How to Pronounce Pinyin: Initials, Finals, and Tones

Pinyin letters lie to you, and they lie in a very specific way. You already know the Latin alphabet, so the moment you see “c” or “q” or “x” or “zh” written on the page, some part of your brain automatically assigns them the sound they make in English. Every one of those assumptions is wrong. “C” is not “k” or “s,” it is closer to the “ts” in “cats.” “Q” is not “k” at all. “X” is not “ks.” “Zh” is not “z” followed by “h.” Pinyin borrowed the alphabet’s letterforms and threw out almost all of its sound-to-letter promises, and nobody warns you about this before you start sounding out your first words.

The usual fix is a chart: thirty-some finals and twenty-one initials in a grid, to be memorized the way you might memorize a multiplication table. That approach works about as well as memorizing multiplication tables usually does, which is to say it produces a learner who can recite the chart and still hesitates on the syllable in front of them. This article does it differently. Pinyin is not a list of unrelated sounds; it is a small set of building blocks (a handful of tongue positions, a handful of vowel shapes) combined by a few consistent rules. Learn the blocks and the rules, and the chart stops being something you memorize and becomes something you can reconstruct on the fly.

What Pinyin Actually Is

Pinyin is young. Mandarin has been spoken for a very long time, but Hanyu Pinyin, the romanization system taught in this article, dates only to the late 1950s. A committee led by the linguist Zhou Youguang developed it as part of a broader literacy and standardization push, and the Chinese government officially adopted it in 1958. It later became the international standard for romanizing Mandarin under ISO 7098 in 1982, which is why it shows up on street signs, in dictionaries, and in every textbook you are likely to open today.

That history matters for one practical reason: pinyin was designed by people who already knew exactly how Mandarin sounds and needed a way to write it down using letters borrowed from a different language’s alphabet. It is a spelling convention, not a pronunciation guarantee. The letter “x” was free real estate in the Latin alphabet (English does not have a distinct consonant sound anywhere close to what Mandarin needed a spare letter for), so the committee assigned it a job. Nothing about the shape of the letter “x” tells you what it should sound like. You have to learn that separately, the same way an English-speaking child has to learn that “gh” is silent in “though” and pronounced “f” in “enough.” Once you accept that pinyin’s letters are labels rather than promises, the rest of this system stops feeling like a series of exceptions and starts feeling like exactly what it is: a code, consistently applied, that you can learn once and use forever.

The 21 Initials, As a System

Mandarin syllables are typically built from an initial (the consonant sound that starts the syllable) plus a final (the vowel, or vowel-plus-ending, that follows it). There are 21 initials in the standard set, and the traditional teaching order groups them by where in your mouth the sound is made, which is a far more useful way to learn them than alphabetical order. Two extra letters, “y” and “w,” show up at the start of certain syllables, but they are not separate consonant sounds in their own right. They are spelling conventions that mark the “i” glide and the “u” glide, and you will meet them properly in the finals section below.

Labials: b, p, m, f

These four are made with your lips, and three of the four (b, p, m) behave close enough to their English counterparts that you can lean on your existing habits. The fourth, f, is just an ordinary English “f.” The real news here is not m or f. It is that b and p are not what your alphabet trained you to expect, and the next section explains exactly why.

Alveolars: d, t, n, l

Made with the tip of your tongue against the ridge behind your upper front teeth, same spot English uses for these same four letters. N and l are straightforward. D and t carry the same twist that b and p do, covered below.

Velars: g, k, h

Made at the back of the mouth. G and k follow the same pattern as the two pairs above. H is the odd one: it exists in English, but Mandarin’s h is rougher and pulled further back in the throat, more of a scrape than a breath. Treat it as a cousin of the English h rather than an identical twin.

Palatals: j, q, x

Made with the flat part of your tongue near the hard palate, with the corners of your mouth pulled slightly back. This trio does not have a clean English equivalent, and reaching for the nearest English letter is, if anything, worse than having no comparison at all, since it anchors your mouth in the wrong place before you’ve even started.

