Vocabulary thresholds for reading have been studied for decades, and the numbers agree across researchers and languages. Short version: about 3,000 word families to read most fiction with some effort, 5,000 for comfortable reading, and 8,000 to 9,000 for the smooth reading you have in your native language. Below: what those numbers feel like, how they vary by language, and why reading itself is the fastest way to reach them.
The Research: Coverage Thresholds That Actually Matter
Paul Nation and Batia Laufer, from the late 1980s through the 2000s, established “text coverage”: the share of words in a text a reader knows. Laufer (1989) proposed 95% as the minimum for “reasonable” comprehension. Nation (2006) set 98% as the threshold for comfortable, unassisted reading without a dictionary.
Vocabulary against coverage, from their work and later studies by Hsueh-chao and Nation (2000) and Webb and Macalister (2013):
| Word Families Known | Approximate Text Coverage | What It Feels Like |
|---|---|---|
| 1,000 | 78-80% | Every 5th word is unknown. Exhausting. You get isolated words but miss the meaning of most sentences. |
| 2,000 | 85-90% | Every 7th-10th word is unknown. You grasp the general topic but lose the thread constantly. |
| 3,000 | 93-95% | 1 unknown word every 15-20 words, roughly 1 per line. Manageable with effort. You follow the story but miss nuances. |
| 5,000 | 97% | 1 unknown word every 30-35 words, roughly 1 every 2 lines. Comfortable. You can infer most unknowns from context. |
| 8,000-9,000 | 98% | 1 unknown word every 50 words, roughly 1 per paragraph. Smooth reading. Unknowns rarely disrupt flow. |
The percentages look close, but at 95% you stop about once a line, and at 98% you may go a page before an unknown word matters. That is the gap between struggle and pleasure.
What These Percentages Actually Feel Like
Take the opening of a Harry Potter book: “Mr. and Mrs. Dursley, of number four, Privet Drive, were proud to say that they were perfectly normal, thank you very much.” That is 22 words.
- 80% coverage: 4 or 5 unknown. “Mr. and Mrs. _____, of number four, _____ Drive, were _____ to say that they were _____ normal, thank you very _____.” Slow, uncertain guessing; over 300 pages you quit by chapter three.
- 95%: about one unknown, maybe “proud” or “perfectly”, recoverable from context. But it happens 10 to 15 times a page, so reading takes effort and focus.
- 98%: perhaps none in this sentence and 2 or 3 per page, mostly not critical. This is how native speakers read.
What “Word Family” Means (And Why It Matters)
A word family is a base word plus its inflected and closely derived forms. “Run”: run, runs, running, ran, runner, runners. “Happy”: happy, happier, happiest, unhappy, happiness, happily. One family each, six forms each.
When researchers say 5,000 words they mean families, which is roughly 15,000 to 25,000 word forms. The model is optimistic for learners: know “inform” and you recognize “information”, “informative”, “informed”, “uninformed” and “misinformation” without study. You learn 8,000 base concepts, not 8,000 unrelated items.
Nation (2001) estimates an educated adult native English speaker knows about 15,000 to 20,000 families. You do not need that: at 8,000 to 9,000, about half, you can read virtually anything written for a general audience.
Real-World Benchmarks: What Can You Read at Each Level?
500 Word Families
Greetings, numbers, colors, common verbs (be, have, go, want, like), question words, some concrete nouns.
What you can read: picture books for 2 to 4 year olds; menus, street signs, labels; very simple textbook dialogues; social media posts without slang.
What it feels like: scattered familiar words in a wall of unfamiliar text. Connected text is out of reach.
1,000 Word Families
The structural core (articles, prepositions, pronouns, basic conjunctions) plus the commonest everyday words.
What you can read: fairy tales and folk stories for children; the lowest-level graded readers; simple headlines; texts and casual posts from friends, if you can parse the slang.
What it feels like: you follow a simple narrative skeleton (“The man went to the house. He opened the door. The room was dark.”) and reread often. Complex description and abstract ideas stay out of reach.
Example: a simplified “Caperucita Roja” (Little Red Riding Hood) in Spanish. You follow girl, forest, wolf, grandmother, and miss many details.
2,000 Word Families
A significant milestone: the high-frequency core is covered and coverage jumps.
