A learner can hear every word in “Let me turn to the second reason” and still miss what the phrase does. The speaker has closed one section, opened another and announced its place in the argument. If the listener treats those eight words as ordinary detail, the next minute arrives without a map.
For a first-year university student preparing for an English-medium lecture, discourse signposts deserve direct listening practice. The goal is to recognise functions such as topic shift, example, contrast and summary quickly enough to organise the content that follows.
Large signals carry the outline
Craig Chaudron and Jack Richards created four versions of an American history lecture: a baseline, one with small local markers, one with markers of overall organisation, and one containing both. They assigned the versions to classes of pre-university and university ESL learners and used three comprehension measures. Their 1986 study found better recall with macro markers than with micro markers. Macro markers name the lecture's structure. Phrases such as “There are two causes”, “The next point concerns housing” and “To return to the main argument” tell the listener where an idea belongs. Small markers may link adjacent sentences or fill a turn without revealing the larger route.
Its outline lives in the signal.
The study is old, and its lecture was constructed for research. Later findings have varied. The useful classroom claim stays modest: learners can be taught to notice explicit organisational language, then use it as one source of evidence about structure.
Signals help at more than one level
Euen Hyuk Jung tested 80 Korean learners of English. Half heard an academic lecture with discourse signals, and half heard a version without them. Within each condition, learners completed either a summary or recall task. The signalled group recalled more high-level and low-level information accurately. The 2003 paper also notes that earlier studies had produced inconsistent support, so one experiment cannot settle every genre and proficiency level. Signals do some of their work by setting an expectation. “One consequence appears in hiring” prepares the listener for an effect and a new subtopic. “For instance” changes the status of the next material: the example supports a prior claim. “To qualify that point” warns that the speaker will narrow or limit it.
Prediction starts here.
Short words can carry a long span.
Teach functions before inventories
A list of 40 linking expressions creates another vocabulary test. Start with four functions that the learner needs in lectures or workplace presentations: opening a section, adding a point, giving an example and closing a section. Use several phrases for each function and let the learner sort them after hearing them. For opening a section, recordings might contain “The first issue is”, “I want to begin with” and “Let's look at cost”. Example signals could include “for example”, “take Taipei as a case” and “one illustration comes from the 2024 survey”. The phrases differ in wording while sharing a job.
Four functions will do.
Ask for the function first. Meaning details come on a second listen. This keeps the initial task observable: the learner marks S for a new section, E for an example, C for contrast and R for a return to an earlier point. Four labels are enough for a 3-minute recording.
Our article on preserving the route through a talk in listening notes shows how arrows, indentation and headings can keep that structure on paper. Signpost training supplies the moments when the note layout should change.
Sound can hide a familiar phrase
Knowing on the other hand on a worksheet does not guarantee recognition in speech. The phrase may be compressed, unstressed and attached to the next clause, especially when a speaker is watching slides, turning toward the audience and treating the transition as routine rather than new content. A listener who waits for four fully articulated words may hear only fragments. Use short clips with natural timing. Play the signal in context, ask learners to mark its function, then replay a 7-second window and reveal the transcript. A final replay without text checks whether the sound now triggers the function. This sequence joins bottom-up recognition with discourse-level interpretation.
Spoken phrases shrink.
Jeanette DeCarrico and James Nattinger analysed lexical phrases in academic lectures and grouped them by the macro functions they perform. Their 1988 article proposes teaching such phrases to help learners predict incoming information and organise the flow. It offers a descriptive and pedagogical account rather than a controlled outcome trial, a distinction worth keeping.
Where signposts fail
Some speakers announce an outline and abandon it. Conversation contains false starts, delayed examples and topic changes with no tidy phrase. Skilled listeners also use prosody, repetition, prior knowledge and the content itself. A lesson built only from clean “first, second, finally” recordings can make spontaneous speech look much more orderly than it is. Highly proficient listeners may already recognise common signals automatically. They could gain more from implicit cues such as a pitch reset, a longer pause or repeated noun phrase. At A2, a dense university lecture remains difficult even when every section has a label. Vocabulary and sentence processing still set a ceiling.
Speech repairs itself.
I prefer one imperfect recording after two clean ones. Let the speaker say “The third... actually, before that, one more problem with staffing.” Ask learners to revise the outline. Real listening sometimes requires repairing the map.
A seventeen-minute lesson
Select a 3-minute explanation with six audible signposts. On the first listen, learners write only the function labels. They compare in pairs for 90 seconds and identify the exact phrases on the second listen. Show the transcript, underline the signals and ask what content each one governs. For the third listen, learners build a four-line outline with no full sentences. Then remove the recording and give them 2 minutes to explain the route: opening claim, two supporting points, one example and the final qualification. Their wording can differ from the speaker's.
Then change the voice.
End with a transfer clip from a different voice and topic. Include one clear example marker and one topic shift without a verbal marker. The learner should catch the first, then use the pause and content change to infer the second. That blank line in the notes records a real listening decision.