Text
                    Bukhara Arabic: A Metatypized Dialect of
Arabic in Central Asia
Robert R. Ratcliffe

General Issues
The obscure Arabic dialects spoken in Bukhara state of the Uzbek Republic
are known to scholars primarily through the pioneering work of Vinnikov and
Tsereteli, especially Vinnikov (1969), a collection of narrative texts collected
in 1936, 1938, and 1943 in the two villages of Djogari and Arabkhane. The
dialects have generated considerable interest among Arabic dialectologists,
and have been investigated from the point of view of dialectology and classical historical (genetic) linguistics (Fischer 1961, Fischer and Jastrow 1980,
Jastrow 1997a, 1998, Versteegh 1997). But the materials are of contemporary
interest to a broader linguistic audience because of what they reveal about language contact and syntactic typology. While the dialects are conservative in
their phonology and lexicon, they are radically different from other Arabic
dialects (and correspondingly similar to the surrounding Persian and Turkic
languages) in their syntax and to a lesser extent their morphology. They are so
different indeed as to constitute a case of ‘metatypy’ in the sense of Ross
(1996), that is, change of syntactic type. Furthermore they show a configuration of word order properties (specifically RelN alongside NG and NA) which
is apparently quite rare (possibly otherwise non-existent) among world languages, to judge by its lack of attestation in the 149 language sample of
Hawkins (1983).
The aim of this article is to delineate, on the basis of an analysis of Vinnikov’s texts, those morphological and syntactic features of the dialect which
are of most interest from the point of view of typology and contact studies, in
short to bring the data from this dialect to bear on such larger questions as the
following:
— To what extent can specific structural patterns in or among languages be correlated with specific types of contact situations?
— Are there any limitations on what aspects of a language can change due to contact with another language?
— How are reversals of normal word order patterns implemented in the course of a
change of type?

The analysis is based on the texts from Arabkhane. Reference numbers are the
number of the text in Vinnikov’s collection, followed by the line number.
Vinnikov’s transcription is maintained.

1


Phonology and Lexicon The dialects are conservative in phonology, preserving the ‘back’ consonants q, ©, x, ?, and V ( k¢, h¢, æ, g in Vinnikov’s transcription), as well as the ‘emphatic’ (pharyngealized) † and s¢. (One scholar has suggested, however, that the latter is only an appearance due to Vinnikov’s conservative transcription (Jastrow 1997)). One phonological innovation likely due to areal influence is the shift /aa/ > /OO/ kon ‘he was’ < Classical Arabic (henceforth CA) kaana (Tsereteli 1970). The lexicon too appears to be overwhelmingly Arabic in origin, although the texts contain a handful of words of Persian, Turkic, or Russian origin. A count of one text chosen at random turned up only ten percent vocabulary (tokens) of demonstrably non-Arabic or unknown origin. Some examples of foreign vocabulary: (1) bozor sartaros# parc#a amlokdor ‘market’ ‘barber’ ‘piece’ ‘government official’ < Persian < Uzbek < Persian < Turkic < Ar. /amlaak, pl. of mulk ‘property’ + Persian daar ‘holder’ The dialects are genetically related to the dialects of Iraq (Jastrow 1998) with which they share certain innovative lexical and phonological features: — indefinite article /fad/ < CA fard ‘individual’ — shift of interdental fricatives /T/, /D/, /D¢/ to sibilants /s/, /z/, /z/. This shift is found elsewhere only in Northern Iraq. The other dialects have either maintained these sounds (Arabian peninsula dialects) or shifted them to dental stops /t/, /d/, /d¢/ (all other central dialects). (2) Some lexemes (for Iraqi see Woodhead and Beene 1967) ‘to give’ ‘to ask’ ‘head’ ‘kid’ ‘with’ Iraqi ni†a nis#ad kalla saxal ya Most other dialects /?a†a/ < CA //a?†aa/ < CA /naSada/ ‘to seek’ CA /sa/ala/ CA ra/s CA jady CA ma?a Morphology and Syntax In spite of its conservatism in phonology and lexicon, Bukhara Arabic is radically different from CA or the central Arabic dialects in its syntax and (to a lesser extent) its morphology. Where syntactic and morpho-syntactic structures differ from those of other Arabic dialects, they usually match closely 2
