A learner can understand every sentence on a Teams call and still sound absent. Camera delay hides a nod, silence stretches for four seconds, and the speaker begins explaining the same point again. For a 34-year-old Taipei account manager working with colleagues in Singapore and Germany, listening needs an audible surface. Small responses such as mm-hm, right, oh, and really help the speaker judge attention, understanding and stance. They belong in speaking lessons because conversation is jointly managed.
A response can keep the turn open
Linguists often call these brief listener contributions backchannels or response tokens. They arrive during another person's turn without taking control of the floor.
Their size is deceptive.
A quiet mm-hm may invite the speaker to continue. Oh can register newly received information. Really can display surprise and invite elaboration. Right may acknowledge understanding, and in another context it can sound like agreement. Timing, pitch and the next turn decide much of the meaning. Flat bilingual lists handle these tokens poorly. Translating right as 對 and drilling it alone misses the interaction. Learners need to hear what came before, decide whether the speaker is finished, and notice what happens after the response. Placement decides. The same syllable can support a long explanation or interrupt it.
Anne Kari Bjørge analysed 13 video-recorded simulated negotiations involving 51 students from 16 nationalities, with 3 hours and 46 minutes of talk. In the 2010 English for Specific Purposes study, head nods were the most frequent backchannel, while verbal responses relied heavily on yes, yeah, mhm, and okay. Behaviour also varied with the level of conflict. Participants could give or withhold support as part of the negotiation. A nod or yeah signalled participation in context. Neither form carried a fixed promise of agreement. Because the recordings preserved gesture, speech and the surrounding negotiation, the analysis could distinguish a listener's quiet support from the same word used to seize a turn or strengthen the speaker's own claim.
That distinction matters on a work call.
“Yeah” during a status update may mean “I am following.” A manager who hears it as approval could close a decision too early. The repair is specific: follow an important proposal with a full confirmation, such as “So we agree to ship on the 27th.”
Listener habits cross languages
Response timing is culturally learned. Chien-Hui Hsu examined online conversations from 20 participants, ten Taiwan Mandarin speakers and ten American English speakers. The design removed visible gestures and facial expressions so that verbal responses could be compared. The 2024 University of Mississippi thesis focused on cross-linguistic differences and forms that could obstruct comprehension or trigger misunderstanding. Twenty people form a small sample, and a thesis carries less accumulated scrutiny than a mature research programme. Its direct comparison still gives Taiwanese teachers a reason to observe response tokens instead of treating listening as silent comprehension.
Video calls complicate the picture because gestures remain visible, though delay and window layout can weaken them. Francisco Javier Fernández Polo analysed one Spanish-German English-as-a-lingua-franca video conversation from ViMELF. The 2021 Journal of English as a Lingua Franca case study found intense backchanneling around openings and closings, plus heavy reliance on a few weak forms. Look at the sequence, not the token. A response that seems polite in a phrasebook can feel late, excessive or disruptive once it lands inside overlapping video speech, beside a frozen face, after a delayed nod and before the speaker has completed the clause that makes the point intelligible. Some responses arrived in awkward positions, felt excessive, or interrupted the flow. One dyad cannot establish a frequency norm. Timing wins. The case can expose placement problems that a vocabulary quiz will never catch.
Teach four functions in context
One short lesson can separate functions before adding variety.
Use an 87-second clip with two speakers and ask the learner to mark each listener response. Then replay the surrounding five seconds and classify what the response appears to do. Four useful starting categories are continuation, receipt, evaluation and clarification.
- Mm-hm with level or gently rising pitch often leaves the speaker's turn open.
- Right can show receipt of a point. The next sentence reveals whether agreement follows.
- Oh commonly marks a change in the listener's knowledge state.
- Really? displays a reaction and often asks for more detail.
These are tendencies rather than decoding rules. Speakers, regions and relationships vary. Some people use yeah frequently as a continuer. Others reserve it for firmer alignment. A teacher should present several voices and allow learners to compare effects. The aim is a wider set of choices and sharper timing, not imitation of one imagined native norm. This work fits naturally beside pair work with shared control. A pair task can assign one learner a story and the other a listener goal. The listener receives no content questions at first. Their job is to keep the account moving, show one change of knowledge, and ask for clarification once. Swap roles after 71 seconds.
Record the listener
Most speaking recordings preserve only the person holding the floor.
Record both sides instead.
On playback, mute the main speaker for one pass and listen to the response track. The learner can hear whether every token is yeah, whether a response lands on top of a stressed word, and whether silence follows a point that plainly invites uptake. Keep the speaker audible on a second pass. On a noisy call with six windows open, an acknowledgement that arrives half a second too early may cover the customer name, currency or revised date, leaving both people with the pleasant feeling of an active exchange and two different records of what was agreed.
Timing becomes visible in a transcript:
A: We moved the launch to the twenty-seventh... B: Mm-hm. A: ...because legal needs another review. B: Oh, right. Does that change the campaign booking?
The first response lets the reason continue. The second registers the new explanation and then claims a turn with a relevant question. Moving mm-hm two seconds later would cut into “because legal”. Replacing the final question with five more acknowledgements would leave the practical issue untouched. I prefer this two-line transcript to a page headed “active listening phrases.” It shows placement, consequence and a real decision point. When the transcript includes the words immediately before and after each response, teacher and learner can discuss observable timing rather than arguing from a vague impression that one person seemed interested or quiet.
Keep agreement explicit
Listener responses support rapport, yet their ambiguity creates risk when money, dates or responsibility are involved. Close those moments with a full proposition. “Right, I understand the option” reports comprehension. “Yes, I approve the revised budget” records a decision. A backchannel can carry the conversation towards that statement. It should never carry the decision alone. The next Teams call offers a narrow practice target. Choose one 90-second update. Use mm-hm once to keep the turn open, oh once when information changes, and one full clarification question. After the call, write down where each response landed, including the word that came immediately after it.