CREATIVE STANDARDISATION

While people argued vehemently for their own, Prizren orthography, in practice many mix this Romani orthography with other (primarily Albanian) orthography, with little consistency. A sound may be spelled differently by the same person, sometimes within the same text or even word. When I first arrived, I asked how to say ‘pleased to meet you’; Nexhip wrote avlama shukar so pengjargjum tut. But when I read it out, I was told I had pronounced it incorrectly: in pengjargjum (Ro: ‘I met’), the first <gj> represents /ʤ/, using the Albanian <gj>, which in Kosovar Gheg Albanian (unlike the Tosk-based standard) is not differentiated from <xh>; by contrast, the second <gj> represents /Ɉ/ or /gj/. Such orthographic mixing did not cause problems for native speakers, who learn the spelling after learning the word (indeed, no one noticed these inconsistencies), but they were still very vocal in support of ‘their’ orthography.

Many Prizrenites (both Roma and non-Roma) also write like this in Albanian. Shopfronts often have a name spelled in two different ways. This is sometimes due to mixing Albanian orthography with Turkish or BCS, and sometimes to an approximation of local Gheg forms in Albanian orthography, rather than the Tosk-based standard. Equally, although Nexhip had completed a university degree using Bosnian as the language of instruction, he would refer in written communication to the language as boshnjaqki, BCS for ‘Bosniak language’, written with Kosovar Albanian orthography. This was not mere carelessness: he, and others, routinely wrote in this unpredictable, spontaneous way in both formal and informal texts. This suggests that, at least at the level of orthography, consistency of form is not considered central to the standard. Orthographic choices are important, but they need not be consistent.

Literature on online Romani use from across Europe has highlighted the sporadic and creative use of writing, which is marked by consistency of dialect features alongside creative pluralism in orthographic representation (Leggio Reference Leggio2011; Leggio & Matras Reference Leggio and Matras2013; cf. Halwachs Reference Halwachs2012a). In Prizren, the sporadic, creative use of orthography is not limited to online communication, nor to Romani. While in some languages, the internet as a new domain of language use contrasts with well-established standards of consistency and propriety, with Romani (and to a lesser extent Albanian), there is no such contrast. Informal writing practices in Prizren are not an escape from an older elite idea of written language. Rather, they form part of the contemporary need to differentiate Romani writing from alphabets that look Slavic, and to construct a writing system parallel to, yet distinct from, Albanian, as the main dominant language. Thus, the regimenting of Romani to match the uses of the dominant standard language (Gal Reference Gal2006) lacks emphasis on orthographic consistency. It is necessary to show an allegiance with Albanians, as the dominant ethnic group, by avoiding and purging any visible Serbian elements. In practice, then, use of the Romani standard is both creative and contingent, although the idea of a standard orthography is ever-present.

The same goes for lexicon. One of the daily tasks at the radio was to translate news from Albanian to Romani to be read on air. Once, when an NGO journalist got to an Albanian word she didn't know in Romani, i papranueshëm (Shq: ‘unacceptable’), she looked it up in an Albanian-Romani dictionary written by Kujtim Paqaku, a local Roma poet and intellectual. It wasn't there, so she asked Nexhip, the director. He didn't know, so typed pranon (Shq: ‘accept’) into Google Translate to get the English accept. Not knowing English, he said something like ats-tsept, to which he added the Romani suffix -imi to make ‘acceptable’, and the prefix bi- to make it negative, giving biacceptimi. He told me that when the journalists didn't know the ‘real’ Romani word, it was better to make Romani words from ‘international’ words than to use words from surrounding languages. ‘International’ words almost always referred to loans from English, the international language, used by the international community. Such spontaneous techniques of borrowing were common.

