1 

 

The Romani Language: 

Cataloging Ramifications for a Language in the Process of Standardization 

GEOFF HUSIC 

Preprint: Final edited version is published in: Slavic and East European Information Resources. 

Volume 12, Issue 1 (January 2011), pages 37-51: DOI: 10.1080/15228886.2011.556076). 

 

Abstract 

 

A discussion of issues related to the cataloging of a language, Romani (or Romany), which is 

only in the 21
st
 century beginning to achieve some degree of standardization. The discussion focuses on 

issues of Romani orthography, specifically a small number of unusual Unicode characters that may 

cause technical problems in certain automated cataloging environments, such as OCLC WorldCat, the 

OCLC cataloging client Connexion, and online library catalogs. 

 

Keywords: Romani (Romany) language, Cataloging, MARC, Unicode, Numeric Character 

References 

 

Introduction 

 

  The discussion of cataloging foreign language materials has been recently invigorated by a 

number of developments. The ability to add matching non-latin
1
 vernacular fields in bibliographic 

records in online library catalogs for a variety of languages (e.g. Japanese, Arabic, Chinese, Korean, 

Persian, Hebrew, Yiddish, Greek, and some Cyrillic languages
2
, i.e. the so-called JACKPHY 

languages) has somewhat ameliorated the problem of the end-user's lack of acquaintance with the 

romanization schemes used in library catalogs and other databases, as he now has, in some cases, the 

option of searching either in transliteration or in the native script. In addition, especially among the 

languages of the former Soviet Union, there have been changes in the orthographies of several 

languages that have abandoned Cyrillic in favor of romanized scripts (e.g. Moldavian, Azeri, and most 

recently Uzbek), which have certain ramifications for cataloging and retrieval in library catalogs. In 

some cases the change is unproblematic, for example, when abandoning Cyrillic the Moldovans, whose 

language is basically identical to Romanian, just chose to return to writing in standard Romanian, while 

                                                 
Address correspondence to: Geoff Husic, MA, MS, Slavic and Special Languages Librarian, University of Kansas Libraries, 

1425 Jayhawk Blvd, Room 519, Lawrence, KS 66045-7544. E-mail husic@ku.edu.  



 

2 

 

preserving the token Moldovan language designation. In the case of Azeri and Uzbek, complications 

arise, in that the new romanized alphabets diverge somewhat from the ALA romanized transliteration 

schemes for the Cyrillic that have been employed heretofore in library catalogs and many databases. In 

some cases this has resulted in the need to address the Library of Congress guidelines for supplying 

uniform titles in cases of changes in orthography that affect alphanumeric filing and retrieval.
 3

  The 

Slavic and East European Section (SEER) of Association of Research Libraries (ARL) has recently 

convened a taskforce to discuss these and other remaining problematic issues such as mixed 

orthographies, reemergence of pre-Revolutionary orthography in Russian publications, and problems 

with various recensions of Church Slavic. This information will be used to update the widely consulted 

Slavic Cataloging Manual.
4
 In this article I would like to discuss a related but somewhat distinct 

phenomenon, which deals with the ramifications of cataloging and retrieval not of changes in an 

established language, but rather those of a language which is only now beginning to achieve a degree of 

standardization, namely Romani (commonly also spelled „Romany‟, and now sometimes Rromani). 

 Romani, spoken by approximately 4 million Roma
5
 (for a variety of reasons an exact count is 

impossible) in Europe, Asia Minor, and the Americas is considered to be a northern Indic language. The 

great bulk of Romani vocabulary is of Indic origin, and its grammar preserves some archaic Indic 

elements that have been lost in the other modern Indic languages. In a yet to be published article, 

Marcel Courthiade, a preeminent scholar of the Romani language at Institut National des Langues et 

Civilisations Orientales in Paris (INALCO), convincingly traces the original migration of the Roma out 

of the Indian city of Kannauj, in state of Uttar Pradesh, sometime after 1018, as a result of raids on the 

city by the Persian ruler Mahmud of Ghazni (791-1030).
6
 After a lengthy peregrination though Asia 

Minor, the Roma first appeared in Europe in the early 1300s, a portion of the population establishing a 

fairly sedentary life in some areas, while others remained mainly itinerant, travelling freely throughout 

several countries. On its way to Europe, the Romani language was greatly influenced by other Indo-

European languages spoken along the path, such as Persian, Armenian, and Greek. Because of the large 

populations of Roma in the Balkans, Romani has also developed sufficient features shared by the other 

Balkan languages (Serbian/Croatian/Bosnian, Bulgarian, Macedonian, Albanian, Greek, Romanian, and 

