Polish Pronunciation and IPA Transcription Guide

← Back to Polish transcription ↑ All Languages

Table of Contents

About This Tool

This Polish transcription app uses a modified version of the Wiktionary Polish pronunciation module (pl-pron_wasm.lua; on Wiktionary its successor is Module:zlw-lch-IPA, which {{pl-pr}} calls) running inside the browser via Wasmoon (Lua 5.4 WebAssembly). It converts standard Polish orthography into the International Phonetic Alphabet (IPA).

Polish spelling is highly phonetic and rule-governed — far more so than English or even German. The tool simulates the exact phonetic transformations (devoicing, assimilation, palatalization, and nasal vowel decomposition) applied by native speakers, producing both phonemic and narrow phonetic transcriptions without requiring a lookup dictionary. This rule-based approach means it can transcribe any Polish word, including neologisms and rare compounds.

Dialects & Limitations

Interactive Features

Note: Unlike German or French, Polish transcription is entirely rule-based — there is no lexicon lookup. This means every word gets a single deterministic transcription based on orthographic rules. There are no "alternative variants" to cycle through.

Phonemic vs Phonetic

Quick Reference (Glossary)

Palatalization (Zmiękczenie)
The modification of consonant articulation by raising the tongue towards the hard palate. In Polish, it occurs before the letter i or is indicated by a diacritic acute accent (e.g., ć, ś, ń).
Decomposition of Nasals
The process where nasal vowels (ą, ę) are split into an oral vowel followed by a nasal consonant (like [ɔn], [ɛm], or [ɔŋ]) depending on the following consonant.
Final Devoicing (Wygłos)
The neutralization of contrast between voiced and voiceless obstruents at the end of a word (e.g., chleb "bread" is pronounced as if it ended in p: [xlɛp]).
Regressive Assimilation
A phonological change where a consonant adapts its voicing to match the following consonant (e.g., prośba "request" has ś voiced to [ʑ] due to the following voiced b).
Progressive Devoicing
A specific Polish assimilation rule where the voiced consonants w (/v/) and rz (/ʐ/) become voiceless [f] and [ʂ] when they follow a voiceless consonant (e.g., kwiat [kfjat], trzy [tʂɨ]).
Affricate (Afrykata)
A consonant sound starting as a stop and releasing immediately into a fricative at the same place of articulation (e.g., c /t͡s/, cz /t͡ʂ/, ć /t͡ɕ/).
Penultimate Stress (Akcent paroksytoniczny)
The standard stress pattern in Polish where the stress falls on the second-to-last syllable of a word.
Phoneme
The smallest unit of sound that distinguishes meaning (e.g., /p/ vs. /b/ in pas "belt" vs. bas "bass").
Allophone
A variant pronunciation of a phoneme that doesn't change meaning (e.g., aspirated [pʰ] vs. unaspirated [p] in different positions).
Sonorant
A consonant produced with continuous airflow (m, n, ɲ, ŋ, l, r, j, w). Sonorants are "neutral" in Polish voicing assimilation — they neither trigger nor undergo devoicing of neighboring obstruents.
Obstruent
A consonant produced by obstructing airflow (stops, fricatives, affricates). In Polish, obstruents are subject to voicing assimilation. Contrasts with sonorants.
Syllable Boundary
A marker (.) showing where syllables divide. Polish syllabification follows the Sonority Sequencing Principle: consonant clusters are divided so that each syllable rises in sonority toward its nucleus.
Geminate
A doubled consonant sound (e.g., /kk/ in oddychać). Polish orthography often shows this with consonant doubling.
Onset
The initial consonant(s) of a syllable (e.g., "p" in "pan"). Polish allows complex onsets like /pr/, /tr/, /kr/, /sp/, /st/, /sk/.
Coda
The final consonant(s) of a syllable (e.g., "t" in "kot"). Polish codas are subject to final devoicing.
Tie Bar
The combining tie bar (◌͡◌) used in IPA to indicate affricates — two segments forming a single phonological unit (e.g., /t͡s/, /t͡ʂ/, /t͡ɕ/).
Homorganic
Produced at the same place of articulation. In nasal decomposition, the nasal consonant matches the place of the following stop (e.g., ą + b → [ɔm] — both bilabial).

How to Read IPA Symbols

Vowel Symbols

Polish contains six oral vowels and two nasal diphthongs. Unlike English, there is no phonemic distinction between long and short vowels.

