Rhythm in Stockholm’s two working-class varieties:Separate models predict intervocalic durational contrast

Nathan J. Young

Queen Mary, University of Londonnathan.j.young@qmul.ac.uk

Abstract

This study shows that two distinct social models predict speechrhythm variation – measured by the normalized pairwise vari-ability index of vowels (nPVI-V) – for Stockholm’s two workingclasses. The non-white working-class variety (multiethnolect)has less intervocalic durational contrast than the speech of elites(41–51 vs. 49–57) and correlates with the speaker’s neighbor-hood diversity. Incremental increases in neighborhood diver-sity correlate with incremental decreases in nPVI-V. The whiteworking-class variety has more intervocalic durational contrastthan the speech of elites (53–61 vs. 49–57) and correlates to oc-cupational status. Incremental decreases in occupational statuscorrelate with incremental increases in nPVI-V.

The data comes from 31 male Stockholmers, ages 24–49,who read aloud a passage with 285 vocalic elements. Thirteenself-identify as white ‘Swedes’: five working class, eight upper-middle class (‘elites’). Eighteen self-identify as non-white ‘im-migrants’: five working class, seven lower-middle class, sixupper-middle class (‘elites’). Twenty-eight were born in Swe-den; three arrived before age four. They hail from five neighbor-hood types that are representative of Stockholm’s geographicethnic distributions.

The findings add Swedish multiethnolect to a growing listof contact varieties with less intervocalic durational contrastthan their heritage counterparts. The findings also nudge ourfield’s discussion of rhythm away from second-language acqui-sition to the social domain of race and class. At the same time, anew research question emerges whether intervocalic durationalcontrast is a sociolinguistic variable in its own right or a byprod-uct of segment-level variation.Index Terms: multiethnolect, nPVI, Rinkeby Swedish, Stock-holm Swedish, variationist sociolinguistics, vernacular

1. Background1.1. Ethnosocial dialects of Europe and Stockholm

Much of Europe is witnessing the development of new soci-olects that are specifically late-modern and referred to as con-temporary urban vernaculars [1]. Stockholm’s multiethnolectis part of this linguistic phenomenon that has been exclusivelyseen (thus far) in cities of European countries that ran postwarguest-worker programs. These include Berlin [2], Copenhagen[3], Gothenburg [4], Hamburg [5], London (Multicultural Lon-don English, ‘MLE’) [6], Malmö [7], Mannheim [8], Oslo [9],Paris [10], Rotterdam [11], and the South Midlands [12].

1.2. Adult speakers accentuate class and age strata

Most of the aforementioned works are investigations of youthlanguage. However, a growing number of studies have ex-amined sociolinguistic variation among British-Asian adults,

young and old [1, 13], many of whom are MLE speakers. Re-search on Germany’s Turkish diaspora has also improved ourunderstanding of adult ethnosocial linguistic practice [8].

Such research is important because the repertoires typi-cally ascribed to youth have already matured into ethnoso-cial dialects that are spoken by a diverse non-white class ofworking-age adults [1, 14, 15, 16]. Few researchers have in-vestigated this ‘adult question’ in Sweden. One ethnographicstudy of Assyrian-Swedish men in Stockholm (n=3) found thatthey style-shifted from ‘native’ speech at work to ‘non-native’speech in private [17], echoing [1] and [13]. The findingschallenge outdated notions in Sweden about L2 deficiency andimply that the ‘non-native’ register was actually Stockholm’snew ethnosocial dialect. A later study found that six speakersfrom multiethnic suburbs were assessed differently by listen-ers in a perception experiment: two were said to be ‘neutral’and ‘Swedish’ while four were assessed as ‘rough’ and ‘non-Swedish’ [16]. This was attributed to the former two style-shifting like the Assyrian-Swedes of the first study [17].

