1. MULTI-ETHNIC YOUTH IN FRANCE AND THEIR URBAN VERNACULARS

In the 1990s, most of the literature on youth vernaculars spoken in pluriethnic urban areas of France, particularly the Paris banlieues, focused on lexis, whether in the form of glossaries and dictionaries, e.g. Décugis and Zemouri (Reference Décugis and Zemouri1995); Aguillou and Saiki (Reference Aguillou and Saiki1996); Seguin and Teillard (Reference Seguin and Teillard1996); Goudaillier (Reference Goudaillier2001) or commented listings such as Sayah (Reference Sayah1999a), (Reference Sayah1999b) and (Reference Sayah2001). Such work undoubtedly points to the linguistic effects of frequent contact among speakers of a considerable number of varieties, particularly the multiplicity of origins of neologisms in such vernaculars, but with at best only the most schematic indications of the social and geographical currency of the items listed. The impression of ‘convergence within divergence’, perhaps more plausible on the level of phonology, where relatively few features come into play (Jamin, Trimaille and Gasquet-Cyrus, Reference Jamin, Trimaille and Gasquet-Cyrus2006) is, to say the least, more difficult to substantiate with regard to vocabulary. The aim of all such lexicographical work is to describe new forms in French, whether derived from native lexical stock, including traditional argot and English or indeed varieties brought to France by migrant communities, particularly Dialectal Arabic, Berber, Rom and certain sub-Saharan languages, such as Wolof and Bambara. Such glossaries of tchatche, verlan and other forms of youthspeak sometimes (e.g. Goudaillier, Reference Goudaillier2001) give a misleading (although pervasive) impression of a newly emerging pluriethnic variety. That often ephemeral lexical inventiveness and a relatively small number of phonological phenomena, constitute new dialect formation has been justifiably questioned inter alia by Lamizet (Reference Lamizet2004) and Trimaille and Billiez (Reference Trimaille, Billiez, Galazzi and Molinari2007).

Variationist work, whether conducted from a behavioural or perceptual approach, on such phenomena, has drawn attention to the crucial role of interethnic contact in linguistic behaviour. Several studies have shown young people of Maghrebian background to be the highest users of both certain segmental vernacular variants in Paris (Armstrong and Jamin, Reference Armstrong, Jamin and Salhi2002; Jamin, Reference Jamin2005; Reference Jamin2007) and Grenoble (particularly, the work of Trimaille) and Marseille (Gasquet-Cyrus), as exemplified in Jamin, Trimaille and Gasquet-Cyrus (Reference Jamin, Trimaille and Gasquet-Cyrus2006) and suprasegmental features in Paris (Fagyal, Reference Fagyal2010). Both Armstrong and Jamin (Reference Armstrong, Jamin and Salhi2002) and Lehka-Lemarchand (Reference Lehka-Lemarchand2007), who studied the use of certain perceptually emblematic suprasegmental features in Rouen, point to the correlation in the relatively frequent use of such features with involvement in local street culture and in the latter case with social exclusion and low social status (rather than ethnicity) and informal style (peer-to-peer interaction). The title of Jamin, Trimaille and Gasquet-Cyrus’ (Reference Jamin, Trimaille and Gasquet-Cyrus2006) article suggests convergence in the use of features divergent from mainstream usage, such as affrication by adolescents in three quite distant urban areas. The apparent degree of divergence should, however, be tempered by the observed use of palatalisation/affrication (Trimaille, Reference Trimaille2011) and strongly fricative r (Pickles, Reference Pickles, Hintze, Pooley and Judge2001) by non-marginalised speakers. Interactionist studies, e.g. Billiez, Krief and Lambert (Reference Billiez, Krief and Lambert2003), point to the ethnicisation of the way in which young people classify themselves (bourges, i.e. white, middle-class and bouffons ‘uncool’) as opposed to sub-cultural norms (‘langage racaille’ or ‘brounes/brounettes’), while studies of insults, e.g. Baines (Reference Baines, Ayres-Bennett and Jones2007), such as kehba and ta mère exemplify direct and calqued borrowings from Arabic but still fall short of the ‘use of speech varieties associated with other people’ (Rampton, Reference Rampton2005: 1).

