TOEIC Link Listening — Numeric Distractor and Near-Miss Figure Discrimination Under the Financial Summary: The Decoy-Number Architecture That Separates Band-7 Mishears From Band-9 Value Tracking

The financial-summary listening segment — a quarterly-results readout, a budget briefing, a sales recap — is engineered to produce a specific band-7 error. The audio states four or five numbers in the space of two sentences: "Revenue rose to 4.2 million from 3.8 million, and while the board had targeted 4.5, the 4.2 figure still beat the analyst consensus of 4.0." The comprehension item then asks a single question — what was the target, what did revenue actually reach, what was the consensus estimate — and the band-7 listener, who heard a cluster of numbers and retained the most salient one, answers with a figure that was said but is not the answer. The listener did not mishear the number. The listener mistracked which number was bound to which role.

The band-9 listener treats each number the moment it is spoken as a bound pair: not "4.5" but "4.5, the target"; not "4.0" but "4.0, the consensus." When the question arrives, the listener retrieves by role rather than by salience, and the near-miss decoys — the numbers that are numerically close and were genuinely spoken — hold no pull, because they are filed under a different role. This guide formalizes the decoy-number architecture and the value-tracking discipline that keeps every figure attached to its referent through the moment of retrieval.

Why the near-miss decoy works on the band-7 listener

A distractor in a financial summary is not a wrong number invented to trap you. It is a real number, spoken in the same breath as the correct one, differing from it by a small margin. The 4.5 target sits beside the 4.2 actual; the 4.0 consensus sits beside both. Every decoy is a value the passage actually stated, which is what makes it dangerous: the listener who tries to verify the answer by asking "did I hear this number?" gets a yes for the decoy as readily as for the correct figure. Recognition of the number is not discrimination between numbers.

The band-7 failure has a predictable shape. Under the memory load of holding several figures at once, the listener retains the numbers but drops the bindings — the labels that say which number plays which role. What survives is a bag of values with no attached roles, and when the question asks for "the target," the listener reaches into the bag and pulls the most available figure, which is usually the largest, the last, or the one that carried vocal stress. The near-miss decoy is available precisely because it was near the correct one in the audio and in the numeric range. The architecture exploits the gap between remembering a value and remembering what the value was for.

The decoy-number architecture

The architecture has three layers, and the comprehension item can target any of them.

Layer one — the co-mentioned near-miss. Two figures of the same kind are stated close together: last quarter versus this quarter, budgeted versus actual, list price versus discounted price. The decoy is the figure of the same type that is not being asked for. "Down from" and "up to" clauses are the reliable trigger — the "from" number is the decoy when the question asks for the current value, and the "to" number is the decoy when the question asks for the prior value.

Layer two — the aspirational figure. A target, forecast, goal, or projection is stated alongside the achieved value. The decoy is the number that was hoped for rather than realized. Words like "targeted," "aimed for," "projected," "hoped to reach," and "fell short of" mark a figure as aspirational, and the comprehension item frequently asks for the actual result while the aspirational number sits nearby as bait.

Layer three — the third-party figure. A number attributed to an outside source — an analyst estimate, a competitor's result, an industry average — is stated alongside the company's own figure. The decoy is the externally sourced number when the question asks about the company, and vice versa. Attribution phrases ("analysts had expected," "the industry average was," "our competitor reported") are the markers that a figure belongs to a different entity.

The three layers can stack in a single segment, which is how a two-sentence passage produces four plausible-looking answer options that are all real numbers from the audio.

The three tracking rules

Rule 1 — Bind at the moment of utterance, not at the moment of the question. The band-9 discipline is to attach the role to the number as it is spoken, in real time, rather than storing bare values and hoping to reconstruct the roles later. When you hear "4.5," the internal note is not "4.5" but "target: 4.5." The binding costs a fraction of a second and it is what survives the memory decay that strips bare numbers of their roles. Reconstruction after the fact does not work, because the roles are exactly the information the near-miss architecture is designed to erase. This real-time binding is the listening-side analogue of the referent-tracking that governs approximation and rounding discourse under the quantitative summary, applied to discrete labeled figures rather than to running estimates.

Rule 2 — Tag the transition words, because they carry the roles. The roles are almost never stated as bare labels; they are carried by the function words around the numbers. "From X to Y" encodes prior-versus-current. "Targeted X but reached Y" encodes aspirational-versus-actual. "Analysts expected X; we delivered Y" encodes external-versus-internal. A listener who tracks only the content words — the numbers and the nouns — loses the roles, because the roles live in the prepositions and conjunctions. Listening for "from," "to," "but," "versus," "against," "compared with," and the attribution verbs is what makes the binding possible. The dependence on function words for meaning is the same phenomenon that makes accent and reduction handling matter, discussed in accent variation and regional pronunciation — a reduced "to" is easy to miss, and missing it collapses the from/to distinction.

Rule 3 — Answer by role, then confirm the value; never the reverse. At the question, the band-9 sequence is to identify the role the question asks for — target? actual? consensus? — and retrieve the value bound to that role, rather than scanning the answer options for a number that sounds familiar. Familiarity is worthless here because every option is familiar. Retrieving role-first means the decoys never enter the comparison: if the question asks for the actual result, the target and the consensus are simply not candidates, no matter how recently or how loudly they were spoken.

The near-miss resistance drill

The drill isolates the binding skill from the numeric-recall skill, because the two are separate and the near-miss architecture attacks the binding.

Take any financial-summary audio or transcript with at least three numbers. On first pass, do not try to memorize the numbers. Instead, write only the roles in order as they appear: "prior revenue, current revenue, target, consensus." On second pass, fill each role with its value. This forces the roles to be captured first and the values to be hung on them, which is the correct order and the reverse of what the band-7 listener does under pressure. After several sessions the role-first capture becomes the default, and the values arrive already labeled.

The second variation drills retrieval resistance. From a passage you have transcribed with roles bound, write a comprehension question for one role, then list every other number in the passage as a decoy and articulate, for each, why it is not the answer — "3.8 is the prior figure, not the current"; "4.5 is the target, not the actual"; "4.0 is the analyst consensus, external, not ours." Naming why each near-miss is wrong builds the reflex to reject on role mismatch, which is faster and more reliable than trying to remember which number was correct.

What band-9 numeric listening actually looks like

The band-9 listener does not have a better memory for numbers than the band-7 listener. What the band-9 listener has is a discipline that never stores a number without its role, so the memory load is not "hold five values" but "hold five labeled values," and the label is what makes retrieval unambiguous. The near-miss decoy, which defeats bare-value recall by being numerically and acoustically close to the answer, has no effect on role-first retrieval, because it is filed under a role the question is not asking about.

The financial summary rewards this discipline more than any other listening segment because it front-loads several same-type figures into a short window, which is the exact condition under which bindings decay and near-miss decoys dominate. A listener who binds at utterance, tags the transition words that carry the roles, and retrieves role-first converts the decoy cluster from a trap into a set of clearly labeled values, only one of which the question wants — and that one is retrieved without a scan and without competition.