2018 Sat Session A 1130

Saturday, November 3, 2018 | Session A, East Balcony | 11:30am

The acquisition of Mandarin tonal processes by children with cochlear implants
P. Tang, I. Yuen, N. Xu Rattanasone, L. Gao, K. Demuth

Cochlear implants (CIs) have made oral communication an obtainable goal for children with hearing loss. However, acquiring tones is difficult, since CIs cannot transmit pitch effectively[1]. Mandarin has four lexical tones (T1-4), acquired by children with normal hearing (NH) before age 3[2]. In addition, Mandarin has contextual tones: neutral tone (T0) and tone sandhi, involving tonal processes in different tonal contexts. T0 is a short toneless category, occurring after lexical tones, with contextually conditioned pitch (e.g., “ma1 ma0” mother), falling after T1/2/4 and rising after T3. Tone sandhi is a phonological process whereby a sequence T3T3 changes to T2T3 (full sandhi), and T3 changes to a low-falling tone before T1/2/4 (half sandhi). It has been suggested that NH children acquire these tonal processes by 3 years[2].

Children with CIs reportedly face challenges in building the tonal system[3-4]. Although studies using perceptual coding suggest that early implantation benefits lexical tone acquisition[2], there has been no acoustic evidence demonstrating this. Moreover, no study has investigated their acquisition of contextual tones. These issues were addressed here: Experiment 1 conducted acoustic analysis of children’s lexical tone productions with different implantation ages. Experiments 2 and 3 explored their contextual tone productions. We predicted that children with CIs would face challenges in producing both lexical and contextual tones, but that earlier implantation would result in better production.

Thirty-eight 3-6-year-old children with CIs (12 girls, 25 boys) were tested in Beijing. Seven were implanted between 1-2 years (CI 1-2), and other 31 were implanted between 2-3 years (CI 2-3). Forty-two NH children (3-year-olds, 31 girls, 13 boys) were included as controls. Tonal productions were elicited using a picture-naming task.

Experiment 1 elicited children’s lexical tone productions using four monosyllabic TX words. The results showed that only the NH and CI 1-2 groups produced correct lexical tones, while the CI 2- 3 group confused T2 and T3 pitch contours (Figure 1). This raised the possibility that the CI 1-2 group might also show knowledge of contextual tone.

Experiment 2 elicited children’s T0 productions using four disyllabic TX+T0 words. The results showed that both NH and CI 1-2 groups produced contextually conditioned T0 pitch, while the CI 2-3 group produced identical (falling) pitch contours across contexts. Moreover, both CI groups produced longer T0 duration than the NH children (Figure 2).

Experiment 3 elicited children’s tone sandhi productions using 16 disyllabic T3+TX compounds. Results showed that the NH and CI 1-2 groups produced T2 and low-falling tone for full and half sandhi contexts, respectively, while the CI 2-3 again produced identical (falling) pitch for both contexts (Figure 3).

Our results therefore provide acoustic evidence that children with CIs face challenges in acquiring lexical and contextual tones, confusing T2 and T3 and using a fixed pitch to produce contextual tones irrespective of context. They also had problems reducing T0 duration. However, those implanted between 1-2 years demonstrated better production of both lexical and contextual tones. The implications for word learning by children with CIs are discussed.