While the convenient equating of social and linguistic fracture (as in Goudaillier, Reference Goudaillier2001) undoubtedly persists both as a linguistic characterisation and a political diagnosis (Bertucci, Reference Bertucci2010), it may be questioned for a number of reasons. Firstly, studies of La Courneuve from an ethnographic (Lepoutre, Reference Lepoutre1997) and variationist perspective (Jamin, Reference Jamin2005) show that linguistic divergence (in the form of rituals such as verbal jousting or vannes and the use of vernacular variants) is at its greatest among younger adolescents immersed in street culture, even compared to older teens (Lepoutre, Reference Lepoutre1997) and young adults (Jamin, Reference Jamin2005), who use such forms significantly less. Secondly, studies of the social fracture from sociological perspective such as Beaud and Pialoux (Reference Beaud and Pialoux2003) point to young adulthood and the harsh realities of the job market as the crucial factors. Thirdly, it may be argued that all our informants find themselves on the downside of such a social fracture, regardless of ethnic background, even if it is sometimes, and often covertly, ethnicised (Tissot, Reference Tissot2007). That said, the notion of code-crossing implies socio-cultural and ethnolinguistic differences which people such as our informants have to come to terms with on a daily basis. The different levels of competence in Arabic of non-inheritors, as we hope to show, point to some of the processes involved

2. A NEW INVESTIGATION OF LILLE

While the primary overall aim of the project being reported here was to develop the investigation of vernacular French begun in 1995 (reported in Pooley, Reference Pooley2000; 2004) among school-age adolescents in a number of areas of Lille-Métropole,Footnote ¹ an important secondary goal was to describe and evaluate the multilingual repertoire of the informants, expanding the scope of previous work which had concentrated on the ancestral vernacular, Picard, to migrant languages, particularly Arabic, and by extension any instances of code-crossing. Three areas of Lille-Métropole were selected for study, two of which were central: Lille-Sud, and La Madeleine and the third, Lys-lez-Lannoy on the north-eastern side of the conurbation (Roubaix-Tourcoing) (Figure 1).

Figure 1. Location of fieldsites within Lille.

Two of the areas selected are classified as zones urbaines sensibles (ZUS) with all the associations of social-problem areas which that term connotes. While Lille-Sud lies within a ZUS, the school at Lys-lez-Lannoy, situated at the meeting point of three communes, Lys-lez-Lannoy, Leers and Roubaix takes in pupils (including the majority of those in the present study) from two areas of Roubaix, Le Pile (consisting largely of 19th-century terraced housing) and Trois-Ponts (largely a 1960s development of barres and tours), which are classed as ZUS. The third area, La Madeleine, is contiguous with Lille to the north and home to people from a broader spectrum of social backgrounds (Figure 1).

The informants were adolescents, aged 13 to 14 attending the special-needs sections (SEGPA – Section d’Éducation Générale et Professionnelle Adaptée) of the collèges concerned. In terms of career prospects and family background, the students were clearly at the lower end of the social spectrum, being Category 4, as defined by the Ministry of Education's scale of predicting academic success (www.education.gouv.fr/ival), and working class or underclass in Chauvel's (Reference Chauvel2005) social-class categorisations based principally on income. The working classes of the early 21st century are, it should be noted, very different from those of the traditional textile-based heritage, manifesting all too little of the solidarity of yore, favoured by the small terraced houses of the industrial, often behind street-front buildings (courées) and plentiful and stable local work (Pooley, Reference Pooley1996).

