This article reviews the recent complex and somewhatc onfusing evidence on writing-to-learn and discusses why this lack of clarity exists. It then draws on the field of cognitive psychology to offer a way to reconceptualize how researchers might approach the study of the impact of writing on learning. This reconceptualization involves a modification in both how researchers select writing tasks and conditions in writing-to-learni nvestigationsa nd how they assess the possible knowledge changes due to writing. In the selection of writing tasks and conditions, it is suggested that researchers draw on theories of knowledge change to guide their selections. Four basic theoretical mechanisms potentially related to knowledge change due to writing are discussed. In the measurement of knowledge change, it is argued that writing may more likely influence structural than reproductive aspects of knowledge. Five methods for assessing structural changes in knowledge due to writing are considered.

The research on the cognitive processes in writing has led to models of the writing process such as that of Flower and Hayes. The work underlying these models has been carried out on relatively unconstrained writing genres. The purpose of this study was to investigate the types of processes used by journalistic writers when producing texts of varying constraints. A three tier methodology was used to offset certain methodological difficulties. Journalism students wrote either a news story or an editorial under either a pausal procedure, a pausal interview, or a protocol. It was found that news story writers paused more often and carried out more activities per writing session than did editorial writers suggesting greater monitoring activity by news story writers. In addition, news story writers were extensively concerned with accuracy and appeared to use a preorganized structure to guide writing and a priorities list to determine order of mention. Editorial writers paused less often and somewhat longer indicating a more open ended task. They appeared to adopt one of two strategies—treat the editorial as a news story with an opinion paragraph or search for a personal viewpoint. The results indicated that the Flower and Hayes model is applicable in a limited manner to journalistic writing.

The primary objective of this study was to investigate the role of content knowledge on topic choice in writing. Children's knowledge on topics they wanted to write about (want topics), on topics they did not want to write about (don't-want topics), and on topics the teacher chose (teacher topics) was measured using Langer's topic-specific knowledge measure. Results showed that children had significantly more knowledge, as assessed by the fluency and combined knowledge measures, on the want topics compared to the don't-want topics. In addition, children had significantly more knowledge on the want topics than on the teacher topics. There was no significant difference in knowledge between the children's don't-want topics and the teacher topics. These findings not only demonstrate the significant role of content knowledge on one writing process, topic choice, but also add support to students' self-selection of writing topics.

This article discusses how research on the origins of writing from such fields as anthropology, linguistics, philosophy, and history provides new perspectives on current writing research and on the teaching of writing. Four major issues are considered: (1) the functions of writing, (2) the influence of writing and writing systems on the writer, (3) the role of the writing topic on writing, and (4) writing and the decontextualization of knowledge. The implications of these issues for research on and the teaching of writing are considered.