Learning from the input: a corpus-based investigation of Chinese classifiers in children’s books and child-directed speech

In Mandarin Chinese, numeral classifiers form a grammatical category that is syntactically obligatory when a noun is modified by a numeral or a demonstrative. The appropriate choice of a classifier is associated with the semantic properties of its corresponding noun and is context dependent. Experience with language is needed to learn these patterns, but little is known about how classifiers are structured in children’s language environments. We compared the frequency and distribution of classifier phrases in four corpora: child-directed speech, children’s television shows, children’s books, and adult-directed speech. Classifier usage in children’s books was more diverse than in both child-directed and adult speech. Books contained more specific classifiers that co-occurred with a higher proportion of unique nouns whereas everyday speech relied on more generic classifiers. Books therefore provide access to classifier-noun collocations that are rare in speech. Implications for language development and language processing are discussed.