Is Chinese hard to learn? What's actually hard, and what's easier than you think
Mandarin takes English speakers about 2,200 class hours, but the difficulty is lopsided: what's steep, what's easy, and a map for your own background.

The US government gives its diplomats 88 weeks of full-time class to learn Mandarin. Spanish gets 30.
That figure describes full-time adult students, and it hides how uneven the work is. How hard Chinese will be for you depends on three things: which skills you need, what you already speak and how you practise.
If you speak English and you're deciding whether to start, or wondering why Chinese feels so uneven once you have, this guide is for you. We'll take the language one part at a time. Then you can set a map to your own background.
Key takeaways
- It's slow, officially. The State Department's yardstick puts Mandarin at three to four times the class hours of Spanish.
- Speaking keeps pace; reading and listening lag. After two years of college Chinese, students spoke about as well as Spanish learners but read and listened two or three steps behind.
- Tones are real, and trainable. Even advanced learners miss tone errors, but in one study two weeks of listening practice made a lasting difference.
- Listening is hard because of sound-alikes, not speed. Mandarin isn't spoken unusually fast. The trouble is how many words share a sound.
- The grammar is the easy part. Verbs never change, and there's no gender, no articles and no required plural.
- Your background changes the picture. Readers of Japanese, and Koreans who studied hanja, already know most of the first 1,200 characters in some form. Heritage learners start ahead in speaking but not in reading.
Officially, Chinese is one of the slowest languages
The State Department's language school, which trains US diplomats, sorts languages by how long they take. Mandarin sits in its hardest group, Category IV, the "super-hard languages". Its neighbours there are Cantonese, Arabic, Japanese and Korean.
The Defense Language Institute, which trains military linguists, draws the same line. So does data the State Department school published in 1993. Here's how the two schools budget their time:
Source: the State Department page (updated 21 January 2026) and DLIFLC, linked above. The State Department's students are adult diplomats studying full-time: a typical week is 23 hours in class and 17 of self-study, aiming at a professional working level, 3 on the government's 0–5 scale, in speaking and listening. 552–690 hours is the range for Category I, where Spanish and French sit at 30 weeks; until January 2026 the page put it at 600–750 hours. In the school's 1993 data, reported by Chiswick & Miller 2004, students scored 1.5 for Mandarin speaking after 24 weeks, against 2.5 for French and 1.0 for Japanese and Korean.
By that yardstick, Mandarin takes three to four times as many class hours as Spanish. But remember who the students are: adults studying full-time, aiming for professional-level speaking and listening.
Two details are missing from almost every article that quotes these numbers.
First, the page changed this year. Until January it set the goal as speaking and reading, not speaking and listening. An archived copy shows the old version, and six of the 20 ranking pages we read still quote its figures.
Second, the page's own arithmetic doesn't quite close:
| The State Department's numbers | Hours |
|---|---|
| 88 weeks × 23 class hours a week | 2,024 |
| Class hours the page states | 2,200 |
| 88 weeks × 25 hours, the week the old page implied | 2,200 |
Source: the current State Department page and its archived 2024 version. The page doesn't explain the gap.
Treat 2,200 as "about two thousand", not as a precise count.
Speaking keeps pace; reading and listening lag
The most useful number isn't a total. It's which skills fall behind.
A study of more than 2,000 US college language students tested speaking, reading and listening separately. After two years, learners of Chinese spoke about as well as learners of Spanish. Their reading and listening were two or three steps behind. Here's where the typical student stood:
Chinese students spoke as well as Spanish students, but read less
Median ACTFL level after two years of college study. N = Novice, I = Intermediate; L, M, H = Low, Mid, High.
| After two years (median) | Speaking | Reading | Listening |
|---|---|---|---|
| Chinese | Intermediate Low | Novice Mid | Novice Low |
| Spanish | Intermediate Low | Intermediate Mid | Novice High |
| French | Intermediate Mid | Intermediate Low | Intermediate Low |
Source: Winke et al. 2020, Tables 4–6, linked above. Non-heritage students at three US universities, with different students in each year of study. Students tested for speaking, reading and listening: 52, 40 and 36 in Chinese; 341, 334 and 312 in Spanish; 154, 148 and 143 in French. ACTFL scale, Novice Low to Superior.
That pattern runs through the rest of this guide. Speaking keeps up. Everything that depends on linking sounds to symbols and meanings falls behind.
