The “fill in the blank” review questions all seem to be linked to the context in which the lingq was created. For things like YouTube videos (particularly with learner content), these contexts can be wildly unnatural and far more complex than they need be. A random subtitle string scraped off a YouTube video isn’t always going to be a complete or grammatically accurate sentence.
It would be helpful if these types of questions had more natural, standardized, or even AI generated contexts that were more appropriate for these types of questions.