TY - CONF

T1 - Acquisition of allophonic variation in second language speech

T2 - Interspeech 2022

AU - Nagamine, Takayuki

PY - 2022/9/18

Y1 - 2022/9/18

N2 - Acquisition of positional allophonic variation is seen as the foundation of a successful L2 speech learning. However, previous research has mostly focused on the phonemic contrast between English /l/ and /r/, providing little evidence in the acquisition of positional allophones, such as those in English /l/. The current study investigates the acoustics and articulation of allophonic variations in English laterals produced by Japanese speakers, focusing on the effects of syllabic positions and flanking vowels. Acoustic and articulatory data were obtained from five Japanese speakers in a simultaneous audio and high-speed ultrasound tongue imaging recording set-up while they read sentences containing syllable-initial and -final tokens of English /l/ in four different vowel contexts. Acoustic analysis was conducted on 500 tokens using linear-mixed effects modelling and the articulatory data were analysed using generalised additive mixed modelling. Syllable position and vowel context had significant effects on acoustics, while midsagittal tongue shape was more influenced by vowel context, with fewer positional effects. The results demonstrate that differences in acoustics not always be mirrored exactly by midsagittal tongue shape, suggesting multidimensionality of articulation in second language speech.

AB - Acquisition of positional allophonic variation is seen as the foundation of a successful L2 speech learning. However, previous research has mostly focused on the phonemic contrast between English /l/ and /r/, providing little evidence in the acquisition of positional allophones, such as those in English /l/. The current study investigates the acoustics and articulation of allophonic variations in English laterals produced by Japanese speakers, focusing on the effects of syllabic positions and flanking vowels. Acoustic and articulatory data were obtained from five Japanese speakers in a simultaneous audio and high-speed ultrasound tongue imaging recording set-up while they read sentences containing syllable-initial and -final tokens of English /l/ in four different vowel contexts. Acoustic analysis was conducted on 500 tokens using linear-mixed effects modelling and the articulatory data were analysed using generalised additive mixed modelling. Syllable position and vowel context had significant effects on acoustics, while midsagittal tongue shape was more influenced by vowel context, with fewer positional effects. The results demonstrate that differences in acoustics not always be mirrored exactly by midsagittal tongue shape, suggesting multidimensionality of articulation in second language speech.

U2 - 10.21437/Interspeech.2022-11020

DO - 10.21437/Interspeech.2022-11020

M3 - Conference paper

SP - 644

EP - 648

Y2 - 8 September 2022 through 22 September 2022

ER -