IPA Polish Grapheme Example English Approximation Description
/a/ a tak [tak] father (shortened) Open central unrounded vowel
/ɛ/ e ten [tɛn] bed Open-mid front unrounded vowel
/i/ i pić [pit͡ɕ] meet (shortened) Close front unrounded vowel
/ɨ/ y ty [tɨ] roses (unstressed) Close central unrounded vowel
/ɔ/ o to [tɔ] caught (British short) Open-mid back rounded vowel
/u/ u, ó tu [tu], bóg [buk] boot (shortened) Close back rounded vowel
/ɔw̃/ ą (word-final / before fricatives) są [sɔw̃], wąski [ˈvɔw̃s.ki] French "on" + English "w" Nasalized back rounded diphthong
/ɛw̃/ ę (before fricatives) węch [vɛw̃x] French "in" + English "w" Nasalized front unrounded diphthong

Consonant Symbols

Polish consonants are grouped into dental, retroflex (hard), alveolo-palatal (soft), labial, and velar sounds.

IPA Polish Graphemes Example English Approximation Description
/p/, /b/ p, b pan [pan], bok [bɔk] pin, bin Bilabial plosives
/t/, /d/ t, d to [tɔ], dom [dɔm] toy, do (dental articulation) Dental plosives
/k/, /ɡ/ k, g kot [kɔt], gra [ɡra] keep, go Velar plosives
/f/, /v/ f, w fakt [fakt], wóz [vus] fit, voice Labiodental fricatives
/s/, /z/ s, z syn [sɨn], zoolog [zɔˈɔ.lɔk] see, zoo Dental sibilants
/ʂ/, /ʐ/ sz, ż, rz szum [ʂum], żaba [ˈʐa.ba] hash, vision (but retracted) Voiceless/voiced retroflex fricatives
/ɕ/, /ʑ/ ś, si śmiech [ɕmjɛx], ziemia [ˈʑɛ.mja] sheet, azure (hissing/soft) Voiceless/voiced alveolo-palatal fricatives
/x/ ch, h chleb [xlɛp], hak [xak] Scottish loch Voiceless velar fricative
/t͡s/, /d͡z/ c, dz co [t͡sɔ], dzban [d͡zban] cats, ads Dental affricates
/t͡ʂ/, /d͡ʐ/ cz, dż czas [t͡ʂas], dżem [d͡ʐɛm] church, jam (hard/retracted) Retroflex affricates
/t͡ɕ/, /d͡ʑ/ ć, ci, dź, dzi ćma [t͡ɕma], dzień [d͡ʑɛɲ] cheep, geep (soft/palatal) Alveolo-palatal affricates
/m/, /n/ m, n mama [ˈma.ma], noc [nɔt͡s] more, no Bilabial, dental nasals
/ɲ/ ń, ni koń [kɔɲ], niebo [ˈɲɛ.bɔ] canyon, Spanish ñ Alveolo-palatal nasal
/ŋ/ n (before k, g) bank [baŋk], gong [ɡɔŋk] sing Velar nasal
/l/ l las [las] leaf (clear L) Alveolar lateral approximant
/w/ ł ładny [ˈwad.nɨ] wet Labio-velar approximant
/j/ j jutro [ˈjut.rɔ] yes Palatal approximant
/r/ r ręka [ˈrɛŋ.ka] Spanish r (rolled) Alveolar trill

Diacritical Marks & Stress

Symbol Name Function Example
ˈ Primary stress Marks the onset of the stressed syllable Warszawa /varˈʂa.va/
. Syllable boundary Divides phoneme sequences into syllables okno /ˈɔk.nɔ/
ʲ Palatalization marker Indicates secondary narrowing of the vocal tract (softening) pies /pjɛs/
͡ Tie bar Indicates that two letters form a single affricate sound noc /nɔt͡s/ (not t followed by s)

Polish Phonetics & Pronunciation Guide

Oral Vowels

Polish oral vowels are remarkably stable and monophthongal. The language does not have phonemic vowel length distinctions (like English ship vs. sheep or German Staat vs. Stadt). However, minor contextual shifts exist:

Nasal Vowels (ą, ę) in Depth

Polish is unique among Slavic languages in retaining nasal vowels (spelled ą and ę). In modern standard Polish, these are actually pronounced as nasal diphthongs or decompose into oral vowels + nasal consonants based on context. The Lua transcription module mimics this behavior using the following rules:

