Age of Acquisition and the Structure of Dictionary Definition Space

In a dictionary, you can learn the meaning of any word through definition alone, as long as you already know the meaning of the words in the definition. How many words do you need to reach all the rest by definition alone? It turns out that you can reduce a dictionary to a unique "grounding kernel" (GK; about 10% of the dictionary) from which you can learn the meaning of all the rest of the words by definition alone. The words in the GK turn out to have been learned at a younger age (and probably not by definition, but from direct sensorimotor experience). GK words are also more concrete, imageable and more frequent, both in writing and speaking. The GK, however, is not the smallest number of words from which all the rest of the dictionary can be reached by definition alone. That would be the "minimum grounding set" (MGS). We have not yet found the MGS (because the general problem of finding an MGS is NP-complete, but we are coming closer for the special case of dictionaries), but we have already compressed the GK down to a "hard kernel" of strongly connected words. The relationship with the psycholinguistic variables (age of acquisition, concreteness, frequency) turns out to be more complex within the GK. We will describe this relationship for two special dictionaries -- Longman's and Cambridge -- as well as for WordNet, and close with some findings on sensorimotor category learning.