Verb‐to‐preposition reanalysis in Chinese

Abstract

Since the introduction of the MP relegating variation to the lexicon, there has been a growing awareness that syntactic change should in fact be captured by changes in the discrete features of individual lexical items. This insight has given rise to independently developed proposals: Grammaticalization: shift “up the tree” to a functional category (Roberts & Roussou 1999); Reanalysis as relabelling: changes only in categorial or projection ([+max, +min] features and elimination (“pruning”) of unmotivated structure, under preservation of structural (ccommand) relations (Whitman 2000); Inertial Theory: syntactic change i.e., change in the formal features of a lexical item only occurs when caused by other types of change, e.g. phonological and semantic changes including the (dis)appearance of whole lexical items ( Longobardi 2001). While these approaches have in common to locate syntactic change in individual lexical items, they differ, however, in other respects. Roberts & Roussou (1999) for example can only account for changes of a lexical to a functional category, but not for changes from one lexical category to another. This kind of change is however well-attested by e.g. the numerous cases of V>P reanalysis in Chinese and African languages. In most accounts, P is not a functional category and certainly not an FC dominating VP. Roberts & Roussou (2003: 129) (henceforth R & R) take up this problem. Comparing V>P reanalysis with the reanalysis of the first verb in a serial verb construction (SVC) as complementiser, they suggest that the preposition resulting from a V>P reanalysis “remains lexical and therefore can have predicative properties”, ie V>P reanalysis is not an instance of grammaticalization, there being no “new grammatical (functional) morpheme”. Furthermore, such a preposition is claimed to still have relation properties in the sense of Hale & Keyser (1993) and no loss of the external argument is said to have occurred, the latter being a crucial component in the V>P reanalysis according to Whitman’s (2000) analysis. Chinese offers an ideal test case here: not only are V>P reanalyses attested (e.g. gen ‘accompany’ > ‘with’, cong ‘pursue’ > ‘from’, dui ‘face’ > ‘towards’, gei ‘give’ > for’), but they are likewise claimed to have SVC as sources. Chinese also provides us with a richly documented history of more than 3000 years. This paper examines Chinese data from different historical periods in order to investigate the precise mechanisms of V>P reanalysis. We will argue that contrary to R & R’s predictions, prepositions in Mandarin and other contemporary varieties do not retain the relational status of VPs: they are not predicative and do not provide evidence for an external argument. (1a) * Ta (shi) cong Beijing (1b) Ta [vP [PP cong Beijing] huilai-le ] 3SG be from Beijing 3SG from Beijing return-PERF (‘He is from Beijing.’) ‘He has returned from Beijing.’ (1c) Ta you ji -ge xuesheng [vP hui shuo zhongwen] /*[PP cong Beijing] ] 3SG have several-CL student can speak Chinese / from Beijing ‘He has several students who can speak Chinese / several students from Beijing.’ (2a) [TopP Youju [TopP [PP cong zher] [TP ni wang nan qu]]] post.office from here 2SG towards south go ‘The post office, from here, you go south.’ (Lu et al. 2000: 130) (2b) [TopP Akiu [TopP[VP chi fan] [TP ta hen hui tVP ]]], [TopP [VP zuo shi] [TP ta bu hui tVP ]]] Akiu eat food 3SG very can do work 3SG NEG can ‘Eat, Akiu certainly can, but work, he cannot at all.’ (Huang 1982: 164) (3a) (*bu /zhi) gei Akiu, wo zuo cha (3b) Wo [vP bu/zhi [vP [PP gei A.] zuo cha]] NEG/only for Akiu 1SG make tea 1SG NEG/only for A make tea ‘I only make tea for Akiu/don’t make tea for Akiu.’