How long does it take to learn a language? Hours of real video to each level, in 12 languages
We counted every word in 4,500 hours of native video in 12 languages to estimate how many hours of input build the vocabulary of each CEFR level.
Chief Language Pedagogy Geek @ Reelang
Our model estimates about 200 hours of Spanish video at your level to build a B1-sized vocabulary (intermediate), and 250 hours for B2 (upper-intermediate). German takes about 245 and 325 hours, Dutch 145 and 180, Russian 400 and 500. These are vocabulary milestones on the way to understanding everyday conversations and native videos, not guarantees of comprehension. How often each word comes up is something we counted, word by word, in 4,500 hours of short videos in these 12 languages. How quickly those words are learned is estimated from published research. Pick the language you are learning and your goal below, and the calculator gives you your own number and the steps on the way.
Two things to know before you trust the number. The videos were measured: we know how many words an hour of real Spanish contains and how often each one recurs. The learning was estimated: we used published research to estimate how many of those words a learner retains. Watching builds other skills too, but we haven't measured those.
The surprise: easy videos use much the same everyday words
An A1 clip is easier than a C1 clip: it is slower, with shorter sentences and simpler grammar. What we expected, and did not find, is that it would also use more common words. In a typical Spanish clip graded A1, 76 per cent of the words you hear are among the 1,000 most common Spanish words; in a typical C1 clip, 79 per cent. In German it runs the other way, 81 against 78. Across the 12 languages the level label moves that share by a few points at most, in no consistent direction, while the speed does change: across our native Spanish clips, A1 runs at a median of about 130 words a minute and C1 at about 165, German A1 at about 100 and C1 at about 140. People talking about real life reach for much the same everyday words whether or not a learner is listening; what an easy label marks is how slowly and simply they say them.
That changes what the hours are for. There is no level of real video where the words are all easy ones. From the first hour you are in the full language, and the time is spent not on meeting fewer words but on meeting the same words often enough to keep them.
Our speech-rate study found the same thing in speed: A1 clips are clearly slower, but from A2 to C1 speed barely moves (Spanish A2 at 5.87 syllables a second, C1 at 5.93). So above A1, speed and common words hardly change, yet a C1 clip is plainly harder. What is left is grammar, how densely ideas are packed, topic, and how much you need to know about the speaker's world. To hear what's left, here are two English clips with the same speed and the same share of common words: a mom plating McDonald's for her family (A2) and a news report on Australia suing the chemical company 3M (C1).
The news report asks for more background knowledge: a $2 billion lawsuit, "forever chemicals", the Attorney General.
After A1, don't judge a video by how fast someone speaks. Try watching it. If you can follow what's going on, even when you miss a few words, it's probably a good fit.
Hours to each level, in 12 languages
Cumulative hours of video at your level, starting from nothing, until your vocabulary matches a learner at that level. Central estimates; how fast you personally pick up words moves them, as the method section explains. Each language links to its clips by level.
| Language | A1 | A2 | B1 | B2 | C1 |
|---|---|---|---|---|---|
| Dutch | 40 | 85 | 145 | 180 | 240 |
| Swedish | 55 | 105 | 160 | 210 | 245 |
| Romanian | 50 | 110 | 165 | 210 | 280 |
| English | 75 | 120 | 170 | 215 | 290 |
| French | 60 | 115 | 180 | 225 | 280 |
| Spanish | 75 | 135 | 200 | 250 | 340 |
| Italian | 80 | 145 | 225 | 280 | 365 |
| Portuguese | 85 | 155 | 245 | 290 | 360 |
| German | 80 | 155 | 245 | 325 | 435 |
| Greek | 90 | 165 | 250 | 315 | 415 |
| Czech | 105 | 205 | 310 | 400 | 560 |
| Russian | 190 | 305 | 400 | 500 | 660 |
Rounded to the nearest 5 hours. Estimates, not measurements: see "How we measured this".
For a B2-size vocabulary, Dutch comes out at about 180 hours; Romanian, Swedish, English and French at 210 to 225; Spanish at 250; Italian, Portuguese, Greek and German at 280 to 325; Czech and Russian at 400 and 500. For most of these languages, that is one to two years at half an hour a day.