with structures found in one of both of the other languages spoken in Bukhara state, namely Uzbek (a Turkic language) and Tajik (an Indo-European, specifically Iranian language very close to standard Persian). Basic Word Orders The table below indicates basic word orders with less frequently attested word orders in parentheses. Bold text indicates identity with the Bukhara Arabic pattern, plain text difference from it. (3) Other Arabics VSO(SVO) NA NG Prep NRel Bukhara Arabic SOV NA (AN) NG (GN) Prep RelN Tajik SOV NA (AN) NG Prep NRel (RelN) Uzbek SOV AN GN Postp RelN This configuration of word order properties (specifically RelN alongside NG and NA) is apparently quite rare (possibly otherwise non-existent) among world languages (Frederick Newmeyer, personal communication). It is not attested in the 149 language sample of Hawkins (1983), although Hawkins’ information on relative clause/noun order is incomplete (reflecting the fact that it was not included in Greenberg’s (1966) article on word-order typology which inaugurated this line of research ). According to the first four word order properties (SOV/Pr/NG/NA) Bukhara Arabic is Greenberg type 17. Hawkins indicates three other Afroasiatic languages, Iraqw, Akkadian, and Neo-Aramaic, as well as Persian and Tajik as belonging to this type. All five are NRel, unlike Bukhara Arabic (for Akkadian see Huehnergard 1997:185 #ff., for Neo Aramaic, Krotkoff 1982: 56–7, for Iraqw, Mous 1993: 277). Basic Clause Order Intransitive verbs always show a simple SV pattern: (4) fat faqir kon a poorman was ‘There was a poorman.’ (56, 1) (5) †ole? rose fat adami min giddamu a man from before-him ‘A man rose up before him.’ (51, 9) 3
This is in contrast with other Arabic dialects, where VS is the unmarked order: Cf. Iraqi (Baghdad Muslim) (Fischer-Jastrow 1980: 155): (6) aku fad wa©id faqir was a one poorman ‘There was a very poor man.’ jiddan very Cf. Cairo (Fischer and Jastrow 1980: 230): (7) kan fi marra wa©id was at time one ‘There was once a barber.’ mizayyin barber For transitive verbs the simple (S)OV pattern is occasionally found (8) fat ?ud xada a stick took ‘He took a stick.’ (59, 27) (9) dabba ijir zarab horse leg struck ‘The horse thrashed its legs.’ (59, 31) But much more commonly transitive verbs also have an encliticized pronoun referring back to the nominal object or objects: SOV-o, SOOV-oo (10) sakina xada-ha knife (he) took-it ‘He took a knife.’ (59, 30) (11) xilaf li-gidday xubza an†-u-a then to-beggar bread gave-him-it ‘Then he gave the beggar a piece of bread.’ (56, 6) This type of construction is by no means alien to other forms of Arabic, where a word can be topicalized by being moved to the first position of a sentence, 4
with its syntactic role indicated by a resumptive pronoun. For object-topic sentences Brustad (2000: 349) gives examples like the following from Egyptian Arabic: (12) il-fustan the-dress ‘I got the dress.’ gibt-u I-got-it On the surface this Egyptian example is structurally identical to 59,30, above. Apparently what has happened in Bukhara Arabic or is happening is the reanalysis of a resumptive pronoun as a verbal inflection agreeing with the object. Whether or not such a reanalysis has been internalized by speakers is perhaps impossible to tell. The important point from a historical linguistic point of view is that shift from VO to OV is not immediate, but is mediated through a variant word-order pattern available to the language in the stage where VO is the unmarked order. Adjective-Noun The adjective generally follows the noun (as in CA), except that the adjective kasir ‘many’, ‘much’ precedes the noun (which is in the singular): (13) kasir sana kasir