K ,-y; the middle radical yields Iw/, the final radical yields zero:
M ' light' 'remedy'
(Suwa duwa
J
C
Qawa dawa
dawa
'bags' 'roo ms'
M
J
C
junat ebar gu!;>a!;>
jenat ebag qebab
jenat ebag qebab
T he singular J/gekba/, C/gekbi/, M/rukbal ' knee' gives JC/gekab/, but M/ rkabf. O n c/jedadl and /'etaq/, see (d) below. (d) There are a few singular substantives and a few plural adjec· tives that have a pattern qettal· in MJC; those listed are the only o nes noted:
M 'rice' 'ladder' ' new (pl.)' 'old (pl.)'
J
C
temman
temman
temman
sellam jed dad
sellam jeddad 'ettaq
sellam jeddad 'ettaq
'ettag
79
One C informant gives for the las t two the alte'rna nt forms Ijedadl and /'etaq/. (e) Nouns having a qlal patte rn in all three dialects ha rk back to OA pattern qlllal and aqlal, rarely also to qital; as a rule, nouns harkIng back to qilti! have qlti! in M onl y, but have been altered through 'ima!a in J, and C a nd are discussed in (f) below. No uns har~lllg back to qaltil have qtti! in 1, usually a lso in M, but usua ll y qala! 111 C. The s lI1gle la rges t category in which a ll three have qtti! IS tha; of .adJectlve plurals; on the reasons for positing a n OA pat. tern quta! for such patterns, see 3.36 above. These include:
<;Iawa
Nat urall y, the M qeral·forms a re undistinguishable from the qetal· fo rms com mon to MJC and reflecting OA qilal or qlltal. (c) Reflexes of OA qilal and qutal patterns. These yield MJC qetal-bases, except for o nc or two cases of qtal- in M, and are most commonly plurals of nouns qetl· and qatlT·; for the singular MJC Ij anla/, M also /j untal 'bag', and the singular M/ebra/, JC/ebgi/, M/gu!;>!;>a/, J/qebba/ , C/qebbil ' room' we get:
'neecpes'
MORPHOLOGY
M 'fat' 'big' 'small' 'long' 'clean'
'broad' 'nice'
'sho rt' 'strong'
J
C
sman
sman
kbar zgar Ilval n(\iif ' ral)
sman
kbag zgag twiil n(\af 'ga(\
kbag zgag tWa I n<;laf 'gii<;l mliil) qsag
mlii~
,
g~ar
,CJ,Wiiy
qsiig qway
qway
. There is no M cognate for JC/mla~/, nor for the corresponding sll1gular ; a ll sll1gulars are reflexes of OAqalEi except for the adjectives 'small' and 'short', ~hich have patterns qlayy!. in the three Baghdadi (as In many other) dJalects. As for substanti ves in which JC as well as M have qlti!, they include: M 'teeth' 'children' 'small fry' 'sleeve'
J
C
snan (weled) frax rd an flan grab
snan wliid frax rdan fla n grab
' rope for headdress' 'donkey'
wiad frax rdan fliin grab
'gal
'country'
blad
'gal 1)1!lag blad
'gal I)mag bliid
'so-and-so' 'crow'
~l!1ar
soan
• , I
~.
.~,
80
COMMUNAL DIALECTS IN BAGHDAD
The first three are instances of plurals· harking back to (or contaminated by) OA 'aqlol form s, and Ird,ml seems to rest on an old plural a lso, the new plural being MJC/redenj. The others hark back to older qUlol, or, exceptionally for JC, to older qilol nouns which for some reason have not undergone the expected 'imola ; /, gall is clearly a borrowing from M or another gelel-dialect, while the word for 'donkey' may rest on an older, albeit unattested,lOS Ibumar/. Note also M/zyara/, JC/zyagal 'pilgrimage' , MJC/zma l/, another term for ' donkey' . F inally, in J practically all OA qOlol nouns now have this qlol pattern, while M preserves the lal in some and C in most such nou ns:
'wine'
' North' 'cloak' 'three' 'day'
'illicit' 'licit'
M
J
C
srab samal 'abaya tlaOa
'gab smal 'bayi tla8i nhag !)gam I)lal
sagab samal 'abayi tlati nhag bagam I)alal
nhar
Qaram balal
For J/ l)wasl 'clothes', MC have no cognate, only IhdGm j. (f) . Nouns harking back to OA qilol have, as a rule, qltil in M but qli!! in C a nd qlil in J. They include many high freq uency singular and es pecially plural substa ntives, and the threefold correspondence is almost, but not quite, complete: M 'quilt'
Isan ll,taf
'belt'
i:tz3ffi
'underpants' 'book'
Ibas ktab I,tsiib rjal " lab jmal sba'
tongue'
'account' 'men' 'dogs'
'camels' 'lions'
81
MORPHOLOGY
J
C
lsin l!)if I)zim Ibis (ktab) I,tsib gjil klib jmil
lsen Il,tef l)zem Ibes kteb (I)sab) gjel kleb jmel
sbi'
sbe'
'candles' 'baskets' 'ropes' 'mountains'
sma' sial I,tbal jba l
smi ' slil l)bil jbil
f .
sme' slel l)bel jbel
The forms C/l)sabj' and J/ ktabl do not conform to the regular correspondence, but cf. .J/ ktibil 'a writing'. With K ,-y and the feminine ending, M has the '.regular Il)caya/, J/ l)kiyyi/ or better Il)kiyi/, and c/l)kayyil with the expected layl for lei before Iy/- As for M/dejajal ' hen, chicken', it has the same pa ttern qlol with an automatically anaptyctic lei, but 't he JC equivalents J/jijil and Cfjeji/ now show different patterns, though clearly harking back to a n OA form Idijaja/ 106 Note also M/bhaml (apparently also used in C) and J/bhim/ 'thumb' , harking back not to a qild! noun but to OA/,ibham/ , with the J 'imola of course still regular. The J word is now homophonous with J/bhiml 'beast of burden, donkey', and in general the J nouns in qlil listed above are indistinguishable from those harking back to older qalll, e.g. Inxi!1 ' palm trees', Ihlibl 'milk', which M and C keep apart; this renders at least one or two etymologies uncertain, e.g. J/zwijl 'marriage', neither ' /zawijl nor */ziwajl being attested; the MC form is Izawaj/. (g) The OA pattern qatil is represented in MJC by a pattern qiflll, in which if is shorthand for lal alternating with unsta ble /eI in M, lal (and occasionally zero) in C and zero (and occasionally la/) in J. On the fate of OA lal in unstressed open syllable, cf. 3.33. The pattern is found in ma ny common singular adjectives, a nd some singular and some plural substantives. They include:
'heavy' 'light' 'long' !broad' 'big' 'poor'
'milk' 'friend' 'donkeys' 'jar'
M
J
C
Gegi! xaftf tewil
faqir balib
8qi! xftf !wi! 'giil kbig faqir I) lib
laqil xafif !awi! 'agi<:! kbig faqir I)alib
~adiq
~(a)diq
~adiq
I)amir brig
I)mig bgiq
I)(a)mig bgiq
'\ ariQ ~ebir
82
MORPHOLOGY
COMMUNAL DIALECTS IN BA GHDAD
The last instance harks back to OA/,ibriq/, not to a qalll noun. Forms in which C has /e/ before / t)/ may be considered as ha ving the same pattern e.g. C/malet)/ ' nice', h ahet)/ 'whole'. With K ,-y , there is e. g. M'/C/qawi/ 'strong' ( M also /guwi/) with diver ge nt phonemic shape (penult-stress, truncated second syllable) In_the masculine but the pattern is normal in the feminine, M/qaWlya/ , J/qwiyi/, i.e. base qmvEy in which y yields zero in the suffixless form , wi th concom itant stress and length sh ifts, but /y/ with suffixes ; cf. also the plural /qwayj. . . (II) The pattern qflil, especially wi th the feminine ending, IS frequent in the three dialects for plurals of qVtl nouns :
. 'houses'
M biyut
J biyiit
'roofs'
~ !ut)
~!ui)
'hearts' 'tables'
glOb
'bastards' 'bottles'
ngu!a !;itO!a
qlub myuza ngula i?lu!a
myiiza
C biyiit Hot) qlub
myuzi ngOli i?luli
This pattern does not seem to occur wi~h K,-y . On: C informant gives the unusual form /I)~uyen/ 'horses, pl._ of / I)~an/, wh.ch has an aggregate MJC/xel/ and a plural MJ/ I) ~ una/ ; my other _C informa nts recognized / I) ~uyen/, state they themselves use /xe,l! or /I)~uni/ a nd thought the unusual form to b: "from the North. On C/o/ for /u/ before /1)/, / q/ a nd /g/ , cf. / malel)/ In (g) above and 3.34. (i) All three dia lects have, to a greater extent than most vane~te-' of A rabic, preserved plurals harking back to the OA ~att~rn qalala, with 'im ala of the final vowel (spelled with 'alif maq~lIra In CI. Ar.) in J a nd C' most but not all corresponding singulars are adject.ves , ' 1y qla-If qatlall: the patterns are respective a or M , on the pattern qliill for J, and qatali for C; such plurals Include:
'lazy' 'drunk' 'tired' 'ca refree'
'thirsty'
M ksala skara ('aba i?lara -liisa
J ksali skagi t'abi i?lagi 'Iasi
C kasali sakagi ta'abi i?alagi
'atasi
'orph ans' 'pregnant'
yt.a ma t)biila
ytiimi I)bali
83 yatami I)abali
For the last example C informa nts also give / I)abeli/, as though a p atte rn qaretel with' K ,-y, cf. /zaweli/ from /zuliyi/ 'rug', and cf. the JC singular / I)eblej. All three dialects have only / najara/ 'Christia ns'. Also note in J'( MC eq ui va lents not noted): / i)zani/ 'sad' (sing. / I)zin/), /myati/ 'dead' (si ng. / mayyet/), /!yabi/ 'alive' (sing. / layyeb/). Some of the above ex~ mples a lso have plura ls in -/in/ , e.g. M/kaslanin/ , M/ sak ra nin/, Mj'a!!iinin/ . (j) Reflexes of nouns on the OA pattern qalil are numerous and va riegated. Many have a pattern qalel in all three dialects, while all Form I participles have a regular correspondence M-qalel vs. JC-qetel, whereas still others show the correspondence M-qatel vs. J-q!lel and C-qiitel. Participles are discussed separately in 4.7; for the ordinal numerals, see 4.6. In a ll cases, the e of the base pattern is elided in the feminine, in M also with suffixes, while in JC it is retained and stressed with pron. suif. (cf. 3.4 above). Cases where a ll three have qatel include: MJC/ kateb/ 'clerk' /tajer/ 'merchant' / Iazem/ ' necessary', /wadi/ 'valley', M/ l)al)1u(\/, J/ l)al)1e9/ , C/ i)iil)1e\f/ 'sour', M/ ~ 9ul/, J ~a gel/, Cf ~ i?e17, 'officer', MJC/wali/ ' provincial governor', etc. With 'ima/a in Je but excluding participles and ordinal numerals, we have for example:
'mosque' 'child' 'frozen' 'cold ' 'corrupt' 'barefo ot' 'high' 'warm' 'dry' ' health ' 'rim '
M jame' jiihel ja med bared fased
J jime' jihel jimed biged fised
haft
hili
I) ii ~ya
'iii difi yibes 'jfi i)iSi
,iili diifi yabes 'iifya
C jeme' jehel i,omed belied fesed i)efi 'eli defi yebes (not noted) (not noted)
With K,.y, masculines (i n J a lso feminines) end in / i/ , in M there is /yl before suffixes, Qut in C informa nts hesitate between two types
,
84
.,,
COMMUNAL DIALECTS IN BAGHDAD
of feminines, /I)efy il and II)efiyi/. For M/nasi/ 'low' there is the regu larly expected C form Inesi/, but J has Inassi/. (k) In OA , or rather in Classical Arabic, one could. distinguish two '0'1101 patterns, homophonous in this shape but qUIte dIfferent in syntactic behavior: one '0'1101 pattern for elatives with many special syntactic features a nd (some) feminine .equiva lents w Ith a pattern qllllti, and another ' oqlal for colors and mfirmltles wIth less syntactic peculiarities and reg ular femin ines in qallti'. The two patterns are still quite di stinct in MJC, th ough to the syntactic differences we must now add a morphophonemic one operating in Je, where the two patterns behave differen tl y wi th K ,-y. There is a pattern aqllil for co lor, a nd infirmiti~s (with Ii merging with K,-y into lif) and a n elative pattern aqlal with unchangi ng pattern elements: JCfa'mil 'blind ' but la' ial 'higher' , cf. JC/algasl 'deaf', lalwall 'longer' . In M, the two patterns are identical: M/a'ma/ , la'la/, r al ras/ , lalwal/. Elatives a re invariant as to ge nder and number, and often come before the noun wit h which they enter mto the same sort of annexation as the ordinals: MJC/a\wal yoml 'the longest day'. Colo;s and infirmities have a fe minine pattern qallti, with the final ti stressed in JC but unstressed (and hence short) in M; on the final vowel, cf. 3.5; on morphophonemics, cf. 4. lb: M/ bamra/, J/bemga/, Cfl)amgal ' red (f.s.)': M/soda/, Jjsuda/, C/sodal 'b!ack (f.s.),; M / be~a/, J /bi~a/, Cfbedal 'white (f.s.)'; M/larsa/, J/legsa/, Cflagsal 'deaf (f.s.)' . Elatives derived from geminate roots (K, and K, Identical) show both the unmodified pattern (MJCfaxfaf) 'lighter' root x.lf ) and a classical-li ke modification aK ,aK, K" MJCfaqa1l1 'less, least' , root qll. Note also J/aqqabl ' nearer', cf. J/qqibl ' near', but Cfaqgabl , Iqagib/.
(I) Reflexes of patterns qOlllil, qilllil, qllll lil. The three dialects ha ve the regularly expected equivalents for the common Arab~c pattern den oting habitual or professional activity, MC-qaltal, J-qeltli/:
'bake r' 'tailor'
M xabbaz xayyal
'carpenter'
najjar
'cobbler
ragga'
J xebbaz xiyyal nejjag geqqa'
C xabbaz xa yyal najjag gaqqa'
MORPHOLOGY
85
Some nouns occur in short phrases with a descriptive complement e.g. MC/ fattab fall; J/fettab falf 'fortune teller', J/qellii' edgubl 'highwayman'. The pattern occurs also in some names of household utensils, e.g. with fem. suff., J/qerrayil 'Sabbath lamp'. A pattern qalltili in MJC (without the lal > l ei shift in J) OCCurs in a few special inva riant forms that seem to be used only predicati vely, of wh ich [ have noted the following : M/ga"iidi I, JCfqa"iidil 'in a sitting position', M/waggafi/ , JCfwaqqafil 'in a stand ing 'position', Ilayyaril 'on the wing, in flight" 07 a nd Cfbaddawil 'in the language Of the Baghdad Muslims', which apparentl y has no MJ equiva lents' o• In J, there is no phonemic difference between this pattern and the o ne
harking back to OA qlllltil, for which MJC have qelltil, with the usual values for e; this pattern is mostly found in plurals of nOllns ha ving qVlel in the singular: M 'oldsters' 'clerks'
'inhabitants' 'barefoot'· 'high'
siyyab kuttab sekkan beffay 'ellay
J siyyab kettab sekkan (Min) ('ilin)
C siyyab kettab sekkiin beffay 'ellay
The last two exa mples, for which J seems to have only plurals in
-/in/, show " regular" behavior in MC in roots with K-,y. Among singular nouns on this pattern, cf. MJC/xe ssiifl 'bat', Isekkiinl 'rudder'. No uns har~ing back to an older qilllil are differentiated by the expected 'ill1tila in JC, so that we have the patterns M-qelltil, J-qell il, C-qelliil ; syn chronically we may include here M/oebban/ , J/oebbin/, Cfdebben/" 'flies' , though historically the pattern was
qitlall . 109
(m)
Reflexes of OA patterns qalltin , qlllilill, and qilloll. The lhre.e
dia lec ts have nouns
in
qV/Wn denoting personal properties or states
of mind or body; in M the pattern is qalltill, in J qelltill as expected, but in C there is usually qelltill rather than the expected preservation of OA/al that is otherwise typical of the dialect. M 'lazy' 'carefree'
kasliin \>a\ra n
J kes liin i;>elgan
C keslan i;>elgan
86 'thirsty' 'h ungry' 'drunk' 'glad'
'atsan
'~lsa n
jo'an sakrii n farl)an
ju'an sekgan feg l)an
87
MORPHOLOGY
COMMUNAL DIALECTS IN BAGHDAD
'etsftn j o'an sekgan fegl)an
All three dialects have qerltin in a number of substanti ve plurals, usually with no 'imtila in JC, so that they may hark back to OA qlllitill plurals even when the classical forms a re on the pattern qilltill: MJC/sedqanj 'friends', j l)ezmanj ' belts', jsebya nj 'boys' , j neswa nj 'women', Jj begqanj 'jars'. However Jjjegdinj 'mice' (MC equivalenis not noted) with the expected 'in/{ila (cf, OAjjirMnj), a nd simila rl y the term for ' flies' referred to at the end of the preceding paragraph. The pattern qelltin'also serves, at least in J, to form ve rbal nouns for some Form 1 K 3-y verbs and some verbs with geminate roots, though in the latter case K , is, in this pattern, replaced by jyj : Jjkefyanj in j kefya n sa ggj 'avoidance of evil', (i . e, payment of a bribe in return for safety, etc.), root kfy; Jjsemyanj 'smelling' (smm), Jjmedyii nj ' stretching' (mdd).
'bird , sparrow'; cf. also Jjxeggufj 'sheep', though Mjxarufj on a di fferent patte rn ; ef. a lso MJC/za'lulj 'small boy'. A special use of this pattern, with added -jij is fou nd in the hypoeoristie form s of certain given name~ : jbar humij (from j brahim/), jl)ammud ij (from jahmadj, jmahmOd/, etc.), j'abbOdij (fro m any name beginning with j'abd - j), etc.
, Reflexes of OA patterns qartilil, maqtilil, etc. These patterns are, in general , to ~e fo und in plurals of disyllabic singulars of wh ich the first syllable is long (i.e, ends in V or VK). The patterns have undergone the exp~cted 'imtila ·in JC, viz. are ' qrilel (mqllel) and qarelel (//laqel el), respecti vely, though some forms with jilj a lso occu r. That the JC 'imdla is "in the pattern" and not in the indi vidual words is sho\yn by the app lication of these patterns to plurals of relatively recent borrowings. Some examples: (0)
.. M
Patterns qVttVI a nd qVrlVI. In addition to patterns 'lVI/til which have been treated separately above, the re are a number of trieo nsonantal a nd q uadrieonsonantal patterns of simil ar syllabic
'eyebrows' 'handkerchiefs' 'tents'
cwader
structure and other com mon features; these are considered briefly
'pistols'
here.
'drums'
'rings'
wrawer danabeg mhabes
jwimc' hwijeb kfili twider wgiweg dnibek mhi bes
'schools'
madares
mdares
'snakes' 'persons' 'lands'
(nOl noted) awfldem arai)i
hyiyi awadem agai)i
(II)
(i) Reflexes of OA patterns qirltil a nd miqltil include: Mjgarbilj, Jjgegbil j 'sieve' (see a lso this item in the lexicon, Chap. 6); with K ,-IV merging with the first pattern element into jij, jij : Mjmiza nj, J/ mizinj, C/mizenj 'scale' (root IVzn) ; note MC/miliidj, Jlmiladj 'bi rth day' (root wid) with out ' imtila , but C/miledj in j'id elmiledj 'Ch ristmas'. On the unusual features of Mjmezribj, JC/megzib j 'gutter', cf. Cl. Ar . mizrdb , see this item in the lex icon. (ii) Reflexes of OA patterns qillif and miqlif include Mjseccinj, JC/sekki nj 'knife', MC/\;>a\li xj, Jj \lelPxj ' melo n', MJC/meskinj ' poor soul', MJC/me' milj 'customer, dealer'. On raqlll, see ('I) below. (iii) Reflexes of patterns 'loll iii and qarliil (on maqllil, see 4.7 below) include several names of a nimals, e.g. Mj bazz un(a)j, Jjbezzu n(a) Cjbazzun(i)j, 'cat', Jj'aqruqaj, Cj'aqroqaj 'frog'; Mj'asfUrj, Jj'esfUgj
'mosques'
jwamc' hwajeb
J
crali
C jawe me'
hawejeb kafefi caweder (not noted) danebek mal)ebes made res, madares hayeyi awedem ageQi
In the last insta nce, the singu lar is Mjarui)j, JjaMj, CjageQj but the plural is fo rmed on! a root with K,-y, which, as in other cases (cf. 'handkerchiefs', 'sria kes') merges with the e of the pattern into jij. The M pattern is qrdrel, but occasionall y qardlel. The C form s show a greater regularity than in J with res pect to 'imtila; this is true also of refl exes , of patterns qaldyil (CI. Ar. qalt'i'i1), which are usuall y plurals of ilOuns havi ng patterns of the type K(V)KYK;
., j
••"n
:j,
-
88
in such cases C often has the ex pected layl < lal before IYI (cf. 3.36a), whereas J, like M, has la/ :
M
J
'brides'
'araye~
'gaye~
'old women' 'taxes'
'ajayel i)ariiyeb
'jayel (\arayeb
C 'agayyes 'ajayyel dariiyeb
Note a lso J/lmiiyel/, Ibhayeml 'donkeys', (M /lmiiyel/), Cj'tayyeql 'old things'. The C plural Ijawegebl 'socks' is regular for the singular Cjj6gabi/ , but the MJ cognates ha ve the long vowe l in the second syllable of the singular (M/jiiriib/, J/juriibj) and regular plurals belonging to a different pattern (see following paragraph). The J plural . Imkinesl 'brooms' is reg ular for the si ng. J/mekensil but C speakers hesitate between the expected Imakenesl (sing. C/ maknasi/) and Imakenis/, cf. the M form in the following paragraph . As to the OA pattern qarcililT, with the fem inine end ing, it is used, as in Classical Arabic a nd in many dialects, for aggregates denoting inhabitants of a town or co untry. In the few examples noted for MJC, however, it appears that the I ii of the OA pattern has left none of its usual traces: no 'into/a, and in J the feminine allomorph is fat, even whe re
the ge neral rules (see 4.32) would lead one to expect M 'Baghdad is' 'people of Mos ul' 'people of Basra' , Egyptians'
J
89
MORPHOLOGY
COMMUNAL DIALECTS IN BAGHDAD
/i/ : C
bgiidda
bgadda
bagiiddi
mwa~ !a
m ~a lwa
ma\Va~1i
b~arwa
b~agwa
(not noted)
m~arwa
m~arwa
(not noted)
m ~a !wa
(1') Reflexes of OA patterns Iqarutil/, Imaqatil/ , etc. These include, by and large, plurals of disyllabic singulars of which both syllables are long, though in J the first may be a phonemically short vowel. As has been seen, (3.36a) there is the expected 'imcila in C, but J has lal rather than an expecled Ii/. Some examples:
., M ,.xbiibiz xyayi! bZ3zin qpiipi! d,wiilib dkiikin jwarib
'bakers' 'tailors' 'cats' 'coats'
'wardrobes'
,
'shops' 'socks'
J xbabil xyayi!
"
C xabebil xayeyi!
bzazin
bazezin
qpapi! dwalib dkakin
qapepi! dawelib dakekin Uawegeb)
jwarib
Slightly irregular, are M/l)wiiwin/, J/l)wawin/ , Cjl)awewi n/, and M/d(a)niinir/ , J/daQanir/ , Cjdanenir/, from si ngulars with K,-y (M //:Ii wiinl 'animal', Idiniirl 'd inar'). For 'brooms', M/ meknasal has the regular plural Imkiinis/, cf. the JC forms in the preceding paragraph. The pattern is also found in plurals of some words with only one long syllable, e.g. M/mfiilisl ' broke' sing. Imeflesl Cja,ebi'l 'fingers', sing. le,'ya'i and I,'yii'a/, C/adefigj ' nails', sing. ledfeg!; this includes nouns of geminate roots having in the singular the pattern maqta! thus for MJ/mxaddal 'pillow', Cjmxaddil MJ/ m/:lallal 'quarter', Cjmal)allij (b ut not MJC/mal)alll 'place, store' which has a plural in - /iit/), M/mga"l, JC/mqa,,1 'scissors' we have:
'pillows' 'quarters' 'scissors'
M rnxiidid .m/:liilil mga,i,
J
mxadid m/:lalil
C maxedid mal!elil
mqa ~i~
maqe~i~
l
(q) Reflexes of. OA patterns taqtil and taqattll!. In the three dialects, verbs of Forms II and V have regularly corresponding verbal nouns with MJC patterns taqtil and tqette!; the latter is used for quadriliteral verbs also. A feature of the taqtil pattern is th at it keeps its lal in 1: MJC(tasli ml 'handing over' from Isallam/ ; Itax\Vifj from Ixawwafl 'to frighten' or Itxawwafl 'to be frightened'; j.ta'liml from j'allaml 'to teach' or It'allaml 'to learn'; M/tkeffer/, C/tkeffegl from Ikaffar/, Ikaffagl 'to curse'; J/tmeddedl from Itmaddadl 'to stretch out', J/tyebbesl from Iyabbasl 'to dry' and Ityabbas/ 'to dry (intr.)'; J/ txellil from Ixallal ' to put'; M/ddebbecl from Idabbacl 'to stamp one's foot'; M/tme(jyegl from Itmai)yag/, 'be bored, annoyed' and Ima(\yagl 'to bore, annoy'; and in absolute construction, Imtalfal)a tteffe/:ll 'round as an apple (cheek)' , Imkal)l)ala tke/:ll)ell
90
COMMUNAL DIA LECTS IN BAGHDAD
MORPHOLOGY
'black with Ikul)ull (eye)'. With K, -y verbs y is usually phonemically ze ro in taqtil, e.g. Ita 'qif)1 from l'qayyaf)1 'to whiten'. In J/tbitl ' food kept warm from Friday through Saturday', we have a development from "/tabitl ve rbal noun of Ibayyatl 'to keep overnight'; the Mosul and Anatolian Jews still call this Sabbath food Ita bit/. (r) The above list of nominal patterns is far from being exhaustive, but does include most of the common, productive patterns as well as some less common o nes that were deemed of interest. Some nominal forms that occur only in conjunction with certain numerals are discussed in the following section. 4.6 T HE NUMERALS . 4 .61 The cardinal numbers. (a) The numbers 'one' to 'ten' exhibit' in MJC a morphology that is characteristic of Eastern Arabic as a whole. While the numerals 'one' and 'two' are essentially adjectives inflected for gender, the numerals from 'three' to 'ten' have, in each dialect, three shapes each, one when preceding the numbered noun (shape A below) a second when not preceding the numbered noun (shape B), and a third bound to one of a limited set of plural nouns (shape C) . The numerals
J weQed wel)di Onen Oenten
M waQed weMa Onen Benten
(m .)' (f.)' (m ,), (f.)'
tlaO art;>a' xames
tliiOa arQa'a
xamsa
xamsi
arf?a'tiyam xamstiyam settiyam
xamsti y~m
...
sabe'tiyam 9mentiyam tese'tiyam
'asertiyam
settiyam
sabe'tiyiim Smentiyam
"
,
setti sab'a
1man
tmeni
tese'
tes'a
'aSeg 'asga tlattiyam arf?a 'tiyam xamstiyam ·settiyam
sabe'tiyam
tese'tiyam
tmentiyam tese 'tiyam
'asegtiyiim
,'asegtiyam
tnen
considered either as one or two words, but note that disjuncture,
when it occurs, usually detaches the ItI from its etymological place and joins it to the noun: Ixams tiyamf. 110 Moreover, the nom inal
in usage remain to be investigated. The numera ls fro m 'three' to
tliiOi agb'a
'asra
tlattiyam
sett sabe l
tenten
' ten' are listed below; in each column, shape A is on the left, shape B on the right, and shape C is listed separately after them, attached to MJC/iyiiml 'days' . J tlaO agba' xams
laser
sett setti sab' sab'a Oman Omini tes' tes'a 'aseg 'asga tlattiyiim agi;>a'tiyiim
C wel)ed wel)di
On reflexes of OA/fardl and their use, see 4.9 below. Many nouns admit both the dua l and the use of 'two' with the plural, e.g. M/betenl or IOnen ebyOtl or IbyOt eOnenl 'two houses'; details and differences
M
setta sab'a Smiinya tesla
As can be seen, shape C consists essentially of shape A with an added Itl, except that no third ItI appears in Isettl and that there are slight modifications in 'three' and 'eight'. As in many other dialects, shape C is thus not based on shape B by replacing the final lal or /if with ItI (as would oeeur in the ordinary fern. suff.), but results from a complex historical reshuffling. The dissimilarity of this alternation from that of the fern. suff. ·T is enhanced by the irregu larity of the li/-/al alternation in J, which here exactly parallels that of C and is conditioned by factors that do not usually operate in J (ef. 3.36a above). In the shape C combinations, the complex might be
'one' and 'two' are:
'one ' one 'two ' two
sett sabe' Sman tese'
91
C tlat arba' xames
tHiti arba'a
xamsi
forms occurring in these combinations often show special features
and have different shapes when isolated : JC/xamstesegl 'five months', Isette~egl 'six months' and so on with -/teseg/, but'months' in isolation is lesheg/ ; M see ms to have both Ixamsteshurl and Ixamstesur/,' and the separate form is lashurf. Shape-A numerals with a pattern qat! and qeli behave normally as to anaptyx is ; MJC/xams ebyOtl 'five houses', MJC/xames daqayeql 'five minutes'. (b) The numerals from 'elevell' to 'nineteen' have, in MJC, only one shape each, and that without any final Ir/ ; this is true of all Mesopotamian dialects examined so far, so that these contrast with the Syrian area (two shapes, one with Irl before nouns and one without Irl else,where) and the Egyptian area (a single shape,
92
with Irl throughout). The differences between M. 1, and C are largely but not wholly due to regular phonetic correspondences, cf. especiall y 'eleven', 'fourteen', 'seventeen', 'nineteen' ,Ill
M
1
da' as Ona'as \laqa'as ar!;>at;\'as
ida's Ona 's \!etta's ag !;>ata',
xr!1u ~ \a' as
xrpe~ta' s
!e\\a'as !!;>a'Was
~!;>ata'$
8f11Cnpl'as
Ol]1enta's
! ~ a ' ta' as
t~a!a's
ida'es tna'es t! ana'es ar!;>ata'es xal]1e!(a'es !eqa'es !!;>a(')!A'es (l]1an(a'es t!a(')\a'e,
(c) The tens from 'twenty' to 'ninety' also have a single shape each and exhibit mostly reg ular correspondences :
M
J
C
'esrin
'esgin
tHiSin
t1i9in geb'in
'esgin tlOtin
•
arba'in
arba'in xamsin scttin
xamsiri setOn sab'in Omanin
xemsin settin seb'ln
9m.inin
sab'in tamenin
les' in
tes'in
les'in
'thousand' 'two thousand'
'three thousand'
'alef alfen , ' t1attalaf
alf elfen tlattalaf
alef alfen tlattaliif
Before a numbered noun, M/miya(, JC/miyi(, and compounds ending in them have the alternant MJC/mit / : 1mit felsl 'a hundred fels ', Ixamesmit felsl 'five hundred fe ls'. ' Hundreds' beyond two are formed with numerals of shape A, 'thousands' beyond two with numerals of shape C attached to MC/tiilar/, Iltalar/. 4.62 The ordinals. The ordinals from 'second ' to ' tenth' are on the pattern qatel in M, qftel in J and qatel in C, in the last case without the expected 'im·ala. They are:
C
~ ena 's
93
MORPHOLOGY
COMM UNAL DIALECTS IN BAGHDAD
M
J
C
Oani OaleO rabe' xames sades sabe" 9amen
mni 9i1e9 gibe' . ximes sides
Lani talet rabe' xames sades sabe'
tase' 'aser
lise'
tarnen tase'
'iSeg
'aseg
sibe'
9imen
,
These function. either as adjectives, i.e. follow the substantive and inflect for gender, number, and determination, or they function as elatives, i.e. precede the substantive and remain invariable: in the
These call for no special comment beyond the J form for 'forty' which can easily ha ve developed from ' /agb'inl > ' /ageb'inl > Igeb'inl , and the J 'imiila in 'thirty' and 'eighty' is discussed in 3.36a. Beyond these, the 'hundreds' and 'thousands' have the following form s:
latter case the article is never present: (9a le8 yoml 'the third day' , 19a1e9 lelal 'the third night'. As for 'first', MJ C/awwall is used in this latter constructiQD; it may also occur in the adjective construction , though in that case there is the more common variant Me
lawwalani/, 1lwlani/, with regular
f~minine
and sound plural. There
are, as in other dialects, no special ordinal form s after 'tenth', the
cardinal numbers being used in that function. M
' one hundred'
miya
'two hundred'
milen
'three hundred'
tla9miya
J miyi miten tla9miyi
C miyi milen
t1atmiyi
4.7 THE PARTICIPLE. 4.71 Form I participles. (a) Probably all Arabic dialects exhibit some formal features distinguishing participles from nouns having similar patterns; 112 an additional differentiation
i
.1.
~ .~
94
95
COMMUNAL DIALECTS IN BAGHDAD
MORPHOLOGY
connected with ' illlliia a nd hence a lready mentioned in 3.36 obtains in the Form I active participle of J, to a lesser extent of C, but not in that of M . That participle a lways bas in M the pattern qate/, which is shared by many nouns; in C it has the pattern qetel; and in J qiite/ is, to all intents and purposes, exclusively reserved for Form I active participles, so that with a very few exceptions J nouns harking back to OA qatil have either qItel or qalel. The various degrees of contrast and non-contrast are illustrated by the following:
to say that all verbs admit of both, cf. 4.73 below) but the phonemic distinction is retained fully in the three forms (m.s., f.s ., and pl.) only by C; in J and M, it is to some extent blurrecj. The details follow. (i) For Form II, from the verb ' to hand, over', perfect base MJC-sallam, imperfect base MJC-sallem, we have in M:
'a clerk' 'having written'
M kateb kateb
'a mosque'
jame'
' having gathered' 'it's possible' 11 3 'having agreed'
jame' qabel qabel
J kateb keteb jime' jeme' qiibel qebel
C kateb keteb
jeme' jeme' qabel qebel
With hollow roots (KI-y or IV) the second radical yields /y/ in MJC but while the a of M-qatel is unchanged, the e of JC-qetel yie lds /ay/ : M/ sfiyel/ , JC/sayyel / ' transferring' (Syf), M/dayer/, JC/ dayyeg/ 'going around' (dIVr, dIVg) . With geminate roots (K2 and K, identical) the JC patterns are unmodified, but the masc. sing. of M exhibits a .collapsing of K2 and K, into a single phoneme: from /zbb we get M /I:tiib/; fern. Il)abba/ , but JCfl)ebeb/ fern. / l)ebbi/ . This M form 114 is typical of gelet and Beduin dialects, and I have noted it also among the Negev semi-nomads, while the JC forms are the ones found throughout the qe/tu area and in the sedentary dialects as a whole. (b) Form I passive participles have the pattern maqliil in MC and meqtiil in J and exhibit no differentiation beyond the regular phonetic ones in MJC; no modifications occur in geminate or hollow roots (MC/mahbub/, J/mel)bub/ 'beloved', MCfmabyu'/, J / mebyu'/'sold') or with KI-IV roots, MC/ mawzun/ 'weighed', though • 3.37b and note 86. With K,-y the pattern is maqtil on J/muwzun/ see e.g. MJCfmaswi/ 'fried', fern. M/maswiya/, J/meswiyi/ , Cfmaswiyi/. 4.72 Participles of Forms II and III. These have, in MJC, the regularly expected patterns involvi ng a preformative /m/- bound to the base of the imperfect fo r the active participle and to the base of the perfect for the passive participle. A ll three dialects distinguish the active participle pattern from ' the passive pattern (which is not
m. S.
AP PP
. msallem !./ msallam
f.s.
pI.
msalma
msalmin
msalma or msallama
msalmin or msallamin
Thus the distinction between active and passive is optionally retained in the fern .: sing. and in the plural. In J, on the other hand, we get: pI. f.s. m.s. AP PP
msallem msallam
msalmi msalma
mselmin mselmin
Thus the distinction is lost in the plural, but retained in the fern . sing. because of the different allomorphs of the fern. suff., depending on whether the base ends in -eK or in -eK (cf. 4.32 above). In C the distinction is retained throughout: m.s.
AP PP
msallem msallam
f.s.
msalmi msallami
pI. msalmln msallamin
Form III participles behave enti rely analogously: ' fro m the verb 'to forgive' , perfect base MJC-sdma/z, imperfect base MJC-sdme/z, we get the masc. sing. AP MJCfmsameh/ , the masc. sing. PP MJC / msiimal:t/ and fem. and plur. forms exactl y anj'logous to the ones listed above. (ii) There are no modifications for hollow or geminate roqts; 115 with K 3-y roots, the last radical merges with the last pattern element so as to yield /i/ in the AP and /a/ in the PP masc. sing.: MJC/mxalli/ 'putting', /rnxalla/ 'having been put' (xly). The fern. sing. of J in such cases obliterates the gender distinction in the AP (fmxalli/ is also f.s.) but retains it in the PP (J/ mxelliiyi/ is the f.s.) and rein-
"
I
••
-~
., 96
97
COMMUNAL DIALECTS IN BAGHDAD
MORPHOLOGY
troduces it in the PP plural Imxellayin/,. Again, Form III is entirely analogous: J/m'afil 'giving health' (m.s. and f.s.) for the AP, (root 'fy), with the PP J/m'Mal (m .s.), and Im'afayil (f.s.), Im'afayin/ (pl.). The corresponding MC forms were not noted. 4.73' Other participles. (a) Verbs of Forms IV-X do not yield pa rticiples as freely as Forms I-III. Many do not seem to have participles at all , and most seem to have lost the active-passive contrast, having a single participle which , in different verbs, may be formed o n the perfect base in - aK or on the imperfect base in - eK ; this seems to apply equally to M, J, and C, but because of the detailed treatment of the J verb in K ohen already referred to several times, data on J are more abundant and will therefore be somewhat disproportionately detailed in what follows. (b) There are no Form IV verbs to speak of in M and especially in JC (see 4.84); the only participial form noted is M/l)len!i/, pI. /l)len!in/, from M/en!a/ 'to give', for which JC ha ve /!a'a/. (e ) Forms V a nd VI seem to have lost the active-passive distincti on and to have participles formed on the imperfect base only, patterns MJC-melqatte/ and melqiilei, respectively: MJC/met'allem/ 'having learned, studied', MCfmet'ahed , /J/ met' ahed / 'having made a contract' . With K ,-y, J shows the expected fiI in the masc. sing. (as do MC), but the fern . and plur. have the -/ay/- infix noted for Forms . I1 and III passive participles: J/ metgaddi/ ' having eaten I unch', fern . / metgeddiiyi/, plur. /metgeddayin/ ; J/ meWiqi/ 'having met with', fern. / metiaqayi/, plur. /metlaqayin/' (d) Form VU participles are rare, a fact which is no doubt connected with the funct ion of Form VII as passive of Form I, so that Form I passive participles can usually be used instead: MJCfnjara!)/ 'to be wounded', / majril!)/ '(having been) · wounded'. However, in Kohen's work I find such forms as J/ men!itfi/ (m.s. and f. s.) '(having been) extinguished', (mensaqq/, fem. / menseqqayi/ '(having been) torn', and /mengad/, fern. /mengiida/ and /mengadayi/ ' necessary', which is perhaps not a true participle; the verb itself is more or less fossilized as JCfye ngad/ M/ye nrad / 'it is necessary'. (e) Form VII I participles are somewhat more common: M / menteQer/ 'waiting, having waited ', MCfmejteme'/, J/ mejtiime'/ ' gathering, having gathered' . They show no active-passive distinctio n, but while most are formed on the imperfect base, J/ mettitham/, fern. / mettahma/ ' (having been) accused' shows' the perfect base.