The selection of schools as field-sites for the present study (the fieldwork was conducted in 2004 and 2005 was based partially on previous fieldwork conducted in La Madeleine (1997) and Lys-lez-Lannoy (1998)) and partially on the objective of studying (central) Lille, Lille-Sud being widely perceived as a socially difficult area. In addition to their social profile, the choice of SEGPA pupils as informants brought a number of other advantages: firstly, their greater availability than mainstream pupils, and secondly, the small class sizes with relatively stable year-on-year membership, which created a strong group solidarity (students had been together in the same class, doing most subjects together for between one and a half and two and a half school years) and thirdly, the perfectly feasible prospect of recording all pupils at least twice in small groups as well as carrying out in-depth individual interviews in school time within a two-week study visit for each of these schools. As Table 1 indicates, a total of 57 school students took part in the study from three schools and five classes, with between seven and fourteen members (one from Lys-lez-Lannoy and two each from Lille-Sud and La Madeleine). Forty of the informants (70.2%) were of Metropolitan or European French background,Footnote ² eleven of Maghrebian origin (19.2%), and six ‘others’. These latter cannot be claimed to form a homogeneous group. Two (3.5%) were of mono-ethnic background (one of Senegalese and one of Serbian extraction) and four (7%) of ethnically or nationally mixed parentage (Franco-Portuguese, Franco-Italian, Franco-Rom and Franco-Moroccan).

Table 1. Ethnicity and geographical distribution of informants i fieldwork locations

MF = Metropolitan French parentage; MAG = Maghrebian parentage; OTHER = neither MF nor Maghrebian; MIXED = of mixed parentage

While the small-group recordings were largely designed to capture the students’ spontaneous vernacular speaking style which they used among themselves, the individual interviews were designed not only to gather relatively formal speech but to administer language tests in Picard and Arabic (see Appendix 1) and tease out ethnographically important information regarding friendship groups, life-style choices and socio-cultural orientation. In what follows, we intend to concentrate on the results of an Arabic language test administered as part of these one-to-one interviews, which were in turn part of a broader investigation into the practices and repertoire of Lillois adolescents before evaluating them in the light of our characterisation of the socio-cultural space in which the informants were moving. Details of the test are presented in Section 3, where the results are reported. All the Arabic data were analysed by one of the co-authors (ZMH, a native speaker of Algerian Arabic). The potential danger of a ‘floor’ effect (i.e. low and undifferentiated scores) among non-Maghrebians and a ‘ceiling’ effect (high and undifferentiated scores) among Maghrebians did not materialise. The degree of differentiation noted among non-Maghrebians corresponded to some extent to gender and location. The significance of location is of course related to the number of opportunities of interethnic interaction. To evaluate the latter, three indices of Social Interaction Distance were devised (Lambert, Reference Lambert2011) based on a friendship survey (Lemel, Reference Lemel2006; Chan, Reference Chan and Chan2010) among the participants. Subjects were asked to name the friends with whom they spent the most time in school (but not necessarily in their class group) and outside school. As first names are generally evocative of ethnic originFootnote ³ (this was checked in cases where the individuals were not pupils at the school being studied), values for two indices of interethnic friendship could be worked out, awarding one point for any cross-ethnic friendship (e.g. a pupil of Metropolitan French background names ‘Rachid’) among in the first three people named and half a point for any further names. Each pupil was assigned an index value for friendships in school (Interethnic Friendship Index at School or IFIS) and outside school (Interethnic Friendship Index Outside School or IFIO). Some of the same information was used to assign values to an Index of Popularity in Class (IPC), which was simply the number of mentions received by each pupil by classmates. It was expected and an earlier study has given partial confirmation to this (Pooley, Reference Pooley, Beeching, Armstrong and Gadet2009) that this would point to leaders in various forms of behaviour, including language. All subjects were asked about their tastes in dress and music as well as their views of the headscarf issue. Subjects of Muslim background were asked about what forms of ritual they observed, such attendance at a mosque, Ramadan or ritual prayers. All subjects were asked three questions on a multiple-choice basis (Appendix 3) designed to work out a Regional Loyalty Index (RLI): 1) whether they liked living in the area (‘par ici’);Footnote ⁴ 2) whether they liked people ‘around here’; 3) where they would like to live in an ideal world. While the primary aim of this index is correlate a measure of regional loyalty with the use of regional-accent features, question 3) was useful in eliciting the attitudes of subjects of Maghrebian background towards the country of family origin.

