To translate a website into Polish, Czech, Hungarian or Romanian properly, use a native-speaking translator with subject knowledge, give them a glossary and context for every string, and make sure your templates can handle grammar that English never needs: noun cases, several plural forms, gendered adjectives and characters such as ł, ř, ő and ș. Machine translation is fine for a first draft, but these four languages expose its weaknesses quickly, and the visible errors (broken plurals, wrong formality, missing diacritics) are exactly what makes a site look untrustworthy.
This guide covers the grammar that breaks templates, diacritics and encoding, formality, local formats, and how to find and manage reliable native translators.
Grammar that breaks templates
Most localisation bugs in CEE languages come from code that assumes English grammar. The table summarises the main traps.
| Feature | Polish | Czech | Hungarian | Romanian |
|---|---|---|---|---|
| Noun cases | 7 | 7 | Many suffix-based cases | Few, but the definite article is attached to the noun |
| Plural categories (CLDR) | one, few, many, other | one, few, many, other | one, other | one, few, other |
| Grammatical gender | Yes | Yes | No | Yes |
| Typical trap | "2 pliki" but "5 plików" | "2 soubory" but "5 souborů" | Nouns stay singular after numbers: "5 fájl" | "20 de fișiere" needs "de" from 20 upwards |
Plurals
English has two plural forms; Polish and Czech effectively use three for whole numbers, and Romanian adds a "de" construction for larger numbers. Never build strings such as "You have " + n + " new messages". Use a proper internationalisation library that supports CLDR plural rules (ICU MessageFormat or your framework's equivalent) and give translators every plural form to fill in.
Cases and concatenation
Because nouns change form depending on their role in the sentence, inserting a variable into a fixed sentence often produces wrong grammar. "Welcome to {site}" or "Message from {name}" can need a different form of {site} or {name} in Polish or Czech. Prefer whole sentences per context, avoid splitting sentences across several strings, and let translators reorder placeholders freely.
Gender agreement
In Polish, Czech and Romanian, adjectives and past-tense verbs agree with gender. A button saying "Saved" may need different forms depending on what was saved. Provide context notes for each string ("refers to profile", "refers to photo").
Hungarian specifics
Hungarian is agglutinative: grammar is expressed through suffixes, so words become long, and suffix choice depends on vowel harmony with the word it attaches to. Brand names and variables therefore need suffixes that change with the word, which templates rarely anticipate. Hungarian also puts the family name first, so "{first} {last}" name fields should be reorderable. Our guide to affiliate marketing in Hungary covers the market itself.
Diacritics and encoding
- Use UTF-8 everywhere: database, files, emails, feeds, CSV exports and PDF generation.
- Check your fonts. Some decorative web fonts lack ł, ř, ő or ű, and browsers silently fall back to another font for those letters, which looks broken. Test a pangram in each language.
- Hungarian ő and ű. These double-acute letters are often replaced by õ and û when text passes through older encodings. Search translations for õ and û before publishing.
- Romanian ș and ț. The correct letters have a comma below (ș, ț). Older systems use cedilla forms (ş, ţ), which look wrong to Romanian readers and can split search results. Normalise to comma-below.
- URLs. Transliterate slugs (łóżko becomes lozko, příspěvek becomes prispevek) but keep diacritics in visible text, titles and meta descriptions. Missing diacritics in visible copy read as careless.
Formality and tone
All four languages distinguish informal and formal "you", and the choice changes verbs and pronouns throughout the copy.
| Language | Informal | Formal | Common practice online |
|---|---|---|---|
| Polish | ty | Pan / Pani, Państwo (plural) | Informal common on consumer sites; formal for finance, legal and B2B |
| Czech | ty (tykání) | vy (vykání) | Formal vy remains common and safe; informal for youth-oriented brands |
| Hungarian | te | Ön | Many consumer brands use te; Ön for official and older audiences |
| Romanian | tu | dumneavoastră | Formal common in commerce and support; informal on casual brands |
Decide once per language, write it into the style guide and check it in every string, including emails and error messages. Mixed formality is one of the most common signs of patchy translation. For adult and dating copy, tone and euphemism matter even more; see localising adult and dating copy.
Local formats
- Dates: Polish and Romanian 18.10.2026; Czech 18. 10. 2026; Hungarian 2026. 10. 18.
- Numbers: all four use a decimal comma. Polish, Czech and Hungarian group thousands with a space; Romanian commonly uses a dot.
- Currency: 49,99 zł, 249 Kč, 4 990 Ft, 49,99 lei. See pricing in PLN, CZK, HUF and RON.
- Addresses and phone numbers: use local postcode formats and do not force a US-style state field.
Let a locale library (the browser's Intl APIs, or ICU on the server) format dates, numbers and currencies rather than hard-coding patterns.
Finding reliable native translators
- Hire native speakers of the target language, living in or closely connected to the market, with experience in your subject.
- Run a paid test of 300 to 500 words including UI strings, a marketing paragraph and something technical. Have a second native speaker review it.
- Provide a glossary and style guide: formality, brand terms that stay in English, preferred terms for key concepts and words to avoid.
- Give context. Screenshots, character limits and string notes prevent most errors.
- Use translation memory through a translation management system so terms stay consistent and you do not pay twice for repeated text.
- Separate SEO from translation. Translators translate meaning; keyword research needs to be done in the target language. See SEO for Polish websites and Seznam SEO for Czech search.
Machine translation plus editing: machine translation followed by full native editing can cut costs for large volumes of informational content. Budget for genuine rewriting, not light proofreading, and keep landing pages, checkout and legal text fully human-translated. Our comparison of machine vs human translation for SEO sets out where each works.
If your content is adult or adult-adjacent, confirm up front that the translator is comfortable with the subject; many agencies decline such work. Our content providers directory lists services that work with the industry.
Frequently asked questions
How much does it cost to translate a website into Polish?
Professional translators usually charge per word, with rates varying by subject, volume and turnaround. Get quotes from two or three native translators based on the same sample, and remember to budget separately for review and SEO keyword work.
Can one translator cover Czech and Slovak?
Czech and Slovak are closely related and mutually intelligible, but they are separate languages with different spelling and vocabulary. Slovak readers notice Czech text immediately, so use a native Slovak translator for a Slovak site.
Which CEE language is hardest to localise?
Hungarian usually causes most template problems because of its suffixes, word length and name order. Polish and Czech plurals and cases come close behind.
Next steps
Before commissioning translation, audit your templates for concatenated strings and plural handling, then pilot one language with a paid test. For the wider process across all European markets, read website localisation for European markets.