Polish pronunciation

This is my attempt to describe Polish phonology and phonotactics. The Wikipedia article has a lot more detail, though.

This mainly applies to contexts where there are native words. Loanwords that are unchanged from the original spelling are pronounced regarding the origin language's pronunciation rules. Thus, weekend is pronounced like in English, but Waszyngton (Washington) is pronounced with a V sound instead of a W sound.

IPA transcriptions for phonemes (annotated /.../) are provided, but I tried to avoid going into phones (annotated [...]) as much as possible. Graphemes are annotated ⟨...⟩.

Alphabet

The Polish alphabet extends the Latin alphabet, and has 32 letters. Q, V, and X are only in loanwords. Providing it makes it easier to understand the phonology.

A Ą B C Ć D E Ę F G H I J K L Ł M N Ń O Ó P R S Ś T U W Y Z Ź Ż
a ą b c ć d e ę f g h i j k l ł m n ń o ó p r s ś t u w y z ź ż

Fun fact! X actually used to be used in native words, e.g. xiądz ("priest"). X was removed from the core alphabet at some point, and was replaced with ks or gz, depending on the pronunciation, such as in ksiądz, egzamin ("exam").

Stress

Most words are stressed on the penultimate syllable (as in, the 2nd syllable, but counting from the end). However, there are cases where the stress is on the last syllable, or on the 3rd or 4th syllable from the end.

Vowels

Tables

Modern Polish's oral vowel system is similar to languages like Ukrainian or Indonesian, having two front, two central, and two back vowel phonemes. These vowel positions alone are significant, other factors like length or tone are not.

The close central vowel can be transcribed /ɘ/, /ɨ/, or (more rarely) /ɪ/. This article will use /ɘ/.

FrontCenterBack
Closeiɘu
Midɛɔ
Opena

While it is debated, there are two nasal vowel phonemes. Some sources support the below table, but others (especially newer ones) don't and consider nasal vowels as combinations of two sounds. In most positions, they would end with /ŋ/ or /w̃/. This article will transcribe those with /w̃/.

FrontCenterBack
Mid(ɛ̃)(ɔ̃)

These are the graphemes used to represent each vowel sound:

SoundGraphemes
/a/a
/ɛ/e
/i/i
/ɔ/o
/u/ó, u
/ɘ/y
(/ɛ̃/)ę
(/ɔ̃/)ą

Fun fact! Polish is one of the few modern Slavic languages to preserve nasal vowels in some way. In other Slavic languages, they have merged with other vowels.

Fun fact! The U/Ó split is a relic of when Ó represented the long O sound. At some point, when Polish stopped distinguishing vowels by length, Ó represented /o/, and eventually /u/. Some words have been respelled with the other letter of the two, such as bruzda ("furrow, groove") and ogórek ("cucumber"). The split still exists in some local dialects, such as the Podhale dialect.

In Ą and Ę, the consonant may be reduced or shifted into another one.

ShiftąęCondition
denasalizationwziął → /vʑɔw/
(he) took
wzięła → /vʑɛwa/
(she) took
before l or ł
addition of /m/dąb → /dɔmp/
oak (n)
dębowy → /dɛmbɔvɘ/
oak (adj)
before b/p
addition of /n/piątka → /pjɔntka/
five (n)
wędka → /vɛntka/
fishing rod
before t/d, c/dz, cz/dż
addition of /ɳ/wziąć → /vʑɔɲt͡ɕ/
take (v)
pięć → /pjɛɲt͡ɕ/
five (num)
before ś/ź, ć/dź
addition of /ŋ/mąka → /mɔŋka/
flour
ręka → /rɛŋka/
hand (n)
before k/g
no changewąsy → /vɔw̃sɘ/
mustache
węch → /vɛw̃x/
(sense of) smell
before f/w, s/z, sz/ż/rz, h/ch
in ę (name of the letter Ę)
no change (Ą)
denasalization (Ę)
są → /sɔw̃/
"they are", be-3.PL
się → /ɕɛ/
reflexive pronoun
word-finally

In careful speech, Ę is pronounced /ɛw̃/ word-finally, except in ę, the name of the letter Ę. Otherwise, it is pronounced /ɛ/.

Because this exception doesn't apply to verbs, it may be confusing if a verb has two present tense forms that end with (1st person singular) and -e (3rd person singular). Despite the fact that Polish is a pro-drop language, they are nonetheless easy to distinguish: dropping pronouns mostly only happens to 1st- and 2nd-person contexts. /piʂɛ tɔ/ by itself is generally understood to be in 1st person: "Piszę to."

