Professional Documents
Culture Documents
General Editor
E.F. KONRAD KOERNER
(University of Ottawa)
Volume 126
Romani in Contact
ROMANI
IN CONTACT
THE HISTORY, STRUCTURE
AND SOCIOLOGY OF A LANGUAGE
Edited by
YARON MATRAS
University of Hamburg
Yaron Matras
Hamburg, November 1994
CONTENTS
Introduction IX
Yaron Matras
the hybrid character of Romani, some arguing that contact was responsible
for the fact that Romani shared distinct features with different varieties of
subcontinental Indo-Aryan. While Turner (1926) advocated a historical
layer-solution, arguing for the emergence of Romani in Central India and a
subsequent migration to the Northwest, where it remained unaffected by
later developments within the Central languages, Sampson (1926: 29)
suggested a merger of various dialects as a possible solution.
Since Gilliat-Smith (1915-1916), dialect classification in Romani has
relied heavily on the distinction between Vlach and non-Vlach varieties, the
Romanian influence upon the former being one of the significant criteria.
Besides reviving this latter distinction (Kochanowski 1963), modern
Romani linguistics has directed much of its efforts toward investigating the
structural and sociological background for the emergence of those Gypsy
varieties which are based on the surrounding so-called host-languages, but
retain, at least in part, a Romani vocabulary (see Bakker & Van der Voort
1991 for an overview). Growing interest in these varieties during recent
years (cf. Bakker & Cortiade 1991) has partly to do with the fact that their
structures are more easily accessible to non-Romanologues than are those of
'inflected Romani' (so the term for Romani proper), but it is also connected
to the theoretical questions which they pose. Much discussion has been
dedicated to seeming similarities between such idioms and Creoles (e.g.
Hancock 1970; Boretzky 1985; Acton 1989), before the term Para-Romani
(Cortiade 1991) was suggested. Romani 'mixed' varieties have been
presented as further evidence for the existence of a 'mixed language' type
which owes its genesis to a process of 'broken transmission' across
generations - evidence put forth in order to challenge the conventional
notion of gradual genetic development as the primary course of linguistic
evolution (Thomason & Kaufman 1988).
But inflected Romani, a genuine genetic relation of Modern Indo-
Aryan, also possesses structural features which confront descriptivists with
a challenge. It has been shown that apart from borrowing lexical items,
Romani also treats borrowed morphological material systematically,
assigning much of it selectively to the borrowed lexicon, a phenomenon
referred to as thematic vs. athematic grammar (Hancock 1993). While this
is generally characteristic of the language as a whole (cf. Kostov 1973;
Boretzky 1989), specific dialects also show their own typical patterns of
adoption (Kostov 1973; Igla 1989; Boretzky & Igla 1991). Besides selective
integration of borrowed morphology, Romani dialects have incorporated
INTRODUCTION XI
whatever social, cultural and political needs are felt necessary to be covered
by it, might feel encouraged by the fact that this collection also aims at
stimulating further discussion and involvement in Romani studies and its
applied domains. By emphasizing the importance of curiosity and
discovery, as opposed to prejudice and ignorance, this book can constitute
but a very modest contribution to normalizing the difficult position still
assigned to the Roma by the majority in our society.
YARON MATRAS
University of Hamburg
REFERENCES
VIT BUBENIK
Memorial University of Newfoundland
0. Introduction
Recently there has been an upsurge of interest among historical linguists,
typologists and sociolinguists in the interference phenomena of European Romani.
olization, and Genetic Linguistics (1988) brought them to the attention of
North-American audiences. One of their achievements is their explicitly stated
borrowing scale (pp.74-6) assigning numerical values from 1 to 5 on the ba
sis of lexical and structural borrowing: category (1), lexical borrowing only,
(2) and (3) slight structural borrowing, and (5) heavy structural borrowing.
Various Romani dialects appear at the heavy end of their borrowing scale
(category 5); they underwent changes through borrowing that range from
moderate to heavy. As is well known, these include lexical items of all types
(to enlarge a very restricted vocabulary of ca. 1 000 inherited words of Indic
origin); borrowed phonemes (including entire contrastive sets such as pala
talized consonants in Russian Romani); development of a typical Balkan peri
phrastic future; Aktionsart prefixes borrowed from Slavic languages; a plural
suffix borrowed from Romanian in Vlach dialects; special paradigms -
modeled on Greek - for inflecting borrowed nouns, adjectives, and verbs,
etc.. This interference has been and is still going on, especially, in the sphere
of grammatical categories and syntactic rules.
dialect). In Marwari the particle la: is indeclinable (as ga: in Malvi), while
Jaipuri declines lo: for gender and number in the same fashion as Marwari
does it in the case of the particle go:.
Given the parallelism with the ga:-future the source of the la:-future
might be the past participle of the OIA (Sanskrit) verb la:- "grasp, take" (this
verb was replaced in MIA by le-, presumably, by non-proportional analogy
with de- "give").
All these forms are surveyed in (2):
But now the crucial question: how old are these constructions in our late
MIA or early NIA texts?
To the best of my knowledge they do not appear in the Apabhramsa
texts. As far as the early NIA texts are concerned, it should be noted that the
ga:-future is not found in the medieval poets from the central area; on the
other hand, it did not replace completely the old synthetic sigmatic future in
the regional dialects (e.g. Bundeli calha" I will go" continues the OIA
TYPOLOGICAL CHANGES 5
The earliest atestations of the above constructions are from the 16th
century in Old Western Rajasthani (cf. Tessitori 1915).2
Having said all this, I am not inimical to the idea of considering the
suffix -a of the future tense in ER as a relic of the particle such as known
from Modern IA languages in (2) and (3); whether it was originally a clitic
form of the copula (as in Bangaru or Rajasthani) or the verb "to go" (as in
Hindi and Rajasthani) or the verb "to take" (as in Rajasthani) we will not be
able to decide with any degree of certainty.3 A propos the former two, the
copula and "to go", it should be in kept mind that their forms merged
6 VIT BUBENIK
phonologically in ER. A propos the verb to "take" Bloch (1933: 288) made
an intriguing comment that 'the convergence with the use of the Russian
Romani la- "to take", should be noted'. A few observations of mine follow.
The North Russian Romani (Wentzel 1980: 95) forms its imperfective future
by combining the verb lava "to take" in the present with the subjunctive form
of the main verb. Boretzky (1989: 369) proposed that the foreign model
could have been Ukranian which - unlike Russian - forms its future by
cliticizing the present forms of the verbjati"to take" to the infinitive.
Alternatively, given our evidence for the "take" -future in medieval Ra-
jasthani (16th c.) and Old Marathi (12th c.) we might propose a genetic
solution in the sense of a common historical source. The problems, however,
might be insurmountable. How old is the "take"-future of Russian Romani?
How far back can we project the "take" -future of early NIA languages? Both
proposals are schematized in (4):
If one identified this *asi with Apabhramsa a:si in the past tense
meaning, this semantic obstacle would presumably be removed but the con
structional dilemma would remain. Sampson (1926: 192) compared the
formation of the Welsh Romani (WR) imperfect and pluperfect with that of
Panjabi. His examples are shown in (6):
8 VIT BUBENIK
Panjabi actually conjugates its auxiliary (san, sy), si, sa, sow, sn, cf.
Shackle 1972: 35) when it forms its imperfect and pluperfect; and in this
respect Grierson (1903-1922: 629) mentions that:
the common form of the past tense of the verb substantive is usually si:
(=3/Sg "was") for both masculine and feminine singular, and for the
masculine plural. This is generally explained as the feminine of sa:, but
much more probably it is a corruption of some old form akin to Prakrit
a: si:.
ma:ra:=ha: (Lahnda)
strike+ 1/Sg/Pres=was
"I would have struck/if I had struck"
ma:ren=ha: (Lahnda)
strike+3/Pl/Pres=was
"they would have struck/if they had struck"
10 VIT BUBENIK
If the ER imperfect marker -as goes back to OIA a:sa "has been"
(Perfect) or a:si:t "was" (Imperfect), then Lahnda would furnish the closest
synchronic parallel. The singular paradigms of these three languages are
surveyed in (9):
3. Perfect
Upon the examination of the suffixes of the perfect, we may ascertain
that:
(i) they are different from the suffixes of the present
(ii) in the 1st and 2nd Pers they are identical with the suffixes of the copula
in the present tense (the 3rd Pers is of a nominal origin; witness the plural
suffix -e shared by the 3rdP1perfect and the plural form of the PP; the 3rd
Sg Perf is kerd'a vs. the singular form of the PP kerdo), and
(iii) they are preceded by the yod (=palatal glide) in Baltic, Balkan and Vlach
dialects (in the latter group -y- is used in Kalderas but not in Lovari, cf.
Hancock 1993: 48). In Slovak Romani, the yod palatalizes the preceding
apical consonant t', d',l' (cf. Hbschmannov et al. 1991) The forms of
Vlach and SlovakRomaniare shown in (10):
pesidamhi (<pesida=mhi)
"I have sent"
The h of the root could be lost through contraction during the cliticiza-
tion process. For s-dialects the segmentation of the copula would have to be
proposed.
However, the major problem still has not been solved. What is the
source of the yod intervening between the PP stem and the suffixes of the
copula? Diachronically speaking, I would prefer to keep this yod separate
from the yod which appears in the stem of the passives and inchoatives (cf.
Section 4). Assuming that the ancestral speakers of Romani left India without
it (i.e. kerdom would be earlier than kerdjom) we would have to provide a
potential source for it in the yod-initial copula of one of those many languages
they encountered en route (as it happens the East Iranian, Greek and Slavic
languages possess a yod-initial copula).4
But whatever its source, the addition of the yod in the perfect enhanced
the contrast between it and the PP. Without it certain forms would come
"dangerously" close to each other; e.g. in Slovak Romani kerdo "made" (PP)
vs. kerd'a "he/she made" (it should be observed that the latter form is often
enlarged by the imperfect marker -(a)s : kerd'as); or, mard'o- in mard'om "I
have struck" differs only by its palatality from the PP mardo "struck". On the
other hand, the ambiguity remains in the plural where kerde "made" (P1 form
of the PP) is homophonous with the copula-less PP kerde "they made" (but
notice Vlach kerd(in)e).
In terms of morphosyntax, ER treats both the agent and the subject of the
perfective event in identical manner; put differently, ER is unique in the
context of Central and North-Western IA languages in not possessing the
ergative construction. To use a simple example, in Hindi perfective events
TYPOLOGICAL CHANGES 13
ma gaya: h:
I go+PP/M be+1/Sg
"I have gone"
me gejl'om
I go+PP+1/Sg
"I went"
gada=mhi
go+PP/M=be+l/Sg
"I went"
14 VIT BUBENIK
Considering the examples for the perfect in (11) kadamhi "I have made"
and pesidamhi "I have sent" we reach a dysfunctional state of affairs pre
sented in (18); that is the same construction of the PP plus the copula may
grammaticalize the active "I have sent" (i.e. I am the one who has sent) and
the passive "I am/ have been sent".
ma:ribo hose
being-struck became+1/Sg
"I was being struck"
We do not have to worry about the third solution since there is no trace
of the "go"-passive in ER. In what follows I will attempt to show that the
synthetic passive of ER goes back to the MIA "become"-passive (solution ii).
The synthetic passive in ER is formed by the suffix -jov which is
attached to the stem of the preterite plus non-final affix, the so-called NFA
(Hancock 1992: 41) The same suffix -jov is used to derive inchoative verbal
forms from adjectives. In Slovak Romani, according to Hbschmannov et
18 VIT BUBENIK
al. (1991: 647), the derivational suffix is -uv and the final consonant of the
verbal stem is palatalized (d,g > d', k,t > t', l > l', n > , st > t'). The
source of palatality is obviously the yod of the suffix as preserved in Vlach
Romani. Some examples are presented in (21):
The causativizing suffix -ar is most likely descended from MIA -a:d
which survived as -ad in Gujarati (cf. Turner 1926: 279). It should be added
that late MIA possessed also the suffix -a:r (e.g. baisa:riyau "he seated him"
in Svayambhdeva's Ritthanemicariu [1.9.3], ascribed to 7-10th century). As
above for the synthetic passive, I want to claim that the yod (seen in Vlach
Romani, e.g. ratjarel) goes back to the OIA derivational suffix -ya. The / seen
in sovl'arel and rovl'arel is found also in Hindi (sula:na:,rula:na:). As a part
of my argument I suggest considering this / as an intrusive consonant
between p (ancestral to v) and the yod', the p in sov ( < svap) belonged to the
root wheras that in rov ( < roda-p(a)ya-) belonged to the causative suffix
20 VIT BUBENIK
Thus, paradoxically, the yod, the original causativity marker has been
reduced to the feature of palatality in Slovak Romani, while the intrusive /
ended up as a full segment which in Hindi co-indexes the causativity (contrast
sona: with sula:na:).
Finally, a few comments regarding the source of -il- in the past passive,
e.g. ker-d-il-em "I was born" (in Vlach Romani, cf. Hancock 1992: 41). The
same -il- is used to express the past of the inchoative verbs, e.g. phur-il-em
"I got old", lol-il-e "they turned red". The finitizing suffixes are those of the
copula in the present, as shown in (10). As mentioned by Turner (1926: 279)
the most obvious candidate is the Prakrit suffix -ilia- (cf. Sanskrit -ila-,
equivalent of -vant). In Prakrits it was used to enlarge the participles (e.g.
a:ga-elliya: "having come" corresponding to Sanskrit a:gata-va:n). In Modern
IA languages it is found in Marathi {dekhila: "seen", vs. Hindi dekha:),
Gujarati and the Eastern languages (cf. Bloch 1933: 267).
5. Conclusion
One of the striking results of our enquiry into the contact phenomena of
European Romani was an overall picture of systemic complication rather than
that of simplification, however the latter is defined. Obviously, the length and
depth of contact play an important role in determining exactly what happens.
Let us remind ourselves of another well documented example, namely, that of
Asia Minor Greek discussed extensively by Thomason and Kaufman (1988:
TYPOLOGICAL CHANGES 21
65) under the motto "Turkish soul in the Greek body" (from Dawkins 1916:
198). The strong cultural pressure of Turkish on Asia Minor Greek, lasting
almost one millenium, resulted in heavy structural borrowing (the highest
point on their scale) affecting the innermost core of language, such as the
introduction of Turkish vowel harmony and agglutinative morphology into
several of the most affected Greek dialects. Data from EuropeanRomanimay
be used to exemplify such a strong degree of structural borrowing that we
may actually talk about massive replacement of large portions of inherited
grammar. On the other hand, we saw some remarkable retention of MIA
grammatical categories under the thick overlay of structural borrowings from
European languages. Their co-existence in ER strikes me as a genuine case of
overall systemic complication: the co-existence of the system of the definite
article with the system of seven cases; the co-existence of inherited post
positions with borrowed prepositions, and the appearance of typologically
rare circumpositions (documented in Pashto, Amharic and Asia Minor Greek,
cf. Campbell et al. 1988); the co-existence of the inherited deflected adjectival
agreement with the borrowed full adjectival agreement. In verbal categories,
ER in addition to the inherited present, perfect and the passive, displays an
innovative reflexive voice; in addition to the inherited indicative and
imperative an innovative subjunctive. In phonology Russian and Czech
Romani are more complex than either the MIA ancestor or the Slavic donors
of the feature of palatalization. The consonantal system of Russian Romani
includes voiceless aspirates descended mostly from the MIA murmured stops
but also a full set of palatalized consonants.
One of the most remarkable archaisms of ER, namely, its nominative-
accusative typology, has not been - to the best of my knowledge - discussed
in pertinent literature. There are two conceivable answers to this problem: (i)
Romani possessed the ergative construction before its ancestral speakers left
India, but lost it later on as a result of its long contacts with European
languages which do not have it; or, (ii) Romani did not have chance to
develop it because the Dmba left India before this construction crystallized
during the later period of MIA. The first solution could be entertained, if we
could prove that theDmbaleft India relatively late during the centuries of the
Muslim expansion, llth-12th century, (cf. Hancock 1993: 1-2 and this
volume) when the ergative construction was more or less established.
Alternatively, one may side with those Romanologues who favor an early
outmigration from India (Kenrick 1976, Kaufman 1984). Either way one has
to conclude that ER in its nominative-accusative typology preserved a
22 VIT BUBENIK
remarkable archaism of late MIA (cf. Matras, this volume). Thus, the
evidence of ER is of cardinal importance in solving one of the fundamental
problems of IA linguistics: that of the emergence and development of the
ergative construction (cf. most recently Bubenik 1994).
NOTES
1
Tessitori (1915:81):
na bolai=li: [Packhyna, 310; 1500-50 A.D.]
"[If] you will not speak"
amhe pachai kar:=la: [Upadesamlblvabodha 288; 1500-50 A.D.]
2
Old Western Rajasthani (Tessitori 1915:78):
ja: cha "I am going" (16th c.)
kahai chai "You are saying" (16th c.)
bhamai chai "He is wandering about" (16th c.)
Old Gujarati (Dave 1935:49):
sium kahau chau "what are you going to say?" (16th c.)
3 According to Jesina (1882:27,31) in Czech Romani the future is formed by attaching
the copula/verb "to go" in the future tense to the root of the main verb:
Sg 1 cor-ava < cor + (v)aba ~ avava
2 -eha + (v)eha aveha
3 -ela + (v)ela avela
"steal" + "I will be "I will go"
4 Paspati (1870: 95) mentions the presence of yod in the perfect as one of the
distinguishing characters of the dialect of the sedentary Gypsies vs. its absence in that of
the nomadic Gypsies: (in his orthography) kerghim vs. kerdm (nomadic).
5 Hbschmannov (p.c.) informs me that mardo in the analytic passive constructions is
used in its secondary meaning "punished", e.g.joj mard i le Devlestar "she is punished by
the God/fate". To translate "I was beaten by the father" one would prefer the synthetic
passive tnard il om le dadestar to the analytic *sotnas mardo. The analytic passive occurs in
the proverb somas buter mardo sar calo "I was more beaten than full".
6 OIA derivational suffix -ya:
kri-y-te "is made" finite passives (deverbative)
tavis-y_- "is mighty" inchoatives (denominative)
ka:r-ya "[which is] to be done" gerundives (deverbative)
mukh-ya "principal" adjectives (denominative)
TYPOLOGICAL CHANGES 23
REFERENCES
Pott, A.F. (1844) Die Zigeuner in Europa und Asien. Halle: Fricke.
Puchmeyer, Anton (1821) Romni cib: Das ist Grammatik und Wrterbuch
der Zigeuner Sprache. Prague: Wildenbrunn.
Sampson, John (1926) The Dialect of the Gypsies of Wales. Oxford:
University Press.
Shackle, C. (1972) Punjabi. London: Hodder & Stoughton.
Tessitori, L.P. (1914-1916) "Notes on the Grammar of the Old Western
Rajasthani", The Indian Antiquary.
Thomason, Sarah G. and Terrence Kaufman (1988) Language Contact,
Creolization, and Genetic Linguistics. Berkeley: University of California
Press.
Turner, Sir Ralph (1926) The position of Romani in Indo-Aryan. Journal of
the Gypsy Lore Society, Third Series 5 (4), 145-89.
Wentzel, Tatjana V. (1980) Die Zigeunersprache. Leipzig: VEB Verlag
Enzyklopdie Leipzig.
ON THE MIGRATION AND AFFILITION OF THE DMBA:
IRANIAN WORDS IN ROM, LOM AND DOM GYPSY*
IAN HANCOCK
University of Texas
0. Introduction
Charles Godfrey Leland (1875:637, see also Andreas 1926:129), re
ported the observation of an Iranian gentleman who, on the subject of Roman
Persian words, such as one hears from peasant grandmothers." About one
hundred are discussed below, only some of which may be accepted incon-
testibly as being of Iranian origin; but there are literally dozens of items in
theRomanidialects which remain etymologically unaccounted for, and it is
likely that a good number of them may eventually also be identified as
Iranian.
Paspati (1870:1), in his oft-repeated quote, said that the key to the his
tory of the Roma is to be sought in the study of theRomanilanguage. Since
the revelation in 1760 to Valyi Istvn that theRomanispoken in the heart of
Europe was actually of Indian origin (subsequently published in 1776),
scholars have to this day been attempting to determine when and why the
ancestors of the Roma left their homeland, whom they are most closely re
lated to there, and under what circumstances, and by what route, did they
enter Europe. A great number of hypotheses have been proposed (sum
marized in Hancock 1988), the most persuasive of which have certainly
relied upon linguistic, rather than purely historical, and sometimes
anecdotal, evidence.
Using linguistic material, the migratory route out of Central India can
be reconstructed, particularly through an examination of the sources of the
lexical items adopted along the way; thus beyond the now well-established
Central and Northwestern core, the presence of Dardic (Miklosich 1872,
Pischel 1883, Grierson 1908) and Burushaski (Berger 1959, Kenrick 1976)
elements suggests - assuming no great changes in linguistic territory over
26 IAN HANCOCK
the past millennium - that their westwardly route passed through the Hindu
Kush, south of the Turkic languages and north of Pashto (neither of which
is well represented in Gypsy), along the southern and southwestern
shoreline of the Caspian Sea through Iranian linguistic territory (Persian
and Kurdish) the southern Caucasus (Ossetic, Georgian and Armenian), on
through the Byzantine Empire (Medieval Greek) and thence into Europe.
The job of the historian is to reconstruct the social events involving
those early migrants which took place within the geographical territory sug
gested by this lexical material. Were the ancestors of the present-day Roma
Rajputs, as one current hypothesis maintains (Hancock 1991) and, if they
were, was it from this area that they moved out to encounter the Afghani
and Turkic Ghaznavids or was it further south - information necessary to
support or disprove that theory. And if, according to the same hypothesis,
the Pratihra did indeed constitute an element in the social (and subse
quently genetic) make-up of those who left India, were they of Scythian
origin, and were those Scythians the ancestors of the Iranian-speaking
Ossetes, as Abaev indicates (1964:xi)? If so, can Iranian linguistic influence
be dated back to the pre-exodus period? It is in these areas of Gypsy lin
guistic and social history that the Romanologist must seek his answers.
He was saying, in other words, that while Romani and Domari are both
Indic, it did not mean that the ancestors of both necessarily formed one pop
ulation while they were still in India. Nevertheless it is still widely accepted
that the Dom, Lom and Rom branches of Gypsy are related in terms of their
original speakers' having left India as one population and separating only
once having passed through Persia. One recent statement in support of this
is found in Ventzel (1991:102):
The route of the ancestors of the Roma passed through Iranian lin
guistic territory. One group established itself in the north and the west
of Persia, from whence it subsequently moved into Syria and Palestine,
and [thence] into Armenia. The majority, however, reached the Byzan
tine Empire. Those whom we call Tsiganes are the descendants of those
nomadic tribes who passed on through to the Balkans.
More recently still, Harcourt Films released a film entitled The Romany
Trail in which this historical account was presented with animated maps
and interviews with Gypsies in Egypt. Their exodus from India was
28 IAN HANCOCK
structed means that such a language could have existed" (p. 38), and on that
basis go on to reconstruct such non-lndic Proto-Gypsy forms as
'lord" and "pear" (from Kurdish xod and Persian amrd respec
tively) though at the same time maintaining that in India, several hundred
miles to the east of those languages, Proto-Gypsy "was an actual language
... in the Central Zone or on the border between the C[entral] and W[estern]
zones" (1984:39-40).
Some years before Turner voiced his suspicions that Dom Gypsy had a
different linguistic history from Rom Gypsy, Finck (1907:49-50) had also
made the same claim for Lomavren, which he believed was probably of
sauraseni descent, unlike Romani, which he saw as a Dardic language.
However, since Lomavren survives only as a lexicon (like Angloromani or
Cal), a fact first mentioned by Papasian (1901:126), conclusions regarding
its specific Indic affinity cannot be addressed on the basis of its structure. It
is significant, however, that not one word in the overwhelming Armenian
component of Lomavren appears among the ca. 45 Armenian-derived items
in European Romani (Redzosko 1984).
There are reasons why a separate origin, at least for Domari, seems
likely: First, if the ancestors of its speakers left India during the fifth
century AD (the hypothesis which rests upon Firdausi) and entered Europe
in the early Medieval period, where were they located and under what cir
cumstances were they existing during those intervening six centuries spent
in the Middle East before leaving the area of Arabic linguistic influence and
entering the Byzantine Empire? And why, during such a long period of
time, was there such scant lexical impact from Arabic upon European
Romani3 when just two centuries of contact with Medieval Greek - a mere
third of that time - has resulted in over two hundred items from that lan
guage being adopted - on the other hand, Domari is more than half Arabic
in its contemporary lexicon. Secondly, although Sampson disputes this
(1926:125, para. 272), according to Macalister (1914:9), Domari retains
vestiges of a third, neuter grammatical gender, which reflects the time of its
exodus from India; the neuter gender became lost in the neo-lndic lan
guages before the ancestors of the Rom (and presumably the Lorn) left, but
after those of the Dom left. The present paper discusses a third argument:
the impact of the Iranian lexical content, that is words adopted from
Persian, Kurdish, and Ossetic. If the Dmba had moved into, and resided in,
Iranian-speaking territory as one population before dividing into the Syrian,
Armenian and European groups, we would expect the lexical material
30 IAN HANCOCK
2. The languages
All sources for the following items are listed in the acknowledgements
and the bibliography; only in particular cases are they cited for individual
items. Given the evident migratory route of the early Romani population, a
closer examination of other languages besides those dealt with here (e.g.
Pamir, Pashto, Zebaki, etc.) is clearly in order, and will be part of a larger
study.
IRANIAN WORDS 31
The only attempt to date to list and identify the various languages and
ethnonyms which occur in the literature is Kenrick (1976). Those dealt with
here are, besidesRomani,Lomavren, Domari, Karaci and Mitnp:
Lomavren is the language of the Lorn, or Bosa as they are called in Tur
kish. Because it is now only a variety of Armenian, it has probably existed
in intimate contact with that language for a very long time. According to
Narodi Kavkaza, Vol. 2, page 40 (cited in Kenrick 1976:32), there are few
Lom now remaining in Armenia, but they constitute a viable community in
neighboring Azerbaijan. Benninghaus (1991) lists a number of localities in
eastern Turkey which are also inhabited by Lorn.
Domari is the language of the Dom or Nawar, who inhabit Syria,
Lebanon, Egypt, Israel and neighboring countries.
