I have mentioned the idea of Sentence Banks before in a few posts. In this post I will explain to you what this concept is, how it will benefit you and most importantly how to actually make your own Sentence Bank.
Thumbnail source: Japanese verbs beginning with た. Can you read them all? Source: https://goo.gl/Czhh3I
Why Use Sentence Banks?
They allow you to:
- Quickly mine thousands of sentences.
- Find great example sentences using Anki’s inbuilt browser.
Below you can find a video of me demonstrating how I use this method to both mine sentences and store/search for hundreds of example sentences.
But what in blazes is a Sentence Bank?!
A Sentence Bank is exactly what it sounds like. It is a massive collection of sentences, the aim here being on quantity (100,000+ easily) and quality (context with audio and imagery). In this tutorial, I am going to show you how to convert entire seasons of TV shows into Anki flashcards and how to study with them for maximum efficiency.
The following post is a full step-by-step tutorial.
To see how I study with this method, check the last section of the page entitled Sentence Mining.
Here is an example of what the sentence cards look like (they also have audio on the backside of the card).
Each sentence has the front sentence along with the scene of the show that it comes from. As you will remember what situation the sentence is being used in, this gives you a high amount of context. And context is incredibly powerful.
Each sentence also has audio on the back which tells you exactly how the sentence is pronounced. I’m not a linguist but I feel that this has helped me to improve my accent subconsciously, as I have heard the correct pronunciation for nearly all the sentences I have learnt. At the very least, having this is a nice aid.
Every sentence also has normal furigana readings that can be generated with the Japanese Support plugin for Anki. Support for readings are also included for dictionary definitions as well (use Anki’s “Add” option to generate furigana for dictionary sentences then paste them into the “Meaning” field, see Step 4.5 in the second half).
Of course, you are free to change the layout and I will show you exactly how to do this later on.
How do I get thousands of these sentences and how can I use them in my sentence mining efficiently?
The second question I will answer is how to use Sentence Banks efficiently. I mention this briefly at the bottom of the post in the section “Sentence Mining“.
There are some prerequisites that you will need before starting this process.
- A film, TV series or other form of video in your target language.
- Accurate and correctly timed subtitle file(s).
Unfortunately, I can’t help you with obtaining the video files of whatever it is you want but a couple of sites you can use for subtitles if you are learning Japanese.
Anime – http://kitsunekko.net/
Dramas – http://www.d-addicts.com/
Once you have a video file and its corresponding subtitles file then we can start the process.
If you need to change the timing on your subtitle files then use http://www.aegisub.org/.
How to Create Your First Sentence Deck with subs2srs
This program is one of the main reasons my reading is so good. If you can use it in the right way, it will become an incredibly powerful tool for you.
Here is the page for the software (it looks a bit old but it is perfectly safe)
This page also has more detailed instructions which you may wish to look at. I will be explaining the bare minimum that you need.
Once that has downloaded and you have opened the software, you should end up with a screen that looks like this↓
Take some time to have a look at the features and see what you can do with the program. It can also produce video for flashcards but I have yet to get this working.
Step 1: Add L2 Subtitles
To add subtitles to subs2srs you will first need to click the Subs1 button near the top of the program.
Once clicked, a file browser will appear. You will then need to find the folder that your subtitles are in.
From here, click the first file and click the open button, (I apologise if you can’t read my systems buttons, but it is the one highlighted blue that says 開く(O)).
The top part of the program should show that it has added the subtitles correctly.
Now here is where it gets interesting. For adding multiple video files we are going to need the star wild card which is just an asterisk, (*). We are going to use this to get the entire seasons’ subtitle files as well as video files. Don’t worry if you don’t know what a wild card is or how to use it, just follow the instructions and you will be fine.
First, click on the subtitles path and use the mouse, or arrow keys, to get to the end of the file path.
Then highlight all the way up until you reach the “\” (or in my case “¥”), which indicates which folder you are in, making sure NOT to delete the file extension (in this case the .ass part ( ͡° ͜ʖ ͡°) ).
Now simply add the asterisk before the file name and this will automatically select all the “.ass” files that are in the folder you are pointing to.
Step 2: Set Output Directory
This is very self-explanatory. Here you will decide where the files that this program generates are going to be saved. Click the Output button and choose a location. I recommend creating a new folder for this or it is going to get messy.
Step 3: Adding Video(s)
This is going to be the same process as the subtitles. Click the “Video” button, find your video file(s) and select one that is in a folder. Then if you want to add all the files of the same file type in that folder, use the asterisk (*) again followed by the file type.
Tip: I don’t recommend overloading the software with a lot of video files at a time. It is liable to crash and you will have to start over. Usually, a series of about 20-30 episodes will be fine. If it does crash then just start again.
Common Error & Fix
It is common to get this error.
The reason this occurs is usually because you have forgotten to use (*) for one of the fields or because there are more subtitles than there are video files or vice versa. So make sure that there are exactly the same number of subtitle files as there are video files.
Step 4: Naming the Deck
The next step is to give the deck a name so you can find it in the output folder. I’m just going to call it the name of the series I am using.
