Principles and Parameters of Long-Distance Reflexives

One of the most striking differences between Chinese & English is the fact that in English, reflexives must have strictly local antecedents, but in Chinese, as well as in other East Asian languages like Japanese & Korean, long distance antecedents are possible: (1) Zhangsani renwei [Lisij zhidao [Wangwuk xihuan zijii /j/k]]. 'Zhangsan thinks Lisi knows Wangwu likes self', & (2) John thinks [Tom knows [Billi likes himselfi]]. In Chinese (1), the antecedent of the reflexive can be local, the subject of the clause containing the reflexive, or long-distance, the intermediate or matrix subject. In contrast, in English (2), only a local antecedent is possible. From the perspective of the government & binding theory, the differences between Chinese & English raise the following problem: on the assumption that reflexives in Chinese & English are both instances of bound anaphora, is it possible to formulate a tightly constrained theory of conditions on antecedency for bound anaphora that will apply to both Chinese & English? An attempt is made here to show that certain recent innovations in government & binding theory provide a solution to this problem, & to predict that reflexives must be long distance in Chinese & local in English. E. Battistella ("Chinese Reflexivization," paper presented at the 2nd Harbin Conference on Generative Grammar, Heilongjiang U, Harbin, People's Republic of China, 1987) proposes that reflexives in Chinese be analyzed along the lines previously proposed for English by David Lebeaux (see LLBA 18/2, 8402481), & later adopted in modified form by Noam Chomsky (Knowledge of Language: Its Nature, Origin, and Use, New York: Praeger: 1986). According to the Lebeaux-Chomsky-Battistella analysis, reflexives move in LF to INFL. This analysis provides a natural account of the subject orientation of reflexives, but the differences between Chinese & English with respect to long distance reflexives appear to remain unexplained. It is argued that the movement-to-INFL analysis, when embedded in the context of the account of the ECP found in Chomsky's Barriers (Cambridge: MIT Press, 1986), provides an automatic solution to the problem of why reflexives can be long distance in Chinese but not in English. The leading idea of Barriers is that certain but not all maximal projections constitute a barrier for antecedent government. Non L-marked maximal projections are barriers, but L-marked maximal projections are not. (L-marking is direct theta government by