Babaii, E., && Ansary, H.

Babaii, E., & Ansary, H. (2001). The C-test: a valid operationalization of reduced redundancy principle? System, 29, 209-219. Tehran, Iran: Shiraz University.

Population: 32 engineering students taking an English for Science and Technology course. 10 females and 22 males of heterogeneous background. Ages ranged from 20 to 30.

Purpose: As the title implies, this study was to explore whether or not the C-test, as it was claimed, serves as a valid operationalization of the reduced redundancy principle (RRP).

Instrumentation: Used were the 1995 TOEFL test (without the Listening Section) and the recommended format for the C-test (Klein-Braley, 1997), i.e., an original set of eight short texts (that were later narrowed down to five) with 20 mutilations each. The excerpts were taken from the textbooks Practice and Progress (Alexander, 1968) and To start you practicing (de Freitas, (1974). The estimated Flesch reading ease for the text was rated 69.9 (an average difficulty level). Pre-testing showed that all had high reliability values. Five subsets whose IF index values were above 0.32 were selected for the final version. Furthermore, to adjust to the overestimated reliability of the C-test (resulting from the application of measures of internal consistency), these selected five were again treated as Super items with Crombach’s alpha formula. Its reliability was again estimated at a=0.93.

Details: In the process, an attempt was made “to investigate the frequency and type of micro- and macro-level cues that EFL learners employ to restore the mutilations in the C-test.” The two tests were administered concurrently. The retrospective data collection was done orally.

Results: Analysis of the protocols indicated that “there exist four major types of cues with varying frequencies: (1) automatic processing; (2) lexical adjacency; (3) sentential cues; and (4) top-down cues” . . . automatic processing (n=110, 16.6%) because the intact half of a mutilated word reminds an examinee of some familiar and highly frequent words and expressions as in [kn _ _] and [of co _ _ _ _] ([know] and [of course]); lexical adjacency (n=363, 54.9%) through reliance on the syntactic and lexical cues provided by the words immediately before and/or after the mutilated word; sentential cues (n=148, 22.4%) because of reliance on the grammatical features such as tense, referencing, and cohesive ties with surrounding sentences; and through top-down cues (n=40, 6.1%) such as background knowledge, topic familiarity, and similarity chains . . . items discovered elsewhere in the text.

Comments: The texts used for the C-test are stored at http://clozeonline.us/Babaii.htm. The article’s keywords are: C-test, Language proficiency, Reduced redundancy principal, Language testing, and Verbal protocol analysis.