1. Before Obstruent Stops & Affricates (Decomposition)

The nasal vowel loses its vowel nasality and splits into a standard oral vowel (/ɔ/ or /ɛ/) followed by a homorganic nasal consonant matching the place of articulation of the following stop/affricate:

Context (Next Sound) Type ą becomes... ę becomes... Example Grapheme Phonetic Output
b, p Bilabial [ɔm] [ɛm] dąb, zęby [dɔmp], [ˈzɛmbɨ]
d, t, c, dz, cz, dż Dental / Retroflex [ɔn] [ɛn] kąt, chętny [kɔnt], [ˈxɛntnɨ]
ć, dź, ci, dzi Alveolo-palatal [ɔɲ] [ɛɲ] pięć, dociąć [pjɛɲt͡ɕ], [ˈdɔt͡ɕɔɲt͡ɕ]
g, k Velar [ɔŋ] [ɛŋ] mąka, ręka [ˈmɔŋka], [ˈrɛŋ.ka]

2. Before Fricatives (Retained Nasal Diphthong)

Before fricatives (f, v, s, z, sz, ż, rz, ch, h), the nasalization cannot resolve into a distinct nasal stop. Instead, it is pronounced as an oral vowel followed by a nasalized labiovelar approximant: ą → [ɔw̃], ę → [ɛw̃]. E.g., wąski [ˈvɔw̃s.ki], węch [vɛw̃x]. The alveolo-palatal fricatives ś and ź are the exception: before them the module gives an oral vowel followed by [ɲ], as in gęś [ɡɛɲɕ] and gałąź [ˈɡa.wɔɲɕ].

3. Before Laterals and Approximants (Denasalization)

Before the letters l and ł, the nasal vowels completely lose their nasality. ą becomes standard [ɔ] and ę becomes [ɛ]. This is handled at the grapheme-to-phoneme conversion stage — the letters2phones table has explicit entries for ą + l, ą + ł, ę + l, ę + ł that produce denasalized output directly. E.g., wziął [vʑɔw], wzięli [ˈvʑɛ.li].

4. Word-Final Position

Palatalization & Soft Consonants

Palatalization represents the division between "hard" and "soft" sounds in Polish orthography. There are two primary mechanisms:

  1. Diacritics: Consonants with acute accents (ć, dź, ś, ź, ń) are inherently soft alveolo-palatal sounds (/t͡ɕ, d͡ʑ, ɕ, ʑ, ɲ/).
  2. The letter i:
    • If followed by another vowel, the i is silent and acts purely as a softening marker for the preceding consonant. E.g., pies [pjɛs].
    • If followed by a consonant or at the end of a word, the i is the vowel nucleus [i]. The consonant before it is slightly palatalized in speech, which the module does not mark. E.g., pili [ˈpi.li].

When dental fricatives/affricates (s, z, c, dz) are palatalized by i or converted to diacritics, they transform completely: s → ś [ɕ], z → ź [ʑ], c → ć [t͡ɕ], dz → dź [d͡ʑ].

Voicing & Devoicing Rules

Polish exhibits pervasive voicing assimilations within consonant clusters and at word boundaries:

Consonant Cluster Assimilations

Polish contains complex consonant clusters. In some contexts, progressive voicing/devoicing rules apply to specific consonants:

Stress Patterns & Exceptions

Polish stress is predominantly fixed on the penultimate (second-to-last) syllable, e.g. domu [ˈdɔ.mu], warszawa [varˈʂa.va]. The Lua module recognises one group of grammatical exceptions, the conditional endings, by matching the end of the word against two lists (third_last_syllable_stress and fourth_last_syllable_stress):

Orthography to IPA Mapping

Below is a highly detailed, comprehensive spelling-to-IPA mapping guide reflecting the rule structures of the pl-pron_wasm.lua engine. These tables list every possible Polish grapheme, digraph, and context-dependent sound modification.

