2.1 Corpora

The main body of data this study rests upon is a spoken language corpus collected by the Oslo branch of the Upus-project. The corpus consists of video-recorded interviews and conversational data from adolescents born and raised in Oslo, living in two areas with an immigrant population of 34% and 44%, respectively. The adolescents have various backgrounds. Some have Norwegian-born parents and some have foreign-born parents, and there is variation regarding their reports on multilingualism. The majority consider Norwegian to be their mother tongue, either as their sole mother tongue or in addition to another language (see Svendsen Reference Svendsen and Hvistendahl2009). The data were collected between 2006 and 2008 and are being made available through an Internet-based interface developed by Tekstlaboratoriet [The Text Laboratory] at the University of Oslo. Data from 49 adolescents are so far part of the corpus. The database and interface are still under construction and, at the time of writing, consist of interview data from 41 adolescents and conversational data from 30 of these.Footnote ¹ In addition to the Upus-corpus I will include data from the NoTa-corpus (Norsk Talespråkskorpus – Oslodelen – ‘Norwegian Spoken Language Corpus, the Oslo part’). The NoTa-corpus consists of interviews and conversational data from 166 respondents from Oslo and surrounding areas. The corpus is statistically representative regarding the respondents’ age, place of residence and socio-economic background.

Both the Upus- and the NoTa-corpora are transcribed with the support of the transcriptional software Transcriber (http://trans.sourceforge.net/en/presentation.php). These original transcriptions are meant to meet the needs of several research interests and represent the most adaptable format to be handled by a search engine. Hence, they are not suitable for conversational analyses, where features like pause lengths, stress and overlapping speech are of importance. Therefore, the excerpts serving as illustrations throughout this paper have been re-transcribed, following a partly adjusted version of the transcription guidelines presented in Hutchby & Wooffitt (Reference Hutchby and Wooffitt1998:vi–vii, 73–92). The transcription key is given in the appendix along with access information about the Upus- and the NoTa corpora. All speakers’ names are fictitious for ethical reasons.

2.2 Some methodological remarks on the nature of the Upus data

Given the fact that the Upus corpus was to be built from in-group, multiethnic youth's verbal interactions, avoiding the observer's paradox (Labov Reference Labov1972:209) was a major practical issue for the researchers to resolve in designing data collection methods. The adolescents participated in two different speech situations: an interview and a peer conversation where no adult researcher is present. Evidence for the adolescents’ susceptibility towards the two different speech situations is the fact that the amount of structural characteristics associated with a multiethnolectal speech style – such as, for instance, violations of the syntactic verb-second (V2) constraint (cf. Kotsinas Reference Kotsinas, Linell, Adelsvärd, Nilsson and Pettersson1988:267; Quist Reference Quist2000:152; Ganuza Reference Ganuza2008; see section 4.3 below) – is significantly greater in the peer conversations than in the interviews (Opsahl & Nistov to appear). The same is true for the frequency and distribution of the discourse markers discussed in this paper. The adolescents’ actual physical behaviour also indicates that there were significant differences between the two speech situations: For instance, Mike, a 15-year-old boy with parents of West-African origin, was a paragon of calmness and common sense in the interview, but when the interviewer left the floor, he suddenly turned lively and started dancing, joking and singing with his friend (see Opsahl & Røyneland Reference Opsahl and Røyneland2008).

Where no adult researcher was physically present during the peer conversation, there was a camera running. Some incidents which took place during the recordings point towards the camera as a weak or even non-intruding factor: At one point two boys try to slip some soda bottles under their sweaters, seemingly unaffected by the fact that they are being captured on videotape. When one of them points to the camera, the other one says: ‘That doesn't matter, they won't notice until they get home’. An awareness of the camera was of course there, but it varied in its intensity. The duration of the recordings gave the participants a chance to forget the presence of the camera from time to time. Researchers behind the audio recordings of Norwegian youth language in a previous project, the UNO-project (Drange & Hasund Reference Drange, Hasund, Stenström, Kotsinas and Drange2001), have reported similar experiences. Another – somewhat contrasting – point to be made here is that recorded exposure has probably become increasingly common in recent years. Young people are nowadays often used to exposure through pictures and texts of various kinds, such as YouTube, Facebook and MSN. The fact that a camera is running establishes the ground for a general mode of expression, so to speak, where the adolescents are given the opportunity to display their readymade performances as spokesmen for adolescents in a multiethnic urban reality. Quist (Reference Quist2005:76) argues that adolescents actually enter into performance acts and stylization more often than they do not.