Karai is a variety of Domari, as the higher incidence of shared forms
between the two (below) demonstrates. The population refers to itself as
Dom. The name Karaci may derive from the town of Karai near Isfhan;
the proposed origin in Turkish kara "black" + -ci"agentive nominalizer" is
not grammatical in that language (although kara-ci "land-dweller," is; there
is also a Turkish form karaca, [karad3a] which means "somewhat
swarthy"). It is applied to groups inhabiting western Turkey, Syria, Iran and
Iraq also. Kenrick (loc. cit.) lists a North and a South dialect which differ
appreciably, and there are probably others. Jochelson (1928:172) estimated
a population of half a million Gypsies in Asiatic Russia 65 years ago.
Andrews (1989:139) list a figure of "10,633 registered peripatetics," in the
Turkish census called gebe ve gezginci ingeneler ("nomadic and trav
elling Gypsies").
Mitrip is also a variety of Domari, and may be a name of Arabic origin
meaning "musician," cf. Ar. Motribiyya, "Gypsy." The population's self-
designation is Dom. The speakers of this dialect live along the Turkish
borders with Iran and Iraq in Batman, Elmayaka, Van and other towns. All
Mitnp data here are from Benninghaus.
variable, but the perfective marker (i.e. the short forms of the BE verb)
changes for person and number: raft am "I have gone", raft i "you have
gone", rafl ast "he, she has gone" etc..Romanidiffers from each of these,
in that the person/number marker is retained, and the third person singular
imperfective/pluperfective morpheme has been generalized for all persons/
numbers: dikhv-as, dikhs-as, dikhl-as, etc.. The loss of final /t/ from the
hypothesized underlying Persian ast is a normal phonological feature of the
language (cf. [vas] for <vast>, "hand", [bus] for <bust> "skewer", etc.)7.
The second grammatical acqusition from Iranian is the comparative
suffix {-der}, enclitic to the oblique adjective: baro "big", bareder
"bigger". While Sanskrit had the related form -tara, Sampson has noted
(1926:150) that the structure "is lacking in A[siatic] Gyp[sy] (i.e. Domari)
as well as in the Indian vernaculars", though he provides parallels with Skt.
at para. 125.3 on p. 53. Hindi uses the oblique adjective followed by
postpositional se: bora "big", bore se "bigger", and Banjara does the same:
moto "big", mote si "bigger". Gujarati adds -kartm or -th to the nominal
preceeding the adjective: mot "big", str mo t-kartm hokr he "a
woman is bigger than a girl", while Sindhi "uses the Iranian comparative in
-taf' (Campbell 1991: 1235). In the Iranian languages, however, the
Romani construction is found. In both Pehlevi (Middle Persian) and
Modern Persian, the comparative suffix is -tar: bozorg "big," bozorgtar
"bigger." Kurdish has -ter: gewre "big" gewreter "bigger," while the Ossetic
form -der comes closest of all toRomani:saw "black" sawder "blacker."
The possibility of an Ossetic origin is strengthened by the fact that the
Common Romani comparative feder (or fededer) "better" may also be
Ossetic; see discussion for this entry below. Pashto has the cognate
comparative morpheme tor, but this functions prepositionally only.
Kostov (1963:99) has suggested that the model for forming the nume
rals between eleven and nineteen (with des "ten" + -u-: desu-jekh "11," des-
u-duj "12," etc.) is either Persian or Armenian; the R. model is neither Skt.
or New Indian; cf. Skt. ekdasa,dvdasa, Hi. gyarah, brah, "11, 12," but
cf. Arm. tasn-u-mek, tasn-u-erku "11, 12," P, bist-o-yek bist-o-do "21, 22"
(Romani has conjunctive ta for compound numerals above 20). In some
Vlach dialects, conjunctive -u- is not used in compounds with the non-
Indian numerals efta "7," oxto "8" and enja "9." Further discussion of
numerals is given below.
34 IAN HANCOCK
R26 erxan, sky arx "sky." Miklosich's Ar. source seems to be a ghost
erxaj, akano word.
R27 i, ii, i anything, ci, ih "what." Bloch (1926:139) dismisses Skt.
nothing, not indefinite enclitic -cid because it cannot exist unbound
(Sampson, 1926:60).
R28 inri plane (tree sp.) enar "plane-tree" (Albanian R.)
R29 ukni, upni whip buk "active, quick, nimble"
R30 desto handle; stick dastah "handle, haft"
R31 divno conversation dvan, K diwan "council", also Tk, Balkan lgs.
Non-final stress suggests later adoption in Europe.
R32 diz town, fortress diz, diz "fort," K diz "fort"
R33 dorjv, sea, lake, river dary "sea"
dorjvo
R34 do-, du to milk K du-, du- "to milk"
R35 duman, enemy dolman "enemy," also Rum, Tk, Ar, Psh, Ur.
dumano
R36 damutro son in law dmd "son-in-law," cf. D potra "son", but
almost certainly I, with convergence, cf. Peh damat,
Skt mtr.
R37 di, (pi. da), ozi stomach, heart, soul K jn"life," dzagr "stomach."
Convergence of unrelated Ir forms in R?
R38 dukel dog aghl "jackal." This is usu. listed as < Skt jakuta
"dog," but Sampson says this has no descendants in
neo-Indic. Soravia ( 1988:5) however lists Mynwle
jukel, Sansi chhkkal, Kanjari jhu nkil (these lan
guages are discussed in Grierson, 1922).
R39 feder, fededer better Oss fid "strong." Sampson tentatively suggests
Skt bhadratara "better," but development of R HI
would be hard to explain, {-der} may itself be < Oss
{-der}. Feder is the only comp, form surviving in the
Vlach dialects, in which it means (as adverbial maj
feder) "rather, sooner."
R40 gertjno, throat, gullet K gerdin "throat," cf. Rum dial, girtan, poss. Slav.
girtno See also R. 50)
R41 harbuz, melon xrbuz "kaveh melon, donkey-mouth melon,"
arbuz Gk and Tk have karpuz(i); also Rum.
36 IAN HANCOCK
R42 kam- to desire Peh. kmitan "wish," cf. Skt kmyate. Turner says this
is not IR; cf. Hi kam "desire, wish," although we would
have expected the R form *kav- if it were < I.
R43 kanzavri hedgehog xrndaz "porcupine" (?) (Balkan R. only, and almost
certainly < Gk skanzxoiros
R44 katun sheet, cloth qutn "cloth" < Ar. Also Gk & Slavic, see
Sampson, (1926:136)
R45 kermso, rat karmu "rat"
kermsa, kermso
R46 khangeri church kangura "battlement, turret". Discussion at D 6 below;
see Sampson (1926:133)
R47 kirmo worm kirm "worm," + Hi, cf. Skt kpnih, Pkt kimi. (Arabic
adopted the Skt form as qirmiz, from which the P form
is derived. Cf. E kermes "cochineal insect," also
carmine, crimson). Turner believes this to be directly
from Skt.
R48 ki , ke , ke silk kaz "raw silk"
R49 kisi pouch ks, ksa "pouch," < Ar.
R50 korr neck, gullet K karrhk, P grdn "neck"
R51 korro, koro blind kr "blind," + Arm kuir, K kor, Hi kor
R52 kui cup kze "pot, vessel," K kz
R53 kun, kunsus corner kunj "corner"
R54 kta, kti wrestling match koti "wrestling," K koe "effort." Albanian R.
R55 kutik belt kuti "girdle worn by Parsis"
R56 lalo, mute, dumb lal "mute," + Hi, cf. D lala (cf. the Lallere Sinti).
lavoro, lavodo
R57 1av word lafz "word," K lebz (<Ar). Sampson (1926:191) ..
rejects Skt lpa/Hi sallp. Tk has laf, a common
form in many Balkan R dialects
R58 lis, li terror larzdan "tremble," cf. D rzri "tremble."
R59 lono, loeno joyful ron "bright, splendid" K rro in "delighted, happy,
lit up." Sampson (1926:200) says not < Skt las, but cf.
Skt rocana "bright, pleasing, lovely." See also L22,
L29
R60 majmno monkey meymun "monkey," K, Tk maymn, Rum momuie,
and other Balkan lgs.
R61 mjna plain, savanna K mejdan (< Ar)
R62 mamux, sloe mahmiz "sloe fruit," but cf. Tk mahmuz (< Ir?)
IRANIAN WORDS 37
mamuxo
R63 mjve, fruit mlveh "fruit," K mewe, but cf. Tk meyva.
mvo, mibao
R64 mero person K merdum "people," Old P merah "husband"
Greek & Spanish Romani
R65 mom, mum wax mom "wax," + K + Hi
R66 momeli, candle The P word for this is am, in K it is mom as
memel i, mum(b)ali above. The R form, according to Sampson (1926:82,
233), is formed from mom + the adjectival suffix
{-al-), the word therefore meaning "(the) waxy (one)."
The form might however be Arm, thus momeln "of
wax."
R67 muso, musos, mouse mus "mouse," cf. Hi ms, msak
muakos, mia
R68 na, anav name nam "name," + Hi
R69 nian sign, signal nean "sign," K nian "sign, mark, spot, meeting place"
R70 orde, here Oss orth "here"
vorde, arde
R71 pata, footwrap ptva, pataba "leggings," cf. Hi; E "puttee"
patay, patavo
R72 paver-, rear, foster parvardan "nourish", + Hi. Cf. Skt. prati "protect,
parvar- nourish"
R73 pendex, nut banduk "filbert", cf. K bendak,
penax, penaxa Gk fountouiki "hazelnut, filbert."
R74 perde curtain prdeh "curtain." Cf. Hi,Urpurda "veil," hence E
"purdah." But R form prob, immediately < Tk perde
R75 phurt, bridge purt, purd "bridge," but cf. Skt *prtu- "bring across"
phurd, phurt
R76 por feather par "feather" (prob, not cognate with R patrin "leaf."
See Sampson, 1926:282 and Bloch, 1926:140), though
cf. Pkt pama-
R77 poom, wool pam "wool," "down"
poum
R78 poti hide, skin post "skin," postn "fur," + Hi
R79 poxtan, cloth paxta "cotton"
potan
R80 rad- set out on a K rra-y "set out on journey" (R form probably
journey compounded with d- "give;" Hi rahn unlikely).
38 IAN HANCOCK
R103 xandz an itch xrdan "itch," xrndan "scratch," xrxr "an itch"
R104 xanrrnd- scratch as above. Cf. also K xuran "itch," Arm xantel "itch"
See Sampson (1926:177).
R105 xar valley, gully xr, xre "slope, valley," but cf. Skt khadda "hole"
R106 xeljlax, xeljal herbs kirkir "kind of bean or pea", cf. Gk kekris
R107 xer donkey xar "donkey," but cf. Pkt khara-
R108 xevis, xejic, cornmeal mush K xav "cold, uncooked meal"
xivici, xovici
R109 xirpa term of abuse; K xirpo "term of abuse"
asshole
R110 xolov, sock, stocking Oss xalaf. Wolfs comparison with Bulgarian
xoliv, xoluv holev (1960:134) reveals no such word found in that
language. Tk has orab, do.
R111 xulano, chief, leader, K xola "landlord"
xulaj host
R112 xumer dough, dumpling xamr "dough", Also Ar, Arm
Rl 13 xurdo, small, petty, xurd "small." Ashiti (an archaic Iranic dialect
xuredo insignificant between Kurmanci and Persian) has xori "it's a
waste of time." H khurd less likely.
R114 zen saddle zin "saddle," K zi n
R115 zeja back Plural of zin (< zinja > zija > zeja, with
characteristic Vlach loss of intervocalic nasal) but
not with this meaning in P or K.
Rl16 zet, diet, oil zeit "oil"
zetino
Jin "a spirit, genii"
R118 zor, ruz-, power, strength zor "power," + Arm + K
o
Rl 19 zumav- try, test, prove azmudn "try out, test" + K. A doublet with R13 azb-:
Gjerdman & Ljungberg (1963:203) attempt no
etymology for azb-, while Sampson (1926:409)
suggests a Romanian source for zumav- but does not
provide one.
Romani Domari
amal friend beli, sahib
angutri ring ngli, wirg
avin steel stm
azb- touch mnr
baxt luck miritk
berk breast i, xri
bust/buzex spit; spur sx
buzno goat kli
erxan sky, star lhii, xy
do- milk xlaur
katun tent xmi, kuri
khangeri church ktirnkki
kisi pouch, pocket b
kugtik belt kli
lig terror bi, xf
momeli candle mixri
pendex nut bgnn
IRANIAN WORDS 43
K 1 alav flame lw
K 2 angudari ring angutar
K 3 bafir snow barf
K 4 banir cheese panr
K 5 bugandus he dug kandan
K 6 ia hole, well ah
K 7 dava camel dav- "run"
K 8 deh village dih
K 9 dost friend dst
K10 dara candle iry
K11daver barley Jaw
K12 gand sugar qand, kand "candy"
K13 gyrmyzi beautiful; red girmiz "crimson," cf. kirmo
K14 hafta week hafta (also "seven")
K15 hanaq trick, joke hanaq-myqa "do not trifle"
Kl6 hargu everywhere haru
K17 erimda graze ardan
K18 urabaura diverse Jr-ba-Jr
K19 ku beard kusa
K20 kuli cap kulah
K21 luleh leg lla "tube"
K22 mahni they stopped mandan
K23 mahudi cloth mht
44 IAN HANCOCK
4. Analysis
The following breakdown of shared forms includes items which are
probably of Indic rather than Iranian origin ("stone", "eat", "needle",
"rain"), or which might have been adopted from Arabic ("monkey," "dark
blue," in Karaci). Not included are cognates which, because of semantic
shift, were probably inherited independently and cannot support common
acquisition (for example R khangeri "church" and D kangri "waggon," R
lono "happy" and L rowan "bright," R kirmo "worm" and K gyrmyzi
"red," R efta "seven" and K hafta "week").
13 poom "wool" - pm -
14 rad- "depart" - raw- -
15 res- "reach" - ras- -
16 suv "needle" ? suzan su -
17 tang "narrow" - tng -
18 xa- "eat" ? xar - -
19 xer "donkey" - qar qar
20- "apple" sib, ansev - sib
21- "god" xod xuja xuja
22- "after" - pai pa
23- "ass" - qar qar
24-- "barley" - dzau dzaver
25- "beard" - ku ku
26- "dark blue" - nila nila
27- "foot" - pau paf
28- "mouth" - zari zever
29- "or" - ya ya
30- "spittle"? - f tu
31- "uncle" - xal xalum
32- "village" - de deh
33- "yellow" - zerda zardari
Of the above items, Romani and Lomavren share eight, Romani and
Domari/Karaci share sixteen, Lomavren and Domari/Karaci share three, and
Domari and Karaci share sixteen, to be expected since they are both Dom
languages. None of the eleven Persian-derived items in Mitnp occur in any
of the other languages, and there are no Iranian items at all shared by all
three (Rom, Lorn and Dom).
Among the total of 119 items of Iranian or possible Iranian origin in
Romani, the seven shared with Lomavren constitute about one fifteenth or
roughly 7 %; the sixteen shared with Domari/Karaci just over one sixth, or
about 15.5 %. Of the total of 36 items of Iranian or possible Iranian origin
in Lomavren, those it shares with Domari/Karaci constitute one eighth, or
about 12 %. Of the sixty items in Karaci, sixteen are shared with Domari,
which is about 26 %. These ratios would be even lower if doubtful items,
such as those followed by a question mark, were not counted. The corre
spondences would then be two items shared by Romani and Lomavren or
2.3%, eleven items shared by Romani and Domari-Karaci, or 11.5%, and
IRANIAN WORD 47
5. Conclusion
It may be assumed that the low incidence of shared Iranian-derived vo
cabulary between Romani and Lomavren or Romani and Domari/Karaci,
and the absence of any items shared by all three, argues strongly against
their separation after having coexisted in Iranian-speaking territory. Even
Sampson (1923:164) admits that "lacking in Nuri [ie. Domari] are several
important loanwords [from Persian occurring inRomani],which may per
haps be regarded as evidence that the two bands had separated before these
later Persian borrowings were absorbed into the speech of the western Gyp
sies." Whether these figures prove separation within Indian territory, or
outside of India before reaching Persian territory, however, remains to be
demonstrated using other criteria.
NOTES
I should like to thank Ali Jazyery and Mohammed Ghanoonparvar for their help with
Persian, Ms. Corinna Leschber for her help with Kurdish, Ms. Naciye Kunt for her help
48 IAN HANCOCK
with Turkish, and Angus Fraser, Anthony Grant, Norbert Boretzky, Victor Friedman and
Donald Kenrick for their valuable comments on earlier drafts of this paper.
1 Although Leland (1873) gave himself the credit for making this connection, as he did
for the "discovery" of Shelta sixty-five years after McElligott first wrote about it (Han
cock, 1984:385), it was in fact Hermann Brockhaus who first suggested it, in a letter to Pott
dated July 16th, 1841. The relevant section of that document (found in Pott, 1844, Vol. I,
on page 42) is important, and bears reproduction here:
In the collection of fairy tales of the Somadeva edited by myself, we find Tar. 13 l.
96 (page 169), and in Khlana's History of Kashmir, e.g. V. 353, the word Dmba
(with retroflex d), and Wilson makes at this point the comment that this name in
dicates a kind of pariah. Since this word is missing from the Sanskrit dictionaries, and
thus is not considered to be classical by Indian grammarians, it must therefore belong
to the words borrowed from the colloquial vernaculars. In Hindi, we actually find the
word d'ma, feminine d'mn, with which a person of the lowest class is labelled.
Might not this word dom be the same as the Gypsy Rom? Doesn't this perhaps refer to
a tribe originally living in north-western India which, being subjugated, were
degraded to the status of pariahs? The fact that a people don't call themselves by a
name indicating something dishonorable is obvious; only through subjugation can the
name of a people become a name of opprobrium among the victors.
2
By Kaufman, 1984.
3 The position (first mentioned by Groome, 1963:xxxiii, and discussed in Hancock,
1988:206) which assumes that the Dom and the Rom left India together and passed through
Persia before the 7th century rests upon the supposed lack of Arabic influence in the latter,
the argument being that after that date, the spreading influence of Arabic upon Middle
Eastern languages would have been reflected in Romani. However, while Arabic began to
affect the liturgical vocabulary of Persian at this early date, it did not have any considerable
impact upon the colloquial speech until the 11th or 12th centuries. In addition, the route of
the Dmba through Persia seems to have been along the shoreline of the Caspian Sea,
geographically remote from Arabia and, presumably, the original linguistic influence of
Arabic. Nevertheless, Arabic has made some impact upon Romani. De Goeje (1903:55)
claimed that the Arabic items he lists "all occur in European Gypsy dialects...and they
sufficiently establish the theory that all the Gypsies of Europe have lived for a long time
among Arabic-speaking people." Although Miklosich (1877 ff., vol. 6, pp. 63-64) has
made a convincing case against most of them, and some are found neither in Romani nor
Arabic, they are repeated here because of the inaccessibility of the original list, and for
future reexamination: agor "end," Ar xir, alikati "time," Ar al-ikt, axal- "understand,"
Ar 'aqal, baxt "luck," Ar baxt, berk "breast," Ar berka, aro "dish," Ar ahn, eni
"earring," Ar odhni ("ear"), oro "deep," Ar ghr, handako "ditch,"Ar xandak, jar "heat,"
Ar harr, kghed "paper," Ar kghed, kha "house," Ar kha, kx, katuna "tent," Ar
qaitun, ke, kez "silk," Ar ke, kazz,, kisi "purse," Ar ksi, ko tor "piece," Ar kot'a, kurko
"Sunday," Ar kurki, ma(j)muno "monkey," Ar maimun ("happy"), mom "wax," Ar mom,
moxto "box," Ar motn, pendex "nut," Ar bondoq, tremo "vestibule," Ar trima, xasar-
"lose," Ar xasar, xev "hole," Ar kav, xud-, ud- "seize," Ar axadi, xumer "dough," Ar
kamr, zeiti "oil," Ar zeiti.
IRANIAN WORDS 49
4
While Vlach e "what" only coincidentally resembles the Persian, CommonRomaniCi
and its variants meaning "nothing, anything" may be of Iranic origin. This item is discussed
at R 25.
5
Abbreviations in the lists are as follows: Afg Afghani (Pashto), Ar Arabic, Arm
Armenian, D Domari, Dar Dardic, E English, Gk Greek, Hi Hindi, I Indian, Indic, Ir
Iranian, K Kurdish, Ka Kashmiri, Kar Karai, L Lomavren, OP Old Persian, OI Old
Indian, Oss Ossetic, P Persian, Pan Panjabi, Peh Pehlevi, Pkt Prakrit, Psh Pashto, R
Romani, Rum Rumanian, Sh Shina, Skt Sanskrit, Tk Turkish, Ur Urdu.
6 This form is only found in literary Persian today, but may have been a part of the
colloquial language a millennium ago.
7 For the treatment of prothetic lv-1 in vast and other thematic items in Romani, see
Turner (1975). For an alternative analysis of {-as}, see Bubenik, this volume.
REFERENCES
Mann, Stuart (1933) Albanian Romani. Journal of the Gypsy Lore Society,
Third Series, 12(1), 1 -32.
Marre, Jeremy, producer/director (1992) The Romany Trail Pt. 1: Into
Africa. Beats of the Heart Series, Shanachie Records, SH 1210.
Harcourt Films, Inc.
Miklosich, Franz (1872-1881) Ueber die Mundarten und die Wanderungen
der Zigeuner Europa's. Vienna: Karl Gerold.
Papasian, Vrtanes M. ( 1901) Les Boschas (Tsiganes) armniens. Erivan:
State Printer.
Paspati, Alexandre G. ( 1870) Etude sur les Tchinghians ou Bohemiens de
VEmpire Ottoman. Constantinople: Karomla.
Pischel, W. (1883) Die Heimath der Zigeuner. Deutsche Rundschau, 36,
353-375.
Pott, Augustus F. (1844-5) Die Zigeuner in Europa und Asien. Two
volumes. Halle: Heynemann Verlag.
Pott, Augustus F. (1846) Ueber die Sprache der Zigeuner in Syrien. Zeit
schrift fr die Wissenschaft der Sprache 1, 175-186.
Redzosko, Y. le (1984) Armenian contributions to the Gypsy language.
Ararat 25 (4), 2-6.
Sampson, John (1923) On the origin and early migration of the Gypsies.
Journal of the Gypsy Lore Society, Third Series, 2 (4), 156-169.
Sampson, John (1926) The Dialect of the Gypsies of Wales. Oxford: The
Clarendon Press.
Soravia, Giulio (1988) Di alcune etimologie zingariche, Archivio Glotto-
logico Italiano 73, 3-11.
Trudgill, Peter, ed. (1984) Language in the British Isles. Cambridge:
University Press.
Turner, Ralph (1927) The position of Romani in Indo-Aryan. Journal of the
Gypsy Lore Society, Third Series, 5 (4), 145-183.
Turner, Ralph (1975) So-called prothetic v- and y- in European Romani,
ln:Collected Papers, I9I2-I973. London: Oxford University Press.
331-335.
Wolf, Siegmund (1960) Wrterbuch der Zigeunersprache. Mannheim:
Bibliographisches Institut.
Ventzel, Tatiana V. (1991) Le 'Bosa,' parler 'insulaire' des Roms d'Armenie.
In: Bakker & Cortiade (1991), 102-105.
Wahby, Taufiq & C.J. Edmonds (1966) A Kurdish-English dictionary.
Oxford: The Clarendon Press.
PLAGIARISM AND LEXICAL ORPHANS
IN THE EUROPEAN ROMANI LEXICON*
ANTHONY P. GRANT
University of Bradford
0. Introduction
A considerable amount of attention has been paid to the extraction from
the European Romani lexicon of elements from the genetic Indo-Aryan
component, and to profiling the broad etymological structure of the voca
bulary. This paper discusses two issues in the documentation of the European
Romani lexicon which have hitherto received scant attention. The first is that
of the authenticity of many of the data themselves, their status as
representative samples of the dialect under discussion and as bona fides Ro-
mani. Much "Romani" data is actually fabricated from collections made from
several dialects, or contains a number of metanalysed (and often invented)
words, and several instances of this are discussed. The second issue is that of
lexical orphans, genuine words of whatever origin, but especially firmly-
established and integrated borrowings, in the way in which Poplack and
Sankoff (1984) used the term, which are attested only in one dialect in one or
more reliable sources, and which do not derive from the lexicon of the host
language. Such words - and not least early loans from the Greek and South
Slavic strata common to all dialects of European Romani- are important for
tracing the migrations of the speakers of a particular dialect and in
documenting the earlier history of the dialect.
searchers), the temptation to adopt the relevant Hindustani words tallo and
kunara into a recognised Indic language must have been strong indeed
(though the latter word is a Farsi loan in Hindi!). Other common book-loans,
deriving from Hindustani via Grellmann like those two, such as banduk
"rifle, gun", a word of Turkish origin, are equally spurious as tokens of
Romani.
The first known instance of plagiarism ofRomanidata is that committed
by Samuel Bjrckman, a Swedish pastor and scholar of the early eighteenth
century, who wrote a Latin work, Dissertatio Academica de Cingaris (1730),
in which he passed off a number of items lifted bodily from Johannes
Scaliger's or Bonaventura Vulcanius' late sixteenth-century vocabulary as his
own, despite claims that he had checked it over with aRomani-speakerwho
was then in prison. The fact that Bjrckman's glossary is copied from
Vulcanius-Scaliger can be shown by the fact that both share the same
orthography, often spelling the same words identically, thus both use <ch>
for several sounds including /x/, // and //, both list the same words, such as
<buchos> "book" and both share mistakes, such as final <-t> for /-1/ in
<tzuket> "dog" (Romani dzuke) or <for> for por "feather". As can be seen,
the similarities in transcription and the shared mistakes give him away.
