How to actually read in another language

Research into extensive reading puts the comfortable zone at 95–98% known words: enough unfamiliar words to learn something, not so many that you're just decoding. If you're stopping on more than one word in twenty, the text is too hard for this kind of reading; save it for later and pick something easier now.

Stopping for every unknown word breaks the sentence you were in the middle of understanding. Guess from context first. The evidence on whether guessing alone teaches more words than a dictionary would is genuinely mixed, but guessing is reliably faster, and staying in the sentence is what turns a word from new to familiar in the first place.

Context guessing works until it doesn't: if a word is blocking the sentence's meaning, or the same word keeps resisting a guess, look it up. The goal was never "use the dictionary as little as possible"; it's staying in the text as much as possible. A lookup that gets you back to reading is a win either way.

Translating word-by-word turns reading into decoding: you can finish a paragraph without having actually read it. Read for the gist first. A handful of unclear words rarely costs you the sentence if the rest of it makes sense, and chasing perfect understanding of every word is usually a worse use of the same time.

A single paragraph studied carefully teaches less than several pages read at a comfortable level. Extensive reading research consistently finds that how much you read, and how often, matters more than how thoroughly any one page gets picked apart. Fifteen or twenty minutes most days outperforms an occasional long, careful session.

A word you recognize but haven't fully learned usually needs another exposure, not a flashcard. Meeting it again inside a sentence you already understand is often exactly what moves it from "familiar" to "known." Rereading isn't wasted time, even though it can feel like it.

Extensive reading only works if you keep doing it, and a textbook dialogue rarely survives contact with real motivation. Pick the kind of thing you'd read in your first language (a hobby, a writer you like, the news you'd check anyway): the content matters less than whether you come back to it tomorrow.

Comprehension and production use different muscles. Merrill Swain's Output Hypothesis grew out of watching learners who could understand a huge amount while still struggling to produce even simple sentences: recognizing a word doesn't require getting it exactly right, only using it does. Summarize what you just read out loud, or write one sentence with a new word in it, because producing the language forces you to notice gaps that reading alone lets you quietly skip past.