Retroflex: zh, ch, sh, r

Made with the tip of the tongue curled backward, toward the roof of the mouth but further back than English “r” ever goes. This is the series that gives Mandarin its distinctive “shhh”-with-a-curl texture, and it is also the series where copying the spelling (“zh” looks like it should sound like “z” plus “h”) does the most damage.

Sibilants: z, c, s

Made with the tongue tip close behind the upper front teeth, producing a hissing, slightly buzzy quality. S is ordinary. Z and c are the ones that surprise people, since they are not the plain English “z” and “c” sounds at all.

Here is the full set, in the order above, with a real spoken word for each so you have something to anchor the sound to rather than an abstract letter.

Notice the pattern already forming in that “sounds like” column: six of these pairs (b/p, d/t, g/k, j/q, zh/ch, z/c) keep showing up with the word “aspirated” attached to one member and “unaspirated” to the other. That is not a coincidence, and it is worth its own section, because it is the single idea that explains more of this alphabet than any other.

Aspiration: The Concept That Explains Half the Alphabet

Here is the piece of the puzzle that most learners never get told directly, and once you have it, roughly a third of the 21 initials stop being individual memorization problems and become one rule applied six times.

English organizes its stop consonants around voicing: b is “voiced d,” p is its “voiceless” partner, and so on, at least according to how the letters are usually described. Mandarin does not organize its consonants that way at all. The b/p, d/t, g/k, j/q, zh/ch, and z/c pairs are distinguished by aspiration, meaning whether a small burst of air escapes right after the consonant releases, not by voicing. An aspirated sound is one where you can feel, or even hear, a little puff of breath right after the consonant; an unaspirated sound releases cleanly with no puff at all.

The genuinely useful part is that your mouth already does this distinction, in English, without your ear ever noticing. Say the word “pin” out loud, then say “spin.” In “pin,” the p is aspirated, a real puff of air. In “spin,” the p that follows the s is unaspirated, almost no puff at all, even though you spell it with the same letter and have never consciously registered a difference. Try it again holding the back of your hand an inch in front of your lips: “pin” hits your hand with a small burst of air, “spin” barely stirs it. English does have this contrast, it is just allophonic (an automatic side effect of spelling, not a distinction that changes word meaning) rather than phonemic, so nobody ever taught you to listen for it. Mandarin promotes exactly that same physical distinction to the most important job an alphabet can give a sound: it is the difference between two completely different words.

So: b sounds like the unaspirated p buried in “spy,” while p sounds like the aspirated p that starts “pie.” D sounds like the unaspirated t hiding in “sty,” while t sounds like the aspirated t that starts “tie.” G sounds like the unaspirated k inside “sky,” while k sounds like the aspirated k that starts “kite.” The same aspirated/unaspirated split carries over to j versus q and to zh versus ch and to z versus c, even though those three pairs don’t have quite as clean an English minimal pair to borrow. If you only take one exercise away from this article, make it this one: say “spy” and “pie” back to back, hand near your mouth, and feel the difference your own mouth has been making for your entire life without telling you about it.

Mandarin didn't invent a new sound your mouth can't make. It just promoted a distinction your mouth already makes, without your ear ever being asked to notice it.

Merry Mandarin

The Finals, As a System

If initials are about where in the mouth a consonant starts, finals are everything that comes after it: the vowel, or the vowel plus a nasal ending. Textbooks often present the finals as a list of roughly thirty-six items to memorize in one sitting, which is exactly the wrong way to encounter them. Almost all of that list is built from six simple vowels, combined by a small number of predictable rules. Learn the six, learn the rules, and you can reconstruct nearly the entire list yourself instead of memorizing it as thirty-six unrelated facts.

The 6 Simple Finals

These are the true building blocks. Every other final in Mandarin is either one of these six standing alone, or two-to-three of these combined with each other or with a nasal ending.

4 Compound Finals

Four more finals are simple, stable combinations of two of those vowel sounds glided together, close enough to English diphthongs that you can lean on real English words: ai sounds close to the English word “eye” (爱 ài, love), ei sounds close to the “ay” in “day” (累 lèi, tired), ao sounds roughly like the “ow” in “how,” though not an exact match (猫 māo, cat), and ou sounds close to the “oh” in “go” (头 tóu, head).

