iTranslated by AI
ATOK for PC: Why English Word Conversion Can Degrade Typing Accuracy
I have been a long-time user of ATOK.
As an IME for Japanese input, I still have high expectations for ATOK. On the other hand, the PC version of ATOK has become packed with too many features intended to make Japanese input more comfortable, and I sometimes feel that some of them actually degrade conversion accuracy.
What I find particularly troublesome is the feature that converts Katakana or Japanese words into English words.
If you want to input English words, you can simply input them directly. Additionally, ATOK has a feature called "Half-width English Conversion (Post-conversion)." If you assign a key binding to this, you can convert to English only when necessary.
In other words, you already have the means to input English words. There is basically no situation where it is helpful to have English words mixed into your standard Japanese conversion candidates.
Furthermore, the problem is not just that unnecessary English word candidates appear; it is that they are accumulated as automatically registered words. As unnecessary candidates build up, your conversion candidates become cluttered the more you use the software. This is quite serious for a Japanese IME. I have been suffering from the problem of ATOK's conversion accuracy declining year by year.
In this article, I will summarize the settings to suppress unnecessary English word conversions in the PC version of ATOK and prevent your conversion candidates from being polluted by automatically registered words.
Removing English Dictionaries
First, remove the dictionaries that provide English candidates.
Open the ATOK Properties (Environmental Settings).
Dictionary/Learning
→ Accessory Dictionary Set
Remove the following dictionaries:
Katakana-English Dictionary
Japanese-English Dictionary
This will make it less likely for conversion candidates from Katakana or Japanese to English words to appear.
Turning Off English Input Assistance Data
Next, turn off the English input assistance data (predictive text data).
Settings can be found here:
Input/Conversion
→ Predictive Conversion
→ Input Assistance Data
Turn off the following:
English Input Assistance Data
This will make English predictive candidates and input assistance candidates less likely to appear.
This will make it less likely for English predictive candidates and input assistance candidates to appear.
Deleting Automatically Registered Words
ATOK has an "automatically registered words" feature. This is a function where ATOK automatically registers words based on your input and conversion results.
When unnecessary candidates accumulate in the automatically registered words list, your conversion candidates become cluttered. Especially if English candidates or incorrect conversions have been registered, they may persist as candidates even after you change your settings.
Open the ATOK Dictionary Utility.
ATOK Dictionary Utility
→ Automatically Registered Words
Delete any unnecessary words.
If you want to delete them in bulk, select the entire list and delete.
Stopping Future Automatic English Word Registration
Even if you delete the automatically registered words, the system will return to the same state if they are automatically registered again.
Disable the automatic registration for English words.
In the ATOK Properties, go to "Dictionary/Learning" → "Learning" → "Automatic Registration" settings, and change the following to "Do not register":
English Post-conversion: Do not register
English Compound Words: Do not register
"English Post-conversion" is a setting that automatically registers words you have converted to English.
"English Compound Words" is a setting that automatically registers compound words containing English letters.
If you do not need English candidates, setting these two to "Do not register" will help prevent unnecessary English candidates from accumulating.
Summary of Settings
Finalize the settings by checking the following:
Dictionary/Learning
→ Accessory Dictionary Set
→ Remove Katakana-English Dictionary
→ Remove Japanese-English Dictionary
Input/Conversion
→ Predictive Conversion
→ Input Assistance Data
→ Turn off English Input Assistance Data
ATOK Dictionary Utility
→ Automatically Registered Words
→ Delete unnecessary automatically registered words
Automatic Registration Settings
→ English Post-conversion: Do not register
→ English Compound Words: Do not register
Conclusion
If you find English word candidates unnecessary in the PC version of ATOK, you can improve the situation by reviewing the following:
If you feel that ATOK's conversion accuracy has declined, it is possible that not only the performance of the dictionary itself, but also the accumulation of unnecessary candidates in the automatically registered words list, is the cause.
By suppressing unnecessary English candidates and cleaning up the automatically registered words, you can use ATOK more comfortably as an IME for Japanese input.
I still have high expectations for ATOK. That is exactly why I believe it is important for users to be able to suppress candidates that interfere with the basic functions supporting Japanese input, as well as the contamination of candidates caused by automatic registration.
Actually, I have a feeling that having this behavior enabled by default is a bit strange. Am I the only one who feels this way?
Discussion