2016. június 26., vasárnap

Ian Hancock - ON ROMANI ORIGINS AND IDENTITY

http://www.oocities.org/romani_life_society/indexpics/HistoryOrigins.pdf

Ian Hancock - ON ROMANI ORIGINS AND IDENTITY 

The Romani Archives and Documentation Center The University of Texas at Austin


“A geneticist’s summary of [our] data would describe the Gypsies as a conglomerate of Asian populations . . . unambiguous proof of the Indian ancestry of the Gypsies comes from three genetic marker systems . . . found on the same ancestral chromosomal background in Gypsy, Indian and Pakistani subjects. While confirming the centuries-old linguistic theory of the Indian origins is no great triumph for modern genetic research, the major, unexpected and most significant result of these studies is the strong evidence of the common descent of all Gypsies regardless of declared group identity, country of residence and rules of endogamy . . . The Gypsy group was born in Europe. All marker systems suggest that the earliest splits occurred 20 to 24 generations ago, i.e. from the late 13th century onwards” (Kalaydjieva et al., 2005:1085-6)."

Those of you familiar with my work know that it has taken a circuitous route over the years in an ongoing effort to refine it, and no doubt it will be modified further as it continues. Thus in my earliest writing I supported a fifth-century exodus from India and accepted the established three-way Rom-Dom-Lom split; I no longer do. I argued for a wholly non-“Aryan” ancestry, but no longer believe this to have been the case nor, indeed, that “Aryan” is even a genetically relevant label. I saw the migration from India to Anatolia as having been a slow one, consisting of a succession of military encounters with different non-Indian populations; I no longer think that it happened in that way. I have argued, sometimes strenuously, that our people were one when they left India, one when they arrived in Anatolia, and one when they entered Europe. My findings are leading me more and more to believe that they were not.

Working especially closely with three other scholars, themselves also Romanies— Kenneth Lee at Newcastle University, Ronald Lee at the University of Toronto and Adrian Marsh at Greenwich University—I have come to modify these positions considerably. The ongoing research of Marcel Courthiade and Jan Kochanowski in France has also been most useful in reaching these newer interpretations1 , and I am especially grateful to Vardan Voskanian for generously sharing his materials and his ideas regarding Lomavren. Though this newer perspective differs considerably from my earlier one, I find myself obliged to accept it because I am wholly convinced that the history that is coming to light is, in its broad form, the correct history.

 That the Romani people had a military origin is not in fact a new hypothesis; it was first adressed over a century ago by Burton, Leland, de Goeje, Clarke and others. My own contribution, besides attempting to flesh out the details, addresses rather the origins of the Romani language as a military koVnJ (Hancock, 2000). Certainly not everyone is persuaded by the direction my work is taking; Matras says :
"In a number of recent publications, Hancock claims that Romani was formed as a military koïné by a caste of warriors assembled to resist the Islamic invasions of India. In some circles, this view is gaining popularity as it pretends to revise what is referred to as potentially racist, or at least stereotypical images of the Rom. There is, however, neither linguistic nor historical evidence to support it (2004a:301; see also 2004c for a harsher criticism2 ).  "

Matras’ own position—though itself accompanied by neither “linguistic nor historical evidence to support it”—adheres to the traditional account: 
"Indic diaspora languages [are] spoken by what appear to be descendants of itinerant castes of artisans and entertainers who are spread throughout Central Asia, the Near East and Europe. They include . . . Romani (1999:1) Proto-Romani was carried from India westwards by migrants who appear to have been members of service-providing castes, similar in status and occupational profile to jatis or service groups known in some parts of India as dom . . . the řom settled in the Byzantine Empire some time around the tenth century CE (2004b:278).  "


More recently, Tcherenkov & Laederich (2004:13) have leant toward the same traditional origin: 
"[It] is but a small step to support the hypothesis that these Indian Dom are the ancestors of the European Rroma. The professions exercised by the Dom in the Indian subcontinent—musicians, dancers, smiths, basket weavers, sieve makers, even woodworkers, are transmited from father to son. From their similarity to the ones of the European Rroma these could or may be considered as the origins of the traditional Rroma trades! . . . some authors claim that Rroma originated from either one of the upper castes such as the Rajputs or from a mix of different castes. With our current knowledge, this cannot be settled to satisfaction."
 The present monograph raises a number of questions, some of which are posed at 12, below. A central position underlying the discussion is that three salient, and hitherto not adequately considered, aspects of the contemporary Romani condition rest upon the facts of our history detailed here:  

First that the population has been a composite one from its very beginning, and at that time was occupationally rather than ethnically-defined; Second that while their earliest components are traceable to India, Romanies essentially constitute a population that acquired its identity and language in the West (accepting the Christian, Greek-speaking Byzantine Empire as being linguistically and culturally ‘western’), and Third that the entry into Europe from Anatolia was not as a single people, but as a number of at least three smaller migrations over perhaps as much as a twocentury span of time. 

Together, these account in part for the lack of cohesiveness among the various groups selfidentifying as Romani, and for the major dialect splits within the language. We might see each major post-Byzantine group as evolving in its own way, continuing independently a process of assimilation and adaptation begun in northwestern India. Thus the descendants of those who were held in slavery until the 19th century, and those whose ancestors entered Spain in the 15th century are today very different, the former—the Vlax Romanies—having been heavily influenced genetically, culturally and linguistically by Romanian and the Romanians; the latter on the other hand—the Kalé Romanies—having been influenced in the same way by Mozarabic and Spanish, and both populations have furthermore been separated by a more than six century span of time. Thus any originally acquired characteristics they might still share, which constitute the genetic, linguistic and cultural so-called “core of direct retention,” are greatly outweighed by characteristics accreted from the non-Romani world. The reunification (or more accurately unification) movement urged by such organizations as the International Romani Union or the Roma National Congress seeks—as I do myself—to emphasize the original, shared features of each group rather than those acquired from outside which separate them; yet for some, that original material is now scant, and creating for them any sense of a pan-Romani, global ethnicity would require the kind of effort that is, sadly, very far down on the list of day-to-day priorities and, pragmatically, would be difficult to instigate. It also calls into question the legitimacy of the exclusionary and subjective position taken by some groups who regard themselves as being “more Romani” than others.3

1. Who Were the Ancestors of the Romanies?  

Using lexical data I demonstrated (Hancock, 1995) that the assumed early single migration out of India with a subsequent split into Domari, Lomavren and Romani (i.e. Middle Eastern, Armenian and European Gypsy) once it had passed through the Persian language territories, could not be maintained in light of the percentages of shared and non-shared Iranic items evident in each today. This confirmation is already finding a place in the new scholarship; thus Windfuhr, in his entry on Gypsy languages in the Encyclopædia Iranica (2000:415) refers to that 1995 study when stating that the Iranic items “reflect three distinct historical layers of Indo-Aryan innovations, which suggests three successive westward migrations, rather than a single one.” It is of some significance since it overturns the generally accepted historical scenario current over the past one and a half centuries, and impacts directly on our understanding of early Romani. Higgie (1984) for instance, attempted a reconstruction of Proto-Romani by comparing Romani with Domari, as did Kaufman (1984); Higgie’s work points to something like the 6th century BC as the time of the split from Indic, while Kaufman posits the separation from Indo-Aryan by 300- 400 BC. This would be comparable to attempting a reconstruction of the original Latin by including (say) Umbrian or Oscan along with the modern Romance languages in the comparative data.

An examination of the earliest words in the Romani language suggests a number of things: firstly that there is little in the original, ‘first layer’ Indian vocabulary that reflects a nomadic or itinerant population, but rather it points to a settled one; and secondly that while there are not many original words for e.g. artisan or agricultural skills, there are quite a few military terms. There are Indian words for soldier and attack but not for farmer or harvest; there are words for sword and spear but not for plough or hoe; there is a word for horse but not for buffalo and so on4 . Given these lexical clues and the likely time period (both discussed below), and given that the Indian words and grammar in modern Romani point to the languages spoken in the north-western part of India and to nowhere else, an examination of Indian history for evidence of any military activity during that time and in that area is a natural next step—but first, the time period must be established. 

2. The Date of Departure

 It has been claimed repeatedly that the speakers of the language that developed into modern Romani left India some time between the fifth and ninth centuries; those who support the traditional Shah Nameh explanation, which is routinely repeated in even the latest books on Romanies, would place it in the 5th century. Others, like myself, see military activity as the reason for leaving, but still argue for an earlier date of departure: “they left perhaps as early as in the sixth century A.D., probably due to repeated incursions by Islamic warriors”5 (Barany, 2002:9)   

On the basis of lexicon, Kaufman (1984:12) has asserted that:
There is no way that Romani could have avoided Arabic loanwords unless it had entered Iran before 700 AD. Speculations that do not operate within these constraints as axiomatic are idle; it is totally irrelevant that there may be some historical evidence of troubles in, and outmigrations from, India around 1000 AD, and I am getting bored with hearing again and again the speculation that the Gypsies may have left India at such a late date.

Vekerdi (1988:13) says : The Gypsies’ ancestors began leaving northwest India probably about the seventh century AD. They are characterized as robbers, murderers, hangmen and entertainers. These professions were prescribed for them by the rules of the Hindu caste system. Thus they belonged to the so-called ‘wandering criminal tribes’ of India and were obliged to lead a parasitic way of life. Among the numerous outcast groups, they occupied the lowest rung on the social scale. 

Halwachs (2000: 5, 24): is also persuaded that the lack of adoptions from Arabic is a decisive factor in dating the time of departure: As Romani lacks Arabic loans, it is to be assumed that the Romani speakers left the Persian area before its arabization . . . and following this moved on to the Byzanthinian area of influence . . . Experts still disagree on the point of time of the Gypsies’ emigration from the north-west of India. If we consider all the different statements, the resulting period of time is somewhere between the 5th and 10th centuries after Christ. In the second half of the first millennium, emigration most probably did not happen all at once but took place in the course of various waves.  

In an earlier monograph (1977: 3), Kenrick too believed that [t]he Romanies of Europe must have come through Iran before 600 ADC—the first Arab invasions—this is the only possible explanation for the large number of Iranian words and the small (infinitesimal) number of Arabic words found in the Romani vocabulary, 

though in a more recent statement (Patrin, 14:viij:00) he moves that estimation two or three centuries forward: My basic theory at the moment is that the Roma of Europe are mainly offspring of the defeated Zotts of Zottistan [in AD 855]. These were divided by the Arabs into two groups; one was sent to Ain-Zarba where they were in due course massacred by the Byzantine Greeks—maybe the women taken as slaves. The other group went to Khaneikin and thence to Europe. They were mainly buffalo keepers (see Rishi’s article “Panjabi love of buffalo milk” [1976]) but obviously in Zottistan had developed other trades. We know there were musicians there. Some other Indians joined them and adopted Romani as their language, intermarried, &c. 

