Home > Romani > Syntax


Contrary to New Indo-Aryan languages with the verb in the final position or SOV order, Romani generally has (S)VO order in so-called neutral declarative sentences:

maiñne laṛkī dekhī
S O V 'I saw a girl.' Hindi
I girl saw
(me) dikhlom rakľa
(S) V O 'I saw a girl.' Romani
I saw girl

As indicated by the brackets in the example above, pronominal subjects are optional in Romani, because they are contained in the verbal ending. The position of the subject shows variance between categorical contrastive SV word order which, as in the example above, corresponds to the neutral declarative sentence, and thetic continuative VS word order, as demonstrated in the following final sequence of a Lovara fairy tale.2

Kadej šudas o Rom e benges perdal. Taj tradine penge. Šingerdas pe o beng ande xoli, sa mundardas pe ande stanki. 'Thus the Rom outwitted the Devil. And they left. The Devil, however, exploded with rage and dove into the rocks.'

The change in the general word order from SOV to SVO is the result of Romani’s Europeanisation through contact with Greek and subsequently the languages of the Balkans. This alteration can also be seen in the dichotomy of factual and non-factual subordinating conjunctions mentioned above. Additional innovations include the relative clause, the use of prepositions and the definite article. Generally, these phenomena do not occur in the New Indo-Aryan languages on the Indian subcontinent.

Noun Phrase

In noun phrases, which can comprise a head NOUN (a noun or pronoun) and [optional] constituents in a generally fixed order, the preposition is always in the initial position:

[Preposition] + [Determiner] + [Quantifier] + [Adjective] + NOUN + [optional]

The following example, a prepositional phrase with a postposed noun phrase as an option extension, demonstrates all possible constituents in a noun phrase:

'between my father’s two elder brothers'
maškar o duj phureder phrala mre dadeskere
Prep Det.nom.pl. Num Adj.comp SUBST.nom.pl Pron.obl Subst.obl.gen.pl
between the two elder brothers my father’s

As shown in the example above, the noun is generally in the nominative form when following prepositions. Exceptions are the aforementioned bi 'without' and vaš 'because of', which require the genitive and dative respectively:

bi grasteskero 'without horse'
vaš brišindeske 'because of rain'

Contrary to the nouns, which are in the nominative form in the majority of cases and varieties, pronouns following prepositions are usually locative-marked:

maškar amende 'between us'
upre tute 'above you'

Another distinctive feature of Romani which was already discussed in the chapter dealing with morphology is the genitive noun phrase with its "double case" where the case of the determining article correlates with the genitive attribute, which in turn correlates with the head noun in case, number and gender:

le dadeskeri angrusti 'the father’s ring' / 'the ring of the father'
le dad-es-ker-i angrusti
det.obl subst-obl.sg-gen.sg-nom.sg.fem subst.nom.sg.fem

As shown in the introductory example, genitive nouns can also take the optional place after the head noun which does, however, not affect the "double case". This is also demonstrated in the following example from the Lovara-Romani which shows the genitive plural suffix –ger- contracted to -g-:

kher le dilengo 'house of the mad' / 'madhouse'
kher le dil-en-g-o
subst.nom.sg.mask det.obl subst-obl.pl-gen.pl.-nom.sg.mask

While the postnominal position of genitives is quasi systemic in some Romani varities, the attributes following the head noun of the noun phrase usually carry discourse-pragmatic functions:

... lengere rakle čore mule. '... their (pl) children (the) poor (ones) died.'

Verb Position

The position of the verb depends – as shown initially – on discourse-pragmatic factors with regard to the subject. A similar, if softened, form applies to its position with regard to the object. Generally, the verb is, as defined in the conservative word order specified by Boretzky (1996), positioned before the object – (S)VO. This applies to pronominal objects in particular: Even in Burgenland-Romani, which like all other Vend varieties of the south central dialects shows a contact-induced tendency towards verbs in final positions, direct pronominal objects are usually positioned after the verb:

diklom len 'I saw them (pl)'

In contrast, nominal objects in Burgenland-Romani are often found in front of the verb, most likely due to the influence of Hungarian contact varieties:

Idž leskero nevo auteri diklom. 'Yesterday I saw his new car.'

Indirect objects are usually positioned after direct ones. The verb position in interrogative sentences is the same as in declarative sentences:

Dikhav le grasten. 'I see the horses.'
Dikhes le grasten? 'Do you see the horses?'

Contact-induced word order is a characteristic which Romani shares with many other dominated languages. It is primarily rooted in the plurilingualism of adult Roma and the dominance of the primary contact language, i.e. the local majority language. The resulting contact-induced variance in Romani introduced in this section is further discussed in the following description of complex clauses.