In the following three sections we propose firstly to describe and evaluate the degree of multilingualism noted among the subjects, before presenting a more detailed breakdown of the results of the Arabic (Section 3). In Section 4, we first assess the individual profiles of the most prolific acquirers of Arabic from non-Maghrebian backgrounds, and secondly, we seek to explain the variation in test scores of the ‘inheritors’. In the case of non-Maghrebian subjects, relatively high test scores as well as the data derived from the Speaking Test are construed as indicative of potential for code-crossing. Finally in Section 5, which is more qualitative in approach, we attempt to evaluate the importance of factors within the socio-cultural space which favour the maintenance of Arabic among young people of Maghrebian background and make Arabic and indeed other languages and cultures, particularly Rom, attractive to non-inheritors.

3. MULTILINGUALISM, KNOWLEDGE OF ARABIC AND SOCIAL FACTORS

The multilingualism of the sample population was, at least potentially, considerable. The shared variety or parler normal was the students’ vernacular variety/ies of French, spoken with varying degrees of regional markedness. For the Metropolitan French subjects, Picard was (again at least potentially) a heritage language, of which few were totally ignorant (although no subject used or claimed to use it spontaneously), while the non-Metropolitan French subjects were inheritors of Dialectal (and sometimes Standard) Arabic, Berber, Rom, Italian, Portuguese, Serbian and Wolof. Contrary to the findings of studies in the UK and Germany, no use of ethnolectal varieties of the mainstream language, particularly what one, taking inspiration from Stylised Asian English, might call Stylised Maghrebian French, was observed, although use of a variety/varieties containing features lending some justification to such a label was noted in public places in peer-to-peer exchanges and even during witness interviews on FR3 regional news. The similarity of the labels should be considered with regard to form and not to use. The similarity of form concerns conscious use of L2 features. In terms of use, however, Stylised Asian English is used very much as the language of the Other and has been observed as an instrument of crossing, whereas Stylised Maghrebian French is, in our observations, use as an I- or we-language to exhibit difference from mainstream usage.Footnote ⁵

The language tests used a methodology designed to tease out any remaining knowledge and awareness of Picard in what may fairly be called members of a post-shift generation (Pooley, Reference Pooley2003; 2004). While the breakdown in intergenerational transmission of Arabic is by no means as advanced as that of Picard, it is nonetheless the case for most inheritors that sibling-to-sibling and peer-to-peer conversations take place predominantly in French. Indeed, even with parents with limited active competence in French, asymmetrical exchanges (parent speaks in Arabic, child responds in French) are claimed to occur frequently. While switches to Arabic (or indeed other languages) were considerably more likely than to Picard, only a very few in fact occurred in previously recorded data, and it was therefore deemed preferable to tease out and evaluate knowledge and awareness of Arabic using similar methods (See Section 2 for details of how the test was administered and Appendices 1 and 2 for the vocabulary tests).

The tests were each divided into three parts: firstly, giving French equivalents of ten Arabic or Picard words (Appendix 1); secondly, giving Arabic or Picard equivalents of ten French words (Appendix 2); thirdly, inviting the informants to say something in Arabic. For the first two parts, a point was awarded for any plausible answer. In the Speaking Test, a maximum score of 10 was awarded for any grammatically acceptable string of at least five syllables or a series of shorter sequences; five points were awarded for shorter sequences manifesting appropriate grammatical combinations and up to three points were given for isolated words. The scores out of a maximum of 30 are presented as percentages.