As local figures of authority on Romani language, Nexhip and Kujtim often debated language issues. Once, Kujtim told me not to say beginava (Terzimahalla Romani: ‘I like’) since it was korohani (Terzimahalla Romani: ‘Turkish’)—from beğenmek (Tu: ‘to like’). Kujtim suggested man plaqe instead, but Nexhip thought this sounded too much like the Albanian më pëlqen. In fact, man plaqe is from Romanian, and Kujtim probably learned it through his involvement with Romani activists outside Kosovo, some of whom speak Vlax Romani, which has borrowings from Romanian but is often deemed ‘pure’ in international circles. However, to Nexhip it sounded too Albanian.

Then they discussed how to say ‘be careful’. I had asked Nexhip how to say pazi (BCS: ‘be careful’) a few days before; unsure, he had responded: Mi kažemo ‘javaš’, ama to je turska reč. Nije naša reš (BCS: ‘We say yavaş, but that's a Turkish word. It's not our word.’). The primary meaning of yavaş is ‘slowly’ in Turkish, but it can be used in any language in Prizren to mean ‘be careful’, or ‘go slowly’. Nexhip then suggested as a Romani term dikkat. Dikkat in Turkish means ‘attention’, but it resembles dikh, the Romani word for ‘look’. It would therefore mean something like ‘look out’. Nexhip thought it would work, as it is recognisable, from Turkish, and sounds Romani. But Kujtim again rejected it as Turkish. He was clearly unimpressed with Nexhip's attempts at making standard Romani words, and felt he had more authority. Kujtim turned to me and said didactically, using BCS to ensure I understood: Lingvistika se ne uči u kadi, ‘you can't study linguistics in the bath’. Noticing my confusion, he explained that you learn about languages by doing research in the field, by going to the forests of Albania and finding isolated groups speaking ‘pure’ forms of Romani. He had conducted research in the field, not alone in the bath, so he was more qualified to make new words.

Numerous words used in Terzimahalla are deemed ‘foreign’ by NGO staff, or ‘not ours’, and therefore incorrect. For example, gindinava ‘I think’ (from Romanian), is preferred to mislinava, which is from BCS misli- ‘think’; shajnava ‘I can’ is preferred to mozhinava, from BCS mog- ‘can’. Similarly, words perceived as Turkish-origin are considered non-Romani: jeftisar ‘sorry’ was preferred to afikya (from Turkish affi ‘forgive’, with the Romani imperative ker ‘do’). On several occasions, I learned a word used in everyday speech, recognisably of BCS or Turkish origin, and was told instead to use a Romanian-origin word. For example, I was told that numay (from Romanian numai ‘only’) should be used for ‘but’ rather than ama, which is from Turkish but has also spread into other Balkan languages.

Though the specifics of lexicon and orthography were contested, the aim of standardisation was not: everyone wanted the purest, most authentic Romani words from which to construct a standard language. If I asked why a certain word or spelling should be used, I was told either that it was gramatika (‘grammar’ in BCS, Shq, and Ro), a term for an implied standard or proper way of speaking, or simply that this was may laqhi or may shuzhi (Ro, standard: ‘the most beautiful’ or ‘the most pure’). The use of may to form a superlative is itself a standard form, from Romanian mai; the usual Terzimahalla form is po laqhi/po shuzhi, from the Slavic (for a discussion of borrowings of comparatives and superlatives, see Matras Reference Matras2002; cf. Friedman Reference Friedman1985b). Purity and beauty are not clearly delineated and are based on subjective judgements, but subjective judgements made within a particular social setting. When Prizren Roma talk about purity and beauty, they mean what sounds least like the other languages they know. One of the tasks of those involved in standardisation is thus to identify, and purge, foreign-sounding words.

Hence the emphasis on purging Romani of non-Romani words focused on othering words from local languages. Romanian-origin words learned by intellectuals from Vlax Romani were acceptable because they were not recognisably foreign—most people in Prizren do not know Romanian. The rejection not simply of alien words, but of words from the surrounding languages in particular, is apparent with man plaqe, which Nexhip thought sounded too like Albanian. Similarly, biacceptimi is acceptable despite its obvious foreign origin. International words can become Romani, and as they are less familiar in the Prizren context are interpreted as more authentically Romani.

