Typing vs Handwriting Notes: Which Is Better for Learning?

If you have ever looked up whether to type or handwrite your notes, you have met the same study. A 2014 paper called The Pen Is Mightier Than the Keyboard reported that students who took notes by hand outperformed students who typed, and the finding travelled a long way. It is quoted in study-skills guides, in university orientation materials, in productivity writing, and by a great many people who have never read it. Note-taking is not the only place where typing under divided attention gets discussed with more confidence than evidence — what dispatch typing actually demands is another.

It has since been directly replicated twice. Neither replication found the effect.

That is not a gotcha, and it does not mean handwriting is useless or that laptops are fine. It means the popular version of this question — which medium is better for learning — has an answer most people would find disappointing, and that the genuinely useful findings are buried underneath it. This article works through what the research actually established, what failed, what separately turns out to be true about laptops in classrooms, and where typing speed fits into any of it.

What the original study actually found

Mueller and Oppenheimer ran three studies in which students watched lectures and took notes either longhand or on laptops, then answered factual and conceptual questions. Laptop note-takers did worse on the conceptual questions. The original paper is indexed on PubMed if you want the abstract.

The mechanism they proposed matters more than the headline, and it is the part that has held up best. Laptop users, they argued, tend to transcribe the lecture close to verbatim because they can type fast enough to nearly keep up. Longhand writers cannot keep up, so they are forced to select, condense and reframe — and that reframing is itself the learning. They called this shallower processing on the laptop side.

Note the structure of that claim. It has two parts: a behavior (verbatim transcription) that is bad for learning, and a medium (laptops) that supposedly causes it. The first part is well supported. The second is the part that did not survive.

Then it did not replicate. Twice.

In 2019, Morehead, Dunlosky and Rawson ran a direct replication and extended it, adding a group using eWriters and — cleverly — a group that took no notes at all. Their replication and extension is indexed on ERIC. Some trends leaned toward longhand, but performance did not consistently differ between any of the groups.

Read that last clause again, because it is the more alarming result. It was not merely that laptops matched longhand. The participants who took no notes whatsoever performed comparably too. When students were allowed to review their notes, the differences shrank further. The authors concluded that deciding which method is superior looks premature.

In 2021, Urry and colleagues published a second direct replication of the original Study 1, run with 145 undergraduates, along with a mini meta-analysis across eight similar studies. The full paper is available as a PDF, and the Association for Psychological Science wrote it up under the headline that the laptop does not need ditching yet. They found only small, statistically non-significant differences in quiz performance. The meta-analysis agreed: the effect of typing versus handwriting was not statistically significant.

None of this is unusual. A great many well-known psychology findings have had a harder time under direct replication than their popular reputations suggest, and the broader story is documented under the heading of the replication crisis. What is unusual here is how completely the original claim has outrun its evidence in public. The 2019 and 2021 results are barely known outside the field, while the 2014 result is still repeated as settled fact. Teaching-focused writeups such as the Oregon State summary of the replication failure, the British Psychological Society research digest and The Learning Scientists have tried to correct the record with limited success.

The finding that explains the whole mess

Buried in the 2021 replication is the most useful result in this entire literature, and almost nobody quotes it.

Urry and colleagues measured two things about the notes themselves rather than only the medium. They counted how many words each student wrote, and they measured how much of it overlapped word for word with what the lecturer said. Then they checked each against quiz performance.

More words predicted better performance. More verbatim overlap predicted worse performance. Those point in opposite directions, and typing increases both at once.

That is why the medium comparison keeps coming out null. Typing hands you a benefit and a cost simultaneously, they are roughly the same size, and they cancel. Averaging across a room full of students, half of whom are transcribing and half of whom are summarizing, produces a difference of approximately zero — which is exactly what two replications and a meta-analysis found.

It also means the question everyone asks is the wrong one. Typing versus handwriting is not the variable. Processing versus transcribing is the variable, and it is not determined by your device.

Where typing speed actually enters

Here is the part that is specific to this site, and it is a genuine trade-off rather than a sales pitch for typing faster.

