The link between anterior lingual gesture delay and loss of coda /r/: an ultrasound study

It has long been recognised that consonants sound different and have different diachronic sound-change trajectories depending on their location within the syllable (Trager 1959). Articulatory phonetics has sought to discover why liquid consonants in syllable-coda position have a tendency to lenite or be lost over time. The possibility that a delayed anterior lingual gesture in coda liquid consonants might result in auditory weakening has been suggested by (Recasens and Farnetani 1994), who noted that the alveolar gesture of phrase-final /l/ in Catalan and American English was found to occur on occasion partially or completely after the offset of voicing, leading to apparent loss at the acoustic level, if not at the articulatory level. A similar phenomenon has been observed for Dutch /r/ (Scobbie et al. 2009; Scobbie and Sebregts 2011). We investigated the contribution that anterior lingual gesture delay is making to the lenition of postvocalic /r/ in Scottish English. This study uses a socially-stratified, audio-ultrasound corpus of adolescent Glaswegian English (16 speakers), containing word-list recordings from two sociolects; one with postvocalic /r/ weakening (working-class speech) and the other with /r/ strengthening (middle-class speech). The word list used had 45 monosyllabic items containing postvocalic /r/, 25 of which were CVr words, 5 of which were CVrC words and 15 of which were CVrC nonsense words. There were also 98 distractors, some of which were real words and some nonsense words. The inclusion of nonsense words does not directly relate to the design of the current study; another aim of the data collection was to obtain a set of stimuli for a subsequent nonsense-word mimicry experiment. The /r/-ful nonsense words were included in the current study in order to increase the number of tokens of /r/. Ultrasound was recorded at a frame rate of circa 30 fps, increased to 60 fps after deinterlacing. We quantified auditory strength of rhoticity for all tokens of /r/ using a Praat-based rating experiment and three phonetically-trained raters, familiar with the variety being studied. Articulatory measures involved quantification of the temporal difference between the maximum of the anterior lingual /r/ gesture and either the offset of voicing in CVr words: bar, bore, fur, or the onset of a following labial consonant in CVrC words: farm, herb, burp. This articulatory measure we refer to as 'lag'. We considered both voicing offset and final consonant onset to be events that could auditorily mask the /r/ articulation to some …