Extensive Reading

Graded readers are the highest-return, least-used tool in English learning: these are novels, thrillers, classics and non-fiction rewritten within a controlled vocabulary for each of seven stages from A1 to C1, most with audio, so you can finish real books long before you could read the originals. Reading several at a comfortable level does more for fluency than grinding through one text that is too hard. They are paid, priced per title, and the simplifications strip out much of the original style — the point is volume of comprehensible reading, not literary appreciation.

Dozens of illustrated PDF booklets sorted into levels, from near-zero Japanese up to comfortable intermediate reading, published free by the tadoku (extensive reading) organisation. Perfect for the phase where textbook dialogues are boring but native material is still impossible. The catalogue is small compared to the paid printed sets, and the stories are simple by design.

Proper graded reading: sci-fi, crime and thriller stories with bolded vocabulary, chapter glossaries, plot summaries and comprehension questions, so you can read a whole book without a dictionary at your elbow. It is the standard recommendation for the moment textbook dialogues stop being enough, and there is a Volume 2 plus an intermediate edition when you outgrow it. Published by Teach Yourself; the audiobook is sold separately, which is an annoying split if you want to read and listen together.

Over 1,600 episodes of original serialised stories and articles, each with human-recorded audio, per-word definitions in context, and editor notes explaining why a sentence works the way it does — that last part is what separates it from generic reading apps. You can dial furigana down as you improve, and it syncs with your WaniKani level so known kanji lose their furigana automatically. Some content is free; full access is $9/month or $89/year, and it is aimed squarely at the awkward post-Genki stage rather than absolute beginners.

A Chrome extension that shows target-language and native subtitles simultaneously, lets you click any word for a definition, replay a line, and save phrases for review. The free tier already covers dual subs, dictionary lookups and the CEFR-graded video catalogue; Pro (about $6/month) adds machine translation, speech recognition and better AI lookups. It transforms passive Netflix time into study time, but you need enough of a base — roughly B1 — for subtitle mining to beat just watching.

A streaming service where every show has clickable dual subtitles and saved words become flashcards, covering 15 languages with apps on phone, Roku and TV. It solves the 'what do I watch' problem for languages where Netflix's catalogue is thin. The library is much smaller than a mainstream streamer and it is subscription-only — if you already pay for Netflix, Language Reactor's free tier does most of the same job.

Import podcasts, articles or ebooks, tap unknown words for definitions, and LingQ tracks your known-word count over time. The free tier caps how many saved words you can keep, so serious use means paying (~$13/month or less annually). The interface is clunky and the built-in SRS is weaker than Anki, but the reading-plus-audio workflow with 50+ languages is hard to replicate elsewhere.

Hans Ørberg's Familia Romana starts with a sentence you can understand knowing no Latin at all and never uses another language, building grammar and vocabulary by context and pictures across thirty-five chapters. It is the single most recommended way into Latin among people who actually end up reading Latin, and the reason is that you finish it able to read rather than able to parse. Slow at first; the pay-off arrives around chapter ten.

Ghalib, Faiz, Mir and tens of thousands of others, each poem displayed in Urdu, Devanagari and roman with word meanings on hover and often a recitation. Urdu's centre of gravity is its poetry, and this is the reason a learner can reach it years earlier than they otherwise could. Also hosts a large free ebook library. Literary register throughout — it will not teach you to buy a train ticket.

Dawn publishes in Urdu here and in English at dawn.com, frequently covering the same story in both. Reading the pair against each other is the cheapest parallel-text practice available for Urdu, and it costs nothing to set up. The Urdu edition's register sits between the BBC's plainness and Jang's density.

Well developed, with an active community, and written in Batua throughout. Read what you already know: Basque shares no vocabulary with anything, so context has to do all the work that cognates do in other languages.

State-run and free: levelled materials from A1 to C1, grammar reference, self-assessment tests, graded readers and two dictionaries. It is the most complete structured Georgian course on the open web, and it exists because the government funded it rather than because a market did. Institutional presentation and clumsy navigation; the material underneath is solid.

Daily news stories rewritten for children and non-native residents, with furigana on every kanji, definitions on hover, and a slow narrated recording of each article. Free, updated every weekday, and the single best bridge from textbook Japanese to real news. Note that NHK moved the site to news.web.nhk and renamed it NHKやさしいことばニュース; old www3.nhk.or.jp links redirect. Articles are short and the vocabulary skews toward politics and disasters.

A digital magazine that publishes everything bilingually in parallel columns — interviews, essays, learner material, poetry. Parallel text is the cheapest way to read above your level without stalling, and almost no other language has a publication built entirely around it. Free, and unusually generous about covering learners at every stage.

Tufts's classics library, where every word of every text is linked to a morphological analysis and a dictionary. For reading unadapted Latin this is the difference between fifteen minutes a sentence and two — you look words up without losing the thread. The interface is twenty years old and it does not matter.

