German practice — list credits This learning deck contains 5,000 distinct German dictionary headwords with short English translations. Inflected forms are consolidated, German spelling and capitalisation are normalised, duplicates are removed, and common English translations are simplified. Ordering uses aggregated word-form occurrences in the German OpenSubtitles 2018 frequency list. This is a corpus-based ranking, not a claim that one ordering is universal across speech, news, and writing. Frequency data: FrequencyWords / OpenSubtitles 2018, by Hermit Dave and contributors. https://github.com/hermitdave/FrequencyWords Data license: Creative Commons Attribution-ShareAlike 4.0. https://creativecommons.org/licenses/by-sa/4.0/ Lemmatization lookup: spaCy lookups data, Explosion and contributors. https://github.com/explosion/spacy-lookups-data German lemma lookup includes TIGER-derived resources; see upstream notices. Lookup project license: MIT. Translations: Deutsch-English FreeDict+WikDict, edition 2025.11.23, maintained by Karl Bartel; Wiktionary contributors via DBnary. https://www.wikdict.com/page/download https://www.wiktionary.org/ https://dbnary.getalp.org/ Dictionary's embedded license: Creative Commons Attribution-ShareAlike 3.0. https://creativecommons.org/licenses/by-sa/3.0/ The adapted word dataset is available under CC BY-SA 4.0. Changes: frequency aggregation, lemma merging, filtering, spelling normalisation, shorter glosses, deduplication, and translation corrections. No dictionary example sentences, pronunciations, or long definitions are included in the trainer. German translation review, 3 October 2026: all 5,000 displayed word pairs were reviewed. 426 English translations were corrected or clarified for ordinary usage, 13 headwords were corrected, and alternate dictionary meanings were updated to keep valid translations out of the wrong-answer box. Card positions and saved progress were preserved. Detailed changes are recorded in the source repository at scripts/german-translation-review.json. Disputed meanings were cross-checked with Duden and Langenscheidt; their full definitions and examples are not reproduced in this learning deck. https://www.duden.de/rechtschreibung/Vater https://www.duden.de/rechtschreibung/Herzschlag https://de.langenscheidt.com/englisch-deutsch/daddy Sentences 1000 This deck contains 1,000 distinct short German sentences and everyday expressions. Ordering follows descending cleaned sentence occurrence counts in OpenSubtitles2018. One-word entries, names, subtitle noises, bare noun fragments and malformed lines are filtered out. This represents subtitle usage; it is not a universal ranking of all real-life German conversation. The simple English translations and contrasting wrong choices were authored for this trainer. The current aim game displays only the correct translation. Sentence frequency list: orgtre and contributors, Top OpenSubtitles Sentences. https://github.com/orgtre/top-open-subtitles-sentences https://github.com/orgtre/top-open-subtitles-sentences/blob/main/bld/top_sentences/de_top_sentences.csv Upstream notice: the lists retain the underlying corpus's terms. The upstream Python code is CC BY 3.0; that code is not included in this trainer. Underlying corpus: OPUS OpenSubtitles2018, sourced from OpenSubtitles. https://www.opensubtitles.org/ https://opus.nlpl.eu/legacy/OpenSubtitles-v2018.php P. Lison and J. Tiedemann (2016), OpenSubtitles2016: Extracting Large Parallel Corpora from Movie and TV Subtitles. Proceedings of LREC 2016. MIT License — spaCy lookups data Copyright (c) 2019 ExplosionAI GmbH Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. Japanese Kana and Kanji Kana: the 46 modern basic Hiragana and 46 modern basic Katakana. Romanized sounds use the usual Hepburn-style spellings (shi, chi, tsu, fu). を / ヲ are shown as "o (wo)": wo identifies the spelling, while o is the usual sound. No voiced forms, small kana, combinations, or obsolete wi/we are included. Reference charts: https://www.tofugu.com/japanese/learn-hiragana/ https://www.tofugu.com/japanese/learn-katakana/ Kana frequency counts: Chikamatsu, N., Yokoyama, S., Nozaki, H., Long, E., and Fukuda, S. (2000). A Japanese logographic character frequency list for cognitive science research. Behavior Research Methods, Instruments, & Computers 32, 482-500. Appendices B and C list character occurrence counts in the 1993 Asahi Shimbun corpus: 20,711,361 Hiragana and 3,608,288 Katakana. https://doi.org/10.3758/BF03200819 Author's copy: https://researchmap.jp/YOKOYAMA_Shoichi/published_papers/622371/attachment_file.pdf Only the basic characters' factual counts were transcribed; the paper and its page images are not distributed by this website. ヲ has no listed count. Each new local calendar day has all 92 characters once, then 408 repetitions: 204 Hiragana and 204 Katakana, apportioned by each script's observed counts using the largest-remainder method. This gives 250 answers per script in a 500-answer day. A minimum weight of 1 handles an absent frequency count; each character is covered in the opening 92 regardless of its repeat weight. The order is shuffled deterministically for the day, so reload keeps the current card. Extra rounds use weighted repetitions; daily totals are additive. Jōyō Kanji: a curated KANJIDIC2 extract, downloaded 3 October 2026 (local time), source file date 2 October 2026. Exactly 2,136 distinct characters marked elementary-school grades 1-6 (1,026) or secondary-school grade 8 (1,110) are included. Name-only Kanji (grades 9/10) are excluded. English labels use dictionary meanings, with brief labels and parentheses removed where suitable; other meanings are retained to reject ambiguous distractors. 