Foreign language reading comprehension: Text representation and the effects of text explicitness and reading ability.

Abstract

The present study investigated text reconstruction performance of EFL university-level students reading a fairly long naturally-occurring popular magazine article taking two factors into consideration: degree of text content explicitness and EFL reading ability level. More specifically, it attempted to examine a deeper level of text representation, or what Kintsch and his associates label the situation model (Kintsch & van Dijk, 1978; van Dijk & Kintsch, 1983; Kintsch, 1974, 1988, 1992, 1994, 1998), by subjects of two reading ability levels in EFL [high and low]. Subjects performed an immediate written reconstructive recall after reading one of the two versions [fully explicit and less explicit ] of a popular science article. This recall or text representation reflected (1) the comprehension or the reconstruction of the text as a whole, (2) the distribution of information in the text, i.e., in terms of its macro and microstructure, and (3) any correct inferences that may have been generated. In addition, the study considered the influence of text difficulty, topic interest, and topic familiarity in the reconstructive representation. Ninety-two Brazilian university-level subjects participated in this study. Comprehension was measured quantitatively in terms of: (1) the number of propositions recalled from reading one of the two versions, (2) textbase recall, and (3) inferential recall. In both text versions, six hierarchical levels of information were considered. There were four main findings of the present study: (1) Text version had an impact on the reconstructive process. Readers benefited from reading a less explicit version of the text regardless of their reading ability level, although high reading ability level readers outperformed low reading ability ones. (2) The fully explicit version had an advantage over the less explicit version only with respect to the construction of the textbase representation. (3) Results respected the Hierarchy Principle, that is, higher-level propositions were better and more frequently recalled than lower-level ones. (4) Text difficulty and topic familiarity were not determining factors in the reconstructive representation. Topic interest, however, was shown to be a significant factor in the construction of the textbase as well as in the reconstructive process as a whole of low reading ability subjects. The findings of the present study are broadly consistent with those reported in earlier cognitive research in the area of text representation, particularly with those which examined text comprehension in the context of the Kintsch & van Dijk (Kintsch & van Dijk, 1978; van Dijk & Kintsch, 1983, Kintsch, 1974, 1988, 1992, 1994, 1998) model of reading comprehension.