If you want those hours turned into weeks at 30 minutes a day, see how long it takes to learn Mandarin.
What's hard about Chinese for English speakers?
Four parts are steep: tones, listening at natural speed, characters and vocabulary. Three more take practice but aren't walls: measure words, word order and the particle 了.
Tones are real, and trainable
Mandarin uses pitch to tell words apart. "Buy" is 买, and "sell" is 卖. English uses pitch for mood and questions, never to separate words, so you're building a category your ear has never needed.
"Mom" 妈 · "hemp" 麻 · "horse" 马 · "scold" 骂
The difficulty shows up wherever researchers look. Among 305 beginners whose first languages were European, 32% of tones in words and sentences came out wrong.
The third tone was the hardest, right only 58.8% of the time. When listening, the second and third tones are the pair learners confuse most, across 18 studies.
It doesn't fully go away, either. Advanced learners were shown made-up words that differed from real ones only in tone. They rejected them 35% of the time; native listeners rejected them 91% of the time.
So do tones matter if people understand from context? In a quiet room, mostly not. In noise, they do:
Source: Patel, Xu & Wang 2010. In quiet, sentences with flattened pitch were understood as well as natural ones; the noise was background chatter.
Context rescues you at a quiet table, not in a busy restaurant.
The good news is that tones respond to training. In one study, eight listening sessions over two weeks raised American learners' tone identification by 21% on average. The gain held six months later.
Practise with the tone-pair trainer, and read our guide to learning tones as an adult for a full plan.
Listening is hard because of sound-alikes, not speed
Mandarin isn't spoken unusually fast. In recordings from a 17-language study, here's how quickly people read aloud, and how much each syllable carried:
Source: our computation from Coupé et al. 2019 and their published data. Read speech, 10 speakers per language, simple averages.
Speed isn't the problem. The problem is how few distinct syllables carry all that information.
By our count, all 3,000 characters in China's official standard for learners are pronounced with one of just 396 syllables, if you ignore tone.
Counting tone, there are 1,110. So sound-alikes pile up as your vocabulary grows:
Source: our computation over the GF 0025-2021 character list. A character with two readings counts once per reading.
Two things follow. Tones do real work: through HSK 4, they cut the number of readings behind each sound by more than half. And context has to do the rest.
Eleven different characters in the first three HSK levels are pronounced exactly shì, from "to be", 是, to "try", 试.
Spoken Mandarin copes by building words from two syllables: 71% of the 3,245 words through HSK 4 are two characters long. That's why listening improves with vocabulary, not just with ear training, and why it lagged in the college data. Graded stories read aloud let you hear words you can already read.
Characters are a long road with an early payoff
Reading is the largest single block of work. China's learner standard expects 1,200 characters by level 4 and 3,000 at the top level.
The early payoff is fast, though. Just 717 characters cover 90% of the characters in Chinese film and TV subtitles, as we computed in our characters guide from the SUBTLEX-CH corpus.
Characters are also built from recurring parts. "Woman", 女, appears in "good", 好. It also appears in "mom", 妈, where "horse", 马, hints at the sound.
Writing by hand is a separate, smaller job. The new HSK syllabus lists 1,096 characters to recognise through level 4, but only 400 to write, as counted in our HSK 3.0 guide. If you're deciding which script to start with, see simplified vs traditional.
Every word starts from zero
Apart from a few loanwords such as "coffee", 咖啡, Mandarin words don't resemble English ones. Each word starts from zero, and China's standard has 500 words at level 1 and 3,245 through level 4.
Word choice, not grammar, is where learners slip most. In a corpus of 33,835 annotated learner errors, choosing the wrong word was the largest category, at 39%.
Word order? Just 7%. Short daily review with flashcards is the usual answer to the vocabulary load.
Three grammar points take practice
Mandarin grammar asks for something new in three places, and none of them is a wall.
Measure words. A number needs a classifier before the noun:
- "three books": 三本书
- "a sheet of paper": 一张纸
- "a fish": 一条鱼
Classifiers are obligatory in Mandarin and absent in English. By our count, China's standard tags 21 at level 1 and 100 by level 4.
English speakers lean on the general one, 个. They used 一个 4.9 times per thousand characters, against 2.9 for Japanese speakers, often where Chinese doesn't allow it. The measure word finder gives the right one for any noun.
