Computer spell checkers and collaborative peers: Intellectual partners.

Abstract

Over a semester, four groups of fourth grade students wrote essays as part of their regular class assignments. The students attempted to find and correct spelling errors in their own writings and in three prewritten stories. The four groups can be arranged in a 2 x 2 factorial design. The first factor is whether a collaborative peer partner (another fourth grader) is present to help with the tasks (yes or no); and the second is whether a technological partner (a computer spell checker) is present to help with the tasks (yes or no). This arrangement allows an extension of the Salomon, Perkins, and Globerson (1991) "partners in cognition" model (Salomon et al.'s claim is that a computer can be used as a "partner in cognition," a device to support and aid a person in doing a cognitive task). In this study, both technological partners (computer spell checkers) and same-age human partners (peers) are viewed as "partners in cognition." The design allows a test of the effects of each type of partner, and of the effects of the two partners in combination, namely, a test of whether the two effects (from a technological partner and a human partner) are additive. A model of error correction in spelling is presented, and data are collected in support of various phases of the model. Data include identification of misspellings, correction of misspelled words, the rate at which identification and correction were performed, efficiency (number of corrections per unit time spent on task), correctness of final products, and post-test spelling performance on a test customized for each student, containing only words that the student had actually misspelled and corrected during the study. Overall, the students in the two spell checker groups (working either with or without a human peer) outperformed the students with no technology partner; they identified and corrected a higher percentage of both their own misspellings and those in the prewritten stories. Correction rates however, were not significantly different among the groups. The two groups working without a peer were significantly more efficient than those working with a peer. The three groups with either or both types of partners had significantly better final products (in terms of the proportion of correctly spelled words) than did the group working with neither type of partner. No significant differences among the groups in individualized post test spelling scores were found. On none of the measures were there significant additive effects from a human or a technological partner, but for the individualized post test spelling scores, there was a nonsignificant trend in this direction. While both types of partners contribute to the tasks in this study, a technological partner contributes more toward improved performance than does a human peer. One might thus designate a spell checker as a 'super' partner in cognition.