The Limits of Observation: Can the Situations in Which Words Occur Account for Cross-linguistic Variation in Vocabulary Composition?

Crosslinguistic variation in vocabulary composition is usually attributed to differences in the structural properties of languages and correlated differences in the frequency and salience of different word classes in the input. Here we explore whether another property of the input, the extralinguistic contexts of word use, can account for the similarities and differences in the early vocabularies of English and Mandarin learners. Early word learning is limited by the child’s initial representation of the input. Since novice language learners know few words and little syntax, they must initially learn words from the extralinguistic contexts in which they appear. By simulating this learning procedure with adults we can determine whether the information in word-to-world pairs is sufficient to account for the composition of early vocabularies in both linguistic environments. We assessed the potency of word-to-world mapping in each language by asking adult native-speakers to identify nouns and verbs from their extralinguistic contexts in maternal speech. We found that verbs were identified more often in Mandarin (p < .001) while nouns were identified more often in English (p < .001). Even after the data were corrected to compensate for potential differences in response bias, there was still a large and persistent difference in verb identification. Subjects in both groups performed well on action verbs and poorly on verbs referring to unobservable relations or mental states. Since Mandarin mothers used more action verbs than the Americans, the Mandarin subjects were better able to identify their target verbs. These results indicate that the information available in extralinguistic contexts can account both for the early acquisition of verbs in Mandarin and for the predominantly nominal initial vocabulary of English learners. Cognitively mature adults, forced to rely on the evidence of their senses, were able to identify only those words that can be learned by children who are similarly impaired.