Complex Clauses

The word order in subordinate clauses is generally the same as that of main clauses and displays the (S)VO order described above.

Relative Clauses

As with other European languages, Romani also uses relative clauses. The most commonly occurring elements introducing relative clauses ('relativiser') are kaj 'where' and so 'what'.

i zumi, so kerďa lenge, ... 'the soup which she made for them...'
o murš, kaj alo idž, ... 'the man who arrived yesterday ...'

If the relative clause’s head noun does assume non-subject roles, resumptive pronouns are obligatory. In this function we find kon 'who' and savo 'which', which correlate with the reference-noun of the main clause.

le gadžes, kaskero le grasten si, ... 'Gadžo, whom the horses belong to, ...'
panč džene, saven khera si, ... 'five people who own houses ...'

Verb Complements

Complements of epistemic verbs describing independent and real processes and conditions are marked by the factual or epistemic conjunction kaj; modal complements are introduced by the non-factual or modal conjunction te, which occurs as ti in some varieties.

phenen, kaj lakero phral dikhla la, ... 'they say, that her brother saw her, ...'
o X. mangela ti čumidel la, ... 'X wants to kiss her'

While te/ti are rarely substituted by loans, kaj is often replaced: In Vlax varieties by the functionally equivalent Romanian or ke; in varieties under Greek influence by the also equivalent Greek oti; in central varieties by the Hungarian hod/hodž/hod'/hot/hoj < hun. hogy.

According to Matras (2002: 180) the conjunction te/ti marks modal and aspectual verb complements as well as direct statements.

kamen grasten te bikinel 'they want to sell horses'
astaren te khelen 'they begin to dance '
adava te keres 'you should do that'

In the first example, te marks the complement of the modal verb 'to want', in the second one that of the inchoative verb astaren 'to begin'. The last example is a directive speech act which in English is expressed by the use of the modal verb 'should'.

Adverbial Clauses

Generally, adverbial clauses are introduced by semantically specified conjunctions which can be roughly divided into three categories: The first two consist of the conjunctions kaj and te which, as in their function as verb complements, differ with regard to their factuality. The third category comprises all those subordinate conjunctions which are based on interrogatives or which correspond to these. The non-factual te/ti introduces final, conditional and consecutive clauses.

phenďom lake, te anel amenge mol 'I tell her to get us some wine'
te sas man baxt, ... 'if I were lucky, ...'
buti kerel, bi te kerel love 'he works without making any money'

In contrast, kaj introduces causal clauses:

avav, kaj akarďan man 'I come because you called me '

An example for the third category is the interrogative pronoun kana 'when' in its function as a conjunction with temporal, simultaneous meaning:

arakhav, kana soves 'I stand watch while you sleep'

Additionally, variety-specific loans which either completely replace the original conjunctions or which occur simultaneously with these; combinations of elements of the three categories with loan conjunctions, prepositions and other particles may also occur. The following table shows an incomplete list of types of subordinate clauses and conjunctions based on purely semantic criteria:3

final te/ti, kaj te, hot te, kə te, ja te, či te in order to, so as to, so that
conditional te, bi/by, ako, -se, kana, kada, an te; ob te, či/čy, dali, li, mi if, in case, provided
consecutive te/ti, kaj te, hot kaj; bi te, oni te so that, so; without
causal kaj, kə/ke, vajl, anda kodo ke, sostar, soske, sar, adake, sar, sebepi kaj, afu, jati, zere, bo, mer, jer, lebo, pošto because, as
concessive xoč, hjaba kaj, trocdem kaj, sa jekh ke, jeva despite, in spite of, although, even though, whereas,…
local kaj; katar where (to/from)
temporalanterior sar/syr/har, angla sar, angla kodo ke, bi te na, prin te; dži kaj, dži te, bis te, džikim, bisko, medig before; since
temporalsimultaneous kana, kada/keda, sar/har/syr, kaj, so, afu while, whilst, as, when
temporalposterior kana, kada/keda, sar/har/syr, kaj, so, posle, čim, pala kodo ke, akana, jekh kaj, jekh ta after

1. ^ The treatment of syntax is largely based on Matras (2002: 165–190).

2. ^ Matras‘s (2002: 169 f.) discussion follows up on Holzinger (1993: 274) regarding the discourse-pragmatic and text linguistic function of this differentiation with regard to continuity, connectivity and consecutivity.

3. ^ Individual varieties may show features differing from this list.