More recently (2004:10) he presents a social origin but (wisely?) avoids speculating as to dates: My own belief, as stated earlier, is that Indian immigrants from various tribes intermarried and intermixed in Persia, forming into a people there using the name Dom, and that a large number of them then moved into Europe; their descendants are the Romany Gypsies of today. 

In his UNESCO-sponsored book, Alain Reyniers (1998:25) writes of . . . Une sortie étalée le Ve et le XIIe siècle . . . Après une première étape en Perse, les Tsiganes se seraient divisés en deux groupes. Le premier se serait dirigé vers le Moyen-Orient et l’Egypte. Le second se serait déplacé vers le nord-ouest. 

Another recent publication (Marushiakova & Popov, 2000:5) supports the traditional view, and places the presence of Romanies in Persia before AD 900: According to most linguists, the formation of the Gypsy language began sometime in the 6th or 7th century, while from the 8th-9th centuries onwards, it developed as a separate language under the influence of the majority languages spoken in the area: Persian, Armenian, Greek. Wandering for several centuries throughout the lands of what are today Pakistan, Afghanistan and Iran, and to the south of the Caspian Sea, the Gypsies (and their language) divided into two separate branches, speaking the so-called “ben” and “phen” dialects respectively, this being an important stage in the development of the Gypsy language and the Gypsy community as a whole. Reaching the land of northern Mesopotamia and the eastern boundary of the Byzantine Empire towards the end of the 10th and beginning of the 11th centuries, the Gypsies split into three major migration groups—the ben-speaking Dom, who took the southern route, or stayed in the Middle East, and the phenspeaking groups of Lom, who took the northern route, and Rom, who took the western route.
  
Achim (2004:7-12) also accepts a ninth-century departure: The migration took place over an extended period of time and was not dramatic in nature . . . [i]t is generally accepted that the migration of Gypsies from India to Europe took place between the ninth and the fourteenth centuries, in a number of waves. It is believed that the Gypsies arrived in Persia in the ninth century. Persian sources call them Luli or Luri. 

Blachut (2005:26), using Barany (2002) as his source, says Gypsies had begun leaving the southern part of India in 1500 BC, when the Aryans invaded the country . . . Gypsies originated in the Punjab region of northwestern India. They began leaving in the sixth century AD because of constant invasions by Islamic warriors.  

Wogg (2006:3), basing his position on the combined works of Hübschmannová (2004), Kenrick (2004) and Matras (2002) says: The [Shah Nameh] legend about the Luri could thus very well refer to the Roma, who left Persia already in the 5th century, and India even earlier on their westward journey . . . most scientists today assume it was over a long period of time—between the 3rd and 10th century—during which the Roma left India, most likely between the 8th and 10th century

In discussing Romani history, Price (2000:207) says “[a]t some indeterminate period, not later than the ninth century AD, the Romanies were on the move again [out of India].” Miklosich (1874) put the date of departure at somewhere between AD 500 and AD 700, while Sampson (1923:157) argued for the ninth century. Fonseca (1996:94) provides an account that concludes “the earliest Gypsies would have left India at least by 720 AD.” A recent interpretation has Romanies leaving India between AD 1017-1030 as a result of Ghaznavid invasions, but splitting into the three-way Dom-Rom-Lom (Jordan, Europe, Armenia) division somewhere between Afghanistan and Persia (Knudsen, 2003:22-23), while Tcherenkov & Laederich (2004:14), who also support the one-migration three-way-split position, have most recently placed the migration back six hundred years: “[A] time frame for the migrations of Rroma from the Indian subcontinent . . . at what we believe is a reasonable departure date, around the fourth and fifth centuries.”

Djurić (2003), on having determined that Romani has a middle voice, has argued that the language must date back to before the time of Christ. Any claim to a pre-AD 1000 date of departure, however, must be challenged on the basis of the historical development of the IndoAryan languages. Woolner’s (1916:123) analysis of the derivation of the first-person singular personal subject pronoun me, and Bloch’s (1953:24) similar examination of kon “who” both point to a post 7th century development. We must also examine the reassignation of neutergender nouns after that category began to disappear from the Apabhramśas by the end of the Middle Indic period. This is accepted as about the year AD 1000; Masica (1991:8) gives the New Indo Aryan period as “1000 AD - present . . . the modern Indo-Aryan languages properly and henceforth called New Indo Aryan . . . date from approximately AD 1000.” The transition was clear-cut, and the date significant. Bloch (1965:29) says “it is of great importance to indicate the chronological break, which isolates the whole of neo-Indian [from Middle Indic].” “The three genders [of Old Indic] continue [in Middle Indic] but the masculine and the neuter come closer together” (Sen, 1960:75). The OIA neuter gender was systematically lost, the change spreading towards the northwestern part of India, where some three-gender NIA languages are still found to this day, e.g. numbers of Central ( auraseni) languages, such as Bhili, Gujarati and Khandeshi as well as some Southern (Maharasthri) languages such as Marathi. Nevertheless. 

. . . the most widespread NIA system is a two-gender system, in which the old masculine and neuter have merged. (That is not to say that there have not been some reassignments of OIA gender . . . e.g. the NIA descendants of OIA agni- ‘fire,’ which is masculine, are mostly feminine”), as is Romani jag, as well (Masica, 1991:221). 

According to Burton (1851:90-91), the neat shift of the neuter to the masculine set did not happen everywhere: “In the Játaki dialect, nouns are of two genders, masculine and feminine. The neuter is not used, and words which properly speaking belong to that gender are made masculine and feminine, as usage directs without any fixed rule.” He describes Jataki, the language of the Jats, as “a corrupt form of the Multání, itself a corruption of the Panjábī, tongue.”

 It is significant that the languages most like Romani—Hindi, Panjabi, Rajasthani, &c., are not three-gender languages. If pre-Romani had left India before the end of the first millennium AD, which is to say during the MIA period, it would have retained its three-genders, and the fact that it is a two-gender language today would oblige us to accept that the loss of the neuter, and its reassignation to either masculine or feminine, took place outside of India. Kenrick is of this opinion, believing Persian to have been the factor of change: Il y avait trois genres (comme en allemand), au moment où les Tsiganes ont quitté l’Inde, mais le neutre a disparu au Moyen-Orient, sans doute sous l’influence du parsi (1994:54).

He maintains this position in (Kenrick, 2004:104): “There were three genders (like German) when the speakers left India, but the neuter disappearted in the Middle East, probably under the influence of Persian.” Out of contact with other Indian languages, such reassignation would have been random; however, comparing those Romani nouns deriving from neuter sourceforms in Sanskrit and/or Prakrit, with their equivalents in Hindi, we find that the match is 98.7% (one mismatch out of 35 items compared) for the masculine set, and 60% for the feminine set, 86% for both masculine and feminine matches. The approximately 2:1 ratio of masculine to feminine Indian-derived nouns in Romani also accords with the reassignation of OIA neuters mainly to the masculine set. While he did not discuss the date of the presence of pre-Romani in India or recognize its relevance to ascertaining the time of its separation, Lesný had already noted the reassignation of OIA neuters in MIA nearly a century ago: Die mittelindischen lautlichen Prozesse haben bekanntlich bewirkt, dass auch das Geschlecht eine Änderung erfahren hat. Marati und Gujarati haben noch die ursprüngliche Einteilung in drei Geschlechter behalten, Bangali unterscheidet eigentlich kein Geschlecht, Hindi, Panjabi, Bangali, Sindhi, Kashmiri und Naipali unterscheiden nur nur das männliche und das weibliche Geschlecht. Diese nordwestlischen Gruppe reiht sich auch die Zigeunersprache an, indem sie auffallenderweise in bezug auf die Änderung des Geschlechtes mit derselben überinstimmt. Ich will die zahlreichen Neutra, die zu Maskulinen geworden sind, übergehen und erwähne nur zwei Substantiva, die sowohl in der genannten Sprachengruppe als auch in den Zigeunermundarten Feminina geworden sind: agni masc. ‘Feuer’, MIA aggi, Marati, Gujarati, Hindi ag f., Panjabi agg, f., Sindhi agi, f., Kashmiri agun, m., Romani jag, f. aksi ‘Auge’, MIA acchi, n. oder f., Marati aksi n., Gujarati ankh f., Hindi ank f., Panjabi akkh f., Sindhi akhi f., Kashmiri acchi f., Romani jakh f. (1915-1916:422).
Since the loss of the neuter gender had begun to take place while the NIA dialect groups were still in formation, this means that pre-Romani was still in India at the time that this was taking place, i.e. still a part of the Middle Indo-Aryan cluster. Even if pre-Romani were derived from various Indian languages, as I maintain, the case still holds; a gender match with Sindhi or Panjabi yields the same result. If we assume that Sampson’s (and my own earlier) “single race speaking a single language” remained intact until it had passed through Persia, then we would expect the Persian words it picked up during that period to be shared by Romani, Domari and Lomavren; but they are surprisingly few: just 16% between Romani and Domari, 7% between Romani and Lomavren, and 12% between Lomavren and Domari. By way of comparison, on the other hand, over 50% of the Persian words in Romani are shared by Urdu: ..... 

... 
4. Domari 

That Romani, Domari and Lomavren constitute three branches of an original proto-language has remained the conventional wisdom in Romani historical linguistic studies for over a century, and continues to be repeated. That they had independent origins had already been suspected by Colocci (1907:279), who urged caution in drawing too sweeping a conclusion from the available sources:  
To imagine that just because the Gypsies of Europe and their brothers in Asia share a common linguistic core, one should therefore conclude that there was a single exodus of these people [out of India], and furthermore that the unity of their language argues against more than one migration, 13 seems to be a conclusion which is only slightly weakened by the still nebulous state of the documentation. Unity of language might well prove unity of origin; but there could still have been different migrations, chronologically and geographically, without that fact being too apparent from the lexical adoptions acquired by the mother tongue in the countries through which they passed; all the moreso since those migrations were very rapid. To conclude, therefore, that the unity of their exodus rests upon the recognition of the unity of the substrate of their language, strikes me as a proposition which shouldn’t be universally accepted without [first incorporating] the benefit of a [lexical] inventory. 

