| I think you would find it difficult to conjugate _the translation_ as the conjugation tables/rules differ between the languages and the mapping is not one of one. For example, English only has 'you' but French has both singular and plural forms. Russian has the declension as well (Именительный, Родительный...) I think it would make more sense to normalize at the _source_ language as I wrote originally. See also the other comments in the same sub-thread. As to the other ideas, like I said "10 pages". And a lot of them about the system that really accumulates the knowledge about you to give you the best training. See, for example, https://solaresearch.org/ for academic foundation of that and https://french.kwiziq.com/. But here is a couple of other thoughts. Feel free to contact me if they spark anything: * Synchronized text and audio (e.g. Amazon Whispersync). Has been done by many people but was usually done by hand (was too expensive) or by text-to-speech on the fly (which was awful at the time); yet it was always welcomed. TTS is now much better, but also maybe some work can be pre-calculated (rather than immediate). * Use the Google API mentioned before (syntax tree) to extract sub-sequences (named entities, stable expressions, etc) and present those instead of individual words; this helps to see them over and over within different sentences. Even showing a verb and its dependencies is useful. This also allows to include grammar references for tenses/irregular verbs/expressions, etc. * Dictionary entries by the normalized form (lemma) of the verb, noun. Flashcards that first show the multiple contexts line of the same verb in the same form, then same verb in other forms and only then translation. * Color code the text being read by word type. So all verbs as red, all nouns as blue, all adjectives as green, etc. Also see https://langliter.com/, they do (and lemmas) that as "highlights". * Color code the text by words known. Green as those learned already in that exact form. Yellow as those learned in different form (different tense, different conjugation, etc). Red as those unknown. * Pre-flash. Analyze text (small chunk), compare again flashcards and compile the list of words/expressions and let person practice them first. Then read the actual text with the hope to increase the speed of comprehension. * Slightly further along are things like "Words of the day" emails that analyze your flashcards to give you more words in the same category (arm+leg=>offer head). Or touch typing game using your own flashcard words. Or conjugation table training using collected context. |