Count/mass syntax and superordinate categories: Evidence of a conceptual distinction and its cognitive implication

The semantic and conceptual status of the count/mass distinction has been a topic of extended debate among philosophers, linguists and psychologists. Some have seen the syntactic distinction as grounded in an ontological distinction between objects and substances. Others have argued that form class assignment is often arbitrary, pointing to counterexamples to the typical correspondence between count nouns and discrete objects on one hand, and mass nouns and substances on the other. Mass superordinates (e.g., furniture, wildlife) have often been cited among such counterexamples.;Recent evidence from Wisniewski, Imai and Casey (1996) suggests that the superordinate count/mass distinction maps on to a cognitive distinction between individuals united by taxonomic relations and unindividuated groups united by thematic or functional contiguity. The present thesis explored this hypothesis in greater detail, and investigated its potential implications for cognitive processes and for everyday communication.;In Experiment 1, it was found that people can instantiate count superordinates more easily than corresponding mass superordinates under time pressure. In Experiment 2a, novel group-appropriate properties were considered more plausibly true for domains which typically take mass syntax, whereas novel individual-appropriate properties were considered more plausibly true for domains usually labeled with count nouns. In Experiment 2b, subjects were more likely to assign novel group-than individual-appropriate properties to mass superordinates. In Experiment 3, inferences concerning the relative numerosity and heterogeneity of overlapping count/mass superordinates varied depending upon the typical syntax used in a given domain. In Experiment 4, analysis of spontaneous written language found that people are more likely to invoke count superordinates in discussing interactions with individuals, but mass superordinates when relating interactions with groups. In Experiment 5, a second text analysis found that mentions of parts of individual category members occur more frequently in the context of count than mass superordinates. These findings constitute converging evidence for the view of Wisniewski et al. (1996) concerning representational differences between count and mass superordinates. In addition, they suggest that these differences are consequential for both communication and cognitive processing.