The late Angus Fraser also cautioned (1992:39) that 
despite Sampson’s insistence that both sprang from a single source, some of Domari’s dissimilarities from European Romani create doubts about how far we can assume that the parent community was uniform.

In Hancock (1995) I demonstrated that Colocci’s and Fraser’s doubts were justified, and the fact that Domari and Romani had separate origins is gradually moving toward general acceptance; Matras concludes, in the most recent overview of Domari, that together with Romani they “were, to begin with, two distinct, albeit related Indo-Aryan idioms” (1999:55). The additional claim I make that the Domari language and its speakers left India earlier than did Romani and its speakers, might also be supported by the evidence of gender. Macalister (1914:9,11) says : [t]here are three genders [in Domari], masculine, feminine, and neuter. The last is now all but obsolete, but recognisable only by the form of the accusative singular . . . As in most Aryan languages, neuter substantives have no accusative form different from the nominative. This is now the only criterion for distinguishing neuter nouns. But even here they appear to be in process of assimilation to the masculine or feminine declension, and developing analogous accusative forms.  

Sampson (1926:125) has contested this, though it has to be assumed that he is only querying Macalister’s claim that the modern Domari language has three genders; he would have known that three genders existed in the speech of the original population, which he maintains left India “at least as early as the end of the ninth century” (Sampson, 1926:28-29). He says “the Nuri ‘neuter’ of Macalister has no historical basis, and is to be understood merely as a term applied by this collector to nouns denoting inanimate objects in which, as in Eur. Gyp., the form of the acc. sg. is identical with that of the nominative.” Unfortunately, Macalister does not provide genders in his Domari vocabulary, though he lists some examples of each in his grammatical outline (pp. 15-16): béli, záro, “friend”, “boy,” m., cóni, júri, “girl,” “woman,”f., páni, ag, “water,” “fire,” n. Kenrick, however, in his current series of Domari lessons (2000:2), says “some dialects also have a traditional (historical) neuter gender, ending in a consonant.” 

Besides losing the neuter gender, Indo-Aryan also lost the dual number that characterized its Old period. Romani lacks this entirely, but according to Macalister (op. cit., p. 9), in Domari “faint traces are not wanting of the former existence of a dual, but this is almost wholly obsolete.” It is a pity that Macalister did not provide actual examples of these, since if they had indeed existed in Domari, it would suggest an improbably much earlier separation from India; thus Masica (1991:226) says “[t]here are only two numbers, singular and plural, in NIA at best. OIA had three, but the old dual quietly disappeared at the beginning of MIA,” but he has MIA beginning around 600 BC (op. cit., p. 51)—far too early to match with the rest of the linguistic data we have on Domari. If Domari does indeed show evidence of a dual number, this is probably influence from Arabic, which has it (and not Persian or Kurdish, which don’t).

5. Lomavren 
On the basis of its lexicon Lomavren, the language of the Lom or “Bosha” in eastern Turkey and the Caucasus would seem to stand somewhere between the two migrations that gave rise to Domari and Romani. On the one hand it shares items with Romani which differ from their Domari equivalents, thus  and on the other, it shares items with Domari which are absent (or which have not been replaced) in Romani: .......

The complete lack of Greek lexical items in Lomavren shows that the ancestors of the Lom never made it into Anatolia, or else that they passed through it before Greek was established there. It furthermore shares only five items from Persian with Romani (Some nineteen have been identified in Lomavren altogether (Voskanian, 2002), and over 100 in Romani—Hancock 1995). Significantly, not one such item in either language is from the Middle Persian (Pahlavi) period; all are from the modern period, which dates from the early 10th century, thus further undermining the argument for a 5th century passage through the area. Furthermore, there is only one indisputable item of Kurdish origin in Lomavren (cf. perhaps ten in Romani). Fraser (0000:14) has already noted that Lomavren and Romani share practically none of the same Armenian loanwords. Those Armenian-derived items in Romani for which both languages have words are as follows: 
.......

Another feature that distinguishes Lomavren is that in that language, New Indo-Aryan /a/ was not raised to /e/ as it was in Romani: (LOM khar, par-, phan-, saj, thar-, ROM kher, per-, phen-, šel, ther- “house”, “fall”, “say”, “hundred,” “have”, cf. Hindi ghar, par-, bhan-, sau, dhar-); however numeral “ten,” which is las in Lomavren (cf. Hindi das, Romani deš), has the eform de(s) in its combinations: de’-hu-dui “twelve,” cf. Romani deš-u-duj). A further indication of its later date of separation from India is in the behaviour of initial Middle Indo-Aryan /v/, which became /b/ in New Indo-Aryan (including Romani) but not in Lomavren or in Domari:......... 


Nevertheless, some Lomavren items appear not to have undergone this: banth-, bakhot-, “shut”, “break”, cf. Romani phand-, phag-, Domari ben-, bFg-. It is intriguing that both Romani and Lomavren share the secondary meaning of the verb “sit” to mean “reside”, not paralleled in the modern languages of India (see '10, below), and that the early speakers of both languages relexified the original Indian tru ula “trident” (presumably in its religious context as the one held by the god Shiva) into new religious contexts: Romani tru ul “cross” and Lomavren tcrusul “church”.6 

 It is also the case that the Romani word xulaj “host” (from Persian xudāy, and not, as Voskanian (2002:182) has convincingly shown, from Kurdish xola ‘god’) exhibits the same phonetic rule that is general in Lomavren, i.e. the shift of /d/ to /l/ (cf. LOM xula, do., and the items level, lei, las above), suggesting a common point of separation—though it is the only Romani item that does this; the possibility exists that the word may have entered each language independently from separate sources. Since /r/ does not go to /l/ in Lomavren, it can be argued too that lom is from *dom(ba) rather than from Rrom or Rum.

 If my argument is maintained that Romani only crystallized into an ethnic mother tongue under the influence of Byzantine Greek and that prior to that it was a military koïné and not a native language, then we might suppose that this nativization did not happen to pre-Lomavren but rather that its speakers were quickly assimilated into the eastern Armenian speech community, retaining Indian words solely as lexical items conforming to Armenian morphosyntax and phonology. Though the processes giving rise to each may or may not differ, this has resulted in an ethnolect similar in many ways to the Angloromani dialect of the British Romanichals (Hancock, 1984).  

The present work is supported by Courthiade’s independent research in France which even more specifically places the origin of Romani in Kannauj which, together with Ayodha further east, was the city in the central area which served as the home for the Rajput armies. A comparison of modern Kannauji with Romani shows less similarity between the two than that shared by Romani and Hindi; the same is true for Sindhi, the language of Sindh—the place of origin proposed by Marsh (2003), Marwari, proposed by Hübschmannová (2004), and Jataki (Jatki, Siraiki), the language of the Jats, a people proposed by Leland (1882) as the ancestors of the Romanies. While the Rajput conscriptees and their camp followers may well have spoken these and many other languages, the fact that Romani is nevertheless closer to Hindi/Urdu supports the shared origin these latter languages have in Rajputic. Domari equivalents have been included by way of comparison.  

7. The Military Factor 

A military origin for Romanies, generally as captives, is not a new idea; de Goeje (1876: 32) wrote that “in the year 1000, we find bands of Zotts in the army of Abû-Naçr ibn-Bakhtiyâr, in Persia and Kirmân (Ibno-’l-Athîr, ix., p. 114). In 1025, al-Mançûra was conquered by Mahmûd al-Gaznawî, because the prince of this town had forsaken Islamism;” Clarke (1878: 134) wrote that “it was from the Ghaznevide conqueror and at home that the independence of the Jats received its death-blow. The victorious army of Mahmoud, when returning laden with spoil from the Somnauth expedition of 1025, was attacked and pillaged by them on the banks of the Indus. Their temerity was chastised with exemplary rigour. Broken and dispersed by the resistless arms of the Sultan of Ghazni, they were not, however, annihilated.” Leland, (1882: 24) that “Jat warriors were supplemented by other tribes . . . they were broken and dispersed in the eleventh century by Mahmoud” and Burton (1898: 212) that “Sultan Mahmoud carried with him in A.D. 1011 some two hundred thousand [Indian] captives, the spoils of his expedition.” Kochanowski later agreed (1968: 27-28) that “our own inter-disciplinary studies have shown that the Gypsies are Rajputs who left northern India,” and Vijender Bhalla’s serological studies undertaken in India concluded that “Rajputs occupy the [genetic] position nearest the Gypsies” (1992: 331- 332). Over a decade ago the Polish scholar Lech Mróz had also considered a specific connection with the Islamic raids into India: “Podsumowują: uwaŜam za prawdopodobne Ŝe Cyganów dostali się do Iranu w czasach Mahmuda z Ghazny, w resultacie jego wypraw do Indii” (1992:40; “I consider it likely that the Gypsies’ ancestors arrived in Iran in the time of Mahmud of Ghazni, as a result of his raids into India”). Bajram Haliti (2006:6) has come to the same conclusion: “Some time between the tenth and eleventh centuries, the largest groups of Roma left India and the main cause was invasion of the great emperor Mahmud Gazni, who led 17 raids in western India. Running away from terror, Roma first stopped in Iran, and then separated in two groups, the first moving toward Spain, and the second toward Byzantium and Greece.” Nevertheless there continues to be resistance to this; in 2004 in his own interpretation of Romani history Viorel Achim (loc. cit.) wrote “The distinguishing feature of the Gypsy migration is that it was not of a military nature.” An examination of the circumstances of Indian and Middle Eastern politics and warfare during the relevant timeframe is thus called for. 

For roughly the first quarter century of the second millennium, north-western India came under a series of attacks by Muslim troops led by General Mahmud from his headquarters at Ghazna (today called Ghazni and located in Afghanistan). Between AD 1001 and AD 1026 these Ghaznavids, as they were called, made seventeen forays into the Hindu-Shahi kingdom as far as Kashmir with the intent to spread Islam; ultimately the Indian kingdoms of Nagarkot, Thanesar, Kannauj and Kalinjar were all conquered and left in the hands of Hindu vassals. There were seventeen battles altogether; the main ones during the Shahiya Dynasty being

.......... 

