The part that isn't spelled
Every language uses pitch, loudness and timing for something — but not for the same thing. In Cantonese and Mandarin pitch is lexical: it belongs to the word. In English, Spanish, Italian and German it is stress that changes words. In Georgian pitch does neither — it marks questions, and nothing else. In French there is no word stress at all. Getting this layer wrong is what makes a sentence with perfect vowels still sound foreign.
| Language | Pitch does… | Stress does… | Rhythm |
|---|---|---|---|
| Georgian | marks questions only | nothing contrastive | Syllable-timed, heavy clusters |
| English | intonation only | changes words | Stress-timed |
| French | intonation only | nothing — no word stress | Syllable-timed |
| Spanish | intonation only | changes words | Syllable-timed |
| Portuguese | intonation only | nothing contrastive | Stress-timed (Portugal) |
| Italian | intonation only | changes words | Syllable-timed + long consonants |
| German | intonation only | changes words | Stress-timed |
| Malay | intonation only | nothing contrastive | Syllable-timed |
| Japanese | intonation only | nothing contrastive | Mora-timed |
| Russian | intonation only | nothing contrastive | Stress-timed, heavily reduced |
| Hindi | intonation only | nothing contrastive | Syllable-timed, with deletion |
| Mandarin | carries the word (tone) | nothing contrastive | Syllable-timed |
| Cantonese | carries the word (tone) | nothing contrastive | Syllable-timed |
| Vietnamese | carries the word (tone) | nothing contrastive | Syllable-timed, one tone per syllable |
| Korean | intonation only | nothing contrastive | Syllable-timed, phrase melody |
| Turkish | intonation only | nothing contrastive | Syllable-timed, rising to the end |
Georgian — No tone at all — pitch is reserved for asking questions
Pitch
Georgian is not tonal: no word changes meaning because of its pitch. Pitch does one job instead, and it is a big one — a yes/no question is marked by INTONATION ALONE. Georgian does not invert the word order, has no question particle like Mandarin 吗, and keeps the same words in the same order; the sentence simply rises at the end. Get the intonation wrong and a question becomes a statement.
Stress
Weak and non-contrastive. Two-syllable words take it on the first syllable; longer words tend toward the third-from-last. No pair of Georgian words is distinguished by stress alone, which is why dictionaries do not mark it. To an English ear Georgian can sound evenly stressed, almost flat — that is correct, not a lack of expression.
Rhythm
Syllable-timed-ish, with famously heavy consonant clusters (გვფრცქვნი) that take time to get through without inserting vowels. The vowels never reduce to schwa the way English ones do.
Hear it
The ◈ button is the synthesiser drawing the pitch contour directly — same words, different melody, which is exactly the point. ▶ uses a real voice if your device has one.
| Pattern | Sentence | What to listen for | Hear |
|---|---|---|---|
| Statement — falling | ის მოდის. is modis. | "He is coming." The pitch settles down at the end. | |
| Question — rising | ის მოდის? is modis? | "Is he coming?" Same words, same order — only the final rise makes it a question. | |
| Question word — falling | ვინ მოდის? vin modis? | With a question word (ვინ 'who') the rise is not needed — it falls, like a statement. English does the same: "Who's coming?" does not rise. |
English — Stress is spelling — move it and you change the word
Pitch
No lexical tone, but intonation carries an enormous amount: rise for yes/no questions, fall for statements and wh-questions, and a whole grammar of contrast ('I didn't say SHE took it').
Stress
Contrastive, and unmarked in writing — the single most common cause of not being understood. Dozens of noun/verb pairs differ only in where the stress goes, and every unstressed vowel collapses to /ə/.
Rhythm
Stress-timed: stressed syllables come at roughly even intervals and everything between them is squeezed. This is why 'a cup of tea' becomes /ə kʌp ə ˈtiː/ and why learners who give every syllable equal weight sound wrong even with perfect vowels.
Minimal pairs by stress
| One | The other | Why it matters |
|---|---|---|
| record a record (noun) — REcord “That is a new record.” | record to record (verb) — reCORD “I want to record this.” | Same six letters. Stress on the first syllable = the thing; on the second = the action. The rule holds for dozens of pairs. |
| present a present (noun) — PREsent “She gave me a present.” | present to present (verb) — preSENT “I will present the results.” | Also: object / object, contract / contract, permit / permit, increase / increase. |
| desert desert (the sand) — DEsert “They crossed the desert.” | dessert dessert (the pudding) — desSERT “We had ice cream for dessert.” | Spelled differently, but the difference you HEAR is which syllable is strong. |
The pairs that share a spelling are played inside a sentence — a voice reading the bare word has to guess which one you meant, exactly as a reader would.