1.3. Geography matters in a segregated Stockholm

In this paper, multiethnolect [18] refers to the variety often spo-ken in Stockholm’s multiethnic neighborhoods. Sundry termscirculate, but Rinkeby Swedish [19, 20, 21, 22] remains the mostwell known, although it no longer is solely bound to the immi-grant suburb, Rinkeby. It has other 21ST-century geographicindexicalities, including the outer northwestern and southwest-ern suburbs [21, 23] (Figure 1).

Like in other segregated European cities [24], ‘neighbor-hood matters’ in Stockholm and is stereotyped to certain socialclasses and types. In this manner, multiethnolect is like other va-rieties: working-class Södersnack (aka Ekensnack) is tied to Sö-dermalm and the inner southern suburbs; ‘posh’ Stockholmianis tied to Danderyd and Östermalm [15, 23, 25] (Figure 1).

1.4. Two varieties in symbolic opposition

Söderssnack was spoken by working-class Stockholmers formost of the twentieth century until other speech styles began toemerge, including multiethnolect. Although both Södersnackand multiethnolect ‘belong’ to Stockholm’s working class, thesimilarities end there. The former is often attributed to whitespeakers in the inner southern suburbs and the latter to non-white speakers in the outer northwest (Sw. Västerort; En. West-side) and southwest (Sw. Söderort; En. Southside). The formeris often described by the media with endearment [26] and thelatter with alarm [20, 22].

Manufacturing’s decline and migration’s rise have stokedtension between the white working class and the multiethnicworking class, aiding the rise of the national socialist Swe-den Democrats (Sw. Sverigedemokraterna) [27, 28]. Indeed,one striking finding of the ethnographic part of this study was

9th International Conference on Speech Prosody 201813-16 June 2018, Poznań, Poland

448 10.21437/SpeechProsody.2018-91

that all white working-class participants vote Sweden Democratwhereas the multiethnolectal participants vote Left (Sw. Vän-sterpartiet) or Social Democrat (Sw. Socialdemokraterna).

Two studies have examined multiethnolectal features injuxtaposition to white working-class speech [6, 10], but to date,no study has examined systematic variation in Stockholm. Asfar as prosody is concerned, one pilot study has examined multi-ethnolect (see 2.2) [16]. Prosody in Södersnack has never beenstudied, but some of its long vowels are more diphthongal thanthe standard [25, 29], which might mean more intervocalic du-rational contrast than in multiethnolect and standard styles.

2. Speech rhythm2.1. Defining the variable

The study of prosodic rhythm emerged as a way to measure de-grees of isochrony, a now-disproven concept [30] (see [31] for amore complete review). It was assumed that there is “a straight-forward relationship between duration and abstract phonolog-ical categories, such as syllable structure, vowel weight andvowel reduction patterns” [31]. Therefore, languages with lowintervocalic durational contrast were assumed to be ‘syllable-timed’ and languages with high intervocalic durational contrastwere assumed to be ‘stress-timed’.

This study intentionally disengages from the taxonomy oftiming per se. Its catalyst are the many works that describeRinkeby Swedish as ‘staccato’ or ‘jerky’ (Sw. stötig) [15, 16,19, 22]. Plainly stated, this study seeks to investigate whether“the regular alternation of strong and weak elements [32]” is initself a sociolinguistic variable in late-modern Stockholm.

One could speculate that ‘strong’ and ‘weak’ would mani-fest themselves in F0 and intensity. However, these two featuresrely heavily on the progression of time in order to operational-ize contrast. This claim is supported by research that has foundthat speech prominence – while influenced by F0 and intensity– is nonetheless highly reliant on duration [33, 34]. Therefore,this paper defines rhythm as local durational alteration.

Aggregate or global metrics [35] like percentage of vowelduration to segmental duration [36], standard deviation [36],and variation coefficients [37, 38] do not model the localessence of strong and weak elements [35]. Therefore, I use alocal metric [35] known as the normalized pairwise variabilityindex for vowels (nPVI-V) [39]. It creates a numerical represen-tation for the durational alternation of consecutive vowels andnormalizes for speech rate:

nPV I =|dn+1 � dn|

dn+1+dn

2

· 100 (1)

where d = duration of the first vowel in the pairwhere dn+1 = duration of the second vowel in the pair

2.2. Rhythm in contact varieties and Swedish

Most analyses of rhythm in contact varieties have conceptual-ized contrast as durational, and many have used the nPVI-Valgorithm to model this (cf. [3, 7, 10]). Most of these studiesshow reduced contrast for contact varieties than their heritagecounterparts [6, 35, 39, 40, 41, 42] (cf. [35, 41]).

