Something I’ve thought about a lot: a mode for pro users that only involves pushing one button. It translates 12 words at a time and doesn’t create any cards while doing so.
Reason: I often use Readlang right before bed or when my cat is on my lap and occupying one of my hands but still want to read my current book, and streching my hand to select a passage is not always easy. Would be nice if you could just hit the right side of the whole screen to move to the next 12 words.
Also sometimes when reading a boring stretch of a particularly long fantasy book I just want to get through it without missing any details (or might be in a noisy place where it’s hard to concentrate) so instead of read-and-click-on-new-words mode it’s more of a select-all-and-then-look-out-for-new-words mode.
If I’m understanding you correctly then while in this mode you’d end up seeing a translation of every single word you are reading. Kind of like you might with a parallel text or an interlinear text. But chunked into sections of 12 words at a time. Is this right?
Are you suggesting the 12 words just because that’s the current limit? Maybe a sentence would be a more logical length for this kind of reading experience?
What if there was an “interlinear” mode where Readlang showed translations of all words (and chunking them into phrases as appropriate) in the text without requiring you to interact with them? Would you prefer that or what you are suggesting? (@Marius suggested something along these lines in a recent thread: We Are Entering the Golden Age of Language Acquisition!)
This would obviously be a big departure from how Readlang currently works, which is to hide the translations until you actually need them. With either of these proposed approaches an extra click/tap could be used after seeing the translation if you wanted to create a flashcard or view an explanation of a specific word.
If we just use an LLM interlinear translation, maybe one way to do it is to push a button and generate an interlinear translation for the entire page. So, kind of like now, but put the translation under as opposed to above but for every word (though above it would be just fine as well if that’s easier).
I figure something like this would be much easier to do with the current tech. Using a human translation in addition to LLM would probably require a bit more work.
It is a bit of a departure from the way the product is but it’s useful for the beginner intermediary level. I will use both: interninear at the beginning/ intermediate and the current product at the more advanced level.
The main downside is that it takes so long to just tap on half of the words on a page.
So maybe the quick and dirty way would be to add a button and write a script that would press on all button son the page for us.
Yeah, 12 words was just because it’s the current limit. A sentence could work but sometimes on mobile a sentence can be greater than the length of a screen.
Interlinear mode would be good too! Yeah, as you mentioned it would not add cards by default unless you tap on them. Although with context-aware translations now for pro users even just chunking into 12 words at a time would probably still produce pretty readable translations. I wanted to suggest something that didn’t take too much effort to implement which is why I thought just 12 words at a time but whatever works!
Regarding both “12 words at a time” and “button to generate an interlinear text”: I’ll bear these in mind as I plan future work. A new interlinear text mode with “click on word to save” sounds intriguing to me. No promises though!
Readlang is not designed to be used on texts where you need to tap half the words on a page (as you’ve probably noticed!). I certainly get the appeal of using tools like interlinear texts or parallel texts to attempt to read these, but another approach would be to find simpler texts to read and gradually ramp up the difficulty of the texts as your skills improve. Have you tried using LLMs to rewrite complex texts using simpler language?
Yes, I’m working my way through graded readers, but the use case of reading more advanced texts remains.
I get that Readlang is geared towards the graded readers approach. What I’ve been suggesting though is to make some tweaks to expand the use case of the tool.
There’s a few methods for reading and learning a language. One is the slower, progressive, graded readers approach. Another is a grammar upfront and reading advanced texts approach. I get that Readland is built for the first approach. But the second approoach remains popular. It’s been around for a long time, there are a lot of books based on it. And for good reasons. Sometimes people don’t have the time to take it slow. They might need to read an advanced text righ now, for a paper they’re working on, or for a book. So they read a grammar introduction and start reading the text they need to read now, with the dictionary. What I’ve been suggesting is to use current technology to help people read advanced texts sooner. Interlinear texts are just a more efficient way to read advanced texts (whatever the reasons people chose to read them).
What I’m suggesting is why not expand the use case to bring these the people who read like that on as premium users.