How to Track Comprehensible Input So It Actually Adds Up

"I've been watching a lot of Japanese lately" is not a number. Six months later you still won't know if "a lot" was 40 hours or 400, which means you can't tell whether your plateau is real or whether you've simply under-counted what you've actually done. Our guide to comprehensible input covers what counts as input and why the 70–95% comprehension zone matters. This one is about the much more boring, much more useful problem: the actual mechanics of turning hours of watching and reading into a number you can trust.

On this page

Rule one: log clock time, not content units

"Watched 3 episodes" is a content unit. It's useless for tracking because episode length varies (a 22-minute sitcom and a 45-minute drama are not the same unit), and because it tempts you to round up. "Watched 68 minutes" is clock time, and it's the only unit that adds correctly across a messy, mixed week of 12-minute videos and 90-minute films.

The fix is mechanical: check the actual runtime before you log it, or run a timer while you watch or read. Video players show elapsed time; for reading, a phone timer does the job. If you paused for 20 minutes to make dinner, don't count that 20 minutes — log the watching time, not the wall-clock time the show was open.

Rule two: weight for attention, not just duration

A focused half hour and a half hour with your phone in the other hand are not the same input, even at identical comprehension levels. Most trackers don't have a built-in attention slider, so build a simple personal rule instead:

• Full attention, no second screen: log 100% of the time.
• Mostly attentive, occasional glance away: log around 75%.
• Background with the target language on while doing chores: log around 25–40%, and tag it separately from active sessions if your tracker allows multiple activity tags.

The exact percentages matter less than being consistent with yourself. The goal isn't precision to the minute — it's making your weekly total mean roughly the same thing every week, so a slow month actually looks slow in your data instead of being padded out by three hours of half-watched TV in the background.

Reading: the page-count trap

Reading time is harder to estimate than watching time because nobody tracks "pages per minute" the way video players track elapsed seconds. The common mistake is estimating backwards from pages ("I read 20 pages, that's probably 40 minutes") using a reading speed from your native language. Early in a new language, reading speed is wildly variable — a page of a graded reader at your level might take 3 minutes, the same page six weeks later might take 90 seconds, and a page of native content well above your level might take 8.

The reliable fix is the same as for video: time yourself. Start a timer when you open the book, stop it when you close it, log that number. If you forget, it's fine to estimate afterward — but estimate from how long the session actually felt, not from a page count multiplied by a speed you're guessing at.

Tag by subtitle level, not just by activity

"Watching" isn't one activity. Native audio with no subtitles, native audio with target-language subtitles, and native audio with your own language's subtitles are three very different amounts of actual target-language processing, even though they all look identical in a log that just says "watching – 45 min." If your tracker supports notes or sub-tags, record which one it was:

• No subtitles: full input, hardest, most valuable per minute once you can follow it
• Target-language subtitles: reading + listening simultaneously, good for connecting sound to script
• Native-language subtitles: mostly reading your own language with the target language as background audio — still useful for ear training, but be honest that very little of the comprehension is coming from the target language

Six months in, this distinction is often the single most useful thing in your data: it tells you whether your "200 hours of watching" was actually 200 hours of input, or 150 hours of reading English subtitles with Japanese playing in the background.

Why raw input hours are a legitimate thing to count

Counting input hours isn't a vanity metric invented by trackers. Dreaming Spanish, one of the largest dedicated comprehensible-input programs, built its entire learner roadmap around cumulative input hours rather than course levels or chapters: its published roadmap places milestones at 150 hours (following topics adapted for learners), 300 hours (understanding a patient speaker), 600 hours (understanding people speaking normally) and 1,000 hours (comfortable with daily conversation). It also says to roughly double those numbers for a language unrelated to yours, such as Mandarin or Arabic for an English speaker. Those numbers won't transfer exactly to every language or every learner, but the underlying method — count the hours, don't count the weeks — is exactly what this system is for.

A sample weekly ledger

Here's what an honest week looks like for someone doing short, frequent sessions rather than long weekend binges — a common pattern for input-heavy learners fitting study around a job:

• Monday: 2 sessions, drama with target-language subs, 35 min total, logged at 100%
• Tuesday: 1 session, podcast while commuting, 25 min, logged at 60% (noisy train, partial attention)
• Wednesday: 3 sessions, graded reader (18 min) + drama (20 min) + vocab review (10 min)
• Thursday: background drama while cooking, 40 min, logged at 30%
• Friday: 2 sessions, no-subtitle video, 30 min, logged at 100%
• Saturday: longer session, native podcast, 50 min, logged at 90%
• Sunday: reading, timed, 24 min

Raw watching/reading time that week: roughly 242 minutes. Attention-weighted input: closer to 200 minutes. Neither number is "wrong" — the raw total tells you about consistency and habit, the weighted total tells you more honestly how much actual input you got. Logging both, even informally, is what lets you tell the difference between a week that felt busy and a week that actually moved you forward.

Common mistakes that quietly inflate the log

• Counting muted or ignored TV. If it's on in the room but you're not following it, that's ambience, not input. Don't log it, or log it at near-zero weight.
• Re-watching the same episode and counting it twice at full value. A second watch is real (and useful for catching things you missed), but it's not equivalent to fresh input — weight it down.
• Rounding every session up to the nearest 30 minutes. Over a year, consistent rounding up by even 5 minutes a session adds dozens of fake hours to your total.
• Treating "I had it on in another tab" as a session at all. If you can't say what happened in the last two minutes of audio, it wasn't a study session.

Making this sustainable

None of this needs to be precise to be useful — it needs to be consistent enough that your weekly and monthly totals mean the same thing over time. In LangTrack, logging a session by activity type takes two taps, which matters more than it sounds: the system that survives six months is the one with the least friction between finishing an episode and logging it, not the one with the most detailed attention-weighting spreadsheet you abandon in week three.

Pick your weighting rule once, write it down somewhere you'll actually see it again, and apply it the same way every day. The honesty is what makes the hours mean something — a slightly conservative real number beats an inflated one every time you're deciding whether to trust your own progress.

For the theory behind why comprehension level matters as much as hour count, see Krashen's input hypothesis, tracker-friendly, and for how input and output hours should balance against each other, see tracking input hours vs output hours.