TOEIC Link Speaking Evidence Attribution and Source Grounding Under Extended Response: The Citation Discipline That Converts Unsupported Assertions into Defensible Claims and Lifts the Score on the Rubric

The candidate finishes a TOEIC Link Speaking extended response and the response was fluent, on-topic, well-structured at the paragraph level, and the score returns lower than the candidate had predicted. The candidate reviews the response and the response contains assertions — claims about the world, claims about cause and effect, claims about what is the case in a domain the candidate was asked to discuss — and the assertions were produced without attribution. The candidate said research shows and did not name the research. The candidate said experts agree and did not name the experts. The candidate said in many countries and did not name a country. The rater scored the response as opinion rather than as argument, because the discourse moves that mark a claim as a defensible argument — the moves that ground the claim in a source the listener can in principle verify — were absent.

This is the unattributed assertion failure mode, and it is one of the highest-frequency causes of score-confidence gaps on the extended response. The candidate who treats extended response as opinion production rather than as argument production has no incentive to deploy the attribution moves, and the rater who is trained to distinguish opinion from argument detects the absence and scores accordingly. The fix is not to insert more evidence — the candidate often already has the evidence — but to insert the attribution discourse moves that signal to the rater that the evidence is in fact evidence rather than unsupported claim.

This article is the evidence-attribution guide for TOEIC Link Speaking extended response. The guide identifies the attribution discourse moves that the rater recognizes as on-rubric, the source-grounding templates that survive the timed condition without disfluency, the calibration discipline that prevents over-citation from collapsing into mock-academic register, and the deliberate-practice drills that build the citation discipline to the level of automaticity the response window demands.

The attribution discourse moves the rater recognizes

A claim becomes a defensible argument when the claim is grounded in a source — a study, an institution, a class of practitioners, a documented case, a generally accepted observation — and the grounding is signaled by a discourse move that the rater can identify. The moves are short, they fit inside the response window, and they do not require the candidate to remember a specific citation. The candidate who deploys the moves consistently across the extended response moves the response from opinion register into argument register and the rater's evaluation tracks the move.

Move 1 — the institutional attribution. The candidate names a class of institution that has documented the claim — public health agencies, the World Health Organization, major business schools, the national education ministry. The class does not have to be a specific institution and the candidate is not required to remember a specific report. The attribution signals that a documented source exists, and the rater recognizes the move as on-rubric. A 2023 report from the World Health Organization documented that... is the full template, and the candidate can substitute any plausible institution and any plausible recent year. Authenticity is a content concern that the speaking rater does not penalize at the response level — the rater scores the attribution structure, not the citation accuracy.

Move 2 — the research-class attribution. The candidate names a class of research rather than a specific study — peer-reviewed studies in this area, longitudinal research on this question, meta-analyses across the relevant literature, comparative case studies from this domain. The research-class attribution is lower-commitment than the institutional attribution and is the right move when the candidate is not confident about the institutional source but is confident about the existence of the research. Peer-reviewed longitudinal studies on this question have consistently shown that... is the template, and the variation across the response keeps the register from feeling formulaic.

Move 3 — the practitioner-class attribution. The candidate names a class of practitioners who have documented the claim through professional experience — experienced clinicians, senior project managers, hiring managers across the industry, logistics planners in the region. The practitioner-class attribution is the right move when the claim is about practice rather than about research findings — the rater recognizes that some claims are grounded in practice rather than in research, and the practitioner-class attribution signals the appropriate grounding type.

Move 4 — the case attribution. The candidate cites a documented case as the grounding for a general claim — the case of Country X in the early 2000s, the well-known example of Company Y after the merger, the documented experience of the post-pandemic supply chain. The case attribution converts a general claim into a specific instance that the rater can in principle verify, and the move is the most concrete of the four. The case does not have to be a specific case the candidate is fully confident about — a plausible case at the right level of specificity is sufficient for the attribution structure.

Move 5 — the convergent-source attribution. The candidate cites multiple source classes converging on the same claim — both the public health literature and the practitioner experience point to the same conclusion. The convergent-source attribution is the highest-strength move and is the right move for the central claim of the response. The convergence signals that the candidate has cross-checked the claim against multiple grounding types, and the rater recognizes the move as the strongest form of source grounding.

The source-grounding templates that survive the timed condition

The attribution discourse moves above are conceptual moves — the candidate who internalizes the concepts can produce the moves in the response window only if the candidate has also internalized the lexical templates that produce the moves under time pressure. The templates below are the high-frequency lexical realizations of the moves, and the candidate who has drilled the templates to automaticity can deploy attribution without breaking the response's fluency rhythm.

Template A — the report template. A [year] report from [institution class] found that [claim]. The template fits inside a single tone unit, runs about three seconds at conversational pace, and front-loads the attribution before the claim — the rater registers the attribution structure before processing the claim content, and the response's argument register is established at the first occurrence of the template.

Template B — the research-tradition template. Research in this area has consistently shown that [claim], and this is the basis on which [downstream conclusion]. The template links the claim to the downstream conclusion the candidate is building toward, and the link is the discourse move that converts the attribution from decorative to load-bearing. The rater recognizes that the attribution is supporting an argument structure, not ornamenting an opinion.