Based on these observations and the differences in the frequency of linguistic patterns regarding the two different speech situations, I claim that we managed to get a glimpse into some of the linguistic practices used in peer-group settings. This naturalistic quality of the data set represents its strength, but is, at the same time, its limitation: I have no control over the adolescents’ actual linguistic contributions in the way a more experimental setting would hypothetically provide. However, I believe an experimental setting to be unsuitable for capturing the characteristics of a multiethnolectal speech style. As pointed out by Opsahl & Nistov (to appear), there is, for instance, a discrepancy between adolescents’ reported and actual use of multiethnolectal features.

4.1 Wolla background

4.1.2 Distribution of wolla in the two corpora

Among people in general there seems to have been developing a metonymical reinterpretation from the USE of wolla to the USERS of wolla in that wolla denotes the users themselves. The difference between the two corpora investigated (Upus and NoTa; cf. section 2 above) is typical in this respect. In the larger NoTa-corpus, containing data from speakers from all over Oslo, wolla is used as a noun or an adjective to describe adolescents in multiethnic areas and their linguistic practices. Such uses of wolla are typically connected to descriptions of the eastern parts of Oslo, where we find the largest proportion of immigrants. Two boys discussing school districts, for instance, refer to peers from the western part of Oslo as sosser ‘posh-es’ and from the eastern part as wollaer ‘wolla-s’. Another boy reports on the use of wolla-uttrykk ‘wolla-expressions’ as sånn stkant ‘such East end’. This matches Svendsen & Røyneland's (Reference Svendsen and Røyneland2008:71) claim that we are dealing with a shibboleth (cf. also Opsahl et al. Reference Opsahl, Røyneland and Svendsen2008:40). In the Upus-corpus, on the other hand, the majority of the wolla tokens are instances of wolla used as a discourse marker.

The distribution and frequency of wolla differ tremendously between the two corpora. The larger NoTa-corpus provides nine wolla tokens, whereas the Upus-corpus provides us with 137. Wolla is in fact the most frequent loan word from languages other than English in the Upus-corpus (Opsahl et al. Reference Opsahl, Røyneland and Svendsen2008:40). There is a rather clear distributional pattern regarding the adolescents’ gender and parental background, as can be seen in Table 1.

Table 1. The distribution of wolla in the Upus-corpus arranged by parental background and speakers’ gender.

The boys alone provide us with 87% of all the wolla tokens (119 of 137). The majority of the tokens found among the girls, both with foreign- and Norwegian-born parents, resemble the ones found in the NoTa-corpus: They are for the most part descriptions or meta-examples pursued in the interviews. The girl Maja, for instance, provides us with an example of what she regards to be a typical utterance used among adolescents in her surroundings:

1. (1)

The distributional pattern outlined above matches Svendsen & Røyneland's (Reference Svendsen and Røyneland2008:71) claim that the use of loan words, slang and discourse markers like wolla appears to be a masculine phenomenon (cf. also Jørgensen & Quist Reference Jørgensen and Quist2008:102). The differences regarding parental background are also striking. Of the 137 tokens, 117 come from adolescents with two foreign-born parents, and 111 of these have been produced by boys, as can be seen in Table 1. In the vast majority of these cases, wolla serves as a discourse marker; no more than 19 of these tokens are unquestionably meta-linguistic (that is cases where wolla is mentioned and/or discussed as a typical feature).

4.2 Functions of wolla

We encountered above Svendsen & Røyneland's (Reference Svendsen and Røyneland2008:71) description of wallah as ‘an intensifier and emphasizer’. I believe their claim to be indisputable, but it still leaves us with the question of WHAT it is that is being emphasized, and WHY. In the following sections, I will attempt to provide some answers to this question. We shall start by looking at the cases where elements of the original meaning, ‘to swear by Allah’, are recognizable, as in (2):

1. (2)

In (2) Farid and Samir are discussing the number of bad marks on their school report cards. When Farid claims to have as many as twenty-six, Samir challenges him to swear, and Farid's reply, wolla (line 5), may very well be translated with I swear by Allah, as it is used to perform the speech act of swearing.