Swedish prosody has been examined with nPVI-V twice.In a small pilot study of male speakers (n = 8, 50 vocalicelements), speech assessed as multiethnolectal had lower in-tervocalic durational contrast (n = 4: 39.2, 40.6, 45.3 &47.3, respectively) than speech assessed as the received stan-dard (n = 4: 52.6, 52.7, 54.8 & 54.9, respectively) [16].

Another study of female speakers (n = 10, 120 vocalic ele-ments) showed that standard Central Swedish has an nPVI-Vthat ranges between 46.0 and 57.4 and that L2 Swedish spo-ken by L1 Estonians ranges between 44.7 and 53.0 [42]. Thesestudies were preceded by one other that measured the durationof vowel phonemes, hypothesizing that long vowels shorten andshort vowels lengthen in Malmö multiethnolect. Both hypothe-ses were invalidated [7].

3. Definitions and research questionsDEFINITIONS: In this paper, the upper-middle class are some-times referred to as elites. I use white instead of ‘Swede’ or‘native’ and non-white and of color instead of ‘immigrant’ or‘non-native’. This very topic has been tangentially [11, 14, 16,17, 20] and explicitly [43, 44] explored elsewhere but will notbe handled here.RESEARCH QUESTION A: Is the intervocalic durational con-trast of Stockholm’s non-white working-class variety, i.e., mul-tiethnolect, lower than both the elite variety and the whiteworking-class variety?RESEARCH QUESTION B: Is the intervocalic durational con-trast in Stockholm’s white working-class variety higher thanboth the elite variety and multiethnolect?

4. Method4.1. Speakers

Thirty-one adult male speakers, ages 24–49, participated in thestudy, recruited via the snowball method. For the purposes ofcontrolling for social class and ethnicity, women were excluded(see discussion in 8.4). The speakers are categorized by socialclass according to the Swedish Standard Classification of Occu-pations 2012 (SSYK) [45].

Thirteen self-identify as white ‘Swedes’ (Sw. svensk): fiveworking class, eight upper-middle class. Eighteen self-identifyas non-white ‘immigrants’ (Sw. invandrare: five working class,seven lower-middle class, and six upper-middle class. Elevendifferent ethnic heritages are represented: Azeri (1), Chilean(1), Egyptian Arab (1), Iranian (3), Kurdish (3), Mbembe (1),Peruvian (1), Qashqai (1), Somali (1), Tigrinyan (4), and Turk-ish (1).

All speakers were born in Sweden except for three who ar-rived before the age of four. All speakers reported Swedish astheir strongest language for speaking and understanding1.

4.2. Speech material and phonetic analysis

The participants read an adapted version of the The Circus (Sw.Cirkusen) [46], a reading passage that includes multiple exam-ples of all Swedish phonemes and pitch accents. For the anal-ysis, I removed disfluencies and hesitation markers, resultingin data ranging from 215 to 277 vocalic elements, depending onelisions. Earlier benchmarks for the number of vocalic elementsanalyzed are n = 237 [6], 80 [39], and 200 [40].

The final foot before a pause was included in the calcula-tion [6, 35, 38, 39], cf. [40, 41]. Syllable-coda /r/ and /j/ wereincluded as part of the vowel, whereas intersyllabic /r/ and /j/were not [40]. Breath groups were delineated by pauses of 70ms or more [40] (cf. 150 ms [35]).

1Some elites reported English as equally strong.

449

4.3. Social predictor variables

Age, social class and race/ethnicity are typical social factors totest. In Stockholm’s context, neighborhood is also important,given the strong geographic indexicality of the city’s varieties.

