Chinese middle cosntructions [i.e. constructions] : lexical middle formation

Abstract

This dissertation is an exploration of Chinese middle constructions, which starts with distinguishing the middle construction from the middle voice, considering that these two terms have long been used interchangeably with different connotations. Consequently, middle data presented in the literature vary tremendously, both intra-linguistically and inter-linguistically. In order to delineate a clear boundary for the middle construction, this study defines it as a generic semantic category with the obligatory non-realization of a verb’s highest argument. By contrast, the middle voice is a morphological category with various syntactic and semantic features.

Starting from the working definition arrived at, the study identifies several middle constructions, viz., the qilai middle, the hao middle, the rongyi/nan middle, the de middle (and the te middles in Chengdu Chinese), the bu middle and transitive middles. Most of these middle constructions come with middle markers, though these are both morphologically and syntactically diversified. In the case of transitive middles, middle formation depends on an idiosyncratic argument realization of a small group of verbs and does not include markers. In this sense, the presence of middle markers is not a design feature of Chinese middles. Moreover, the occurrence of the identified middle markers does not necessarily lead to a middle analysis, because they can also mark other constructions, e.g., the qilai unaccusative construction, the rongyi/nan tough construction, and the de-resultative unaccusative construction.

Syntax-wise, Chinese middles can have both complex predicates (e.g., [V-qilai AP] and [rongyi/nan V]) and simple predicates (e.g., hao-V, V-de-A, V-bu-A, V-te-A, V-te), both of which are proved to be unergative in nature. Consequently, these middle constructions are collectively termed “unergative middles”, which stand in contrast with “transitive middles” in terms of transitivity. However, “unergative middles” and “transitive middles” share one property: they do not involve any movement mechanisms. Therefore, both of them are subsumed under the rubric of “lexical middles”, which in addition exhibit the cross-linguistic lexical middle properties proposed by both Marelj (2004) and Lekakou (2005).

This study is the first to adopt a parametric approach to Chinese middle constructions. It adds Chinese to languages like English, Dutch and German on the list of lexical middle languages, to be distinguished from syntactic middle languages like French, Italian, Portuguese, and Serbian/Croatian.