Intensive vs extensive listening: how to learn a language from real videos

There are two ways to listen to a language you are learning. Intensive listening means working closely through a short piece of audio: you replay it, check what you understood, and look at the words you missed. Extensive listening means listening to a lot of material for meaning, at a level where you follow most of it without stopping. Both work, but for diffrent learners at different moments, and with real videos it is easy to get the balance wrong: hours of watching that is too hard to learn from, or slow study of videos you could simply enjoy.

This post explains the difference, shows how to tell which one a video needs from you, and then walks through both with real videos: intensive listening step by step, and extensive listening in a feed. The examples come from Reelang, where the two are called Focus Mode and Flow, and every step comes with the research behind it.

Intensive vs extensive listening at a glance

Which one a video needs depends on how much you understand

To follow speech well, you need to know around 95% of the words you hear, and about 90% is workable with video because the picture carries part of the meaning (van Zeeland & Schmitt, 2013). Below that line, the audio is mostly noise, and just watching does not teach you much. We wrote about where that line sits for each language in how many words you need to understand native videos.

So the same learner needs both. A slow cooking vlog may already be extensive listening for you, while a fast panel show in the same language still needs intensive work. A useful test: watch 30 seconds without subtitles. If you could retell what happened, watch it through. If you could not, work through it closely.

Intensive listening with real videos, step by step

The usual problem with intensive listening on real videos is subtitles. You press play, the speakers go too fast, you turn subtitles on, and you end up reading instead of listening. Focus Mode is how we built intensive listening into Reelang to avoid that. It takes a clip one sentence at a time and holds back help until you ask for it: first you listen, then you decide how much support you need, and the support takes a little effort to reach. You can copy most of these steps with any video and a replay button. The app does the bookkeeping for you.

All the examples use one simple Spanish sentence from a beach clip: Mañana vamos a la playa con mis primos. (Tomorrow we're going to the beach with my cousins.)

Try it first

This is the Focus panel as it appears in the app, under the clip. Press the buttons as a learner would. The box underneath lists what the app records from each tap.

Mañana vamos a la playa con mis primos.

The short version of what you just did:

1. Hear the sentence with no text.
2. Say how much you understood. Nobody grades this answer.
3. Read it, if you need to, in Spanish, with Smart Subtitles.
4. Break it down word by word, if you still need to. Some translations are blurred and you tap to see them.
5. A quick quiz on the word that gave you trouble.
6. Every ten sentences or so, the whole part plays again in one go, at full speed and without text.

Most sentences only need two of those steps. Here is each step and the research behind it.

Step 1: choose how to start each part

Mañana vamos a la playa con mis primos.

Focus works through a video in parts of about ten sentences. Each part opens with a choice. Watch this part in full first plays it straight through with no text and no stops, then asks if you understood it. If you did, you go to the next part. If not, you go sentence by sentence through what you just heard. Go sentence by sentence starts the loop right away.

Why we built it this way. Listening for the gist before working through the details is a standard sequence in listening teaching, and hearing the same audio twice helps by itself: a 2024 study found that a second play improved recall and comprehension and lowered anxiety (Holzknecht, 2024). Your answer to "did you get it?" also tells the app how this clip lands for you before it has given you any help.

New learners never see a screen explaining any of this. The second clip they watch opens straight in the loop for a single sentence, and only then asks: Keep going like this? By that point they have tried it once and know what they are choosing.

Step 2: a short warm-up (first time only)

Mañana vamos a la playa con mis primos.

The first time you go sentence by sentence in a clip, the app shows the two or three words the clip relies on that are probably above your level. It picks them by frequency rank. Playa is the 1,346th most common word in Spanish and primos is the 2,292nd, so both fall outside the 800 or so words a beginner is assumed to know. Tap the words you're unsure about to see what they mean, or skip with I know all these words. The whole thing takes about ten seconds.