Vlax) to be also considered as part of the Balkansprachbund.
7
  As is the case with all languages of 

Europe, Romani has been present in Europe long enough now to develop a number of definable 

dialects, which differ among themselves in points of grammar, vocabulary, and pronunciation, often 

rather substantially.
8
 Until recently, orthographic variations used to write Romani have been utterly 

chaotic, and the most common pattern was to use the orthographic system of the host country, resulting 



 

3 

 

in Romani being written in a variety of non-latin scripts (Cyrillic, Greek, and Arabic) as well as a 

multitude of latin scripts, based on, e.g. Czech, Croatian, Hungarian, Albanian, German, Spanish, and 

English, often with no internal consistency, even in writings by the same author. A survey of Romani in 

Internet websites, chat rooms, and Facebook shows that this situation still obtains to this very day. 

  This discussion is intended to be somewhat forward looking. This author probably has much 

more contact with Romani materials than librarians in most other research libraries. Because of my 

close working relationship with INALCO, the University of Kansas Libraries receive most of the 

Romani-language publications (books, journals, and multi-media) published by this institute and others 

collected by the institute.  Despite the lack of any authoritative cataloging documentation in how to 

deal with this language, as cataloger as well as bibliographer, I have had to develop my own strategies 

for processing Romani-language materials. We must however acknowledge that there has not 

traditionally been a large body of printed literature in the Romani language, which is perhaps why the 

problems associated with Romani in the scholarly and library context have not been raised until this 

point. Until most recently, educated Roma generally have deferred to the languages of their host 

countries for scholarly and educational purposes, whereas among the uneducated and illiterate, oral 

literature and song have been the usual methods of transmitting their history and culture.  The 

environment has change dramatically in the last 20 years, as a cadre of Romani intellectuals, historians, 

and linguists have perceived an urgent need to try to achieve some kind of standardization of the 

European Romani variants, in order to promote education by creating culturally-inclusive instructional 

materials in Romani and to foster and preserve their rich culture. 

  

Warsaw Agreement 

 

In 1990, the 4th World Romani Congress, an umbrella cultural organization that represents 

Roma in 25 countries, convened in Warsaw Poland in order to accomplish two goals: to validate 

Romani as a European language on par with other European languages, and to attempt to achieve a 

degree of standardization, making it a viable vehicle for primary education and publishing.
9
 A team of 

linguists forged a new standard alphabet, quite different from any that had ever been used before. There 

was quite a bit of debate and disagreement between the various team members. Some wished to adhere 

to something closer to the common latin-script usage in those countries with the largest Romani 

populations, i.e. the Balkan countries. Others saw an opportunity to showcase the unique features of the 

Romani language, and they viewed a truly distinctive alphabet (see Appendix for the complete 



 

4 

 

alphabet) as a means to achieve this end. The chosen alphabet exhibits a careful attempt to avoid 

adhering too closely to the orthographic systems of the countries in which the majority of Roma reside, 

and is a rather unfortunate product from my perspective, especially from the point of view of 

cataloging and data retrieval. This is not purely a value judgment on my part, for practical reasons I 

will explain further below. 

  In order to expound on some of the special complications that Romani presents, I must first 

delve into a bit of linguistic minutiae that explain how this new alphabet became endowed with some 

rather exotic features that will be problematic. Romani is an Indic language, descended from Middle 

Indic languages that had already lost the elaborate case and verbal systems we see in the older Indic 

languages such as Sanskrit and Pali. It therefore shares certain syntactic features with other modern 

Indic languages such as Hindi, but has subsequently undergone certain innovations since the Roma left 

the India subcontinent. One salient feature concerns what has been traditionally viewed as a case 

system in Romani. This syntactic feature can indeed be interpreted quite differently when comparing 

Romani to its sister modern Indic languages.  In Hindi, for example, case relationships, such as 

possession, indirect object, and location, are formed by placing a postposition after the modified noun, 

which, in turn, appears in the oblique form (which may or may not change from the direct form), e.g. 

the direct form of the Hindi word for 'boy' is larka, and larke ki ankh means „boy's eye', and „boys‟ 

eyes‟ will be larkoŋ ke ankhe, with the noun changing to the oblique form with the placement of the 

possessive postposition ki. We see a very similar reflex in Romani. The Romani word for boy is 

raklo,
10

 and 'the boy's eye will be rakleski yakh, and 'the boys' eyes' will be raklǎnge yakha
11

