Belarusian Pronunciation and IPA Transcription Guide

← Back to Belarusian transcription ↑ All Languages

Table of Contents

• Getting Started
  • About This Tool
  • Phonemic vs Phonetic
  • Dialects & Limitations
• Quick Reference (Glossary)
• How to Read IPA Symbols
  • Vowel Symbols
  • Consonant Symbols
  • Diacritical Marks
  • Interactive Features
  • Multiple Pronunciation Variants
• Belarusian Pronunciation Guide
  • Stress and How to Mark It
  • Hard and Soft Vowel Letters (е ё ю я і)
  • Soft Sign and Apostrophe
  • Consonant Letters
  • Affricates: дз, дж, ц, ч
  • Г, Ґ and the Stop in зг/жг
  • Ў (Non-syllabic U)
  • Devoicing and Voicing Assimilation
  • Sibilant and Palatal Assimilation
  • Gemination (Long Consonants)
  • Hard and Soft L
• Orthography to IPA Mapping
  • Comprehensive Mapping Tables
• Using the Tool: Practical Notes
• For Developers
  • Implementation Details
  • Common Issues & Limitations
• Related Pronunciation Guides

About This Tool

This Belarusian transcription app uses the Wiktionary Belarusian Pronunciation Module to generate phonemic transcriptions for Belarusian text. IPA (International Phonetic Alphabet) is a standardized system for transcribing speech sounds. This tool converts Belarusian Cyrillic spelling into IPA, helping learners, linguists, and developers see how a word is pronounced. Phonemic transcription represents the abstract sound system (phonemes), while phonetic transcription shows actual speech variations (allophones).

Unlike the German or French pages, there is no lexicon here. Every word is transcribed purely by the module's rules: the text is split on spaces, each word is cleaned (everything except letters, combining marks, apostrophes and hyphens is dropped) and handed to the module's toIPA function. Because the rules only see spelling, the quality of the output depends on the spelling you type, and above all on whether you mark the stress (see the box at the top of this page).

Belarusian spelling already records the vowel reductions of the spoken language (akanye: unstressed о is written а; yakanye: unstressed е is often written я). For that reason the module maps letters to sounds almost one-to-one and applies no vowel-reduction rules of its own. What it does model is consonant palatalization, voicing assimilation, final devoicing, sibilant assimilation, gemination, the hard/soft distinction of л, and the placement of the stress mark.

Dialects & Limitations

• Supported: Standard (literary) Belarusian as written in the official orthography, one style only: Default.
• Not supported: Dialects, and the classical orthography (Taraškievica) as such. Its soft-sign spellings (сьнег) are read through the ordinary soft-sign rule and happen to give the same result as the official spelling (снег): /sʲnʲex/ vs /sʲnʲex/.
• Cyrillic only: a word that contains any non-Cyrillic letter (for example Latin letters or the modifier-letter apostrophe U+02BC) is rejected by the module. In the web app the word is then shown unchanged.
• Edge cases: Loanwords, proper names, and compounds with more than one written о need an explicit stress mark; see Common Issues.

Phonemic vs Phonetic

• Phonemic (the only form offered for Belarusian): A broad transcription generated by rules, with no dictionary lookup. Although it is labelled phonemic, it already shows some surface detail that Wiktionary's Belarusian transcriptions include: palatalization ʲ, the non-syllabic u̯, hard ɫ and consonant length ː.
• Phonetic: not available for Belarusian.

Quick Reference (Glossary)

Phoneme

The smallest unit of sound in a language that distinguishes meaning (e.g., /p/ vs. /b/ in English "pat" vs. "bat").

Allophone

A variant pronunciation of a phoneme that doesn't change meaning.

Palatalization (soft consonant)

A consonant pronounced with the middle of the tongue raised towards the hard palate, written ʲ after the consonant, e.g. ліс /lʲis/. A consonant followed by е ё ю я і or ь is soft.