The same may be said to be true about the several instances where wolla appears in argumentative passages following disagreement, as in (3):

1. (3)

Here, in (3), Waqar and Lars disagree about the length of their recording session. Waqar claims (line 1) that the interviewer said the recording lasts till the clock turns (an hour and) fifty six. Lars negates this claim (line 3). Waqar repeats his claim (line 5), this time by changing to a direct quotation of what the interviewer is alleged to have said, introduced by wolla. Lars no longer disputes Waqar's claim; on the contrary he completes Waqar's utterance cooperatively (line 7), pointing towards the time Waqar presumably has in mind. In other words, in argumentative passages, the emphasizer wolla seems to upgrade an assertion or an assessment (cf. Pomerantz Reference Pomerantz1984:65), and appears to be an efficient verbal device for winning an argument (see also example (10) below).

In the two previous examples, questions concerning the truth value of an utterance are pointed out in discourse; concerning the reliability of the reported number of bad marks on Farid's school report card in (2) and the accuracy of the remaining time mentioned in (3). This kind of presence of what we may call epistemic challengeable elements is not always the case, as can be seen in (4):

1. (4)

Farid inserts Si wolla ‘Say wolla’ in what seems to be a routine practice after each of Samir's utterances. Samir complies with Farid's request, by introducing each new utterance with wolla (lines 5 and 11). In line 11, however, his wolla is stressed and with a prolonged final vowel. Before continuing his account of his father's reaction to his bad grade in Norwegian, he is apparently emphasizing the lexical contribution, wolla, itself, more than complying with a request to perform the speech act of swearing (by Allah). It is not obvious in each case that his conversational contributions are of such a nature that the truth value of his utterances should be questioned by Farid. The fact that his father klikka litt ‘snapped a bit’ (line 5) may of course be surprising and somewhat out of the ordinary, hence provoking Farid to question this fact. On the other hand, the fact that Samir's father may give them a lift (to a soccer game mentioned earlier in the conversation) (line 1) is more difficult to interpret as questionable to such a degree that it should automatically call for the need to swear by God. There may of course be reasons, unknown to us, why this should evoke such a reaction, but an important point to be made here is that as many as 16 times throughout their conversation, Farid utters si wolla as a minimal response to Samir's utterances. It seems in many of the cases to be an automatic minimal response rather than an actual request for the performance of a specific speech act (‘to swear by God’).

The fact that wolla doesn't necessarily involve the act of swearing, in a strict sense, is also clear in examples such as (5):

1. (5)

In (5), Mike claims that if he wins in the lottery, he will share his prize with Lukas. Lukas immediately disputes Mike's claim (line 3), with an initial wolla. When Mike challenges him to swear on a handshake (line 9), Lukas denies; he cannot swear on something relying on Mike's future actions (line 11). The interesting fact is that he has presumably already sworn (cf. line 4), if wolla is to be interpreted literally. As the previous examples also illustrated, such literal interpretation is not always possible or necessary. Discourse markers often undergo a grammaticalization process. In the case of wolla, there seems to have been a typical loss of some of its original semantic content paralleled by a pragmatic strengthening. Wolla's function as an emphasizer in (5) seems to be the same as in (3) above: to strengthen an assertion.

In addition to assertions, wolla is also present in connection with assessments and evaluations, as can be seen in (6a–c) below. As is typical of the expression of interpersonal meaning, the emphasizing is done through several means in parallel (Halliday Reference Halliday1970; Lee Reference Lee1987), e.g. repetition and the addition of several emphasizers and, consequently, the choice of extreme case formulations (cf. Pomerantz Reference Pomerantz1986):

1. (6)

The ‘what’-part of the question posed at the beginning of this section, ‘what is being emphasized and why?’ may be answered as follows: Wolla emphasizes assertions, typically in cases where the truth value of an utterance is questioned by the conversation partner, as we saw in (3), or when the conversation partner challenges the interlocutor's actions or intentions, as in (5). Wolla also emphasizes assessments and evaluations, as seen in (6), typically in combination with other features. In both cases there seems to be an increase in the speakers’ affective involvement.

