Mandarin Chinese Guide: From Pinyin to the HSK

Pinyin uses the Latin alphabet, which is exactly why it catches people out: plenty of familiar letters stand for unfamiliar sounds. The letter c sounds like a breathy ts, q is close to an English ch, made with the tip of the tongue tucked down against the bottom teeth and the lips spread, and x sits near a light sh. For the retroflex group zh, ch, sh and r, the front of the tongue lifts and bends a little backwards, toward the hard palate. For ü, the lips round as if to whistle while the tongue stays where it would be for i.

The trouble lies in near pairs. The j, q, x series sits close to zh, ch, sh, and z and c sit close to zh and ch. Learners tend to collapse each pair into whichever sound their first language already has, so 四 (sì, four) and 十 (shí, ten) end up sounding identical to a listener. One spelling rule adds to the confusion: after j, q, x and y the two dots over ü are dropped, so ju, qu and xu are still said with rounded lips.

Practice works pair by pair. Say zi, zhi, ji, then ci, chi, qi, then si, shi, xi in front of a mirror while watching the lips, then record the set and compare it with the audio from a pinyin chart. Aspirated sounds (p, t, k, c, ch, q) can be checked with a tissue held in front of the mouth: it flutters when there is enough breath and stays still for the unaspirated partners b, d, g, z, zh and j. The sounds are settled when a learner reads a new pinyin word from the dictionary and matches the audio on the first try.

Tone sandhi is the automatic tone change that happens when certain syllables meet their neighbours. The rule used most often: when two third tones sit side by side, the first is spoken as a second tone. 你好 appears as nǐ hǎo in the dictionary and comes out as ní hǎo in conversation. The same pattern applies to 很好 (hěn hǎo), 可以 (kěyǐ) and 水果 (shuǐguǒ).

Two common words follow rules of their own. 不 (bù) becomes a second tone before a fourth tone, so 不是 is said bú shì and 不要 is said bú yào. 一 (yī) takes a fourth tone before first, second and third tones, as in 一天 (yì tiān) and 一起 (yìqǐ), and switches to a second tone before a fourth tone, as in 一个 (yí gè) and 一样 (yíyàng). When it is read as a number in counting, phone numbers or years, 一 goes back to its first tone.

Some textbooks print the original tones and others print the changed ones, which puzzles many beginners. A calmer approach is to learn the most common phrases as single sound packages, colour-code the changing syllables in your notes, and play the audio until ní hǎo simply sounds right. This part is in place when a learner says 一个, 不是 and 你好 correctly without pausing to recall the rule.

The Kangxi dictionary sorts Hanzi under 214 radicals, components that recur across thousands of characters and often point to meaning. Nobody needs all of them at once. The thirty to fifty most frequent radicals already unlock a large share of basic characters: 氵 (water) in 河 (river), 海 (sea) and 洗 (wash); 木 (tree) in 林 (forest) and 树 (tree); 口 (mouth) in 吃 (eat), 喝 (drink) and 叫 (call); 女 (woman) in 妈 (mother) and 姐 (older sister).

Most Hanzi also carry a sound component. 马 (mǎ, horse) appears in 妈 (mā), 吗 (ma), 骂 (mà) and 码 (mǎ). The radical on the left hints at meaning while 马 hints at pronunciation, even though the tone often shifts. The same pattern shows in 湖 (hú, lake), which pairs 氵 with 胡 (hú). Sound hints are imperfect, yet they still narrow the guesswork whenever a learner meets an unfamiliar character.

Practice begins by grouping new characters by radical and turning their parts into a short story. 好 (hǎo, good) combines 女 with 子 (child), and a simple picture of a woman with her child is enough to tie shape and meaning together. One character, one short story, written on the side of the flashcard. The skill has landed when a learner sees an unstudied character such as 江 or 池 and immediately suspects it has something to do with water, before reaching for the dictionary.

A basic Mandarin sentence runs subject, verb, object, much like English: 我是学生 (I am a student), 我喝茶 (I drink tea). Verbs never conjugate, nouns have no plural form, and time is shown with time words. A yes-or-no question just adds 吗 at the end, so 你是学生吗 asks whether you are a student. Question words such as 什么 stay where the answer would go: 你叫什么名字 (what is your name).

Two word orders tend to trip up learners. Time and place come before the verb: 我明天在家学习 means I will study at home tomorrow, with 在家 placed ahead of 学习. Modifiers also come before the noun they describe, linked by 的: 我的书 is my book, and 很贵的书 is a very expensive book. Sentences translated word for word from another language usually stumble at exactly these two points.

Numbers connect to nouns through measure words: 一个人 (one person), 一张纸 (a sheet of paper), 一本书 (a book), 一只猫 (a cat). 个 is the general-purpose measure word and the safe choice when unsure. The particle 了 marks a completed action or a change of state, as in 我吃了 (I have eaten). Each new pattern gets five home-made sentences, then its parts are swapped one at a time. A pattern is in place when a learner can give a three or four sentence self-introduction without stopping to rearrange the words.

The HSK format still used by many institutions has six levels with a steep vocabulary climb: about 150 words at HSK 1, 300 at HSK 2, 600 at HSK 3, 1,200 at HSK 4, 2,500 at HSK 5 and 5,000 at HSK 6. HSK 1 and 2 test listening and reading only, with pinyin printed above the characters. HSK 1 has 40 questions, 20 listening and 20 reading, takes around 40 minutes, and plays each listening item twice.

The biggest jump comes at HSK 3. Pinyin disappears from the paper and a writing section begins, first arranging words into sentences and writing characters, then paragraphs and essays at HSK 5 and 6. The newer HSK 3.0 standard, whose exams are scheduled for full rollout from July 2026, divides ability into nine levels across three bands, with word lists that differ from the six-level format, so learners should ask their test centre which format it uses before building a study list. Learners under 15 can take the YCT, which has four written levels. The paper-based exam requires handwritten Hanzi, while the computer-based version uses pinyin typing.

Choosing a level starts with one timed sample paper, with the result compared against the official pass mark on the organiser's website. Once a single level is set as the target, timed mocks follow at regular intervals, and every mistake gets sorted: a word that has not stuck, a tone misheard, or an instruction skimmed past. A reasonable sign of readiness is mock results clearing the pass mark again and again, with more than eighty percent of that level's word list recognised.