adami ‘many years’ (43, 5) ‘many men’ (43, 20) In Tajik too the normal order is NA, though determiners of quantity, superlatives, etc. may come before the noun (Rastorgueva 1992: 94–5). When the noun precedes the adjective it is linked to it by a particle /-in/ which is formally identical to the Classical Arabic indefinite genitive suffix (more exactly a genitive suffix -i, plus an indefinite suffix -n) but which functions like the Persian/Tajik ezafe (cf. Fischer and Jastrow 1980: 96). The adjective agrees with the noun in gender. (14) ?eys#-in aswad rice-N black ‘black rice’ (at a funeral ceremony) (40, 2) (15) ba©ar-in kabir river-N big.MASC ‘a big river’ (51, 20) 5
(16) xubzat-in zina bread-N good.FEM ‘a good piece of bread’ (56, 24) (17) xubzat-in dawonoka bread- N poisoned.FEM ‘a poisened piece of bread’ (56, 23) Cf. Tajik (18) kitOb-i xondagi book-EZ read ‘a read book, a book which has been read’ (Rastorgueva 1992:39) Genitive Construct This is generally of the Arabic type with a simple juxtaposition of N and G, or with pronouns linked by the preposition /ila/ ‘to’: (19) mazar graveyard ?arabin [of the] Arabs (40, 1) (20) raqabet neck dabba [of the] horse (59, 30) (21) wald son amir [of the] prince (56, 9) (22) mart wife ila-y to-me (57, 4) Turkic type constructs GN-pro, with a resumptive pronoun are also rarely found: (23) amir prince wald-u son-his 6
‘the prince’s son’ (56, 8) (24) s#ib?an qus#-u oldman hawk-his ‘the old-man’s hawk’ (59, 5) (25) ilay ism-i to-me name-my ‘my name’ (51, 12) Cf. Uzbek (Sjoberg1963: 140–141): (26) qiz Ona-si girl mother-her ‘girl’s mother’ (27) men-iN bir I-GEN a ‘friend of mine’ dost-im friend-POSS1SG Adpositions Adpositions are almost exclusively prepositions, as in all forms Arabic. One example of a complex circumfix was found: (28) fi beyt-u adrun-u in house-his inside-its ‘inside his house’ (53, 22) Relative Clause Lengthy modifier clauses are headed by a form of the Arabic active participle preceding the modified noun. Not only the word order but also this usage of the participle is quite foreign to other forms of Arabic. As yet unexplained elements sometimes intrude between the participle and the modified noun As noted, it is this feature which makes Bukhara Arabic typologically unusual. (29) Iskandar muqul-un fad amir kon Alexander saying-PL(?) one prince was ‘There was a prince (whom they) called Alexander.’ (43, 1) 7
Cf. Turkish, traditional story-telling style (Bernt Brendemoen, personal communication) (30) Iskandar Alexander derler say-PL bir one amir-di prince-was The normal Arabic dialectal structure for this expression would be something like: (31) kaan was (fii) (exist) /amiir prince ismuh name-his Iskandar Alexander (32) fi ide-h woquf qus#-u il-dabba zarabu in hands-his standing hawk-his against-horse he struck ‘His hawk which was standing (i.e.sitting) on his arm, he struck against the horse.’ (59, 5) (33) min nahar Vodi-yam s#ajar kon from river going-(?) grass was ‘There was grass which extended (lit. went) from the river.’ (53, 5) The normal pattern in Uzbek is the same (Ismatulla 1995, p. 430): (34) kitåb oqiyåt-gan yigit özbek tilini book read-PART youth Uzbek language-POSS-ACC ‘The young man who is reading a book knows Uzbek.’ biladi knows In Tajik also relative verbs with -gi precede the modified noun (Rastorgueva 1992: 95): (35) Omada-gi Odam came-PART man ‘the man who came’ Other Complex Clause Structures An infinitive formed with suffix -(a)han (see below for discussion of the possible origin) has a wide range of syntactic roles. In Classical and dialectal 8
