ABSTRACT
Concept compounds play an important role in expanding language through the coining of new terms. The understanding of novel compounds also provides a key perspective on the inherent generativity of natural languages. In this paper, we report three experiments examining the effects of familiarity on the understanding of noun-noun compounds. We show that familiar compounds are judged to be sensible/comprehensible more quickly than unfamiliar ones. We also show, contrary to some previous proposals, that the production of relational/property interpretations to these compounds is equally “natural”. These findings have important implications for computational models of concept combination, as they support models that can produce both interpretation types as a matter of course.
