TOEIC Link Writing — Source Citation and Attribution Management

The TOEIC Link Writing module includes two integrated tasks — the email-response task and the opinion-essay task — in which the candidate must work from one or more source texts and produce a written response that uses, refutes, or extends what the source said. The single largest performance discriminator on integrated tasks is paraphrasing skill, which is covered in detail in the guide on paraphrasing and summarization. The second-largest discriminator, which most candidates underweight, is attribution management — the ability to keep clear, unambiguous markers of what the source said and what the candidate is adding.

Attribution management has a much larger impact on band score than candidates expect because the rubric reads attribution failures as evidence of comprehension failures. A candidate who paraphrases a source claim cleanly but does not mark the claim as coming from the source is read as either confused about the source or attempting to pass off the source's idea as their own. Both readings cap the response at the same band ceiling, regardless of how clean the prose is otherwise. This article walks through what attribution the rubric is measuring, the four patterns that high-band candidates use, the three failure modes that cap intermediate candidates, and a three-week drill protocol that installs the skill.

What the rubric scores under attribution

The TOEIC Link Writing rubric for integrated tasks scores attribution on three dimensions that are evaluated jointly under the "source integration" axis of the scoring grid.

The first dimension is attribution presence, which is the binary question of whether the response makes it visible to the reader that source material is being used at all. A response that paraphrases the source without any attribution markers, however cleanly, scores below the B2 floor on source integration. The rubric requires that source material be marked, not merely that it be present.

The second dimension is attribution clarity, which is the question of whether each source-derived claim is unambiguously tied to its source. In responses that draw from two source texts, attribution clarity also requires distinguishing which claim comes from which source. A response that uses generic markers ("the article says" / "according to the source") without naming which article or which source loses points on attribution clarity even if every claim is correctly drawn.

The third dimension is attribution-to-original ratio, which is the question of whether the response includes enough original analysis to be classified as a response rather than a summary. A response that is 80 percent attributed source material, even if the attribution is clean, caps at the same band as a response with no attribution at all, because the rubric reads it as a summary task rather than an integrated response. The high-band target ratio is roughly 30 to 45 percent attributed source content and 55 to 70 percent original analysis, response, or extension.

The four attribution patterns

High-band candidates use four attribution patterns, deployed selectively based on what the source material is doing in the response. The patterns are not equivalent — each one signals a different relationship between the candidate's analysis and the source material — and using the wrong pattern reduces the rhetorical force of the response even when the attribution itself is grammatically clean.

Pattern 1 — Direct attribution

Direct attribution names the source and the source's claim explicitly. The phrasing is typically "The article argues that X" or "Smith claims that Y" or "The first source presents X as a primary cause." Direct attribution is the strongest signal of source comprehension and is the right choice when the candidate is introducing a major claim from the source for the first time, or when the response is about to disagree with the source and needs to make the disagreement target unambiguous.

The cost of direct attribution is that it is verbose and that overuse makes the response sound mechanical. The high-band candidate uses direct attribution two or three times in a typical integrated response — usually at the introduction of each major source claim — and uses lighter patterns for the surrounding text.

Pattern 2 — Embedded attribution

Embedded attribution integrates the source marker into the syntax of the candidate's sentence, typically through a reporting verb that compresses the attribution. Phrasings like "according to the article, X" or "as Smith notes, Y" or "the source's argument that X" are embedded attributions. The pattern is more efficient than direct attribution and lets the candidate maintain forward momentum while still keeping the source visible.

Embedded attribution is the workhorse pattern in high-band responses, appearing roughly five to eight times in a typical 350-word integrated response. The candidate learns to vary the reporting verb across the response to avoid lexical monotony — "notes," "argues," "presents," "frames," "characterizes," "treats" — and to choose verbs that signal the candidate's stance toward the source. "The article argues" is neutral, "the article assumes" implies skepticism, "the article demonstrates" implies endorsement. Verb choice does evaluative work that the rubric rewards under both source integration and stance clarity.

Pattern 3 — Trace attribution

Trace attribution uses a minimal marker to keep the source connection visible without restating the attribution. Phrasings like "this view" or "such an argument" or "the source's framing" point back to material that was attributed earlier in the response. Trace attribution is the right choice when the candidate has already introduced a source claim with direct or embedded attribution and is now developing the response to it across several sentences.

The pattern is essential for maintaining the right attribution-to-original ratio. Without trace attribution, the candidate either has to repeat the full attribution every sentence (which inflates the source content beyond the high-band ratio) or drop the attribution entirely (which loses the source connection). Trace attribution lets the candidate develop two or three sentences of original analysis while keeping the source visible through minimal markers.

Pattern 4 — Contrastive attribution

Contrastive attribution explicitly positions the candidate's view against the source's view. Phrasings like "while the article argues X, a more accurate framing is Y" or "the source treats X as primary, but this overlooks Y" are contrastive attributions. The pattern is the highest-leverage move in the integrated response because it accomplishes attribution, original analysis, and stance signaling in a single clause.

Contrastive attribution is what most candidates miss. The default move when disagreeing with a source is to state the disagreement without naming what is being disagreed with — the candidate writes "Y is the more important cause" without marking that the source said X. The rubric reads the unmarked disagreement as a weak claim because the reader cannot evaluate the response against the source position. The marked version — "while the article argues X, Y is the more important cause" — converts the same claim into a structured rebuttal, which scores roughly one band higher on the integrated-task rubric for the same lexical inventory.