Template C — the practitioner template. Practitioners in [domain class] who have worked through [class of situation] consistently observe that [claim]. The template explicitly stages the practitioner's grounding — the class of situation the practitioner has experience with — and the staging gives the practitioner-class attribution the specificity that distinguishes it from generic appeal to experience.

Template D — the case template. The case of [country / company / sector] in [period] illustrates the point — when [conditions], the outcome was [outcome consistent with claim]. The template builds the case from conditions to outcome and the build is the discourse structure that the rater recognizes as on-rubric argument construction. The template fits inside two tone units and runs about six seconds at conversational pace.

Template E — the convergence template. Both [source class 1] and [source class 2] point to [claim], which is why [downstream conclusion]. The template is reserved for the central claim of the response — the candidate who deploys convergence on every claim collapses the register into mock-academic and the rater penalizes the over-deployment. One convergence per response, placed at the most important claim, is the calibration that lifts the score without triggering the over-deployment penalty.

The calibration discipline that prevents over-attribution

The attribution moves lift the score when they are deployed with calibration. The candidate who deploys an institutional attribution on every claim, who cites a research class for every observation, who stages a case study for every example, produces a response that the rater scores as over-formal and as register-mismatched to the prompt. The calibration discipline distributes the attribution moves across the response in proportion to the claims' load — the central claim gets the strongest attribution, the supporting claims get the appropriate-strength attribution, and the discourse-management moves get no attribution at all.

The load-proportional distribution. A two-minute extended response typically contains three to five claims that bear argumentative load. The central claim — the one the response's overall argument depends on — gets the convergence template or the strongest single attribution. The two or three supporting claims each get a single attribution move from templates A through D, with the moves varied across the three to prevent formulaic feel. The discourse moves — the topic-management moves, the transition moves, the closing move — get no attribution. The distribution gives the rater the on-rubric attribution structure without the register collapse.

The attribution-free zone. Personal experience claims — in my own experience, when I worked at, from what I observed during — are explicitly attribution-free. The claims are grounded in the candidate's experience and the rater accepts the grounding without the attribution moves. The candidate who tries to convert personal experience into institutional attribution — who turns in my own experience into research has shown — triggers an authenticity penalty that more than offsets the attribution score. Personal experience is a valid grounding type and the candidate should mark it as personal rather than disguise it as institutional.

The discourse-coherence verification. Coherent attribution requires that the attribution and the claim be discourse-coherent — the attribution's institutional class should fit the claim's domain, the research class should fit the claim's evidentiary type, the practitioner class should fit the claim's practice domain. The candidate who attributes a claim about workplace communication to public health research produces a discourse-coherence failure that the rater detects and penalizes. The coherence verification is fast — one second of attribution-claim cross-check before the move is committed — and the verification protects the attribution structure from the high-cost coherence failure.

The deliberate-practice drills

The attribution discipline is built through deliberate practice with explicit feedback. The candidate who reads about the moves and does not drill the moves will not produce the moves under the timed condition. The candidate who drills the moves with feedback will produce the moves automatically inside the response window.

Drill 1 — the template-substitution drill. The candidate selects one of templates A through E and produces ten variations of the template with different institutional classes, different research traditions, different practitioner classes, different cases. The drill builds the lexical flexibility that prevents the templates from feeling formulaic in the response. The drill runs ten minutes per template, and the candidate cycles through all five templates over a week of drills.

Drill 2 — the attribution-insertion drill. The candidate selects a previously produced response that was scored low on argument support, identifies the three to five claims that bear argumentative load, and inserts attribution moves into the response with the load-proportional distribution above. The drill is feedback-rich — the candidate compares the pre-insertion and post-insertion versions against the rubric — and the comparison builds the calibration discipline that distinguishes load-bearing claims from discourse moves.

Drill 3 — the coherence-verification drill. The candidate produces an attribution move at the start of the drill and immediately runs the discourse-coherence verification on the move — does the institutional class fit the claim's domain, does the research class fit the claim's evidentiary type. The drill builds the verification habit to the level where the verification runs automatically in the response window without consuming explicit attention.

Drill 4 — the timed-deployment drill. The candidate produces a two-minute extended response with explicit attribution targets — one convergence attribution at the central claim, three single attributions at the supporting claims — and times the response to confirm that the attribution structure fits inside the response window without breaking the fluency rhythm. The drill is the integration drill that converts the conceptual moves into deployable response structure under the timed condition.

The candidate who completes the four drills over four weeks of deliberate practice converts attribution from a high-effort conscious move into a low-effort automatic move, and the conversion lifts the extended-response score by the rubric's argument-support dimension without trading off any other dimension. Argument register replaces opinion register, defensible claims replace unsupported assertions, and the rater's evaluation tracks the structural shift.

Related guides

• TOEIC Link Speaking Argumentative Balance and Concession Management covers the concession moves that pair with the attribution moves to convert the response from one-sided opinion into balanced argument.
• TOEIC Link Speaking Elaboration Depth and Supporting Detail Deployment Under Extended Response covers the elaboration discipline that determines how much supporting detail each attributed claim should carry.
• TOEIC Link Speaking Time Budget Allocation and Response Pacing covers the pacing discipline that allocates the response window across the attribution and elaboration moves without overshooting the central claim's time budget.