Key points are not available for this paper at this time.
The implications of three versions of the contrastive analysis hypothesis (CAH) are explored. The strong and weak forms are rejected in favor of a more moderate version which predicts the results of a spelling error analysis on the dictation section of the UCLA placement examination in English as a second language. Spelling errors of foreign students whose native language employed a Roman alphabet (Group H) were compared with spelling errors of foreign students whose native language used some non‐Roman system (Group NR). An analysis of covariance with non‐spelling errors as the covariate, and spelling errors as the dependent variable showed NR superior to R (p < .025) with a significantly smaller percentage of spelling errors. These results support a more moderate CAH which predicts that spelling errors are based on “interference” of similar patterns due to false generalizations.
Oller et al. (Tue,) studied this question.