2,037 characters have the source's Mainichi Shimbun frequency rank; these come first in rank order. The remaining 99 follow by grade and stable order. Frequency ranks describe newspaper usage, rather than every type of Japanese. https://www.edrdg.org/wiki/KANJIDIC_Project.html Data: https://www.edrdg.org/pub/Nihongo/kanjidic2.xml.gz Documentation: https://www.edrdg.org/kanjidic/kanjidic2_dtdh.html KANJIDIC2 copyright: James William Breen and the Electronic Dictionary Research and Development Group (EDRDG). This derived Kanji dataset is distributed under Creative Commons Attribution-ShareAlike 4.0. https://www.edrdg.org/edrdg/licence.html https://creativecommons.org/licenses/by-sa/4.0/ https://creativecommons.org/licenses/by-sa/4.0/legalcode No endorsement by EDRDG is implied. Dictionary data is supplied without warranty. scripts/update-kanji.mjs provides the monthly refresh procedure for the curated extract; it downloads the latest KANJIDIC2, checks the exact 2,136-character set, and retains only meanings, frequency ranks and grades. SKIP, other dictionary search codes, and proprietary descriptors are omitted. Economics 3500 An original easy-to-hard trading and economics curriculum with 3,500 distinct prompts: 720 terminology cards, 720 definition-to-term questions, 720 applied concept scenarios, and 1,340 numerical questions spanning 100 calculation skills. The opening 70 terms emphasize small-cap trading; the later topics broaden to execution, risk, financial statements, valuation, portfolios, bonds, derivatives, microeconomics, macroeconomics, monetary policy and international economics. Ordering is curated by practical relevance and learning prerequisites. It is not a measured frequency ranking of a universal list of economic items. Definitions, scenarios and contrasting incorrect choices were written for this trainer. Numerical problems use hypothetical amounts and explicitly stated model assumptions. Unless another precision is given, numerical choices are rounded to at most two decimals; m means million and bn means billion. Stop levels are triggers, not guaranteed fills; planned-risk problems explicitly exclude slippage and fees. No account-specific broker, tax or leverage rules are assumed. Economics uses Monday-to-Sunday weeks in the browser's local time. The weekly count and card position restart on Monday; replaying within a week retains that week's totals. Earlier daily records are combined into week totals. Concepts cross-checked on 3 October 2026 against primary educational references: SEC / Investor.gov — order types, execution, corporate actions and accounts. https://www.investor.gov/introduction-investing/investing-basics/how-stock-markets-work/types-orders https://www.investor.gov/introduction-investing/investing-basics/how-stock-markets-work/executing-order https://www.sec.gov/about/reports-publications/beginners-guide-financial-statements https://www.investor.gov/introduction-investing/investing-basics/glossary/reverse-stock-splits FINRA — low-priced stocks, liquidity and concentration risk. https://www.finra.org/investors/insights/low-priced-stocks-big-problems https://www.finra.org/investors/insights/concentration-risk https://www.finra.org/finra-data/browse-catalog/short-sale-volume https://www.investor.gov/introduction-investing/investing-basics/glossary/securities-lending https://www.sec.gov/rules-regulations/staff-guidance/trading-markets-frequently-asked-questions-8 LULD Operating Committee — U.S. stock price bands and trading pauses. https://www.luldplan.com/ Cboe Options Institute and CME — option rights, Greeks and futures margin. https://www.cboe.com/optionsinstitute/defining-options https://www.cmegroup.com/education/courses/understanding-the-benefits-of-futures/the-benefits-of-futures-margins https://www.cmegroup.com/education/courses/option-greeks/options-gamma-the-greeks.hideSubnav.educationIframe.html?hideAddThisExt=y&hideFooter=y&hideHeader=y&hideRightRail=y Federal Reserve — monetary policy and interest-rate concepts. https://www.federalreserve.gov/monetarypolicy.htm BEA and BLS — output, inflation and consumer-price measures. https://www.bea.gov/resources/learning-center/what-to-know-gdp https://www.bls.gov/cpi/questions-and-answers.htm IMF — supply, demand and