Neither drill is universally better: use listen-and-write to expose what your ear missed, and listen-and-repeat to rehearse sound, timing, and output once the wording is clear enough.
Suppose the audio says, “I didn't say he stole it.” You write every word correctly. Useful—but the page still cannot show which word the speaker emphasized. Now you repeat the line with a plausible rhythm. Also useful—but that performance does not prove you consciously decoded every word. Same clip, two drills, two kinds of evidence.
What Each Drill Actually Trains
Think of listen-and-write as an X-ray and listen-and-repeat as a rehearsal. An X-ray gives you something visible to inspect. A rehearsal gives you a chance to reproduce what you heard. Asking which one is “best” without naming the problem is a little like asking whether an X-ray or a rehearsal is the superior life form. Different jobs.
In a basic repetition drill, you hear a model and repeat it. The British Council describes drilling this way and notes that pronunciation and intonation can be targets—but also that learners need to understand what they are being asked to say for the drill to be meaningful. Its guidance on drilling is useful here as a definition and teaching boundary, not as proof that repeating automatically improves listening or pronunciation.
Also, ordinary pause-and-repeat is not exactly the same as shadowing. In shadowing, the learner follows the audio near-simultaneously. Much of the research on sound-focused repetition studies shadowing, so we can borrow cautious clues about perception, timing, and spoken rehearsal—but we should not pretend every shadowing result automatically applies to a simple “pause, then repeat” routine. If you want the full method rather than this comparison, use the full English shadowing practice method.
| What you care about | Listen and write | Listen and repeat |
|---|---|---|
| Exact words decoded | Leaves a visible candidate transcript you can compare. | Can rehearse the line, but a convincing imitation does not by itself prove every word was decoded. |
| Word boundaries | Makes missing, merged, or extra words visible on the page. | Lets you reproduce chunks, but the output may not reveal exactly where your lexical uncertainty was. |
| Grammatical endings | Can expose that an ending was omitted from your written version. | Lets you practise saying an ending once you know it is there. |
| Spelling | Forces a spelling choice—which is useful, but can mix spelling knowledge with listening evidence. | Avoids spelling as a task demand. |
| Short sound sequence | Requires you to hold enough of the line to write it, but writing changes the task into an orthographic one. | Directly rehearses holding and reproducing a short heard sequence; research support here comes mainly from shadowing and related speaking tasks, so keep the claim modest. |
| Pronunciation and rhythm | Does not rehearse vocal timing by itself. | Gives you an immediate chance to imitate sounds, stress, and rhythm. |
| Meaning recall | Does not prove you can explain or retell the idea. | Does not prove you can explain or retell the idea. |
What Repeating Hides
Repeating feels satisfyingly active. You are making sound. Your mouth is busy. The replay button may become the most highly trained participant in the session. But spoken imitation gives you a different kind of evidence from a written transcript.
Here is a hypothetical example, not observed learner data: a learner hears “I didn't say he stole it,” repeats the overall cadence fairly well, and then cannot confidently say whether they decoded every small word. The repeat may still be useful rehearsal. What it does not give us is a visible record proving that each word was identified correctly.
This matters because familiarity can masquerade as clarity. After five repetitions, a line can feel less alien simply because you have heard its shape five times. If your real problem is “Which word vanished?” rehearsal may be happening before diagnosis.
A 20-second rehearsal check
After one repeat, ask yourself:
- Can I say the exact words I believe I heard?
- Am I copying the rhythm while one small word is still fuzzy?
- If the audio disappeared now, could I identify the uncertain part rather than just perform the whole sound shape again?
If those questions expose uncertainty about the wording, switch tools. That is where the X-ray earns its keep.
What Writing Exposes
Listen-and-write leaves evidence behind. A missing word is visible. A merged boundary is visible. An ending that never reached the page is visible. That makes writing especially useful when the problem is not “Can I imitate this?” but “What exactly did my ear give me?”
A 2021 System study used dictation responses from 60 undergraduate ESL learners in Hong Kong to classify 640 perceptual errors in connected English speech. That supports a narrow but valuable point: a written response can give researchers—and learners—an inspectable record of what was recovered from the signal. It does not mean every written error is a listening error. See the study on perceptual errors in connected speech.
