Pronunciation Is Shared Work: What Speakers and Listeners Each Control

Pronunciation is shared work: see what speakers can change, what listeners can do, and how both sides can repair a missed word without blame.

Chloe Hart

FunFluen editor · Vocabulary and learning

Aug 31, 2026 9 min

The short answer

Pronunciation-related understanding is jointly produced: speakers can shape and repair the signal, listeners can attend, adapt, request and confirm repair, and both can fix the smallest failed unit without assigning a blame score.

“Just pronounce more clearly.” “Just get used to accents.” Lovely slogans. Terrible repair strategies. Neither tells you what to do with the one name, number, or phrase that just failed.

Use two control panels and one shared loop: SPEAKER PANEL → LISTENER PANEL → SHARED REPAIR. The speaker shapes the signal; the listener completes the understanding. That does not mean every breakdown is 50/50: neither person controls noise, technology, the other person’s processing, or every social and institutional condition.

Shared work is not shared blame

A 2025 System review argues that pronunciation research and teaching can put too much responsibility on L2 speakers and miss the mutual nature of communication. Its authors recommend paying more attention to how interlocutors successfully repair difficulty and recognizing responsibility for intelligibility across the interaction, rather than assuming one prescribed pronunciation norm should carry the whole job. Read the 2025 review on mutual responsibility and intelligibility.

That does not mean “the speaker never needs to change.” Speech properties still matter. Nor does it mean “the listener is at fault if they do not understand.” Listener experience and adaptation have limits.

Shared work means both people have useful controls. It does not give us a responsibility percentage.

Control panel 1: what the speaker can do

The speaker controls the signal they produce—not the listener’s brain. Useful speaker-side actions include:

Make key information prominent: stress or highlight the name, number, contrast, or decision that carries the message.
Phrase the message: break dense information into manageable chunks rather than delivering one breathless ribbon.
Adjust pace when the task needs it: especially for unfamiliar names, numbers, codes, or instructions.
Repair one local feature: a repeatedly misheard sound, word, ending, stress pattern, or phrase.
Repeat or rephrase: saying the same unclear string louder is not the only repair available.
Confirm high-value payload: “That’s fifteen—one-five,” or “The surname is Nouri, N-O-U-R-I,” when confirmation is appropriate.

Speaker practice is not pointless just because communication is shared. Saito and Plonsky’s meta-analysis of 77 pronunciation-teaching studies found that instructional outcomes depend on what is being measured and whether speech is controlled or spontaneous. In other words, pronunciation can improve through instruction, but a drill-room gain is not automatically the same thing as effortless transfer in real conversation. See the pronunciation-teaching meta-analysis.

So practise what repeatedly matters. You do not need to rebuild an entire accent because one project code keeps disappearing.

Control panel 2: what the listener can do

The listener cannot manufacture a clearer signal after the fact, but they are not a passive microphone either. Useful listener-side actions include:

Attend to the message before judging the accent: noticeable accentedness is not the same thing as failed understanding.
Use context: narrow the plausible meaning before deciding the whole utterance is lost.
Ask for a local repair: “I missed the number after ‘room’,” is more useful than “I can’t understand your accent.”
Ask for a rephrase when repetition fails: a new wording can succeed where a second identical attempt does not.
Confirm the payload: “Did you say fifteen?” closes the loop instead of leaving a guess untested.
Allow adaptation: repeated exposure to an unfamiliar pronunciation pattern can sometimes make later listening easier.
Check unfamiliarity and bias: difficulty may involve limited exposure or expectations about who “sounds clear,” not only the speaker’s pronunciation.

ASHA’s current accent-modification guidance explicitly treats communication as affected by both speaker and listener characteristics. It describes breakdown-and-repair strategies for speakers and also a listener-training role, noting that listener beliefs and behaviors can affect judgments and interaction. See ASHA’s Accent Modification guidance.

Listener adaptation is real—but it is not magic

A useful listener role is adaptation: your perceptual system can become more familiar with a pronunciation pattern through exposure.

In a 2025 Foreign Language Annals experiment, L2 listeners repeatedly transcribed unfamiliar L2-accented sentences over two days. Adaptation varied with speaker intelligibility and input variability, and the tested multiple-speaker condition produced the largest improvement. Read the listener-adaptation study.

That is evidence that listeners can learn. It is not a command to understand every unfamiliar speaker instantly. The study used particular listeners, speakers, tasks, and controlled exposure. “Listeners can adapt” and “this listener should already understand me” are not the same claim.

The speaker’s signal still matters

Balanced does not mean pretending speech itself is irrelevant.

Munro, Derwing and Morton had Cantonese-, Japanese-, Mandarin- and English-background listeners evaluate the same set of foreign-accented English utterances from 48 speakers. The listener groups broadly agreed on which speakers were easier and harder to understand, although some between-group differences appeared. The authors concluded that properties of the speech itself were a potent factor in perception even across diverse listener backgrounds. Read the mutual-intelligibility study.

That is the balance in one paragraph: the signal matters, and the listener matters.

A tiny distinction that prevents a lot of blame

You only need three terms here:

Accentedness: how different speech sounds from a listener’s reference or expected variety.
Comprehensibility: how easy or difficult the speech feels to understand.
Intelligibility: how much of the message was actually understood.

They can move together, but they are not the same. A listener noticing a strong accent has not yet proved a message was misunderstood. A speaker sounding close to a familiar model has not guaranteed that a critical number came through.

For this page, that distinction has one job: repair what actually failed instead of treating accentedness as a diagnosis.

