i+1, Made Measurable: Turning Krashen's Idea into a Number

By Beeblio · November 10, 2026

Every language learner who has spent time reading about second language acquisition has encountered the formula: i+1. It is elegant, intuitive, and — when you try to act on it — completely useless. How far above your level is "just beyond"? Which sentences count? How do you know when you have crossed from comprehensible into overwhelming? Stephen Krashen's 1982 formulation gave teachers and learners a powerful mental model, but it left the actual calibration entirely to guesswork. Forty years of vocabulary research, however, has quietly done the work Krashen never finished: it has translated "a little beyond your level" into something you can count. This article shows how frequency bands make i+1 comprehensible input measurable — and what that means for how you should choose what to read today.

What Krashen Actually Said (and Didn't Say)

Krashen (1982) famously argued that learners acquire language "in only one way" — by understanding input that is just beyond their current level of competence, the i+1 principle. He defined i+1 so that i is the learner's current level of proficiency and +1 is the level of proficiency just beyond it. The intuition is sound. Reading material that you understand nothing of is wasted time. Material you already understand perfectly gives you fluency practice but no new acquisition. The sweet spot is somewhere in the middle.

The problem is that Krashen never defined where, exactly, the middle sits. The inability to operationalize i+1 in research made it an untestable construct. Critics argued that Krashen's claims were vague, untestable, and overstated, lacking clear definitions and robust empirical support. Even sympathetic reviewers acknowledge that what exactly "one level above current competence" is remains hard to measure, and Krashen's framework does not give clean tools for doing so. The core insight — that comprehensible input is necessary for language acquisition — is essentially universal in modern SLA research — but the formula i+1 itself sits on foundations of sand.

The vocabulary research tradition stepped into this gap.

The Vocabulary Coverage Threshold

Starting in the late 1980s, researchers began asking a precisely measurable question: what percentage of words in a text does a reader need to know in order to comprehend it? The answer turns out to be both specific and somewhat uncomfortable.

Laufer's (1989) study was the first to compare reading comprehension at differing amounts of lexical coverage. She found that most L2 learners had reasonable comprehension of an academic text — operationalized as a score of 55% or higher on a comprehension test — when they knew 95% of the words in that text. In other words, one unknown word in every twenty is already enough to threaten comprehension for many learners. However, more recent studies tend to indicate that knowing 98% of the words in a text provides a much better chance that readers will be able to understand L2 written text. Nation (2006) reports that 98% coverage is most appropriate for incidental vocabulary learning.

The landmark study was Hu and Nation (2000). Hu and Nation (2000) investigated the relationship between unfamiliar vocabulary and comprehension, establishing an influential basis for considering the proportion of known words in a text. Their finding is sobering: learners generally require at least 95% known-word coverage for minimal reading comprehension (Laufer, 1989), and 98% for independent comprehension (Hu & Nation, 2000; Schmitt et al., 2011).

What does 98% coverage actually feel like? At 98%, you encounter roughly one unknown word in every fifty. At 95%, it's one unknown word in every twenty. The gap between those two numbers is small on paper but significant in practice — the 98% coverage would reduce more than half of the unknown words in the text that would be encountered if readers only had 95% coverage.

It is worth being honest about the limits of these numbers. Research has found a broadly linear relationship between the percentage of vocabulary known and the degree of reading comprehension, with no clear indication of a vocabulary "threshold" where comprehension increased dramatically at a particular percentage. The 95% and 98% figures are probabilistic guidelines, not bright lines. Text type, genre, background knowledge, and individual learner variation all matter. But as a practical target for choosing reading material, the 95–98% window is the best tool the field currently offers.

Frequency Bands: Turning Coverage into a Countable Number

Coverage percentages become actionable when you map them onto word-frequency bands. The key insight from corpus linguistics is that word frequency follows a steep power law: a small number of very common words do an enormous share of the work in any text.

The most frequent 1,000 words account for around 75% of the running words in formal written texts and around 84% of informal spoken use. From there, the gains diminish quickly. Research on spoken and written corpora found that the first 1,000 word-family level provides cumulative coverage of around 81–82%, climbing to roughly 88% at the 2,000-word level, 92% at 3,000, and 94% at 4,000. To reach the 98% threshold needed for truly independent reading, Nation (2006) found that the most frequent 8,000 to 9,000 word families plus proper nouns provided 98% coverage of newspapers and novels.

These numbers, plotted as a curve, give you something genuinely useful: a map. If you know your position on the map — if you can say "I know the top 800 most frequent German word families" — you can immediately compute roughly what percentage of any given text you are likely to understand. That position is your i. The band just above it, with the words you do not yet know, is your +1.

