Abstract
Along with personal interviews, individual word translation tests from the target language to the mother tongue are recognized as a reliable method of determining students’ actual lexical knowledge. However, as most English as a foreign language teachers are aware, the marking of these tests can be a laborious task. A far easier vocabulary testing format is the Yes/No (YN) checklist test, which can examine a large number of words while not over-burdening the marker. Pseudowords, which look like real words but do not bear meaning, have been added to the YN format to check for evidence of overestimation of lexical knowledge by test-takers. Four scoring formulae, which adjust YN results according to the number of pseudoword reports, have become established in the literature. Of these, the h-f formula has become recognized as the simplest to use for adjusting YN scores. This study presents a regression-based prediction formula derived from the h-f results in a pilot study, which was then applied to the YN h-f adjustments in a second study (the main study) to predict actual vocabulary knowledge as demonstrated by a meaning recall translation test of the same items. This prediction formula, labeled h-fRF, was compared with another regression-based formula as well as the original h-f formula. Results showed that 54% of the 455 individual h-fRF predictions were within 5% (4.8 of 96 words) of matching translation test scores, and 82% were within 10%, which were better than the other formula predictions. These results may be of interest to classroom teachers as they suggest that by using the h-fRF, the burden of marking translation tests can be reduced by the far easier YN test format.

Suggested citation