Every classical Jewish text with a parallel English translation you can read line by line, plus word lookups into Jastrow and Klein and a growing modern-Hebrew section — an enormous free corpus for reading practice with a safety net. Genuinely useful if your interest in Hebrew includes the sources, and the interface for toggling and comparing translations is excellent. Be clear that most of it is Biblical and rabbinic Hebrew: the grammar and vocabulary differ from what Israelis speak, so it supplements a modern-Hebrew path rather than being one.

Around a quarter of a million articles written with simple words and short sentences, covering the same subjects as full Wikipedia — an endless supply of free reading on whatever you are already interested in, which is the hardest thing to find at lower levels. Reading about your own field here first makes the full English article much easier afterwards. Coverage is uneven and articles are much shorter and sometimes less accurate or less current than the main edition, so treat it as reading practice rather than a reference.

This is the hub of the Tagalog learning world: a dictionary of roughly 85,000 words with 66,000 native audio clips and 20,000 example sentences showing syllable stress and word-by-word breakdowns, alongside a text reader with click lookups, a daily-updated library of YouTube videos with transcripts, spaced-repetition decks including the well-known 2k frequency deck, and an active learner forum. If you use one Tagalog site, use this one. Much is free with an account, but the deepest dictionary and reader features sit behind a subscription, and the interface is dense rather than elegant.

State-funded courses from beginner level upward, plus a very large library of Tamil literature and reference works, all free. The presentation is institutional and the navigation is poor, but the material is real and there is a great deal of it — including classical Tamil texts with commentary.

Esperanto's Wikipedia is disproportionately big for the number of speakers, and unusually readable for a learner because the language's word-building means you can often decode a compound you have never seen. One of the earliest points at which reading in Esperanto stops feeling like an exercise.

News, science and culture stories written with a limited word list and read at about two-thirds of normal speed, each with the full text alongside the audio, plus long-running series like Everyday Grammar and Words and Their Stories. It remains free with no registration and the site was still publishing daily in August 2026. The register is deliberately simplified American English, so it builds comprehension rather than natural idiom, and its YouTube channel has been dormant since 2025 — go to the website rather than the video feed.

Indonesian has no tenses, no gender, no cases and a Latin alphabet, so an encyclopedia becomes readable within weeks rather than months. The affix system — meN-, ber-, -kan, pe-an — is the real grammar, and expository writing is where you meet it used properly.

Afrikaans is close enough to Dutch and shares enough with English that an encyclopedia becomes readable weeks rather than months in. Use that: read about things you know, and let the shared vocabulary do the work while you absorb the grammar's differences.

Read about what you already know. Azerbaijani switched from Cyrillic to Latin after 1991, so almost everything online is Latin — but older books and Iranian Azerbaijani use other scripts, which is worth knowing before you go looking for literature.

Sean Banville publishes current news stories graded into seven versions from very easy to near-native, each with audio at several speeds and a stack of downloadable exercises — gap-fills, dictations, discussion questions. The multi-level format means you can read the same story at your level and then reread it a level up, which is an unusually effective way to push comprehension. It is entirely free, but the site is ad-supported and visually cluttered, and the worksheets are aimed at classrooms so self-studiers will ignore half of them.

Full commentaries on Caesar, Ovid, Sulpicia and others, with running vocabulary, grammatical notes and often audio. The core vocabulary list — the thousand-odd words covering most of classical Latin — is worth an afternoon on its own. Free and openly licensed, from a US liberal arts college.

Thousands of short lessons across six difficulty tiers, each with clear recorded audio, a pop-up dictionary and toggleable pinyin, so you can read comfortably from your first hundred characters. A few free lessons refresh regularly; a subscription unlocks the library. The stories are written for learners rather than native content, which is exactly what makes it work early on but also why you should graduate to real material later.

Pairs a very active YouTube channel of short subtitled videos with easy readers, a weekly magazine and vocabulary courses on the site — the combination of listening and level-appropriate reading is unusual for Hungarian, where graded readers barely exist. Videos and grammar explanations on YouTube are free; the magazine subscription, books, workbooks and courses are paid. Uploads were current in August 2026. Best for A1–B1; advanced learners will find it too light.

Each episode tells a short story set in Brazil, then goes back through it one sentence at a time alternating Portuguese and English, so you get the meaning without reaching for a dictionary. Over 340 episodes and a new one most days, which solves the usual beginner problem of running out of material at your level. The podcast is free; a paid Plus tier adds transcripts and vocabulary lists. The site does not say who writes or narrates the stories, and the format is formulaic by design — useful as daily volume rather than as your only listening.

Read what you already know. Galician sits between Spanish and Portuguese and a reader of either gets a long way on day one — which is exactly the trap, because the differences are small, constant and where all the mistakes live.

