Ask HN: What app features actually help vocabulary stick long-term?
I want to thank everyone who commented on my previous Ask HN post about building a calm, non-gamified learning app.
I read through all the comments, and one theme kept coming up again and again: Most people don’t struggle with starting language learning — they struggle with making vocabulary actually stick long-term.
Streaks, flashcards, progress bars, and “feeling productive” came up a lot. But so did frustration: people doing everything “right” for months, yet forgetting words when they actually need them.
So I’d like to zoom out and ask a more fundamental question:
For those who learned a language seriously — what actually helped vocabulary stick long-term?
• Was it audio? • Personal sentences? • Immersion? • Writing? • Teaching others? • Something else entirely?
Related to that, I’m curious about something else:
Why do people stop using language learning apps, even when they genuinely want to learn? Is it motivation, friction, lack of personal relevance, or something else?
And more broadly — do you think tools that let learners fully control their own content (words, sentences, context) and turn that content into audio they can listen to regularly actually help with long-term retention?
I’m especially interested in what worked after the initial motivation wore off. Unless you’ve got an eidetic memory, the only things that will really make them stick long-term are consistent reinforcement either through practical application (e.g., immersion) or spaced repetition (SRS). I personally take the time to create visual and often highly inappropriate mnemonics for each new vocabulary word I learn, connecting the foreign language’s sound to an English homophone. The upfront cost is significantly higher, but it pays off since my long-term retention rate significantly improves though you won't see me publishing a shared Anki deck for the public any time soon. Quick SFW Example: The Russian word for "bed" is "кровать" which sounds a bit like cravat which is a kind of scarf. Then a sample story might be, "A distinguished and dapper gentleman with a luxuriously thick fur cravat lays down in his bed. He proceeds to unroll the cravat to fashion a comforter blanket to keep him warm before going to sleep." This makes a lot of sense.
I really like the point about the upfront cognitive cost
that matches my own experience too. What you’re describing feels very close to why “personal” content sticks better: the story, the emotion, and the effort are all doing the heavy lifting, not the tool itself. Yeah the personal part makes a huge difference. I experimented with automating LLM generated mnemonics about a year ago but when I shopped it around with a couple of friends for internal testing, we found that it wasn't nearly as effective as when they came up with the stories themselves.