Data basis: 48 separate pulls from Steam’s public appreviews endpoint — eight languages
(English, Simplified Chinese, Korean, Russian, German, French, Japanese, Brazilian Portuguese)
across six games — read 2026-08-27, most recent first, 100 reviews per language per game where
available. Rates below are of those samples. Steam lets you filter reviews by language one at a
time; it does not lay them side by side.
Every review score on Steam is an average over languages. Splitting six games’ reviews by language turns up gaps large enough that the average stops meaning much.
Project Zomboid reads as two different games
| Game | Highest language | Lowest language | Spread |
|---|---|---|---|
| Project Zomboid | French 97% | Chinese (Simpl.) 40% | 57 points |
| Escape from Tarkov | Portuguese (BR) 79% | Chinese (Simpl.) 31% | 48 points |
| Mistfall Hunter | Portuguese (BR) 59% | Russian 17% | 42 points |
| Corsair Cove | Portuguese (BR) 98% | Chinese (Simpl.) 69% | 29 points |
| X4: Foundations | Portuguese (BR) 89% | Chinese (Simpl.) 71% | 18 points |
| Songs of Syx | French 99% | English 85% | 14 points |
97% positive in French. 40% in Simplified Chinese. Same game, same week, a 57-point spread.
Tarkov runs 48 points apart (79% Brazilian Portuguese, 31% Simplified Chinese), Mistfall Hunter 42 (59% Portuguese, 17% Russian). At the other end, Songs of Syx spans 14 points and every language sits between 85% and 99% — the only game here where the communities essentially agree.
The obvious reading is that some language communities are harsher. The data does not support that cleanly, because the direction flips. Simplified Chinese sits 44 points below English for Project Zomboid — and 20 points above it for Mistfall Hunter. A blanket “Chinese reviewers are tougher” cannot produce both.
There is a language-level tendency, but it is smaller than the game-level swing
Averaging each language across all six games flattens the game-specific noise. Brazilian Portuguese comes out highest at 87%, Simplified Chinese lowest at 58%, with everything else between 67% and 79%.
So a tendency does exist — roughly 29 points between the most and least positive review cultures. But Project Zomboid alone swings 57 points across languages. The game-specific gap is about twice the size of the language-culture gap, which means most of what you see in the first figure is about the games, not about who is reviewing them.
Where a gap is large, it is usually pointing at something specific. Project Zomboid’s Chinese reviews are dominated by a single complaint: updates that break mods. That is also the leading complaint in its English corpus, at roughly half the rate — see who actually stops enjoying it. A separate piece examines the Chinese review text directly.
What this means for reading a store page
- The percentage shown to you is not filtered to your language. If you read reviews in one of these languages, Steam’s own language filter will show you a different picture than the headline.
- A wide spread is a signal to go looking. Songs of Syx’s 14-point range means the game is experienced similarly everywhere. Project Zomboid’s 57 means at least one community is hitting something the others are not.
- Server-region and mod-ecosystem differences travel by language. Where these gaps have an identifiable cause in this dataset, it is usually infrastructure or modding, not translation quality. The game with no gap worth reporting is also the one with no infrastructure complaint — the Songs of Syx curve.
Method, and its limits
Rates are simple: positive reviews divided by total, per language pull. That part is language- neutral and reproducible.
Topic analysis is not applied in this piece, and deliberately so. Keyword classification needs a validated keyword set per language; ours exists for English, Simplified Chinese and Russian only. Running English keywords over Portuguese text produces near-zero matches, which would look like “Brazilian players do not complain about performance” when it actually means the method did not look. Where this site does report topics by language, it names the language and the keyword set.
What could not be confirmed
- Why any given gap exists, except where a per-game piece examines the review text directly.
- Whether sample sizes are equivalent. 100 reviews per language was the target; some pulls returned fewer and were excluded below 25.
- Whether language equals region. A Simplified Chinese review can be written anywhere.
- Whether these gaps are stable over time. This is one read on 2026-08-27; the endpoint returns recent reviews, so no historical comparison is available.
- Anything about the other twenty-odd Steam languages not pulled here.
Sources
- Steam appreviews endpoint — 48 pulls, appids 108600, 1162750, 392160, 3932890, 3282300, 1368140, eight languages each, all read 2026-08-27
- Project Zomboid on Steam — read 2026-08-27
- Songs of Syx on Steam — read 2026-08-27