.  When 

comparing the Romani structure with Hindi, we can clearly see that what may look like a case ending 

can, in fact, be interpreted as a postposition.
12

 What differs in Romani from Hindi is that the phonetic 

nature of the postpositions in Romani changes based on the consonant to the immediate left. If that 

consonant is voiced, then so will be the consonant of the postposition, and conversely if it is unvoiced 

then so will be the consonant in the postposition. This can be seen in the example above. The singular 

oblique stem of raklo is rakles-, while the plural oblique stem is raklen-. The phonetic form of the 

possessive postposition can thus be either /-ke/ or /-ge/ depending on whether the preceding constant is 

unvoiced or voiced. The same reflex occurs with the postpositions -te „to‟ and -tar „from‟, which 

phonetically become /-de/ and /-dar/ after voiced consonants. Some Romani linguists considered this 

syntactic feature so distinctive and significant that it should be conspicuously accounted for in any 

standardized orthography chosen for Romani. They therefore came up the following solution: rather 

than write the consonant of the postposition with two different consonants, in each instance they would 



 

5 

 

choose a rather obscure grapheme to represent them. The postpositions added to the oblique stem 

would thus be written as below: 

 

Pronounced Written as Meaning 

 

-ke/-ge  -qe  „of‟ 

 

-te/de  -θe  „to‟ 

 

-tar/-dar -θar  „from‟ 

 

-sa/-tsa * -ça  „with‟ 

 

*The sound /ts/ is written Romani with the letter c. 

 

 Thus, the above examples would now be written as raklesqi yakh and raklǎnqe yakha. 

 

In order to try to further distance Romani from the graphical representation of the neighboring 

Slavic languages, the diacritic hachek, i.e. caron (ˇ), was eschewed in favor of the acute accent, thus ć, 

ś, and ź for the aspirant sounds ch (as in Charley), sh (as in ship), and s (as in pleasure) for the aspirant 

sounds. Practically speaking, this is a rather modest modification. However somewhat more 

contentious was the discussion of how to represent the frequent Romani consonant corresponding to the 

j in judge. This particular sound (and closely related sounds) are represented by a wide variety of 

graphemes in the languages of Europe, e.g.: j, g (English), đ, dž, dj (Bosnian/ Croatian/ Serbian), rz, ż 

(Polish)
13

, c (Turkish), gy (Hungarian), and gj, xh (Albanian). Previous to the 2009 Warsaw agreement, 

treatment of this sound varied widely, as it still does in the non-standard Romani orthographies, being 

variously represented as ž, ƶ, ź, as the number 3, and even a very quirky square box minus the left side 

and an acute accent above.
14

 While it would have been logically consistent to choose ź to represent this 

sound, the committee, rather inexplicably, decided to select the obscure International Phonetic 

Association character „ezh‟ ʒ. The ezh, alongside the theta that appears in the Romani postpositions, 

will be problematic from the perspective of cataloging. 

  



 

6 

 

Cataloging Ramifications 

 

The problem of languages that have undergone changes in the official orthography, e.g. Russian 

after the Soviet Revolution, is well known to catalogers in research libraries. However the Roma live in 

a number of countries and therefore are not able to claim to have any governmentally sanctioned 

“official orthography.” Thus, there is some uncertainty whether the AACR2 stipulation about changes 

in official orthography is applicable, or even desirable in the cataloging of Romani materials. 

Nevertheless, as more and more materials are being published in Romani, many, but not all adhering to 

the new standardized orthography, it is important to consider having a systematic approach to 

cataloging materials in the Romani language. 

AACR2 cataloging records consist of two main components, the descriptive elements, 

transcribed from the item being cataloged, and additional controlled metadata elements assigned by the 

cataloger, such as names and subject headings, which are subject to authority control. In the context of 

Romani, the latter are generally unproblematic, as authors‟ names are usually established based on the 

orthographies of the countries in which they reside, and therefore do not normally contain the 

problematic characters ezh and theta
15

. The descriptive elements of the bibliographic record, on the 

other hand, present very little flexibility. AACR2 clearly stipulates that the descriptive elements should 

ideally be transcribed exactly as they appear in the item being cataloged, although with further caveats. 

Until recently, for languages in non-latin scripts the only option in bibliographic utilities, such as 

OCLC, was a romanized transliteration scheme, which at least for North American libraries, was based 

on the ALA norm.
16

 While we are now able to add matching vernacular fields for many scripts in 

OCLC WorldCat, this is not the case for all valid Unicode characters.
17

 In addition, AACR2 stipulates 

some rather elaborate rules about how to deal with special characters such as ezh and theta, which 

seem inappropriate to apply in the case of a language for which these characters are part of the 

established orthography.
18

 We are therefore left with a dilemma when it comes to Romani. 

