Acoustic And Semantic Processing In Children With Cochlear Implants Revealed By Event-related Potentials

Objective Although cochlear implants(CIs)can help children with congenital sensorineural recover auditory and speech functions,these abilities vary greatly.In the daily communication,the cognitive processing of speech sounds includes low-level acoustic analysis and higher-level semantic comprehension.We should focus not only on whether CI children can hear sounds clearly,but also on whether they can understand.However,the dynamic processing of acoustic and semantic information by CI children and its related neural basis are still unclear.In this study,the dynamics of speech processing in CI children were explored by using event-related potentials(ERPs),a tool widely used in speech cognitive processing.Methods We recruited 38 Mandarin-speaking children,including 19 prelingually deafened children with cochlear implants(12 males;age range,8-13 years)and 19 children with normal hearing(NH)(10 males;age range,8-13 years).All CI children were diagnosed with profound sensorineural hearing loss before cochlear implantation.Of these CI children,15 were implanted unilaterally(10 on the right side and 5 on the left side)and 4 bilaterally(3 implanted on the right side first and 1 on the left side first).Auditory and speech abilities of CI children were assessed by Meaningful Auditory Integration Scale(MAIS),Speech Intelligence Rating(SIR)and Categories of auditory performance(CAP).All 38 children took part in the EEG experiment.Sound stimuli used in the EEG experiment were four-syllable Chinese idioms or monosyllables.Idioms used as standard stimuli were /ba2 miao2 zhu4 zhang3/ and /huo3 mao4 san1 zhang4/.By changing the lexical tone of the last syllable of each idiom,two novel pseudo-idioms(/ba2 miao2 zhu4 zhang4/ and /huo3 mao4 san1 zhang3/)were formed as deviant stimuli,respectively.Each idiom has a rich implicit meaning in addition to its literal meaning.Although each syllable of the pseudo-idioms has its literal meaning,two pseudo-idioms do not make sense in Chinese.Therefore,the tone contrast between standard and deviant stimuli includes both acoustic and semantic differences.The last syllables of the idioms(/zhang3/ and /zhang4/)were used as auditory stimuli in the monosyllable condition.The sounds were passively presented to the children in an oddball paradigm from loudspeakers,and the auditory event-related potentials were recorded simultaneously.Results 1.MMN responses could be evoked in both NH and CI children.In the monosyllable condition,both NH [F(1,18)= 35.932,p < 0.001] and CI children [F(1,18)= 12.195,p = 0.003] showed significantly more negative responses to deviants than to standards at 200-240 ms and 190-230 ms time windows respectively.In the idiom condition,NH [F(1,18)= 12.138,p = 0.003] and CI children [F(1,18)= 4.549,p = 0.046] also showed significantly more negative responses to deviants than to standards at 220-260 ms and 240-280 ms time windows respectively.The results indicated the presence of robust MMNs in both NH and CI children.2.Smaller MMN amplitude and longer MMN peak latency for idiom condition in CI children than those in NH children.In the monosyllable condition,there was no significant difference in MMN amplitudes between CI and NH children [F(1,36)= 0.379,p = 0.542].However,in the idiom condition,MMN amplitude was significantly smaller in CI children than in NH children [p = 0.008].In the monosyllable condition,no significant difference in MMN peak latencies between CI and NH children was found [F(1,36)= 3.702,p = 0.062].However,in the idiom condition,MMN peak latency was significantly longer in CI children than in NH children [F(1,36)= 11.166,p = 0.002].3.LDN responses could be evoked only in NH children,but not in CI children.NH children showed significant differences of ERP amplitudes between standards and deviants at 546-586 ms time window in the monosyllable condition [F(1,18)= 23.907,p < 0.001] and at 540-580 ms time window in the idiom condition [F(1,18)= 6.050,p = 0.024] respectively,indicating the existence of LDNs in NH children.However,we failed to find the LDN for CI children in the monosyllable [F(1,18)= 1.735,p = 0.204] or idiom [F(1,18)= 0.846,p = 0.370] condition.4.Speech performance was positively correlated with duration of CI use in CI children.There was no significant difference in accuracies for identifying phrases or sentences consisting of one conditional word [p = 0.574] or two conditional words [p = 0.080] between CI and NH children.However,the accuracy identifying phrases or sentences consisting of three conditional words was significantly lower in CI children than in NH children [p = 0.032].Furthermore,a significantly positive correlation between the accuracy for three conditional words and duration of CI use in CI children [r = 0.64,p = 0.003].Conclusion Our results indicated two stages of speech processing,namely early acoustic analysis and late semantic processing.CI children's ability acoustic analysis is related to context: CI and NH children have no significant difference in MMN when processing monosyllables.However,in the complex context(such as idiom),CI children show deficit in this ability.Furthermore,the LDN was only elicited in NH children,indicating that CI children have deficit in high-level semantic analysis.The level of speech comprehension increases with the use of CI device,suggesting that long-term auditory and speech training is very important to speech rehabilitation.