Word order. The basic order is English's, but descriptions come before the noun, question words stay where the answer goes, and sentences often open with the topic:
- "the book I bought yesterday": 我昨天买的书, literally "I yesterday bought book"
- "Where are you going?": 你去哪儿? Nǐ qù nǎr?, literally "you go where?"
- "I've read this book": 这本书我看过。 Zhè běn shū wǒ kàn guo., literally "this book, I have read"
It feels hard, but it isn't where most mistakes are. Word order accounts for 5.7–8.9% of annotated errors across eight tallies from two large learner corpora, TOCFL and HSK, compiled in our grammar-mistakes research.
了 and time. There's no past tense. Time words like "yesterday", 昨天, set the time, and particles add aspect: 了 marks a completed action, and 过 means "have done before".
Missing or extra aspect particles are about 8% of the annotated errors in the TOCFL learner corpus. Learners tend to use 了 sparingly but mostly correctly; the classic slip is attaching it to state verbs. The grammar coach checks sentences like these.
The grammar is easier than you think
Most of the grammar machinery that makes European languages slow for English speakers simply isn't there.
- Verbs never change. One verb, 吃, covers "eat", "eats", "ate" and "eaten": "he eats" is 他吃, and "I ate" is 我吃了.
- No gender, not even in speech. "He", 他, and "she", 她, are both pronounced tā.
- Plurals are optional, and only for people. 书 is "book" or "books"; "students", 学生们, is the main kind of plural you'll meet.
- No articles. There's no "a" or "the" to choose. English speakers sometimes use 一个 as if it meant "a", which shows how much they miss it.
- Simple sentences work like English. "I drink tea" is 我喝茶. Add 吗 to ask: "do you drink tea?" is 你喝茶吗? nǐ hē chá ma?
- Pinyin is consistent. The official romanisation since 1958 spells sounds the same way every time, and you only need 396 base syllables. A few letters don't match English: q, x, c and zh. The pinyin chart has audio for every one.
Here's how Mandarin's grammar compares with English, Spanish, Japanese and Korean:
Grammatical categories per verb
- Mandarin
- 0–1
- English
- 2–3
- Spanish
- 4–5
- Japanese
- 4–5
- Korean
- 6–7
Past tense
- Mandarin
- none
- English
- yes
- Spanish
- yes
- Japanese
- yes
- Korean
- yes
Grammatical gender
- Mandarin
- none
- English
- three (pronouns)
- Spanish
- two
- Japanese
- –
- Korean
- –
Plural marking
- Mandarin
- optional, people only
- English
- always
- Spanish
- always
- Japanese
- optional, people only
- Korean
- –
Classifier needed with numbers
- Mandarin
- yes
- English
- no
- Spanish
- –
- Japanese
- yes
- Korean
- yes
Basic word order
- Mandarin
- SVO
- English
- SVO
- Spanish
- SVO
- Japanese
- SOV
- Korean
- SOV
Relative clause
- Mandarin
- before noun
- English
- after noun
- Spanish
- after noun
- Japanese
- before noun
- Korean
- before noun
Question words
- Mandarin
- stay in place
- English
- move to front
- Spanish
- move to front
- Japanese
- stay in place
- Korean
- stay in place
Yes/no questions
- Mandarin
- particle 吗
- English
- word order
- Spanish
- word order
- Japanese
- particle か
- Korean
- verb ending
Tones
- Mandarin
- complex
- English
- none
- Spanish
- none
- Japanese
- simple (pitch accent)
- Korean
- none
Source: WALS Online (Dryer & Haspelmath, CC BY 4.0), features 22A, 66A, 30A, 34A, 55A, 81A, 90A, 93A, 116A and 13A. "–" means WALS has no data point for that language.
Mandarin drops most of the machinery English and Spanish carry, and shares its word-order habits with Japanese and Korean. If you've ever memorised Spanish verb tables, the first row is the one to look at.
How hard will Chinese be for you?
It depends on what you want to do with it and what you already speak. Set your own profile:
Your Chinese difficulty map
Pick what you already speak and your goal. For an English speaker who wants to speak, 3 parts are steep (tones, listening at natural speed and vocabulary), 3 moderate and 4 gentle.
