Reference date: 2026-09-06. This guide was compiled as a knowledge foundation for absolute beginners of Chinese (Mandarin/Putonghua), written for an international audience. English is only the medium — nothing here assumes your first language is English, and comparisons that use English are always marked as such.
Who this is for
You are considering Chinese, or you have just started, and you want reliable answers to the questions every beginner runs into: What exactly is "Chinese"? How do the sounds work? What is pinyin? How many characters are there, and how many do I need? How is the grammar different from languages I know? What are HSK levels, and are they changing? How long does this take? What should I do first?
This guide answers those questions and shows you how confident each answer is. It does not teach the whole language — it gives you the ground to stand on while you learn it.
How to read the claims in this guide
Every factual statement was curated against a source register, graded for confidence, and written to make its status visible:
- Stated flatly → established fact, anchored in primary or authoritative sources (laws, official standards, ministry publications, or well-replicated research).
- "Official figures report…", "the 2020 national survey found…" → supported claims: real sources, attributed, but single-lineage or secondhand.
- "Reports vary", "unverified", "one estimate puts…" → uncertain material kept visible rather than deleted, because beginners encounter these numbers in the wild and deserve to know why not to trust them yet.
- A practical rule: / Decide by asking: → advice. Editorial guidance, clearly labeled, never dressed up with citations it does not have.
- Numbers always carry their year, what they count, and where. A naked statistic is treated here as a defect.
- Anything that can change (exam formats, rollout statuses, laws) is marked "as of 2026-09-06".
Where credible sources disagree, this guide preserves the disagreement instead of picking a comfortable winner — see the open questions at the end.
What this guide is built on
The content was compiled through a structured pipeline: a scope definition, a curated corpus of beginner questions (from an audited internal learner-question dataset plus systematic gap analysis), 92 researched problems, parallel evidence curation into graded registers (492 evidence rows, 220 statistical rows, 113 documented source conflicts, 262 logged gaps, a registry of ~1,190 sources), and per-problem synthesis with binding conflict and gap rules. Consequential claims were additionally verified by direct primary-source fetches where possible. The full machinery — methodology, incident log, and registers — lives in the corpus under knowledge/.
What this guide never does
- No product names, apps, textbooks, or tutors are endorsed or ranked — only selection criteria. (Where a named platform's own reported data is the evidence for a claim, the platform is named with full scope labels — attribution, not endorsement.)
- No "how many learners asked this" counts — internal demand signals never surface as numbers.
- No false precision: quarantined figures (like total Unicode character counts) stay quarantined; ranges stay ranges.
- No Vietnam-specific assumptions — this edition is deliberately international. (A separate Vietnamese-language edition carries local material for Vietnamese learners.)
Open questions — where the evidence still disagrees
Twenty-eight documented conflicts remained unresolved after curation (positions preserved side-by-side in the evidence registers). The ones a beginner is most likely to trip over:
Exams (fast-moving — re-verify before acting):
- The HSK 3.0 rollout status: official communication describes a trial of levels 1–6, while other reporting describes levels 7–9 as operational since 2025-12-07. As of 2026-09-06 neither picture can be confirmed as the whole truth; treat any "HSK is now X levels" claim with its date attached.
- Whether HSK 2.0 still runs in parallel with 3.0 anywhere.
- HSK 3.0 vocabulary quantities per level (including what "HSK level 3" requires) — competing figures circulate without a retrievable official syllabus.
- Whether an official HSK–HSKK co-registration rule exists, and for which levels — reported but not confirmed against a primary document.
- The 2024 global HSK taker count: 819,000 foreign candidates is the fetched, attributable figure; a 1.28-million variant circulates unverified.
- TOCFL band/level counts and HSK global test-site counts (2024 vs 2025) — inconsistent reporting.
Sounds and pinyin:
- Whether the neutral tone changes word meaning — analyses differ on how to frame the evidence.
- The exact tone-mark placement rule on multi-vowel syllables — competing conventions are taught; the standard's own wording could not be retrieved.
- Whether verified research exists on tone-perception transfer from a tonal first language (e.g. Vietnamese) to Mandarin — a widely cited advantage percentage could not be located in any primary study; treat all such claims as hypothesis.
Characters:
- The total number of Chinese characters in Unicode — competing totals exist across Unicode versions and counting conventions; by rule, none of them is printed in this guide.
- The number of radicals (201 vs other indexing systems) and the size of the 1956 first simplification batch.
- How strong the research basis is for stroke-order/handwriting benefits and for similar-shape character confusion — claimed effects outrun verified studies.
Vocabulary and grammar:
- Where modern disyllabic abstract vocabulary came from — Japanese-mediated coinage vs native development; the honest answer is "both, in proportions nobody has measured".
- Whether 是…还是… or 呢 is "the" marker of alternative questions — competing grammatical analyses.
History and encoding:
- How much graphic continuity links oracle-bone script to modern characters — competing continuity percentages exist with incompatible measurement definitions, and neither is adopted here; also whether modern literate readers can recognize oracle-bone script at all (an earlier claim was retracted).
- What GB 18030-2022 actually requires of implementations, and whether the empty box for unencoded characters is standard-defined behavior or an implementation choice.
Policy and society:
- The exact scope of the Putonghua requirement in broadcasting, and the attribution of a widely quoted 31-province dialect-attitude survey.
Limitations of this guide
- Evidence horizon. Sources were retrieved in Chinese, English, and Vietnamese up to 2026-09-06. Some paywalled academic studies (including at least one tone-transfer paper) could never be located for verification; claims resting on them stay hedged. A number of secondary sources were dead or unreachable at check time; where a consequential claim lost its anchor, it was re-anchored to a fetched primary source or downgraded.
- Conservative grading. Where a claim's citation lineage could not be fully verified against primary tables, it was capped at a lower confidence grade — so this guide hedges more than the underlying literature might justify. That is deliberate.
- Time-criticality. Exam logistics, rollout statuses, fees, and laws move. Anything marked "as of 2026-09-06" should be re-verified against official channels before you act on it.
- Not a course. This is a knowledge foundation for the first 92 questions; the language itself is learned by using it.
Lineage
Compiled through the pipeline documented in knowledge/METHODOLOGY.md: scope → question discovery → 354 curated needs → 92 problems → research waves → 24-group evidence curation → lead verification patches → synthesis → this master. Registers: knowledge/evidence/ (SOURCES.jsonl, evidence-rows.jsonl, STATISTICS.jsonl, CONFLICTS.md, GAPS.md, EVIDENCE.md). The Vietnamese-language edition (BEGINNER_CHINESE_VI.md) shares this evidence base and adds Vietnam-specific chapters governed by knowledge/synthesis/VI_ANALYSIS.md.