 

The Problematic Characters 

 

In the context of 1990, when the new alphabet was adopted, computer automation and the 

Internet were not as developed as they are today, and therefore the Romani linguists did not see the 

introduction of new exotic letters as unduly problematic, as typewriters could be ordered with specific 

keys as needed. However in the current computer environment these characters present a particular 



 

7 

 

problem. OCLC WorldCat is still limited to the smaller Marc-8 Unicode subset and limits what 

characters can be entered in certain contexts. While the vast majority of letters of the new Romani 

alphabet are MARC-8 compatible, the ezh is a character that is still unrecognized in that character set, 

and so we must replace this character with some alternative.
19

 The introduction of the Greek character 

theta (θ) also presents some distinct problems for cataloging. This character is disallowed in OCLC 

cataloging in otherwise latin-script fields and is generally replaced with the bracketed name of the 

letter, i.e. [theta], or a phonetic equivalent. If we do attempt to input this character in OCLC Connexion 

(other than in a matching vernacular Greek field) it will force Connexion to assume that the record 

being input is a non-latin script record and will automatically code the resulting record as such. This 

code is system supplied and cannot be removed. Connexion will also attempt to add matching fields for 

romanization, as we see in OCLC WorldCat records for Cyrillic, Arabic, etc., which cannot be left 

blank and must be filled with text of some sort in order to validate and add the record to OCLC 

WorldCat. 

Alongside these problematic consonants, the new orthography contains a number of vowel 

graphemes ǎ, ě, ǐ, ǒ, ǔ. These so-called pre-jotizing vowels are pronounced with a faint “y” sound after 

the consonant preceding them.  From the descriptive aspect of cataloging these characters are not 

unduly problematic and are compatible in OCLC WorldCat. The most important thing to bear in mind 

is that these diacritics are the caron, and not the breve, which is more commonly seen above vowels in 

European languages. 

In addressing the descriptive aspect of the bibliographic record, how then do we deal with the 

theta and ezh? Computer alphabetical sorting and keyword searching is normally good at regularizing 

variants of roman-alphabet letters occurring with various diacritics, so that, e.g. z, ź, ż, ž, etc., will all 

be retrieved when searching the stripped z. The theta and ezh, however, are not characters for which 

such a correlation is normally encoded. As far at the descriptive elements of Romani bibliographic 

records that contain these problematic elements are concerned, we must turn to practical solutions. 

There seem to me to be two acceptable solutions for dealing with the ezh, and, for the sake of 

consistency, I believe it would be prudent to treat the theta as well as the ezh in the same fashion when 

cataloging in OCLC WorldCat.  The approaches I will suggest below, while admittedly not extremely 

aesthetic, will nevertheless assure that important graphical information will be retained until catalogs 

and bibliographic utilities are compatible with all Unicode characters. For the time being, it is likely 

that these steps will be mainly necessary in the descriptive areas of the bibliographic record and will 

mostly affect the transcription of the titles. 



 

8 

 

 

Option 1:  

 

Except for the theta and ezh all other Romani letters have ALA character equivalents and can 

be transcribed as they are. The ezh has two acceptable variants according to the Warsaw Agreement, ƶ 

and з (the Cyrillic /z/). However, neither of these is a valid ALA character, and therefore cannot be used 

in OCLC WorldCat to transcribe the title proper (MARC 245 field).  Although not commonly seen in 

OCLC cataloging, there is a provision in the Library of Congress Rule Interpretation documentation for 

replacing special characters that cannot be used in the MARC record. According to this list, the ezh, 

which can also appear in some African languages, should be transcribed as z ̳(z with double underscore) 

and theta as t ̳(t with double underscore). Using this substitution for ezh, Kote ʒàna e Kosoviaqe 

Rroma would be transcribed as:  Kote zà̳na e Kosoviaqe Rroma. Since titles are generally not 

unwieldy, I believe it would also be helpful to provide a note, such as “z ̳ in the word z ̳ana appears as the 

IPA symbol ezh.” Such notes will be helpful to users unfamiliar with the vagaries of transcribing 

unusual characters in OCLC cataloging. 

 

The use of  t ̳ for theta seems a little less satisfactory, since this letter has two phonetic 

outcomes, /t/ and /d/ depending on the preceding consonant. Nevertheless, it makes sense to consider 

employing this substitute character. In the case of theta there will also be cases where added title 

entries will be helpful. Regrettably, the treatment of theta will require a bit of knowledge of Romani 

phonology as revealed in the discussion of postpositions above. However, with this knowledge in hand, 

when constructing added title entries (MARC 246 field), I would recommend, in cases where the theta 

is indeed pronounced as the voiced d, that the theta be transcribed according to its phonetic value, i.e. 

d or t depending on whether the preceding consonant is voiced or unvoiced respectively, e.g. Sar me 

vastesθar xutǐlav tu would be transcribed as Sar me vastestar xutǐlav tu, while But lenθar 

ʒivdinèna would be transcribed as But lendar zi̳vdinèna.  