They were successful; with only a couple of exceptions the Ghaznavids were able to win each confrontation with the Indian armies, sometimes taking many hundreds of prisoners, as in the encounters at Kabul and Peshawar. In addition to being prisoners of war, Indians themselves also fought in ghulams or special units with the Ghaznavids as mawālī, i.e. ‘client’ soldiers. Following Bosworth, Patricia 25 Crone (2003) describes the special Qiqaniyya regiments, Hindu Indians in ethnic units fighting as Ghazis in the armies of Islam from the earliest periods. Indeed, Indians would not have reached Trans-Oxiana (on the plain of Dandanqan near Marv) had it not been that they were also a major and important part of the Ghaznavid army and palace guard. If Mahmud’s son and successor Mas‘ud had not been sidetracked by the Oghuz Türkmen, he would not have lost the western empire, and the beginnings of Romani history would have stopped there. 

Ghulams were highly trained slave-soldiers, mostly Indian in origin, but including Khurasanis and others. The Ghaznavid army included an elite palace guard, consisting of from four to six thousand ghulam heavy cavalry. The balance of the standing army was also comprised of ghulams, bringing the core force to an estimated 30,000 people. The cavalry were armed with bows, maces, battleaxes, lances and long curved swords, though their horses went unarmored (Haider, Nicolle). Wink (1991:23) describes the “large numbers of Indian captives [who . . .] under the Ghaznavids did become important.” 

That they were used to fight for the Ghaznavids is documented by Ikram (1989:31) as well, who writes of the “Hindu contingent” of the army of Mahmud’s son Mas‘ud “fail[ing] conspicuously against the Seljuqs” during the 1038 confrontation (see also Reynolds (1858), Pipes (1981) and (2000), Bosworth (1961), Crone (1980) and Haider (1990) for descriptions of medieval Muslim armies, and Lal (1994), Levi (2002a/b) for soldiery in India). It is to those Hindus, both captives and militia, that we must look for the ancestors of the Romanies. 

Those Indian military detachments were made up of the fighters and their camp followers, the iviranugama, people recruited to tend to the duties associated with war. They generally outnumbered the soldiers themselves, and like the soldiers came from many different backgrounds and spoke many different languages and dialects. That Romanies have a mixed Indian origin is not a new idea; over a century ago de Goeje (1876) wrote of the ancestors’  

consisting in large part of the tribe of Jats, which occupied lands in the Indus Valley near Multan . . . during the Omayad Dynasty they took a great number of their families, together with their camp followers, into the regions of the Lower Tigris . . . organizing a resistance to al-Motacim’s government they were subdued and taken to Baghdad, and then taken to various places along the borders of the Byzantine Empire . . . it is from these Jats that the European Gypsies originate,  


This would have placed the date of departure at ca. AD 855. Although the Jat language, Jataki, is considerably less like Romani than is Hindi-Urdu—see Burton (1849)—a mixed (principally Jat) origin was also supported by Leland over a century ago, who wrote (op. cit. 332-333) that Romanies

speak an Aryan tongue, which agrees in the main with that of the Jats, but which contains words gathered from other Indian sources. This is a consideration of the utmost importance, as by it alone can we determine what was the agglomeration of tribes in India which formed the western Gypsy.  

Woolner (1914:123-126) has also referred to Mahmud of Ghazni’s forays, leading, he said, to the “wiping out” of the Indian dialects of Ghandāra (now northwestern Pakistan and eastern Afghanistan) and the destruction of the Jats. The soldiers themselves, whatever their social backgrounds, were given honorary warrior, or kshattriya, caste status and were called Rajputs, or “sons of princes.”
 The administrative language of both the government and the military in the Hindu Shahi kingdom during that period was medieval Persian, though the local population spoke different Indian and  Dardic languages natively; it is already widely accepted that such a situation gave rise to the Urdu language as a military lingua franca, combining elements from Persian and a number of different Indian languages. Bailey (1938:1), critically discussed at length by Faruqi (2001:45- 62) writes that “Urdu was born in 1027; its birthplace was Lahore, its parent Old Panjabi; Old Kha#ї was its step-parent.” Its very name Urdu in fact means “battlefield,” although the word did not appear in print until the late 1700s, and we can speculate that Romani began to emerge under the same circumstances; for want of a name I have called this hypothesized contact language Rajputic elsewhere (Hancock, 2000). As demonstrated above, it shares over three times as many of the same Persian words with Urdu as it does with Domari. Military terms (or terms with a military application) of Indic and Persian origin in Romani, and which have thus been a part of the language from the very beginning, include: “arrow” (sulica) < Skt śūla, Hi sūl, - + Gk –ιτζα (Balkan dialects use the Turkish derived okja for this). “axe” (tover) < cf. Hindi and Persian tabar “axe,” and Hi tarvar “sword,” Kurdish taver “battle” (kurripen) < Skt ku- + -tvana “confront, oppose” (nikl-) < Skt nik layati, Hi nik ln~ “conqueror” (idjavno) < Skt -nayati + karoti- “decamp; move out” (rad-) < Skt rah- + dad ti “defeat in battle” (vidjav-) cf. Hi vijit, vijetā “fight” (kurr-) < Skt kuayati, Hi k n~ “gaiters” (patava) < Skt patta-, Hi pa, cf. E. “puttees” “horse” (khuro) < Skt ghoa-, Hi gho~ “military” (lurdikano) < Skt l ati + “plunder” (lur-) < Skt l ati, Hi l n , cf. E. “loot,” “Luri” “set up camp” (lod-) < Skt lagyati Aslaughter” (manušvari) < Skt m nuam rik~ “soldier” (kuripaskero) < Skt kuayati + -tvana + kro “soldier” (lur, lurdo) < Skt l ati, Hi l n~ “spear, lance” (bust) < Skt v cika-, bhr ti- “spear, stab” (pošav-) < Skt spar ay , Hi phasn~ “sword” (xanrro) < Skt khaaka-, Hi khã:~ “trident” (trušul, but now meaning “cross”) < Skt triÑãla- “whip” ( upni, ukni) < Skt uknuti “battleaxe” (ni ako) < Persian na ak, cf. also Kurdish nijakh “halter” (ašvar) < Persian abz r “spur” (buzex) < Persian sbux “saddle” (zen) < Persian zin “stirrup” (bakali) has no located etymology.

Given the comparatively small number of Indic items in Romani it is significant that there are
two words in the language for silk, viz. phanrr (cf. Panjabi patt “silk,” from the same root as 
 27 patavo “gaiter,” and kež (cf. Kashmiri kheš) “silken cloth,” but also Urdu (
silk”). The military distinguished between two kinds of silk: that with a fine weave for outer
clothing, and another with a coarse weave worn as an undershirt, and designed specifically to
entangle and impede arrow-heads fired into the body (information courtesy of Adrian Marsh). 

One connection with the Ghaznavids is found in the word for a mattock (a tool with a head consisting of an axe-blade on one side and a hammer on the other), one of the symbols of authority carried by the Rrom baro. This is nižako in the Vlax dialects and njako in Balkan Romani. According to Nicolle (1996:153) . . . as usual there was considerable variety among the troops of the eastern Muslim countries, ranging from the Ghaznavids’ elite heavy cavalry armed with nachakh (axes) to the Turkish horsearchers of 13th century northern India  

The word is Iranic (cf. Gurani Kurdish ni āk “axe”, Mokri, 1951:134), and has passed into the Perso-Arabic military lexicon as nachakh (Nicolle, op. cit., 306). Various Romani populations in Europe and America also maintain nacijange semnura or group symbols, such as the sun (representing e.g. the Serbian Romanies) and the moon (representing the Lovara), which may be found drawn or carved onto the stago or ‘standard’ at a wedding, and on the sèmno or rupuni rovli (‘silver baton’), i.e. the clan leader’s staff, and which are appealed to at the consecration of the mulengi sinìja or ‘table of the dead’ at a pomàna (plural pomèni) or wake. Here, the invocation is “Khama, hona thaj Devla, ašun(en) man!” which means “Sun, Moon and God, hear me”. The significance here is the fact that the Sun and the Moon were the two symbols worn emblematically on the armour and tunics of the Rajput warriors to identify them in battle from all others. The Rajputs’ religious restriction on eating vegetables that grow below ground at a funereal feast is also maintained amongst Vlax Romanies, where potatoes and peanuts are forbidden at a pomàna. 

The military connection has been explored from the historian’s perspective by Marsh (2004) who, rather than (or perhaps in addition to) seeing the koïné as having originated within the Rajput’s own environment, says (op. cit.) 

The development of the military koiné happened, I suggest, in the Ghaznāvid armies amongst mawālī troops (‘client’ soldiers) of Indian units, Qīqanīyya or Kīkanīyya, as a result of the need to communicate across dialectical and regional differences . . . The assaults on the region of northern India following the accession of Amir, later Sultan Mahmud of Ghaznā (Mahmūd b. Sebüktigin, sometimes Mahmūd-i Zābulī 998-1030), [and his part in] the dislocation of Indian peoples as infantry, elephant-drivers, prisoners of war, craftspeople and artisans, and their incorporation into the Ghaznāvid state, centered in what is now eastern Afghanistan. These were a continuation of his father’s policy of raiding the sub-continent in assaults against Shi’ite Multan and other centres of the Ismā’īlī Muslims of Sindh, and also the pagan Hindus. The resources of India were needed in order to finance the professional, multi-ethnic standing army of the Ghaznāzvids. The expansion of Ghaznā during Mahmūd’s rule followed the important razzias of 1018-1019 against the cities of the northern Ganges, including Kannauj.    

The significance of Kannauj has been explored by Courthiade, who goes so far as to claim it to be the ultimate home of the Romanies (2004:105-124); this has been pointedly—though unconvincingly—attacked by “Im Nin’alu,” who argues elsewhere for a Jewish origin for Roma (2004:3): A recent theory that is having some success among the intellectual environment interested in the subject, and that is destined to be proven fallacious like all the preceding hypotheses, pretends to 28 have discovered the original “city” from where Roma might have come: Kannauj, in Uttar Pradesh, India . . . the author founds the entire argumentation on an alleged linguistic proof that is quite insufficient to explain the Romany cultural features not related to language and that are undoubtedly much more relevant, and not any reliable evidence is given to support his theory . 

The mixed nature of the Romani lexicon is exemplified by the numerals; one, two and three are traceable to the Central group, four is Dardic, while five and six are of mixed origin (John, 2006: 6). The language includes numbers of synonyms traceable to separate Indian dialect groups, i.e. it cannot be linked with any single Indian language but has features from several of them. There are three different words for “burn”: xačar-, thab- and phab-. The first descends from OIA ksāti, the second from daghda- and the last from *bhabh-. The first is mainly represented by the Central neo-Indic languages (Panjabi, Pahari, Jaunsari, &c.), the second mainly by members of the eastern group (Bengali, Oriya, &c.), while there are no descendants from the last other than in Romani. Except for Romani, no Indian language has descendants from all three forms, though the first and second exist in Shina, Sindhi, Panjabi, Kashmiri, Nepali and Gujerati.