The publication of the economist Grellmann's work, itself largely derived
from secondary sources (both published and unpublished) of varying accu
racy and dialect affiliation, rather than a great deal of independent fieldwork
(Wolf 1960: 39), and coloured by what he had heard and read about Sanskrit
and Hindustani, was a significant event which had often rather dire con
sequences for Romani linguistic scholarship, as he introduced a number of
ghost-words into the published material onRomanithrough the perpetuation
of printing errors, especially into the records on Central European and Sinto
dialects, for example such nonexistent words as feizrile "tomorrow" (for
*teizrile, compare Sinto tisrla), or telel "animal" (where German Tier
"animal" has been misread for tief "deep"; tlal means "underneath" in
Romani; "deep" is the Armenian loan xor). Grellmann's book was plagiarised
and excerpted more than any other single work, and its mistakes and falsities
were copied along with the more genuine data.
dialect" out of Romungro, Lovari and invented data, was still felt in the
1960s, and one of his works was used and noted in Wolf (1960), who
evidently thought the work genuine, while words from the other works of
Wlislocki were included in a commentary on Wolf (1960) by Johann
Knobloch (Knobloch 1964), together with a number of words, mostly
derived from Hungarian, from texts which Knobloch had collected at Nazi
concentration camps in Austria as material for his doctoral dissertation
(Knobloch 1953), from speakers of various forms of Romungro and Lovari,
shortly before they were taken away to the gas chambers. Wlislocki seems to
have preferred to invent his own lexemes (for instance glete "tongue",
authentic Romani chib), rather than to draw from Grellmann's polluted well
like his predecessors.
Sinto dialects have often been documented in works largely derived from
other works whose lexical content is of questionable authenticity. The first
extensive source for Sinto lexicon is the dictionary compiled by the
Thuringian judge Wilhelm Ferdinand Bischoff (Bischoff 1827), who drew
on printed sources, including Grellmann's work (which contains material
from Romungro and Lovari as well as Sinto), and on his own fieldwork,
some of which was evidently carried out with Vlach-speaking Roma. This
mlange is perpetuated through the nineteenth century by authors who stole
data from Bischoff s book: a word recorded by Bischoff or maybe taken by
him from Grellmann's book would be taken from his dictionary and included
by the magistrate Richard Liebich (Liebich 1863), and others working from
Liebich's book would add it to their list.
Eventually Rudolf von Sowa sorted out the problems in this field,
producing a sensible dictionary of Sinto, based on his own fieldwork in
Westphalia and East Prussia and on printed sources, and keeping separate the
Baltic Romani dialect of East Prussia, which others had lumped with Sinto
simply because East Prussia was German territory. In his dictionary of 1898
von Sowa printed the more reliable words in italics and the others, including
words recorded in no other dialect and other more dubious ones, a good half
of the ones collected, in plain type. This list was incorporated bodily into
Finck's grammar, complete with errors, where Finck included dubious
words. Some words listed in Finck (1903), taken from earlier sources,
whose authenticity I doubt are: <karaw> "I pull out" (first listed by Bischoff
as "Kahraf, ich rcke aus", and attested for no other dialect, but included in
Rishi 1974); <k'eledo> "penance", that is, Bue, for Rue, also meaning
"soldier, beau, lover"; see a form such as Northern PolishRomanixalado
58 ANTHONY P. GRANT
"soldier", Kraus and Zippel <chellado> "lover" in von Sowa (1898) and in
Wolf (1960: 133); <kora> "hour" (which some authors trace back to
Hungarian koran "late" (!) although German Sinto has no Magyar loans, and
any such listed for German Sinto are a sign that the source under consultation
is false; it probably derives from metanalysis of jekh ora "one hour" as *je
kora); and probably also: <dsajel> /dajl/ "to freeze", which is reputedly of
Indic origin (allegedly from Sanskrit jaayati), but unknown outside Sinto and
first noted by Bischoff; elsewhere "freeze" is expressed by forms of mraz-
from South Slavic, for instance Welsh Romani mzin "it's freezing", or on a
verb derived from pho "ice", from Greek, e.g. Lovari pahosarl).
Quite why Finck perpetrated such a fraud is a mystery, but deceive people
he did, and he attempted to cover his trail, as one sees from the preface to his
book (especially pages VH-VI11), where he tells of having consulted with
Sint during his time as a professor at Marburg (1896-1900), then checking
his material with von Sowa's dictionary. The modern published lexica of
Sinto dialects show the soundness of von Sowa's instincts as to which items
were bogus, as they are not noted by scholars working with native speakers,
for example, being unlisted by Calvet and Formoso (1987).
A curious instance of misrepresentation of lexical data through erroneous
decontextualisation occurs in Wolf (1960), with several proper names from
Sinto. These names are taken from a published conversation in Sinto, taken
down in the late 1830s by the seminarian Tielich at the artificial Romani
colony set up by the Lutheran Church at Friedrichslohra in Thuringia, and
reproduced in Pott (1884-1845: 1: 491-497). The speakers to whom the cues
are attributed have German and Sinto names, thus the Sinto Anton is known
as Polla in Sinto, the speaker Wilhelm's name is given as Hater, the speaker
Franz has the Sinto name Kringla, and so on. These were individual people,
and they had individual Sinto names as well as German ones; however, Wolf
translates Polla as "Anton", for example, as if it were the stock translation of
the name, and as if it could be applied as a Sinto name to anyone who was
called Anton, which is not the case.
Borrowing at second and third hand was taken a stage further, when one
Eduard Hrkal compiled a dictionary of "Central European Romani" from a
series of dubious sources - Jesina for Carpathian, Finck for Sinto, Wlislocki
for the spurious Transylvanian dialect (Hrkal 1940), thus succeeding in
incorporating a number of forms, such as kunara, whose history of falsity
extended over a century and a half.
PLAGIARISM AND LEXICAL ORPHANS 59
dialect, is that they follow their more cautious instincts and list such terms
separately, drawing on their knowledge of the migration routes of the group
of Roma in question (bearing in mind, for instance, that there are no loans
from Hungarian orRomanianin the Kalajdzi dialects), and do not use them
as evidence in trying to clinch any theoretical point, while noting and
monitoring forms of dubious authenticity and cross-checking them with more
reliable sources and data from other dialects.
this may in turn be based on Northern English /nut/ "nut"), stigl "alone",
pii "certainly" (the last two are of unknown etymology).
It must be pointed out that not all words of unknown etymology are
confined to one dialect. There are a number of pandialectal words, such as
gadzo "non-Rom", or por "tail", for which satisfactory etymologies have not
been found. And there are words confined to one subgroup, which form part
of the shared lexicon of that group, whose origins are mysterious.
Naturally there are loan orphans (often from French or Italian) and unique
lexical innovations in German Sinto, too, words such as matrli "potatoes",
for example, and an incautious approach to the study of the Sinto lexicon
might cause such words to be discarded, simply because they do not occur in
other dialects. In this case, evidence from the lexica of subdialects, such as
the Sinto dialects of France and northern Italy, is important. There is much
false material in many of the published Sinto lexica which must be sifted from
the genuine material, but we do not want to throw the baby out with the
bathwater.
There are also a number of lexical orphans in Finnish Romani. This
dialect shares many of the Greek and Slavic loans which are to be found in
Sinto and Welsh Romani, as well as having maybe 700 stems borrowed from
Swedish (e.g. vella "evening" from kvlljego "one's own" from egen,
dzenom "through" from genom, ilako "evil" from archaic elak "loathsome"
crossed with illa "evil", and msava "to roam" from msa "to roam";
(Thesleff 1901). Furthermore, there are some strays, isolated loans from
Scots (banaka "breadcake" from bannock), and Norse zurupos "settlement"
from porp.Some words from earlier strata, such as jinderdi "rainbow"
(recorded in the speech of the Lajenge Roma and rather fancifully claimed by
some to be related to Sanskrit indradhanu "Indra's bow" and celadoin
"swallow" (< Greek) are not found outside Finnish Romani.
Swedish was a source language for loans into Finnish Romani which, of
course, was not exploited by other Romani dialects, the Roma having reached
Finland after a long stay in Sweden; furthermore, Swedish was long a
prestigious language in Finland, and most Finish Kaale (as they call
themselves: "black", like the Spanish and Welsh Gypsies), bear Swedish
surnames. Thus all Swedish loans can be said to be "loan orphans". The
words from Norse and Scots are the only words from these languages in
Finnish Romani.
Finnish Romani also had its unique unetymologisable elements, such as
phab "cheek, side of face", also found in Scandinavian Romani mixed
64 ANTHONY P. GRANT
languages but not elsewhere, and also ummi "women's defensive weapon".
The penultimate example is a word of unknown etymology not found in any
other dialect, which usually have reflexes of cham. The final one is also of
unknown etymology and seems to have been regarded by Thesleff as an odd
word, and it is not unique to FinnishRomani,by a hair's breadth. In a case
which shows the value to Romani philology of pre-modern materials, it
otherwise occurs only as <schiimije> "skewer", in the pre-1570 wordlist of
Johan van Ewsum of Groningen, Netherlands, our earliest example of a
Sinto dialect, and one of our earliest samples of Romani (Kluyver 1910-
1911). Thus it is noted both for Sinto and for FinnishRomani,and could be
attributed back to the Northern Division of European Romani, as a word of
unknown source like paijar "case, cover, border, edge", shared by Finnish
Romani, Romnimos and Sinto, would be an item of shared lexicon in
Northern Romani.
Other dialects, apart from those here discussed, have their share of old
loanwords which are otherwise unknown in other dialects. Examples will be
drawn from the Greek-derived part of the lexicon. Thus SlovakRomanihas
ten "an equal, a counterpart" from a form connected with the Greek verb
teriazo, which I have otherwise only seen in terenices "stalemate, a draw; an
equal result between two marbles-players", attested in the Kalajdzi dialect of
the Burgudzi in Macedonia, and kindly brought to my attention by Norbert
Boretzky; nor have I met FinnishRomaniguduni "bell", also from Greek (the
etymon is kuuni)elsewhere apart from North RussianRomani.Yet another
Greek loan, timin "price, cost", occurs only in North Russian Romani, and
as a loan in Russian Kalderas (Demeter 1990). The Greek loan kockarida
"hiccough", from kloksos, is only found in CarpathianRomani,while forms
for "heel" derived from patuna are to be found only in Carpathian and some
Balkan dialects.
All of these are valid if not widely dispersed early loans into Romani,
from the Greek stratum of lexicon, the first truly European stratum in the
Romani lexicon, an important element which bridges the gap between
thematic and athematic grammar (older loans being assimilated into the
essentially Indic grammar), and which still awaits full documentation and
study.
Kaufman's criterion, that a European loanword only be included in a pan-
dialectal dictionary of Romani if it is attested in at least two of the twenty
groups, is fundamentally sound, but it might be tempered with some
historical hedging, especially in the case of loan orphans which are obviously
PLAGIARISM AND LEXICAL ORPHANS 65
relics of the early migration of the Roma in that they are taken from Greek or
South Slavic. One may argue that a hard-line approach to the admission to the
shared lexicon of EuropeanRomaniof loans noted only in one dialect group
can obscure matters of historical or cultural interest, as can overreliance on
data from a betterrecorded subdialect.
I shall close with examples relating to the Greek stratum. Even a brilliant
work such as Sampson (1926) does not give a complete picture of British
Romani at all times, and omits some elements not recorded by Sampson and
which need to be supplied from other sources. Thus, we see that Welsh
Romani has the Greek word for "heaven", ranos, in ravnos, as well as such
words as valgra "market", from Greek ayor. Neither of these words is
found in other dialects. The pan-dialectal word for "town", foros, is not
found in WelshRomani,where gav, elsewhere "village", is now "town", and
"village" is vlija, but Andrew Boorde collected it in 1547 for English Romani
in our oldest sample of Romani.
The Greek numerals borrowed intoRomaniare usually felt to be eft oxt
enj trinda sarnda pennda, "7, 8, 9, 30, 40, 50", which are still used in
manyRomanidialects (though the word for "forty" has oddly dropped out of
Kalderas and is replaced by a compound form starvardes, while Alexandre
Paspati also noted eksinda for "sixty" in a Drindari-type dialect spoken at
Yambol in Bulgaria). There is no evidence for the Greek numerals in
Sampson's WelshRomani,apart from desto xori "eighteen pence", though
"7, 8, 9" are listed in England by Jacob Bryant in 1784 (see Sampson 1910-
1911) and were apparently heard by him rather than copied from elsewhere.
But WelshRomanihas drika "dozen", from Greek /eka/, known in no
other dialect, while Kotel Drindari has dekapnde "fifteen", unattested
elsewhere (Donald Kenrick, personal communication). Could it be that once
most of the Greek cardinal numeral system (apart from terms for 1-6, 10, 20
and 100) was used in Romani?
5 . Conclusion
As Grellmann pointed out over two centuries ago, the history of the
Roma is to be found in their language, and the length of their stay in various
countries can be gauged by the proportion of loans into Romani from each
source. An exploration of the mustier corners of the lexica ofRomanidialects
will tell us much more about the history and travels of the various groups,
and of the relations obtaining between particular groups. However, when
studying the lexica of one or moreRomanidialects, it is important to bear in
66 ANTHONY P. GRANT
NOTES
I would like to thank Peter Bakker, Jette Bolle, Norbert Boretzky, yit Bubenik, Jean
Cooney, Angus Fraser, Victor Friedman, John Green, Ian Hancock, Milena
Hbschmannov, Birgit Igla, Donald Kenrick, Corinna Leschber, Christopher Sheppard,
and Hein van der Voort for their assistance and encouragement in the research of which this
paper is an offshoot.
Najs tumnge!
1 It is through the dialect mixture in Grellmann's work that the Hungarian and most of
the Romanian loans (and the false Indicisms) "entered" the Sinto lexicon. There are no
Magyar loans in any Northern European Romani dialects apart from Carpathian ones, and
only a few Romanian ones. The early speakers of Northern Romani dialects apparently
went via Croatia and Slovenia into Germanophone territory in what is now Austria (and
maybe briefly into Bohemia) before entering Germany.
2 An opposite example is a word allegedly meaning "false", vingro, which is listed by
Puchmayer (1821: 51) as a word which he could not elicit from his consultant, but noted as
occuring in Grellmann's book; Grellmann also noted a word latschila, also supposed to
mean "false" and also unrecorded by careful observers. Grellmann took these words from an
early vocabulary, apparently an Angloromani list, which listed : Lachilo, vingro "false".
The intrusive comma led Grellmann to believe that there were two words there when there
was in fact a single word, a form of * lahelavngero "good-word person", that is, one who
says good words but who does not support them with good deeds, and thus a false person.
As one might expect, Jesina (1886) swallowed the two words whole, as lailo and vingo.
REFERENCES
Bischoff, Ferdinand (1827) Deutsch-Zigeunerisches Wrterbuch. Ilmenau.
Calvet, Georges and Bernard Formoso. (1987) Lexique tsigane 2: le sinto
pimontais. Paris: PoF.
Colocci, Adriano (1889) Gli Zingari. Torino: Loescher.
Decourdemanche, Jean-Adolphe (1908) Grammaire du tchingan ou Langue
des Bohmiens errants. Paris: Geuthner.
PLAGIARISM AND LEXICAL ORPHANS 67
Rdiger, Johann Chr. Chr. (1782) Von der Sprache und Herkunft der Zigeu
ner aus Indicn. Neuester Zuwachs der teutschen, fremden und all
gemeinen Sprachkunde in eigenen Aufstzen, Bcheranzeigen und
Nachrichten. 1,37-84.
Sampson, John (1910-11) Jacob Bryant. Journal of the Gypsy Lore Society,
2nd series, 4, 162-194.
Sampson, John (1908-1909) Review of Prince (1907). Journal of the Gypsy
Lore Society, 2nd series, volume 2 (1), 74-84.
Sampson, John (1926) The Dialect of the Gypsies of Wales. Oxford Univer
sity-Press.
Smart, Bath Charles and Henry Thomas Crofton (1874) The Dialect of the
English Gypsies. London: Asher.
Sowa, Rudolf von (1898) Wrterbuch des Dialekts der deutschen Zigeuner.
Abhandlungen fr die Kunde des Morgenlandes, XI. Band., 1. Teil.
Leipzig: Deutsche Morgenlndische Gesellschaft.
Thesleff, Arthur (1901) Wrterbuch des Dialekts der finnlndischen Zigeu
ner. Helsinki: Acta Scientarum Societatis Fennicae XXIX.
Wlislocki, Heinrich von (1884) Die Sprache der transsilvanischen
Zigeuner.Leipzig: Friedrich.
Wolf, Siegmund A. (1960) Groes Wrterbuch der Zigeunersprache (romani
tsiw). Wortschatz deutscher und anderer europischer Zigeunerdialekte.
Mannheim. [reprint: 1987, Hamburg: Helmut Buske Verlag)
Wratislaw, Rudolf, Graf von Mitrowic (1868) Versuch einer Darstellung der
Lebensweise, Herkunft und Sprache der Zigeuner. Prague.
INTERDIALECTAL INTERFERENCE IN ROMANI
NORBERT BORETZKY
Ruhr-University Bochum
0. Introduction
Romani is split up in a great number of dialects, and we very often find
speakers of different dialects living in the same community or even in the
same neighbourhood. Therefore, we would assume dialect mixture to come
about on a rather large scale. The following is a first attempt at collecting and
systematizing the phenomena of dialect interference in Romani, but before
going into detail let us discuss the factors favouring or impeding interference.
It is plausible that linguistic contact should occur at places where speakers
of various groups of Roma live together. We are not in a position to claim
that this has been the case in Central and Western Europe at all times, but in
the south-eastern part of our continent where stable urban settlements existed
for many centuries different groups living together in one area can be
considered the normal form of settlement up to our days. In Serbia and
Macedonia we normally come across no less than three groups in one town.
In general, Gypsies live in towns rather than in small scattered villages.
However, today we find small groups or single families in typical villages as
well. Apparently, the type of settling depends on the Gypsies' way of earning
their living. Thus, in Kosova the Bugurdzides, traditionally blacksmiths, are
found in villages as well. There is rarely a village without at least one
Bugurdzi family. Greater towns have their ciganske mahale (Gypsy quarters)
where separate groups live next to each other but in different streets or sub-
quarters rather than totally mixing.
According to what has been said here we might expect a far-reaching
levelling of dialects or, at least, quite chaotic variation, but this is not at all the
case. On the contrary, it is more difficult to find instances of mixture or
mutual influence, even random mixture, than to determine recent dialect dis
tinctions. In the following we want to discuss some factors that might have
controlled the Roma's linguistic behaviour.
70 NORBERT BORETZKY
between the dialects, because it is not at all clear what the major dialects were
like in early times. We do not know how to define the dialects linguistically,
which traits are old or which are regional innovations etc. It will be an impor
tant task for the near future to develop a classification that is based on lin
guistic criteria rather than on geographical factors and tribal denominations.
Obviously, this would be the best basis for analysing interference. In the
following, we want to discuss the importance of some isoglosses for
identifying the major dialects.
1.1.1. s-/h-
This alternation is found in the copula, and moreover in sar/har "how",
savo/havo "which", so/ho "what", i.e. in function words (interrogatives), h-
must be derived from s-, since this type of change occurs quite often in
various languages. In all likelihood it goes back to India, because the same
distinction is found between the new Indo-Aryan languages. Aspiration of s-
is most characteristic for Sinti and closely related dialects; in Arli1 it appears
in the copula only, along with s-forms:
hinjum, hinjan, hi (i) besides sinjum, sinjan, si/isi
In some varieties of Arli we have free variation, others seem to prefer one
or the other form. It is almost impossible to determine which form is the basic
one for which variety. Where forms with h- appear in other dialects, they
may have been introduced by interference (cf. 2.2.l.d).
There was ocurrence of h- in the speech of one Gurbetka (Pozarevac). In
this case, it is totally unclear how it has become part of the dialect.
This phenomenon must not be confused with the aspiration of final
(grammatical) -s occuring, for instance, in Bosnian Gurbet.
but there also seem to be Sinti-like dialects displaying -an - in all probability
an old interference (cf. 2.2.2. i).
1.1.4. -nr-/-nd(r)-/-rn-/-r-
This variation seems to be less old than the aforementioned, but it is not a
recent one either. The various forms can be derived from one another phonet
ically in the following way:
a) -nd- (Indian) > -nd (preserved in Prilep and in one of the varieties
described by Paspati 1870, the dialect of the nomads; cf. Tableau
comparatif 118ff);
b) -nd- > -nr-, cerebrality being preserved in the sound of [r] (in Kalderas,
partially in Lovari and Gurbet), and further
c) -n- > -nr- with loss of cerebrality (Dambazi, older Sinti; cf. Bischoff
1827);
d) -nr-/-nr- > -ndr- (occasionally found in Greece, Bulgaria, and Ukraina);
e) -nr- > -rn- (in northern Gurbet, in Bosnian Gurbet; cf. Uhlik 1942);
f) -nr- > -r (Arli, Bugurdzi; central and northern dialects, among them Sinti,
cf. e.g. Finck 1903; Drindari, which is closely related to Bugurdzi, as
well as Erli have the strong cerebral instead, cf. Gilliat-Smith 1914).
On the whole the reflexes of -nd- have a clear-cut and plausible distribu
tion, but some details are rather puzzling: southern Gurbet has -nr-, whereas
the northern variant has -rn-. Can this be explained as a late metathesis of -nr-
INTERDIALECTAL INTERFERENCE 75
> -rn-? The same distinction can be found within certain subdialects in Greece
(Vlach-type;cf.Igla l989).
1.2. Innovations
The problem with innovations in general is that they may either have
come about independently in different dialects, or else they owe their exist
ence to contact. It is less convincing to assume a simultaneous initiation of a
given sound change in an area of interrelated dialects. If similarities are not
the result of common change, but of simple chance, then, of course, they
cannot tell us anything about (historical) dialect coherence.
in Arli and in the dialect of Prilep, but also in southern Gurbet and Dzambazi,
and there perhaps even more consistently. It is, however, absent in Prizren. It
is remarkable that speakers of Arli acknowledge -el(a) as a possible form,
although they prefer -ol(a). How is this process to be reconstructed? Did it
start in Arli and spread as an ongoing change over the neighbouring dialects
of Gurbet/Dzambazi and others, or has -ol(a) been borrowed from Arli in a
later process? In the speech of southern Gurbet -el seems to be absent, but
this must not mean that the change is older in Gurbet than in Arli (cf. -el in
the northern variety of Gurbet). The behaviour of Prilep speakers is rather
unusual: they consider -el to be the correct form, but nevertheless prefer -ol in
their speech. I am tempted to take this as an indication of a recent influence on
the part of Arli and/or Dzambazi. It should be noted that the Paspatian dialect
of Greece, which is close to that of Prilep, shows no traces of -ol (cf. Paspati
1870). From all this we can conclude that at least in Prilep -ol came up by
interference, maybe after a longer-lasting influence from Arli and Dzambazi,
whereas the speakers of the Prizren dialect (Kosova) preserved -el because
they immigrated in more recent times.
The reasons for the change -el > -ol are by no means clear. The process
may also have been an assimilation of e to velar /.
2 . Interference phenomena
2.1.Planned, intentional, conscious interference
2.1.1. Written sources
Interference can be brought about consciously, or it can occur as the con
sequence of an unintended act. We find instances of the first type when in
dividuals or groups try to standardizeRomaniin order to use it as a broader
means of communication, mixing elements from different dialects. We also
find it when local standards are to be developed on the basis of one dialect,
incorporating elements from other dialects that have become obsolete in the
basic dialect or that do not seem to be very distinctive; cf. the following cases:
A. Saip Jusuf of Skopje, a Dzambaz by origin, chose Arli as the basis for his
grammar (Kepeski & Jusuf 1980), but he managed to avoid extreme forms of
Arli, and he adopted some elements from other dialects, although in quite a
random way; cf. gndinela "to think" from Kald. gndil (a romanism); khonik
"nobody" and khanci "nothing", extinct in all Arli varieties known to me,
from Dzambazi; po, pi "on, at, above" also from Dzambazi; maj (compara
tive) from Vlach, along with po- (comparative) and naj-(superlative), which
INTERDIALECTAL INTERFERENCE 77
are borrowed from Macedonian3. As for the copula, forms beginning with h-
have been avoided, since they differ too much from Vlach and other dialects.
B. Contrary to this procedure, Ali Krasnici of Prishtina used his local variety
of Gurbet when writing his short stories (1981 and 1986). He even preserved
such idiosyncratic forms as munro "my" instead of the more widespread and
better understandable miro, moro, mo. To be sure, there are foreign elements
in his language, but it is difficult to determine what their status is:
a) are they older elements that have become firmly integrated into the
dialect; or
b) are they elements taken up unconsciously; or
c) are they utilized consciously in order to enrich the language?
In the following we will try to evaluate the importance of some of these
elements.
a) Demonstratives with word-initial k-, e.g. kava, fem. kaja "this", kova
"that", kate "here" appear along with others beginning with g-: gova goda
godo "that", godolese "therefore", g(e)ja "thus", gasavo "such", gothe
"there". It is possible that forms containing k- gradually have been assi
milated into the dialect which originally had g-forms only (cf. northern
Gurbet). On the other hand, we find k-forms in Bosnian Gurbet (Uhlik
1942), but here as well it is not clear whether this is a result of dialect mixture
or not. Also, it is possible that Gurbet never lost the k-forms totally.
Although it is difficult to present clear-cut criteria, I believe that the k-forms
in the language of Krasnici are of foreign origin.
A morphological case is the past of inkle l "go out, go up"; in Vlach it
should be inklisto (northern Gurbet, Kalderas), but here we have ikislo,
ikisli, which in all likelihood derive from Arli.
b) In hutlo "he jumped" the cluster -tl- is borrowed from Arli, xuklo
being the correct Vlach form. It appeares to be a rather unconscious borrow
ing, brought about by the author's familiarity with the other dialect. The
normal Vlach form of the verbal (and deadjectival) noun is -ipe (-ape), but in
the texts -ipa is also used; cf. darajpa "fear", sastipa "health", -ipa might be a
contamination of -ipe and Arli -iba, but since even -ipa is not unkown in Arli
(e.g. in the Arli of Gnjilane), it may have been borrowed as such. Again, we
cannot say if forms ending in -ipa are in common use in southern Gurbet, or
if we are confronted with individual borrowing here.
78 NORBERT BORETZKY
For "forget" we find bistrol, but also a metathesized form bristol too. The
latter is an idiosyncratic form, the better known bistrol may have been used
unintentionally.
c) punro "foot" is the form we expect in this dialect, and it seems to be
used throughout, but for the derivation "step" we find prnalo <prno, which
is common in northern Gurbet. It is likely that prnalo is taken from another
dialect, since the derived word was not available in the autor's own dialect.