French — No word stress at all — the group is the unit
Pitch
Intonation only, at the level of the phrase. A yes/no question can be marked by a final rise alone (Tu viens ?), which is why the rise matters more in French than the est-ce que you were taught.
Stress
This is the one that surprises everyone: French words have NO fixed stress. Prominence lands on the last syllable of the rhythmic GROUP, not of the word — so the same word is stressed differently depending on where it sits. 'Un chocolat' stresses -lat; 'un chocolat chaud' moves it to chaud and the word chocolat loses it entirely. There are no French minimal pairs by stress, because there is nothing to contrast.
Rhythm
Syllable-timed: every syllable takes roughly the same time and vowels do not reduce. Hearing a French speaker give equal weight to English syllables — and an English speaker stress the first syllable of French words — is this difference in both directions.
Watch the stress move
| One | The other | Why it matters |
|---|---|---|
| un chocolat stress on the LAST syllable of the group: choco-LAT | un chocolat chaud the group grew — now it is choco-lat-CHAUD | The word did not change; the group did. This is why 'French has stress on the last syllable' is a half-truth — it is the last syllable of the phrase. |
Spanish — Stress is contrastive — and the written accent tells you where
Pitch
No lexical tone. Intonation marks questions, helped by the opening ¿ so you know a rise is coming before you start reading aloud.
Stress
Fully contrastive, and mercifully written: if a word breaks the default pattern (stress on the second-to-last syllable for vowel/-n/-s endings, on the last otherwise) it carries an accent. Three-way minimal sets are common and the accent is the only thing separating them.
Rhythm
Syllable-timed, with five pure vowels that never reduce — the clarity that makes Spanish easy to hear and hard to speak fast.
Minimal pairs by stress
| One | The other | Why it matters |
|---|---|---|
| término term (noun) — TÉR-mi-no | terminó he/she finished — ter-mi-NÓ “Él terminó el trabajo.” | And in the middle: termino 'I finish'. Three words, one spelling skeleton, three stress positions. |
| papa potato / the Pope | papá dad | One syllable of difference in prominence, and you have changed the subject entirely. |
| hablo I speak | habló he/she spoke | Present or past, decided by stress. This pattern runs through the whole verb system. |
Portuguese — Stress is contrastive, and unstressed vowels get eaten
Pitch
No lexical tone. Intonation carries questions on its own — Portuguese changes nothing else about the sentence, so the rise is doing all the work.
Stress
Contrastive and written. The default is the second-to-last syllable; anything else takes an accent, and the accent also picks between open and close vowels — avó /ɔ/ is a grandmother and avô /o/ is a grandfather.
Rhythm
This is where European and Brazilian Portuguese part company, and it is most of why they sound unrelated. European Portuguese is strongly stress-timed: unstressed e collapses to /ɨ/ and is often deleted outright, so pequeno comes out closer to p'queno and the language sounds clipped and consonant-heavy. Brazilian keeps its unstressed vowels open and lands much nearer syllable-timed, which is why it sounds slower and more musical to most ears.
Minimal pairs by stress
| One | The other | Why it matters |
|---|---|---|
| sabia he/she knew — sa-BI-a | sábia wise (f) — SÁ-bi-a “Ela era uma mulher sábia.” | And a third: sabiá, a thrush, stressed on the last. Three words, one spelling skeleton, three stress positions — exactly the Spanish término/termino/terminó pattern. |
| avô grandfather — close o | avó grandmother — open ó | Not stress but vowel quality, and the accent is the only thing marking it. Getting this wrong in a family conversation is a rite of passage. |
| falamos we speak (present) | falámos we spoke (past, Portugal) | Portugal marks the past with an accent; Brazil writes both the same and relies on context. |
Italian — Stress is contrastive but usually unwritten — you learn it per word
Pitch
No lexical tone; intonation does questions, and Italian leans on it heavily since word order often does not change.
Stress
Contrastive. The default is the second-to-last syllable, and the accent is only written when the stress falls on the LAST one (città, perché). Words stressed on the third-from-last (àncora, prìncipi) are common and unmarked — the dictionary is the only way to know.
Rhythm
Syllable-timed, with the extra wrinkle that double consonants are genuinely held longer (palla vs pala) — length is doing work here that stress does elsewhere.