With K , -y, J/mestitgi/, fern. /mestgiyi/ is both 'having bought' and 'having been bought'; with geminate root, J/mestamm/ 'ha ving breathed' or 'smelled' has the fern. /mestemmayi/ and plur. /m.stemmayin/. (f) Form IX verbs have a single base for the perfect and the imperfect, hence a ~ingle participial form ; in Kohen's wo rk I find J/mel)magg/ ' having become red' , / meswadd/ ' having become black', etc. which, like Form VIII geminate roots, have a fern. /me!)meggayi/ and a plur. / mel)meggayin/. (g) Some Form X verbs seem to have retained the active-passive distinction: MJC/Iflesta'jel/ ' hurrying, havi ng hurried' /mesta'jal/ ' rushed ' (perhaps ao adj. rather than a part.) ; (or geminate roots , Kohen lists /mest'edd/, fern. /mest'eddayi/, plur. /mest'eddayin, ' (having been) prepared', and for K ,-y /mestenqi/ 'ha ving selected' and /mestenqa/ ' having been seletced' . 4.8 THE VERB. 4. 81 Overall view. The usual ten "stems" or derivational types, with their tradi tional designations of "Form I" to " Form X" are retained here. As in other dialects, the dual activepassive pattern oC the base that is found in Classical Arabic for each of the stems except Form IX, is absent, and present base patterns are reflexes of the OA active voice. Each of the ten derivational form s has its characteristic base patterns, one for the perfect and one for the imperfect and imperative ; the former will be called the "perfec t base," the latter the "imperfect base." Once the base pattern is given, the shape of the inflected forms can be deduced by referring to the lists of subject and object pronominal affixes and applying the various morphophonemic rules already discussed (see esp. 4.2). In what foll ows, full model paradigms for the sound root will nevertheless be listed, and the modifications occurring in the base patterns with va rious types of weak radicals wi ll be outlined. 4.82 Form I. (a) In Classical Arabic, ve rbs of Form I h.ave a perfect base either on the pattern qala/ (fkatab/ 'he wrote') or qati/ (flabis/ 'he wore') or qalll/ (fkabur/ ' he grew'). Many Eastern dialects, e.g. almost all Syrian and Egyptian varieties, have preserved a two-way qalal-qili/ contrast, with qiti! usually representing older qali! and qalll/ (fkata b/, / Iibis/, /kibirf) . In the Mesopotamian area, such a two-way contrast is found in some of the qe/Ill-dialects (A natolia, Mosul, and G) whereas other qe/III-dialects CA na , J) and all
98
99
COMMUNAL DIALECTS IN BAGHDAD
MORPHOLOGY
gelel-dialects so far examined have lost it. Thus both M and J have
lakall and laxa51 q.v. below: Iserabl 'to drink', Ikesarl 'to break', IketalJ 'to kill', Isema'i 'to hear' , Ige'adl 'to sit', Ise'all 'to be lit', Ige!a'i 'to cut', /,erafl 'to know', I
a single Form I perfect base pattern and C a dual one; however, it will be recalled (cf. 3.33) that M has introduced an innovation with respect to OAlal where it was in an open syllable and followed by another lal in the next syllable, the M reflex being lei or lui dependi ng on the flanking consonants. Returning to the three verbs just cited, we get the following picture in M1C:
'he wrote' 'he wore' 'he grew'
M
1
ketab lebas kul;>ar
katab labas kabag
C
katab lebes kebeg
The M perfect base. pattern is qelal (where e is lei or lui depending on the flanking co nso nants), the 1 perfect base pattern is qalal, and the two C perfect base patterns are qalal and qelel. The following table illustrates the M and 1 single base pattern; on the optional zeroing-out of M-e in unstressed open syllable indicated by the parentheses, see 3.5; on the obligatory zeroing-out of 1-0 in similar position, cf. 3.33. 1
M Is. 2m.s. 2f.s. 3m.s. 3f.s. Ipi. 2pi. 3pi.
k(e)tabet k(e)tabet k(e)tabti ketab ketbat k(e)tabna k(e)tabtu ketbaw
k(u) 1;>aret k(u) 1;>aret k(u)1;>arti ku1;>ar ku1;>rat k(u)1;>arna k(u)l;>artu ku1;>raw
ktabtu ktabt ktabti katab katbet ktabna ktabtem katbu
In l, Ikabagl and Ilabasl have inflections exactly like that of Ikatab/. Like Iku1;>ar/, i.e. with lui in the first syllable, are all those M verbs with [ul-coloring K, and K" e.g. I!ufarl 'to jump', l!uQaxl 'to cook', Igumazl 'to leap', Ikufarl 'to disbelieve', I~u I;>arl 'to be patientl , Imu!ratl 'it rained', IquQa
Is.
I
I
katabtu katabet katabti katab katabet katabna katabtem katabu
2m .s. 2f.s. 3m.s. 3f.s. 1pi. 2pi. 3pi.
Ibestu Ibeset Ibesti lebes lebset Ibesna Ibestem lebsu
A noteworthy feature of this duaL pattern in C is its instability: my C informants report that one can also say Ilabasl, as well as Ilebesl , Isama'i as well as Iseme'l , Ikabagj as well as Ikebeg/ , etc.
. ,I !
(b) The imperfect base has a pattern MlC-qIE/, in which E yields lei or lal in lC, but lal , lei, or lui in M. The phonemic shape of E is not predictable from the nature of the radicals involved, except that if K, is /1 or ' it .seems to be always lal in the three dialects. Whatever its phonemic shape, it is stable in C, zeroed-out in open unstressed syllable in l , but in M it is zeroed-out if lelor lui ana optionally retained ifit is la/. The addi ti on of object pronouns always results, in lC, inth,; stressing (and hencel non-elision) of the base final syllable: M/te'ruf/ , l Cjte'gefl 'she knows', M/te'urfa/ , lCjte'gefu/, 'she knows him', Morphophonemic clusters resulting -from the various inflectional
affixes are split in ways already described (3.5, 4.2). The anaptyctic vowel preceding K, is lei, though there is some tendency among M speakers to replace it by lui when E is lui : M/te'rufl or Itu'ruf 'she knows', Iteknusl or Ituknusl 'she sweeps'. Inflecti on of the base MlC-ftab 'to open' is as follows: I
~
. .'-..
I
I
100
M
J
M ekteb e[(etbi or ketbi eketbu or .' ketbu
C m .S.
Is. 2m.s. 2f.s. 3m.s. 3r.s. IpJ. 2pJ. 3pJ.
aflab teflal) teflal)in or tfet l)in yeftab teftal:J
neftal) teftabu n or tfell)un yeflal)On or yfetbun
aftab teftab tfc tb en
aflab teflab teftabin
yeflal1 teftal) nefta b tfetl)o n
yefta b teftal) neftal) teftai)un
yfetbon
yfetbun
M Is. 2m. s. 2r.s. 3m.s. 3f.s. IpJ. 2p J. 3pJ.
akteb tekteb tketbin yekteb tekteb nekteb tkelbOn yketbun
J a'rur tc'ruf
t'urrin ye'ruf tc'ruf ne' rur t'urron
y'urfUn
f.s. pJ.
{
When E is lei (e.g. MJC-kteb 'to write') or, in M, 'to know ', for wh ich JC-'geJ), we have:
akteb tekteb tketben yekteb tekteb nekteb tketbon yketbon
101
MORPHOLOGY
COMMUNAL DIALECTS IN BAGHDAD
lui, (e.g.
M-'ruf
C akteb tek teb tektebin yekteb tekteb nekteb tektebOn yektebOn
The imperfect base is also used to form the imperative. [n J, the base is unmodified and the s uffixes !if for the fem. and lui for the plu ra l are added 10 it directly. [n C, the fern. and plur. are as in J, but th e masc. sing. has a prefixed Ie/. In M, this prefixed element is found in all three forms, but since it is stressed the base vowel, being in unstressed open syllabic in the rem. and plur., is zeroed-out, re~ suiting in a KKK cluster and anaptyxis. However, M has variants for th e fe m. and plur. wit hout prefixed lei but lVith retention of what must have been the anaptyct ic vowel, no doubt because of the tendency in M to stress that vowel when in stressable position: M/ketbij < */eketbil < leketbil < */ektbi/.
J kteb kteb i
C ekteb ktebi
ktebu
klebu
With base vowel ,(ul in M, the forms are analogous to the above except that there is a prefixed lui instead of lei. With base vowel lal, the JC forms are exactly analogous to the above, but M again shows the co-existence of two systems, o ne eliding lal in unstressed open syllable (yielding fO~j11s analogous to the above) and one retaining it:
m.s. f. s. pJ.
eftab ertabi or ereti)i or feti)i eftai)u or efetbu or fetbu
When object suffixes are added to the imperative, there are largely predictable modifications th at 'may be summari zed as follows : (i) In the mase. sing., no modification in J, stress shift with
option'al loss of prefixed lei in C, varyi ng position of anaptyctic vowel in M depending on syllable stru cture:
' write it (m .)' 'write it (f.)'
M eketba ektebha
C (e)ktebu (e)kteba
J ktebu kteba
(ii) In the fern. and plur. , th e fi nal vowel is lengthened and stressed, and in J there are the quality changes that have already been noted (cf. 4.2) : M
C
J
~
f.s. f.s. pI. pI.
+ 3m.s. + 3r.s. + 3m.s. + 3r.s.
(e)ketbi (e)ketbiha (e)ketbO (e)ketbuha
ketbinu ketbiya ketbunu ketbuwa
ketbenu ketbiha ketbonu ketbOha
The negative.imperative is formed in the same way in M, J, and C,
,
; ;
102
COMMUNAL DIALECTS IN BAGHDAD
•
viz. by prefixing Iial (often stressed) to the im perfect: MJC/latektebl 'don't (m .s.) write', etc. (e) Modifications of the above patterns in Form I ver bs with KrY roots are illustrated in the following table, taking the roots bny 'to build' and bqy 'to remain'; both are exactly identical in JC and in M the sole difference is the [u)-coloring nature of the firs; two radicals of bqy: M Is. 2m.s. 2f.s. 3m.s. 3f.s. I pI. 2pl. 3pl.
f .
MORPHOLOGY
J
C
bnetu bnet bneti bana banet bnena bnetem banu
banetu banet baneti bana banet banena banetem banD
M Is. 2m .s. 2f.s. 3m.s. 3f.s. I pI. 2pl. 3pl.
103
J
abni a,bga tebni l tebga tebnin tebqin .~ " yebni yebga tebni teega nebni nebga tebniin tebqiin yebni n yebgiin
abni tebni tebnen yebni tebn i nebni tebno n yeb non
C abqa tebqa tebqen yebqa tebga nebqa tebgon yebgon
abni tebni tebnen
,
yebni tebni nebni tebnon yebnon
abga tebga tebqen yebqa tebga nebqa tebqon yebqon
, i
b(e)net b(e)net b(e)neti bena benat b(e)nena b(e)netu benaw
b(u)qet b(u)qet b(u)geti buga bugat b(u)qena b(u)getu bugaw
The imperatives of such verbs are best described as formed on the corresponding imperfect fo rms (3m.s., 2f. s., and 3pl.) by dropping the preformative ana (where there is one) the -/nl of the suffix. They thus reveal the historical process more clearly than the sound root im peratives, whose formatio n has been synchro nically described in somewhat different terms in (b) above. In M and in the masc. sing.
of C there is here again the prefixed le/- :
Thus the perfect base patte rn M-qelal, JC-galal, is unmodified in the 3m.s., though K J is, in that case, zero; in the 3f.s. and 3pl. both the second pattern vowe l and th e third rad ical are zero, and in the C
3pl. the pron. suff. is 101 rather than lui. In the other persons there is the eXpected merger of the pattern element a with the radical y into Ie/. As in tbe sound roo t, C has a number of Kry verbs on the pattern "elel, though here again it appears that they have optiona l aiternants in qalal. Thus fro m msy ' to walk' there is C/mesi/, but also Imasa/ ; in the qelel paradigm we get: Imsitu/, ImSit/, Imsiti/, Imesil, Imesyet/, Imsina/, ImSitem/, ' /meSyul (3pl. not actually noted). The second pattern element e thus merges wit h y into /if (Iii in unstressed final position in the 3m.s.) and is zeroed-o ut (with Iyl for the third rad ical) in the 3f.s. and (p resumably) in the 3pl. As for the imperfect base pattern qlE/, it yields two phonemic shapes that are not predictable: one in which E merges with y into Iii, and both are zeroed-out befo re the VII suffixes; and a seco nd in which
E
is
fat, in final position, zero before th e - VI1 suffixes, and y is zero throughout; that C has -/enl and -Ionl in K J-y verbs has already been mentioned (4.2). Using t he same two roots bny and bqy to illustrate, we get:
~
m.s. f.s. pI.
ebni ebni
ebga ebqi ebou ebgu
J bni bne bno
C bga bge bgo
ebni
bne bno
ebqa bqe bgo
Because of the different syll able structure, M has no optional alte rnan ts without the'prefixed le/- . In J, the final vowels, being stressed, have retained their older quality (cf. 4.21 b(iv)) : there is the usual alternation before - fla: J/bninul 'build (m.s.) it (m.s.)', Ibnenul ' build (f.s.) it (m:s.)" but Ibnihal 'bu ild (m.s. and f.s.) it (m .s.)'; Ibnonul 'build (pl.) it (m.s.)', but Ibniihal ' build (pl.) it (f.S.),."7 . (d) Verbs witl" K,-y and K,-w exhibit no modifications in the perfect, except that with K ,-,v M fiuctuates between Iwe/- and Iwu/- ; M/wuga'i and Iwega'i 'he fell', JCfwaqa 'l ; M/yebas/, JC Iyabasl ' he dried (intr.)'. There are, however, a number of modifications in the imperfect, all connected with the combination of the consonantal preformatives with the initial y or IV of the base; in JC, the preformative la/- causes no modification. In M and C the combination K + IV yields MI Kol and C/ Kiil throughout, whereas in J pho nemic IWh occurs througho ut and the resulting clusters
,
.''''
104
COMMUNAL DIALECTS IN BAGHDAD
KIVKV and KIVKK are split regularly into IKewKVI and
IKweKK I
IC/amagl 'to order' . Besides the fact that the first radical yields
throughout. II B M Is. 2m.s. 2f.s. 3m.s. 3f. s. IpI. 2pJ. 3pJ.
Imperatives
aga'
I awqa'
C awqa'
toga' Hig'jn yoga'
tewqa'
tiiqa'
tweq'en yewqa'
tiiqa'In
toga' noga'
tewqa'
yiiqa' tiiqa'
tag' un
newqa' tweq'on
tOqa'un
yog'iin
yweq'on
yiiqa'iin
nuqa'
h~ve been noted only for M and I:
m.S.
M oga'
f.s. pI.
og'i og'u
I
M oybas tebas tobsjn yobas tobas nebas tobsiin yobsiin
(
phonemic zero initially, the first two verbs exhibit a number of special features, while the third behaves more like sou nd root verbs. In the perfect, the C inflection shows no modification except for the automatic assimi lation of the Idl of laxadl to a following ItI ; the I inflection shows a similar assimilation (/ot/ > Itt/) and, in addition, has a stable initial lal even in unstressed syllable ; in M, the verbs 'to eat' and 'to take' have a perfect base qatal rather than the normal qelal, though I am told that one can occasionally hear lekall and lexa51 as well; the verb 'to order' has the normally expected qelal, with e yielding lui before 11111. Thus M/akitlet/, ICfakaJtul 'r ate', M/nklat/, I Cfakletl 'she ate', etc. M/ur[laret/, IC/amagtul 'r ordered', M/Ur[lrat/, IC/a mgetl 'she ordered'. As for the imperfect, the first two of the three verbs have, as in most dialects not a base qlEI but a base aqlE/, with the a of the pattern and the ' of the root merging into lii/, and the E yielding lui in M and lei in JC:
wqe'
wqe'i
M
I
C
wqe'u
While there are several such verbs with K,-IV <M/wu~all, Ilwa~all 'he arrived', M/ wugaf/, Ilwaqafl ' he stood up', and perhaps half a dozen morel only M/yebas/, Ilyabasl has been noted for K,-y. The imperfect is, in M, analogous to that of K,-IV, i.e., has leland in I there is an irregularity whereby KyKV yields not */ KeyK VI, but I KeKV/ ; the C imperfect has not been noted:
Is. 2m.s. 2f.s. 3m.s. 3r.s. IpJ. 2pJ. 3pJ.
105
MORPHOLOGY
I aybas tebas tyebson yebas tebas nebas tyebson yyebson'
<el Verbs with roots K,-' seem to include, in Form I, only three verbs, MJC/akal1 'to eat', MJ/axao/, Cfaxadl 'to take', and M/ ur[lar/,
Is. 2m.s. 2f.s. 3m.s. 3f.s. I pI. 2pJ. 3pl.
akul lakul taklin yakul takul nakul tiikliin yakliin
akel takel taklon yakel takel nakel taklon yaklon
akel takel takJjn yakel takel nakel tiikliin yakJfm
The imperative has the following forms:
m. s.
f.s. pI.
M
I
ukul , ukli uklu
kel keli kelu
C
kel keli kelu '
As for the third verb, its imperfect has a regular base '1IEI, with yielding lui in M and lei in IC: M /tu'r[lur/, JC/te'meg/, Ilt'emgon/, Cfte'megiinf. However, in the 1st sing. I has lamegj; imperatives seem
E
.,'i ~,
106
COMMUNA L DIALECTS IN BAGHDAD
to be regular, M/u'l11Ur/, Jl'megj, however, J also has a form lamegl side by side with I' meg/, and M has the variant l af[lur/. (f) Verbs with hollow roots show less differentiation iQ the three dialects. There is the fami liar monosyllabic perfect base with lal in the 3rd person forms, and an invariant lei (with, in M, some traces of the lei ~ lui alternation"') in the other persons: MJCjnaml ' he slept', M/nemetl, JCjnemtul 'I slept' . The imperfect has the equally familiar base with lal or lui with K,-IV and Iii with K,-y, e.g. MJCjynaml ' he sleeps', Inuml 'he fasts', Iyjibl 'he brings'. A feature characteristic of M and occurring to a lesser degree in JC is the zeroing-o ut of the imperfect base vowel in the forms with the -/inl and -/unl suffixes. This seems to occur only in a few high-frequency verbs and side by side with the full forms: M/tgulinl and Iteglinl 'you (f.s.) say', Itgulunl and Iteglunl 'you (pl.) say'; Iygulunl and lyegliJnl 'they say'; M/tridinl and Iterdinl 'you (f.s.) want' , Itridunl alid Iterdunl 'you (pl.) want', Iyridunl and Iyerdunl 'they want'. My C and J informants state such forms do not occur in their dialects, but I have heard them use them occasionally, perhaps as a result of M influence. The verb 'to say', M/gal/, JCjqalj, exhibits some peculiarities when the morpheme L plus pron. suff. are added. Whenever a long vowel occurs in the base, it becomes shortened (fal to la/, lui to lui in M, lui to lei in JC) and the I of the root is, in such cases, geminated when intervocalic:
'he told me' 'she told me' 'they told me' 'you tell me' 'tell (m.s.) me' 'tell (f.s.) me' 'tell (pl.) me'
M galli gallatli galloli tgulli gulli gullili gull uli
J qalli qelletli qell oli tqelli qelli qelleli qelloli
107
MORPHOLOGY
C qall i qalletli qalluli tqelli qelli qellili qelluli
Other forms of the imperfect and forms with the other object pron. suff. behave analogously. One can occasionally hear such fuller forms as M/gulilij. (g) The verb 'to come' must, as elsewhere, be treated separately. The perfect has a monosyllabic base in lC, but both monosyllabic and disyllabic variants in M in the 3rd pers. forms:
• M Is. 2m.s. 2f.s. 3m.s. 3f.s. I pI. 2pl. 3pl.
(
I
det
o
jet ' jeti eja or ja ejat or jatti jena jetu ejaw or jawwi
1 jitu jit jiti ja jet jina jitem
jo
C jitu j it jiti jii jet jina
jitem jo
The M forms /iawwij and /ialtij are unusual, and may have developed from sandhi forms , e.g. */jawelyom/ > */jawwelyoml > Ijawwilyoml 'they came today'; with suffixes, the anomalies of the M 3rd pers. disappear: Ijakl 'he came to you', Ijatakl 'she came to yo u', /iokl 'they came to you. 120 The imperfect conjugates as follows: '!
M 1s. 2m .s. 2f.s . 3m.s .. 3f.s. Ipl. 2pl. 3pl.
,aji
teji tejin yeJ I teji neji tejun yjun
1 aJI teji tejen yeJI teji ne]1
tejon
yjon
C aji teji tejen yeJI teji neji tejon yjon
As in other varieties of Eastern Arabic, the imperative is formed on a wholly different root: MCjta'al/, Ita'ali/, lta'alu/, but Iltal/, Ita iii , Italul with: unusual loss of the pharyngeal. (h) With geminate roots, the three dialects exhibit little differentiation. The perfect base has the pattern qatl (qalt) and there is insertion of le/ between the base and the pron. sulf. other than those of the 3rd pers. F ~om sdd 'to shut', we get:
,
M >addet saddot saddeti sadd
Is. 2m.s. 2f.s. 3m.s.
.
,
1 seddetu seddot seddeti sadd
C saddotu saddet saddeti sadd
., I J,
"
~ -. .~
lOS 3f.s. I pI. 2pl. 3pl.
COMMUNAL DIALECTS IN BAGHDAD
saddat saddena sadd;;tu saddaw
saddet seddena seddetem saddu
2f.s. 3m.s. 3f.s.
saddet saddena saddetem saddu
M . asedd tsedd t'eddin ysedd tsedd nsedd tseddun yseddun
1 asedd tsedd tsedden ysedd tsedd nsedd tseddon yseddon
C asedd tsedd tseddin ysedd tsedd nsedd tseddun yseddun
The imperative is as in the sound root: M1Cjsedd/, Iseddil , Iseddu/. (i) T here seems to be only one verb with K 2 -' in co mmon use , na mely M/se'all ' to ask' which behaves entirely like a sound root: Ise'lati 'she asked', Ise'lawl ' they asked', Is(e)'altil 'you (r.s.) asked', Itse'lun/ 'yo u (pi.)' ask', etc. The lC equivalent is Isayal/ , a regular Form III verb. 4.S3 Forms II, I ll , V, and VI, and quadriconso nanta l verbs. (a) Verbs belonging to one of these four categories show little differentiation in M1C and few modifications result ing from the presence of special rad icals. Form II verbs have a perfect base pattern MJC-qallal, imperfect base qallel; the behavior of a when unstressed is regular, and the e of the M imperfect base yields either lei or lui depending on the flanking co nsonants. Thus from kml 'to complete' we get the perfect: 121
Is. 2m.s.
M kamlllalet kamma let
1 kemmaltu kemmalt
kammalti kammal kammelat or kamlat kammalna kammaltu kamrnelaw or kamlaw
I pI. 2pl. 3pl.
The imperfect base behaves much as in the sound root, except for the fact that K2 and K3 are never separated (the pattern is qEII instead of ql EI) and that the base vowel, which may be lal, lei, and in M also lui is not necessaril y lal when K2K3 arc pharyngeals (e.g. M/ygul,ll,l/, l Cjyqel,ll,l1 'he coughs').
Is. 2m.s. 2f.s. 3m.s. 3f.s Ipl. 2pl. 3pl.
109
MORPHOLOGY
kemma lti kammal kamlet
kammalti kammal kammalet
kemmalmi kem maltem kamlu
ka mmaltem kammalu
ka~'malna
As for the imperfect, taking the root kllli aga in and for [ul-coloring, in M also the r7?t 'mr ' to build', we get:
( I
•
, I
M Is. 2m.s. 2f.s. 3m.s. 3f.s. I pI. 2pl. 3pl.
akammel tkammel tkamlin ykammel tkammel nkammel . tkamlun ykamliin
C
\
kammaltu ka llllllalet
I
n'ammur
t'amrun y'amrun
1 akammel tkammel tkemlen ykammel tkammel nkammel tkemlon ykem lon
C akammel tkammel tkamlin ykammel tkam mel nka mmel tka mlun ykamlun
The imperatives are formed as expected : M1Cjkammel/, Ikamlil,
Ikaml uj. However, a salient difference between M on the one hand and lC on the other is that the e of the imperfect base is stressed with all obj . suff. in lC, but is zeroed-out when in unstressed open syllable in M, so that we get:
'he'll finish it' 'he' ll fini sh them' 'finish (m.s.) it'
I
a'ammur t'ammur t'amrin y'ammur t'ammur
M ykamla yka mmelhum kamla
1 ykemmelu ykemmelem kemmelu
C yka mmelu ykammelem kammelu
With K ,-y roots, the perfect shows the same modifications as in Form I, and the imperfect shows the same modifications as Form I imperfects of the type Iyebnij. Thus from xly 'to put' we get M1C Ixallal ' he put', M/xallaw/, llx.allu/, C/xallol 'they put', M/xallet/, llxelletu/, Cjxa!letiil 'I put'; and in the imperfect, M1 Cjyxallii 'he ' puts', M/yxallun!, llyxellon/, Cjyxallon/, a nd the regular imperatives
JIO
COMMUNAL DIALECTS IN BAGHDAD
, I
MJCfxall if (m.s. a nd f.s.) f xalluf . However, note that the masculine-
feminine contrast reappears in the J imperative when obf pron. suff. are added : Jf xelilnuf ' put him', vs. Jf xellenuf 'put (f.s.) him' ; as has been seen, f ef does not occur in final unstressed syllables and the unstressed forms with f if or f uf (cf. fxalluf but f xellonuf 'put (pl.) him' may represent older forms in f ef or f of. No modifications take place in verbs having roots with Kt-y or w, with K 2 -yor w, K 1-' or having geminate roots : MJC:fwassal{ 'to escort', fy abbasf 'to . dry (trans.)', fxawwaff 'to frighten', f nayyamf 'to put to sleep', f ajjarf 'to rent', fjannanf 'to madden', (see note 115) conjugate like MJCfkammal{. (b) Quadriconsonantal roots have conjugations ak in to Form II in all respects, except for tho automatic modifications resulting from the presence of the extra consonant: perfect base qarlal, imperfect base qarlel: MJCfda'balf ' he rolled', f yda'belf ' he rolls'. Verbs such as MJCf nesanf 'to betroth', fdoxanf 'to feel dizzy' can be considered as having roots with K 2 -y and K 2 -w, respectively, with the familiar merger of a + y into f ef and a + IV into fof ; such verbs have the normally expected conjugatio ns, except that in J fef and f of remain stable when unstressed: f nesimtuf , f doximtuf. (e) Form III verbs have a perfect base pattern M JC-qiilal and an imperfect base pattern MJC-qiilel, with a behavior differing from Form II only in ways that are entirely predictable from the presence of the long vowel in place of the ilK sequence in the base. Thus fsamabf 'he forgave', Mfsam!)atf, JCfsam!)etf 'she forgave', Mfsamabetf, Jf sama!)tuf , Cf sama!)tuf 'I forgave', etc. In the imperfect M again has either f ef or f uf for e depending on the flanking consonants, MJCfysame!)f 'he forgives', Mfyjawubf, JCf yjawebf 'he answers'; MCf yjawbOnf , Jf yjawbonf "they answer', etc. Modifications with K ,-y are as for Form II, and other radical combinations require no special modifications,I2l (d) Verbs of Forms V and VI bear a formal and usually also a semantic relation to verbs of Forms II and III , respectively, in that they use the same perfect base with a prefixed 1- and are usually the ret1e'xive, passive, o r intransitive equivalents of the forms without 1- .
The perfect and imperfect bases are identical: MJCftkammalf 'was completed', fyetkammalf 'will be completed'; ftsama!)! 'was forgiven', fyetsama!)f 'will be forgiven'. Quadriconsonantals have similar equivalents with prefixed 1- : fdda'bal{ 'he rolled (intr.)', f tnManf 'be-
III
MORPHOLOGY
came betrothed'. With K,-y, modifications are as in Form II, except that since the imp~rfect base vowel is a, it remains f af in the imperfect and imperative fonns without suffixes, is zeroed-out before suffixes, and y is zero throughout: from gdy 'to eat lunch ', MJCfatgaddaf Jf tetgeddonf , Cftetgaddonf 'you (pl.) ' I eat lunch', MftetgaddOnf, , eat lunch', etc.
I (
,
.
4.84 Form IV .. : As in most dialects, Form I'V verbs have left practically no trace ,.in MJC and where they have, can be recognized only by the perfect and the participle (cf. 4.73): M fenletf ' I gave', fenlaf 'he gave' fanlif 'I give', f yenlif 'he gives', fmen!i! 'having given', fen (.if, fenluf 'give', etc. The perfect fenlaf, 'he gave', f enletf 'I gave', etc. has, moreover, the alternant form fnelaf, f n(e)!etf , etc. in the paradigm of Form I, evidently reSUlting from the interpretation of the unstressed f e( of fenletf , etc. as anaptyctic. The. analogy is, of course, with K,-y Form J verbs: f nesetf and fnsetf 'I forgot' , hence also f(e)nletf and f neletf 'I gave', and f ne!af like f nesaf 'he forgot'. Rusafi (1928, p. 422) does not even mention fen!af as a Form IV verb, and mentions only two partially fossilized ones, i.e. with only
some remnants of a paradigm: f aS9al!f ' he became' and fagba!! 'he came forward ', to which we m.ay add fawda'f 'he took leave of'. Even these few forms seem absent from J and C; the JC ve rb 'to give' is f la'af . 4.85 Form VII. (a) Verbs of Form VII relate to Form J verbs in shape and often. also in meaning. They are often passive or intransitive equivalents 'Of Form 1 verbs of the same root, and have a
I J
perfect base patter.n consisting essentially of the Form I perfect base plus a prefixed 11- ; the imperfect base pattern does not relate directly to the Form I imperfect. The M perfect base pattern is thus IIqelal with the usual twin values for e, yet without its elision in unstressed open syllable, the 11+ K, sequence always yielding unseparable clusters; the JC perfect base pattern is IIqalal, with the expected behavior of a for these dialects. This means principally that a yields zero in J in unstr6ssed open syllable, with resulting clusters and anaptyxis as predictable. From jrb and, for [uJ-coloring in M, also from x9~ we get the following paradigms of the verbs 'to be wounded' and 'to become distracted': M
Is.
njer a~et
J
nejral!tu
C njara!)tu
.,
1 112 2m .s. 2f.s. 3m.s. 3f.s. IpJ. 2pJ. 3pJ.
I
COMMUNAL DIALECTS IN BAGHDAD
njeritl)et njera!)ti njera!) njerl)at
nxul;>aset nxul;>asti nxul;>as nxul;>sat
njera~ na
nx ut.>a~na
njera!)tu njer!)aw
nxul;>ast u nxuf;>$aw
nejra!)t nejral:I!i njara!) njar!)et nejral)na nejra!)tem njar!)u
113
MORPHOLOGY
with prefixed n-) MJC/dazz/ 'he sent' , /ndaz'z/ 'h e was sent' ; however, the imperfect base is identical with the perfect base: MJC /yendazz/ 'he will be sent', MCjyendazziin/ , J/yendezzon/ 'they will
njara!)et njara!)ti njara!) njara!)et njara!)na nja ral)tem njara1)u
be sent'. ! (d) With hollow roots, the perfect base is rdated to that of Form
I: M/dar/ 'he turned (tra ns.), /ndfu/ 'he tu fn'ed (intr.)'; however, the /a/ is extended to the 1st and 2nd pers. a's well (> /a/ when unstressed in J), and the subject pron. suffixes are preceded by /e/ in the perfec t as in KrY and gem inate roots;
The im perfect base pattern is nqalef in J, but nqelef in Me. The a of J behaves normall y for tha t dialect, the e's of J and C are always lei, those of M yield /e/ or / u/ depending on the flanking consonants; the initial cluster n+ K, is always unseparable, all vocalic elements yield zero in unstressed open syll able. In J and C the vowel preceding the initia l cluster is never stressed, th e stress be ing on the first pattern
Is. 2m .s. 2f.s . 3m.s. 3f.s. I pI. 2pJ. 3pJ.
element unless the -VII suffix is present, in which case stress is o n 17,
i.e. J/a nha zem/, Cjanhezem/ '[ flee', J/tenhezmon/ , C/tenhezmin / 'yo u (f. s.) flee' . In M there is hesitation betwee n this pattern and stressing the vowel preceding the /n/ : M/ anhezem/ and /anhezemj. Taking up our paradigmatic examples, we ha ve in the imperfect (leaving the flu ctuatin g stress in M unmarked):
..,
M ndaret ndaret nd areti ndar ndara t ndarena ndaretu ndaraw
J ndagetu ndaget ndageti
ndag ndaget ndagena ndagetem ndagu
C ndogetu ndaget ndagoti ndag
,
ndaget
ndagena ndagetem ndag u
, I
M Is. 2m.s. 2f.s. 3m.s. 3f.s. IpJ. 2pJ. 3pJ.
J
a njerel) tenjere!) tenjerl)in yenjere!) tenjere!)
tenxul?~in
nenjerel).
nenxul;m$
anxu9u$
tenxu I;>us yenxuQus tenxuQus
tenjerl)un tenxu t;>$un ye njer!)iin yenX1I 9 $U n
anjare!) tenjare!) tenjer!)en yenj are!) tenjarel) nenjare!) tenjer!)on yenjer!)on
C a njere!) tenjereh tenjer!)in yenjere!) tenjere!) nenjere!) tenjer!)iin yenjerl)iin
(b) With K ,-y, modifications are as in Form I verbs of the type / bana/ (fbena/l /yebni/ : M/n'ema/, JCjn'ama/ ' he was blinded', M/ n'emet/ , J/ ne'metu/ , C/ n'ametu/ 'I was bl inded'; M/an'emi/ , J/an'a mi/, Cja n'emi/ ' I shall be blinded'. There are no special modifications with the few instances noted with K,-' and K,-IV, viz. J/ n'akal/ 'was eaten', J/ nwalad/ 'was born'.'23 (e) With geminate roots, the perfect base is again that of Form I
l
The same base pattern is used in the imperfect : M/a ndar/ , JCjandag!, M/yendariin/ , J/yendagon/ , C/yendagiinj. 4.86 Form VIII. (a) This form is in many respects similar to the preceding: the perfect base pattern is based on Form I, fro m which it differs merel y by the insertion of t- between K, and K,; . however, it bears no regular semantic relation to Form I verbs of the same root. The perfect base is thus M-qtetaf, JC-qtatal. The pattern elements behave regularly, whic h means illle,. alia that there is in J an initial morphophonemic KKK cluster due to the elision of the first a; this cluster splits in the regular fashion into /KeKKf. Thus (for [uJ-coloring in M, e.g. fHuQar/ , cf. Malai ka, 1963, p. 59.) from film, the paradigm of the verb 'to understand ' is as follows:
Is. 2m .s. 2f.s. 3m.s.
M ftehamet ftehamet fteha mti fteham
J fet ham tu fethamt fetha mti ftaham
C ftahamtu ft.hamet ftahamti ftaham
•
, . 114
COMMUNAL DIALECTS IN DAGHDAD
MORPHOLOGY
115
'.
3f.s. I pI. 2pl. 3pl.
ftehmat ftehamna ftehamtu ftehmaw
ftahmet fethamna fethamtem ftahmu
ftahamet ftahamna ftahamtem ftahamu
the 1st and 2nd pers, subject pron. suff. : from .fr (.fg) M /~fa rr/, JCMaggl 'he grew pale', MMarret/, Jhfeggetu/, C/~faggetul '1 grew pale'; M /ye~farr/, JC/ye~failgl 'he grows pale', etc,; similarly M/lra55/, JCJlgassl 'he became deaf' , No roots with weak KI or K, and no geminate roots
The imperfect base pattern is again similar to Form VII, i.e. qtete/ in MC and qtatei in J, with the same stress variations as in the Form VII imperfect. The second e of M is, in this pattern, subject to [u]coloring, which in turn also modifies the first e to luI (cf. 3.52a); in JC sg/ behaves exactly as does film:
Is. 2m.s. 2f.s. 3m.s. 3r.s. Ipl. 2pl. 3pl.