A rare word is aindz "field", normally used only in Bugurdzi. The author
maintained that he heard it from older Gurbet speakers, but he conceded that
today it is unknown to speakers of his dialect. This is clearly a case of con
scious enrichment.4
prandosarel (pe) "to marry" is normally used in Arli and Bugurdzi, but it
did not belong to the inventory of Vlach. In all likelihood the author bor
rowed it from Arli, possibly because the Vlach romanisms mrtil (pe) and
insuril (pe) have become obsolete in his dialect.
a) Phonology:
- bers, -bersengiri (Arli) alongside bres (Dzambazi);
- maw alongside manro/mangw,
- tikni alongside ciknv,
- dive alongside g'ive (phonetically uncertain);
- on, oj alongside von, voj (uncertain);
- k instead of g in demonstratives: kava, kal, okoja, kate alongside gova,
gothar; in all likelihood this variation is quite common.
-pro "Fu" alongside prno,punro
b) Morphology:
- definite article fem. i instead of Vlach e (probably the only possible
form);
- genitive -kiri etc., cf. efta-bersengiri, lakiri alongside lengi',
- preposition tar "from", with article tar-o, tar-i (tari mi chej) alongside
katar (katar o Munster)',
- preposition ko, ki "in" (ki Germanija) besides an-o, an-i (ano lil, ani
strand)', apparently, this alternation has become customary;
- preposition ko, ki "on, at" (ke staklje "on crutches") besides po,pi
(po aviono);
- preposition kare/kara "at, near" (kare mande, kare late) instead of kaj
or the pure locative;
- copula past tense ine besides sa;
- copula subjunctive te ovol instead of avol (uncertain form);
80 NORBERT BORETZKY
a) Phonology:
- mandro "bread" with ndr (an archaism?);
- e borja acc. "the bride" with preserved) (possibly an archaism);
- hijas chivgjas chivde "to throw" (non-Vlach) besides huv chudas',
- haj "daughter" (non-Vlach) instead of hej;
- iklistilo "to go out", ikalel "to take out" instead of inklisto inkalel;
- cid- "to draw" (non-Vlach) besides crd-;
- kaxni "hen" besides kajnv,
-phanle "to bind" ( <phandle) (non-Vlach) instead of phangle.
b) Morphology:
- grammatical -s in: nom.sing. masc, prxos stlos (from Drindari?)
nom.plur.fem. prtes (from Drindari?)
3rd sing.past xaljas (an archaism?)
- definite article fem. i: i lisica, and-ifurnja instead of e;
- definite article plur. o: o (portes), and-o (khera), o (gurv);
- demonstrative plur. kal "these" (Gurbet should be kal; may be an old
form);
- copula som, past 1st sing. -gjom (non-Vlach; cf. the Bukovina texts
of Miklosich l874);
- cidingjas "drew", past tense extended in -in- (typically Arli);
- copula past sins (sic!) (non-Vlach) alongside sas;
- present with -a: dan-a "they go" (non-Vlach);
- loan verbs in -in-: mislinel, cudinel pes, cidingjas alongside isprat-isar-
gjas (Arli);
- verbal noun in -ibe: maribe.
c) Vocabulary:
- xri (and zlak) instead of xanci, cora etc. "something, a little" (xari
is non-Vlach).
For this text we have no indication whatsoever that the storyteller mixed
forms of different dialects in order to make himself better understood. It
seems that this is a dialect in which the quoted interferences occur quite
82 NORBERT BORETZKY
normally. This can be taken for granted in those cases where the text has no
alternatives, as for instance with grammatical -s, with the feminine article i,
with 1st sing.past -om, and the expanded past forms in -in-.
2.2.1. Phonology
a) Demonstratives beginning with g- are considered typical of the Gurbet
group of dialects. Here the older Indian k- has been replaced by g- for
unknown reasons. Now, a Pristina informant for Gurbet gave me the k-
forms as the normal ones in his dialect, for instance kava kova kote, although
in the text of a song forms with g- were used as well. I would like to interpret
this state as a free vacillation that has come about under the influence of the
prevalent Arli. (Cf. what has been said above about the distribution of k- and
g- in the texts of Krasnici; 2.1.1. B)
b) Palatalized k g appears in Macedonian Dambazi as [k' g'], whereas in
the closely related Gurbet either [6 dz] contrasting with /c dz/ (northern
variant) or [c d] (Kosova) is found. Since the Gurbet as a Vlach tribe must
have immigrated from the North, they should have brought to Macedonia an
already palatalized variant of the [6 dz] type. Under the influence of the local
Albanian and Serbian these have merged with old /[c d]/, but in Macedonia
in all likelihood the earlier sounds [k' g'] have been restituted. This recon
struction is corroborated by the fact that the velars are lacking in original
clusters containing s; cf. les-e < les-ke "him", les-i < les-ki "her" possess..
The only way of explaining this discrepancy is to assume an intermediate
stage [lesk'e], since the k could not have been dropped in a form with velar
[k], i.e. in [leske]. The restitution of [k' g'] in Dambazi has been effected
under the influence of Macedonian, but probably supported by Macedonian
Arli, which as a Romani dialect displays the same words (cognates) as
Gurbet. This might be called etymological support. The reconstruction
scheme runs as follows:
kinel > k'inel > cinel (northern Gurbet) > cinel (Kosova Gurbet);
> k'inel (restitution in Dambazi);
leske > lesk'e > *lesce (earlier Dambazi) > lese (both in Gurbet and
Dambazi without restitution).
INTERDIALECTAL INTERFERENCE 83
but a Sinti variety of Piemont (Senzera 1986) presents two paradigms, one of
them with s-:
som sal si besides om al i, apparently < hom hal hi, h- being lost under
the influence of Italian. In the Sinti of the Slovenian Prekmurje, neighbouring
villages differ as to s- and h- (Igla, personal communication; Strukelj 19807),
and other dialects mix both forms in one paradigm. There are Central and
North European non-Vlach dialects, however, that have preserved s-
throughout (for North Russia cf. Wentzel 1980; for Slovak and Hungarian
dialects cf. von Sowa 1887: 92f, and Hancock 1990; for South Poland cf.
Kopernicki 1930). Data are presented in the following table :
interdialectal interference; cf. so and ho "what", sar and har "how", but
perhaps only havo "which". A possible source for the restitution of the s-
forms is Lovari.
In the Hungarian Vend dialect (Vekerdi 1984: 72), which roughly fol
lows the Slovak paradigm, a specializiation of si and hi has taken place: hi is
"is", whereas si means "have" in the construction si man etc. "I have". It may
well be that reduction began with enclitic si "is" but did not extend to fully
stressed si in si man etc..
To sum up, it can be said that aspiration must have taken place on Euro
pean soil, at least in some dialects, and that not all mixed or variating para
digms owe their existence to interference.
In Vlach dialects, for instance in Lovari and in Bosnian Gurbet, a variant i
/-j' < si can be found as well, but this reduction is restricted to the 3rd person.
e) The change from h d> can be taken as a feature distinguishing
Kalderas and Lovari from other dialects. In dialects spoken in West Ukraine
(cf. Barannikov 1934) forms with h d alternate with those containing ,
which clearly points to interference. I am not sure as to the basis of those
dialects, but it seems to me that we are confronted with a deformed Kalderas.
The newly emerged are rather marked sounds, and it is less likely that
they should have been transferred into other dialects.
f) In Bugurdzi the word "day" has the forms zis and zies < dives. A
Kovac from Skopje cited the form zives, which might be a contamination of
indigeneous zies and Arli dive.
g) The normal form for "town" in Bugurdi is ziz. A speaker from
Southern Kosova used diz instead, apparently under the influence of Arli,
which prompted another speaker to blaming him for not speaking the pure
dialect.
"her" and 3rd plur. len-go "their" (cf. Pobozniak 1964: 49). While, in
general, this innovation did not spread to the closely related Kalderas, we
find it in a variety spoken in the Vojvodina. Surely, the speakers do not call
themselves Kalderasa, but they are certain about not being Lovara. Moreover,
they speak a dialect which differs from Kalderas varieties spoken in Romania
and in Serbia south of the Danube river in unimportant details only. I assume
that plur. pen- has been adopted through contact with Slavonian Lovari
dialects.
d) The preposition pe "on, at" became obsolete in practically all varieties
of Arli, and in most of them the preposition an "in" is not in use anymore,
both being replaced by ke, originally "to, at" (with article ko ki ke). This
process appears to have spread to Kosova Bugurdzi, where pe never occurs,
and an is randomly replaced by ke (but not the other way round). What is
noticeable in this case is that distinctions which appear useful if not necessary
have been abandoned because of pressure exerted by a more important dia
lect. Such cases are by no means unusual in language contact.
In Macedonian Dambazi we find a comparable situation. Here ke also
substitutes for an, but ke is used to an even greater extent; cf. ki bolnica "in
the hospital", ki Germanija 'in Germany"; ke staklje "on crutches". Maybe
substitution even operates vice versa, i.e. sometimes an appears where at an
earlier stage only ke was admitted; cf. potpisisajlji ano ljil "she signed the
paper", which may be interpreted "she gave her name on the paper"; ljilja ...
ani strana te phirol "she began to limp", literally "she began to walk to the
side"; ni primin tu ani buki "they don't employ you", literally "they don't take
you to work". The examples are not fully conclusive, though.
e) Dambazi as a Vlach dialect has the Romanian-derived comparative
marker maj, but alongside it po (for the comparative) and naj (for the
superlative) as well. 10 Since po and naj are Slavic elements we cannot say
with certainty whether they are borrowed from Slavic (Macedonian) directly
or from Arli, where they have become the only means of comparison,
presumably a long time ago. Even a third possibility, i.e. a joint influence
from Macedonian and Arli on Dambazi, cannot be excluded.
f) In Bugurdzi sine "was" is the normal form of the copula past and, as
expected, na-sine the negative form of it, but I heard older speakers using na
sas, i.e. a Vlach form. It is rather unlikely that the Bugurdi should have
borrowed it from Gurbet, firstly because the form of the copula is sa, not sas,
and secondly because it is Arli that serves as a source for borrowings in the
first place. (The Arli, i.e. Turk, yerli "resident, local" have been settling in
INTERDIALECTAL INTERFERENCE 87
the South of the Peninsula presumably since the Roma's arrival in Europe,
whereas the Gurbet/Dambaz re-immigrated from the North.) Therefore, we
have to assume that sas in na-sas is an older form of the copula that survived
in combination with negative na- alone. This means, however, that sine is an
innovation in Bugurdi that might well have been taken from Arli. The hypo
thesis of an Arli origin of sine is corroborated by Drindari sjas (Kenrick
1967:78). Judging from a number of common phonetic changes both dialects
must have had a common ancestor. 11
g) In the Drindari of Kotel, the subjunctive of "to be" is formed with ach-
originally "to remain", but besides it ti ovil (past ulu) can be found. Kenrick
(Ph.D. Thesis) assumes that ov- in this variety of Drindari has been
borrowed from other dialects. (Since Bugurdi and Bulgarian Kaljdzi also
have ov-, this form should not be totally alien to Drindari.)
h) In the same dialect the future tense is formed with ma-, mo- etc. <
mang- "to want", but kam- is used as well, again due to dialect mixture. It
has to be noted that the full verb kam- is not normally used in this dialect,
which points to the fact that kam- is probably of foreign origin.
i) The 2nd person sing.past tense of the verb underwent interference as
well. In a dialect spoken in the Austrian Burgenland, the speakers of which
call themselves Louvara, we find forms like dikl-l "you saw" and und-l
"you heard", but for the copula san "you are" with an n (Knobloch 1953).
Since the Louvara may have spoken a Vlach dialect before they moved west
ward, -al in all likelihood has been borrowed from a central dialect.
k) We already mentioned the 1st person sing.past in -om instead of -em in
the Bukovina dialect (Miklosich 1874), which is unexpected in view of the
fact that this variety displays all characteristics of a "good" Vlach dialect
(Romanian and Serbian Kalderas); cf.:
- h d > z (orthographically s z);
- vowel centralization i > % e > o in contact with sibilants etc.;
- weak palatalization of the velars before i: giv "wheat" > [d'iv];
- tl > kl, e.g. in xuklo from xutav "jump";
- contraction of pijen "they drink" >pen;
- loss of j- in jek(h) "one";
- reduction of katar "from" and andar "out of' > kata anda;
- -plurals with masc. nouns of the type dand "tooth", grast "horse";
- obliquus fem, of adjectives on -a, e.g. bar-a; ekh-a grasnja;
- reduction of len "them" acc. > le;
88 NORBERT BORETZKY
- future form of the verb on -a: lel-a "he will take"; kro <krav-a"I
will do" (contractions of this type are known from Hungarian Lovari);
- 3rd person sing.past of motion verbs both on -o/-i and -j-as (as in
Serbian Kalderas);
- sode "how much" instead of kobor, kozom etc.;
- article forms with /: le, la, l
What deviates from average Vlach is article forms like nom.plur. ol,
acc.sg.fem. ola and the like. An innovation not known to me from other dia
lects is the 3rd pers. sing.past tense on -ou, e.g. bel'ou instead of belo or
bel'as.
As for -em instead of -om, I assume that this formant entered the dialect
by interference from non-Vlach dialects of the Ukraine. This is corroborated
by another correspondence with dialects spoken north to it: the negator is na,
both for the indicative and the imperative, whereas in Kalderas we have ci
and in Gurbet ni for the indicative, but na for the imperative. In this the
Bukovina dialect shows the same characteristics as all dialects of the Ukraine
and Russia known to me; cf. na phenel "he does not speak" and na phen!
"don't speak!" Thus, the evidence points to the following: What we are faced
with is a Vlach dialect of the Kalderas type which has become altered in con
tact with northern dialects, but in a few traits only.
1) From Bugurdi a peculiar vacillation can be quoted: in the 1. pers.
sg.past tense we have both -om and -urn, which are used indiscriminately. I
was unable to elicit from the speakers which form was the "correct", the
"good" one, because my informants changed their opinion all the time. Since
-urn is common in Arli (and in the dialects of Prizren and Prilep as well), the
variation may be caused by influence from this dialect. Bulgarian Drindari,
which is closest to Bugurdi, has -im, a contracted form that cannot tell us
much (from -ijom or -ijum?).
m) In Bugurdi the use of the gerund (verbal adverb) is decreasing, tuj +
present is taking its place.11 I was able to find a few instances of the old
gerund though, some of them ending in -indo si-ando s, others in -indoj/
-andoj. Of those only the first variant seems to be indigenous, because Bu
gurdzi is a dialect that preserved final grammatical -s very consistently,
whereas it is lost in Arli and southern Gurbet; -indoj/-andoj may be borrowed
from a Vlach dialect, but it is more likely that there are varieties of Arli
spoken in Kosova which display -indoj as well.13
n) A case of interference in word formation can be found in Bugurdi.
The verbal noun shows much variation, since we have the widespread -ipe
INTERDIALECTAL INTERFERENCE 89
along with -ibe, -iba and even -ipa. I am unable to establish which of these
forms is the indigenous one, mainly because this category is not formed
automatically from all verbs and adjectives, and speakers are uncertain as to
therightform. At least -iba has been borrowed from Arli. Bulgarian Drindari
uses both -ipe and -ibe (the first for denominal, the second for deverbal
derivations; cf. sahcip < sastip "health", xoxeib < xoxa(v)ip) "fraud", so
both may belong to the inherited inventory of Bugurdi. -ipa should then
have come about by contamination of -ipe and -iba.
o) Drindari has pimus "drink" and xamus "food" besides piibi and xaibi
(Kenrick, Ph.D.Thesis), -mus clearly being of Vlach origin. Since -mus does
not occur freely but only in the two words quoted here, this is a case of
lexical borrowing rather than of a transfer of a word formation morpheme.
2.2.3. Vocabulary
At this level of language, interference is especially difficult to detect. All
European dialects have an essentially identical inherited vocabulary at their
disposal, and whenever we come across a word (etymon) in a subdialect
which is not documented in affiliated dialects, the given word may be an
archaism. To give an example: The word bov "stove" is not used in Kosova
Bugurdi, whereas in Kumanovo (northwestern Macedonia) it was known
and, as I was assured, in general use. Is this a relic in a variety of Bugurdzi
or did the speakers adopt it from Dambazi?
In Bugurdi kirvo and kirvi "godfather" and "godmother" are known,
although the institution of godparenthood is seldom practiced. What is more,
in Bugurdi the sequence ki- should have developed to ci-. Both factors point
to borrowing, perhaps fromDambazi[k'irvo].
For Drindari, Kenrick (Ph.D. Thesis) lists a number of words that appear
not to be indigenous: sukar "beautiful" and sukaribi "beauty" besides lachu
(sukar is absent from Bugurdi, too); phen-ke (a word calling someone's
attention), in all likelihood from a Kalderas-like dialect (with the meaning of
"says he/she, said he/she; that means"); suv "needle" (Vlach) alongside siv;
satra "tent" alongside katuna and cerxa; soru (Vlach) "head" alongside seru,
vas "hand" alongside the idiosyncratic as(t) of the dialect; za- alongside
nkja- "go, come"; va "yes". It has to be considered that va and za- are old
words now becoming obsolete. The same perhaps holds for rom with the
meaning "Gypsy" (not with the meaning "husband"), which is rarely used
alongside the derivation romaici < *roman-ici (a deminutive form, cf. rom-
oro in other dialects).14
90 NORBERT BORETZKY
3. Conclusion,
The investigation of interdialectal interference in Romani attempted here
provides convincing evidence for the assumption that there was not only
short-term interference in speech (parole), but also interference with lasting
effects on the language (langue). Many details could not be clarified mainly
because we still know too little about the historical splitting-up and
development of the dialects and the earlier distinctive features of the main
branches of Romani. On the whole, we can state that there is less interference
than might be expected under the given circumstances, with single dialects
having remained astonishingly homogeneous.
I hope that it has become clear that decisions on interference depend to a
high degree upon our insights into dialectal ramification as well as on the
accessibility of data.
Since in most cases it has been related dialects with very similar structures
which influenced one another, the amount of change through interference re
mains rather modest. Basic structures have not been changed, and adaptation
processes of greater extent have not taken place.
In general, it is lexical elements that undergo the most far-reaching
adaptation, not only with regard to their sound shape but also with regard to
inflectional class, syntactic behaviour and meaning. However, since we have
few clear cases of lexical borrowing little can be said about this topic. In
Bugurdzi the verb "live" appears in two variants, vest-iz-ava with the formant
-iz- typical of this dialect, and vest-in-ava, which because of -in- can be
considered to be of Arli origin. This means that morphological adaptation has
only been optional.
With morphological interference the following cases create no problems
for the replica dialect:
- Bukovina Vlach -om instead of -em (2.2.2.k);
- Bugurdzi -urn in addition to -om (2.2.2.1);
- Bugurdzi -indoj in addition to -indos (2.2.2.m);
- Bugurdzi na-sine instead of na-sas (2.2.2.f), making the
paradigm more homogeneous;
- Dambazi po- and naj instead of maj- (2.2.2.e), involving
categorial expansion;
- Kalderas plur. pen- in addition to pes- (2.2.2.c), contributing to
harmonizing morphological structures;
- Gurbet article i instead of e (2.2.2.b); paralleling the article with the
feminine formant -i.
INTERDIALECTAL INTERFERENCE 91
NOTES
1 If no other source is given the forms quoted here are collected by myself. For forms
that can be considered as generally known to scholars of Romani philology no sources are
given either.
2 Only some of the dialects west of the Dnepr have -em.
3 The indigeneous -eder has not been adopted, however, problably because it has fallen
out of use in most dialects of Serbia and Macedonia.
4 This word occurs in Bulgarian Erli too, in the form of alindz (personal communication
B.Igla). Since the form containing -/- must be older than Bugurdzi aindz , an indigenous
Gurbet aindz is impossible.
5 The past form iklilo may be taken from Arli, which has both iklisto and iklilo (cf.
Bugurdzi iilo, too), but it may as well be a process of inner-dialectal levelling, iklilo
being reshaped according to the present form ikljol. This assumption is corroborated by
Prishtina Gurbet iklilo. Both forms have passed into the passive paradigm. Since the same
change can be observed in Arli, in Bugurdzi and, as mentioned above, in Kosova Gurbet,
but neither in Kaldera nor in northern Gurbet, the southern dialects seem to have carried
out this change in a joint action. It is difficult to decide in which of the dialects the process
began.
6 Because of some typical traits (e.g. neg. i), Cerhari is considered by Mszros a Vlach
dialect, but this should not be taken for granted. It can be said with certainty that i occurs
92 NORBERT BORETZKY
not only in Vlach dialects, and that it does not go back to rum. nici. We find it in German
Sinti varieties as well; cf. Holzinger 1993: 40, 45.
If the dialect had a Vlach basis, then the copula paradigm should have been borrowed from
another dialect.
7 In Strukelj we find present som sal, but hi; sinja "was", but -s- > -h- in grammatical
suffixes; cf. zal-a-hi ipf. "went" < *dal-a-si.
8 The Romani of Wales as decribed by Sampson 1926 has om an si, not horn han hi;
this too makes us hesitant to consider Sinti hom han hi as a very old paradigm.
9 It is possible that in Romani e represents the older form, among others because we find
it in dialects that are very similar to Arli, e.g. the dialect of Prizren and that of Prilep,
where it may be a relic form. From e we get to i quite easily, by way of analogical
adaptation to the feminine -i found in nouns, adjectives and pronouns, whereas for an
adaptation in the opposite direction no plausible explanation can be found.
10 The old indigeneous comparative formant -eder survives in relic forms only.
11 We cannot exclude the possibility of sine and sas (sjas) once having existed side by
side, in complementary distribution, but as long as we have no indication in favour of this
assumption we should disregard this possibility.
12 tuj is clearly of Albanian origin. It is interesting to note that Bugurdzi did not copy the
Albanian construction in all parts: whereas Albanian does not distinguish the grammatical
persons, in Romani a personal construction evolved because of tuj being combined with the
finite present.
13 In Macedonian Arli -indor is found rather than -indoj or -indo.
14 Even the Cal of Spain reveals some traces of dialect mixture. The following doublets
are documented: piro and pindro "foot", minrio "my"andmin-daj "(my) mother" against
monro "friend", apparently < munro "my" (Kalderag); the verbal noun in -pe, -pen and -ben
(uncertain); borrowed masculine nouns in -o and -os, cf. ro/iros, foro/foros, druposltrupo;
pokin-andplasar- "to pay" (Boretzky 1992:18f).
REFERENCES
Barannikov, A.P. (1934) The Ukrainian and South Russian Gypsy Dialects.
Leningrad: Izdatel'stvo Akademii Nauk SSSR.
Bischoff, Ferdinand (1827) Deutsch-Zigeunerisches Wrterbuch. Ilmenau:
Verlag B.F. Voigt.
Boretzky, Norbert (1992) Romanisch-zigeunerische Interferenzen (zum
Cal). Erfurt & Jeing & Perl (ed.). Prinzipien des Sprachwandels. I.
Vorbereitung. In: Beitrge zum Leipziger Symposium des Projekts
"Prinzipien des Sprachwandels" (PROPRINS) vom 24.-26. 1991 an der
Universitt Leipzig. Bochum: Universittsverlag Brockmeyer. 11-37.
Duri, Rajko (1980) Prastarare- daleki svet/Purano svato - o dur them.
Beograd: Narodna Knjiga.
Duri, Rajko (1982) A i U/A thaj U. Beograd: Narodna Knjiga.
INTERDIALECTAL INTERFERENCE 93
Finck, Nikolaus von (1903) Lehrbuch des Dialekts der deutschen Zigeuner.
Marburg: Elwertsche Verlagsbuchhandlung.
Gilliat-Smith, Bernard (1914) The Dialect of the Drindaris. Journal of the
Gypsy Lore Society, New Series VII, 260-298.
Hancock, Ian (1990) A Grammar of the Hungarian-Slovak (Carpathian,
Bashaldo, Romungro) Romani Language. Manchaca, Texas: Inter
national Romani Union (USA).
Holzinger, Daniel (1993) Das Rmanes. Grammatik und Diskursanalyse der
Sprache der Sinte. Innsbruck: Institut fr Sprachwissenschaften. (Inns
brucker Beitrge zur Kulturwissenschaft 85).
Igla, Birgit (1989) Das Romani von Ajia Varvara. (Unpublished Diss., Ruhr-
University Bochum).
Kenrick, Donald S. Morphology and Lexicon of the Romani Dialect of Kotel
(Bulgaria). (Unpublished Ph.D. Thesis)
Kenrick, Donald S. (1967) The Romani Dialect of a Musician from Razgrad.
Balkansko Ezikoznanie 11/12, 71-78.
Kepeski, Krume & Jusuf, Saip (1980) Romani Gramatika - Romska Gra-
matika. Skopje: Nasa Kniga.
Knobloch, Johann (1953) Romani-Texte aus dem Burgenland. Eisenstadt.
(Burgenlndische Forschungen, hrsg. vom Landesarchiv und Landes
museum, Heft 24.)
Kopernicki, Izydor (1930) Textes tsiganes. Krakow: Naklad Polskiej Aka-
demji Umiejtnoci.
Kostov, Kiril (1962) The Vixen and Pirusambi. Journal of the Gypsy Lore
Society XLI:1-2,31-38.
Krasnii, Ali (1981) Cergarende jaga/ Cergarske vatre. Pristina: Jedinstvo.
Krasnii, Ali (1986) Iripe ano dzuvdipe/Povratak u zivot. Pritina: Jedinstvo.
Mszros, Gyula (1976) The Cerhari Gypsy Dialect. Acta Orientalia Aca-
demiae Scientiarum Hungaricae XXX:3, 351-367.
Miklosich, Franz (1874) Mrchen und Lieder der Zigeuner der Bukovina.
Mundarten und Wanderungen der Zigeuner Europas IV. In: Denkschrif
ten der phil.-hist.Cl der Kaiserlichen Akademie der Wissenschaften.
Wien. XXIII, 273-327.
Paspati, Georgios Alexandros (1870) tudes sur les Tchingians ou
Bohmiens de l'Empire Ottoman. Constantinople: Imprimrie Koromla.
Pobozniak, Tadeusz (1964) Grammar of the Lovari Dialect. Krakw:
Pastwowe Wydawnictwo Naukowe.
Sampson, John (1926) The Dialect of the Gypsies of Wales. Oxford: Univer
sity Press.
Senzera, Luigi (1986) II dialetto dei Sinti Piemontesi. Lacio Drom 22:2.