Arabic, verbal nouns exists but have a very restricted set of syntactic functions, and are morphologically unproductive. Infinitive subject constructions The infinitive or a clause governed by the infinitive may function syntactically as a noun phrase, hence may be the subject of a sentence: (36) fi wa©t Iskandar mit-ahan ma in time Alexandar die-INF no ‘In the time of Alexandar there was no death.’ (43, 1) kon was (37) pos#o il-A©mat-Tura bint-u on†-ahan fi qalb-u ma kayin Pasha to-Ahmat-Tura daughter-HIS give-INF in heart-his not being.ACT.PART ‘The Pasha did not have it in mind to give his daughter to Ahmat Tura.’ (51, 3) Infinitive plus attached pronoun in indirect speech and temporal clause This is a characteristic Turkic feature, found also in Tajik (Rastorgueva 1992: 97–99), but completely alien to other forms of Arabic. (39) ?anza min xaro-h ?áarfa furux-a ?al kal-ahan-u goat from excrement-his knew.ACT.PART offspring-hers about eat-INF-his ‘The goat knew from his excrement that he had eaten her offspring.’ (50, 23) (40) is#tu te?rifum hat harami kayn-ahan-u how you-know this thief being-INF- his ‘How do you know that this one is the thief?’ (52, 21) (41) hat fi tigara †olahan-u ras kon ?end-u-mi, ma kon ?end-u-mi this in poplar climb-INF-HIS head was to-him-Q not was to-him- Q ‘When this (man) climbed up the poplar, did he have a head or not? (54, 10) (42) wa!©t-in s#i kal-ahan-u, mana-hu kom time-N(?) something eat-INF-his chin-his was ‘When he ate something, his chin was moving.’ (54, 13) mifta!raq moving.ACT.PART (43) ana hat I these xaruf-at story-PL wa©t-in time-N mit?ilm-ahan-i learn-INF-my 9 fat a ?a#Ser ten xams five kunt I-was
‘I was a (boy of) fifteen when I learned these stories.’ (67, n.) Infinitive temporal clauses (44) kasir darb sayr-ahan xilaf many road go-INF after ‘After travelling many roads…’ (51, 14) In Uzbek a similar construction exists using the converb based on the participle in -gan (Sjoberg 155): (45) bir paz otir-gan-dan a while sit-PART-ABL ‘after sitting a while’ keyin after (46) mihmOn kel-gan guests come-PART ‘when the guests came’ waqt-ta time-LOC Serial Verb Constructions Coordinate When a series of coordinate verbs occurs in sequence, no conjunction intervenes. (47) xilaf ?ayyan: iyetimu mou, †oleh, hareb, Vodi after he-saw farmhand not-he rising escaping going ‘Then he saw: the farmhand was not there, he had gotten up, run away and gone.’ (59, 33) (48) xilaf mahrutat-u laqa† xada then pieces-(of)-it gathered took ‘Then he gathered up the pieces of it.’ (59, 16) (49) suxul ?anza min ba†en hays#a bi-ki-ki-ki sowat, †ol?at kids nannygoat from belly wolf bikikiki doing.FEM.PL rosing up.FEM.PL ‘Saying (lit. doing) bikikiki the nannygoat’s kids came out of the wolf’s belly.’ (50, 15) (50) fi in mu water farru, jumped Vadu, went Varq drowning 10 s¢oru. became
‘They jumped in the water, went down, and drowned.’ (43, 13) Subordinate A commonly found compound verb structure consists of the active participle of the verb ‘to stand’/woqf/ (CA waaqif) dominating a finite verb in the present/future: (51) hinti is# mi-s-sin you.FEM what PRES/FUT-you.FEM-do ‘What are you doing?’ (59, 20) woqf-in-ki standing.ACT.PART-N-you.FEM (52) ana c#ipon iley-kum m-a-xit-u I coat to-you PRES/FUT-I-sew-it ‘I am sewing your coat.’ (59, 21) woqf-an-i standing-N-I According to Masica (1976: 155) this is a Central Asian areal feature. It is found in Tajik (istodan), though not in Persian, and more common in Uzbek (turmak) than in Turkish: (53) Tajik man maktub naviSta I letter writing.PART ‘I am writing a letter.’ (54) istoda-am standing-am Uzbek xat yåzib turib-man letter write-CONV stand- CONV-1SG ‘I am writing a letter.’ Morphological Peculiarities Common Arabic morphemes lost: — Unlike any of the core dialects, Bukhara Arabic has lost the definite article. — The conjunction /wa/ ‘and’ was not found in any of the texts I examined. Nouns are conjoined with /ya/ ‘with’. Verbs are simply strung together without a conjunction. I suspect that the use of different conjunction types for nouns and verbs is a typological characteristic of SOV languages, although the question needs research. Use of Arabic morphemes in a non-Arabic way —Use of the reflex of CA kaana (a copula verb required only in the past tense in CA) as an existential. 11