Speech runs at roughly 120 to 150 words per minute in ordinary lecturing. Handwriting cannot come close to that. Typing can get close, and for a fast typist it can effectively match it. That capability is precisely what makes verbatim transcription possible, and verbatim transcription is the behavior on the harmful side of the ledger.

So the honest statement is uncomfortable: your typing speed does not make your notes better or worse, but it does change which failure mode is available to you. A 35 words-per-minute typist physically cannot transcribe a lecture and is forced into selection, the same way a handwriter is. A 90 words-per-minute typist can very nearly keep up, and therefore has to choose not to. If you want to know where you sit, what counts as an average typing speed has the distribution, and you can take a timed test at three minutes or longer for a figure worth trusting.

This is a specific instance of something that recurs across this subject: a higher number is not automatically an improvement, and the benefits of typing speed have a shape rather than a slope. Where it genuinely pays and where it stops paying is worked through in where typing speed still matters and where it does not.

There is a neat historical illustration of the endpoint. Stenographers can capture speech verbatim in real time, which is exactly the capability being discussed here — and the profession exists specifically because verbatim capture is valuable when you need a record rather than an understanding. How stenographers reach 300 words per minute is the extreme case of transcription as a goal, and courts want a transcript, not a student who learned something.

The laptop finding that is real

All of the above is about note-taking. A separate and much stronger body of evidence is about having a computer open in a classroom, and it points the other way.

Carter, Greenberg and Walker ran a randomized controlled trial at West Point across 50 classrooms and 726 students, randomly assigning class sections to technology-free, unrestricted computer use, or tablets-only-flat-on-the-desk. The full paper is available as a PDF, and Education Next covered it under the question of whether professors should ban laptops.

Technology-free classrooms averaged 72.9 percent on the final exam. Unrestricted classrooms averaged 70.5 percent. That is a gap of 2.4 percentage points, or 0.18 standard deviations, from a randomized design rather than an observational one — which makes it considerably better evidence than most of what circulates on this topic.

Two details make it more interesting. The tablet-only condition scored 71.4 percent, a 0.17 standard deviation drop, essentially as bad as unrestricted laptops — even though only about 40 percent of students in those classrooms actually used a device, against 80 percent in the unrestricted ones. And the authors found suggestive evidence that the harm fell more heavily on students with higher entrance-exam scores, not less.

Why that does not contradict the replications

These two literatures get mashed together constantly, and keeping them apart is most of what it takes to think clearly here.

The note-taking studies asked: given that you are taking notes and paying attention, does the medium change what you retain? Answer, on current evidence: not reliably.

The classroom studies asked: what happens when a connected device is open in front of you for fifty minutes? Answer: something measurably bad. The most plausible mechanism is not encoding at all. It is that the device offers a continuous supply of more appealing things than the lecture, and attention leaks toward them.

Sana, Weston and Cepeda demonstrated the sharper version of this, and their study on laptop multitasking is indexed on ERIC. Students multitasking on a laptop during a lecture scored worse — and so did students merely seated in direct view of a multitasking peer. The distraction is contagious, which is a finding with no equivalent whatsoever in the handwriting-versus-typing debate and a much stronger case for classroom policy.

The brain-scan study, and what it actually measured

In 2024 a study appeared that seemed to settle things in handwriting's favor, and it produced a wave of coverage. Van der Weel and Van der Meer recorded high-density EEG from 36 university students and reported that handwriting but not typewriting leads to widespread brain connectivity. The press release framed it as writing by hand boosting brain connectivity, and it was widely reported as evidence that handwriting is better for learning.

The connectivity finding is real. What deserves attention is the comparison that produced it, and this is where a site about typing has something specific to contribute.

In the handwriting condition, participants wrote words in cursive with a digital pen directly on a touchscreen. In the typewriting condition, participants were instructed to type the presented word using the right index finger only — and the typed words did not appear on the screen while they typed.

So the study compared skilled cursive handwriting, with visual feedback, against single-finger key pressing, without visual feedback. Both choices were deliberate and defensible for the authors' research question, which was about motor and sensory activation rather than about classroom practice; restricting to one hand avoids cross-hemisphere contamination in the EEG. But it is not a comparison between handwriting and typing as anyone actually does either.

