The Sentence Note-Taking Method: Capture, Then Convert

The sentence method is a note-taking system in which every new fact or idea goes on its own numbered line, written as a full sentence. It is the best format there is for capturing a fast lecture and the worst one for revising from, so it only pays off if you convert the page afterwards.

That second sitting is the subject of this guide. Below is what the method looks like on the page, the four marks that decide whether the page can be converted at all, the conversion itself, and an honest read of the research people cite when they warn you off writing in full sentences.

What is the sentence note-taking method?

The sentence method is the least structured of the classic systems. You write each new fact, idea, or topic on a fresh line as a short whole sentence, and you number the lines as you go. There is nothing to lay out first and no format to maintain while you write, which is where the speed comes from. The University of York library guide to the sentence method describes an approach in which “every new thought, fact, or topic is written on a separate line”, often numbered, and that is genuinely the whole specification.

It is worth being precise about which problem this solves, because the folk version is slightly off. The usual claim is that the sentence method suits lectures you cannot predict. The sharper claim is that it suits lectures you cannot keep up with. The same York guide calls it “Ideal for lectures that are organised but rapid”, which is exactly right: a structured layout fails you when material arrives faster than you can decide where each point belongs, and a numbered line asks you to make no such decision. Pace is the axis this method solves. Structure is the axis it defers, and deferring is a long way from avoiding. Our overview of note-taking methods sets the six classic systems side by side, including the ones that ask you to settle structure while you write.

Four marks to make while the lecture is still running

Everything that makes a sentence page convertible has to happen live, because none of it can be reconstructed later from a wall of prose. Four marks, all cheap enough to make at speaking pace.

• Number every line, including the ones you abandon. The number is an address, and an address is what lets you write “7, 12, 19” beside a heading later instead of physically shuffling text around. Skipping the numbers on lines you dislike breaks the addressing for everything after them.
• One idea per line, even when it costs you a line. A line holding two facts cannot be filed under one heading, so it survives the conversion as an orphan you have to split under time pressure. When a sentence needs an “and” that joins two separate claims, break it and take the next number.
• Leave an unfinished line unfinished, and mark it. The reflex is to complete a half-heard sentence from memory, and it is the most damaging thing you can do on this page, because your reconstruction will read exactly like something you were told. Put the line down as far as you got, mark it, move on. A visible gap is recoverable from a friend or the slides; an invented ending is not.
• Mark uncertainty in the moment, with two symbols. Two different things go wrong and they need different fixes. Use one symbol for “these are his words, not mine” and another for “I wrote this down and did not follow it”. The first tells you which lines need rephrasing during the conversion; the second tells you which need the textbook. Deciding this afterwards is guesswork, because every line looks equally understood in your own handwriting.

Why numbered lines are the hardest page to revise from

A finished sentence page has exactly one organising principle: the order in which the lecturer said things. That is a real structure, and it is the wrong one. Delivery order is shaped by what the speaker wanted to build up to, which topics needed a story before they made sense, and how long the room took to settle. The structure of the subject, the one an exam will ask about, sits underneath all of that, unmarked.

This is what makes the page feel unusable a week later, and the York guide names the cost directly: “Reviewing notes can be cumbersome and inefficient, as the lack of visible relationships between notes hampers the identification of main concepts versus minor details”. A Rutgers Learning Center handout on note-taking systems lists the same two failures for this method: you cannot separate major from minor points using the numbered sequence, and editing means rewriting to cluster the points that belong together. Both descriptions circle one absence. The page records what was said and nothing about what mattered, and there is nowhere on it for that second kind of information to live.

Which makes the sentence method the only common system whose output is genuinely raw. A Cornell page or an indented outline is already partly revised the moment you stop writing, because you spent attention during the lecture deciding where things went. A sentence page has none of that attention spent yet. Calling the result worse notes misses what is going on: they are unfinished notes, and finishing them is a scheduled task rather than an optional tidy-up.

How do you convert sentence notes into something you can revise?

Four moves, and the first three happen on the original page in a different colour rather than on a fresh sheet. Nothing gets copied out until you know what the structure is.

Do it the same day if you can, and within two days at the outside. The reason is specific rather than motivational: the group letters come from your memory of the lecture rather than from the page in front of you, and that memory is the part that decays. Marking uncertainty at the time buys you a few days of margin. It does not buy you a fortnight.

Keep this separate from plain recopying, which is a different act with a worse reputation. Copying a page out again preserves the order and the wording, which are precisely the two properties that made the original hard to revise from, so the effort buys neatness and little else. Our piece on whether rewriting notes is effective covers the evidence there. A conversion changes the order, compresses the wording, and discards lines. If your rewritten page runs to the same length as the original, you transcribed instead of converting.