Minimal pairs by stress
| One | The other | Why it matters |
|---|---|---|
| ancora again / still — an-CO-ra | àncora anchor — ÀN-co-ra | The famous pair. Both are spelled ancora in normal writing; only context and stress separate them. |
| principi principles — prin-CÌ-pi | principi princes — PRÌN-ci-pi | Identical in print. You are expected to work it out from the sentence. |
| capitano captain — ca-pi-TA-no | capitano they happen — ca-PÌ-ta-no | A noun or a verb, decided entirely by which syllable you lean on. |
German — Stress decides whether the prefix leaves the verb
Pitch
No lexical tone. Intonation marks questions, but German also inverts the verb, so the rise is doing less work than in Georgian or French.
Stress
Contrastive in a very German way: with separable-vs-inseparable prefix verbs, the stress tells you which grammar you are in — and therefore what the sentence means. Otherwise stress sits on the stem syllable, and the unstressed prefixes (be-, ge-, er-, ver-, zer-, ent-) never take it.
Rhythm
Stress-timed like English, with strong reduction in unstressed syllables (-en → /n̩/) and a glottal stop before vowel-initial words that gives German its clipped edge: 'am Abend' = /am ˈʔaːbn̩t/.
Minimal pairs by stress
| One | The other | Why it matters |
|---|---|---|
| Er fährt das Schild um. ÚMfahren — he knocks the sign over (separable, stress on um-) | Er umfährt das Schild. umFÁHREN — he drives around the sign (inseparable, stress on -fährt) | The same verb, opposite outcomes for the sign. Stress is the only spoken clue — and it is why the prefix physically moves in one and not the other. |
| Ich übersetze den Text. überSÉTZEN — I translate the text | Ich setze über. ÜBERsetzen — I ferry across | Same pattern: unstressed prefix = figurative and inseparable; stressed prefix = literal and separable. |
Malay — Nothing hidden — no tone, and stress that never changes a word
Pitch
No lexical tone. Questions are marked by particles (-kah) or plain intonation; the rise is optional rather than load-bearing.
Stress
Light and non-contrastive, usually on the second-to-last syllable — and it shifts when suffixes are added, without any effect on meaning. No Malay pair is distinguished by stress, which is one of the reasons Malay is regularly called the most approachable major language to pronounce.
Rhythm
Syllable-timed and even. Combined with a small sound inventory and no consonant clusters, this is why Malay is easy to hear accurately from day one.
Japanese — Pitch ACCENT — not tone, but it still changes words
Pitch
Japanese sits between the tone languages and the stress languages, and gets its own name for it: pitch accent. A word has at most ONE place where the pitch drops, and where that drop falls is part of the word. 箸 hashi (drop after ha) is chopsticks; 橋 hashi (drop after shi) is a bridge; 端 hashi (no drop) is the edge. Unlike Cantonese there is no contour to learn per syllable — only 'where does it fall, if at all'. Tokyo and Osaka disagree about the answer for many words, and both are correct.
Stress
No stress in the English sense — syllables are not louder or longer, so a Japanese speaker reading English tends to flatten it, and an English speaker reading Japanese tends to hammer syllables that should be even.
Rhythm
MORA-timed, not syllable-timed: every beat gets equal time, and ん, っ and the second half of a long vowel are each a full beat of their own. とうきょう Tōkyō is four beats; きって kitte 'stamp' is three, and kite (kite 'come') is two — different words. This is the single most important thing to get right, ahead of any vowel.
Hear it
The ◈ button is the synthesiser drawing the pitch contour directly — same words, different melody, which is exactly the point. ▶ uses a real voice if your device has one.
| Pattern | Sentence | What to listen for | Hear |
|---|---|---|---|
| 箸 hashi — chopsticks (drop after HA) | はし hashi | High then low. Tokyo speakers call this 頭高 atamadaka, 'head-high'. | |
| 橋 hashi — bridge (drop after SHI) | はし hashi | Low then high — and the drop lands on the particle that follows: hashi GA. | |
| 端 hashi — edge (no drop) | はし hashi | Flat: the accentless pattern 平板 heiban. Three identical syllables, three words. |
Minimal pairs by stress
| One | The other | Why it matters |
|---|---|---|
| きって kitte — stamp (3 beats: ki-t-te) | きて kite — come! (2 beats: ki-te) | The っ is a full beat of silence. Skip it and you have asked for a different word entirely — length, not stress, is doing the work. |
| おばあさん obāsan — grandmother (5 beats) | おばさん obasan — aunt (4 beats) | One extra beat of vowel. Getting it wrong ages someone by a generation. |
Russian — Stress is unwritten, unpredictable, and decides which vowels survive
Pitch
No lexical tone. Intonation carries questions entirely on its own — Russian changes no words and moves nothing — with a sharp rise on the stressed syllable of whatever is being asked about, then a fall. Move the rise and you change the question.