Sowa, Rudolf von (1887) Die Mundart der slovakischen Zigeuner. Gttin
gen: Vandenhoeck und Ruprecht.
Strukelj, Pavla (1980) Romi na Slovenskem. Ljubljana: Cankarjeva Zalozba.
94 NORBERT BORETZKY
Uhlik, Rade (1942) Bosnian Romani (ed. by F.G. Ackerley). Journal of the
Gypsy Lore Society XX, 100-140, XXI, 24-55 and 110-141, XXII,
305-323.
Vekerdi, J. (1984) The Vend Gypsy dialect in Hungary. Acta Linguistica
Academiae Hungaricae 34, 65-86.
Wentzel, Tatjana W. (1980) Die Zigeunersprache. (Nordrussischer Dialekt)
Leipzig: Enzyklopdie Verlag.
VERB EVIDENTIALS AND THEIR DISCOURSE FUNCTION
IN VLACH ROMANI NARRATIVES
YARON MATRAS
University of Hamburg
0. Introduction
A number of scholars have drawn attention, in a descriptive manner, to
a structural split in the formation of the simple past tense in Romani.
Miklosich (1873: 45) observes that in the third person singular, the past
participle can appear alone, without an auxiliary: "asnilo, risit".1 Similarly,
Sampson (1926: 194) notes that "beside the 3rd. sg. in -as, we find that
Gk.Gyp. and other Southern Eur. dialects use the bare participle sg. in -o
(m.) and -i (f.)". Gjerdman & Ljungberg (1963: 114) add a partial restriction
to "a preterite which is qualified by gender", stating that it tends to appear
with intransitive verbs of motion. Boretzky (1986: 205) refers to the split as
a "peculiar phenomenon", which is restricted to a number of verbs in the
third person singular, while the majority of verbs appear only in the
inflected form, carrying the personal suffix -a/-as. Hancock (1993: 48-51)
classifies the active past participle as an "irregular preterite", and suggests,
in order to avoid confusion with the adjectival or passive participle, "to
incorporate a rule into the standardized grammar that would restrict past
participles to the pre-nominal position".
What is the origin of the split in the form of the third person of simple
past tenses in Romani, and what is the motivation behind its retention with a
number of verbs? Romani is a Balkanized Indic language, a classification
justified by the fact that it shows Balkanic features on the level of sentence
organization. The split in the form of simple past tense verbs of the type
avilas "he/she came" (inflected preterite), vs. avilo "he came", avili "she
came" (active past participle), found in Romani overwhelmingly with verbs
of motion and change of state, is connected, I propose, to the process of
language convergence in the Balkans. At the same time it also reflects the
unique position occupied by Romani within the Indic languages,
96 YARON MATRAS
ma y-
I come:PART-FEM
"I (f) came"
vo y- se el-o
he/she come:PART-MAS he/she come: PART-3SG
"he came" "he/she came"
vo y-
he/she come:PART-FEM
"she came"
Romani ||
me avil-em
I come:PART-lSG
"I (m/f) came"
vov avil-o vov avil-as
he come:PART-MAS he come:PART-3SG
"he came" "he came"
with intransitive and with transitive verbs. This pattern corresponds to the
emergence of inflected forms in Romani, as shown on the bottom right-hand
side of the figure.
Thus Romani generally corresponds to the 'Outer' languages - this is in
fact the case with regard to a series of grammatical formations - but it also
preserves the more conservative active participle.
Bubenik (this volume) suggests two possible explanations for the fact
that Romans does not show ergativity: (i) Proto-Romani had shared the
ergative construction with other Indo-Aryan languages, but lost it after
coming into contact with the (non-ergative) languages of Europe; (ii) The
ancestors of the Roma left the Indian subcontinent before ergativity
ermerged in Indo-Aryan.
The co-existence of the inflected past and active participle in Romani
may be used to argue for the first option, if one were to consider the
participle avilo to reflect the old ergative type. Once ergativity was lost, the
language retained active participles that agreed with a beneficiary which
was also the subject. 'Semantic ergativity' would have thus been allowed to
survive within a nominative-accusative syntactic system. This would
explain why the active participle in Romani is restricted to verbs denoting a
change of state: with such verbs it is the syntactic subject, and not the
object, on which the outcome of the process may be observed.
Alternatively, one could maintain that the participle is pre-ergative,
representing yet an older stage in the development toward ergativity. This
development began with the gradual spread in Old Indo-Aryan of the
intransitive participle, leading ultimately to a generalization of the transitive
participle as well, with the participial forms finally replacing the inflected
past tense in late Middle Indo-Aryan (cf. Chatterji 1926: 938-940). The
stage we witness in Romani would then be that of an active participle which
had not yet established itself with transitive verbs.
Either way, Romani, a nominative-accusative language, still shows
traces of the ergative development in Indo-Aryan by possessing the active
participle alongside the inflected forms.
In inheriting two forms for the past tense, one (the adjectival past
participle) denoting a state which is the result of an action, the other (the
inflected preterite) referring to the action itself and its agent, early Romani
was not unique among the languages of the Balkans. Friedman (1986: 179,
184ff.) describes the development of the Common Slavic aorist and
imperfect (inflected past) into the Balkan Slavic (Bulgarian and
EVIDENTIALS IN VLACH NARRATIVES 99
In (3), a passage from a biographical narrative, the fact that one of the
family members was born in the Czech Republic, and in prison, is presented
as a curious detail in the family history. The participle is chosen here to
convey an exceptional and surprising state of affairs. The element of sur
prise in segments a.-b. is transmitted by the speaker's presentation of the
facts as being based on circumstantial evidence. On the other hand, the pre
sentation of the same event in segment d. relies on the preceding back
ground information. Here the speaker is drawing a conclusion on the basis
of detailed proof, and it is here that a switch into the inflected preterite takes
place.
While in examples (l)-(2) the distribution of the inflected preterite and
the active participle is connected to the difference between expertise
knowledge and actual situative evidence, in (3) the opposition is used to
express the contrast between virtual evidence for an unexpected state of
affairs and well-established conclusions. The term 'evidential', if used in a
pragmatic, rather than a structural sense, seems to me to capture the in
variant meaning of the form: It is used to approach an event via circum
stantial evidence rather than from explicit, prepared knowledge.
Friedman (1986: 185) rejects the term 'evidential' for the non-con
firmative in Balkan Slavic and Albanian on the grounds that it does not
mark the source of information, but the speaker's attitude toward it. In this
respect Balkan Slavic and Albanian differ from Turkish, where the
possibilities for choosing a form are much more restricted. With respect to
the opposition in Romani, 'source' and 'attitude' do not seem contradictory to
me. As seen in (3), it is the speaker's 'attitude' that motivates the choice of
the participle, which marks the event as inferred from secondary evidence.
The same event can appear again in the inflected preterite after the
background for the presentation has been prepared in advance.
As argued by Givn (1982), evidentiality does not relate to the actual
truth of the proposition, but to the 'contract' drawn between speaker and
hearer in the communicative transaction. Choosing the adequate form for
the simple past tense is a strategy applied in order to ensure that the
assertion is accepted by the hearer. When the information conveyed is
unexpected or exceptional, and the speaker's assertive authority is at stake,
EVIDENTIALS IN VLACH NARRATIVES 103
the speaker can resort to the evidential form, thereby disclaiming direct
responsibility for the report on the actual underlying event. Rather than
mark the speaker's plain 'attitude' or 'source of information', evidentiality is
related to the speaker's monitoring of the hearer's preparedness to process
propositions and accept assertions.
The following pair of discourse excerpts provide another example:
(4) a. Aj po dujto var po kaver kurko gelem pale,
and on second time on other week I-went again
aj pale sikhadem o vast,
and again I-showed the hand
b. Ta avilas kothe baro reisoro sas aj pinzardas
and came:3SG there big director was and recognized
ma ke simas aba jokvar, hacares?
me because I-was already once you-understand
c. Aj phendas mange ke kamin von te
and said to-me that they-want they that
khelav filmo.
I-play film (NL/1/104-106)
In (4) the speaker is telling about her adventures during a visit to Uni
versal Studios in Los Angeles. The type of narrative can be classified as a
'report' (cf. Rehbein 1984: 89ff.): She reports on the course of reconstructed
events which she has not only witnessed, but already processed, evaluated
and categorized, and she does so by presenting them in a planned and
serialized sequence. Throughout the entire excerpt the inflected preterite is
used, including the form of the intransitive verb of motion avilas "he came".
By contrast, in (5) the speaker is re-experiencing the original situation.
Events are described in the present tense, as if speaker and hearer were
witnessing them in their original scene. In segment d. the participants are
taken by surprise by the appearance of Gurano, indicated by the active
participle avilo "he came". The state of arrival is first captured on the basis
of perceived situative evidence, from which the underlying event is then
inferred. The linear presentation of events is not pre-planned, but is
constantly re-organized as the hearer is guided through various pictures of
the scene. Connectiveness is achieved through attention markers, such as
akana "now" (segments d. and e.) and staging questions (so avel?, "what's
coming?", in segment b., and kon avel:, "who arrives:", in e.), whereas in
(4) it is marked by simple additive conjunctions, the hearer's attention
having already been drawn to an entire chain of assertions.
Comparing the simple past tenses used in (4) and (5), we see that the
opposition inflected preterite/active participle is connected to the organi
zation of knowledge in discourse. The participle is used to mark information
which is not pre-structured and where the speaker needs to resort to circum
stantial evidence in order to maintain assertive authority.
3. Virtual evidence
Aksu-Ko & Slobin (1986) regard the inferential procedure as in
volving sensory evidence. In (3) we saw, however, that the evidential form
can also occur when the event in question is detached from the sensory
EVIDENTIALS IN VLACH NARRATIVES 105
In segment b., the speaker uses the participle or evidential when talking
about the unexpected fact that her father, having had no schooling what
soever, a) learned to write, and b) acquired his skills in solitary confinement
in prison. Once again the participle is used to mark an exceptional, sur
prising event. However, the statement conveyed by the participle is not
based on sensory evidence. In choosing the participle, the speaker avoids re
constructing the actual action of learning, and instead presents the accom
plished fact.9 Rather than assume direct responsibility for reporting on an
unusual event, the speaker concentrates on the results of the event, evidence
for which is less disputable and virtually accessible, as elaborated on in seg
ment c .
In (7) the speaker in a political lecture reports on one of the key persons
responsible for the Nazi-genocide on Gypsies, Robert Ritter, who escaped
trial and conducted a normal working life after the war. The evidential form
in mulo, "died" can be interpreted in this particular context as an expression
of irony, the speaker disconnecting himself and his personal conviction
from the fate of the criminal. This contextual meaning is reinforced by the
reflexive peske, literally "to himself", implying that Ritter had lived and
died at his own discretion, without anybody disturbing his peace.
The ironic interpretation of the verb itself is connected to the inference
procedure triggered by the participle: the speaker was not involved in the
actual event, but was only able to witness its outcome (not by being present
or actually seing Ritter's dead body, but by receiving the relevant in
formation). He is therefore not to be held responsible for the provoking
aspects of the event, i.e. for the fact that Ritter died in peace, and not, as he
would have deserved, as a criminal convicted of genocide.
Again, in choosing the evidential, the speaker is monitoring hearer
participation and acting in anticipation of the hearer's possible reactions. He
does so on the basis of a set of common values and attitudes, which allow
him to foresee some of the hearer's own judgements.
The use of the Romani evidential in the preceding examples bears
certain similarities to specific meanings conveyed by the Balkan Slavic and
Albanian non-confirmative, as described by Friedman (1986), such as
unexpectedness, surprise, irony, or reportedness. Friedman regards non-
confirmation itself only as one of the possible contextual meanings of the
indefinite past. Non-confirmation cannot be seen as the invariant function of
the Romani participle either. Examples (2), (3) and (5) do not support a
non-confirmative interpretation since the speaker is well aware of the results
EVIDENTIALS IN VLACH NARRATIVES 107
of the event. In (6) and (7) uncertainty may be connected to certain details,
but again the actual occurence is not denied.
In its core the form seems to convey a disclaim of direct knowledge
about the internal course of an event, but does not question the event itself.
However, knowledge about the event is restricted to those pieces of
information which the speaker is able to derive from evidence for the ac
complished result that is available to him. Confirmation of the event
therefore relies on a secondary processing procedure, applied to 'primary'
elements of knowledge pertaining to the result.
With the inflected preterite, the propositional content reflects the
speaker's direct knowledge of a completed event. From this knowledge, the
speaker derives his recognition that the event in question has produced a
result, which may have some bearing on the current speech situation.
Knowledge concerning this result is conveyed implicitly by the predicate
acting in the proposition.
In the evidential, direct knowledge concentrates on the result. Know
ledge about the internal structure of the event resurfaces only as a 'by
product'; it is inferred through secondary processing of the circumstantial
evidence provided by the relevant state of affairs. The reconstruction of an
event by means of the evidential form is therefore epistemic: it requires
further, non-verbal development of existing knowledge.
This opposition between different tasks for processing knowledge
figures in the structural formation of the verb: the inflected preterite is
marked for person; it portrays the direct involvement of an agent in the
action or event. This necessarily implies that the speaker has some
knowledge about the actor, and so about the internal course of the event.
The evidential, however, is only marked for gender. Its adjectival formation
allows a detachment of the perceived result from the actor responsible for
its emergence. Rather than name the responsible actor, the gender suffix
helps locate the referent on which the outcome of the event is detectable.
4. Figurative evidence
So far, two types of evidence were discussed which allow inference
about an underlying process, event, or action, as expressed by the active
participle: actual evidence that is part of the sensory dimension of the
speech situation, and virtual evidence which is embodied within the scope
of accessible, general knowledge. Typical of reconstructive narratives is,
following Rehbein (1989: 166ff.), the transposition of the center of attention
108 YARON MATRAS
5. Contextual evidence
The following examples document how evidentials can be used to mark
a turning point in the discourse by combining the 'surprise-effect' - unpre-
paredness for a new piece of information - with clues pertaining to the new
event which have already been established as background information. The
speaker resorts to this background information while performing the
inference procedure, thus sharing with the hearer a common basis of
contextual evidence:
a. All the most important doctors lived there, such eh/ Jews.
b. And when eh/ the war, you know, they deported them all.
c. And there remained a house and we moved in there.
And so, when she gave it to her, she came home, the king
heard about this.
le ande spital.( )
him in hospital
b. Kade kana sastilo, sastilo, line
so when recovered:PART recovered:PART they-took
taj/ von taj nasle duj ene
and they and they-escaped two people
ando Cexo.
in Czech (Mri/1/53-55)
a. And he burned his veins, all these, and they brought him to
the hospital.( )
b. So when he recovered, he recovered, they got going and/ and
the two of them escaped to the Czech Republic.
From that you will have already understood that before the
year thirty three, when Hitler came, nationalism already
prevailed in Germany.
a. And then,
b. UncleDz/Vuncle Roko took her.
c. Roko took her.
d. Right,
e. And then when he came out, Roko was in a
psychiatric hospital, they sent him, your father,
f. Yes.
g. ... after he had beaten you,
h. Yes.
i. He stayed there half a year, and came out.
j. Aha,
k. And he took his wife ( ), he took his wife and left.
1. He left/ to travel.
m. And he wen/ he went to this Iva.
n. To this Iva.
o. And Iva had a boss, Alexander.
p. Iva was talking about getting married ( ), you know.
q. That Alexander...
r. And that one/ and him they deported him to Romania, the
poor guy, and when he came...
s. They gave him an Abschiebung' [=expulsion],
t. Yes, and when he came to Romania...
u. ... because he was there with forged documents.
v. Yes. And when he came to Romania this Bego helped him
out.
w. But he didn't live much longer.
x. Anyway, we heard this story, what Rok/
y. This Iva hid for more than five years.
z. When he hear/ when he heard where we were, he ran away.
How are the evidentials distributed among these positions in the dis
course excerpt? Quite simply: there are two types of evidentials. The first
appears in the recapitulating phase, in segments a.-o., and is used to mark
all verbs of motion or change of state that appear in the summarized story:
anklisto, "went out", asilo, "remained", gelotar, "left". The story itself is
thereby promoted to an accomplished result of the underlying narrative; it is
captured in its entireness as a completed fact. Rather than be developed
progressively from an internal or involved perspective, it is rediscovered
from an external, passive perspective, having already reached its terminal
point.
The second type of evidentials appears in segment r. (avilo,"came"),
and is repeated in segments t. and v.. This type corresponds to the
evidentials discussed in the previous sections, i.e. to those used within a
complex sentential frame for planning the discourse: it appears in a
118 YARON MATRAS
8. Conclusion
Using comparative typology and discourse pragmatics in a combined
approach, I attempted to illustrate the function of a grammatical category in
Romani which has until now been treated as an 'outcast' by most structural
descriptivists of the language. To summarize: the active participle conveys
the speaker's restricted knowledge of the internal course of events, and
concentrates instead on their outcome. The underlying event is approached
via external circumstantial evidence for its occurence, such as the state
resulting from it.
Although evidentials can superficially be interpreted as an indication of
the speaker's access to or attitude toward knowledge, when taken in their
actual communicative context they turn out to appear when the speaker
needs to resort to circumstantial evidence in order to maintain assertive
authority, for instance in anticipation of the hearer's disbelief or criticism.
Operating within the basic functional scope of the form, the speaker is
therefore not indicating doubt, but rather consolidating interactional author
ity in assertions the propositional content of which is, as described by Givn
(1982: 24), "open to challenge by the hearer and thus requires - or admits -
evidentiary justification".
As seen in a number of examples, notably in (3), (5) and in virtually all
the examples in sections 6 and 7, such functions may be typical of specific
positions in interactional patterns, and therefore contribute to the organi-
EVIDENTIALS IN VLACH NARRATIVES 119
verbs of motion, while inflected forms are used with all other verbs. This
could to be taken as a reflection of an older stage of the language, before
evidentiality emerged, but it would indeed be difficult to explain why
Balkan convergent development 'skipped' such varieties.
Furthermore, it is not clear how exactly contact could have contributed
to the emergence of evidentials in the Vlach variety, since Vlach Romani of
the type discussed in this paper has had its most intensive contacts in the
Balkans with Greek and Romanian, both languages that lack evidentiality. It
is possible to counter this doubt with common sense, figuring that there
must have been contact at least with Balkan Slavic before reaching the
Romanian territories. But here the evidence presented by Kostov (1973) on
borrowed evidentials in Bulgarian Romani provide a second challenge,
since it is structural borrowing that is involved, and not functionalization of
the inherited stock of forms.
However the counter-arguments are pursued, I believe enough evidence
was presented in this paper to show that the active participle in Vlach
Romani is a functional category, and as such bears a certain typological
affinity to the 'evidential' categories in Balkan Slavic, Albanian and
Turkish, and that Vlach Romani in this respect shares an areal feature with a
number of historically contiguous languages. Given that documentation on
the early period of Romani which could be relevant to our discussion is
non-existent, it is not surprising that a diachronic hypothesis cannot succeed
in illuminating all points in a fully satisfactory manner.
Appendix
List of the evidential forms found in the Hamburg Romani Corpus
NOTES
1 Miklosich interpreted the personal endings of the inflected past tense as auxiliaries,
derived from the present copula form (cf. discussion in Bubenik, this volume).
2 It was Rdiger, the pioneer of Romani linguistics, who first suggested that Romani
was using inherited Indic forms to "copy" the structure of a European contact language
(Rdiger 1782: 71). Miklosich concluded from his survey of Romani dialects that the
primary contact language for all Romani dialects was Greek, and that a Greek-speaking
area had been the early European homeland of the Roma (Miklosich 1873: 4). Balkanic
features of Romani syntax are discussed by Kostov (1973: 105ff.), Friedman (1985),
Boretzky (1993: 98), and Matras (1994; in print).
3 In their editors' introduction, Chafe & Nichols (1986) regard evidentially as a gram
matical device used by speakers to convey the source and reliability of their knowledge.
Likewise, Anderson (1986) defines evidentials as a justification of a factual claim. Typical
meanings associated with evidentials are hearsay and inference from an observed result (cf.
Haarmann 1970; Aksu-Ko & Slobin 1986). For a discussion of definitions and types of
evidentiality cf. Willet (1988).
4 Exceptions are animate as well as definite direct objects, which take the accusative
suffix -ko. Here the verb assumes a 'neutral' form.
5 According to Kostov (1973: 107-108), the Bulgarian participial suffix -/ is borrowed
into the Romani dialect of S liven, where it attaches to the verb, as in Bulgarian, in various
tenses (imperfect, preterite, anterior future) to indicate hearsay or indirect evidence.
6 Perfect tense forms are generally an attractive source for forming evidentials in lan
guage since their resultative meanings may be drawn upon to focus on the result, rather
than the event itself (cf. discussion in Willet 1988; Haarmann 1970).
7 The data were collected as part of a larger study on Romani Grammar and Discourse
(see Matras 1994). I wish to acknowledge a University of Hamburg grant and support from
the German Research Foundation, provided through the Program on Multilingualism and
Language Contact at the University of Hamburg, which made it possible for me to collect
and analyze the data.
8 Anderson (1986: 277) even claims that evidentials are rarely used when the fact
reported on is directly observable by both speaker and hearer. This has been mentioned by
Givn (1982), who regards "deictic obviousness" as high on the certainty scale of pro
positions where evidentiality is not required.
122 YARON MATRAS
9 sich-, "to learn", is a transitive verb. Nevertheless it involves a change of state, since
the outcome of the process can be observed on the subject, a learned person.
10
See Rehbein (1984, 1989) and Ehlich & Rehbein (1986) for a theory of repetitive
discourse patterns.
REFERENCES
PETER BAKKER
University of Amsterdam
0. Introduction
This paper deals with the different varieties of Romani spoken in or
originating in the Iberian peninsula, especially Cal (Spanish Romani).
Iberian Romani is taken as the collective name for the varieties of Romani
spoken on the peninsula. These are now all reported (perhaps unjustly) to be
extinct or close to extinct. The argument presented here is therefore based on
publications rather than fieldwork. Virtually all the material published shows
the complete loss of the inherited grammatical system and an adoption of the
grammatical system of the host region.
Below, I present data which may help uncover the genesis of Cal. First,
I discuss the mixed Romani dialects in general (section 1) and some historical
factors related to the Gypsies in the Iberian peninsula (section 2). Then, I
briefly discuss the place of Iberian Romani among Romani dialects (section
3). Next, I discuss some structural features of Cal, in order to assess the
nature of the mixture in Cal (section 4), and provide some data on the other
Para-Romani dialects of the Iberian peninsula (section 5). After that I briefly
mention the social functions of Cal (section 6). Furthermore, I discuss the
oldest sources of Iberian Romani in order to assess the possible origin
(section 7). With these historical, historical-linguistic, structural and clas-
sificatory facts in mind, a hypothesis is formulated concerning the genesis of
the Iberian Para-Romani dialects, in particular Cal (section 8).
As yet, there has been no attempt to explain the genesis of Cal itself,
save in some more general overviews. These were based only on a limited
number of sources.
126 PETER BAKKER
1. Para-Romani languages
Until today, about ten varieties of Romani have been identified which
have a Romani lexicon but which have lost the native grammatical system.
Instead, these dialects use the morphosyntax of the surrounding languages
(see Bakker & Van Der Voort 1991 and especially Boretzky & Igla 1994).
These are called Para-Romani languages, a term originally coined by Marcel
Cortiade. They roughly display the following characteristics: the vocabulary
is Romani (hence basically Indic), but nearly all the phonology, morphology
and syntax are non-Romani. Although all Romani dialects borrow heavily
from the languages of the host country, no cases are reported of languages
with a Romani grammatical system and a non-Romani lexicon. Until now,
Para-Romani languages have been described, documented or mentioned in
the literature in connection with the grammatical systems of Swedish, Nor
wegian, German, Catalan, Portuguese, English, Basque, Spanish, Greek,
Persian, Turkish, and Armenian. All these varieties must be seen as lan
guages in their own right, and not as dialects of Swedish etc. or Romani.
Several theories have been put forward to account for the genesis of these
dialects, ranging from saving a dying language by preserving the lexicon,
through gradual massive grammatical borrowing, the conscious creation of a
mixed language, relexification to language intertwining - the combination of
one lexicon with the morphosyntax of another language. Not all of these are
mutually exclusive, but the debate on their genesis is still going on.
The Para-Romani language of the Spanish part of the Iberian peninsula is
relatively well documented. Both speakers and outsiders have written down
and published vocabularies or grammatical studies of this language. Almost
all of this, however, is in languages other than English. This study will in
troduce this language to an English speaking audience.
The language is generally called Cal [kalo] in the literature. Un
doubtedly the Romani word kalo "black" is the source. It is used both as an
ethnic self-designation and as a name for the language. Speakers also call it
Romano (Quindal 1867: 49), a nominalized adjective derived from the noun
Rom 'Gypsy, man' with the Romani suffix -ani/-ano.
Cal as a Para-Romani language should not be confused with other
languages and slang varieties which are also called Cal. All these languages
have in common that they are cryptolectal or in-group languages embedded in
Spanish structures. This name for non-Romani languages is reported from
places as far as California (Polkinhorn et al. 1983). Although such crypto
lectal languages may have borrowed some words from Romani or Gypsy
GENESIS OF CALO 127
Cal, they will not be discussed here. When we mention Cal in this paper, it
refers exclusively to the Gypsy language of that name, as spoken by people
of Gypsy descent in Spain, Portugal and South America.
As a Para-Romani language, Cal is particularly interesting for three main
reasons. First, there is a lot of material, covering two or perhaps even three
centuries. There is no other Para-Romani dialect documented over such a
long period of time.
Second, in contrast to most other Para-Romani languages, Cal also
makes abundant use of cryptic devices, apparently meant to conceal the
meaning of the communication from outsiders.
Third, apart from Cal, which has an (Andalusian) Spanish grammatical
system, there appear to be several other varieties of Para-Romani languages
which came about under contact with languages of the Iberian peninsula, viz.
one with a Catalan grammatical system, one with a Portuguese grammatical
system (in Brazil), and one with the non-Romance language Basque. These
varieties all seem to be derived from one branch of Romani as they all share
some lexical particularities apparently not shared with other varieties of
Romani (see section 4).
