Research
Language is part of the product.
Ethereum is global, but many of its products are still built primarily for English.
We tracked eight months of wallet clicks on ethereum.org, counted the languages that major apps, exchanges and wallets support, and reviewed the quality of translations in the most used apps.
The pattern is simple. Language can affect what people choose. But translating an interface does not guarantee people understand it.
01 · The opportunity
Sometimes, speaking the user’s language wins the user.
We compared wallet clicks across 18 language groups on ethereum.org.
In 4 markets, language clearly changed the choice. 90% of Japanese readers’ wallet clicks went to wallets that speak Japanese. Traditional Chinese, Indonesian and Korean readers showed the same pull, and 4 more languages a smaller one.
In those markets, language is part of the product decision, not just an accessibility feature.
- Traditional Chinese+38%
- Indonesian+24%
- Arabic+21%
- Japanese+20%
- Ukrainian+19%
- Korean+17%
- Turkish+10%
- Italian+9%
No clear difference in the other 10: Czech, Vietnamese, Spanish, German, Russian, Polish, French, Thai, Chinese, Portuguese. Where many readers are comfortable in English, their own language may matter less to the choice.
Translation is not universally a growth lever. But in the right market, it can shape which product gets chosen.
02 · The coverage gap
Global products still have very uneven language support.
We counted the interface languages supported by some of the most-used Ethereum apps, centralized exchanges, and wallets. The differences are striking.
Ethereum apps
Polymarket17 languages
Jumper16 languages
Uniswap13 languages
1inch13 languages
ENS8 languages
OpenSea7 languages
Tornado Cash7 languages
Aave4 languages
Blur3 languages
KyberSwap2 languages
Relay⚠️ English only
Matcha⚠️ English only
Aerodrome⚠️ English only
Across⚠️ English only
Sushi⚠️ English only
Morpho⚠️ English only
ether.fi⚠️ English only
Pendle⚠️ English only
Lido⚠️ English only
Curve⚠️ English only
CoW Swap⚠️ English only
Maple⚠️ English only
Sky⚠️ English only
Farcaster⚠️ English only
Exchanges
MEXC22 languages
Gate22 languages
OKX21 languages
Crypto.com20 languages
Binance18 languages
Coinbase18 languages
Kraken18 languages
Bitget18 languages
Bybit17 languages
KuCoin17 languages
WEEX17 languages
Ourbit16 languages
LBank14 languages
HTX13 languages
Toobit13 languages
BingX13 languages
Hotcoin12 languages
CoinW12 languages
WhiteBIT6 languages
Bitfinex6 languages
Bitstamp3 languages
Wallets
MetaMask43 languages
Phantom26 languages
TokenPocket22 languages
OKX Wallet18 languages
OneKey17 languages
Trust Wallet16 languages
Trezor Suite16 languages
Zerion14 languages
1inch Wallet13 languages
Bitget Wallet12 languages
Rainbow12 languages
Coin9812 languages
Coinbase Wallet11 languages
imToken10 languages
Ledger Wallet10 languages
Rabby9 languages
Exodus8 languages
MEW wallet3 languages
Safe⚠️ English only
Ambire⚠️ English only
* Counted from the product’s iPhone app listing. English counts as one language. See methodology.
03 · The quality problem
Being translated is not the same as being understood.
Crypto products are full of words that look ordinary in English but carry a specific meaning on screen.
Stake. Collateral. Gas. Bridge. Withdraw. Slippage.
Translate the dictionary meaning instead of the product meaning, and an interface can stay grammatically correct while becoming confusing, misleading or simply wrong.
- Everyday meanings on money wordsSome translations use the wrong meaning of a word. In one app’s Spanish version, the button to place a trade uses the word for “shop”, while the list of sell offers uses the word for “questions”.
- Many words for one actionAcross nine products, Spanish speakers meet five different words for “swap”. One wallet’s browser extension and mobile app disagree with each other.
- Languages in the menu, English on the screenSome products list dozens of languages, but many are only partly translated, and key pages fall back to English.
- Translations left switched offTwo of the apps we reviewed ship finished translations that users cannot turn on.
Teams make these decisions on their own. Some translate the term, some keep the English word, some bend it into the grammar of the other language. Each new product starts again.
Localization is product work.
Language can influence which product people choose. Poor terminology can undermine understanding. Coverage remains uneven. And teams keep solving the same language problems in isolation.
Better localization means deciding where language matters, what Ethereum concepts actually mean, and how to express them consistently.
There is a solution
Ethereum is global infrastructure.
Its products should feel global too.
Most Ethereum apps still speak only English, and those that translate pick different words for the same idea. ETHGlossary gives every team one community-reviewed set of Ethereum terms in 24 languages, so users meet the same words in every product.
Explore glossary.ethereum.orgResearch notes
Methodology & sources
Wallet choice on ethereum.org
Data. “Visit website” clicks on ethereum.org/wallets/find-wallet and its language editions, January 1 to September 2, 2026, from ethereum.org’s Matomo analytics. One click per wallet per visit.
Language groups. A reader’s language is their browser language. Traditional Chinese covers zh-TW, zh-HK and zh-MO; Portuguese covers every pt variant. Seven ethereum.org editions (Hindi, Bengali, Urdu, Telugu, Swahili, Tamil and Marathi) had too few clicks and were left out.