Small differences between languages, like French's 225 hours against Spanish's 250, shouldn't be overinterpreted, because the estimates depend partly on how words are counted.
Guides for each language, with what to watch at each stage, will follow.
What your first language changes
If your first language is related to the one you are learning, you start with a head start: many of the words you hear will sound familiar. To see how big it is in real speech, we took the cognates documented by Schepens and colleagues (2013), words whose translation in the other language is spelled and pronounced alike, and counted them in our clips. For a Spanish speaker, about one word in eight heard in Italian videos is one of them; in French videos, one in seventeen. On paper Spanish shares about three-quarters as many of these words with French as with Italian, but in everyday speech the French ones come up less than half as often. Listening tests point the same way: Spanish speakers understood 46 per cent of a spoken Italian text and 28 per cent of a French one (Gooskens and colleagues, 2018).
A familiar-sounding word is not a word you know. It is one you may recognise in context, and sometimes it misleads you: a French magasin is a shop, not a magazine, and a Polish speaker who hears Russian склеп (sklep) thinks "shop", when it means a crypt. How much faster such words are learned has never been measured, so the calculator does not take hours off for your first language; its numbers are for a learner without that head start. If you have one, you may get there sooner, most of all at the beginning.
Why it takes this long
An hour of real Spanish video carries about 8,000 words, so 250 hours is two million words heard. If every word you met twelve times were yours, a B2 vocabulary would arrive after about 100 hours of Spanish, 170 of German, 200 of Russian. That is the floor: the hours in which the words can even be met often enough, pure arithmetic about the videos. Nobody learns that fast from input, for two reasons the research is clear about.
Meeting a word is not learning it. In studies of learning from reading and viewing, a word met eight to twelve times is recognised afterwards by roughly half the learners, and words met fewer than eight times mostly fade within months. So a learner keeps perhaps one word in two of those met ten times, and the hours double.
And a word only teaches you something when you can follow the sentence around it. A beginner who knows 70 per cent of the words in a clip is guessing at every third word, and most guesses do not land; the same word met later, with 90 per cent of its neighbours known, does. This is why the first hundred hours feel slow and the next hundred fast, why the calculator's steps get shorter as they go. It is also why a related first language helps twice: the words you recognise, and the sentences they make followable for the words you don't. It is also the case for watching at your level: not because easy clips have easier words, they don't, but because a word met inside a sentence you understood is a word you keep.
Lágrima, tear, is the 2,845th most common word in our Spanish clips. In B1 videos it turns up about once every 5 hours, so meeting it ten times takes about 50 hours of video, and ten meetings give it about an even chance of being learned in our central estimate. Thousands of words like it are being met at the same time, each one slowly, and that is where the hours go.
Seen that way, a few hundred hours stops being a wall and becomes a plan. The common words you will meet hundreds of times whether you try or not. The hours are for the words in the middle of the list, the ones that come up every few hours of video, which you need to meet across many clips before they stay.
Why the languages differ
Where the differences between languages come from, the counts can only partly say. What they do show is how much of everyday speech the first few thousand words cover. The 2,000 most common Dutch words cover about 88 per cent of the words in Dutch clips, the 2,000 most common Russian words about 78 per cent: with 2,000 words, a Dutch learner meets an unfamiliar word about once in eight, a Russian learner more than once in five. German sits between them at 84 per cent.
Why the curves differ is partly the languages and partly how words are counted. Our word lists count every German compound as a word of its own, so Schuhgröße (shoe size) and Preisvergleich (price comparison) are separate entries, where Spanish says talla de zapato (literally "size of shoe"), words a learner meets anyway. That raises the count for German, and it may overstate the effort: a learner who knows Schuh and Größe can often work out Schuhgröße the first time they hear it. Russian and Czech inflect heavily, and even after grouping forms under their dictionary word, more distinct words share the stream. Romanian comes out below Spanish for a similar bookkeeping reason: its most common words cover a little more of the speech in our Romanian clips, so fewer distinct words reach the same milestone. That reflects how our clips and word list count, not proof that Romanian is easier. Those are the likely explanations; what we have measured is the coverage, not the cause.
