필사 모드: How to Read a Difficult Book — Five Judgment Calls That Matter More Than Finishing

Introduction — I Read the Same Paragraph Four Times and Nothing Stuck

The most common experience in front of a difficult book goes like this. You read a paragraph. You don't understand it. You read it again. Still nothing. A third time, a fourth. And then it hits you: there wasn't a single word you didn't know.

That's the whole problem. A difficult book is usually difficult for reasons that have nothing to do with vocabulary. You know every word, but the paragraph won't click; you get the paragraph, but you can't tell why this chapter exists at all. In other words, the difficulty lives in the structure, not the vocabulary. And because you keep trying to solve a structural problem with a vocabulary-level fix, you're stuck in the same place after four passes.

This piece lays out how to work with that structure. It draws on two strands of evidence. One is Mortimer Adler and Charles Van Doren's How to Read a Book (first published 1940, revised 1972; in Japan it's published as Hon o Yomu Hon, and in Chinese as 如何阅读一本书). The other is psychological research on reading and learning. And wherever popular reading advice has run ahead of what the evidence actually supports, I'll flag it as we go. There are more of those spots than you'd expect.

What to read is a separate question, covered in Novels Worth the Time. This piece picks up after that — what to do once you've picked something and it won't go in.

The Difficulty Usually Lives in the Structure, Not the Words

Pinpoint exactly where you're stuck and the prescription changes. Most readers don't know where they got stuck; they just lump the whole experience under the one word "hard," and so they keep reaching for the same fix — slower, one more pass.

The two rows below deserve special attention.

The fourth row. This is the most common reason a work of scholarship or ideas won't go in. The author is in the middle of an argument, and the reader doesn't even know that argument exists. What's needed here isn't a study guide — it's thirty minutes of background on the field. Knowing just one sentence about who the author is arguing against can suddenly give the whole book a direction.

The fifth row. This isn't difficulty — it's indifference. And indifference doesn't respond to effort. More on this later.

Read Fast Twice Instead of Slow Once

There's one piece of advice that matters more than any other. A difficult book is usually better read fast twice than slow once.

Of the four levels of reading Adler describes, the second is inspectional reading, and its core rule is simple: the first time through a difficult book, don't stop to wrestle with what you don't understand — just get all the way through. Adler and Van Doren put it as "read it through without ever stopping," and they call it the single most important rule of all.

Why this works follows naturally from the diagnosis that the problem is structural. Very often, what's blocking chapter 3 is sitting in chapter 7. Stopping at chapter 3 to reread it four times is standing around waiting for information that hasn't arrived yet. A first complete pass, on the other hand, means the second reading starts from entirely different conditions — because now you already have the whole map.

Here's how it actually works.

First pass. The goal isn't understanding — it's a map. Mark anything confusing in pencil and keep moving. Don't look things up in the dictionary, don't flip back, don't read the footnotes. Go two or three times your usual speed. A 400-page book gets finished in three or four days.

Between the first and second pass. Let a day or two pass. Try writing, in three sentences, what the book as a whole was trying to do. If you can't, that inability is itself information.

Second pass. Now go back to what you marked. Surprisingly, a lot of it has already resolved itself. Read slowly, holding onto only what's left. This is Adler's third level — analytical reading.

There's a common misunderstanding worth clearing up here. The first pass is not speed reading. The claim behind speed-reading techniques — that training lets you multiply your speed while holding comprehension steady — rests on weak evidence. A large-scale review by Keith Rayner and colleagues, published in 2016 in Psychological Science in the Public Interest (paper), concluded that speed-reading claims are overstated and that there's a real trade-off between speed and accuracy. According to that same review, real gains in reading speed come not from eye-movement technique but from vocabulary and cumulative reading experience.

So the first pass is fast not because of some technique, but because you've agreed to give up on understanding, for now. That's an honest trade. Speed reading promises something for nothing; inspectional reading names the price up front.

When a Study Guide Helps, and When It Becomes a Crutch

Adler's position is clear: outside resources like study guides, summaries, and lectures should come only after you've read the book yourself. Reverse that order and you end up mistaking someone else's reading for your own.