economic foundations. https://www.imf.org/en/publications/fandd/issues/series/back-to-basics/supply-and-demand NIST — statistical estimation and confidence intervals. https://www.itl.nist.gov/div898/handbook/pmd/section5/pmd511.htm The curriculum contains independently worded explanations rather than extracts of those references. Reproducible content sources: src/economics-terms.txt and scripts/economics-calculations.mjs; scripts/build-economics.mjs checks counts, unique prompts and distinct rounded choices before generating the static deck. TRIVIA 20000 20,000 curated questions. Science, AI, history, economics, geography, law, medicine, technology, culture, Greek and Cypriot history, geopolitics, business, philosophy, engineering, astronomy, physics, luxury and wealth, pharmacology, inventions, criminal justice, flags and capitals receive the emphasis. Difficulty is an editorial estimate. There is no claim that this is a statistically measured ranking of the 20,000 most frequently asked questions. Some later questions require specialist knowledge. Trivia's position and round accuracy carry across local midnights. Completing all 20,000 unlocks REPLAY, which starts a fresh round while retaining dated practice history. Progress is saved only in this browser. The bank is a collection of separately licensed contributions, not a relicensing of those sources. Each record's source and original source identifier are in trivia/sources.json, in the same order as the ten trivia/N.json question files. Selection counts are in trivia/summary.json. Changes include filtering, removal of exact and similar duplicates, punctuation repair, estimated difficulty, topical selection, and reduction to one correct and one incorrect alternative. The dataset answer keys are retained unless an original question replaces them. OpenTriviaQA — uberspot and contributors. CC BY-SA 4.0. https://github.com/uberspot/OpenTriviaQA https://creativecommons.org/licenses/by-sa/4.0/ Adapted OpenTriviaQA contributions remain available under CC BY-SA 4.0. Full license: licenses/OpenTriviaQA-LICENSE.txt SciQ — Johannes Welbl, Nelson F. Liu and Matt Gardner (2017), Crowdsourcing Multiple Choice Science Questions. CC BY-NC 3.0. https://huggingface.co/datasets/allenai/sciq https://arxiv.org/abs/1707.06209 https://creativecommons.org/licenses/by-nc/3.0/ SciQ contributions are for noncommercial use under their original license. Supporting textbook passages are not republished. Full license: licenses/SciQ-LICENSE.txt MMLU — Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song and Jacob Steinhardt (2021), Measuring Massive Multitask Language Understanding. MIT license as distributed by the authors and CAIS. https://github.com/hendrycks/test https://huggingface.co/datasets/cais/mmlu Full license: licenses/MMLU-LICENSE.txt GeoNames — country and capital records. CC BY 4.0. https://www.geonames.org/export/ https://download.geonames.org/export/dump/countryInfo.txt https://creativecommons.org/licenses/by/4.0/ Ambiguous and recently changing capital designations were omitted from the automatically generated capital set. Names use common English spellings. flag-icons — Panayiotis Lipiridis and contributors. MIT license. https://github.com/lipis/flag-icons 194 country flags are stored as self-contained SVG images so they work in Windows browsers without relying on emoji fonts or third-party image servers. Full license: licenses/flag-icons-LICENSE.txt Original focused questions and economic definitions were written for the game. Selected primary references checked on 3 October 2026 include: Google Machine Learning Crash Course and glossary: https://developers.google.com/machine-learning/crash-course https://developers.google.com/machine-learning/glossary Vaswani et al., Attention Is All You Need (2017): https://arxiv.org/abs/1706.03762 Cyprus Ministry of Foreign Affairs — historical background: https://www.gov.cy/mfa/en/documents/historical-background/ UN Peacekeeping — UNFICYP: https://peacekeeping.un.org/en/node/104030 UNESCO — Choirokoitia, Paphos and the Painted Churches of Troodos: https://whc.unesco.org/en/list/848/ https://whc.unesco.org/en/list/79/ https://whc.unesco.org/en/list/351/ Greek Ministry of Foreign Affairs — independence treaties and protocols: https://200years.mfa.gr/en/international-treaties-en/ U.S. Courts — legal glossary (U.S. procedure is identified in the questions): https://www.uscourts.gov/glossary FDA — pharmaceutical glossary: https://www.fda.gov/drugs/drug-approvals-and-databases/drugsfda-glossary-terms Rolex, Patek Philippe and Ferrari — historical information: https://newsroom.rolex.com/watches/oyster-collection/submariner https://www.patek.com/en/manufacture/a-story-of-independence/the-founders https://www.patek.com/en/collection/nautilus https://www.ferrari.com/en-EN/history Medical questions teach terminology and underlying science rather than personal treatment choices; clinical dosing and patient-management scenarios are omitted. Legal and