Looking for a co-founder to change language learning forever

I have an awesome prototype that re-imagines how people learn a language. I think this has a real chance to change the world by making language learning as achievable as learning to read in your first language and as predictable as following a recipe.

I applied to YC last year and got an interview, but it was still early enough in development that I needed to rely on instinct and intuition to answer questions. I’ve finished the initial R&D and am building the final UI for the prototype. The prototype natively supports any pairing between 20+ languages, including English, Japanese, Korean, Chinese, Russian, and many European languages. Practically these pairs are limited by the availability of bilingual dictionaries.

I’ve been meeting with a couple of people a week, mostly language learners, teachers, and tech people, but I am finding people, perhaps unsurprisingly, have a lot of trouble seeing past the curse of knowledge. We’ve all learned a language, so most of us feel qualified to teach it to others without deeply examining our assumptions.

Meeting someone on HN isn’t ideal, but I’m determined to find the right people and I’m willing to improve my chances in any way that seems productive.

What’s my idea?

I’ve studied Japanese for years and constantly wonder, “how close am I to being able to read and understand this book, academic paper, song, or show?” And there isn’t any tool that can answer that question for me.

Natural language processing (NLP) has advanced to the point where I can build something resembling abstract syntax trees (ASTs) for natural language sentences. By indexing the words and original sentences, I can easily track a user’s learning across any number of pieces of content, link to word references like dictionaries, cross-index words to show other examples of usage, create automatic exercises with full context, schedule reviews (SRS), tagging named entities (proper nouns and the like) and provide concrete guidance.

I want to build language servers (natural language LSPs) and integrated language environments (an IDE for language) for individuals. For any content a user adds she will have access to content reader, dictionaries, cross-references, character information (like kanji) if applicable, review exercises, parallel text tools (not machine translation), and sentence shadowing tools.

My prototype takes the books, articles, or web pages you would like to read and creates a detailed index of all the words and sentences; it then finds the fewest sentences that cover all the words and turns those into fill-in-the-blank exercises to study from. You are given real feedback and correction when answering exercises. It also includes an integrated e-reader and dictionaries.

Who do I want to meet?

You must be:

* Deeply curious and technical (but this need not be in programming);

* Willing to ignore established methods/ideas when reality doesn’t match;

* Deeply believe that at every level, the details matter;

* Able to say “No” in order to ship, even for things with obvious immediate value.

I'm currently preparing my YC application for the winter 2023 batch. If interested, please email me at joshua@solarmist.net. Put "co-founder" somewhere in the subject and include:

1) something awesome you've built

2) why would you be right for this and

3) the best time and way to contact (FaceTime/Zoom/etc) you.

Even if you aren't the right person, please upvote this and help us meet.

For a longer answer:

I also know a lot of people and am constantly meeting more who have successfully and unsuccessfully learned other languages. One problem that comes up constantly is the intermediate plateau or death march where it seems like no matter how much someone studies they never make any observable progress.

To be functionally independent in another language you need to be able to recognize and understand on the order of 25k words. Parsnip is designed to target that by finding and tracking your learning of all those words and automating the work of looking up, practicing and recording those words.

Lingq was an early inspiration, but it has many limitations (for example, it has trouble tracking word conjugations, let alone kanji in languages like Japanese) and relies almost exclusively on manual volunteer work. It provides very little leverage to users.

For a purely leverage-based answer, Parsnip will save hundreds of hours of manual work per user purely by automating manual things like creating fill-in-the-blank exercises, vocabulary tracking and dictionary lookups (a conservative estimate is 70+ hours for this alone).

The intermediate plateau is because of Zipf's law. In a 300-page book, there are ~5500 unique words and ~3000 of them occur once or twice. This isn't a big deal for native speakers because a 300-page book is about 100k words (1 day's worth of content), but for a language learner, that might take weeks or months to cover. To go further, that native speaker will probably encounter those words again in ~40 days, but it might be years before that learner re-encounters all of them (having long since forgotten them).

Your time is best spent focusing on the sentences (30% of the book) that contain those 3000 words because they use almost all of the rest of the words.

Yeah, my first idea was automatic Anki cards, but once you try to use sentences and track words, it's intractable for Anki. It's a complex graph problem even before you get to the level of word vs. form disambiguation.

For dictionaries, my initial source is Oxford bi-lingual dictionaries and a few others.

Re: words. Yup. This was one of the top technical challenges I needed to solve for this project to be possible. I separate each of the possible meanings and map them to a lemma or character that uniquely represents each.

For example, 見ました [Japanese to see (past tense)] maps to several things simultaneously:

みる - the lemma/dictionary form of the word

見る - the kanji representation of the lemma

見 - the kanji in the word

みました - the non-kanji conjugation of the word

Parallel texts are a feature that I hope to implement eventually, but aligning parallel texts isn't trivial, but even that (much simpler than machine translation) ends up being a many-to-many mapping.