turning reading comprehension into listening comprehension: three simple strategies

Once you have kanji at least somewhat tamed, reading is the easiest skill to develop to a high level. You can take all the time you need to get through a passage, no one is rushing you, you don’t need to depend on anyone else to talk to or correct your writing, and you can do it all day long if you like. And especially for those of us using extensive reading as the foundation for our studies (which, it hardly needs repeating, I strongly support), we will most likely arrive at a stage in our learning where reading is quite far advanced above the other three skills. Then when you try to listen to material that’s at the same level that you’re comfortably reading, it’s completely incomprehensible. This can be very frustrating! It’s like solving a Chinese puzzle with mittens on. The question then is, how can we use that reading skill to boost our listening skill?

Certainly massive immersion is foundational to listening skill. Listen as much as you can. Listening is something you can multitask, so take advantage of that. Going out for a run? Headphones in. Punching numbers in Excel at the office? Headphones in. Deep discussion with your significant other? Headphones in. (Kidding … maybe?)

But that’s a little random for our purposes here, and doesn’t really answer the question. How then can we focus our listening a little more closely? Here’s three simple things I’ve found to work well, in order of precision.

Firstly, you might remember I’ve previously mentioned incremental reading combined with audio. This SRS exercise will result in exceptionally complete comprehension and retention of short passages. A quick summary of the method: find a source for short passages where you have both audio and text. You can select passages that are quite difficult as long as you find them interesting, but don’t overdo it because the quantity of repetitions you’ll need for a passage that’s too far over your level will only result in you getting sick of it. Podcasts are perfect material. I like the audio blogs over at japanesepod101. Your card, then, will have the text on the front, the text with readings on the back, and the audio. Have Anki set to not play audio automatically. Reviews are done by reading the card and then listening. Feel free to look up as much as you need to while reading. Particularly unfamiliar words should probably be added to a simple vocabulary deck or an MCD deck. The reading should be fairly easy though, because the point is to bootstrap the difficult listening with the higher skill level of the reading. Grading can be done like any other SRS card.

That’s the most effective method in some ways, but the focus is quite narrow, and there’s a lot of setup required. It also has a fairly high burn-out risk.

Secondly, let’s consider a method for using Japanese subtitles. Now, just watching video with J-subs is something I have found to be not terribly effective. It does aid comprehension, and I’m sure if you could read at the same speed as normal speech it would be a lot more effective than it is for me. But as it is, it often ends up going by all too quickly to be much use. So what I do instead is put the text of the subtitles file in Learning With Texts. (You could just as easily use a plain text file – maybe put them in an e-reader program like iBunko – and make MCD cards for words or grammar points as needed.) The process then is to first watch the video without subtitles, then read through the text carefully taking all the time you need to understand everything, and then re-watch the video. This should be done without too much delay between, but a day or two is fine. You will notice a big difference in comprehension between first and second viewings!

For this method, the video must be reasonably within your grasp before reading the text, otherwise two viewings won’t be nearly enough. And if you’re anything like me, two viewings of anything is already very nearly one too many, no matter how good the material is. Also, it must not be too long, or it will take you all day to study the text, and you won’t be able to retain all of it. I find the 20 minutes of a typical anime episode to be pretty much perfect.

That might still be too much structure for your taste, so lastly let’s consider a method (that’s so loose it can hardly be called a method) that will let you take in the most volume of text and audio of any of these. For tadoku, it’s often recommended that you concentrate on the works of a particular author for periods of time, since you get used to their style and their selection of vocabulary. We can take advantage of this for listening as well. Most anime are based on manga, as are many dramas. So all you need to do is watch your favourite series, read through the manga (SRS as much or as little as you like), re-watch the series, reread, etc., until you’re tired of it, which is your sign to move on to something else. Very often the lines in the anime or drama are lifted directly from the manga, and even when they’re not, they’ll be in the same style and using the same set of vocabulary. Sometimes you can also find audio dramas (”Drama CD’s”) that are also based closely on the manga. ARIA in particular has a great many of these, so you can get a lot of listening immersion that’s all based on material you already know from reading. As well as ARIA, I like みなみけ and ひだまりスケッチ for this approach. Any simple slice-of-life anime or drama is a good bet for practical and realistic everyday vocabulary. But make sure to follow your interests! The brain is like a little kid faced with a plate of broccoli; very good at rejecting what it doesn’t enjoy. Luckily when it comes to language even the chocolate cake is nourishing!

I always remember something loafyi said in the #ajatt channel one day – something like “when I can recognize a word in listening, that’s when I feel like I really have it”. I’ve found this to be very true. I hope that these simple common-sense suggestions can help you to use your hard-earned reading comprehension to raise your listening to a similar level.

make your learning with texts text window easier to read

If you’ve tried Learning With Texts for Japanese, and I recommend you do try it if you haven’t, you’ve no doubt found that the text window uses the same sans-serif font as everything else and is kind of painful to read. Here’s a replacement stylesheet that will fix that. You’ll need the MS Mincho font installed on your machine, which Windows computers have by default, don’t know about Macs.