international-relations questions use stated jurisdictions or general concepts; changes in leaders, prices and undated live rankings are excluded. French practice — Français 5000 and Sentences 1000 5,000 distinct French dictionary headwords and 1,000 short common sentences. Words use base forms and verb infinitives; essential reflexive pronouns are retained. Past-participle adjective forms, alphabet fragments and spelling duplicates are removed. Independent noun meanings are retained. All displayed word pairs were reviewed, with 1,122 curated headword corrections or additions in src/french-word-overrides.tsv. English labels favor ordinary meanings. This is an editorial review rather than native-speaker certification. Word ordering: Lexique 3.83 by Boris New, Christophe Pallier and contributors. https://lexique.org/ https://lexique.org/databases/Lexique383/Lexique383.tsv The freqlemfilms2 field supplies subtitle lemma frequencies. Repeated entries within a part of speech use the maximum, then parts of speech are summed. Counts are factual corpus observations; the ranking is not universal across every kind of conversation, writing, or news. Lexique's website currently links this license for its data: https://creativecommons.org/licenses/by-nc/4.0/ Lexique frequency contributions retain their upstream noncommercial terms. Dictionary: français-English FreeDict+WikDict, edition 2025.11.21, maintained by Karl Bartel; Wiktionary contributors via DBnary. https://www.wikdict.com/page/download https://download.wikdict.com/dictionaries/tei/recommended/fra-eng.tei Embedded dictionary license: CC BY-SA 3.0. https://creativecommons.org/licenses/by-sa/3.0/ Adapted English dictionary glosses and alternate meanings use CC BY-SA 4.0. https://creativecommons.org/licenses/by-sa/4.0/ Changes: meaning selection, shorter glosses, base-form filtering, spelling normalization, deduplication, and English translation corrections. This does not replace the separate upstream terms for the Lexique frequency component. The adapted data is downloadable at french/words.json. English FrequencyWords / OpenSubtitles2018 was used as a build-time heuristic when choosing common English labels; its raw list is not shipped in the game. https://github.com/hermitdave/FrequencyWords Data license: CC BY-SA 4.0. French sentence frequency: orgtre and contributors, Top OpenSubtitles Sentences. https://github.com/orgtre/top-open-subtitles-sentences https://github.com/orgtre/top-open-subtitles-sentences/blob/main/bld/top_sentences/fr_top_sentences.csv Counts follow the same OPUS OpenSubtitles2018 source and upstream corpus terms described for German sentences. Spellings are cleaned, repeated variants and bare noun fragments are removed, and short everyday phrases are selected. Their English meanings were reviewed and simplified for this trainer. Some starting bilingual references: Tatoeba contributors, selected by CK at ManyThings.org. CC BY 2.0 France. https://www.manythings.org/anki/ https://tatoeba.org/ https://creativecommons.org/licenses/by/2.0/fr/ Each reused pair's sentence IDs and contributor names, along with occurrence counts and whether the translation was edited, are in french/sentence-sources.json. The deck is available at french/sentences.json. The collection retains these separate source licenses. Source download URLs, sizes and checksums are in french/source-manifest.json. Rebuild instructions: scripts/french-content.md. Flashcard practice — 8 October 2026 Flashcard games show their prompt and correct answer together. Space or Enter advances one card, with no reveal step. Phones have a small Next Card button at the bottom. It stays hidden on computers, and no keyboard hint is shown. Language cards show English first and the target language below. Aim targets, distractor choices and background-click scoring are removed. Held-key repeats do not advance more cards. Menus, settings and history controls do not count as practice. Skips count as ungraded practice, not correct answers. Earlier correct/wrong and aim-miss records remain saved in the browser. Text sizes stay constant between cards; longer text wraps instead of shrinking. Arithmetic requires a typed answer and advances automatically only when it is correct. Its original generator, operations, ranges, settings, 2,000-answer weekly goal and Monday-to-Sunday history remain. Space and Enter cannot skip it. ArithmeticAim has been removed from the menu and publication. Earlier manually added totals and historical 1,000-card goals are preserved. Earlier ungraded arithmetic practice remains in the saved totals. The green cursor is shared by the game, menus and arithmetic interfaces. Programming — 4 October 2026 5,000 original question-and-answer cards, progressing through ten difficulty bands and 52 continuous course weeks. Weeks 1–8 have 97 new cards; weeks 9–52 have 96. The ten question perspectives for each of 500 authored micro-lessons provide intentional repetition. The course starts at computer basics and