Birth year was used for age. For social class, the status rankof each speaker’s occupation was used [47]. Status rank cap-tures the incremental differences in symbolic capital between,say, a waiter and a demolition worker (both working class) or adoctor and a portfolio manager (both upper-middle class).

Neighborhood was broken into five classes that corre-spond to five tiers of racial diversity: westside working-classcommunities of color (e.g., Rinkeby, Tensta: 90–100% res-idents of migrant origin), southside working-class communi-ties of color (e.g., Botkyrka, Skärholmen: 80–90% migrant),mixed working-class communities (e.g., Fisksätra, Hässelby,Rågsved: 40–60% migrant), white working-class communities(e.g. Bagarmossen, Farsta: 20% migrant), Central Stockholmand affluent villa suburbs (e.g., Danderyd, Saltsjöbaden: 0–10%migrant). The average percentage of residents of migrant originfor each neighborhood was entered into the analysis: 95, 85, 50,20, and 5 for the above, respectively.

For race, white and non-white were entered as binarydummy variables 0, 1 (see definitions in Section 3).

All four predictors were tested for orthogonality with aPearson product moment correlation (n = 31). Occupationalstatus ⇠ race are orthogonal (R = �0.22), but each of themcorrelated with neighborhood diversity (R = 0.69 and R =�0.68, respectively), which implies that neighborhood is a ro-bust factor for capturing the intersection of race and class. Age⇠ neighborhood (R = 0.55) and age ⇠ race (R = �0.57)have medium correlation, which means they can be entered intoa multiple regression together but interpreted cautiously. Occu-pational status ⇠ age have low correlation (R = 0.16).

5. Overview of ResultsFigure 1 presents all the data with occupational status on the xaxis and nPVI-V on the y axis. Race and neighborhood are su-perimposed to provide visual insight into their collinearity withneighborhood. Two models emerge: one for the relationshipbetween elite speech and white working-class speech, and theother for the relationship between elite speech and non-whiteworking-class speech. The positive-sloped least-squares regres-sion line represents the nPVI-V trend between elites and themultiethnic working class; the negative-sloped one representsthe nPVI-V trend between elites and the white working class.

Figure 1 shows that speakers from neighborhoods withlower percentages of immigrant residents tend to speak withmore durational contrast, albeit with one exception. The whiteworking class almost entirely inhabits neighborhoods with ca.20% residents of migrant origin. They are marked with * inthe plot. Despite their moderate proximity to residents of mi-grant origin, they appear to defy the general trend for nPVI-V.For that model, occupational status seems to be a more relevantpredictor than neighborhood diversity.

6. Question A - MultiethnolectAfter removing the five white working-class speakers (denotedwith *) and running the following linear regression, nPVI-V ⇠Neighborhood + Birth_year (n = 26), neighborhood diver-sity shows a significant effect on intervocalic durational con-trast (p < 0.01). Table 1 displays the results. The model indi-

Figure 1: Demographic map of Stockholm and scatterplot of oc-cupational status, neighborhood diversity, and race as predic-tors for nPVI-V. Neighborhood diversity appears also in map.The upper least-squares regression line models occupation andnPVI-V for elites and the white working class. The lower linemodels occupation and nPVI-V for elites and the non-whiteworking class. Asterisks (*) denote the white working class.

Table 1: Regression analysis – Neighborhood and age as pre-dictors for nPVI-V with the white working class excluded.

Dependent variable:

Observations: 26 nPVI-V

Neighborhood_diversity �0.070⇤⇤⇤ (0.017)Birth_year �0.162 (0.106)Constant 375.441⇤ (210.918)

R2 0.649Variance inflation factor 2.851Note: ⇤p<0.1; ⇤⇤p<0.05; ⇤⇤⇤p<0.01

450

cates that a 1% increase in a speaker’s neighborhood diversityresults in a 0.070 drop in nPVI-V. In other words, being froma multiethnic neighborhood correlates with lower intervocalicdurational contrast. Age was not a significant predictor.RESEARCH QUESTION A appears to be confirmed. The inter-vocalic durational contrast of Stockholm’s non-white workingclass speech (multiethnolect) – defined by non-white speakersfrom immigrant-dense neighborhoods – is lower than that ofnon-white speakers from mixed neighborhoods and white andnon-white speakers from affluent white neighborhoods.

