The general view of speech rhythm comprises two main concepts: 1) of periodicity and 2) of isochrony. The former can be realized as periodical occurrence of events like metrical beats or similar pitch accents. The latter is nowadays not strictly defined. One thing is well known about it, namely that the so-called perceptual centers (P-centers) do not depend in a linear way on the perceived temporal units. The latter means that the perception of speech isochrony is due to factors like syllable structure, focus-assignment, intonational curve etc. All of them are well established separately for intensively investigated languages. The experiments of their combined impact on speech rhythm are still going on.
The speech rhythm of Romanes is often considered ‘different’ from that of the main contact language. On the scale of the classical division: stressed timed vs. syllable timed languages, the only investigated Bulgarian Romani dialect, namely Vlach II (after Igla 2004), is defined as ‘essentially syllable timed’ (Dimitrova 2004).
With this contribution I will try to find out the characteristic constituents of the speech rhythm of Sofia Erli, which will extend the representation to levels beyond the scope of a strict phonological theory. I claim that not only syllable codas, but also syllable onsets can contribute to the specific way of ‘how languages sound’. Another source of difference between Bulgarian and Sofia Erli is a matter of focus assignment. I will also try to extend the theory of Bartels (1999), the meaning of phrase tone L- and the boundary tone H%, to one more general view where the opposite tones, H- and L%, become also pragmatically defined. One marginal result of the study is the realization that some Bulgarian speech features began to infiltrate the well-shaped Romanes rhythm. They are much more typical for a stress-timed rhythm, e.g. a syllable reduction can be observed by higher speech tempo.
The paper discusses the formation of prepositions in Early Romani during its contact with Medieval Greek and their development in different Romani dialects. It is known that Early Romani developed a system of local prepositions in contact with Medieval Greek. In order to express specific local relations, Proto-Romani had Layer III postpositions (or adpositions), while Romani has prepositions, both deriving from adverbials. The prepositions te ‘in’, ke ‘in’, tar ‘from’ are etymologically related to the synthetic case suffixes of Locative -te, Ablative -tar and Dativ -ke.
The preliminary comments concern the researchers’ assumptions whether adpositions or even prepositions were available in Proto-Romani before its contact with Greek. In the first part we will examine the grammaticalisation of the local adverbials into prepositions in Early Romani. An important fact is that the prepositional system of Late Medieval Greek is in a transitional stage: simple prepositional phrases compete with new complex prepositional phrases, e.g. simple: apeksothen tis tentas ‘outside the tent’> complex: apekso apo to kastron ‘out of the castle’. We will examine which type of PP prevailed as a model for Early Romani on the grounds of the coherence of the structures and the complement of the prepositions.
The second part looks into the emergence of the prepositions te ‘in’, ke ‘in’, tar ‘from’ in a second stage within Early Romani. We will examine the distribution of the analytic vs. the synthetic Locative and Ablative in different environments and the tendency for analytic structures with the definite article, at least in certain Romani dialects. Our proposal is that the Greek simple PPs with the primary prepositions se ‘in’ and apo ‘from’ + DEF were the ones that facilitated the replication of the suffixes in pronominal position, not as an independent element but in combination with the definite article, cf. Rom. k-o, k-e/k-i on the model of Gr. ntr. st-o, masc. st-o(n), fem. st-i(n) ‘in + DEF’, Rom. tar-o, tar-i on the model of Gr. ntr. ap’ to, masc. ap’ to(n), fem. ap’ ti(n) ‘from+DEF’.
Finally, we will present the last stage of developments in the different Romani dialects: a) the generalisation of the pattern PREP + NOM in dialects that lost contact with Greek, b) the formation of complex PPs in the Balkan dialects on the model of Greek.
This paper deals with the realization of the spatial and temporal locations in Slovak Romani by means of the system of seven ‘postpositional’ cases (= a closed set of unstressed agglutinative markers derived from Middle Indo-Aryan postpositions and postposed adverbs) and a more open set of analytic prepositions (called Layer III adpositions by Masica, 1991:232 ff).The latter set will be presented in terms of 12 binary contrasts expressing basic topological (‘on’ versus ‘off’, ‘in(to) versus. ‘outside’) and projective notions (‘before’ versus ‘after’, ‘under’ versus ‘over’, etc.).
It will be shown that (Slovak) Romani differs remarkably from all the other New Indo-Aryan languages in realizing the Layer III adpositions as prepositions. Typologically speaking, Romani developed ‘circumpositions’ combining postpositions with prepositions. Contrast the Hindi structure (N=Po) Po with Romani Pr (N=Po): ((bari mez)=ke) age ((big table)=GEN+OBL) before ‘in front of the big table’ with Romani andre (bar+e skamind+es)=te into (big+OBL table+OBL) ‘into the big table’. It will be argued that this typological change in the structure of the postpositional phrase was concomittant with the overall typological shift from the Indic SOV (subject-object-verb) to Romani SVO word order.