1. Vowels (Oral & Nasal Contexts)

Unlike English or German, Polish vowels maintain stable lengths. However, nasal vowels (ą and ę) split or lose nasality depending on their phonological environment:

Spelling Primary IPA Context / Phonological Rule Example Word IPA Output
a /a/ Always (open central unrounded) sam [sam]
e /ɛ/ Always (open-mid front unrounded) ten [tɛn]
i /i/ Vowel nucleus (when not preceding another vowel) mit [mit]
o /ɔ/ Always (open-mid back rounded) nos [nɔs]
u, ó /u/ Always (close back rounded) but, bóg [but], [buk]
y /ɨ/ Always (close central unrounded; only after hard consonants) syn [sɨn]
au /aw/ Vowel + glide (loanwords, e.g., auto) auto [ˈaw.tɔ]
eu /ɛw/ Vowel + glide (loanwords, e.g., euro) euro [ˈɛw.rɔ]
ą /ɔw̃/ Word-final position (nasal diphthong) robią [ˈrɔ.bjɔw̃]
Before fricatives (e.g., s, z, sz, ż, rz, f, w, ch; before ś, ź the result is a vowel + ɲ) wąski [ˈvɔw̃s.ki]
/ɔm/ Before bilabial stops (p, b) dąb [dɔmp] (with final devoicing)
/ɔn/ Before dental/alveolar stops/affricates (t, d, c, dz, cz, dż) kąt [kɔnt]
/ɔɲ/ Before alveolo-palatal affricates/stops (ć, dź, ci, dzi) bądź [bɔɲt͡ɕ]
/ɔŋ/ Before velar stops (k, g) mąka [ˈmɔŋ.ka]
/ɔ/ Before laterals and semivowels (l, ł) (denasalized) wziął [vʑɔw]
ę /ɛ/ Word-final position (standard denasalization) idę [ˈi.dɛ]
/ɛw̃/ Before fricatives (e.g., s, z, sz, ż, rz, f, w, ch; before ś, ź the result is a vowel + ɲ) węch [vɛw̃x]
/ɛm/ Before bilabial stops (p, b) zęby [ˈzɛm.bɨ]
/ɛn/ Before dental/alveolar stops/affricates (t, d, c, dz, cz, dż) chętny [ˈxɛn.tnɨ]
/ɛɲ/ Before alveolo-palatal affricates/stops (ć, dź, ci, dzi) pięć [pjɛɲt͡ɕ]
/ɛŋ/ Before velar stops (k, g) ręka [ˈrɛŋ.ka]
/ɛ/ Before laterals and semivowels (l, ł) (denasalized) wzięli [ˈvʑɛ.li]

2. Standard Consonants (Hard / Unpalatalized)

Polish consonants articulated without palatalization or subsequent assimilation:

Spelling IPA Description Example Word IPA Output
b /b/ Voiced bilabial stop bas [bas]
p /p/ Voiceless bilabial stop pan [pan]
d /d/ Voiced dental stop dom [dɔm]
t /t/ Voiceless dental stop ten [tɛn]
g /ɡ/ Voiced velar stop gra [ɡra]
k /k/ Voiceless velar stop kot [kɔt]
f /f/ Voiceless labiodental fricative fala [ˈfa.la]
w /v/ Voiced labiodental fricative woda [ˈvɔ.da]
s /s/ Voiceless dental sibilant fricative sen [sɛn]
z /z/ Voiced dental sibilant fricative ząb [zɔmp]
h, ch /x/ Voiceless velar fricative hak, chleb [xak], [xlɛp]
m /m/ Bilabial nasal mama [ˈma.ma]
n /n/ Dental nasal noc [nɔt͡s]
l /l/ Alveolar lateral approximant (always clear) las [las]
ł /w/ Labiovelar semivowel ładny [ˈwad.nɨ]
r /r/ Alveolar trill rok [rɔk]
j /j/ Palatal approximant (semivowel) jutro [ˈjut.rɔ]

3. Digraphs & Affricates

Polish makes extensive use of digraphs to represent retroflex, alveolo-palatal, and dental affricates/fricatives:

Spelling IPA Description Example Word IPA Output
c /t͡s/ Voiceless dental affricate co [t͡sɔ]
cz /t͡ʂ/ Voiceless retroflex affricate czas [t͡ʂas]
dz /d͡z/ Voiced dental affricate dzban [d͡zban]
dż /d͡ʐ/ Voiced retroflex affricate dżem [d͡ʐɛm]
sz /ʂ/ Voiceless retroflex fricative szum [ʂum]
ż /ʐ/ Voiced retroflex fricative żona [ˈʐɔ.na]
rz /ʐ/ Voiced retroflex fricative (standard context) rzeka [ˈʐɛ.ka]
/ʂ/ Voiceless retroflex fricative (after voiceless: p, t, k, ch) trzy, przy [tʂɨ], [pʂɨ]
ć /t͡ɕ/ Voiceless alveolo-palatal affricate ćma [t͡ɕma]
dź /d͡ʑ/ Voiced alveolo-palatal affricate dźwig [d͡ʑvik]
ś /ɕ/ Voiceless alveolo-palatal fricative śpi [ɕpi]
ź /ʑ/ Voiced alveolo-palatal fricative źle [ʑlɛ]
ń /ɲ/ Alveolo-palatal nasal consonant koń [kɔɲ]