Stress
Contrastive, and NOT written. За́мок is a castle and замо́к is a lock. Worse, it moves within a word’s own declension — рука́ but ру́ку, окно́ but о́кна — so knowing the word is not the same as knowing where to hit it. Dictionaries mark it; native writing does not.
Rhythm
Strongly stress-timed, and the reduction is severe. Unstressed о and а both collapse to /ɐ/, and further from the stress to /ə/; unstressed е and я go to /ɪ/. хорошо́ has three о and three different vowels — /xərɐˈʃo/ — which is the single clearest demonstration of the system. A learner who gives every written vowel its full value is instantly identifiable, and this is what most foreign accents in Russian actually consist of.
Minimal pairs by stress
| One | The other | Why it matters |
|---|---|---|
| за́мок castle — ZÁ-mok | замо́к lock — za-MÓK “На двери замо́к.” | Identical letters. Only the stress separates them, and the writing marks nothing. |
| писа́ть to write | п́исать — an obscenity | The reason stress is worth taking seriously. Textbooks mark this one. |
| рука́ hand (nominative) | ру́ку hand (accusative) | Same word, and the stress jumps when the case changes. This happens throughout the declension and has to be learnt per word. |
Hindi — Stress follows syllable weight — and the vowel you do not write disappears
Pitch
No lexical tone, unlike its neighbour Punjabi, which developed one. Pitch does intonation, and क्या at the front of a sentence marks a question so the rise is optional.
Stress
Not contrastive and not written. Stress falls on the heaviest syllable — a long vowel or a closed syllable outweighs a short open one — and where several are equal it goes to the last of them. No two Hindi words are told apart by stress alone, which makes it the least stressful stress system here.
Rhythm
Syllable-timed, but with a twist no other language on this site has. Every Devanagari consonant carries an inherent अ /ə/ unless something removes it — and in modern Hindi that vowel is DELETED in predictable places, usually word-finally and between certain consonants. So कमरा is written ka-ma-raa and said kam-raa; नमस्ते is na-ma-s-te and said na-mas-te. The script records an older pronunciation and the reader is expected to know which vowels have since gone. Sanskrit keeps them all, which is why chanted Sanskrit sounds so different from spoken Hindi.
Minimal pairs by stress
| One | The other | Why it matters |
|---|---|---|
| कमरा room — written ka-ma-raa, said KAM-raa | कमल lotus — written ka-ma-l, said KA-mal | Schwa deletion, and it is not optional. A learner who pronounces every written अ sounds like someone reading Sanskrit aloud. |
| नमस्ते hello — na-mas-te, not na-ma-s-te | रामायण Ramayana — raa-maa-yan, not raa-maa-ya-na | The final अ goes almost everywhere. This is why Hindi names lose a syllable that Sanskrit keeps: राम is Raam, not Rama. |
Mandarin — Pitch is lexical — see the tone lab
🎵 Pitch belongs to the word here. Open the tone lab →
Pitch
Four tones plus a neutral one; the pitch IS part of the word. Intonation still exists on top (a question can raise the whole phrase) but it can never overwrite a tone.
Stress
Stress is a matter of length and clarity rather than pitch: a stressed syllable holds its full tone, an unstressed one goes neutral and short.
Rhythm
Syllable-timed, one syllable per character, with tone sandhi reshaping sequences (third + third → second + third).
Cantonese — Pitch is lexical — see the tone lab
🎵 Pitch belongs to the word here. Open the tone lab →
Pitch
Six pitch shapes (nine traditional categories). Because the tone space is already crowded, Cantonese leans on sentence-final particles rather than intonation to carry mood — 呀, 喇, 咩, 囉 do the job a rising pitch does in English.
Stress
No contrastive stress; prominence comes from particles and from changed tone (變調) lifting a syllable to tone 2.
Rhythm
Syllable-timed, with checked syllables (-p -t -k) noticeably shorter than open ones.