My data will be taken from the literary corpus in the East Slovak variety of European Romani (Fabiánová & Hübschmannová 1991, Ferková 199?, Giňa 1991, Hübschmannová 1990), and its grammatical presentations by Hübschmannová et al. (1991), and Šebková and Zlnayová (2001).
Two problems are addressed: the origin of the 3rd Plural forms on and ola, and the emergence of the subject clitics -lo, -li, and -le. Of the two forms, on and ola, on appears to be older one for a number of reasons: it is rather irregular; it is unlikely that it has been shaped on amen ‘we’ and tumen ‘you’, since in this case we would expect a form *o-men to have come about; it cannot be made sure that -n in amen and tumen is an old element in Indic languages. On the other hand, ola occurs in a few Balkanic dialects only, most likely formed in analogy to the Balkan languages, where personal pronouns and demonstratives are not really separated even to-date. In both the Balkan languages and in Romani nominative and oblique forms of the pronoun cannot be derived from the same stem, which is in consonance with a general tendency for pronouns to display suppletive inflection. Subject clitics are very rare among the languages of the world, and nearly absent in the languages of Europe and India. In all likelihood, Romani -lo, -li, -le did not come up at the same time as the oblique pronouns les, la, len, but much later, and not immediately derived from the OIA ta-pronoun, but formed in analogy to the oblique forms. Their original function might have been to mark gender and number with predicative elements that are not inflected for these categories (copula and deictic particles). There are a few subject clitic forms in modern Greek, but it is not clear if they have been the model for Romani.
In Romani there are two pronominal elements which refer to “other people or things,” like the anaphoric specific demonstrative ka and the indefinite pronoun aver in Burgenland Romani spoken in Austria. The anaphoric specific demonstrative is originally used to single out some referents contrastively and emphatically from a group of potential referents, and in some dialects it is getting to refer “the other”. (cf. Matras 2002)
Why can these two pronominal elements like ka and aver co-exist though they have apparently the same function? Firstly I will attempt an explanation for this, using data mainly from Burgenland-Romani. I will argue they are forming a complementary relationship with each other concerning whether their referent is definite or indefinite.
Secondly, as for the anaphoric specific demonstrative ka in Burgenland Romani, unlike other demonstratives, it co-occurs with the definite article or definite determiners and is marking definiteness redundantly. Why does the co-occurrence with the definite article differ between the anaphoric specific demonstrative and other demonstratives? I will suggest this is because the meaning of ka “other” is ambiguous and undetermined in regard to its definiteness and therefore there is some need for ka to be marked as definite additionally by other elements like definite article.
Thirdly, in another dialect like Ajia Varvara dialect spoken in Greece, the anaphoric specific demonstrative okova doesn’t co-occur in principle with the definite article, unlike in Burgenland Romani (cf. Igla 1996). Why does the co-occurrence between the demonstrative and the definite article differ from dialect to dialect? I will attempt to explain this by focusing on the stability of the boundary between the anaphoric specific demonstrative and the indefinite pronoun aver concerning definiteness.
The Romani dialects in Sliven and surroundings form a renarrative mood (‘quotative’) by adding the particle li to the indicative forms. Semantically the renarrative mood is used in referring to reported, unwitnessed events. The existence of this mood in so-called ‘Drindari’ had been mentioned almost fifty years ago (Gilliat-Smith, Kostov). There seems to exist no doubt that this category has come into existence under the influence of Bulgarian, mainly since – to my knowledge – it has not been reported in Romani dialects not influenced by that language.
More actual data from the Romani dialect of Sliven slightly diverge from the first sources as to the morphological shape of the mood. The ‘innovations’ partly correspond to the structure of the renarrative in Bulgarian: Like in that language the renarrative (aorist) differs from the indicative only in the third person, namely by the omission of one element. In the Romani dialect of Sliven the mood can be reinforced by doubling the particle li. Hereby the feature ‘doubt’ about the truth of the event the speaker refers to is added to the paradigmatical semantics ‘reported/unwitnessed’. Reinforcement of the renarrative in the same sense is encountered in the Bulgarian mood.