This usage is found otherwise only at the other end of the Arabic dialect continuum in Moroccan Arabic, under Berber or possibly Romance influence. —Use of active (present) participle as main verb with past tense meaning Both of these features are illustrated by the following example: (55) fat pos#o kayin a pasha being ‘There was a pasha.’ (51, 1) — Use of the active (present) participle plus -Vn suffix plus a possive suffix pronoun as a finite verb with present or present continuous sense. (56) ana ma ?arf-an-i I not knowing-N(?)-my ‘I don’t know.’ (59, 10) The same development has occurred in Eastern Neo-Aramaic (Jastrow 1997) under Kurdish influence. In both Persian and Turkic languages the clitic pronouns which mark verbal subject are similar to those which mark possessor. The CA possessor pronoun clitics are similar to object clitics but not to subject marking suffixes. My hypothesis is that the Turkic/Persian isomorphism has motivated the reanalysis of Arabic possessives as subject clitics (or at least as available for use in that role). Use of Clitic pronouns (57) Other Arabic Bukhara Arabic Persian Turkic Possession X X X X Subject — X X X Object X X X — — Use of an uninflected form of kon (CA kaana) plus m- prefix present/future in the sense of past progressive or past habitual: (58) hamat ?eys# kom m-isun, they.FEM food were PRES/FUT-making ‘They were making food and cooking it.’ (40, 26) kom were m-i†baxun-u PRES/FUT-cooking-it The CA pattern is to use the inflected past verb kaana plus an inflected 12
present. (59) hunna kun-na they.FEM were-FEM.PL ‘They were cooking.’ ya†bax-na cook-FEM.PL (60) hum kaan-uu they.MASC were-MASC.PL ‘They were cooking.’ ya†bax-uuna cook- MASC.PL This is possibly due to a tendency of SOV languages to inflect only the last verb in a sequence, although government should be right to left, not left to right as here. — Participial relative construction (see Relative clause) New morphemes and particles gained (61) From Tajik or Persian (ultimately): /ki/ /na/ particle introducing indirect speech (also in Turkic) verb negator (less common than the inherited Arabic /ma/, it appears in the texts examined only in the speech of female speakers) /ham/, /yam/ ‘also’ (also found in Iraqi Arabic) /hast/ present tense existential particle ‘there is’ (also in Iraqi Arabic) /agar/ ‘if’ (in addition to the inherited Arabic /hin/) (also in Turkic) (62) From Turkic: /mi/ /tar/ /c#i/ interogative marker copula ? (or Persian compartive suffix?) profession suffix (also in Iraqi and other Eastern dialects of Arabic) Examples: (63) duki qoyla ki that one.FEM said.ACT.PART that ‘She said ‘I don’t know either.’ (54, 13) ‘ana-yam I -too (64) qol ki xubza ?and-ak-mi HE said that bread to-you-Q ‘He said ‘Do you have any bread’ (56, 11) 13 na-m-a?rif’ no-PRES/FUT-I know
(65) ilay min dámmi te!?bir agar, m-a-qul to me from blood-my you-cross if PRES/FUT-I-say ‘If you spare my life (lit.blood), I will tell (you).’ (43, 30) (66) wa©edu qol ki jild-a a©mar-tar wa©edu qol ki adrun-a abyas-tar one said that skin-its red-COMP one said that inside-its white- COMP ‘One said ‘Its skin is more red’. One said ‘Its inside is more white’.’ (53, 25) Source unknown /-(a)han/ infinitive suffix (related to Tajik infinitive -an or Uzbek -g/k/qan?) Fischer (1961) interprets it as an based on the Arabic verbal noun pattern (one of several) fa?alaan, influenced by resemblance to Tajik infinitive -an, attached to feminine participle form ending in -a, hence requiring -h- in hiatus. /-ak(in)/ only used with past tense of ‘come and ‘go’ (completive aspect?) possibly Tajik -gi participle suffix + Arabic tanwin /m-/ general present/future prefix (related to E. Arabic /b-/ or Pers. /mi-/?