What's your goal?
| What you already speak | |||||||
|---|---|---|---|---|---|---|---|
| Part of Chinese | English only | + a European language | Japanese | Korean | Mandarin at home | Cantonese | Another tonal language |
| Tones | Steep | Steep | Steep | Steep | Moderate | Moderate | Moderate |
| Listening at natural speed | Steep | Steep | Steep | Steep | Gentle | Steep | Steep |
| Reading characterscan wait for this goal | Steep | Steep | Gentle | Moderate | Steep | Moderate | Steep |
| Handwritingcan wait for this goal | Steep | Steep | Moderate | Steep | Steep | Moderate | Steep |
| Vocabulary | Steep | Steep | Moderate | Moderate | Moderate | Moderate | Steep |
| Measure words | Moderate | Moderate | Gentle | Gentle | Gentle | Gentle | Gentle |
| Word order | Moderate | Moderate | Gentle | Gentle | Gentle | Moderate | Moderate |
| Time and aspect | Moderate | Moderate | Moderate | Moderate | Gentle | Moderate | Moderate |
| Verb forms | Gentle | Gentle | Gentle | Gentle | Gentle | Gentle | Gentle |
| No gender or articles | Gentle | Gentle | Gentle | Gentle | Gentle | Gentle | Gentle |
| Pinyin | Gentle | Gentle | Gentle | Gentle | Gentle | Gentle | Gentle |
| Simple sentences | Gentle | Gentle | Gentle | Gentle | Gentle | Gentle | Gentle |
- S: steep
- M: moderate
- G: gentle
Under each part: whether it is useful or can wait for the goal picked (speaking). Other goals: for reading and writing too, tones, listening at natural speed and pinyin are useful rather than core; for an HSK exam, handwriting is useful rather than core. Levels are qualitative (steep, moderate, gentle). Each one rests on the evidence line shown with it. No level is a measured score for an individual learner.
Quiz answer key
- 买 mǎi means "buy" and 卖 mài means "sell". Which pitch shape is mài (sell)? Answer: Falls sharply. Tone 4 falls from high to low. Tone 3 dips low (and rises at the end when said alone). Mix them up and you've offered to sell the thing you meant to buy.
- Fill the gap: 三___书 sān ___ shū (three books) Answer: 本 běn. 本 běn is for bound volumes. 张 zhāng is for flat things (一张纸 yì zhāng zhǐ, a sheet of paper), 条 tiáo for long thin things (一条鱼 yì tiáo yú, a fish) and 只 zhī for many animals (一只猫 yì zhī māo, a cat).
- 明 míng means "bright". Which two parts is it made of? Answer: 日 rì (sun) + 月 yuè (moon). Sun plus moon. The other two are real characters too: 女 + 子 make 好 hǎo (good), and 人 + 木 make 休 xiū (rest).
- How many forms of 吃 chī (eat) do you need for eat, eats, ate and eaten? Answer: 1. One. 他吃 tā chī is "he eats"; 我吃了 wǒ chī le is "I ate". 了 marks a completed action, not the past as such.
- How many different characters in HSK levels 1-3 are pronounced exactly shì (fourth tone)? Answer: 11. Eleven, from 是 (to be) to 试 (try). In speech, two-syllable words and context sort them out: 71% of the words through HSK level 4 are two characters long.
Here's the short version for four starting points.
If your goal is speaking
Characters can wait, and the steep parts shrink to three: tones, listening and vocabulary. The State Department's current benchmark is itself speaking plus listening. In the college data, too, speaking kept pace with Spanish while reading lagged.
Pinyin is enough to start. You'll want characters eventually for menus, signs and any HSK level, so most learners add them gradually.
If you already read Japanese or Korean
Reading is far gentler for you. We matched the first 1,200 characters, everything through HSK 4, against Japan's and Korea's official lists:
| Through HSK 4 (1,200 characters) | Characters | Share |
|---|---|---|
| Written exactly as in Japan's 2,136 jōyō kanji, like "study", 学, and "country", 国 | 684 | 57% |
| Match once you know the traditional form, like "love", 爱, written 愛 in Japan | 251 | 21% |
| Japanese readers, both together, as a lower bound | 935 | 78% |
| Korean readers who studied hanja: among Korea's 1,800 education hanja, in traditional form | 1,003 | 84% |
Source: our computation from Unicode's Unihan database and the official HSK list. The Japanese total is a lower bound: shapes that differ from both, like Japanese 気 for Chinese "air", 气, aren't counted.
Readers of Japanese, and Koreans who studied hanja, already know most of these characters in some form. Pronunciations differ either way.
