How to Save Words from YouTube Subtitles (and Actually Remember Them)

You watched a 20-minute video in English. You understood maybe 90% of it. Somewhere in there were eight words you didn't know — words you'll meet again in podcasts, in meetings, in the next video. So you paused, typed them into a note, told yourself you'd review later.

We both know how that ends.

The note becomes a graveyard. Not because you're lazy, but because a bare list of words is nearly impossible to study from. This article is about the two things a saved word needs to survive, context and review, and the tools that get you there with the least friction. Full disclosure up front: one of the tools is mine (Piccard, in the comparison table below). The others get fair treatment, and if you'd rather use none of them, there's a notebook method near the end.

Why most saved words die

There are two failure points, and almost every workflow breaks at one of them.

The first is capture without context. You save "reluctant" as a translation: reluctantly = 不情愿地, 마지못한 (that same "reluctant" in Chinese and Korean). A week later the word is an orphan. You know it means something negative-ish. You don't know who says it, to whom, or that "a reluctant yes" means something other than "reluctantly agree". Translation-only lists produce exactly the fragile recognition that falls apart the moment a real person uses the word at real speed.

The second is capture without review. You save the word with a full sentence, into a system you never open again. Saving feels like progress, so your brain stops tracking the word. Nothing schedules it to come back. Three weeks later it's gone and you save it again. This is the loop most learners live in for years.

A word you meet on YouTube has one huge advantage over a word from a list: it arrived attached to a voice, to a moment on screen you half-remember. Keep that attachment and the word has roots. Lose it at the capture step and you've thrown away the best part.

What a surviving vocabulary list needs

A YouTube vocabulary list that actually works has to answer yes to three questions:

1. Does the word arrive with its sentence? Full subtitle line, ideally the line before and after. Context is the memory.
2. Does saving cost less than five seconds? If the answer is pause → switch app → type → translate → paste, you'll do it for three days. Then never.
3. Does anything schedule the word to return? Not "you'll review when you feel like it". An actual algorithm that decides: this word in 1 day, that word in 12 days.

That third one is the quiet killer. Spaced repetition isn't a study hack, it's the difference between a list and a system. The standard algorithm is FSRS, which predicts when a specific word is about to fade from *your* memory and reviews it right before. Anki implements it; so does Piccard.

The honest options, compared

Dual subtitles help comprehension. That's why the reading-aid category is popular. But comprehension isn't retention; look-ups evaporate unless something pulls them back. That gap is the whole reason I built Piccard (free to start; you buy energy top-ups, not a subscription). If you want the wider field compared honestly, I did that in Language Reactor Alternatives in 2026.

The Piccard workflow, concretely

1. Install the extension, open any English YouTube video with subtitles.
2. Click a word you don't know, right in the subtitle line. You get the meaning, the pronunciation in IPA, the CEFR difficulty level, and the full subtitle sentence it came from.
3. Click save. The card now exists, with its context, in your review queue. The pause was two seconds; you barely left the video.
4. Open the web app and review. FSRS shows each card right before you'd forget it, and your 15 cards from this week's videos stay in rotation instead of rotting in a note.

The cards are English-focused by design: the IPA and CEFR tagging matter most when English is the target. And capture doesn't stop at YouTube. Reading an article? Same click-to-save. A word on any web page, an image with text in it, a sentence you want to keep from anywhere: it all lands in the same review loop. YouTube is just where most people start, because video is where vocabulary meets a voice.

Choosing the best YouTube channel to learn English vocabulary

Not every video makes good cards. A few things that matter more than people expect:

• Interviews and vlogs beat edited explainers for natural phrasing. Scripted narration uses bookish transitions you'll rarely need to say; real people talking give you the connectors ("the thing is", "to be fair") that make your English sound lived-in.
• Channels made for learners are fine at the start. But the sooner you move to content made for native speakers about something you actually care about, the faster the vocabulary becomes yours.
• Ten words saved and reviewed beat forty words saved and abandoned. The saved-but-never-seen pile is worse than nothing: it teaches you that saving doesn't work.

If you'd rather not install anything

The zero-tool version of this loop, for the record: pause the video, write the full subtitle line in a notebook (the whole line, not the word), and write your own new sentence under it. Then take the last five pages you wrote every Sunday and quiz yourself. It works. It's just expensive: fifteen minutes of manual labor per video, and FSRS-grade scheduling is impossible on paper. That's the trade the tools above are trying to buy down, Piccard included.

Questions people ask

How do I save words from YouTube subtitles? With a browser extension that handles it (that's the Piccard workflow above), or manually: pause, copy the full subtitle line into a note, add the translation yourself. The extension route exists because the manual route dies around day three.

Is watching YouTube with English subtitles enough to learn vocabulary? Subtitles build recognition. Recognition fades without review, and when it fades you're back to saving the same word three weeks later. You need both halves: context at capture, spaced review after. Subtitles alone give you the first.

Do I need Anki for spaced repetition? No. Anki is the classic and it's excellent; Piccard has FSRS built in; a notebook works if you're willing to schedule yourself. What you need is the algorithm's behavior, not one specific app.

The short version

Words from YouTube die in exactly two places: at capture, when the word arrives without its sentence, and at review, when nothing ever schedules it to come back. Fix both. A list with context and a review loop stops being a graveyard and starts working like a memory. Click the subtitle, keep the sentence, let an algorithm bring the word back before it fades — with Piccard, with Anki and discipline, or with a notebook and stubbornness.