Although the overall scores for both Arabic and Picard are extremely modest, they were remarkably similar: 19.9% (Table 2) compared to 19.8%. The areas of relative strength, however, diverged. For Arabic, the overall mean for the Speaking Tests was significantly higher (32.8%), whereas for Picard the highest average section score was 30.7% achieved in the Picard-French part of the test. The breakdown of the results by ethnicity in Table 2 shows that the scores of the inheritors are, unsurprisingly, significantly higher, particularly in the vocabulary tests, whereas the open-ended Speaking Tests produced a rather smaller, albeit still significant, difference (in both cases p<.01 by χ² test). While these relative disparities are worthy of note, the overall mean score of the inheritors at around 50% is perhaps surprisingly low, although the analysis of individual profiles (Section 4) brings to light some plausible reasons for this. The difference between the Metropolitan French and the others is hardly significant, given that one of the others is a Franco-Moroccan girl, also discussed later, who, at the time of fieldwork, took particular pride in her Moroccan heritage. Within the two most numerous ethnic groupings (Metropolitan French and Maghrebian), boys scored significantly higher than girls, a finding slightly obscured in the ‘others’ category by the small numbers and the presence of this same Franco-Moroccan girl (Table 3).

Table 2. Results of Arabic Skills Test by ethnicity

Table 3. Results of Arabic Skills Test by ethnicity and gender

The clearly pertinent differential social factor for the Arabic test scores of non-inheritors was interethnic friendship. This was calibrated using two indices based on the naming of the friends that subjects claimed to frequent the most, both at school and outside school: firstly, Interethnic Friendship Index in School, IFIS and secondly, Interethnic Friendship Index Outside School, IFIO, supplemented by an Index of Popularity in Class (IPC), based simply on the total number of citations for each subject among their classmates (as opposed to the school as a whole). Clearly, the index values for IFIS and IFIO were related to the potential number of cross-ethnic friendships within the school, and possibly to the lack of interethnic crossing in the names proffered, although this could be readily verified, bearing in mind that the non-Metropolitan French part of the background for subjects of mixed European parentage was within the school context, with one exception, socially invisible.Footnote ⁶

From the results in Table 4, the school is the main locus of interethnic friendships, particularly in two of the five classes studied, i.e. Lys-lez-Lannoy and Lille-Sud V^e, where such friendships were the most frequent. As Table 5 shows, high scoring non-inheritors also stand out as having a positive value for both IFIS and IFIO indices, i.e. their greater knowledge of Arabic appears to be fostered by friendships in and out of school.

Table 4. Comparative mean IFIS and IFIO indices

IFIS = interethnic friendship index at school

IFIO = interethnic friendship index outside school

Table 5. Comparison of IFIS and IFIO indices to Arabic Skills Score

For subjects for whom Arabic is a heritage language, the results from Lys-lez-Lannoy and Lille-Sud are not convergent. As there is much evidence to suggest that spontaneous use of Arabic in peer-to-peer conversations is unusual, it is more plausible to suggest that competence in Arabic, at least as reflected in the test score, depends more on family background and the individual's life history, a theme which we develop in the following section.

4. INDIVIDUAL PROFILES AND CROSS-ETHNIC LANGUAGE ACQUISITION

In this section we first examine the radically different nature of the ‘correct’ answers given by subjects of Maghrebian and non-Maghrebian background in the Speaking Test, before considering other responses, mostly in Berber or Rom, and offering explanations of the divergences on the basis of individual profiling. Comparison of the sample answers given by Maghrebians in Table 6 with the near-exhaustive list of responses from non-Maghrebians (Table 7) reveals a stark contrast of subject matter, indicative of different acquisition processes.

Table 6. Examples of Arabic quoted by Maghrebian subjects in Speaking Test

Table 7. Arabic expressions most frequently quoted by non-Maghrebian subjects in Speaking Test

The responses listed in Table 6 evoke family life and exchanges with relatives and friends and possibly sojourns in North Africa. In examples a), b) and c), the subjects are quoting their mothers directly, from whom they receive instructions or correction. Example d) bespeaks the end of a family meal and e) talk of a relative or friend in North Africa. Example f) is an example of a greeting and a response, rather less well known, as the vocabulary test shows, to L2 speakers than [salaːmalɛːkʊm] ‘Hello’, as well as manifesting features evocative of the Oranais region. Arguably, the most striking difference between the responses of the inheritors and non-inheritors is the lack of repetition in the former, whereas in the latter a limited range of responses tend to recur.