How to use the hours well
Watch at your level, and move up before you feel ready. The words you retain are the ones met in sentences you mostly follow, so clips where you catch nine words in ten are where the learning is. At A1 that means slow clips with pauses and short lines. From A2 up, speed stops being the signal, so choose by topic and by whether you can hold the thread, and when clips at your level stop surprising you, go up a level even if you still miss things. A clip you find slightly hard is the one teaching you.
Clips at the same level vary a lot, so judge a clip by watching it, not by its label: if you miss fewer than one word in twenty, it's at your level; more than one in ten, it's still above you.
Half an hour a day beats three hours on Sunday. What keeps a word is meeting it again in a different clip on a different day. At 30 minutes a day, a B2-size Spanish vocabulary is about a year and five months away; at an hour a day, about eight months. The hours are the same; only the calendar moves.
Turn on subtitles in the language you are learning. The hours above are for watching without captions. Studies of learning from television find that same-language captions raise vocabulary gains substantially, mostly because they show where one word ends and the next begins and let you check the ones you missed. Translated subtitles help you follow the story more than they help you learn words.
If native video is too much at first, start gentler. Videos made for learners by comprehensible input creators are slower and lean on pictures and repetition, which makes the first hours easier to follow; move on to native clips as soon as you can follow them.
Let the picture do its work. Cooking, shopping, getting dressed, feeding the cat: clips where you can see what the words mean are a beginner's friend, because the understanding arrives before the vocabulary does.
Watch the vegetables go in: tomate, papa and cebolla land in the pan as they are named, with a label on screen. Only 84 per cent of this clip's words are common ones, and the picture carries you through the rest.
Mix extensive and intensive. Most of your hours should be extensive: watching for the story, not stopping. Some should be intensive: a short clip replayed, the missed words checked. Our guide to intensive and extensive listening covers when to do which, and the listening-first roadmaps for Spanish and German, and our guide to the best way to learn a language, walk the stages.
Use your first language, and know its limits. If it is related to the language you are learning, listen for the patterns: once you have noticed that Spanish -ción is Italian -zione, hundreds of words open at once. The sound-alikes are also where the traps are, so when a familiar-looking word makes a sentence strange, check it.
Where this sits next to the numbers you have seen
| Source | What it counts | Spanish, roughly | What it leaves out |
|---|---|---|---|
| US Foreign Service Institute | Classroom hours for diplomats to reach professional working proficiency | 690 class hours; German 828; Russian, Czech, Greek 1,012; Arabic, Japanese, Korean, Chinese 2,200 | Anyone not studying full time; all four skills bundled |
| Cambridge guided learning hours | Lessons plus supervised study to each CEFR level | A2 180–200 · B1 350–400 · B2 500–600 · C1 700–800 | Input outside class; one figure for every language |
| Dreaming Spanish roadmap | Hours of their own slowed, teacher-made video, from learners' reports | 300 hours ≈ 3,000 words · 600 ≈ 5,000 · 1,000 ≈ 7,000 | Other languages; real native speech |
| This page | Hours of real native video at your level until your vocabulary matches a level | B1 vocabulary 200 · B2 250 · C1 340 | Grammar, speaking, speed; an exam result |
The classroom figures are higher partly because an hour of lessons carries far fewer words than an hour of video, and partly because they count the whole level, speaking and writing included.
Dreaming Spanish is the one place we could check our model against learners. Their roadmap says learners know about 1,500 words at 150 hours, 3,000 at 300, 5,000 at 600 and 7,000 at 1,000, from thousands of learners' reports. We analyzed their catalog with our model, without adjusting it to fit their figures, and it predicted 1,480, 3,420, 5,580 and 7,400: within about 15 per cent. Its ordering of languages also matches the Foreign Service Institute's, which it was never shown (German about 1.2 times Spanish, the Slavic languages 1.5 to 2 times), though the size of each gap is less certain than the order. The model's predictions were broadly consistent with Dreaming Spanish's published roadmap and the Foreign Service Institute's language-difficulty categories. These are encouraging plausibility checks, not direct validation of viewing hours or learner proficiency.
