关键词: 非宾格性假设, 混合语法, 语言习得与加工, 病理语言学, 机器学习

Abstract: The Unaccusative Hypothesis (Perlmutter 1978) initiated four decades ago within the framework of the Relational Grammar proposed a split of intransitives into two subtypes, the unergative and the unaccusative, which has triggered an extensive exploration in Generative Grammar ever since. The split of intransitives demonstrates by nature mixed properties of a language grammar. The related cross-typological linguistic research abroad in the past four decades falls into three primary fields: syntactic and semantic descriptions and interpretations, psycho-cognitive processing, and neurolinguistic and patholinguistic justifications. But here in China our interest is still primarily limited to the theoretical construal with occasional L2 processing, while neurolinguistic and patholinguistic areas are left untouched. This paper revisits what has been done so far and addresses some controversial issues as well. More importantly, we intend to point out that related research based on the corpus and big data seems to be largely neglected, not to mention machine learning. The paper suggests that it is high time for us to be cross-disciplinary.