C-Tests: Four Kinds of Texts, Their Reliability and Validity

Page No.: 
41
Writer(s): 
Akihiko Mochizuki, Aichi University of Education

This study examines test appropriateness for constructing the most effective CTest.
The following tests were administered to 42 college freshmen from April
1992 through May 1992: the Second Grade Test of the Society of Testing English
Proficiency (STEP), A Comprehensive English Language Test/or Learners o/English
(CELT) Listening, a Dictation test, and four C-Tests whose tests used Narration,
Explanation, Description, and Argumentation. Results indicate the following:
First, the reliability of the Narration C-Test is the highest ( r c: 0.928). Second,
there is a fairly high correlation between the scores of the STEP test and the
Narration C-Test. Third, there is a very low correlation between the scores of CTests
and the CELT Listening Comprehension test. Fourth, there is a very low
correlation between the scores of the C-Tests and the Dictation test. The study
therefore indicates that a C-Test which uses a long narration text seems to be a
promising means of measuring a language leamer's overall language proficiency,
and what a C-Test measures seems different from what a listening test and a
Dictation test measure. Further research is needed to investigate the correlations
between the Narration-based C-Test and a more reliable criterion test like the
TOEFL.

PDF: