General Introduction





Nouns and Pronouns






Derivation and Valency


Mood as an Analytical Category


Infinite Forms


Special Formations






Dialects I

Dialects II


PDFs available in: flaggeEngland

Forms / Morphology

Romani-Project Graz / Dieter W. Halwachs

As for many other dominated languages, whose bilingual speech communities experience great pressure to assimilate, morphology is the most stable structural area. It does, however, also show certain aspects of contact-induced language change.




Romani has two genders, masculine and feminine, two numbers, singular and plural, and eight cases, which are also found in many other European languages. The case system is therefore typically European. However, the way in which the cases are formed is typically Indian.




The double-stage nominal inflexion consists of three primary cases – nominative, oblique and vocative – as well as five secondary cases derived from the oblique: dative, ablative, locative, instrumental/sociative, and genitive (Ill.1). It is most often the same as the nominal form in varieties influenced by contact languages which lack a synthetic vocative.

The oblique functions as accusative with entities that have high referential status. Otherwise, the accusative has the same form as the nominative. Semantically, entities that have high referential status are generally characterised as animate. This has led to the dichotomy of accusative=nominative : accusative=oblique with the semantic feature of [± animate].






This correlation is not fundamentally wrong, but it does not go far enough, because the independent oblique also has further functions. In possessive construction, for instance, the possessor, whether animate or inanimate, is always marked by the oblique, while the possession is expressed by the nominative.




The oblique forms the basis for the five secondary cases: dative, ablative, locative, instrumental/sociative and genitive. Additionally, many varieties have developed analytic case formation, often replacing the locative. The ablative (case of descent and origin) is also affected by this development. In the course of this development triggered by contact with languages of the Balkans, “old” synthetic forms are replaced by “more recent” analytic formations.




Nouns of pre-European origin differ from European loans in their declension (Ill.2). The declension of articles and adjectives is characterised by the dichotomy of nominative : oblique. The noun phrase is always governed by the head noun.




In the case of a noun phrase with a genitive functioning as attribute, government deviates from this rule: here, the article correlates with the attribute, which in turn correlates with the head noun.






The forms of the definite article show variety-specific variation. The article forms used in the above examples are marked in the following illustration. Generally, however, a tendency of reduction and coincidence of forms is observed primarily for the oblique. Only the differentiation between nominative singular masculine and nominative singular feminine shows some stability.





This tendency of formal reduction also affects adjective endings. Frequently, there are three distinct forms for six functions, with a fourth form for the oblique feminine singular in cases where gender is obvious or stressed.





There are only a few indeclinable adjectives, e.g. šukar ‘beautiful’, godžar ‘intelligent’.

Comparison of adjectives is variety-specific. Besides the inherited suffix -eder, borrowed particles and affixes are used to form both comparative and superlative.




Adjectives of European origin are characterised by an even smaller stock of forms than those of pre-European origin, or else are indeclinable, as in the case of Burgenland Romani :

lungo < ron. lung ‘long’ / dlgo < srb. dial. dlgo ‘long’  / brauni < deu. dial. brauni ‘brown’.




The table above (Ill.3) presents an overview of the personal and possessive pronouns of Romani with variations specific to individual varieties.

Most Romani varieties have clitic personal pronouns for the third person in anaphoric function. These are the regular nominal forms of the oblique forms of the personal pronouns listed above.




As a rule, Romani has four demonstrative pronouns, from which articles and personal pronouns of the third person are also derived. Along with relative distance [± near], the demonstratives also encode specificity [± specific]. This makes it possible to choose an intended referent from a group of possible referents: the feature of [± specific] thus serves to disambiguate or explicitly contrast.



The interrogative pronouns so ‘what’, and ko(n) ‘who’ are pronominal nouns and thus decline in the same way as nouns.

The inherited negative pronouns khoni(k) ‘nobody’, and khanči ‘nothing’are among others conserved in Vlax varieties. Many other dialects have replaced them by more recent loans, such as the Slavic ništa ‘nothing’. The same is true for indefinite pronouns, which also for the most part originate from European contact languages and display a great range of variation.




As with the nouns, a morphological distinction between elements of pre-European and European origin can also be observed with Romani verbs. Unlike the pre-European verb stems, the verbs that have been more recently adopted from European languages are characterised by morphemes of adaptation and integration (Ill.4).




