I’ve tried out using Anki for language learning on two separate occasions.
Once was using the Fluent Forever method with German where you go through a frequency dictionary and add vocabulary cards (recognition, production, and spelling). I did this when I was at an A2/B1 level for maybe around six months.
Production front: picture of an onion
Production back: die Zwiebel | die Zwiebeln + audio pronunciation + IPA
Recognition front: Zwiebel
Recognition back: picture of an onion + die Zwiebel | die Zwiebeln + audio pronunciation + IPA
Spelling front: picture of an onion + audio + IPA
Spelling back: die Zwiebel | die Zwiebeln
Problems included:
- Pattern recognition
- Huge amounts of time spent creating the cards
- Failure to recognise/produce the word when speaking to someone or reading, etc. (i.e. in a real context)
I was using Anki as my sole method for language learning then. If you do other language learning methods simultaneously (eg. input, etc.), it goes without saying that your experiments might not pick up the true effects.
The positive effects were:
- I rote learnt the genders and plurals of all the nouns
- I mostly rote learnt the various conjugations for common irregular verbs (eg. fährt | fuhr | ist gefahren)
- My pronunciation of individual words was very good (as I’d speak the word for every card out loud)
- I also did some minimal pairs, which I think was quite useful too (but can’t say with certainty)
My second trying of using Anki for language learning was doing the Mass Immersion Approach (MIA) aka the Refold method, where you do something similar to above (but doing only the recognition vocabulary cards because your focus is initially only on recognition) while watching TV series and movies without subtitles (they call this immersion, which is just Stephen Krashen’s comprehensible input of guessing from context). They also want you to use some mnemonics to help remember the words. I did this for Russian at an A0 level after going through a few lessons on DuoLingo.
The problems with this were:
- Cards were incredibly hard to get correct and I just kept failing card after card
- When a new card was added which was similiar-ish in pronunciation/spelling to an older card, I would continuously get both cards wrong because they would never appear side by side so I could compare the differences between the two words (think of two words like sand vs. send)
- (On the ‘immersion’ side, guessing meaning from context when I knew so few words was incredibly rare.)
I didn’t get very far with this method. I can’t remember how long I did it for but perhaps only a month or two. It is an incredibly frustrating method as I just kept getting card wrong after card wrong due to all the words sounded or looked similar. Re-reading the Mini Stories on LingQ is so much more satisfying and felt like so much more progress than all the words mixing together in my head using the flashcards for a language with very foreign sounds as a complete beginner.
The ways how I used flashcards are not ways I’d recommend to anyone. Sure, the way you are using flashcards is different (fill-in-the-blank cloze cards), but I still don’t imagine that if you used this technique in isolation, you’d have great transfer unless you are already at an advanced level. I imagine sentence cards are better than isolated vocabulary cards, but from my experience and thoughts, the transfer of understanding and being able to use the word in many contexts comes from variability (the same concept is called variable practice in the scientific literature, though I haven’t read much about it). Unless you are using Anki in a rare, unorthodox way, you don’t get variability from Anki at all. The reason you wouldn’t be experiencing a failure to recognise and produce words outside of Anki is not because of Anki, but because you are getting variability from LingQ (i.e. input + ability to look up definitions). I just think that if input is doing the heavy lifting of deep understanding and application of the words, what benefit is Anki providing? And is it worth it? Can I get the same benefit from other techniques?
My methods to grow my vocabulary are reading while listening to YouTube videos with dual subtitles for Russian and reading bilingual books in Italian (because my level is higher in Italian, so I need content with more unknown words). These are methods which do many things at the same time (drilling in words you already know, exposing you to grammar, learning cultural nuances, etc.) and that’s part of the reason why vocabulary remains a key weakness as other aspects of the language get better at the same time as I’m focusing on it.
To learn grammar, my preference is either a grammar textbook (including all the exercises) or grammar videos by YouTubers, but I kinda procrastinate these things as they are kinda boring. Plus input to really drill all that grammar theory in due to encountering it in a large variety of contexts and usages.
To improve speaking, you need to speak a lot. For Italian, if I randomly meet an Italian, I switch to Italian and try and meet up with them again. I also text or sometimes send voice messages to people I know who speak Italian. Alas, I do this minimally for Russian. To see a level up in my Russian, I need to get around to hiring a tutor/conversation partner, as I’m at a level where I would be a frustratingly incompetent conversation partner. Though, production of Russian is still done when I’m trying to speak Italian, as sometimes Russian words pop into my head and I have to ignore them. 
For pronunciation (which is not great in either Italian nor Russian as I’ve never focused on it), I should get around to doing lots of listening to a sentence and repeating it out loud. I was considering shadowing as a method, but after hearing Alexander Arguelles speaking Russian, I reconsidered it as a good technique to improve pronunciation.