How many hours of comprehensible input do you need?

If you learn a language mainly by listening, you will eventually ask how many hours it takes. The honest answer is “it depends”, but there are numbers worth knowing, because they turn an open-ended project into something you can pace.

The short answer. For an English speaker learning a closely related language such as Spanish, French or Portuguese, the most widely quoted roadmap puts comfortable understanding of everyday native speech at around 1,000 hours of comprehensible input. The first real milestone comes at about 300 hours, where easy native content starts to open up, and the last at about 1,500, where nearly everything is followable. For a distant language such as Japanese, Korean or Mandarin, learners usually double every figure, which puts the same landmarks at roughly 600, 2,000 and 3,000 hours. Those totals only apply to time that was genuinely in the target language, genuinely playing, and mostly understood, so the count has to leave out ads, pauses, videos in your own language and content you could not follow. At thirty minutes a day the first landmark takes under two years; at an hour a day it takes under one.

The roadmap numbers

The most widely used figures come from the Dreaming Spanish roadmap, built from learners who logged their hours. For a language close to your own, such as an English speaker learning Spanish, the levels are roughly:

Hours	What it feels like
50	You follow slow, simple speech made for learners.
150	Learner content at normal speed is comfortable.
300	You follow easy native content on familiar topics.
600	Most native speech on everyday topics, with effort.
1,000	Native content is comfortable; you start speaking.
1,500	You follow nearly everything, including fast or slangy speech.

For a distant language, Japanese, Korean or Mandarin for an English speaker, people usually double those figures. The US Foreign Service Institute’s estimates point the same way: its “hardest” category is about four times the classroom hours of its easiest.

Treat these as landmarks, not promises. But they have a real use: a learner at 40 hours who feels lost is exactly where the roadmap says they should be, and knowing that is the difference between quitting and continuing.

Where the figures come from

The roadmap above is not a study. It was assembled from the hours that Dreaming Spanish learners logged and the abilities they reported at each stage, which makes it a large body of self-reported experience rather than a measurement. That is weaker evidence than it sounds when quoted as a table, and stronger than it sounds when dismissed, because there is nothing else built from actual counted hours.

The second source is the US Foreign Service Institute, which has taught languages to diplomats for decades and publishes how long its courses take. Its easiest category, which includes Spanish, French, Dutch and the Scandinavian languages, is about 600 to 750 hours of classroom time to reach professional working proficiency. Its hardest, which includes Japanese, Korean, Mandarin, Cantonese and Arabic, is about 2,200. Those are class hours in an intensive programme with homework on top, so the totals do not transfer to a learner watching videos in the evening. The ratio between the categories does transfer, and it is why the immersion community doubles its figures for distant languages rather than adding a little.

The third source is the immersion communities themselves. AJATT, the site that popularised all-day Japanese exposure, talked in terms of ten thousand hours. Refold and the learners around it tend to talk in milestones of a thousand and two thousand hours. None of these are claims about what is necessary; they are what people who kept counting found it took them.

Why distant languages take longer

The doubling is not arbitrary. An English speaker starting Spanish already recognises thousands of words on sight and by ear, shares an alphabet, and hears a sound system with only a handful of unfamiliar parts. The same learner starting Japanese shares almost nothing beyond loanwords, meets a word order that puts the verb at the end, and has to learn to hear pitch and vowel length as meaningful. The first hundred hours of Japanese are spent building the foundation that an English speaker gets for free in Spanish. The roadmap is the same shape; it is simply stretched.

Only some hours count

Here is the catch the roadmaps quietly assume. The hours on those tables are hours of comprehensible input in the target language. Three things people log that shouldn’t be logged:

• Time in your own language. An English recap of a Japanese show is English. It goes in the day, not the total.
• Ads, pauses and the tab you forgot about. A player that says “3 hours” was open for three hours. That is not the same number.
• Input you didn’t follow. Forty minutes of native-speed news at 30% comprehension is not forty minutes on the roadmap. It is closer to zero, and possibly worse than zero if it made you feel you were failing.

