I'm currently making vocab and kanji lessons using a transcript for a visual novel game, but I'm still trying to figure out the best way to do this.
Currently I'm trying to split the transcript that there will be ~200-400 unique words per lesson. Each chapter is split into like 3-10 parts like this.
That seems like a lot of words but I don't filter out words from previous lessons, so those would be repeated. Also it contains all the "small stuff" like particles and stuff (I try to manually filter out grammatic expressions), maybe that makes out 5-10% of each lesson.
Do you think that's a reasonable size or should I split it more / merge it? Is there a way to filter out duplicate words from previous lessons in a set? Haven't found it in Japanese Index so far, also it doesn't offer (I think) the feature to "remove filtered words from a lesson".
If I was doing it, I’d worry less about the size of each custom list and more how coherent each one is. Custom lists can be large or small in renshuu, and it doesn’t really seem to matter for usability. But tying each list to a concrete entity such as a chapter or subsection will make them easier to manage.
Well the game is split into 5 cases and each case has sections where you change locations. It's not always linear, so sometimes you would be able to different locations in a different order but with the transcript I'm going at least it doesn't seem to break midway through. I'm trying to put the progression in the description without giving too much spoilers.
Visual novels have a lot of text so 200-400 words per lesson, even if you filtered out particles & basic words, sounds about right. Whether you're okay with that size depends on how you plan to use those lessons yourself (or how other users might use them, if you want them public for the community). Also keep in mind that someone who's at the level where they're comfortable tackling a visual novel probably already has a solid vocabulary base, and words will be reused between chapters, so 400 words per chapter won't necessarily be 400 NEW WORDS TO LEARN per chapter and it's not as daunting as it sounds (though, anything over 200 is still pretty big for a Renshuu lesson, and I'd recommend removing the "small stuff" and any beginner language as bloat).
The way I make my media lessons is I make frequency lists of the thing overall and each individual chapter first, subtract particles & grammar expressions & basic vocab, then pick out the top most frequent words across the whole thing and any words that show up in the majority of the chapters. That list becomes my "Core" lesson for that media, i.e. the must-know words to get anything out of it. After that I go through and make multiple lessons for each chapter (minus the words that were in Core): an "A" lesson full of the most frequent words in that chapter, and a "B" lesson full of any other words that come up that are potentially useful or that I don't recognize. This way there's a clear separation in how useful each set of words is for any other users who might study with them, and I have a lesson in each chapter that I can add random words to while I play/read/watch it without bloating the lists that are more important to comprehension. It's not necessary to capture every single unique word in the lessons, so I just focus on sorting them between what's objectively useful (i.e. frequently repeated vocabulary) and what I find interesting myself.
In terms of filtering out terms from other lists, I do all my list-making in a spreadsheet first and use formulas in the spreadsheet to flag duplicates, rather than trying to control it through Renshuu. You could use the Japanese Index to load 2 lessons at a time and find the terms that are common to both lists, then manually remove them from each from one or the other using the + button, but that's a lot more time consuming.
Oh that's very helpful, I don't know if I'm able to implement it but maybe for my next thing! May I ask how you create those frequency lists, do you use a specific tool?
I use cb's Japanese Text Analysis Tool to generate frequency lists. It's kind of old and unsupported and janky now (IIRC it fails if there are spaces in any of the folder names in the executable path), so I'm not sure if there's anything else around that works better? But it's useful in that it allows you to exclude lists of words from the results (like particles & grammar) and the frequency breakdown has a column for cumulative % comprehension, so you can see e.g. that learning the top 200 most frequent words will bring you to 50% comprehension.
I think 200-400 is quite large for a lesson, but if you're covering, say, 95% basic conversational words and 5% unusual words, I think that could be fine. I would personally sort the words based on their use and put the common terms in one lesson and the unusual, story-based words in another.
You may want to decrease the lesson sizes if you want more content to review each day (I think 6 lessons of 100 words will have way more review content than 2 lessons with 300 words.) The higher numbers can be super intimidating, though. (If I estimated the review sizes incorrectly, please correct me.)
It seems that this (frequency sorting) is a good strategy that was also proposed earlier. Though I still struggle a bit with splitting because I want to have a lesson cover at least like 10-20 min of reading time so somewhat chronology needs to be preserved. At my level I know like 75% of the words already though so I guess it should be okay? I guess in later parts some specific vocab might be repeated.
I guess in later parts some specific vocab might be repeated.
I think active schedules may not require you to review the same word twice in one day, but I'm not completely certain. I have the N5 vocab and Renshuu Japanese basics schedules active and I've seen the review numbers drop for N5 vocab when I study the overlapping terms from the Renshuu class schedule.
