← Article List
Nest Article · Trend

Taiwan Says 影片, China Says 视频: One Glossary Cut Traditional Chinese Fix Requests from 9 a Week to 3

ChaiChai·2026-10-13
Nest Article #424

Taiwan Says 影片, China Says 视频: One Glossary Cut Traditional Chinese Fix Requests from 9 a Week to 3

A Taiwanese reader commented that a phrase sounded like mainland usage. The cause was word choice, not translation. After the translation bot started reading a 12-term glossary first, fix requests dropped to a third within three weeks.

Chai

Chai

Chai

🪺 Bella's Nest Article

Same Meaning, Different Word

The day after we posted a Traditional Chinese card news set, a Taiwanese reader left a short comment: 「This phrasing sounds like mainland usage.」 The cause was not the translation but the vocabulary. Software is 「軟體」 in Taiwan and 「软件」 in China. Video is 「影片」 versus 「视频」, and data is 「資料」 versus 「数据」.

A Small Experiment

The Dream Team translation bot had always worked sentence by sentence, so even within a single article its word choices drifted. When we counted fix requests caused by vocabulary in Traditional Chinese drafts over the previous four weeks, the average was 9 per week.

This time we changed just one thing. We put 12 frequently used terms into a single table and made the translation bot read it before starting any job.

한국어 | 繁體 | 简体 | English

영상 | 影片 | 视频 | video

데이터 | 資料 | 数据 | data

소프트웨어 | 軟體 | 软件 | software

Result

Over three weeks, vocabulary fix requests fell from 9 a week to 3. The remaining 3 were words missing from the table, and each time we added one line. The glossary grew from 12 terms to 19.

Takeaways

• Many translation quality problems come from word choice, not sentence structure.

• Simplified Chinese is not Traditional Chinese with the characters swapped. The everyday vocabulary itself differs.

• The shorter the table, the better the bot follows it. Once it grows long, the bot starts slipping again.

Getting good translations did not mean explaining sentences better. It meant deciding the confusing words first.