4.2. Phonology
The phonology of Cal is Spanish; all Romani words are adapted to the
phonology of Andalusian Spanish. This means that Romani phonemes not
existing in Spanish are lost in Cal. For instance, in Cal there are no
aspirated stops. Romani aspirated stops become unaspirated stops in Cal
(except /th/ which becomes //). It is clear, however, that the aspirated stops
were still used when the Gypsies arrived on the Iberian peninsula. The
130 PETER BAKKER
aspirated /th/ of Romani became // rather than /t/, for instance chem /em/
'land' < Romani them, chute /cute/ Romani thud 'milk' (Boretzky 1992).
Further, the voiced affricate /d/ became // in Cal and NI became /b/, etc.,
following Spanish phonology.
One of the features of Cal inherited from Romani is its stress pattern, or
at least some aspects of it. Whereas Spanish words rarely have stress on the
final syllable, this is commonly so in Romani and Cal (inherited from
Romani). See for instances the stress markers in example (1) below. This
may give Cal a distinct flavour.
Phonotactic constraints seem, at first sight, to be the same in Cal and
Spanish, in that the syllable structure is identical. However, there are ex
ceptions as well: in Cal there are words ending in -m, but these do not seem
to exist in Spanish. A recent study suggests that the phonotactic constraints of
Spanish and Cal are not completely identical. This thought-provoking study
by Dietz and Mulcahy (1988) compares the combination of letters in a Bible
chapter in Castilian and Cal. They conclude on the basis of statistical
evidence "that Cal and Castilian differ greatly in the way they form and
distribute basic linguistic particles" (i.e. 'letters' or perhaps, by implication,
phonemes). For example, in the samples of the same text in Spanish and Cal
versions, the phoneme // (the digraph "ch") was counted 2.056 times in the
Cal text and only 242 times in the Spanish text, a difference of nearly 850 %
(Mulcahy & Volland 1986: 144 n. 4).
How can we explain this? There are several possibilities. With very few
exceptions, Cal phonemic structures are possible in Castilian. This is to be
expected, since the Romani words are adapted to Spanish phonology. It may
be, however, that the frequency of certain combinations of phonemes differ
in both languages because the lexicon of Cal remains basically Romani and
follows some non-Castilian features of Romani. For instance, Romani /c/,
//, /h/ and /th/ all became // in Cal, which could explain the high pro
portion of //. In fact, this is what the results of Dietz & Mulcahy's study
suggest.
Whatever the explanation is, we agree that Cal cannot simply be a
'Gypsified' version of Castilian (Dietz & Mulcahy 1988: 15).
4.3. Morphology
Cal uses Spanish derivational affixes, such as the diminutives -ico, -ito,
the superlative -isimo, the adverbial ending -mente, etc.. Apart from that, it
GENESIS OF CALO 131
4.4. Lexicon
McLane (1977, 1985) recorded a few hundred words considered Cal by
the Gypsies of Guadix and other parts of Andalusia and Spain. He calculated
(1985: 188) that 68 % of the lexical items in his corpus are of Indic (hence
Romani) origin and 22 % of Spanish origin (many of those distorted). We
also find the usual sources of non-Indic words in Cal from the pre-European
layer of loan elements present in all Romani dialects, such as Armenian,
Persian, Greek and Slavic languages. Boretzky (1992), using a wider
vocabulary, detected words from Slavic languages, in some cases specifiable
as South Slavic, Czech or Polish (Cal dosta "enough"< South Slavic dosta;
Cal kornes "laced boot"< Czech skornje; Cal stajnia "horse-stable" < Po
lish estanja). There are also a few Hungarian words in Catalonian Romani,
such as arany "silver" < Hungarian aranj and cin "colour" < Hungarian szn.
Many of these words, however, are not specific to Cal; in fact they can be
found in other Romani dialects as well. Boretzky further identified a number
of words which Cal has in common with the Sinti and Angloromani dialects
(Boretzky 1992: 16), so called 'northern dialects'. Further there are words
taken from Germania (Spanish Cant) in Cal (see below 6.3).
It should be mentioned that Iberian Romani also seems to have Greek
(and perhaps Persian) words not attested in other Romani dialects. A few
others have been identified as Arabic (e.g. (j)azari "ten" < Arabic ?ara;
Moroccan variety). Overall, the Cal lexicon is highly aberrant among
Romani dialects because of the singular phonological development, the
cryptic devices used (see below) and the relatively large number of words of
unknown origin.
4.6. Innovations
Although there is a clearRomanicomponent and a clear Andalusian Spa
nish component, there are also elements which are neither. I focus on two of
these: the use of innovative place names and the use of cryptolalic forma
tions.
Some of these are clearly circumscriptions, such as len bar "big river"
for Guadalquivir, and chim ye manr '"and of the bread" for Extremadura.
Others are distortions of the existing place names, such as madrilati for
Madrid and llundun for London. Some of them are cryptolalic formations,
such as Boban for Habana. It is a pun on Spanish haba "bean" and Cal bobi
"bean": one part of the word is replaced by the Romani word with the same
meaning (see below for other distortions of this type).
This indicates that speakers intended to make their speech unintelligible to
outsiders by avoiding borrowing, and instead making up new words. We
find the same phenomenon with place names in Angloromani (Hancock
134 PETER BAKKER
illustrative. The text and its Catalan translation are both from Vallmitjana
1908 (as cited in Leblon 1982: 63). The analysis and interpretation are mine2.
Interpretation:
The Romani-derived words are mutzi (< R. morthi "skin") dical (<R.
diklo "neckkerchief"), nyisquerba (<R. (n)ikalav "to take off), busn "non-
Gypsy" (<R. buzno "goat"), endiquelava (<R. dikhel "he sees"), panant
(<R. penel "he says"). The words aranu and sambanban are of unknown
origin.
The Gypsies in Portugal are reported to have spoken (or speak) a variety
close to Cal, though slightly influenced by Portuguese (Coelho 1892).
These Gypsies must have come to Portugal from Andalusia. In the following
example from Coelho (1892: 9), the Calao example has Spanish rather than
Portuguese function morphemes such as the verbal inflection and the
possessive pronoun mi vs. meu, the personal pronoun yo vs. eu and the
question word quien vs. quem:
Interpretation
Here the content words are also Romani: d (< R. dai), merinhaste (< R.
merel "he dies"), mana (< R. manca "with me"), nachadipem (< R. nasul
"evil" or nasvalo "ill", plus abstract noun suffix -pe(n) ),mena (< R. manca
"with me"), jalou (<R. dial "he goes"). The -inh- suffix is unclear. It may be
related to the -in- element found in some words in Cal between the
(Spanish-derived) stem and the inflection, which is also used in other Rom
37-38).
It is unlikely (but not at all impossible) that the people who had already
lost the Romani grammatical system would adopt a different grammatical
system when moving to a different region. The Portuguese Gypsies in any
case preserved the Spanish grammatical system with no shift to Portuguese.
The fact that the pronouns are taken from different inflected forms, also
points to an independent development: once lost, the other pronouns could
not have been recovered. It is further apparent that the forms selected for the
root of the verbs (imperative or first or third person singular) are not parallel
in the different languages (although there is not enough data from Catalonia).
This strongly suggests that these dialects came about independently -
except for the Portuguese Para-Romani, which is an extension of Andalusian
Cal.
tuai borrowing and it is often hard to keep the two languages apart. This is
discussed in (6.3).
This sounds already very similar to what McLane wrote almost a century
later, when he stated that the language is "in its final step towards extinction"
(McLane 1977: 303). It therefore seems that this language has been
languishing for a long time. It is dying because the words are less and less
used and increasingly forgotten and replaced by Spanish words.
GENESIS OF CAL 139
ENG SPA ROM 1818 1841 1848 1853 1870 1900 1915
BR. BORR. CAMP. JIM. SALES TINEO PABANO
All the functional elements are of Romani or Spanish origin except for the
italicized forms somia for "for" and nu for first person singular, both of
which are of unknown origin.
One result is clear: the sources show no evolution from a more Romani-
oriented variety towards a more Spanish-oriented variety. The first text
(Bright 1818) has almost exclusively Spanish function words, whereas later
texts have more function words from Romani. In fact, the number of Ro-
mani-derived function words used in sources after 1850 is greater than before
1850, contrary to what one would expect, if Cal is seen as a gradual
evolution away from Romani towards Spanish. This could result from
regional differences, however, or increasing purism.
For the final analysis, all sources should be studied, and in a more
thorough way than was possible within the scope of this paper.
dates from the 17th century, but he does not indicate how he arrived at this
conclusion.
The list contains 61 words, almost all of them clearly Romani. Although
there are no phrases in the Hst, it is also clear that the Romani inflection is not
inherited. Indeed the words have inflection: verbs end in -ar, and plural
nouns end in -s, both in the Spanish and in the Cal columns. There are also
some shared derivational endings in both sections, such as -ador (or rather its
Andalusian form -aor) for a person who performs an action. Although there
are some Romani grammatical morphemes, these are most probably lexi-
calized forms, such as gach and gach for "servant" and "maid", with the
Romani masculine and feminine endings. The -o and -i endings are pro
ductive in Romani proper for adjectives, but not for nouns and we only find
them used on nouns in this list. The same for pux and puxy for "old man"
and "old woman" respectively. The x is probably a misreading for r, since the
Romani source word is the adjective puro (M), puri (F). If this list is really
from the 17th century, it means that without any doubt there were already
completely hispanicized varieties of Romani by then.
7.2. Bright
The first dated publication concerning Cal, and the first one which con
tains sentences, is Bright's European travel account. In an appendix, he
compares the Gypsy languages of Hungary, Britain and Spain (Bright 1818:
lxxviii-xcii). Bright gives roughly 150 words and some 25 sentences. In
view of the fact that Bright's book is so hard to obtain, his text material is
presented in an appendix to this paper.
Bright did not collect the material himself, but he received it from one of
his friends (Bright 1818: ix). In the word list many plural nouns end in -s and
verbs in -ar. More important, he gives a number of sentences which clearly
show that it is Cal we are dealing with and not Romani with inherited
inflection.
The proportion of Spanish and Romani lexical elements used here differs
considerably from one sentence to another, as we see comparing (1) and (2),
in which Spanish elements are italicized.
(1) ochanaba mangue loque chik (Bright, Cal)
know-3 1SG that-which tell-2SG
"I know not what you tell me" (Bright, English)
(Romani dzanav "I know", man-ge "to me", chile ??;
perhaps dlav 'I sing'?)
GENESIS OF CAL 143
In short, there is no doubt that in 1818 Cal was already the mixed lan
guage known from later sources. It even has the Romani dative form used in
all grammatical cases, as in the later sources (see also section 5).
of the early sources. It may have been a conscious creation, perhaps related to
an attempt at the reversal of language shift.
9. Conclusions
On the Iberian peninsula and Latin America, at least four Para-Romani
languages have come into being, most likely independently of one another.
Cal combines Iberian Romani vocabulary with Andalusian Spanish
grammar, Errumantxela in the Basque Country combines Iberian Romani
vocabulary with Basque grammar, around Barcelona an unnamed variety
came into being combining Catalan grammar and Iberian Romani, and in
Brazil a Portuguese grammar variety came into being. The Para-Romani
variety of Portugal is derived from Spanish Cal and is called Calo. An early
genesis for Cal is suggested by some archaic Spanish remnants, pointing to
conservatism on the part of its speakers. Perhaps the language already existed
in the 16th century, several generations after the arrival in Spain. Regional
differences remained, some being closer to Spanish, others to Romani, as far
as the use of function words is concerned. Only in Catalonia the inflected
language survived into the 19th century, apparently beside a Para-Romani
variety.
Further research on Cal is needed on a number of subjects. An
etymological dictionary is badly needed, as well as a critical assessment of the
sources, including the question who took over what from whom. Moreover,
a thorough grammatical study is also needed, whereby both the Spanish and
Romani source dialects should be taken into account.
NOTES
1 Boretzky (p.c.1993) has observed that in some Balkan dialects (notably Arli) both -har
and -ha are used beside one another, and the Gurbet dialect of Romani has the suffix -sar,
and he thinks that the preposition sa(r) and the case ending -sa(r) existed side by side for a
longer time (as with most of the other prepositions/ case endings). Cal shows traces of
this.
2
I thank M. Lpez Abelln for his help with Catalan.
3 Paban wrote: "decia que era un habla inventada por ellos para suplir su idioma nativo,
que se les haba olvidado" (Paban 1915: 179). In a superficial search in Delrio's volumi
nous work, I was not able to locate the exact page for this quotation.
146 PETER BAKKER
Appendix 1
THE CAL SENTENCES FROM BR (1818).
Spanish-derived elements are italicized.
Gitano Song:
Del estaribel me sacan They take me from the prison
Montadito en un jun mounted on an ass
Yme van acurrubando and flog me
GENESIS OF CAL 147
Utterance:
Chavo gillate. Be off boy.
que vienen los Doms a cogerte The officers are coming to take you
Date con los carcos en el Buerengi Give your shoes against your breech
Appendix 2
The two texts below were taken from Cnac Moncaut (1855: 345), whose informant was
somebody named Sansberro. Sansberro is one of the Cascarots in Ziburu (Ciboure), a
fisherman's village on the northern Basque coast, close to Donibane Lohitzun (Saint Jean de
Luz). The Cascarots are said to be descendants of marriages of Gypsies with Basques (see
Webster 1889). The first text is a Catholic prayer.
This short text contains a few words typical of Iberian Romani: bato for "father" (Romani
dad. The word apelinguet for "Holy Spirit" is unclear, but it may have to do with penice
(written peniche), the Cal form for 'Holy Spirit', presumably derived from Greek pneuma.
The interpretation of the second text is more complex, since the translation does not seem
to fit the text and it contains some words which are hard not clear.
REFERENCES
CORINNA LESCHBER
Free University, Berlin
0. Introduction
The subject of Romani borrowings in Romanian generally touches upon
different areas of past and present Romanian colloquial speech. It concerns
the language of the youth (school children and students) as well as that of
soldiers, historical argot and contemporary slang, so-called vulgar language,
and the language of newspapers in Romania from the end of the 1800's and
into the 1930's. In this contribution I deal with the adaptation of Romani
words into Romanian and their semantic developments and stylistic
changes. I focus on the question which Romanian words originating from
Romani are still in use today, by whom, and with which semantic content,
and attempt to find out whether the users of these borrowings are conscious
of their origin, and whether the words they use are applied only in certain
circumstances.
characteristic effect through the use of certain items and phonetic traits that
suggest to the reader a Gypsy context.
There have been incidents ofRomanianjournalists introducing artificial
imitations of Romani to make their newspaper articles more interesting.
They often employed aspiration of initial vowels, additional suffixes such as
-os, -ete, -engher, and verse form for these artificial passages. Around the
beginning of the twentieth century one finds such artificially created
Romani texts and Gypsy anecdotes in cabaret and revues in Veselia. From
1922-28 a certain Pribeagu worked for the weekly magazine Pardon and
produced fake Gypsy texts, although he spoke no Romani. These texts later
appeared in the newspaper Dimineaa(Morning) on the "funny page".
Assuming that the material Graur presents is from authentic sources, it
is apparent, when scrutinizing a word with the help of any dictionary, in this
case the three volume Romanian-German dictionary by Tiktin & Miron
(1986-1990), that Romani etymology is usually not recognized, and words
of Romani origin are traced to other languages. Schroeder (1989) dedicates
a total of eleven and a half lines to the subject of "Gypsy words". According
to Schroeder, the lexemes found in the Romanian language which derive
from Romani are seldom neutral, but for the most part are depreciatory and
vulgar.
An important contribution to the study of Romanian argot was written
by the Frenchman Juilland, who pointed to semantic changes which words
of Romani origin have undergone (Juilland 1953: 433), as well as to the
discussion aroused by Vasiliu (1933/34a-b) with regard to a number of
Romani etymologies mentioned by Graur (cf. also Graur 1937).
In the post-1947 era, the subject of Romani etymologies was taboo in
Romania. This is exemplified by Graur (1960: 368 ff), who complained that
"certain classes are changing the language as they like", and characterized
this process as "a tasteless development of the language, much like Jargons
and Argots" (p. 373), concluding that "these useless varieties are to be
eliminated" (p. 321). Such varieties, he claims, are "irrevocably con
demned" (p. 373) by the working classes (cf. also discussion in Bochmann
1980: 23). Graur's position is especially dramatic when seen in the light of
twenty-five years of his own carefully documented research on the Romani
element in Romanian. Graur, Wald & Stati (1971) state that "to occupy
oneself with socially dependent lexical and semantical changes is to deny an
interest in linguistics". It is exactly these aspects with which the present
contribution is concerned.
ROMANI IN COLLOQUIAL ROMANIAN 153
2. Sociolinguistic remarks
In looking at Romani lexical borrowings in Romanian, we are dealing
with two types of phenomena:
(1) Individual lexical elements of Romani as a native language were
preserved in the course of language-loss and adopted into Romanian;
(2) Native speakers of Romani transmitted elements of their mother tongue
into Romanian in order to enlarge the scope of expressions.
Transfer of such lexical and stylistic elements from Romani into
Romanian led to a large number of derivatives in Romanian, consisting of
the basic Romani lexemes, to which Romanian derivational morphology is
added. Such linguistic creativity may be observed in the colloquial variants
of many other languages as well (cf., for example, Matras 1991, or Bakker,
this volume). Another relevant issue is the adoption of the Romani variants
by the Romanian youth, which may have been prompted by their tendency
towards non-conformity. Their need to uncover and employ Romani
expressions may have been a response to the great pressure put upon them
during the Ceauescu era to conform to its social dictates. Part of the reason
for the adoption of Romani elements to express dissatisfaction with the
Ceauescu regime may have been the prevalent belief, based on the
154 CORINNA LESCHBER
stereotype of the Roma, that this people would be most likely to escape the
socio-political pressure of theRomaniandictatorship.
Many of the Romanian informants for the present study, however,
knew nothing about the etymology of a large part of the four-hundred
lexemes that were elicited in the interview. It can therefore be assumed that
Romani elements were introduced step-by-step into Romanian, first by
bilingual Roma who allowed Romani elements to enter into their Romanian,
and then by theRomanianswho copied them. The inherent expressive value
for the Romanians of the Romanian Romani borrowings has had its own
dynamic impact upon their further development.
3. Methods
The field work for the present study was carried out with Romanians
and Romanian Roma who claim that they cannot speak Romani. A control
group was formed of Roma whose mother tongue is Romani and who
consider themselves as bilingual. Twenty-two Romanians and Romanian
Roma participated in the interview. Fifteen of them were willing to take part
in a longer interview involving a thirteen-page questionnaire. The
questionnaire is based on lexemes discussed by Graur (1934) and Juilland
(1952), the only significant studies on this subject. The discussion of the
circa four-hundred lexemes took about four hours for every person and was
divided into two sessions of two hours each. There was an almost equal
number of women and men, and of Roma andRomanians.About half of the
participants have lived in Germany for short periods, while the other half
permanently lives in Romania. Most of the interviewees are under thirty.
The interviews were conducted in 1992 and 1993, and three of them in the
spring of 1988.
The interviewees were asked to describe and more clearly define the
context in which the lexemes were used, and reflect upon possible
synonyms. The atmosphere was purposely kept cheerful and informal in
order to elicit as many instances as possible.
A selection from the accumulated material was made for this
presentation. One of the criteria for the choice of the lexemes in the section
Lexical adoptions from Romani is that these examples have undergone a
semantic development. The most important selection factor, however, was
the extent to which the borrowings were known to the informants, reflecting
an important aspect of current language usage in Romanian today. The
material included in this section is generally considered to reflect usages
ROMANI IN COLLOQUIAL ROMANIAN 155
bfta
Romani baxt (subs.) "happiness, luck, benefit, blessing, joy", baxtal (adj.)
"happy, lucky".
Romanian bft "success, good luck" (for encouragement before
examinations and fishing in the language of the pupils and students), bfta
also means "bye-bye", as well as "fortune, chance, influence, glory". Tiktin
& Miron (1986-1990,I: 267) point out "the Gypsy origin of this Romanian
argot word". Romanian bafts, baftosa (adj.) "happy", with the Romanian
adjective ending masc, -os, fem. -osa, is used "in order to make luck stay".
Cf. Graur (1934: 124), Juilland (1952: 157).
benga
Romani bendza-, cf. beng (subs, sg.) "devil", beng (subs, pl.) "epilepsy",
bengal, bendal (adj.) "devilish, mad, insane, angry, wild", also as a noun
"epileptic".
156 CORINNA LESCHBER
Romanian has bnga "devil", but this meaning has become rare. It was
explained as "a kind of name". We can find Romanian bnga in the
expressions (sta) a dat n benga (street jargon) "somebody who cannot
control his movements", "clumsy", or du-te n bnga "piss off!", and
Romanian beng (subs.) "(something) arrogant", "something which is
growing", bnga (surprisingly as an adj.) "bad". Another variant is a fi
bengs "to be clumsy" with the Romanian adj. suffix -os for derivatives of
nouns that mark the possession of a quality.
A curse is Romanian bengsule! with the Romanian masc. vocative
ending -ule, and Romanian bengals!, an adjective used as a noun with no
vocative ending, cf. Romani bengal "devilish". A further derivation is
zbengut (adj.) "naughty", with the Romanian participle ending -it. One
informant mentioned that South Romanians made fun of North Romanians
using the expression bongose "magical spell"; cf. Graur (1934: 131),
Tiktin & Miron I (1986-1990: 360), bonghen, bongheni, with the
Romanian fem, diminutive suffix -i. Cf. also Graur (1934: 128 ff.),
Juilland (1952: 158).
be, be tele
Romani bes, imperative of beel (vb. intr. 3. Sg.) "to sit (down), to stay, to
remain", best (part.) "sitting"; Romanian:
1. be "stay, sit"
2. be "to sit down" (reference to a position for coitus), also be in flaut.
3. betele "sitting", but also "sit down", cf. Juilland (1952: 158).
4. Popular etymology related to a be, a ba "to suffer from flatulence, to
break wind", and the correlated noun be "poop".
5. a betel pe cineva "to grumble, to criticize somebody" (also in the usage
of intellectuals).
6. In crude usage be! "take off, get out of here!", be de aicy! "piss off!".
Cf. also Romanian be ndra, "stay inside!", and Romanian belean "(you)
were sitting". Cf. Graur (1934: 129), Juilland (1952: 158).
bitri
Romani bsto "twentieth".
Romanian bitri "money", with the Romanian masc. Nomina agentis
suffix -ar, (pl. -ari), usually used for forming derivatives from nouns. In this
case, a numeral is taken as the basis. Cf. Graur (1934: 130), Juilland (1952:
159).
ROMANI IN COLLOQUIAL ROMANIAN 157
a buli
Romani bul "buttocks, ass", adj. buljan "back, arse", buldel (vb. 3. sg.)
"to have intercourse".
Romanian a buli is a productive verb, the original meaning of which is
"to make love; to fuck", as was mentioned already by Juilland (1952: 159),
but its current meaning is closer to "to fail in a task, to screw up, to blow it,
to make a mess of things". Cf. Romanian am bulit-o "I destroyed it". The
simple meaning is still expressed in Romanian te bulsc. It is used towards
women in a sexual context and towards men in a fight to mean "I'll beat you
up". One person even mentioned abid "to fuck" in one word.
Romanian bulel (subs.), as in bulela asta vine de la el "this crap
came from him", bulel and bulangel also mean "to screw like a rabbit",
with the Romanian fem, abstract suffix -el, used as in this example to
form derivatives from verbs, bulendr "rag, whore", with the Romanian
fem, augmentative suffix -ndra, is strongly depreciatory.
Cf. also buln, with the Romanian augmentative suffix -an for nouns:
1. "chicken leg",
2. "pretty woman's legs" in ce bulne mit are (neutr. pl.)
3. "the thighs of a whore",
4. "truncheon".
A derivation is bulangu "idiot", with the Romanian suffix -giu for
nouns and adjectives to form masc. Nomina agentis. Cf. Graur (1934: 131
ff., 1936: 196 ff.), Juilland (1952: 159).
a se bungh
Romani bango (adj.) "bent, crooked, slanting", but also "guilty, unjust".
Romanian a se bungh (vb. refl.) "to stare, to goggle, to gawk", a se
zbungh (vb. refl.) "to look", grammatically adapted to the fourth Romanian
verbal class with the help of an infinitive /-ending, see below zbanghiu
(adj.) "squinting, cross-eyed" etc.. Cf. Romanian a bungh "to grap, to catch,
to find a solution, to spy out", as in am bunghit-o "I caught him", also
Romanian bonght (part.) "achieved, succeeded". Cf. discussion in Graur
(1934: 195) and (1937: 222), Juilland (1952: 160).
cnci
Romani khnci "nothing".
158 CORINNA LESCHBER
car; crici
Romani kar "penis".
Romanian car, and with the Romanian suffix for nouns -ici with a
different grammatical function [or -ciu, cf. Pascu (1916: 324 ff.)]: crici
"male sex organ", carici "female sex organ", but cf. Boretzky & Igla (1994:
136) Romani karci "man, lady's man", and as masc, noun in: Romanian
freac cariciul (not a literal translation) "he masturbates" (with Romanian
masc. def. article). It shows a high lexical productivity: caricioic, (fem.)
"nymphomaniac" (in Romanian denominated as pulrist), with a Romanian
fem, composite suffix -oaic for adjectives and nouns, with which
denominations for female beings are formed from masc, nouns. Romanian
cariciometru or cariciopd "huge sex organ", also used among children in
word-games, with the aid of two suffixoids, a carici (vb.) "to fuck",
grammatically adapted to the fourth Romanian verbal class with an
infinitive /-ending, cariceal "fucking, masturbation", with Romanian fem.
ROMANI IN COLLOQUIAL ROMANIAN 159
abstract suffix -el, caricii "sex-maniac" or "large male sex organ", with
Romanian masc, augmentative suffix -oi, slightly derogatory. Cf. Graur
(1934:134 ff.), Juilland (1952: 161).
a cardi
Romani kharl (vb. tr. 3. sg.) "to call", cf. Boretzky & Igla (1994: 157),
akhardem "I called", akharl (vb. tr. 3. sg.) "to call, to invite", akhard
(part.) "invited, called", (cf. also Romani vacardol (vb. passive, intr.) "to
agree upon", vacard (part.) "said, promised", vacarl (vb. tr.) to speak, to
talk, to say, to promise").