Hard consonant

A consonant with no palatalization, e.g. the ɫ in ла /ɫa/.

Soft sign (ь)

A letter with no sound of its own that makes the preceding consonant soft: соль /solʲ/.

Akanye / Yakanye

The Belarusian reduction of unstressed о to a and of unstressed е to ʲa. Belarusian spelling already writes it, so the module does not apply it: it reads а as a and я as ʲa exactly as written.

Affricate

A consonant that begins as a stop and ends as a fricative, e.g. t͡s (ц) and d͡z (дз).

Tie bar

The combining mark ͡ joining two symbols into one affricate: t͡ʂ, d͡ʐ.

Retroflex / hard sibilant

The symbols ʂ and ʐ are what Wiktionary uses for the hard ш and ж (and t͡ʂ for ч).

Velar fricative

The Belarusian г is a fricative ɣ (the voiced partner of x), not a stop. The stop ɡ is written ґ.

Non-syllabic u

The glide u̯ in кроў /krou̯/. It is the letter ў right after a vowel.

Final devoicing

Voiced obstruents become voiceless at the end of a word: зуб /zup/, горад /ˈɣorat/.

Voicing assimilation

Neighbouring obstruents agree in voicing: вакзал /vaɡzaɫ/ (к becomes voiced before з), лёгка /ˈlʲoxka/ (г becomes voiceless before к).

Sibilant assimilation

A sibilant takes over the place of a following sibilant: сшыты /ʂːɨtɨ/.

Gemination

Two identical consonants in a row are written as one long consonant with ː: каменне /kamʲenʲːe/.

Primary stress

The mark ˈ is placed before the stressed syllable. It comes from a combining acute (U+0301) after the stressed vowel, or from the automatic о rule.

Secondary stress

The mark ˌ, produced only from a combining grave (U+0300) typed by the user.

Word boundary marker

An internal # placed around each word during processing; it never appears in the output.

Foot boundary

A comma or dash in the module's input is converted to | (the web app removes punctuation before calling the module, so you will not normally see it).

NFKC

Unicode normalization applied by the app to each word before transcription; it composes letters such as й and ё and leaves the combining acute in place.

How to Read IPA Symbols

The tables below list the symbols that the Belarusian module can actually produce. Each example was generated by the real engine. English approximations are rough.

Vowel Symbols

Consonant Symbols

Diacritical Marks

Interactive Features

• Click words to cycle variants: not applicable here. The Belarusian module returns a single transcription for each word, so there are no alternatives to cycle through.
• Audio playback: Click the speaker icon next to any word or line to hear text-to-speech pronunciation (Belarusian voice code be-BY; requires browser or Edge TTS support). TTS reads the Cyrillic text, not the IPA, so it is independent of the stress marks you type.
• Export results: Use PDF or CSV buttons to save transcriptions in your preferred format.

Multiple Pronunciation Variants

For some languages (such as German or Russian) a word can have several valid transcriptions, and clicking the word cycles through them. Belarusian does not. The module is a deterministic function: the same spelling always gives the same single result. If a word has two stress patterns (homographs that differ only in stress) you choose the one you want by placing the acute accent yourself.

Belarusian Pronunciation Guide

This guide explains the spelling-to-sound correspondences as the module implements them. Belarusian spelling is close to phonemic: most letters have one value, and the module's job is mainly to apply palatalization and consonant assimilation and to place the stress mark.

Stress and How to Mark It

Belarusian stress is free (it can fall on any syllable) and is not marked in ordinary text. The module cannot predict it, so it relies on what you type.

Explicit Marks

• Acute accent (U+0301) after a vowel means primary stress: вада́ /vaˈda/, чалаве́к /t͡ʂaɫaˈvʲek/, сястра́ /sʲaˈstra/.
• Grave accent (U+0300) means secondary stress: ра̀дыёста́нцыя /ˌradɨjoˈstant͡sɨja/.
• The stress mark is moved to the left so that it stands before the syllable's onset consonant(s), as in ма́ці /ˈmat͡sʲi/ and кастры́чнік /kaˈstrɨt͡ʂnʲik/.