What you can read: young adult novels in plain language; headlines and short news summaries; blog posts on everyday topics; most song lyrics; upper-level graded readers; social media, fluently.
What it feels like: you follow the main events; motivation and emotional nuance come through unreliably. Reading is work, but productive rather than guesswork.
Example: “El Principito” (The Little Prince) without much trouble. “Cien Anos de Soledad” (One Hundred Years of Solitude) is a struggle: García Márquez uses rich literary vocabulary full of botanical terms, archaic words and elaborate description.
3,000 Word Families
For many learners, reading turns from study into entertainment here.
What you can read: most contemporary fiction (thrillers, romance, science fiction, mystery); adapted classics; popular non-fiction (self-help, popular science, biography); full newspaper articles on general topics; Wikipedia in your target language.
What it feels like: you read for the story. An unknown word turns up about once a line but rarely blocks comprehension, and you can read 30 to 60 minutes without exhaustion.
Example: in French, a Marc Levy novel or a translated Dan Brown thriller: direct language, mainstream vocabulary, a plot that carries you.
5,000 Word Families
Most native-audience content opens up.
What you can read: unadapted native novels; quality journalism (Le Monde, El Pais, Der Spiegel); specialized non-fiction on your interests; classic literature, with patience for archaic words; professional documents in your field.
What it feels like: close to reading in your native language. You note an unfamiliar word and move on, read for hours, prefer certain authors and notice good or bad writing.
Example: in German, a contemporary Daniel Kehlmann novel or a translated Stephen King without aids, with 1 or 2 unknown words a page, mostly inferable.
8,000+ Word Families
Nation’s threshold for comfortable, unassisted reading: 98% coverage.
What you can read: academic texts; experimental literary fiction; technical writing outside your field; poetry (which has challenges beyond vocabulary); historical texts with period vocabulary; essentially anything for a literate adult.
What it feels like: you forget you are reading a foreign language. Unknown words feel like unusual words at home, a mild curiosity. Your speed approaches native.
Language-Specific Differences: Not All Words Are Created Equal
The numbers above come mostly from English research. Other languages change the equation.
Chinese: Fewer Families, Different Counting
Chinese words do not inflect, so there are no word families in the usual sense. Characters combine predictably: “electricity” plus “brain” is computer, plus “speech” telephone, plus “shadow” movie. Each new character gives partial access to dozens of compounds. Liang et al. (2015) suggest Chinese learners reach comparable coverage with fewer base items than English learners. About 3,000 to 4,000 high-frequency characters cover 99%+ of characters in general text, though they form many more multi-character words. Count characters and common combinations, not words: fewer units, but more effort per unit.
German: Compound Words Inflate the Count
“Handschuh” (hand + shoe = glove), “Krankenhaus” (sick + house = hospital), “Geschwindigkeitsbegrenzung” (speed + limitation = speed limit). A German novel has more unique word forms than the same novel in English, but many are combinations of words you know. Bauer and Nation (1993) and later work on German suggest word-formation awareness gives bonus coverage: 5,000 German base families may work like 6,000 to 7,000 in English.
Romance Languages: The Cognate Advantage
Thanks to the Norman Conquest and centuries of exchange, English shares a huge cognate pool with French, Spanish, Italian, Portuguese and Romanian. An English speaker recognizes “important”, “different”, “possible”, “information”, “communication”, “situation” and thousands more in French. Estimates put recognizable cognates at 30 to 40% of written French and 20 to 25% of spoken French, where pronunciation hides the link.
Laufer and Goldstein (2004) found cognate awareness significantly speeds acquisition in related pairs. An English speaker effectively starts Spanish with 1,500 to 2,000 recognizable families. False cognates (“embarazada” means pregnant, not embarrassed) and partial cognates confuse, but the time to reading thresholds drops sharply. To reach 8,000 effective Spanish families, an English speaker actively learns about 5,000 to 6,000, since 2,000 to 3,000 come free. Weigh these differences in difficulty across languages when you set goals.
