Research on the effects of schematic knowledge on comprehension dates back to the ’70s -- the seminal works in the first language by Bransford & Johnson (1972), Anderson et al. (1977), and Steffenson et al. (1979). Ever since, a great number of studies have investigated the same phenomenon in the context of L2 reading and listening comprehension (Carrell, 1983, 1987; Anderson & Lynch, 1988; Long, 1990; Schmidt-Rinehart, 1994; to name but a few). It is imperative to acknowledge the contribution of these studies to our enhanced understanding of the processes involved in comprehension, yet two issues seem to have received almost no attention: first, the extent of the predictive power of schema, and second, the question of defining schema in terms of its constituents.
The majority of research findings in this area have focused on investigating the effect of existence vs. non-existence of stereotypic schema, making use of conventional and predictable linguistic input. These results, however, may not be generalizable to real language use that often exceeds the limits of pre-established frames. With this in mind, the primary purpose of the current study was to explore the effects of employing nonconventional and atypical input on listening comprehension of EFL learners at different levels of language proficiency.
With regard to the second issue, it should be pointed out that despite the existence of a large body of related literature, schematic knowledge, itself, is still treated as a broad and blurred concept. That is, schema often has been operationally defined in general and fuzzy terms such as cultural knowledge, religious knowledge, technical knowledge, etc. with little attention to its underlying factors. In the present study, nonetheless, it was attempted , as the secondary purpose, to bring schematic knowledge under scrutiny, analyze it into its possible components and determine the informative weight of each and every component in listening comprehension. To this end, Brown & Yule’s (1983) model was utilized, in which speaker, listener, place, time, genre, topic and co-text are supposed to be the features of schematic knowledge.
To provide empirical evidence for the study, the following steps were taken. Two 31-item multiple-choice tests were developed. In both tests, the input was preceded by some pieces of schematic information. In the first test (called compatible), conventional and in the second test (called incompatible), nonconventional situations were incorporated. The items of the two tests were randomly distributed to form one 62-item test which was then pretested with 70 EFL adult learners. It proved to enjoy acceptable psychometric characteristics ( reliability index = .79, averaged item facility = .39, and averaged item discrimination = .41). It was then concurrently validated against the TOEFL with the participation of 100 EFL learners (reliability and validity indices .83 and .77, respectively). The listening section of the TOEFL also served as the no schema part of the project.
Concerning the first purpose of the study, multivariate analysis of variance (MANOVA) for repeated measurement was applied to the data and all Fs turned out to be significant (98.03 for high and low levels of language proficiency, 446.17 for compatible, incompatible, and no schema tests, and 45.04 for levels by tests, at the p<.01 level). Due to the significance of Fs, Scheffe test was conducted to locate the source(s) of the observed differences. Concerning the second purpose, to check the construct validity of components of schematic knowledge, three varimax rotated factor analyses, and to determine their informative weight, three multiple regression analyses were applied to compatible, incompatible and combinations of the two tests (the 62-item test).
The results of the Scheffe test indicated that the low proficient group performed quite differently on the compatible and incompatible tests, whereas the difference between these two tests was not statistically significant for the high proficient group. This points to the fact that the higher the level of language proficiency, the less difficulty the learner will have in handling novel and non-conventional situations. Moreover, activation of schema did not prove to be the single most important factor in improving the subjects’ listening comprehension. Overall, the findings indicated a crucial role for linguistic knowledge and cast doubts on recent theories that underestimate the importance of bottom-up factors in comprehension.
The results of factors analyses of the components of schema as proposed by Brown & Yule (1983) were not conclusive enough to either support or reject the psychological reality of those components. While the results of regression analyses indicated that the number of components presented to activate schematic knowledge does not have much to do with their informative weight in listening comprehension.