Step 5: Tweaking Subtitle Timings
Note: there is another option further up to “Span” and “Time Shift” the subtitles. I have never used this feature but if you experiment then it might be useful. I personally use a piece of software called “Aegisub” to correct my subtitles.
Most of the time you will be using subtitles that are synced up with the video correctly, but if they are slightly off then you can use this feature to quickly fix this issue. You will have to experiment with what works best.
Step 6: Removing Hard Subtitles and Image Sizing
Second is the image sizes and cropping. I recommend a width of 480px and a height of 320px, this is not too big nor too small. If you have hard subtitles on your video file then you can use the crop bottom option to remove pixels from the bottom upwards. Under Crop Bottom (on the right), the more pixels you put the more will be deleted.
Step 7: Testing
Okay! Now that you have all your settings adjusted and you make sure that all the options you want to use are ticked, go ahead and click the “Preview…” button.
Once you hit the preview button a menu will appear. From here you can check that image is the right size, the audio is synced up correctly and that the subtitles are showing up properly. You can also click the “Episode” drop down list which allows you to check other episodes if you are doing more than one video.
Step 8: Go!
Once you have checked that everything is fine, click the green “Go!” button! Leave subs2srs to do its thing (could take up to an hour or more) and go read a book or do some sentences 😉
Once subs2srs has finished, you will get a message like the one below. DO NOT CLOSE THIS! We will be needing this in the next part…
You may get a message about long lines. This depends on the person but you may or may not want to keep them. I usually keep them.
Importing to Anki
Step 1: Importing the subs2srs Template
Once in Anki, go to File -> Import.
Anki should now have a new deck or will have replaced the default deck with the name of the template file, like below. This is just importing the “layout” that is needed to support the types of cards made by subs2srs. If you don’t see any change but that little yellow dialogue box appears then it has probably imported fine.
It maybe a good idea to make a new deck at this time for the TV show or set of videos you are making.
Step 2: Adding Furigana Reading by Creating a New Note Type (Japanese Learners Only)
If you want to add furigana readings to your sentence cards then you will need to do this step as well. We have just imported a new note type from subs2srs but this note type doesn’t support furigana reading support that is provided by the Japanese Support plugin. To fix this we are going to create a hybrid note type which will use both.
Then once it is installed and you have restarted Anki, go to Tools -> Manage Note Types…
The Note Types manager should then appear on-screen with a list of note types. You will need to click the “Add” button which will take you to a separate window.
I have been using Anki for a very long time and have experimented with lots of different decks and card layouts. As a consequence of this, mine is very much a mess and you should have a lot less note types.
Find the note type “Clone: Japanese” and click “OK”.
Next, find the new note type in the Note Type window and then click the “Fields…” button.
A new window should appear that will show you the different fields that the note has in it. So far yours will just look like the following image.
Now, to make this note type work with subs2srs, we are going to need to add a few fields.
To do this, simply click the “Add” button and add the following fields to your note type, keeping the original three the same.
If you get video working then feel free to add that as well. Also add any other fields you may want.
Here’s the note type I already use:
Purpose of each field
Expression – Contains the original sentence you want to learn.
Meaning – Allows space to contain dictionary definitions and explanations of words or phrases.
Reading – Provides another copy of the original expression but with furigana above the kanji so you know how to pronounce the sentence.
Snapshot – Space to store the image of the scene where the sentence was said.
Audio – Space to store the audio pronunciation of the sentence.
SequenceMarker – Stores the timing of when the sentence was said (useful if you want to refer back to the scene).
Step 3: Importing the subs2srs Generated Deck
Again you are going to want to go to Anki and go to File -> Import. Then find the output directory for subs2srs. From there, click the name of your series (which should be a .tsv file) then click open.
Then, Anki will bring up a menu that looks like this ↓
Make sure that the note type is the one you created earlier. If not click the note type that is selected and you can change it to the one you made (CTRL+N). Also make sure that the deck is the correct deck you wish to save the cards to. If not then select that and change it as well.
We are then going to get the message that subs2srs output and make sure it matches up with this window.
We can see that they are not in order so we are going to need to make sure that they are. As the first outputted line by subs2srs is “Tag”, we are going to change “Expression” in Anki to “Tag” by clicking the “Change” button. A new menu should appear. From here click the correct field and click “OK”.
Once you have done this for each field, Anki should match up with the output message from subs2srs. Make sure you also check the “Allow HTML in fields” otherwise your images will not show up! Mine is below.
Note; Field names may not match up entirely with the message. “Expression” should be used in place of “Line from Subs1”.
Once this is done, click “Import”.
Step 4: Importing the subs2srs Generated Media so that Audio and Images Work
If you try the deck now you will notice that audio, images and readings (if you are learning Japanese) do not yet work. To get audio and images working we will need to add the media generated by subs2srs to Anki’s media collection folder. This folder is in Documents -> Anki -> User 1 -> collection.media.
Once found, keep this open and find the output folder that subs2srs used earlier. Find the folder that is named the same as your TV show with .media after it, then open the folder.