7. Question B - White working-class speechA clear trend emerges in Figure 1 if one excludes the speak-ers from neighborhoods with 50, 85, and 95 diversity percent-ages. With only two neighborhood diversity tiers remaining –20% and 5% – occupational status emerges visually as a morerelevant predictor2. The trend suggests that low occupationalstatus coincides with higher nPVI-V’s, which was tested witha new regression model whereby all non-white working-classparticipants were removed: nPVI-V ⇠ Occup_status_rank +Birth_year (n = 17).

Table 2: Regression analysis – Occupational status and age aspredictors for nPVI-V, non-white working class excluded.

Dependent variable:

Observations: 17 nPVI-V

Occup_status_rank 0.072⇤⇤⇤ (0.018)Birth_year �0.160⇤⇤ (0.065)Constant 368.081⇤⇤ (129.004)

R2 0.707Variance inflation factor 3.413Note: ⇤p<0.1; ⇤⇤p<0.05; ⇤⇤⇤p<0.01

Table 2 shows that occupational status has a significant pos-itive effect on nPVI-V when non-white working-class speakersare removed from the data set. Birth year emerges as significantand has a mild negative effect. The model indicates that a singu-lar drop in occupational status rank will result in a 0.072 highernPVI-V. For every decade younger, the nPVI-V of a speakerwould decrease by 1.6.RESEARCH QUESTION B appears to be confirmed. The inter-vocalic durational contrast in Stockholm’s white working-classspeech is higher than of elite and multiethnolectal speech.

8. Discussion8.1. Descriptive findings

The findings suggest that for male speakers, multiethnolectranges between 41 and 50 for nPVI-V, and elite Swedish ap-pears to range between 50 and 57, which is consistent with ear-lier findings [16, 42]. White working-class Swedish ranges be-tween 53 and 60.

8.2. Addressing the research questions

Two separate models operate. One model is the relationshipbetween the speech of elites and that of the multiethnic working

2Interestingly, status rank has caused an SSYK-designated [45]middle-class speaker (special-education teacher, rank 52) to be super-seded in status by a working-class speaker (hotel chef, rank 45).

class, in which geography appears to be especially relevant. Anincrease in neighborhood diversity results in lower intervocalicdurational contrast than of those who have high-status work andlive in affluent neighborhoods. This is particularly the case forwestside and southside communities. Non-white working-classspeakers from mixed communities appear to have an nPVI-Vthat is consistent with speakers from affluent communities.

The other model is the relationship between the speech ofelites and white workers, in which social class appears to bemore relevant. A decrease in occupational status results inhigher intervocalic durational contrast than of those who arehigh-status and live in affluent neighborhoods. For the whiteworkers, the increased neighborhood diversity does not appearto reduce intervocalic durational contrast.

8.3. White working-class results: legacy or innovations?

The results for the white working class are difficult to interpretbecause the subsample is small (n = 5). Are they speaking aversion of Stockholm’s older Södersnack or are they engagingin an oppositional speech act in reaction to their multiethnicrivals? More speakers should be added to the corpus to informthe analysis.

8.4. Future directions

Examining other linguistic variables that co-occur with speechrhythm can help paint a more accurate picture of what actu-ally constitutes these repertoires. Examining other speech stylesaside from reading, such as casual speech, can also illuminateto what degree these findings constitute style-shifting and towhat degree they actually reflect late-modern vernacular stylein Stockholm. Furthermore, the use of a single occupationalstatus metric [47] may not be the best way to capture socialclass. Metrics that take into account parental occupation, ed-ucation, occupational status, income and taste should also beconsidered.

