Cross-situational observation and the semantic bootstrapping hypothesis

How do children learn their first words? Everyone who has thought about the problem long enough to consider one has come up with essentially the same solution. Initially learners must: pair each word with the extralinguistic contexts in which it occurs, collect several such pairs, and then identify the common element in the scenes. We may disagree on the parents’ role in structuring the input or the range of hypotheses that the child entertains but we all agree on the data source. After all, what other information does the novice language learner have? But reasonable people do disagree about how much children can or must learn in this manner. Both empiricist philosophers (Locke, 1690; Hume, 1758) and nativist linguists (Wexler & Cullicover, 1980; Pinker, 1984) have suggested that children use extralinguistic contexts to learn the meanings of a large and diverse set of words. In contrast, the proponents of syntactic bootstrapping claim that this word-to-world mapping procedure can only provide the child with a small starter lexicon, consisting mostly of concrete nouns. This initial vocabulary is a wedge that the child uses to begin constructing representations of the sentences in which novel words are used. These representations constrain her interpretation of extralinguistic context, allowing her to acquire verbs, adjectives and abstract nouns