Derivation and Valency


The synthetic coding of valency in Romani is identifiable as an Indo-Aryan inheritance. While the intransitive forms are formally uniform and display only functional variation, transitive inflexions vary both formally and functionally:




Intransitivity is expressed by means of the suffix {ov}, which is often accompanied by palatalization of the stem‘s terminal consonant:






Verb conjugation is based on the present stem, which is identical with the verb stem: ker- ‘make/do-’, phuč- ‘ask’, pisin- ‘write-’ trajisar- ‘live-’, dandar- ‘bite-’. The so-called perfective stem is formed by extending the present stem with a perfective marker – ker-d- ‘make/do-PFV-’, phuč-l- ‘ask-PFV-’, pisin-č- ‘write-PFV-’, trajisar-d- ‘live-PFV-’, dandar-d- ‘bite-PFV-’.  The intransitive verbs usually use the suffix {/il/in/}, with the addition of the same gender-specific forms used with the adjectives in the third person singular:




The use of different present and perfective stems corresponds to the aspectual differentiation [± perfective]. States and actions that are completed from the perspective of the speaker are [+ perfective]; states and actions that are not completed, or whose state of completion or non-completion the speaker does not intend to specify, are marked [– perfective]. Similarly, the categories of number (singular, plural) and person (first, second, third) are also expressed by two different morpheme sets:



The morpheme sets exhibit variety-specific variation. The non-perfective endings additionally vary within varieties with respect to their vocalism: when the verb stem ends in a vowel, the vowel of the ending is assimilated to it.




The morpheme {/as/ahi/a/e/ys/s/} expresses remoteness in time and thus it functions as a tense marker in the form of the characteristic [± remote]:






The [– perfective] [– remote] forms have so-called long forms; these are the short forms extended by the morpheme {a}. The functions of the short and long forms are variety-specific: in Kalderaš Romani the short form is used as the present indicative, the long form for the subjunctive. In Arlije and Bugurdži Romani, the long forms are generally used for the present indicative, and the short forms are used for the subjunctive or as alternative present indicative forms. In Burgenland Romani the short forms are used for the present and the long forms for the future. In contrast, the Balkan varieties form the future analytically, by combining the particle {/ka/kam/kama/}, derived from the verb kamel ‘love, want, wish’ and the present: ka ker-av ‘I will make’. This is a contact phenomenon: analytic future formation is a regional characteristic of the Balkan languages.

The table above (Ill.5) presents an overview of the conjugation and the verbal suffixes of Romani.


Mood as an Analytical Category


The modal categories of ‘being able’, ‘needing (to)’ and ‘wanting (to)’ are generally formed analytically and are partly variety-specific. ‘Wanting (to)’ is the most conservative and consistent modal expression in Romani and is usually expressed using the verb kamel ‘he/she wants (to)’, In the Balkans, kamel is often replaced by the verb mangel ‘he/she desires/asks for’.




The modal particle šaj ‘be allowed to’ expresses permission. Its negative counterpart naštig ‘cannot’ serves as the negation both of being allowed (to) and of being able.




The positive sense of being able can be expressed by verbs such as džanel ‘can/be able’ < ‘know’ or, as in Sinti Romani, hajevel ‘can/be able’ < ‘understand’.




‘Needing (to)’ is expressed in several varieties by a particle that has developed from si te ‘it is, that’ by lexicalisation.



In many other varieties, ‘needing (to)’ is expressed by more recent loans – including fully inflected verbs, impersonal verbs and modal particles – and sometimes also by functional extension of inherited verbs:








The traditional description of the verb system of an Indo-European language is centred around the category of tense. The subcategories used in this type of scheme are listed in the row “function” in table above. Usually, the main differentiation is between present and past, with imperfect, perfect and pluperfect being grouped under the general heading of “past”. However, in Romani verbs are organised primarily by aspectual differentiation, something which has generated a good deal of discussion and controversy. This was resolved by Matras (2002: 151ff.), who provided a cogent explanation in terms of the TMA system (TMA = Tense, Modality, Aspect). The next table and the notes appended to it (Ill.6) summarize the functional arrangement of the TMS categories in Romani.




Infinite Forms


The “inherited” infinitive of Romani has probably been lost under the strong influence of Byzantine Greek in which use of infinitives had almost died out by the time they came into contact and of the reduction of infinitives in the southern Slavic languages. Present forms are used as “new” infinitive  in analytic formations of modal verbs by adding the non-factitive particle, without inflexion for person or number.