It is unclear whether intervocalic durational contrast is avariable in its own right or whether these results are a byprod-uct of diphthongs in white working-class speech and monoph-thongs in multiethnolect. The findings call for a phonologicalinvestigation into the role of underlying segmental variation.

A final point is that the speech of women in Stockholm isnot included in this study. Until data on women’s speech iscollected and analyzed, no full conclusion on prosody in late-modern Stockholm Swedish can be reached.

9. AcknowledgementsThis work was supported by Queen Mary, University of London,the British Economic & Social Research Council, the Centrefor Research on Bilingualism at Stockholm University, Sven ochDagmar Saléns stiftelsen, and the guidance of Devyani Sharma,Erez Levon, and Erik Thomas. All errors are my own.

10. References[1] B. Rampton, “From ‘multi-ethnic adolescent heteroglossia’ to

‘contemporary urban vernaculars”’ Language & Communication,vol. 31, no. 4, pp. 276–294, 2011.

[2] H. Wiese, Kiezdeutsch: ein neuer Dialekt entsteht [Kiezdeutsch: Anew dialect emerges]. Munich: C.H. Beck, 2012.

[3] G.F. Hansen & N. Pharao, “Prosody in Copenhagen multieth-nolect,”. in P. Quist & B.A. Svendsen (Eds.) Multilingual Ur-ban Scandinavia. New Linguistic Practices. Toronto: MultilingualMatters, pp. 79–95, 2010.

451

[4] J. Gross, S. Boyd, T. Leinonen & J.A. Walker, “A tale of two cities(and one vowel): Sociolinguistic variation in Swedish,” LanguageVariation and Change, vol. 28, no. 2, pp. 225–247, 2016.

[5] I. Dirim & P. Auer, Türkisch sprechen nicht nur die Türken. Überdie Unschärfebeziehung zwischen Sprache und Ethnie in Deutsch-land [Not only Turks speak Turkish. On the Blurry Relationshipbetween Language and Ethnicity in Germany]. Berlin: De Gruyter,2012.

[6] E. Torgersen & A. Szakay “An investigation of speech rhythm inLondon English,” Lingua, vol. 122, no. 7, pp. 822–840, 2012.

[7] P. Bodén, “‘Rosengårdssvensk’ fonetik och fonologi” [RosengårdSwedish phonetics and phonology] in L. Ekberg (Ed.), Språket hosungdomar i en flerspråkig miljö i Malmö (Nordlund 27). Lund:Lund University Press, pp. 1–47, 2007.

[8] I. Keim, “Socio-cultural identity, communicative style, and theirchange over time: A case study of a group of German-Turkish girlsin Mannheim, Germany” in P. Auer (Ed.) Style and social iden-tities. Alternative Approaches to linguistic Heterogeneity. Berlin/New York. De Gruyter, pp. 155–186, 2007.

[9] B.A. Svendsen & U. Røyneland, “Multiethnolectal facts and func-tions in Oslo, Norway,” International Journal of Bilingualism, vol.12, no. 1–2, pp. 63–83, 2008.

[10] Z. Fagyal, “Rhythm types and the speech of working-class youthin a banlieue of Paris: the role of vowel elision and devoicing” inD.R. Preston & N. Niedzielski (Eds.) A Reader in Sociophonetics.Berlin: Mouton de Gruyter, pp. 91–132, 2010.

[11] L. Cornips & V.A. de Rooij, “Selfing and othering through cate-gories of race, place, and language among minority youths in Rot-terdam, The Netherlands” in P. Siemund, I. Gogolin, M.E. Schulz& J. Davydova (Eds.) Multilingualism and language diversity inurban areas: Acquisition, identities, space, education 1. Amster-dam: John Benjamins, pp. 129–164, 2013.

[12] B. Rampton, Language in late modernity: Interaction in an urbanschool. Cambridge: Cambridge University Press, 2006.

[13] D. Sharma & B. Rampton, “Lectal focusing in interaction: a newmethodology for the study of style variation,” Journal of EnglishLinguistics, vol. 43, no. 1, pp. 3–35, 2006.

