HomeGamesUpdatesPricingMethodology
Steam News27 January 20251y ago

【1.5 Japanese Optimization】The Japanese version has been Fixed line by line!!!

What completely took me by surprise was that the Japanese version of the translation was extremely, extremely poor.

Full notes

Full Subway Exorcist Girl update

Read the full published notes in a cleaner layout. The original post stays linked below.

What changed

1 fix0 additions2 changes0 removals
  • UI and audio
  • Gameplay
  • Fixes
changedWhat completely took me by surprise was that the Japanese version of the translation was extremely, extremely poor. The original translation quality was bad, but since I don’t understand Japanese, I didn’t notice the issues. As a result, I received some negative reviews in Japanese, and I am very concerned about this. The poor translation quality was not intentional on my part. It’s just that I set up 10 languages for the game and had to handle all the translation work alone. So the workload was disastrously huge, far beyond what one person could handle.
changedI thought that even if machine translation results were a bit odd, they would still be understandable and wouldn’t overly affect the gameplay. So I wasn’t too worried. But when I input the game’s text line by line into DeepSeek and ChatGPT today for direct translation, I found that the Japanese translation results were a disaster. I laughed so hard because the translation was so bad. Many parts were completely opposite in meaning, and there were many hilarious mistakes. The characters’ titles were also all mixed up.
fixedI resubmitted each sentence to DeepSeek and ChatGPT, had them translate directly, corrected the errors, provided the revised translations, and then filled them into the game engine. Each sentence took a minute. I even subscribed to ChatGPT’s monthly plan for this.

Subway Exorcist Girl changes

changedWhat completely took me by surprise was that the Japanese version of the translation was extremely, extremely poor. The original translation quality was bad, but since I don’t understand Japanese, I didn’t notice the issues. As a result, I received some negative reviews in Japanese, and I am very concerned about this. The poor translation quality was not intentional on my part. It’s just that I set up 10 languages for the game and had to handle all the translation work alone. So the workload was disastrously huge, far beyond what one person could handle.
changedI thought that even if machine translation results were a bit odd, they would still be understandable and wouldn’t overly affect the gameplay. So I wasn’t too worried. But when I input the game’s text line by line into DeepSeek and ChatGPT today for direct translation, I found that the Japanese translation results were a disaster. I laughed so hard because the translation was so bad. Many parts were completely opposite in meaning, and there were many hilarious mistakes. The characters’ titles were also all mixed up.
fixedI resubmitted each sentence to DeepSeek and ChatGPT, had them translate directly, corrected the errors, provided the revised translations, and then filled them into the game engine. Each sentence took a minute. I even subscribed to ChatGPT’s monthly plan for this.

What completely took me by surprise was that the Japanese version of the translation was extremely, extremely poor. The original translation quality was bad, but since I don’t understand Japanese, I didn’t notice the issues. As a result, I received some negative reviews in Japanese, and I am very concerned about this. The poor translation quality was not intentional on my part. It’s just that I set up 10 languages for the game and had to handle all the translation work alone. So the workload was disastrously huge, far beyond what one person could handle.

I thought that even if machine translation results were a bit odd, they would still be understandable and wouldn’t overly affect the gameplay. So I wasn’t too worried. But when I input the game’s text line by line into DeepSeek and ChatGPT today for direct translation, I found that the Japanese translation results were a disaster. I laughed so hard because the translation was so bad. Many parts were completely opposite in meaning, and there were many hilarious mistakes. The characters’ titles were also all mixed up.

I am very grateful to the Japanese players who endured such translations and still managed to finish the game. I truly admire their comprehension skills and feel deeply sorry.

I resubmitted each sentence to DeepSeek and ChatGPT, had them translate directly, corrected the errors, provided the revised translations, and then filled them into the game engine. Each sentence took a minute. I even subscribed to ChatGPT’s monthly plan for this.

As a result, I used up all of ChatGPT’s quota, and DeepSeek also went down due to too many users, making my translation work even more difficult.

However, I have now completed line-by-line proofreading for the 【Clown Carriage】 and 【Ghost Carriage】, as well as the opening narration and the dialogue between the protagonists. These parts are crucial to the game and must not have translation errors, as they would severely impact the gaming experience.

So in the current version, you should see correct and natural translations in the key parts. However, there may still be translation errors in some less important parts, and I ask for your understanding and patience.

Of course, I will also re-translate and proofread the key parts for English and Korean in the future. Special thanks to the Korean and English players for their tolerance and support! But now I must go to sleep.

Source

Steam News / 27 January 2025

Open original post

Changelog.gg summarizes and formats this update. How we read updates.