Word Auto-Numbering Not Translated

After translating a policy document or technical specification into English, chapter headings may still look like this: 第三部分 Corporate Culture. The words after the prefix are translated, but “第三部分” remains—and you cannot select those characters by dragging over them.

This is not a missed sentence. In Word, prefixes generated by automatic multilevel lists are not the same type of content as text typed into a heading.

A heading is made of two separate parts

In 第三部分 企业文化, 企业文化 is paragraph text that a standard translation workflow can access. 第三部分, however, is generated from a numbering definition—for example, a Chinese numeral format with the template 第%1部分. A translation tool changes the text in the paragraph, while the prefix can remain in the source language.

This differs from simply keeping the list structure intact after translation. Many tools promise that the structure is preserved: add a new section, and later items 4, 5, and 6 still update in sequence. What users are concerned about is the visible label: 第…部分 in a Chinese document has not become Part 3. The numbering structure may be intact while the label remains Chinese—a common partially finished result.

You could convert numbering to static text before translation, but then you can no longer add or remove chapters afterward. Dozens of headings would need manual edits.

Some conversions are rule-based; others need the full template translated

SimplifyAI handles automatic numbering in two categories:

• Formats with only numbers and punctuation are converted by rule. 一、 becomes 1.; Chinese numerals become Arabic numerals, and the full-width enumeration mark is changed to a period commonly used in English. This step does not go through the translation model.
• Templates containing words such as 第, 部分, or 章 are sent through the normal translation workflow as translatable segments rather than mapped through a hard-coded language table. English may use Part %1, French Partie %1, and Spanish Parte %1. The translation must retain the original numbering placeholder. If it is removed, changed, or returned unchanged, that entire numbering level remains untouched.

The second rule is intentional. A partial result such as 第3部分 looks worse than the original 第三部分: the number has been localized, but Chinese characters remain, making it look as though translation stopped halfway through. It is better to leave the full level in Chinese until you handle it in pre-translation terminology review or post-translation QA.

A few additional details affect delivery quality:

• Only numbering definitions and levels actually used in the document are processed. Some files carry hundreds of numbering definitions while the body uses only a dozen or so. Sending unused definitions for translation consumes quota and fills the review list with irrelevant items.
• Prefix fonts follow the target language. Once a template is changed to Part %1, its number is no longer displayed in a Chinese font beside an English heading. If Chinese remains in the heading, the Chinese font slot is retained.
• This conversion is skipped when the original numbering label is already natural in the target language. For example, 第%1章 also works naturally when translating Chinese into Japanese, so there is no need to force it into Chapter.

The table of contents may contain another already-rendered instance of 第三部分. Changing the numbering definition can make the body heading become Part 3 while the TOC in WPS still shows the old prefix. For how TOC entries update—and when it is better to leave an untranslated prefix—see Word TOC Not Updating After Translation.

What works today—and what still needs review

SimplifyAI can automatically identify numbering formats and template words in automatic numbering, convert formats that can be safely converted, send word-based templates for translation while validating their numbering placeholders, and align successfully rewritten levels with Latin fonts.

You should still review translated templates that contain words before translation is finalized (whether Part %1 matches your organization’s style; some teams may prefer Section %1). Levels the model cannot resolve confidently remain unchanged, and you can manually edit the numbering definition when needed.

What is not handled this way: a manually typed 第一章 in body text. That is ordinary text and follows the body translation workflow, not numbering logic. Special numbering formats whose values cannot be read reliably are not forcibly redrawn, to avoid incorrect chapter numbers.

Best suited to these documents

This is useful for policies, contracts, and bid documents with dense multilevel headings; consulting reports that cannot be delivered with source-language prefixes mixed in; and manuals that need multiple Western-language versions without maintaining separate 第 → Part/Chapitre/Capítulo mapping tables for every language.

Conclusion

The characters you cannot select are often not in the paragraph text. Upload a Word document with automatic numbering to SimplifyAI, then check whether the left side of each heading becomes Part 3 or whether the entire level still shows 第三部分. Either outcome is easier to review than a mixed, halfway-localized result.