抄録
This study examined the scale of ideal second language (L2) self (Dörnyei, 2009)
for college English learners by applying the Graded Response Model (GRM)
(Samejima, 1969), one of the major models within item response theory. While Likert-scale
questionnaires continue to be a popular method for investigating ideal L2 self,
there have not been enough discussion on the improvement of the scale and the types
of ideal L2 self that are relevant for a particular population. This study reports on an
application of the GRM and analyzes data from college English learners who may not
be particularly motivated to learn English. The GRM analysis, together with an
analysis on an open-ended question, demonstrated that some items were more
discriminating and informative than others, and that certain types of ideal L2 self were
more relevant and realistic for the participants, pointing to the importance of careful
item construction and selection.
