Articulatory characterisation of vowel length contrasts in Australian English

Many languages contrast long and short vowels, but the phonetic implementation of vowel length contrasts is not fully understood. This dissertation explores the articulation of vowel length contrasts in AusE to better understand their phonological representation and phonetic implementation. In particular, this dissertation aims to understand how the contrast between long and short vowels may arise both from intrinsic differences between long and short vowel gestures and their timing relationships with flanking consonants. Furthermore, I wish to explore how speakers maintain the articulatory differences between long and short vowels preceding voiced and voiceless codas and across different speech rates. To accomplish this, tongue dorsum and labial kinematics were examined during production of two long-short vowel pairs /i:-I/ and /ɐ:-ɐ/ in /pVp/ and /pVb/ syllables, produced at normal and fast speech rates by nine speakers of Australian English (AusE) using electromagnetic articulography. Our results suggest that in short vowels, the movement towards vowel target (formation interval) is shorter, has a smaller displacement, but is not stiffer than formation intervals of long vowels. Instead, in short vowels, shorter duration and smaller displacement primarily arise through a truncation of the latter portion of the formation interval. Syllables containing short vowels are also organised differently from those containing long vowels. Short vowels are more overlapped with following coda consonants than long vowels. Consonant transitions into and out of short vowel gestures are also shorter in short vowel syllables. Overall the differences between long and short vowel gestures were maintained in different contexts. The kinematics of vowel gestures generally did not di_er preceding voiced and voiceless codas; instead, changes to coda voicing primarily conditioned the duration of coda consonants and their relationship with the vowel. Normal-rate and fast-rate vowels differed in stiffness, with fast-rate vowels exhibiting shorter time to peak velocity and higher velocities than normal-rate vowels. However, speech rate did not condition the displacement of vowels such that displacement differences between long and short vowels were maintained across speech rates. Finally, I observed that, short vowel gestures' articulatory properties were not generally incompressible compared to their long equivalents. Instead, the timing relationships between short vowels and surrounding consonants tended to remain stable across changes to coda voicing and speech rate. These results suggest that vowel length in AusE is a syllable-level phenomenon. Instead of differences in intrinsic stiffness or target, the difference between long and short vowels appears to primarily be a difference in vowel-coda organisation, with short vowel gestures truncated by following coda consonants. Our results also suggest different articulatory mechanisms and scopes for the implementation of vowel length, coda voicing and speech rate such that some degree of articulatory contrast remains between long and short vowels across different linguistic contexts. Our results are also consistent with split-gestural models in which there is independent control of movements towards and away from gestural targets.