There are two words for “wash,” xalav- and thov- (from OIA ksātlayati and dhauvati respectively). In India the first is restricted to Pahari and Kumauni; the second is widespread in all dialect groups. Only Kumauni (besides Romani) has both. There are two words for “sing”: gilab- and bag- (from OIA gīta- and vādyāte respectively), the former restricted to Dardic and Sinhalese, the latter to several mainly Central and Eastern Indian languages, but no language in India includes both. There are three Romani words for “to scare,” trašav-, darav- and šas-, from OIA traśati, dāryati and śāsati-; only Romani has all three. The first is restricted to Sindhi, Lahnda, Panjabi and Kashmiri, the second to Assamese and Gujerati, and the third to Bengali. The first and third occur in Nepali and Oriya, and the second and third in Hindi alone. Numbers of these synonym clusters in Romani have been collected, and their analysis is still in progress. To these, and as evidence of koïnéization may be added the two words for “two,” duj and do. These are distinguished by case, do being the oblique form (an innovation—no languages in India distinguish numerals by case). Duj is found in Nemadi, Kannauji, Pahari and Siraji, while do is found almost everywhere else; only Chhattisgarhi and Pangwali have both, though as synonyms, not as case-contrastive forms. 

We might assume that there were even more such lexical clusters among the speakers of Rajputic, some items from which were ultimately selected and others of which were discarded. This would account for the uneven distribution of some Indic items in the European Romani dialects—some restricted only to the Northern dialects (which includes Iberian) for example, and would explain why Lomavren selected e.g. hath for “hand” while Romani (and for that matter Domari) have the earlier (and Dardic) forms with /-s-/ (vast, xFst), or why Lomavren selected, inter alia, the Indic ansev, solav and pantri (“apple,” “dog,”“bird”) while Romani selected the Indic phabaj, d ukel and iriklo. 

Typical of contact languages was the development of analytical numerals in Rajputic (cf. New Guinea Pidgin English two-pela-ten-one, two-pela-ten-two, Cameroon Creole English twotali-one, two-tali-two = 21, 22, &c.) instead of its having retained the original Indian fusional forms; in the case of Romani the model was probably supplied by Persian, although for numbers above twenty the linking word u (< Persian o) is replaced by ta (< OIA atha “and, also”). In some dialects, u and ta are not used to link the Greek numerals seven, eight and nine, following the Greek model instead (e.g. 19 = δέκ’ƒννέα); Indian models for six, seven, eight and nine are also lacking in Domari and Lomavren. In all three languages, the link-word for the teens differs from that for all higher numbers:  ...... 

8. Appearance in the West

 Having established a date for a continuing presence in India, we need now to look for the earliest documentation of a Romani presence in the West, because the window of time between both dates must cover the timespan during which their exodus took place. While most earlier scholars have placed the migration out of India some time well before AD 1000, some have placed it as late as the 12th century—most recently Kochanowski, who argues for the date of departure of Rajputs following the Muslim invasions led by Mohammed Ghori in AD 11917 (2003:341). There are two likelier and earlier possibilities, the first, dated AD 1068, from Byzantium reported the presence of “Lors” in that city but that may have been a reference to Luri, i.e. Dom, rather than Romanies, but the second, dated some time in the latter part of the 1100s clearly refers to Atsinganoi and Æguptoi, then as now the most usual names for Romanies. Fraser’s important lexico-statistical analysis of Romani puts the beginnings of its linguistic split into the different dialect groups in the Byzantine Empire at around AD 1040 (Fraser, 1989), while Kalaydjieva et al.’s more recent findings place the migratory divisions into Europe beginning in the 1200s (op. cit., 6): “The Gypsy group was born in Europe. All marker systems suggest that the earliest splits occurred 20 to 24 generations ago, i.e. from the late 13th century onwards.” Yet further evidence supporting the dates of movement out of India and into Europe argued for in this study is provided by a recent examination of a number of “private” genetic mutations that are exclusively responsible for the specific disorders in the Romani people (Morar et al., 2004). One in particular is a mutation that causes congenital myasthenia which is shared by Indian, Pakistani and Romani patients and which was clearly brought out of India. Those researchers used their genetic data to posit the date of the founding of the entire Romani population, and estimated it to have taken place about 800-1000 years ago, with the subsequent splits (out of Anatolia) into individual groups occurring between 400 and 600 years ago, with no evidence of any such substructure prior to their estimated earliest date. These findings have generated some interest in the scientific community—see for example Sellah (2004) and Brownlee (2004).

9. The Seljuq Factor 

If this provides an explanation for where and how the pre-Romani population may have begun, we are left having to explain how it reached the Byzantine Empire, the period of its history barely ever addressed in the scholarship8 ; nor has the crucial period spent in Anatolia been properly acknowledged, though this is now being redressed with Adrian Marsh’s current research and Nadia Demeter et al.’s “new approach” to Romani history, which includes “a profound analysis of the Romany people’s three century sojourn in Byzantium . . . [which] period is always very cursorily dealt with in scholarly works” (2000:321-334). The Seljuqs, or rather the nomadic Sunni Türkmen who were the main military component in the Seljuq polity during this period, provide this link. They were the force behind the proto-Romani migration, driving the original Hindu army, defeated after Dandanqan (1040), ahead of them: On May 24th, 1040, the Seljukid army attacked the Ghaznavids during their advance towards the castle of Dandanakan . . . Sultan Mesud[’s . . .] command led to the disruption of the order of the Ghaznavid army and to the defeat . . . slaves from the palace left the Ghaznavid army to pass to the Seljukid side, joining the ones who had escaped before. Later, they attacked. This led to the collapse and dispersal of the already exhausted, tired and destititute Ghaznavid army (Ghzel et al., 2002: 105).

They spread out into Khorasan, Armenia and eastern Anatolia because of the complete collapse of the Ghaznavids—which, it has been argued, was “because of Mahmud’s excessive reliance on Hindu soldiers and generals” (Rikhye, 2006:2). They were part of an increasingly composite group consisting of, beside themselves, Persians, Armenians, Greeks and other “refugees” from what the 12th century Armenian historian Matthew of Eddesa described as the “perfidious nation of the Turks”. It was standard practice for the conquering armies in Mongol-Turkic warfare to push defeated populations ahead of them, in order to create fear and disruption to a maximum degree. The Seljuqs were not especially interested in Anatolia, except to distract the Türkmen from depredating the 

Persian lands and causing enormous problems for the sedentary population—which they succeeded in doing nevertheless, ultimately undermining the very fabric of their empire. Only those based in Konya (Iconium) in central Anatolia remained powerful enough eventually to set up the Rûm (i.e. “Rome”) sultanate, and only after the Mongols had invaded at Baghdad and dealt a severe blow to the Great Seljuq Empire. The small Seljuq ruling class in Rûm governed a population that was mostly Greek-speaking Anatolian Christian. Ibni-Bibi describes how the Seljuqs forced the ghulams that surrounded the sultan to flee from the field of battle during their last confrontation in his Seljukname (Bosworth, 2001: 37-39). 

As the core of the Ghaznavid military machine, and infidels as far as the Muslim Seljuqs were concerned, they could have expected no mercy upon the defeat of the Ghaznavid sultan. The camp of the Ghaznavids was sacked after their defeat by the Seljuqs and their Turkoman allies, and the remaining Hindus—the armorers, the grooms, the tent-makers, cooks, entertainers, elephant keepers, shield-bearers, women and children all of whom would have been present, since mediæval armies were societies on the move—fled westwards from the Seljuq onslaught and would seem to have arrived in eastern Anatolia fairly soon afterwards using the Silk Road from Marv, Rayy and through Khorasan into Armenia. The next encounter in the sources also indicates the presence of nomads in tents outside the city walls of Ani, the centre of the Bagratid kingdom, when the Seljuqs destroyed it in 1064 AD. Marsh suggests (in p.c.) that this was likely the last point at which the Hindus from Ghazna, together with a host of other refugees who joined them along the roads in eastern Anatolia are defeated and lose their military capacity, being reduced to the artisans, entertainers, servitors and ‘hangers on’ that any army carried along with it during this period. Matthew of Edessa describes how the this conglomeration of princes, noblewomen and the mass of people moving up and down the highways of Asia Minor after this event constituted what he described as a ‘vagabond nation.’ The destruction of the Armenian Bagratid kingdom marked the end of Byzantine control of Asia Minor and the beginnings of what Vyronis has called the Islamification of the Anatolian region, and the settlement of the Turks. The references to the arrival of the ‘Egyptians’ in Sulukule (the Mesotechion section of the walls around Constantinople near the River Lycus) is well documented and is described by Soulis (op. cit., 142-165). The loss of the Hindus’ primary military function and capacity was due to their defeat by the Seljuqs and the decimation of the warrior elite that were very close to the Ghaznavid sultans, though they always maintained their religion, or at least elements of it, and some military involvement was still in evidence at the time of the move across into Europe (see below).

Bosworth writes that “Indian troops passed from the Ghaznavid to the Seljuq armies; troops, if not formally made prisoners of war, often joined the bandwagon of the winning side” very willing to turn against their captors, while Leiser surmised that “after the Seljuqs defeated the Ghaznavids they ‘appropriated’ their prisoners of war; such action was fairly commonplace in those days” and, citing the work of the Turkish historian Köyman, which provides several sources, goes on to say that “after the victory at Dandanqan, soldiers from throughout Khurasan, ‘some of whom may have served the Ghaznavids,’ joined the Seljuqs.” Marsh (2003) also notes that Sindhi warriors : . . . had been present in the Persian lands since the early fifth century AD, often as auxiliaries to the Sassanid armies of Persia or as remnants of a defeated and ‘decapitated’ military society . . . subsequently the Seljuks of Rûm had acquired large numbers of these troops and their retinues, in the aftermath of their defeat of Mahmud’s heir, Mas’ud on the steppes of Dandanaqan, 23rd May 1040. These combatants were, as all armies in the early mediaeval period, effectively societies on the move, with the fighting force making up approximately one third of the total number. The rest would have been the armourers, grooms, smiths and metalworkers, carpenters, military engineers, servants and servitors, tent-makers, cooks, bakers, washer-women, slaves, camp-followers and children.   