The use of the renarrative mood in non-reported speech, i.e the neutralization between the indicative and the renarrative compulsatorily adds the feature ‘surprise’ (admirative) to the indicative. The changes mentioned took place within a rather short stretch of time; bearing in mind, however, that Romani is not a codificated language and develops without the influence of normative grammar this doesn’t very much come as a surprise. Apart from the morphological pecularities/innovations of the renarrative I will attempt at a more precise description of its semantics.
Synthetic valency changing morphology is an outstanding New-Indo-Aryan feature of Romani (Matras 2002: 119). Historically there are different morphological resources to derive secondary verbal stems from verb roots and new verbs from adjectives and nouns. This process of ‘grammatical derivation’ (Matras 2002: 119) is the most important means to enrich the verbal lexicon of Romani. Synchronically the productivity and functional range of these morphological markers differ to a great extent across the dialects of Romani. The distribution of the productivity of these markers strongly suggests that language contact has a strong impact on the productivity of these morphological processes as well as the functional range of individual markers.
East Slovak Romani is a particularly interesting case with respect to synthetic valency changing morphology. The overall picture in East Slovak Romani is that grammatical derivation is very productive in the creation of new verbs and new verbal meanings. While most markers express their typical functions (such as transitivizing intransitive verbs, e.g. xatj-ol ‘to burn ITR’ > xatj-ar-el ‘to burn something, to make something burn’ or factitives from adjectives, e.g. mato ‘drunk’ > matj-ar-el ‘to make somebody drunk), we also find the expression of a totally new function: the marker -ker- is very productive in the expression of iterative aktionsart (ker-el ‘to do’ > ker-ker-el ‘to do often, on a regular basis’). The marker seems to have lost its original transitivising character, since it can also derive iteratives from intransitive verbs, e.g. tašlj-uv-kerel ‘to gag for a long time/often’.
The paper explores the development of a transitive marker to a marker expressing an aktionsart distinction. I will explain why this particular marker was chosen to express this new function by discussing how its original semantic content fits into its new “job” (if at all). Furthermore I will illustrate the range of aktionsart meanings expressed by this marker and discuss them in the light of prototypical expressions of iterative aktionsart in other languages, especially in the contact language Slovak. Finally I will highlight the role language contact plays in the positive reinforcement of valency changing morphology in general and the emergence of new functions in particular.
In most North Central Romani dialects, iterative verbs are synthetic, being derived by the suffix -ker-, e.g. mar- ‘beat’ > mar-ker- ‘beat more times’. It is usually assumed that such synthetic iteratives have developed from synthetic causatives through a semantic shift, i.e. that the Early Romani transitivizing suffix -ker-, which had resulted from a grammaticalization of the verb ker- ‘do, make’, was re-interpreted as a marker of iterativity (cf. Hübschmannová & Bubeník 1997). However, in the North Central Romani dialects of Liptov, a region in the northeast of Central Slovakia, iterativity may be expressed by a free iterative auxiliary, the verb ker- ‘do, make’, in bi-verbal collocations, e.g. murdarnas kernas [they killed they did] ‘they were killing many people’, te prepadavinel te kerel [to rob to do] ‘to rob more times/people’ (Liptovský Mikuláš), te kerel te viľinel [to do to shoot] ‘to shoot many people’ (Liptovská Teplička/Hranovnica).
There are (at least) two alternative scenarios for the development of the iterative collocations in Liptov Romani. First, the iterative collocations may represent an early stage of a secondary, independent grammaticalization of the verb ker- ‘do, make’ that has been restricted to Liptov Romani. Or, second, the iterative collocations may represent an earlier stage of the North Central synthetic iteratives, in which case one has to assume a grammaticalization of an iterative auxiliary into an iterative suffix. The second scenario would throw serious doubts on the causative-to-iterative shift hypothesis. In my paper I will evaluate both scenarios, and discuss their implications with regard to the diachrony of the category of iterativity. The Liptov Romani data are based on my fieldwork in the region.
Loan verb markers are found in all Romani dialects. They are originate from Greek tense markers, borrowing during the time that Romani first came into contact with this southern-Balkan language. We find such markers preserved in the Romani dialects throughout Europe. The dialect of concern here, Sofades Romani, is situated in mainland Greece in the region of Thessalia. It is part of the Southern-Balkan branch, and is an Arli-type variety. As this dialect has not left Greek-speaking territory, the system of LVM is of particular interest: it is still in constant contact with Greek verbs and so with these same markers used in the original donor language.
We find that the choice of LVM in Sofades Romani is not only dependent on valency but also on ‘telicity’, or the completion of an action. Thus, Sofades Romani also employs an inherited valency marker -ker-. However, this marker is only found with some transitive verbs and not with others. I hope to show that the transitive marker -ker- is being employed not only as a transitive marker, but as a ‘telicity’ marker.