Mandarin Chinese speech recognition by pediatric cochlear implant users
Introduction
Cochlear implants (CIs) can restore hearing sensation to patients with profound sensorineural hearing loss. Many post-lingually deafened adult CI users are capable of high levels of speech understanding. CIs allow deaf children to acquire similarly high levels of speech understanding [1]. Similar to adult CI populations, there is a wide variability in pediatric patient outcomes [2]. While many studies have explored pediatric CI patient performance [2], [3], [4], [5], most of these studies have focused on English speech perception by English-speaking pediatric CI users. Previous studies with French [6], [7], Belgian [8], and Dutch [9], [10], [11] pediatric patients showed that cochlear implantation greatly benefited the speech development of these profoundly deaf children.
The majority of native Mandarin-speaking CI users were implanted as children. Most Chinese pediatric CI users are pre-lingually deafened, i.e., they did not acquire speech before deafness, typically because of congenital hearing loss. There is a sensitive period for development of the human central auditory pathways, beyond which there is limited plasticity [12]. Therefore, implantation at an early age is crucial for pre-lingually deafened CI users’ speech development. Other pediatric CI users may have acquired speech through acoustic hearing (aided or unaided) before deafness. For these (typically) late-implanted pediatric patients, CI outcomes may depend on the duration and quality of acoustic hearing during auditory development. Extended auditory deprivation due to hearing loss may limit the benefit of cochlear implantation. Alternatively, post-lingually deafened pediatric CI users may greatly benefit from implantation if there was sufficient exposure to acoustic hearing. In general, little is known about differences in speech performance between pre- and post-lingually deafened Mandarin-speaking pediatric CI users.
English, French, German and Dutch are non-tonal, Indo-European languages. In contrast, Chinese (which is spoken by the greatest number of people in the world) is a tonal, Sino-Tibetan language. Mandarin Chinese syllables are produced with one of the four lexical tones: Tone 1 (high-flat), Tone 2 (rising), Tone 3 (falling–rising), and Tone 4 (falling). The same syllable produced with different tones can have vastly different meanings [13], [14]. Fundamental frequency (F0) cues contribute strongly to Chinese tone recognition, which in turn contributes strongly to Chinese sentence recognition. Because CI devices typically provide only 12–22 spectral channels (too few to support good frequency resolution), Chinese tone recognition is difficult and CI users must rely more strongly on amplitude and duration cues [15].
Most CI speech perception studies (with adults or children) research have been conducted with English-speaking subjects. There are comparatively few CI studies with Chinese CI users. Several Taiwanese studies have investigated Mandarin Chinese-speaking CI users’ speech perception. Huang et al. [13] evaluated speech performance and auditory function in four adult, Mandarin-speaking, post-lingually deafened users of Cochlear's Nucleus-22 device; they found significant and continuous improvement in all subjects’ speech perception and auditory capabilities after implantation. Wu and Yang [14] reported a negative correlation between Chinese vowel, word, sentence recognition and age at implantation for Chinese pediatric CI users. Peng et al. [16] found that pre-lingually deafened pediatric CI users exhibited poor Mandarin Chinese tone recognition and tone production. Wang et al. [17] found long-term benefit for cochlear implantation in a longitudinal study with Taiwanese pediatric CI users. Recent studies have focused on the contribution of Chinese tone recognition to CI users’ understanding of Mandarin Chinese [18], [19], [20].
To date, most Chinese pediatric CI studies have evaluated speech understanding in terms of language awareness [21], lexical tone production and perception [22], [23], Cantonese word recognition [24], as well as closed-set Mandarin early speech perception [25]. Because of general difficulties associated with testing children (e.g., limited language development, subject attention, etc.), very few studies have evaluated Chinese CI users’ open-set word or sentence recognition with multiple talkers. Understanding words and sentences is essential to daily life, and may better reflect the ultimate benefit of implantation for pediatric patients. Some Mandarin-speaking pediatric CI users may be old enough to understand more difficult or complex speech materials (e.g., words and sentences produced by multiple talkers). Older pediatric CI users may also have different etiologies of hearing loss. Some may be congenitally deafened and implanted at an early age, while others may have experienced significant amounts of acoustic hearing (aided or unaided) before implantation. As such, speech pattern development may be different across pediatric CI users. Pre-lingually deafened CI users develop speech patterns exclusively via electric hearing. Other CI users may adapt novel electric stimulation patterns to previous acoustic patterns; depending on age at onset of hearing loss and/or the severity of hearing loss, these acoustic patterns may (or may not) be sufficient for adaptation. Different types of speech tests may elicit differences between pediatric CI users. Word recognition may be more sensitive to the peripheral representation, while sentence recognition may depend greatly on contextual cues.
