The Contos reading report: what 1,536 free chapters tell us about graded reading

What exactly was measured?

Everything on this page comes from one source: the text of the Contos library as exported on 16 October 2026, after that day's corrections to the story text. It contains no reader data at all, so nothing here says how people read, how far they get or what they remember. It describes the texts, not the learners.

In the Contos library (October 2026), every one of the 16 learning languages has 48 series, 12 at each CEFR level from A1 to B2. That makes 768 series. Each language has one flagship series written separately at all four levels, which runs to eight chapters per level; every other series runs to six. In all, the 768 series hold 4,736 chapters, and chapters 1 and 2 of every series are free to read, so the free text is 1,536 chapters, or about a third of the library.

The text measures below use 767 of the 768 series: one Spanish B1 series is held back from this site while a fault in its second chapter is fixed, so Spanish B1 counts 22 chapters instead of 24.

How we counted

• Words are runs of letters, with internal apostrophes and hyphens kept inside the word. Numbers written as digits are left out of the vocabulary counts.
• Japanese and Chinese are counted in characters (kanji, kana or hanzi), not words. Word spacing in the Japanese text is not yet consistent across the catalogue, so a word count would measure our spacing rather than the language.
• Korean is counted in the space-separated units Korean uses (eojeol), which usually carry a particle or ending, so one Korean "word" often corresponds to two or more English ones.
• Sentences end at a full stop, question mark, exclamation mark or ellipsis followed by a capital letter. A line of dialogue followed by "she said" stays one sentence.
• Word forms count each spelling once: habla and hablaba are two forms. Researchers usually count word families instead, which group these together, so our totals run higher than a family count of the same text would.

How long is a chapter at each level?

Average length of the free chapters, by language and level. Shortest and longest chapters are in the second table.

Three things stand out. First, each level adds a similar amount: about 70 words from A1 to A2, 62 from A2 to B1 and 65 from B1 to B2, so a B2 chapter is about 2.2 times as long as an A1 chapter. Second, the levels overlap at the edges: the longest A2 chapter is longer than the shortest B2 chapter, so level is about more than length.

Third, Turkish, Polish, Russian and Ukrainian chapters have fewer words at every level. That is the languages, not lighter stories. None of the four uses articles, and all of them pack more into each word with endings: the average A1 word is 5.5 letters long in Turkish and 4.8 in Polish, Russian and Ukrainian, against 4.3 in Spanish. Fewer, longer words carry roughly the same story.

If you want to turn these lengths into a reading plan, our extensive reading page has the weekly volume research suggests, and choosing your level turns them into a budget of unknown words per chapter.

How long are the sentences?

Average words per sentence in the 11 Latin-script languages, chapters 1–2:

Unlike chapter length, sentence length does not climb evenly. The step from A2 to B1 (6.3 to 9.9 words) is more than three times the size of the step from B1 to B2 (9.9 to 10.9). Much of the extra length comes from joined clauses. In German, the subordinating words weil, wenn, dass, ob, obwohl, während, nachdem, bevor and damit appear 0.7 times per 1,000 words at A1, 4.2 at A2, 10.6 at B1 and 13.3 at B2. That is the jump our page on the intermediate plateau describes from the reader's side.

Russian and Ukrainian, counted the same way, follow the same curve: 4.1, 5.6, 9.0 and 10.4 words in Russian, 4.2, 6.1, 9.4 and 10.0 in Ukrainian. In characters per sentence, Japanese goes from 10.3 at A1 to 16.1 at B2 and Chinese from 7.2 to 16.8.

The Romance languages write the longest sentences at B1 and B2, around 12 to 13 words, while German, the Scandinavian languages and Polish sit nearer 10. Part of that may again be how each language builds words: German compounds, for instance, fold into one word what Spanish often spreads across a phrase.

How fast does the vocabulary grow from level to level?

We counted the different word forms in each level's free chapters, for the 13 languages that use spaces between words. Because longer texts always contain more different words, we also counted the different forms in every 100-word stretch of text, which makes levels with different chapter lengths comparable.

The totals grow at every step, but the density tells a different story. Different forms per 100 words rise from A1 to B1 and then barely move: B2 chapters are longer than B1 chapters but, stretch for stretch, no more varied. The extra difficulty at B2 comes from length and sentence structure more than from denser vocabulary.

The last column shows how much of each level is new territory. About half the word forms in the B2 free chapters never appear in the A1, A2 or B1 free chapters of the same language. Some of that is names and story-specific words, and the free chapters are only a sample of each level, so read it as an upper bound.

Polish, Russian, Ukrainian and Turkish show the most forms from fewer running words. Their nouns change with case and their verbs carry more endings, so one word turns up in several spellings. A learner of those languages meets more forms for the same story, which is one reason a word-family count is a fairer measure of what you know than a list of spellings.

Japanese and Chinese, measured in characters, show the same climb. Japanese free chapters use 368 different kanji at A1, 494 at A2, 676 at B1 and 859 at B2. Chinese chapters use 565, 729, 918 and 1,251 different characters.

What kinds of stories are in the library?

Every series carries one or two of six genre tags. Mystery is the most common: just over half the series carry it. The mix shifts with level.

Comedy is concentrated at A1 and A2, mystery peaks at B1, and family stories are most common at B2. Crime stays cosy throughout: thefts, forgeries and secrets rather than violence. Genre counts are almost identical from one language to the next because most series concepts are written for every language.

What can't these numbers tell you?

• How hard a chapter is for you. Difficulty depends on which words you already know. Two chapters of the same length can feel very different.
• Anything about later chapters. Only chapters 1–2 are measured. Later chapters are written to the same level but were not counted here.
• How languages compare in general. Every language in the library was written to the same level guidelines, so the cross-language patterns describe this library, not Spanish or Polish at large.
• Quality. Counting words says nothing about whether a story is good, natural or error-free.
• A fixed picture. New series arrive every week, so the totals will move. This is a snapshot of 16 October 2026. Our Spanish-only pages used slightly different counting rules and can differ by a word or two.

Graded readers are usually levelled by headwords, a fixed list of permitted word families, and Rob Waring notes that publishers' lists differ, especially at higher levels. Contos levels are CEFR levels, and a learner moves up by known words rather than by a headword list. The figures above are the best way we have to show what that means on the page; our graded readers and CEFR levels pages give the background.