This is something that will improve over time - as it gets better at identifying longer phrases I can implement rules so it won't omit a neighbouring noun if the phrase contains a verb.
Learning vocabulary and learning grammar can be split apart.
I'd prefer it go the other way, to translate the current language's grammar into the other language's grammar. If you are learning Japanese, for example, having the English grammar match the grammar of Japanese would be a smoother learning experience (I think) than flipping grammar at the translation point.
But in any case, I would want the grammar to match if I am learning vocabulary. I don't want to have to guess, or presuppose the rules of the grammar.
Because it's not teaching grammar? It's teaching words. I already know English grammar: I already am going to have trouble with grammar conversion. But before I get there I need to learn a basic set of nouns and verbs.
Optimising the tradeoff between more frequent repetition of vocabulary versus better quality grammar examples is where a lot of the work is.
The approach I use should work if there are sufficient 'scaleable' ways to improve accuracy and choose the right phrases. Language analysis, NLP, LLMs etc all help here, and there are many methods I can still use and will add over time. I'm very much at a proof of concept stage right now!
This approach won't work if accuracy improvements quickly reach diminishing returns and an explosion of rules and/or human proofing to handle edge cases are required.
I'm hoping for the former!
Does it just replace words on a word-by-word basis? But the ordering of words in English is different from the ordering of words in other languages. How is this not going to teach you terrible grammar?
Looking at one of the screenshots, for instance, it translated "they had met their dead father" as "they had recontré leur mort father." But in French "mort" would come after "père."