The examples we have summarized so far support Quist's (2008:48) description of Danish multiethnolect, where wallah serves as an intensifier to underline the importance or value of a statement. This was not, however, quite clear in example (4) above, where the exchange Si wolla/Wolla ‘Say wolla/Wolla’ appears to be a routine or an automatic strategy, seemingly independent of the epistemic status of the utterances involved. Regarding semantic content, this routine may be said to represent the opposite to the cases where wolla serves to express the speech act of swearing (by Allah), as in (2) above. A point to be made, highlighted by Quist (Reference Quist2008:48) by her choice of the phrase ‘value of a statement’, is that it is not always necessarily the epistemic status which is being challenged or emphasized; rather it can be the news value of the utterance involved. The fact that one is being challenged to swear may, at first glance, seem as a face-threatening act (cf. Brown & Levinson Reference Brown and Levinson1987), but it may at the same time serve as a routine way to convey a consecutive positive evaluation of the news value of the conversation partner's utterances. This notion may contribute to an understanding of cases like (4), where the presence of an epistemic challengeable element is absent from discourse. I will return to this point after looking at some other wolla-related discourse markers; later I will turn to the ‘why?’ part of the initial question.

4.3 Wolla and/or related markers

The metonymical reinterpretation of wolla from the use to the users, resulting in a shibboleth-status, may have nourished the development of new markers like the Norwegian counterparts jeg sverger ‘I swear’ and the imperative sverg ‘swear’, which seem to have taken on similar discourse functions. I have no diachronic data to confirm whether jeg sverger has evolved later than wolla. The chronology of this development is, in other words, speculative, but the main point is that we find markers other than wolla which have discourse functions similar to wolla. Quist (Reference Quist2008:48) mentions that wallah often appears in fixed phrases in Danish multiethnolect, Wallah jeg sværger ‘Wallah I swear’ being one of them, as well as wallah billa, where billa is a contracted form of bi ism Allah ‘in the name of God’. In (6b) above, we saw another example of a fixed phrase: wolla Koran’ ‘wolla Quran’’. These cases, where wolla appears in combination with other elements with the same or related meanings, such as the repetition of the swear-element or the addition of the name of God or the Holy Scriptures, may also be understood in light of a grammaticalization process: Wolla seems to have grammaticalized into functions where the semantic content is bleached (Hopper & Traugott Reference Hopper and Traugott1997:121) to such an extent that the element by God – and to some extent also the element swear – is absent; the most extreme case is that of wolla being used as a noun or an adjective (cf. section 4.1.2). The addition of similar markers may thus be seen as a reinforcement of the original semantic content. The increase of the speakers’ affective involvement may be related to the increase of subjectivity which has been described as typical in connection with grammaticalization processes (Traugott Reference Traugott1989). However, as indicated in this section's heading and introduction, we are facing not only wolla in combination with other markers in the corpora but also the fact that markers such as jeg sverger ‘I swear’ and sverg ‘swear’ carry similar discourse functions on their own.

The frequency of occurrence of jeg sverger and sverg in the two corpora investigated supports the assumption that we are dealing with features characteristic of the conversational style of adolescents in multiethnic areas: There are fifteen instances of (jeg) sverger and sverg in the Upus-corpus, whereas in the larger NoTa-corpus we find only five. Interestingly enough these five instances all show up in a conversation between two boys with minority background. Let us take a closer look at a conversation between two boys from the Upus-corpus, in an attempt to convey some functional aspects associated with jeg sverger:

1. (7)

In (7), we witness an argument between Gabriel and Ömer about the highest possible number of stars representing the quality of a hotel. Ömer puts forth arguments for his view; ‘it's been found out’ that it is seven (line 3) because he has seen it on television (line 7), and specifies the channel in question (line 13). Gabriel is seemingly unstoppable in his simple but effective counter-argumentative strategy; he repeats and repeats his claim ‘[It is] six’ (lines 1, 9 and 11). In line 13, Ömer tries a late – if not the last – argumentative solution: He swears (jeg sverger). Gabriel's response is another repetition of his claim (line 17), this time in a raised voice and with an added ‘No’, acknowledging that Ömer's contribution is a strong one. Ömer now gives in (line 19) and agrees with Gabriel's assessment.

