How to Mark a Shadowing Transcript for Stress, Pauses and Linking
Turn a plain transcript into a simple shadowing map with three marks: phrase boundaries, strongest prominence and useful word-to-word links.
Use three marks and add them in three listening passes: / for a phrase boundary you actually hear, CAPS for the strongest prominence in that phrase, and ‿ for a useful connection across words you want to reproduce.
If your shadowing transcript needs a legend longer than the sentence, the transcript has won. Three marks, three jobs: Boundary first. Beat second. Bridge third.
Before you mark anything, listen once with no pen
That first pass has one job: hear the line as a performance before your eyes start telling your ears what “should” happen.
British Council pronunciation guidance recommends comparing a model for features such as rhythm, pace, stress and intonation when practising imitation. That is the right order here too: the audio is the evidence; the transcript is only where you record what you noticed. See the British Council pronunciation guidance.
Do not mark a word as prominent because it looks important. Do not insert a slash because there is a comma. Do not draw a link because English can theoretically link there. First hear this speaker, in this sentence, in this recording.
Pass 1: mark the boundaries first
Start with the largest structure. Where does the voice seem to finish one small unit and begin the next?
Pronunciation resources describe thought groups as coherent stretches that can be phrases, clauses or short sentences, often signaled by pauses, prominence and intonation. They also make an important point: there is not one compulsory division for every utterance. See Iowa State University’s thought-group overview and the University of Kansas pronunciation chapter.
So a slash means: I hear a useful boundary here. It does not mean: there is a comma here, therefore English demands a pause.
Original practice example
If you have a minute, could you look it over before lunch?
One possible model reading could be marked If you have a MINUTE / could you LOOK‿it‿OVER before LUNCH? The slash separates a condition from the request. Look it over means review it; with the pronoun object in this separable phrasal verb, look it over is the natural order for this meaning. The exact boundary and prominence must still come from your chosen audio.
Use the sentence to make a polite work or school request when you want someone to review something if they have time.
Listen once without the transcript. Add the slash only if you hear the condition and request as separate groups. Then replay and test whether your slash helps you follow the speaker rather than forcing an artificial pause.
Explore more language-learning guides in Media-Based Language Learning.
Pass 2: mark the strongest beat inside each phrase
Now replay the line and ask: Which word carries the strongest prominence in this phrase?
For this article, CAPS means sentence-level prominence: the word your ear experiences as the strongest beat in that phrase. It does not mean every syllable is louder, and it is not the same thing as dictionary word stress.
That distinction matters. Dictionaries can mark which syllable is stressed inside a word—for example, where the main lexical stress falls. British Council dictionary guidance explicitly teaches learners to read those word-stress marks. Sentence prominence is a different question: which word is made prominent in this particular message and context? See the British Council dictionary guidance.
Original practice example
I didn't mean to interrupt.
A possible reading might be marked I didn't MEAN‿to interrupt. Here, mean to + verb expresses intention. “I didn't mean interrupt” is wrong in standard English; to is required before the verb. Do not capitalise every content word automatically—mark the strongest prominence you actually hear in your model.
Use this to apologise when an interruption or action was not intentional.
Replay your model and identify one strongest beat. Then transfer the pattern to “I didn't mean to sound rude,” noticing that a different speaker or context may put the main prominence somewhere else.
If you have capitalised half the sentence, the mark has stopped doing its job. CAPS should guide your eye toward a useful beat, not turn the transcript into a ransom note.
Pass 3: add only the links you can actually hear
Only after the phrasing and prominence are clear should you mark connections across word boundaries.
The ‿ symbol here simply means: these words travel together in the model in a way I want to reproduce. It does not claim one specific phonetic process. Sometimes the connection involves a consonant moving smoothly into a following vowel; sometimes it is just a noticeably continuous transition. If you cannot hear it, do not decorate the page with it.
Original practice example
Could you send it by Friday?
One possible learner map is Could you SEND‿it / by FRIDAY? By Friday gives a deadline: no later than Friday. Until Friday is grammatically valid in duration contexts but is unusual for this one-time deadline meaning. The connection in send‿it should be marked only if that smooth transition is actually useful and audible in your model. The slash is illustrative too; many speakers may deliver the whole request as one flowing group.
Use this for a polite deadline request at work or school.
Listen specifically to “send it.” If the words flow together in the model and you tend to separate them, add ‿. If the mark gives you no useful information, remove it.
Original practice example
You don't have to decide right now.
A possible contextual reading is You DON'T have‿to decide / right NOW. Don't have to means there is no necessity. Mustn't is also grammatical, but it usually expresses prohibition, so the two are not interchangeable. If the context contrasts now with later, NOW may carry strong prominence; another context may emphasise a different word.
Use this to reassure someone that an immediate decision is unnecessary.
Mark the negative phrase and any audible connection only after listening. Then change the context—imagine you are contrasting TODAY with tomorrow—and notice whether the strongest prominence would shift.
Watch one sentence become a performance map
Take this plain transcript:
If you have a minute, could you look it over before lunch?
Do not attack it with every symbol at once. Build it in layers.
- Plain transcript: If you have a minute, could you look it over before lunch?
- Boundary pass: If you have a minute / could you look it over before lunch?
- Beat pass: If you have a MINUTE / could you LOOK it OVER before LUNCH?
- Bridge pass: If you have a MINUTE / could you LOOK‿it‿OVER before LUNCH?
That final line is not “the official pronunciation.” It is an example of what a learner map might look like for one plausible performance. Your actual model is allowed to disagree with it. In fact, if your audio disagrees, your audio wins.
Is this mark helping—or just decorating the transcript?
Check the mark against the audio
I added a slash because there is punctuation, but I cannot hear a reset.
Remove it. Punctuation can help organise written language, but it is not proof of a spoken phrase boundary.
I capitalised four or five words in one short phrase.
Simplify. Replay the phrase and mark only the strongest prominence you need as a visual cue. If several words matter, you can hear them without turning all of them into equal visual alarms.
I marked a link because English can link there, but I do not really hear it from this speaker.
Delete the link. Your transcript should describe this performance, not list every connected-speech process English might permit.
The marks make me stare at the page instead of listening.
Remove the least useful layer first. Your performance map has become heavier than the journey.
I can hear the boundary, beat or connection without looking now.
Fade the mark. Temporary visual support has done its job.
Shadow from the map—then start deleting the map
The annotation is not the final product. Your listening is.
Once the marked line helps you hear the pattern, try the same sentence with one layer removed. You might erase the linking marks first, then the CAPS, while keeping one slash that still helps. Or you may find the boundary is obvious but one difficult link still needs a cue.
Mark less, hear more
A good shadowing transcript is not the most detailed transcript. It is the one that directs your attention to the performance without competing with it.
Boundary first. Beat second. Bridge third. Describe the speaker you actually hear, not English in theory. And when the marks stop teaching your ears anything new, delete the stage directions.
The transcript is a temporary performance map. The audio stays in charge.