How we measured this
The clips. The estimates come from 85,365 clips (4,526 hours) in 12 languages from the Reelang Native Video Corpus, snapshot 2026-11-07: native video with speech, in a single language, and not language lessons. We took up to 2,000 clips per level in each language, so the less common levels are not swamped by the common ones.
Counting the words. Every word was looked up in the frequency list Reelang's Smart Subtitles use for its language, with forms grouped under their dictionary word, so dice, dijo and decir are one word with one rank. Names, noise and words beyond the list, 2 to 9 per cent of the running words, are left out when the model sets its milestones; wherever this page gives a share of the words you would know, in the calculator and in the data table, they count as unknown, so those shares are lower by that amount. From this come the exposure figures: for each level, how often each word turns up per hour, and in how many different clips.
Estimating the learning. This part is a model built from published research, not a measurement. Each clip in which a word appears gives it a chance to be learned: 6.7 per cent in the central estimate, so a word met in ten clips is about 50 per cent likely to be known, in line with studies of incidental learning. The chance is smaller when the learner follows little of the clip, with a floor for beginners helped by pictures. When the expected number of known words reaches a level's milestone, the learner moves on to the next level's clips.
The milestones are the vocabulary sizes James Milton measured in more than 10,000 Cambridge exam candidates: about 1,000 words at A1, 2,000 at A2, 2,900 at B1, 3,500 at B2 and 4,100 at C1. Each is carried to another language by coverage: English B1's 2,900 words cover 92.7 per cent of the words in English clips, so the Spanish B1 milestone is the number of Spanish words covering 92.7 per cent of Spanish clips, about 2,840. C2 is not estimated, because Milton's test stopped at 5,000 words.
How fast you pick up words. The calculator gives the central estimate. A learner who picks words up quickly (10 per cent per clip) needs about 35 per cent fewer hours; one who picks them up slowly (4 per cent) about 65 per cent more. The research is itself that wide, and attention is a large part of it: replaying, pausing and checking a word can raise the chance it is retained.
Comparing languages. Each language has its own clips and its own word list, and no two lists are built from exactly the same sources. Sources, normalization and what counts as one word (a German compound, a Russian verb form) all affect how many words a milestone takes, so a 10 to 15 per cent difference between two languages should not be read as real. Some levels are thin in hours: Czech B2 and C1 rest on about 26 and 5 hours of video, Russian C1 on about 10, Swedish C1 on about 19, Romanian and Greek A1 on about 7 each. Resampling the clips moves any B2 estimate by 4 per cent at most. Where a learner needs more hours at a level than the sample holds (Russian A1 takes about 190 hours of watching, against 11 hours of A1 clips in the sample), the model assumes further clips at that level bring words at the same rates per hour as the sampled ones.
Speed is words per minute of talk, from a clip's first word to its last, with pauses of up to six seconds counted as talk.
What it does not measure. Grammar, speaking, reading and coping with speed; captions and glosses (the hours are for plain viewing); words you may already recognize from languages you know, since the model counts every word as new; and the learning itself, which no study has yet measured against a proficiency test. Levels are a model's reading of each transcript, and the catalog leans toward clear speech and broad appeal.
Use this data
You are welcome to reuse the numbers and tables on this page in articles, teaching materials and research; please link to this page as the source. The hours are in the tables above. The milestones, and what they cover, are here.