4. Palatalized (Soft) Consonants Before "i"

Before a vowel letter, i is not a vowel of its own. After the coronals c, dz, s, z, n it turns them into the separate soft consonants t͡ɕ, d͡ʑ, ɕ, ʑ, ɲ; after labials and velars the module writes it as the glide j. Before a consonant or at the end of a word, i is the vowel i. Consonants before i and j are slightly palatalized in speech, but the module does not mark this with ʲ:

Spelling IPA (Before Vowel) Example (Before Vowel) IPA (Before Consonant/Final) Example (Before Consonant/Final)
ci /t͡ɕi/ ciasto [ˈt͡ɕas.tɔ] /t͡ɕi/ cicho [ˈt͡ɕi.xɔ]
dzi /d͡ʑi/ działo [ˈd͡ʑa.wɔ] /d͡ʑi/ dziki [ˈd͡ʑi.ki]
si /ɕ/ siarka [ˈɕar.ka] /ɕi/ siła [ˈɕi.wa]
zi /ʑ/ ziarno [ˈʑar.nɔ] /ʑi/ zima [ˈʑi.ma]
ni /ɲ/ niebo [ˈɲɛ.bɔ] /ɲi/ niski [ˈɲis.ki]
bi /bj/ biały [ˈbja.wɨ] /bi/ bitwa [ˈbit.fa]
pi /pj/ pies [pjɛs] /pi/ piwo [ˈpi.vɔ]
wi /vj/ wiatr [vjatr] /vi/ wino [ˈvi.nɔ]
fi /fj/ fiord [fjɔrt] /fi/ figa [ˈfi.ɡa]
mi /mj/ miasto [ˈmjas.tɔ] /mi/ mili [ˈmi.li]
ki /kj/ kiedy [ˈkjɛ.dɨ] /ki/ kino [ˈki.nɔ]
gi /ɡj/ giąć [ɡjɔɲt͡ɕ] /ɡi/ gips [ɡips]
hi, chi /xj/ hiena [ˈxjɛ.na] /xi/ chichot [ˈxi.xɔt]
li /lj/ liana [ˈlja.na] /li/ list [list]

5. Voicing & Devoicing Assimilation Contexts

Consonants shift their voicing state systematically in clusters and at the end of words:

Spelling Underlying Phoneme Resulting IPA Triggering Context / Position Example Word IPA Output
b /b/ /p/ Word-final (devoicing) ząb [zɔmp]
Regressive (before voiceless consonant) babka [ˈbap.ka]
d /d/ /t/ Word-final (devoicing) sad [sat]
Regressive (before voiceless consonant) wódka [ˈvut.ka]
g /ɡ/ /k/ Word-final (devoicing) róg [ruk]
Regressive (before voiceless consonant) lekki [ˈlɛk.ki]
w /v/ /f/ Word-final (devoicing) krew [krɛf]
Progressive (following voiceless consonant) twój, kwiat [tfuj], [kfjat]
z /z/ /s/ Word-final (devoicing) mróz [mrus]
Regressive (before voiceless consonant) rozkaz [ˈrɔs.kas]
ż, rz /ʐ/ /ʂ/ Word-final (devoicing) nóż, talerz [nuʂ], [ˈtalɛʂ]
Regressive (before voiceless consonant) drzewko [ˈdʐɛf.kɔ]
dz /d͡z/ /t͡s/ Word-final (devoicing) ksiądz [kɕɔnt͡s]
dź /d͡ʑ/ /t͡ɕ/ Word-final (devoicing) łódź [wut͡ɕ]
ź /ʑ/ /ɕ/ Word-final (devoicing) gałąź [ˈɡa.wɔɲɕ]
k /k/ /ɡ/ Regressive voicing (before voiced obstruent) także [ˈtaɡ.ʐɛ]
p /p/ /b/ Regressive voicing (before voiced obstruent) jakby [ˈjaɡ.bɨ]
t /t/ /d/ Regressive voicing (before voiced obstruent) liczba [ˈlid͡ʐ.ba] (with cz → dż)
ś /ɕ/ /ʑ/ Regressive voicing (before voiced obstruent) prośba [ˈprɔʑ.ba]
ch, h /x/ /ɣ/ Regressive voicing (before voiced obstruent) niechże [ˈɲɛɣ.ʐɛ]

