The operation of rendaku in the Japanese specifically language- impaired: A preliminary investigation.

Rendaku is a well-documented phenomenon in Japanese phonology in which a word-initial voiceless obstruent becomes voiced when it is the second member of a compound (e.g., ori + kami --> origami 'paper folding'). It was hypothesized that Japanese specifically language-impaired (SLI) children who appear to rely on explicit declarative memory as opposed to implicit procedural memory to learn language would have difficulty forming such compounds: word-initial voiceless obstruents would remain unvoiced in the second members of non-frequent and novel compounds. Six Japanese SLI children, ranging in age from 8;9 to 12;1, 6 age-matched non-SLI children and 4 younger non-SLI children were given a word formation task involving three different categories of compounds. A significant difference in performance between the groups was found. The data indicate that the SLI children did not in fact voice most of the initial obstruents of the second member in non-frequent and novel compounds, whereas the age-matched non-SLI children did voice the appropriate obstruents of all the compounds and the younger non-SLI children voiced some initial obstruents of all the compounds. Qualitative differences in the responses provide evidence that the SLI children did not have or were unable to construct a productive procedural rule of voicing.