Romanian a cardi "to talk", cf. noi cardim "we are talking, we are
having a chat", cardel "talk", with the Romanian fem, abstract suffix
-ed. Cf. Graur (1934: 133), (1936: 198), (1937: 223), Juilland (1952: 160).
crl; caro
Romani kherl (adv.) "from the house", "out of the house", cf. Romani kher
(subs.) "house", or Romani kerl (vb. tr. 3. sg.) "to make" etc.But cf. also
Romani kor, kor, kar (prep.) "in, at", Romani kori (adv.) "there, in that
direction", (prep.) "at, to", cf. Boretzky & Igla (1994: 148).
Romanian carel; caro in: car de aici, carl de aici, crl potca "get
lost!". See also in Juilland (1952: 161) car "flee! run away!". Cf. Graur
(1934: 134), Juilland (1952: 161).
ciangalsule!
Romani chagl, chaglin (part.) "ugly", cf. Boretzky & Igla (1994: 55) also
chang-, Romani changal s le "he is ugly", possibly overheard by
Romanians as a vocative-form. Another explanation is a derivation from
Romani changal (adj., masc.) with the aid of the Romanian suffix -os, for
the formation of nouns from adjectives to mark possession of a quality, in
Romanian masc. vocative form with masc. vocative ending -ule:
*ciangal()- (o)s- ule > ciangals- ule.
The meaning could not be further explained by the informant (cf. also
ginglule, cialapadiule, janghinsule), but these words were insulting.
cialapadiule
Romani capldile "stupid"
Romanian cialapadiule "an insult for men", "clod, idiot", with the
Romanian masc. vocative ending -ule.
160 CORINNA LESCHBER
a ciord
Romani corl (vb. tr.), part, cordo "to steal", cf. cor "thief", cordi "female
thief".
Romanian a ciord "to steal", cf. ciordel "stealing", with the
Romanian abstract suffix -el, which forms fem, nouns from verbs and
adjectives, for describing the result of a previous act; ciorditr "thief", with
the Romanian masc, suffix for Nomina agentis -tor, which serves to derive
verbal adjectives from other verbs; substantival use. Cf. Graur (1934: 139;
1936: 198), Juilland (1952: 162).
ciriclu
Romani cirikl (fem. subs.) "bird", cf. also cirikl (masc, subs.) "bird,
sparrow".
It is used very often by all Romanians, but, as one informant pointed,
they do not know its meaning. In: "Ciordt-i ciricliu, barburt-i ciripi" it
was perhaps a kind of magical spell.
Romanian ciricliu (masc. subs.) is probably based on Romani fem.
subs. cirikl; this is indicated by the yet conserved female form of the
adjective in: ciordt-i ciricliu... The i-ending is adapted to the Romanian
morphological system with the help of Romanian masc. substantive ending
-iu, but the form could suggest also a fem, vocative with the ending -io. Cf.
Graur (1934: 139).
a ciumid
Cf. Romani cumdel "to kiss" (vb. tr. 3. sg.), (prt.) cumidja, (part.) cuntido,
cumidino "beloved", cum, cumi (subs.) "kiss".
Romanian a ciumid "to kiss", but also a ciumid (pe cineva) "to
degrade, to humiliate somebody", cf. the rough expression ciumda-mi bul!
"kiss my ass!". But Romanian ciumte in bi, ciumte! "hey, dummy!" is
related to slav. um "pestilence", Romanian cim "id." with Romanian
substantive suffix -ete, with a diminutive or depreciatory nuance. Cf. Graur
(1934: 140).
a se coflei
Romani kovlo (adj.) "soft, gentle, weak", kovljl (vb. passive, intr.) "to
soften, to slacken" etc..
Romanian a se cofle(vb. refl.) "to get squashy" (melons or fruits), "to
get slack or loose" (skin). The adaption to the fourth Romanian verbal class
ROMANI IN COLLOQUIAL ROMANIAN 161
dvia
Romani devla "o God!", vocative of del, devel "God", "sky, heaven"
Romanian dvia "head, mind". For a semantic discussion see Graur
(1934: 149), Juilland (1952: 163).
dic
Romani dikh, imperative of dikhl (vb. 3. sg.) "to look", dikh mo! "hey,
look!"
Romanian dic mo\ "hey look!", (cf. mo), die la el ce obraznic este "get
a load of that guy!", die fza! (slang) "hey, take a look at that!" or "get a
load of this!". Cf. Graur (1934: 150).
diliu
Romani dil (adj.) "crazy, mad", (subs.) "madman, fool"
Romanian diliu "a crazy guy", it is formed with the Romanian masc,
adjective suffix -iu, used here as a noun. Cf. Romanian a dili "to destroy".
Romanian diliu is not verified in Graur (1934), Juilland (1952).
dita; dtai
Romani dita! (interject.) "here! look!", cf. Boretzky & Igla (1994: 73) <
dikh-ta!
Cf. Romanian dtai! (it is used when the Romanians want to point to
something big and extraordinary), dtamai (mocking, ridiculing) "look at
him". For mai see Boretzky & Igla (1994: 173). Cf. Graur (1934: 150 ff.).
gagu, gagca
Romani gadz (subs. masc.) "non-Gypsy man, husband, peasant, master,
landlord". Romani gadz (subs. fem.) "non-Gypsy woman, woman, wife".
Romanian gagiu 1 means:
1. "a handsome guy" as in ce gagiu mit "what a handsome guy",
2. "man",
3. "a tough guy",
4. "gigolo",
5. as a rough greeting-formula ce faci, bi, gagule! "(hey man!) how are
you doing?",
162 CORINNA LESCHBER
gen
Romani dzan "they go", dal (vb. 3. sg.) "to go, to rush, to hurry".
Romanian gena 2 in: e gen pe tine "he runs after you" or "he is after
you", cf. Graur (1934: 155 ff.), and see a gin.
ginglule
Romani dzungal "bad, mean, dirty" (about people and their actions).
Romanian ginglule! "curse", with the Romanian familiar
augmentative suffix for nouns and adjectives with depreciatory character
-u, and Romanian masc. vocative ending -ule. Cf. Graur (1934: 155).
ROMANI IN COLLOQUIAL ROMANIAN 163
a gini
Romani verbal-stem dan- in danl (vb. 3. Sg.) "to know"; Graur (1934:
156) also mentioned the variant din- (gin-).
Romanian a gini "to notice, to look, to steal a glance". Nowadays, a
gini means "to stare", and especially "to spy, to observe" in connection with
the police and the secret service. Cf. Romanian a le gini "to be in the know,
to know what one is doing with...". Its Romanian derivations are: ginela
"spying", with the Romanian abstract suffix -el, which forms fem, nouns
from verbs and adjectives, for describing the result of a previous act, and
ginitr "spy in a political context", and especially "store-detective", with the
Romanian masc. suffix for Nomina agentis -tor, which serves to develop
verbal adjectives from other verbs. Within the control group of Roma whose
first language was Romani, the Romanian verb a ti "to know" was
associated with a gini, which confirms the etymology mentioned above. Cf.
Graur (1934: 155 ff.; 1936: 198), Juilland (1952: 165).
hacan
Romani akan (adv.) "now, immediately, there!"
Romanian hacan "get away!" or "(get) on the side". Also acan,
harcan. For a semantic discussion see Graur (1934: 157), cf. also Juilland
(1952: 166).
hai
Possibly Romani ajl (interjection)"yes!"
Romanian hai is often considered generally Balkan, but nevertheless
mentioned by Graur (1934: 158). Although its origin is obscure, it is still
worth mentioning, and the interviewees explained it to mean:
1. "scandal",
2. as in a face hai (often, vulgar) "to make fun of someone". This expression
has entered the literary language and is known to all Romanians,
3. "to make a good mood, to make people happy",
4. "to cause a fight, to create a scandal",
5. "to laugh loudly",
6. "to bother others".
As a derivation with the Romanian adjective ending -os, -oas, there is
an adjective hais (masc), haiosa (fem.) "happy", in: un om hais "a
happy man" or este o femeie haios "a cool lady". Hais, haios cm also
mean "pretty", for example, in connection with new clothing. Lupu (1972)
164 CORINNA LESCHBER
hafarl
Romani av ord "come here!".
Romanian hafarl', haord "come over! come here!". The following
observation of a Romanian informant is an example of how a neutral
expression, overheard by a Romanian, suddenly assumes a new meaning:
"In the suburbs of Bucharest there were Roma, who raised pigs, and they
said to the pigs hafarl, and they also used that word to insult each other." It
appears that we are dealing not with hafarl, but with misinterpreted
Romanian haord or Romani av ord. Probably the Roma were calling their
pigs to feed them. When the Romanian children heard the Roma call other
people to come over, they had the impression, they were using a curse, since
they related it to the pigs. For haord see Graur (1934: 161 ff.), Juilland
(1952: 166).
a hali
Romani xal (vb. 3. sg.) "to eat".
Romanian a hal (vb.) "to eat" in the following expressions:
1. trebuie s hlesc ceva "I've got to eat",
2. s nu faci aa, c astfel eti halit "don't do it, or you'll get devoured,
caught".
3. In a more gentle application: ce frumoas eti, haliti-a gura "you're cute,
could eat you up".
4. taci, c te hlesc "shut up, or I'll eat you up", whereby two major
meanings emerge: the derived "to get somebody, to hit somebody", and the
original "to eat (much and quickly), to pig out", also in a figurative sense.
Cf. Romanian hal "hunger". Derivations thereof are: Romanian haleal
(generally) "a large casual meal or a large crude dinner party with several
persons and with simple food". Some informants explained that if need be,
one person can hold his own haleal dinner-party. As a nominalized verb
haleal means "eating, meal, feeding", with the Romanian abstract suffix
-el, which forms fem. nouns from verbs and adjectives, for describing the
result of a previous act. A derivation using a Romanian word formation
suffix is halimatr "quarrel", but also "bread", with the Romanian suffix
-tur, resp. -atur, which forms fem. nouns from verbs. Cf. Graur (1934:
159 ff.), Juilland (1952: 166).
ROMANI IN COLLOQUIAL ROMANIAN 165
halimi
Romani xalimata (subs. pl.) "ruin" etc..
Romanian halimi "scandal, quarrel", as in a face halimi "to quarrel",
which is not connected to Romani xalma "quarrel", as there is a different
accent, but rather to Romani xalipe, pl. xalimata "ruin, destruction, itching,
boundless desire". According to one informant it is possibly related to Ro-
mani xalm (vb. 1. sg. pret.), to xal "to eat". Cf. Graur (1934: 161).
ta
Romani ta! (interj.) "just look!"
Romanian ta! "look!", ta - dic la el,ta-mi,ta-mi,tala el "look (at
him)!". Cf. Graur (1934: 150, 163).
janghinsule
Romani dangl, danglino (part.) "famous, known, intelligent", cf.
Boretzky & Igla (1994: 82), budangl (adj.) "experienced, learned", (subs.)
"expert".
Romanian janghn/janghins "clever, jealous, envious"; the meaning
of Romanian janghinsule is "clod, idiot", in the Romanian masc. vocative
form with the suffix -ule. The origin of these words (ginglule,
cialapadule, ciangalsule and janghinsule) was not explained, but the
informant was sure that it can be traced to Romani. He frequently heard
these expressions from Romani children, who used them in addressing other
children.
lovle
Romani lov (subs, pl.) zu lov "money".
Romanian lovle "money"3' with a Romanian artic. fem. pl. -le. Cf.
Graur (1934: 165), Juilland (1952: 166 ff.).
macht
Romani makhiv "to get drunk"; macarl, makjarl (vb. 3. sg.), cf. Boretzky
& Igla (1994: 172) "to make drunk".
Romanian machit, maht (part, pret.) "drunken", cf. a se maki (vb.) "to
get drunk", but also "to stare at", with the Romanian participle ending -it.
Graur (1934: 166), Juilland (1952: 167).
166 CORINNA LESCHBER
mndea
Romani mnde (pers. pron. 1. sg. locative), to me "I" (pers. pron. 1. sg.
nom.).
Romanian mndea "to me" in cite pe mndea "throw it to me".
Romanian mndea can express "something familiar/trusted" as in vnu la
mndea instead of vinu la mine "come here to me". Its meanings are:
1. "come to daddy" (as if doing a favor, patronizing),
2. "I dare you to come here" (before a fight, to a weaker opponent),
3. "come to me" (in the sense of "I'm the boss, I can help you"). It is also
used by women. Cf. Graur (1934: 167), Juilland (1952: 167).
mangti
Romani manghn "treasure, property", manginal (adj.) "wealthy, rich". But
cf. also the Romani verbal stem mang- "to beg".
Romanian mang ti (subs. Pl.) "money", with a Romanian neutr. ab
stract suffix for nouns -ot, here in the plural: -ofi. Romanian manghitr,
manghitr, "thief, beggar", a mangli, a mangli "to steal, to beg". Cf. Graur
(1934: 167), Juilland (1952: 167).
mardi
Romani mard (subs.), "(little) coin".
Romanian mardi (subs. pl.) "money", mardu (subs. sg.) "leu, franc" (the
singular form is never used). Cf. Graur (1934: 168; 1936: 199), Juilland
(1952: 167).
a mardi
Romani marl (vb. 3. sg.) "to beat", part. mard "beaten", but also as subs.
"coin", cf. Romanian mardi.
Romanian a mardi "to beat", as in te mardsc "I'll beat you up", also
mard "to beat", mardel "walloping", with the Romanian abstract suffix
-el, marditr who is beating", with the masc. suffix for Nomina agentis
-tor, mardei "pugnacious", with the Romanian suffix for Nomina agentis
-a. Cf. Graur (1934: 168), Juilland (1952: 167 ff.).
matl
Romani mat (adj.) "drunk", cf. Romani macl, dial, matl "to get drunk",
cf. Boretzky & Igla (1994: 172), see macht.
ROMANI IN COLLOQUIAL ROMANIAN 167
a mierli
Romani merl (vb. 3. sg.) "to die", muljarl (vb. 3. sg.) "to kill, to murder".
Romanian a mierli (vb.) "to do in, to murder, to bump off', but also "to
die", as in o mierlte "(I'm sure) he will die", "he's going to kick the
bucket", a mierlit-o "he died", "he kicked the bucket". For discussion see
Graur(1934: 170; 1936: 199), Juilland (1952: 168).
mit
Romani mist (indeclin. adj.; adv.) "good".
Romanian mit "nice, good, pretty". Every Romanian knows this
word. Derivations are: mitocreal in the usage lsa mitocrel la o
parte, treci la subiect "knock it off, let's get down to business", with the
abstract suffix -eal, mitocar "somebody, who often jokes at the expense of
others" 4 , with the masc. suffix for Nomina agentis -ar. Romanian mit also
appears in connection with "to make jokes", or "to make stupid or bad
jokes", and in a lu la mit "to make fun of" and a face mit de cineva "to
talk ironically about somebody". Cf. Graur (1934: 171 ff.; 1936: 199),
Juilland (1952: 168).
moln
Romani mol "wine".
Romanian mol, moln "new wine", with a Romanian augmentative
suffix -an, which forms masc. nouns from other nouns, adjectives, verbs
and interjections. Cf. Graur (1934: 173; 1936: 199), Juilland (1952: 169).
mie
Romani muj "mouth, face".
Romanian mie "face" or vulgarly "gob, mug", and muin "id.", with a
Romanian masc. augmentative suffix -an. Cf. a da la mide (in a sexual
context), a duce cu mida "to lie". Cf. Graur (1934: 174), Juilland (1952:
169).
nasl, nasl
Romani nasl (indeel, adj.) "bad, damned".
168 CORINNA LESCHBER
Romanian nasl (adj.) "ugly, dumb", (subs.) "dirty fellow", nasl (adj.)
"something ugly". To drive away a person, it's enough to say mai, naslule!
(in the Romanian masc. vocative form with -ule) "get away, ugly!", eti un
nasl! "you are ugly!". Derivations are Romanian o nasulie "a bad, horrible
or ugly event ", with the abstract suffix -ie, for the fonnation of fem. nouns
from other nouns, adjectives and verbs which describe properties and
conditions, and nasolel "id.", with the fem. abstract suffix -eal, a refl.
verb: a se nasuli "to get worse" (for example a situation), in: s-a nasolit
treb. The informants were not familiar with nasoale5. Cf. Graur (1934:
176), Juilland (1952: 169 ff.).
nasfarliu, nasparliu
Cf. Romani nasulip (subs.) "evil, vice, damnation, epilepsy" and nasvalip
(subs.) "illness", nasval, nasfal (adj.) "ill, sick".
Romanian nasfarliu, nasparlu, naparlu was explained to mean "an
unserious type, whom no one can trust; a man who likes to fight; a dandy",
but also "something pretty, interesting, stylish", with the Romanian masc.
adjective suffix -iu, here used as a substantive. Romanian nasfarliu was also
considered a synonym for nasl, and nasl the antonym to mit. A number
of Romani informants said that nasfarliu derives from nasfal, meaning
"sick". Cf. Juilland (1952: 169).
a parad
Romani phaad (part.) "split, ragged", phaavl (tr. vb. 3. sg.) "to make
something burst, to split, to wear out" etc., cf. also Romani phadol (vb.
pass. intr.) "to burst, to split, to explode" etc..
Romanian a paradi (vb.) "to break, to spoil" or "to smash up a car"; this
verb is formed from the Romani part. (see above), and morphologically
adapted to the forth Romanian verbal class with an /-ending; Romanian te
paradsc! "I'll beat you up!"; Romanian parade l "break-in, burglary". Cf.
Graur (1934: 178), Juilland (1952: 171).
a pili
Romani pijl (vb. tr. 3. sg.) "to drink", pilo (part.).
Romanian a pili "to drink", cf. am pilt "I got tanked up" and a se pili
"to get tanked up". Derivations are: pilel "boozing", with the fem. abstract
suffix -el, pilangu "alcoholic", with the masc. suffix for Nomina agentis
-giu, in this case with a depreciatory connotation. Graur (1934: 180) and
ROMANI IN COLLOQUIAL ROMANIAN 169
pirnda
Romani phirad (adj. fem.) "like a slut",phirvan,phiran"sweetheart, slut".
Graur proposed pirand (part.), to the verb pirav- "to make love", not
mentioned in Boretzky & Igla (1994).
Romanian pirnda "name", "name of a Romani woman",. Cf.
discussion in Graur (1934: 181 ff.), Juilland (1952: 171 ff.).
puradl, puradu
Cf. Romani poadchavo "naughty child"',poad(part.) "opened wide",
from the verb pofavl (vb. tr. 3. sg.) "to open wide", chavof (subs. dim.)
"small boy, son", to chaa (subs. pl.) "brats, naughty children". According
to an informant with a shift of meaning, so that poado > puradu means
"naughty child".
Romanian puradl, puradu "child", especially "Romani child", "brats".
Some informants quoted Romanian Roma saying am ase puradei "I've got
six children". In a fight, Romanians challenge each other: ia-ti puradii i
pleca!" - "take your brats and get out!". Juilland (1952: 172) related
puradl toRomanipurde "naked".
puriu
Romani phur (adj. masc), phuri (adj. fem.) "old".
Romanian puriu (adj. masc.) "old" or "old man", pure (adj. fem.) "(my)
old woman". We even encountered: o purie btrn (!) "an old purie".
Romanian purisnca "comic old woman", with a fem. composite diminutive
suffix -nc, here with an ironic touch. Cf. Graur (1934: 183), Juilland
(1952: 172).
a soil
Romani sovl (vb. intr. 3. sg.) "to sleep", sovip, sojip (subs.) "sleep, bed".
Romanian a soili (vb.) "to kip, to take a snooze", as s substantive soi in:
a trage un soi "to kip a little bit, to take a little snooze/ forty winks",
soilela "snooze, nap", with the Romanian fem. abstract suffix -el. Cf.
Graur (1934: 187), Juilland (1952: 174).
170 CORINNA LESCHBER
ucr, ucr
Romani ukr (invar. adj.; adv.) "beautiful, pleasant".
Romanian ucr, ucr (adj. masc), ucr (adj. fem.) "good, pretty,
beautiful", as a compliment to women: eti ucr "you are so pretty".
Compare this to the Romani name Sukari "name of a pretty woman". In
contrast, consider the verbal derivations: a ucr (vb.) "to make somebody
angry" and a se ucr (vb. refl.) "to get angry", as in: m ucrsc "I get
angry". Cf. Graur (1934: 188 ff.), Juilland (1952: 174 fff.).
uru
Romani suri, chur "knife".
Romanian uru and ciuru "knife", specifically "knife, worn on the
body, and one used for criminal purposes". This noun is productive and
forms the Romanian verb a ciurui "to perforate, to poke holes into". The
word uriu is well integrated into Romanian and is rarely associated with
Roma. One Romani informant corrected uriu to uri "knife", which is
phonetically and semantically identical with the Romani word. Cf. Graur
(1934: 189; 1936: 200), Juilland (1952: 176 ff.).
a uti
Romani chuvl (vb. tr. 3. sg.) "to put, to put into, to stuff, to throw" etc.,
(pret.) chut]a.
Romanian a uti "to pinch, to nick things" (in an elegant manner, as
assured to me by an informant), also "to pinch somebody's girl-friend" as in
-am utt gagc', utel "theft", with the fem. abstract suffix -el;
utitr and ut "thief" as in el e ut de buzunare "he is a pickpocket". Cf.
Graur (1934: 189 ff.; 1936: 300), Juilland (1952: 177).
ticlos
Romani tikalos6 "an unhappy person, vagabond", perhaps related to
Romani tka "father", cf. Saru (1992: 160).
Romanian ticls "bad, sad, disgusting, false" (adj.), un ticlos (subs.)
"a lazy person, a man with two faces, dishonest person, liar, cheater, bandit,
immoral person", intentii ticlose (adj.) "bad intentions", ticloenie
"falsehood", with the fem. abstract suffix -ie for the formation of fem. nouns
from other nouns, adjectives and verbs which describe properties and
conditions; in a verbal construction a ticloi pe cineva "to cheat, to deceive
somebody", as in te am ticloit. Cf. Graur (1934: 192).
ROMANI IN COLLOQUIAL ROMANIAN 171
trla, tral
Romani tras1 (subs.) "fear" , trasl (vb. intr.) "to be afraid of",
(perf.) me trasjlem "I was afraid".
Romanian tr l, tral, with the variant -l from the Romanian fem.
abstract suffix -el, means:
1. "fear, anxiety",
2. a psychological condition, as in mi este trl, s fac ceva "something's
keeping me back from doing...", mi-e trla "I'm afraid" and generally "to
be afraid; the fear to do something, to be frightened, to avoid",
3. "embarrassment",
4. "the feeling, to suffer from loneliness".
Cf. Graur (1934: 193; 1936: 200), Juilland (1952: 177 ff.).
a se uch
Romani ustl, ucl (vb. intr. 3. sg.) "to get up, to wake up", (pret.) utlo,
cf. in Graur (1936: 199) Romani ustiav "I go away".
Romanian a se uch (vb. refl.) "to buzz off", Romanian uchta (adj.)
"something crazy", further explained as "his mind has gone away, he
remains with an empty head", uchela "flight", with the fem. abstract
suffix -el. Cf. Graur (1934: 194; 1936: 200), Juilland (1952: 178).
vast
Romani vast "hand", cf. also vastalo (adj.) "long-armed, thievish".
Romanian vast "hand", but also "paw, pocket" as in a bag vstul "to
paw, to steal", strictly speaking "to grasp", and in ia vstul de aici! "hands
off!". In earlier sources (Graur 1936: 200) documented as "punch in the
chin, push, knock". Cf. Graur (1934: 194), Juilland (1952: 179).
zbanghiu
Romani bango "bent, crooked, slanting" etc., cf. Tiktin & Miron (1986-
1990, III: 917), DEX (1975: 1040).
Romanian zbanghiu (adj.) with Romanian the prefix z- and masc.
adjective suffix -iu:
1. "squinting, cross-eyed"
2. "with a sick eye",
3. "a man, who beats his wife",
4. "an unreliable fellow".
172 CORINNA LESCHBER
Cf. eti zbanghu? "are you nuts?" Cf. Graur (1934: 195), Juilland (1952:
179).
zurlu
Romani zural (ajd.) "strong, powerful, violent".
Romanian zurliu (adj. masc.) "with the head in the clouds; crazy; in a
happy mood", but also "naughty, impertinent", "impetuous, wild",
"something noisy, barmy, retarded", with the adjective suffix -zw, resp. fem.
-ie. Cf. dar criz e cam zurle (adj. fem.) "it's a horrible crisis", o femeie
zurlie "a crazy woman". Cf. Graur (1934: 195; 1936: 200).
the word was frequently very close to the original meaning in Romani. It
was evident that the Romanians knew relatively few lexemes, usually with a
semantically strong and special connotation. The Romanian Roma, on the
other hand, although without any knowledge of Romani, were familiar with
many lexemes, but related to them semantically neutral meanings.
Judging by their behavior in the interviews, the two groups, Romanians
and Romanian-speaking Roma with no knowledge of Romani, may be
considered as representing two stages in the semantic development of
Romani-origin loans in Romanian. The first, the Romanian group, has
knowledge of selected items which have shifted quite significantly from a
semantic point of view. The second, the Roma, show traces of a mixed
lexical system, where Romanian-speakers still had direct access to an active
Romani lexicon and could draw on items in their original Romani meaning.
Although Romani borrowings are recognized and known to various
degrees, the interviews have shown that none of the interviewees, in neither
one of the groups, including the Romanian Roma who consider Romani to
be their dominant language, understood all words mentioned by Graur
(1934) and Juilland (1952). This raises the question of the authenticity of at
least some parts of the material used by the authors from Veselia or other
newspapers such as Pardon or Diminea ta, which furnished a part of the
sources for Graur and Juilland.
Attempts to conduct a statistical analysis on the basis of elicited data
show that knowledge of the elicited lexemes differs most strongly among
women from both ethnic groups, Romanians and Romanian Roma, while
the degree of knowledge of lexemes among men of both groups is nearly
the same. This means that Romanian women generally tend to know the
least Romani loanwords. Romanian Romani women, on the other hand,
know the most lexemes. This can indicate a stronger language conservatism
among women, but it could also mean that while Romani women are still
strongly embedded in a traditional cultural context, Romanian women have
little access to those domains on the fringe of Romanian society where
taboo-related items, including Romani loans, are used.