The Automatic О Rule

When a word carries no ˈ at all, has exactly one о (the letter ё counts as an о here, because it becomes ʲo), and has at least one other vowel, the module puts the primary stress on that о. A grave accent does not prevent it.

Because the rule only looks at the letters о/ё, it is wrong for loanwords that keep an unstressed о in spelling (фотаграфія gets its mark on фо). Always add an accent for such words.

Hard and Soft Vowel Letters (е ё ю я і)

Belarusian has five "hard" vowel letters а о у э ы and five "soft" ones я ё ю е і. A soft letter after a consonant makes that consonant soft and then gives the plain vowel; at the start of a word or after a vowel it gives the glide j plus the vowel.

The hard letters never soften: ла /ɫa/, лы /ɫɨ/, лэ /ɫɛ/. The letter э stays ɛ; the module never produces ɔ or ʲɛ (the source comments mention them as possible alternatives but they are not implemented).

Soft Sign and Apostrophe

• Soft sign ь becomes ʲ on the preceding consonant: соль /solʲ/, сталь /stalʲ/. After a soft sign, a soft vowel letter gives j: бельё /bʲeˈlʲjo/.
• Apostrophe (straight ' or typographic ’) becomes j and blocks softening: аб'ява /abjava/, п'я́ны /ˈpjanɨ/, сям'я́ /sʲaˈmja/. The modifier letter apostrophe U+02BC is not accepted (see Common Issues).

Consonant Letters

• ж, ш, ч are always hard (ʐ, ʂ, t͡ʂ) and ц is t͡s; before a soft vowel letter ц becomes soft: цяпер /t͡sʲapʲer/, чалавек /t͡ʂaɫavʲek/.
• й is j: зімо́й /zʲiˈmoj/.
• х is x: хле́б /ˈxlʲep/.
• ф is f: фатограф /faˈtoɣraf/.

Affricates: дз, дж, ц, ч

The letter pairs дз and дж are always read as one affricate with a tie bar, whenever the two letters stand next to each other (the module looks at two characters at a time): дзе́ці /ˈd͡zʲet͡sʲi/, джаз /d͡ʐas/, дзякуй /d͡zʲakuj/. Before a soft vowel or ь, дз is soft: дзень /d͡zʲenʲ/. At the end of a word or before a voiceless consonant the affricate is devoiced: будзь /but͡sʲ/, дзед /d͡zʲet/.

Г, Ґ and the Stop in зг/жг

• г is the fricative ɣ: га́лава /ˈɣaɫava/, кніга /knʲiɣa/.
• ґ is the stop ɡ: ґанак /ɡanak/.
• г directly after з, ж, с or ш is a stop ɡ: мазгі́ /mazˈɡʲi/, жгу́т /ˈʐɡut/, зго́да /ˈzɡoda/.
• At the end of a word or before a voiceless consonant, ɣ becomes x, and ɡ becomes k: лёгка /ˈlʲoxka/, снег /sʲnʲex/.

Ў (Non-syllabic U)

The letter ў is w in the middle of the module's processing and then turns into the glide u̯ when it stands after a vowel and is not followed by another vowel: праўда /prau̯da/, кроў /krou̯/, аўтобус /au̯ˈtobus/. At the start of a word or after a consonant it stays w: ўчора /ˈwt͡ʂora/, ўсход /wsxot/.

Devoicing and Voicing Assimilation

Voiced obstruents lose their voicing at the end of a word and before a voiceless consonant; voiceless obstruents gain it before a voiced one. The pairs the module knows are:

The sonorants (м н л р й) and в take no part: в is neither devoiced nor does it block assimilation, so встаць comes out as /vstat͡sʲ/ with v (see Common Issues). Each word is transcribed on its own, so there is no assimilation between words: a word-final voiced consonant is always devoiced, even if the next word starts with a vowel.

