August 11, 2000 (Preliminary Report)

7.  Types of Errors Counted
In this experiment, it was decided to deal only with two types of errors: those directly related to articles and prepositions. This decision was made not only because it is neither realistic to deal with all the possible errors nor it necessary to do so in view of the purpose of the current experiment, but also because these are the two major areas which the average Japanese EBP learners/users find most difficult to deal with (Someya 1999a). It should also be mentioned that, in this author's view, many, if not most, of the errors related to these two grammatical categories can be effectively avoided by reference to the BLC Concordancer.

Method of Error Count: In general, the number of errors in a given text is directly related to the length of that text. The longer the text, the more errors there usually are. In counting the numbers of errors in the current experiment, therefore, all the numbers (row counts) were standardized to percentage figures, or the number of errors per 100 words, so that a meaningful comparison between texts of different lengths can be made.