I mostly agree with this, with one exception. The missing context from the table above is better solved in advance. Getting four hundred pages into a book without knowing who the author is arguing against is a waste. But what you need there isn't a guide to that specific book — it's background on the era and the field. The two are different things.

Here's how I draw the line.

When it helps. You skim just the background, in under thirty minutes, before you start reading. You look up one specific concept, and only that one, when you hit a wall mid-book. You compare your own reading against someone else's after you've finished. All three cases share one thing: your own reading either comes first, or isn't interfered with.

When it's a crutch. You read the chapter summary right before reading the chapter itself. You search the moment you hit any snag. You read an explanation, think "oh, that's what it meant," and never go back to the original text. What these three have in common is that they erase the experience of being stuck, entirely.

Why is erasing that experience costly? Learning research has a concept called desirable difficulty — a load that hurts performance in the moment but helps long-term retention. There's a caveat worth being honest about here, though. This concept has mostly been confirmed in memory and learning tasks, and the evidence that it carries over cleanly to reading literature or works of ideas isn't nearly as strong. So there's no need to treat "you have to struggle to grow" as an article of faith. In my experience it's a useful rule of thumb — that's enough.

One practical compromise I'd suggest: don't look anything up until the chapter ends. Save it all for when the chapter is done. That way you keep the experience of being stuck, while avoiding the waste of wandering lost through four hundred pages.

The Notes That Survive After You Close the Book

Underlining diligently while you read, then never looking at those underlines again. That's where most reading notes end up.

There's a study often cited here — John Dunlosky and colleagues' 2013 review of ten learning techniques in Psychological Science in the Public Interest (paper), which rated highlighting and rereading as low utility, and self-testing (retrieval practice) and spaced practice as high utility. The finding that the two techniques students use most are the two that work least became widely known.

But when this research gets carried over into reading advice, two exaggerations tend to come along with it.

First, the claim that "rereading is pointless." The rereading this review examined was cramming — reading the same textbook material repeatedly right before a test. That's a different animal from rereading a novel years later, or from the second pass over a difficult book — done with a map, as described above. This research doesn't reject rereading in general. It only says that mindlessly skimming the same page again doesn't do much for memory.

Second, the expectations around note-taking systems. Elaborate note-taking systems like Zettelkasten and Second Brain have become hugely popular in recent years. But there's essentially no controlled evidence for how much these systems actually improve comprehension of literature or the humanities. The evidence that does exist backs something much humbler: pulling something out of memory without looking. It's retrieval that does the work, not the tool.

So instead of a system, I use just three things.

One, close the book, then write. When a chapter ends, close the book and write its gist in two or three sentences. Write while looking at the book and you're just copying; close it first and it becomes retrieval. That's the entire difference. A badly written summary isn't a bad note — it's an accurate diagnosis.

Two, write down only objections and questions. Skip anything you agree with. You'll remember it anyway. What's worth writing down is things like "I don't buy this part" or "doesn't this claim contradict that other chapter?" Notes like these are still alive years later.

Three, a one-sentence index. Once you finish, write one sentence in the margin of the first page describing what the book actually argues. Underneath that, jot down three or four page numbers worth revisiting later. That's all a person reopening this book three years from now actually needs.

How to Read a Translation Honestly

As long as you're reading in your own language, a large share of difficult books are translations. Not glossing over that fact tends to make for better reading, as it turns out.

When you hit an awkward sentence, hold two possibilities at once. Either the author genuinely wrote it that way, difficult on purpose, or the translation carried the original's structure over directly and it collapsed in the new language. Inexperienced readers always assume the former and blame themselves; cynical readers always assume the latter and blame the book. Both are lazy diagnoses. There is one useful signal, though: if the awkwardness is spread evenly across the whole book, the translation is more likely the culprit; if it clusters in one particular chapter, that chapter's content is more likely genuinely difficult.