This is the move that makes Krashen's formula practical. Instead of an abstract inequality, you get a frequency number.

Reading in the Zone: What a Band-800 Text Looks Like in German

To make this concrete, consider a learner whose known-words count sits at the 800-word band in German. At that level, they already have solid command of the most common pronouns, verb forms, prepositions, and everyday nouns. What percentage of a typical passage do they already know? Beeblio measured the following short passage against the top-800 most frequent German word families:

The result — 100% of the tokens in that passage covered by the top-800 band — means this particular text is well within the learner's current zone. It is excellent for fluency practice and confidence-building, but it sits closer to i than to i+1. For genuine acquisition, the learner needs texts where perhaps 5–10 new words per page fall just outside their known band: words from the 801–1,000 range, introduced in context rich enough to allow intelligent guessing. That is the operational definition of i+1 comprehensible input that Krashen never provided.

Why "Just Beyond" Requires Both a Ceiling and a Floor

One subtlety that gets lost in many discussions of i+1 is that the idea has two constraints, not one. Input must not be too far above your level — that much everyone remembers. But it also should not be so far below your level that nothing new is encountered. Nation's framework distinguishes intensive reading, targeting less than 98% coverage for language-knowledge development; extensive reading for language growth at around 98% coverage, optimized for incidental vocabulary learning; and extensive reading for fluency development at 99–100%, aimed at reading quickly.

The practical implication: a single coverage number does not prescribe a single text. It prescribes a purpose. The same learner reading the same text at 95% coverage might be doing hard but productive acquisition work. Reading at 100% coverage is not wasted — it builds the kind of automatic, fluent processing that allows vocabulary to shift from recognized-with-effort to truly known. Both are legitimate activities, but they serve different goals, and conflating them is what leads learners to either avoid challenging texts entirely or to grind through texts so difficult that every other sentence requires a dictionary lookup.

Incidental Vocabulary Learning and the Role of New-Word Density

When you read in the 95–98% coverage zone, something important happens beside comprehension: you pick up vocabulary incidentally. Incidental learning of vocabulary from context while reading is a major avenue to lexical growth in either a first or second language. The mechanism is not magic: you encounter an unknown word in a context that makes its meaning inferrable, you infer it, and — given enough repeated encounters over time — it enters your passive vocabulary.

How many encounters does it take? Vocabulary acquisition research suggests the number is higher than most learners hope. The consensus figure is typically in the range of 10 or more encounters before a word is reliably retained, though this varies considerably with word type, learner attention, and the richness of context. This is why volume matters alongside difficulty calibration: even at the right coverage level, acquisition is slow if you read too little.

This is also where spaced repetition enters the picture. Research on the spacing of encounters has revealed that spaced learning consistently leads to greater retention of vocabulary knowledge than massed learning — that is, learning without an interval between repeated encounters. The benefits of spacing are particularly large in L2 vocabulary learning. Reading at your band gives you the initial encounters; flashcard-style review, timed by a spaced-repetition algorithm, provides the consolidation encounters at the intervals that maximize retention.

The Practical Argument for Measuring Your Band

None of the above is actionable unless you know where you currently stand. This is the deep problem with self-reported proficiency levels (Beginner, Intermediate, Advanced) and even with standardized tests that report a broad CEFR band: they tell you a category, not a coordinate. Knowing that you are "B1 in German" tells you roughly how well you can communicate, but it does not tell you what percentage of a news article you are likely to understand — or which 200 words, if you learned them, would push you from 92% to 95% coverage.

A frequency-band count solves this. If the app tells you that you have a confirmed vocabulary of 1,200 word families in your target language, you can immediately reason: at approximately the 1,200-word level in German or Spanish, general texts will be around 90–92% covered. You need roughly 400–600 more high-frequency word families before the 95% threshold becomes routine. That is a concrete, plannable goal — far more useful than "get better at German."

It also makes your reading choices more deliberate. Instead of picking texts by feel or prestige ("I should be reading a newspaper by now"), you pick texts by measured fit. A text with 3% unknown words is probably right. A text with 12% unknown words is a wall, not a challenge.

What the Research Does Not Settle

Intellectual honesty requires flagging what the vocabulary-coverage framework cannot do. First, it measures tokens (individual word occurrences), not comprehension of sentences or discourse structure. A text at 98% coverage can still be grammatically opaque if the syntax is complex. Second, the thresholds — 95%, 98% — were established mostly with narrative and academic English texts. The vocabulary size necessary for reading comprehension may vary depending on the type of text and the degree of comprehension required. A poetry collection, a legal contract, and a thriller novel at the same coverage level are not equally comprehensible. Third, the often-cited 98% figure was inferred from conditions with 80%, 90%, 95%, and 100% known-word coverage; a later replication and extension found that the original results did not fully replicate in a different group.