Why we built it this way. If you already know the two words that carry a sentence, you can match much more of the sound to meaning on the first listen. Each tap also records whether you know a specific word at a specific rank, before the clip has even started.

Step 3: listen with no text

Mañana vamos a la playa con mis primos.

The sentence plays once, padded by half a second on each side so you never miss the first or last syllable. You get no subtitles at this stage, only the sound and the picture.

Why we built it this way. Subtitles help: watching with captions beats watching without them for both comprehension and vocabulary, and meta-analyses keep confirming it (Montero Perez et al., 2013; Kurokawa et al., 2025). The catch is what beginners do with them. Learners with a low level lean on the text and stop processing the audio (review in Frontiers in Psychology, 2022). Reading is easier than listening, so when text is on screen your eyes do the work and your ears don't. Your listening only gets trained if your ears get the first try. That is also the skill you actually want: understanding people who are talking to you, without subtitles.

Step 4: say how much you got

Mañana vamos a la playa con mis primos.

You get three answers, and none of them is wrong. Understood this sentence goes to the quiz, which checks whether you really did. Again replays it, free and as often as you like. Help me with subtitles moves on to step 5.

Why we built it this way. Again is free because repetition is the cheapest effect we have. Many sentences click on the second listen (Holzknecht, 2024), and a sentence you work out from the sound alone is worth more than one you read. The self-check is honest by design: nothing is scored, so nothing pushes you to claim you understood something you didn't.

Step 5: Smart Subtitles

Mañana vamos a la playa con mis primos.

The same sentence plays again, now with the Spanish text. The underlines show how common each word is: solid for the 500 most common words, dashed for the top 1,000, dotted for the top 2,000, and no line for anything rarer. You can tap any word for its meaning. After the replay you can say you understood it, play it again, or ask for a word-by-word breakdown.

Why we built it this way. The text is in Spanish, not your language, so you are matching sounds you have already heard to their written form. Captions in the language you're learning are where the meta-analyses above found the benefit. You come to them after trying with your ears, so the text confirms what you heard rather than replacing the listening. The underlines tell you which unknown words are worth your attention: missing a top-500 word is a gap to fix now, while missing a word ranked 5,000th rarely matters.

Step 6: word by word, with a little effort

Mañana vamos a la playa con mis primos.

Each word gets a translation underneath it. You'll notice they are not all the same. Under words that are probably above your level (playa, primos) the translation is shown. Under the words you probably know, it is blurred, and you need one tap to peek. Reveal all shows everything at once, and the app records it as a sign that the sentence was hard.

Why we built it this way. This is the effort-based reveal, and it comes from one of the most useful ideas in vocabulary research. The Involvement Load Hypothesis says you remember a word in proportion to how much a task makes you need it, search for it and evaluate it (Hulstijn & Laufer, 2001). A translation that appears without you asking costs nothing and leaves little behind. One that you reach for costs a moment of thought, and that thought helps the word stay. The format matters too:

• Translations in your own language beat definitions in the target language (Kim, Lee & Lee, Language Teaching Research).
• Putting the translation directly under the word (the interlinear format) lowers cognitive load compared with glosses in the margin (Bonilla Carvajal, 2025).
• Captions that show only the hard words worked about as well as full captions while showing far less text (Montero Perez, Peters & Desmet, 2014). Leaving the easy words alone keeps your attention on the sound.

A peek is useful information too. If you tap vamos, the app learns that a word it assumed you knew is not solid yet.

Step 7: one quick quiz

Mañana vamos a la playa con mis primos.

The sentence comes back with one word missing, and you choose which word you heard. The missing word is the one that gave you trouble: the rarest word above your level, preferring the ones you tapped or peeked at. The other options come from the same clip, and about a third of quizzes include a word that looks similar to the answer (premios next to primos), so you have to read the word itself. It is one question and takes a few seconds. If you pick the wrong word, the right one is shown and the loop moves on.