A Hong Kong community project publishing story books written in real Cantonese (粵文, not standard written Chinese), graded step by step and split into a series for all ages and one written specifically for teens and adults, with a video and an audio recording for every book plus lesson plans for teachers. Reading it all online is free; printed sets are sold to fund the project. This is the graded-reading resource Cantonese has always lacked, but the site is Cantonese-first and awkward to navigate, and written Cantonese looks alien at first even if you read Chinese.

Classic stories (Sherlock Holmes, Jane Eyre, Chinese folk tales) rewritten with strictly controlled character sets and reset in Chinese settings, so a beginner can finish an entire book without constant lookups. Finishing one is a genuine motivation milestone that app-based reading rarely delivers. Books are cheap but paid, available in simplified and traditional, and the Level 0/1 range is narrow — you outgrow the series around HSK 4.

Long-form reporting on Nepal published in both languages, which makes it a ready-made parallel text: read the English, then the Nepali, on a story you now already understand. That trick is worth more than a textbook chapter and costs nothing.

A joint Academy and Fundación Barrié project writing about Galician vocabulary — etymologies, usage notes, words in danger of being lost to Castilian. Short pieces, and the closest thing Galician has to pleasure reading for a learner.

A self-publishing platform with a very large Bengali library — short stories, serialised novels, poetry — in modern, plain, conversational prose. That register is the gap between a textbook and Tagore, and almost nothing else fills it. Quality varies enormously because anyone can publish; sort by popularity and abandon anything that bores you.

The Portuguese entry in the Teach Yourself graded-reader series: sci-fi, fantasy, crime and thriller stories written for high-beginner to low-intermediate readers, with bolded vocabulary, chapter glossaries, plot summaries and comprehension questions. It is the most reliable way to get from textbook sentences to reading whole chapters. Brazilian Portuguese only — there is no European Portuguese edition, which is why Storyglot exists.

Two volumes — Beginners (A2–B1) and Intermediate Learners (B1–B2) — each with eight short stories, bolded glossary words, plot summaries and comprehension questions per chapter. The deliberate vocabulary control means you can read a whole chapter without stopping, which is the point. Story quality is functional rather than literary; treat it as reading-fluency training, not literature.

Olly Richards' graded readers solve the classic beginner problem that real Italian books are far too hard: sci-fi, crime and fantasy plots written within a controlled vocabulary, with per-chapter word lists and summaries so you rarely need a dictionary. Two volumes exist at beginner level, plus an intermediate one. The prose is deliberately plain — it's training material, not literature.

Olly Richards' graded reader for Russian solves the gap between textbook dialogues and real books, which is especially wide in Russian: controlled vocabulary, per-chapter word lists, plot summaries and comprehension questions so you can read without stopping every line. Good for building reading stamina at A2–B1. The language is deliberately simplified — it prepares you for real Russian prose rather than being it.

Pratham Books' open library holds tens of thousands of illustrated stories graded Level 1 to Level 4, with a very large Hindi collection — the closest thing Hindi has to graded readers, and all of it free to read online or download as PDF. Filter by language and level and you have months of extensive reading at genuine beginner difficulty. The content is written for children, so topics and register skew young, and the site covers many Indian languages, not Hindi alone.

Thai is written without spaces between words, which makes early reading brutal; this tool segments a pasted passage and glosses each word with tone-marked transliteration and audio. That makes it the fastest way to work through a Thai post, menu or news snippet before your parsing is automatic. Core lookup is free with a paid tier for extras, and its segmentation occasionally mis-splits names and compounds, so sanity-check anything surprising.

A news-based graded reader adding articles daily across levels roughly mapped to HSK 1–6, each with human-recorded audio, one-tap dictionary and exportable vocabulary. Because the content is current events rather than invented stories, it keeps intermediate learners reading when textbook dialogues get boring. Free sample lessons exist but the useful volume is behind a subscription, and simplified characters dominate.

Because The Daily Star publishes in both languages, you can often find the same story in English and Bengali and read them against each other — the cheapest parallel-text practice available for this language, and it costs nothing to set up. Coverage is national news and business.

Small but well tended, and strongest on Scotland, the islands and Gaelic literature. Read about what you already know; the initial mutations that make dictionary lookup hard are also what make reading volume so valuable.

Read what you already know. Lithuanian is the most archaic living Indo-European language and its vocabulary rewards anyone who has met Sanskrit or Latin — an encyclopedia is where those resemblances start jumping out.

Modest in size, and one of the larger bodies of freely available written Latvian. Read subjects you already know; the long marks on vowels are meaning-bearing and reading a lot is how you stop skipping them.

Read what you already know so context carries the vocabulary. Estonian compounds heavily, so a long unfamiliar word is usually two or three familiar ones — an encyclopedia is where that habit becomes visible and then automatic.

The Catalan Wikipedia is far bigger than the speaker count would predict and covers general topics well, not just Catalan ones. Read articles on subjects you already know; the shared Romance vocabulary means a Spanish or French speaker gets further on day one than they expect.