[14] N. Young, Suburban Swedish Maturing – Variation and perceptionsamong adult speakers of Swedish contemporary urban vernacular.(MA thesis). Dept. of Linguistics, Stockholm University, 2014

[15] E. Bijvoet & K. Fraurud, “What’s the target? A folk linguistic studyof young Stock-holmers’ constructions of linguistic norm and vari-ation,” Language Awareness, vol. 25, no. 1–2, pp. 17–39, 2016.

[16] N. Young, “Talrytmens sociala betydelse i det senmoderna Stock-holm – Vokaldurationskontrast som ett indexikalt drag” [The socialmeaning of speech rhythm in late-modern Stockholm – Intervocalicdurational contrast as an indexical feature], accepted & under revi-sion, Nordand – Nordisk tidsskrift for andrespråksforskning, 2018.

[17] J. Eliaso Magnusson & C. Stroud, “High proficiency in markets ofperformance,” Studies in second language acquisition, vol. 34, no.2, pp. 321–345, 2012.

[18] M. Clyne, “Lingua franca and ethnolects in Europe and beyond,”Sociolinguistica, vol. 14, no. 1, pp. 83–89, 2000.

[19] U.B. Kotsinas, “Rinkebysvenska – en dialekt?” [Rinkeby Swedish– a dialect?] In P. Linell, V. Adelswärd, T. Nilsson & P.A. Petersson(Eds.) Svenskans beskrivning, 16. Linköping: Linköping Univer-sity Press, pp. 264–278, 1988.

[20] C. Stroud, “Rinkeby Swedish and semilingualism in language ide-ological debates: A Bourdieuean perspective,” Journal of Sociolin-guistics, vol. 8, no. 2, pp. 196–214, 2004.

[21] E. Bijvoet & K. Fraurud, “Rinkeby Swedish in the mind of the be-holder. Studying listener perceptions of language variation in mul-tilingual Stockholm” in P. Quist & B.A. Svendsen (Eds.)Linguisticpractices in multiethnic urban Scandinavia. Clevedon: Multilin-gual Matters, pp. 170–188, 2010.

[22] T.M. Milani & R. Jonsson, “Who’s Afraid of Rinkeby Swedish?Stylization, Complicity, Resistance,” Journal of Linguistic Anthro-pology, vol. 22, no. 1, pp. 44–63, 2012.

[23] E. Bijvoet & K. Fraurud, “Studying high-level (L1-L2) develop-ment and use among young people in multilingual Stockholm,”Studies in Second Language Acquisition, vol. 34, no. 2, pp. 291–319, 2012.

[24] L. Wacquant, Urban outcasts: A comparative sociology of ad-

vanced marginality. Cambridge: Polity, 2008.[25] U.B. Kotsinas, “Snobbar och pyjamastyper: Ungdomskultur, ung-

domsspråk och gruppidentiteter i Stockholm” [Snobs and Pyjamatypes: Youth culture, youth language and group identities in Stock-holm] in J. Fornäs, U. Boethius, M. Forsman, H. Ganetz & B.Reimer (Eds.) Ungdomskultur i Sverige. Stockholm: BrutusÖstlings Bokförlag, pp. 311–336, 1994.

[26] F. Lindström, F (host), “Stockholmska” [Stockholmian] (televisionseries episode) in M. Hellberg (producer) Svenska dialektmysterier[Swedish dialect mysteries], season 1, episode 4, Stockholm: Swe-den’s Television, 2006.

[27] D. Mulinari & A. Neergaard, “The Sweden Democrats, racisms andthe construction of the Muslim threat” in G. Morgan & S. Poynting(Eds.) Global Islamophobia: Muslims and Moral Panic in the West.Farnham: Ashgate, pp. 67–82, 2012.

[28] Open Society Foundations, Europe’s white working-class commu-nities: Stockholm. New York: Open Society Foundations, 2014.