Purism is directed against the perceived lexical incursions of a dominant neighbouring language. With more than one dominant neighbouring language, the need for differentiation from different neighbours can lead to a lack of solidarity when it comes to standardisation on a wider scale, since ‘purists will not agree on the targets for puristic censorship and intervention’ (Thomas Reference Thomas1991:126). Here, rather than contesting which local language needs purging, all local loans are targeted, while words from further afield are accepted and encouraged.

Reactions against loanwords are historically situated. Mixed codes are products and symbols of domination, making purification an act of resistance (Jaffe Reference Jaffe1999). Since foreign-origin words in Romani represent the incursion and domination of foreign elements, people seek to purge them. The commonly used word anglaynava (Terzimahalla Romani: ‘I understand’), from Turkish anlamak, is thus considered impure. But in Prizren, it has been replaced not by the more international Romani haqarava (or other cognate terms), but instead by lava khan, calqued from the Albanian marr vesh ‘I agree, I understand’; marr, like lava, means ‘take’, and vesh, like khan, means ‘ear’. Standardisation often relies on calquing to replace loanwords (Thomas Reference Thomas1991; cf. Hill & Hill Reference Hill and Hill1980), though lexical enrichment sometimes also relies on loans from the dominant language (cf. Wexler Reference Wexler, Jernudd and Shapiro1989). The use of a calque from Albanian, rather than a borrowing from Turkish, or indeed a more international Romani form, shows the importance of correlating with, yet distinguishing from, Albanian. Thomas (Reference Thomas1991) argues that separation, solidarity, and prestige are central to language standardisation. Despite the use of international Romani forms, I suggest that here the relationship with local languages is more important than solidarity with a wider Romani speech community. Rather than looking to international work on Romani standardisation, they focus on the affinity with and distinction from local languages.

Local standardisation processes have different priorities from larger-scale processes. The preference for words not loaned from local languages is sometimes explained in terms of comprehensibility in neighbouring countries. When a colleague explained that I should use hramina rather than pishina (from BCS pisati) for ‘we write’, he did not appeal to gramatika, but said that “everyone [all Roma], even if they don't know Serbian, will understand this; people in Macedonia, for example, will understand this”. However, international communication is rarely cited as a priority. In fact, as seen in the discussion on orthography, it is more important to show allegiance to those in power (the Albanian authorities) than to ally with Roma from other countries.

Loanwords are deemed necessary to compensate for the ‘poverty’ of Romani vocabulary. For Nexhip, the fact that the single Romani word Rom could signify three different words in BCS—čovek ‘man’, muž ‘husband’, and narod ‘nation’—was a sign of the poverty of Romani. Similarly, he once claimed that the verb fulela had three meanings: fulava means ‘I'm going down’, fulavava means ‘I take off’, and fulinava means ‘I throw’. Despite the different derivational infixes, he took the common root as an example of the language's poverty—a quantitative lack of words.

By contrast, Kujtim would extoll the richness of Romani, listing, for example, the words for egg in three different dialects, all of which are etymologically identical, and citing a high rate of synonymy in Romani. From his perspective, the problem was not lexical poverty but the ignorance of many Romani speakers. While dialect variation is often considered an insuperable problem for language standardisation (Friedman Reference Friedman, Hyltenstam and Fraurud2002), for Kujtim it evidenced the strength and richness of Romani. Purity is valued, but so is richness. If language is understood as a collection of words, richness is understood as the sum of these words, excluding words deemed foreign.

Debates over standard Romani are couched in terms of correctness, rather than consistency or uniformity, and relate primarily to words and orthography rather than to other aspects of language. While people would often help me learn Romani by telling me to use different words, which they considered to be standard, I was rarely corrected for misusing morphosyntax, where my problem was speaking like a nonnative rather than not speaking the standard. Morphosyntactical corrections, unlike correcting nonstandard vocabulary, were not legitimised through a metalinguistic discourse.