The examples show the commonest form, which is the third person singular of the short forms. Participles in Romani are a [+ perfective] participle and a gerund, which functions as its [- perfective] counterpart.

Verbs of pre-European origin form the perfective participle with the perfective stem and the adjective endings -o / -i // -e.



In contrast, the perfective participles of loan verbs from European languages are formed using the suffix {/ime(n)/ome(n)/ame(n)/}:




The counterpart of the perfective participle is an unflegded [– perfective] gerund. Bugurdži Romani:






The intransitive derivation is also used to construct a synthetic passive, in the form perfective stem + {ov}. Varieties with unproductive intransitive derivation mostly have only a few lexicalized forms – e.g. maťojav ‘I am drunk’ in Burgenland Romani – and form the passive analytically by using the perfective participle with the verb ‘become’: av- or ov-




Another possible way of making the passive is to use reflexive forms. For example, Kalderaš does this, with recent loans:




Special Formations


Special formations of verbs are common enough in Indo-European languages, and Romani is no exception in this respect. There are a number of irregular constructions and suppletive forms such as, for example, the verb ‘to go’. The verb stem, which ends on a vowel, dža- ‘go-’ assimilates the vowel of the ending. On the other hand, the perfective stem is a suppletive formation, gel- ‘go-PFV-’, and takes the gender-specific adjective endings in the third person singular, as is usual for the intransitive verbs.






Other special formations cannot be discussed in detail here, because they exhibit a lot of variety-specific variations.

Similar rules apply to the special form and functions of the verb ‘to be’. Some of the variety-specific present tense forms and their Sanskrit equivalents are listed in the table above (Ill.7).

With regard to synthetic forms, the verb ‘to be’, only has a present and past form which formally correspond to perfect and pluperfect:




As a suppletive form for the future tense and/or conjunctive, the verbs ovel ‘to become’ and avel ‘to come’ are used depending on the variety. Balkan varieties form the future analytically, as shown above.




In this section we will describe the core set of particles that is conserved in most Romani varieties. However, a complete treatment of all adverbs, prepositions, conjunctions and other non-inflected words in Romani cannot be undertaken in this article because of the high incidence of variety-specific variation. Some particles are explained in the section on syntax.




Adverbs can be subdivided into the derived modal adverbs on the one hand, and “inherited” or loaned adverbs of time and place on the other.

Modal adverbs are derived from adjectives by adding the morpheme {/es/eh/e/} as a suffix: bar-es ‘big’, šukar-es ‘beautiful’.

The large majority of the adverbs of place belong to the Indo-Aryan core vocabulary. The deictics of place ‘here’ and ‘there’, like the demonstrative pronouns, have a set of forms that express the permutations of the characteristics [± near] and [± specific]:



In the Romani varieties spoken in the Balkans, often only the locative versions of the specific forms have survived, but they no longer have their original specifying function, and also exhibit wide variation:




Many varieties have ablative forms as well as locative forms:




Similar locative-ablative pairs are also found with other adverbs of place. In contrast to the deictics of place, these have ancient locative and ablative suffixes inherited from Old Indo-Aryan:




These adverbs of place often also serve as prepositions. If the preposition ends in a vowel and is followed by a definite article, they are fused:



If the particle ends in a consonant, it can act as a preposition without changing:




Romani has preserved only a few adverbs of time from the Indo-Aryan. Some adverbs of time have arisen endogenously in Romani, but the majority are loans from European languages:






The negative particles na (< inc. na) and ma (< inc. ), which are inherited from Indo-Aryan, have different functions based on the characteristic [± indicative]:




 In several varieties, including Kalderaš Romani, this functional separation is fundamentally modified by the loan či (< rom. nici):




An additional particle of negation inherited from Indo-Aryan is the prefix bi- (< inc. vi-), which is found in almost all varieties:






The general coordinating conjunctions are thaj ‘and’ (< inc. tathāpi) and vaj (< inc. va) ‘or’:




The subordinating conjunctions kaj (< inc. kasmin) and te < inc. tad) are also inherited from Indo-Aryan. They differ in the characteristic [± factual]:




An extended description of the morphology of Romani is presented in
Yaron Matras (2002) Romani. A linguistic introduction. Cambridge: Cambridge University Press.