/) /ta-/ jussive prefix (possibly contracted from Arabic ©attaa or Pers. taa) /-am/ participle suffix (?) Conclusion: Parameters of Contact Although it is not always traditionally defined so, historical linguistics can be defined as a discipline which attempts to correlate structural patterns within or among languages with the history of the communities speaking those languages. Traditionally historical linguistics has focused on only one such correlation: the genetic relationship. In this case systematic correspondences (in phonology and morphology) between languages make possible the inference that the languages had a common source. But genetic relationship is only one possible type of historical relationship which can obtain between languages, and there is no particular reason why it should be privileged. Rather than trying to strip away the residue of contact as a hindrance to discovering genetic relationships, the question which students of language contact should be asking is ‘is it possible to correlate specific types of structural similarity between languages with particular types of contact situations?’ If a positive answer can be given, then the study of language contact can aspire to the same degree of rigor as traditional genetic linguistics. As a rough outline, let me suggest the following: Contact situations can be defined along two parameters: direction of influence and degree of bilingualism. By direction of influence I mean the dichotomy of ‘shift’ and ‘borrowing’ as these terms are used by Thomason and Kaufman (1988). Speakers may either shift to a second language, in which case the first language is assumed 14
to influence how they speak the second language. Direction of influence is L1 to L2. Or speakers continue to speak their original language but their speech is influenced by a second language which they know: They ‘borrow’ from the second language (Direction of influence L2 to L1). Degree of bilingualism refers to the extent to which the whole speech community commands the second language. This parameter is clearly a continuum rather than a dichotomy. Within these parameters four cardinal points can be described. (67) Level of L2 knowledge maximal ⇑ ⇓ minimal Language Contact situations Direction of Influence L1 > L2 (‘shift’) L1< L2 (‘borrowing’) ‘substrate influence’ ? ‘pidgin/creole’ ‘adstrate influence’ ‘cultural borrowing’ Maximal bilingualism with shift would describe the classic substrate situation, where a population (usually a politically weak one) over the course of centuries shifts to another language (usually that of a politically dominant group): Celtic to Romance or Germanic in NW Europe; Berber, Coptic, Aramaic to Arabic in the Middle East and North Africa. The expected linguistic result is primarily in phonology, particularly in suprasegmental phonology, stress and intontation. Typically lexical influence is minimal, morphological influence non-existent, although oddities of word order and expression may be found. If the shifting population is large enough the language of the orignal L2 community may be affected as well. The second case minimal bilingualism with shift describes the more drastic situation of pidgin formation. Speakers do not have adequate time or opportunity to acquire native-like L2, but are forced to communicate with L2 speakers. In these cases lexicon is clearly taken from L2 (the ‘lexifier-language’), morphology is generally supposed to be lost altogether. Whether syntax and phonology are necessarily derived from L1 or supplied by universal principles is a topic of debate. The third case, borrowing with minimal bilingualism describes cultural borrowing or ‘adstrate influence.’ A small group of scholars or traders are in contact with and may have a high knowledge of the L2, the majority of the community does not know it. This is the case with the influence of Chinese on Japanese, Latin and Greek on English, Arabic on Turkish. Somewhat higher levels of bilingualism were no doubt involved in the influence of Arabic on Persian, Arabic on Berber, or French on English. Finally we come to the case of borrowing with maximal bilingualism. In this case the entire community is fluent in an L2 (which is thus only a ‘second’ language from the historical point of view) and this influences their L1. I 15
