A multi-dimensional corpus study of mixed compounds in Chinese

Certain mixed compounds, consisting of two synonymous components (one classical and one non-classical), are ideal for investigating lexical stylistics, as they are semantically synonymous but stylistically contrastive with both of their components. They are investigated using Biber’s (1988) multi-dimensional framework for register variation and the statistical method of Correspondence Analysis, which has been used by Zhang (2012, 2013) to show that there are two dimensions in written Chinese, namely the Classical and the Literate. The results, consistently replicated with different corpora, are rather surprising, being partially at odds with the intuitive prediction that mixed compounds should be in-between in stylistic value compared with their components on both dimensions. On one hand, mixed compounds are indeed less classical than their classical components but more classical than the non-classical components; on the other hand, they are interestingly more literate than either of their components. The unexpected results are tentatively attributed to the association of di-syllabicity with the literate style and possibly the propensity for parallelism in Chinese. Pedagogical benefits may be reaped from the present research, as mixed compounds may be ideal for alerting students to stylistic variation in Chinese. They exemplify the two stylistic dimensions most clearly in their embodiment of the tripartite distinctions of non-literate/non-classical vs. strongly classical vs. strongly literate.