In an attempt to reconstruct the process following the Roma’s migration into the Persian Empire, research is dependant on linguistic evidence. John Sampson, the first scholar to classify the Romani dialects (1923), based his studies on the assumption that the Indian immigrants arrived in Persia simultaneously, eventually forming a homogenous group that shared the same language. In his view, during its long-term residence in Persia, this initially homogenous group split up in two. Subsequently, the two subgroups went in different directions and developed distinct linguistic features based on their surroundings:
One group went from Persia to Syria and the Mediterranean. The Dom/Nawar presently still living in the Near East (Palestine, Syria, Jordan) as well as the Kurbat (Northern Syria), the Halebi (Egypt and Libya) and the Karachi (Asia Minor, Persia) are held to be their descendants.
Depending on their variety of the word "sister" influenced by Arabic, Sampson summed up their dialects as the so-called ben group. His theory moreover claims that the predecessors of today’s Armenian and European Roma formed the other part of the "Proto-Roma", who moved on to the North and North West, in the direction of Armenia.
Through contacts with their Armenian-speaking surroundings, their dialects over time took on phonetic features from Armenian, as in the case of the Sanskrit word bhen, which eventually changed to phen, the designation for the second group of dialects.
Turner (1926) made a similar distinction. His terminology, however, grounded itself on different varieties of the word domba. He accordingly called the language of the
Dombari, the dialects of the groups inhabiting the Near East Domari. Within the phen group, Turner differentiated between Lomavren of those who stayed in Armenia, and Romani, spoken by the Roma who later moved on to Europe by way of Byzantine
More recent investigations give reason to doubt the notion that the two or three dialect groups mentioned above ever actually formed a unit. The ben dialects (Domari), for instance, show traces of a lost third grammatical gender that the phen dialects (Lomavren and Romani) do not account for in any way. This could mean that the predecessors of the Lomavren/Romani group left India at a later point in time, when this gender had already disappeared entirely (→ Origin and India). Today, the existence of a homogenous phen group is likewise doubted.
Apart from these problems, questions arise as to when the assumed split up actually took place and how long each group stayed in Persia. The significant linguistic differences existing between the ben and phen Roma allow for the assumption of a split-up at a relatively early point of time. The lack of Persian loanwords in the ben dialects gives reason for the conclusion that this group left Persia first. When exactly they did so and under which circumstances they arrived in Syria, remains unclear.
Emigration from Persia
Due to lack of evidence, any answer to the question of when exactly the predecessors of the European and Armenian Roma left Persia, depends on the assumption of how fast and to what degree the language of the Arab conquerors entered into the lexicon of the Persian people.
For a long time, the lack of Arab words in European Romani was held to prove that the Roma must have left Persia before or shortly after the Arab invasion, at the latest, around the middle of the 8th century. Yet, recent studies contradict this theory.
One reason is that, as opposed to former claims, Romani in fact does contain several, if not many, Arab loanwords. Secondly, the assumption that the conquest of Persia and the introduction of Arabic as an official language within only a few years or decades brought about severe changes in the Persian people’s lexicon, simplifies complex linguistic processes and thus proves to be unreliable.
It is necessary to take into account that the Arab influence manifested itself mostly in the bigger cities and higher social spheres. Arab was spoken most of all by the few intellectuals and holders of religious offices. At no point in time did it ever actually replace Persian in its function as every day language. It certainly must have taken some time until the culture of the new leaders even to a small degree reached more distant regions, and individual Arabic expressions entered into the language of the lower levels of society.
Moreover, the Arab seize of power in Persia did not necessarily result in increased cultural contacts for the Roma. Even the fact that some Roma served in the Arab armed forces as metal workers, musicians or salesmen, is no proof of fundamental Arab influence. In the Arab army, only the commanders were actually Arabs, the different peoples of Persia made up the remainder. (→ Origins of Roma)
Taking into account the situation of the time, and considering the duration of linguistic processes, it is possible that the Roma left Persia at a later point than generally assumed. It can only have taken place at a time when the Arab influence was strong enough for the dialect of the former Indian immigrants to have taken on Semitic words. However, on the basis of the information presently available it remains impossible to establish more exact time frames.
Thanks to linguistic evidence, it can be stated with certainty that the phen Roma’s migratory route to Europe led through Armenia, where they remained over a longer period of time. For along with a number of Persian loanwords, European Romani also contains several expressions from Armenian. An even more significant indication is the fact that the Armenian language did not only influence the lexicon, but also the pronunciation of Romani. The change of the aspirated, voiced sounds in Sanskrit (bh) to voiceless aspirated sounds (ph) in Romani can only have taken place on the basis of long-term close contact with speakers of Armenian.
Curiously, the Lomavren of the Bosa living in Armenia and the Southern Caucasus lacks these Armenian loanwords adopted by Romani. Although both dialects are characterised by Armenian influence, they do not share even one common word of Armenian origin.
This again suggests that the phen group may have split up previously to being fully exposed to Armenian influence. It can not be ruled out that linguistic processes shared by both Lomavren and Romani actually took place separately at a later point. On the other hand, the existing differences can be used as an argument against the notion of a homogenous phen group.
The emigration of the Roma from Armenia seems to have happened gradually. At the beginning, the devastation of the country by the ongoing conflicts between Greeks and Arabs may have played an important role. When at the beginning of the 11th century Western Armenia was integrated into the Byzantine Empire, the conditions were especially favourable, since at that time the Roma were able to move west without having to cross a border. This may explain the lack of written documents about the Roma’s immigration to the Byzantine Empire.
Later, the attack of the Seldschuk on Armenia must have triggered a wave of migration to the Byzantine Empire. With the defeat of the Byzantine army at Manzikert (1071) and the consequent loss of Anatolia, the refugees seem to have moved further west to Constantinople and Thrakien. From there, they went on to the Balkans and Central Europe.