Lexical effects on spoken word recognition performance among Mandarin-speaking children with normal hearing and...

This investigation aimed to examine the effects of word frequency and lexical neighborhood density on spoken word recognition of monosyllables and disyllables in Mandarin by normal hearing children and children with cochlear implants. The lexical characteristics were incorporated from the Neighborhood Activation Model (NAM), which suggests that words in the mental lexicon are organized into similarity neighborhoods. The difficulty of a listener's task is affected by the frequency of the target word and the density of the lexical neighbors from which that word must be identified. The Monosyllabic Lexical Neighborhood Test and the Disyllabic Lexical Neighborhood Test in Mandarin Chinese (Mandarin LNT and MLNT) were developed to take into account the effects of these linguistic and cognitive demands on speech perception performance.

Three stages were conducted in this investigation. In the first stage, Mandarin words of monosyllables and disyllables were selected and their lexical properties were calculated from the CHILDES database. Four lexically "easy" and four lexically "hard" word lists in Mandarin LNT as well as two word lists across lexical properties among disyllables were determined based on their relative word frequencies and neighborhood densities. In the second stage, word stimuli were verified by 30 children of the NH group and 36 children from the CI group. In the third stage, the inter-list equivalency and test-retest reliability of word lists across lexical properties were determined, and the correlations of Mandarin LNT and MLNT with other measures and inter-rater reliability were also investigated.

Word recognition scores were higher among disyllables than among monosyllables. Lexically "easy" disyllabic words were better recognized than their "hard" counterparts and the monosyllables among two groups of children. However, no lexical effects on word recognition of Mandarin monosyllables were observed for either group. No significant differences were found among word lists in each combination of syllable structure and lexical property. Inter-rater reliability, inter-list equivalency, and test-retest reliability were revealed. The Mandarin LNT and MLNT were found to be highly reliable measures of spoken word recognition (r = 0.84; p < 0.01) with acceptable equivalency between lists (r = 0.638-0.876).

Lexical effects on Mandarin word recognition were only demonstrated among disyllabic words by NH and the CI children, while Mandarin homophones appearing in monosyllabic words were suggested. Lexical effects on spoken word recognition in Mandarin are not substantially demonstrated as in English, but the Mandarin LNT and MLNT provided reliable information on the spoken word recognition of pediatric CI users in the initial stage after implantation as well as in the rehabilitation progress.