Located to the south-east of the Byzantine Empire, Armenia fell to armies directed by the Seljuqs in AD 1071 and the foundation was laid for the establishment of a new sultanate called Rum, occupying former Armenian and some Byzantine territory in Anatolia—the area that is today Turkey. Fraser, supporting the conclusion reached in the important earlier work of Soulis 32 wrote that “the appearance of the Gypsies in Byzantine lands is undoubtedly connected with the Seljuk raids in Armenia” (emphasis added), though he would have been nearer the mark if he had called these instead Thrkmen raids. It is also probably more accurate to regard the incoming Indians as part of the co-opted Ghaznavids and as allies, with the Armenians as the antagonists, and not as joint protagonists along with the Seljuqs. 

Marsh goes on to suggest that the establishment of groups who were to become Romanies in Anatolia was the result of the Seljuqs’ policy of establishing beyliks, that is to say granting autonomous fiefdoms within Rûm to bands of their warriors. While this early connection with Sindh is well documented, it must be taken as geographical rather than as necessarily linguistic, since many languages besides Sindhi are (and were) spoken in that part of India. The comparative wordlists above do not demonstrate a particularly close lexical relationship between Romani and Sindhi. 

While it is documented that “Indians” were brought into Byzantine territory by the Seljuqs “usually in a military capacity,” nowhere are those Indians referred to specifically as either Rajputs or Rom. We would not expect the former, since it is an Indian word and only a minority of the Indians would have been Rajputs in any case, and if, as is proposed here, the Romani population did not come into existence until the Byzantine period, then “Rom” had not yet become a label.

One self-designation found among Romanies in the Balkans is Romi—a Slavic plural, though the word is also found in non-Slavic-speaking Romania. Romi (< √:≅4, √:∀4≅4) seems to have been used consistently in Byzantium after about 1070 to refer to all the inhabitants of Rum, especially after the history of Michael Attaliates, who wrote from the perspective of a military official in a twelve-year period leading up to Manzikert (1071). The Arabs used it in its Arabic form (Rûmi) from about the same time, as part of the titles in diplomatic correspondence between the Mamluk Sultans and the Byzantine Emperors, according to a recent article by Korobeinikov (2004). It was probably used regularly in Europe in the aftermath of the Great Schism (1054), as focus turned to concerns for crusades and the Holy Lands. Ibn Battuta refers to Rum in his mid-14th century account of travels through Asia Minor to Central Asia (Gibb & Bechingham, 1994) and specifically mentions that from the earliest times (i.e. since the contact between Arabs and Byzantium), it was called Rum to designate the lands ruled, or once ruled in his time, by the Romans. In that sense, it seems that the Arab invasions and conflicts with the Byzantines in the 7th and 8th centuries probably resulted in the emergence of the notion of the Rhomaioi or Rhomoi by the eleventh century. Marsh argues that the Indian population in Anatolia became ethnicized into the Romiti by the 1300s, i.e. into the forerunners of the Romani people. Significantly, there is a population of nomadic metal workers living in the West Bank who are referred to as ‘Kurds,’ but who call themselves r m or r mat (Matras, 1999:7).9 

IoviŃă & Schurr (2004: 275), whose very valuable article on genetic evidence for Romani identity goes far to support the proposals in the present paper, do question specifically my military koïné hypothesis (Hancock, 2000), arguing that “the defeat of the Rajputs by the Muslim Ghaznavids in the 12th century . . . is difficult to reconcile with historical data that places Gypsies in the Byzantine Empire before this time—around the 10th-11th centuries.” However, they have confused Kochanowski’s later dates with my own; the Ghaznavid invasions took place between AD 1000-1027, not in the 12th century as Kochanowski has it; furthermore, we cannot be sure that the earliest references to people identified as “Gypsies” in the Byzantine Empire were Rom rather than Dom. 

The late and much missed Milena Hübschmannová (2000; 2004) too was bothered by the 33 time frame, questioning the linguistic evidence because it supports an 11th century exodus when the presence of Romanies in Byzantium is also recorded for the same century: Roma professor Ian Hancock le Redžosko of Texas University in Austin (USA) believes that Roma – originally Rajput fighters whose army was composed of a great variety of castes – left India because of the Muslim invasions. Troupes led by General al Qasim began the invasions with the conquest of Sindh in 712. The wars peaked with twenty-one border raids led by Mahmud of Ghazni (beginning of 11th century). Hancock dates the departure of the Roma from India back to exactly the time of Mahmud. Hancock bases his theory on linguistics: according to him, modern Indic languages lost their neuter gender, and neuter words were absorbed by the masculine gender. Hancock (2001) presents a table in which he compares the genders of Romani words with related Hindi words. His theory is interesting, but if Athingani (very probably Roma) were already in Byzantium in the eleventh century, they would scarcely have left India in the same century      

Of course the loss of the MIA neuter isn’t according to Hancock, but to the specialists in this area such as Masica and Bloch, mentioned above. I have simply applied it to the Romani case. Historical evidence points to the Seljuqs, or their main fighting force the Thrkmen, as the link accounting for this span of time (which easily allows for such a migration to have taken place within a century); they defeated the Ghaznavids in AD 1038 and AD 1041 and took their prisoners of war to use as their own fighting force and as a front-line “buffer” in their move towards Anatolia. Tsaggas (2006:21-23) writes about the Seljuqs’ arrival in Bagdad, Ani, Manzikert, Edessa, Nicaia, Antioch, Tripoli and Jerusalem in company with Indians from the Panjab (“Pentopotamia”), probably acquired in Khurasan or Manzikurt from the Ghaznavids. Lee has recorded seldjuko meaning “Turk” in the speech of his principal Kalderash-speaking informant Russell Demetro.

Like the Shahi administrators before them, the Seljuqs too used Persian as their lingua franca. Kjeilen (2003:1) says The Seljuqs made Esfahan their capital, and they started to use the Persian language in the administration of their new state. The Seljuq sultans also sponsored Persian, and they were effectively propagators of the language to the entire Persian continent.  

10. Anatolia and the Emergence of the Romani People and Language 

Almost a century ago Colocci (loc. cit.) saw the move from India to the Byzantine Empire as having been “very rapid;” but if that took only two or three decades, the stay in Anatolia itself lasted for over two centuries, and was crucial to the emergence of the Romani people. As an already ethnically and linguistically mixed population, bound together by former occupation and now social circumstance, the Indians not only intermarried with each other but with the local people as well. Social and caste barriers to marriage that are strictly maintained within India become relaxed in diasporic Indian populations; this is clearly evident in Indian communities in e.g. Fiji, Mauritius or Trinidad. 

Byzantine society was ethnically diverse and included many different peoples and languages, though the lingua franca was Greek and the national religion Orthodox Christianity. It may be relevant that while Greek was the everyday language of Constantinople (Byzantium), situated on the European side of the Bosporus, and was the administrative language throughout the Empire, it was not the only language spoken at the popular level throughout the rest of the land. Children newly born into this community must have been exposed to a variety of  languages, including the Rajputic of their own parents and the Greek being spoken all around them. We may well suppose that the Romani language, and the Romani people, came into existence in the Byzantine Empire during this time; this being the case, reconstructing protoRomani as a discrete pre-Byzantine Indian language is not possible, though a more detailed description of Rajputic is underway. 

The influence of Byzantine Greek in the makeup of the Romani language cannot be underestimated (see Grant, 2003); not only does it constitute the second largest percentage of the pre-European vocabulary after the Indian words, being found in every semantic area (even in the numerals), but it has also contributed to fundamental areas of the grammar, such as the different words for the definite article ‘the’, losing the Indian grammatical feature of ergativity, and the change of the basic NIA syntactic ordering from subject-object-verb to subject-verb-object. The middle voice which Rajko Djurić argues is evidence of Romani’s great age may equally well have been acquired from Greek, as well as the shift of the Indic dative to the Balkan accusative. Athematic final and non-final affixes of Greek origin include inter alia –in, -os, -is, -mos, -mata, -itza (also Slavic), -itko, -me(n), verbal –as, -is, -azo, -izo, -isar- and -ar-. The synthetic construction modeled on Greek Β∴≅ (in Romani po, relexified by maj, meg, &c., in other dialects) before comparative adjectives (po-baro “bigger”) replaces—or was selected—in some dialects rather than the Iranic/Ossetic enclitic –der (bareder “bigger”; see Hancock, 1995:33 for further discussion of this). A semantic calque is found in the Romani verb be -, which means both “sit” and “reside,” a dual reference not found in any Indian language, but paralleled in Byzantine Greek 6∀2∴.Τ “sit; settle in a place; (of an army) encamp” (Liddell & Scott, 1980:339)10 .

11. Into Europe 

The main move up into Europe was also the result of Islamic expansion, this time initiated by the Ottoman Turks, who eventually sacked Byzantium in AD 1453 and extended their influence up into the Balkans, though it would be wrong to think that this migration happened all at one time. The bubonic plague (the “black death”) had reached western Anatolia by 1347 for instance, and forced a general migration across into Europe that surely included some Romanies, since they were blamed for having introduced it. Linguistic evidence points to the Romani language existing in three distinct overlapping strata across Europe (see Courthiade, 1994); there are very few Greek words, including the definite articles, in at least one European Romani dialect (Istriani, spoken in Slovenia, see Cech & Heinschink, 2001), suggesting a very early move out of Anatolia before the heavy lexical impact of Greek had affected it11 .  

Not only was Islam a key factor in the move into Europe, as it was in the move out of India, but both events also shared a military aspect, since the Ottoman Turks used the Romanies “as direct participants (in their militia), mainly as servants in the auxiliary detachments or as craftsmen servicing the army” as Marushiakova & Popov have written. By the 1300s, there were specifically military garrisons of Romanies at both Modon and Nauplia, in Venetian Peloponnesia, today southern Greece. The Romanies had arrived in Europe. We do not know how the various groups of Romanies first entered Europe. Most presumably crossed the isthmus at Constantinople, though it has been suggested that others left Anatolia by boat across the Aegean or even the Black Sea (Gheorge, 1983: 13). In whatever way they reached the Balkans, they continued to move on in all directions, being reported in almost every country in Europe by 1500.

