Annotation strategies that help you remember what you read

To annotate a book for later recall, underline selectively, add a note in your own words, and turn an important passage into a question you can answer without looking. John Dunlosky and colleagues' 2013 review rated highlighting and underlining as low-utility learning techniques, with practice testing rated high. Use a mark to begin a task, then make time to attempt the answer. Dunlosky et al., 2013.

Start with one passage you want to remember. Identify its main claim, write a question about the reasoning, then revisit that question with the passage covered. The annotation strategies below are suggested applications of generation and testing research. They are not a tested formula for how many marks to make on a page.

Choose a method according to the work it should do

Use the table as a set of options for your next reading session. Its recommendations are practical tasks, not a ranking established by a head-to-head experiment on these four annotation methods.

Method	What to use it for	Suggested next action
Underline	Locate a definition, claim, or sentence to quote	Add a reason for selecting it
Margin note	Record an interpretation or a connection	Check the note against the surrounding passage
Question	Prepare a prompt for explanation or recall	Cover the text and attempt an answer
Summary	State the scope of a section in your own words	Write from memory, then correct it against the source

Dunlosky and colleagues' Improving Students' Learning With Effective Learning Techniques (2013) found that highlighting often failed to improve learning reliably. They also rated summarization low overall, while acknowledging benefits under some conditions. A summary is therefore no automatic guarantee of remembering. The closed-book step in the table deliberately adds a recall task. Dunlosky et al., 2013.

Before marking a sentence, decide what you'll do with it. If you want to quote its wording, underline it and record the page. If you want to explain its reasoning, write the question your explanation must answer. If neither purpose matters to you, give yourself permission to leave the passage unmarked.

Generate a response instead of only copying

Norman Slamecka and Peter Graf's The Generation Effect: Delineation of a Phenomenon (1978) found better memory for words participants generated than for words they read. Those experiments concerned word-learning tasks, rather than annotations in whole books. Treat applying that result to margin notes as a reasoned suggestion, not a direct experimental finding about marginalia. Slamecka and Graf, 1978, with Slamecka's retrospective account.

Try a note that requires you to contribute an explanation. Use a short prompt such as “depends on,” “example of,” or “conflicts with,” then finish the thought. Keep the original passage beside the note while you check whether your interpretation follows from it.

Consider a hypothetical essay that argues a library should extend its opening hours. An underline might select the recommendation. A generated note could say, “The proposal assumes people who cannot visit now would come in the evening.” Add a question asking where the essay supports that assumption. This is an invented example for practice, not a claim about library use.

Avoid changing the author's claim while simplifying it. If the passage says the change could help some visitors, preserve both qualifications in your note. After writing, compare the strength of your verbs and the scope of your nouns with the source. Mark your own inference as an inference.

Use marginalia to keep a conversation tied to the page

AIReader is designed around the same practice: mark claims, evidence, assumptions, and special uses of terms, and put questions in the margins. It also asks readers to check whether proposed connections hold. That is the basis for the marginalia routine here: place your response next to the passage that prompted it.

For a first attempt, use a small set of marks. Circle a term you need to examine, underline a claim, and put a question beside a step you cannot follow. Write a brief explanation alongside any symbol whose meaning might be ambiguous. Treat the system as something to revise after using it.

Make disagreements specific enough to investigate. In the hypothetical library essay, replace “weak argument” with “Does the evidence distinguish preference for later hours from ability to attend?” Leave space to record an answer if a later paragraph supplies one. Let the note document a question in progress.

For fiction, try adapting the task to narration and language. Mark a shift in who appears to know something, or a repeated description whose meaning you want to compare. Ask what the wording supports before assigning a motive to a character. Use these as optional reading prompts, without turning every scene into an exercise.

If you cannot write in the book, put the note in a notebook with the edition and page number. For a digital text without stable pages, record the section heading and a short identifying phrase. Keep exact quotations visibly separate from your paraphrases so you can check the wording before quoting later.

Turn selected annotations into recall prompts

Henry Roediger and Jeffrey Karpicke's Test-Enhanced Learning (2006) found that recalling prose passages improved delayed retention compared with restudying them. This is the testing effect: retrieval can be a learning activity as well as an assessment. Roediger and Karpicke, 2006.

To apply that finding, choose a question from your margin at the end of a section. Cover the passage and write an answer without checking. Then reopen the book and mark what you omitted, overstated, or misattributed. Keep that correction next to the attempted answer.

For the library example, a recall prompt could ask which evidence the author offered for extending the hours. A separate interpretation prompt could ask whether that evidence supports the proposed change. Attempt both, but check them differently: verify the first against the passage, then defend the second with reasons.

Return to a few selected questions in a later session. If you want a suggested starting point, try the next day, then choose another occasion according to what remains difficult to explain. This is a routine to adjust, rather than an interval established by the cited experiment.

Keep the system small enough to use

Try setting a modest annotation budget for a chapter: a central claim, a supporting passage, and an unresolved question. This is a suggested constraint, not an evidence-based quota. Add more only when you can name what you intend to do with the extra notes.

During review, assess your annotations by attempting their tasks. Explain the claim, locate its support, and answer the question. Revise notes that are too vague to use. If an annotation only records that you liked a sentence, keep it for that purpose without treating it as a memory exercise.

For your next reading session, turn one existing highlight into a question and answer it with the page covered.