You should see lots of music and image files. You are going to want to copy ALL of these from this folder over to the collection.media folder in Anki.
To copy all use CTRL+A (to select all) then CTRL+C (to copy) and to paste in the target folder use CTRL+V.
Depending on how many episodes you have done, this could take a while. Once it has finished transferring, you can test the Anki deck again to see if audio and images work. It might be worth restarting Anki if it doesn’t work. If you try the deck now you should see something like mine. You should have sentence, an image and some audio on the card.
Step 4.5: Adding Readings for Japanese Learners
To add readings open up the browser by click “Browse” at the top of the main page of Anki or by pressing “B”. Then find the name of your deck and click it.
Once you click the deck, the contents of it will appear on the right. Here you need to select every note in the deck by clicking CTRL+A. Then go to Edit -> Bulk Add Readings. Anki may freeze for a bit but just be patient. Once it has finished you can check in the browser whether readings have been added by clicking a card and scrolling down. Or you can check the deck again.
Once done the deck should have readings for every card like so:
To generate furigana for the Meaning section we are going to have to do something slightly different. Anki will only generate Readings for anything that is in an “Expression” field. We can make use of this by clicking “Add” and putting our dictionary definition into the “Expression” field of a “Japanese” note type. Click on the “Reading” field and it should generate the furigana in brackets. From there, simply copy and paste it over to the card you want it in.
Tip: For efficiency it might be a good idea to create a new deck to save these expressions in. That way you can just click “Add” and this form will reset to a new blank “Add” form. This means you don’t have to delete both fields with your mouse. 🙂
Step 5: Checking/Changing Card Layout
I have made a lot of changes to this over the years and I vaguely remember that the layout for these cards was completely different when I first used this method. I have since changed the code on the cards to make the layout suit me. Personally I prefer having the sentence on the front with the image to give context. Then on the back I will have audio, dictionary definitions and how the sentence is pronounced. I will show you where to go to change this and my code so that you yourself can use my layout (if it is different).
First, go to the deck and press “E” to edit the card. A new window should appear like the one below. From here you need to click the “Cards…” button.
Once you click “Cards…” this window should appear. It has code for the layout on the left and a viewer on the right to see changes that are made.
To make sure your layout is the same as mine, just copy and paste the following code into the correct sections.
If you want to re-arrange parts then take the bits that are in curly brackets and move them to a different location, making sure that they are included inside a
can be moved to the Back Template and you will have the image on the back instead. If you have other fields with Japanese writing in them and you want to have furigana readings for that field then you can add “furigana:” to the curly brackets, like this;
Step 6: Repeat
Now do this again for as many shows as you want and you will have yourself a massive Sentence Bank!
Step 7: Sentence Mining
I spoke at the beginning of this post about why Sentence Banks are so powerful. The main 2 reasons being,
- Able to quickly mine thousands of sentences.
- Able to find great example sentences using Anki’s inbuilt browser.
The first reason is because you can go to a deck that you have made in Anki and you can go through the cards one by one as if you are watching it live. As you do lose some contextual and background information that is not included in speech, I do recommend watching the video first and then using Anki to go over it again later.
My process when Sentence Mining these decks.
- Open deck, press “Space” to view back of the card and play audio.
- If I understand the sentence completely, I press “Delete”.
- If I don’t understand the sentence or a word/phrase then I press “Shift+1” which “Suspends” the card. Suspending means Anki will prevent you from seeing it until you un-suspend it in the browser. I use this to “Save” cards for later.
- If I understand all the words but come across a new grammar structure, I hit “Space” and learn the sentence without editing it.
- Rinse and repeat!
I do this process REALLY quickly. I want to make it seem like I am just re-watching the show again, not like I am studying. I have done this for well over 30+ Anime and TV shows and it has helped my reading drastically.
After I have finished a series or a film I will add those sentences that have been suspended to a side deck I call “New Sentences” and then “un-suspend” them. In here I go through them again, adding definitions to the ones I want to keep in my main deck (which has the 10k) and delete the rest that are too easy or I don’t want to add. After I have done this I then simply move the cards in bulk to the main deck.
I don’t recommend learning while going through the deck the first time. The first pass should be enjoyable and should just be a process of “saving” the sentences that you don’t know. If you do this for 20+ decks then your reading is going to get pretty good. 🙂
The second reason to use this technique is because Anki can act as a brilliant database. If you accumulate a few decks of different TV shows then you will have thousands of example sentences. If you then find a word you don’t know in your immersion environment, and want to study it, you can search the browser in Anki for other sentences with the word in. This will give you a better understanding of the word and give you quick access to high quality contextual sentences, making remembering words much easier.
I hope this tutorial was helpful. If there are any issues or problems you don’t understand then comment below.
UPDATE: I am in the process of making a website called “The Japanese Sentence Bank” which will contain a variety of example sentences from native Japanese sources. If you want more information about the site then please go here. I am looking for people to help out with adding example sentences and possibly web design. Please email me at email@example.com if you are interested!
Thanks for reading.
2017/21/01By Matthew Hawkins