Why we built it this way. This is the testing effect: pulling a word out of memory strengthens it much more than seeing it again (Barcroft, 2007; Cognitive Science, 2022). Most video-based study tools stop at showing you the word. That step builds recognition, but retrieving the word yourself is what makes it stick.

Step 8: zoom out

Mañana vamos a la playa con mis primos.

At the end of each part, the whole part replays at full speed with no text and no stops. The sentences you just worked through one at a time now go by at normal speed, and you can follow them. This is where the effort of the loop pays off.

Why we built it this way. It trains the exact skill you are after: following natural speech in real time. It also completes a sequence that starts with a lot of support and ends with none. A 2025 study found that a staged sequence like that, compared with any single fixed subtitle mode, lowered cognitive load and improved recall of word forms and meanings (Yuan et al., 2025). Focus runs that sequence for every sentence, and again for every part.

The loop adapts to you

Mañana vamos a la playa con mis primos.

Focus watches how you're doing and adjusts:

• Three sentences in a row understood on the first listen, and it offers to stop stopping. Let it run plays the next sentences straight through with subtitles and checks in with you every four sentences. Keep stopping keeps the full loop, and the offer comes back after another three first-listen wins.
• Two "Reveal all" in a row, and the next sentences start at the subtitles step instead of the no-text listen. Speech that is genuinely fast (above 170 words per minute) plays at 0.9× speed. Slower speech is left alone.
• Skip is one tap away at every step. The loop never keeps you stuck on a sentence, and a skip also tells the app something.

Why we built it this way. If a task is too easy, you're only reading. If it's too hard, you're hearing noise. Learning happens in between, and that middle zone moves from clip to clip, so the loop recalculates it from your answers on every sentence.

The end of the clip

Mañana vamos a la playa con mis primos.

The end card shows how many sentences you cleared, the words you reinforced, and one number worth watching over the weeks: the share of sentences you understood on the first listen, without text. Tap any word to save it for review.

Extensive listening: a feed at your level

Once you understand most of what you hear, a different kind of learning starts to work. Studies of extensive viewing found that learners pick up a real share of unknown words just from watching a lot of TV, and that how often a word comes up is the best predictor of which ones stick (Rodgers & Webb, 2019; Pujadas & Muñoz, 2019). At this stage the amount you watch matters more than working through each sentence, as long as the videos stay at your level.

That last part is the hard part. Real videos vary a lot: one creator speaks slowly about daily life, the next one talks fast about politics. In Reelang, extensive listening happens in Flow: a vertical video feed, much like TikTok, YouTube Shorts or Instagram Reels, except that it learns your level and shows you clips around it, by vocabulary and by speaking speed. Smart Subtitles are on and you can tap any word for its meaning, but nothing stops the video.

Flow learns from what you do, and from the life buoy on the right side rail. Tap it and say how the clip felt: Too fast, Too hard, Just right or Too easy. The clips that follow move toward what you told it. You can try it here: tap the buoy, give an answer, and watch the level.

Mañana vamos a la playa con mis primos.

How to combine intensive and extensive listening

• Let the video decide, not a schedule. Use the 30-second test above. If you follow it, keep watching. If you don't, switch to intensive work for that clip.
• Keep intensive sessions short. Ten to twenty focused minutes beat an hour of half-attenion. Stop when you notice you are pressing "again" without listening.
• Do intensive work on material slightly above your level, not far above it. If almost every sentence needs the word-by-word breakdown, the clip is too hard to be worth it today.
• Shift the balance as you improve. Early on, most of your time goes into intensive work on easy, clear speakers. Later, most of it is extensive, with intensive sessions for the hard clips you still want to understand.
• Track one number. The share of sentences you understand on the first listen, with no text, is the clearest sign your listening is improving. Focus shows it at the end of every clip.

In Reelang you can switch between Focus and Flow at any time, and the same learner often uses both on the same day. Focus isn't only for beginners: years in, it's still the best way through a clip that is above your level.

Questions and answers

What is the difference between intensive and extensive listening?