M
J
C
aftehem astugu! teftehem testugu! teftehmln te,tug!in yeftehem yestugu ! teftehem . testugu! neftehem nestugu! teftehmun testug!un yeftehmun yestug!iin
aft ahem teftahem teftehmen yeftahem teftahem neftahem teftehmon yeftahmon
aftehem teftehem teftehmln yeftehem teftehem neftehem teftehmiin yeftehmun
(b) Modifications in roots with K,-y, in geminate roots and in hollow ro"ts are as outlined for Form VIl above. Thus from ! ry, . sgy, we have M/stera/, JCfstagal 'he bought', M/yesteri/, J/yestagi/, Cfyestegi/, 'he buys', etc.; from smm, weget MJC/stamml 'he smelled', M/stammet/, J/stemmetu/, C/stammetu/, '[ smelled', MJC/yestamml 'he smells', etc. From 'IVZ, we get MJC/,tazl 'he needed', M/' tazet/, J/, tazHu/, C/' tazetul '[ needed', MJC/ye'tazl 'he needs', etc. No verbs with K,-y or K t -' have been noted; in the few instances of Kt-w, the IV yields ItI as in CI. Ar., resulting in a Itt/ cluster which is never separated , and in other respects the conjugation is regular: from wfq we get M/ttefaq/, JC/ttafaql 'he agreed', M/yettefeq/, J/yettafeq/, C/yettefeq/ ' he agrees', etc. 4.87 Form [X. This form is entirely connected with names of colors and infirmities. It has a single base pattern qtall for the perfect and imperfect, and inserts leI in the perfect between the base and
( i
I •
hav~
been noted, and there are no modifications with
hollow roots: MJC/swadd/, /yeswaddl 'to become black' (root swd), 4,88 Form X, , Only a few verbs of Form X have been noted, and little can be said about them beyond the fact that they show a perfect base pattern MJC-staqta/ and an imperfect base . pattern MJC-staqte/: MJCfsta'jall ' he hurried', Iyesta'jell ' he hurries'. With K,-y, note J/stanqa/ ' he se lected', Iyestanqil ' he selects' (root nqy). With geminate anel hollow roots, modifications, are similar to those of Forms VII and VIII , except that the element a of sta- is now in unstressed open syllable and so is zeroed-out in J : MCfsta'add/, J/st'addl 'he got ready', MCfyesta'edd/, J/yest'eddl ' he gets ready'; M/starah/, J/stgah/, 'he rested', J/ yestgah/ ' he rests', There is thus in J an unusual initial KKK cluster /stK/ which seems unseparable, though there is something sim ilar in Form VIII verbs, e,g. J/stgaltu/ ' I worked', /stgetu/ ' [ bought', VS, /fethamtu/ ' I understood'. 4.89 Verb modifiers. (a) Present markers. The , morphology of the verb would be incomplete without some reference to the preposed markers that may. accompany the imperfect, imperative, and perfect, especially since there are some differences among M, J, and C in
,
this respect. Perhaps the most co mmon of these modifiers are what we may call the " present markers," i.e. the morpbemes J-qad, C-qa, M-da, and gd'ed: Some details on their morphophonemics and function follow . (i) [n J and C, the imperfect may be preceded by the morpbemes J-qad and C-qa to specify, roughly speaking, present time and non-contingency, [n J the allomorphs are /qad/ before the /a/of the 1st sing., /qa/ with gemination of a following t or II before KV, /qa/ in other cases;12' in C tbere is /qa/ througbout with optional /'I in the 1st sing, : J/qadamsi/, C/qa{,)amsijl" 'I'm walking', J/qattemsi/, C/qatemsi/ 'she's walking', J/qayemsi/, C/qayem ~i/ ' he's walking'; JC/qatgid/ 'she wants', /qayqul/ 'he says'. This morpheme seems always absent when an imperfect is syntactically d'ependent on a preceding verb, JC/qaygld yemsi/ 'he wants to walk', (ii) In M, there is a (relatively infrequent?) marker /ga' ed/ preceding the imperfect with a function similar to that of the JC
116
MORPHOLOGY
COMMUNAL DIALECTS IN BAGHDAD
117
morphemes just mentioned, and a more common Ida/- with more ~omplex functions .. Thus M/,endak fass! zen enta maga'ed tetqaddaral yo u have somethmg good that you don 't appreciate'; this could a lso be rendered with Imadatetq addara/ , and Ida/- (with optional merger with the lal of the 1st pers.) is quite common where we would expect to find the present markers in IC: Idam si! or jda'amsi/, 'I'm wa lking', Idayemsil 'he's walkin g', etc. However, I have the impression that the M im perfect occurs more commonly without Idal
why don't you wri(e me the receipt so I can go' ; I~!abru yam'awdin! dogfu dafahhemkum/ i30 'wait, please, stop so I can explain'. From one C informant I have similar uses of Ida/, perhaps under M influence: Ijitu da'is~fakl or /jitu asiifakl or /iitu !)attanasOfakl ' I came to see you'; and l~aketOlak hale~kayyi date'gef sejjaweb/, or I ~at tante'gefl ' I told you this story so you know what to answer'. My other C informants use I ~attal, I ~attan/, or II.lattani/, a nd in J only II)ettil seems to' He' used, with Idal restricted to the optative use just
contexts ImplYing present time than do parallel constructions in
referred to. Of co urse th e contrast between o ptative and purposive
I and C; thus J/asqatgid/, C/sqatgidl seem the preferred constructions for 'what do yo u wa nt 7', whereas M seems to prefer Isetridl without
may not be clearly marked, cf. Idafahhemkuml :above, and the need
idaf. Even more striking is the fact that M/dal serves a number of other funCti ons, some of which are filled by a IC morpheme Ida/, on which see (b) below. (iii) Other Mesopotamian sedentary dialects so far investigated
(e) There is a n additional optative morpheme MIC-xal/{i) preposed to the 1st 'and 3rd persons of the imperfect with allomorph s as illustrated by the following: MJ C/xa llinaml ~ I et him slee p', Ixalletnaml 'let her sleep', MC/xallinamOn/, J/xallinamonl 'let them sleep',. MJC/xalli'anaml ' let n;e sleep', Ixallennam/, Iet us sleep'. In the first person this may precede Ida/- as indicated above: Ixaldanaml 'let me sleep', Ixaldannaml 'let us sleep'. The reduction of gemi nation is normal and automatic before KY: MJC/xaltektebl 'let her write' . The fully inflected imperative of MJC/xallal is used in a similar sense, though only when addressing, and making a request of, a specific person: Ixallini a naml '(you there) let me sleep'. (d) All three dialects have a preposed Imal with the 2nd person
In
for more investigation is obvious.
have a present mar ker of some sort, as seems to be the case with
sedentary dialects throughout all or most of the Arabic area: Mosul has a Iqal appare ntl y identical wi th that of J and thus similar to that of'.C, I=6 but in Anatolia there is Ikwa/- (Qarabas) or Iwal (Urfa, Swerek), 'Ana has Ijayl and Hit has Iqa'ad/. Rural Lower Iraq has Idal a nd /ia'ed/ , in Qal'at $iile!) Ihalj- or Ida/- with gemination of the following consonant. l27 The Beduin dia lects inves tigated by Cantlncau do not have any sort of present marker, and its presence or a bsence correlates fai rly well with the dichotomy Beduin vs.
of the imperfect used in directives th at seem somewhat more ener-
sedentary;I28 va riou s interm ed iate dialec ts, either "Bed uinized" or "sedentarized," natura l1y may be expected to present intermediate
features, as seems to be the case in M. (b) In J and C, there is a pre posed jda/- with the 1st pers. of the imperfect that has optative function: JC/daqO lj, Ida'aqiIiI ' let me say', IdanqOl1 ' let's say'. In M, there is a simi lar use of Ida/-, so that M/dangOlj corresponds to both JC/danq Ol1 'let's say' and Iqanqo l/ 'we say'; the tl istinction is, however, maintained in M in at least one direction, in that M/xal/- can precede Ida/- and make it unambiguously optative: IxaldangOI/- 'let's say', Ixalda'a l.lcil 'let me speak'; cf. the other form s of xal/(i) in (e) below. In addition, M/da/is also used to lin k an imperfect with a preceding verb in construction s
referring to purpose or intention: M/ ral) daysOfhuml 'he went to see them', side by side with /ra~ ysOfh uml perhaps with a shade of sema ntic difference; cf. a lso / dek tebli asCI-lwa~e l da'arubJl 29 'now
getic than the plain imperative : MJC/matektebl '(come on and) write'. This is phonemically not the same as the negation of the imperfect, which usually stressed the Ima/, MJC Imatektebl 'you don't write'.
1 ...
(e) The imperative is often accompanied, in MJC, by a preposed de, with allomori'hs Idl before Y, Idel otherwise: MJC/deniiml 'sleep', M/duklu/, JC/dekelul 'eat (pl.)'. Such imperatives wiih de are perhaps a shade more energetic than the plain imperative and a shade less energetic than the imperfect plus Imaf.
(f) Preposed to the imperfect, M/ra!)l, J/ga ~/, C/gal)1 denotes futurity; this is either in va riable for the various persons and numbers.
or admits of a plural M/ray ~in/, J C/gay ~in/ : M/ra~anam/, J/gal)anam/ , C/gahanaml 'I'll sleep'. In addition, J also uses Isal and Issa/ : J/ssa'anam/; /sa 'anam/'l'lI sleep', as well as the full Ihassal from which
, I
118
MORPHOLOGY
COMMUNAL DIALECTS IN BAGHDAD
the shorter forms are no doubt derived. Malaika, 1963, p. 62, gives for M the future markers / ral)/ , Ilal)l and {hassa/. (g) It has been seen that / Ial plus the 2nd person of the imperfect serves as negation of the imperative: Ilatnami 'don't sleep'. There are a number of additional uses of {Ia/ with the imperfect, notably in expressions implying apprehension or doubt: M{ma'arldak etsafer batta laY?lr 'alek sil 'I don' t want you to leave lest something happen to you'. The foregoing could also occur without Il)atta/, apparently with the same meaning; cf. also M/abu ?aber, layedrukna Iwaket/ 'A. S., (I'm afraid) time may run out on us'; J/wha5a Ixellag laykOn yestal)i/ 'and that guest may feel shy'. Another use of Ila{ is to underscore the apodosis of a tern poral or conditional sentence: {madam za'imna I'amln belwujOd, lay~lr e!'eraq janna men jannat 'adanl 'as long as our trusted leader is alive, Iraq will be a garden of Eden'. This may be limited' to higher styles and is, at any rate, not a true verbal prefix: Iwda'tak lal'eraq Y?lr jannat 'adanl 'you can be sure'31 Iraq ,,,ill become a Garden of Eden'. (iI) In C, there is a preposed Iken/ accompanying the perfect which does not seem to result in a meaning contrasting with that of the plain perfect : /sme'tu kenqalu hekkil '[ heard that they said so', apparently same as /sme'tu qalu hekki{. At any rate this sets C apart from M and J, and again marks the similarity of C with the Mosul and Anatolian dialect '32 4.9 OTHER MORPHEME CLASSES. 4.91 Determination markers. (a) All three dialects have, in slightly different shapes, the characteristically Mesopotamian 'indetermination marker', M/fared/, J/fagad/, C/fagedl and MJC{fadd{. Its presence contrasts fairly clearly with that of the article III or other determination marks, but the degree to which it contrasts wi th absence of any mark is yet to be determined. It occurs before singular nouns (M/jani fadxabarl and /jani xaba rl ' [ got word') but also before dual and plural nouns and before numerals : M/fadyomen tlaSal 'a couple or three days', J/ kanu fagad eSnen ?edqan/ 'there were two friend s', M/fared-weld eSnen ezgarl 'two little boys', M/fared-cam I)aja/ 'a couple of things'. Its presence seems most common and most stable in expressions such as M/faredwal)edl 'someone', J/fagadwel)ed/, Cjfagedwel)ed/ 'someone', M/ faredsi/, / fad s!/, IfaSS//, J/fagad sen/, Cjfagedsel ' something', M/fadmarra/ 'once', / fadyom{ 'one day' (both in beginnings of
I'
119
anecdotes), etc. However, note M/fadmarra/ (usually contrasting in stress with Ifadm"Fal 'once') in the sense of ' al\ at once, completely' . Details of usage are obviously complex; here are some more or less random examples: M/,endi fadbetl 'I have a house' (in answer to question 'what ha,ve you got ?') but /'endi bet/, 'I have a house' (in answer to question ' who has a place we can meet at ?'); M/esma' menni fareds! arid' agOla elakl 'listen to something [ want to tell you' and from the same speaker, same text: Ibass SI wiil)ed arid agOlal 'but there is on~ thing I want to say' ; note the following dialogue : A: Itfai)i)al jlgiiral ' Have a cigarette' B: /Ia' , a5kurakl 'No, thank you' A: Ifadgahwa?1 'A cup of coffee?' 1
Note also M/gawwiiden ffadbfllad maY~lr{ 'you can't have two pimps in one town'; and the J proverb (also found in M and C with appropriate changes) her wseti ffad?a!1)1 'summer and winter on one roof' . In negative sentences, {faddl and its variants are usua lly replaced by M/kull/, JCjkell{: M/kull 51 maku bihal 'there's nothing the matter with her' or Imaku biha (kull) 51/. The negative parallel to M/faredwiil)edl and the JC equivalents is Ikullal)l)adl JCjkellal)l)ad/, . or simple lal)l)adl plus negation. (b) This indeterminacy marker is found in other parts of the Mesopotamian area, in dialects both of the gelet and the qeltu type, and in Central Asia, where it has the shape fat . My Siirt informant gives the form [fAll as the sale one in use, and my Swerek informants seem to use only a form [faqad) in a few restricted expressions such as [faqad yowm) 'one day'. Taken together with the form Ifagad/, these suggest a possible merger of two etymologically different forms : MJCjfadd/, M/ fared /, Cjfagedl from OA/fard/, which normally should have yielded J/fagdl and some such form a~ Ifaqa\1 ' only', which seems to occur also as / faqad/. Mosul and 'Ana have Ifadd l and / faredl (lfagod/) . It is worth noting that, though a somewhat d ifferent "indefinite article" occurs in other Arabic dialects, notably in North Africa, ' the other languages of the Mesopotamian area have particles that ~have quite similarly, Turk. bir, Pers. ye(k), NeoAramaic xa. (c) All three. dialects have the usual "determination marker" or . "definite article" that has the allomorph III before vowels, sem}vowels, labia Is, and back consonants, and the allomorph 'gemination
,
.' .
!
120
COMMUNAL DIALECTS IN BAGHDAD
of following consonant' elsewhere: / Ibet/ 'the house', /ssex/ 'the shaykh'. Before vowels a /'/ may optionally intervene : /Iadmi/ or / I'admi/ 'the person'; anaptyctic vowels are inserted in the usual fashion: / Ieblad/ 'the country'. In IC, /g/ behaves "synchronically" i.e. ;eqUlres /II no matter what its etymology: / Igas/ 'the head', / Igem/ the clouds' , vs. M/rras/ and / Igem/. A detail of its use in which C di.ffers from M and J wi ll be given in Chapter 5. It tends to merge wIth the /1/ of a preceding / mall : MIC/ mal el'eraq/ is usually MC /mal'eraq/, l / mal'eraq/, 'of Iraq'. 4.92 The relative L ' . The three dialects ha ve, in addition to /elli/, a relative pronoun L ' which behaves exactly like the article, i.e. has the allomorphs /1/ and gemi~at i on in precisely the same cond itions: M/lbena/, JC/lbana/ '(he) who bUilt', MJC/(e)nniim/ '(he) who slept'. It is sometimes difficult to decide whether we have this L' morpheme that IS replaceable by /elli/ Or whether we are faced with the article: M/ lwalad Imuzen/ 'the bad boy' or ' the boy who is not good'. 4.93 Reflexes of OA/ li/ and / 'ila/. (a) The enclitic alternants of the ~lOrpheme L 'to, for' have been discussed in 4.23a above. The non-enclitic alternants have a base el in MC and I/e in J with suffixes' when preposed to nouns and particles, however, aU three dialecl~ have the sa me two allomorphs as the article and the relative L ' namely /1/ and gem ination of following consonant: MJC/lbetna/ 't~ Our house', /~.adiqna/ 'to our friend' , / Iebladna/ 'to Our country'; before vowels I have not noted any intervening /,/, viz. only / Ia\lmad/ 'to A.'. ' The forms with pron. suff. are regu lar, the J base req uiring the post- vocalic alternants with the exception of the 2nd pers. fem. sing., which is / llek/ rathe r than an expected ' / lleki/ and thus the same as the 2nd pers. rn ase. sing.:
M eli elak e1ec ela elha
Is. 2m .s. 2f.s. 3m.s. 3r. s. Ipl. 2pl. 3pl. On the use of
elna
elkum elhum L
1 lIeni lIek lick lIenu lIeha lIena Ilekem lIehem
C el i elak elki elu ela e1na elkem elem
for marking the direct o bject, see 5.4. An added
'
.. ,
MORPHOLOGY
, ,
121
difference betw en J on the one hand and MC on the other is the existence of a special 1 alternant, or mor", probably a different 1 morpheme use4 in expressions of belongi,!g and sometimes interchangeable wif~ the particle /,end/. In MC, the same morpheme L is used in such ;cases; in J, there is instead ta morpheme which differs from L only in the non-enclitic allomorph with pron. suff., which is /Ie/ rather i han /lle/ ; M/ela \laqq/, C/el'!,.I)aqq/ vs. l / lenu I)aqq/ 'he has the right, is right'. With the negation / ma/ : M/ ma'eli xu!ug/, l / maleni xelq/, C/ma'eli xeleq/ 'I don't fel well', and ' so on. (b) The three , , dialects have a morpheme /Ii/, for some speakers /Ii/, that contra~ts with the preceding but occurs only in some limited environments, namely only before K V; it usually conveys such notions as 'u~tii, up to, right up to' , but the meaning contrast with L is not always very evident: / menbet libet/ ' from house to house', but also / menbet elbet/ ; however, /Iwen/ 'where to', liwen/ 'up to where, how far '; in some combinations it seems fixed: MIC/lihassa/ 'up to now, hitherto' ; M/ Iigeddam/, JCjliqeddam/ 'in advance'. Before KK onl y L occurs and any contrast is neutralized: M/mnessugu! lelbet/ 'from work to the house'. In MIC there is also a for"m / ila/, (for some speakers only /ela/) used instead of L or flit in mildly formal or semi:literary style, and in some fixed expressions that also belong to the )Jigher style ranges, e.g. M/ilamata/ 'until when, how long' (plain colloquial M/leswaket/), and it is regular and ve ry common in the special higher style used by 1 speakers in written or oral Bible translations and the like. 'J3 The use of / Ii/ with pron. suff. (base form flit) seems rare: J/liha/, / lina/ ' to her, to us'. 4.94 Some prepositions. (a) The three dialects have a morpheme B that has a\lomorphs / bi/ before pron. suff.,. otherwise /b/ with full assimilatioI;l before /f/ and /p/ and regular behavior in consonant clusters; before V, no intervening /'/ seems to occur: / bbetna/, 'in our house', / bebladna/ 'in our country', / bnafsi/ 'by myself.., / ppara/ 'worth one /para/" /ffels/ 'worth one /fels/" / badmi/ 'with or in a person'. Add'ilion of suffixes to /bI! poses no special problems: M/biya/, l /biyi/ ' in me', MIC/bik/ 'in you (m .s.)', M/ bic/, l Cjbiki/ ' in you (f.s.)" M/ bi/, lC/ binu/ 'in him', Ml/ biha/, C/biya/ 'in her', etc. This morpheme is used in most but not all the Mesopotamian area to the exclusion of any reflex of OA/fi/, though MJCjfi/ occurs in semi-literary styles and in some fixed expressions; the Central Asian dialects,' on the other hand, do have Ii.
122
COMMUNAL DIALECTS IN BAGHDAD
MORPHOLOGY
(6) The three dialects have a morpheme 'alay with allomorphs /,ala/ without suffixes, varying freely with j'a/ before the article, and with - ay- yielding /ay/ , /e/ and in J also /i/ as shown in the following table:
Note JC/sob/ and M/yamm/ 'toward, near' which partly overlap with /'end/ in the sense of 'to or at someone's place'. (e) All three dialects have a morpheme lI'iyd 'with', to the exclusion of any reftex of OA/ma'/ except in some fixed expressions; it has the expected allomorphs /wiya/ with suffixes and enclitics and /wiya/ in other cases : MJCjwiyani/ 'with me', M/wiyiiman/, J/wiya mani/ 'with whom'. Some of the Mesopotamian dialects e.g. Mosul, do have / ma'/, but other dialects, both qelill and gelel have this \Viya morpheme; Central Asian has the related lVey. (f) The morpheme MJCfjawwa/ serves both for 'inside' and for 'under' ; for the latter M/tal)at/ occurs in some proverbs, J/tal)t/ in "Judeo-literary." For 'over' M/fog/ and JCjfoq / behave normally, but there is a related fixed form with inserted /a/ in M/fogaha/ , J/fuqaha/, CffOqiiha/ 'in addition, on top of all that', contrasting with M/fogha/, JCjfOqa/ 'over her'.
Is. 2m.s. 2f.s. 3m.s. 3f.s. Ipl. 2pl. 3pl.
M 'alayya 'alek 'a le~
'ah~
'aleha 'alena 'alekum 'a.lehum
J 'Iayyi 'Iek 'Ieki 'lenu 'liha 'lena 'Iekem 'Iehem
C 'alayyi 'alek 'aleki 'alenu
'alayya 'alena 'alekem 'alayyem
The form /,a/ occurs before the article, e.g. /,alga'/ 'on the ground', i. q. /,alalga'/ , but also in the fixed formula /,a bal/ as in e.g. MJC /,abali/ 'it seems or seemed to me', /,abalak/ 'you (one) would think', /,abal axiik/ 'your brother thinks', etc. (e) The morpheme men has the allomorphs / men/ before KV and / mn/ before V or KK with, in the latter case, regular anaptyxis: MICjmenbetna/ 'from our house', / mnelbet/ 'from the house'; before V there is / men/ with intervening /'/ : /men'al)mad/ 'fro m A.' . With suffixes, M and C have the allomorph /menn/ (with automatic loss of geminatio n before KV) whereas J, as with L, has a fo rm with initial gemination: MC/ menni/, J/ mmenni/ 'from me' M/menkum/ l / mmenkem/, C/ menkem/ 'from you (pl.)'. Note the composite forms MJCjmnen/ 'from where', M/menni/ (and / menna/, offering an instance of the rare /a/-/a/ contrast in final position, cf. / menna/ 'from him') JCjmnoni/ 'from here, this way'. In MJC, / men/ also occurs before verbs: M/ mensafna gal/ 'when he saw us, he said .. .'; M/~a rla hwaya mensafna/ 'it's a long time since he's seen us', and analogously in JC. (d) The three dialects have a morpheme 'elld which has the allomorphs /'end/ before V and, in M, also /,ed/ before KV: MJC/'endi/ ' chez moi', M/,edna/, JC/,endna/ 'chez nous'; forms without / n/ may also, more rarely, be heard in JC, and forms with /dd/ before V occasionally occur in M (e.g. /,eddif) though my informants consider them provincial; cf. also M/,edman/, J/,endmani/ 'chez qui'.
123
,• <
t
•I
-~I
l
;
I,
SOME SYNTACTIC FEATURES
125
of my C informants (male, born 1940) who spontaneously produced most of the above examples in a free conversation. 5.22 This copula is not only lacking in J and M, but also, by present evidence, in the Mosul dialect, i.e. the one that is most similar to C. I have, however, some fragmentary evidence of similar copulas
5
SOME SYNTACTIC FEATURES 5. 1 INTRODUCTORY. The inquiries on which this work is based were far more extensive in matters of phonology , morphology, and lexicon than in matters of syntax. Accordingly, while I did gain the impression that the three Baghdadi dialects do not differ considerably from each other in this respect, the evidence is not abundant and the claim may eventually have to be modified. Tn what follows, I have singled out a number of phenomena differentiating M, J, .and C from each other; so me are uniquely characteristic of one or two .of the dialects, while others are matters of relative rather than abso lute differentiation (i.e. are "more common" in one dialect and
"less common" in another). These features are : the use of a post· posed copula in C (5.2), the use of the article in some noun-plus. attribute constructions (5.3), and the use of the anticipatory pron. suif. plus L to mark the definite direct object (5.4). 5.2 THE POST POSITIONAL ·COPULA. 5.21 The nominal sentence in J and M calfs for no particular remarks as compared with Arabic usage in general. On the other hand, the nominal sentence is, in C, subject to the optional use of a copula placed alter the predicate and consisting of the element Iyal plus pron. suff.: /,alakell!) iil enta eda sefet maye'jebak ennadi, ta!;>'an. hada seglak ya nul 'anyway, if you feel yo u don't like the club, then of course it's your business'; I te'gef betna Win kelles jaml! yanul 'you know how very beautiful our house is' ; j kalebna kelles zen yanul 'our dog is very nice'; I!)elwi yahal 'she's pretty'; Iwladem ham !)elwin yaheml 'their children are good-looki ng too'. In my rather limited stock of connected utterances from C informants, the majority of nominal sentences are without this copula. Informants state this usage is old-fashioned and obsolescent, and seem not to find any semantic difference between absence and presence of the copula. It was, incidentally, the youngest 124
in the Anatolian dialects: my Swerek a nd QarabiiS informants (the former Jewish, the latter Christian) have a postposed copula based not on Iya/ , but on a truncated form of the free personal pronoun, namely lwei or Iyej (jiyej): jha(\a mesli-wel 'this is my comb' Ihaye o<,let nawm-iyej 'this is a bedroom' Imene-wel 'who is it?' jnayyem-wel 'he's asleep'; Inayme-yej 'she's asleep' jakid-wej 'it is certain' There is a construct ion teminiscent of this in the Central Asian di~lects, namely a post posed element in plus pron. suff., e.g. 1I0XUSinni ' I'm not well', lIoxus-inki 'you (f.s.) are not well'.134 Similar postpos~d copulas are, of "course, found in Turkic and Iranian, b~t also in Syriac and Neo-Arama ic ; what accounts fbr its presence In
C is, at present, impossible to conjecture . 5.3 ABSENCE OF ARTICLE. 5.31 Constructions with Imal/. In M and J we find such roughly equivalent pairs as jbetil a nd jlbet malil ' my house', Ibet abuhal and Ilbet mal abuhal ' her father's house'; in the jmalj construction, the noun preceding Imalj has the article. Constructions where the article is absent occur, albeit infrequently, but in M and J they contrast in meaning with the above: jbet malij 'a house of mine', Ibet mal abuhaj 'one of her father's houses' or the li ke. In C, on the other hand , this contrast seems to be completely absent, ~nd the normal Imal j constructions equivalent to noun-pluspronoun-suffix or to noun-plus-noun annexations is without the
article: jhiida kaleb malna ger lagwa belbetl 'that dog of ours is nothing but trouble in the house' jasu-jii Ifallai:l dabb mes!):! maletu we nhazamj 'so the gardener threw away his spade and ran off'
126
SOME SYNTACTIC FEATURES
127
COMMUNAL DIALECTS IN BAGHDAD
Ixa<)ar malna qayzid yom waga yoml 'our vegetables are increasing from day to day' Ihada mudir mal pasportat l3 ' sawwali fadlagwal 'the man in charge of passports gave me trouble' I!ariq mal !ayyara maloti ja ila braksell 'my plane went by way of ' Brussels' fj a Mmi malu wqallul ' his manservant came and said".' My C informants agree that in such examples, the governing noun can also take the arllcle, w ithout any semantic distinction; they ~eem to prefer the form without article. This usage has, so far as • know, practICally no parallel in other dialects."· 5.32 Noun plus q~aliHer. (a) On the whole, the prevalent pattern that reqUires the adjecti ve to agree in determination with the noun It q~alIfies applie~ in,MJC : Iwalad zenl or Ifadwa lad zenl , ' a good boy, Ilwalad ezzenl the good boy'. Now and again, however, one meets equivalent constructions without the article on the governing noun, more especially III J , but also in C, and apparently only in more or less fixed combinations in M, My M informants balk at !;valad eZ,~en/ , though agreeing (not without hesitation) that it is possible, . whereas C a nd J informants have been heard to use it (or ItS equivalents J/walad lemlil)/ , C/walad elmalel)1l side by side with the form Wlt~ the article. This has the effect of assi milating the noun-pius-qualIfier sequence to a noun-plus-noun anne xation or "construct phrase," as can ~ seen from the behavior of nouns ending in the feminine morpheme T; some details fo llow. (i) In M, this construction is common in such fixed phrases as, ~Iace, names: 1Mb essarjil 'the East Gate (quarte r)" Imal)allat ~I atlgal ,the Old Quarter', cf. such contrasting phrases as Ibab essexl the Sex s Gate' and Imahallat enna~aral ' the Christian quarter'; these names of quarters are used in Je as well. The construction is also common in expressions of time, e.g. Isahr ejjayl 'next month', I~ant ellux/ 'next year' /,am elmaQi/ 'last year', Imart elluxl ' next lime, once more'. With Ilux/, Ilaxl 'other' it seems somewhat more productive, e.g. Inawesni hay junu!! elluxl ' hand me that other ~alise, perhaps bec~,use the membership of Ilu x/ in the category qualIfYlllg adject ive IS moot; the form lejjun!a lIuxl does, however, occur. Sim Ilarly, lid elyemna/, lid elyesral 'the right (left) hand'·131 note also ,l'id eccebirl 'the Great Feast (Feast of the Sacrifice)', ~nd III proverbs, e.g. Isabb 'enab laswadl 'he cursed the black grapes'l38
.. (ii) . This construction has been noted in C in expressions similar to those listed above for M, wi th perhaps a shade greater frequency and freedom: lid elyemnal and lelyesral 'the right (left) hand' j'id lekbigl 'the Great Feast (Easter)' Isant ellexxil ' next year' Iqondart ellexxil 'the other shoe' I~l?ii'a lekbigil 'the big finger (the thumb)' (note Iqondart/- but I~ba'al definitely not I~~ba' a/), Ibent lekbigil 'the older daughter' . It may not be accidental that in all these article-less constructions the qualifiers have something in common, namely contrastive meaning: 'the other', 'the older' as, opposed to the younger, 'the big' as opposed to the smaller or lesser 'the right' as opposed to the left, etc. (iii) In J lye again find the 'equivalents of the 'right hand'; 'next year', etc., lid elyemna/, Isant ellex(xi)1 but freer constructions are not uncommon, though again often with qualifiers suggesting contrastive meaning: Ibab leqqibl 'the near(er) gate', Iwalad luwlanil 'the first boy', Isuwal eemnil 'the second question'. It is common in more or less congealed expressions, e.g. I~ ! at lekbigil ' the Great Synagogue', I~!at lejdidil 'the New Synagogue', (specific places in Baghdad), Isahg en naqe~1 'the incomplete month (22 days after death)" Isahg ettmaml the complete month (full month after death)' ; 139 it is extremely common in J proverbs: Idrater el'ettaq/" 'the old notebooks', l xall elkaskin/ ' strong vi negar' , lid elmaq~(i~al 'the amputated hand', Ileqmet lekbigil 'the big morsel' .I4O (b) A closely related but more problematic construction is that involving a noun plus relative clause. As in most varieties of Arabic, the prev¥iling MJC construction is to mark determination on both the noun and the relative clause (using presence or absence of article on the former, presence or absence of L ' on the latter): Iblad sefnahal 'a country which we saw' vs. Ileblad essefnahal 'the country which we saw' . There is some tendency, apparently more widesp read in J than in M or C, to leave off the article, viz. to say Ibliid e$Sefnaha /. The similarity with the noun-plus-qualifier construction is rendered even more striking by the near-identity of the phonemic shapes of the allomorphs of L ' with those of the article ; the identity is not complete only because L ' has an a lternant lelli/, in some cases also Illailil (cf. 4.92 above) ;I41 the equivalence to construct phrases is here again sometimes shown by the allomorphs of the feminine T: JC/sa nt ejjinal 'the year we came'. In C and M such constructions may be restricted to fixed expressions, especially of rime (M /yom essafarl
,• ,
·• • ••
128
COMMUNAL DIALECTS IN BAGHDAD
:the day he left'), whereas in J there is not only / waqt elkent bejjays/ the tl1~e ~~u were In. the army', but also /sen lagidu/ 'the thing that I want, /se~ essuwwetu/ 'the thing that you did', / ~al)iir elliiai qayezzawwa]/ t~e young ~an \~ho is getting married'. In J proverbs, cf. /s~hg e!ma endak bInu I)slb/ 'a month in which you have no account, / I)Jaga elmateg(\iiha/ 'a stone you don't like .. .' (or / l)jiigt/).142 As shown by some of the examples, it is not always certain whether the antecedent IS determinate or indeterminate, and the contrast may be neutralized; cf. / mayniim bemkiin elyebgadlu/' he doesn't sleep III a place where he'll be cold ', i.e. 'he knows how to get along'. (e) Absence of the article in such cases as have just been discussed. and the merger or near merger of noun-plus-qualifier phrases with construct phrases are attested in older stages of Arabic and in a number of present-day dialects. 143 Such constructions seem nowhere to have replaced the dominant pattern, and in present-day dialects they occur predominantly in more or less lixed phrases. If, as would appear, J and to some extent C have here again preserved more traces of an older usage than M, the differentiation may be due to greater mfluence of Cl. Ar. and to greater dialect levelling among my M mformants.
5.4 ANTICIPATORY PRONOUN SUFFIX PLUS
L.
5.41 Object of verb.
(a) Another phenomenon that occurs in all three dialects, but with
va rymg degrees of frequency, is the formal marking of the delinite dIrect object of a verb. This consists in affixing to the verb an anticIpatory pron. suff. (= AP) referring to the object and preposing the morpheme L to the object: thus MJCfbii' elbet/ 'he sold the house' has the equivalent M/ba'a lelbel/, JC/ ba'u lelbetf. Where the object ha_s the a rhcle, the L IS sometimes missing, e.g. M/ bii'a lbet/ , JC/ba'u Ibet/ ; that thIS IS probably due to haplology is indicated by the fact that nouns without the article always 'have L: JCfbe'niinu Ibetna/ ' we. sold our house'. When the direct object is indelinite (i.e. marked neIther by the artIcle, apron. suff., membership in the class of proper nouns, etc.) the verb object relationship is unmarked: / be'na bet/ or /fadbet/, 'we sold a house'. There are some differences, largely of frequency, III M, J, and C usage; some details follow. (6) [n M, this procedure seems rarer by far than the unmarked construction. Among examples noted : /sedha lbab/ 'close the door' (noted once; noted many times /sedd elbab/), /fukka "ebbao/ 'open
SOME SYNTACTIC FEATURES
129
the window', /kallafta laxiiya/ '1 entrusted my brother', / mate' urfa Ifa(\el/,don't you know F.?'; with the interrogative 'whom' the L is necessary, though there is no pron. suff.: /elman sefet/ 'whom did you see?'. With intonation conveying special emphasis (high on verb, low on object): /xallensiifhum lejjamii'a/ 'l et's see them, those boys!', and / dansiifa bennadi Igayliin ramez/ 'we see him at the club, that G.R.'; in extraposition: /arjiik lhalmaw(\ii' sedda/ 'please close that subject of discussion'. Though in the last examples it is the special intonation or word order that lend emphasis to the utterance, it may be that the comparative rarity of the marked construction in M mak;es ,it .especiall y suitable for use in special cases; the construction
per se does not, 'however, seem to carry any extra emphasis as compared to the unmarked construction. (e) In C, this usage is much less marginal than in M ; among examples noted:
/sama'ta laxbar malriyaQa/ 'did yo u hear the sports news?' /ma'gef slon gal)idabbega Il)aletu bamrika/ ' I don't know how he is going to take care of himself in America' /ye<.\gebu ddelu 'bweona/ 'he wags his tail in our face' /seftiinu jjeddi/ ' I saw my grandfather'. /slon I)agaqtiya 1gejleki/ 'how did you burn your (f.s.) feet?' / nel)na nensanu Jada/ 'we'll forget about that one' (liada/ = / Ihada/)
•
••
•
•
••
, I "
(d) In J, this construction is so common that it gives one the impression of being the normal one. However, in an actual count of occurrences of definite direct objects in recorded texts by two informants totalling about forty-five minutes, I find the number of objects so marked to exceed unmarked objects only slightly: 21 marked vs. 19 unmarked. A few out of the very many examples noted:
/asisemmiiha layi/ 'what do they call that?' (flayi/ = /Ihayi/ ) / hezzu Igiisak/ 'shake your head' /yjibonu Ima8al/ 'they quote the proverb' / qla'a le!!rumbat mahpayy kelleta/ 'he cut off all the water pipes' l1aq yeb ni ba'd wkemmela lqebba/ 'he was able to go on uuilding and complete the room' /Imaga testagel' wtesgelu Imexxa/ 'the woman works and uses her brain'
" :jj ~
130
COMMUNAL DIALECTS IN BAGHDAD
/ mabezmana i)ebbeta lkenneta/ 'she never liked her daughter-in-law' /ebna qayi)ebba lemgatu/ 'her son loves his wife' /kef eti)ebba wtekraha lkenneta/ 'because she loves her daughter and hates her daughter-in-law' /qamu ybi'iiha besmakan legragem/ 'they started selling their things for whatever they could get' /qayfeti)iiha jjenal/ 'they 're opening up the suitcases' jQiyya'u lbabiic i)alqu/ 'he lost the slipper of his mouth' i.e. 'the cat's got his tongue' With the interrogative /mani/, L seems mandatory, the pronoun (no longer anticipatory, but rii cklVeiselld) optional: /lmani mrejjei)u/ or /nrajjei)/ 'whom shall we prefer?'. In extraposition, the pron. suff. is apparently mandatory, whereas L mayor may not be present, with . no doubt some difference of connotation: /hMa ~~aleg ktabu li)axam belwaqqa/ 'that line, the rabbi wrote it down on a piece of paper' ; /wlab'an elbenta tl,ebba azyad menkenneta/ 'and of course she loves her daughter more than her daughter-in-law'. Note that without the L the last example would be ambiguous. With epexegetic repetition of the pronoun, either free or appended to L: /safni anal or /siifni lleni/ 'he caught sight of me'. (e) This sort of marking of the definite direct object is quite common in the Mosul dialect, though it is for some reason absent from Socin's texts. In the Central Asian dialects, both definite and indefinite direct object seem to be marked by apron. suff. (which is however not usually anticipatory, since the object generally precedes the ve rb) and a frequent equivalent of our morpheme L, namely i, with suffixes i/a_. 144 A more strictly analogous construction is found in Lebanon \4S and both the use of L and that of the anticipatory pron. suff.' as direct object markers occur in older Arabic and in other dialects, at least sporadically!46 The occurrence of the construction in the Syrian and Mesopotamian areas has naturally - ·and unanimously - been attributed to the Aramaic substratum. l47 Though it seems more common in J and (to a lesser extent) in C than in M, Ganima (see note 147) implies that it is general Baghdadi usage; he is, however, himself a C speaker. The use of L as a direct object marker is specifically though briefly reported as Baghdadi usage by an eleventh century source, so that J and C seem again to have preserved older features to a greater degree than M.14S
SOME SYNTACTIC FEATURES
131
5.42 Object of preposition and noun . (a) Noted in J only so far are instances of the construction preposition AP + L + noun as equivalent to preposition + noun, i.e. /,endu lax uk/ i.q. /,end axuk/ 'at your brother's'. Moreover, I have noted this construction only with t end/ and /b/ : /gai)u 'endu Hiaa/ 'they went over to that fellow's house' /as'endu biha layi/ 'what's there in it for him?' /thayyag biha layi/ 'he was perplexed by it' I lazem aku biha lei)kiyyi fad ion/ 'there must be something in the story' Ibala mabahjad ye'gef biha lei)kiyyij 'without B.'s knowing anything about the matter' Similarly, indirect objects of verbs, i.e. objects of the preposition which thus appe,ars twice:
L,
Iqalla lemgatul 'he said to his wife' Iwoi)ed yqellu llaxl 'one tells the other' Iybiyyenla lkell wei)di ~e!!etal 'he explains to each what her sphere of authority is' As some of the examples clearly indicate (esp. the type Ibiha layil 'in it') it might be preferable to speak of object pronoun plus epexegetic object introduced by L, and not of "anticipation" of the objectl49 (b) l(we view the governed noun of construct phrases and their equivalents as "object of a noun," we may say such objects are occasionally also marked by this AP + L construction: J/abiinu ssalmanl 'S.'s father', roughly the same as labu salman/, though the latter can also serve as a fixed kUllya or teknonym while the former canD:0t. 150 Indeed, the construction seems particularly common with
kinship terms and the like in which annexation via / mall seems uncommon; Imagt, 'amma Ibenti/ 'my daughter's mother-in-law (or paternal aunt)'; ,/ mgatu ssalim/ 'S.'s wife' ; Isdiqu lelmudirl 'the director's friend'; however, note /kella Iba~gal 'all Basra'. Analogous examples occur in MC again largely with kinship terms though note M/wda'ta lmai)mud/, cf. note 131. (e) Constructions of the latter type, i.e. /,ammu leflan/, are common enough in Mosul and throughout the Syrian area, especially with kinship terms; they have, again, been attributed to Aramaic influence. lSI The Central Asian dialects have a parallel construction,
,
•
132
COMMUNAL DIALECTS IN BAGHDAD
though in reverse order (viz. i/olon 'ommu), which seems to be the prevalent form of annexation. l52 As for constructions of the first type, viz. /,endu lefHin!, I have no direct parallels III the other dialects though repetitions of L (/qallu leflan!) are common.
