This dissertation aims to provide a detailed description of the primary and secondary cues associated with Norwegian phonological vowel quantity and explore the link between speech production and perception. While vowel duration is the primary, obligatory cue to long-short vowel distinctions in production, the existence and role of secondary cues in Norwegian vowel length is less thoroughly explored. Secondary cues described in Norwegian include longer consonants after short vowels, long vowels that are more peripheral in the vowel space, and long mid vowels that are diphthongized compared to their short counterparts. There also a gap in understanding how listeners use cues in the perception of long and short vowels, particularly knowledge about if and how listeners use secondary cues like vowel quality and postvocalic consonant duration. Experimental studies of the acoustics, hyperarticulation, and perception of long and short vowels were conducted to investigate these topics. The production, enhancement, and perception of long and short vowels as well as theoretical implications thereof are the focus of this dissertation. Three main questions were asked and explored in three experiments: (1) what are the acoustic correlates of Norwegian vowel quantity, (2) how is quantity enhanced by speakers to increase their intelligibility for listeners in Norwegian, and (3) how do listeners use multiple acoustic cues when perceiving the vowel quantity distinction. In production, Experiment 1 demonstrated that listeners systematically produce long vowels with: (1) longer vowel duration, (2) a different spectral quality than short vowels, and (3) shorter postvocalic consonant duration. Furthermore, long mid vowels were shown to diphthongize in a centralizing direction. Experiment 2 tested how listeners adjusted their speech in clarifying an apparent misunderstanding of their intended vowel length by a simulated interlocutor. Acoustic analysis showed that speakers adjusted their speech by lengthening the duration of long vowels and shortening that of short vowels, producing long-short pairs further apart in the vowel space (except for /ɑ/, and producing consonants after long vowels shorter. Speakers did not, however, enhance spectral movement in long mid vowels. This demonstrates that of the four cues investigated, only three are enhanced for clarity-motivated reasons. Furthermore, there were vowel-specific patterns in that speakers did not enhance the quality difference between long and short /ɑ/. Critically, Experiment 2 also supported accounts that speakers will adjust their articulations in a targeted way that is aimed at eliminating perceptual confusability of phonological contrasts. In perception, Experiment 3 tested how listeners used the acoustic correlates of quantity found in Experiments 1 and 2 and whether there would be vowel-specific perceptual strategies. Analysis showed that listeners did have vowel-specific patterns in their perception of long and short vowels. While vowel duration was used for all six pairs, vowel quality was not reliably used for long and short /ɑ/ and postvocalic consonant duration was not reliably used for long and short /u/.
The results of the experiments outlined in this dissertation suggest that (1) speakers produce long and short vowels with multiple acoustic correlates, (2) these acoustic qualities are enhanced in targeted ways to increase intelligibility of speech, and (3) listeners use secondary cues in perception with both cue- and vowel-specific patterns. The theoretical implications of these findings were further discussed in relation to cue weighting, clear speech, the production-perception link, and cross-linguistic patterns.