The first two are mechanical, and a tool can fix them. The third is a judgement only you can make, and the usual advice stands: pick content where you follow most of what is said, and move up when it gets easy.

How to tell whether you followed it

There is no instrument for comprehension, but there is a working test. At the end of a video, could you say what the last minute was about, in your own words, without having read subtitles in your own language? If yes, it was input. If you were carried along by pictures and tone and could not have summarised the speech, it was exposure, which has some value for getting used to the sound of the language and very little for the roadmap.

Subtitles complicate this. Subtitles in your own language turn a listening hour into a reading hour in your own language, with the target language as background noise. Subtitles in the target language are more defensible, especially for a language whose word boundaries are hard to hear, but an hour spent reading them is not quite an hour of listening either. A reasonable rule is to count the time if you were mostly listening and glancing at the text, and not if you were mostly reading.

The practical advice for choosing content has not changed in decades: pick material where you follow most of what is said, and move up when it feels easy rather than when it feels possible. Falling asleep to a podcast in the target language is a pleasant habit and not a countable one. For more on why raw watch time misleads, see why “minutes watched” is the wrong number for learners.

Why the count has to be honest

A padded total is worse than no total. Two learners with “300 hours” logged, one honest and one counting everything, will have wildly different abilities, and the second will conclude the method doesn’t work. The value of the number is in comparing yourself to yourself over time, and that only works if the units stay the same.

The padding is also invisible from the inside. An ad break here, a tab left open there, an English video counted as Japanese because nobody was checking; each is small, and together they add a fraction to every week, spread so evenly that nothing ever looks wrong. You only find out when your ability does not match the roadmap’s description of your hour count, and by then there is no way to know how many of the hours were real.

Hours that are not listening

Most learners also read, review flashcards, take lessons or practise speaking, and it is tempting to put all of it in one total. The roadmaps built from listening logs are about listening. Some allow reading to count, often at a discount and only past the beginner stage; others keep it separate entirely. Whichever you choose, keep the columns apart. A listening total that is secretly part reading and part grammar study is not comparable to the tables above, and the comparison is the whole reason to count.

A pace that works

The average learner who reaches 1,000 hours does it at about an hour a day over three years, or two hours a day in eighteen months. Set a daily goal you can hit on a bad day, not a good one, and let the streak do the work. Thirty minutes a day is 180 hours a year. That is the first two levels of the roadmap, from nothing, by showing up.

Working out your own timeline

The arithmetic is short enough to do in your head. An hour a day is about 365 hours a year. Two hours a day is about 730. So the 1,000-hour landmark is three years at an hour a day and eighteen months at two, and for Japanese or Korean the doubled figure of 2,000 hours is six years at an hour a day or three at two. That is not meant to discourage. It explains why the people who reach the top of the roadmap did not find a shortcut; they changed what a normal evening looked like. The daily goal matters more than any technique, and a goal you can hit when you are tired will beat a bigger one you abandon in March.

Counting it without a spreadsheet

That is the whole reason Kiku Hours exists. It is a free Chrome extension, with no account, that counts from the video player’s own clock rather than a timer, so a hidden tab counts and a laptop that went to sleep doesn’t. It excludes ads, pauses and skipping about, and it credits the minutes a video actually took, so an hour of video at 2× counts as the thirty minutes you spent on it. It listens to each video’s audio to identify its language, one of 106 it can recognise, so an English video on a Japanese channel is recorded but held out of the total, with the held-back time shown beside the number it came out of. When a video’s language cannot be identified its time is not counted; nothing is guessed. What’s left is the number the roadmaps mean, with a streak, a calendar of every day and achievements dated the day you crossed them, all stored in your browser and exportable whenever you want your own charts. Setting it up takes two minutes; the guide to tracking on YouTube walks through it.