So this isn’t theoretical, really, or supported by a sample size of more than one (me). But I found it an interesting realization.
I’ve been trying to learn/acquire (mostly what’s available is geared toward learning, not acquisition) a certain Native American language for some time now. Like most traditional students of language, my efforts were sporadic: I’d find a new PDF online, or some new source of language, and get enthusiastic about it, but never got anywhere because there simply was not enough repetition or really anything more than a few short, overcomplicated dialogues. (You can hardly even find a verb conjugation chart, that’s how bad the situation is!)
So right now, I’m involved in a new online course teaching this language. It’s still pretty much learning-based, with a heavy emphasis on flashcards. I wanted to make my flashcards as CI-friendly as possible, so I have taken to writing a large number of sentences using the new words, getting a teacher to find the mistakes, then using these large sets as flashcards. It’s actually been — well, I wouldn’t say effective, exactly, but much better than previous efforts. While the pace of progress is glacially slow because the input is so non-rich (it’s correct, but not varied and broad, so I only get precisely what structure and vocabulary is on the list), at least there’s actual progress.
One of the things on the list is a seemingly endless list of pronouns. This language has at least 45 pronouns that occur in roughly four different forms, so — based on the method being used by this course — there’s a lot of pronouns to memorize. I already had flashcards of all the pronouns from my previous efforts, but they worked no better than before.
Finally, I simply abandoned the pronouns and replaced them with flashcards based on the drill-and-kill concept of “fill in the blank with the correct verb form”. Since we have only been given one verb (!) one verb the past month of class, it wasn’t hard to choose. I made up a set of cards with that verb stem combined with those pronouns, but instead of using pronouns or similar non-specific language in the English prompt (“he and I”, “those three males”, “those two females”, etc.) for some reason I randomly assigned names to the “people”: “Bob and I”, “Sally, Anna and Brenda” and so on.
These are not names that are meaningful to me (though I’m sure I could use fictional characters if I wanted to). They’re just random names. And they add an added cognitive step to the processing: I still at first had to stop and think “Bob and I, that’s the exclusive we form, so…” when coming up with the answer in the target language.
But after a couple of times through the stack, I found something strange. I was starting to “feel” the pronouns. Much more strongly than I had doing the same thing in the past with just the meanings and no names.
When I train TPRS teachers, I always emphasize that it doesn’t matter about whom you ask a story or have a conversation. If it’s the real students, that’s fine. If it’s customized input talking about fictional characters they like, that’s fine too. But TPRS is always humanized in some way (well, sometimes “animalized”, I guess…) with names or at least characters. It makes sense that humans would “get it” quicker when actual communication is involved. Maybe throwing some names in there moves language, even in the mundane form of really short, almost meaningless sentences, into the realm of communication instead of memorization?