You read the transcript and understand almost everything. Play the recording again, and the same sentence sounds obvious. Yet a minute earlier, without the text, you could barely catch it.

Knowing a word in writing does not guarantee that you can recognise it quickly in speech. You also have to match its spoken form, find where it begins and ends, and keep up with the message. If the transcript is easy but the recording is hard, practising those connections is a better starting point than automatically adding more vocabulary.

What the transcript reveals

Try a short recording with a reliable transcript in the language you are learning. Listen before reading. Then compare the words you caught with what was actually said.

The result can point you towards different problems:

  • The written words are unfamiliar too. Vocabulary is part of the difficulty. Look up the important gaps or choose easier material.
  • The words look familiar, but you missed them in the recording. Work on recognising their sounds in that phrase.
  • You caught the words but lost the meaning. The sentence structure, an expression, or the effort of keeping up may need attention.

These are clues, not a formal diagnostic test. Several problems can happen in the same sentence. The useful question is more precise than “Why am I bad at listening?” It is “Which part did I lose?”

That distinction has a research basis. In a 2024 study by Aoiz Pinillos, 209 adult Spanish-speaking learners of English showed smaller aural than written vocabularies on the tests used. This does not establish the same gap for everyone, but it shows why a written vocabulary test cannot tell the whole story.

Speech does not arrive with spaces

On a page, you can stop, reread, and inspect the written forms. In languages that use spaces, word boundaries are visible too. A recording usually does not put a pause between each word. You have to work out the boundaries from the sound, rhythm, surrounding words, and meaning.

Consider this ordinary English question:

Did you eat yet?

You might know all four words and expect four clear units. In some casual English speech, “did you” can sound roughly like “didja.” That spelling is only an illustration, not a universal pronunciation. Different speakers will say the question differently.

If you are waiting for a careful dictionary pronunciation of “did,” then a separate “you,” you can miss a phrase whose meaning you already know. The useful practice unit is often the phrase, including the words around the one you missed.

Lange and Matthews (2020) found that recognising spoken vocabulary and locating words in connected speech were both related to listening performance among 130 Japanese learners of English. The study measured relationships, not the effects of a training routine.

Casual pronunciation adds another difficulty. Verbeke and colleagues (2025) found that Dutch-speaking and Spanish-speaking English learners recognised reduced word forms less accurately and more slowly than unreduced forms in their experiments. This is one explanation for missed words, not the explanation for every listening problem.

Use text to discover what you missed

A transcript gives you something specific to compare with the recording. You can find the small word that disappeared, the phrase you divided incorrectly, or the familiar word whose pronunciation surprised you.

Research offers a reason to use that support. Mitterer and McQueen (2009) found that English subtitles helped proficient Dutch learners adapt to unfamiliar English accents, measured in a later audio-only repetition task. That narrow result supports trying matching written help; it does not prove that every subtitle method works for every learner.

For this exercise, use text in the language being spoken. A translation can clarify meaning, but it cannot show exactly which target-language words made up the sound you missed.

After checking the text, listen without it. Following a sentence with your eyes and recognising it from sound alone are different checks. Neither is cheating. They simply tell you different things.

A ten-minute exercise with one short clip

This is a suggested practice routine, not a research-validated dose. Ten minutes is a convenient limit for keeping the work focused.

Choose roughly 20 to 40 seconds of clear speech with an accurate transcript. A learner podcast, a course recording, or a short interview can work. Pick a passage whose written version you can mostly understand. If choosing that material is the hard part, start with our guide to comprehensible input.

  1. Listen for the message. Play it once without text. Write one sentence about what you think happened or what the speaker meant. Leave gaps instead of inventing an answer.
  2. Locate one missed phrase. Listen again and note a few words or a timestamp. Choose a manageable problem rather than trying to transcribe the whole clip perfectly.
  3. Check the transcript. Find the exact phrase. Was a word genuinely new, or did you know it in writing? Check the meaning now if necessary.
  4. Replay the phrase in context. Keep a little speech before and after it. Follow the text once, paying attention to how the words connect. Slow the audio modestly if that helps, then return to its original speed.
  5. Hide the text. Replay the phrase and then the whole clip. Say what the message means. Notice whether you actually recognise the sound or are mainly remembering the sentence.
  6. Try a fresh sample. Later, use another short passage from the same speaker. This gives you a better check than endlessly improving on a recording you have memorised.

Keep one brief note: “I knew the words, but I missed…” followed by the phrase and what you noticed. One clearly understood difficulty is enough for a session.

What that looks like in practice

Imagine a learner who hears “Did you eat yet?” and catches only “eat.” They open the transcript and immediately understand the question.

Adding “did,” “you,” and “yet” to a new vocabulary list would not address this particular miss. The learner already recognises the words. Instead, they replay the whole question and compare its rhythm with the four words on the page.

They listen to the beginning as a connected phrase, then replay the question without looking. Finally, they explain its meaning in their own words.

Suppose the question is now clear. That is useful feedback about this recording. It is not proof that every similar question will be easy tomorrow. A fresh example from the same speaker, and eventually another speaker, is the next check.

If the learner still cannot hear the phrase after comparing it carefully, they can ask a teacher to point out the sounds or choose clearer audio. Repeating the same confusion indefinitely is not the goal.

How to use LingoUnify for this

In LingoUnify, you can play a saved Library text aloud with word highlighting. Choose a passage you already understand, follow the words during playback, and then replay it while looking away. Playback speed can be adjusted when you need a slower first pass.

You can also hear individual saved words. That helps you check a word's isolated pronunciation, but include the sentence when you practise the listening problem. A word on its own does not reproduce all the connections around it.

Your Learning or Learned status is a useful record of how you classify a word. It is not a test of whether you can recognise it in speech. A word marked Learned can still deserve listening practice.

This route uses synthetic speech. It is convenient for pairing text with clear audio, but natural conversations contain variation, interruptions, and reductions that a generated voice may not reproduce. Add recordings of real speakers too. Word highlighting only helps with sound when you actually turn playback on and listen.

Mistakes that make progress hard to judge

  • Choosing audio you cannot understand even in writing. You are tackling vocabulary, grammar, and sound at once. For this exercise, choose something easier.
  • Leaving the transcript visible for every replay. It becomes hard to tell what your ears can recognise independently. Include a pass without text.
  • Trusting every caption. Automatic captions can mishear words, and subtitles may shorten dialogue. A mismatch can be a text problem rather than a listening failure.
  • Only practising one voice or one memorised clip. Include fresh material so that success means more than remembering what comes next.
  • Expecting one exercise to solve conversation. Following another person also involves meaning, attention, turn-taking, and responding. This exercise works on one part of that larger task.

Start with the words you already have

Choose one short recording today. Listen, inspect the transcript, and find one familiar phrase you did not catch. Compare the text with the sound, then listen again without reading.

Keep the first attempt and try a new passage later. Look for a concrete change: a phrase you can now identify, fewer places where you lose the message, or less need to check the transcript.

Measure what you can understand in fresh audio. If you can read the words already, you have a useful starting point. Give those words some deliberate listening attention.