I have an idea for a language learning app and web platform based off some pet theories I have about learning. Before I go and try to make this I will definitely look into the existing schools of thought more, but I thought it would be interesting to record my starting assumptions to be able to read back later.

The main idea is to create a body of texts and exercises to be as efficient a means of learning a language as I can manage. It would have minimal explicit lessons on grammar and mostly rely on learning via inference. Questions would have to be carefully designed to properly challenge the aspects that create the most growth, rather than optimizing for rote internalization or other undesirable behaviors. It would feature a pool of vocabulary that starts from the very basics and "bootstraps" into new terms. It would focus on high-usage, conversational material in order for it to be as applicable as possible. Ideally this would just serve as a bridge from starting to the point where you can just about comprehend native material to a manageable level, at which point you would simply transition into only using that.

theories

My working theory is that someone who is decent at self-directed learning will have the most efficiency in language learning by far by just signing up for as many iTalki calls with native, non-credentialed speakers as they possibly can, and just naturally expanding their pool of expressions as needed as they communicate about themselves. That is, no lesson plans, no flashcards, and minimal grammar instruction outside of basic and widely applicable rules. Simply starting by learning to introduce yourself, then ask basic questions about the native speaker, and gradually learning the means of expressing yourself, to me seems like the most natural and effective way to learn to speak. We hardly understand the formal rules to our native languages, after all, and I feel that our natural means of apprehending something like language, is best done through the "natural" means (I have the feeling that this point is brought up a lot and rather polarizing between grammar proponents and detractors...).

Naturally, even if you achieve a measure of comfort or even fluency in this manner, you would still be illiterate. But this is another assumption that seems reasonable to me: becoming literate is relatively trivial in languages that have a phonetic alphabet, if you can speak it. I mean, all you have to do is sound out the words. You don't even needed to know where the boundaries of words are necessarily--after all, in speech we often don't have any audible pause to delineate words.

Consumption of native material, once one is at a reasonable level for it, would likely be helpful as a supplement and means of solo study. But a good lookup tool/dictionary on hand is all I believe should be used, rather than making flashcards out of everything. I don't have as certain an opinion on notes, but certainly if they are used a deliberate observation of some sort should be made, rather than just noting content, such as "mortgage: real estate loan - mort -> death, gage -> pledge", rather than simply "mortgage: real estate loan". The notes would not be something to look back on as much as just a means of mkaing a deliberate action to underscore a connection or observation of some sort that you have made.

why reading then?

If I have the belief that conversation with native speakers is the most efficient method, why even bother with texts at all? There are a few things that this helps to address: the most straightforward one is cost. On iTalki, it seem like you can have these sort of calls with native, but otherwise uncredentialed speakers, for 10-15USD an hour, which can add up quickly. I don't know how much time learning a language takes, and it varies in any case, but this will at minimum cost 1K USD--at 10 USD/hr, that's only 100 hours of practice. Reading and exercises can provide supplementary practice for much cheaper.

Additionally, scheduling a decent amount of calls might be difficult for any number of reasons relating to logistics, availability of either parties--maybe some native speakers are much better at facilitating this sort of learning, so you would only want to call with them, I don't know. This is all hypothetical still.

Then there is the issue of non-completely phonetic languages in Chinese and Japanese, which inconveniently are two that I want to learn. For these languages, being able to speak isn't enough to be able to read, so deliberate reading practice is necessary.

Additionally, people have varying degreees of interest in speaking to strangers, all while hardly being able to talk themselves. Even if someone does some calls, they might not want to undertake that for several hours a day, switching speakers, even if cost wasn't an issue at all. Having the solitary exercises here is a means of alleviating that.

Finally, it can provide more of a structured pathway than native conversement. As mentioned earlier, it's on the learner in these calls presumably to determine what they should try to focus on and do. If one isn't as accustomed or proficient in self-direction, I could imagine that going pretty poorly, potentially. If I design the program well, it should naturally force you along a highly useful route where if you just keep doing what it asks, you will learn fairly effectively.

conclusion

I guess that about does it. Will be interesting to see how right/wrong I was, if I go through with this project at some point.