What you convert into depends on what the lines turn out to be, which is something you only discover once they are grouped.

The second row is the Cornell method arriving late. When your lines are mostly facts, the conversion amounts to writing the cue column you had no room for during the lecture, and our guide on how to take Cornell notes covers what to do with that column once it exists. When a term of notes turns out to be a mixture of all three shapes, the wider problem is filing rather than format, which our guide to organising your notes takes on.

Is your page convertible at all?

Not every sentence page can be converted, and finding out costs about four minutes. The value of checking is that a page which fails these is telling you to fix something in the next lecture rather than to try harder on this one.

1. The line test. Pick five numbered lines at random and read each one on its own. If you cannot say what a line is about without reading its neighbours, the lines were never self-contained, and what you have is a transcript broken into pieces. The fix is one idea per line next time.
2. The grouping test. Try to put a letter beside every line inside five minutes. If a large share of them refuse to belong anywhere, the lecture had a structure you did not hear, and no amount of rearranging will invent it. The fix is to get the slides or the recording before you convert.
3. The question test. Take each group and write one question you could plausibly be asked about it. A group that yields no question was a digression, and the honest move is to cut it rather than promote it to a heading.

Does writing in full sentences hurt what you remember?

Probably less than you have been told, and the evidence is more contested than the headlines suggest. The worry is legitimate, though, and it has a specific source. In Mueller and Oppenheimer (2014), students who took notes on laptops did worse on conceptual questions than students writing longhand across three studies, and the authors attributed the gap to a tendency to transcribe lectures verbatim instead of processing the information and reframing it in their own words. One of those three is the one to know about here, because it tested the very thing this guide recommends. It moved the test a week out and let people study their notes first, and the write-up of that condition reports that "When participants were given an opportunity to study with their notes before the final assessment, once again those who took longhand notes outperformed laptop participants." The sentence method is the one system that openly invites verbatim capture, because a full sentence written at speaking pace is very often the sentence the lecturer just said.

The follow-up work complicates it. A direct replication led by Heather Urry and colleagues at Tufts (2021) reproduced the note-taking difference without reproducing the learning difference. With 74 laptop and 68 longhand participants, laptop notes again carried more verbatim overlap and more words, yet laptop participants did not perform better than longhand ones on the quiz, and the authors concluded that their results do not support the idea that longhand note taking improves immediate learning through better encoding. Exploratory meta-analyses they ran across eight similar studies echoed that pattern. The one correlation that would most directly indict this method, more verbatim overlap going with worse quiz scores, did show up in both the original study and the replication, but the authors report that it was not robust in their own data.

One design detail is worth knowing about the replication specifically. In that replication, participants watched the lecture, were briefly distracted, and sat the quiz with no opportunity to study in between, which makes it a clean test of capture alone and says nothing either way about a later conversion. The 2014 work is the one that did test a review pass, and that is the finding to take seriously here, because verbatim-heavy notes still lost even when their owners were allowed to study them first. Rereading a wall of transcription does not rescue it. That is an argument for making your conversion a real rewrite rather than a reread, which is the same test set out further up this page: if your rewritten page runs to the same length as the original, you transcribed instead of converting. One direct replication plus a set of exploratory meta-analyses is reason to hold the original claim loosely, and no reason at all to reverse it. For the medium argument itself, pen against keyboard, our comparison of handwriting versus typing notes takes it on directly.

Testing whether the rewrite actually worked

The conversion leaves you holding a question it cannot answer for you. Your rewritten page is now the most familiar document you own: you chose its headings, you decided which line was a claim and which were support, and rereading it will feel like knowing the material whether or not you do. Fluency with your own structure is not evidence of recall, and a page you built by hand is unusually good at producing that feeling.

GeniusPal sits after that step and nowhere near it. It reorganises nothing, restructures nothing, and tidies nothing you hand it; there is no outliner anywhere in it, and the grouping, ranking, and rewriting described above stay entirely your work. Its one job is to read the converted page and write questions whose answers are all somewhere on it, which only makes sense once that page is finished. Upload the raw numbered lines instead and the questions come back shaped like the transcript, which is the same problem one step further along.

A new GeniusPal account comes with two study-set generations. That pair is everything it receives without payment, nothing renews it later, and the quiz is the only format either of them runs in. GeniusPal puts written active recall and flashcards behind the Student plan, where the two open together at 14.99 dollars per month with one hundred generations in the allowance. Genius bills once a year at 59.99 dollars, where the word unlimited stands in for a ceiling high enough that the fair-use policy has to name it.

The method asks almost nothing of you while the lecture is running, which is precisely why it asks so much afterwards. Number the lines, group them the same day, rewrite them into a shape that has a hierarchy in it, and then find out whether the shape went in. Everything the sentence method is worth happens in that second sitting.