 

Option 2: 

 

The second option involves the use of a so-called “lossless” solution. Such a proposal for 



 

9 

 

dealing with problematic characters in MARC cataloging records already exists in the MARC 

documentation.
20

 Although accepted in 2006, its use with characters that fall outside the MARC-8 

character repertoire has not be widely used in OCLC cataloging, the reasons for which might become 

evident from the examples below.  In short, this lossless method recommends replacing a Unicode 

character that cannot be mapped to MARC-8 with a placeholder that contains the Numeric Character 

Reference (NCR) for the character. The NCR consists of the hexidecimal representation of the code 

point of the character (four ASCII characters), preceded by #x and all surrounded by & and ; 

 

Using this method yields the following:  

 

The lower case theta can be substituted with θ and the capital letter as Θ  

 

The lower case ezh will be ʒ and the upper case &#x01B7  

 

There are advantages and disadvantages to each of these options.  

 

The first option will allow for searching and alphanumeric filing based on a more natural 

representation of what a speaker of Romani, or one researching Romani might expect, in the case of the 

ezh because z ̳will regularize out with the many forms of the letter z frequently used to represent this 

sound in the various common manifestations of the Romani alphabet. However it is not ideal in this 

respect, as diagraphs, such as dž, dj, xh are also frequently used to represent this sound in Romani 

texts. In regards to the theta, this approach is somewhat better suited, as it represents a form used by 

the majority of Romani writers, many of which still shun the use of the theta. On the other hand, the 

second option will allow for eventual automatic conversion of the characters back to the proper graphic 

representation when these Unicode characters become available in the MARC format. Some newer web 

interfaces are already able to render these characters correctly when they are manually input in the local 

catalog database, even if it is not yet possible to input these non-MARC-8 characters in OCLC 

WorldCat directly.
21

 Nevertheless there are obvious drawbacks to this approach as well. First, since 

these NCRs are mostly likely to occur conspicuously in the title of the bibliographic record, any user, 

other than those familiar with cataloging limitations and hexidecimal coding of Unicode characters, 

will likely assume that a coding error has occurred, resulting in garbled characters.
22

  Second, this 

method will profoundly affect keyword searching and alphanumeric sorting of main titles. Finally, even 



 

10 

 

if the characters are rendered correctly in the catalog and browser, another unfortunate result is that the 

catalog user may not have the ability to input the correct characters for the purpose of searching. 

 If the second option is chosen, as in the case of the first option, it will still be desirable to 

provide some alternate title headings to the bibliographic record, based on a reasonable assumption of 

what the user might expect. This is clearly not a trivial issue. As in the case of cataloging all foreign 

language materials, the cataloger must have at an adequate working knowledge of the language being 

cataloged in order to be able to discern which added entries are most appropriate and useful for the end 

user.  In the Appendix I discuss my personal preference for treating these problematic characters but I 

feel both approaches are equally legitimate. 

 

Authority Control Implications 

 

Turning now to the portion of the bibliographic record subject to authority control, i.e. the 

assignment of controlled vocabulary and standardized names, we are left with a further dilemma.
23

 

Many Romani writers are either unaware of the standardized alphabet adopted by the 1990 World 

Romani Congress or find the system so bizarre that they simply choose to disregard it, opting for the 

orthographic system of their home country or an idiosyncratic one of their own design. However, as a 

cataloger of these materials, I must make a decision, one way or another, how to approach this issue. As 

a linguist, I recognize that Romani is very different from most other European languages in that it is 

represented by no nation state and therefore has no governmentally-sanctioned official orthography, 

and is represented by a wide variety of dialects. However as a result of my experience as a cataloger I 

recognize the importance of collocation in the context of the library catalog and strive to find a means 

to make Romani materials as accessible as possible. This raises the question of whether it is desirable 

to accept the 1990 Romani Congress decision as authoritative enough to apply the AACR2 guidelines 

concerning change in orthography to Romani. I would argue that this is indeed desirable, at least in 

certain cases where possible, and where there a desire to bring together related manifestations of a 

specific work.
24

 