This emphasis on lexical items suggests an understanding of language as a collection of words. For those without formal linguistic training, words, especially their roots, are more tangible than morphosyntax and therefore more liable to purification. Language contact can lead to structural convergence while maintaining surface-level separation (Matras Reference Matras2000), and Romani has been shaped by its place on the Balkan linguistic Sprachbund; but this kind of convergence, unlike loanwords, is much less obvious. By contrast, words can be used as a symbolic resource that can be identified as foreign and purged, or as not foreign and valorised as Romani.

When people disavowed knowledge of their own language, they meant that they could speak fluently but did not know gramatika: they lacked command of the full range of ‘pure’ words that purportedly exist. ‘Knowing’ the standard means knowing the ‘pure’ variant, rather than (or as well as) their own ‘impure’ vernacular. In this respect, Roma are like other Prizrenites, who also describe a feeling of not knowing their language. This feeling of ‘not knowing’ belies an awareness of the difference between the way people speak in Prizren and the standard variants. Turkish and BCS speakers express similar disjunctures between their speech and gramatika (cf. Thomas Reference Thomas1978, Reference Thomas1988).

Many people, then, are aware that their everyday speech practices are considered a nonprestigious and impure form of the language. This is what Milroy (Reference Milroy2001) terms a standard language culture: the idea that a standard is correct becomes recognised as common sense, rather than as an ideological stance. This can lead to a reification of the standard, as it comes to be understood as natural and imbued with truth (Silverstein Reference Silverstein, Brenneis and MacAulay1996). Albanian, Turkish, Serbian, and Bosnian all have varying levels of formal codification of their respective standards, which are endorsed and legitimated by state institutions. The source of legitimacy of standard Romani is less obvious, raising the question of how, where, and why the standard is diffused from the intellectuals discussed above to other Roma. The feeling of ‘not knowing one's own language’ thus relates to the prevalence of an idea of a standard, which makes the spoken variant seem nonstandard or even substandard (cf. Hill & Coombs Reference Hill and Coombs1982). Prizren Roma primarily experience standard language ideology through their use of surrounding languages. In creating a Romani standard, then, those involved must not only make a pure standard, but also turn the pure standard, something quintessentially artificial, into a seemingly natural object. This is what people mean when they say they do not ‘know’ their language.

AUTHORITY AND PURITY

In a group interview about standard language with my colleagues at the radio station, one of them explained:

My colleague consciously uses the standard here: he uses the formal/nonderogatory terms albanyuni ‘Albanian’ and albayna ‘Albanians’ rather than their everyday equivalents gaunani and gaune (used to mean both ‘Albanian’ and ‘peasant’/‘villager’); the standard feri so rather than soske for ‘because’; the formal, international terms standardizaciya ‘standardisation’, adekvatyuno ‘adequate’, and teritoriya, ‘territory’, which are concepts with no obvious everyday equivalents. J ivdinena ‘they live’ is probably from BCS, but is accepted as standard Romani in contrast to zhivinena, which sounds closer to the Serbian žive and is used by Roma in Serb areas. The only nonstandard word he uses is sebepi ‘reason’, which others avoid because of its Turkish origin (see below).

He discusses ‘barbarisms’, the words from surrounding languages that have entered Romani, making it impure. Hard work is needed to purify the language of such barbarisms, in particular by those anointed with the title of linguist, whose role is to pinpoint impure words and replace them with ‘pure’ forms, or to make new forms without drawing on other local languages.