Listed in Table 7 are all the answers proffered by more than one non-inheritor. The most frequent item [wala] is a discourse marker that is arguably becoming adopted in French, and has certainly been brought to the attention of a wider public, at least since the publication of Azouz Begag's book Dis Ouallah in Reference Begag1997. Example e), although not figuring in the vocabulary test is an item of which many non-speakers of Arabic are aware, not least as a jocular excuse for normally socially unacceptable behaviour such as burping. Examples b)-d) may be used as code switches to insult and to interrupt an interlocutor, again in most cases in jest.Footnote ⁷ The plausibility of this interpretation is strengthened by the indicators of L1 influence in certain answers, particularly the French uvular approximant or fricative /r/ in [naʀdinəmʊk] and [naʁʤɪnbɛbɛk].Footnote ⁸ That conceded, other Metropolitan French informants produced more native-like realisations.Footnote ⁹ Example f) appears to be an example of code-mixing with French nique ‘f. . .k’ combining with Arabic [ʊm] ‘mother’ augmented by the possessive suffix based on [k]. While f) and b) to d) manifest signs of grammatical combination, for which credit was given in the Speaking Test, it is probable, as Rampton notes for non-inheritor use of Panjabi, that these strings are repeated as sequences learned as set phrases rather than being produced as a result of rule-governed combination. This interpretation was reinforced by some of the interpretations suggested by informants which erred heavily on the side of pragmatic equivalence, compared to the more literal glosses given in Table 7. The small number of items proffered by only one subject were mainly insults, e.g. [ʕaɹuːʒ] ‘idiot’ and [ħmɑːɹ] ‘donkey’, which when considered alongside such pseudo-Arabic forms such as [həmdubevɛk], apparently a blend of hamdullah ‘thanks to Allah’ and [bɛbɛk] ‘your father’ as well as French-Arabic combinations such as [wɛʃtχkiːl], literally ‘what, tranquille (in French)’ and [naʕadinjəmmɛʃkɛkɛt] ‘Shame on your mother's religion + willy’, point strongly to playful (and solidarity-building?) usage.

The data suggest some divergence from a common diaspora form of Dialectal Arabic, which is shared by heritors of different national and regional backgrounds within Algeria and Morocco and non-inheritors. Indeed, as already mentioned, one subject of Maghrebian background manifested clear traits of Oran Arabic and one non-inheritor of the classical variety. Moreover, informants of Moroccan parentage tended to propose answers in Berber, while conscientiously believing that it was Arabic (cf. the difference in test scores between inheritors of Algerian and Moroccan descent shown in Table 8).Footnote ¹⁰

Table 8. Arabic test scores of Maghrebian subjects

The subjects of Algerian descent scored significantly higher (p<.01 by χ²) than their Moroccan counterparts, despite two low scorers (FT and IR). Both these girls were reluctant to speak Arabic at all, possibly because they perceived French as the language of social advancement and modernity, an attitude often strongly encouraged the point of non-maintenance of the ancestral tongue by parents whose own French was limited. It may be too that the interethnically shared elements, such as those shown in Table 8 were perceived as unfeminine, although no inheritor specifically suggested this. F.T.'s low overall score, however, appears to reflect a breakdown in family-based transmission.

Another finding that emerges from the data presented in Table 8 is that the three highest scorers are from Lys-lez-Lannoy and of Algerian background. Those of Moroccan descent are concentrated in Lille-Sud, apart from the isolated individual Y.D. in one of the La Madeleine classes. This skewed distribution also corresponds to markedly different scores in the vocabulary tests among non-inheritors, the Lys informants producing 9/11 (82%) of all correct answers in the Arabic-French test and 18/28 (64%) in the French-Arabic task. Moreover, the Lys non-inheritors gave a greater range of acceptable responses, covering half the items (ami, garçon, homme, bonjour and merci) whereas [salaːmalɛːkʊm] ‘bonjour’ accounted for 8/10 correct responses by non-Maghrebians in the other schools (Appendix 2).

