Previous studies of language contact in multilingual urban neighborhoods in Europe claim the emergence of new varieties spoken by immigrant-background youth. This paper examines the sociolinguistic conditioning of variation in allophones of Swedish /ε:/ of young people of immigrant and nonimmigrant background in Stockholm and Gothenburg. Although speaker background and sex condition the variation, their effects differ in each city. In Stockholm there are no significant social differences and the allophonic difference appears to have been neutralized. Gothenburg speakers are divided into three groups, based on speaker origin and sex, each of which orients toward different norms. Our conclusions appeal to dialectal diffusion and the desire to mark ethnic identity in a diverse sociolinguistic context. These results demonstrate that not only language contact but also dialect change should be considered together when investigating language variation in modern-day cities.

This paper examines the effects of housing segregation on variation in the vowel systems of young speakers of Swedish who have grown up in different neighborhoods of Gothenburg. Significant differences are found for variants of the variables /i:/ and /y:/, which are strongly associated with the local dialect; these two vowels also exhibit coherence. Another vowel pair, /ε:/ and /ø:/, are involved in a coherent leveling process affecting many of the central Swedish dialects but differing in degree of openness in different neighborhoods of Gothenburg. The results show that the variation is not simply a reflection of foreign background, nor of groups of youth adopting single variants; rather, a number of social factors conflate in housing segregation, which interferes with the transmission of more abstract aspects of the local dialect's vowel system to young speakers in certain neighborhoods.

Standard language cultures are characterised by beliefs in idealised standard forms of the language in question. In this paper, these beliefs are connected to the concepts of referee design and speech community, through analysis of how Swedish adolescents reflect upon and selfassess their language proficiencies. The data consist of interviews where 111 participants self-assess their Swedish, English and additional home languages. During the self-assessment, participants use different points of reference when reflecting on the different languages in their repertoires. Four main categories of answers are found, all relating to an absent referee in some manner: the participants' evaluations of other people's language proficiency compared to their own; their proficiency in other languages; their evaluation of their proficiency in relation to formal grading and feedback given in school; and their own experiences of their limitations and abilities in different situations. When assessing Swedish, participants display attitudes towards 'good' and 'bad' language and contextualise their proficiency in a way that focuses on standard language ideologies and their speech community. The same pattern does not occur when participants reflect on their other languages, indicating the important role that the peer group and speech community have in creating and facilitating these ideologies. ARTICLE HISTORY

Background/Aims: This study investigates a possible merger in the early stages between /i:/ and /y:/ among young speakers in Gothenburg, Sweden. Methods: (1) A large-scale online perception experiment testing listeners' abilities to identify the two vowels and (2) acoustic analysis of 705 vowels from 19 speakers. Results: The perception study shows that listeners classify the horizontally centralized /y:/ as /i:/, both in isolated vowel items and in items containing the full word. This indicates that /y:/ is moving into the perceptual space of /i:/. Listeners also classify the unmerged /y:/ as /i:/ when listening to [y:] in isolation, indicating that lip rounding is a perceptually weak feature, for this centralized vowel, in this variety. The acoustic analysis shows that /i:/ tends to be produced as [ɨ:], and that there is no acoustic difference between /i:/ and /y:/ in measurements correlated with the first two formants, i.e. lip rounding is the most important distinctive feature. Conclusion: Results point in the direction of an incipient vowel merger, following a merger-byapproximation model. These results indicate a lack of perceptual strength of an articulatory feature in the disappearing phoneme, namely lip rounding, and the consequent perceptual similarities between the horizontally centralized [ɨ:] and /y:/.

No abstract