One problem, until very recently, has been the lack of any authoritative reference source to 

resolve questions of orthography for Romani. Such reference sources are frequently consulted by 

foreign-language catalogers, for example, when cataloging Arabic to ensure a consistent transcription 

of the short vowels, which are normally not written. In 2009, a dictionary, of which I was coeditor of 

the English content, was published and which may be very helpful in resolving problems. This 



 

11 

 

UNESCO sponsored dictionary is the first international dictionary of the proposed standardized form of 

the Romani and it employs the new orthography discussed above. 
25

 

While it is difficult at this stage of Romani publication to point to many examples to support 

this argument, I can present at least one real-life example where assigning uniform titles based on the 

current form of the orthography can bring together related works. The Romani scholar, Rajko Đurić, 

has published many books on the topic of the Roma, several in Romani. His earlier Romani 

publications were written in an orthography that mirrored Serbian latin-script orthography, whereas he 

later chose to adopt the 1990 Romani Congress orthography. Đurić published two related books 

treating the history of the Roma, the first, in 1988 under the title Bibahtale breša [Unhappy years]. 

Then in, 1996, he published a companion volume, using the new orthography, under the title: Bibaxtale 

berśa].
 26

  Clearly these works will regularize differently in a title browse and a keyword search. In 

cases such as this, it seems that it would be prudent to provide uniform titles (MARC 130 or 240 fields) 

based on the new orthography on the bibliographic record for the title in the old orthography, in order 

to provide access to the new established orthography. An authoritative source, such as the dictionary 

mentioned above, can facilitate in determining the appropriate form.   

 Other than examples such as above, where there is a desire to bring together related 

manifestations of a work, I fear that it may impractical to try to routinely construct uniform titles for 

Romani works for a number of practical reasons. First, the new orthography is only a recommended 

standard, with no legal weight. While being used by several Romani organizations and scholars in 

Europe, others may either totally disregard it, or use it only partially (especially avoiding the q, ç, ʒ, 

and θ). While above I have concentrated on the problem of the consonants, similar problems exist with 

the new Romani vowel characters. To give just an example for ǎ and ǒ, these have been written as ja, 

ia, ya, я and yo, io, yo, ё respectively in various kinds of orthographies. While books being printed by 

publishers that specialize in Romani materials may be more likely to adhere to the new orthography, 

there will continue to be a great variety in how Romani is written in books and journals, the Internet, 

etc. The burden will therefore be on the cataloger to attempt to construct uniform titles for a language 

that is generally little known outside the Romani community and among Balkanologists. 

 

Conclusion 

 

As a non-territorial and transnational language, it is very difficult to envision a time in the near 

future when even European variants of Romani will achieve the degree of standardization of other 



 

12 

 

European languages. Nevertheless, considering that there are now UN-affiliated and other cultural 

groups that are attempting to promote the new standard orthography and are introducing it more and 

more in printed publications and Internet resources, it should be approached by the cataloger with the 

same degree of attention that we pay to other complicated issues of orthography. The goal of our 

diligence is always to keep the potential end user in mind, and to anticipate how in the future 

automated algorithms may be able to bring together related works, as envisioned in FRBR models
27

. 

Even if we conclude that certain issues of authority control, such as assigning uniform titles, are too 

complicated or ambiguous to be applied systematically, we are nevertheless left with the problematic 

characters that presently prevent their exact transcription in some cataloging environments. It is my 

hope that this discussion will be of help to other catalogers who may be presented with Romani 

materials to understand the complicated cataloging issues involving this fascinating language. Please 

feel free to contact me at husic@ku.edu if you should need any assistance in cataloging Romani 

materials.

mailto:husic@ku.edu


 

13 

 

Appendix: The Unified Romani Alphabet (Based on the 1990 Warsaw Agreement) and Summary 

of Suggested Cataloging Approaches 

 

Most Romani letters are pronounced very similarly to Croatian. The exceptions are:  ç (as s or 

ts, q (as k or g), and θ (as t or d); x (as the Russian х);  kh, ph, and th (aspirated k, p, and t), rr ( 

uvular r), and ʒ (as the j in jam.) The letters ć, ś, ź are very close the Croatian č, š, and ž in most 

dialects. 

 

Unproblematic characters: The following letters (in Romani order) present no special problems to the 

cataloger in any cataloging environment. The only possible point of confusion is whether rr 

(pronounced as a uvular sounds, such as r in French Paris), when initial, is written as Rr or RR and is 

of little concern, although Rr is preferred. 

 

a, b, c, ç, d, e, f, g, h, x, i, j, k, kh
28

, l, m, n, o, p, ph, r, rr, s, t, th, u, v, z. 