Nasal Codas: -n vs -ng

A large share of the remaining finals are one of the six simple vowels followed by a nasal ending, either -n or -ng, and the good news is that English already trains your tongue to make exactly this distinction. Say “sin” and then “sing.” For “sin,” your tongue tip touches the ridge behind your front teeth. For “sing,” your tongue pulls back and the sound closes in the back of your throat, near your soft palate, and never touches that ridge at all. Now say “ban” and “bang” the same way. That is the entire -n versus -ng distinction, a difference your mouth already makes automatically and correctly every time you say those English word pairs; Mandarin just asks you to apply it deliberately and to notice it as meaningful.

With that distinction in hand, a run of finals that look like separate vocabulary become one pattern applied to different vowels: an (三 sān, three), en (很 hěn, very), in (心 xīn, heart), un (春 chūn, spring), ün (群 qún, group) on the -n side, and ang (忙 máng, busy), eng (冷 lěng, cold), ing (明 míng, bright), ong (红 hóng, red) on the -ng side.

Medial Glides: Building Bigger Finals from the Blocks

The last building block is the medial glide: an i-sound, u-sound, or ü-sound placed in front of another final, gliding into it the way “y” glides into the vowel that follows it in the English word “yes.” This is where the “y” and “w” letters from the initials section finally earn their keep: when one of these glides opens a syllable with no other consonant in front of it, pinyin spells it starting with y (for an i- or ü- glide) or w (for a u- glide) purely as a spelling convention, not because a new consonant sound has appeared.

Rather than memorize every resulting combination as its own entry, it helps far more to see the pattern doing the work. A handful of examples make the rule visible:

Once you see medial glides this way, the roughly thirty-six-item final chart stops being a memorization task and becomes something closer to a small set of Lego bricks (six vowels, a couple of nasal endings, three medial glides) that recombine according to rules you can predict. You will still meet finals in the wild that this article hasn’t spelled out combination by combination, but you will meet them able to guess correctly, which is the entire point of learning a system instead of a list.

The Buzzing Exception

There is one genuine irregularity worth knowing by name, because it trips up almost every learner at least once: the “i” in 是 shì is not the same sound as the “i” in 米 mǐ, even though it is spelled with the identical letter. After the retroflex series (zh, ch, sh, r) and the sibilant series (z, c, s), the letter “i” does not represent the “ee” vowel at all. It represents a buzzy, consonant-like vowel that is really just an extension of the consonant sound before it, held out and shaped into a syllable.

For the retroflex version, in shì, zhī, chī, and rì, try saying the English word “shirt” and cutting yourself off the instant before the “r” coloring would normally kick in. What is left, that held, buzzy residue right after the “sh,” is close to the sound you want. For the sibilant version, in zì, cí, and sì, the buzz is shorter, tighter, and has no lip rounding at all, more like a prolonged “zzz” pulled into the shape of a vowel than anything resembling “ee.” Compare 是 shì “to be” against 四 sì “four”: both are spelled with a final “i,” and neither one sounds anything like the “i” in 米 mǐ “rice.” This is the one place in the whole system where the spelling genuinely does not tell you what to do, and simply knowing that it’s an exception, rather than a rule you’ve somehow missed, is most of the battle.

The Four Tones (Plus a Fifth That Isn’t Really One)

Everything above describes how a syllable is shaped. Tone describes the pitch that syllable rides on, and in Mandarin, pitch is not decoration. It is load-bearing: the same shape ma, said with four different pitch contours, produces four completely unrelated words, plus a fifth, toneless version that does yet another job entirely.