Sibilant and Palatal Assimilation

• Sibilant clusters merge: сшыты /ʂːɨtɨ/, расчэска /raʂt͡ʂɛska/, бяззубы /bʲazːubɨ/.
• з, с, н before a soft consonant become soft: зняць /zʲnʲat͡sʲ/, сцяна /sʲt͡sʲana/, ланцюг /ɫanʲt͡sʲux/, слёзы /ˈsʲlʲozɨ/.
• з/с before б м п ф в + soft vowel become soft too, when at least two vowels follow in the word or the cluster is at the start of the word: размяня́ць /razʲmʲaˈnʲat͡sʲ/, свята /sʲvʲata/, звязда́ /zʲvʲazˈda/.
• ц/дз before в + soft vowel become soft: цвёрды /ˈt͡sʲvʲordɨ/.

Gemination (Long Consonants)

Two identical consonants in a row (counting a following ʲ) are shown as one long consonant: ссылка /sːɨɫka/, каменне /kamʲenʲːe/, сёння /ˈsʲonʲːa/, жыццё /ʐɨˈt͡sʲːo/, аддзел /ad͡zʲːeɫ/. The same happens when assimilation makes two consonants identical (ззад /zːat/).

Hard and Soft L

Every л that is not followed by ʲ is written as the dark (hard) ɫ: лодка /ˈɫotka/, стол /stoɫ/. Soft л (before е ё ю я і ь) stays lʲ: ліса /lʲisa/, альтанка /alʲtanka/.

Comprehensive Spelling-to-IPA Mapping

Every example below was produced by the engine. A stress mark appears in the output only if it was typed (or added by the automatic о rule).

1. Vowel Letters

2. Simple Consonant Letters

3. Digraphs, Signs and Special Characters

4. Consonant plus Soft Vowel (Palatalization)

5. Final Devoicing and Assimilation Examples

6. Letters Outside the Belarusian Alphabet

Using the Tool: Practical Notes

Why does the output often have no stress mark?

The tool has no dictionary and cannot know where the stress falls, so it prints a stress mark only if you type one. The one automatic guess is that a word with exactly one о (or ё) and at least one other vowel is stressed on that letter: го́рад /ˈɣorat/ and горад /ˈɣorat/ give the same result, whereas вада /vada/ gets no mark. See Stress and How to Mark It. Open in the converter

How do I mark the stress?

Type a combining acute accent (U+0301) directly after the stressed vowel, as in вада́ /vaˈda/. A combining grave accent (U+0300) gives secondary stress. The letter ё is not treated as stressed by itself: ёсць /josʲt͡sʲ/ has no mark, while ё́сць /ˈjosʲt͡sʲ/ does.

Why is only one transcription shown for each word?

The Belarusian module is deterministic: the same spelling always gives the same single result, so there are no variants to cycle through. If a word has two stress patterns, choose the one you want by placing the accent yourself; see Multiple Pronunciation Variants.

Why can the output differ from a pronunciation dictionary?

Rules see only spelling, so the result depends on the spelling and stress you type. Each word is transcribed on its own, which means a word-final voiced consonant is always devoiced, even before a following vowel. The automatic о rule is wrong for loanwords that keep an unstressed о in spelling, so add an accent for such words. The output is labelled phonemic but includes some surface detail; see Phonemic vs Phonetic and Common Issues.

Implementation Details (for Developers)

The transcription is produced by lua_modules/be-pron_wasm.lua, a copy of the Wiktionary Belarusian pronunciation module. The web app calls window.be_ipa.toIPA(cleanText) (utils.js, group "Direct Generation"); there is no lexicon and no post-processing in JavaScript. Per word, the app first applies sanitize() (keeps only \p{L}, \p{M}, ', ’, ‿, -; turns ’ into '; NFKC), and if the module raises an error the original text is shown. The module's entry point is export.toIPA(term). It is a fixed pipeline:

1. Script check (checkScript(term, "Cyrl")).
2. convert(): punctuation, lowercasing, letter mapping, word boundaries, ʲ → j, г → ɡ after sibilants, the о stress rule, w → u̯.
3. assimilate_voicing, then assimilate_sibilants, then assimilate_palatals.
4. Hard ɫ, then gemination.
5. move_stress, then removal of the # markers.

