ryanheise's experiment log - Experiment 5 (Japanese)

[ryanheise]

I've been working on a way to track my listening progress. I came up with this scorecard which I fill in while doing a listening session to record how many lapses of concentration I had, or in other words, how many times I had to listen to a segment again:



This is for a session dated 27/09/2019 in which I listened to a portion of 消滅 from 02:40 to 03:37 (57 seconds). The first row of numbers break down how many times I had to repeat any segment of audio, where I used AudioWorkBook to cut the audio up into segments of between 0.5-3 seconds in length. There were 7 segments that required a repeat hearing. Of those 7, there were 2 that required a second hearing. Of those, there was 1 that required a 3rd hearing, and that one also required a 4th hearing. Breaking it down this way strikes a balance between the amount of useful data I can extract out of a listening session, and how much data is actually practical to write down without being too disruptive. (Although I'd welcome any other ideas on what I should be tracking!)

It was 4 months ago when I began this experiment and took the first measurement (yes, it's still going!). Today I decided to go back and take the same measurement on the exact same segment of audio, and the results show a staggering improvement:



Note that I hadn't listened to this particular audio in the intervening time so as to simulate the conditions of the first measurement. I've also been doing the same attention stretching exercises every day for the past 4 months, working my way through different podcasts, although midway through I ended up writing my own program to use instead of WorkAudioBook, since it made it easier to build into it the desired behaviour for my exercises.

I think I'll sign off this experiment as a success. I haven't been able to measure improvement in my language exchanges yet since my language exchange partners have been either busy or on vacation (and I was a bit lazy in finding more exchange partners), but I certainly feel like my listening attention span has improved, and look forward to getting back to language exchange. If not, I'll start planning a new experiment that can be done independently of language exchange.

[ryanheise]

Reflections on Experiment 2

I'd like to journal a few of my reflections on this "expanding your listening attention span experiment" while they're still fresh in my mind.

To briefly restate the listening exercise, we use a tool to cut the audio of a podcast or audiobook into short segments of around 3-4 seconds and listen to each on repeat until comfortable before moving on. After reaching the end of the section you want to practice (around 2 minutes in length), we return to the beginning but cut into segments that are twice as long. After that, we again start from the beginning and double the segment length yet again. And so on, each time doubling the segment length. We do 4 rounds in total, and in the first 2 rounds we also verbally recite back the phrase from memory (while the segment length is still short enough).

1. Overcoming the fixation on one missed word

It became apparent early on that the main obstacle to expanding my listening attention span was where I would get fixated on a single word in the sentence that I didn't understand immediately, and meanwhile would miss the next 15 words due to that fixation.

One "micro-skill" I developed to cope with this was to forget about the one word I didn't know, or couldn't bring to memory in that moment, and just keep on listening. Sure, if I were to pause the audio and think on it for 10 seconds, I might remember the word, but what use is that if I miss the next 15 words? I end up understanding much more by removing this fixation.

2. You get more used to it over time

If you stick with this approach, your mind gets used to keeping up with the pace, and in this more relaxed mental state, you ultimately find it easier to bring to mind those words as you hear them. You won't pick up on everything 100% of the time, but worrying about perfection is counterproductive so it's a no brainer to just focus on keeping up with the pace.

3. Chunking is the way

Another "micro-skill" I developed was to perceive the meaning of what is being said in chunks, or groups of words where I hold a slot in my head for the general meaning of that chunk as a whole, rather than remembering the precise individual words. This higher level thinking becomes possible once you get over the individual word fixation since you can then start to notice the structure of the sentence and use that as a scaffold to absorb meaning. In the same way that it is not effective to perceive a word in terms of its individual letters, it is not effective to perceive a chunk in terms of its individual words.

4. Listening to sentences in context is powerful / The "story" technique for improving memory

This could be the most powerful thing that I discovered in the experiment, and it specifically had to do with the fact that I was stretching my listening beyond the sentence unit to two sentences, four sentences and onward.

For most of the experiment, I was studying new audio content with new vocabulary that I had to learn. Of course, there is a plethora of memory techniques advocated by different people for memorising vocabulary, such as mnemonics, roman room, other kinds of association etc. You name it, someone's used it for vocabulary. But merely putting a word or even a sentence into the context of a "story" is a very powerful technique because if you study a story and you can remember what happens next in the story, you can use that context to help you remember the meanings of the words. And as it turns out, remembering a story is something that comes quite naturally. Prior to this experiment, I had been learning sentences in isolation through spaced repetition (Anki-style), but would regularly forget the nuanced meaning of a sentence because I didn't have the context. But with the kind of time stretching experiment I did, I was remembering every nuance on every repetition because every repetition occurred in its original context.

