Phonological Variation in Child-Directed Speech is Modulated by Lexical Frequency

We investigate the hypothesis that child-directed speech (CDS) contains a higher proportion of canonical pronunciation than adult-directed speech (ADS) by analyzing Korean noun stem-final obstruent variation. In a word-teaching task, mothers used a higher ratio of canonical pronunciation when addressing infants than an adult. In a follow-up experiment, adults used a greater ratio of canonical pronunciation when pronouncing high- than low-frequency words. Additional analyses conducted with only the high-frequency monosyllabic words from the two experiments found no evidence for simplified phonology in CDS when lexical frequency was controlled for. Our results suggest that CDS contains a higher ratio of canonical forms than ADS in the context of morphophonological rules, which is mediated by the frequency of word-usage. Thus, the didactic function of CDS seems a serendipitous result of mothers using easy words to children. These results highlight the importance of considering word-usage in investigating the nature of CDS.