Intensive listening means working closely through a short piece of audio: replaying it, checking what you understood and studying the words you missed. Extensive listening means listening to a lot of material for meaning, at a level where you already follow most of it. Intensive listening suits material above your level; extensive listening suits material at or just below it.

How can I improve my listening comprehension in a foreign language?

Combine the two kinds of listening. Work through short clips that are a little above your level sentence by sentence: listen first without subtitles, then read, then check the words you missed. Spend the rest of your time watching a lot of material you can mostly follow. Choose material by how much you understand without subtitles, and track how many sentences you understand on the first listen.

When should I use Focus Mode instead of Flow in Reelang?

Use Focus Mode, Reelang's intensive listening mode, when you can't follow a clip on a first listen. Use Flow, the feed for extensive listening, when you understand most of it. The same learner can use both: Focus on a fast, dense clip and Flow on a slower one.

Why does Focus Mode hide the subtitles at first?

When text is on screen, learners, beginners especially, read it and stop processing the audio. Playing each sentence once with no text gives your listening the first try. Subtitles are one tap away when you need them.

Why are some translations blurred in Focus Mode?

In the word-by-word step, translations are shown under words that are probably above your level, and blurred under words you probably know. Revealing one takes a tap. Research on vocabulary learning shows that a little effort in finding a meaning helps you remember it, and each tap also tells the app which words you are not yet sure of.

Does Focus Mode work in languages other than Spanish?

Yes. Focus Mode works the same way in every language on Reelang. This article uses Spanish only for the examples.

Sources

1. van Zeeland, H. & Schmitt, N. (2013). Lexical coverage in L1 and L2 listening comprehension. Applied Linguistics. doi:10.1093/applin/ams074
2. Montero Perez, M., Van Den Noortgate, W. & Desmet, P. (2013). Captioned video for L2 listening and vocabulary learning: a meta-analysis. System. doi:10.1016/j.system.2013.10.007
3. Kurokawa et al. (2025). Meta-analysis of captioned viewing and incidental vocabulary learning. Language Learning. doi:10.1111/lang.12697
4. Review of on-screen text in second language video viewing (2022). Frontiers in Psychology. doi:10.3389/fpsyg.2022.904523
5. Holzknecht (2024). On repeated (double-play) listening. TESOL Quarterly. doi:10.1002/tesq.3249
6. Hulstijn, J. & Laufer, B. (2001). Some empirical evidence for the involvement load hypothesis in vocabulary acquisition. Language Learning. doi:10.1111/0023-8333.00164
7. Kim, Lee & Lee. On first-language versus second-language glosses. Language Teaching Research. doi:10.1177/1362168820981394
8. Bonilla Carvajal (2025). On interlinear translation and cognitive load in vocabulary learning. Íkala, Revista de Lenguaje y Cultura. doi:10.17533/udea.ikala.356253
9. Montero Perez, M., Peters, E. & Desmet, P. (2014). Is less more? Effectiveness and perceived usefulness of keyword and full captioned video for L2 listening comprehension. ReCALL. doi:10.1017/S0958344013000256
10. Barcroft, J. (2007). Effects of opportunities for word retrieval during second language vocabulary learning. Language Learning. doi:10.1111/j.1467-9922.2007.00398.x
11. On retrieval practice in vocabulary learning (2022). Cognitive Science. doi:10.1111/cogs.13135
12. Yuan et al. (2025). On sequencing subtitle support and cognitive load. British Journal of Educational Psychology. doi:10.1111/bjep.12740
13. Rodgers, M. & Webb, S. (2019). Incidental vocabulary learning through viewing television. ITL International Journal of Applied Linguistics. doi:10.1075/itl.18034.rod
14. Pujadas, G. & Muñoz, C. (2019). Extensive viewing of captioned and subtitled TV series: a study of L2 vocabulary learning by adolescents. The Language Learning Journal. doi:10.1080/09571736.2019.1616806