know of two documented cases— the famous case of Kupwar in India (Gumperz & Wilson 1971), where Kanada (Dravidian), Marathi (IndoAryan), and Urdu (Indo-Aryan) come into contact and a more recently documented set of cases (Ross 1996) in Papua New Guinea, where Austronesian languages (Takia and Maisin) come into contact with Papuan languages. For the Kupwar case Gumperz & Wilson (1971: 154) observe: “A historical linguist would readily identify particular texts as from a deviant dialect of Kannada, Marathi, or Urdu. What would be missed is that sentence-by-sentence comparison of natural conversation texts in all three main local varieties reveals an extraordinary degree of translatability from one local utterance to the other. …The sentences in this example are lexically distinct in almost every respect, yet they have identical grammatical categories and identical constituent structures… What seems to have happened in these informal varieties is a gradual adaptation of grammatical differences to the point that only morphophonemic differences (differences of lexical shape) remain.” And Ross (1996) observes for the New Guinea cases: “It is important to note here that all the Papuan features of Takia have to do with morphosyntactic structures. Where grammatical morphemes participating in these structures can be sourced (by way of cognates in other languages), their forms are invariably W[estern] Oc[eanic].” (1996: 188) “We could continue this catalogue of syntactic parallels further.. However, it is more important to conclude by pointing out what has not occurred in Takia. There has not been much lexical borrowing from Waskia [a Papuan language]. And there is no sign that Takia phonology has undergone any major changes.” (1996: 192) “Extensive though the Papuanisation of Maisin has been, its Papuan features (like those in Takia) are all morphosyntactic structures. Careful reconstruction reveals that the forms are of P[apuan] T[ip] origin.” (1996: 194) In short the pattern observed in these cases is that morphosyntactic structure (word order and use of morphological formatives) and perhaps semantic function conforms with surrounding languages while the phonological form of words and formatives are cognate with those of genetically related languages located farther away. This description largely fits the Bukhara Arabic case, although there does appear to be some borrowing of the form, not just function of morphological formatives and particles. Moreover, the situation is complicated by the fact that the two surrounding languages, Uzbek and Tajik, are not entirely convergent in syntactic structures. In general it seems that Bukhara Arabic is converging toward the Uzbek pattern, that is toward a consistently head final or harmonic SOV type. There is only one syntactic structure found in Bukhara Arabic and Tajik but not in Uzbek or most other forms of Arabic. That is the Noun-linker-Adjective pattern. As regards the sociolinguistic situation which this linguistic situation reflects, the Indian and New Guinea cases would lead us to suspect that there is extensive bi-(or multi-)lingualism and that the original L1 is maintained prinicpally as a vehicle of in-group communication and a symbol of identity: 16
“As far as can be determined, almost all local men are bi- or multi-lingual….There is every indication that the Kannada-speaking Jain cultivators and the Marathi-speaking service castes have both been in the region for more than six centruies. The Urduspeaking Muslims date from the days of Moghul domination threee or four centruies ago. Bilingualism in Kupwar is thereore a long-standing tradition. Why has it been maintained for so long? Information obtained from living in the village over a period of several months suggests that the major factor in language maintenance is that the local norms or values require strict separation beween public and private (intra-kin group) spheres of activity.” (Gumperz & Wilson 1971: 153). “For many New Guinea bilinguals (‘bilingual’ here also subsumes ‘multilingual), one of their languages is emblematic of that ethnicity…. Both the emblematic language and the intergroup languages are ‘native’ in the sense that they are acquired together in infancy.” (Ross 1996: 181). Whether this expectation is borne out in the Bukhara Arabic case awaits