Most people find pure forms, as the anecdote at the beginning of the article suggests, by asking someone in a position of authority. Interestingly, however, those such as Kujtim, at the top of the social hierarchy, consider that the purest forms are found at the bottom of the hierarchy: among the so-called wild people in the forests of Albania. These Roma are deemed the most excluded, most isolated, and least ‘cultured’, and are therefore treated as untouched specimens, yet to be defiled by barbarisms. Natural language must therefore be preserved and defended against invasion, and while this natural, pure language is spoken by the natural, pure folk, untainted by contact, it must be defined and legitimised by intellectuals. Here purism focuses not on the Indic origins of Roma, but on the contemporary purity of isolation (cf. Leggio Reference Leggio2013). This understanding of language is typical of the Herderian tradition, which spread from Germany throughout central and then southeastern Europe in the nineteenth century (Bauman & Briggs Reference Bauman, Briggs and Kroskrity2000). Western scholars of the time viewed the mixture of ‘small’ ethnic groups in the Balkans negatively, and subsequently local scholars resorted to linguistic purism to find the authentic, untouched, pure speaker (cf. Petrović Reference Petrović2003). This scalar reproduction of hierarchies in language ideology is a form of ‘fractal recursivity’, the reproduction of binary differences in different places and at different scales (Irvine & Gal Reference Irvine, Gal and Kroskrity2000). Here the opposing definitions of self as cosmopolitan, urban, and civilised, and the Other as pure, isolated, and uncivilised come together to posit the Romani intellectual as a paternalistic researcher and defender of the purity of the ‘wild people’.

When people turned to those who they felt knew the standard, at the radio this meant Nexhip, unless Kujtim was visiting, in which case he generally had the higher authority. But, as the extract above illustrates, there was an even higher authority than Kujtim: Marcel.

Marcel Courthiade, known locally as O Marceli, is talked about in almost legendary terms. Marceleski qhib ‘Marcel's language’ is considered the ideal language. Marcel Courthiade lives and works in France. He identifies as Albanian Roma, an identity that was upheld by my participants, but is contested by others (Matras Reference Matras2015). He visited Prizren once to conduct research, and I heard that he went to Roma settlements in Albania where even Kosovo Roma were afraid to go, to seek authentic Romani. He has also compiled an orthography (Courthiade Reference Courthiade, Fodor and Hagège1989), which uses ‘archegraphemic symbols’ such as <θ> and <З>, to be pronounced variously according to the dialect (Friedman Reference Friedman, Hyltenstam and Fraurud2002; Matras Reference Matras2002; Leggio Reference Leggio2013). The orthography was intended to unite the dialects of Romani, a longstanding task for Romani activists; thus in 1990, the IRU (International Romani Union) adopted this orthography. Kujtim is the only person I met who uses this orthography; others say they find it too difficult. Nevertheless, Marcel's international renown is well-known in Prizren, even if the orthography is not.

Kujtim is also seen as an authority on the language, as he has conducted research at home and abroad, writes poetry in Romani, and has even written a dictionary that people in Prizren use. His authority stems both from his research of obscure dialects among ‘wild people’ or ‘real gypsies’ in Albania—he refers to this experience to show he is a real linguist, like Marcel—and from the widespread knowledge that he studied in Paris alongside Marcel. Kujtim's proximity to Marcel clearly bolsters his prestige.

I often detected a sense of competition between Nexhip and Kujtim about their knowledge of the standard, which sometimes led to criticisms. When a friend suggested Nexhip should go to Paris to learn Romani, Nexhip responded angrily, saying: Me vakarava romani sako dive. Na manglape te nakhav ano Parizi (Ro: ‘I speak Romani every day. I don't need to go to Paris’).

While the pair vied for a position of authority in public, in private these debates were taken less seriously. Asking Kujtim for a lighter to make coffee, I used the phrase normally used in Terzimahalla, s itu chakmako? (Ro: ‘Do you have a lighter?’). He corrected me—Chakmako hito korohani lafi (Ro: ‘Chakmako is a Turkish word.’) to je od turskog. (BCS: ‘It is from Turkish.’) Vakar ‘yagaripe’ (Ro: ‘Say yagaripe’). Yagaripe is derived from the Romani yag ‘fire’. Later, when Kujtim and Nexhip had left, the other staff rolled their eyes and laughed. This degree of purism, though publicly respected, was treated with humour in more private settings. Indeed, in private conversations some people criticised the way I was being taught the standard.

