Input languages tell Convert which writing-system or spelling rules it may apply. English is always preserved as the base language. You can enable up to three non-English conversion languages at once.
Enabling a language does not translate the complete message into that language. The model first considers the full sentence, then changes only the spans that match an enabled rule. Clear English, product names, code, URLs, email addresses, and other protected values remain subject to the preservation contract.
Available conversion languages
| Conversion type | Language | Illustrative input | Illustrative output |
|---|---|---|---|
| Script conversion | Chinese (Pinyin) | nihao | 你好 |
| Script conversion | Japanese (Romaji) | arigatou | ありがとう |
| Script conversion | Hindi (Latin) | namaste | नमस्ते |
| Script conversion | Arabic (Arabizi) | marhaba | مرحبا |
| Script conversion | Русский (Translit) | privet | привет |
| Script conversion | Ελληνικά (Greeklish) | kalimera | καλημέρα |
| Tone reconstruction | Tiếng Việt | khong dau | không dấu |
| Accent and orthography repair | Español | como estas | ¿cómo estás? |
| Accent and orthography repair | Français | deja | déjà |
| Accent and orthography repair | Português | obrigado | obrigado |
| Accent and orthography repair | Deutsch | grusse | Grüße |
| Accent and orthography repair | Italiano | perche | perché |
Examples show the intended kind of change, not a guarantee that every ambiguous fragment has only one correct result.
Configure languages
- Open UnimeType Settings and choose Languages or Conversion languages.
- Enable one to three non-English languages you actually type.
- Choose a Preferred language from the enabled set.
- If Japanese is enabled, choose Natural Japanese or Kana output.
- Type a short example in a normal text field and run Convert.
The preferred language resolves genuinely ambiguous Latin fragments after the model considers sentence grammar. It does not force clear English or every Latin word into that language.
Choose Natural Japanese when ordinary Japanese orthography is wanted. Choose Kana when the converted Japanese span should avoid kanji. This option affects Japanese conversion only.
Input languages versus Language Packs
| Setting | Changes | Does not change |
|---|---|---|
| Input language | Which conversion rules Convert may apply | Provider, model, or terminology catalog |
| Language Pack | Extra terminology guidance or exact local shortcuts | The supported-language list |
A Language Pack cannot add an unsupported input language. Read Language Packs before diagnosing a missing term or shortcut as a language-selection problem.
If Convert chooses the wrong language
- Disable languages you do not use; success means the ambiguous span is no longer eligible for those rules.
- Set the intended language as Preferred language; failure means an unambiguous span is still being misclassified, not merely tied.
- Select only the target phrase when surrounding text points to another language.
- Keep deliberate English spelling clear and complete. UnimeType preserves clear English but cannot infer every intentionally ambiguous fragment.
- For a product name or preferred spelling, enable the relevant Language Pack instead of adding another input language.