In spite of the large amount of research that has been done on first language reading, very little concerns the investigation and analysis of the processes in second language. A reading theorist, Goodman (1967) suggests that reading is a selective process in which the reader makes efficient use of minimal language cues in order to produce guesses. With this line of thinking, Smith (1978) indicates the significant importance of non-visual information in understanding printed information. Some psychologists, such as Carton (1971), Wildman and Kling (1979), and Olshaysky (1976-77), analyze cue systems and strategies the reader uses in the receptive processes.
This study was designed to examine how the reader's expectancy in the prose was affected by 1) cultural familiarity with the passage and 2) English proficiency. The main hypotheses to be tested were: the culturally familiar passage would facilitate the reader's expectancy regardless of his English proficiency; the reader with high English proficiency would expect more language elements than the reader with low proficiency regardless of the cultural familiarity of the passage. Sixty students enrolled in English 1 classes at Hiroshima University were given CELT (S—A, V—A, and L—A Forms) prior to three doze tets.
The doze tests, each of which included 49 items, were constructed by deleting every sixth word from three culturally different passages (Japanese, American, and Arabian) adapted from two books. The syntactic complexity of the passages was controlled. The passages described in common a young couple's and their parents' marital arrangements in each culture. The responses were scored by three different procedures: Exact replacement (Exact), Semantically acceptable (SEMAC), and Grammatically acceptable (GRCO) scorings.
The results of CELT showed that Group A (N=30) was superior to Group B (N=30) in English grammar, vocabulary, and hearing skill. A two-factor analysis of variance with groups (A and B) and conditions (Japanese, American, and Arabian cultural backgrounds) was done on each doze test scores of the two groups.
As predicted, both Groups A and B dozed significantly more items, exactly (P < .05) and semantically (P < .01) for the Japanese passage than for the American and Arabian passages, although the differences in the GRCO scores for the three passages proved insignificant. Group A was significantly more successful in filling the deleted words, exactly (P < .05) and semantically (P < .01), as well as grammatically (P < .01) than Group B in all passages.
On the basis of the results, we conclude tentatively that the culturally familiar passage tends to facilitate the reader's expectancy of semantic constraints regardless of his English proficiency. It is also confirmed that the reader with higher English proficiency is more successful in expecting successive language elements in terms of his knowledge on semantic and grammatical constraints.