TOEFL Preparation using Recreational Reading
Beniko Mason
Proceedings of 2007 (9th) International Conference and Workshop on TEFL and Applied Linguistics
Department of Applied English, Ming Chuan University, Taiwan

Can students prepare for the TOEFL just by doing recreational reading? I present here the results of a study that may have profound theoretical, practical and financial implications.

Previous research shows a clear correlation between the amount of reading students say that they have done and performance on the TOEFL: (Gradman and Hanania, 1991; Constantino, SY Lee, KS Cho, and Krashen, 1997; see also Constantino, 1995, for a suggestive case history).

Subjects were six university level students of English as a foreign language in Japan. All were aware of the "power of reading," having taken courses from the author, and were enthusiastic participants in the study. This is, thus, a study that is "loaded for success": Failure to achieve results with these subjects probably makes experimentation with less enthusiastic subjects unnecessary.

All students were interested in improving on the TOEFL, and all agreed to engage in free reading during a time when they had no other source of English input. Three read during their summer vacation, two during the spring break, and one during the academic year (an Arabic major who was taking no English classes at the time).

Students had access to a library of about 4000 graded readers (about 700 different titles), and were advised to begin with very easy graded readers, and read about 70 to 100 pages per week. Some students read the graded readers. Others read authors such as Sidney Sheldon, Paulo Coelho, Judy Blume, and Bertice Berry.

No book reports or summaries were required; students were only asked to keep a record of what they read. The readers were entirely on their own during this time; they did not meet with the researcher.

Students agreed to take versions of the TOEFL test available from the Educational Testing Service (TOEFL, 2005). These versions are constructed from previously administered tests and their reliability is high (total reliability = .95).

1