| Language | Clips in sample | Hours in sample | Words per hour | Milestone A2 / B1 / B2 / C1 (words) | Share of words known at B2 |
|---|---|---|---|---|---|
| Dutch | 7,113 | 303 | 8,999 | 1,588 / 2,542 / 2,985 / 3,559 | 90% |
| Swedish | 5,574 | 313 | 7,821 | 1,225 / 1,953 / 2,449 / 2,711 | 90% |
| Romanian | 5,607 | 326 | 7,844 | 1,636 / 2,343 / 2,838 / 3,286 | 88% |
| English | 9,490 | 410 | 9,054 | 2,001 / 2,901 / 3,501 / 4,101 | 92% |
| French | 7,449 | 453 | 9,299 | 1,846 / 2,876 / 3,483 / 4,111 | 91% |
| Spanish | 9,824 | 672 | 8,152 | 1,919 / 2,837 / 3,412 / 3,972 | 89% |
| Italian | 7,639 | 298 | 7,815 | 2,263 / 3,448 / 4,126 / 4,794 | 89% |
| Portuguese | 8,643 | 470 | 8,001 | 2,111 / 3,198 / 3,625 / 4,125 | 90% |
| German | 10,683 | 783 | 6,441 | 2,129 / 3,391 / 4,258 / 5,057 | 88% |
| Greek | 5,269 | 217 | 7,135 | 2,121 / 3,239 / 3,991 / 4,710 | 86% |
| Czech | 3,532 | 133 | 6,336 | 2,939 / 4,422 / 5,268 / 6,130 | 88% |
| Russian | 4,542 | 149 | 6,305 | 3,658 / 5,228 / 6,164 / 6,961 | 87% |
Clips and hours in sample: the stratified sample this study ran on, not the whole catalog. Words per hour: running words per hour of video across the sample, pauses included. Milestone: the receptive vocabulary, in dictionary words, that marks each level in that language. Share of words known at B2: the share of running words in real clips a learner with the B2 vocabulary knows, counting names and unlisted words as unknown; it is a measure of words, not of understanding.
Dataset snapshot: 2026-11-07.
!callout How to cite this Maksimyuk, R. (2026). How long does it take to learn a language? Hours of real video to each level, in 12 languages. Reelang Research. Based on the Reelang Native Video Corpus, snapshot 2026-11-07 (85,365 clips, 4,526 hours, 12 languages). Estimates, not measured learner outcomes. https://reelang.com/blog/how-long-does-it-take-to-learn-a-language
Questions people ask
How many hours does it take to become fluent in a language? In our model, a B1-sized vocabulary takes about 145 hours of Dutch video, 200 of Spanish, 245 of German and 400 of Russian. For a B2-sized vocabulary, the estimates are 180, 250, 325 and 500 hours respectively. Understanding and speaking fluently require more than vocabulary alone.
Is 30 minutes a day enough? Yes, if it is every day: that is about 180 hours a year, a B2-size vocabulary in Dutch or close to a B1 one in Spanish by our estimates. Meeting the same words on different days matters more than long sessions.
Does it help if I already speak a related language? Yes: a Spanish speaker hears a documented sound-alike in about one word in eight of Italian video. How many hours that saves has not been measured, so the calculator's estimate is for a learner without that head start.
Can you learn a language just by watching videos? You can build the vocabulary and the ear for one; research finds people learn words from viewing at about the same rate as from reading. Speaking and writing need their own practice.
Is Spanish faster to learn than French? In our model they are close, about 250 hours of video to a B2-size vocabulary for Spanish and 225 for French, a gap too small to read much into. Both come out well below German or Russian.
What level do you need to watch YouTube or Netflix without subtitles? Usually more than a B2 vocabulary. Our B2 milestones cover about 86 to 92 per cent of the words in native clips, depending on the language; for 95 per cent, our vocabulary study found the typical Spanish clip needs about 6,800 words. Knowing the words is the first condition for following the speech, not a guarantee of it.
Are these hours of lessons or hours of watching? Watching. An hour of lessons carries far fewer words than an hour of video, so classroom figures from Cambridge or the Foreign Service Institute measure something else.
Sources
- Milton, J. (2010). The development of vocabulary breadth across the CEFR levels. In I. Bartning, M. Martin & I. Vedder (eds.), Communicative proficiency and linguistic development (EUROSLA Monographs 1), 211–232. eurosla.org
- Milton, J. & Alexiou, T. (2009). Vocabulary size and the Common European Framework of Reference for Languages. In B. Richards et al. (eds.), Vocabulary studies in first and second language acquisition, 194–211. Palgrave Macmillan.