Touch typing is a ten-finger, bimanual, rhythmically chunked motor skill, which is a completely different thing from hunting for keys with one index finger — the difference is roughly the whole subject of how touch typing is learned, and the two produce different speeds, different error patterns and almost certainly different neural signatures. Whether skilled touch typing generates the connectivity patterns handwriting does is an interesting open question. This study did not ask it.

I am not the only one with reservations. A 2025 commentary by Pinet and Longcamp, published in the same journal, raises several methodological caveats about the mechanistic inferences drawn from the work. Treat the study as evidence that the two motor acts differ, which is not in doubt, rather than as evidence that one teaches you more.

How fast can you actually write by hand?

The speed gap is the engine of the original theory, so it is worth getting right, and it is an area where the numbers circulating online are unusually bad. Most of the figures you will find on blogs cite nothing at all.

The credible material sits in occupational therapy assessment literature, where handwriting speed is measured clinically. Instruments such as the Handwriting Assessment Battery for Adults exist precisely because handwriting speed needs standardized norms. Broadly, functional adult handwriting lands somewhere around 20 words per minute, but the figure swings substantially depending on whether the writer is copying or composing, on age, and on how legible the output has to be. Anyone quoting a single precise number for adult handwriting speed is over-claiming, and in keeping with our editorial standards we would rather say the range is wide than repeat a tidy figure with nothing behind it.

What is not in dispute is the ordering. Speech is far faster than handwriting. Typing sits in between and, for a practiced typist, close to speech — the large-scale figures on that are in the Aalto study of 136 million keystrokes, and the arithmetic behind converting any of these numbers is in what WPM actually measures.

The practical implication is the one the original researchers got right even if their headline did not replicate. Handwriting imposes selection because it has no choice. Typing makes selection optional. Optional is the problem.

What notes are actually for

Note-taking research has long distinguished two functions, and they pull in different directions. The encoding function is the benefit you get from the act of writing itself — deciding what matters, compressing it, putting it in your own words. The external storage function is the benefit you get later from having a record to review. The general framing is summarized under note-taking.

Verbatim transcription maximizes storage and minimizes encoding. Sparse paraphrase maximizes encoding and may leave you with a record too thin to revise from. Neither extreme is what you want, and the Urry finding is exactly this trade-off showing up in the data: word count helped because storage helped, and verbatim overlap hurt because it meant encoding had been skipped.

The 2019 finding that people who took no notes at all did comparably well is worth sitting with too. It suggests that in a single-session lab study with an immediate test, notes matter less than everyone assumes. Their value shows up over weeks, through review — which is why the intervention with the best evidence in this whole area is not a note-taking medium but retrieval practice: testing yourself on the material rather than rereading it.

Method beats medium

If the medium is not the variable, the obvious question is what is. The answer is method, and methods port across devices better than people expect.

The Cornell system is the best-known structured approach, and its design directly targets the failure mode identified in 2014: a narrow cue column that forces you to generate questions, and a summary line that forces you to compress. Nothing about that requires paper. It works in any text editor with a table or two panes.

What matters is that a method imposes a constraint your typing speed has removed. Handwriting supplies that constraint through physics. On a keyboard you have to supply it deliberately, which is harder but also more flexible, because you can choose a constraint that fits the material instead of accepting the one your hand imposes.

A protocol for typing notes without transcribing

This is the practical core. It is built to reproduce the encoding benefit that handwriting gets for free, while keeping the storage benefit that typing is genuinely better at.

The discipline this requires is closer to editing than to stenography, and the mental shift is the same one described in typing fewer keystrokes rather than faster ones: the goal is not throughput, it is getting the right things down.

When handwriting genuinely wins

None of this makes handwriting pointless. There are situations where it is clearly the better tool, and they are not the ones usually cited.