This is a somewhat unusual outcome in that the most common effect of the use of wolla, jeg sverger or sverg in an argument is immediate success, in that the opponent withdraws his or her claim. We saw this in (3) and (5) above, and we see it here in (8):

1. (8)

In (8), Lars and Waqar disagree about the placement of Waqar's soccer team on the score board. Waqar claims that they are already in last place (line 1), justifies the assessment by mentioning the result of their latest match (line 5), before challenging Lars to swear on his contradictory assessment of them being in second place (line 11). Lars immediately withdraws his assessment by admitting that he must have mixed up Waqar's team with another team (line 13). As was the case with the previous example, the element swear is brought in late and can be seen as an extreme case formulation (Pomerantz Reference Pomerantz1986) or a last resort argument.

The example in (8) is included here to illustrate also the co-occurrence of the markers in question and another characteristic feature: In line 7, Lars violates the Norwegian syntactic V2 (or: verb-second) constraint, which requires a finite verb to occur immediately after a topicalized element. Lars’ statement [forrige dagen dere ikke hadde spilt]_CPvi _NPsjekka _VP has a structure typical for learners of Norwegian as a second language (see e.g. Brautaset Reference Brautaset1996). The utterance is introduced by a topicalized sentence, forrige dagen dere ikke hadde spilt ‘the other day when you hadn't been playing’. The finite verb sjekka ‘checked’ occurs after – not before – the subject, vi ‘we’. Lars has two Norwegian-born parents, and there is no indication of him being a learner of Norwegian as a second language. He displays several of these structures in his conversation with Waqar, but none in the interview situation. This pattern – i.e. verb-second violations in peer-group conversations and fewer, if any, in the interviews – is typical for many of our respondents regardless of their language background (see Opsahl & Nistov 2009). The violations of the verb-second constraint seem to be a part of the adolescents’ peer-group multiethnolectal practice alongside the use of discourse markers like wolla and sverg.

Another wolla-related marker in Norwegian is helt ærlig ‘quite honestly’. Its frequency of occurrence in the NoTa- and the Upus-corpora, shown in Table 2, may be taken as evidence of this being a typical marker for the conversations of adolescents in multiethnic areas. There are only two instances of helt ærlig among the speakers older than 26 years of age in the NoTa-corpus, both instances of helt ærlig being part of the fixed expression for å være helt ærlig ‘to be quite honest’. In the group of speakers under 26 years of age, there are nine instances of helt ærlig. Four can be regarded as discourse markers in that they do not enter into a fixed expression, serve as an adjective or take on an obligatory syntactic function in, for instance, a predicate position. Three speakers – two of whom have, interestingly enough, a minority background – provide us with these four examples. For instance, a girl with parents from Morocco says Helt ærlig jeg vet ikke ‘Quite honestly I don't know’ with a continous intonation contour, i.e. there is no pause after helt ærlig. This may be taken to indicate that we are dealing with another instance of violation of the V2 constraint (cf. example (8) above) in that the verb (vet ‘know’) is in the third place in the structure (after helt ærlig and jeg ‘I’). However, if helt ærlig is a discourse marker, then it does not have to be considered as an integral part of the syntactic structure of the sentence in which it appears. In the Upus-corpus there are as many as 25 instances of helt ærlig used as a discourse marker. An extract from the conversation between Aswan and Thomas in (9) reveals the functions that helt ærlig can have in interaction

Table 2. Distribution of Helt ærlig in the NoTa- and Upus-corpora.

^aAs mentioned in the introduction, the Upus-database is still under construction and, at the time of writing, contains interview data from 41 adolescents and conversational data from 30 of these.