6
SOME LEXICAL FEATURES 6.1 INTRODUCTORY. In this chapter an atte mpt is made to tnalacterize the lexical relationships bctwcen the three dialects by (a) assessing the degree to which they differ in "basic vocabulary," (b) examining terms of non-Arabic origin, notably the Hebrew vocabulary of J, and (c) listing and discussing a number of terms of special historical and comparative interest. A problem that cou ld not be resolved quite satisfacto rily is that of asymmetrical usage ; by this is meant the fact that, for example, the normal term for 'good' in M is !zen!, in J !mli~!, in C !male~ !, but J and C speakers know and sometimes use !zenr as well, whe reas M speakers use no cognate of the JC forms. In such cases it is not al\~ays possible to
•
,,••
determine when we have a borrowing -from the other dialect, more
or less well integrated, or a native synonym. Another problem is that it was not always possible to obtain exactly equivalent terms in the three dialects, or to ascertain whether a term heard in one of them also occurs in the others. For these reasons, the formulas used for word citations throughout this work, e.g. J!kalb! 'dog' , mean primarily that this f9rm and meaning combination occurs in J, and only secondarily, depending on other evidence, that it occurs in J only, or that it is the sale nOfm for J, and the like. 6.2 BASIC VOCABULARY. 6.21 The major differentiation we have observed in phonology and morphology leads us to expect considerable divergencies in cognate words. Indeed, it is precisely to these divergencies that most of Chapters 3 and 4 have been devoted. We have constantly been dealing wi th such triads as M!clabec, J!klibek!, C/klebki! 'your (f.s.) dogs' and investigating the phonological and morphological differences they present. This major differentiation does, not however, necessarily imply similarly divergent
vocabularies. In fact, we would rather expect the basic vocabu lary 133
.~
, "
134
COMMUNAL DIALECTS IN BAGHDAD
(however defined) to be made up largely of cognates, and such is indeed the case. If we examine the "basic word list" made up by M. Swadesh for his lexicostatistics, 153 which contains 200 common words denoting parts of the body, basic kinship terms, cardinal colors, the numerals one to five, etc., we obtain the following results. (i) Out of 200 words, 192 or 96 per cent are cognates, the eight triads of non-cognate terms are: M 'bad' 'fire '
muzen
'to give' 'husband'
neta
'many'
'nose' 'to see' 'to, throw'
naf
rajel hwaya xasem saf oabb
J duni (laww ta'a zoj kelg enf 'ayan farr
135
SOME LEXICAL PEA TURES
C mu ma le~
nag 1a'a
zoj ktlg xasem saf dabb
In five of the eight cases J stands alone as against M and C, whereas in three cases J and C go together and M stands alone: in no case is there a 1M combinatio n as against C.
cognates is that there are a number of what might be called "distant cognates." By t)lis is meant words for which the ultimate etymon is the same or fqr which part of the etymon is the same, but in which divergent develqpments have produced unusual or irregular alterations, so that kins hip is recognizable to the investigator but may be moot for the speakers. Such distant cognates are e.g. the items 'here', M/hnii/, JC/ hon(i)/ ; 'to kill' M/ketal/, JC/qatal/ ; ' near', M/gerlb/ and /qarlb/, J/ qqlb/, C/qaglb/ ; 'root' M/,ereg/ J/,eqq/ C/,egeqj. Other irregular cognates are some of the personal pronouns (cf. 4.2) some of the ordinal numerals (cf. 4.6) and some of the demonstratives and interrogatives (cf. 6.3). 6.22 Turning now to a list of some 280 items compiled specifically for lexical comparison between Moroccan, Syrian, Egyptian, and Iraqi Arabic 154 and founded not on considerations of basic vocabulary but on high frequency and maximum differentiation in the four dialects, we find substantially similar results. There are only
• ,• c
•
7 non-cognate items, 4 instances of asymmetrical cognates and
some 20 irregular cognates. (i) The non-cognates include the items 'bad', 'to give', 'nose' and 'much' already obtained in :'1e previously considered list and in addition:
(ii) This bald statement of 192 cognates vs. 8 non-cognates does not, however, tell the whole story. In the first place, there are, even in the above list, some asymmetries. Thus some C speakers seem
'to -look'
to use the M verb for 'to give' 'side by side with the C form, at least in the imperative. Further, while M/rajel/ ,is the common pariance term for ' husband', M/zawj/, a cognate of JC/zoj/, is used by M speakers in " mildly formal" style. To the last non-cognate triad, 'to see' , (ve must add that J has, in addition to /,ayan/: a verb /saf/ identical with the MC verb, though the contrast /,ayan/ 'he saw' vs. /sM/ 'he caught sight of, found' (durative vs. ingressive) occurs in J only and thus sets the J form apar.! (cf. also note 173). Among the 192 cognate words (a complete list of which is omitted here) we find further cases requiring qualification. As already indicated in 6.1 , it is a toss-up whether the term for 'good' should be listed under cognates (MJC/zen/, MJCjxosf) or under non-cognates (M/zen/, J/mlii)/, M/maleh/). The item 'mouth' yields the three cognates M/i)aleg/, J/ i)alq/, Cji)aleq/, but J a lso has the synonym /eemm/ . (iii) .Another qualification that must be added to the 192
'tomorrow'
'beside'
M bawa' bacer yamm
J safan gada sob
C bawa' gada ,
~ob
As in the case of the terms for 'to see', terms fo r 'to look' are only partly equivalent in function. Moreover, Cjbawa'/ may be a loan from M. One of my C informants used /bacer/ for ' tomorrow', a patent M borrowing. (ii) Asymmetrical cognates include the item 'cup', for which MJC have several common terms, e.g. /fenjan/, but for which there is also the special C/pyala/ and M/ kub/. For 'grandmother', there is MJCjjedda/, but M also uses /blbi/. For 'to close' there is MJC/sadd/ but J also uses-/taras/, /tagas/, which also means 'to fill' as do the MC cognates. For 'to open', there are the cognates JCjfatai)/, M Ifetai)/, but M also uses / fakk/, which also means 'to loosen' as does the cognate JCjfakkj.
.,•,., ~l ?,
136
COMMUNAL DIALECTS IN BAGHDAD
SOME LEXICAL FEATURES
(iii) As for irregular cognates, they include the demonstratives and interrogatives already yielded by the first list as well as the following items:
'yesterday' 'thing' 'in order to' 'socks'
M Ibar!:>a
si
!:>atta jwarib (s . jiirab)
'spoon'
xasiiga
'to ask'
qasiiga se'aI
'broom' . 'children' 'fruit'
muknasa 154a
weled mewa
J (I)b6!:>i sen(i) !:>etti jwarib (s. jurab) qaSiiga xasuga sayaI mokensi wIad miwa
C mbe!:>a se !:>attan(i) jawegeb (s. jogabi) qasoga sayaJ maknasi wI ad mewa miwa
6.3 INTERROGATIVES AND DEMONSTRATIVES. 6.31 Probably any pair of Arabic dialects yield, when compared, differences and asymmetries in the inventories of interrogative demotlstrative particies, 1 55 This is tbe case in MJC as well: most of the forms listed below will be seen to be wholly or partially cognate, yet irregular in their correspondence. In the demonstratives, J and C group together as against M, but in the interrogatives, M and C group together as against J nearly throughout. 6.32 Interrogatives. (0) Terms for 'what' have, in MJC as in many other dialects, an element lsi, presumably from OA/say'/ 'thing', formerly preceded by an interrogative that is now lost or blended with it. Each of the three dialects exhibits, though in varying degrees, an alternation among a free, a ' preposed, and a postposed variant, that have the following shapes : M free : preposed: postposed:
senu
s
es, wes
J askiin as es, as
C senu
s es
(i) T he free form occurs either in isolation or as subject of a
137
nominal sentence of which the predicate (which may either follow or precede it) is grammatically determinate: M/ha5a senu/, Ilha5a ",kiin/, Cjhada senul 'what's that?" M/senu le!:>caya/, J/ askiin le!:>kiyi/, Cjsenu le!:>kayyil 'what's the story?'. (ii) The preposed form occurs otherwise (except as indicated below): MCjsesmak/, J/ases makl 'what's your name ?', M/~~ar/ , l /assag/ , Cjssagl 'what happened?' (with assimilation to following lsi or /!f) . The l?reposed form also occurs in a number of compound interrogatives: MC/ Slon/ , l /aslonl 'howT, M/sgad/, l lasqad/, Cjsqad/ '!low-much l' M/swaket/ , C/ swaqet/ 'when?' cf. also /skam/ side by side with MC/karn/, M/cam/, J/ kem/ 'how manyT. It also occurs in a number of fixed expressions with rbetorical question tinge, e.g. MCjsda'wa/, l /asda'wa/, roughly 'how come l' (surprise and indignation, esp. at an overly high price and the like); MC/s'aza/, l las'azal in e.g. M/ma'ruf s'aza/, l lma'gef as' azal 'und was weiss ich noch'; M/s!:>ala/, lit. 'what is his condition l' is also used , apparently without interrogative tinge, as an expression of commiseration, roughly 'poor fellow'. Before Ima/ in tbe sense of 'whatever', M /smatrid/, l lasmatgid/ , Cjsmatgid/ 'whatever you want', in M also /skulmatrid/. (iii) As for the third form , it is most commonly postposed to prepositions, with which it mostly forms inseparable and fixed formulas: MJCjbes/ 'in what' and 'bow much?' e.g. / bes essa'a/ 'what time is it?' or 'wbat does the watch cost?'; MICjles/ 'why?', M/xa!res/, J/xa!gasl and /x'd!as/, C/x'd!ges/ 'why, wbat for?' , M/elwes/ for what purpose, reason?'; M/'alawes/, IC/,ala(')es/ 'on what, about wbat 1'; MIC/ men(')es 'from what?', MJ/megeI(,)es/, C/meteI(,)es/ 'like whatT In all the forms with optional intervening /'I there may be two separate stresses, i.e. two separate words; tbis is regularly the case in such construct phrases as e.g. MJ/ maw(jii' esl 'what subject ?' . (iv) The preposed form of I, though for once not of C, is identical with that of the Mosul and Anatolian dialects, which have /as/ though 'Ana has /s/ as MC and the gelel-dialects. However, 'Ana and Hit hilVe the free form /skiin/ reminiscent of J, while Mosul has /assiin/. The postposed les/ seems common to the whole Mesopotamian area. The Central Asian dialects have iis and sim ilar foim s"· as do many other dialects; note 'Ana /Ils/ 'why', also found in some rural gelel-dialects. (b) Terms for 'who' show a single form for I , a double one in M; data for C are insufficient. The free form (in I also the sale form)
,•
·
I
.'"g I
"
138
COMMU NAL DIALECTS IN BAGHDAD
SOME LEXICAL FEATURES
is M/menu/, J/mani/, Cfmenu/ ; in addition , there is a postposed this time truly enclitic, form M/manf. Thus M/elman/, l / lmani/ 'to whom' ; M/,edman/, l /,end mani! 'chez qui ?', M/,aleman/, J/,ala mani/ 'on or about whom ?', M/ wiyiiman/ , J/wiya mani! 'with whom?' Mosul, like MC, has jmenu/ , the Anatolian dialects / menu/ and /menej, 'Ana and Hlt have / man/ , the rural gelel-dia lects / menhu/ and / menhif. (e) As for 'which (of several, either persons or things)' there is M/yaj, J/ hayyi/, Cfhayyu/ . The JC forms are used both adjecti vally and pronominally, the M form is adjectival only, the pronominal form being jyahu/ either invariable, or masc. sing. with fem . sing. /yflhi/ and a plur. /yiihum/ . Thus M/yii ktab/ , J/ hayyi ktiib/ , Cfhayyu kteb/ 'which book?', but M/yiihu lakbar/, J/hayy i lakbag/ , Cfha yyu lakbag/ 'which one is the oldest 7', with optional variation for number and ge nder in . M as indicated. (d) For 'where ?', MJC have /wenj, and /mnenj 'whence ?'. For ' how many', Mjcamj and the "mildly formal" /kam/, Cfkam/, l /ke'm/ ' :' For ' when?', the cognates MCjyamtaj,/yemta/, J/emta/ l58 are "asymmetrical" in that J/emta/ is the normal term, whereas M prefers /swaket/, C prefers /swaqetj. Again, qellll and gelel-dialects align themselves in no discernible pattern: Mosul has /wen/ like MJC, bu~ also jesab/, (not esob as in Socin, 1882, p.II). Mardin has / an[/ (ibid., p. 260) Carmiic /aynij, 'Ana and the gelel-dialects, /wenf. In Central Asia there is hln, hen (apparently from OA/ayn!) but also _,db and variants (Fischer, 1961 , p. 260). As for 'when ?', Mosul has /emati/ , 'Ana · /wemta/ , Kweres has jswaket/ as in M according to Meissner, 1903b, p. xxx iii, to which Weissbach, 1904, p. 939, adds jyemethaj and / wemethaf. In Centra l Asia, there is mala and Imil (Fischer, 1961, p. 260). 6.33 The demonstratives. (al I have lumped together here demonstratives belonging to several fonn classes that share some semantic and some formal features; here is a comparati ve table:
'this' (m.) 'this' (f.) 'these'
'these' (f.) 'this, these.'
M hiiila hiiy h(a)Mla (ha)5anni hal-
J ha5a hiiyi haMli
C hiida hiiyi hadoli
hal-
hal-
'that' (m.) 'that' (f.) 'those' 'here' 'there' ' thus'
'now'
(ha)6iik (ha)51" (ha)Mlak hnii(na) hnak(a) hlci hassa
ha5iik ha5lk haMlak hon(i) wniki heke5 hassa, ssa 'a
139 hadiik hadlk hadolak hOn(i) honlki hekki hassa
The invariant jh'al/- (with assimilation of j l/ to following consonants under the same conditions as the article) is common in a number of more or less congealed expressions, e.g. / hal yo m/ (i.q. /Iyom/) 'today', Ml/hassana/, Cfhassanij 'this year', Mjhalgad/, l Cfhalqad/ 'so much'; in the last item of the list it is at least etymological, / hassa/ < jhassa'a/. Its productive use is extremely common, and the degree and manner in which jhahvaladj contrasts with jhii5a Iwalad/ 'this boy' remain to be investigated. In JC, the initial /hj tends to be elided when immediately preceded by a short closely bound morpheme, e.g. L or IV: M/lhiiyj, JCfliiyi/ ' to that one' ' to this (f.) one', M/wha5aj, 1/ wii5a/, Cfwiidaj 'and this (m.)'; though M usually retains j hj , it does have jmenna/ 'from here, this way' as sole combination of / men/ + / hniij. This /hl-dropping tendency is probably a lso responsible for the J form / wnik ij.1 59 The most salient difference between the M forms on the one hand and the JC ones on the other is the greater retention of a reflex in the latter of the OA element j haj-, j hiij-; Mjhniij < OA/hunii/ , but lCfhon(i)/ < OA/ hflhunii/ , and cf. all the forms in which M has an optional but Je; an obligatory / ha/-. The fern. sing. /(ha)5annij is optional, M/(ha)Mla/ serving for both genders as well, but appare ntly more in use among urban M speakers than other fem. pI. pronominal forms (cf. 4.2). (b) The differences exhibited by M vs. JC recur, but only roughly, in the overall gelel vs. qeltu split. The qellu-dialects as a rule only have forms in /0/ (or the equivalent diphthong) for 'here' and 'there', whereas the gelel-di alects usua ll y have forms only in / hnj- : Mosul jhon(i)j, 'Ana [hown], Anatolia [hawne] and [hawni], and Syrian sedentaries have analogous forms; For ' there', Mosul has jhonekj (m y Christian and Jewish informants and Socin's texts) but also jhniikaj (my Muslim informants), whereas 'Ana has jhniik/. While the gelel-dialects of the countryside go a long with M, the Beduin
• ,,•,
140
COMMUNAL DIALECTS IN BAGHDAD
SOME LEXICAL FEATURES
dialects studied by Cantineau (1936, p. 110) have form s such as Ihanl and Ihanl with long vowels. A form Ihawnal is attested for eleventh century Iraqi vernacular and Ihunil for fourteenth century Baghdad,l60 no doubt the ancestors of the various sedentary form s reported above. Terms for 'thus' have initial Ihl everywhere except in Anatolia (Qarabas I k~5i/), with J/hekeill (also / hekelf) a ligning itself with Mosul Ihakeil/, while Cfhekkil is, so far, isolated; Mlhicil recurs in the gelef-dialects, but lines up also with 'Ana Ihie/ , Hit Ihikf. !6! For ' now', forms like MJCfhassal are fairly widespread throughout the area, with Mosul lhassal, Ihassa'l, the latter also in Hit and the gelef dialects; 'Ana has [alhaz], wh ich lines up with scvcral of thc area's Beduin dialects (Cantineau, 1936, p. 110: hal!uizz, al!ll;zz in seve ral tribes, others haying hassa'). Finally, note that the ending Mlal, JCfil which is optionally added in some demonstratives, fixed in others, has its counterparts in other dialects, but does not, despite
specifically religious and ritualistic vocabulary has nct been examined in detail , such an examination would, no doubt, reveal many more Hebrew terms, as would the speech of rabbis and other traditionoriented speakers. On the whole, it may be said that Hebrew terms
first sight appearances, behave like reflexes of the feminine
SUffiX. 162
6.4 . ELEMENTS OF NON-ARABIC ORIGIN. 6.41 Words of Turkish and especially of Persian origin are strikingly common in MJC and in other Meso potamian dialects. From my data and from the long list of Persian words published by Calabi, 1960, I infer tentatively that the three Baghdadi dialects do not differ from each other markedly in this respect. Individual lexical items do occur in one or two of the dialects only, others are treated so mewhat differently in one or the other dialect; these will be mentioned in 6.5, but they seem to indicate no particular trend . Much the same , may be said of Aramaic loans (studied by Ganima, 1926) except for the Judea-Aramaic to be discussed below: one would cY.pect C speake rs to use Syriac terms,
especially in the sphere of rel igion, not found in' M and J, but the relatively few religious terms gathered so far in C are of Arabic origin. I take it this is due to the non-systematic character of this aspect of the investigation. 163 The one non-Arabic set of lexical items by which one of the three dialects is clearly marked off from the other two is, as in other Judea-Arabic dialects, the Hebrew a nd Judea-Aramaic element in J, to which we turn below. 6.42 (a) The list appearing in (d) belo w, gives an idea of the sort of Hebraic vocabulary in use among the generality of J speakers as an integral part of their Arabic dialect. A good many of these have direct and in direct reference to religious matters, and though ttte
141
in common, everyday use are not very numerous; many have Arabic
equivalents, though the use of the Hebrew terms usually carries a special tone or stylistic value of its own. Before listing the items themselves, some remarks about the phonetic treatment of H ebrew words ' are in ortler . (b) As in other Jewis h vernaculars, one must distinguish between "whole Hebrew" and "merged Hebrew."!64 Whole Hebrew (henceforth WH) IS Hebrew as pronounced in wholly Hebrew utterances either in the reading of Scripture, in prayer, or in qu oting; on th~ other hand, merged Hebrew (henceforth MH) refers to Hebrew items used as an integral part of the vernacular. In WH , pronunciation is guided by Masoretic spelling and a set of spelling-pronunciation rules; in MH, there is fairly systematic deviation therefrom in the direction of greater similarity with the Arabic dialect, notably 10 the generalIZIng of penultimate stress and the reduction of unstressed vowels. No phonemes ·alien to the dialect seem to occur in WH or, a forfiori, in MH, though frequency and distribution are naturally somewhat differenL I6 5 Consonantal equivalences are the expected ones; ~hose requiring no special comment are . h IV Z h ( Y 1m II S' $ q s; the others are discussed below. . .
Of the six consonant symbols that may be differentiated by the presence or absence of ddus lene, only four show twin values in the Baghdadi pronunciation: g is Igl and fgf; k is Ikl and Ix/; p is Ip/ and IfI ; t is ItI and /9/ . There is only /bl for b, and only /dl for d, except that two words spelled with lenis d are pronounced with the emphatic spirant /i)/ ; these are lai)anaYI 'the Lord' wherever it occurs, and in the word Iehai)l only at the end of the phrase lai)anay elohenu ai)anay el).i)1 'the Lord our God the Lord is one'. There is a tendency toward fafxim , though' in precisely what cases remains to be investigated , so that Hebrew words often ha ve I ~/ , 1f[1/, l'v/, And M' IOstead of IZ/, Iml, Ibl, and lsi indicated by the spelling pronunciation rules. Hebrew r is always Irl, never Igl, except in the proper names Igahell and /i!,ahmin/, cf, WH/ ral)el/, Irahamim/. Vowel eqUivalences are somewhat more complex and some details
remain unclear ; mobile SIVa, as well as $~r~ and segol are lei in all
i
I f,
~
~
142
COMMUNAL DIALECTS IN BAGHDAD
positions in WH, but in MH mobile sIVa is zero, final unstressed lei is replaced by fiI and unstressed leKI is replaced by leK/ : thus WH/menassel (pr. n.) MH (with shift of stress) Imnassi/ ; WH/seferl 'scroll', MH/sHer/. Much the same holds for boltim, always 161 in WH, reduced in MH: WH/sel6m61 (pr. n.), but with shift of stress in MH/s16mu/· As for pallab and qamr$, they do not seem to contrast, both being rendered lal in certain positions a nd lal in others; at any rate, in MH there is a reduction of lal to lal and lal to lei similar to that described for the dialect: with shift of stress to penultimate position, final Iftl > lal (MH/ braxal for WH/ beraxal 'blessing') unstressed lal before KK becomes lei, e.g. MH/sekkanal for WH Isakkaniil 'danger'. As for !,ireq, it is rendered lei, Iii, and /l/, while qllbbll$ and SlIrllq are both rendered lui and 101; here again, the varia nt renditions seem to depend on position. '(e) As to morphology, some Hebrew nouns take Hebrew plu rals Ul!a ber! pI. l\.Iaberiinl 'friend(s)') but some take the Arabic broken plurals (fseddOrl 'prayer book' pI. Isdadir/). The word h lal 'synago'g ue', though not of Hebrew origin, takes the Heb. plur. IslawoO/. The fern. ending without pron. suff. is always unstressed lal (in WH stressed Ift/), but with suffixes it behaves like the dialectal morpheme T, e.g. Insamal 'soul' (WH/ nesamllf), Insametul 'his soul', etc. In general, the morphophonemics of the pron. suff. is that of the dialect. The same holds for verbal inflection: the root morpheme is Hebrew, but the patterns are dialectal. (d) The following is a list of words gleaned from J speakers, some in free utterances, some through elic;:itation. Only a dozen or so can actually be heard in common, everyday utterances, as will appear from t~e accompanying comments. They are listed in the order of the Arabic alphabet, with all vowel initial words listed under 'alif; Ipl follows Ibl, Igl follows Ikl, 17,1 follows Izl, etc. All forms are MH, unless otherwise stated "; the Hebrew etyma are given
in a transliteration of the Masoretic spelling. /a/)el/ . a term of general disapproval and disgust, 'terrible, awful', 'someone or something terrible or awful'; cf. Heb. 'aPrl 'in mourning, a mourner'. Thus / meOel al;>el/ 'like a fool, like an idiot'; /a/)el e'lek/,go to hell ; a plague on you'; la/)el wheffel/ '·· and /a/)ol la/)el/ 'very bad, very terrible', e.g. / kan 'asabi aswad , ya'ni a/)el la/)el/ 'he was awfully nervous, something terrible'.
SOME LEXICAL FEATURES
143
This term is very frequent, and one of the few Jewish terms well-known to M and C speakers, who use it as a mock J word and pronounce it lubbel/. lasOrl 'forbidden, unlawful' , Heb. 'aSlIl', same meaning; fairly common as a sentence word meaning 'don't do it' or 'that isn't done', i.q.
M/\.laram/, C/\.lagam/. Cf. also lessOrl, Heb. 'isslIl' 'a prohibition', esp. in tbe1c$$urJ, 'in an unlawfu l manner' lafe llul 'nevertheless, still', Heb. 'afillu 'even', e.g. lasma suwwit61u mli\.lat, afellu lala' mamegla(lil ' no matter how many favors [ did him, still he was dissatisfied' 11;>arbennan/ 'God forbid ', Jud.-Aram. bar-mi/lllan 'exclud ing ourselves'; used when mentioni"ng something
unpleasant or un-
lucky Ibraxal pI. Ibrax601 'blessing, benediction', Heb. b"axa, e.g. Isawwa braxal 'he made a (ritual) blessi ng'. T he verb 'to bless' is / begak/ , /ybegek/, in ' which the root is Arabic, judging from the /g/ and the / k/ , but the pattern qelal instead of the expected qalal (cf. OA/ b'rak/) may be due to the Heb. verb b~rax / l;>ahOr/ 'young man', Heb. bdbur, same meaning / peOahayyim/ 'cemetery', Heb. bre ha!lOyyim 'cemetery' (lit. 'house of the living') ; for IP/ < /bl in loanwords, cf. 3.21b above; whether this is reinterpreted as Heb. peeab bayyim 'opening of life' is doubtful ' / paSOq/ 'verse, a 'line of writing', Heb. pasuq 'verse' ; this word ha s the Arabic plural /ppasiq/ ; on /pp/ (viz. [ppJ) for / pw/ , cf. 3.21f Ip6qar/ , Iypoqer/ 'to deprave, render irreligious', apparently from Heb. epiqoros 'a freethinker' /\.Iabor/, 'a friend' , pI. /\.laberim/, Heb. bap~r, same meaning Ihax.m/ 'rabbi', pI. /haxa mim/, Heb. bdxam 'wise man' ; / \.laxa m basil 'chief rabbi' /I)eremj ' ban, ritual exclusion, excommunication' , Heb. b{!l'em, same meaning, e.g. /slehu bhereml 'pay no atte ntion to him , forget about him' Idasexl 'repelli ng, disgusting'; 1 have no reason to doubt the etymology given by informants, namely the initials d-$-k given in . the Passover Haggadah as part of the mnemonic device for the Ten Plagues. These three initials refer to the first three of the plagues, namely dam ' blood', $"rard~a' 'frogs' and kinnim ' lice', with regular change of /k/ to /XI in · word-final position, as
144
SOME LEXICAL FEATURES
COMMUNAL DIALECTS IN BAGHDAD
normal in wo rds made up of initials as well. The word has the sa~e pattern as such J adjectives as Idanes! 'impure', and phonic Simi Ian ty wlth/we~exl 'dirty' I ~~~~nl '~n old man, an elder', Heb. zdq~n, same meaning Iressanal New Year's day', Heb. ros hasSlina IseddOrl 'a prayer book', Heb. siddur; the plural is on the Arabic pattern, Isdadirl Isekkal in f'id essekkal 'the Feast of Tabernacles', called in Heb. sukkof! , pI. of sukkd 'a tabernacle'; however, J/sukkal 'a tabernacle (built for that feast)' Isekkfinal 'danger', Heb. salckdna Isammem,nveOI said of bad taste in food and the like Heb. sam hamllllilVe(} 'poison of death': Ihayi Ifelfeldara l)idda meOel sammemaweOI 'this pepper is as hot as all hell' IsebbaOI 'Saturday, the Sabbath', Heb. sabbdf! Isettax.a! ' t?e :a~s~ver ceremony', from the Judeo-Aramaic phrase . hasatta haxa tillS year we are here' which, followed by 'next year m Jerusalem', is one of the central passages of that ceremony; ·the term Iseder/, Heb. s~l>er, is also in use. Note the verb isattaxl 'to celebrate the Passover', e.g. Iwen qatsattex hassanal __ wh~re are celebratmg the Passover this year?' Isol)adl to bnbe , Heb. solwg, 'a bribe'; the pattern qOla! seems due to the pattern of the Heb. noun Iso!il 'a fool, a madman', Heb. sole, same meaning he~Idl 'prayer shawl', Heb. ~i~if! 'fringes (of the prayer shawl)" with unusual Idl for f!; i.q. l!emO/, Heb. 'Iallif! 'prayer shawl'
rou
I~a'ar/ 'sorrow, 'grief', Heb. $Q'ar, same meaning ; hence the verb
I!!a"arl 'to be grieved' /,arelil 'a C~ristian', pI. j'arellm/, Heb. 'dr~! 'u~circumcised'-; syn. of Ine!ram/, pI. Ina!ara/, the Hebrew term being the more pejorative
/qeddusj 'consecration', Heb. qiddus, same meaning; contamination
of the Arabic and Hebrew roots, with the Hebrew pattern Iqayyaml 'to ca rry out, fulfill (a religious commandment)" Heb. qiyy~m, same meaning; the verb has the Hebrew root and the Arabic pattern jkuww.ana/ 'religious concentration" Heb. kalVwand , same meaning. with the regular shift of unstressed lal to lui before Iww/, e.g.
145
Isawwa kuwwana qabel lebraxal 'he concentrated before making the blessing' IkabOdI 'honor, glory', Heb. ktipod, same meaning IgelgOl1 'reincarnation', Heb. gilgul, same meaning; severe pain or grief . is called I~a 'ar malgelgulf ; on the same root, Itgalgalj'was reincarnated', e.g. Itgalglet ensametul 'his sou l was reincarnated' /gennab/ 'a thief', Heb. ganndp, with convergence of the Hebrew and Arabic nominal patterns, cf. J/keooabl 'a liar' ; hence also " the 'terb Igannab/, fyga nneb/, 'to steal' Igoyl 'a Gentile', esp. 'Muslim', pI. IgoyIm/, Heb. goy 'a Gentile' flax/, Iyllxl ' to run away' : if from Heb. italax, 'to go (away)', imptv. l~x, this would be the sole term widely used in M and C having a Hebrew etymology, hence borrowed from J: fro m a C speaker, f<;lagabu facjcjageb wliix fadlexal ' he hit him once and beat it'. Needless to say, the etymology is far from certain; Ganima (1926, p. 584) includes it among words of presumed Aramaic origin, but his etymology is even less convincing. l!]labOII 'flo od, deluge', Heb. mabbul, same meaning; unusual loss of gemination Imezzall 'luck', Heb. mazztil, same meaning. A frequent curse is l\Vaqa' mezzalak! ' may your luck fail'; the term is in frequent use, and is ~nown to some
Me speakers
l!]lasall 'an example', Heb. maMl, same meaning I me~wal ' commandment, a good deed', Heb. mi$IVG, same meaning Imgellal 'the Scroll (i.e. the Book of Esther)" Heb. m'gilla 'scroll , esp. the Book of Esther' l!]la!]lzerl 'a bastard', i.q. Inagel/ ; Heb. mamz~r, same meaning Insamal 'soul', Heb. n.sdma, same meaning: Iwe~ed 'ensametu bellaxl 'they are very close friends' /naggaf/ 'to die', in a pejorative sense, cf. Fr. crever, Germ. krepieren; apparently from Heb. niggaf 'he was struck down by God' InlHal 'ritual washing of hands', Heb. n'lila, i.e. 1I.{ilaf! yaoayim, same meaning
Iharagl 'to beat', Heb . hdrag, 'to kill' 6.5
SELECTED LEXICAL ITEMS.
Below is a list of MJC words
selected for their historical or comparative interest; some are com-
"i,
146
COMMUNAL DIALECTS IN BAGHDAD
mon to MJC and typical of the Iraqi area, others are peculiar to one of the dialects involved. They are listed in the order of the Arabic alphabet with Ipi following Ib/, Icl following /j/, and Ig/ following /k/. Unless otherwise specified, Aleppo forms are from Barthelemy's Dictionary, Palestinian and Egyptian forms from my own informants, Central Asian forms from Vinnikov, 1949, Glossary, and Kweres forms from Meissner, 1903b, Glossar. I was unable to make fu ll use of Vinnikov 1962, which I received while this work was being printed. MJC/abu/ in construct phrases, besides the usual meanings and the "automatic kunya" (see note 150), also denotes owner or agent :
MJC/abu Ibet/ 'the landlord', tabu Ixiin/ 'the innkeeper', tab u ttaksi/ 'the taxi driver', etc.; fern. M/ umm/, IC/emm/ MJC/iidmil 'man, a human being', pI. MJ/awiidem/, C/awedem/, contrasting with MI/niis/, C/nes/ as enumerative vs. aggregate. Common to the Mesopotamian area as a whole, both gelet and qeltll, and is the normal term in Central Asia as well. In other . areas , e.g. Syr.-Pal., only as adjective, 'affable'. This latter usage occurs at least in C, Ines awedem/ 'nice people' MJC/asfi/, /asu/- ; here are some illustrations of its use, all from M, JC usage being quite similar: /asu-jet/ 'so, you did come', viz. 'ah , here you are'; /asu-majet/ 'so you didn't come, as it turns out'; /aso. aso.! lalSiI enta kull si !/ 'now, now, don't you pick up anything! (let me carry the valises), Iha ' ad nan, asu-max l;>u~ , sbik?1 'hey, 'Adniin, you seem preoccupied; what's the matter?'; /as u-trid et'arek wiya/ 'it looks as tho.ugh you want to pick a fight with him'; /asu-gumet; sbik? matug'ud!1 'you got up; what's the matter? sit down! ' Some speakers connect this particle with /asuf/ 'I see', not an implausible etymology; though cf. las hii(wa)/ 'what is it'? Malaika, 1963, p. 37 and passim, writes lassul MJ CfakuI 'there is', occurs throughout the Mesopotamian area except in Anatolia, including Mardin, where there are such forms as /fihul , l fiyu/. The MJC negative /miiku/, Imiiku/ is not quite co-terminous with it: 'Ana has /maku/, some of the rural dialects of Lower Iraq have Imiimes/. A variant of /aku/ heard from M speakers is /akusf. A general inquiry upon meeting acquaintances is Isaku smiiku/ 'how are things in general', cf. Turk.. ne var, lie yok. Barthelemy (Diel ., p. 776) writes akll and o
SOME LEXICAL FEATURES
147
gives the etymology hak-hu, an ingenious conjecture. Malaika, 1963, p. 58, adds a C form /akun/ M/ames/ 'yesterday', less common synonym of /Ibiirl)a/ ; the contrast occurring in some Beduin dialects,16 7 whereby the former is -'yesterday' and the latter ' last night', seems absent from M. In J only /b6l)i/, in C/mbel)a/, as in Mosul ; in Anatolia / mberl)a/ and /ams/ J/enf/ 'nose', for which MC/xasem/, the latter being the typical Mesopotamian term, found both in gelet and the qeltu-dialects; Aleppo has /enf/, and my Swerek and Carmuc informants use /efen/, with metathesis, though they know /xasem/ as well. In Central Asia, 'nose' is un/, whi le xasim is 'mouth' as in Egypt; the forms cited suggest an unattested OA ' /unfl rather than Cl. Ar. /'anf/ as the ancestor
MC/englizi/, J/englezi/ 'English'; I have no explanation for the difference M/ahad/, J/ahd/ 'contract', MJ/t'iihadl 'to make a contract'; (C form not noted) cf. OAr ahad/ ; the shift seems isolated, but occurs in the same root in Aleppo, cf. Barthelemy Diet. , who cites a similar phenomenon in Western Syriac
C/baddiiwil '(in) the language of the Baghdad Muslims', a term unknown to my MJ informants, who have no special designation for any of the communal dialects; see note 108 M/l;>arii;um/ ' thick lips', cf. M/ml;>ar;um/ 'pouting'; in 'Ana I l;>rii;em/ is the ordinary word for 'lips' , which is MJC/sefaf/, sing. M/seffa/, JC/seffi/ M/bazzun(a)/, I/ bezzuna/, C/bazzuni/ 'a cat'; in M the form / bazzun/ seems preferred as a generic term, in which case it is nevertheless feminine: M/etrek elbazzu n latxarmsak/ 'leave the cat alone or it'll scratch you'; the same holds for 'Ana [b~zzunl, whereas in JC and Mosul the form with T is generic. This is the term used throughout most of the Mesopotamian area and in Central Asia (bazllna); in Anatolia the term is [~'nnorl, reHecting the name of the cat in older Iraqi vernacular literature (cf. Flick, 1955, p. 168; -:ra1iqani, no. 174, sillnawr, no. 263, sinnawra, but
a'lso no. 110., hirra) M/ biicer/ see /gadal MJC/balil 'yes'; on 'inllila of CI. Ar. bola, cf. 3.36e(iii) The IiI in the M form suggests the word may have been re-borrowed from
.'
