[T-953] Is this hack for increasing the frequency of T-1 sentences legit?

Dear Rine,

This is concerning my complaint from the other day about active immersion feeling too stilted and tiresome.

I got an idea today to make the process more fluid.

The problem was that long sentences messed up my flow.

Whenever a long sentence came up during a TV show, I could either just skip it… or I could try and look up all the unknown words in it.

If I chose to skip, I would get a long succession of sentences I didn’t understand and also I would have a long unproductive stretch of not mining anything.

On the other hand, looking up everything also took a lot of time… and the utility was unclear, because in the end I still couldn’t add the sentence in my SRS because it wasn’t T+1.

Last time I wrote about this, my idea was to try and add other sentences to cover for all the unknown words.

But my sentence search engine of choice (this one) kept failing me, returning 0 results for what were supposed to be fairly common words.

So out of frustration I came up with another approach:

Chop up the long sentences and mine the resulting sentence fragments.

It’s so obvious when I write it like this.

It’s consistent with the principle from the 20 rules too keep items atomic.

And it’s a practice that was encouraged by Japanese-learning OG Khatz in one of his old blog posts. I don’t have the reference at hand but I can produce it if you like. I just happened to remember it today, randomly.

In practice, this is what I have been doing today and what I mean by “mining sentence fragments”:

Say a line of subtitle goes like this:

そのかわり今まで裏に流していなかった作品を渡してやる

In it is a whole bunch of words I don’t understand.

But instead of skipping the sentence altogether and waiting for a nicer one, I make items for each clause/standalone phrase within it, like so:

そのかわり

裏に流していなかった作品

作品を渡してやる

Note the overlap between the second and third one.

For me, each of these three fragments is T+1.

Now I only got this idea today, so I didn’t get to test it with actual reviews.

But my conjecture is that I could perhaps also put another card with the entire sentence, and schedule it some time in the future, say, 10 days from now.

Only downside of this approach is that the Migaku plugin will still extract the audio from the entire sentence. But I don’t mind that, if anything, it’s extra context.

To recap: the idea is to break up long sentences into shorter clauses in order to be able to mine more T+1 material per immersion session in order to make the sessions more engaging.

You think this makes sense?

P.S. My doom emacs keeps crashing when using evil snipe to navigate (eg. g s j). It works for a while then suddenly it crashes.

What I did/planned today