An Investigation of Combinational Productivity for Abstract and Concrete Nouns
Proceedings of the 28th Annual Conference of the Cognitive Science Society, (2006)Abstract
Although various theories of conceptual combination have been proposed in the past, these models have addressed interpretation issues rather than addressing the circumstances in which combinations are used. As a result, existing theories make no explicit predictions about what kind of nouns will be used more often in combination and why this might be the case. In this study we address the issue of combination use and investigate whether differences in productivity exist for two very different types of noun, namely abstract and concrete nouns. The slotfilling view of conceptual combination (e.g. Wisniewski, 1997) motivated our hypothesis that differences in combinational preference for both the modifier and head role would be observed between these two types of noun. We extracted combinations involving a sample of abstract and concrete nouns from the British National Corpus and obtained type and token frequencies for how often each noun occurred in both the modifier and head positions. Results revealed that abstract nouns were more likely to occur in the head position while differences for concrete nouns were only partially supported. We also discovered that abstract nouns occurring in the BNC were far less likely to occur in combination than concrete nouns. We interpret these results in light of what they reveal about the use of compound phrases and discuss the implications for existing theories of conceptual combination.