Wallet language support. The languages each wallet reports to ethereum.org (wallet-data.ts), as it stood each month. ethereum.org does not verify these.
The comparison. On the English page the wallet list is shuffled for every visitor, so list position cannot explain a difference. For each language, we compared the share of clicks going to wallets that support it from readers with that browser language against readers with an English browser. Ranges are day-level bootstrap 90% intervals. The bars show the relative difference between the two shares. A language is charted when the interval sits above zero and the shares differ by 4 percentage points or more. Japanese and Traditional Chinese rest on about 298 and 207 clicks, so treat them as strong signals rather than precise sizes.
Why not the language pages. On a language edition, wallets that support the language are moved to the top of the list, so their higher share mixes language with position. Portuguese is the clearest case: the Portuguese-supporting wallets get 52% of clicks on the English page and 81% on the Portuguese page, yet Portuguese-browser readers on the English page pick them slightly less often than English readers do (−5 points). Most of that jump is position. The same browser-language comparison on the language pages points the same way but has too few English-browser readers to chart.
Limits. People who read the English page may be more comfortable in English than those who use their own edition, so these numbers probably understate the preference. A preference can also reflect regional brand loyalty rather than language. Clicks show which wallet was chosen, not what happened after. A new catalog layout went live on September 3, 2026 and is not included.
Translation quality and terminology
Quality example. Polymarket’s Spanish interface (polymarket.com/es), read October 15, 2026. Asks, Trade and Positions were seen on public market and profile pages without logging in; Balance (“Polymarket Balance” → “Equilibrio de Polymarket”) ships in the live production code of the deposit flow, which needs a login. Polymarket uses the correct words elsewhere (Operar, Posiciones, Saldo). Each term was checked by translating the shipped word back to English in context. Uniswap, MetaMask and ENS were checked the same way and came back largely clean.
Divergence. Translation files that users can select in the live product, October 2026: MetaMask (extension and mobile), Rabby, Trezor Suite, Ledger Wallet, Uniswap, 1inch, OpenSea, Jumper and ENS. Only the main swap action and the standalone wallet label or Connect wallet button were compared, within one language; capitalisation and noun or verb forms count as one word. Spanish mixes Latin American and European catalogs, but the wallet split also appears within European Spanish alone. In Brazilian Portuguese and Russian the same terms came back consistent.
Language coverage
Apps. Twenty-four widely used Ethereum ecosystem apps that people use through their own interface, chosen editorially. Onchain activity (30-day active addresses on growthepie, fetched October 15, 2026) is one signal of relevance, not a ranking: for aggregators such as Jumper, Relay and Matcha it includes use inside other apps, and for apps such as Polymarket and Farcaster it covers only part of their contracts, so those were included on reach and category. Launchpads and memecoins are left out. Polymarket runs mostly on Polygon PoS. Interface languages verified from locale files in public repositories, production JS bundles or the live language picker (July 2026 for apps in the first edition, October 2026 for the rest). A language counts only if users can select it: CoW Swap ships Spanish, French, Brazilian Portuguese and Russian behind a switched-off flag, and Sky ships Spanish and Korean but forces English, so both count as English only. Farcaster was checked in its web app and iPhone app, Tornado Cash at tornadocash.eth.limo, and Maple from an archived copy of its app (September 27, 2026) because the live app blocked automated access. English-only verdicts for closed-source apps come from inspecting their live bundles and markup; high confidence, short of certainty.
Exchanges. Eleven widely known large venues, chosen without a ranking, plus the ten largest others by CoinMarketCap 24-hour spot volume on October 15, 2026 that pass both CoinMarketCap’s spot score (8.0 or more) and CoinGecko’s trust score (8 or more), leaving out institution-only and single-country venues such as Upbit and Bithumb. Counts are the languages each website declares to search engines, read October 5 and 15, 2026 from Europe; Binance, Bybit, Bitfinex, Bitstamp and WhiteBIT from their language menus (WhiteBIT’s Georgian and Turkish entries point to separate regional sites). Coinbase and Gate block automated reads, so their counts come from their iPhone app listings. Exchanges can offer different languages by region.
Wallets. Twenty widely used Ethereum wallets: the ten from the first edition and TokenPocket, imToken, Coin98 and Ledger by reported user count, plus six well-known Ethereum wallets (OneKey, Trezor, 1inch Wallet, MEW wallet, Safe and Ambire). Counts are the languages each wallet reports to ethereum.org. OKX Wallet is not listed there and Exodus’s entry says English only, so both use their iPhone app listings. Trezor’s entry lists 5 languages but Trezor Suite ships 16, so its count comes from the Suite source code. Safe’s and Ambire’s English-only entries were confirmed in their public source code.
Counting. English counts as one language. Simplified and Traditional Chinese count as two. Counting languages says nothing about quality.
Spot-check. Exchange and wallet counts are what products declare, so in October 2026 we checked 10 of them (Bybit, MEXC, OKX, Crypto.com, Kraken, MetaMask, Phantom, Trust Wallet, Zerion, Rabby) against the languages users can actually pick. The live menus offered more languages than declared in 6 cases and fewer in 4, so declared counts do not systematically inflate the gap with apps. Some offered languages are only partly translated: about 15 of MetaMask’s 52 extension languages are close to complete, and Crypto.com’s prices page stays in English after switching language. The counts measure breadth, not completeness.