Building Meaning: Constructing New Word Knowledge from Simple Statistics
We can understand and express an unlimited variety of meaningful ideas using language. This remarkable ability depends on the fact that as we learn words, they become organized according to meaningful, semantic links, such as those connecting apple, juicy, eat, and pear. Extensive computational evidence attests that everyday language is rich in statistical regularities that could, in principle, drive the formation of these links: (1) Direct co-occurrence (e.g., eat - apple) may foster links between words that can be combined to express meaningful ideas, and (2) Shared patterns of co-occurrence (e.g., apple and pear both co-occur with eat) may foster links between words similar in meaning. Here, we investigated whether humans can harness these simple but powerful statistics to integrate new words into their existing networks of organized word knowledge. In three reported experiments (N=128), participants came to link novel with familiar words based on both direct and shared patterns of co-occurrence following mere exposure to sentences containing these statistics. This novel finding highlights a potentially key role for co-occurrence regularities in building the organized word knowledge that is fundamental to our unique fluency with language.