Tips for English learners: /r/ their shapes accent‑specific? - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Tips for English learners: /r/ their shapes accent‑specific?

Résumé

The articulatory complexity hidden behind the lips for English /r/ is well-established. Native speakers use an array of tongue shapes, ranging from bunched to retroflex, resulting in equivalent acoustic outputs, at least up to the third formant. Typically, bunched configurations are tip-down, with the dorsum retracted, while retroflex ones are tip-up, and raised towards the hard palate. The use of one or another partly depends on vowel context, as retroflexion is more compatible with low back vowels than with high front ones. However, it is unclear why some speakers develop multiple lingual gestures, and others only one, or why more bunching is reported in rhotic Englishes than in non-rhotic Englishes. Heyne et al. (2020) suggest that rhotic and non-rhotic varieties may encourage the acquisition of different strategies. The pronunciation of postvocalic /r/ in rhotic varieties (e.g., American English (AE)) may favour the development of bunching, as this configuration prevails in coda for speakers who use both tongue shapes. While most research on /r/ has focused on native speakers' productions of /r/, this paper addresses Heyne et al.’s hypothesis by looking at the understudied production of learners of English. In most manuals for learners, the bunched pronunciation of /r/ is completely absent as only tip-up shapes are described. Although Standard Southern British English (SSBE) is traditionally the variant taught in French schools, the lack of pronunciation classes makes it unlikely that the dialect used in teaching predicts the configuration developed by French learners. We examine whether the variety acquired by French learners, rhotic AE or non-rhotic SSBE, correlates with their tongue shape strategy. 10 native speakers of French, all university students of English, participated in the study. Midsagittal tongue images were recorded using an EchoBlaster 128 ultrasound system with a 5-8MHz transducer with acoustic data in Articulate Assistant Advanced. Lip videos were also collected with a front-facing NTSC micro camera. The ultrasound and camera were stabilised relative to the head using the UltraFit headset. To improve tongue shape interpretation and comparison, we used a bite plate to adjust the probe-to-chin angle to the occlusal plane [10]. Prior to the experiment, we categorised each student’s pronunciation as American (A) or British (B), based on a series of 12 sentences designed to classify their accent. Results from 4A and 6B are presented here, but data collection is still in progress. Participants read a randomised list of words in isolation including 20 different monosyllabic r-words and 10 distractors. Prevocalic /r/ occurred before /i:, ɪ, e, æ, ɑː, ɒ, ʌ, ɜː, ɔː, u:/ and postvocalic /r/ after /ɪ, e, ɑː, ɜː, ɔː/. All tokens were repeated 3 times. We manually extracted the ultrasound frame at the point of maximum constriction for each /r/ production and classified tongue shapes as tip-up (TU) or tip-down (TD). 3 tokens had no consistent strategy across the 3 repetitions and were labelled with the majority pattern found. The results indicate a great variety of tongue shape patterns similar to that observed in native speakers, but which could not be predicted by SSBE or AE accents. Across both groups, TU was largely preferred, and 4 participants used it exclusively (1A, 3B). 5 learners showed both strategies (2A, 3B), but as in many native speakers with allophonic variation, TD only occurred with front vowels or in postvocalic /r/. The only speaker with exclusively TD tongue shapes was in group A. It should be noted that 3 learners out of 6 in group B still produced /r/ in postvocalic position, 2 with a TD and 1 with a TU configuration. The observed patterns are therefore highly heterogeneous and reflect those found in native speakers across vowel contexts and in postvocalic /r/. However, these preliminary results did not reveal robust differences in lingual gesture acquisition between rhotic and non-rhotic varieties of English. Except for the TD-only participant, similar distributions of tongue shapes were found across groups. While a larger dataset will provide more robust insights into tongue shapes for /r/, it will also allow us to study the potential role of various degrees of accentedness.

Domaines

Linguistique
Fichier non déposé

Dates et versions

hal-04139003 , version 1 (23-06-2023)

Identifiants

  • HAL Id : hal-04139003 , version 1

Citer

Alice Léger, Hannah King, Emmanuel Ferragne. Tips for English learners: /r/ their shapes accent‑specific?. Ultrafest X, Nov 2022, Manchester (UK), United Kingdom. ⟨hal-04139003⟩
21 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More