Consonants

LabialAlveolarRetroflexPalatalVelar
Nasalmnɳŋ
Nasal approximant(w̃)
Plosivep bt dk ɡ
Affricatet͡s d͡zt͡ʂ d͡ʐt͡ɕ d͡ʑ
Fricativef vs zʂ ʐɕ ʑx ɣ
Approximantljw
Trillr

All consonants use the same graphemes as the IPA, except:

SoundGraphemes
/ɳ/ń, n (before ⟨i⟩ + consonant), ni (before vowels)
/v/w
/w/ł
/t͡s/c
/t͡ʂ/cz
/d͡ʐ/
/ʂ/sz
/ʐ/ż, rz
/x/h, ch
/t͡ɕ/ć, c (before ⟨i⟩ + consonant), ci (before vowels)
/d͡ʑ/dź, dz (before ⟨i⟩ + consonant), dzi (before vowels)
/ɕ/ś, s (before ⟨i⟩ + consonant), si (before vowels)
/ʑ/ź, z (before ⟨i⟩ + consonant), zi (before vowels)

The sounds /ŋ/, /w̃/, /ɣ/ don't have separate graphemes. Instead, they can only happen in certain circumstances:

Fun fact! Polish used to use the /ɫ/ sound (same as the dark L in English, but as a separate phoneme), but as delateralization took its course, it shifted to /w/. Despite this, it is still written Ł, a letter suggesting similarity to L.

We have a name for this shift: wałczenie. This is a fairly old phenomenon, which started with 16th-century peasants. It was rejected by the upper class, and the stigma lasted until around the 1950s. It was described by realizing Ł as a "non-syllabic U". Today, pronouncing Ł as [ɫ] is considered conservative.

Serbo-Croatian (except the Croatian dialects) also underwent L-vocalization, but only for syllable-final Ls, which changed to Os, cf. Београд Beograd ("Belgrade"). Some other Slavic languages, such as Bulgarian, are reportedly having the same L-vocalization that affected Polish.

Fun fact! RZ used to be a fricative R (/r̻/), but has merged into Ż phonemically. This is another leftover of old Slavic languages, preserved in Czech and Slovak. In other Slavic languages, it has become R instead.

In modern Polish, this is preserved in inflection and derivation, where RZ can become R and vice-versa, e.g. powietrze ("air") and wiatr ("wind").

In dialects with mazuration, chiefly the Mazovian and Lesser Poland dialects, ⟨ż⟩ is pronounced /z/, but ⟨rz⟩ is pronounced /ʐ/. Therefore, morze ("sea") and może ("maybe"), which are homophones in standard Polish, are not homophones in those dialects.

Phonotactics

The Polish syllable structure appears to be (C)(C)(C)(C)V(C)(C)(C), as exemplified by the words wstrząs ("shock, impact") and rejestr ("register" [noun]).

Final obstruents are devoiced: dług ("debt") → /dwuk/. Inflection explains the spelling: dług-∅dług-u.

Some words have difficult consonant clusters in the middle that get simplified: sześćset ("six hundred") → /ʂɛɕ.sɛt/ or /ʂɛj.sɛt/, jabłko ("apple") → /jap.ko/. Inflection and derivation often explains the spelling of those consonant clusters: sześćsetsześciuset, jabłkojauszko.

Assimilation

Onset consonant clusters

Idiosyncracies like rdza ("rust"), rtęć ("mercury, quicksilver"), or źdźbło ("blade [of a grass or cereal]") are mostly preserved for historical reasons. For this reason, Poles often insert a very short vowel between the consonants in everyday speech. This shows a fossilized trace of otherwise reduced yers, very short vowels that existed in Proto-Slavic, marked with *ь/*ĭ and *ъ/*ŭ.

Modern PolishGlossOld PolishProto-SlavicCognates in other Slavic languages
rdzarust(unknown)*rъďaBulgarian ръжда́ rǎždá
Ukrainian іржа́ iržá
rtęćmercuryrtęć*rьtǫtьCzech rtuť
Slovak ortuť
srebrosilverśrzebro*sьrebroRussian серебро́ serebró
Silesian strzybło
źdźbłoblade of grass or cerealśćbło*stьbloUkrainian стебло́ stebló ("stalk, stem")
Serbo-Croatian ста́бло stáblo ("tree, trunk")

In Czech and Slovak, r and l may be syllabic, which led to the tongue twister strč prst skrz krk. In Polish and most other Slavic languages, only vowels may be syllabic.