1. (9)

The preferred response (Pomerantz Reference Pomerantz1984:65) to utterances with an initial helt ærlig seems to be instant and affirmative, here in (9) illustrated by Thomas’ immediate response to Aswans question (cf. lines 1 and 3). When Aswan's assertion in line 13 is opposed by Thomas’ minimal negative response, Aswan is astonished and elaborates on his assertion. On another occasion (not illustrated here), when Aswan poses a question with an initial helt ærlig, there is a long pause, indicating that he expects Thomas to take over the turn in alignment with the preferred response pattern for utterances emphasized by helt ærlig. When this doesn't happen, Aswan makes a statement, also with an initial helt ærlig. He himself takes on Thomas's preferred, and expected, response by reinstating his own initiative. This time Thomas’ response is immediate and affirmative.

As we have seen, helt ærlig occurs in both questions and assertions. Like the other discourse markers, it has the effect of demanding truthful involvement in interaction from both the speaker and the interlocutor(s): We find both jeg sverger ‘I swear’ and the imperative Sverg! ‘Swear!’ in the Upus-corpus, and Quist (Reference Quist2008:47) shows how wallah in Danish multiethnolect can be used with a rising intonation indicating a question, ‘is it true?’ and falling intonation confirming ‘it is true’.

As is the case with jeg sverger, helt ærlig also occurs in combination with wolla, as in (10) below. This excerpt also offers us an opportunity to repeat and summarize some of the functional aspects we have looked at in the previous sections.

1. (10)

The conversation in (10) provides a good illustration of our earlier observation (see section 4.2) that the presence of the discourse markers is connected to the speakers’ affective involvement. Moreover, we see that Wolla helt ærlig det er sant da (line 15) is an upgraded, late argument in Farid's attempt to convince Samir of his immense knowledge of a specific television series. Samir's initial, potentially dispreferred response (his sigh in line 13) now changes to an upgraded agreement (Pomerantz Reference Pomerantz1984:65) by bringing in evidence from another television series (line 17). Farid says that this series has been taken off the air, and emphasizes his assertion with an utterance-final wolla (line 19). Samir immediately agrees, as expected from the previous examples we have looked at. It seems that what triggered Farid towards impersonating characters form the television series in the first place (lines 9–11) is Samir's initiative in line 1, when he encourages Farid to ‘relax’ because one of the characters ‘doesn't die today’. Indirectly Samir claims to know things about the series that Farid does not. Farid is challenged and uses several means to convey his knowledge, wolla is used as one of the means to underline the news value of his utterances.

4.4 Grammaticalization revisited

On several occasions throughout this paper, the discourse markers have been connected to various aspects of a grammaticalization prosess. In this section I intend to draw the previous remarks together and summarize the assumptions underlying them.

To begin with wolla, we have seen that it is doubtful whether or not there is always a semantic correlate meaning ‘swear by Allah’ present in the conversational context (cf. example (4) above). In (11) below, the speaker (Samir) touches, somewhat indirectly, on a more generalized meaning of the phrase:Footnote ²

1. (11)

People are so stupid that they do not know that they swear in Arabic, he says. This observation is correct in that the original semantic content seems to have undergone a classical weakening of meaning typical for a grammaticalization prosess, i.e. semantic bleaching (Hopper & Traugott Reference Hopper and Traugott1997:121). This may, as suggested in section 4.3, have nourished the rise of combinatory markers conveying similar meanings in an attempt to reinforce the original semantic content. However, some of these combinatory or alternative markers also seem to be examples of elements undergoing grammaticalization. In the case of helt ærlig, we may be facing a development from the adjectival and predicative use via the fixed expression for å være helt ærlig ‘to be quite honest’ ending in the discourse marker. The distribution in the different generations represented in the NoTa-corpus (cf. Table 2 above) supports this idea, but as long as I have not had the opportunity to consult any diachronic data it must remain a hypothesis.

The markers’ favourite position seems to be utterance-initial, but wolla may also occur in utterance-final position or as a separate tag, as seen in (10), lines 3 and 19. Earlier I mentioned the possibility of regarding the speakers’ affective involvement as related to the increase of subjectivity which has been described as typical in connection with grammaticalization processes (Traugott Reference Traugott1989). Another point, not mentioned earlier, is that some of the wolla tokens in the Upus-corpus are phonologically reduced and sound like something close to wo'a. Bearing in mind that not all of the steps in the grammaticalization process are of equal relevance for every marker, a cline illustrating the grammaticalization processes that the markers discussed in this paper have undergone may be as follows:

1. (i) Independent syntactic unit (CP) with propositional content (Wolla spelled out as ‘I swear by Allah’, Jeg sverger ‘I swear’, Jeg er helt ærlig ‘I am quite honest’).

