I think the devil is in the details in terms of input. I’ve tried many forms of it (eg. reading on LingQ in Page View, reading on LingQ in Sentence Mode, reading while listening in Page View, reading bilingual books, reading while listening to YouTube videos with the likes of Language Reactor, not to mention the various variations of these methods if you add in repetition as well) and I notice that some are better than others. Not that reading speed / content density is the be all and end all, but a reading speed too low is an issue. This goes back to my very first post, where removing non-language can play a significant role in efficiency. For example, one recent YouTube video I watched had 3k wph, whereas my usual content has 9k wph. That is 3x as much language learning content! This may sound obvious, but the same goes with all the clicks and loading dictionaries when just reading. I’ve noticed over a doubling of reading speed by removing useless clutter.
From another thread over a year back:
If the content is at the ‘perfect’ level, a mostly but slightly modified ‘comprehensible input’ approach works wonders. Really. If you get a native speaker to use pictures and point while saying what they are, you can really learn a lot of vocabulary. It’s not immediate, but gradually you start to accumulate words. After that, you can move onto comic books and the native speaker focuss on describing the images and what’s happening (not dictating the dialogue). During this practice, you can repeat back various words you start to learn and even start to jump in and start contributing to the describing of the story too. The added influence of a native speaker/tutor doing this for you personally, and you can interact with them to indicate confusion is how you get everything at the ‘perfect’ level (if they are good at their job). Furthermore, the fact that you are interacting with a person really adds something extra to the whole situation (especially if you get along with the person). It’s not just reading or watching or listening to some content. It’s a human interaction. This somehow electrifies the situation with increased attention. It’s also a noticeable difference doing this in person as opposed to online. This is probably my favourite variant of all, especially at the beginner to intermediate stages. The downsides are it requires a physical person, which probably costs money.