Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Show HN: Parsnip – a tool to learn a language using books and NLP
1 point by solarmist on March 15, 2021 | hide | past | favorite | 3 comments
I created Parsnip (http://solarmist.net) and I'm getting ready to share the prototype I've been working on for a while now.

Its an e-book reader with a parser and built-in references along with a review system.

I think of it like an IDE and version control for reading books in a foreign language.

Things HN might appreciate about it:

* parses all of the text with NLP to provide rich-meta to the user. Conjugations, part-of-speech, and readings/pronunciation (for character based languages)

* it also parses it to make an example sentence database; to create exercises out of later while reviewing

* then it automatically looks up those words for you in a dictionary and other references showing them while you are reading

* it also tracks the words the user has seen by lemma (dictionary form) grouping conjugations together

* it creates exercises and reviews using the content in the user's library.

* finally because it tracks words I'm able to solve the problem that flashcard apps have for language users; having to choose between single word cards vs sentences that inevitably have redundancy.

Once it's live available I'll updated the thread.

Ask me anything. I'd love to hear feedback from people as well.



What's, according to you, the difference with LingQ? The premise looks the same to me: import your own content (text, audio, video, ...) and support the user in understanding it with dictionaries, conjugation guides, review sessions, ...


LingQ was actually an inspiration for me in designing this. For Parsnip I took inspiration from LingQ, Anki and ReadLang.

I feel like it was just too early for its time though, because of this LingQ has you manually match up definitions that were community sourced, words aren't kept together, etc. So I'm trying to follow through on the ideas that you can see in LingQ.

Also the LingQ UI is hard to use. Instead of refining the ideas that set them apart they've added more and more features. They feel like memrise or any other flashcard app with this one neat reader function.


I've learned a ton while writing it.

One of the most interesting things so far is when aligning parallel texts (such as HP 1 in English and Japanese) one of the primary ways of doing that is looking for correspondence between named entities (proper nouns and the like).




Consider applying for YC's Summer 2026 batch! Applications are open till May 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: