Analysis of Student Revisions on a State Writing Test

Abstract

Type and quality of revisions made by students between first and final drafts of a state writing test were scored using a revision taxonomy. Scorers categorized revisions first by unit (e.g., word, phrase, sentence), and then by type (e.g., addition, substitution, spelling). They then evaluated the impact of each revision on the readability of the final draft as “improved,” “decreased,” or “neutral.” Descriptive analyses revealed three major findings. First, in terms of unit changes, word revisions were the most prevalent (approximately 40%) across students in Grades 5 and 8. Second, when coding for type of revision, substitutions accounted for approximately 45% of all revisions, while revisions classified as “additions” had the most positive impact on students' final drafts (i.e., positively impacting readability 66% to 76% of the time). An interesting third finding was the relationship between rate of writing and score on the writing test: Students in special education with Individualized Education Programs goals in writing made approximately the same number of revisions per 100 words as students in the general population but wrote 100 words less, on average.

What and When Second‐Language Learners Revise When Responding to Timed Writing Tasks on the Computer: The Roles of Task Type, Second Language Proficiency, and Keyboarding Skills

“…The conditions of writing are also likely to have affected the participants’ revision behaviors. It is possible, as Crawford, Lloyd, and Knoth () have argued, given the time pressure, that some participants opted to make ‘safe’ revisions that improve form and mechanics without necessarily changing text organization and content (cf. Porte, ).…”

A product- and process-oriented tagset for revisions in writing

“…Moreover, some studies described whether the starting point was correct: in other words, whether an error or non-error was revised (e.g., from a correct non-error into an error; Wobbrock & Myers, 2006). In contrast, Crawford, Lloyd, and Knoth (2008) evaluated the quality of all revisions. As it is harder to determine whether semantic changes are "correct," they renamed the labels into increase, decrease, and neutral.…”

“…Some researchers focus more on the smaller domains, e.g., sub-grapheme, grapheme, morpheme, word, and above-word . Others added domains such as theme (extended statement of one idea; Sommers, 1980) or punctuation (Crawford et al, 2008). Xu (2018) took a lower-level approach by merely counting the letters deleted or produced.…”

“…For these manual annotations, different types of data were used to identify revisions. Some studies compared drafts (Crawford et al, 2008;Min, 2006;F. Zhang & Litman, 2015) or Wikipedia revisions (Daxenberger & Gurevych, 2012), while others used keystroke logs (Barkaoui, 2016;Wengelin, 2007), sometimes combined with thinking-aloud data (Stevenson et al, 2006).…”

Determination of Hearing-Impaired Students’ Requirements for Editing and Revision of Written Texts

“…Revision Taxonomy: Revisions made to written text can be analyzed in three stages (Crawford, Lloyd, & Knoth, 2008;Faigley & Witte, 1981), consisting of revision unit, revision type, and quality of revision. Sound/syllable and affix correction elements were included as revision types and revision units, since each affix added to the base of a word has an alteration feature that functions in different ways (Hengirmen, 1995).…”