Fifth Graders Writing From Sources: Copying Isn't Citing
Mr. Bell's fifth graders spend three 45-minute periods writing an explanatory essay from two short articles about later school start times. The prompt: 'Using both articles, explain the arguments for and against later start times.' The rubric has four traits - focus/claim, use of evidence, organization, and conventions. Results across 24 essays: 19 state a main idea; 4 quote or paraphrase from both articles; 14 contain at least two consecutive sentences copied verbatim from a source; 21 end with a conclusion that merely restates the introduction. Six students are English learners at the bridging level, and three of those six wrote no more than five sentences. Mr. Bell has one instructional day left before the unit ends, and the district writing benchmark is in two weeks.
1. Which rubric trait should be the target of the one remaining day, and why not the others? Use the numbers.
Read the worked answer
Target: use of evidence. Only 4 of 24 students drew on both sources, and 14 of 24 copied at least two consecutive sentences. Copying is not evidence use - it is evidence location without selection, paraphrase, or connection to a claim. That is the largest single gap between what the task asked for and what students produced.
Why not focus/claim: 19 of 24 already state a main idea. Reteaching it buys little.
Why not conventions: it matters for the benchmark, but 45 minutes of grammar will not change the essays' substance, and mechanics errors are cheaper to fix later than a missing evidence habit.
Organization is a real secondary problem - 21 of 24 conclusions just restate the introduction - but the fix (a conclusion that returns to the claim and says why it matters) is short and can be folded into the same lesson as the evidence work, because a conclusion that uses evidence is naturally different from a restated intro.
2. Design the single-day reteach: what you model, what students do, and how you would know within the period whether it worked.
Read the worked answer
Model (10 min): With one short shared paragraph on the board, think aloud through three moves - choose one fact that actually supports the claim, say it in my own words with the source closed, then connect it with a sentence like 'This matters because…' Deliberately show a copied sentence next to a paraphrased one so the difference is visible.
Do (25 min): Students revise one paragraph of their own essay rather than start a new piece - the goal is replacing a copied sentence with a paraphrase plus an explanation. Supply a two-part frame: '[Source] found that ___ . That shows ___ because ___ .' Circulate and check against three criteria: own words, tied to the claim, one sentence of explanation.
Know it worked (10 min): Collect every revised paragraph as an exit ticket and sort into three piles - paraphrase + link, paraphrase without link, still copied. Success criterion: 18 of 24 in the first two piles, with no more than 3 still copied.
Why revise instead of rewrite: revision works on the exact error the data showed, takes less time than new drafting, and lets the evidence-use move be practiced inside a text students already understand.
3. What two supports would you build in for the English learners, and what would you deliberately not change for them?
Read the worked answer
Supports:
- Oral rehearsal before writing - the student says the paraphrase aloud to a partner, then writes it. This separates the language-production load from the transcription load and is exactly what the three students who wrote five sentences or fewer need.
- Sentence frames plus a bilingual glossary of the two or three key terms (e.g., 'start times,' 'adolescent'). Frames supply the academic syntax; the glossary keeps the content accessible so the evidence-use task stays the target.
What not to change: the task demand. These students should still be selecting evidence and linking it to a claim - lowering that expectation would remove the very skill being taught. Scaffolding the route to the task is appropriate; substituting a recall task is not.
Also: let them work from one source at a time with a graphic organizer split into 'for' and 'against' columns, rather than holding both articles in mind at once. Reducing the working-memory load is a fairness adjustment, not a lowering of the standard.