Holistic vs. decompositional storage of Chinese words: An electrophysiological study

In Chinese, character and syllable boundary is much clearer than word boundary. This leads to a debate on whether Chinese words are stored holistically or they are decomposed into the constituent characters and syllables for storage. Previous behavioral experiments that aimed to investigate the issue have produced inconsistent results, some of which are also difficult to interpret because of methodological concerns and uncontrolled confounding factors. Recently, an electrophysiological technique called event-related potential (ERP) recording, which allows monitoring of brain processes with high temporal resolution, has been adopted to investigate the issue of word storage in Indo-European languages. Specifically, holistic and decomposed storage of words have been shown to affect an ERP index called mismatch negativity (MMN) in opposite directions. Therefore, by examining how Chinese words of different properties affect MMN in one or the other direction, it is possible to infer whether the Chinese words are stored holistically or in decomposed forms. Moreover, MMN can be elicited even when participants are not paying attention to the stimuli. MMN is thus typically considered to reflect relatively early language processes that are free from response-bias, which makes it an ideal measure to supplement previous behavioral experiments in Chinese.

This proposed study utilizes MMN as a tool to clarify the nature of Chinese word storage. Specifically, five experiments will be conducted to examine how different lexical variables may modulate the balance between holistic and decompositional storage. Affirmative results may help resolve previous discrepancies by identifying potentially uncontrolled confounding factors. The variables tested include 1.) word frequency (high vs. low), 2.) semantic transparency (i.e., whether constituents contribute to word meanings; transparent vs. opaque), 3.) word structure (monomorphemic vs. bimorphemic), 4.) constituent productivity (i.e., how many words can be formed with each constituent; high vs. low), and 5.) morphological boundedness (i.e., whether the constituents can stand alone as individual words; free vs. bound). Results of these experiments will contribute to a better understanding of Chinese word storage and the development of models of Chinese word processing. They also help establish the validity of using MMN to study high level language processing in Chinese, which can be extended to other aspects in future research.