- Nation, I. S. P. (2006). How large a vocabulary is needed for reading and listening? Canadian Modern Language Review, 63(1), 59–82. doi.org/10.3138/cmlr.63.1.59
- Nation, P. (2014). How much input do you need to learn the most frequent 9,000 words? Reading in a Foreign Language, 26(2), 1–16. nflrc.hawaii.edu
- Webb, S. & Rodgers, M. P. H. (2009). Vocabulary demands of television programs. Language Learning, 59(2), 335–366. doi.org/10.1111/j.1467-9922.2009.00509.x
- van Zeeland, H. & Schmitt, N. (2013). Lexical coverage in L1 and L2 listening comprehension. Applied Linguistics, 34(4), 457–479. doi.org/10.1093/applin/ams074
- Schmitt, N., Jiang, X. & Grabe, W. (2011). The percentage of words known in a text and reading comprehension. Modern Language Journal, 95(1), 26–43. doi.org/10.1111/j.1540-4781.2011.01146.x
- Schepens, J., Dijkstra, T., Grootjen, F. & van Heuven, W. J. B. (2013). Cross-language distributions of high frequency and phonetically similar cognates. PLoS ONE, 8(5), e63006. doi.org/10.1371/journal.pone.0063006
- Gooskens, C., van Heuven, V. J., Golubović, J., Schüppert, A., Swarte, F. & Voigt, S. (2018). Mutual intelligibility between closely related languages in Europe. International Journal of Multilingualism, 15(2), 169–193. doi.org/10.1080/14790718.2017.1350185
- Uchihara, T., Webb, S. & Yanagisawa, A. (2019). The effects of repetition on incidental vocabulary learning: A meta-analysis. Language Learning, 69(3), 559–599. doi.org/10.1111/lang.12343
- Waring, R. & Takaki, M. (2003). At what rate do learners learn and retain new vocabulary from reading a graded reader? Reading in a Foreign Language, 15(2), 130–163. nflrc.hawaii.edu
- Peters, E. & Webb, S. (2018). Incidental vocabulary acquisition through viewing L2 television. Studies in Second Language Acquisition, 40(3), 551–577. doi.org/10.1017/S0272263117000407
- Feng, Y. & Webb, S. (2020). Learning vocabulary through reading, listening, and viewing. Studies in Second Language Acquisition, 42(3), 499–523. doi.org/10.1017/S0272263119000494
- Montero Perez, M., Van Den Noortgate, W. & Desmet, P. (2013). Captioned video for L2 listening and vocabulary learning: A meta-analysis. System, 41(3), 720–739. doi.org/10.1016/j.system.2013.07.013
- Stæhr, L. S. (2008). Vocabulary size and the skills of listening, reading and writing. Language Learning Journal, 36(2), 139–152. doi.org/10.1080/09571730802389975
- Dreaming Spanish (2025). Language Learning Roadmap, v1.3. dreaming.com
- U.S. Department of State, Foreign Service Institute. Foreign Language Training. state.gov
- Cambridge English. Guided learning hours. support.cambridgeenglish.org
- Reelang Research (2026). How many words do you need to understand native videos? and How fast do native speakers talk?
Start watching
Pick a language and start watching. Real people talking, sorted by level. No level test, no sign up.
Pick a languageRead next
How many words do you need to understand native videos?
With the 5,000 most common English words, the typical native clip is 95.6% words you know. Spanish needs about 6,800 words to reach 95%, and in 13 of 23 languages even 10,000 are not enough. Data from 35,615 native clips, per language and per level.
How fast do native speakers talk? Speech rates in 44 languages
In native videos, Tamil, Japanese, Turkish, Portuguese, Italian and Spanish speakers say about 6 syllables per second, English and German speakers about 4.7. Measured on 16,293 clips in 44 languages, with the speed by CEFR level, by dialect, and against the classic lab study.
Intensive vs extensive listening: how to learn a language from real videos
Intensive listening works closely through short clips; extensive listening covers a lot at your level. When to use which, shown with real videos and the research behind each step.