The grammar overlaps too. As the grammar table shows, Japanese and Korean also require classifiers, put descriptions before the noun and leave question words in place. Tones are still new, though Japanese listeners hear the falling fourth tone better than English listeners do. For a full comparison, see Chinese vs other languages.
If you grew up hearing Mandarin or Cantonese
Heritage learners are ahead where home exposure counts. In one classroom study, they did significantly better than other students in listening, speaking and grammar. They did no better in reading comprehension, vocabulary learning or character writing. And in a first-year study, language background made no difference to learning characters.
If Cantonese is your home language, expect help from the written language. It varies little across China, even though the spoken languages differ sharply. Expect less help from tones than you'd think: Cantonese listeners also confuse Mandarin's second and third tones. See our guide for heritage speakers and Mandarin vs Cantonese.
If you speak another tonal language
Pitch-as-meaning isn't new to you, but your tones won't map neatly onto Mandarin's. Experience with native tones "does not necessarily facilitate" hearing a new tone system, the same tone study found.
Most of this research tested Cantonese speakers; Vietnamese and Thai speakers are less studied. Classifiers will feel familiar, since Vietnamese and Thai require them too.
Four habits that make Chinese easier
Train your ear, speak early and get feedback. Each has good evidence behind it.
- Train listening deliberately. Across 25 years of studies, perception training improved how learners heard new speech sounds, with an effect size of 0.92. It carried over to how they pronounced them, at 0.54.
- Practise both directions. In one experiment, learners who only listened or only spoke did much worse on the skill they hadn't practised. Say things out loud from the first week.
- Get feedback on tones in sentences, not just single words. Isolated syllables are the easy case. The advanced learners in the made-up-words study were strong on single syllables and weak in context.
- Read things you can almost read. Graded stories with pinyin over every character let characters build up in context. If reading is your main goal, compare graded reader apps.
What this page can't tell you
- The levels in the map are judgments, each tied to the evidence shown with it, not measured scores for any one learner.
- The research leans on particular learners: US college students, beginners with European first languages and, among tonal-language speakers, mostly Cantonese speakers.
- The speech-rate figures come from read speech, with 10 speakers per language and simple averages rather than the original paper's statistical models.
- We didn't use SayMei's own learner data. We checked it first. Our old tone checker passed only 32–42% of correct second- and third-tone recordings, as our audit found. Our lesson quizzes are answered correctly 90–96% of the time on the first try, which is too easy to rank anything.
What to do next
- Decide which skills you need. If it's speaking, start with pinyin and leave characters for later.
- Train tones on purpose, in sentences. Short listening practice works, and feedback in context matters most.
- Grow your vocabulary to fix listening. Sound-alikes are the problem, and known words are the cure.
- Use what you already know. Japanese and Korean readers and heritage speakers each start ahead somewhere different.
Start with the tone-pair trainer, or set your own profile in the map above.
FAQ
Is Chinese the hardest language to learn?
For English speakers it's in the hardest group, not alone at the top. The State Department puts Mandarin in Category IV with Arabic, Cantonese, Japanese and Korean, all at about 88 weeks of full-time training. In the school's 1993 data, students actually progressed more slowly in Japanese and Korean speaking than in Mandarin after 24 weeks.
Is Chinese harder than Japanese?
They sit in the same official category, but the hard parts differ. Mandarin's tones make early listening and speaking harder; Japanese grammar and its three writing systems weigh more later. In the State Department school's 1993 figures, English speakers reached 1.5 in Mandarin speaking after 24 weeks, against 1.0 in Japanese. Our Chinese vs other languages guide compares them in detail.
Is Chinese grammar easy?
The forms are easy: verbs never change, and there's no gender or required plural. A Mandarin verb carries 0–1 grammatical categories, against 4–5 for a Spanish one. The usage takes practice, especially measure words, word order and the particle 了. Even so, word order makes up only 5.7–8.9% of the errors annotated in large learner corpora, while choosing the wrong word makes up 38–46%.
Do I need to learn characters to speak Chinese?
No. Speaking and reading are separate skills, and pinyin is enough to start talking. In US college programs, learners of Chinese spoke about as well as learners of Spanish after two years, while their reading lagged far behind. You'll need characters for signs, menus and any HSK level, so most learners add them gradually.
I'm tone deaf. Can I still learn Mandarin?
Almost certainly. True congenital amusia, the clinical form of tone deafness, affects about 1.5% of people. Even people with amusia learned new tones in a training study. Tones respond to practice: eight listening sessions raised learners' accuracy by 21% on average.
