I’ve always thought it’s hard to find statistical information regarding software localization, so I decided to gather some of them in one place. Here are some data that I collected over time regarding localized text. I hope you may find them useful.
These are for English and are referring to application strings, so it does not apply for documentation or marketing materials:
- Average words/string: 3.6 words
- Average characters/string: 22 characters
- Most characters/strings: <1000
Text size expansion when translated
These data were published by IBM, in their National Language Design Guide Volume 1:
| Characters in English | Average expansion |
|---|---|
| <10 | 200-300% |
| 11-20 | 180-200% |
| 21-30 | 160-280% |
| 31-50 | 140-160% |
| 51-70 | 130-140% |
| Over 70 | 150% |
You may find this very useful for UI strings because it allows you to do pseudo-localization with strings length expanding, in order to evaluate how your interface will behave.
I encountered the biggest expansions on Russian language, followed by French. If you rely on word-wrap, remember, that there are languages with very long strings, German being one of them.
Translation metrics
- IT/documentation - ~2000-2500 words/day (considering 8h workday)
- Law/licensing – 50% speed decrease
- On easy text you can boost the productivity a lot, reaching even 5000 words/day.