12. Questions 

I am well aware that these hypotheses have been challenged by some of my colleagues, and I welcome that. We are all working towards the discovery and documentation of Romani history, and if theories can be shown to be baseless, then we can eliminate those lines of pursuit and move on in other directions. So far, however, I have not seen any specific counterarguments (though some relevant questions are raised in Matras, 2004b), and would like the following points to be addressed. Perhaps a future conference might be organized to deal solely with these: •If the migration out of India pre-dated AD 1000, how may we account for the reassignment of formerly neuter nouns in Romani and their matching reassignment in languages still spoken in India, as well as for other neo-Indic characteristics of the language? •If the migration through Persia and the acquisition of Persian words took place in the 5th century, why are all such items in Romani, Lomavren and Domari from 
Modern (i.e. post-9th - 10th century) Persian?12 •If the ancestors of the Romanies were not a military force, how may we account for the significant number of military terms of Indian origin in Romani, and the corresponding paucity of e.g. agricultural terms? If they were military but not Rajputs, who else could they have been? Consider also the further non-linguistic arguments for Rajput identity made in Hancock (2000), summarized here: 


Ahmad, Fazl, 1999. Mahmud of Ghazni. Karachi: Sh. Muhammad Ashraf.
Bailey, Grahame, 1938. Studies in North Indian Languages. London: Lund Humphries & Co.
Bakker, Peter, 1995. “The Northern branch of Romani: Mixed and non-mixed varieties,” in Y. Matras (ed.), 1995.
Romani in Contact. Amsterdam & Philadelphia: Benjamins.
Barany Zoltan, 2002. The East European Gypsies. Cambridge: The University Press.
Bhalla, Vijender, 1992. “Ethnicity and Indian origins of Gypsies of Eastern Europe and the USSR: a bio-
 Anthropological-perspective,” in K. Singh, ed., Ethnicity, Caste and People, Moscow & Delhi.
Blachut, Kenneth B., 2005. Transient Criminal Subcultures and the Crimes they Commit Against the Elderly:  Implications for Continuing Criminal Justice and Community Education. Unpublished D.E. dissertation,  University of Illinois.
Bloch, Jules, 1953. Les Tsiganes. Paris: Presses Universitaires de France.
Bloch, Jules, 1965. Indo-Aryan, from the Vedas to Modern Times. Paris: Maisonneuve.
Boretzky, N., & B. Igla, 1994. W`rterbuch Romani Deutsch Englisch. Wiesbaden: Harrassowitz.
Bosworth, C. Edmund, 1961. “Ghaznevid military organization,” Der Islam, 63:37-77.
Bosworth, C. Edmund, 1963. The Ghaznavids: Their Empire in Afghanistan and Eastern Iran 994-1040. Edinburgh: The University Press.
Bosworth, C. Edmund, ed., 2001. The History of the Seljuq Turks, from the Jani al-Tawarik: an Ilkhanid Adaptation  of the Saljuq-nama of Zahir al-Din Nishapuri. Translated by K. A. Luther. Richmond: Curzon Press.
Burton, Sir Richard, 1849. “A grammar of the Jataki-Belochki dialect,” Journal of the Bombay Branch of the Royal Asiatic Society, 3(1):81-125.
Brownlee, Christen, 2004. [“Genetic evidence of Gypsies’ Indian origins”], Science News: The Weekly News Magazine of Science, September 20th
.
Burton, Sir Richard, 1849. A Grammar of the Játakí Dialect. Bombay: American Mission Press.
Burton, Sir Richard, 1851. Sindh, and the Races that Inhabit the Valley of the Indus. London.
Burton, Sir Richard, 1875. “The Indian affinities of the Gipsies,” The Academy, 7:324-325.
Burton, Sir Richard, 1898. The Jew, the Gypsy and EI Islam. Hollywood: Thew Angriff Press.
Cech, Petra & Mozes Heinschink, 2001. “Sinti Istriani, Lički Šijaci, Gopti und Cigani Brajdiči: Die Dolenjski Roma  in Slovenien,” in Igla, Birgit & Thomas Stolz, eds., Was Ich Noch Sagen Wollte . . . A Multilingual  Festschrift for Norbert Boretzky on the Occasion of his 65th Birthday. Berlin: Akademie Verlag, pp. 341-368.
Clarke, A., 1878. AOrigin and wanderings of the Gypsies,@ The Edinburgh Review, 148:117-146.
Colocci, Adrian, 1907. Review of De Goeje (1903). Journal of the Gypsy Lore Society, new series, 1:278-280.
Cortiade, Marcel, 1994. Phonologie des Parlers Rom et Diasystème Graphique de la Langue Romani. Unpublished doctoral dissertation, University of the Sorbonne.
Cortiade, Marcel, 2002. The Rromani People’s Origins: Chronicles and Legends. Paris: INALCO.
Courthiade, Marcel, 2004. “KannauZ on the Ganges, cradle of the Rromani people,” in Donald Kenrick, Gypsies: From the Ganges to the Thames. Hatfield: The University of Hertfordshire Press.
Kjeilen, Tore, 2003. “Seljuqs.” Encyclopedia of the Orient. Stockholm: LexicOrient., pp. 105-124.
Crone, Patricia, 1980. Slaves on Horses: The Evolution of the Islamic Polity. Cambridge: The University Press.
Crone, Patricia, 2003. “The pay of client soldiers in the Umayyad period,” Der Islam, 80(2):284-300.Demeter, Nadia G., N.V. Bessonov & V.K. Kutenkov, 2000. Istorija Cigan: Novij Vzgljad. Moscow: Rossiskaja Akademija Nauk.
Djurić, Rajko, 2003. “The language of the Roma: the oldest language still spoken.” Berlin: PEN Centrum press rrelease, January 25th.
Edessa, Matthew of, 1869. Patmouthiun-Mattheosi Ourhayetsuos. [12th century chronicles]. Jerusalem : Athorhoy S.
Yakovbeants, 1869.
ElÓRk, Viktor & Yaron Matras, 2006. Markedness and Language Change: The Romani Sample. Berlin & New
York: Mouton de Gruyter.
El-Liethy, Imam, 2003. “Iraq’s Gypsies struggle for life after Saddam’s fall,” Islam on Line, Baghdad, May 6th
.
Faruqi, Shamsur, 2001. Early Urdu Literary Culture and History. Oxford: The University Press.
Fonseca, Isabel, 1996. Bury Me Standing: The Gypsies and their Journey. New York: Knopf.
Fraser, Angus, 1989. Looking into the Seeds of Time. Paper read at the annual meeting of The GLS, Toronto, 7-9 April.
Fraser, Angus. 1992. The Gypsies. Oxford: Blackwell. 
 40
Gafarová, Svetlana, 2003. “Central Asian Gypsies,” The Times of Central Asia, July 16th issue, pp.1-2.
Gheorghe, Nicolae, 1983. “Origins of Romas’ slavery in the Rumanian principalities,” Roma, 7(1): 12-27.
Gibb, H., & C.F. Bechingham, 1994. The Travels of Ibn Battuta 1325-1354. London: The Hakluyt Society.
Gjerdman, O., & E. Ljungberg, 1963. The Language of the Swedish Coppersmith Gipsy Johan Dimitri Taikon. Uppsala: Lundequist.
Goeje, Michael-Jan de, 1876. “Bijdrage tot de geschiedenis der Zigeuners.” Koninklijke Akademie van Wetenschappen: Verslagen en Mendeelingen, 2(5):56-80.
Grant, Anthony, 2003. “Where East meets West: Observations on a list of Greek loans in European Romani,” in A.P. Grant (ed.), Papers in Contact Linguistics (= Interface: Bradford Studies in Language, Culture and Society, 6), pp. 27-70.
Ghzel, Hasan Celal, C. O uz & Osman Karatay, eds., 2002. The Turks: Middle Ages. Ankara: Yeni Thrkiye Research and Publishing Center. 2 vols.
Haider, Syed Zafar, 1990. Islamic Arms and Armor of Muslim India. Lahore.
Haliti, Bajram, 2006. India and Roma. Kosovo: Central Office of Roma.
Halwachs, Dieter, 2000. “Romani: Attempting an introductory overview,” Romani in Austria. Graz: Institut für Sprachwißenschaft [www-gewi.kfunigraz.ac.at/romani].
Halwachs, Dieter & Florian Menz, eds., 1999, Die Sprache der Roma. Klagenfurt: Draga Verlag, pp. 172-209.
Hancock, Ian, 1984. “The social and linguistic development of Angloromani,” in Thomas Acton & Donald Kenrick (eds.), Romani rokkeripen to-divvus: The English Romani dialect and its contemporary social, educational and linguistic standing. London: Romanestan Publications, pp. 89-122.
Hancock, Ian, 1988. “The development of Romani linguistics,” in Ali Jazyery & Werner Winter,
eds., Languages and Cultures: Studies in Honor of Edgar C. PolomJ. Mouton: The Hague., pp. 183-223.
Hancock, Ian, 1995. “On the migration and affiliation of the Rom, Lom and Dom Gypsy,” in Y. Matras (ed.), Romani in Contact. Amsterdam & Philadelphia: Benjamins, pp. 25-51.
Hancock, Ian, 2000. “The emergence of Romani as a koïné outside of India,” in Thomas Acton (ed.), Scholarship  and the Gypsy Struggle, Hatfield: The University of Hertfordshire Press.
Hancock, Ian, 2002. We Are the Romani People: Ame Sam e Rromane Džene. Hatfield: The University of  Hertfordshire Press.
Hancock, Ian, 2003. “Language corpus and language politics: the case of the standardization of Romani,” in Daftary, Farima, & FranHois Grin, eds., Nation Building, Ethnicity and Language Politics in Transitional Countries. Budapest: LGI Publishers, pp. 269-286.
 Hancock, Ian, 2006. “Gypsy languages/Zigeunersprachen,” in Ulrich Ammon, Norbert Dittmar, Klaus Mattheier and Peter Trudgill, (eds.), Sociolinguistics: An International Handbook of the Science of Language and Society. Berlin & New York: Walter de Gruyter. Vol. III, pp. 1870-1873.
 Higgie, Brenda, 1984. Proto-Romanes Phonology. Unpublished doctoral dissertation. Austin: The University of Texas.
 Hübschmannová, Milena, 2000. “On the beginnings of Roma history,” in Zdeňka Žařabová & Eva Davidová, (eds.), Life in Black and White, Prague: Gallery Publications., pp. 17-28.