 

Vowels with the acute (representing unpredictable stress: à, è, ì, ò, ù) 

 

Possibly problematic characters: The pre-jotizing vowels, i.e. vowels with a caron (hachek) written 

above. For bibliographic utilities such as OCLC Connexion these characters no longer present a 

problem. Each vowel can be input with the necessary caron above. Library catalogs may display them 

correctly or strip out the diacritics, depending on the system, but searching and filing will not usually 

be affected. Nevertheless, it behooves the cataloger to input these letters with the proper caron, rather 

than to accidentally substitute the diacritic breve. 

 

ǎ, ě, ǐ, ǒ, ǔ 

 

The Latin characters with the acute diacritic: The characters representing these letters are available 

in OCLC Connexion and should search and display correctly in any Unicode-compliant library catalog. 

The characters may be downgraded when imported into some bibliographic utilities, such as Endnote, 

to their stripped forms c, s, z, or may be replaced with blank or filler characters, depending on how 

character importing is set. However this particular problem is beyond the scope of this article. 

 



 

14 

 

ć, ś, ź  

 

The Postpositions q and ç, 

 

 In the examples with the ezh and theta, I have recommended making title added entries in 

certain cases where problematic characters occur in a title proper. I would also recommend constructing 

such added entries in the cases of the characters, q and ç, used in the postpositions, based on their 

phonetic outcome, i.e. k/g and s/c. Again, while requiring a bit of knowledge of Romani phonology, 

these added entries will aid a user by providing keyword searching based on a more natural form. 

 

Problematic Characters and their Numerical Character Reference Codes. These characters may 

present considerable problems for the reasons discussed above. 

 

Theta θ and ʒ 

 

Although visually unaesthetic, my preferred recommendation for transcribing these characters 

in OCLC is to use the lossless method discussed above by replacing these characters with their 

appropriate Numerical Character References, i.e. θ (θ) and Θ (Θ) ʒ (ʒ) and 

&#x01B7  (Ʒ). I am hopeful that in the near future we will have all Unicode character sets available to 

us in MARC cataloging, and therefore this problematic situation should not last indefinitely. Until then 

some catalogs can already display these characters correctly when encoded in this way. 

 If this solution is locally considered to be unacceptable, then the use of the t̳ and z ̳(t and z with 

double underscore) is a good pragmatic alternative. At some future point these replacement characters 

could be searched and replaced in the database with the proper Unicode characters. 

 

 One further note: if attempting to search OCLC WorldCat for examples of records for items in 

Romani you will discover that a large number of records have the MARC fixed-field language code for 

Romanian (rum) miscoded as rom, which is in fact the language code for Romani. This will result in 

many false hits when attempting to filter by language. 

 

Romani Samples on the WWW 



 

15 

 

 

Below I have listed a few examples “in the wild” Internet sites that use the new orthography or a 

permutation of it, that may interest the reader. 

  

A website of news from Kosova, using the new orthography in totality:  

 

http://rroma.courriers.info/spip.php?article394 

 

A website devoted to Romani rights. Some articles are in the new orthography, which others are in a 

variety of the idiosyncratic variants: 

 

http://romarights.wordpress.com/category/nevipenewsvijesti/nevipe/page/3/  

 

Website of the European Roma Rights Center. Except for the notation of the aspirated consonants, i.e. 

kh, ph, and th, the orthography is the same as Croatian. 

 

 http://www.errc.org/cikk.php?cikk=2255 

  

http://rroma.courriers.info/spip.php?article394
http://romarights.wordpress.com/category/nevipenewsvijesti/nevipe/page/3/
http://www.errc.org/cikk.php?cikk=2255


 

16 

 

 

NOTES 

 

                                                 
1 . In cataloging literature and documentation, the terms latin and roman are often used interchangeably when speaking of 

scripts. However the process of transliterating a non-latin script into latin script is almost always called “romanization.” 

  

2. The complete character sets for certain Central Asian and other Cyrillic languages are not yet available in OCLC. Further 

information on available character sets can be found at: 

http://www.oclc.org/support/documentation/worldcat/records/subscription/4/4.pdf 

3. Library of Congress Rule Interpretation based on Anglo-American Cataloging Rules. 2nd ed., 2002 Revision. (Chicago: 

American  Library Association), 2002 via Catalogers Desktop: http://desktop.loc.gov:  „Section 25.3A‟:  “For monographs, 

on the bibliographic record for any edition of a work whose title proper contains a word in the old orthography provide a 

uniform title reflecting the new orthography, although no edition with the reformed orthography has been received.” 

