Kokkota, V. (1988). Letter-deletion procedure: a flexible way of reduced text redundancy. Language Testing, 5(1), 115–119.
Population: 18 – 19 year old Estonian first-year students at the Tallinn Polytechnic Institute.
Size: Six groups of 51 – 56 testees each, with each group taking eight different tests in one term.
Details: Average English proficiency levels and score distribution curves were equalized using the All-Union standardized test ITA-83. ITA-83 consisted of 135 items: vocabulary, 50 items; multiple-choice cloze, 50 items; grammar, 25 items; reading comprehension, 10 items. When pre-tested in 75 Soviet universities and institutions on about 7,500 testees (with ITA-82), it had reliabilities of .89 – .91 on different forms, and validities of .86 – .89 (Einväli et al. 1986).
The inter-item distance (IID) and the number of undeleted letters (NUL) of the two L-tests were between those of the classic cloze and the standard C-test. The intent was to monitor the extent of influence NUL and IID had on item difficulty.
For reliability purposes, each group was exposed to each test twice, each time taking a different test format or the same test format with a different item deletion rate. An interval of about two months was left between the two administrations to diminish the memorization effect. In each case, the classic cloze or the test with the lower item deletion rate was administered before the C-test or the L-test.
Conclusions: The letter-by-letter deletion of LDP is a more flexible and powerful means of controlling reduction of text redundancy than a cloze procedure or the C-test. It is also more general than either, because (1) it is possible to derive both cloze procedure and C-test from LDP; (2) the parameters of LDP are between those of the cloze procedure and C-test, which are its extreme modifications; and (3) LDP has a number of the advantages of the cloze, RD and C-test, while overcoming some of their disadvantages.” The facility curve of the L-tests was between the C-test and classic cloze test curves, with the C-test facility curve at 60 – 70 and the cloze curve from 36 – 48.
Comments: Kokkota regards RD, the rational deletion system (Bachman, 1982, 1985), and the C-test (Klein-Braley, 1985) the “major improvements on the cloze procedure” of the decade. His aim was to outline a comparatively new letter-deletion procedure (LDP) that would flexibly match the advantages of the RD and C-test, while avoiding their main disadvantages.
It was this introduction of the L-test that caused me to develop ClozeOnline.[1]
[1] ClozeOnline™ creates and administers, online, this L-test, all forms of the C-test, and the classic cloze.