Part 4 — Short Talks: reading the format before you hear the answer

One speaker, ten talks, thirty questions

Part 4 drops the back-and-forth of Part 3 for a single continuous monologue: a voicemail, an announcement, an advertisement, a tour, a meeting excerpt, or a radio broadcast. Ten talks, three questions each, thirty questions total. The questions and answer choices are printed on screen exactly as in Part 3, so your job during the audio is confirmation, not memorization — read the three stems before the talk starts and let the recording fill in what you're already looking for. Because there's only one voice, the talk moves in a straighter line than a conversation does, which is exactly what makes format recognition such a reliable shortcut.

The format tells you what the questions will ask

• Voicemail: almost always ends with a callback request, a deadline, or both ('Please call me back at your earliest convenience') — expect a question about why the speaker is calling and one about what the listener should do next.
• Announcement (transit, workplace, safety): opens with the reason for the announcement, then gives a cause and an instruction — expect a 'main purpose' question, a cause/reason question, and a 'what should X do' question.
• Advertisement: leads with the product or event, stacks numbers (discounts, hours, quantities), and closes with a call to action — expect at least one question built entirely around a number.
• Tour talk / guide introduction: establishes location early, then sequences what happens next ('next we'll head to...', 'please remember to...') — expect a location/inference question and a 'what will happen next' question.
• Meeting excerpt / briefing: states the topic, lists two or three items to work around, and hands off responsibility at the end — expect a purpose question built from the sum of the items, not any single one of them.
• Broadcast (news, weather, traffic): states the subject in the first sentence, then adds detail — expect the detail questions to test whether you can find a fact quickly, not whether you understood the gist.

Purpose stated early, action item buried late

This site's own warehouse shift-change briefing is a clean example of the shape: 'Before you head out, let's do a quick shift-change briefing' announces the topic in the first sentence, and the payload — the closed east dock, the two broken pallet jacks, the unfinished aisle-twelve count — is delivered in the middle, with the actual instruction ('pass this information along to the next crew') arriving only in the last line. The client voicemail from Marcus Lee follows the same arc in miniature: the reason for the call ('I need to reschedule our meeting') comes in sentence two, and the action item ('Please call me back at your earliest convenience to confirm, or just send me an email') doesn't land until the final lines. If you tune out after the opening sentence because you think you've already got the purpose, you'll miss exactly the detail question that asks what the listener is supposed to do.

How 'main purpose' questions are built — and the trap inside them

Part 4 purpose questions are rarely answered by one sentence; they're answered by the sum of the talk. The wrong options are built by taking one true fragment from partway through and inflating it into the whole purpose. The shift-briefing question 'What is the main purpose of the talk?' is a textbook case: the correct answer, 'to pass on conditions the incoming crew will need to work around,' covers all three items raised (the count, the jacks, the dock). Each wrong option grabs just one of them and overstates it — equipment trouble becomes 'to report equipment failures to the maintenance department,' the unfinished count becomes 'to review the findings of a completed inventory audit' (note the count is explicitly incomplete, not reviewed), and the one-night dock closure becomes 'to introduce a new procedure for loading outbound trucks.' The tell is always the same: a wrong purpose option sounds true because a piece of it is true. Before you commit, ask whether the option covers everything the speaker said or just the one detail you happened to latch onto.

Inference questions: more common here than in Part 3

'What does the speaker imply about...' and 'What does the speaker mean when he says...' show up more often in Part 4 than in Part 3, because a single uninterrupted speaker has more room to leave a conclusion unstated. The pallet-jack question from the shift briefing shows how these get built: the talk never says outright that the jacks are unavailable all night, but it says maintenance won't have them 'back in service until tomorrow morning' and that the speaker is briefing the crew handing over to the night shift — put those two facts together and the implication (they'll be out for the entire night shift) follows even though no single sentence states it. The advertisement set does the same thing with the grand-opening promotion: the twenty percent discount runs 'from nine A.M. until closing' on 'your entire purchase' with no cap mentioned, while the free tote bags are explicitly capped at the first fifty customers — the trap answer borrows the fifty-customer limit and wrongly attaches it to the discount, because the two offers are announced back to back. Inference questions reward keeping separate facts separate, not blending whatever was mentioned near each other.

Numbers, dates, and times are favorite targets

• If a talk mentions more than one number, assume at least one question tests whether you attached the right number to the right thing, not whether you heard a number at all.
• Watch for numbers that get restated in a different form: 'twenty percent off your entire purchase' becomes an answer option worded as 'an additional discount' or 'a discount on selected items' — the trap swaps the scope, not the figure.
• Deadlines and time windows in voicemails and announcements ('within the next two hours', 'starting at ten p.m.') are almost always the answer to a 'what will happen' or 'by when' question — flag the moment you hear a time expression.
• When two numbers compete for the same question (fifty customers vs. an uncapped discount; a fifteen-minute delay vs. a twenty-minute single-track window), the correct answer is the one attached to the noun the question actually asks about — re-read the question stem, not just the numbers you jotted down.

Graphic-based talks

Some Part 4 talks pair with an on-screen graphic — a schedule, a floor plan, a price list, a program — the same way some Part 3 conversations do. The audio deliberately never states the answer directly; it gives you a name, a time, or a condition, and you have to cross-reference that against the graphic to find the one line it matches. Preview the graphic along with the three questions before the talk starts, and identify which column or row each question is likely to point to. The trap here isn't a sound-alike or a paraphrase — it's picking the row that seems most prominent on the graphic rather than the one the speaker actually pointed you to.

Pace: use the steadiness against the format

Because Part 4 is one speaker reading a prepared script rather than two people improvising a natural exchange, the pace is more even than Part 3's — no interruptions, no false starts, no overlapping turns. That steadiness is an advantage: once you've identified the format in the opening line, you can predict roughly where the purpose, the cause, and the action item will fall in the timeline and aim your attention accordingly, rather than tracking a shifting conversation between two voices. Don't mistake steady for slow, though — the words-per-minute is the same as Part 3. The gain is entirely in predictability, not in extra time.