Japanese: Kanji Knowledge Compounds
Most Japanese words are written with 1 to 3 kanji, each carrying meaning, so knowing “work” and “place” decodes “workplace”, “working hours”, “homework” and more. Most kanji have at least two readings, one Chinese-derived and one native, which doubles the load per character but makes each more productive. Mori and Shimizu (2007) found kanji knowledge predicts Japanese reading ability better than word knowledge: a learner with 2,000 kanji and a smaller measured vocabulary may out-read one with more words and fewer kanji. Comfortable reading takes about 2,000 to 2,500 kanji (the school list through high school) plus 6,000 to 8,000 vocabulary items.
Agglutinative Languages: Turkish, Finnish, Hungarian
These languages stack suffixes: Turkish “evlerinizden” = “from your houses” (ev + ler + iniz + den: house + plural + your + from). One Turkish verb can generate hundreds of forms, so “word family” breaks down, and 3,000 Turkish base words reach far more forms than 3,000 English ones. Mustonen (2015) on Finnish suggests 2,000 to 3,000 base lemmas reach 95% coverage of general fiction, against 3,000 to 5,000 for English. Each lemma costs more (you learn the morphology), but there are fewer to memorize.
How Reading Itself Builds Vocabulary: The Virtuous Cycle
Reading needs vocabulary and is also the strongest way to build it, so each feeds the other.
Incidental Vocabulary Acquisition
Waring and Takaki (2003) and Pellicer-Sanchez and Schmitt (2010) found readers learn 5 to 15% of unknown words from context alone. A 60,000-word novel with 600 unknown families yields 30 to 90; ten novels yield 300 to 900 without flashcards. The rate depends on:
- Context quality: informative contexts teach faster than ambiguous ones.
- Frequency: one sighting rarely sticks; 6 to 12 across a book often do.
- Existing vocabulary: more known words give better inference, so your 5,001st word is easier than your 1,001st.
- Engagement: words in compelling, emotional text stick better than in dry text. Novels beat textbooks.
The Multiple Encounter Threshold
Researchers converge on 6 to 12 meaningful encounters to know a word (Nation, 2001; Webb, 2007). Meaningful means processing it in a sentence, not seeing it flash by on a list, and each new context enriches the memory. Authors reuse vocabulary heavily: Stephen King’s “It” uses “sewer” more than 100 times. Several books by one author give repeated exposure to a consistent word set; a new author then adds fresh words on top of the shared core. That alternation of depth and breadth is the rhythm of extensive reading.
The Annual Acquisition Rate
- Nation and Waring (1997): about 1,000 new families a year through extensive reading at the right level.
- Webb and Chang (2015): 500 to 1,500 a year, depending on amount and level.
- At B1 (about 2,500 to 3,000 families), 30 minutes a day yields roughly 800 to 1,200 a year from reading alone; with conversation, media and deliberate study, 1,500 to 2,000.
From 2,000 families, 5,000 takes about 2 to 3 years and 8,000 another 2 to 3. That fits the US Foreign Service Institute’s estimate for professional proficiency in Category I languages (600 to 750 hours, spread over 2 to 3 years). Comfortable reading is a 2 to 4 year commitment of daily reading, not a decade, and enjoyable with the right material.
Practical Strategy: From 0 to 3,000 Words Through Reading
Stage 1: The First 100-250 Words (Week 1-2)
Method: a high-frequency list of the 100 or 250 most common words: articles, pronouns, prepositions, basic verbs, numbers, question words. Lingo7 includes pre-made word collections organized by frequency for this.
Goal: recognize them on sight (production can wait). Flashcards are efficient here because you will meet these words constantly. Reading does the rest.
Stage 2: First 500-1,000 Words with Parallel Translation (Week 2-8)
At 250 to 500 families you cannot follow even simple texts alone, so read with a translation beside the original or on tap.
Method: read each foreign sentence first, check the translation when stuck, then reread the original with the new understanding. This is scaffolding, not cheating, and you check less over time.
What to read: children’s stories, very short stories, opening chapters of familiar books. If you know Harry Potter’s plot, reading it in French removes one layer of difficulty.
Goal: 500 to 1,000 families, mostly from repeated encounters, plus deliberate review of frequent words context does not explain.
Stage 3: Graded Readers and Simplified Content (Month 2-6)
Method: read extensively, for volume, 20 to 30 minutes a day. Look up only recurring, important words and keep 5 to 10 per session for review.
