Differences in the Semantic Content of Abstract and Concrete Words: Evidence from a Linguistic Feature Analysis.

Although much is known about the differences in representation and processing between abstract and concrete words, few studies have investigated this issue from the perspective of linguistic feature analysis. In this article, we first analyze the differences in semantic content between abstract and concrete words according to four linguistic properties: Activity (Q), linguistic category model (LCM), moving average type-token ratio (MATTR), and second thematic concentration (STC), when performing a lexical definition task in Experiment 1. We then employed a lexical decision task in Experiment 2 to further investigate whether these four linguistic features could predict the differences in behavioral performance (i.e., response times and accuracy) between abstract and concrete words. The results of Experiment 1 show that, compared to concrete words, the semantic content of abstract words involves fewer action-related expressions (lower Q), greater psychological distance (higher LCM), and less lexical diversity (lower MATTR), but no significant difference in STC between the two. The results of Experiment 2 indicate that concrete words are responded faster than abstract words, and that the Q, LCM, and MATTR can significantly predict their recognition performance. These results suggest that linguistic features can effectively quantify and distinguish the semantic content of abstract and concrete words. This finding increases the possibility of using quantitative linguistic methods to study the semantic content of words at a text level.