These caveats do not overturn the usefulness of frequency bands as a planning tool. They mean you should treat them as calibration, not as guarantees. A coverage estimate tells you the vocabulary load is manageable; it does not promise you will understand the text.

FAQ

Is 98% coverage really necessary, or is 95% enough?

It depends on what you are trying to do. Researchers have suggested that 95% coverage might be needed for minimal comprehension, while 98% might be necessary for optimal comprehension. For effortful but productive reading with intentional study, 95% is workable. For relaxed, incidental acquisition where you want to infer word meanings easily and retain them, aim for 98%. Most learners find 95% noticeably harder than the numbers suggest, because the unknown words tend to cluster rather than distribute evenly.

How many words do I actually need to know to read comfortably?

It depends on the text type. Nation (2006) found that learners would need to know 6,000–7,000 word families to have 98% coverage of spoken texts, and 8,000–9,000 word families to reach the same coverage of written text. For graded readers, simplified novels, or casual conversation, the threshold is far lower — often 2,000–3,000 word families is enough for 95% coverage of simplified material. The number that matters is not an absolute target but the gap between where you are now and where your target texts sit.

Does frequency-band reading replace explicit vocabulary study?

No, and the research is clear on this. Nation (2001) identifies four pathways for learning high-frequency words: direct teaching, direct learning (e.g., from word cards), incidental learning through extensive reading, and planned encounters through vocabulary exercises. Frequency-calibrated reading is primarily an incidental learning environment. For the top 2,000–3,000 word families of any language, explicit study is typically faster and more reliable than waiting for incidental exposure. After that point, the words become rare enough that only extensive reading provides sufficient encounters.

What is the difference between a word family and an individual word?

A word family groups a base word together with its inflected and derived forms. In English, teach, teaches, taught, teaching, teacher, teachers, teachable all belong to the same word family. Research on vocabulary thresholds typically counts in word families because knowing the base form of a word generally allows you to recognize the inflected forms. This matters for practical targets: "know 3,000 words" measured in word families represents considerably more coverage than 3,000 individual tokens.

Can my frequency-band score go backwards?

Your verified known-word count should only increase over time, since it reflects words you have demonstrated you can recognize. What can happen is that a formal measurement — encountering words you thought you knew in a genuinely contextualized test — reveals gaps in words you assumed you had. This is not regression; it is calibration. A band score based on actual evidence is more useful than one based on self-report, even if the evidence-based number is initially lower than you expected.

Sources

• Krashen, S. D. (1982). Principles and Practice in Second Language Acquisition. Pergamon.
• Hu, M., & Nation, I. S. P. (2000). Unknown vocabulary density and reading comprehension. Reading in a Foreign Language, 13(1), 403–430.
• Nation, I. S. P. (2006). How large a vocabulary is needed for reading and listening? Canadian Modern Language Review, 63(1), 59–82.
• Laufer, B., & Ravenhorst-Kalovski, G. C. (2010). Lexical threshold revisited: Lexical text coverage, learners' vocabulary size and reading comprehension. Reading in a Foreign Language, 22(1), 15–30.
• Nation, I. S. P. (2001). Learning Vocabulary in Another Language. Cambridge University Press.
• Schmitt, N., Jiang, X., & Grabe, W. (2011). The percentage of words known in a text and reading comprehension. The Modern Language Journal, 95(1), 26–43.
• Kim, Y., & Webb, S. (2022). The effects of spaced practice on second language learning: A meta-analysis. Language Learning, 72(1), 269–319.
• Lichtman, K., & VanPatten, B. (2021). Was Krashen right? Forty years later. Foreign Language Annals, 54(2), 1–21.
• Frontiers in Psychology (2025). Beyond comprehensible input: a neuro-ecological critique of Krashen's hypothesis in language education.

In Beeblio, your known-words count is the number that answers the question Krashen left open. Every text you read is profiled against that count so you can see, before you start, whether it sits in your fluency zone, your acquisition zone, or beyond your comfortable reach. When you tap a word you do not know, it moves into a spaced-repetition queue calibrated to surface it again at the interval most likely to push it into long-term memory — transforming a single incidental encounter in a text into the series of spaced encounters that the research says acquisition actually requires. Reading at your level, with just enough stretch beyond it, is not a soft recommendation. It is what the numbers say.