Usually Roma speak dialects of Romani, an Indo-Aryan language, but also speak the languages of the countries they live in and incorporate loanwords and calques into Romani from the languages of those countries, mostly words for terms that the Romani language doesn’t have. For example the Gitanos of Spain and the Romnichal of the UK, have lost their knowledge of pure Romani, and respectively speak the patois languages Caló and Angloromani.               

Currently in Romania, Serbia, Montenegro, USA, and Sweden are independent groups that work to standardize the language, but not in India where Romani is not currently spoken.                                        

The Romani language is very related to those spoken in central and northern India, Pothwari in particular. In 1753, Istvan Valyi, a young Hungarian, studied in the Netherlands where he met fellow students from India and they taught him about 1000 words of their language. In 1776, in an article he exposed the fact that part of the vocabulary he learned in Leiden was understood by Rroma marking the beginning of Rroma studies. This linguistic relationship indicate the Roma's and Sinti's geographical origin. The pattern of their migration can be established with the help of the loanwords. Originally they came from the Indian subcontinent, today northern India and parts of Pakistan. Romani is part of the Central Indo-Aryan languages, beside Western Hindi, Bhili, Gujarati, Khandeshi, Rajasthani etc. The Romani language was comparate with various north Indian dialects and languages that showed the origins of the Roma people back in India.              In the course of history and time, the Indian retroflexive consonant D was deformed in a L in Armenia and to Rr, R or Rh in Europe. Here is a simple example, the Sanskrit word manda(ka) is pronounced men(d)a in Asia, manlav in Armenia and by European Gypsies as manrro, ma(n)ro or manrho (bread in Rromanes), clearly showing this evolution. Romani, Panjabi, and Pothwari share some words and similar grammatical systems. In 2003 was published in Nature a study that suggested that Romani is also related to Sinhalese, language spoken in Sri Lanka.                                                  

From the grammatical point of view, Romani is conservative in maintaining almost intact the Middle Indo-Aryan present-tense person concord markers, and in maintaining consonantal endings for nominal case – both features that have been eroded in most other modern languages of Central India. It shares an innovative pattern of past-tense person concord with the languages of the Northwest, such as Kashmiri and Shina. But there are other proofs that atest the Romani origins in the Central region, then migrating in the Northwest. Characteristic for Romani is the fusion of postpositions of the second Layer (or case marking clitics) to the nominal stem, and the emergence of external tense morphology that attaches to the person suffix. Romani and Domari shere these features, which has prompted much discussion about the relationships between these two languages.           Sometimes the Romani language is considered a group of dialects or a collection of related languages that comprise all the members of a single genetic subgroup.                                        

The language is nowhere official, but there are attempts currently aimed at the creation of a standard language out of all variants (such as those from Romania , the USA , Sweden ). In Slovakia for example, country with a large number of Roma population, variants of the language are now in the process of being codified.                         Since their departure from Anatolia, the dialects of Romani are differentiated by the vocabulary as well as through divergent phonemic evolutions and grammatical features. Today many Roma speak various new contact languages from the local language with the addition of Romani vocabulary.                       

One of the categorisations was a division between the Vlax (from Vlach ) from non-Vlax dialects. Vlax were the Roma who lived in the territory of Romania for many centuries. The primary distinction between these groups is the degree to which their vocabulary is borrowed from Romanian . Vlax-speaking groups include the great number of speakers (between half and two-thirds of all Romani speakers). In 1915, Bernard Gilliath-Smith in made for the first time this distinction in his book The Report on the Gypsy tribes of North East Bulgaria. Afterwards, based on geographical and vocabulary criteria, other groups of dialects were recognized, such as:Balkan Romani : in Albania , Bulgaria , Greece , Macedonia , Moldova ,Montenegro , Serbia , Romania , Turkey and Ukraine Romani of Wales

A bunch of scholars have worked out a categorisation of Romani dialects from a linguistic point of view on the basis of historical evolution and isoglosses . Linguist Norbert Boretzky, has carried out this work who pioneered the systematic plotting of structural features of Romani dialects onto geographical maps. In 2005 appears an Atlas of Romani Dialects and plots numerous isoglosses onto maps. Similar work has been carried out at the University of Manchester by Yaron Matras, linguist and former Romani-rights activist and his associates. The largest compilation of data on the dialects of Romani is the Romani Morpho-Syntax database, made by Matras together with Viktor Elšík. Parts of this database can be accessed online via the webpage of the Manchester Romani Project. Matras (2002, 2005) has argued for a theory of geographical classification of Romani dialects, which is based on the diffusion in space of innovations. According to this theory, Early Romani was brought to western and other parts of Europe through population migrations of Rom in the 14th-15th centuries. During the 16th and 17th centuries these groups settled in various regions of Europe, acquiring fluency in a variety of contact languages. After that  emerged some changes, which spread in wave-like patterns, creating the dialect differences attested today. Two major centres of innovations existed according to Matras: some changes emerged in western Europe (Germany and vicinity), spreading eastwards; other emerged in the Wallachian area, spreading to the west and south. A complex wave of language boundaries were created by many regional and local isoglosses. Matras points to the prohtesis of j- in aro > jaro 'egg' and ov > jov 'he' as typical examples of west-to-east diffusion, and of addition of prothetic a- in bijav > abijav as a typical east-to-west spread. He concluded that dialect differences formed in situ, and not as a result of different waves of migration.                                  Starting with 1982 Marcel Courthiade proposed in a series of articles a different kind of classification. Using the criteria of phonological and grammatical changes he concentrates on the dialectal diversity of Romani in three successive strata of expansion. He find common linguistic features of the dialects and presents the historical evolution from the first stratum (the dialects closest to the Anatolian Romani of the 13th century) to the second and third strata.Those which have only a Romani vocabulary grafted into a non-Romani language were named "pogadialects" (after the Pogadi dialect from Great Britain )