Multilingual perspective in perceptions of Norwegian accents

Kamil Malarski, Magdalena Wrembel, Kamil Kaźmierski, Witosław Awedyk

The 13th International Conference on Third Language Acquisition and Multilingualism

29-31 August 2024

norwegian-accents-iam2024.netlify.app

Accent and dialect variation in Norway

  • High dialectal variation across all linguistic domains i.e. morphosyntax, phonology, lexicon
  • Four broad dialect areas, in fact many more, including sociolects (Helleland and Papazian 2005)
  • South-Eastern accents seen as the most standard (Johnsen 2015)
  • Inclusivity towards dialects

Dialectal areas

  • Western Norwegian (Vestnorsk)
  • Eastern Norwegian (Østnorsk)
  • Trøndelag dialect (Trøndersk)
  • Northern Norwegian (Nordnorsk)

Kristoffersen, Gjert. 2000. The phonology of Norwegian. Oxford: OUP.

Why some accents are seen as better or worse than others

  • Accents as indexes of how people from the region are seen
  • Dependent on salient, easily identifiable linguistic features (Llamas et al. 2016)
  • The person “down the street” will always have a stronger accent than the person you are talking to (Preston and Niedzielski 2003)
  • Voice parametrics (f0 especially, creaky voice)
  • Intelligibility + mergers potentially causing misunderstandings (Labov 2010)

Perceptions of Norwegian accents (1/2)

  • Stratified socially (Johnsen 2015)
  • Western Oslo accents more prestigious than working-class or multicultural Oslo accents (Aasheim 1995, Johnsen 2015, Svendsen and Røyneland 2008)
  • Northern Norwegian accents perceived more negatively than Southern (Sollid 2014)

Perceptions of Norwegian accents (2/2)

  • Mixing or switching between dialects seen negatively (Røyneland 2017)
  • Non-Oslo dialects viewed as more Norwegian (Røyneland 2017)
  • Boys with foreign appearance seen as less Norwegian when using Oslo dialect than when using other dialects (Røyneland 2017: 101)

Previous studies :: methods

  • A lot of studies on how upper / West Oslo accents and dialects are perceived, both using the verbal guise technique, as well as surveys in different forms (Dahl 2002; Jensen 2006: 73, Lund 2006, Hult 2008, Kristiansen 1995)
  • Nine Norwegian accents tested on a scale from ‘nice’ to ‘ugly’ (Voje 1979)

Previous studies :: findings

  • Positive attitudes in Tromsø towards other dialects (Satermo and Sollid 2021)
  • Changing one’s dialect seen as incorrect (Satermo and Sollid 2021)
  • Negative attitudes by urban Vika speakers towards upper Oslo dialects (Jensen 2006, Lund 2006)

Our Study

Research questions

  1. Are some accents of Norwegian perceived differently than others?
  2. Do L3/Ln learners of Norwegian attribute similar aesthetic judgments to Norwegian speech as L1 Norwegian listeners do? Does acquisition setting play a role?
  3. Are L3/Ln learners able to identify the origins of the accents in line with L1 speakers?

Design

  • Online survey in Qualtrics
  • Reading passage in Norwegian (Nordavinden og sola from the Norwegian dialects database www.hf.ntnu.no/nos)
  • Selected speech samples \(N=14\)
    • 10 middle-aged native Norwegians (5 f, 5 m) from five dialect areas: the Tromsø area, Trondheim, Stavanger, Kristiansand, and Oslo
    • 4 controls: non-native accents of Norwegian of different strength (L1 Polish)

Our data (L1 NO)

      
 



Our data (L3 NO)

   


Procedure

7-point Likert scale:

  • Education level
  • Pleasantness
  • Intelligence
  • Prestige
  • Friendliness
  • Self-confidence
  • Distance / proximity

Open questions:

  • Likes / dislikes
  • Characteristic features
  • Region of origin of speaker

Samples presented in randomized order
Duration: ca. 20 mins

Participants :: 3 groups

  • Polish instructed L3/Ln learners of Norwegian living in Poland \(n = 47\)
  • Polish naturalistic L3/Ln learners of Norwegian residing in Norway \(n = 25\)
  • Norwegian native speakers as controls \(n = 45\)