For related coverage of how rebuttal structures work at the paragraph level, see the guide on rebuttal and counterargument structure.

The three failure modes that cap intermediate candidates

Internal practice-corpus data on candidates plateaued in the 20-to-22 band on integrated tasks shows three recurring attribution failure modes. Each is independently fixable, and most plateaued candidates have all three at once.

Failure mode 1 — Silent paraphrase

Silent paraphrase is the most common failure mode. The candidate paraphrases a source claim cleanly but provides no attribution marker, so the reader cannot tell whether the claim is source-derived or candidate-derived. A response with three or more silent paraphrases caps at the 18-to-20 band on source integration even if the paraphrases themselves are at the C1 level lexically.

The fix is to install a default attribution rule — every claim that came from the source text gets an embedded attribution marker on first appearance, no exceptions. The rule produces over-attribution at first, which the candidate then trims back through trace attribution in revision. The over-attribution phase is necessary because most plateaued candidates have under-developed attribution instinct and need to over-correct before reaching the right ratio.

Failure mode 2 — Generic attribution

Generic attribution uses non-specific markers like "the source," "the article," "it says" without distinguishing which source in a multi-source task or what specific claim is being attributed. The pattern is one band-step better than silent paraphrase but still caps responses in the 20-to-22 band because the rubric requires that attribution be unambiguous, not merely present.

The fix is to specify the source by descriptor on each major attribution — "the first source," "the second article," "the policy paper," "the editorial" — and to anchor each attribution to a specific claim through the embedded-attribution pattern. The fix typically adds two to three words per attribution but lifts the response by half a band on source integration.

Failure mode 3 — Attribution inflation

Attribution inflation is the inverse failure mode and appears in candidates who have over-corrected on the silent-paraphrase problem. Every sentence has a direct or embedded attribution, the response is 70 to 80 percent attributed material, and the rubric reads the response as a summary rather than an integrated response. The candidate scores high on source comprehension and low on original analysis, with a net cap in the same 20-to-22 band.

The fix is to install the trace-attribution pattern and apply the explicit ratio rule — count attributed clauses and original clauses in each paragraph and target a 1-to-2 ratio. The counting is mechanical at first and feels artificial, but it forces the candidate to develop the original-analysis muscles that the attribution discipline was crowding out.

The three-week drill protocol

The three-week drill protocol installs attribution control through structured daily practice. Each week targets one component of the skill and the components stack into a working integration by the end of week 3. The protocol assumes 30 to 40 minutes per day on weekdays.

Week 1 — Attribution presence and variety

Week 1 focuses on installing the embedded-attribution pattern as the default. The candidate writes one integrated response per day using a single source text and reviews each response for two metrics — number of embedded attributions and variety of reporting verbs. The week-1 targets are at least four embedded attributions per response and at least three distinct reporting verbs across the response.

By the end of week 1, the candidate should produce responses with visible source attribution without thinking about it. If attribution still feels effortful, the drill needs another five to seven days of practice before week 2 begins.

Week 2 — Trace attribution and ratio control

Week 2 introduces the trace-attribution pattern and the explicit ratio rule. The candidate continues the daily one-response practice but now uses trace attribution to develop two or three sentences of original analysis between each major source citation. The week-2 metric is the attribution-to-original ratio — count clauses, calculate the ratio, and adjust if it is outside the 30-to-45 percent range.

The most common week-2 failure mode is the candidate producing original analysis that drifts away from the source rather than developing the source's argument. The fix is to require that every paragraph have at least one contrastive or extending move that explicitly references the source — "the source's argument suggests" / "extending this framing" / "the implication of this position is" — so the original analysis stays anchored to the source material it is responding to.

Week 3 — Contrastive attribution and integration

Week 3 introduces the contrastive-attribution pattern and integrates all three patterns into responses to multi-source tasks. The candidate writes one two-source integrated response per day and reviews each for the three patterns — embedded, trace, contrastive — and the source-distinguishing requirement. The week-3 target is a response that uses all three patterns, distinguishes the two sources clearly, and includes at least one contrastive move that names a specific disagreement with one of the sources.

Candidates who reach the week-3 target consistently across five consecutive practice sessions typically demonstrate the new band on the next live administration. Candidates whose responses still cluster on one or two patterns should extend the protocol by another week of contrastive-pattern drilling.

Why attribution is the unlock for the 22-to-26 band shift

Most candidates plateaued in the 20-to-22 band on integrated tasks invest their next round of practice in vocabulary expansion or grammar refinement. The investment rarely shifts the band because vocabulary and grammar are not the binding constraints at the 20-band level — attribution management is. A candidate who installs the four attribution patterns and reaches the target ratio typically sees a 1.5-to-2 band shift on integrated tasks within three to four weeks. The same investment in vocabulary or grammar typically yields a 0-to-1 band shift on the same timeline.

The asymmetry exists because the integrated-task rubric weights source integration roughly twice as heavily as it weights lexical range, while the candidate's intuition treats them as equally important. The mismatch between rubric weighting and candidate intuition is why attribution drilling is the highest-yield investment in this score band. For related guidance on which writing-module skills to drill first based on diagnostic data, see the guide on writing task types and scoring criteria.