the results of further research. (68) Contact situation Max. shift (substrate) Min. shift (pidgin) Min. borrowing (adstrate) Max. borrowing Phonology L1 (L2) UG? (L1) L1 L1 Structural correlate Morphology Lexicon L2 L2 UG? L2 L1 L2 (L1) L1(L2) L1 Syntax L2 (L1) UG? L1? (L2?) L1 L2 References Brustad, K.E. (2000) The Syntax of Spoken Arabic: A Comparative Study of Moroccan, Egyptian, Syrian, and Kuwaiti Dialects, Washington, D.C.: Georgetown University Press. Cowan, W. (1967) ‘Sound Change in Central Asian Arabic’, Der Islam 43: 134–138. Fischer, W. (1961) ‘Die Sprache der arabischen Sprachinsel in Uzbekistan’, Der Islam 36: 232– 263 Fischer, W. and Jastrow O. (1980) Handbuch der Arabischen Dialekte, Wiesbaden: Otto Harrasowitz. Greenberg, J. (19662) ‘Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements’, in J.H. Greenberg (ed.) Universals of Language, Cambridge, Ma.: MIT Press, PAGES Gumperz, J.J. and Wilson R. (1971) ‘Convergence and Creolization: A case from the Indo-Aryan/ Dravidian Border in India’, in D. Hymes (ed.) Pidginization and Creolization of Languages, Cambridge: University Press, PAGES Hawkins, J.A. (1983) Word Order Universals. San Diego: Academic Press. Huehnergard, J. (1997) A Grammar of Akkadian, Atlanta, Ga.: Scholars Press. Ismatulla, Kh. (1995) Modern Literary Uzbek, Bloomington: Indiana University. Jastrow, O. (1998) ‘Zur Position des Uzbekistan-Arabischen’, in Annäherung an das Fremde: XXVI. deutscher Orientalistentag vom 12. bis 29.9.1995 in Leipzig, Stuttgart: Franz Steiner Verlag, 173–184. — (1997a) ‘Wie arabisch ist Uzbekistan-Arabisch?’, in E. Wardini (ed.) Built on Solid Rock: Studies in Honour of Professor Ebbe Egede Knudsen. Oslo: Novus Forlag, 141–153. — (1997b) ‘The Neo-Aramaic Languages’, in R. Hetzron (ed.) The Semitic Languages, London: 17
Routledge. PAGES Krotkoff, G. (1982) A Neo-Aramaic Dialect of Kurdistan: Texts, grammar, and vocabulary, New Haven, Ct.: American Oriental Society. Masica, C.P. (1976) Defining a Linguistic Area: South Asia, Chicago: University of Chicago Press. Mous, M. (1992) A Grammar of Iraqw, Hamburg: Helmut Buske Verlag. Rastorgueva. V.S. (1992) A Short Sketch of Tajik Grammar, trans. by H.H. Paper. Bloomington: Indiana University. Ross, M. (1996) ‘Contact-induced Change and the Comparative Method: Cases from Papua New Guinea’, in M. Durie and M. Ross (eds) The Comparative Method Reviewed: Regularity and Irregularity in Language Change. Oxford, New York: Oxford University Press. Sjoberg, A.F. (1963) Uzbek Structural Grammar, Bloomington: Indiana University. Thomason, S.G, and T. Kaufmann (1988) Language Contact, Creolization, and Genetic Linguistics, Berkely: University of California Press. Tsereteli, G.V. (1970) ‘The Influence of the Tajik Language on the Vocalism of Central Asian Arabic Dialects’, Bulletin of the School of Oriental and African Studies 33: 167–170. Versteegh, K. (1997) The Arabic Language, Edinburgh: University Press. Vinikov, I.N. (1969) Jazyk i folklor Buxarskix Arabov. Moscow: Nauka. Weinrich, U. (1963) Languages in Contact, The Hague: Mouton. Woodhead, D.R. and W. Beene (1967) A Dictionary of Iraqi Arabic: Arabic-English, Washington, D.C.: Georgetown University Press. Acknowledgements Thanks to Akio Nakano for alerting me to the existence of these texts and for considerable help in deciphering them, to Olga Kapeliuk for updating my bibliography and sending me a number of articles I could not readily find in Tokyo, to Tooru Hayasi, Éva Á. Csató, and Lars Johanson for first allowing me to present this material in a forum they organized in Tokyo in 1997 on peripheral Altaic languages, and especial thanks to Éva Á. Csató (again), Bo Isaksson, and Uppsala University for making possible my participation in this extremely stimulating forum. List of abbreviations ABL ACT CA COMP CONV EZ FEM INF LOC MASC N PART PL POSS PRES/FUT Q SG ablative active Classical Arabic comparative converb ezafet feminine infinitive locative masculine ???? participle plural possessive present/future question particle singular 18