Vietnamese — Pitch is lexical, and the spelling writes it — see the tone lab
🎵 Pitch belongs to the word here. Open the tone lab →
Pitch
Six tones in the north, and every one of them is written: the mark on the vowel IS the tone (ma, mà, má, mả, mã, mạ are six words). Two of the six are made with the throat as much as with pitch — ngã breaks in the middle and nặng is cut off short — so a Vietnamese tone is a shape plus a voice quality, not a note. Intonation still rides on top for questions and emphasis, but it can never overwrite the tone: a question is asked with a particle (không?, à?) rather than with a rise.
Stress
No lexical stress. Every syllable carries its own tone and its own full vowel, and none is reduced. What sounds like stress to an English ear is a syllable holding its tone longer or louder for emphasis, and grammatical words (đã, sẽ, được) being said lighter and faster than content words.
Rhythm
Syllable-timed, and unusually evenly so: every syllable is written as a separate word, is one beat long, and ends either open or in a clipped, unreleased consonant. The clipped endings (-p, -t, -c, -ch) and the two glottalised tones give it a staccato that sounds abrupt to European ears and is simply the rhythm of the language.
Korean — No tone, no stress — the melody belongs to the phrase
Pitch
Not tonal, and not a pitch-accent language either in the standard: Seoul Korean gives its pitch to the PHRASE, not the word. Each accentual phrase rises over its first two syllables and falls at the end (low-high … high-low), and the very first syllable starts higher after an aspirated or tense consonant than after a plain one — which is how listeners tell 달 from 탈 at the start of a phrase. Questions rise at the end, as in English. The Busan and Daegu dialects DO have lexical pitch accent (가지 is 'branch' or 'aubergine' by pitch there), which is why southerners sound sing-song to Seoul.
Stress
None at word level. No syllable is reduced and no vowel becomes a schwa; every syllable is said in full. What an English ear hears as stress is the phrase-initial rise.
Rhythm
Syllable-timed, with a strong tendency to even length, and every syllable ending either in a vowel or in one of seven unreleased consonants (ㄱ ㄴ ㄷ ㄹ ㅁ ㅂ ㅇ). The sound changes between syllables — nasalisation, tensing, liaison — keep the stream smooth: a Korean sentence has fewer audible consonant clusters than its spelling suggests.
Turkish — Stress at the end of the word — until a suffix or a place name says otherwise
Pitch
Not tonal. The stressed syllable is marked by a rise in pitch more than by loudness, which is why Turkish sounds even and melodic to an English ear. A yes/no question carries the particle mı/mi/mu/mü, and the pitch peak lands on the syllable just BEFORE it — on the word being asked about: Ahmet mi geldi? 'Was it AHMET who came?' against Ahmet geldi mi? 'Did Ahmet COME?'. The particle moves, and the melody moves with it.
Stress
Regular by default: the last syllable, and it travels rightwards as suffixes are added (kiTAP, kitapLAR, kitaplarDAN). The exceptions are what make it a real system. Place names are stressed early, on a pattern borrowed from their shape (ANkara, İsTANbul, İZmir); many adverbs and old loans stay on the first syllable (ŞIMdi, YArın, LOkanta); and some suffixes refuse the stress and push it onto the syllable before them — the negative -me-/-ma- above all, so that the same letters can be a noun or a command.
Rhythm
Syllable-timed: unstressed vowels keep their full quality, nothing reduces to a schwa, and long words built from many suffixes are said at an even pace, each bead of the chain getting its beat. Vowel harmony makes those chains sound internally rhymed — evlerimizden 'from our houses' is all front vowels, okullarımızdan 'from our schools' all back ones.
Minimal pairs by stress
| One | The other | Why it matters |
|---|---|---|
| yazma yazMA — writing; a printed headscarf (noun, final stress) | yazma YAZma — don't write! (the negative -ma pushes the stress back) | Identical letters. The negative suffix never takes the stress, so the command is stressed on the root and the noun on its last syllable. |
| bebek beBEK — a baby | Bebek BEbek — the Istanbul neighbourhood on the Bosphorus | Place names are stressed early. The same word becomes a district the moment the stress moves. |
| okuma okuMA — reading (noun) | okuma oKUma — don't read! | The same pattern on a three-syllable verb: the stress sits on the syllable just before -ma. |
Why this page exists
It started as a request for a tone lab for Georgian. Georgian has no tones — but it does have a prosodic system worth a page: it is the only language here where a yes/no question is marked by pitch and nothing else. Sorting out which languages use pitch for words, which use stress for words, and which use neither turned out to be more useful than pretending they all do the same thing.