What to read: graded readers from Oxford, Penguin or language-specific publishers, simplified classics, young adult fiction where available.
Goal: 2,000 families. When graded readers feel easy, approach simpler native content.
Stage 4: Native Content with Support (Month 6-18)
The hardest stage psychologically: real books after graded readers can feel discouraging.
Method: native books with support: parallel translation when you lose the thread, word lookup when a recurring word blocks you, audio narration for pronunciation and speed. Stay in the foreign language and use support only when needed.
What to read: contemporary fiction in accessible prose. Thrillers, romance and young adult novels use direct language, short sentences and modern vocabulary. Skip literary fiction, poetry and historical novels for now. More in our guide on choosing books by language level.
Goal: 3,000 families, checking the translation once every few pages instead of every few sentences.
Stage 5: Independent Reading and Spaced Repetition (Month 18+)
Reading now sustains itself: each book makes the next easier.
Method: read widely and rotate genres. Save useful words for spaced repetition. Context plus spaced review is the most effective vocabulary strategy documented in the research.
Goal: 3,000 to 5,000 and then 8,000+ families, over years, through books you want to read.
Common Misconceptions
”I Need 10,000 Words Before I Can Start Reading”
The most damaging myth: study for years, then read. The reverse is true. With parallel translation, word lookup and graded readers you can read productively at 500 words, and starting early matters because reading builds the vocabulary for more reading. Waiting to “know enough words” is waiting to be fit before going to the gym.
”Knowing More Words Always Means Better Reading”
Vocabulary size is the strongest single predictor of L2 reading comprehension, but depth matters as much as breadth. A learner with 5,000 families known as single translations (happy = feliz) reads worse than one with 4,000 known deeply: that “feliz”, “contento” and “alegre” are all “happy” in different ways, with different collocations and registers. Reading builds that depth: a flashcard gives one L1 link, a novel gives dozens.
”All Languages Require the Same Vocabulary Size”
False. An English speaker learns fewer Spanish words than Finnish ones thanks to cognates, and Chinese reaches high coverage with fewer base units. Use 8,000 to 9,000 families as a benchmark for most European languages and adjust: subtract 2,000 to 3,000 for Romance languages, count characters for character-based languages, count lemmas for agglutinative ones.
”Frequency Lists Are All You Need”
Lists are valuable for the 500 to 1,000 words that form the backbone. Beyond about 2,000 they give diminishing returns, because low-frequency words are domain-specific: the 5,000th French word may belong to law, medicine, cooking or politics. A reader of French crime fiction needs different rare words than a reader of French cookbooks. After the core, read in your domains, and the words you need will find you.
”Children Learn 10 Words a Day, So Adults Should Too”
The figure (3,000 to 4,000 a year) comes from Nagy and Herman (1987) on first-language acquisition. It misleads for adult L2 learners:
- Children get 12 to 16 waking hours of exposure a day, against your 30 minutes.
- It is an average over a whole childhood of fast and slow periods.
- First-language acquisition involves partly different cognitive processes.
A realistic adult target with 30 minutes of daily reading is 2 to 4 families a day, roughly 800 to 1,500 a year. After 3 years that is 2,400 to 4,500 from reading alone, well within the comfortable reading range.
Building Your Reading Vocabulary with Lingo7
Lingo7 is built around this strategy: high-frequency words first, reading with parallel support, scaffolding removed gradually.
- Books in 49 languages with sentence-level parallel translation, native audio narration and tap-to-translate word lookup, so comprehension holds even at low vocabulary.
- One-tap saving: words are saved with the sentence you found them in.
- Spaced repetition schedules reviews at optimal intervals, moving words from recognition to long-term memory.
- Pre-made word collections by frequency give you the first 100 to 250 words before you start reading.
You pass through the thresholds in this article while reading real books, from “I know 200 words and cannot read anything” to “I just finished my third French novel and checked the translation twice in the last chapter”.
The Bottom Line
You do not need to reach a threshold before you start reading; you read to reach it. Reading and vocabulary grow together. Start with 250 words and a parallel text. Read something short, save and review the words that recur, then read something slightly harder. Within a few months the first texts will look trivially easy, and eventually you will forget you are reading in another language.