Imagine the target sentence is “He worked late.” A hypothetical learner writes “He work late.” Classification: as the standard-English transcription of that past-tense target, “He work late” is wrong; the natural target is “He worked late.” In another context, a listener might guess that the learner meant the habitual present “He works late,” but that would still require works in standard English. Here, the learner’s likely intent is simply to reproduce “He worked late.” The important diagnostic point is that the mismatch does not tell us why it happened: the learner may have failed to hear or identify the ending, or may have perceived enough of the sound and then omitted -ed while writing. Replay has to settle that question.
| Target | My version | Possible issue | Next check |
|---|---|---|---|
| He worked late. | He work late. | Could be perception of the ending, writing/grammar during transcription, or both. | Replay only the end of worked; then say what you heard before writing it again. |
So do not punish every miss with the same diagnosis. Mark it heard differently, spelling uncertain, or not sure yet. Then replay the smallest part that can settle the question. That turns dictation from a red-pen ritual into actual debugging.
Choose by Symptom
You do not need to join Team Dictation or Team Repeat. Choose the output that makes today’s problem easier to see or rehearse.
Which symptom is actually yours?
I keep missing tiny words or grammatical endings.
Write first. One short transcription gives you visible evidence of what reached the page. Compare with a trustworthy transcript, then replay only the uncertain word or ending. If the mismatch is purely spelling, label it spelling rather than automatically calling it a listening failure.
Two or three words blur into one chunk.
Write first. Put down exactly how many words you think are there, then compare. Your goal is to locate the boundary problem before you rehearse the chunk aloud.
I know the word, but I’m not sure whether I heard it wrong or spelled it wrong.
Write first, then isolate the uncertainty. Say the word you believe you heard before looking at the spelling. If your spoken identification was right but the written form was wrong, that is different from failing to decode the word.
I know the exact words, but my timing or rhythm feels stiff.
Repeat first. The wording no longer needs an X-ray. Listen to one short chunk and imitate one delivery feature—perhaps stress placement, timing, or how the words connect. Do not try to “fix your whole accent” in one pass.
I can hear the line, but I lose the short sequence before I can say it.
Repeat a shorter chunk. Cut the line down until you can hold and reproduce it, then rebuild it. Research on shadowing and related speaking practice suggests links with phonological processing and short-term verbal handling, but that does not justify promising a general memory improvement from a simple repeat drill.
I can repeat it, but I cannot explain or retell the meaning.
Neither drill is the main fix. You have moved from decoding or imitation into meaning retrieval and speaking. Switch goals instead of adding ten more echoes.
If you land on “both,” the order is simple: X-ray first, rehearsal second.
Combine Them in One Session
Here is the same-input comparison, using one verified FunFluen-served clip. The transcript is “I didn't say he stole it.”
Clip provenance
This asset was generated on by Microsoft Azure AI Speech using the en-US-AvaMultilingualNeural voice and the audio-24khz-160kbitrate-mono-mp3 format. Rights status: founder-confirmed API voice-use rights.
| Use the clip this way | What you can inspect | What remains hidden |
|---|---|---|
| Listen and write | You can compare your written words with “I didn't say he stole it.” A missing didn't, he, or another item becomes visible. | A correct transcript does not record which word carried the strongest emphasis in the audio, and it does not rehearse your spoken timing. |
| Listen and repeat | You can imitate the line’s sound, timing, and emphasis as closely as you can hear them. | A smooth repeat does not leave a word-by-word record proving that every item was consciously decoded. |
One-line X-ray → rehearsal
If you already read the transcript above, treat this as a method demonstration rather than a blind test. On a future unfamiliar line, keep the transcript hidden until the comparison step.
- Play the clip once. Do not pause after every word.
- Write exactly what you believe you heard.
- Compare your version with the transcript: “I didn't say he stole it.”
- Mark one mismatch or uncertainty. If it is only spelling, call it spelling—not automatically a listening error.
- Replay the line and repeat it once, focusing on one delivery feature such as the emphasized word or overall rhythm.
- Now say one controlled variation without audio: “I didn't say she broke it.” That is a small production step, not proof of spontaneous-speaking mastery.
Notice what happened: the written pass gave you an inspectable record; the spoken pass gave you rehearsal. Same input, different evidence. When both problems are present, this order prevents you from polishing the rhythm of a sentence you still have not fully decoded.