On the other hand, subjects from Lille-Sud V^e were more likely to produce responses in Berber, irrespective of their ethnic background, and those in Lille-Sud IV^e either gave answers in Rom or claimed that they would find it easier to do so than in Arabic. This unexpected finding is explicable by the presence of P.D., whose father is an L1 speaker of a variety of Rom and an apparently deficient L2 speaker of French. Although P.D. himself clearly separated French, Arabic and Rom, his popularity within the class, substantiated by both his IPC score (5 in a class of 12) and by a number of unsolicited remarks, pointed to a charisma sufficient to influence classmates. While occurrences of Berber among non-inheritors are arguably more difficult to attribute to a single source, the IPC values of the three boys of Moroccan background from Lille-Sud V^e were very positive (4 or 5 in a class of 14). As, however, the primary aim of the friendship indices was to contribute to the explanation of differential use of variants in French, the utility of which is borne out by results to date (Pooley, Reference Pooley, Beeching, Armstrong and Gadet2009), such indices should be used with precaution with regard to responses concerning varieties which do not exhibit generalised vitality as social practices among these peer groups.

The profiles of the most prolific potential code-crossers using Arabic as the outgroup code, summarised in Table 9 suggest that they were likely to be male, live (or have lived) in certain areas of Roubaix – M1 had only recently moved to the area of L’Épeule from Le Pile – and to have friends of Algerian descent both in and out of school.

Table 9. Summary profiles of the highest scoring non-Maghrebian subjects in Arabic test

These results suggest that cross-ethnic language acquisition among these school populations is neither a major nor a unified phenomenon. Although Arabic is the leading trans-ethnic code, there is no strong evidence of a unified diaspora variety, comparable to Creole in UK. Adolescents of non-Maghrebian parentage may equally acquire through non-institutional means scraps of other languages, and such acquisition may apparently derive from contact from a small number of individuals. Moreover, at ages 13 to 14, the subjects were in a period where enforced interethnic contact through school was likely to be at its most frequent and prolonged but also to take place in a carefully nurtured atmosphere of mutual acceptance. Given the extremely modest career prospects of the subject populations, the supportive atmosphere of the SEGPA may be assumed to foster a more positive attitude than may be the case in later life, when the reality of the consequences of poor educational qualifications have clear potential to undermine the by no means invulnerable interethnic tolerance created within the education system. What is more, as Auer and Dirim (Reference Auer, Dirim, Androutsopoulos and Georgakopoulou2003) argue regarding Turkish, non-inheritor competence does not correspond to a stable set of factors, although acquirers are rarely hostile to Turkish culture. They may well, however, harbour neutral, rather than clearly positive, attitudes towards the inheritors of the target variety of code-crossing.

5. SKETCHING THE REMAINDER OF THE SOCIO-CULTURAL SPACE

While cross-ethnic language acquisition clearly correlates with friendship clusters, it seems to us worthwhile to consider a wider range of factors which go to make up the overall socio-cultural space in which these adolescents were moving at the time of the fieldwork. In this section, therefore, we propose to present a qualitative analysis of a number of factors, firstly the attitudes of the Maghrebians towards their own heritage culture, and secondly, how such cultural elements are perceived by their Metropolitan French peers.

As regards the Maghrebians’ attitude towards their heritage culture, we have already mentioned a number of indicators which suggest that language loyalty is not particularly strong within their community, particularly compared to the Turkish community in Germany and, in terms of their protectiveness of Creole, the Afro-Caribbean community in UK. Comparison of Arabic and Turkish speakers born in France with inheritors of other languages, e.g. the INED survey of 2002 (Héran, Filhon and Deprez, Reference Héran, Filhon and Deprez2002) suggests that the Turks show greatest language loyalty of any migrant or indigenous linguistic minority of any magnitude in France, whereas Arabic speakers are merely slightly above average overall, but feature among the least linguistically loyal of the migrant communities, although some allowance should be made for a correlation between recency of arrival and frequency of intergenerational transmission. None of the subjects of Maghrebian descent claimed to use Arabic with their peers or siblings, and in some cases testified to the active discouragement of their mothers and fathers, although the parents used Arabic among themselves and with people of their own, or their parents’, generation. They, however, readily tolerated asymmetrical conversations with their children. In the case of subjects of Moroccan parentage, exposure to Berber complicated the situation, at times causing a fuzziness in the separation of heritage tongues.