[29] J. Öqvist, “Riktig stockholmska? Ekenssnacket som varietet ochfenomen” [Real Stockholmian? Ekenssnack as a variety and phe-nomenon], Studier i svenska språkets historia, 11 (Acta AcademiaeRegiae Gustavi Adolphi CXIII), pp. 253–260, 2010.

[30] R.M. Dauer, “Phonetic and phonological components of languagerhythm,” Proceedings of the 11th International Congress of Pho-netic Sciences, vol. 5, pp. 447–450, 1987.

[31] A. Arvaniti, “Rhythm, timing and the timing of rhythm,” Phonet-ica, vol. 66, no. 1–2, pp. 46–63, 2009.

[32] E. Lerdahl & R. Jackendoff, A generative theory of tonal music.Cambridge, MA: MIT Press, 1983.

[33] R.E. Cumming, Speech rhythm: the language-specific integra-tion of pitch and duration (Ph.D. dissertation), University of Cam-bridge, 2010.

[34] S. Al Moubayed, G. Ananthakrishnan & L. Enflo, “Automaticprominence classification in Swedish,” Speech Prosody, 2010.

[35] R. Fuchs, Speech rhythm in educated Indian English and BritishEnglish (Ph.D. dissertation), Westfälische Wilhelms-Universität,2013.

[36] F. Ramus, F. Nespor & J. Mehler, “Correlates of linguistic rhythmin the speech signal,” Cognition, vol. 73, pp. 265–292, 1999.

[37] V. Dellwo, “Rhythm and speech rate: A variation coefficient fordeltaC,” Language and Language Processing: Proceedings of the38th Linguistic Colloquium, Peter Lang, pp. 231–241, 2006

[38] L. White & S.L. Mattys, “Calibrating rhythm: First language andsecond language studies,” Journal of Phonetics, vol. 35, no. 4, pp.501–522, 2007.

[39] E.L. Low, E. Grabe & F. Nolan, “Quantitative Characterizations ofSpeech Rhythm: Syllable-Timing in Singapore English,” Languageand Speech, vol. 43, no. 4, pp. 377–401, 2000.

[40] E. Thomas & P. Carter, “Prosodic rhythm and African AmericanEnglish,” English World-Wide, vol. 27, pp. 331–355, 2006.

[41] P. Sarmah, D. Verma Gogoi & C. Wiltshire, “Thai English –Rhythm and vowels” in L. Lim & N. Gisborne (Eds.) Typologyof asian englishes. Philadelphia: Benjamins, pp. 75–96, 2011.

[42] D. Krull, “Rhythmic Variability and Swedish-Estonian LanguageContact,” Proceedings of FONETIK 2012, pp. 81–84, 2012.

[43] C. Lacatus, “What is a blatte? Migration and ethnic identity in con-temporary Sweden,” Journal of Arab & Muslim Media Research,vol. 1, no. 1, pp. 79–92, 2007.

[44] T. Hübinette, H. Hörnfeldt, F. Farahani & R.L. Rosales, “Om rasoch vithet i det samtida Sverige” [On race and whiteness in con-temporary Sweden] in (same Eds.) Om ras och vithet i det samtidaSverige. Stockholm: Mångkulturellt Centrum, pp. 11–36, 2012.

[45] SSYK, Standard för svensk yrkesklassificering [Standard forSwedish occupational classification]. Statistics Sweden, 2012.

[46] U. Morris & H. Zetterman, Från bondgård till cirkus. Konstruktionav en högläsningstext för bedömning av röst-och talfunktion ochtalandning [From farm to circus. Constructing a reading passagefor assessment of voice and speech faculties] (MA thesis). Dept. ofSpeech Pathology, Karolinska Institutet, 2011.

[47] Y. Ulfsdotter Eriksson, Yrke, Status & Genus - En sociologiskstudie om yrken på en segregerad arbetsmarknad [Occupation, Sta-tus & Gender - A sociological study of occupations in a segregatedlabour market] (Ph.D. dissertation), Gothenburg University, 2006.

452