ends in practical extension, TTS, screener and desktop application engineering. Subject counts match the requested percentages exactly. This is an authored learning sequence, not a statistically measured ranking of popular questions. Course data, the plan, counts and selected primary technical references are available in programming/cards.json, programming/curriculum-plan.json, programming/summary.json and programming/sources.json. The references check technical facts; short examples and project scenarios are original. Rebuild and review behavior are documented in scripts/programming-content.md in the source checkout. Speech voices, browser capabilities and packaging prerequisites depend on the target environment and are identified as such. Programming also shows each answer immediately and advances with Space or Enter. Its 5,000 cards continue in curriculum order across course boundaries. Completed new cards never reset on a calendar rollover. Replay starts again at card 1 while keeping total course completion. Earlier knowledge ratings, review dates and mastery records stay saved; skips do not change those ratings. All personal progress is kept only in this browser using a separate storage key. QUANT — 8 October 2026 5,000 original question-and-answer cards, from 500 authored skills with ten different teaching views each. Continuous difficulty progresses from simple percentages and probability to combined professional quantitative reasoning. Exact topic shares: statistics 30%, probability 15%, arithmetic 10%, EV/Kelly and decisions 10%, papers 10%, calculus 10%, trading/betting/poker 7%, models 4%, product analytics 2%, information theory 2%. These are curriculum choices, not a measured popularity ranking. Definitions and worked examples are original. Selected primary/reference sources verify mathematics and research methods; the source manifest, curriculum plan and counts are in quant/sources.json, quant/curriculum-plan.json and quant/summary.json. QUANT uses the current simultaneous question-and-answer style. Space or Enter advances once, with no aiming, reveal, ratings or calendar resets. Long text wraps at fixed sizes. Progress, practice history and per-card visits use the separate browser-only quant5000:v1 key. Replay after 5,000 retains 100% course completion. Ungraded practice cannot establish accuracy, weak topics or mastery; these are shown as not rated rather than inferred from skips. Review consists of authored recurrence across views and later applications. Additional languages — 16 October 2026 Spanish, Italian and Modern Greek: 5,000 unique base words and 1,000 short bilingual sentences each. Korean: 5,000 dictionary-form words and 40 Hangul letters (24 basic and 16 combined/tense letters), with a 500-card daily symbol practice goal. Every card shows English/sound above and the target language below simultaneously. These selected corpus rankings are not a universal ranking of all speech and writing. Words can have other meanings in context; labels give short learner meanings. Dictionary and ranking sources include Wiktionary contributors, WikDict, Kaikki/Wiktextract, subtitle frequency lists, Morph-it and learner lists. The exact input files, licenses, hashes, changes and sentence contributor attributions for each deck are in the public files: languages/spanish-sources.json languages/italian-sources.json languages/greek-sources.json languages/korean-sources.json The adapted word and sentence payloads are alongside those manifests in languages/. Source component licenses apply separately; attribution and share-alike terms remain with the corresponding dictionary contributions. Korean dictionary glosses: English Wiktionary contributors, adapted from the Korean dictionary extracted by Kaikki/Wiktextract. CC BY-SA 4.0: https://creativecommons.org/licenses/by-sa/4.0/ https://kaikki.org/dictionary/Korean/ https://en.wiktionary.org/wiki/Appendix:Basic_Korean_Vocabulary_List The source word entry is https://en.wiktionary.org/wiki/#Korean; its history is available through that entry's History tab. The Korean manifest retains the selected sense IDs and identifies edited labels. The derived Korean word payload is offered under the same CC BY-SA 4.0 terms. No affiliation with Wiktionary or NIKL is implied. Hangul inventory and romanization checked against the National Institute of Korean Language; compact sound labels are original learning aids: https://www.korean.go.kr/eng_hangeul/principle/001.html https://www.korean.go.kr/front_eng/roman/roman_01.do Romanized sounds are approximate and depend on syllable position/context. The review weights count letters in OpenSubtitles Korean surface tokens, decomposed into initial/medial/final letters. FrequencyWords contributors, CC BY-SA 4.0, https://github.com/hermitdave/FrequencyWords New decks load only when selected; all personal progress remains in this browser. Rebuild details are in scripts/languages-content.md and each language's content notes in the source repository.