Those with the social and cultural capital to position themselves as intellectuals, at the top of the hierarchy, are in a different position to that of some journalists, whose task it is to reinforce and diffuse ‘pure’ language directly. This is not always straightforward. In an interview with a Roma journalist who provides Romani language services at Radio Television Kosovo, the state broadcaster, we discussed how words are selected. He gave the example of the word for cause/reason, for which the usual word is sebepi, from Turkish. He told me that he used the word kondiciya, despite having been told by Kujtim that the correct term is ash. The journalist accepted Kujtim's authority about the word ash. However, he questioned whether to use this Romani form, or kondiciya, an ‘international’ loan (rather than a BCS or Turkish one), which would be more understandable to his listeners. He decided that comprehensibility was more important than purity, although he still wanted to avoid the impure sebepi. Kondiciya was an effective compromise.

While Kujtim, from his position of authority, was free to insist on more exotic, ‘purer’ forms, those working in media such as the radio station were in a more difficult position, as they had to translate things at short notice, while also making them comprehensible to the Romani-speaking public, many of whom are less familiar with the idea of a standard Romani. Kujtim displayed his knowledge by speaking the ‘purest’ form possible, while those working in media had to negotiate a ‘standard’ form, which would grant the media authority while also ensuring its intelligibility.

Both levels of language use (the higher, intellectual level, and the in-between media level) are purist, but in different ways. While Kujtim's ideal is a preserved, isolated form, those working in the media focus on expunging loan vocabulary specifically from surrounding languages. An English-origin word is acceptable where a Turkish-origin word is not. Their construction of the standard is much more negotiated, seeking to create an educated, yet widely understood, variant of Romani.

In sum, the everyday tactics journalists and intellectuals use for making new terms are creative and spontaneous. Those considered Romani intellectuals can learn ‘pure’ forms, reinforcing the social hierarchy. Kujtim's authority over Nexhip, however erratic, is not just based on their respective research experience and publications, but also on Kujtim's time with Marcel. What is crucial, then, is not just going to the periphery to find isolated people and words, but also remaining connected to the centre, which is imagined to be where Marcel and his favoured group of intellectuals are.

Learning ‘pure’ Romani is not the work of a lone scholar. It cannot be done in the bath. It is necessary for the linguist to brave the wild people of the forests of Albania to ‘know’ what is pure, while also using one's place in a social hierarchy to maintain proximity to the western intellectual. The next section discusses how these political wranglings affect speech practices.

DIFFUSION OF IDEOLOGY

Nexhip once discussed a possible project to increase attendance at Romani classes, but he wasn't sure if the schools would agree. I suggested using the NGO's office or a children's centre instead. Nexhip was unimpressed and said the classes must be held in schools—otherwise, they would be pointless.

Not all of the children who attend Romani classes know Romani, and their backgrounds vary widely. In class, those with little or no Romani learn some basic phrases and words, while those who are already familiar are taught standard Romani words and practise writing in Romani. The aim is not to teach them a Romani orthography (Albanian orthography is generally used), but rather for them to speak and write Romani in a classroom setting, and therefore to associate it with education, formality, and the public sphere, and to use it outside the private spaces of the home or Romani quarter. Writing Romani in the home or a children's centre would not be the same. Institutionalising Romani through education and media is a means of establishing it in the public sphere. Establishing Romani as a language of education and a written language, and making a place for it in the public sphere, is at the forefront of its standardisation in Prizren.

In terms of most understandings of language standardisation, neither the classes, the Romani theatre group, nor the radio appear particularly effective at diffusing the standard. The plays do not reach a wide audience, and relatively few people own and listen to radios. Low attendance at Romani language classes renders questionable their effect in spreading knowledge of the standard. They do, however, teach children that Romani can be used beyond the home or Romani quarter, and that they can be Roma while in school. The Romani-language textbooks may lack a clear methodology or curriculum, but their presence in classrooms is important for both Nexhip and the children. They act as a symbol, showing that Romani can be used for formal education. Similarly, while attendance at the dramas is limited, their very existence is crucial to showing that Romani can be a literary language.

