Word frequency of the CHILDES corpus: Another perspective of child language features

AbstractBased on the corpus of the Child Language Data Exchange System (CHILDES),the current research explores the grammatical composition of child language interms of word classes. Unlike past studies, the research reported in this articleexamines not only nouns and verbs but also adjectives, adverbs, prepositions,pronouns, determiners, and interjections. We investigated the word frequencypatterns of these word classes in both child and maternal language in order toexplore the correlation between input word frequency on the mother’s side andthe output word frequency on the child’s side. Our results show that the fre-quency patterns for word classes differ between child language and maternallanguage. Due to children’s mental development, young children use many morewords with concrete and imageable referents such as nouns and pronouns thanless concrete or imageable words such as adjectives, prepositions, interjectionsand conjunctions. Data prove that nouns are the most frequently used word classin child language. In addition, children acquire monosyllabic words more easilyand use them much more frequently than multi-syllabic words except for thosewith extremely high frequency. Moreover, our study reveals a positive correla-tion between the token volume of maternal language output and that of childlanguage output, suggesting the important role of maternal language in chil-dren’s language acquisition. A principle of comprehensible input should behighlighted in adults’ speech to children in order to make them achieve largervocabulary. This discovery in first language acquisition will hopefully help fur-ther studies in second or foreign language acquisition, learning and teaching.