Is it too late to learn Chinese as an adult?
No. The State Department's 88-week benchmark describes adult diplomats reaching professional speaking and listening. Adults with no Mandarin can already tell tones apart better than chance, and they improve within a month of classes. Age matters less than hours and feedback.
How long does it take to learn Chinese?
About 2,200 class hours for professional speaking and listening at the State Department's full-time pace, plus roughly 1,500 hours of self-study across the 88 weeks. Conversational goals take far less. Our guide to how long it takes to learn Mandarin turns the official estimates into weeks at your own pace.
Is Mandarin easier if my family speaks Mandarin or Cantonese?
For speaking and listening, yes: heritage learners did better than classmates in listening, speaking and grammar. For reading, no: they did no better at reading comprehension or writing characters. Cantonese speakers start with tone awareness, but they still confuse Mandarin's second and third tones like everyone else.
Sources
- US Department of State, Foreign Language Training (modified 21 January 2026; checked 6 October 2026); archived version, 29 December 2024.
- Defense Language Institute, Languages at DLIFLC (checked 6 October 2026).
- Chiswick, B. R. & Miller, P. W. (2004). Linguistic distance. IZA Discussion Paper 1246 (reports Hart-Gonzalez & Lindemann 1993, FSI).
- Winke, P. et al. (2020). The proficiency profile of language students. Second Language Research & Practice 1(1), 25-64.
- Chen, N. F. et al. (2015). iCALL corpus. Interspeech.
- Cui & Zhao (2026). Meta-analysis of Mandarin tone perception. Frontiers in Psychology 16:1670858.
- Pelzl, E. et al. (2019). Advanced learners' perception of lexical tone. Studies in Second Language Acquisition 41(1).
- Patel, A. D., Xu, Y. & Wang, B. (2010). The role of F0 contours in Mandarin intelligibility. Speech Prosody.
- Wang, Y. et al. (1999). Training American listeners to perceive Mandarin tones. JASA 106(6).
- Sakai, M. & Moorman, C. (2018). Perception training meta-analysis. Applied Psycholinguistics 39(1).
- Li, M. & DeKeyser, R. (2017). Perception practice, production practice and Mandarin tones. SSLA 39(4).
- Peretz, I. & Vuvan, D. T. (2017). Prevalence of congenital amusia. European Journal of Human Genetics 25(5).
- Zhu et al. (2023). Lexical tone learning in amusia. JSLHR 66(7).
- Wang, Potter & Saffran (2020). Tone discrimination in naive adults and learners. Language Learning and Development 16(3).
- So, C. K. & Best, C. T. (2010). Cross-language perception of non-native tonal contrasts. Language and Speech.
- Coupé, C. et al. (2019). Different languages, similar encoding efficiency. Science Advances 5(9); data.
- Ministry of Education of the PRC (2021). GF 0025-2021 standard; Chinese Testing International, Standard Query database; CLEC, new HSK syllabus.
- Ministry of Education of the PRC (1958). Hanyu Pinyin Scheme.
- Unicode Consortium. Unihan database, version 18.0 (kJoyoKanji, kKoreanEducationHanja).
- Cai, Q. & Brysbaert, M. (2010). SUBTLEX-CH. PLOS ONE.
- Dryer, M. S. & Haspelmath, M. (eds.). WALS Online, CC BY 4.0.
- Li, C. N. & Thompson, S. A. (1976). Subject and topic: a new typology of language.
- Lee, L.-H., Tseng, Y.-H. & Chang, L.-P. (2018). TOCFL learner corpus. LREC.
- Mochizuki et al. (2015). Use of 一個 by L2 learners. IJCLCLP.
- Xu, Lu & Brezina (2019). Use of 了 by learners.
- Xiao, Y. (2006). Heritage learners in the Chinese language classroom. Heritage Language Journal 4(1).
- Ke, C. (1998). Language background and learning Chinese characters. Foreign Language Annals 31(1).
Methods
The research behind this guide (reading the sources above, sampling what search engines and chatbots say, and running the computations marked "our computation") used public datasets: China's official HSK word and character lists, Unicode's Unihan database, WALS and the published data from Coupé et al. Each table says how its numbers were computed. Facts were checked against the linked sources, and the Mandarin examples against the CC-CEDICT dictionary. No SayMei learner data is used. If you spot a mistake, tell us.