For those involved, what is at stake is that Romani has a pure, standardised form, appropriate for use in the public sphere. This must be distinct from the mixed, nonstandard form used in the home or Romani quarter. The diffusion of a standard language ideology thus takes precedence over the diffusion of content. Matras (Reference Matras2002:254) describes how ‘(t)he overwhelming majority of texts produced in Romani do not form part of coordinated language planning efforts. Rather they are the outcome of local and individual text-production initiatives’. Certainly, in Prizren, despite Marcel's role as an external reference point, local rather than international efforts are central to the formation of a standard. While with other languages in Prizren the standards are set by state institutions, Romani standardisation is an ongoing process. This is not to suggest that other standards are fixed, but there is at least a point of reference from which to debate standardisation. Correctness of, for example, Albanian or Serbian is managed institutionally, in Tirana or Belgrade. By contrast, the selection and codification of Romani norms occurs sporadically, at a local level, structured by competing authorities on the language.

With the surrounding languages, enregisterment—the ‘processes through which a linguistic repertoire becomes differentiable within a language as a socially recognized register of forms’ (Agha Reference Agha2003:231)—occurs through state institutions. Without state structures, the diffusion of Romani is performed by disparate governmental and nongovernmental initiatives and individuals. In Haugen's (Reference Haugen1966) terms, the classes, media, and drama are all sites of the simultaneous selection and diffusion of norms.

The vertical diffusion of standard Romani in Prizren—that is, the attempts of those in institutions to inform people about the standard—brings Romani into the public sphere, and spreads the idea that there is a pure, standardised, literary form of the language. This is intended to show both Roma and non-Roma that the language is equal to the surrounding languages, in a sense mimicking the role of the state. It is thus indicative of the domain change undergone (to some extent) by the surrounding languages at an earlier stage (cf. Paugh Reference Paugh2012; Romaine Reference Romaine1994).

Language contact therefore affects Romani in three ways. First, there is structural convergence (Matras Reference Matras, Boeder, Schroeder, Wagner and Wildgen1998, Reference Matras2002:ch. 8, Reference Matras and Siemund2011; Friedman Reference Friedman1985b, Reference Friedman1986, Reference Friedman1991, Reference Friedman2001). Second, those formulating the standard differentiate it from those contact languages through the lexical purism discussed above. Third, standard Romani itself is conceptualised on the model of other languages—a sort of horizontal diffusion, the diffusion of the idea of a standard across languages, as opposed to its vertical diffusion from intellectuals to ordinary people. From this perspective, a standard is needed not just to qualify Roma as a nation, with Romani as a national language; the existence of the standard also shows that Roma, and their language, can be educated and cultured. While those involved in the media, drama, and school classes seek to diffuse the standard to other Prizren Roma, the ideology of a standard is being diffused from speakers of the surrounding languages. There is thus a kind of ideological convergence. In the opening vignette, then, when my hostess said ‘we do not know our own language’, she was replicating the ideological stance of speakers of other local languages: that the spoken variant is an impure, inferior version of a perceived standard.

The ideology behind BCS (and to a lesser extent Albanian) standardisation in the nineteenth century was a reaction to imperialist domination (see Wilson Reference Wilson1970 on Vuk Karadžić’s codification of Serbian, Despalatović Reference Despalatović1975 on the Illyrian movement's codification of BCS; cf. Heyd Reference Heyd1954 on Turkish). In contexts of language contact, purism can be seen as a reaction to the hegemonic purism of the dominant language (cf. Hill & Hill Reference Hill and Hill1980). A similar ideology is now mobilised as a response to the dominated position of Romani. Standardisers aim to have a standard language, as a cultural artefact; and standardisation also preserves and promotes Romani in the face of the specific problems faced by Roma in postconflict Kosovo, namely, the need to counter charges of collaboration.

In Prizren, then, a metalinguistic discourse of what is correct and incorrect Romani is being spread. Language ideologies do not simply misrepresent actual speech practices: linguistic engineering can also alter these practices (Golovko Reference Golovko, Matras and Bakker2003). Those familiar with the speech practices defined as standard correct one another, and me as a language learner, who should learn the best variant. However, they only do so in the more institutional contexts of the radio, theatre, and classes.