Experience shows that it is the intellectuals and students in Romania
who have the most natural and positive relationship to the Roma. An
interesting phenomenon can be observed in the language of school students,
particularly, as far as we can tell, among pupils in Bucharest since the early
1980's. Those informants who applied a new interpretation to the
expression be (see above) were all pupils or students between the ages of
174 CORINNA LESCHBER
sixteen and twenty-six, who had attended school in Bucharest (with the
exception of one informant who went to school in Ploeti, sixty kilometers
to the north). Other examples of typical expressions used among pupils are
bft "success" (for encouragement before examinations), barosn "big fat
man or woman", and also in a figurative sense "important person", and
gagu "a guy, a dude" (among the youths). A central position in the
language of the pupils is occupied by the antonymic pair nasl "stupid,
bad", and mito "good, marvellous, super, pretty".
To conclude, it can be stated that the Romanian lexical items of Romani
origin dealt with in this study adhere to the criteria for various non-literary
varieties, such as Colloquial Style, Slang, or Jargon. However, what
Domaschnev (1987: 314) mentions as features of Argot - to conceal
meanings from the outside world and serve as a means of identification for
insiders - has disappeared. It remains to be seen whether the Slang
component - a large variety of synonyms (such as expressions for "to steal")
- might not disappear as well, giving way to full integration of the items in
question as 'neutral' expressions in general Romanian colloquial speech.
NOTES
1 It exists also in Bulgarian as the invariable neutral noun gde, denotating female and
male persons. With the definite neutral -article -to: gdeto. It means "lover, girl-/boy-
friend, dear". No Bulgarian, whom I asked, knew about the origin of the word.
2 Romanian gen "(eye)lash" is not related to this.
3 One informant said that there are other words used by the young Romanians for
"money": mardi (<Romani), bitri (<Romani), mangti (<Romani), marafti (<Turk.).
4
Lupu (1972) mentions mitocr "crafty customer, smart lad".
5 C f . Lupu (1972) a bgpe nasoale "to slander; to rabbit on, to blether", afipe nasoale
"to be enough to drive you up the wall". But Lupu saw only a possible connection to nasl.
6 As suggested by aRomaniinformant from the control group.
' It seems that the Romani word is borrowed from an Iranian language, but Romanian
adapted it from Romani.
ROMANI IN COLLOQUIAL ROMANIAN 175
REFERENCES
VICTOR A. FRIEDMAN
University of Chicago
0 Introduction
Romani is one of the few widely-spoken languages of Europe for which
basic issues of standardization (orthography, dialectal base, etc.) are in the
process of resolution. In Haugen's (1966:16-26) terms, Romani is at the
stage of selection of a norm. Due to the fact that the Roms are a transnational
people, problems of language planning are additionally-complicated by the
fact that they are being confronted in the frameworks of various and varying
state mechanisms. For the Roms of the Republic of Macedonia, issues of
identity maintenance and sociopolitical integration must be viewed in the
context of an educational policy that has included multilingualism for the past
half century in a state that has only recently achieved independence and is
surrounded by overt and covert threats to its integrity. This paper will
examine a specific event in the current efforts to standardize Romani, namely
a conference held in Skopje, Macedonia on November 20-21, 1992, and the
document that resulted from it. The document itself will be presented together
with commentary on its context significance.
GENERAL PRINCIPLES
1. This codification is for the Romani language as a course of study in the
Republic of Macedonia. This codification is viewed as a necessary step to
ward the international Romani literary language and not in competition with
it.'
This statement was intended to address Cortiade et al. (1991). The
Romani participants in the conference felt that the situation in Macedonia
required a regional standard for use in Macedonian elementary schools, with
a view to study of the international standard later. See also comments on the
Alphabet.
ROMANI STANDARDIZATION IN MACEDONIA 181
'2. In view of the fact that the majority of Roms in the Republic of Macedonia
use the Arlija dialect, this dialect shall serve as the basis of the Romani
literary language in the Republic of Macedonia, but with certain grammatical,
phonological, and especially lexical additions (and modifications) from all the
Romani dialects of the Republic of Macedonia such as Dambaz, Burgudi,
Gurbet, and others.'
Throughout the history of Romani standardization efforts in Macedonia,
Arlija has served as the basis, but, as indicated above, this question had be
come a politically divisive issue. This compromise was satisfactory to all
present at the conference.
ALPHABET
The Romani alphabet in the Republic of Macedonia consists of the following
letters in Latin transcription. The corresponding Macedonian orthography is
used for Cyrillic.
Aa, Bb, Cc, , h/h, Dd, D/d, Ee, Ff, Gg, Hh, Ii, Jj, Kk, Kh/kh, Ll,
Mm, Nn, Oo, Pp, Ph/ph, Rr, Ss, , Tt, Th/th, Uu, Vv, '
The corresponding Cyrillic alphabet would be the following (the order
follows that of the Latin alphabet):
contact with Romani, e.g. Serbo-Croatian and Polish, was also felt to favor
the use of the hacek for strident palatals for Romani in Macedonia.
The situation can be compared to that of the Albanian alphabet congress
of 1908. The crucial decision of that congress was the adoption of the
principle that Albanian would be written in a Latin alphabet rather than Arabic
or Greek, although the two major Latin alphabets then in use - one based on
the principle of one letter per sound the other using digraphs - were both
endorsed. Eventually a single alphabet became official. Similarly, while most
Roms agree that the alphabet used for Romani should be Latin (the mention
of Cyrillic is simply for contexts where transliteration of individual items
might be desirable in Macedonia, as opposed to the exclusive use of Cyrillic
in Malikov 1992), there is not yet a general consensus concerning the details
of orthography. See especially points 4 though 8 below.
COMMENTARY
In some Romani dialects, the uvular fricative /x/ is distinguished from the
glottal aspirate /h/ and/or the rolled Ivl is distinguished from another related
type of sonorant, but in view of the fact that such distinctions are not made in
the Arlija dialect special letters [for these distinctions] have not been
introduced into the alphabet. Such pronunciations are permitted as literary for
those speakers who have such distinctions in their native dialects.'
These are two major contested issues. Jusuf and Kepeski (1980),
Kenrick (1981) and Cortiade et al. (1991) all prescribe a graphic distinction
for /x/ and /h/, but Jusuf and Kepeski (1980) fail to make the distinction in
practice, using both <x> and <h> in the same roots, e.g. xiv, hiv "hole", xor
"depth" but horadaripe "deepening", an illustration of the problem that would
be encountered by speakers of dialects without the distinction. Given that
minimal pairs are extremely rare and that the sounds themselves are in free
variation in dialects where they are not phonemic, the majority of Roms at the
meeting felt that a single grapheme should be used. In the case of the two
types of /r/, Cortiade et al. (1991) makes the distinction facultative while both
Kenrick (1981) and Jusuf and Kepeski (1980) give only one Ivl in their
standard orthographies.
Romani dialects. In the rare instances of schwa in the Arlija dialect, the
corresponding form in Dambaz or some other Romani dialect with a
different vowel will be taken as the literary norm, e.g. instead of Arlija
vrdon ["wagon"] Dambaz vurdon is accepted.'
Schwa is of foreign origin in Romani and only occurs in dialects
influenced by contact with languages where schwa is phonemic. Its
occurrence is generally limited to borrowings from those languages. Thus,
for example, in Jusuf and Kepeski's (1980) vocabulary of 2,292 entries,
only 31 items, representing at most 21 roots, contain schwa. Although Jusuf
and Kepeski (1980) use the sign <> for schwa, Kenrick (1981) and
Cortiade et al. (1991) both exclude it as dialectal. In view of the fact that
many occurrences of schwa in one dialect correspond to some other vowel in
another dialect, this was viewed as a good opportunity for expanding the
vocabulary of standardized Macedonian Romani beyond the limits of the
Arlija dialect.
'2. Where there is aspiration in the root of a word, it will always be written,
e.g.jakh ["eye"].'
Although some dialects have deaspiration of underlying aspirates in
some positions, the adoption of the morphophonemic principle of re
presenting the underlying morphophoneme in spelling which is common to
many of the languages of Eastern Europe was adopted.
'3. Automatic devoicing is not spelled at the end of a word, e.g. dad
["father"].'
Same as point 2 above.
'4. Where an underlying dental or velar stop or sonorant occurs before a
front vowel or jot, i.e. t, d, k, g, 1, n plus i, e, j , the underlying consonant is
used in spelling, e.g. buti ["work"], kerdjum ["I did"], geljum ['T went"], lil
["letter"], pani ["water"].'
This is an area of both considerable and salient dialectal variation and
morphophonemic alternation in Romani. Underlying or historical dental
and/or velar stops in these positions can be pronounced as palatals and/or
with affricated or fricativized articulation in various dialects of Macedonia and
elsewhere (see Ventcel' and Cerenkov 1976 for details), e.g. Arlija buti,
Dambaz buki, Burgudi buci, Gurbet buci; singular buti/plural buka, etc.
Similarly, HI and /n/ can become palatals or lost, e.g. Arlija pani but Dambaz
pai (< *pani). Jusuf and Kepeski (1980) show considerable variation, e.g.
writing both <k> and <kj>, <1> and <lj>, etc. before front vowels in the
same lexical items at different occurrences. Cortiade et al. (1991) articulates
184 VICTOR A. FRIEDMAN
this same principle for velars, but has special graphic symbols for alternating
dentals and velars in their function as case markers (also called postpositions,
see Friedman 1991), viz. 6 and q, respectively. Thus in the orthography of
Cortiade et al. (1991) the same morphophonemic alternations have different
spellings, while the same graphic symbols have different pronunciations, as
illustrated in the following table:
There has also been confusion in prepositions and adverbs, e.g. and-o
"in the" (Cortiade et al. 1991) but and-o "in the" anddro "inside" (Saru
1992).
It is important to emphasize that in the various dialects of Romani the
same phonological changes that effect the dental case endings also effect the
dental participial marker, and, similarly, the same processes affect velars
before front vowels in both roots and grammatical endings in those dialects
with fronting of velars. To this can be added the fact that voicing is
distinctive. The Roms present at the 1992 meeting were unanimous in their
decision to follow morphophonemic practice and spelling using underlying
consonants. Cf. comments on the alphabet above.
'5. In writing the first person singular aorist the final consonant is preserved
according to the root, e.g. kerdo ["done"] > kerdjum ["I did"].'
This is a specific example of point 4, but at the same time it specifies the
Arlija form of the first singular aorist, which can also be -em or -om, with or
without jotation, and is a salient dialectal feature, i.e. one which is taken by
speakers themselves as indicative of dialectal affiliation. In Cortiade et al.
(1991), a hacek is used over a vowel that follows a consonant that is jotated
ROMANI STANDARDIZATION IN MACEDONIA 185
in some dialects but not in others (except in case suffixes, see point 4), e.g.
kerdm.
'6. Where there is jotation, it is written with j ; the letter i is only written as a
vowel:
Romni, Romnie, Romnja, Romnjatar, etc.
["nom. sg., voc. sg., nom. pl. abl. pl."]'
Jusuf and Kepeski (1980) are inconsistent in writing <i> for <j> in final
position in some words, e.g. saj/sai "it is possible", muj/mui "mouth".
Similarly, there has been variation in the spelling of feminine obliques under
the influence of the nominative e.g. abl. Romniatar/Romnjatar for
[romnjatar]. Given that /i/ can contrast with /j/ as indicated in the vocative
singular and nominative plural forms of Romni "(Romani) woman, wife"
cited here, it was agreed that there was a basis for a consistent distinction.
See also point 5 above.
7. In suffixes where Dzambaz and other dialects preserve an older s which
has been lost in Arlija, s is written: devies, devlesa, ["god" acc. sg., instr.
sg.] manges, mangesa, mangas, mangasa ["want", 2 sg. short/long, 1 pl.
short/ long].'
Jusuf and Kepeski (1980) do not address this issue directly but rather
mix forms with and without original /s/ throughout the work. In Cortiade et
al. (1991) this problem is addressed in the instrumental case but not
elsewhere. Thus, for that suffix there is a special grapheme, viz. , but no
prescription for other positions where /s/ alternates with /j/, 0, etc. Hence
devlea, but mangesa or mangea. This was another area of important
compromise for the Roms present at the conference. The fact that the forms
with /s/ are older while those without /s/ represent dialect-specific innovations
gave greater authority to the principle of adopting the compromise.
'8. The instrumental case is always written with s, e.g. mansa ["with me"],
Romensar ["with the Roms"].'
Although /s/ >/c/after /n/ in the instrumental, the alternation is automatic
and so the underlying /s/ is kept in spelling. Moreover, in Arlija and some
other dialects the instrumental plural has a variant in/-r/.Although this is not
etymological, theRomaniparticipants included it.
'9. The personal pronouns are the following:
me, tu, vov, voj [1 sg., 2 sg., 3 sg. m., 3 sg. f.]
amen, turnen, von or ola [1 pl., 2 pl., 3 pl.]'
186 VICTOR A. FRIEDMAN
4. Conclusion
Haugen (1966:16-26) defines four stages in the development of language
planning and standardization: 1) selection of norm, 2) codification of form, 3)
elaboration of function, and 4) acceptance by the community (cf. also Ismalji
ROMANI STANDARDIZATION IN MACEDONIA 187
REFERENCES
Byron, Janet (1985) An overview of language planning achievements among
the Albanians of Yugoslavia. International Journal of the Sociology of
Language 52, 59-92.
Cortiade, M. et al. (1991) I alfabta e standardone Rromane chibaqiri,
Dcizia "I Rromani Alfabta". Informaciaqoro Lil e Rromane Uniaqoro
1-2, 7-8.
Friedman, Victor A. (1985) Problems in the codification of a Standard Ro-
mani Literary Language. In: Grumet, Joanne (ed.) Papers from the
Fourth and Fifth Annual Meetings: Gypsy Lore Society, North Ameri
can Chapter. New York: Gypsy Lore Society. 56-75.
Friedman, Victor A. (1991) Romani nominal inflection: cases or post
positions?. Problemy opisu gramatycznego jzykw sowiaskych
(=Studia gramatyczne, vol. 11). Warsaw: Polish Academy of Sciences.
57-64.
188 VICTOR A. FRIEDMAN
MILENA HBSCHMANNOV
Charles University, Prague
0. Introduction
Romani is the language of a minority subjected to genocidal and
ethnocidal aggression for centuries, a dispersed minority sub-divided into
jati-like (cast-like) subgroups which, like the Indian jatis maintain relations
of social distance. Only the intensive as well as extensive pressure of the
mainstream gadikane societies brings Romani jatis together, at least at the
level of ideology, as expressed by sayings shared by all of them: Rom
Romeha - gado gadeha (Rom with Rom - gado with gado) or Sem
Roma saml (We are Roma!).
Under what political and social conditions did Slovak Romani in the
Czech and Slovak Republics start to develop as a literary language? I would
like to give a short survey of this process and to show how Romani
"behaves" on the pages of periodicals edited by Roma.
The forms of ethnonymic terms will be used on the basis of Romani
grammar: Rom (sg.), Roma (pl.), Romani (as adjective; also stands for the
Romani language), romipen - romhood, Romani tradition, culture etc.. In
Slovak Romani the suffixes -iben, - ipen, -ben, -pen are highly productive,
and they help derive nouns not only from verbs, adjectives, numerals,
adverbs but also from substantives: kher - kheriben (house - dwelling), lav -
laviben (word - way of talking, way of using words), phral - phraVipen
(brother - brotherhood) etc.. In contrast to romanipen used by Mirga (1987),
the Slovak Roma use the expression romipen. For a non-Roma the term
gado (sg.), gade (pl.) is applied; gadikano (adjective). In agreement with
Hancock (1993) the terms "internal/ external differance" between Romani
dialects are used. The term "calque", more common in French or Czech
linguistics, is applied in correspondence with the definition of the Random
House English Dictionary (1973): " ... (translation) resulting from bilingual
190 MILENA HBSCHMANNOV
housing. Most of the ca. 200 000 Roma living in the Czech Republic today
are post-war migrants from Slovakia and/ or their descendants.
The prevalent majority of Roma in Slovakia (the Slovak and Ungrika
Roma) have been settled in outskirts of villages for three or four centuries.
They provided the peasants with blacksmith and musical services; well-
digging, adobe production, basket weaving, seasonal field-work was also
the labour domain of Roma (Hbschmannov 1984). Specific traditional
production and services of Roma became an integral part of the economy of
semi-feudal, agrarian Slovakia. This is perhaps the most important reason
why during World War II, when Slovakia was a pro-Nazi but formally
independent state (March 1939 - May 1945), Roma were not annihilated as
a group. Though sent to camps of forced labour, evicted from villages to
deserted places, not allowed to enter big towns, they survived. The
estimated number of Roma in Slovakia in 1945 amounted to 100 000.
The Slovak Roma are the most numerous group in the Slovak as well as
in the Czech Republic (65 - 70 % of total Romani population). The older
generations used the attribute slovenska/ slovaika (Slovak) to distinguish
themselves from the Ungrika Roma, Vlachi, eventually also other groups
who were not amare Roma (our Roma). Slovak Roma who were born in the
Czech part - many of them have never visited Slovakia - do not call
themselves slovenska any more. Today it is: My sme Romov-Cesi. (We are
Roma-Czechs.) It is quite interesting to note that the attribute serbika
(Serbian), which was quite common among the Slovak Roma four or five
generations ago, has not fallen into oblivion completely, though several
centuries divide the Slovak {serbika!) Roma from their pre-Slovak
transitory home-land - Serbia and/ or Croatia. Rumungre is an appalation
used for the Slovak Roma as well as for the Ungrika Roma by Vlachi,
though on grounds of etymology it should only be used for the latter: Rom
Ungro - Hungarian Rom.
The Humenn-variety (town in Eastern Slovakia) of Slovak Romani has
been described by Lpa (1963). The nominalisation system of Slovak Roma-
ni was analysed by Hbschmannov (1984). The Romani-Czech/ Czech-
Romani dictionary (Hbschmannov et al. 1991) is also based on (East)
Slovak varieties of Romani.
One regional variety of Slovak Romani is spoken in Southern Poland.
However, here its speakers are called Carpathian or Highland Roma
(Bartosz 1981) or Bergitska Roma (Mirga, personal communication). They
share with the Slovak Roma not only the language but also similar tradition-
192 MILENA HBSCHMANNOVA
of life would be conserved, as they would become even more isolated from
other members of the working class."
At this session the policy of strict assimilation was adopted and it was
implemented till November 1989, with a short interval during the "Prague
spring".
2.3. The new wave of assimilation and further attempts at ethnic emanci
pation (1973-1989)
When in March 1973 the Union of Gypsies-Roma was banned, a new
wave of assimilation swept away all ethnic emancipation efforts. Romani
poems piling up in the editing offices of Romano Vil remained unpublished
until 1979, when the Cultural House of Praha 8 consented to publish them
as "auxiliary material for cultural activists". The bilingual booklet Romane
gil'a was issued in 200 copies.
From the beginning of the eighties the echo of Gorbachov's "pere
stroka" was alleviating the pressures of the assimilation policy. Romani re-
194 MILENA HBSCHMANNOV
a - b c hdddd e - f g h c h i - j
a b c d d e f g h c h i j
k kh 1 I' m n o - p ph - r - s s t t th u -
k -- 1 - m n o p - q r s s t t - - u u
v - - - z z
v x y y z z
In Romani the diacritic for prolonging the vowel (rka-acute accent: ' )
is not used since the vocalic length is not distinctive as it is in Czech (plat -
"salary"; plt - "plate"). Acute accent is applied only to mark shortened
forms of the future or imperfect tense: kerava/ ker (I shall do), keravas/
kers (I was doing).
Palatalisation of d, I, n, t is marked only by a ciriklo () (term coined by
M. Courtiade), while in Czech it alternates with / . y in Czech:
a) is a historical element; b) denotes non-palatalised d, n, t. In Romani it has
no function and therefore it was not included in the alphabet.
palatalised non-palatalised
Romani di, l'i, i, ti di, li, ni, ti
Czech di, --,ni, ti dy, -, ny, ty
5.4. Shift from the personal to the reflexive pronoun in verbal constructions
In Czech the reflexive pronoun se, svuj is used where in older Romani
personal pronouns were applied. Alternative use of the older form and the
calque occurs in RLN 74, p. 5: bisterav man u dav andal peste avripeskere
vakeribena (lit. "I forget myself and I speak out my narrations"); RLN 70, p.
3: chude pes andre lachi buti (lit. "get yourself into a good job"). But on the
same page: arakhl'om man andro foros ("I found myself in the town").
There is not sufficient space here to list all of the grammatical categories
in which calquing occurs. I have mentioned only the most conspicuous
ones.
Many Roma do not recognize calques as a foreign element. On the other
hand loan words are more widely felt as improper in the zuzi chib ("pure
200 MILENA HBSCHMANNOVA
7. Conclusion
Out of the four Romani dialects still spoken in the Czech and Slovak
Republics (Czech Romani is extinct), the Slovak Romani is most elabo
rated, though in contrast to the Vlachi Romani it is being forgotten by the
youngest generation. The first attempts at written Romani were published in
Romano L'il, the bulletin of the Union of Gypsies - Roma (1969-1973). The
202 MILENA HBSCHMANNOV
Appendix 1
ROMA: monthly; Romani kultura; Bratislava; Nov. 1990; Slovak; Hungarian (Czech), 10-
20 % R; R Hung, R Vx, R SI; Dezider Banga M.A. (Hung. Rom. poet, former teacher).
ROMANO LIL NEVO: weekly; Association JEKHETANE; Presov ; Sept. 1990 (with
intervals); Slovak (Hungarian, Czech); 30-40 % R; R SI; Anna Koptov (professional Rom
journalist, ex MP), Dana Silanov (Slovak poetess).
AMARO LAV: monthly; Rompress; Brno; Jan. 1991 (under name LACHO LAV since
Jan. 1990); Czech (Slovak); 20-25 % R; R SI; Jan Horvth (Slovak Rom, poet, former
worker).
ROMANO KURKO: weekly; Rompress; Brno; Sept. 1991; Czech (Slovak); in the
beginning no R, now app. 20 %; R Sl (other dialects: Kalderas, Russian R); Jan Horvth
(cf. AL).
Appendix 2
Romsk psne - Romane gil'a, gendi romana poeziatar (Romani songs, a book of Romani
poetry), The Cultural House Praha 8 1979, 57 p.; bilingual, R Sl, SCR orth., 200
copies.
Dobr slovo je jako chleba - Lacho lav sar maro (A good word is like bread; collection of
Romani proverbs), The Cultural House of the Captal Praha 1984, 110 p.; bilingual, R
Sl, SCR orth., 200 copies.
Ceho je na svt nejvc ? - So hin pro svetos jekhbuter ? (What is most numerous in the
world ?; collection of Romani riddles and anecdotes), The Cultural House of the
Captal Praha 1987,110 p.; bilingual, R SI, SCR orth., 500 copies.
Kale rui (Black roses; a selection of Romani poems and short stories by 11 authors), The
Regional Cultural House in Hradec Krlov 1990, 119 p.; bilingual, R Sl, SCR orth.,
2000 copies.
Demeter, Gejza; Mule maskar amende (The spirits of the dead among us), Romani chib,
Praha 1992,16 p.; SI R, SCR orth., 1000 copies.
Reiznerov, Margta; Kal'i (Kal'i), Romani chib, Praha 1992, 20 p.; R SI, SCR orth., 1000
copies.
Rusenko, Arnost; Trin phea (Three sisters), Romani chib, Praha 1992, 13 p.; R SI, SCR
orth., 1000 copies.
204 MILENA HBSCHMANNOV
Ferkov, Ilona; Mosarda peske o divipen anglo love (She spoilt her life for money),
Romani chib, Praha 1992,18 p.; bilingual, R Sl, SCR orth., 1000 copies.
ervek, Stefan; Romani chajori the beng (A Romani girl and the devil), Romani chib,
Praha 1992, 16 p.; R Sl, SCR orth., 1000 copies.
Go daver lava phure Romendar - Moudr slova starych Romu (Wise words of old Roma;
Romani proverbs), Apeiron, Praha 1992,45 p.; bilingual, R Sl, SCR orth., 2000 copies.
Gia, Andrej; Bijav - Svatba (The Wedding), Apeiron and The Cultural Union of Romani
Citizens, Praha 1992, 64 p.; bilingual, R Sl, SCR orth. 2000 copies.
Fabinov, Tera; Cavargos - Tulk (A vagrant), Apeiron, Praha 1992, 68 p.; bilingual,
R Sl, SCR orth., 2000 copies.
Lackov, Elena; Rmske rozprvky - Romane paromisa (Romani tales), Vychodoslovensk
vydavatel'stvo, Kosice 1992, 90 p.; bilingual, R SI, the author's own spelling.
Fabinov, Tera; Sar me phiravas andre kola - Jak jsem chodila do skoly (How I used to
go to school), DO Cesk Budjovice and The Association of Roma in Moravia, Brno
1992, 23 p.; bilingual, R Sl, SCR orth.
Ravasz, Jzsef; Jileskero kheroro - Domcek v srdci - Szvhzik (The house in the heart;
short tales), Romani kultura, Bratislava 1992, 44 p.; trilingual, R Vlax, Slovak,
Hungarian, author's own spelling.
Bun mluvi ke svym dtem - O Del vakerel ke peskere chave (God speaks to his children;
stories from the Bible), Ceska biblick spolecnost and Romani daj, Praha 1992, 183 p.;
bilingual: stories from the Old Testament translated by Vlado Olh into Presov-Slovak
Romani, stories from the New Testament translated by Gejza Demeter into Humenn-
Slovak Romani, SCR orth.
REFERENCES
Bartosz, Adam (1981) Carpathian Gypsies and the rural community. Ethno-
logia Polona 7, 27-33.
Halwachs, Dieter (1994) Romov v Burgenlandu. Romano daniben,
casopis romistickych studii 1:1, 38-41.
Hancock, Ian (1993) The emergence of a Union Dialect of North American
Vlax Romani, and its implications for an international standard. Inter
national Journal of the Sociology of Language 99, 91-104.
Holzinger, Daniel (1993) Das Rmanes. Grammatik und Diskursanalyse
der Sprache der Sinte. Innsbruck: Verlag des Instituts fr Sprachwissen
schaft der Universitt Innsbruck.
Hbschmannov, Milena (1984) Nominalisation in Slovak Romani.
Rassegna della facolta di Lettere e Filosofa dell' Universta di Catania
XIV, 27-70
TRIAL AND ERROR IN WRITTEN ROMANI 205
Anthony P. Grant
Dept of Modern Languages
University of Bradford
BD7 IDP
England