Participants :: profiles

Analyses

  • Likert scale ratings
    • Joint group analysis
    • Across-group comparison
    • Across-accent comparison
  • Attitudes to individual accents (likes/dislikes)
  • Accent identification
  • Characteristic features of accents (qualitative)

Statistical analysis (1/2)

  • A Principal Component Analysis (implemented in FactoMineR [Lê et al. 2008])
    • to identify the underlying dimensions behind ratings on the six traits
  • Three relevant components determined based on a visual inspection of a scree plot

Statistical analysis (2/2)

  • Three mixed-effects linear regression models fitted with lme4 (Bates et al. 2015),
    • response variable: one of the respective components
    • fixed effects: region, rater group, region by rater group interaction
    • by-participant and by-voice random intercepts
  • Post-hoc pairwise comparisons (with Tukey correction) performed with emmeans (Lenth 2024)

PCA :: Component 1

tinytable_br3i0mzk53mg8h8g5l0a
trait R² p-value
intelligent 0.87 < 0.001
educated 0.80 < 0.001
pleasant 0.79 < 0.001
self-confident 0.73 < 0.001
prestigious 0.73 < 0.001
friendly 0.67 < 0.001

PCA :: Component 2

tinytable_w4dgzz460yqrxyeqjoij
trait R² p-value
friendly 0.63 < 0.001
pleasant 0.40 < 0.001
intelligent -0.12 < 0.001
educated -0.37 < 0.001
prestigious -0.47 < 0.001

Open answers (identification)

  • The majority of correct identification came from the Norwegian native raters
  • Few L3/Ln learners able to identify the place of origin
  • Over 100 responses per voice
  • Mandal :: (corr) : Mandal (2), Kristiansand (8), Flekkefjord (2), Vest-Agder (7), Sørlandet (22), (incorr) : Bergen (6), Stavanger (5)
  • Kristiansand :: (corr) : Kristiansand (11), southern Norway (18), Vest-Agder (7), (incorr): Bergen (6), Stavanger (5), Trondheim (8), Hamar (2), Lindesnes (1), Hordaland (1), Førde (1)

Open answers (attitudes)

  • In all cases, comments were :: positive > neutral > negative
  • Over 100 comments for each accent
  • Many listeners commented on uvular [ʁ] sounds when they occurred (Kristiansand, Mandal, Stavanger)
  • Other features noticed: prosody, ka - hva replacement
  • In general, comments on likes, dislikes, “flow”, “melody”, softness and friendliness

Trondheim

Mandal

Oppdal

Kristiansand

Discussion :: RQ1

  • Some variation in how different Norwegian accents are perceived, but the differences are smaller than predicted
  • Oslo accent evaluated as the most positive of all (see Component 1)
  • Component 2: the higher an accent wrt warmth, the lower the prestige, e.g. the Tromsø accent

Discussion :: RQ2

  • Reported differences between the three groups of raters
  • Larger for Component 2 - perceived warmth and prestige for Tromsø and Kristiansand
  • Differences in speech / reading rates controlled for
  • No statistically significant difference in how male and female voices were perceived by the respondents

Discussion :: RQ3

  • Accent origin identification proved challenging
  • Native Norwegian raters > naturalistic learners > formal learners
  • Only samples with a rather high success rate: Oslo accents

Conclusions

  • Novel insights into accent perception both from the native Norwegian and multilingual perspective of language learners
    • Different accents in Norway are given different aesthetic and social attributions
  • Perceptions are aligned along different criteria
  • With more exposure and social context, non-native speakers seem to be more inclusive towards accentual variation
    • Potential implications for non-native pedagogy

Acknowledgments

Thanks to Jørn Almberg for access to the speech sample database and to Jacques Koreman for consultations.

This research is supported by Norway funds/NCN project grant GRIEG-1 (UMO-2019/34/H/HS2/00495) ADIM “Across-domain investigations in multilingualism: Modeling L3 acquisition in diverse settings”.

Thank you!

norwegian-accents-iam2024.netlify.app

magdala@amu.edu.pl