The shared repair loop: fix the smallest failed unit

When something breaks, try this:

LOCALIZE: What exactly was missed—the number, name, word, phrase, or instruction?
REPAIR: The speaker repeats, slows, highlights, spells, or rephrases that unit.
CONFIRM: The listener says what they understood.
CONTINUE: Once the payload is correct, stop auditing the accent and move on.

Example:

Listener: “I missed the number after ‘room.’”
Speaker: “Room fifteen—one-five.”
Listener: “Fifteen. Got it.”

Treat that as an illustrative repair pattern, not a universal script; wording should fit the relationship, task, and stakes.

That three-turn repair is usually more useful than a five-minute debate about whether the speaker “has a difficult accent.”

Who can act next? Try the scenario before opening the answer

You hear fifty, but the speaker meant fifteen. Who can act next?

Both. The speaker can make the number more distinctive or add digits; the listener can confirm what they heard. The shared finish is explicit confirmation: “Fifteen—one-five.” This is a local intelligibility repair, not evidence that the entire accent is unclear.

A new colleague uses an unfamiliar regional or nonnative English variety. Most of the message is clear, but you need extra effort. Who can act next?

Mostly listener-side adaptation now, with shared repair when something actually fails. Keep listening, use context, ask locally when needed, and allow familiarity to build. The speaker can still repair specific missed items. Do not demand a global accent change merely because the variety is new to you.

A presenter rushes through dense information and the key contrast disappears. Who can act next?

Speaker-side first. The speaker can slow locally, chunk the message, and make the contrast prominent. The listener can ask which item was being contrasted. The practical target is presentation delivery, not the speaker’s identity or whole accent.

The call is breaking up and both people keep mishearing each other. Who can act next?

Fix the channel/context first. Move rooms, change connection, call back, use text for the critical payload, or otherwise improve the channel. Do not diagnose an accent problem from degraded audio. Sometimes the Wi-Fi is the third party everyone is unfairly prosecuting.

A surname is repeatedly misheard across several conversations. Who can act next?

Both. The speaker can choose a clearer local production, stress cue, spelling, or rephrasing strategy; the listener can repeat the name back and learn the recurring pattern. Repetition across events gives you a real local target—still not a verdict on the whole accent.

A listener misses one phrase and says, “I can’t understand your accent.” What is the next move?

Listener-side repair language should get smaller. “I missed the last phrase” or “Could you say that another way?” identifies the failure without claiming the whole accent is unintelligible. The original sentence is grammatically valid, but after one missed phrase it overgeneralizes the evidence.

A speaker is easy to follow for familiar colleagues, but a new listener has very little exposure to the variety. Who can act next?

Both, but do not assume equal cause. The listener can build familiarity and request local repair; the speaker can choose extra confirmation for important payload while the relationship is new. The pattern may change as familiarity grows.

The speaker mumbles one key project code, although the rest is clear. Who can act next?

Speaker-side local repair first. Repeat the code more distinctly, perhaps chunk or spell it; the listener confirms. Do not turn one mumbled item into “your accent is the problem.”

Your no-score repair sheet

Use this after any ordinary pronunciation-related breakdown. There is no blame percentage.

What failed?	Speaker can try	Listener can try	Shared confirmation
Name	Stress clearly; repeat; spell if useful	Ask for the name only; repeat it back	“Yes, that’s ___.”
Number/code	Chunk; add digits; slow locally	Say what you heard; ask for confirmation	Repeat the final number/code once
One word	Repeat or rephrase the sentence	Name where the gap occurred	Confirm the recovered meaning
Dense phrase	Phrase/chunk; highlight the key word	Ask for the key contrast or rephrase	Restate the decision/point
Unfamiliar variety	Repair only real missed items	Use context, exposure and local clarification	Confirm high-value payload
Bad audio	Change channel if possible	Report the audio problem specifically	Move critical details to a clearer channel

Longer-term work is different from immediate repair

After the conversation, either side may decide there is a recurring pattern worth practising.

Speaker-side long-term work

Choose a target that repeatedly affects your actual communication: a high-value word, segment contrast, prominence pattern, pacing habit, or phrasing issue. Practise it in controlled speech, then test it in more spontaneous use. Do not assume one perfect drill means transfer is finished.

Listener-side long-term work

Broaden listening exposure across speakers and varieties. Repeated exposure may improve adaptation under some conditions, but do not turn this into “I must understand everyone immediately.” Familiarity is a skill, not a moral grade.

Shared work has limits in high-stakes settings

Ordinary conversational repair is not a substitute for institutional duties or formal safety procedures. In exams, healthcare, aviation, legal settings, customer-service systems, accessibility contexts, or other high-stakes environments, responsibilities may be asymmetric and specific protocols may apply.

Use the appropriate professional standards, qualified language support, safety procedures, or accessibility arrangements where required. “Communication is shared” is not permission to improvise away those obligations.

If you choose a speaker-side target to practise

Once the immediate repair is done, you may discover one recurring piece of the signal that you want to work on. FunFluen can support general English speaking practice for that chosen target. It does not assign fault, measure listener bias, or decide why a conversation failed.

Practise the part of the signal you can control

Repair the message, not the blame

The speaker can shape the signal. The listener can work to receive it. Both can identify the smallest failed unit and repair it together. Sometimes the signal is the bigger problem. Sometimes unfamiliarity or bias is. Sometimes the room, microphone, connection, or institutional setup matters more than either person.

You do not need a verdict before you can act.

Two control panels. One conversation. Localize, repair, confirm, continue.

Sources

Explore more language-learning guides in Media-Based Language Learning.