If we restrict ourselves to dialects spoken by Rroma groups which are of Rromanes origins, up and including the Southern Slavic influences, and neglect the creolised languages, the question of the "degree of purity" of the dialect spoken by different groups of Rroma arises.
Be it Rroma themselves or experts, there have been discussions on that topic and several statements about which dialect was the "purest". This is a rather ridiculous approach and leads to quite a lot of nonsense. The status of Rroma groups relative to one another comes to a large part from the perception of the purity of the Rromanes. However, this does not mean that one dialect is better that the other. For many Rroma the only criterion of "purity" of a given kind of Rromanes is the absence of local acquired vocabulary - that is, words which are immediately recognised as being non-Rromanes and understood by the local population. For example, Kaldera‰a in Russia consider themselves and are considered by many Russian Rroma as speaking a "better" kind of Rromanes as their dialect contains a relatively limited Russian vocabulary. For those Kaldera‰a who do not speak Romanian any more, a visit to Romania and Bessarabia often constitute a "shock".
Many of the word they use in their Rromanes and which are by now considered as being truly Rromane, are found in Romanian. Another amusing example is the variants used by various Northern Rroma to describe a window. Broadly speaking, there are two main variants: one German and one Slavic, For Sinti, the term fenstra [window], used by Russian and Baltic Rroma is purely gadžikano, that is, non-Rromanes. The Sinti themselves use the term voxnin of Slavic origins a term which is recognised by Russian and Baltic Rroma as being gadžikano.
For an objective analysis of the "purity" of a dialect, one must turn to the common trunk of that language. Words, morphological and syntactical elements that were acquired later do not allow one to make a judgement. The interesting fact is that in any given variant of Rromanes, one finds words or expressions stemming from the common trunk which have disappeared among many other variants. The verb nakhav [to cross], found in Vlax and Balkan variants is absent in Carpathian, Nordic and Ukrainian variants. Masxari [the Virgin Mary] for some Nordic groups is of old Indian origins but is absent in all other groups. Another factor enters this equation: the total vanishing of certain "important" terms from the common trunk.
For example, for most Vlax Rroma dialects, there is no common trunk word for life or for to live. They use the lexems trajo [life] and trajiv [to live] both of Romanian origins while the Nordic use the terms džijpen [life] and dživav [to live] of Indian origins. Interestingly enough, all Vlax dialects nevertheless use the term džuvindo [alive] stemming from the same roots as in the Nordic dživav. Among nowadays Balkan Rroma, džijpen has also almost vanished but in the 1860's dÏivav - to live, džijbe - life were both present. Even the reflexive form as in nowadays Polska Rromanes - dživdjuvav was recorded in the form of dživgjovava. Another such term is lodav in the sense of to reside, to live. Boretzky and Igla have recorded the use of this term among Bosnian Gurbeti but not among Balkan Rroma while it is mostly absent for the "new" Vlax vocabulary. The Nordic Rroma are still using it in the sense of to camp or among Sinti as lodopen [living quarters].
There are also common trunk terms whose meaning have changed according to the groups. Morav is such an example. Among Vlax Rroma, morav means to scrub, to rub while for the Nordic groups morav means to wash. Paspati notes the use of morav as to scrub and to rub, in connection with washing showing a dual use among these Rroma.
In Vlax and even in some Balkan dialects, naj nowadays means a finger while for Nordic Rroma, naj is a fingernail. Paspati notes the use of naj exclusively as fingernail and of angušt as finger as the Nordic (an)gušt. This shows once again the switch of signification found in some dialects while others have retained the original signification.
Different groups have also selected different common trunk lexems for a given meaning entirely loosing the other term. The Nordic groups use the words xačuvav [to burn oneself] and xačerav/xačkirav [to burn something] while the Vlax use phabuvav and phabarav; in the Balkan dialects one finds thabljovav and thabarav/tharav for the same meanings. Another such example is given by men [neck] for Balkan and Nordic dialects while Vlax and Ukrainians use kor for the same meaning.
This, in view of the fact that in the dialect spoken by Sliven Rroma, both terms exist, men as neck and kor as back of the neck, suggest that these alternative terms were present in the common trunk with a somewhat different meaning and that during the creation of the dialects, one or the other term was chosen over the other, often thus loosing the "specialisation" of the terminology.
Finally, one should note that certain lexems which are by now associated with given specific groups were still to be found among Balkan Rroma in the XIXth century and sometimes have still survived although in many cases, these also vanished by now. Terms such as urjavav [to dress]; mamuj [against, in front of and sometimes alongside (Russia)]; udžakerav [to wait] are nowadays found mostly among the Nordic metagroup although they can be traced back to the Balkan.
This evolution is not limited to Rromanes: many other languages, such as German, in its Austrian and German forms - and even in Germany itself - shows many such phenomena as does English between its British and American variants although they are recognised as being one language as Churchill once said namely that England and the USA were two countries separated by a common language. So what are the implications of these facts for the purity of a given Rromanes? De facto, for all the dialects where the common-trunk substrate is strong, it is not possible to pass value judgements on their degree of "purity". All such dialects contain and show elements which have either vanished or changed signification rendering their classification in various "degrees of purity" impossible and irrelevant.