Acceso usuarios
Joel Walters, Yuval Wolf
Ninety‐three subjects participated in a series of experiments investigating how the number of errors from different linguistic sources affects evaluative judgments about the need for revision in a non‐native language. In the first three experiments, groups of non‐native and native writers of English as well as EFL teachers were exposed to bifactorial combinations of syntactic and lexical errors incorporated in passages from an English composition textbook. Subjects were exposed individually to all factorial combinations of errors from both sources and asked to judge how much effort was needed to make the passages well written. Results from all three experiments show lexical errors having a greater effect. Employing the framework of information integration theory and functional measurement, it was found that non‐native writers used an additive rule to integrate information from both sources, while native writers used differential averaging. Non‐native writers participated in two additional experiments, where cohesion errors were combined bifactorially and trifactorially with syntactic and lexical errors. Lexical and cohesion errors showed greater effects than syntactic errors. An additive rule was used to integrate syntax with either lexicon or cohesion, while a differential averaging rule was used for the integration of lexicon with cohesion. The procedure was adapted for a classroom experiment; it included actual error correction along with metalinguistic judgment. The findings conform to those of the previous experiments.
