General Introduction





Vowels and Diphthongs

Pronunciation and Stress



Dialects I

Dialects II


PDFs available in: flaggeEnglandflaggeaus


Romani-Project Graz / Dieter W. Halwachs

  The sound inventory of Romani differs insignificantly from that of other European languages, which for the most part also belong to the family of Indo-European languages. Merely the distinctive function of the aspirated voiceless plosives /ph, th, kh/ is hardly found in any other European language.  


Ill. 1 Frequency spectrum of the word Romani











Ranging at about 60, the number of phonemes (semantically distinctive sounds) of Romani is relatively high compared to other Indo-European languages. As Romani is a cluster of varieties without any homogenising standards, this high number of phonemes can be seen both as a result of the its heterogeneous character and the generally strong influence of contact languages on dominated ones. Illustrations 2 and 3 present an overview of the Romani sound inventory. The individual phonemes are represented by spelling system as used in RomLex, the multilingual and multidialectal lexical Online-Ressource, to represent all known sound phenomena of Romani1. The sounds in blue colour form the so-called core of Romani, whereas all the others are either loans from contact languages or result from developments within the variety itself that were partly also triggered by contact.








The consonant system of Romani differs in one significant aspect from those of other European languages: it disposes of the aspirated plosives (aspirated stops) characteristic of Indian languages. In the case of Romani, these are the voiceless aspirated plosives /ph, th, kh/, which for the majority of Romani variants at least in initial position have semantically distinctive function. The following examples show the distinctivity of the aspirated plosives as well as their Old-Indo-Arian (inc) correlates, the aspirated voiced plosives:

Some varieties also dispose of the aspirated voiceless affricate /čh/ in distinctive function:
Additionally to /čh/, Romani displays three more postalveolar affricates, /c/, /č/ and dž:
In Lovara-Romani, a Northern Vlax variety, the affricates /čh/ and /dž/ are reduced to the fricatives /š/ and /ž/:

In Kalderaš-Romani, another Northern Vlax variety, these fricatives are additionally palatalised:

Palatalisation, sometimes called jotation due to the quasi simultaneous articulation of the palatal approximant /j/, appears in all Romani varieties in contact with Slavic languages. The palatal affricate /ć/ is characteristic of Gurbet-Romani, a Southern Vlax variety:
The affricate /ć/ results from the palatalisation of /k/ before /i/ and /e/. The palatalised consonants of other Romani varieties can be explained in the same way:
Vlax varieties primarily dispose of two vibrants with distinctive function:
Gemination (doubling) as a result of lengthening is a more recent phenomenon of contact, appearing among others in the Finnish variety of Romani. The following examples are loans from German (deu):
The two dental fricatives /þ/ and /ð/ appear exclusively in loans from English in the Welsh variety described by John Sampson in 1926. They correspond to sounds represented in English by <th>.
  1 The writing conventions developed for RomLex <> are based on purely practical considerations and do not have any intentions of standardisation.   






All Romani varieties dispose of the five cardinal vocals /i, e, a, o, u/. The Old Indo-Arian vocal system of Sanskrit, in which vocal quantity (length) and quality are distinctive, has come to be purely qualitative.
Today, purely qualitative vocal systems are characteristic of varieties spoken on the Balkans. This holds for both Vlax and Balkan varieties. Central varieties spoken in regions with Hungarian influence show vocal lengthening resulting from regional contact. However, the phonemic status of the length is often arguable. The same holds for Sinti-Romani, Finnish-Romani and Welsh-Romani. As the examples illustrate, vocal lengthening affects both lexemes of Pre-European and European origin.
Depending on the individual contact situation, some varieties dispose of centralised vocals adopted from Romanian (ron) or Turkish (tur). Some varieties furthermore dispose of rounded vocals also resulting from contact with the Turkish language:
Diphthongs in words from the Pre-European lexicon of Romani result from intervocal consonant elision.
In the Northern Vlax varieties, the diphthong /aj/ is replaced by /ej/ in most words of Pre-European origin:
In some varieties, this process of elision is still going on. In Burgenland Romani, for instance, variants with consonant elision and the resulting diphthong are used alongside older full forms.
Other diphthongs appear in individual Romani varieties due to language contact or as a result of internal processes partly also triggered by language contact.


As realisation or, respectively, pronunciation of the sounds of individual dialects always also depends on the respective dominant contact language, it is difficult to make any generalizing statements on this matter. As to the so-called core stock, illustration number 4 roughly describes the pronunciation of individual sounds compared to German.
   This coining effect of contact languages similarly holds for stress. So-called conservative varieties with stress differentiation between the Pre-European and European parts of the lexicon can be distinguished from varieties with quasi homogeneous stress.
   Kalderaš-Romani can be attributed to the first group. Words of Pre-European origin generally stress the final syllable. This holds for both uninflected and simply inflected forms.

Inflected forms with secondary suffixes generally stress the next to last syllable.

Forms with primary suffixes and uninflected words adopted from European languages are characterized by penultimate stress, that is, they stress the next to last syllable. Inflection forms of European loanwords with secondary suffixes stress the next to last syllable just like those of Pre-European origin.

Like several other varieties, Burgenland-Romani has most likely developed homogeneous penultimate stress under the influence of Hungarian.

Except for forms with bi-syllabic secondary suffixes that stress the last syllable preceding the suffix, penultimate stress also holds for words of Pre-European origin with secondary suffixes and loanwords from European languages.

As suggested by the different spelling of corresponding lexemes of Kalderaš- and Burgenland-Romani, the individual varieties differ in sound structure. These differences primarily result from internal phonological processes. It would go too to discuss these processes in detail at this point. Some relevant processes concerning the classification of Romani and resulting differences between individual variety groups are treated as part of dialectology.

2 pal stands for Middle-Persian Pahlavi,sla means Slavic in the following examples.









An extended description of the Phonology of Romani is presented in chapter 4 "Descriptive Phonology" in

Yaron Matras (2002) Romani. A linguistic introduction. Cambridge: Cambridge University Press: 49-71.