4.  Slavic Cataloging Manual (http://www.indiana.edu/~libslav/slavcatman/smtocs.html) 

 

5. The term „Gypsy‟ is considered pejorative and is no longer used in scholarly writing.  In English both Roma and 

Romanies are used for the plural. 

6.  This information has been included with the kind permission of Prof. Courthiade. 

 

7.  A linguistic concept that examines the syntactic, morphological, and lexicological commonalities that have arisen in the 

territorially contiguous, but linguistically quite different Indo-European language groups spoken in the Balkans. 

 

8. The Romani dialects are broadly grouped into Vlax, Balkan, Carpathian, and Sinti. 

 

9. Courthiade, Marcel, ed. Morri angluni rromane ćhibǎqi evroputni lavustik. Budapest: Fővárosi Onkormányzat Cigány 

Ház-- Romano Kher, 2009, pg. 496-497. 

 

10.  Both raklo and yakh are cognate with the Hindi larka and ankh. 

 

11.  I am intentionally not using the new standardized alphabet in these examples, as it would obscure my argument. 

 

12. Victor Friedman. “Case in Romani: old grammar, new affixes.  Journal of the Gypsy Lore Society, 5th ser., vol. 1, no. 2 

(1991, pg. 85-102.  

 

13. In fact in standard Bosnian/Croatian/Serbian and Polish there are two distinct “zh” phonemes, one more palatalized than 

http://desktop.loc.gov/
http://www.indiana.edu/~libslav/slavcatman/smtocs.html


 

17 

 

                                                                                                                                                                        
the other. In Romani there is generally just one such phoneme, although there is a great variety in the actually pronunciation 

of the phoneme. 

 

14. Courthiade, Marcel. Gramatika e gjuhës rrome. Tirana: [s.n.], 1989, pg. 19. 

 

15. Since theta only exists in the postposition, it of course cannot occur in the non-oblique form of a name. 

 

16. http://www.loc.gov/catdir/cpso/roman.html 

 

17.  As of this writing OCLC Worldcat is limited to the Marc-8 character set, plus four additional scripts (Thai, Bengali, 

Devanagari, and Tamil), which use the UTF-8 character sets. 

 

18.  Anglo-American Cataloging Rules. 2nd ed., 2002 Revision. (Chicago: American Library Association), 2002  via 

Catalogers Desktop: http://desktop.loc.gov: „Section 1.0.E:  Language and script of the description.‟ 

 

19.  In order to not become overly technical in this article and to concentrate on the current cataloging implications, I have 

chosen not to go into specific details about the variety of MARC formats and ALA, MARC-8, and Unicode character sets, 

as they are out of scope. 

 

20. http://www.loc.gov/marc/marbi/2006/2006-09.html 

 

21. This is the case in, e.g., with the Voyager catalog. 

 

22.  An example can be seen on the OCLC WorldCat record # 233230871 (an example with Mongolian and Kazakh in 

Cyrillic script). 

 

23.  An excellent explanation of authority control in the context of bibliographic records can be found at: 

http://en.wikipedia.org/wiki/Authority_control 

 

24.  The new requirements for how related manifestations are to be handled will be more clear when RDA, the cataloging 

code that is to replace AACR2, is published  in 2010, but this is not likely to greatly affect the suggestions made in this 

article. 

 

25. Courthiade, Marcel, ed. Morri angluni rromane ćhibǎqi evroputni lavustik. Budapest: Fővárosi Onkormányzat Cigány 

Ház-- Romano Kher, 2009. This dictionary has not yet been widely circulated, but questions concerning its availability can 

be directed to: Mr Zsigo Jenö, Roma Parlament, 1084 Budapest, Tavaszmezö ut 6. (Roma.parlament@chello.hu) or to Mr 

Lakatos Laszlo (lakatoslaszlo@frokk.hu). 

 

http://www.loc.gov/catdir/cpso/roman.html
http://desktop.loc.gov/
http://www.loc.gov/marc/marbi/2006/2006-09.html
http://en.wikipedia.org/wiki/Authority_control
mailto:Roma.parlament@chello.hu
mailto:lakatoslaszlo@frokk.hu


 

18 

 

                                                                                                                                                                        
26.  Admittedly this is an imperfect example, as these two items, although they share the same main title, are not variant 

editions, but different books covering different time periods. The discrepancy between breša and berśa is not a typo. It is a 

dialect variation for this word. 

 

27.  FRBR (Function Requirements for Bibliographic Records) is an entity-relationship model for displaying related works 

in a catalog in a more holistic manner.  

 

28.  The h after another consonant indicates aspiration. These represent separate phonemes and are thus also considered 

separate digraph letters.