~ '"
148
COMMUNAL DIALECTS IN BAGHDAD
Turk. beli, Pers. bale. Its use as a plain affirmative rather than a counte r affirmative to a negation (Fr. si) is dec ried as an error by J:lariri (d. 1122) pp. 119- 20; in MJC it is used more or less synonymously with / na'am/ and IiI, as it is over most of the area. Meissner reports it for Kweres but Weissbach (1904, p. 938) adds for the same dialect the term mbala with th e meaning of Germ. doch, Fr. si as in man y Syr.-Pal. varieties MJC/ bab/, 'door, gate', is fern. in ordinary usage in M (despite /bab e55arji/ 'the East Gate (qua rter),) but masc. in lC. M/ lbab tasa' jemal /, the door is wide enough for a ca mel' (hint to a vis itor to depart) M/ bibi/ term for addressing the grandmother, MJCjjedda/ ; in Central Asia, ' lady, mistress', Pers. bibi ' matron, grandmother' C/ bi'a/ 'a chu rch', the normal term in C, for which M/ kanisa/ J/ knisi/ M/ta ra/, JCjtaga/ a "fi ller" particle with little semantic content of its own, usually occurring immediately before the subject of an imperfect or of a no minal sentence; if the subject is a pron. affix,
it' comes before the ver b. Some examples from M (JC usage is analogous) : / tara da 'ag ullakj'I'm telling you'; / bass elmuhemm, tara
da'awa~~ik
menhassa/ 'now the main thing, I'm making
my request as of now'; /web na tara sgad matu' mur ba(\rin/ 'as for us, whatever amount you request, we're at your service';
/tara hfIwa dayman yel,awwar nafsa, ya'ni , sayeg mal 6ahab/ ' he always imagines he is, well , a goldsmith' MJCjtufga/ 'rifle, rifle shot', pI. /tufag/ ; the mock J form /tefqayi/ showing the regularly expected M-J equiva lences (ju/- /e/, /g/- /q/, /a/-/ayij) occurs in a Muslim jibe imitati ng the Jewish dialect: /1,1esqel , asqal qalbak mentaqqet ettefqayi ?/ '/ besqel/, what did your heart say when the rifle went off?' The J-like effect is achieved by piling up /qj's; the name jbesqelj 'Ezekiel' is one of the most common among Baghdadi Jews. The jibe has beyond it an anecdote purporting to show the pusillanimity of Jews, especially when faced with firearms,168 Pees. lofiing from wh ich also Kweres Ifugga, Aleppo Ifang MJCjtemmanj 'rice' so called when uncooked ; an Iraqi wo rd , of unknown etymology. When prepared for serving, it is called by various other names, including jplawj, (Pers. paloII', T urk. pilav) and , in some dishes, by the common Arabic name Mjrezzj, JCjgezzj, e.g. the dish called Jjgezz beblibj 'rice with milk'
SOME LEXICAL FEATURES
149
MJCjjamj 'glass (the material)', Pers. jam, Turk. cam MJjederj see jgeder/ Mjjrectij, JCJjgedi/ 'a mouse, a rat', pI. MJjerdan/, Jjjegdinj C not noted ; on j d/ for etymological j6j, see 3.22 MJCJjigara/, Jjjigara/ 'a cigarette', pI. MJCjjegayerj. Com mon to most of the area, including Anatolia. T urk. cigara in more modern usage sigara; the initial affricate is probably a result of the .Turkisl;> treatment of [ts], as in Greek tsigaro ' 6 ' MJCjjaj '(also M/ejaf) j yejij ' to come'; for parad igm, see 4.8 . Mjja leddenyaj, JCjja leddeni/ is the ordi nary expression for 'to be born', though M also has / nwelad/, JCjnwaladj MJCfjabj, jyjibj 'to bring' as in practically all Arabic dialects, also in the sense of 'to bear (a child)'; in the latter sense, the ve rbal noun is MjjebUbaj, Jfj ibfIba/ , CjjebfIbij Mjjiirabj, I jjurabj, Cjjagabij 'a sock , a stocking', pI. Mfjwaribj, Jjjwa ribj, Cfjawegebj . Mosu l has the same form as C, 'Ana has (jariiba], pI. (jwarib] . Various other forms are fo und in the nearby areas, e.g. Pal. jjra9j,Aleppo fj rabj, Cairo jsura9j and jsara9j . Cf. Turk. ~orab . CI. Ar. jalVrab, Pers. gorab MJCj~araj, IC also j~agaj, 'remedy', especially in j maku caraj 'there's n"thing to be done'. Pers. care, Turk. ~are, 'remedy' Mjcarakj, JCjcagakj 'a quarter', i.q. Mjrubu'/, JCjgebe'/. Pers. tarok Mjcarpayaj, JCjcarpayij 'bedstead', Pers. carpaye 'stool; bedstead' MJCjcardaxj 'hut, cabin' (esp. for bathing on the Tigris), cf. Pers. cartaq
J/cereg/ (MC not noted) ' fl imsy, no good' (cloth, merchandise, piece of work) ; cf. Aleppo j corok j , 'damaged', Eg. j surukj ' unfit (for m.ilitary service), defective' ; from Turk. riiriik, wh ich covers all
these meanings MJjcaqlabj 'to tumble, fall over', and Jjceqelbayij 'a tumble, a fa ll '; perhaps from *jtsaqlabj ; cf. Barthelemy, Dict. , s. v. saq/ab MJCjecaliabj, JC also /eallabj, 'to hang on to, cling to', Mjmenjatti '. bibiti ~e nna neecallab biha, tebcinna bcayaj 'when my grandmother came, we (child ren) would cling to her until she told us a story'. If from Mjcaleb/ 'dog', it is a borrowing from M in JC; but cf. Mosul jtkalbacj, same meaning, cf. MJC jkalepcaj 'handcuffs' Mjcenga!j, JCjcengalj 'a hook' in JC also 'a fork', for which M
150
COMMUNAL DIALECTS IN BAGHDAD
speakers prefer /eata!!; for 'fork' also J/~ekkiixa/. Pers. tangal, Turk. rengel, ratal M/J:tebla/, JC/i)eble/ 'pregnant' ; for 'imtila in older forms of this word, see 3.36e(v). Mosul has the same form as JC, and the stress and length may be connected with the fact that these three dialects have fern. adjectives of infirmity and color with a final stressed long vowel, J!~emM!, C and Mosui /i)amga/. Other dialects have ' imtila in this word, e.g. Aleppo and Damascus /i)Qble/ but Jerus. Ii)Qbla/ ; in these dialects color adjectives end in lal and ' imtila of final /a/ is rare except in the fern. ending; cf. Cantineau, 1946, p. 152, who finds this word "avec une legere 'imtila, difficile it expliquer" in a region of the Horan where there is otherwise no 'imtila of final /a/ . M/i)un(a/,.JCji)en(a/ 'wheat'; so throughout the Mesopotamian area, whereas Syria, Egypt, and dialects further west usually have a reflex of OA/qami)/. Such a distribution is hinted at as early as the ninth century by Jal)i? (ed. 1948, p. 17) who says that iiin{a is used in Lower Iraq (Iuga kuflya) whereas qamb is used in Syria (/uga stimlya) JC/I)wiis/ 'clothes' M/hdiim/, both plural; Mosul has /i)was/, 'Ana /hdiim/, Qarabii~ / bawis/, the rural gelet dialects / hdiim/ M/oxu1;>a~/, JC/oxa1;>a~/ 'to be distracted, preoccupied, have some· thing on one's mind': M/hay ujrat elmu'ayana Jigeddam axiif tuoxu 1;>u~ bacer/ 'here is the fee for the consultation in advance, in case you have other th ings on your mind tomorrow'; M/defukkila 1Mb 'iid, le~ eoxu I?a~ti?/ 'go on and open the door, what are you dreaming about?' MCjmax\;>ii~/ J/mexQii~/ 'preoccupied, having something on his mind' . M/xetal/, /yextel/ 'to hide (intrans.)', for which J/x taba/; the game of hide-and-seek is /xettela/ in M, / mexta Qoya/ in J M/xa~em/ 'nose'; see /e nf/ J/xliyi/ 'a sin', commonly a sentence-word used to deplore regrettable acts; /jiinu x!iyi 'alaI 'he felt sorry for' (cf. ex. s. v. /garag/ below). Aleppo /xlayyel has much the same sense, as has M /xa!iya/ MJCjxo/, /xob/, particle used at the head of phrases to imply the speaker's hope or conviction; M/xo mlmeset?/, 'you didn't forget, did you?'; J/xo mat'eMet?/ 'I hope you didn't hurt yourself?'; C/xo magei)tem?/ 'you didn't go, did you 1'170 C/xo ana ~'alayyi?/ 'what do 1 care?; C/xob kan jit honi bwai)dak/
I
SOME LEXICAL FEATURES
151
'you should have come here by yourself'; C/xo ma'aji bwal)di/ '[ don't (like to) come by myself'. Cf. Pers. xl/h, older XDb 'good, well', and Meissner, 1903b, Glossar, 'ho(u)b' 'natiirlich , doch' MJC/xoja/ is used for addressi ng a Jew;- some informants also know the literal meaning 'merchant', and it occurs, at least in
stories and proverbs, as a traditional honorific Muslim title, e.g. /xoja 'ali, mulla 'ali/, used proverbially in the sense of six of one and half.a dozen of the other, where /xojal is the equivalent of /mulla/ (!:Ianafi, 1962, p. 169, no. 757); Meissner's stories have the doublet /xawaja/ as an honorific title (e.g. 1903b, p. 57), and one C informant tells me that in Baghdad /xawaja/ followed by the given name is, as in other regions, sometimes used for addressing Christian men . Cf. Pers. X(IV)aje and the article /xawiija/ in Barthelemy's Diet. MJC/xM/ 'good', invariable, precedes the substantive: /xos walad/ 'a good boy', /xos bent/ 'a good girl', /xos nasi 'good people'; with article, /Ixos walad/ ' the good boy'; often stressed , esp. in fixed expressions: M/x611ebi/ 'a jolly good fellow'; Cf. Pers. xos, though the special positioning is more reminiscent of Turkish! 71 MJCjdazz/, /ydezz/ 'to send', cf. Barthelemy, Diet. MJCjdiid/, term , for add ressing one's brother M/dumbug/, pI. /danabug/, JC/dembek/, pI. J/dnibek/, Cjdanebek/, 'a sort of drum', cf. Pal. /durbakke/, Aleppo /d.rbakke/, Eg. /darabukka/. The Pers. diinbiik seems to be at the root of all of these, though the JC forms are the closest. Cf.. Barthelemy'S Diet., S.Y. drhk '72 MJCjdoxan/ 'to feel faint, dizzy', perhaps related to MJCjdax/ ' to faint' M/oabb/ C/dabb/ 'to throw down, throwaway', for which J/farr/' Cf. Barthelemy, Diet., p. 229, top M/rawa/, /yrawi/, J/gawa/, /yegwi/ 'to show' for wh ich C seems to have only the common MJCjsawwaf/, same meaning MJC/zaqnabiit/ ' poisoned food, anything detestable', used in speaking of food (and, by extension, anything else) disapprovingly 'or derisively J/zalez/, Cjzelez/ 'naughty (child)' M/se'al/, /yes'al/, JC/s"yal/, /ysayel/ 'to ask'; the /'I of the M form suggests this may be a loan from CI. Ar. , cf. also M/su'al/,
"
::1
'"
152
COMMUNAL DIALECTS IN BAGH DAD
M1C/suwal/ 'a question'. The two roots vie with each other in other areas as well, cf. Pal. /sa'al/ (Jerusalem, Galilee) and /sayal/ (Central Pal. village) ; cf. Barthelemy, Diel. s.v. sa'al III M1C/sangin/ 'strong (of tea)' ; Pers. siingin 'strong, hard, heavy' from sang 'stone'
M/si!, l /sen/, /seni/, C/se/ 'a thing, something'; M1C/si 'ala si/ 'all in a ll, all things considered' M1C/meswar/ 'soon, in a little while', e.g. /yeji meswar/ 'he'll be right along'. This word, which in the Syr.-Pal. and Eg. areas is a noun with meanin gs centering about 'a trip', has, to my know~ ledge, received no satisfactory historica l or comparative treat-
ment; cf. Barthelemy, Diet. M1C/saf/ , /ysilf/ 'to see', though usage is not precisely the same in the three dialects, cf. 6.2 abo ve. In M and C, this contrasts with / bawa'/ 'to look', also /dal)l)aq/, 'to look'; in 1 it has the meaning 'to catch sight of' and contrasts with /,ayan/ 'to see' and /~afan/ 'to look'. Mosul has /qese'/ and /gese'/ 'to see', /dal)l)aq/ 'tq look'; 'Ana has / saf/ 'to see', /dal)l)aq/ 'to look'; Qarabas has [ara] 'he saw' [areytu] 'I saw'. On qr, cf. also Barthelemy, Diet. Central Asia has both 'ayyan and sal, used more or less interchangeably! 73 M/~udug/, lC/~edeq / 'true, right' commonly as a sentence word 'it's true; that's right' M1C/~al)iJ:>/ 'true, right, correct'; as in other areas, this seems to be a classicism, the natural dialect forms being the doublets meaning 'whole," healthy', l/~l)il)/, C/~al)et)/. (M not noted); cf. Barthelemy /~l)il)/ 'entier' vs. /~al)il)/ 'vrai', and in Eg. /~el)il)j vs. hal)il)/ M1C/~affal/ 'to put away (shelf, pocket, etc.), arrange in its .place'; M /~affal leHus ebjeba/ 'he put the money away in his pocket'; also l h affal lemwa'in/ 'to do the dishes' MC/~a ! a/ 'prayer', l/~ !a/ 'synagogue', M/kanis/' MjC/~ u c/ 'fault', e.g. M/e~~u6 mu 'alek/ 'it's not your fault', /hay ~Ocna/ 'it's our fault'; Turk. sur, same meaning Mj~ aya/ especially in j !;>~iiyat a!!a/ 'by God's protection, thank God' ; Turk. saye, Pers. saye, 'shade, protection' M1C/~al)/, j~il)/ 'to caU, name' : M/axuya da~il)la dadj 'my brother, I call bim /diidJ'; MlC/~ayyal)/ ' to shout'· . In Central Asia, ,aYYlI!:z has the meaning 'to call, name, cry'
SOME LEX ICAL FEATURES
153
M/<>aQlQlj, /y<>UQlQlj 'to hide (trans.), for which l /xaba/, jyexbi/, C not noted . Central Asia has zamma (regular reHex of OANamm!) for 'to hide (trans.), l /<>aww/ 'fire', for which M/nar/, C/nagj, In Syr.-Pal. /<>aww/, /<>ayy/ is common for 'light', but for 'fire' I find only the Central Asian form zalV, glossed niir in Tsereteli, 1956, p. 12 MNuwa/, lNawa/, C/c)awa/ 'light' M1C/\araf/ 'neighborhood', smaller and less official than M/mal)allaj, •. l jml)>llla/, C/mal)alli/ 'quarter of town' lC/la'a/, jyel'iJ 'to give', for which Mjenlaj (jnelaj) /yen!i/. The lC form is also that of Mosul and apparently more typical of qellll dialects, the !vi form is found in the other gelet-dialects and in Central Asia. Forms similar to the latter are common in many localities of Syr.-Pal. (e.g. Galilean and l;Ioran /an!a/), and Barthelemy lists it for ' Ia langue des nomades" ; I have noted /la'aj among the Negev Beduins l jnlagaq/ 'to get scared'; /lagqaj 'fright' M1C/lawli/ 'backgammon'; the same term recurs throughout the Syr.-Pal. and Eg. areas as /lawla/ or /lawle/, and the origin of the ji/ is puzzling: if from Turk. lavla, It. lavold, the ja/ might have been reinterpreted as a fern. suff. , with the allomorph fiJ quite plausible in this environment for C, nearly as plausible for 1, but in M explainable onl y as a loan from 1 or C (cf. 4. 3) . The Greek name of the game is lavli; could it have passed into Iraq, perhaps t hrough an unattested Turk. *lavli?!74 l /laq/,lY!iq/ 'to be able', i. q. lC/qadag/, jyeqdeg/, M/gedar/, /yegderj MC/'erbi/ and I' urbi/, 1l'rubi/ 'a village Arab ; a provincial' also as adj. 'provincial'. The collective Ml'ara!;>/, lC/'aga!;>/ is used to denote the village population as a whole, to the exclusion of the full nomads (MJC/ badu!) in contrast to other parts of the Arab world, where terms derived from the root 'rb refer to Beduins, whether nomadic or semi-nomadic. No doubt the Beduinized character of the sedentary rural popUlation has something to do with this! 7' M/,a ru~j, l/,gu~/, Cj'ag6~/ 'bride', pI. M /, araye~/, J/,gaye~j , C/,agayye~/ M/,ari~j and !,arri~j l/,eggi~/, C/, agj~/ 'bridegroom', pI. M /,erarj~/, l/, gagi~/, C/,agayye~j
M/,ag urba/ and /,agraba/, l /,eqquba/, C/'aqgabi/, 'scorpion', pI. M/'agarub/, l /,qagib/, Cj'aqegeb/ ; on /qqj < /qgj, cf. 3.24e
,. "
:1
154
COMMUNAL DIALECTS IN BAGHDAD
M/,agrugga/, J/,eqruqa/, 'frog' , pI. M/,agrugg/, J/,qariq/, C not noted ; cf. Aleppo 'qarqa, call. 'qarroq, and Syriac 'aqruqa (Barthelemy, Diet.); Kweres has 'agriig MJC/,alamOdl 'for the sake of, because of'; Mj'alamOd hay ma'ejawl 'that's why they didn't come'; Ijet 'alamOdakl 'I came for your sake'; /,am elrpa(\i cent ad fa' mit dalar beSsahar 'alamOd gurfa we!)dal 'last year I paid a hundred dollars a month for one room'
JCfgada/, for wh ich M has Ibaeerl 'tomorrow'; the gelet-dialects have Ibacer/, the qeltu-dialects are split: Mosul Igada/, Anatolia [gade], 'Ana [1;>ACor], Hit [bakor] M/garbil/, J/gegbil/, Cfgarbell 'sieve', pI. M/grabil/, J/gegabil/, Cfgarebil/ ; an OA Igirbal/, cf. late Classical girhiil, could account for the J form, for Mosul Igubel/, (cf. 3.24d), Aleppo Igorbel/, Pal. and Eg. /f!,irbal/, but C and M pose a problem ; Central Asian galhir CTsereteli, 1954, p. 256) in a dialect without 'imala, taken together with the M and C forms, implies alternative OA forms such as */garbill or the like; on possible etymologies, cf. Dozy (Lat. harpelus?) and Fraenkel, 1886, p. 91 MJCfffadl 'heart, entrails'; Iwaj' effadl 'stomach ache, indigestion'; JCfgas effad/, i.q. Iffad / ; for Mardin, Socin 1882, p. 240, writes fUIVIid, flVlid ; in Central Asia, fod 'heart' (Vinnikov, 1957, p. 429). On Iffl < IfwI, cf. 3.21f J/qebba/, Cfqebbi/, M/gu1;>1;>al 'room', pI. JCfqebab/, M/gu1;>a1;>/ ; in somewhat formal speech, M/gurfa/, Igerfal may be used; Mosul has Iqebbi/,' but Anatolia has reflexes of Turk. oda, as in Syr.Pal.' and Eg.; Iqobbel occurs in Aleppo for some kinds of rooms J/qerrayil 'oil lamp lit on Sabbath eve', apparently unknown to MC speakers; note Irr/, though qgy 'to read' JCfqemag/, M/gemarl 'cream, esp. clotted cream', Mosul Iqemag/, Aleppo Iqaymaql from Turk. kaymak, same meaning; the M form seems to be derived via the JC or Mosul form, reinterpreting the Iql and Igi as equivalent to those of such words as JCfqamag/, ' moon', M/gumar/. A doublet is MJCfqaymag/, in Idonderma qaymag/, a sort of ice cream; cf. also Pers. qeymaq, from the above Turkish form MJCfkefl in MC/slon kefak/, J/aslan kHaki 'how are you?', but never used for 'how?' except in semi-literary style; MJC/,ala kHakI 'take it easy' ; used for ' because', at least in J, but now-
SOME LEXICAL FEATURES
155
adays less common than the literary loan IIi'ann/ : J/maysig wniki yedxel walad sabb, kH lem'elmat kelletem sefUrl 'a young man is not supposed to go in there, because the women teachers are all unveiled' ; in thi s sense M has lee fl (Malai ka, 1963, p. 37) J/gederl 'pot', pI. Igdura/ ; the Igl and Irl would suggest a loanword from M, but for the fact that M has {jeder/, pI. Ijdura/, which in turn suggests th at M borrowed this from a rural dialect (cf. 3.26) e.g. Kweres gid(ii)r. Possibly M once had Igi in this word , whence the J form. Central Asia has gidir. This term (qidr, pI. qlld!lr) is given by Ja!)i, (10th century) as a cha racteristic Iraqi term (ed. 1948, p. 18) JC/garag/, roughly i.q. MJCflazeml in the sense of 'probably'; my M informants say they never heard this word. Ex.: J/garag mallelwal 'probably she's not pretty'; J/ ne!)kilak e!)kiyetna, wenta garag yejik exliyi 'Ienal 'we'll tell you our story and yo u surely will have pity on us'; Cfgarag I)assa l ektigl 'he must have made a lot of money' ; Cfgarag kensiifal 'surely he's seen her' . Turk. gerek 'necessary' J/ganal 'a duck', call. Igan! MC not noted ; in most dialects, l 1;>al\(a)1 MJC/gaddal 'to beg for alms', Imgaddil 'a beggar', Igedyal ' beggary'; M also has {jad da/, Imjaddi/, Ijedya/ ; Barthelemy Diet. has Imgaddi/, considered as a Mesopotamian loan, for Aleppo, and cf. Central Asian gidday 'beggar'.I76 This old borrowing from Pers. gedye appears in older Arabic Sources spelled with k (e.g. Taliqani, nos. 444, 460, and cf. the dictionaries, e.g. Freytag, quoting the Qamiis; and MUl1jid, Dozy, etc.) or with q (Dozy, one source, from Spain) both probably intended for IgI MJC/g!asl 'a glass for drinking', Eng. glass J/genzl 'a treasure' (MC not noted), cf. Pers. giillj which, no doubt in some other form ending in IzI has passed into late Classical as kallz and into other dialects as Ikinz/, with the Ikl reflecting the Arabic spelling MJCf laket/, Ibalaket/, variants of Ilakenl ' but'. Note the absence of 'imala in the JC form s, despite the fact that o'ider 'imala is attested in this wo rd (Blau, 1961 b, p. 19, from a tenth century lraqi source) and in Pers. valiken, whence modern vali
J/lemmall 'ants', IJemmalayil 'an ant' (MC not noted) ; most dialects have regular reflexes of OA /nam l/, but cf. Mosul Inem mall 'ant s'
'"
156
•"
157
COMMUNAL DIALECTS IN BAGHDAD
SOME LEXICAL FEATURES
MJC/161 'or', e.g. M/tamam 16 la?1 ' right, or not?' MJC/lax/, Iylixl 'to run away', cf. 6.43 MJC/lekk/, pI. Ilkiikl 'a large quantity, specified by some as ten thousand, by others as a hundred thousand ' . lhaila yemlek elkiikl 'he's tremendously rich'; cf. also Ilaklaki 'to get rich', e.g. J/mal)l)ad laklak mneleqmagj ' no one ever got rich from gambling'. Cf. Barthelemy, Diet., s.v. lakk (2) a nd the Indian term rendered in English and French as lac, lakh, 'one hundred
to Fraenkel, ibid., marzlpii and cf. Mishnaic Heb. marzeP. Cf. Pers. me,zab, probably from the Arabic MJC/mesl):l1 'shovel', common throughout Iraq; Jal)i ~ (ed . 1948, p. 19) says that the people of Kufa (in contrast, by implication , to those of Basra) replace this word by bdl, cf. Pers. bil, older bet, a word of which I find no trace in Arabic today, except for Central Asian bll MJC/me'mill 'one with whom one habitually trades, i.e. either a
thousand' MJC/mal/, particle denoting possession, connection, etc., one of the most uniformly characteristic features of the Mesopotamian area, common to all qeltu and gelel-dialects examined so far, including Anatolia, Khuzistan, and all points within Iraq ; it is found even in modern Persian (ma/~e soma 'of yours, belonging to you'; mal-e koja, 'from where') but apparently not in Central Asia. With suffixes, it can occur indifferently with or without
the fern. T, e.g. Innad i malnal or Innadi malatnal 'our club'. Some examples of its manifold uses (all in M, JC being analogous): Isayeg malilahabl 'a goldsmith', Imez malxasabl 'a wooden table', Imudir malmadrasal 'a school principal', Il)amui)I)elu maleblad / 'imported sweets' Ubladl = 'abroad') MJC/l11ardil 'pole for pushing a boat' ; M/dfa'ha bgusl;>a wala tedfa' ha I;> l11ardil 'push it with a reed (now) and don't push it with a pole (later)', i.e . 'do it while the doi ng is good;' Idaf'at l11ardi w'asat kurdil 'a shove by a pole and a blow from a Kurd' used in describing a prompt departure or expulsion M/mezrib/, JC/megzibl 'gutter, waterspout off roof', pI. M/ mzarib/ , J/mgazib/, C/magezib/ . Cf. Aleppo Imnreb/, and in non-'imdla dialects Imezrabl (No . Pal.) or Imezrabl (l:Ioran, Cairo). These Syr.-Pal. and Eg. forms, are regular reflexes of OAimizrab/ ; the J form offers no particular difficulty, especially since OA Imirzabl is also attested (Freytag, citing the QamOs); in fact , Imirzabl may be primary and the form Imizrab/, more co mmonly represented today, a metathesis (Fraenkel , 1886, pp. 24-25). The difficulty is in the /i/ of C instead of th e expected lei, and especially in the Iii of M instead of the expected la/. Either M a nd C have in this wo rd preserved the same ancient 'imiila as J, or the three forms har k back directly to the Aramaic ancestor of th~
OA form ; the most likely form of that ancestor is, according
steady customer or one's regular merchant'
MJC/magazal 'a large store with varied articles, larger than a Imaxzan/, itself larger than a Idekkanl or Itekkan/, ; Pers. miiqaze 'store', from Fr. magasill, itself ultimately from Af. maxazill ,
pI. of maxza" 'store' M/munxul/, l lmuxell 'fine sieve, strainer', pI. M/ mnaxel/, J/ mwixel/ (C not noted), cf. CI. Ar. mUllxul, with regular reflexes in M, in Pal. Imunxul/ , Aleppo Imonxol/, but in some Central Pa l. villages, Imiixell as in J (Bauer, Wort. , s.v. Sieb). The forms in 10/ no doubt stem from an intermediate -[muxuIJ, nasaliza tion of vowels preceding Inl plus velar being ve ry co.mmon in J and other dialects: Jldanqii lj -' let's say', lenxill 'palm trees' are normally· [diiqul, ;;xil] MJC/mawil 'bl~e', partly synonymous with M/azrag/, JC/azgaq/ , which covers other shades as well; from MJCjl11ay/, Il11ayyl 'water'; cf. Pefs. abi 'blue', from ab 'water', and Turk. ma vi
J/medil 'table set for the Sabbath , Passover, or other festive meal', apparently no MC equivalent. The J usage thus antedates the remarks of !;Iariri (d. 1122) who criticizes his Baghdadi audience for using md'ida and mayda for any table , rather than for a table on which a meal is set, as correct usage requires (l:Iariri,
ed. 1299 A. H., p. 10)
MJC/mezl 'a table', pI. MJ /myuza/, C/myOzij. From Pers. miz, older mez. Used throughout the Mesopotamian area and in the Eastern Neo-Aramaic dialects (Zakho meza); even in l:fariri's time, the
word he suggested for a plain table (cf. preceding item) was ·Persian, namely x(lV)an, in later CI. Ar. xilrall or XlI\ral1. The Persian term miz, mez no doubt reflects Portu guese mesa, just as other dialects reflect other Mediterra nea n lan guages ( Lat.
tabliia and its descendants, etc .). Oussani, 1901, p. 110, gives
158
COMMUNAL DIALECTS IN BAGHDAD
'dinner table' for mez, and lists {Gwle as 'table'; the latter is un·
known to my informants; see /!fnvli/ above M/ mewa/, JC/ miwa/, though C also /mewa/ 'fruit'; Pers. mive, older meve M/nages/, JCjnajes/ 'dirty, unclean, filthy' and M/naggas/, JCjnajjas/ 'to render unclean'; I have no other instance of M/g/ for OA/j/, nor any explanation for this one M /na~i/, Cjne~i/, for which J/na~~i/ 'low', pI. MC/ne~~ay/, J invariable. Meissner, 1903b, Glossar: na$ls 'niedrig'; !;fanafl, 1962, p . 143 /n~ayye~/ 'low' M/ nagalJ, JC/nagel/ 'bastard', pI. M/ngu!a/, J/ngula/, C/nguli/ ; cf. the J proverbial expression /essa!! mani l)fagu wennagel mani bzagu/ 'who dug the Tigris and who begat the bastard?', said of or to an inquisitive person (CI. Ar. nag!) MC/ niSan/, J/nisan/ 'betrothal, engagement'; MJC/ nesan/ 'to betrothe', ftoe-san/ 'to become engaged'. Pees. nisan 'token, sign',
perhaps via / mal)bas niSan/ 'engagement ring', at first 'token ring' MJCjhalhalj'to produce a trilling noise with the tongue', that sound itself being called MJC/ halhula/, pI. M/halahil/, J/ hlahil/, C/halehU/ ; this noise, usually produced on joyful occasions, is the same as that referred to in other dialects by the roots zgrd, zgrt, zgll, and other variants; the root hIhi occurs in other parts of the M~sopota mian area (Mosul , Anatolia) with the same meaning ; for a different meaning in Lebanon, cf. Barthelemy, Diet. MJCjham/ 'also', with variants MJCjhammen/, / hammena/, J/ hamzed/, /hamzedif. Pers. ham, whence also Turk. hem, an old loan into Arabic (Jawaliqi, p. 146, Bariri, pp. 113, 233) .today appa rently restricted to the Mesopotamian area and Central Asia, even more widespread in tiD latter, where it also serves other functions (cf. Fischer, 1961 , p. 261); however, Barthelemy, Diet. , lists it without localization, hence presumably used in Aleppo, but only as /ham ... ham/ 'both .. and' M/hwaya/, i.q. J/keig/, C/ktlgj 'much, many' invariable: / hwaya nasi or Inas ehwayal 'many people', /f1us ehwaya/ ' much money' ; jhwiiya yesta'emluha/ 'they use it a lot' ; occurs in other geletdialects, though to the south (Basra, Kuwayt) /wajed/, / wiiyed/ occur instead. Etymology unclear; Malaika, 1963, p. 35, suggests
SOME LEXICAL FEATURES
159
OAjhawiiya/' mein Wunsch', but cf. Pal. jhwiiyej 'a blow', and Fr. beaucollp M/hiiysa/, JC/hasa/, pI. MJjhwiiyes/ and jh05j (C not noted) ; Kweres has the M form, but my own informant from the same region has / hasaj ; Weissbach, 1930, p. 330, adds for Kweres the more familiar jbgaraj ; 'Ana has [bagra]; a Muslim from Amara has jhaysa/, and for Qal'at Salel), Van Wagoner, 1944, gives jbaqaraj 'head of cattle' (p. 107) and jhiiysal 'cow' (p. III). Barthelemy, Diet., cites the M form for Iraq. In late Classical Arabic hall'a'i! occurs in the sense of 'cattle, large animals' (Dozy) MjCjhi~j 'nothing'; M/saku bidak? hie, kulSi miikul 'what's that in your hand? - nothing, there's nothing'; Jjhiiyi Imaswa hie, segbet rnayj 'this hike is nothing, a child's play (lit. a drink of water),. Pers. hie. Mjwuje(h)/, jwuccj, MJCjwecej 'face'; the last form recurs in the qeltll-dialects, the first in the gelet-dialects; Central Asia has lie; Barthelemy, s. v. IVaee, points out that Jawiiliqi (p. 149) condemned IVII}h for IVa}" as vulgar, and that all modern dialects have forms harking back to an OA form with lui MJCjyezil ' that's enough', rarer than Ibassl also occurs in Kweres (Meissner, 1903b, p. xlix). Meissner's surmise that this comes from OA/yajzi j is supported by form s such as yadzi, yazzi found with the same meaning in some North African dialects (Mar"ais, 1956, p. 112 ; Cantineau, 1960, p. 60, where the same etymology is proposed)
SUMMARY AND CONCLUSION
7
SUMMARY AND CONCLUSION SUMMARY OF FINDINGS. 7. 11 Muslims vs. non-Muslims. (0) T he Muslims, Jews, and Christians of Baghdad (and, so. far as I can tell, of the other cities of Lower Iraq) speak lhree different dialects each fully correlated with community affiliation. The majo r diehoto'my divides the Muslim majority fro m the two minority groups: the majo rity di alect (M) is of the ge/et type, the two minority diale~ts (J and C) are of the qe/tu type (on the terms ge/et and qe/tu see 1.31 and 1.32). Although our knowledge of the Mesopotamian dialect area is still quite li mited, it seems a fair guess that dialects of the qe/tu type are spoken by old sedentary populations o nly (Muslims and non-Muslims in Upper Iraq and Anatolia, only non-Mushms in Lower Iraq), whereas ge/et-dialects are spoken by semi-sedentary and recently sedentarized populatio ns, and by Muslim sedentanes in Lower Iraq ; fully nomadic populations speak dialects closely akin to the 'ge/ef type. Muslims and non-Muslims alike speak dialects that for a ll the differences they exhibit, all belong to the Mesopotamia; area; there is thus no prima facie basis fo r an explanation of these differences by immigration from outside the area. To what extent migrations within the area may be considered will be discussed below. . (b) The features that set apart M from JC are summarized below. The reader is warned that this summarization is schematic and leaves out details, qualifications, and reservations, for which he is referred to the body of the work. . (i) Phonology . In M , OA/II is often N , in JC usually fIf: M/xa!/, JC/xal1 'mother's brother'. In M, OA/ql IS often Ig/, hI JC a lmost always Iq/ : M/gaJ/, JCjqii 11 ' he said'. In M, OA/kl has been affricated to I~I in given instances, whereas JC have preserved Ik/ : M/can/, JC/kan j'he was'. In M, OA/rl is apical Ir/, while in JC it has, in many instances, become a velar spirant Igj: M/ras/, JC/gasl ' head'. 7. 1
160
161
As for the vowel system, OA/ul and Iii are both represented by JCfe/, whereas M has both lei and lui depending on a variety of factors: OA/kull/, M/kull/, JCfke1l1 'all' ; OA/sidq/, M/~u dug/, JC/~edeq/ ; OA/sitt/, MJC/settl 'six'. In given syllabic sequences, OA/al yields M/ul or lei , JC/a/ : OA/qamar/, M/gumar/, JCfqamagl ' moon'; dAtjamal/, M/jemalf, JCfjamal1 'camel'. Final, stressed lal is preserved to a greater degree in JC tha~ in M: J/suda/, Cfsodi/ , M/sodal 'black (f.s.)'. The 'imti/o of OA/a/, which already made itself felt in Abbasid Baghdad, is very common in given cases in JC, totally absent from M; thus M/wiihed/, JCfwehedl 'one'; M/~eta/ , JCf~etil 'winter'. In matters of word stress, there are some diffcrcnccs that are more strictly of a morphophonemic character: MJC/ ~ahebl 'friend', M/sahbil vs. J/sahebi/, Cfsahebil 'my friend'. (ii) Morphology. Subject pron01l1inal suffixes attached to perfect verbal stems differ in phonemic shape and in some olher respects : M/geletl ' I said' and 'you (m.s.) said', JC/qeltul 'I said' vs. J/qelt/, C/qeletl 'you (m .s.) said'; M/galat/, JCfqaletl 'she said' ; M/geltul , JC /qelteml 'you (pl.) said'; M/galaw/, JCfqalul ' they said' . Object pronominal suffixes differ as to the morpJlOphonemics of suffixation (cf. e.g. the last item in the preceding paragraph) and ill phonemic shape: M/abuya/, JCfabuyil 'my father'; M/abiJ./, JC labunul 'his father' ; M/uljlIpa/, JC/emmul 'h is mother' ; M/umha/, JC/emmal ' her mother' ; M/umhum/, JCfemmeml 'their mother' ; The feminine suffix T has, in the isolated word, the allomorphs lal and Iii in JC, to which corresponds a single allomorph lal in M: M/maftuJ:!a/, J/mefluhai , Cfmaftiihal 'open (f.s.), but M/cebira/, JC/kbigil ' big (f.s.)'; the allomorphs of T in sandhi are also different: M/gubti/, JC/qebbetil ' my room' ; M/gu\J\Jatna/, JCfqebbetn al ' our room'. Unit nouns are usually formed by appending the suffix -/ayil in JC, the suffix T in M: MI l;> us!a/, J/bes!iiyi/, Cfba!aJayil 'an onion' . Morphemes preposed to the imperfect are similar in· shape and function in JC and differ from M usage: JCfqanqiil1 'we say', Idanqii ll 'let's say', vs. M/dangiill 'we say' and 'let's say'. (iii) Syntax. Few points were investigated ;' determinate nounplus-adjective phrases more often lack the initial article in JC Uwalad lekbigf) than in M (flwalad eecebir/, ' the big boy'); determinate direct objects are marked by an anticipatory pro nominal suffi J( plus the morpheme L more often in JC than in M: JC/hazzu 19asul 'he shook his head', M/hazz rasa/.