Additional Key Processes

Helpers

rsub is mw.ustring.gsub keeping only the string result. do_sub_repeatedly(term, fun) and rsub_repeatedly(term, pattern, repl) apply a rule again and again until the string stops changing, so rules propagate along a cluster (for example through бяззубы).

Tables Used by the Pipeline

1. Entry and Script Check

2. Normalization (start of convert)

3. Letter-by-Letter Mapping

A while loop walks along the string. At each position it first looks up the next two characters in correspondences (this is how дз and дж are found), and otherwise the single character. A character without an entry is copied unchanged.

The source comments next to е and о (-- or ɛ, -- or ɔ) are only notes: no alternative vowels are generated.

4. Word Boundaries

Every run of spaces is wrapped in # ((%s+) → #%1#) and the whole string is wrapped in #…#. The marker lets later patterns recognise word starts and ends. It is removed at the very end.

5. Soft Sign Becomes j Between Vowels

6. Stop ɡ After a Sibilant

7. Stress Marking (the о Rule)

Two rsub_repeatedly calls place ˈ after a lone o. The character class [^#o ˈ] excludes the word boundary, other o and an existing primary stress, so each pattern matches only if the word has no primary stress and exactly one o.

This stage runs before ў is converted, so the glide does not count as a vowel: кроў /krou̯/ has one vowel and is left unmarked, здароў /zdaˈrou̯/ gets a mark.

8. w → u̯

9. assimilate_voicing

A loop (do_sub_repeatedly) that applies two rules until nothing changes. Characters in [ʲː͡]* (softness, length, tie bar) are skipped between the two consonants.

10. assimilate_sibilants

11. assimilate_palatals

Another do_sub_repeatedly loop. Its sub-rules, in the order they appear in the code:

12. Hard ɫ

13. Gemination

14. move_stress

Up to this point the stress mark sits after its vowel. move_stress shifts it to the syllable boundary. Its six rsub steps run in this order:

Known limitation of step 3: consonant does not contain ɫ, and perm_syl_onset lists kl/pl/bl with plain l, but hard л has already been converted to ɫ at this point. Clusters ending in hard л are therefore not recognised as onsets: прыкла́д /prɨkˈɫat/, апла́та /apˈɫata/.

15. Finish

All # markers are removed and the string is returned. The app displays it between slashes (phonemic form). Two other functions in the file are not used by the app: export.remove_pron_notations(text, remove_grave) (strips grave accents from annotations) and export.show(frame) (the Wiktionary template entry point, which adds brackets and formatting). The file also contains a leftover print(export.toIPA("Злу́чаныя Шта́ты Аме́рыкі")) that runs when the module is loaded.

Not Implemented in the Module

• No dictionary lookup, no stress dictionary, no stress prediction beyond the single-о rule.
• No vowel reduction (akanye/yakanye), no ɔ, no alternative ʲɛ.
• No devoicing of v, no voicing assimilation across word boundaries.
• No special handling of loanwords, proper names, Russian letters (щ ъ и pass through unchanged) or numerals.
• No clitic handling: every space-separated word is transcribed on its own, so prepositions and particles get their own stress logic and their own final devoicing.

Common Issues & Limitations

Known Transcription Problems

This table shows known issues where the automatic transcription may be incorrect. "Correct IPA" is the engine's output for the corrected input shown in brackets, unless it is a description.

Related Pronunciation Guides

Help pages for related languages in this tool:

• Russian
• Ukrainian
• Bulgarian
• Polish
• Czech
• Mongolian
• All languages

For technical issues or suggestions, please visit our GitHub repository.