5. Collecting data

I recorded data throughout the experiment and was able to analyse how many reps I typically needed for each of the 4 listening rounds. Typically each successive round would require fewer and fewer reps, dropping off quickly and approaching zero. I hope to use this data to inform a future experiment.

Conclusions

All of the micro skills I developed for listening in the process are probably not novel in any way, but even if more experienced language learners had guided me that X/Y/Z are the skills I needed to work on to increase my listening attention span, I doubt that it would registered in my own head without the actual attention span stretching exercises themselves. This gave me the playground in which I could practice and notice and improve.

[ryanheise]

Experiment 4

I experimented previously with Poly-glot-a-lot's 1-on-1 language exchange method, and what I loved about it is that he laid out some specific procedures and activities and rules of engagement for how to have a language exchange session that leans towards noticing rather than explaining. Comprehensibility was increased through visual aids, and memory was enhanced through a quick listen/speak cycle.

But for me the limiting factor was the 1-on-1 part. If I could arrange a meetup with at least a couple of speakers of the other language (e.g. it could be 2 native Japanese speakers and 2 Japanese learners), this could open the door to new kinds of language exchange approaches.

What are the problems?

Given my primary way of learning is through listening, when my Japanese language exchange partner asks me a question, what I really would love to hear is how another Japanese person might answer that question before I give my own answer so that I can mimic some of their sentence structure and even intonation, but the 1-on-1 format doesn't give me natural exposure to that sort of interaction. Various other nuances such as the "Ums" and "Ahs" would not occur if the same person were answering their own question rather than a completely different person who reacts authentically on the spot.

What will I try?

I will develop the details as I go, but in general, the goal will be to practice the following types of exchanges.

Pattern 1:

• Japanese person A asks a question.
• One or more other Japanese people answer.
• One or more Japanese learners answer.

Pattern 2:

• A Japanese learner asks a question.
• Each Japanese person gives their response.
• Each Japanese learner gives their response.

Other useful patterns might be discovered through this experimentation process.

Since this is a language exchange, we also do the same in reverse to practice the other language.

In preparation, learners will come with a set of discussion topics they would like to practice using these patterns, and a set of language features (it could be specific words or specific grammatical patterns to practice). These topics and language features could be written on cards or done digitally, and for each round, you pick a topic and a language feature to focus on. The native speakers will ask and answer questions in line with that discussion topic and making use of that language feature, and learners then have an opportunity to mimic the sentence structures and phraseology they hear in their own questions/answers. The idea is to get a quick turnaround time between the learner hearing something and being able to use it.

Note: Although I am really interested to try this out, the main obstacle is how easy or difficult it will be to get a group of language learners together, either via a meetup or some other means. If it turns out not to be practical, I'd be open to trying it with Skype or even try to emulate most of this via text chat.

Edit: This experiment has been put on hold due to the coronavirus situation.

[ryanheise]

Experiment 5

This experiment was inspired by an interview with Richard Simcott that I listened to on John Fotheringham's podcast, The Language Mastery Show.

I will quote the relevant except from the show before I get into the experiment:

=== BEGINNING OF EXCERPT ===
Richard: One thing that I read somewhere, and I think it's true actually, is: Language Learning is Over Learning. And that makes sense to me. It is over learning, it's repeating things over and over again.

John: Which I think is one of the reasons why children, you know granted their brains are wired a little bit differently until puberty, but I think a lot of it is that they don't mind repetition. I mean, my little nephews will watch the same cartoon 50 times, and love it every time. I'm sure you know as a father.

Richard: oh yeah yeah.

John. Where as adults, we watch something twice and we're like, yeah.

