통사적 복잡성 가설의 반례

Counter-evidence of Syntactic Complexity

  The role of syntax is to expand utterances by combining them via recursive mechanisms or to shorten outputs by deleting redundant elements. But, if our goal is to be understood, we should use the simplest syntactic arrangement possible, keeping our utterances easy and short. It might be that the more we elaborate on a sentence, the longer and more complicated it gets. When this happens, the more there is for us to hold in short-term memory, such that more complexity is meant to incur greater comprehensibility. Interestingly, however, a wider variety of structures under test show that there is no simple correlation between syntactic complexity and parsing difficulty. In this paper, we examine two syntactic processes, VP deletion and scrambling, to see how one is related to the other in parsing. Given the syntactic complexity hypothesis, it is expected that the reading time will be the longest when a sentence implemented both processes. Our experiment, however, shows that the reading time declines sharply when scrambling is added to VP deletion. This indicates that parsing is not directly related to the degree of complexity, but possibly to the types of transformations involved. It further implies that to make a sentence syntactically more complex can be a way of having economy satisfied when we relax the notion of economy to performance.