In this study, speech recognition was measured in Mandarin-speaking pediatric CI users to see how patient demographics influence performance with relatively difficult test materials and methods. Subjects were grouped according to etiology of deafness and previous acoustic hearing experience. Group 1 subjects were all congenitally deafened with little-to-no acoustic hearing experience. Group 2 subjects were not congenitally deafened and had substantial acoustic hearing experience prior to implantation. Open-set multi-talker (2 male and 2 female talkers) disyllabic word and sentence recognition was measured in each subject. Multiple linear regression analyses were performed to see whether patient demographics (e.g., age at implantation, age at testing) predicted speech performance.
Section snippets
Subjects
Thirty-seven pediatric CI users participated in this study. All subjects were implanted at the Shanghai Eye, Ear, Nose & Throat (EENT) hospital; this study was conducted in the Vision and Audition Center of the Shanghai EENT hospital. Subject inclusion criteria consisted of bilateral profound sensorineural hearing loss, no evidence of mental retardation, at least 6 years old and Mandarin Chinese as the native language. The profound hearing loss was confirmed using auditory brainstem response
Results
Fig. 1 shows mean disyllable and sentence recognition performance with different talkers for Group 1 (left), Group 2 (middle) and all subjects (right). For Group 1, mean performance was 82.3% correct for disyllables and 82.8% correct for sentences. A two-way repeated measures analyses of variance (RM ANOVA), with talker (Female 1, Female 2, Male 1, and Male 2) and test (disyllable or sentence recognition) as factors, showed a significant effect for talker [F(3,78) = 11.94, p < 0.001], but not for
Discussion
In general, the present Mandarin-speaking, pediatric CI subjects scored well on the relatively difficult open-set, multi-talker disyllable and sentence recognition tasks. Age at implantation best predicted performance, especially for Group 1, while both age at implantation and age at testing predicted Group 1 sentence recognition performance. Performance was generally better with the female rather than the male talkers. Below we discuss the results in greater detail.
All of the present subjects
Conclusions
The present study revealed several significant findings regarding Mandarin-speaking pediatric CI users’ speech perception:
- (1)
In general, the present pediatric CI subjects (aged 6 and older) were capable of good open-set, multi-talker word and sentence recognition, suggesting that more difficult speech materials and tests may be appropriate for older pediatric CI users.
- (2)
Speech recognition was significantly better with the female talkers than with the male talkers.
- (3)
Age at implantation strongly
Acknowledgements
We are grateful to all the children who participated in this study. We are also grateful to Justin Aronoff for assistance with the statistical analyses. The work was supported by National Natural Science Foundation of China (Grant No. 30872867) and NIH grant R01-DC004993.
References (47)
- et al.
Determinants of speech perception in children after cochlear implantation
Lancet
(2000) - et al.
Speech perception of Mandarin Chinese speaking young children after cochlear implant use: effect of age at implantation
Int. J. Pediatr. Otorhinolaryngol.
(2003) - et al.
Pediatric cochlear implantation in Taiwan: long-term communication outcomes
Int. J. Pediatr. Otorhinolaryngol.
(2007) - et al.
Mandarin Chinese tone identification in cochlear implants: predictions from acoustic models
Hear. Res.
(2008) - et al.
Tone production of Mandarin Chinese speaking children with cochlear implants
Int. J. Pediatr. Otorhinolaryngol.
(2007) - et al.
Assessment of Mandarin-speaking pediatric cochlear implant recipients with the Mandarin Early Speech Perception (MESP) test
Int. J. Pediatr. Otorhinolaryngol.
(2010) - et al.
Concurrent-vowel and tone recognition by Mandarin-speaking cochlear implant users
Hear. Res.
(2009) - et al.
Effect of age at cochlear implantation on open-set word recognition in Mandarin speaking deaf children
Int. J. Pediatr. Otorhinolaryngol.
(2006) - et al.
Relationships among types of speech intelligibility in pediatric users of cochlear implants
J. Commun. Disord.
(2001) - et al.
Modeling spoken word recognition performance by pediatric cochlear implant users using feature identification
Ear Hear.
(2000)