Richard: And not just the same episode. They will rewind to hear the dialogue - my daughter will rewind to hear the dialogue until she knows it off by heart, and that is interesting for her, and then she'll say it, and see what the reaction is from me or whoever's listening, to see what the reaction is to these new sort of combinations of words that she's learned. And people talk about children being really, you know, they're so quick and as adults we're a lot slower. But I don't know if I really believe that. I think that children just repeat things so often. As an adult we think, if we heard the same thing, like more than five times and you go "Oh I'm so stupid. Why can't I remember it?" Well you're not. You need to hear it a lot of times to be able to actually use it. There are some words that will stick in your mind because they they mean something to you but read over the same things over and over again. You need to do it. And the thing is as adults we get bored. I think what you're saying is right. We get bored because we think "Oh I should know this by now. If it were interesting, I'd remember it." But of course we don't. And it could be because the words are completely alien in a language. It could be because the sounds are completely alien, I think that is one of the big factors that you know makes it difficult to remember words.
=== END OF EXCERPT ===

The goal of this experiment is similar to Experiment 3 in trying to find ways to help transfer my passive language ability into active language ability. While Experiment 3 aimed to accomplish this through language exchange, Experiment 5 aims to accomplish this through over learning.

What are the problems?

• My current diet of intermediate input alone is not transferring into intermediate speaking ability.
• I struggle to find the right expression to say in conversation, and fail to even fall back to simpler language.

What will I try?

My theory is that I simply need to get my memory to the point where expressions and phrases I've heard before are actually flowing through my head, and I can easily hear them in my mind. That is, I can easily imagine someone else is saying it to me in my mind. I believe some people do have better memories than others, and have a higher than average ability to remember and reenact lines from movies perfectly, and this is a skill that language learners can leverage. But even if you don't have a particularly good memory like this, I believe that you can compensate for that simply through a higher volume of repetitions, and the end result should be the same.

I will try to achieve this through over learning of simple audio material. That is, listening to the same audio over and over again, and speaking along with it, until I can actually say the dialogues from memory. It is important that the dialogues be simple. For example, learning more advanced vocabulary or sentence patterns will not address the problem. I already know quite a lot of these, but the goal is to develop fluency of output, and I need to target the input for this experiment to the level where my "output" actually is, which is the beginner level.

I hope to see that with enough over learning, common ways of saying things will come to mind more easily because the "after images" of the dialogues I've heard many times before will continue to stay with me and be available to me when I need them.

[DaveAgain]

I will try to achieve this through over learning of simple audio material. That is, listening to the same audio over and over again, and speaking along with it, until I can actually say the dialogues from memory. It is important that the dialogues be simple. For example, learning more advanced vocabulary or sentence patterns will not address the problem. I already know quite a lot of these, but the goal is to develop fluency of output, and I need to target the input for this experiment to the level where my "output" actually is, which is the beginner level.

I hope to see that with enough over learning, common ways of saying things will come to mind more easily because the "after images" of the dialogues I've heard many times before will continue to stay with me and be available to me when I need them.

Yizy uses memorization as part of his routine.

https://forum.language-learners.org/vie ... 96#p169070

[ryanheise]

Yizy uses memorization as part of his routine.

https://forum.language-learners.org/vie ... 96#p169070

Thanks for pointing me to his log. I hope I can do what he did with reading out aloud, but without the reading since I am learning Japanese entirely through audio. So I will very much be doing something similar to what Richard's daughter did, listening to the same story 100 times and just copying what they say after hearing it. I think it'll be fun because I'll get to copy the emotion behind the words along with the pronunciation and intonation.

I really like step 7 in Yizy's routine, and it is similar in some ways to what I tried to do in Experiment 2. The other interesting element is the Glossika-style over learning schedule. In the Glossika schedule, you study new sentences on a moving window of 5 days where each day you're listening to new sentences and repeating a certain number of times. Then, you're going to review those sentences for 5 days before they move out of that window. So on any given day, you are repeating everything you have encountered over the past 5 days, but as a sentence approaches the end of the window, it starts to get repeated less until it is dropped off and you move the window forward. Yizy appears to have a similar sort of structure if we look purely at the repetition schedule. Combining the repetition schedule with a gradual expansion of the length of audio is something I've also been experimenting with as a variation on Experiment 2. But that primarily benefited my listening (I still continue to do this style of practice today). I only incorporate speaking in the early repetitions, though, while the segment lengths are short enough to hold entirely in memory, so in my case I think I did not do enough speaking reps in total to make the sentences sticky. Over learning really means "over" learning, so I'll be going over the top this time to ensure I get "more" than enough repetitions.

[ryanheise]

Update on Experiment 5

I haven't written any status update yet, but things have certainly happened, so I will try to recap my experiences so far.

