How to Learn Chinese — Mandarin flashcard data
Prepared 2026-10-29. 5,000 distinct simplified headwords; explicit word pinyin,
Chinese examples, example pinyin and English translations for every card.

300 cards: original teaching meanings and examples prepared for this edition.
4,700 cards: meanings, headword readings and parts of speech adapted from the
English Wiktionary, by Wiktionary contributors, via Kaikki.org / Wiktextract.
The pinned dictionary was extracted 2026-10-25 from the 2026-10-01 dump.
Dictionary source: https://kaikki.org/dictionary/Chinese/index.html
Wiktionary contributors: https://en.wiktionary.org/
License: Creative Commons Attribution-ShareAlike 4.0 International
https://creativecommons.org/licenses/by-sa/4.0/
Every derived card links to its specific Wiktionary page. Page history gives
the contributors and revision history. The selected dictionary adaptations
remain available under CC BY-SA 4.0.

3,227 of those cards use sense-linked Wiktionary usage examples under the same
CC BY-SA 4.0 license. Quotations were excluded from the import.
1,473 use Chinese-English sentence pairs from Tatoeba, by Tatoeba contributors,
under Creative Commons Attribution 2.0 France (CC BY 2.0 FR):
https://creativecommons.org/licenses/by/2.0/fr/
https://tatoeba.org/en/downloads
Each example links to both its Chinese sentence and English translation, where
contributor attribution and history can be found. Pinyin comes from Tatoeba's
Latn transcriptions with a named reviewer; it is not automatically generated.
Sentence pairs were obtained from this pinned mirror of the raw Tatoeba data:
https://github.com/Roxaleen/hsk-annotated-corpus/blob/dd6bc245038fdb0c93f7a29e380be15be671848b/data/raw/sentences/tatoeba_cmn_sentence-pairs.tsv
Transcriptions: https://downloads.tatoeba.org/exports/per_language/cmn/cmn_transcriptions.tsv.bz2
The source repository preserves both sentence IDs, original text, numbered
transcription and reviewer. Tatoeba contributions are community-authored;
a named transcription reviewer is not a guarantee of linguistic correctness.

Adaptations: OpenCC t2s converts traditional forms to simplified; source strings
are retained. Reviewed numbered pinyin is converted to tone marks. Classifier
cross-references are removed from display glosses. Topic groups are inferred
from selected meanings; study difficulty combines editorial grouping with a
wordfreq selection signal. Headwords are deduplicated by simplified spelling.
No manufactured inflections or duplicated traditional/simplified cards are
used to increase the count. Examples may illustrate more than one headword.

Selection aid: wordfreq 3.1.1 by Robyn Speer, whose code is Apache-2.0 and data
is CC BY-SA 4.0. The imported selection scores and derived ordering/grouping
are provided under CC BY-SA 4.0. This is not a definitive frequency ranking.
https://github.com/rspeer/wordfreq
wordfreq includes data from Wikipedia, SUBTLEX, OPUS, Twitter, Reddit,
OpenSubtitles and other sources; full upstream attribution is reproduced in
/wordfreq-data-notices.txt. No upstream corpus sentences are imported from
wordfreq into the flashcards.

Review: automated coverage, source, sense-link, pinyin and highlight checks,
plus editorial sampling and exclusions. External Tatoeba examples also pass
segmentation, target-reading and English sense-agreement checks. These checks
do not prove every sentence correct. This edition has not received an
independent Mandarin-teacher review. Word senses, dialect usage, punctuation,
pinyin spacing and treatment of tone changes can vary across sources.
Levels are suggested study groups, not HSK or CEFR ratings.

Download: https://howtolearnchinese.com/practice/data.json
Source manifest and selected raw records: scripts/chinese-vocabulary/
The inherited JSON field "spanish" holds Mandarin text in this edition.