When asked to say where they would prefer to live in an ideal world (Appendix 3, question 3), two Maghrebians mentioned either Algeria or Morocco but backtracked somewhat when they realised the implications of permanent resettlement.

Proficiency in Arabic did not correspond in any patterned way with religious observance. Subjects of Maghrebian descent generally observed Ramadan and preferred halal meat. In other respects, there was something of a gender divide. Some boys did daily prayers and studied the Quran, whereas for girls’ dress, particularly the headscarf, seemed most important. Only one girl claimed to wear one outside school since the wearing of such an overt religious symbol was not allowed on school premises. For some girls, the choice of a future spouse meant marrying someone of the same religion, whereas boys answered the same question in terms of ethnicity. No Maghrebian subject was prepared to go beyond expressing openness to a spouse of any ethnicity, but none specified a French or European person (cf. Flanquart, Reference Flanquart2003). There was also something of a gender divide with respect of musical taste, with girls being more favourable to the North African genre, rai, whereas boys preferred a range of western styles, two of the most frequently mentioned bands being Sniper (a Beur-Black group of Parisian origin playing a mixture of RnB and hip-hop) and Tragedy (a white American band who play RnB).

A question which may be regarded as a variant of the ‘Tebbit test’ framed in terms of which team they would support when watching a football match between France and Algeria or Morocco, only half opted for the appropriate North African team with most others preferring to remain neutral. As regards French nationality, however, no subjects expressed any hesitation about its desirability.

The subject who embraced Maghrebian culture most enthusiastically was E.D., a Franco-Moroccan girl from Lille-Sud V^e, who seemed to want to outdo her Maghrebian classmates in terms of religious observance, particularly the wearing of the headscarf (out of school) and of cultural loyalty in her liking of rai and her expressed desire to live in Morocco. On her own admission, she was consciously behaving differently from her sister, who disapproved of her conduct, as did many of her classmates, as indicated by her low IPC score. These latter may have felt that she was not, in hip-hop parlance, ‘keepin’ it real’ (cf. Cutler, Reference Cutler2003).

For non-Maghrebian subjects, the question on regional loyalty suggested that North Africa carried few attractions as a place to settle. Nor was rai spontaneously cited as a preferred musical genre, although some pupils expressed some degree of positivity towards it when asked specifically. The religious practices of their Maghrebian classmates were alien to them but readily tolerated within the imposed pluralism of the school environment, although a small number did express approval of headscarves for aesthetic reasons.

As with Turkish code-crossing in Germany, any positive score in the Arabic test implied at least a degree of tolerance towards people of migrant background, whereas any mention of Jean-Marie Le Pen (Appendix 4) as the favoured candidate for the presidency or approving attitudes to people who vote for him corresponded to a nil or low score and membership of a class, where Metropolitan French pupils were in a clear majority.

While it would be rash to claim that no elements that help to define the socio-cultural space are favourable to the cross-ethnic acquisition of Arabic, none of them seems comparable in importance to personal friendships, both within and outside school, although there are clearly indicators which reinforce gender differentiation among subjects of Maghrebian background. Nor does there appear to be any evidence that crossing is favoured by ethnic musical genres, as is the case with Turkish pop stars in Germany, and for reggaeFootnote ¹¹ in UK and hip-hopFootnote ¹² in the USA. Even the non-mainstream musical style enjoyed by Panjabis known as bhangra seemed to hold more attraction for Anglos than rai for Metropolitan French youngsters (Rampton, Reference Rampton2005).

Of course, much of Maghrebian culture is largely impenetrable for MF subjects. Arabic was at best only acquired to a very modest level and the religious practices that few Maghrebian adolescents would dare disavow, in practice excluded mostFootnote ¹³ of their peers from other ethnic backgrounds.