162
COMMUNAL DIALECTS IN BAGHDAD
(iv) Lexicon. Basic vocabulary is almost entirely cognate, but a number of common terms are not; many but not all non-cognates oppose JC to M: JC/gada/ , M/ bacer/ 'tomorrow'; J/kSig/ , C/ktig/, vs. M/hwaya/ 'much' ; this holds for most personal pronouns (JC/ana/, M/ ani/ 'I'; JC/entem/, M/entu/ 'you (pl.)" (and cf. the pron. suff. in (ii) above) and demonstratives (JC/hOni/, M/hna/ 'here') which illustrate instances of "imperfect cognates." 7.12 Jews vs. Christians. Despite the fact that J and C belong together in the same qeffll group and share many features as against M, the two dialects exhibit many differences from each other. On the whole, C is closer to Mosul a nd, to a lesser degree, the Anatolian dialects. The differences between J and C are summarized below. (i) Phonology. The / r/ > /g/ shift does not quite cover the same ground: J/ qaga/, Cfqaral 'to read'; J/jurab/, Cfjogabil 'a sock'. The interdentals have been preserved in J, as in most of the Mesopotamian area, but replaced by the corresponding stops in C, as in some of the Anatolian dialects and in the Syrian urban dialects as a whole. In the vowels, C goes along with Mosul in the retention of most l a/, s in unstressed open syllables, whereas J goes along with 'Ana in their zeroing-out in open syllable and in their shift to /el in closed pre-stress syllable, with M standing somewhere in the middle: J/8qil/, Cftaqilj, Mj8egilj 'heavy'; Jjkelbenj, Cjkalbenj, Mjcalbiinj 'two dogs'. Related to this is the treatment of OAjawj and jayj in unstressed syllable, which are reduced to j uj and j ij in ], while C has joj and j ej: Jjbitenj, Cjbetenj 'two houses'; Jjyumenj, Cfyomenj 'two days'. For OAj u/ and /i/ C has j ol and jej in given environments: Cfsoq/, ] /suq/ 'market', C/male1,J.j, ] j rrili1,J.j 'good'; here again C goes with Mosul. There are many points of difference in the degree and detail. of ' imiifa : in given cases, J has ji j for Cfej: Jjklibj, Cjklebj 'dogs'; in others, ] has /a/ : ] I skakinj , Cfsakekinj 'knives'. In C, long vowels retain their phonemic length in unstressed syllables (as in M and Mosul), whereas] has, SO to speak, no long vowels in unstressed syllables (as 'Ana): Jjdulabj, Cfdfiliib/ 'wardrobe'; Jj1,J.alfibj, C/1,J.alfJbj 'hail'; Jjmizinj, Cjmizen/ 'scale'. Final consonant clusters are, on the whole, maintained in J, separated in C: Jjdagbj, Cfdagebj ' road'; Jjktabt/, Cfkatabetj 'you (m.s.) wrote'. (ii) Morphology. In the subject pronominal suffixes attached to the perfect base, J and C differ iii morphophonemic details: JC jqeltul 'I said', but Jjqeltoluj vs. Cfqeltuluj 'I said to him'; JCfqaluj
SUMMARY AN D CONCLUSION
163
'they said' but JjqalOluj vs. Cjqiiluluj 'they said to him'. Other differences are due to the fact that C treats K ,-y verbs unlike others, while J does not : J/qaguj 'they read' (like Jjkatbu/ ' they wrote'), but Cfqaroj vs. jkatabu/ ; Jjteqqenj 'you (f.s.) read' (like jtketbenj 'you (f.s.). write'), but Cfteqrenj vs. jtektebin/ ; Jjteqqonj 'you (pl.) read' (like jtketbonj 'you (pl.) write'), but Cjteqronj vs. jtektebunj . As for object pronominal suffixes, C has the invariant -jki/ for the 2f.s., as Mosul and Anatolia, whereas J has j k j ~ j kij : Cjabukij 'your (f.s.) father', jemki/ 'ydur (f.s.) mother', Jjabukij, jemmekj. The morl?hophonemics of the suffixes - hii and - hem differ: JCjemmaj 'her mother', ]jabuhaj, Cjabuwaj ' her father'; JCjjabaj 'he brought her' but ]Jjabuhaj (cf. Ijjabonuj 'they brought him') vs. Cfjabuwaj 'they brought her' ; lCfemmemj 'their mother', but Ijabuhemj, . Cfabuwemj 'their father'; JCjjebnahaj 'we brought her', jjebnahem/ 'we brought them'; C again goes along with Mosul. Though both 1 and C have jaj and j i j allomorphs of the fem. suff., their distribution follows entirely different principles: ]j1)elwaj, Cf1)elwij 'pretty (f.s.)'; l/ l;>a~ilij , Cjl;>a~ilaj 'trifling (f.s.)'. In Form I verbs, Chas preserved the two conjugations qalaf vs. qatif, like Mosul and Anatolia whereas J has a single qalaf conjugation, like 'Ana (and like M, which has a single qelaf where e is j ej ~ j uj depending on the flanking consonants) : Cfkatabj 'he wrote', jlebesj 'he wore', Ijkatabj, jlabasj. The imperfects of Forms VII and VIII show differences beyond the regular phonemic ones and place C somewhat closer to M : Ijaftahemj, Cfaftehemj 'I understand' ; Jjanhazemj, Cjanhezemj 'I flee'. The imperative of the verb 'to come' is peculiar to J, C going along with most Eastern dialects. The morpheme L 'to' has the presuffixal allomorphs Cjelj- (as in M) vs. I j llej- ; double pronominal objects have a special form in J, a more common one in C: Ijjabelyakj 'he brought it (him, her, them) to you', Cjjablak yfmuj 'he brought him to you'. Preverbal morphemes differ : l /qadaqulj, Cjqaaqulj 'I say'; Ijssa'aqulj Cfga1,J.aqulj 'I will say', and especially Cjkenqeltuj 'I said', as in Mosul and Anatolia, with no parallel in J. (iii) Syntax. The optional use of a copula in C sets it apart from J (and ' M) and links it with the Anatolian area. The absence of the article in the first term of jmalj constructions is peculiar to C. The marking of the determinate direct object by an anticipatory pronoun plus L is less common than in J. (iv) Lexicon. The Hebraic element in J sets it apart from
164
C, M, and other dialects., In the free personal pronouns and demonstratives, there are a few differences between J and C (e.g. J/ hekeo/, Cfhekkil 'thus') but in interrogative particles C is closer to M than J: C/S/, Isenul 'what', M/as/, liISkunl 'what'; a number of non-cognate nouns and verbs set C together with M apart from J: MCfxaseml 'nose', J/enf/; MC/sMI 'to see', J!'ayan/, etc. 7.13 Christians vs. Mosul. (a) The dialects of Mosui" and of tbe Mosul area are too imperfectly known to permit a categorical answer to the question as to wbether C is purely and simply a dialect of immigrants from that area. Similarities with Mosul are many, but
tbere are enough differences to consider C a related but separate dialect. Inasmuch.as data point to fairly recent and fairly considerable immigration of Christians from the north to Baghdad, and because of some suggestive similarities between C and tbe Anatolian dialects we may surmise until all the evidence is in that C has at least bee~ influenced by dialects from the northern regions. (b) One set of differences between C and Mosul is of a special nature: they consist of pairs in which Mosul has unusual features peculiar to itself whereas C has, as its equivalent, an exact or approx-
imate replica of ,the M form. These arouse the suspicion that peculiar Mosul-like features bave been suppressed in favor of something more acceptable. The suspicion is reinforced by the fact that tbese are essentially lexical items, in which this sort of substitution is
notoriously easier than in phonology or grammar ; in the following, C has the same item as M:
'four'
Mosul
C
oba'a
arba'a
(and the related terms 'fourteen', 'forty', 'Wednesday') 'he' 'they'
hlnu hiyem
'to see'
qese'
'what'
as, assun
'where'
e!ab
huwa humma sM 5, senu wen
In the following cases, C has a form similar to that of M, Mosul an unusual or special form:
165
SUMMARY AND CONCLUSION
COMMUNAL DIALECTS IN BAGHDAD
'three'
'when' 'twenty: ~
'sieve' \1 ' ,j
SaSi emati 'essin
giibel
I·,
tlati swaqet 'esgin garbel
However, some differences between MosJI and C do not . C'.nearer to M; tbey are either peculiar to C or common to C brmg and sOl1]e Anatolian dialects; tbey are, summarized below. (i) .Phonology. The interdentals are preserved in Mosul, replaced by stops in C. Tbe 'imiila of Mosul is different in several respects from that of C: Mosul Isakikin/, C/sakekinl ' knives', Mosul Ibestenci/, C/bestiin6i! 'gardener'. (ii) Morphology. Pronominal affixes are remarkably similar in shape and behavior, the sole difference noted being the double object suffixes : C/la'etlu yanul 'you gave it (m.s .) to him', Mosul/la'etlilwa/. The allomorpbs IiI and lal o( the fern. suff. do not have quite the same distribution: Mosul Ikabiga/, C/kbigif. The vocative suffix leI of Mosul is absent from C: Mosul !,amme/, Ixalel (terms for addres, sing paternal and maternal aunt). (iii) Syntax. Mosul has no copula like tbat of C , nor is the C article dropped in Imatl constructions. (iv) Lexicon. A number of lexical differences, have been mentioned in (b) above. Others include: Mosul/hnilkal and Ihonekl 'there', C/boniki/; Mosul Ihakeol 'thus', C/hekki/; Mosul/mbor/ , MJC!'alamildl 'because of, for tbe sake of' ; Mosul/gaqqabi/, C/geqbil 'neck'; the Mosul particle Ize/, roughly in the function of MJC/asul and JC/baqa/, is absent from C. Mosul/ta'iil zel 'come here now', C/ta'al baqaf. 7.14 Muslim vs. tbe countryside. '(a) While M is closely akin to ,the urban and rural gelel-dialects of Lower Iraq, it differs ' from them 'in 'many respects. It is closest to 'the urban dialects on Wi,ich some data are available (Basra, Qal'at Salel» so tbat one dimly fore- ' sees a possible classification of urban vs. rural ' gelel-dialects, as yet not solidly established. (b) It appears useful to summarize the differences separating M from the rural gelelCdialects tbat are closest to it: for this purpose we shall make use of the detailed data available on Kweres and tbose from my own informant from a nearby point in the same Musayyab district; this combination of data shall be designated by the symbol R. (c)
COMMUNAL DIALECTS IN BAGHDAD
SUMMARY AND CONCLUSION
Phonology. OA/};./ is more often affricated in R: M/akel/, 'food' . There is greater retention of OA Iql in M, and relatively few instances of a Ijl reflex, whereas R has regular affrication to Iii in some positions, coupled with morphophonemic alternation with Ig/: M/ge'adl 'he sat', /ga'edl 'sitting', R/ge'ad/, Ija'ed/. A /pi phoneme is well established in M, not so in R. In R there is the typical Beduin syllabic reshuffling and alternation absent from M: M/gahwa/, R/ghawal 'coffee'; M/ge~aQ/, 'reeds', Igu~Qal 'a reed', R/gu~aQ/ , Ig~uQaf. The distribution of Ie/ and lui differs in M and R : M/kull/, R/ kell/, 'all'; M/ge~aQ/, R/gu~al:i/ 'reeds' . (ii) Morphology. In subject and object pronominal affixes, R has a masculine-feminine distinction in the 2nd and 3rd person plural which is only marginal in M. Addition of suffixes to certain nominal bases yields completely different morphophonemics in R than in M because of the syllable reshuffling mentioned in (i) above. Similar differences occur in Form I conjugations: M/ketbatl 'she wrote', R/ktebat/; M/,erfawl 'they knew', R/,rufaw/. Form II perfects show vocalic alternations in R, none in M: M/sallam/'he greeted', l~addagJ 'he believed', /xallaf/'he begat', Rlsallem/,/~addagJ, Ixalluf/. The verbal systems' exhibit many other differences of detail. (iii) Lexicon. A check through Meissner, 1903b, Glossar, reveals little lexical differentiation between M and R; where M differs from J and/or .C, R has the same or very nearly the same forms as M; instances of slight deviations can be seen in R/~enhul 'what', M/~enu/; R/ menhu/ 'who', M/menu/. . ' 7.15 Uniqueness of J. Within the qellLl group, there IS no dialect so far investigated that bears a particularly close relationship to J. Some salient features ('imlila, II!.I < It/) link it with C and Mosul, others ('imlila, some lexical items) with Anatolia, and two features (shortening of unstressed long vowel, treatment of OA/a/) are shared witn 'Ana. All these in addition to features shared by all or most qeIILl-dialects. The Jews of the other cities of Lower Iraq speak dialects that are virtually indistinguishable from J. 7.2 . CHARACTERIZATION OF THE DIALECTS. 7.21 Conservatism in JC. (a) If we combine the findings summarized above with comparative data from other dialects and with whatever is known of medieval Iraqi vernaculars, we conclude that J and C are direct descendants of dialects spoken by the urban population (both Muslim and nonMuslim) of Abbasid Iraq. More precisely, J and C, no less than the
other qeIILl-dialects, have preserved or continued several basic phonological and morphological features of the older vernaculars,. It is problematic whether J and C actually' continue older dialects of Baghdad' itself or whether their characteristic features have been. imported f,om the north, the present home of most of the qellll-dialects. With respect to C, the likelihood of northern influences is reinforced by the immigration of Christians from the north.l 77 With respect to J, on the other hand, I have no linguistic or historical evidence suggesting immigration fro,m the north or elsewhere; if, as seems possible, the Jewish community in Baghdad has had a continuous exis-
166
(i)
R/a~ell
.. , '1
167
tence in or around that city from its foundation to the present,178
features that are at present peculiar to J may hark back to the vernacular of medieval Baghdad. (b) The Jews and Christians of Iraq spoke Aramaic until they were completely Arabized,l79 yet most of the main features difl'erentiating JC from M are not related to the Aramaic substratum. We may assume that, by' the time Arabizalion was completed, there were some minor communal differences in speech, such as the use of some Hebraic elements in J, but I find no evidence of early major differentiation. I so It is as yet impossible to say with certainty when and how the present major differentiation arose; there is, however, linguistic evidence pointing to Beduinization in M as a crucial factor in that differentiation, and some historical evidence suggesting rough answers as to "when" and "how." These points are taken up in the
following two sections. 7.22 Beduinization in M. (a) While M shares with J and C a number of features, especially lexical, that evidently continue older urban Iraqi forms, its phonology and. morphology are permeated with elements similar to and, in all likelihood, imported from, the ' rural dialects of Lower iraq. Together with other urban gelel-dialects, however it differs from the rural dialects in a few crucial respects, and in ;hese very respects it resembles the dialects of sedentaries on the other side of the Syrian desert (Transjordan, !:loran). in general, the gelet-diale~ts and their congeners are,- outside Arabia,
spoken by Beduin and Beduinized populations. These include fully nomadic camel herders, semi-nomadic sheep and goat herders,
recently sedentarized nomads and semi-nomads, and groups in various intermediate stages of sedentarization. Some dialect characteristics correlate fairly well with the various stages of nomadism and seden-
168
SUMMARY AND CONCLUSION
COMMUNAL DIALECTS IN BAGHDAD 1
tarization. In the Syrian .desert and its fringes, which include the Mesopotamian area, some features (e. g. affrication of OA/ql to Idzl and OA/kl to Its/) belong to full nomads, others (e. g. affrication of OA/ql to iii and OA/kl to lei) to semi-nomads and the Lower Iraq village population, whose sedentarization is of recent vintage, or who have at least been strongly influenced by recently sedentarized and semi-sedentary groups;!81 and others (e.g. non-affrication of OA/q/ > Ig/, less frequent affrication of OA/kl to lei) are shared by groups that arc fully sedentary, but that have, in centuries past, absorbed non-sedentary influences. It is to the last that the present Muslim dialect of Baghdad belongs .. (b) It must be pointed out that there is an alternative possibility: the gelet influences in M might be due to sedentary immigrants from Arabia . Dialects of the gelet type are Beduin or Beduinized dialects only outside of Arabia; in Northern and Eastern Arabia, rather similar dialects are spoken by sedentaries as well. Such townsmen could have migrated to the Lower Iraqi towns and brought their dialects with them. 182 For Baghdad, however, immigration from the
169
monograph. Still, a cursqry glance at the available sources!84 tends to corroborate some of the hypotheses outlined above, and may suggest questions and directions for their further elucidation. I find very little on the history and composition ,of the Baghdad popUlation itself, but from the history of Lower Iraq as a whole, it does seem permissible to draw two conclusions: (I) that Lower Iraq was, from the thirteeflth or fourteenth century onwards, subjected to a pro.cess of de-urbanization and re-tribalization which has only recently been reversed, ~nd has left deep marks on its population; (2) that the population of both villages and towns has continually (but especially since the downfall of the Abbasid state) been replenished by waves of Beduin ·immigration from Northern Arabia via ·the Syrian desert. These processes can be outlined as follows: (i) From the seventh century, when Arabization bega n as a result of the Muslim conquest, until the tenth century, when centralized government began to decline, Arabization and urbanization went '
hand in hand. Cities were founded, became populated, and prospered. The non-Muslims adopted the Arabic speech of the Muslim
surrounding countryside seems both linguistically and historically more likely and the fact that Muslim townsmen, in given times and
townsmen. ~
circumstances, should absorb Beduin or semi~Beduin elements from the countryside is neither surprising nor unprecedente?183 A similar
(ii) From the eleventh to the thirteenth centuries, centralized power was on the decline. The ruling groups were increasingly dominated by ' Persian and Turkic speakers. Urban centers were still large and populous, but Beduin nomads and semi-nomads increased
process is going on in present-day Baghdad: a part of the population is readily identifiable as of rural origin, both by its speech and by its personal history, and though largely belonging to the lower classes, some influences from their dialects have been absorbed into M. The assumption is that something similar occurred in previous centuries,
at a time when the social and linguistic prestige of the Muslim population of Baghdad , and hen~e its resistance to rural infiuences~ was at its lowest. (e) While most features of M are shared by all M speakers, some show greater Beduinization than others (greater use of Ij/, for OA/q/, retention of the fem.-masc . distinction in the plural of verbs and pronouns, etc.) . There is also some evidence, as yet fragmentary,
that some older urban features, now found in JC only, may have bee n preserved in some remote corners of the M community and , in some spec ial usages . ISJa . 7.23 A glance at Iraqi hiStory. (a) A detailed inquiry into Iraq!
history, wllile obviously desirable, is beyond the scope of the present
their raids and migrations into the cultivated area. 185
(iii) \ The decline of the caliphate and of urban life was brought to a clil)lax by the Mongol sack of Baghdad in 1258, by the subsequent devastation and anarchy, and by ·the second sack of Bagqdad in 1400. The irrigation system, and the agricultural life dependent on it, were',ruined, 1 86 and the effects of that ruin are. felt to this day.
Baghdad was greatly reduced in population, the countryside subjected to further Beduin inroads, and Iraq became "a country of few and small towns .. . while around and between them lay tracts grazed and dominated by the tribes alone."!87 Until well into the eighteenth century, Iraq was a zone of constant strife between alternating
Turkic-speaking and Persian-speaking conquerors. (iv) The establishment of Ottoman control in the seventeenth to eighteenth centuries marked the beginning of a gradual and slow
170
COMMUNAL DIALECTS IN BAGHDAD
return of centralized authority, though that authority now had to contend with solidly established Beduin tribes and tribal confederations. Many of these were sedentarized in Lower Iraq, but still tribally organized l88 In the eighteenth and nineteenth centuries, new towns were built and old ones gradually repopulated. 180 The process of re-urbanization and de-tribalization has, with various tribulations and backslidings, been going on ever since, and has probably been accelerated in the present period of independence. (b) It seems a fair inference that these populations shifts are intimately connected with the dialectal development outlined in 7.21 and 7.22. In the fourteenth century, the Baghdad Muslims were still speaking a qeJtu type dialect and were, presumably, undifferentiated from the non-Muslims, at least not in the way they are today. Some time after the re-establishment of a relatively stable regime under the Ottomans, i.e. from the seventeenth century onwards, Baghdad underwent a repopulation process which must have drawn heavily on the village and tribal population; other towns underwent the same
., "
SUMMARY AND CONCLUSION
and do not seem to have been subjected to the same depopulation and repopulation seesaw as those of Lower Iraq. For another, the topography of Upper Iraq seems more conducive to polarization between sedentary and non-sedentary life.' The steppe is suitable for grazing only, the narrow banks of the river courses (which are here fixed, not variable as in Lower Iraq) are suited for permanent se-
dentarism. In contrast,lOO the lowlands of the center and south, for centuries deprived of flood control and regulated irrigation, more easily invite a type of semi-sedentary population which, according to our conjecture, served as a reservoir of Muslims for the repopulation of its towns. 191
process, some were entirely Dew creations. This new urban popula-
tion was Muslim and presumably spoke a gelet type dialect, indeed the ancestor of present-day Muslim Baghdadi. The non-Muslims obvio usly could not come from the tribal population ; the Christians came, at least in part, from the north (see note 177) and the Jews seem to have received no massive accretion from outside Baghdad ·(see note 178). That the Jews and Christians did not adopt the speech of the new Muslim population is probably due to two factors: on the one hand, the minorities were socially isolated, and on the other
I
hand, Muslim speech was neither the language of the governing circles, nor that of their social or cultural elite. Since 1918 social and economic power has been returning to the Baghdadi Muslims, and with the growth of the new Arab culture, the majority dialect is acq uiring prestige (cf. 1.42). (c) What has been said of Baghdad seems to hold for the other cities of Lower Iraq, at least grosso modo. We have seen that matters are different in Upper Iraq: there 'is no qeltu-gelet split correlating with community affiliation, and urban dialects are all of the qeltu type. Why this should be so is a m~tter .for separate investigation, but several factors come to mind. For one thing, the small towns of Upper Iraq seem to have had a more continuous urban population
171
\.
.>
REFERENCES C ITED ABBREVIATIONS USED IN REFERENCES AND. NOTES
American Anthropologist. Menasha, Wis. de I'Instifut d'Etudes Orientales de 10 FaclIite de Lettres de I'Universile d'Alger Acla Orientalia Hungaricae, Budapest AOH ArOr Archiv Oriel/lalni, Prague Bei/riige zur Assyr.i%gie lind Semi/ischer Sprac}J)vissenscha/t, Leipzig BASS Bulletin de ['Illstitlll Fra,,~ajs d'Archeologie Orientale du Caire BlFAO Bul/etill de fa Societe de Lil/guislique de Paris BSL Bulletin of the School of Oriental [alld A/rican] Studies, London BSOS Ene. of lsi. Encyclopaedia of Islam , lst ed. 1913/38; 2nd ed. 1954- , Leiden International Journal of America" Linguistics lJAL JA · Journal Asiatique, Paris Journal oftlte American Oriental Society. New Haven JAOS Journal 0/ Jewish Studies, Manchester JJS Journal of Semitic Swdies , London JSS Lugat al-'Arab, Baghdad LA language, Journal of the Linguistic Society of America, Baltimore Lg. Middle East Journal, Washington MEl Mitteilungen des Semillars fiir Orielltalische StIldiell zu Berli" MSOS Neuphilologische Mitteilungen, Helsinki NM Revue Africaille, Algiers RA R evue des Etudes Jllives, Paris REl Studies ill Lillguistics, Buffalo SlL SOY. Vost. SOlleckoe Vostokovedellie, Moscow-Leningrad Word, Journal of the Linguistic Circle of New York Wd. Wiener Zeitschnjt fur die Kunde des Morgelllalldes ' WZKM Zeitschrift der Delltsc"e,~ Morgellliindischetl Gesellschaft, Leipzig ZDMG (since 1950, Wiesbaden) Zeilschnjt fur Semitistik, Leipzig ZS
AA AlEO
I Atmales
172
,
Works are cited by author's surname and year of publication. In cases of several works by the same author in the same year the letter "a" after the year indicates the first work listed here, the letter "b" the second, etc. A few' incidental references are cited in full in the appropriate notes and a re not listed here. Abrahamian, R" Dialects des Israelites de HamadQlI et dt/spahal/ et dia/ecte de Baba Tahir, Paris, 1936. al-CaJabi al-Maw~i li , D" al-'AOar al-'iiriimiya Ii fugat al-Maw$il al-'iimmiya , Mosul , 1935. - - -, Kalima, fc"irislya Inusta'mala Ii 'iimmiyat al-Maw#1 wa-li 'al/~l(i' al-'Iraq, Baghdad, 1960. al-Dabbag al-HuOaIi, A. X. , Mu'jam 'am()iil al-Maw$il al-'Ammlya, 2 vots" Mosul, 1956. • a i-Banaei, J. al-AmOiil af-Bagdiidiya (Pt. I) Baghdad, 1962. - , MIl'ja", al-Jllga al-'iimmiya al-Bagdiidiya (vol. I, letter 'nit/) Ba'ghdad , 1963. al-J:lariri , 'Abu Mul)ammad aI-Qasini b. 'Ali , Kitiib Durrat al-Gall)\V{"i$ /i ' Awhiim aJXawii$$ , fl'a-fl 'axirihi al-sarl,r Ji-'A~lIl1ad Sihiib aI-Diu al-XafaF, Istanbul , 1299 A.H. al-I:JiIli, see Hoenerbach. al-Ja\:li" 'AbO Veman 'Amr b, Ba\:lr, Kitiib al-Bayliu wa-I-Tabyil/ , ed. Han]n, vols. I and II, Cairo, 1948, 1949. al-Jawaliqi, see Derenbourg. al-KarmiJI, A.M. , "al-Lugat wa-I-Iaeagat," al~Masriq 6: 529- 36; 589-93 (1903). al-Ru~ari, M., "Oaf' al-Muraq fI Lahjat 'Ahl al-'Craq" (t it les vary wi th each installment, under general heading of "Lugat 'Awammal-'l raq"), LA 4:84- 88, 140-46, 211 - 14, 333- 35 (1926); 403- 05, 460- 64, 52i- 25 , 596- 99 (1927); 5: 94-96,147- 50,347- 50,541-43 (1927); 6,: 203- 07, 521 - 24, 683- 88 (1928). al-Taliqani, 'Abu al-l;Iasan 'Ali b, al-Fac.tl al-Mu'ayyadi. Risiilar al-'AmOiilaJBagdiidiya 'allati tajr; baYIl al-' Amma 1i-I-QiirJi... ,jam~ ·alll"ifj Sallat 421 H, ed. Massignon, Cairo, n.d-, (1914 ?). AI-Toma , S. J" Tlte Teachil/g of Classical Arabic 10 Speakers of the Colloquial ill Iraq: A Study of the Problem of Linguistic Duality and its Impact all Language £dllea/ioll , Unpublished Ed. D. dissertation, Harvard University , Cambridge, Mass. , 1957. al-Xaf3.ji, see al-l:lariri. Anastase , see al-Karmili.
173
174
COMMUNAL DIALECTS IN BAGHDAD
Arabian Ame>rican Oil Company. Pocket Guide 0/ Arabic, Dhahran , 1955 (also various other undated books : e.g. Spokell Arabic, Basic Arabic, Conversational Arabic, etc.).
Barthelemy, A., "Observations critiques sur un article du P. Pourriere intitule 'Etude sulr Ie langage vulgaire d'Alep',': JA 10.6: 170- 89 (1905). - - - , "Notice su r l'ouvrage intitule 'Der vulgararabische Dialekt von Jerusalem' etc. ," JA 10.8: 197- 258 (1906). - - -, Dictiollnaire arabe-!ranfois, 5 fase., Paris, 1935- 56. Bauer, L. , Dos paiaslinische Arabisch, Leipzig, 1913. - - -, Worterbuch des paiiislinischell Arabisch: Deutsch-Arahisch, LeipzigJerusalem, 1933. Binder, F., " Zur Aussprache dec k und q bei den Beduinen Arabiens," WZKM 46: 209-18 (1939).
Blanc, H.o Swdies ill North Palestinian Arabic: Linguistic J,Iqlliries omolrg the Druzes of Western Galilee and MI . Carmel, Jerusalem, 1953. - - - , " Iraqi Ara bic Studies," H. Sobelman ed. Arabic Dia/ecl Studies (Washington , 1962) pp. 48-57. - - - , "Stylistic Variations in Spoken Arabic: A Sample of lnterdialectal Educated Conversation," Contributions to Arabic Linguistics , cd. C. A. Ferguson, Cambridge, Mass., 1960, pp. 81- 159. Blau, J., "Die arabischen Dialekte der Juden des MiUelalters im Spiegel der jtidisch-arabischen Texte," Orbis 7: 159- 67 (1958). - - , The Stilt us of Arabic as used by Jews in the Middle Ages: Do Jewish Middle Arabic Texts Reflect a Different Language?" JJS 10: 15- 223 (1959). - - -, " The Importance of Middle Arabic Dialects for the History of Arabic," Studie s ;1I Islamic History alld Civilizatioll, Scripta Hierosolymitalla 9 : 206-28 «(961). - - -, Dikduk ha-'aravit ha-yehildit sel yemey ha-beYIlayim, Jerusalem, 1961. Bloch, I., "Castes ct dialects en tamoul/' Memo ires de la Societe de LiIlK.uistique 16: 1- 30 (1909). Blum. S., Qissah l1Iusa : Eill Beitrag zum bagJu!adisehen Dialekt des Neuarabischell, Hanover, 1927. Bravmann, M., Materialien ulld Untersuchungen zu den phonetischen Lehren der A:raber, GoUingen, 1934. Brockelmann, c., Grwldriss der 'IIergleiehellden GrammaJik der semitisc/ten Sprachell, vols. I and II, Berlin , 1908. 1913. Brunot, L., "Notes sur parler arabe des JU,i fs de Fes," Hesperis 22: 1-32 (1936).
Ie
- - -Introduction d {'arabe marocain, Paris, 1950.
REFERENCES CITED
175
Cantineau, I. ' "Etudes sur quelques parJers de nomades arabes d'Orient ," AIEO 2 : 1- 118 (1936),3: 119- 237 (1937). - - , " Les parlers arabes du departement d'AIger," RA 81: 703-11 (1937). - - - , "Une alternance quantitative dans des pronoms suffixes semitiques," BSL 38: 148- 64 (1937). - - - , "Le parler des Dnlz de la montagne ~odi.na ise, " AIEO 4: 157-84 (1938). - - -,. "Reinarques sur les parJers de sedentaires 1 syro-libano-l;mlestiniens," BSL 40: 80-88 (1939). I . - -, - - -, _ _, --,
"Les parlers arabes du departement d'Oran," RA 84: 220-31 (1940). Les Parlers arabes du lIoron , Paris, 1946. (review of Van Wagoner, 1949) BSL 49: 148-50 (1953). "La dialectologie arabe," Orbis 4: 149-60 (1955).
- - - , "Nqtes sur les parlers arabes des oasis syriennes Qariten, Pahnyre, Soukhdt, "Stud; Orielltalistiei ill Ollore di Giorgio Levi della Vida 1: 120-31 Rome, 1956. - - -, Etudes de linguistique arabe, Paris, 1960. Carbou, H. , Methode pratique pour [,etude de l'arabe parU au Ouadayet d l'est dll Tchad, Paris, 1913. Chiha, H. K., La province de Baghdad, Cairo, 1908. Chowdhury, M., "The Language Problem in East Pakistan," . see FergusonGumperi, cds. , (1960), pp. 64-78. I Cohen, D. "Koine, langue commune et dialectes arabes", Arabica, 11: 119-144 (1962). Cohen, H. , see Kohen, H. Cohen, M., Le parler arabe des juifs d'Alger, Paris , 1912. - - - , Pour une sociologle du langage, Paris, 1956. Coke, R., &ghdad, City 0/ Peace, London, 1927. Cole, J. T., Vocabulary of the Mesopotamian Dialect of Arabic, with Grammatical Notes and Easy Phrases, Lahore,1917. Colin, G. S., " The Arabic Dialects in Morocco," ~Ilc. 0/ Is.! 6: 334-40 (1936). CaIabi, see al-Calabi.
Dabbag, see a l-Dabbag. Derenbourg, H. , ed., "Le livre des locutions vicieuses de Djawaliki," Morgellliindische Forsehungen, Festschrift/iir H. R . Fleischer, Leipzig, 1875, pp. 107-66. Documentation Franc;aise, "Aperc;u sur l'evoluti0!l politiqu~ de l'[raq ," Notes et Etudes Documentaires (5 juillet 195 1), pp.I-40. Dowson, E., All Inquiry inlo Land Tenure and R elated Questions, Letchworth , 1932. Dozy, R., Supp/emellt aux dietionnaires arabes, 2nd ed., 2 vols., Leiden-Paris, 1927. Duri, A. A. , "Baghdad," Ene. of /s. ,2 I: 894-908 (1960).
Brunot, L. and E. Maika, Textes judeo-arabes de Fes , Rabat, 1939. - - - , Glossaire judeo-arabe de Fes, Rabat, 1940.
Epstein , 1. N. , Der Gaonische Kommelliar zur Ordllung Tohoroth: Eilte kritische
176
•
COMMUNAL DIALECTS IN BAGHDAD
Einleitullg Zil dem R. l-!ay Gaon zugeschriebenen Kommenlor, Berlin, 1915.
Farman,
Garbell, I. , "Remarks on the Historical Phonology of an East Mediterranean Arabic Dialect," Wd. 14: 14: 303- 37 (1958) . . Gibb, H. A. R., " The Future for Arab Unity" , The Near East - Problems and Prospects, C hicago, 1942, pp. 67- 99. (Joitein, (S. D.) F., "lemenische Gesch icnten," ZS 8: 162- 81 (1932), 9: 19-43 (1933). - - - , Jemellica --:- Spriclnvorter ulld R edensarten aus Celliral-Jemen, Leipzig, 1934. . Goldziher, 1. , "Linguistisches aus der Literatur der mohammedanischen Mystik" ZDMG 26: 755ff. (1872). - - , "Melanges judeo-arabes," REJ 43: 1- 14 (1901). Grar, G., Der Sprac~lgebralich d~r iiltesten christlich-arabischell Literat,lf: Ein Beitrag zlir Geschichte des Vulgiirarabischen, Leipzig, 1905. Gr iinert, M., Ober die IlIlala. Wien . 1875 . Gumpertz, .Y. G. F., Mivla'ey Sfaten" , me~lkarjm fOlletiyi", historiyim, Jerusalem, 57 13, (1952/53).
,
,
REFERENCES OITED
177
Gumperz, J. J. , " Dialect Differences and Social Stratification in a North Indian Village', " AA 36: 668- 82 ( 1958). Gamma, Y.' R ., "al-'Am9al al-'ammiya fi al-bilAd al-'Iraqiya," al-MaSriq 9:297302 (1906). - - -, "a l-'Al fa~ al-'anlmiya fi al-luga al-'arilmiya al-'j-r aqiya," LA 4: 265- 74, 339-42',406- 10 , 465- 70,53 1-32,584- 88 (1926).
,,
f..,
Harris, G. et al., Iraq : Ir s People, its Society , 'its Clliture. New Haven. 195.8. Hava, J. G., Arabic-English Dictionary for the Use of Students, Beirut, 1915. Hertzler. l: 0., "Toward a Sociology of Language." Social Forces 32: 109- 19 (1953). Hocnerbacli , W .• Die vulgiirarabische Poetik: al-Kitiib al-'Alil al-f:/iili wa-lMlIra.llba$ ai-gail des Saf;yaddill f:/illl; kritiscll herausgegebell lI11d erkliirl VOll •••• Wiesbaden. 1956. Hubbell , A., The Pronunciation of English ill New York City. New York, 1950. l:Ianafi, se'e al-l;Ianafi. 1;Iariri, see al-l;Iariri. 1;IilH. see Hoenerbach. Ibn XaldOn, 'Abd al-Ra.omnn, al-Muqaddima . Beirut, 1886 . . Ivic P.• "Ober den spezifischen Charakter der mundartlichen Ausgliederung des serbokroatischen Sp,rachgebiets," Orbis 7: 134-140 (1958).
,
Jacobsen , T. and R. 1:1. Adams. "Salt and Silt in Ancient Mesopota'm ian Agriculture," Science 128: 1251-58 (1958). Ja\:ti?. see al-Ja.oi? Jawad, M., "al-Luga al-'ammiya al-' iniqiya," LA 8: 11 5- 17,119- 20,610-12 (1930). Jawaliqi, see Derenbourg. Jeannier, A., "Lettre de M. Jeannier, chancelier du consulat de France a. Baghdad, M. Barbier de Menard," JA 8.1 2: 331-48 (1888).
a
Karmili, see al-Karmili. Kelly, T. J .• Notes 011 Colloquial Arabic 0/ Lower Mesopotamia, Bombay, 1917. Khalafallah, A. A., A Descriptive Grammar of Sg'idi ColloqUial Egyptiall Arabic. unpublished Ph. D. dissertation. Austin , Tex., 1961. Kofter, H., "Reste altarabischer Dialekte," WZKM 47: 60- 130, 232-62 (1940), 48: 52- 88, 247- 74 (1945), '49: 15- 30,234-56 (1946). Kohen. H .... 'Uxlusiya ve-l:Ievra bi-m\:toz Bagdad.... Ha-MizralJ. he-l;/adas 8: 16-24, English summary, pp. ii-iii (1953/54). - - -, Hapo'al ba-dyalekt lIa-ye/lUdi ba-'ir 'Am(lra, unpublished Hebrew University paper, n.d. (19527). Kuwait Oil Company, Ltd.• Handbook of Kuwaiti Arabic, Hertford, Eng., 1952.
i
178
COMMUNAL DIALECTS IN BAGHDAD
Lew is, B., The Arabs in History.-London , 195 0. Longrigg, S. H. , Four Centuries of Modern Iraq, Oxford, 1925. ___ , Iraq, 1900- 1950: A Political, Social and Economic History, Oxford, 1953.
I
Malaika, N., Grundziige der Grammatik des arabischen Dialekles von Bagdad, Wiesbaden, 1963. Malick, A. P., " A Comparative Study of American English and Jtaqi Arabic Consonant Clusters," LAngllage Learning 7: 65-87 (1956/57). Mansour, J., "Hagiyat ha-res be-fi yehudey Bagdad," LeIonenu20: 47-49 (1955/56). _ __ , " The Arabic Dia lect of the Jews of Baghdad and the Pronunc iat~on of Hebrew," JJS 8: 187-98 (1957). Man;a is, Ph. , Le parler arabe de Djidjelli (Nord Co" s fa"linoi .... Algerie) , Paris, ( 1956). _ __ , " Les parters arabes," Initiatioll d I'Afgerie , pp. 215- 237, Paris, 1957. ___ , Tf;xtes arabes de Djidjelli, Paris, 1954. Man;ais, W ., Le diaiecte arabe parU d Tlemcell, Paris 1902. Martyais , W. and A. Guiga, Textes arabes de Takro/ma, Paris, 1925. Mass ignon, L., "Notes sur Ie dialecte arabe de Bagdad ," BIFAO 11: 1- 24 (19 14). McCormack , W., "Social Dialects in Dharwar Kannada," Ferguson-Gumperz, ods. (1960), pp. 79-91. McD avid , R., " Post-vocalic "r" in South Carolina: A Social Analysis," American Speech 23: 194-203 (1 948). _ __ , " Some Social Differences in Pronunciation," Language Leaming4 : 102- 16 (1952/53). Meissner, B., "Neuarabische Sprichworter und Ratsel aus dem Irak," MSOS 4: 137-74 (190 1) . _ __ , "Neuarabische Gedichte aus dem lrak ," MSOS 5:2, 77- 131 (1902), 6:2,57- 125 (1903), 7:2, I- II ( 1904). ___ , "Neuarabische Geschichten aus dem lrak,'.' BASS 5.1 :i-lviii, 1- 148 (Leipzig, 1903). Mieses , M. , Die Entstelllmgsursache der judischell Dialekte, Vienna, 1915. Millon, G., "Les parlers de la region d'Alger," RA 81: 345- 51 (1937) Morag, S. , "Seva' kfulot bgd kprt ," SeIer N. I f.. TI" Sinai , (Jerusalem, 1959/60), pp. 207- 42. Mus il , A., The Middle Euphrates, New York , 1927.
Obermeyer, J., Die Lalldschafl baby/olliell im Zeilalter des To/muds lind des Gaonals: Geographie IIIId Geschicille nacll talmlldischen , arabischellulld allderen Quellell, Frankfurt-am-Main, 1929. Oppenheim, M. von , Vom Mittelmeer ZlIm Persischen Golf, vols. I and n, Berlin,
1899, 1900. ___ , Die Bedllillen. Bd. I , Die BedlljllellSlamme ill Mesopolamien lind Syrien
REfERENCES CITED
179
(Ullfer M itarbeitlmg von Erich BrUlllich IIlld Werner Caskell), Leipzig, 1939. - - -, Die Bed"inell. Bd. Ill, Teil 1 (bearbeitel und herausgegeben von Werner Caskell), "Iraq", pp. 165-486, Wiesbaden, 1952. Oppert, J. Exp~dition scielltijique ell Mesopotamie , Pa ris , 1863. Oussani , G., "The Arabic Dialect of Baghdad," JAOS 22: 67- 114 (1901).