6. Foreign Letters & Loanword Graphemes

Polish handles recent loanwords and non-standard letters through phonetic approximations matching standard Polish orthography rules:

Spelling IPA Output Phonetic Mapping Principle Example Word IPA Output
x /ks/ Decomposed to /k/ + /s/ xero [ˈksɛ.rɔ]
q /k/ q maps to voiceless velar stop; u follows as /v/ quiz [kfis] (q+u always gives /kf/ in cluster)
v /v/ Mapped directly to voiced labiodental fricative (same as w) video [viˈdɛ.ɔ]
th /t/ + /x/ Not a digraph; t maps to /t/, h maps to /x/ theatre [txɛˈat.rɛ]
ph /p/ + /x/ Not a digraph; p maps to /p/, h maps to /x/ graph [ɡrapx]
qu /kf/ q+u mapped to /k/+/v/; /v/ devoices to /f/ after voiceless /k/ quasi [ˈkfa.ɕi]

Using the Tool: Practical Notes

Why is the stress in a verb form not where I expect it?

The module works from spelling alone and places stress on the penultimate syllable by default. It recognises the conditional endings -by, -byś, -byśmy, -byście but not the past-tense endings -śmy and -ście, so zrobiliśmy is output as [zrɔ.biˈliɕ.mɨ] rather than with the prescriptive antepenultimate stress. See Stress Patterns & Exceptions. Open in the converter

How do I force the stress onto a particular syllable?

Type an apostrophe directly before the stressed syllable: zro'biliśmy gives [zrɔˈbiliɕmɨ]. A word marked in this way is output without syllable dots, and an acute accent on a vowel is not understood as a stress mark. Details are in Stress Patterns & Exceptions.

Does the tool use a dictionary?

No. Polish transcription is generated entirely from spelling rules, so every word receives exactly one transcription and there are no variants to cycle through. Words whose stress or pronunciation is not covered by the rules, such as some loanwords and compounds, are transcribed by the default rules; see Known Issues & Limitations.

Why are consonants before i not marked as palatalized?

Consonants before i are slightly palatalized in speech, but the module does not mark this with ʲ; pili is output as [ˈpi.li]. See Palatalization & Soft Consonants.

How the Lua Module Works

The Lua file pl-pron_wasm.lua transcribes Polish words through a pipeline of character-by-character grapheme-to-phoneme conversion, contextual nasal resolution, voicing assimilation, and syllabification. Unlike the German or French modules, it does not use a lexicon — all output is generated from orthographic rules.

Module Execution Flow

  1. Stress Identification (get_stressed_syllable):

    Analyzes the verb suffixes (like -byśmy, -liście) and grammatical rules to establish which vowel nucleus receives the primary stress. If none matches, it defaults to the second-to-last vowel nucleus.

  2. Phoneme Mapping (convert_to_phones):

    Processes the string from left to right. Characters are matched against the letters2phones table. digraphs (like sz, cz, ch) are prioritized over single letters. When i is encountered, it sets a soft flag on the preceding consonant and determines if the i itself should be kept as a vowel or discarded as a spelling marker.

  3. Nasal Resolution (simplify_nasals):

    Scans for nasal vowels and looks ahead to the next phoneme. It replaces the abstract nasals with their contextual combinations (e.g., converting nasal to oral + velar nasal before velars).

  4. Terminal Devoicing (terminal_devoice):

    If the final phoneme is marked as a voiced obstruent, it is converted to its voiceless counterpart using a direct lookup table (e.g., d → t, g → k).

  5. Consonant Cluster Voicing (process_consonant_clusters):

    Identifies consonant clusters (sequences of non-vowel phones). For each cluster, finds the "determining consonant" — the last obstruent that is neither a sonorant nor a forward assimilant. The voicing state of this consonant propagates to the entire cluster. Forward assimilants (w/v and rz) are devoiced progressively — they become voiceless when preceded by a voiceless consonant (e.g., kwiat: k→voiceless, so w→f).