Don't build an interpretation on a single sentence. Building an elaborate reading on the fact that "the author chose this exact word here" is risky when you're reading a translation — that word might be the translator's choice, not the author's. If your argument needs sentence-level evidence, check the original; if you can't, the honest move is to hold the claim loosely.

For books that matter, check a second translation. This costs less the shorter the book is. Lay the same passage from two editions side by side, and you can roughly tell which parts are the translator's interpretation and which are the bones of the original. For concrete guidance on which edition to pick, see the translation section of the novels piece.

And when you say you've read it, say you read it in translation. It sounds like a small thing, but it changes your posture. A reader who stays aware that what they accessed wasn't the original tends to hold their own interpretation a little less tightly — and that's usually the right call.

How to Quit Well

This might be the section this piece needs most. How to tell "this book doesn't fit me right now" apart from "I'm just being lazy."

Confuse the two and you lose in both directions. Either you cling to a book that isn't a match and drift away from reading altogether for months, or you treat a book that was one push away from opening up as evidence of your own laziness and close it every single time. Fortunately, there are signals that let you tell the difference.

Signal 1: Are other books going fine? If it's just this one book that won't move while everything else reads fine, it isn't laziness. Laziness doesn't discriminate between books. If nothing is going in right now, on the other hand, that's not a problem with the book — it's the state of your life — and swapping books won't fix it.

Signal 2: Which layer is the resistance coming from? Sit with it for thirty minutes and watch. Are you sliding off at the sentence level, over and over? Or do the sentences all go by fine while your mind is somewhere else? The first is a skill problem, solvable with method. The second is a matter of interest, and no method will solve it.

Signal 3: Why are you trying to read this book in the first place? You need to answer this honestly. Curiosity? Wanting to be able to say you've read it? Not being able to stand the fact that you haven't read it? If the motive is either of the latter two, you'll come out the other side of a full read with almost nothing to show for it. Finishing a book while getting nothing out of it is a real thing that happens.

Signal 4: How many times, and at what intervals, have you tried? If you've tried three separate times, at different points in your life, and stalled in roughly the same place each time, that's not a coincidence — it's a structural mismatch. If you've only tried once, and it was on an exhausted night, you don't have enough data to judge yet.

Signal 5: What changes thirty pages further in? It's only fair to try the methods from the earlier section once before you judge. Did you attempt a first pass? Did you go back to the table of contents and look at the map again? Did you get your thirty minutes of background? If you've done all of that and it still isn't working, that's a well-founded decision to quit.

And when you do quit, keep a record. The book's title, the date, how far you got, one line on why you stopped. Keeping this record gets you two things. First, when you run into the same book again years later, you won't have to flounder from scratch all over again. Second, once the list builds up, you start to see what kind of book you keep stalling on. It was only after looking back at mine that I realized I have an unusually low tolerance for abstract, throat-clearing introductions. Since then, whenever I run into a book like that, I skip the preface and start at chapter one. That alone noticeably raised how often I finish books.

One last thing to add. A book you didn't finish isn't nothing. A book you read to page 120 is different from a book you never opened. Those 120 pages quietly do their work the next time you run into that author, that field, that era. The accounting that only counts finished books as reading is, in fact, inaccurate.

Closing — Knowing Where to Stop Is Harder Than Reading to the End

The ability to handle a difficult book is less about patience and more about judgment.

The judgment to work out whether what's blocking you right now is vocabulary, structure, context, or motivation. The judgment to work out whether what you need right now is one more pass, thirty minutes of background, or an entirely different book. And the judgment to admit, without self-blame, that this book doesn't fit who you are right now. Method is just a tool that supports that judgment.

So don't make your goal a completion rate. Someone with ten books they quit well on tends to get further, in the end, than someone with ten books they forced themselves through. The first person is still finding out what they want; the second is just crossing things off a list.

If you'd like to pick something different to read, Novels Worth the Time is there, and if you'd rather take a break from text for a while, Where to Start with Classic Films is the other option. Both stand on the same principle: a guide that also tells you when it's fine to stop is a guide worth trusting more.