., Petermann , J . if., Reise im Orient, 2 vols.,
Leipzig, 1820. Pickford, G . ~R ., " American Linguistic Geography: A Sociological Appraisal;" Wd. 12 ,211- 33 (1.956). Pourri ere, L.~ :'Etude sur Ie langage vulgaire d'Alep," 'MSOS 4: 202- 28 (1901). Putnam, o. ~. and E. M. O' Hern , The Status S igllificance of all Iso laled Urban Dialect, Language Dissertation no. 53 (supplement to Lg. 33.4, 1955), Rabin , Ch. , Ancient West Arabian, London , 195 1. Ramaswamy-Aiyar, L. V. , ' 'Tulu Prose Texts in Two Dialects," BSOS 6: 897- 931 (1932). Reichstein, R" " Etude des variarites sociales ct geograph iques des faits Iinguistiques," Wd. 16:55-95 (1960), wit h " Note en conclusion" by A. Martinet , pp.95- 99. Ross, A. S. c., " Linguistic Class Indicators in J,>resent-Day English," NM 55 : 20- 56 (1954). Rossi , E., L'arabo parlato a $all'a, Rome, 1939. RuSari, see al-RuSari. Salim , S. M., al-Cibayil: Dirasa 'anOropolojiya Ii-qarya fi ·aJllvar 01-' Iriiq; Baghdad, 1956. Sapon, S., "A ,Methodology for the Study of Socia-Economic Differentials in Linguistic Phenomena ," S IL II: 57-68 (1953). Sassoon, D. 5., The Hislory of the Jews ill Baghdad, Letchwort h, Eng. , 1949. Schaade, A., Sibawaihi 's Lautlehre, Leiden , 19 11. Schramm , G. M., Judeo-Baghdadi : A Descriptive A nalysis of t lte Colloquial Arabic of the Jew~' of Baghdad, unpublished Ph, D. dissertation, Dropsie Co llege, Philadelphia, 1954. ' Smeaton, B. H., Lexical Expal/sion due to Technical Challge , as I11l1strated by the' Arabic Did/eci of al-Hasa, Saudi Arabia, unpu'bl ished Ph. D. dissertat ion, Columbia University, New York . 1958. Socin , A. , " Ocr arabische Dialekt von Mosul und Merd in," ZDMG 36 : 4-12, 26-53, 237- 77 (1882). Stephens, E. S., Folk Tales of Iraq, Oxford, 1931. ' Streck , M., Die alte Lalldscha!t Babylolliell IIach dell arabischell Geograplien. vats. I and II~ teiden, 1900, 19OJ. Sa labi, 'Abd al-FatUil;l ' Isma'n, Fi al-diriiSiit al-qu/"'r"illiya wa-I-Iugaw/ya: al-'imala fi al-qira'at wa-Maltajat al-'arabiya, al-Fajjala, Egypt , 1957.
, ., 180
COMMUNAL DIALECTS IN BAGHDAD
Sarbatov, G. S., "Nekotorye os'obennosti irakskogo dialekta arabskogo jazyka", Kratkie soobUenie Iltstftula vos/okovedelJija, 40: 80- 86 (1960).
0/ Words alld Phrases iI/ the Basrah DialeCI of Arabic, Simla, 1915. Thnayyan, 'Abd aI-Latif, "al-'AmOal al-'ammlya al-bagdadiya, LA 5: 11 - 15, 77- S1 (1927). Tsereteli , G. Y. , " K xarakteristikc jazyka sredneazijatskix arabov: predvoritelnoe socincnie, "Trudy 2- oj sessii asosiacii arabistov, 19- 23 Okt. 1937 (MoscowLeningrad, 1941), pp. 113-4S. - - - , "K izuceniju jazyka sredneaziatskix arabov: obrazcy feci I$:as~adarjinskix ambov," Trudy Insriluta JazykozrzQllija , Scrija VostolllYx Jazykol', 1: 251 - 71 (Tbilis i, 1954). - - -, Arabskie dlafekty sredllej Azi/: Tom I, Buxarskij arabskij dialekt, Tb ilisi, 1956.. T<1liqani, see al~ Taliqani.
Thompson, R . C.,A List
•
Van Ess, J., The Spoken Arabic of Mesopotamia, Oxford, 1917. - - -, The Spoken Arabic of Iraq , London , 1918. Van Wagoner , M. Y. , A Grammar of Iraqi Arabic, unpublished Ph. D. dissertation, Yale University, New Haven , 1944 - - - , Spoken Iraq; Arabic, New York, 1949; vol. II (mimeographed), Washington , D.C. , 1958. Vinnikov, 1. N. , " Materialy po jazyku i folkloru buxarskix arabov," Sov. Vost. 5: 120-45 (1949). - - , "Folklor buxarskix arabov," AOH 6: 181- 206 (1956). - -, "Obrazcy folklora buxarskix arabov," ArOr 25: 173- 89,426-51 (1957). - - -, Slovar dialekta buxarskix arabov ( = Palestinskij Sbornik 10), MoscowLeningrad , 1962. Weinreich, M., "Prehistory and Early Histo~y of Yiddish : Facts and Conceptual Framework," The Fiel4 of Yiddish, ed. U. Weinreich (New York , 1954), pp.73- 102. Weissbach, F. H. , ( review of Meissner, 1903b) ZDMG 58: 931-48 (1904). - - - , Beitriige zur Kunde des Irak-arabischell , vol. I, Leipzig, 1908; reprinted together with II , Leipziger Semitistische Swdiefl IV (1930). WeuJersse, J., Paysolls de Syrie et dll Proche-Oriellt 2, Paris , 1946. Xafaji see al-l;Iariri. Yahuda, A.S., '''Bagdadische Sprichworter," Orielllalisclie Studiell Th. Niildeke gewidmet (Giessen, 1906), pp. 399-416.
, I
NOTES I. In this work, the term "Iraq" designates the present political entity, with "Lower Iraq" and "Upper Iraq" separated from each other by a line ruiming roughly between al-FaJlilja on the Euphrates to Samarra on the Tigris. The term "Mesopotamia" is used here more or less in its literal geographical sense, i.e. to cover all the Tigris and Euphrates valleys and the a reas between them , from their sources on the 'Anatolian plateau down to the Persian Gulf.
2. The med ieval sources consulted are silent on the existence of dialectal differences among the socio~religious communities. Scattered data begin to emerge in the travel and descriptive literature of the nineteenth century. Oppert, 1863, and Jeannier , 1888, indiscriminately cited forms be longing to the dialects of 'the several communities merely as "Baghdadi," with a noteworthy preponderance of Christian or Jewish forms, unidentified as such. Oppert says nothing whatever of communal differences; Jeannier explicitly (and mistakenly) attributes two Muslim features to the Christians, showing at least a dim awareness of the existence of communal diffe-rences. In 1900, Oppenheim makes a brief but accurate characterization of the Muslim dialect (p. 266): "Die Aussprache, namentlich der Muhammedanischen Bev61kerung [of BaghdadJ, kommt der Beduinischen im , Vergleich zue Syrischen .und Agyptischen weit naher." In 1901, Meissner noted correctly that the Jews·~nd Christians of Baghdad speak a dialect different from that of the Muslims (p. 137, rn. 1), that the Christian dialect is close to that of Mosul (ibid.) , and that the rural sedentary dialect he_was investigating was closer to the Beduin dialects than to the uroan Muslim dialects of Baghdad or al-l:lilla (pp. 138- 139). The earliest article on Baghdadi Arabic as such (Oussani, 1901) presents largely" Christian data, a wholly Christian text, but aside from the observation that nonMuslims often have 1f!.1 where Muslims have Irl, a fact also noted by Anastase, 1903,p.592,and a few, vaguer statements, the data and text are not communally identified. Yahuda, 1906, whose proverbs are explicitly couched iri Jew ish Baghdadi, indicates briefly (p. 400, and fn. 1) that there are dialect differences among the three communities, that the Christian dialect isc10se to that of Mesul , and lets the lI1at~er rest there. Iq the same year, Ganima, in a collection of Christian proverbs from Baghdad, supplies many valuable and accurate comments on Christian Baghdadi and its differences from Muslim Baghdadi. In his description o fthe province of Baghdad, Chiha, 1908, p. 101 , notes "une assez grande difference entre Ie dialecte parle par les Musulmans et celui [sic] des Chretiens et des Israelites." in 1914, the problem of dialectal differences within Baghdad was recognized but not too successfully treated in Massignon's " Notes." Based on some notions acquired during his residence in Baghdad in 1906-1907, they give a bewildering account that speaks of no less than seven different dialects (three Sunni, two Shiite, and a Christian and a Jewish dialect) within Baghdad; except for a list of half a dozen features contrastlSI
182
COMMUNAL DIALECTS IN BAGHDAD
ing Christian and Muslim (which dialect 1) usage, and some vague statement about the Jewish dialect, the sevenfold subdivision is unburdened by substanstiatmg data. Many of the features he lists as common Baghda~i are Christian, some Muslim. It remained for P. Anastase Marie de St. Elie to publish in his Llljal al-'Arab large excerpts of Ma'riif al-Ru~fi's excellent description of Muslim Baghdadi (first written in 1919, published in installments in 1926,1927, and 1928) as well as a number of other articles on the Iraqi coUoquials (Ganima, 1926; JawAd, 1930; etc.) and to stale unequivocally in a comment on Ru~fi. 1926, p.141 , Cn. 2: uA peculiarity of the people of al-'Iraq is that every community ('umma) has a dialect of its own, so that no sooner does a single word cross the speaker's lips than he can be recognized as a Muslim, a Christian or a jew." Beyond this, the literature on the Iraqi colloquials (see note 3 below) consists of descriptions of Muslim Baghdadi so identified or not, of nondescript mingling of dialects, and some description of the Jewish dialect correctly identified but with little or no comparative comment. To non-linguists writing on contemporary Iraq, the situation is well known and deemed worthy of brief comment, thus Coke, 19~7, p. 20: "Her [Baghdad's] ill-fated history, too, accounts largely for the strange way in which her little communities have gone on from century to century leading their own little lives in their own little ways ... [They] have all gone on living, so to speak, in watertight compartments, each with their own system of law, their own way of working, their own preferences in family living , even their own peculiar oddities of dialect ." Cf. also Stephens, 1931, p. xxiii: "]n Baghdad, the Christians, Jews and Muslims all pronounce their words differently, and employ their own corruptions of speech." Duri,"describing 19th century Baghdad (1960, p. 908) is tersely accurate: "Though people of the three religions spoke Arabic, their dialects differed." The late Cantineau, though he did not deal with the Iraqi dialects directly, made a few brief but penetrating remarks aptly characterizing the position of the Muslim dialect: "lis {the nomads] ont fait de !'uaq une Algerie dans laquelle les parlers de sedentaires De sont que des Hots entoures de tous cotes par les pariers de nomades, et Ie parler des musulmans d'une grande ville comme Baghdad, de meme que celui d'Alger, est tout penetre de mots et de formes empruntes aux nomades ." (1937a, p. 226; cf. also 19~3, p. 148). 3. For a list and d iscussion of most of the relevant literature, see Blanc, 1962. To the works on Muslim Baghdadi and similar dialects (Ru~afi 1926-1928; Jawad, 1930; Van Wagoner, 1949, 1958; Malick, 1956/7; AI-Toma, 1957) several items can now be added: Sarbatov; 1960 (main characteristics and brief text with translation); J:lanafi, 1·962 (proverbs in fully vocalized Arabic script, with many accurate and valuable linguistic comments); Farman, 1959 (short stories in modern classical Arabic but with dialogues in the vernacular, though unvocalized). l;Ianafi, 1963 (first volume ofa dictionary, letter 'ali/only, but very detailed and useful); Malaika 1963 '(brief but valuable sketch of phonetics and grammar). The last tW?, both by native speakers, were received aftetthe present monograph had been set In trve; because of their importance, ] have tried to make as much use of them as possLble, chiefly in the notes, but this could not be done exhaustively. There is much less published material on Jewish Baghdadi: to Yahuda, 1906, we C3I! now ~dd t~e dozen Jewish proverbs included in l;Ianafi, 1962, and the score of Items listed m l:Ianafi, 1963, all correctly identified and rather well transcribed and interpreted;
NOTES
183
Kohen, n. d., and Schramm, 1954, are unpubli; hed. There are only fragmentary notes on Christian Baghdadi (Oussani, 1901; Ganima, 1906). A rural dialect of al-J:lilla province has been fairly well covered (Meissner, 1901 - 1904; Weissbach, 1904, 1908,' 1930), and the dialect of Qal'at Salel:t in the Amara region was the object of Van Wagoner's unpublished dissertation (1944; J am grateful to W. Cowan for calling it to my attention). Data on the Mosul dialect may be found in Socin, 1882; Calabi, 1935; and Dabbag, 1956: and on Mardin , in Socin, 1882. 4. On the phonemic value of leI, retained from a typescript where it had to be used instead of /,a/, see 3.21 below. All gelel-dialects actually use the . form 1ge1et/, but one dialect that otherwise belongs to the qe.11ll group seems to lack the final luI of the 1st pers. sing. (see 4.2) and another has Ikelt4/, viz. exhibits a Ikl vs. I~I contrast in place of the IqJ vs. Ikl contrast of the other qeltu-diaJects (see 3.26). 5. The term "non-Mus lim" is here equivalent t o "Christians and Jews." . The Mandaeans seem to speak the dialect of the surrounding Muslim population. 6. Cantineau, 1956. The oases in question are QariH~n, Sukhne, and Palmyra. 7. Data on Kuwayt are from my recordings and from Kuwait Oil Company, 1952. On the Persian Gulf area, cf. the various handbooks publish"ed by the Arabian American Oil Company, in the 1950's, and Smeaton,.1958. 7a. Such a relationstJ-ip is hinted at by Tsereteli (1941, p. 147}'and was the subject of a paper, as yet unavailable to me, by Wolfdietrich Fischer at the 25th Congress of Orientalists held in Moscow in 1960. On Central Asian Arabic, cr. Fischer, 1961, and bibliography listed there, p. 232. 8. Throughout this work, the abbreviation OA ("Old Arabic") has been used to designate the putative ancestor of the present colloquials; it is a noncommittal, blanket term, does not refer t o any single, concrete dialect and does not purport to so lve questions of reconstruction (see Ferguson, 1959; BJau, 1961a: Cohen, 1962). A symbolization such as "OA/q/" shou ld be construed merely as "the ancestor of the"initial consonant of present:.day Iqall or Ikilll in the. various dialects." What OA/ql might have sounded like or whether i.t was uniform throughou t OA is quite another matter; see 3.25. 9. rite Iraq Statistical Abslracl-1959 (Baghdad, 1960) ·gives, on t he basis of the 1957 census, a total of 784,763. On the immigration from country to city, see Kohen, 1953/54.
10. Coke, 1927, p. 299. II. Duri, 1960, p. 906, Coke, 1927, pp. 286, 299. In 1953/54 Kohen (p. 13) stiJlestimates the Jewish population at 17 percent in the years immediately preceding the exod,us. 12. The handbooks of Iraqi Arabic published for the benefit of the British
184
COMMUNAL DIALECTS IN BAGHDAD
troops during World War I (Thompson, 1915; Cole, 1917; Kelly 1917; Van Ess, 1917) all show an admixture of many features st rongly resembling Christian Baghdadi, but all are based on data suppl ied by Christian informants from Basra or Amara. The evidence for the informants' communal affiliation is, it must be added, indirect, but nevertheless quite compelling, e.g. the name (Elias Georges in Thompson) and certain passages in the other works. [n all, Muslim features are liberally mixed with the Christian dialects, often occurring side by side and interchangeably. 13. This is based on statements by my informants, all of whom were quite conversant with the Baghdad situation and aU of whom stated that differentiation does not run along communa l lines in Mosul. The fact that my Jewish and Christian Mosul informants say e. g. [honak] 'there' whe reas my Muslim informants say [hnDka] o r that my Jewish informant has [elrather than [i] in the feminine ending of such words as [bazziine] (non-Jewish informants : [bazzDui]) raises a question which [ cannot answer, viz. how far these differences actually correlate with community affiliation. Dabbag, 1956, p. 8, speaks of differences in pronunciation, including greater vs. lesser 'imala (i. e. possibly the matter of the feminine ending just referred to) as ex.isting between different quartcrs, not different communities. Caiabl, 1935, includes a score of items (mostly but not exclusively words of Aramaic origin pertaining to religion) in use among Mosul Christians only; cf. notes 126 and 163 below.
NOTES
185
distinctions . On Jewish dialects in present-day Arabic, see Cohen, 191 2 ; Brunot, 1936, and 1950, pp. 17ff.; Brunot-Malka, 1939, 1940; cf. a lso notes 17-21 below · on Jewish dialects in Iraqi Arabic, cf. note 3 above; on medie'va l Judeo-Arabic' cf. Blau 1 9~5, 1961a, 1961b. On the language of Christian Arabic literature s~ ~r~f~ 1905. On Jewish dialects in Persian, cr. Abrahamian, 1936. On the differe~ces dlVldmg Orthodox from Catholics and Muslims in Serbo-Croatia, cf. Ivie, 1958. 16. The effect of migrations on the rise of communal dialects in the SerboCro~tian area is neatly summarized by l vic, 1958, pp. 138-139; both t his and the JeWish-Muslim split described for parts of Persia by Abrahamian 1936 bear • " striking s ~milarities to our Iraqi situation . . 17. This ske~ch of communal differentiation in Jerusa lem Arabic is heavily mdebted to M. Plamenta, who was kind enough to send me a deta iled preview of his work on this subject. Both Piamenta and I doubt whether [1.] vs. [~) (Barthelemy, 1906, p. 199) or the structural differences reported by Garbell, 1958, pp. 306, 322, . 326, truly oppose Christians to Muslims, viz. a re fully correlated with communal affiliation. . IS. Barthelemy, 1905, p. IS3 , indicates a single difference between Christians and MUSlim's in Aleppo, viz. Christian lay] and raw] vs. Muslim [eJ an(l [0]. One wonders whether even this correlates fully with community affiliation, cf. Barthele~y's DictiOllllaire, s: v. Itor/. where " les femmes chretiennes" are said to pronounce 7tor/ and Itawr/ . For information on Cairo, I am indebted to N. Safran, M. Farid, and G , Abi-Saab.
14. A neat characterization of social dialects is given in the introduction to Ferguson- Gumperz, 1960, pp. 9-11; a more extensive and fa r vaguer discussion may be found in Cohen, 1956, esp. pp. 168- 213. The relevance of Hertzler, 1953, seems restricted to its title. A critique of the American dialectologists' unawareness of social dialects and of modern sociologica l techniques, especially with regard to sampling, may be found in Pickford,1956,and the application of such techniques in a particular case (Mexico City) in Sapon, 1953. Linguistic correlates of socioeconomic class in Engl ish have been studied by Ross, 1954 ; Putnam-O'Hern, 1955; McDavid, 1945, 1952/53; Fries, 1940; Hubbell, 1950; in French, by Reichstein, 1960. Perhaps the most interesting studies and the most clear-cut cases in po int are those dealing with the Indian caste dialects, nearly all in the Dravidian area: Bloch, 1909; Ramaswamy-Aiyar, 1932; Gumperz; 1958; McCormack, 1960. On the sedentary-nomadic sp lit in Ar.abic, see note 21. On specifically religious grouping , cf. note 15 below.
20. On the Algiers region, see Millon, 1937; Ca~tineau, 1937b, pp.709ff.; on Fez, Brunot, .1936; on Tlemcen, Mar.;ais, 1902, and Cantineau, 1940, pp. 222ff. The somewhat different account of Muslim-Jewish differences I find in Ph. Mar~is , 1957, ~p. 223ff., though it does not invalidate my conclusions, is puzzling, as it is ?stenstbly based on the same sources. A complicating factor (which, incidentally, 10 no ~ay solves t~e puzzle) is that Cohen, 1912, leaves out of account the many ru~a~ IOfluenc.es which seem to pervade present-day Muslim Algiers speech. lowe thIS IOformatJon to A. F. R. Brown, who wrote a dissertation on.this dialect.
15. For a bird's-eye view, sec Mieses, 1915 ; Cohen, 1956. It is noteworthy that Mieses who, in his discussion of the origin of Jewish dialects, tries very hard to show that «die Konfession" per se (as opposed to race, etc.) can create dialectal splits, produces illustrations of " minor" differentiation only. Similarly, Pickford, 1956, suggests a difference between upper-class Protestants on the one hand and, on the other hand , lower-class Protestants plus Catholics of all classes in the U. S. in such matters as mentioning the name of the Deity or discuss ing religion. Another so rt of mino r differentiation is mentioned in Bengali as between Muslims and Hindus by Chowdhury, 1960, p. 69, namely in the adaptation of Persian phonemic
21. On s.~?entary v~. nomadic dialect t ypes in North Af.rica, cr. Mar.;ais-Guiga, 1925 , pp. "-'Sill ff.; Colin, 1936; Millon, 1937; Brunot , 1950; Ph. Mar.;ais, 1957, pp. 21Sff.; on Greater Syria, -Cantineau, 1939. Note t hat (a) while all nomads talk " nomadic ~ype" dia~ects, not all sedentaries t~lk "sedent ary type" dialects; (b) for North Afnca, the view commonly held, foll owing W. Mar~is, is that this sp lit is due to two phas.es of Arabization: one in the eighth century, resulting in presentday sedentary dialects, and a second beginning in the twelfth century via wholesale Beduin invasions and resulting in present-day nomadic dialects, some of whose speakers have gradually become sedentarized.
19. Goitein, 193~ , 1933, 1934; Rossi, 1939. My own experience. so far as it goes, confirms the small degree of differentjation.
186
COMMUNAL DIALECTS IN BAGHDAD
22. On Oran, cr. Cantineau, 1940; on towns of the Algiers region, 1937. 23.
cr. Cant ineau,
NOTES
cr. Millon,
1938, 1946.
24. For al-l;Iill a province, d. Meissner, 1903b, p. ix; this is confirmed by my own rural Iraqi informants and recordings. 25. I am indebted to W. M. Erwin for pointing out this contras"ting pair. 26. Cf. Oussani, 1901, where this is attributed to the Baghdad dialect without specification, but where the context makes it clear that Oussani is describing Chr istian Baghdadi. 27. From my Siirt informant I have such forms as {hava] 'this', {fllfe] 'three' (cf. Mosul {eaSe], [ellGin, [abyav] 'white', and so on throughout; it is quite certain that this is not a speech peculiarity· of this particular individual. A Kurdish informant from Zakho (Upper Iraq) volunteered similar forms with labiodentals which he attributes to unspec ified Christians from the North, though he does not have them in his own Arabic, which was learned as a second language in Mosul; Zakho itse lf seems to have only Kurdish and Nco-Aramaic speakers. A similar phenomenon, though apparently restricted to [v] for [6], is reported for "some speakers," apparently of the Shiite minority, in al-Hasa by Smeaton, 1958. Because of tbe erratic transcr iption , it is hard to know wha t to make of the v in yerkov (= lyerkef) 1 'he runs') in Kelly. 1917, side by side with ethrobll ( = Ir,rebul 'hit him!') ibid. , p. 4 1, from a Basra Christian. 28. For further examples of III in M, see Ru~afi. 1926, p. 405. 29. On the 'presence of III in some roots in t he l:Ioran, d. Cantineau, 1946, pp. 107-109; I have noted it among the Negev (Southern Palestine) semi-nomads .in e.g. Inagal l ' he transferred' and throughout this root, /galebl 'heart ', etc. For a general discussion of III in Arabic, d. Ferguson, 1956. 30.
cr. Mansour,
1955/56, 1957.
31. The etymo logy of these fo rms is, of course, OA/albii riba/; the history of the C form is self-evident , and must include such stages as /lberilJal > IlbeglJa/ > ImbelJa/, but that of the J form is puzzling because of the 10/. 32. Socin, 1882, has only Ir/ for OA/r/ in his Mosul texts, does not mention a 181 reflex ; Dabbag, 1956, p. 8, says that Igl for OA/rl is common to members of all the communities in Mosul and is not a mark of communal ditTerentiation as in Baghdad, but that "some quarters pronounce the r as such (kama hiya)." 33. cr. Ph. Maro;ais, 1956, pp. 16-17. 34. Cohen, 1912, p. 27. Ph. Mar~is, 1957, p. 224, has matters reversed, perhaps through an oversight (see note 20). 35. Colin, 1936, p. 500.
187
36. Tsereteli , 1954, p.254; 1956, pp. xiii, xxii. 37. Jai:liz, ed. Harlin, 1948, pp. 15 . 37. It is not entirely clear whether the pronoun in the words awjaduhii and aqalluhii qub(lf,lII refers to all the lu8gal or to all the '/lOgat,that befall the If/: the latter interpretation seems the more likely. Kofler, 1949. pp. 81-82, reports ItJ for If/ in Baghdad al an unspecified time; his references come from Howell's dictionary. where they are unclearly cited; they seem to include Jawhari (d. 1010) and the Qamils of Firuzabadi (d. 1414), but Howell's abbreviation "Jh" may just as well refer to J al:li~ as to Jawhari, and the Qamlis yields nothing of interest upon inspect ion of the articles ra' and /tIOga.
38. Cf. the rather full d iscussion and verbatim quotes in Gumpertz 1952/53, pp. 114fT., and the confrontat ion with pfesent-day Jewish ,Baghdadi data in Mansour, 1955/56, 1957. Cf. also the discussion , in Morag, '1959/60, pp. 221ff. and p. 224, fn. 59. Gumpertz thinks , plausibly enough, that makriix may be onomatopoetic reference to uvular pronunciation but also notes that Ibn Durayd (d. 933 in Baghdad) believes the . term al-karixa or al-k(irNw (both versions in the original) to refer to the throat. Ot herwise, the root k-r-x has, in the Arabic dictionaries meanings that seem unrelated to pronunciation. 39. Bravmann, 1934, pp. 128ff. 40. l owe this info rmation to the kindness of I. Avineri, who sent me the manuscript of his master's' thes is on the Arabic element in the Nco-A ramaic of the Jews of Zakho. 4 1. It is noteworthy that among the speech defects and dialectal pronunciat ions listed by the medieval Arab grammarians, such as kaskasa, kaskasa, talla/a, etc. (for a list, cf. Anastase, 1903), there seems to be no spec ific term for 1&I < leI; in these sou rces the te rm lufJga covers a number of speech pecu liarities (despite Hava who thinks it refe rs only to "misp ronuociatfoo of Ir/"), as can easily be seen from t he many examples in Bravmann , 1934, pp. 62ff., the discussion in Ja lJi~ (see note 36), etc. Among the several unexplained terms for speech peculia~ities, howeve r, t here is one which might poss ibly refer to this; it is the term !utatjyat al-'iraq, of which Xafllji (d. 1659), ed. 1299A.H., p. 234, says l hat it is a lso called laxlaxonfyat al-'iraq. These terms appear in an anecdote that has come down in several slightly variant versions, t he earliest of which seems to be in the Komi! ofal-Mubarrad (d. 898) (quoted in Xafaji, ibid" and Rabin ; 1951, p. 21), where it is listed with other pronunciation defects by a man who claims that one particular ' dialect is free of.a ll of them. Its meaning has, so far as 1 know, never been satisfactorilyexplained; it is the only such peculiarity which is attached not to a tribal name (such as kaskasat rabi'a, etc.) but to an area, that of al-'Iraq (viz. what we have been calling Lower Iraq ; Upper Iraq bore the name of al-Jazira) and presumably implying a sedentary population. Whether it applies to opr/gl < Irl shift , or to the famous .accent of the "Nabateans" (Aramaic speake rs) of that area or to someth ing else altogether cannot at present be ascertained.
188
COMMUNAL DIALECTS IN BAGHDAD
NOTES
42. For additional examples and for a characterization of "plain colloquial," "mildly formal ," and "semi-literary" styles, cr. Blanc, 1960, pp. 83, 85.
one of M (lqJ > /g/ with only occasional /j/ in loanwords) and another for the countryside Ig/ or fjJ according to environruent, as in the Beduin dialects).
43. On reflexes of OA/kj in ru ral Lower Iraq, cr. Meissner. 1903b, Weissbach, 1904, 1908, 1930; Van Wagoner, 1944; data were corroborated by my informant from al-J:Iilla province. On the classification of Beduin dialects having a leI reflex and those having a jel reflex , see Cantineau , 1936, 1937a.
53. Meissner, 1903b, p. ix, indicates only /q/ inisuch words for his rural dialects but my recordings have /&/, rather than /g/ in those words for which M has /q/ : /ga$er/ 'castle', /gesma/ 'fate', /gabbot/ 'coat', (M/qappii\/),/ga¢iya/ 'issue, cause', and /&at/ 'suit of clothes' .
44. Cf. Cantineau , 1939, for the distribution of affrication of OA/k/ , and for the different iation between conditioned affrication in nomadic dialects vs. unconditioned affrication (coupled with fronting of OA/q / to /kf) in some sedentary dialects (Central Palestine. some Syrian oases, some North African sedentaries). The coverage in Binder, 1939. is inadequate and has been superseded by Cantineau's. The semi-nomad ic Beduins of the Negev, Sinai, and (presumably) Egypt show no affrication o f OA/kL, whereas those of Transjordania follow the pattern of the Baran sedcntaries and the Syrian desert nomads. 45. Cf. Kofler, 1940, p. 116, and notc 47 below. 46. Cf. Anastase, 1903, and references in note 41 above. On an other alterat ion of this pron. suff., called by the same names, cf. Cantineau, 1960, p. 87. 47 . Cr. e.g. 1the comments by Suyuti (d. 1503) quoted by Anastase in RU$afi, 1926, p . J47 , fn. I, and those by Xafaji, ed . 1299 A.H., p. 214.
48. Schaade, 1911 ; p. 18. 49. The opinion t hat there a re no more than three instances of OA/q/ > Ikl in M is expressed by RU$afi, 1926, p. 144; this is also the source for the form /kufax/, which issorroborated by my informants; the forms /ketal l and /waket/ are, of course, mo re common and came up without elicitation. A /kl in the root corresponding to OA q-I-l is common to Beduin dialects in general , cr. Cantineau, 1936, p. 29, and is found also among the Negev and Sinai semi-nomads. Among some of the Syrian desert Beduins and in rural Lower Iraq, (Meissner. 1903b, p. ix) that root also has alternants with initial lei. ·tt also occurs in the west (W. Mar,ais, 1902 , p. 1 I). 50. ·Cr. additional examples in RU$afi , 1.926 , p. 144. Some are listed as having the alternate form with /g/, in other instances the forms w.ith /j/ are explicitly stated to be rural (x a ~$a bi-'abna' al-badiya). ·Cf. also J:lanafi , 1963, p. 18. 5 1. Cf. Meissner, 1903b, p. ix ; Weissbach, 1908 ; and my own informant for the region of Musayyab; for the Amara region , my own recordings and Van Wagoner, 1944; cr. also reference in note 50 above. 52. Cf. Cantineau , 1936, pp. 3Qff. In his review of Van Wagoner, 1949 (Cantineau, 1953), he wonders at the lack of affrica tion of OA/q/ to /j/ in Baghdad, in view of the other Beduinizing features of the Baghdad (Muslim) dialect , and offers the form /I)arij/ (my info rmants know only /I)arijiya/) as evidence that such affrication does exist . Cantineau's guess was in the right. direction, but he did not anticipate that there would be, so to speak, two degrees of Beduinization,
54. Cf. Cantineau, 1939. 55. Cf. Kofler, 1940, p. 232, wherc the phrase " all grammarians" must be taken, in the light of the sources he quotes, as referring to writers frool the eleventh century onwards, which is corroborated by my own findings and , in a negative way, by the lack of any such distinction in Brockelll1allll, 1908, p. 21. 56. Bravmann, 1934, p. 121. 57. Ibn Xaldun, 1886, p.509; cf. also the citation of this passage by Anastase in his comment to RU$afi , 1926; p. 143, fn. I. 58. Bravmann, 1934, p. 128. 59. Cf. Scbaade, 1911 , p. 14 and fns. 14' and 19.
~, /e/
compl~l~ntary dist~ib~ti~n
60. In and /u/.are in nearly or, to put it another way, the contrast /e/-ju/ is neutralized in a good many pos itions but maintained in some limited environments, e.g., fjerQal /'camel', /jumal/. 'sentehces', /I)ebb/ 'jar', Il);ubb/ 'love', /rekbat/ 'she rode', /rukba/ 'knee'. Neut ral ization is primarily connected with the existence of two types of consonantal environments, one [u]-coloring and one [e]·coloring (see 3.32), e.g. Idarub/ 'road' but Ibaredl 'cold'; /waguf/ 'stand ing' but {lazem/ ' ne~essary'; /tufar/ 'he jumped' but /tela'/. 'he went out': For M speakers, t he ident ification of a given phone as belongir)g to lu/ or to /el can be seen most clearly in the informants' spell ing, with lui classed as tjamma and lei as kasraj the latter is, on the other hand, kept dist inct f.rom /i/ by some ql!alifying adjective. e.g. the frequent term used in Ru~afi, passim, of kasra gayr ~ar;l.za. J and C speakers make similar verbal responses' when asked to id~ntify a given allophone; several say that allophones of lei are " a sound 'like the French e ' in je," a comparison also used by Anastase in his ' comments on Ru.safi, and .Dy Dabbag, 1956, p . 8, in his discussion of the Mosul dialect. Many of my phonemicizations (e.g. JC/I;lete!/ ys. M/I;lutulf) are l:5ased on such explicit . speaker identifications. 61. This on-gl ide has been noted by Me issner, 1903b, p. ix, though he conceives of it as a "moui llierung" of the preceding consonant; this is corrected by Weissbach, 1904; p. 932, who interprets it as a falling diphthong and rewrites Meissner's m'ez, z'en, a's miez, zien, and so forth. 62.
Ru~ fi , 1926 ,
p. 313,andpassim .
190
COM.MUNAL DIALECTS IN BAGHDAD
63. This dist inction between final unstressed tal and lal is apparently not sufficient to distinguish, in pronouncing CI. AT. , between the singular and dual in such forms as Ikataba/ 'he wrote' and /katab
191
NOTES
74. lowe t he admirably plausible suggestion as to the unattested qural pattern for OA adjective plurals to Ferguson , 1959, p. 627, fn. 21. 75. On 'correct' 'imala as prescr ibed by various authorities from the eighth to the twelfth centuries, cf. the monographs by Grunert, 1875, and Salabi, 1957, and addit iona1 material in Schaade, 1911; Bravmann , 1934; Kofler, 1940; and Rabin, 1951. On colloquial 'imiila as a hallmark of the Iraqi vernacular, there are some indications in the above, as well as in J:lariri (see note 79) and especia lly in Bill! (d. 1348149). cf. Hoenerbach, 1956. I;filli, himself a Baghdadi, states that 'imiila is one of the characteristics of Baghdad speech (p. 55 and p. 12 of Arabic text) and hence a characteristic of Iraq i vernacular poetry (p. 33, p. 170, and pp. 135- 36 of Arab ic text}..I am indebted to J. Blau for calling my attention to the last named source. 76. Sibawayhi (d. 793) in Schaade, 1911, PP. 38ff; and 'ZamaxSari (d. 1144) in GrUnert , 1875, pp. 22ff. 77. Siba":ayhi in Schaade , . ~911 , pp. 43-44{the action of back consonants is not too clear 1.0 the examples gIVen: mi.Jbafl can be said with or without 'imiila, qiirib on ly with 'imlila butfariq only without. In JC, despite such forms as Il).magl (see also note 105 below) back consonants do not seem to prevent ' imala. 78. Schaade, 1911, p. 42 , though the point is not too clear; at any rate, the examples just quoted from I;filli do show 'jmiila of qiitil forms; including.participles.
68. Cf. Rusafi , 1926, pp. 212, 596, on the M treatment of OA/a/. 69. Plurals of the OA pattern qitlall also seem largely immune to 'imala, e. g. JC/sedqanl ' friends:, I!)ezmanl 'belts', J/begqanl 'ja rs', but there is the J plural Ijegdinl ' mice' (C form not noted) and the fo rms J/Oebbin/, C/debbenl 'flies' to ind icate that non-'imiila in this pattern is not as far- reaching and not due to the same cause as non-'imala in the pattern qtiil. 70. From Kohen's exhaustive treatment of the verb in the dialect of the Amara Jews (practically identical with J) it ·appears that aliForm I active participles have this leI except Iyibesl "drying', i. e. the same as the adjective Iyibesl 'dry'; I know o f no other Form I verb with K1-y. . 71. Thompson, 1915,. p. 16: aymree. (See note 12 above.) 72. From an informant from Jirja district of Suhaj province, Upper Egypt, 1 have such forms as [kalbi] 'bitch', [salli] 'basket', ('andeni] 'chez nous', [heni) 'herc'. Howevcr, I note that in a similar dialect of Qina province (Khalafallah, 196 1) such forms have IiI in the pause only, lal otherwise. 73. The dialect of the oasis of Sukhne in the Syrian desert has fil and IiI for OA/al under conditions similar to those of J, but also in some cases where the qellu-d ialects have only la/, e. g. [cib] "he brought', MJC/jab/ ; cf. Cantineau,
1956, pp. I 28ff.
79. Schaade, 1911 , pp. 38-40; l:Iariri (ed. 1299 A. H.) p. 105, on not pron.ouncing bauii with 'imala, and Xafaji's comment on this, ibid., p. 221, indicating on the contrary that this is allowed by some authorities because "some Arabs" do so pronounce. The form hiilli (read Iha ni/ ?) occurs in Hoenerbach , 1956, p. 169,1.9.
80. Fiick, 1955, p. 69. 81. Salabl, 1957, pp. 238ff. The quote from Sibawayhi's Kiliib, II , p. 270, referring to one speaker's pronunciat ion of the· fem. ending in pause, reads in part : sabah 'a flui' bil'allffa'amal ilia qablahii kama yumfl mi; qabl 'al'ali/ 'he likened the" [of the fem. ending] to an 'alii. and so pronounced the preceding [fattl] with 'ill/dla as he does [for afatfz] preced ing an 'alif' 82. Sa labi, 1957, p. 238; Bravmann , 1934, p. 98. 83. GrUnert, 1875, p. 78; Schaade, 1911 , pp. 38-40. 84. Quoted in Rabin, 1951 , p. 123, fn. 13, from the Xitiib, II , p. 28 1. Kofler, 1940, p. 250, reports statements indicating some tribes pronounced this word with M~ ~ ~~
.
85. The contrasts may be even more tenuous than they seem, since some
192
COMMUNAL DIALECTS IN BAGHDAD
NOTES
speakers seem to have /p l ratlier than /jay'f./; others may not use /zawil,/zawja/ at all, the common parlance terms for 'husband' and wife' being /rajel/ and /mara/. For my MJC informants, however. the contrasts are unmistakable, even if their functional yield is very low. Van Wagoner, 1949, 1958, and, presumably following him, Malick 1956/57. write ay for both lay! and leI while differentiating aw from 00: since no adjustments or explanations are added, the results are misleading : bayla 'his house' and mayta 'dead (f.)' are undifferentiated. For a discussion of this and sim ilar problems, see Ferguson, 1957.
95. In fourteenth century Baghdadi vernacular poetry, which st udiously avoids the final short vowels of C lassical Arabic. the 1st pers. sing. of the perfect is consistegtly vocalized with t;lamma or {Iamma~wiiw (Hoenerbach, 1956.passim). What is ~ore , this vocalization is explicitly stflted by t~e contemporary sources to be de rigueur in Baghdadi vernacular genres, just as its absence is de rigueur in Western genres, because those are the respective colloquia l usages: Baghdadi qultu vs. Western quit (ibid., p. 31 of Int roduction). 96. RUl}afi, I926 , p. 247, says the dropping of Inl is wiijib (obligatory) before ·nl and ·mi, but jii';z (optional) before other pronouns.
