TOEIC Link Listening — Backchannel and Acknowledgement Marker Decoding Under Cooperative Segment: The Continuer-vs-Agreement Distinction That Separates Band-22 From Band-25

Backchannels — the short interjections that a listener produces while another speaker holds the floor (mm-hm, right, exactly, I see, sure, yeah, huh) — are one of the most under-decoded features of the TOEIC Link listening module at the CEFR B2-to-C1 transition. Band-22 listeners treat every backchannel as a uniform agreement signal and collapse the four functional categories (continuer, agreement, surprise, doubt) into a single "the listener agrees" interpretation. Band-25 listeners decode the function — distinguishing the mm-hm that merely keeps the speaker talking from the exactly that endorses the substantive claim from the huh that signals doubt — and the decoded function is what answers the comprehension questions that target speaker-listener stance alignment.

This guide formalizes the four-function backchannel taxonomy, catalogues the four failure modes that hold candidates at band-22, and outlines a four-week drill routine that installs functional decoding to automatic recognition. For broader listening-module preparation, see the listening discourse marker and turn management decoding guide and the listening agreement and alignment marker decoding under collaborative segment guide.

Why backchannel decoding discriminates so strongly

TOEIC Link listening passages frequently contain cooperative two-party segments — a project handoff, a status update, a mentoring conversation, a customer briefing — where one speaker holds the floor for an extended turn and the second speaker produces backchannels at the conversational beats. The comprehension questions that target these segments often ask not what the floor-holder said but what the backchannel-producer thought about it, and the answer is encoded in which backchannel the listener selected at which beat.

A backchannel-producer who issues mm-hm at every beat is signaling "I'm tracking — keep going" and is not committing to agreement. A backchannel-producer who issues exactly at a substantive claim is signaling "I endorse that specific claim." A backchannel-producer who issues huh at a substantive claim is signaling "that claim surprises me — I'm not yet convinced." A backchannel-producer who issues I see at a procedural step is signaling "I now understand the step — I previously did not." Each carries a distinct discourse function, and the function is what generates the question targets.

The band-22 candidate hears mm-hm and exactly and I see as undifferentiated agreement, picks the answer choice that says "the second speaker agreed," and misses the question that asks "did the second speaker actually endorse the proposal." The band-25 candidate decodes the function, recognizes that the second speaker was tracking without endorsing, and picks the answer choice that captures the actual stance — typically "the second speaker was listening attentively but did not commit to the proposal." The discrimination is what separates a band-22 outcome from a band-25 outcome on the small but reliable set of cooperative-segment questions.

The four-function backchannel taxonomy

Function 1 — Continuer

A continuer backchannel signals "I'm tracking — keep going" and carries no substantive stance toward the content. The canonical continuer inventory includes mm-hm, uh-huh, yeah (with rising or level intonation), okay (with level intonation), and right (with level intonation, not stress). The continuer is the most common backchannel in cooperative segments and is what permits the floor-holder to extend the turn without losing audience engagement, but it does not commit the backchannel-producer to anything beyond attention.

A candidate who decodes a continuer as agreement will over-attribute commitment to the backchannel-producer and will pick the answer choice that says "the second speaker accepted the proposal." A candidate who decodes a continuer as continuer will recognize that the second speaker has not yet committed and will pick the answer choice that says "the second speaker is still evaluating."

Function 2 — Agreement endorsement

An agreement-endorsement backchannel signals "I endorse the specific claim you just made" and carries substantive stance toward the content. The canonical agreement-endorsement inventory includes exactly, absolutely, that's right, precisely, that's the issue, and yeah (with falling or emphatic intonation that contrasts with the continuer use). The agreement endorsement is what marks the moments at which the backchannel-producer is committing to the floor-holder's substantive claim, and the commitment is what answers the question that asks "what did the second speaker actually agree with."

A candidate who decodes an agreement endorsement as a continuer will under-attribute commitment to the backchannel-producer and will pick the answer choice that says "the second speaker remained noncommittal." A candidate who decodes the endorsement correctly will recognize the substantive commitment and pick the answer choice that says "the second speaker endorsed the specific claim."

Function 3 — Surprise or doubt

A surprise-or-doubt backchannel signals "I find that claim surprising — I'm not yet convinced" and carries skeptical stance toward the content. The canonical surprise-or-doubt inventory includes huh, really?, is that so?, wait, hold on, hm (with falling intonation that contrasts with the continuer mm-hm), and oh? (with rising intonation). The surprise marker is what flags the moments at which the backchannel-producer is signaling that the floor-holder's substantive claim has hit an unexpected gap, and the flag is what answers the question that asks "what did the second speaker question."

A candidate who decodes a surprise marker as a continuer will miss the doubt and will pick the answer choice that says "the second speaker accepted the explanation." A candidate who decodes the surprise correctly will recognize the doubt and pick the answer choice that says "the second speaker was skeptical of the explanation."

Function 4 — Comprehension acknowledgement

A comprehension-acknowledgement backchannel signals "I now understand what was unclear before" and carries epistemic-update stance toward the content. The canonical comprehension-acknowledgement inventory includes I see, ah, got it, that makes sense, now I understand, and oh (with falling intonation, distinct from the rising-intonation surprise oh). The comprehension acknowledgement is what marks the moments at which the floor-holder has successfully repaired an information gap, and the marker is what answers the question that asks "what did the second speaker learn from the conversation."