The ‘purest’ forms, promoted by intellectuals, are not widespread. However, the idea that people should speak a particular way has diffused. There is a developing awareness not that everyone should speak the same, but rather that the way people speak in the Romani quarter is inferior or inadequate, which distances the standard from the vernacular (Thomas Reference Thomas1991). The standard is understood as prestigious, not merely uniform, and thus appears neutral and common-sensical (Milroy Reference Milroy2001; Gal & Woolard Reference Gal and Woolard2014). Why it is correct or pure or standard does not need qualifying: it is legitimised by recognised authorities.

Certainly, standard forms themselves are diffused. In more formal settings, such as my interviews about language use with journalists exposed to the standard, I found it common for people either to use an everyday word before correcting themselves with a standard word, or to use the standard word followed by the everyday word as an explanation. Although I often understood ‘international’ terms better, they still felt the need to offer a more usual, less prestigious form. When they wanted both to use the standard and to be understood, they used both.

This corrective or explanatory doubling-up of terms could perhaps be explained as an interim phase, whereby the intellectuals educate ‘the masses’ about the impurity of their language use, gradually raising them to the level of pure, educated Romani. However, though this may be the obvious Herderian explanation, I would suggest that this is not merely a phase, and that neither term is likely to die out soon. Rather, as with other languages in Prizren, efforts to make a standard create a diglossia, whereby ‘incorrect’ mixed spoken forms exist alongside pure gramatika. The ‘correct’ forms are used in institutions, while the ‘incorrect’ forms are acceptable in the home. Other situations, like an informal discussion at the radio, may allow either: even in institutions, the informal/‘incorrect’ forms are used in certain contexts.

Standard Romani is discursively constructed not as something that must undergo processes of selection and codification before diffusion, but rather as something that exists but requires better diffusion. When people talk about ‘not knowing our own language’, they imply that Romani exists in a pure form but must be relearned and better disseminated, like, for example, standard Albanian. Informal, mixed variants are contrasted with formal, distinct standards. With Romani standardisation, the diglossia apparent in other languages and the attitudes attached to it, are being replicated.

Rather than domain change meaning a shift from the private to public domain, requiring lexical enrichment, the public use of gramatika has resulted in its bifurcation and stratification from the private domain and the Terzimahalla idiom. Making a standard thus also entails producing a nonstandard; legitimising one type of speech delegitimises another, thus reproducing hierarchies at different scales. As with Corsican, where domain change reproduced dominant (French) language ideology (Jaffe Reference Jaffe1999), a dominant language ideology, whereby speech practices are understood as hierarchical (and attached to the cultural capital of education), is being reproduced within Romani.

The gramatika (the educated, cultured variants) of the surrounding languages is contrasted with the local variants, perceived as mixed and impure. Impurity and peripherality become intertwined, opposed to the pure, standardised gramatika. While Romani lacks a physical centre such as the other languages have, I would argue that by making a standard, and by making Terzimahalla Romani peripheral, intellectuals position themselves vis-à-vis an imagined centre. The mixed, impure, peripheral variant is contrasted to the pure variant promoted by intellectuals, who, positioning themselves at the top of the social hierarchy, also put themselves at the centre of the standardisation process.

Figures of authority are attempting to create, and diffuse, a standard. They also often insist that all Romani dialects are mutually intelligible. On one level, they seem to be creating and spreading an inclusive standard for people who speak various slightly different variants. However, by creating a diglossic situation with two unequal variants, they are actually forging new forms of inequality in language use. Whereas previously knowledge of another national language signalled social stratification among Roma, by introducing this diglossic situation in Romani, intellectuals are unwittingly creating a layer of inequality within Romani itself. By rendering everyday Romani more peripheral, they position themselves as gatekeepers of purity and correctness. This is not, of course, to say that they intend to increase negative perceptions of poorer Roma, but rather to suggest that the generalisation of the language ideology in the area reinforces such perceptions.