86. Kohen, in the remarkab ly accurate phonemic transcription he uses in his work on the verb. regularly writes ~w and ~y not only in mawjiid sayydbu, mxawfin, but also in cases such as nawladtu 'I was born'. However, in stressed forms, such as tawzan 'she weighs' he gives the alternant tiizall. This may be due to the fact that in unstressed syllables the phonetic diphthongizat ion is more common, whereas in stressed syllables the monophthongal long vowel is the normal allophone. When consulted by me on this point, Kohen agreed that [mlljild] occurs and that the aw of mawjiid· sounds very much like lIW.
97. Dlau, 1958, p. 164; Hoenerbach, 1956, p. 66 of Introduction and p. 88 of Arabic text. 97a. Tn l;fanafi, 1962, p . 27, proverb 44 reads as follo\V.s: luxOi zzcn matxel}renl 'pick what's good , you (£.s.) won't lose by it' ; this J·like len! instead of M(ml in an M proverb calls forth this comment by the author: "This is a proverb in the: dialect of t he A'(}amiya quarter, a dia lect which is practica lly extinct." It would, of course. not come as a comp lete surprise if J·like features were to be unearthed among some conservative Baghdadi Muslims.
87. For a detailed and accurate list of M consonant c1usters,see Mal ick 1956/57. 88. Cf.
Ru~ii n ,
1926, p. 21J.
89. More accurately: when K) is y, as in 'my, the phonemic result of pattern element if plus radical y is IiI. This sort of "merger" of a pattern element with a radical (esp. y or w) into a single vowel phoneme is very common and will often reappear in this text. 90. These form s may be due to an unusual development with / hiya/'she' as the starting point: a metanaIysis of Ihiyal on the analogy of / biya/' jn her' /xalliyal ' leave her' as composed of a base hi· plus the 3f.s. pron. suff. - hii. (yielding -/yal after base final - i). The equivalent masc. sing. and pIur. forms a re then constructed in the regular fashion: just as we have Ibiyal 'in her', °/binu/' in him' Ibiyem/' in
them', so we now have /hiya/ 'she', /hinu/ 'he' and /hiyem/ 'they'. 91. Jawaliqi, p. 138, cites the vulgar forin nbn' . 92. cant ineau, 1939, p. 81. 93. Ru~~ifI , 1926, p. 460, lists the feminines throughout without mentioning their marginal character. AI·Toma. 1957. p. 42 , may be reflecting Karbala usage; for villages of a l ~ 1:iilla province, Meissner. 1903b, p. xiv, agrees with my informant. 94. The notation 0, e.g. kitbo 't hey wrote' in Al-Toma, 1957, and Ferguson, 1957, fn. 18. is puzzling. At any rate, Al-Toma might have intended [ow] and his dialect is, moreover, not quite the s.a me as M; and Ferguson info rms me that, though ht? be lieves he heard no diphthongization, he used. a C informant for M usage.
193
•
98. Cantineau had already noted (1 937c, p. 153) this invariant -/kil as a feature of C and of a number of " paders de vieilles populations sedentaires" (Palmyra , Sukhne, Derez-ZOr, etc.). However , I have note~ it also among the Negev sem i·nomads. l;Iilli has both zawjik (Hoenerbach, 1956, p. 158) and 'ummiki ibid., p. 160), with the former type perhaps more frequent. 98a. l:Ianafi, 1963, p. 33,cites achildren'srhyme containing the forms /tal;ltu/ 'under him'/waQ-dul "by himself'; commenting on the unusual -/u/ suffix instead of M/a/, the author explains it as taken either from the Mosu1 dialect or from that of the A'Oamiya quarter of Baghdad; cf. a lso n. 97a above, 99. On the overlapping of lal and lal in final unstressed position in M, and on the possibility of an allomorph Ia! after 1t).1and rI. see 3.31. The same holds for an optional allomorph -/hal rather than -/hal for the object suffix of the 3rd pers. fern. after base final /a/. 100. One wonders whether this is connected'with the prevention ofhomc;mymy: if the fern. ended in -Iii , there would be homonymy with Ixali/ 'my maternal uncle" /,ammi/' my paternal uncle', Ijeddi/ 'my grandfather". Cf./kanni/'daughter· in·law, (no possible homonymy, fern. has the regularly expected -liD. lOt. In other words, where the base fina l syllable Kad IiI in OA, the present fern. ending is -IiI, a d iachronic statement whi~h would include many other examples cited so far. If we add that the remaining cases reflect /if 01" Iyl in the OA form, we have an instance of a diachronic statement which is s.impler, shorter, and more inclusive than the synchronic one. It would not cover such forms as Ibgadda/, or words not attested in OA, principally loanwords, so that some supplt~mentary statement would still be necessary. 102. One would expect the allomorphs of T in the prod~ctive, inflectional function of creating fern. adject ives. as here, to be more regularly conditioned than
194
COMMUNAL DIALECTS IN BAGHDAD
in words in which it is merely a fixed termination U~wayya/. I~arika/. Ibonta!). All o ther adjectives in- lib/-noted have the aJlomorph-/i/ :/~l)il:ti"·wholet./~I:til:lil 'st ingy', /saril;li/' fran k', /qabii)i/ 'brazen', c1. also Imli'i/'obedient ', /wsi'i/ 'wide', /gfi' i/ 't hin'. and nouns such as /gi'r)i/ 'smell ' . 103. C f. Canima, 1926, p. 267, on t he Aramaic origin of-f Ont· 104. Cohen , 101 2, p. 355, gives bl/lta fo r Jewish Algie rs. 105. Cf. the absence of 'imiila in this wo rd in other 'imiila dia lects, e.g. Aleppo /bsebl vs. Il:lfllar /; and there a re instances of l'criplio plena with waw in an eleventh century Judeo-Arabic tex t from Iraq (Epstein , 1915, p. 71). cr. also Eg. /~u f!lii ~ / .
106. An archaic Mjjij / may have been preserved in a rhymed proverbial imprecation: Ib.emmat ejjij wrej fat lefrii rij /' thc hens' fever and the chicks' quaking {upo n you-r (Hanafi, 1962, p. 157, no. 697). I:'anafi's laconic comment (" /jijl ~nd /jaj/ are the same as Idejaj /,,), coupled with this awareness of J .features OCCu~rlDg in the J proverbs he Quotes (e.g. nos. 11 6. 274, 512, etc.) clearly Imply he conS iders this fjU/ as M. That M may once have sounded mo re like J has already been seen (3.300, 4.21b, notes 95 a nd 97a.). 107. 'Heard o nl y in the J proverb /,~ fUg kafal zarzl1r we6nenem tayyari/ 'a sparrow was suret y for ' a starling and both are on the wing'. However, Sassoen, 1949. p. 195, has th is J proverb end in I~iyyar/ 'flying, having, flo wn off (pl. )'. and the M version in l:Ianafi, 1962, p. 190, no. 899, is /zarzl1r kefal 'asffir we8nenhum tayyara/. On /wag~afi / etc .• cf. a lso Jawad, 1930, p. 612. 108. This C term for the speech of the Baghdadi Muslims. which I've heard in such comments as /qayei)ki baddawi tamaml 'he speaks the Muslim dialect to perfection', shows that t he Bedu inized character of that dialec~ is .quite evident. Though it is presumably not less evident to M or J. speakers. thiS term seems unknown to them a nd they seem to have no specia l name for the dialect. Asked whether this term does not a lso refer to Beduin speech, C informants say that ir " reall y" does, l::lut that they "actually" only use it as indicated. 109. T he term M/jirant. J/jigan/, C/jigan/ seems to be used both as a plural and as a singular, 'neighbor(s); in J there .is the new p lural fjwagin/. 110. The fusion of the Itt wit h the following noun is common to very many Arabic dialects (d. Ferguso n, 1959. pp. 624-25) and attested in Middle Arabic and medieval vernacular writings. For fourteent h century vernacular Iraqi, cf. I;lilli in Hoenerbach, 1956, p. 35, xamas-Iasya 'five things', and many ot her examples passim; cf. also Blau, 196 1b, p. 56. 111. On It! and general velarizat ion in the teens as a feature common to most if not all Arabic dialects, cf. Ferguson, 1959, pp. 625- 26.
195
NOTES
11 2 ~ Each participal pattern bears a direct relation to a specific verb and a specific conj ugation t ype. Participles ta ke the a lternant -/ni/ for the 1st pers. object pron. suff.• as do verbs, but unlike nouns. Participles do not take the "broken .plural" patterns, whereas nouns do: M /~aye f/' having seen', pl. /~aynn /, vs. M/M.yeb/ 'old man', pI. /~iyyab/. Individual words may belong to both categories : M/masjunl 'having been imprisonejd ' (p I. /masjfrnin/) or 'a prisoner' (pI. Imsajin/). The distinction is, however, of,ten blurred, espec ially in passive participles, e.g. the noun /masjun / can also take the sound plural.
11 3. MJC/qabel / is a predicat ive expression, coniOlon ly heard in rhetorica l question , e. g. /hassa qabe l enjGz?/ 'does it make sense for us to give up now?'; cf. the similar use given in Barthelemy's Dictiollary for Aleppo , and the similar lack of 'illla/a. 114. Malaika. 1963. p. 53, reports forms without co ll apsi ng, viz./hubeb/. for his variet y of M. 115 . .In other fo rms of geminate roots. the onl y modification is t he different behav ior of the KKK cluster: fromjllll we get MJC/mjannen/ 'maddening', fern. M/mjannena/, JC/ mjanneni/, etc. This holds for the var ious forms of the Form II verb, e. g. M/yjannenun/ vs.· /ykamlUn/. 116. Most of my M informan ts and RU$a~i have only If)erab/, but one of my informants has /Qurab/. Fluctuation between -lui and le/ in some envi ronments has already been a llude!i to ; W. M. Erwin tell s me that some of his' M informants show, without entire consistency , such alternations as / rckab/ 'he rode' but /rukbat/ 'she rode', /geraf/' rooms ' but /gurfa/'a room'. 117. [n J, the verb 'to read', root qgy. presents an anomaly wherever the parad igm requ ires the sequence KIK 2 , in that therc 'is /qq/ rather than /qg/, e.g. /qqelu/ ' Illave read' ,/aqqa/' I shal l read', imperative /qqa/./qqe/. /qqO/. cf. 3.24(c). In M and C, the root is qry and the behavior normal. 118. Kohen gives both liJwz;m and tiiziJ", etc. , i.e. note 86.
{~wJ"'"
[uw], [11], cf.
119. Some speakers may have only Ie; (cr. RU$afi, t927, p. 148),othersseem to have lui in a few environment s: /gumet/, but /te'Qct/. Malaika, 1963. p. 56, reports lu/ in practically every verb with K 2 ...-<.w (a notable exception is cellet . so that the ru le given , ibid., is incorrect as it stands), e.g. /gulet /./sufet/./rui)et/, etc. My sources and W. M. Erwi n's informants have on ly /gelet/, /~e fe t /. / rebet / , etc., but /gumet/, /buget/. 120. RU$afi, 1927, p. 96, lists no forms with in it ial Je/- . and says that "some" say /jit/. etc. instead of /jet/, but in his treatment it is neve r clear whether he is rest rict ing himse lf to Baghdadi Mus lims. Mosul has /jit/. etc. li ke JC. as has 'A na ; the latte r a lso has the unusual3rd pers. plur. of the pe rfect [jam], 't hey came' though with suffixes agai n (jowk) 'they came to you'. Cf. Cairo Igum/. /guk/, though in Cairo /ml in the 3rd pl ur. is not restricted to thi s verb.
196
COMMUNAL DIALECTS IN BAGHDAD
121. In M . t he second a, when unstressed, yields zero Ukamlawn or an anaptyctic vowel (fkammeiaw/), which may be luI in the appropriate e~vironment: fjarbaw / or /jarruhaw! ' they tried'. This holds for Form V and the quadnconsonantals as well; cr. Malaika, pp. 46, 48, 61.
122. Kohen gives for the J verb r ayan! 'to see' the somewhat irregular imperfect forms fl 'ayn/. In'ayn!, etc., It'anon/. Iy'anon!. etc. My informants give forms both wit h and without Iy! fo r the lattcr. i.e. both It 'anon/ .a nd It'aynon/. but in the suffixless forms only the regularly expect ed /t'ayenl. Iy'ayen/. etc. 123. Kohen lists for KI -w a lso J/ntakal/' torely·. a sort of merger of a hypothetical*/nwakal / and an actual Form VIII /ttakal/, also found in MC. One wonders whether one can still speak of a root wkl; the verbal nouns are /tekliin/ and /netkal/. Cf. a lso J:l anafi , 1963, s. v. /nteCal!. p. 335. 124. Somel speakers have /qal/-in the 1st pers. sing.: /qalaqul/'l say" /qalam~i! ' I'm walking' . Kohen gives neither /qadl nor /qal/, but only /qa/: !qa'aqul/, /qaqul/, as in C ;.this may be onc of the rare differences between his Amara speech and true J. 125. In C, a lthough there is usually no
1'/
bet ween /qa/ and the following
la/, a merger into la/ is o rten prevented by the phonetic differences bet,,:,een the
two !a/'s, thus /qaam~i/, viz. [qAam~il However , when the second la/ Is back , merger does occur : Iqaaqull is usually the same as /qaqiil/. 126. My Mosul ·informants (two Muslims, one Jew , one Christian) all h~ve this /qa/ ,..... /qad/, but CalabI , 1935 , p. 72 , says that this is in use among Muslims only, whereas Christians h~ve / ka/ ,..., /kad/ (~ews are not ~e?tioned); several other present markers used III Mosu! are also listed there. Socm s texts have kethroughout , eJg: ke-ja'mif ' he docs', p. ll , as well as a past marker spelled identically, e.g. ke-gilla 'we came', p. 7 ; my informants have [b), [kan) in the latter case. 127. Van Wagoner, 1944, pp. 56- 57. 128. Cf. Cantineau , 1936, p. 83; as a gene ral characteristic of nomadic vs. sedentary dialects , 1939, p. 82. Among the Negev semi-nomads, who otherwise speak an unmistakabl y "Bedu in type" dialect. there is a fbI preposed . to the imperfect in non-subord inate verbs, t hough its function rema ins to be invest igated. 129. On the /d/ of /dekteb/ see4.8ge; /a~ui is a frequent MlC "tone modifying" particle, often corresponding to Fr. alors, o r to some uses o f Russ. till , Eng. well, so , tltel/. See this item in 6.5. 130. MJC!yam'awwad/, fern. /yam'awda/, plur. /yam'awdinl is used to underscore a request. roughly like Syr.-Pal. /daxlak/. Idaxilak/ ; on the Id/ of Idogfu /, see 4.8ge. 131. M J ~/wda' tak / is used to unde rscore affirmations or- oaths; the pron. suff. is variable: M/wdu'ta Imal:1mDd/ 'I swea r by Mal:\lllUd'.
197
NOTES
,
132. A similar Iken/ or {ka/ occurs i~ Anatolia and Mosul; on kllll , k im in the Central Asian dialects. see F ischer, 1961,<1>. 256, and rn. 2. 133. In Baghdadi Judeo-Arabic translations of Hebrew scriptural or liturgical texts. /ila/ regularly serves to render the Hebrew accusative particle 'eO . This type of Arabic. which we may call "Judeo-literary," has many features of its own and differs from J as much as it differs from C lassica l Arabic. It naturally does not fall within the scope o f this monograph. A rough idea of its features can be gained from a perusal of Blum, 1927, though Blum was unaware of t he spec ia l nature and position of this idiom. 134. Tsereteli, 1956, p. 5; also Fischer, 1961, p.254. In Cairo, the demonstratives are sometimes used as copulas, e.g. /gozik da-gabi! 'your husba.nd is a fool', fax iik da-dallu'a/ your brother is a spoiled brat '; on some copula-like uses of a particle /d/. apparently of Berber origin, in an Algerian dialect, cf. Ph. Mar~is . 1956, pp. 462-466. 135. On /mal elpaspOrtat/ > /malpaspo rtat /. cf. 4.91c. 136. J-Iowever. in a sub-Saharan- dialect. I find kitiib hanak, halla'i, 'your. my book' (Carbou, 1913, p. \7). 137. In MJC. /yemna/ and /yes r:a/ often seem to hav~ substantive status. Le. 'the right. left side'. However, /idak elyemnal 'your right hand', is just li ke fidak eecebira/ 'your big hand .' J
138. Thnayyan, 1927, p. 78. A host of examples may be found in I;lanafi. 1962, e.g. p. 79,/id e lmag$u~/ ' the amputated hand', p. 104. /het el'atig/ ' the old house'; p. 138. /ealb elmii6i/ 'the harmful dog' . and !11any more. 139. Sassoon, 1949, p. 187. The construction is part icularl y .common and product ive in the quasi-literary idiom of the Baghdadi Jews. I
140. Cf. the many examp les in Yahuda, 1906 , and Sassoon, 1949, pp. I95ff. 141. On the cl ose connection between definite articles and relat ive pronouns, cf. Feghali, 1928, pp. 90- 100,190, 310 ; Brockelmann, 1913.pp. 536ff.; as a problem in general linguistics , E. Benveniste, "~1. phrase relativc-probleme de syntaxe generale," BSL 53: 39c 54 (1958). 142. Cf. l;Ianafi , 1962, p. 145, /\:Ijart e lmate'ejbak/ ... 'a slone you don't like' the equivalent of the 1 proverb cited ; l;Ianan's collection contains many more M proverbs illustrating this ·const ruction: p. · 82, Ibab ettcjik menha rib/, ' the door through which wind comes in'; p. 105. Ibir eUeSrab menha/'a well from which you drink'; p. 138, /talb elyenbal)/ 'a barking dog' ; p. 138. /celmat ettestelJ i menha/ 'a word you're ashamed o f' . etc. Fo r Kwerd in the Musayyab district. a geletdialect , cf. /~ fl:lat elli mabiha qcb la/ ·a side not facing towards Mecca' (Meissner,
.~
•
.
198
NOTES
COMMUNAL DIALECTS IN BAGHDAD
1903b, p. xxvi); for MardIn, a qe/t/tMdialect, cL /ani larai:tet laqulti/ ... 'where is the carpet which you said .. .' (Socin, 1882, p. 260). 143. Brockelmann, 1913, p. 208; Feghali, 1928, pp. 134-36, pp. 211- 12; Bauer, 1913, par. 70b. 144. Tsereteli , 1956 , is replete With examples, e.g. i had qatalu 'he killed him' i .salldiiq kasaru 'he broke the chest', zaraba i billl 'he hit the girl'; cf. also Tsereteli. 1941 , pp. 143ff., and Fischer, 1961 , pp. 262-63. 145. Fcghali, 1928, pp. 298, 332- 33. 146. Brockelmann , 1913 , p. 316. 147. For Lebanon , Feghali, 1928, p. 362; for Mosul. Calabi, 1935, p.S, Dabbag, 1956, p. 15; (or Jewish Baghdadi, Schramm . 1954, p. 110 ; on Baghdadi in genera l •. Ganima, 1926, p. 586. On Middle Arabic Jewish texts from Iraq, er. Blau 1961a, p. 215 , and 1961b, pp. 172, 179.
148. AI-Wat:aidi (d. J075) in his Sarb Dlwdfl al-MIlIQllabbi, ed. Dieterici . (Berlin , J 861- 64), p. 384, line 3: "wa-'i1btiq 'aI-lam bit-mart'll qabT11 jiddan wa-6iilik mill fah 'al-bagiladiyin"; Brockelmann, 1913, p. 316, thinks these words are a continuation of the immediately preceding sentence from Ibn Furraja quoted by WabidI; if so , the statement is from the tenth rather than from the eleventh century. 149. lowe this distinction to some classroom remarks of H. J. Po!otsky's concerning similar constructions in Syr iac and Geez; cf. his review of R. Schneider's L'expressioll des complbnellts de verbe et de nom et la place de ['adjecti! epilhete eu Cueze (paris, 1959) in JSS 6:251- 56 (1961). 150. It is perhaps even commoner in I raq that a person addressed as labu salmanl should not be the father of Isalman/, because of the custom of using. a n " automatic kunya" for persons having certain given names: anyone oalled Idft.wo.d/, whether he has children or not , is likely to be addressed as labu salmanl jllst as anyone called Isa lmanl is likely to be addressed as labu dawud/. Son~e other fixed (and reversible) pairs: Isukurl or Isakerl is labu mabmiid/, lyOsef/ ls labu ya'gOb/, Imbammadl is labu jasem/, labmadl is labu shab/, etc. This is said to be most prevalent among Muslims, but occurs among Jews and Christians as well. cr. !:lanafl, 1963, s.v. !abb!. 151. Brockelmann, 1913, pp. 34. 237,243; Feghali, 1928 , p. 363.This constru~ tion as well as the types lsafu leflanl and Iqallu leflan/, are att ributed to AramaiC inft~ence by these authors and those mentioned in note 147, but somewhat similar constructions in North African dialects are thought by Fischer, 1907, and p:h. Man;ais, 1956, pp. 413, 611, to be due to Berber influence; these occur with kinship terms on ly. 152. Tsereteli, 1941, p. 145, and for examples , Tsereteli, 1956,passim.
1 199
153. Cf. lJAL 21: 121- 37 (l955); and AI-Toma, 1957, who uses this list for a lex ical c~rnparison of classical Arabic and Iraqi. I have not used his entries where they seemed to deviate from M and reflect another Iraqi colloquial, perhaps that or Kafbala. 154. C.A. Ferguson and M. Said, Lexical Voriallll' ill Arabic Dialects, 1958, 18 pp. (muJt \lith). 154a. However, Malaika, 1963, p. 26, has the regular cognate M/makensa/. 155. An interesting historical and comparative study of the demonstrat ive constructions in the Arabic dialect , based on the ava il able literature, may be found in Fischer, 1959. 156. Fischer, 1961, p. 259. 157. The vowel of J/keml is probably nothing'but a lengthening of the original J as in MC would have been (k£ITL], with the characteristically Baghdadi mid or lower mid vowel; it is this vowel, lengthened , t hat we have in Ikern/. As for the lengthening itself, it has parallels jn Cairo Ikarnl and elsewhere .
Ikam/, which in
158. On C /emti/ , cf. 3.360 (iii) and note 71. One wonders whether the initial syllab le of M/yamta/ [yemt£] could not have developed from an original /cmtal (i em1E;] through stabilizat ion and phonell1iciza~ion of t he on~glide (cf. 3.31e (iii]). A more certain instance of sllch a shift can be seen in IbOfya/ 'buffet'; the nonoccurrence of final unstressed /el in the dialect caused something like -[biW£) to be resolved into [bOJYf:]. which is more in accordance with existing patterns, cf. !barnya! 'okra'. 159. Positing - [hawniki] as the basis of both the J and C forms, the·J development would look something li ke this : [Ihawniki] 'thither' [Iawnik i) [I~wnikil and [Iuwniki], which is' reinterpreted as /11 plus Iwniki/, just as Iluwl5.dl is III plus Iwladl 'the ch ild ren'; a simi lar deve lopment took place in Iwlani/ 'first', ultimately from lawwaian i/. cf. 4.6. That the initial phoneme is Iwl and not luI (o n, the instabi lity o~ the dist inction , see 3.21e) is confirmed by the "Judea-li terary" version of t his word. This idiom replaces initial /KKI by I KeK/, e.g. IbeUidl for Iblad/, and Iweladl for Iwlad /, and similarly Iwenik i Ifor (wnik i/. 160. JavJaliqi, ed. 1875, p. 139: Jllvll' ; Hoem!rbach, 1956, p. 169, 1.9. : Jui ni.
,
161. C/hekkil « - /hekkil < - !hekdi/?) may perhaps be compared with Moroccan Ihakkakl and with a different sort of ass imilat i~n, Cent. Pal. village hMOal (Bauer, Wort., s. v. so). The Ihc/- or /hi/- initia l of most d ialects does suggest an OA etymon -/ha k iM /, cr. J. Greenberg, Wd. 17: 64 (1961); as do the forms with in itial /ke/-, Iki/- ; Mosul/ hakc()1 poses a problem, however, as one would have expected 'imala.
200
COMMUNAL DIALECTS IN BAGHDAD
162. Cf. the Iii o f M/ hlci/, Qa " at ~a 'e~ Ih.MIiI (Van Wagoner, 1944, p. 139; repeated in his 1949 course ostensibly as Baghdadi) in dialect s with no IiI a llomorph of the fem. ending; and even the IiI ofthe J forms does not conform to t he morphophonemics of the fern. cnding (cf. 4.3). 163. The study of Aramaic elements in Mesul Arabic (Dlabi. 1935, does list a score of items , largely connected with religion, that are in usc among Christ ians only. Some of these, incidentally, arc probably not of Aramaic origin at all (d. notc 126 above) but they arc nevertheless of interest in that they contribute to our understanding of communal differentiation in Mesu!. . 164. The dist inction was initiated by M. Weinreich, 1954. pp. 85- 86 J 65. There is as yet no detailed study of the tradit ional pronunication of Hebrew used by the Ba ghdadi Jews. This section is therefore based on my own preliminary observations and on some details kindly passed on to me by S. M~rag and A. Levin. The latt er, who did a seminar paper on the Hebrew elements to J in 1962, also supplemented the Hebrew lexical items listed here.' ~or w?ich I am especialiy indebted t o D. Khazzum, a J speaker. For some addlllo~allt~~s an~ their occas iona l use in M , see now l;Ianafi, 1963, esp. s.v. lalmanal widow, lawir/ 'ai r', l(e)~baI:t6al ' hy mn s'. 166. The rhymed phrase may echo the Biblical ,!JPf!1 wa(ulfuy rOl, 'mournful and bareheaded' (Esther 6: 12); at any rate, Il:teffcll has no meaning by itself. 167. Cantineau, 1939, p. 85. 168, Cf. the va riant given by l;Janafi, 1962, p. 42, no. 11 6: leSqal qalbak sason mentaqqat eHeqqaqa/ 'what did your heart say, S., when the ~un went off', which has M/(e)sl fo r J/asl and M/atl in the verb fo r J/et/, and a different mock J form for gun. The given name Isasonl is as typically J as Il:lesqel/.
169. I am indebted to U. Heyd for this suggesti~n as to the etymology of the Turk ish term .
I
170. The exp ression M/xo majabaw esmi?j 'I hope they didn't mention my name?' (l::Ianafi, 1962, p. 169, no. 759) is proverbiall y put in the mouth of a man wishing to insinuate he might be included among certa in persons wanted by the pol ice or feared by the authorities. '
J7 J. This Ixosl is simila rl y dev iant (i.e. precedes noun, is invariable) in the Nco-Aramaic dialects of the area, e.g. Zakho. Turkish influences are fa r less ~oll1mon in those dialects than in Baghd'ad, so that the so lution must be looked for elsewhere.
172, Barthelemy . Diel. , S.y. darbakke, gives dumburko for "Mesopotamia," i.e. Upper Iraq (or the Syrian Jazira) and dumbug for "I raq septent.rional.''' i.e. Northern Lower Iraq, viz. what we would today call Central Iraq, mcludmg of
NOTES
201
course Bag~dad. On the terminology "Iraq" for Lower Iraq and "Mesopotamia" for Upper Iraq, see Blanc, 1962, p. 52, fn. I. 173. The:. J vs.
Me
split in usage is
eviden~iy
more complex, for Fa rmAn
19~9.' p ..11, ~oe.s have Igamaw y'aynun 'aJayya/ 'they started looking at (or ogling) me In hl~ genumel.y M dialogues, though he regularly uses lsafl and Ibawa'j with the. mea~ mg ~esc flbed here. Ferguson, 1959, p, 629, is only partially right when he 11llp iJe.o; that OA/ra'al has eve rywhere been replaced by Isarl' in addition' to Qarabas lara), l::Ianafi's M proverbs often have M/re'al for ' to se;' e.g. p. 63, no. 226 ljre'akl 'he saw you') and Inre'aj for 'to be seen' (e.g. p. 58, no. 195). 174. The Turkish dictionaries seem to list no such form, but 1 note that M. Ben-Cheneb, Nfols lures el persons cOl/serves dans Ie parler arabe d'Alger (Algiers, 1922), p. 57., gives as etymon for Alg. 1awla "turc lavla, IQvli , grec lavli,"
175: Op~nheim, 1952, p. 478, classifies *~ Beduin (tribally organized) PO!,~lat lon of Iraq into four categories, called respect ively /baduj (camel herders), lsawlyal (s heep and goat herders), Ifelel:tl (peasants, viz. sedentarized tribesmen), and Ime'danl (water buffal o he rders). He gives I'arabl as a synonym for the second group, but in Baghdadi usage, so fa r as I can tell, the term covers the Ifeleh l as well ; certa inly j'erbij does. Similarly, Ime'danl covers loosely aU marsh dwellers. In Talmudic Aramaic, 'arbii'e des ignates the sedentary Arab ic-speaking population of Babylonia, the nomads being referred to as toyyii'e, cf. SYi'ia~ rayyti'e and the Arab triba illame Tayyi' (Obermeyer, 1929, p. 235). 176. Fischer, 1961 , p: 236 , fn. 1, derives this from Cl. Ar. j-d-y 0;1 t he basis of the Iraqi rura l fo rm Ijaddal cited by Meissner 1903b Glossar. Hanafi 1962 twice cites proverbs containing the form Ijedyat" bcgga;y' (p. 18, 'p. 130) and seems to imp ly a similar derivation. Surely the existence of the forms in Igi and the common affrication of Igl t o fjf (cf. 3.26a) prec lude such a der ivation. . . 177: The Questi?n of the composition and or igin of the Baghdadi Christian population has receIVed but scanty attent ion, My informants are a ll aware of "large" numbers of fairl y recent immigrants from t he Nort h some of them Aramaic speaking. The name ItelkH/, sing. Itelkcfi/. derived fr.~m an Aramaicspea.ki~g Christian vi ll age near Mesul, is loosely used by J and M ' speakers for Chnsttans as a whole. However, the features characterist ic of C do not date from this ce~tury (cf. Oussani , 1901; Ganima, 1906), nor are they attr ibutnblc 10
..
178. Sassoon, 1949, gives documentary evidence for the continuous existence of a Jewi~h community in Baghdad from its foundation to the present', except for a peflod from the middle of the fou rteenth to the middle of the sixteenth century (pp. 100- 101). 1 know of no reports of large-scale migrations of Jews from ot her regions,to Baghdad, though among ind ividuals I have known surnames or ;,
202
COMMlJNAL DIALECTS IN BAGHDAD
other evidence occas ionally indicate Persian or Aleppo origin; I cannot corroborate Peterman's supposition (1820, II, p. 281), quoted by Coke, 1927, p. 262. that most o f the oldest famili es of Baghdadi Jewry emigrated from 'Ana. For a useful historical summary. see now G. Vajda, "Le Milieu Juif a Bagdad", Arabica 9.389393 ( 1962). 179. The Arabization of the Iraqi Jews must have been well-nigh complete by the tenth century, if on ly because it was then that Sa'adya wrote his Arabic translation of the Old Testament. Traces of Aramaic speech (not restricted to Jews) can st ill be found in the e leventh century (Epstein, 1915, pp. 53ff.). 180. [ find no reports of communal differentiation in Baghdad before the nineteenth century (cf. note 2). Flick, 1955. pp. 87- 88, thinks the minorities did, on the whole, speak Arabic like the major ity; on the other hand, his statement that the Jews of MedIna spoke, in the seventh century, an "un inte lligible" dialect of t heir own (ibid.), repeated by Blau . 1961b. p. II , fn. 26,is surely somewhat hasty. It is based on a single s tatement reporting that someone was able to jabber bi-Iyahiidiya (ibid. , p. 88, fn . 1). On the term '''Kal 'al-yalllid referring to unspecified peculiarit ies of the Jews' speech, andlor to Hebraisms in their Arabic, cf. Goldziher, 1872, p. 755, fn. 5; 1901 , p. 14 and fn. 4; the latter reference also cites a single specific feature said to be characteristic of the speech of the MedIna Jews, viz. ItI for 19/. Weinre ich , 1954, p. 83, fn. 27. apparently bases his assumption of an early differentiation of Jewish dialects with in Arabic entirely on these reports of Goldziher's. 181 . Cf. Weulersse, 1946 , p. 64. The figures for the population of Iraq given in Docum entation Franltaise, 1951, pp. 6-8 (estimated for 1930) show approximately 55 percent non-'Sedentaries vs. 45 percent sedentaries, but many village dwellers are in various stages of sedentarization , and very few townsmen (estimated at 5 percent) are truly urbanized. The preponderance of semi-sedentary and recently , sedentarized populations is considerably greater in Lower Iraq thap in Upper Iraq. These figures, as G. Baer pointed out to me , are only rough estimates at best ; they are based on Dowson , 1932, p. 12, where the author states that they are "the best estimates that the /iwa authorities were able to furnish, and the censUS . was not in a position to give me any better material." For a list and account of sedentarized groups in Lower Iraq, cf. OpvCnheim. 1952; for the different situation in Upper Iraq , cf. Oppenheim, 1939. A detailed history of the sedentarization of o ne large group is given in SalIm, 1956, pp. 197-248. Cf. also note 175 above. 182. Oppenheim , 1952, pp. 428ff., reports that the towns of Sfik al-Suyfix and al-Xamisiya in Southern Iraq were, in the nineteenth century, partly peopled by Najada (anparentl y immigrants from Najd). 183. C f. the discuss ion in 2.23 ·and 2.24 and note 21. 183a. Cf. notes 97a, 98a, 106. Malaika , I963, whose variety of M is of the " less Beduinized" type, has one or two forms that may represent this sort of conservatism : Iluxxil as a variant of Iluxl 'other (f.)' (pp. 13, 70) is unexpected
203
NOTES
for M , where the form s so far noted a re Ilux/. Iluxra/. but cf. JC/lexxi/, with the no~m~l ~C " i!lIala and Ixxl < Ix&!
I
184. The main works consulted were Coke, 1927 ; Lewis, 1950 ; Longrigg, 1925, 1953; Obermeyer, 1929; Opp«nheim, 1939, 1952; Streck, 1900, 1901; and the relevant articles of the Ellcyclopaedia of Islam ; especia lly Duri, 1960. . IS5. On tho re-Beduinization of Iraq in the eleventh and the thirteent h centuries , cf. Oppenheim, 1952, pp. 185- 18S. IS6. It is possi ble that the ruin of the flood cont rol and irrigation system , on which Lower Iraqi agriculture depends, and which in turn depends on stable ~dmi?ist ration for regulation and maintenance, actually p receded the Mongol I~vaslons about a century; an interesting hypOthesis connecting that ruin with Silt depos its and the inabil ity' of a weak administrat ion to cl ear them , may be found 1Il Jacobsen-Adams, 1958, espec ially pp. 1257- 125S. lowe this reference to M. Brosi.
br
IS7. Cf. Longrigg, 1925, p. 13, where the author summarizes the effect of the Mongo l invasions on the course of Iraqi history ; fo r a brief b\lt exCellent summary of this effect, cf. Lewis, 1950, pp. 154, 158- 159. Cf. a lso Oppenheim , 1900, p . 356; 1952, pp. 187- IS8. On the downfall and in some cases the disappea rance o f towns from the 15th century onward, cf. Coke, J927, pp. 182- IS3; Longrigg, 1925, pp. 2- 3; Lew is, 1950, pp. 154, 15S- 159. A check through the articles summarizing the history of the various towns in the Encyc/opaedia ofblam shows almost uniformly the same picture of decay and depopulation between the fifteenth and the nineteenth centuries. 188. Longrigg, 1925, pp. 120-128, 154-162,200--211 ; Coke, 1927, pp. 194ff., 230ff. ; and Oppenheim, 1952, pp. 415ff.• are replete with accounts of the struggles . between tht pashas and the Beduins. 189. The process of re-urbanization, which is still going on (cf. note l SI a nd note 9), can only be pieced together from various scattered notes and figures. In the s ixteenth century, Baghdad was "a large carav.an center'~ (Longrigg. 1925, p. 30) had an estimated population o f 14,000 in the seventeenth century (Streck, art. "Baghdad" in Ellcycl. of lsI. I), of 40,000 to 100,000 in the early nineteenth century (Duri , 1960), and today num be rs over 7S0,000 (th is and a ll present-day figures. for which I am indebted to G. Baer, a re rrom the Ira.q Statistical Abstract 1959, Bag hd~d , 1960). In the sixteenth century , Hi ll a was "a large tribal market and center or exchange" as it has been a lmost to the present (Longrigg, ibid.) and Basra consisted mostly of reed huts "whose owners were but lightly tied to city life" (ihid.); it was rebui lt , though not on its ancient site, in t he e ighteenth century (Pellat. a rt. " B~ra" in EIIcyc/. of 1sf.2) and today numbers ove r 164,000 inhabitants.
204
COMMUNAL DIALECTS IN BAGHDAD
Kut was non-existent in the fifteenth century (Longrigg, 1925. p. 2), was a "miserable little village" down to 1860 (Kremers, art. "Kut" in Elleyc/. of lsi. 1), now numbers 26,500. Amara was built in 1862 (Longrigg, a rt. "Amara" in Elleyc/, 0/ Isl.2), now numbers 53,300. Nas iriya, present population 39,000, was founded about 1870 (Oppenhe im , 1952. p. 204); al-Xamisiya was founded about 1880 (ibid., p. 428) and Satra seems to have been founded in 1778 (ibid.). The literary revival reported in eighteenth cent ury Baghdad by G ibb, art. "Arabic literature", Elleyc/. of Isf.2 , p. 596, may also be con nected with the fe- urban ization process. 190, On the contrastive topography of Upper and Lower Iraq cr. Documentation Fran9aise, 1951 , pp. 3- 5; Dowson, 1932, p. II. On the stability vs. instability of towns in the two areas, cf. the disappearance of Lower fraqi towns such as Wasit, 'al-Mada' in , l;Iulwan, Qadisiya, a.nd many others that can be found e.g. in Streck, 1900, pp. IOff.; and compare this to the relative stability of the Upper fraqi towns whose history is summarized in the El/cyclopaedia of Islam or in Musil, 1927, pp. 345, 350, 353, 363; 'al-'Anbar, which was on the boundary between Upper and Lower Iraq near present-day Falliija, is ruined (M usil, ibid., p. 353). Naturall y, matters must not be oversimplified: Mosul, in Upper Iraq , has had a rather turbulent history, while Hilla in Lower fraq has had a rather stable o ne, though the core of its population seems to have been Beduin since its foundation.
19 1. The socio-li ngu istic data and hypotheses set forth in this work thus tend to bear out an opinion exp ressed twent y years ago by H. A. R. Gibb: " ft is the contin ua l recruitment from fresh nomadic stock-operating of course in a series of stages-that has preserved the Arab population of the settled areas in the Middle East from the ph~sical degeneration and decay to which the populations in its lowland areas have been especially liable. It is reasonably certain that no people could halle come through the long centuries of economic impoverishment and enderuic disease which have been the fate of the Eastern lands, and yet have retained such a measure of vit ality, had it not been able to draw fresh powers of resistance from it s desert reservoir." (G ibb , 1942, p.. 88).