Thank for your recommendation of Glossika I've never heard of it before but it looks good. I'm a beginner I've been using Pimsluer (for Mandarin) for a while while.

Although it's not perfect. I prefer Pimsluer over Micheal Thomas Method, which I've done previously. I'm just using it as a source of comprehensive input to develope my ability to able to understand spoken Mandarin (which MTM was awful for).

I'm sure I could use other programs similarly. Though Pimsluer probably has the most material. That might not matter if I get good enough to use films without subtitles and podcasts for my sources.

Glossika looks like it's geared more to how I'm trying to learn. I'll probably give it a try at some point.

>It is the current generation's Rosetta Stone

To be fair. Rosetta Stone got it's "reputation" because of it's insane price point comparative to what you learnt. Duolingo on the other hand is completely free.

It’s freemium, which I don’t feel is “completely” free.

There are only marginal benefits to Duolingo plus that can't be obtained with a proper adblocker. Many people who choose to subscribe admit they only do so to support the company.

I think Duolingo is fine if you're looking to learn a language that's really close to yours. I know French, took up Spanish and within a month of practice I could cobble simple sentences in a casual conversation or text my Spanish-speaking friends. The approach where you just "figure out" from trial and error works well when your source and target languages kind of work the same way so you can just feel your way through it. On the other hand, I suspect that if I wanted to learn, say, Russian or MSA, I would probably be better off with actual formal courses.

I have only given Rosetta Stone and Pimsleur a cursory attempt but they seem awfully slow and expensive for what they do. Duolingo, for all its flaws, is free and fun.

As for your "purely from marketing" comment - Duolingo solely relied on word of mouth for all of its early phase, apart from a single TEDtalk. It's true it may not be as good as many alternatives, but as far as the free and fun factor goes - it is simply unmatched.

Interesting. I strongly dislike Duolingo (because it seems very focused on passive recall, aka multiple choice), but think pretty highly of Rosetta Stone.

I haven't looked at either of Glossika or Pimsleur, but Glossika's page title is "AI-based effective language learning" which immediately reeks of marketing blather. Why do you recommend them?

Yeah, their new branding is a little strange. But the original Glossika method (which the new software still essentially does) is to teach a language only via phrases - not grammar trees, isolated vocabulary words, or anything else.

I’ve found this approach to be more fruitful than the abstract one most books and programs take.

> the original Glossika method (which the new software still essentially does) is to teach a language only via phrases

Based on the screenshots, it looks like they maintain a "memory level" for each phrase and presumably do spaced repetition to keep that level above a certain target. Am I right in assuming that if you fail a review because you forgot what one specific word means, you'll get the exact same sentence next time, rather than a different sentence using the same word?

I have a system for personal use that does the latter, but I haven't yet found a good interface for indicating why I got the sentence wrong. My current implementation is to have a huge list of checkboxes for every aspect of each word (dictionary form, spelling, pronunciation, etc.) that could potentially be forgotten independently.

How's it built? What languages do you use this with? My biggest concern before starting something like this is the I've thought about scripting the generation of example sentences for myself, but I'm unsure I'd be able to do it in a grammatical way (since I don't know the language, a catch-22).

When I started building it for Japanese, I knew a little (half-forgotten hiragana and katakana, also many kanji because I speak Mandarin), but most of the language-specific features come from existing tools I've just glued together.

I have two main sources of example sentences: 52114 from Tatoeba https://tatoeba.org , most of which have translations, and 114591 extracted from public-domain ebooks from Aozora Bunko https://www.aozora.gr.jp/ , some of which use rather archaic language.

I process each sentence using Kuromoji https://github.com/atilika/kuromoji to segment it into individual words and get details like part-of-speech tags, dictionary forms (lemmas) and pronunciation for each one. Then I add them to a SQLite database.

Some sentences on Tatoeba have audio recordings contributed by native speakers, but there aren't so many for Japanese, so I also use Open JTalk http://open-jtalk.sourceforge.net/ to get a robotic pronunciation. That pronunciation may differ from the one determined by Kuromoji, which I use as a consistency check.

I have tried various criteria for determining what to learn next. One that worked well is to keep track of the least frequent unknown detail of each sentence (e.g. the spelling of a particular word) and pick the sentence where that frequency is highest. That way, sentences with common words come first. Lately I've modified it a bit to also consider how many other sentences have the same least frequent unknown detail, so I always have multiple different examples available to help me generalize.

Reviews are scheduled for each detail based on the time since the last review and the time when I last had to relearn it. I've been collecting data to later tune the review schedule, but right now it just doubles the interval between reviews every time I remember correctly.

I used to just pick a random sentence for each detail with a scheduled review, but that would sometimes randomly schedule the same sentence often enough to be noticeable, so I've changed it to prefer sentences I've never seen or not seen in a long time.

The two kinds of review are dictation, where I listen to the recording and write it down on paper, and reading, where I read the sentence aloud. In each case, I manually check whether I made any mistakes and indicate them by unticking a checkbox (as mentioned in my previous comment). Those details will then be scheduled for review more frequently.

After a year of daily use, there are now 35611 sentences in my database which could potentially be scheduled for review (which means that, theoretically, I should be able to understand them), 21.3% of the total. The ebook which I should understand the best (35% of all sentences) is the Japanese translation of Karel Čapek's play "Rossum's Universal Robots" https://www.aozora.gr.jp/cards/001236/files/46345_23174.html 35% understanding is ... not enough to allow me to read it from beginning to end, but occasionally I get a nice sentence e.g. about the worldwide robot uprising.

I think it should be possible to adapt the system to work for other languages, but it would require significant refactoring to remove the assumption that everything is in Japanese.

Thanks for the detailed reply! I'd be interested in reading more technical details about the whole pipeline if you have written about it.

One question that sticks out in particular: how do you actually perform the reviews? Did you create a custom GUI for it or are you using an off-the-shelf software?

My comment above was the most I've ever written about how it works, except for one very long comment in my database migration code, which I had to write to remember how it's supposed to work.

I created the GUI in Qt. It's not beautiful, but gets the job done. I decided to finally open-source my WIP and put a screenshot in the README: https://github.com/Yorwba/alphabet-soup

The comment on database migration is here: https://github.com/Yorwba/alphabet-soup/blob/4f68057ed5d7dbf...

If you don't care about all the Japanese-specific stuff, you could try running make data/tatoeba_sentences_XXX.csv where XXX is the ISO 639 three-letter code of the language you're interested in. That collects all the sentences in that language contributed to Tatoeba by native speakers and puts them into a TSV file.

Thanks!

> Am I right in assuming that if you fail a review because you forgot what one specific word means, you'll get the exact same sentence next time, rather than a different sentence using the same word?

Correct. It’s just based on the sentence, not the words within the sentence.

I agree, I'm happy that I managed to buy two courses with books and mp3s before they switched to the online version.

I enjoy Pimsleur, especially for learning bits of a language before travelling somewhere. It's fun to listen and practice with my partner. We both sound silly, and we can encourage each other to improve our pronunciation.

I think that's the hard part about learning languages: it works way better if you don't do it alone, and that's why apps can be pretty limiting.

can you elaborate on what the differences are, in terms of language presentation and recall?