I got off to a great start in the first week by memorising a 4 minute episode of Peppa Pig. My daily ritual during that week was to play the episode's audio on repeat through my shower speakers (waterproof!). I guess I am slow at showering etc. because that gave me a good 30 minutes of listening each day, during which I would try to mimic what I was hearing. I also spent some time each day transcribing the episode and figuring out the grammar. But it wasn't until the last 2 days of the week that I began to consciously start memorising the story, and to do that, I used the same kind of technique I use to memorise music which I described here.

At the end of the first week, I had a language exchange session and I was really excited to have found a few opportunities to use some of the expressions from Peppa Pig that came to mind. One thing I learned was that it was really useful during those shower listening sessions to notice when a sentence or expression or pattern might actually be useful. Even if a sentence may not be directly copied and pasted into a conversation, it was useful to recognise that a certain construct or expression was potentially really useful if I switched this word or that word.

What happened after the first week was not really expected. I was a bit worn out from doing the grammar study, and so rather than jump straight into the next story, I just decided to become lazy and for the whole next month, I just continued listening to the SAME episode of Peppa Pig in the shower, and nothing else. This turned out to actually be great. I was able to tap this material for much more than I originally would have, and I guess I really did the "over learning" thing in true style! In the next language exchange session, I was able to use (by my count) around 10 different expressions from Peppa Pig, just because I had been thinking about these same expressions a lot, over and over again.

It was about this time that I realised I probably should be doing some sort of measurement for this experiment. I created a spreadsheet to track how many expressions I was able to use in each language exchange session, but then the next couple of weeks I wasn't able to do language exchange anyway due to schedule conflicts.

I will say that I did recently move onto the second episode of Peppa Pig, and I've done the shower listening thing for at least a week now and am planning to start the active memorisation process soon.

My plan to write an app

Speaking of starting the active memorisation process, I have finally found some time to build an app to help me with this. By the way, I'm going to figure out a way to share this app here through the Google Play Store and Apple App Store BETA channels, so if anyone wants to try out this new app that I'm building, you are welcome to try it. I plan to develop it out in public so you'll be able to see each feature as it gets built, and hopefully people can also give suggestions and help influence how I actually build it. I've written a separate post about the app idea over in the Language Programs and Resources forum.

[ryanheise]

Update (of sorts)

Just thought I'd give an update since I've been completely quiet on the general and language log forums for a couple of months. In this time, I haven't learnt a whole lot of Japanese, but I have learnt a lot about the limits of my health.

It all started when I set myself a goal to build this app for the memorisation experiment in 2 weeks. In principle, I knew that if I set a goal, and let others know the goal, it would help to keep me accountable. And you could say it pretty much worked. I pushed (i.e. overexerted) myself, literally working on it all waking hours regularly until 4-5am, and then, 2 weeks later, I found I had somehow implemented all of the algorithms and data structures, a basic UI, and the low level audio code across two different platforms. At the end of it, I was feeling burnt out, but nothing that a bit of rest wouldn't have been able to fix.

I would like to have taken some time off at this point, but the universe was not on my side as I was just given a separate 2 week deadline for an unrelated job, almost equally intense. At the end of this prolonged 4 week period of intense work, I came down with a virus. The first virus kept me sick for about 2 weeks, but I never really fully recovered, and since my body was in a weakened condition, that made me susceptible to catching the next virus, and the next one, and so on.

Since this had continued for months, I decided to see a doctor, got some blood tests done, and fortunately that all turned out fine. But what the doctor prescribed me is simply: downtime. I'm the sort of person who is constantly doing mentally taxing activities, and my idea of downtime is learning a new language, or listening to a lecture on machine learning or linguistics. Ironically, people who are addicted to using their brains might not realise that a brain will be healthier and perform better if you switch it off every now and then.

So I really need to examine my lifestyle and figure out how to achieve the right balance of activity and rest. I plan to use the pomodoro technique to force myself to take a 5 minute break every 20 minutes, and I will also aim to go for a walk every day to get my exercise, relaxation, and (as a bonus) some vitamin D which should help improve my resistance to viruses.

As for my language learning, it feels like so long since I've done some serious Japanese study. I'm not worried since I'm sure I'll be able to easily get back into the swing of it when I'm ready to, but for now, I just want to finish off the app, and recover my health. I am dreaming of the day when I'll have the freedom (and health) to do a serious immersion experiment with Japanese, and if things turn out well, maybe even travel to Japan next year.