• Anything spatial. Diagrams, molecular structures, circuit sketches, mathematical working, annotated graphs. Notation that is not linear text is slow and awkward to type and fast to draw, and this is the strongest and least controversial case.
• Environments where a screen is inappropriate or distracting to others — a small seminar, a client meeting, a one-to-one conversation where a laptop puts a wall between you and the person talking.
• When you know you cannot resist the device. The randomized evidence on classroom computers is real, and a paper notebook is a commitment device. Choosing paper for self-control is a good reason, and it is a different reason from the one everybody quotes.
• When the act of writing is the point rather than the record — planning, thinking a problem through, drafting. Many people report handwriting suits this better, and the research is not settled enough to argue with a preference.

When typing genuinely wins

And the reverse, with the same caveat that these are practical advantages rather than learning claims.

• Volume you will actually revisit. Searchable, greppable, syncable notes are worth a great deal over a semester, and the storage half of note-taking is where typing is straightforwardly better.
• Anything you will restructure later. Reordering, merging and rewriting are near-free in a text editor and expensive on paper.
• Accessibility. For anyone whose handwriting is slow, painful or illegible, the medium debate is settled by physiology rather than by psychology — a point that also applies to dictation as an input method.
• Fast-moving technical material, where accuracy of symbols and terms matters and legibility later matters more than the feel now. Typing symbol-dense material has its own difficulties, which is the subject of why code is harder to type than prose.

What about tablets, styluses and note apps?

The obvious middle path is handwriting on a screen, and the honest answer is that the evidence is thin. The 2019 replication included an eWriter group specifically to test this, and it did not differ consistently from the others — which is unsurprising given that nothing differed consistently from anything.

What can be said is that a stylus keeps the speed constraint that forces selection while adding searchability and reordering. If the mechanism in the original theory is the real one, that combination should be good. Nobody has demonstrated that it is, and you should be suspicious of any product page telling you otherwise. The pattern of confident claims resting on nothing is one that recurs throughout this subject, catalogued in the typing myths that refuse to die.

For teachers, and for parents

The policy question is genuinely different from the personal one, because the classroom evidence is stronger than the note-taking evidence and points more clearly.

A blanket claim that students learn better writing by hand is not supported. A claim that unrestricted laptops in a lecture hall lower average exam performance is supported, by a randomized trial, and the tablet-only result suggests half-measures do not help much. Those two statements are compatible, and conflating them is how the debate has gone wrong.

For younger students the calculus differs again, because keyboarding is itself a skill being acquired. A child who types at fifteen words per minute is not choosing between media in any meaningful sense — the keyboard is a bottleneck, not an option. What to expect at each stage is in typing speed by age and grade, and the sequencing question is covered in how to teach a child to type.

How to tell whether your notes are working

Rather than choosing a medium on faith, test it. The measurement is cruder than a lab study and considerably more relevant to you.

After a session, close the notes and write down everything you can recall, unaided. Then open them and check. What you produced from memory is the encoding half; what you missed but had recorded is the storage half. If your recall is thin but your notes are complete, you transcribed. If your recall is decent but your notes are unusable a week later, you under-captured.

Run that once with each medium if you like. It will tell you more about your own habits than the entire literature will, largely because the literature keeps finding that the average effect is zero — which is another way of saying it depends on the person.

If typing speed turns out to be your bottleneck rather than your temptation — if you genuinely cannot get enough down — that is a fixable and separate problem, and a practice routine that moves the number is where to start. If your speed has been flat despite practice, why progress stalls covers the mechanism. And if you are unsure what the site's own numbers mean, the tutorial documents all of it.

The honest answer

Type or handwrite, whichever you will actually do consistently. The evidence does not support a reliable advantage for either, and two direct replications plus a meta-analysis is a considerably stronger basis for that conclusion than one widely-quoted paper is for the opposite.

What the evidence does support is narrower and more actionable. Copying words you have not processed is bad for learning, on any device. Having a connected screen open in a lecture is bad for learning, and that finding comes from a randomized trial rather than a lab analogue. Reviewing and self-testing beats both media and every note-taking format anyone has compared.

The reason the medium question persists is that it is easy to answer and easy to act on: buy a notebook, feel virtuous, change nothing about how you listen. The real answer asks more of you, which is to say something about why it lost the popularity contest to a study that did not replicate.

And if you type, the thing to watch is not your speed. It is whether the words arriving on the screen went through you on the way.