Differences of Mean Dependency Distances of English Essays Written by Learners of Different Proficiency Levels

This study investigates the differences in the mean dependency distances (MDDs) of the English essays in a learner corpus, focusing on the different proficiency levels of learners, and the different dependency types. This study is based on the following three assumptions. Firstly, the MDDs of learners' production increase as proficiency levels increase. Secondly, there is an upper limit over which MDDs do not exceed, as predicted by the Dependency Distance Minimization principle. Finally, different types of dependencies show different tendencies across learners of different proficiency levels. This study attempts to verify these assumptions with substantial learner corpus data, categorized into subcorpora according to learner proficiency. Corpus analyses yield results that support these assumptions. These results are expected to constitute a prerequisite for employing the MDD of an individual learner's production to evaluate his or her proficiency level.