A candidate who decodes a comprehension acknowledgement as agreement will over-attribute substantive commitment and miss that the second speaker was previously confused. A candidate who decodes it correctly will recognize the epistemic-update structure and pick the answer choice that captures the learning sequence.

The four failure modes that hold candidates at band-22

Failure 1 — Uniform-agreement collapse

The first failure mode is collapsing all four functions into a single "the listener agreed" interpretation. The candidate hears mm-hm, exactly, I see, and huh in the same passage and answers as if all four signaled agreement. The collapse is the standard band-22 default and is what the TOEIC Link listening module is engineered to detect. The repair is to install the four-function taxonomy as a default decoding lens and to tag every backchannel with its function before integrating across the passage.

Failure 2 — Intonation-disregard error

The second failure mode is ignoring the intonation cue that distinguishes the continuer use of a token from its substantive use. Right with level intonation is a continuer; right with stress and falling intonation is an agreement endorsement. Yeah with rising or level intonation is a continuer; yeah with falling intonation is endorsement. Oh with rising intonation is surprise; oh with falling intonation is comprehension acknowledgement. The candidate who disregards intonation will mis-categorize the token and will produce a wrong answer on a question that targets the function. The repair is to drill intonation contrasts on the same lexical token across functions.

Failure 3 — Density misreading

The third failure mode is misreading the density of backchannels in a segment as a stance signal. A backchannel-producer who issues continuers at every conversational beat may be signaling intense engagement, casual attention, or active listening without commitment, and the density does not by itself disambiguate. A candidate who reads high backchannel density as high commitment will over-attribute, and a candidate who reads low density as low commitment will under-attribute. The repair is to weight by function rather than by count: one agreement endorsement at a substantive claim carries more stance signal than ten continuers across the passage.

Failure 4 — Position-of-occurrence misreading

The fourth failure mode is misreading where in the floor-holder's turn the backchannel occurs. A backchannel produced immediately after a substantive claim attaches to the claim and commits the producer to that specific content. A backchannel produced after a transitional discourse marker (so, and then, which is to say) attaches to the transition and does not commit the producer to the upcoming content. The candidate who collapses position will mis-attribute the backchannel's scope and will answer questions about scope incorrectly. The repair is to drill backchannel-scope identification on transcripts with the floor-holder's turn segmented into claim-sized units.

The four-week drill routine

Week 1 — Function-tagging drill

The candidate works through 40 transcribed TOEIC-Link-style cooperative segments and tags every backchannel with one of the four functions (continuer / agreement endorsement / surprise or doubt / comprehension acknowledgement). The week's output is a tagged transcript log that surfaces how often the candidate defaulted to the uniform-agreement reading.

Week 2 — Intonation-contrast drill

The candidate works through 50 audio pairs that contain the same lexical backchannel under different intonation contours (right as continuer vs right as endorsement; yeah continuer vs yeah endorsement; oh surprise vs oh comprehension) and identifies the function from the prosodic contour alone. The week's output is a prosodic-discrimination log.

Week 3 — Position-scope drill

The candidate works through 30 transcripts segmented at the claim level and identifies which claim each backchannel attaches to, distinguishing claim-attached backchannels from transition-attached ones. The week's output is a scope-attribution log.

Week 4 — Integrated-comprehension drill

The candidate works through 25 full TOEIC-Link-style cooperative segments and answers the comprehension questions that target speaker-listener stance alignment, drawing on the function tags, intonation contrasts, and position-scope analyses from the prior three weeks. The week's output is a hit-rate log against a band-23 baseline of 72% and a band-25 baseline of 86%.

Backchannel decoding under listening-pace pressure

The cooperative-segment listening passage moves at natural speech pace, and any decoding the candidate is asked to perform must run in under one second per backchannel. The four-week drill works specifically because it pushes the function tagging, intonation contrast, and position-scope identification into automatic recognition where they cost zero marginal cognitive budget. A candidate who has completed the drill produces a function tag on every backchannel within the listening turn itself, and the tagged stream is what generates the correct answer to a question that asks "what did the second speaker actually commit to."

The interaction with the broader listening hedge and tentative-recommendation marker decoding under advisory segment discipline matters: cooperative segments frequently embed hedge-laden advisory turns from the floor-holder, and the backchannel-producer's selection between endorsement and continuer at the hedge marks the moment at which the producer either accepts or withholds commitment, which is exactly the question target on a substantial fraction of the items.

Closing — Function-decoded backchannels as a band-25 marker

Backchannel function decoding is one of the cleanest illustrations of the TOEIC Link listening principle that above band-22, the rater is no longer scoring whether the candidate can hear the content but whether the candidate can decode the stance the second speaker carries toward the content. The four-function taxonomy (continuer / agreement / surprise / comprehension) is one such decoding dimension. Installing it over four weeks produces a robust band-25 floor on the small but reliable set of cooperative-segment questions that target speaker-listener stance alignment.

For adjacent listening-module disciplines, see the listening counter-proposal and alternative-suggestion decoding under negotiation segment guide and the listening clarification request and information-gap repair marker decoding under troubleshooting segment guide.