Sandhi of /rs/ sequences in North Norrland Swedish

Abstract

This dissertation investigates the phonology of /rs/-sandhi in North Norrland, the northernmost 40% of Sweden. North Norrland is of particular phonological interest because it is the locus of a neutralization between the phonetic product of /rs/-sandhi, [s], and the phoneme of Swedish known as sje or /f/, which in this region is most often also pronounced as [s]. The phonology of /rs/-sandhi includes four overlapping areas of variability: the choice of variants of sje (Chapter 3), the occurrence of internal /rs/-sandhi (Chapter 4), the occurrence of external /rs/-sandhi (Chapter 5), and the production of the copula ar 'is/are/am' with an unreduced coda-/r/ (Chapter 6).I collected new data in North Norrland with two stories, and examined A using three population factors (the age, gender and dialect area of speakers) and several linguistic factors (including the lexical items, syntactic categories of words, position and types of adjacent stress, and c-command relationship between words). This dissertation focuses on describing the patterns of variation in this data, but also rams many issues of theory. I found a wide range of variation between the various lexical items for each of the four areas of interest, and sought to explain them by other factors The population factors of age, gender and dialect played a strong role in regard to the choice of variants of sje, a smaller role in regard to the unreduced, copula ar, and had little influence on internal or external /rs/-sandhi. Both internal and external /rs/-sandhi appear to be influenced by adjacent segments, although data limitations reduce confidence levels. Closeness of the two elements favored sandhi. C-command by the word on the right favored sandhi. For each of the four sets of data, stress was a major factor Primary stress on the following vowel had a range of influences; it facilitated the choice of dark-sje variants, favored internal /rs/-sandhi, disfavored external /rs/-sandhi, and disfavored the maintenance of coda-/r/ on the copula ar. There was no evidence that frequency of occurrence of the words was a meaningful factor in any choice of variants.