Mandarin Numbers: 二 vs 两 and Counting in Ten-Thousands

In week one of Mandarin you learn that "two" is 二 (èr). In week two someone asks you to meet at 两点 (liǎng diǎn, two o'clock), orders 两杯咖啡 (liǎng bēi kāfēi, two coffees), and your 二 has quietly vanished. Then a landlord quotes the rent as 一百五十万 and your brain stalls trying to count the zeros.

Two number walls, and most learners hit both in the same week, right after tones stop tripping them up. The good news: each one is a single rule once you see what's actually going on. Here's the position rule for the two words for "two," the regrouping trick that stops the big-number freeze, and the spoken quirks your textbook either scattered or skipped.

两 and 二 mean the same thing and don't swap

Say 二个人 or 二点 to a native speaker and you'll get a small wince. Both are wrong, even though 二 is the word the textbook taught you for "two."

The rule is about position. 两 (liǎng) goes in front of things you count: 两个人 (liǎng gè rén, two people), 两本书 (liǎng běn shū, two books), 两杯咖啡 (two cups of coffee), 两点 (two o'clock). Anything with a measure word after it takes 两. 二 (èr) is the number itself, used for counting out loud (一二三, "one two three"), ordinals (第二, dì èr, "second"), compounds (二十 èr shí, twenty; 十二 shí èr, twelve), and labels like 二楼 (èr lóu, the second floor). The Chinese Grammar Wiki lays out the same split.

The cleanest minimal pair is the building: 二楼 is "the 2nd floor," a label, so it takes 二. 两层楼 (liǎng céng lóu) is "two floors," a count with the measure word 层, so it takes 两. Same building, two different "twos," and the only thing that changed is whether you're naming the floor or counting the floors.

So the instinct to build is one question: am I counting something, or naming the number? Counting things, telling time, anything with a measure word, reach for 两. Saying the number on its own, an ordinal, a digit in a bigger number, reach for 二.

The one place where it's genuinely fuzzy

Hundreds are the exception nobody admits. "200" is said both 二百 (èr bǎi) and 两百 (liǎng bǎi), and native speakers don't agree on which is better. Don't lose sleep over it; both are heard, and you'll be understood either way.

From "2,000" upward the wobble stops. It's 两千 (liǎng qiān) for 2,000, 两万 (liǎng wàn) for 20,000, 两亿 (liǎng yì) for 200 million. Once a number is big enough to need its own bracket, 两 wins.

The real wall: counting in ten-thousands

Here's the one that actually makes fluent-feeling English speakers stall mid-sentence. Someone says 一百五十万 and you stand there mentally counting commas.

The problem isn't vocabulary. It's that English and Chinese chop big numbers in different places. English groups by thousands: a million is a thousand thousands, a billion is a thousand millions. Chinese groups by ten-thousands. There is a dedicated word for 10,000, 万 (wàn), and another for 100 million, 亿 (yì). The whole myriad system is built on those cut points instead of the Western comma.

So the ladder looks like this: 十万 (shí wàn) is 100,000, 百万 (bǎi wàn) is one million, 一千万 (yī qiān wàn) is ten million, and 一亿 (yī yì) is one hundred million. English "billion" lands on 十亿 (shí yì), confirmed by Cambridge's dictionary.

The trick that stops the freeze: stop marking a comma every three digits and mark a cut every four instead. Write the number, then put your cuts at 万 and 亿. A million is 1,000,000 in English, but re-chop it as 100|0000 and it reads straight off as 一百万 (one hundred ten-thousands). 1.5 million re-chops to 150|0000, which is 一百五十万 (one hundred fifty ten-thousands). The digits never moved. You just stopped counting in the wrong base.

Drill it both directions until it's automatic. Hear 三万五 and land on 35,000. Hear 一百五十万 and land on 1.5 million. Hear 两亿 and land on 200 million. The freeze is never the words; it's the comma-shift, and the only fix is reps.

零: the zero you say exactly once

Learners do one of two things with internal zeros: drop them, or stack one for every zero they see. Both sound off.

The rule is that a gap of zeros inside a number gets a single 零 (líng), no matter how many zeros there are. 108 is 一百零八 (yī bǎi líng bā). 1,005 is 一千零五 (yī qiān líng wǔ). 10,020 is 一万零二十 (yī wàn líng èr shí). The Grammar Wiki on zeros covers the edge cases, but the one-零 habit handles almost everything you'll say out loud.

The spoken shortcuts textbooks skip

Three forms show up constantly in real speech and rarely in chapter one.

俩 (liǎ) is 两个 squeezed into one syllable. 我们俩 (wǒmen liǎ) is "the two of us," 你们俩 (nǐmen liǎ) is "you two." Because the 个 is already baked in, you never add another one; 俩个 is the kind of thing that makes a native speaker tilt their head. There's a matching word for three, 仨 (sā), as in 我们仨, "the three of us."

幺 (yāo) is the one that surprises people. When you read a string of digits out loud, like a phone number or a room number, 一 (yī) gets read as 幺 (yāo) instead, which keeps it from getting heard as 七 (qī, seven) on a crackly line. Room 108 becomes 幺零八 (yāo líng bā), not 一零八. You'll hear it the first time a hotel clerk reads your room number back to you, and the myriad-system reference notes the 幺-for-一 swap as standard for digit strings.

Where this leaves you

None of these is hard once you stop translating digit by digit out of English. The two "twos" come down to one question: counting something, or naming the number? 两个 for the count, 二十 for the number. And big numbers come down to where you put the cut: ignore the thousands comma your eye wants, and slice at 万 and 亿 instead.

If you want reps without a tutor on the clock, this is exactly the kind of thing an AI conversation partner like Conversa is good for: practice numbers out loud, drilling prices and phone numbers until 一百五十万 stops being a math problem and starts being a number. The next time a landlord rattles off the rent, you'll mark the cut instead of freezing.