If you practise with shows or videos, FunFluen can reduce the friction of moving line by line and replaying a difficult subtitle on supported video pages, so you can run the same listen → check → repeat loop without turning navigation into the exercise. Subtitle-based controls require available subtitles, and the tool supports deliberate practice rather than proving that you decoded a line correctly or guaranteeing listening or pronunciation gains. Review the FunFluen extension for short-line practice before installing.
What Neither Drill Trains
There is a trap at the end of a successful drill: assuming success on the drill means success at English. It does not.
You can copy a sentence perfectly and still be unable to explain it. You can repeat it beautifully and still freeze when someone asks a follow-up question. That is not evidence the drill “failed”; it means the goal changed.
Neither drill, by itself, guarantees that you have learned register, collocation, or what sounds natural in a particular situation. For example, “Could you give me a hand?” is common in neutral-to-casual everyday conversation. “Could you assist me?” is also grammatically valid and communicates the same basic request, but it tends to sound more formal and may feel less conversational in an ordinary friendly exchange. The best choice still depends on the speakers, setting, and surrounding wording. Repeating either line accurately does not teach you that context by itself.
Collocations create the same problem. English normally says “make a decision.” A learner may produce “do a decision.” Classification: for the intended meaning in ordinary standard English, “do a decision” is unusual and non-idiomatic. A listener would probably infer that the learner means “make a decision,” which is the natural alternative. The original wording can still appear deliberately in quotation, wordplay, or discussion of learner language; the correction here is about ordinary reuse. Hearing and repeating the non-idiomatic phrase accurately would simply make you more accurate at repeating the wrong collocation for that context.
If the third box is the real one, switch methods. The guide on retelling versus repeating for real speaking covers that boundary without pretending more imitation automatically creates spontaneous output.
The Evidence, Honestly
The evidence does not justify declaring a universal winner between ordinary listen-and-repeat and listen-and-write. The research bodies are related but different, and the strongest studies do not form a clean head-to-head trial of these two exact self-study drills.
What the dictation evidence actually says
- A 2002 TESL Canada Journal study compared two groups of 30 male elementary EFL learners in Tehran. The experimental group completed 11 dictations during a 20-session term in addition to textbook listening and showed larger listening-comprehension gain scores. That supports a bounded classroom claim, not “one dictation makes you a better listener.” Read the dictation study.
- A university study published online in 2019 and later in Computer Assisted Language Learning compared classes of 37 and 33 learners. The experimental condition used daily dictation through a WeChat group alongside the same listening course and decoding instruction. It improved more on the listening test, but the intervention bundled dictation with mobile/social delivery, so we cannot assign the whole effect to writing alone. Read the mobile-supported dictation study.
- The 2021 System study of 60 Hong Kong undergraduates used written dictation responses to classify 640 connected-speech perceptual errors. That makes dictation valuable as a diagnostic record, but a written error can still contain a spelling or orthographic component. Read the connected-speech error study.
What the repetition and shadowing evidence actually says
- A study of 43 Japanese university EFL learners completing nine shadowing-based lessons found phoneme-perception improvement in lower- and intermediate-proficiency groups, while improvement on the listening questions used appeared only in the lower-proficiency group. Useful evidence, but shadowing is near-simultaneous and the results were not uniform across outcomes. Read the shadowing study.
- A 2025 systematic review included 44 shadowing studies and reported promising findings for comprehensibility, intelligibility, accentedness, and some suprasegmental features such as fluency and prosody, while segmental-pronunciation findings were inconclusive and the authors highlighted methodological weaknesses. That supports cautious use of shadowing research for sound-focused rehearsal, not a promise about ordinary pause-and-repeat. Read the systematic review.
- A four-week study that began with 119 young adults compared time-compressed L2 listening, shadowing, reading aloud, and an active control. The key behavioral comparison combined the speaking groups for some analyses and found changes on working-memory measures alongside imaging findings. Because it does not isolate ordinary listen-and-repeat, it is a reason for cautious interest in phonological processing—not permission to advertise a memory boost from repeating one sentence. Read the working-memory and neural-systems study.
That is why the decision in this guide is practical rather than magical. If you need to expose missing words, boundaries, endings, or spelling-versus-hearing uncertainty, start with writing. If the wording is already clear enough and you need immediate sound-to-output rehearsal, start with repeating. If you need both, diagnose first and rehearse second.
For other drill types, see more English listening exercises by skill.
Writing is the X-ray. Repeating is the rehearsal. Stop looking for a universal champion and reach for the output that gives you the evidence you need from this line today.