Hübschmannová, Milena, 2004. “Origin of Roma,” Rombase Network, history section, Graz (http://
 www.unigraz.at/rombase/net/).  Ikram, S.M., 1989. History of Muslim Civilisation in India and Pakistan. Lahore: Star Books.
 Im Nin’alu (= Sándor Avraham), 2004. The True Origin of Roma and Sinti. The Romanestan Webring (http://www.geocities.com/romanestan/).
 John, Vijay, 2006. Indian Origins of Romani Numbers: A Case for KoVnJization. Romani Studies term paper, The  University of Texas, Fall semester.
IoviŃă, Radu, & Theodore Schurr, 2004. Reconstructing the origins and migrations of diasporic populations: the case of the European Gypsies,” American Anthropologist, 106(2):267-281.
Kalaydjieva, Luba, Bharti Morar, Raphaelle Chaix & Hua Tang, 2005. “A newly discovered founder population: the Roma/Gypsies,” BioEssays, 27(10):1084-1094.
Kaufman, Terrence, 1984. Explorations in Proto-Gypsy Phonology and Classification. Paper presented at the Sixth South Asian Languages Analysis Round-Table. Austin.
Kenrick, Donald, 1977. A contribution to the Early History of the Romani People. Romani Institute Occasional Paper No. 3. London.
Kenrick, Donald, 1994. Les Tsiganes, de l’Inde à la Méditerranée. Paris: Centre de Recherches Tsiganes.
Kenrick, Donald, 2000. “Learning Domari: Unit 2.” Kuri-DR Journal, 1(3):2-5. 
 41
Kochanowski, Vania de Gila, 1968. “Black Gypsies, white Gypsies,” Diogenes, 63:2 -47.
Kochanowski, Vania de Gila, 1990. “Aryan and Indo-Aryan migrations,” Diogenes, 149:122-145.
Kochanowski, Vania de Gila, 1994. Parlons Tsigane: Histoire, Culture et Langue du Peuple Tsigane. Paris: ’Harmattan.
Kochanowski, Jan, 2003. Précis de la Langue Romani Littéraire. Paris: L’Harmattan.
Korobeinikov, Dimitri, 2004. [Article in] Al-Masaq: Islam and the Mediaeval Mediterranean.
Köyman, A., 1957. “Büyük Selçuklu imparatorluğu’nun kuruluşu,” DTCF Dergisi, 15:29-150.
Knudsen, Marko, 2003. Die Geschichte der Roma. Hamburg: RomaBooks.
Lal, Kishori., 1994. Muslim Slave System in Medieval India. New Delhi: Aditya Prakashan.
Leiser, Gary, 1988. A History of the Seljuks. Carbondale: Southern Illinois University Press.
Leland, Charles Godfrey, 1882. The Gypsies. New York: Houghton, Mifflin & Co.
Levi, Scott C., 2002a. The Indian Diaspora in Central Asia and its Trade, 1550-1900. Leiden: Brill.
Levi, Scott C., 2002b. “Hindus beyond the Hindu Kush: Indians in the Central Asian slave trade,” Journal of the Royal Asiatic Society, third series, 12(3):277-288.
Levy, V., 2000. “Proti Rom m stejn jako kdysi proti n
Lesný, Vincinc, 1915-1916. “Über die langen Vokale in den Zigeunerdialekten,” Zeitschrift der Deutschen Morgenländischen Gesellschaft, 70:417-422.
Liddell, A.K., and Robert Scott, 1980. A Lexicon, Abridged from Liddell and Scott’s Greek-English Lexicon. Oxford: The Clarendon Press.
Macalister, R., 1914. Language of the Nawars or Zutt, the Nomad Smiths of Palestine. Edinburgh.
Marsh, Adrian, 2003. Ottoman Gypsies or Gypsies in the Ottoman Empire: A Brief Summary and Some New Perspectives on Romani History. Unpublished monograph, The University of Greenwich.
Marsh, Adrian, 2004. History, Historiography and Identity; Gypsies in the Late Ottoman Empire. Unpublished doctoral dissertation. London: Greenwich University.
Marushiakova, Elena, & Vesselin Popov, 1997. “The Romanies in the Balkans during the Ottoman Empire,” Roma, 47:63-72.
Marushiakova, Elena, & Vesselin Popov, 2000. Gypsies in the Ottoman Empire. Paris: Interface.
Masica, Colin, 1991. The Indo-Aryan Languages. Cambridge: The University Press.
Matras, Yaron, 1999. “The state of present-day Domari in Jerusalem,” Mediterranean Language Review, 11:1-58.
Matras, Yaron, 2002. Romani: A Linguistic Introduction. Cambridge: The University Press.
Matras, Yaron, 2004a. “Typology, dialectology and the structure of complementation in Romani,” Trends in Linguistics, Studies and Monographs: Dialectology Meets Typology, 153:277-304.
Matras, Yaron, 2004b. “The role of language in mystifying and demystifying Gypsy identity,” in Nicholas Saul & Susan Tebbutt, eds., The Role of the Romanies: Images and Counter-Images of ‘Gypsies’/Romanies in European Cultures. Liverpool: The University Press, pp. 53-78.
Matras, Yaron, 2004c. “A Conflict of Paradigms,” Romani Studies, 5th ser., 14(2):193-209.
Miklosich, Franz, 1874. Beiträge zur Kenntnis der Zigeunermundarten. Vienna.
Mokri, Mohammed, 1951. Gurani. La Grande Assemblée des Fidèles de Vérité. Teheran.
Morar, B., D. Gresham et al., 2004. “Mutation History of the Roma/Gypsies,” The American Journal of Human Genetics, 75:596-609.
Mróz, Lech, 1992. Geneza Cyganów i Ich Kultury. Warsaw: Wydawnictwa Fundacji ‘Historia Pro Futuro.’
Nicolle, David, 1996. Medieval Warfare Source Book: Christian Europe and its Neighbours. London: The Brockhampton Press.
Nicolle, David, 1997. The Armies of Islam, 7th-11th Centuries. London: Osprey Books.
Nicolle, David, 1998. Armies of the Caliphates, 862-1098. London: Osprey Books.
Pipes, Daniel, 1981. Slave Soldiers and Islam: The Genesis of a Military System. New Haven: Yale University Press.
Pipes, Daniel, 2000. Military Slaves: A Uniquely Muslim Phenomenon. Paper presented at the Conference on the Arming of Slaves from the Ancient World to the American Civil War, Yale University, New Haven, November 16-18.
Price, Glanville, 2000. “Romani,” in his Languages in Britain & Ireland, Oxford: Blackwell, pp. 207-212.
Reyniers, Alain, 1998. Tsigane, Heureux Si Tu es Libre!, Série Memoire des Peuples. (Avec CD-ROM). Paris: Editions Unesco.
Reynolds, James, 1858. The Kitab-i-Yamini. London.
Rikhye, Ravi, 2006. Mahmud Ghaznavi’s Seventeen Invasions of India. [http://orbat.com/site/kings_master/ kings/Mahmud_ghaznavi/Mahmud%20Ghaznavi.html]. 
 42
Rishi, Weer R., 1976. Roma: The Panjabi Emigrants in Europe, Central and Middle Asia, the USSR and the Americas. Patiala: Panjabi University Press. Chapter 13, “Panjabi love for Buffalo’s milk”.
Sampson, John, 1923. “On the origin and early migration of the Gypsies,” Journal of the GypsyLore Society, third series, 2(4):156-169
Sampson, John, 1926. The Dialect of the Gypsies of Wales. Oxford: Clarendon Press.
Sellah, Anna, 2004. “Study: Gypsies came out of India,” ABC Science Online, September.
Sen, Sukumar, 1960, A Comparative Grammar of Middle Indo-Aryan, Poona: Linguistic Society of India.
Sinclair, Albert T., 1909. “The word ‘Rom’,” Journal of the Gypsy Lore Society, n.s., 3:33-42.
Soulis, George, 1961. “Gypsies in the Byzantine Empire and the Balkans in the late Middle Ages,” Dumbarton Oaks Papers, 15:142-165.
Sway, Marlene, 1988. Familiar Strangers: Gypsy Life in America. Urbana: University of Illinois Press. Tcherenkov, Lev & StJphane Laederich, 2004. The Rroma, Otherwise Known as Gypsies, Gitanos, ∋βνϑ≅4, Tsiganes, igani, Gingene, Zigeuner, BohJmiens, Travellers, Fahrende, etc. Basel: Schwabe Verlag. 2 vols.
Tod, James, 1920. Annals and Antiquities of Rajasthan, or the Central and Western Rajput States of India. Oxford: The University Press. In three volumes.
Tsaggas, Nikos, 2006. 9ς<ϑ.46γ∆ϑ - / Υ∆Π0 ϑ≅∴ ΙΞ8≅¬Η ϑ≅∴ 9γΦς4Τ<46 athens:="" div="" ekdoseis="" gkovosti.="">
Turner, Ralph, 1927. The Position of Romani in Indo-Aryan. Monograph of the Gypsy Lore Society.
Voskanian, Vardan, 2002. “The Iranian loanwords in Lomavren, the secret language of the Armenian Gypsies,” Iran and the Caucasus, 6(1/2):177-188.
Vekerdi Jószef, 1981. “On the social prehistory of the Gypsies,” Acta Orientalia Academiae Scientiarum Hungaricae, 35(2/3):243-254.
Vekerdi József, 1988. “The Gypsies and the Gypsy problem in Hungary,” Hungarian Studies Review, 15(2):13-26.
White, Karin, 2000. “Metal-workers, agriculturalists, acrobats, military people and fortune tellers: Roma (Gypsies) in and around the Byzantine Empire.” De Gouden Hoorn 7(2):1-9.
Windfuhr, Gernot L., 2000. “Gypsy dialects,” in Ehsan Yarshater (ed.), Encyclopædia Iranica. London: Routledge and Kegan Paul, Vol. XI, fascicle 4, pp. 415-421.
Wink, André, 1999. Al-Hind: The Making of the Indo-Islamic World. Leiden: Brill, 2 vols. Vol. I, “Early Medieval India and the Expansion of Islam, 7th-11th Centuries.”
Wogg, Michael, (ed)., 2006. Roma History: From India to Europe. Strasbourg: Council of Europe Directorate General IV, Education of Roma Children in Europe.
Woolner, A.C., 1913-1914. “The Indian origin of the Gypsies of Europe,” Journal of the Panjab Historical Society, 2:118-137.
Woolner, A.C., 1915-1916). “Studies in Romani philology,” Journal of the Gypsy Lore Society, 2(9):119-128.