Understanding middle graders’ language borrowing: how lexical and demographic characteristics predict similarity

Language borrowing from sources is a phenomenon used by developing writers as they are learning academic language, though there is much to be learned about how younger students borrow from sources. This study explores student writing, from a sample of 166 diverse middle graders, across topics to determine patterns in borrowing from instructional source texts. Computational techniques were used to identify borrowing, defined as similarity in pairs of sentences from student texts and instructional texts. Qualitative analyses of the sentence pairs that included borrowing showed that position statements, borrowed in part or whole, represent the majority of borrowing. Students also made superficial edits to sentences from instructional text by substituting or deleting words, but rarely adding words. Quantitative analyses of the sentence pairs revealed that more difficult words (longer, less frequent) were more likely to be borrowed. We found no differences in amounts of borrowing among students with varying English fluency levels (as measured by grade, standardized English test scores, and English language designation). Lastly, we examined whether borrowing was related to writing quality and found no effect. Using a unique combination of methodological approaches, we provide information about patterns of borrowing and point toward considerations of how students integrate instructional text into writing.