EFFECTS OF GENERALIZATION INSTRUCTION ON THE WRITTEN LANGUAGE PERFORMANCE OF ADOLESCENTS WITH LEARNING DISABILITIES IN THE MAINSTREAM CLASSROOM

7

US high-school students with learning disabilities

The University of Kansas Error Monitoring Strategy, taught with the COPS mnemonic - Capitalization, Overall appearance, Punctuation, Spelling - is the one packaged intervention in education whose explicit target is conventions in the student's own writing. This study is the only one anywhere that tracked mechanics error rate across a school year INTO SETTINGS WHERE THE STRATEGY WAS NEVER TAUGHT. Multiple-probe across strategies: four strategies (Sentence Writing, Paragraph Writing, Error Monitoring, Theme Writing) taught in sequence by the students' own resource-room teachers, with writing sampled in general-education English and social studies classes as well. Design weaknesses are disqualifying for anything but a hint: n = 7, single-case, no control group, and four strategies plus a full year of maturation and course-placement change are all confounded together. The standardised-test and GPA gains have no counterfactual at all. Numbers taken from the authors' own 2009 retrospective review (Schumaker & Deshler, Learning Disabilities Research & Practice), not from the primary report.

Mechanics error rate in resource-room writing fell from .27 errors per word to .04 - roughly one error in every four words to one in every twenty. The transfer observation is the reason the study is recorded: writing collected in mainstream English and social studies classes, where the strategies were never taught, showed that "most of the students also made the same kinds of gains", and "the two students who did not generalize... did so quickly after they had been taught to do so" - i.e. generalisation happened for most, and for the rest it had to be taught explicitly as its own step. Woodcock-Johnson writing grade equivalent moved 6.2 to 8.2; the district minimum-competency writing exam score was 3.5 out of 5 against a district mean of 2.5; GPA 2.1 to 2.7. The earlier validation study of the same strategy (Schumaker et al. 1982, n = 9, Grades 8-12) took error detection on teacher-written passages from about 29% to 92-97% and errors in the students' own writing from about one per four words to one per 33.

Not applicable in the usual sense - participants serve as their own controls, which nets out stable individual differences by construction. The corresponding cost is total: nothing separates the strategy from maturation, practice, teacher attention or regression to the mean.