Save this file into the css subfolder in your LWT installation directory, overwriting the existing version, and enjoy readable text.

Thanks to Kanjius and Kiriyama for their help with this.

getting text from image-based subtitles

I’ll try to expand on this post later, but just a quick note: today I watched an excellent movie (The Abacus and the Sword) and wanted to take some of the subtitles to use as SRS cards. But the subs were in the image-based .sub format, so Aegisub couldn’t handle it. The solution is to use Subtitle Edit. This doesn’t have a Japanese dictionary built in so you need to get it from the Tesseract project. Unpack that file to the \Tesseract\tessdata folder under Subtitle Edit’s install folder. Then when you open such a .sub file, it will ask you if you want to import it, then ask you what system to use (Tesseract) and what language (Japanese). Then you wait, some magic happens, and you have text. It marks the ones it isn’t sure of in a different colour so you can correct them manually, but it gets most of them pretty much right.

Comments Off on getting text from image-based subtitles Posted in Learning Tools

when can i delete this srs deck?

“I don’t really SRS anymore” is something one occasionally hears from advanced learners, usually with a twinge of envy. Compared to reading, watching movies, conversing with natives, SRS feels like (and is) a pretty dry and mechanical thing. Some days it would be a nice relief to drop it entirely. Of course no native speaker uses SRS for their own language.

Then again some will insist that one should never stop SRS, even when the card intervals stretch into the years and no new cards have been added for a long time.

It’s actually very easy to tell approximately when you can permanently delete or suspend any given SRS card. SRS is designed to remind you of a fact just before you forget it; so if you see the fact (be it a grammar point, word, or kanji) in the course of your normal day-to-day use of the language with a greater frequency than the card interval, you don’t need that card anymore. In fact, keeping it around is a waste of time. You can see that this is always going to be approximate, going by intuition not exact statistics.

Thinking of it mathematically like this it’s obvious that an advanced learner will both have his average card interval much longer than a beginner, and will see the facts on those cards far more often, due to faster reading, better listening comprehension, and involvement with more advanced materials. Then the time inevitably comes when new material comes in so infrequently, is so easily remembered due to the massive context already absorbed, and old material is so ingrained that the whole SRS process can be dispensed with. You could graph the whole process if you wanted.

This is partly why I like to keep separate decks that are at least somewhat homogeneous. It is a time saving to suspend an entire deck of easy grammar, for example, rather than continuing to review that deck and suspending cards bit by bit. One also needs to learn to take a somewhat detached view of the individual cards. A dragon might know the details of every treasure in his pile, but it’s a bit much for a human; and anyway there’s so much treasure already there and flowing in all the time that a gem or two or a dozen will never be missed.

how to stay monolingual: making use of recursive lookups

If you advance beyond the beginner stage in your language learning, the question will arise soon enough: “How do I switch over to monolingual study?”  Thinking about Japanese in English is of course how we start, and even at advanced levels an English explanation of some abstruse point can quickly get you unstuck, but there’s no denying the multiplicative power of thinking about Japanese in Japanese.  Usually the suggestion will be made, “Do recursive lookups”; with the idea of looking up every word in a Japanese definition that you don’t understand, again in a Japanese dictionary; looking up any unknown words in those second level definitions; and so on, until you’ve branched out enough that no more unknown words remain.  Depending on the case, you could get five new words, or twenty, or fifty out of this.  But that sounds like a lot of work, and then once you’ve done that, what do you do with them?

In case you’ve somehow been missing it till now, there’s a brilliant Japanese learning resource in a location you might not immediately think of; the IRC channel #ajatt on Rizon.  It’s a sort of blender for ideas and pretty often some excellent stuff comes out.  Today, the subject was nesting definitions and massive context cloze deletion (MCD) SRS cards.

It got a little long, so the rest is behind the jump, posted as written.  There’s examples too.

Continue reading

template spreadsheet for tracking your readthekanji progress

Due to overwhelming popular demand (two people), I’ve made a template version of my spreadsheet to track progress on readthekanji.com.  The link is here.

Row two is your start row.  Change the values to whatever your current stats are, and change the date in cell A2 to today.  Then, tomorrow and every day thereafter, enter the quantity of reps you did, and your current number of untested words and kanji for each category.

The completion date, after the first two weeks, takes your running average as of the last two weeks to calculate the end date.  This is to allow for variations in pace over longer periods of time; i.e., if you are busy for a while and can only do fifty reps a day, that slow period doesn’t forever after push your completion date back past what it actually will be.