2. (ii) Fixed expression with propositional content and relatively free placement options (Wolla, for å være helt ærlig ‘to be quite honest’).

3. (iii) Discourse marker with weakened propositional content and fixed placement, typically utterance-initial (wolla, helt ærlig ‘quite honestly’).

4. (iv) Morpho-phonological reduction (wo'a; potentially also short forms sverger ‘swear’, ærlig ‘honestly’).Footnote ³

5. (v) Renewal by addition of reinforcing elements with propositional content (wolla Koran ‘wolla Quran’) or by combining several discourse markers (wolla jeg sverger ‘wolla I swear’, wolla helt ærlig ‘wolla quite honestly’).

4.5 Why?

The question asked before examining the data was ‘what is being emphasized and why?’. We dealt with the ‘what’ part of the question in connection with the analysis of example (10) above. Included in that discussion are also some of the ‘why’ answers: The discourse markers serve in many cases as a rhetorical instrument in argumentative conversational exchanges to upgrade assertions and assessments. However, in the quest for ‘why?’ answers we may also broaden our horizons by investigating some wider sociolinguistic assumptions. The speaker (Samir), who in (11) above indirectly pointed to the semantic bleaching that wolla seems to have undergone, may again be on the right track in (12):

1. (12)

Samir explains Farid's extended use of wolla by pointing to the fact that he is an Arab; hence he cannot help it (line 7). (It is a bit ironic that Farid's heartfelt assertion that he says too much wolla is actually emphasized by wolla (line 1).) By including this example, I certainly do not wish to express a deterministic view regarding a speaker's ethnic origin as a one-to-one correlate with his or her use of wolla. But it is a fact that, in the corpora investigated, the speakers using wolla most frequently have parents from Morocco and the Middle East. The same is true for the users of jeg sverger and helt ærlig. The fact that Samir chooses the phrase Du er bror ‘You're brother’ (line 7) is also interesting, because the distribution of wolla tokens in Table 1 above indicated that we seem to be dealing with a primarily masculine phenomenon. Each of the two girls who use a multiethnolectal conversational style most extensively in my data have two Norwegian-born parents. They express affection for ‘guys with brown hair and brown eyes’ and an intense sense of pride in belonging to the eastern part of Oslo, where we find the largest proportion of immigrants (Opsahl Reference Opsahl2009, Aarsæther to appear). Another young man in the data, Gabriel, refers to girls like these, i.e. girls using a multiethnolectal speech style, as ‘gangster girls playing tough’. He seems to associate the use of a multiethnolectal speech style with aggressiveness, toughness and what may be seen as prototypical masculine features.

Using a different perspective, a possible answer to why discourse markers like wolla occur frequently in our data, may be that the use of these discourse markers is shaping and is shaped by a specific conversational style, and is an act of expressing personal identity. By entering into this specific conversational style speakers express their shared background and shared (masculine) orientation towards a late-modern urban, multiethnic reality. Another fact to support the idea that we are dealing with identity construction is data taken from a questionnaire (Opsahl & Røyneland Reference Opsahl and Røyneland2008) showing that the vast majority of the Upus-adolescents share an affection for a specific style and music genre, namely hip-hop and rap. Within many segments of the hip-hop culture and rap lyrics, there are performers and stories highlighting the urban, multicultural reality. Hip-hop culture may be seen as a forum for sharing common experience of what is, to some extent, a marginalizing society (Opsahl & Røyneland Reference Opsahl and Røyneland2008). Moreover, the essence of credibility in hip-hop is to be real and genuine. Again, without regarding their affection for hip-hop music as a one-to-one correlate to their use of discourse markers like wolla, I venture to say that a conversational style with emphasis on truthfulness and epistemic focus fits well with a culture where authenticity is a key concept.