What roles do constituents play in the identification of Chinese compound words? A meta-analytic review

Abstract

Because compound words comprised of two (or more) constituents make up the majority of Chinese vocabulary, understanding how they are processed and identified is critical for understanding the mechanisms that support the reading of Chinese. This meta-analytic review thus investigated whether Chinese compound words are processed and identified in a compositional versus holistic manner. Our meta-analysis includes 268 constituent effect sizes derived from 81 studies involving 5,911 participants. Overall, we found a statistically significant, albeit small, constituent effect (Hedges’ g_rm = 0.22, 95% CI [0.18, 0.25]). The magnitude of this effect varied across study designs, being significantly larger when words were presented in isolation than within sentences, and in studies using preview methods rather than natural reading. Consistent with compositional processing, evidence from both behavioral and neurophysiological experiments also suggested that constituent effects occurred at the orthographic, phonological, and semantic processing levels. These findings highlight the significant role of constituents in Chinese compound-word identification, encompassing both form and semantics. Finally, although these discoveries have significant implications for existing and future models of Chinese reading, we should warn the readers that the average effect sizes should be interpreted with caution. The present study showed substantial heterogeneity, which calls for future studies to understand the underlying reasons for the variabilities across studies.

Public Significance Statement

This meta-analysis suggests that, in Chinese reading, individual constituents of compound words have a small but significant facilitative effect on their orthographic, phonological, and semantic processing across different word presentation methods and tasks. These findings are important for understanding how compound words are processed and identified in Chinese but also shed light on the universal mechanisms of compound-word processing across languages. These findings also suggest that the unique properties of a language’s script influence the cognitive mechanisms that support word identification. In the future, the findings from this study may help improve the efficiency of Chinese language instruction.