  6. Syllabification (collect_syllables):

    Segments the phone array into syllables using the Sonority Sequencing Principle. Vowels form nuclei, and valid clusters are shifted to syllable onsets.

  7. Formatting & Post-Processing:

    Applies stress markers (ˈ) and syllable boundaries (.), then performs final IPA cleanup: removes redundant ʲ before i/j, converts remaining ʲ to j glide, and replaces nasal vowels with diphthong notation (ɛ̃ → ɛw̃, ɔ̃ → ɔw̃).

Execution Pipeline

Raw Polish Word
       │
       ▼
┌─────────────────────────┐
│ 1. get_stressed_syllable │  Identify stress from verb suffixes
└────────────┬────────────┘
             ▼
┌─────────────────────────┐
│ 2. convert_to_phones     │  letters2phones lookup, digraph priority
└────────────┬────────────┘
             ▼
┌─────────────────────────┐
│ 3. simplify_nasals       │  ą/ę → oral V + nasal C (contextual)
└────────────┬────────────┘
             ▼
┌─────────────────────────┐
│ 4. terminal_devoice      │  Final voiced obstruent → voiceless
└────────────┬────────────┘
             ▼
┌─────────────────────────┐
│ 5. process_consonant_    │  Regressive voicing assimilation
│    clusters              │  Progressive devoicing of w/rz
└────────────┬────────────┘
             ▼
┌─────────────────────────┐
│ 6. collect_syllables     │  Sonority-based syllabification
└────────────┬────────────┘
             ▼
┌─────────────────────────┐
│ 7. Format output         │  Stress ˈ, syllable ., post-processing
└────────────┬────────────┘
             ▼
       IPA Output

Key Data Structures

Structure Type Purpose
letters2phones Nested table Maps graphemes to IPA phones. Uses [false] key as fallback for unknown characters. Supports multi-character lookahead (e.g., c → h → i → vowel chain).
valid_phone Set All recognized IPA phone strings. Used to distinguish phones from boundary markers during syllabification.
vowel Set {a, ɛ, ɛ̃, i, ɨ, ɔ, ɔ̃, u} — vowel nuclei for syllable detection.
devoice / voice Map Voicing pairs. Sonorants (m, n, ɲ, ŋ, l, j, r, w) map to themselves — they are "neutral" and do not trigger assimilation.
nasal_map Map Following consonant → homorganic nasal: p/b→m, t/d→n, k/g→ŋ, ć/dź→ɲ, etc.
forward_assimilants Set {v, vʲ} (and rz via flags). These consonants are devoiced progressively (by a preceding voiceless C), unlike standard regressive assimilation.
third_last_syllable_stress Suffix list Conditional mood suffixes (-łbym, -łabym, -łbyś, etc.) that shift stress to antepenultimate syllable.
fourth_last_syllable_stress Suffix list -libyśmy, -łybyśmy, -libyście, -łybyście — stress on 4th from end.

Post-Processing Rules

After syllabification, the module applies final transformations to the IPA string:

Rule Input Output Description
Palatalization cleanup ʲi, ʲj i, j ʲ before i or j is redundant — removed.
ʲ → j bʲa bja All remaining ʲ converted to j glide.
Nasal diphthongization ɛ̃ ɛw̃ Nasal vowels rewritten as oral + nasalized approximant.
Nasal diphthongization ɔ̃ ɔw̃ Same for back nasal vowel.

Extending the Module

To add custom grapheme mappings (e.g., handling recent loanwords), insert entries into the letters2phones dictionary at the top of pl-pron_wasm.lua. The table uses nested lookahead: each key maps to either a phone string, a phone array, or a sub-table for the next character.

Example structure for a simple consonant:

["b"] = {
    ["i"] = { "bʲ", "i" },  -- before i → palatalized
    [false] = "b"            -- default/fallback
}

Known Issues & Limitations

Input System Output Expected Cause
Rzeczpospolita [ʐɛt͡ʂ.pɔs.pɔˈli.ta] Stress on -po- Compound boundary not detected; defaults to penultimate
quiz [kfis] [kfis] (some speakers) Loanword qu mapped as /kv/ by module
partner [ˈpar.tnɛr] [ˈpartnɛr] Correct — but stress may vary in casual speech
wziął [vʑɔw] [vʑɔw] Correct — ą denasalized before ł

Help pages for other Slavic languages in this tool: Czech, Russian, Ukrainian, Belarusian, all languages.

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