First tone, mā (妈, mother), is a high, flat pitch held steady for the whole syllable, the way you’d hold a single sustained note while singing rather than letting it drift up or down. Second tone, má (麻, hemp), rises, the same rising pitch your voice makes on an incredulous “What?!” when you can’t believe what you just heard. Third tone, mǎ (马, horse), dips low, the low, drawn-out, skeptical pitch of saying “Well…” when you’re not quite convinced. Fourth tone, mà (骂, to scold), falls sharply from high to low, the same sharp, clipped fall as barking out “No!” or “Stop!” And then there is the neutral tone, ma (吗, the question particle), which is short, unstressed, and carries no fixed pitch contour of its own at all; its actual pitch depends entirely on the tone of the syllable that came right before it.

Those pitch-to-feeling comparisons are a starting scaffold, not a substitute for listening. Your own voice already produces a flat sustained pitch, a rising incredulous pitch, a low skeptical dip, and a sharp falling command dozens of times a day in English, just never attached to a single, meaning-bearing syllable the way Mandarin attaches them. The work ahead is transferring pitch shapes you already make instinctively onto syllables where you’ll need to make them on purpose, which is a much smaller task than learning to produce four entirely new sounds from nothing.

Tone Sandhi: When Tones Change on the Fly

The four tones above describe how a syllable sounds in isolation. In connected, natural speech, certain tones shift automatically depending on what sits next to them, a phenomenon linguists call tone sandhi. These are not exceptions to memorize case by case; they’re rules, and once you know all four, you can predict the shift correctly every time you meet it, rather than being caught off guard by why a phrase you learned as bù shì keeps coming out of native speakers’ mouths as bú shì.

1. Two 3rd tones in a row

   The first one becomes a 2nd tone. 你好 (nǐhǎo) is actually pronounced ní hǎo, even though both syllables are written and learned as 3rd tone.

2. Three or more 3rd tones in a row

   Only the very last syllable keeps its 3rd tone. Every 3rd tone before it becomes a 2nd tone instead, so the shift cascades backward through the whole run.

3. 不 (bù) before a 4th tone

   不 switches from its usual 4th tone to a 2nd tone. 不是 is pronounced bú shì, not bù shì, because 是 is a 4th tone syllable.

4. 一 (yī) changes depending on what follows

   一 becomes a 4th tone before a 1st, 2nd, or 3rd tone syllable, so 一天 is pronounced yì tiān. Before another 4th tone, it becomes a 2nd tone instead, so 一样 is pronounced yí yàng. Said alone, or at the end of a number like 第一, it stays its original 1st tone.

Notice that the last two rules share the same underlying logic as the first two: Mandarin doesn’t like stacking two of the same “heavy” tone shape back to back, whether that’s two 3rd tones bumping into each other or 不/一 running into a 4th tone that would otherwise clash the same way. Once you notice that shared logic, all four rules read less like four separate things to memorize and more like one instinct (smooth out an awkward tone collision) applied to four specific situations. Reading these rules is the easy half, though. The hard half is training your ear to catch the shift happening in real, fast speech, and no amount of reading about sandhi substitutes for that. There’s a reading-side version of the same instinct worth checking too: our timed test of how automatically you actually recognize written Chinese includes a tone-pair category that measures whether tone distinctions register instantly when you see them on the page, not just when you hear them.

Hearing It For Real

Everything above gives you the map: which part of your mouth makes each initial, which six vowels build every final, and which four rules bend tones in connected speech. A map is not the territory, though, and pinyin’s entire premise, that letters can stand in for sounds, only holds up once you’ve actually heard the sounds the letters are standing in for. Reading that q is aspirated and j is not will get you partway. Hearing the two side by side, and then saying them back, is what actually rewires your ear and your mouth together.

The tone sandhi rules above reward the same treatment even more, since sandhi only really clicks once you hear a real sentence shift a written 3rd tone into a spoken 2nd tone in real time, rather than reading a rule about it in isolation.

Pinyin was built, from 1958 onward, to be a bridge: a way to write Mandarin’s sounds using letters a much wider world already recognized. It was never meant to be a promise that those letters would behave the way they do in any other language, and treating it as a system of building blocks rather than a foreign alphabet in disguise is what turns thirty-six finals and twenty-one initials from a wall of memorization into something you can actually reconstruct, syllable by syllable, the next time you meet one you haven’t seen before.