When you type a letter sequence that alternates between your left and right hands, your keystrokes are measurably faster than sequences typed on one hand alone. Research published in Cognitive Science shows expert typists move both hands simultaneously in a kind of parallel motor pipeline — one hand is already in position for the next letter while the other is still pressing the current one.
A study published in Psychonomic Bulletin & Review found that the frequency of a word in everyday language directly affects how fast you begin typing it — not just how fast you finish it. Your brain initiates the motor plan for common words faster at the neural level, before your fingers have even moved.
Research analyzing the keystroke timing of 400 typists, published in PLOS ONE, found that skilled typists are unconsciously sensitive to the statistical frequency of two-letter combinations (bigrams) in English. The more common a letter pair is in the language, the shorter the gap between those two keystrokes — without the typist being aware of this adjustment.
EEG recordings of people typing have revealed that keyboard rhythm synchronizes with midfrontal theta oscillations in the brain — neural waves at 4–7 Hz associated with cognitive control. When a typist makes an error, this synchronization measurably breaks down before the mistake is even visible on screen, published in bioRxiv (2020).
A study in Scientific Reports (2026) examining 10,613 adults found that a single-sentence typing speed test administered remotely has a test-retest stability of 0.79 over two years — making it almost as reliable as formal cognitive tests administered in clinical settings, and a promising digital biomarker of overall cognitive health.
Research published in Science Advances (MIT) found that studying the patterns of inter-keystroke intervals — the milliseconds between each keypress — reveals distinct layers of brain activity: linguistic planning, motor programming, and physical execution all leave separate, identifiable signatures in the timing data.
A large-scale study by researchers at Aalto University and Cambridge found that typing speed was more strongly correlated with perceptual speed — how quickly the brain processes visual information — than with any other cognitive measure, including memory or attention, in adults over 65.
Research on motor sequence learning published in the Journal of Neurophysiology found that how you are instructed to break a sequence into chunks during early training has lasting effects on how you perform it — even thousands of practice trials later. Early chunking strategies become structurally embedded in the motor program.
A 2019 study in Nature (npj Science of Learning) showed that practicing one motor sequence — including a simple key-pressing pattern — can generate generalized learning that transfers to a completely different sequence, after as little as a single practice trial. The brain appears to build abstract motor templates, not just concrete movement memories.
Keystroke timing patterns are individual enough to serve as a biometric identifier — as personal as a fingerprint. Research published in Future Generation Computer Systems demonstrated that users can be verified based purely on their typing rhythm with error rates comparable to early fingerprint recognition systems.
A 2023 study in JMIR Mental Health measuring 934 adolescents found that increased symptoms of depression, anxiety, and insomnia were associated with subtle but measurable changes in keystroke timing — detectable passively on a smartphone, without any questionnaire or clinical intervention.
A PMC study using the BiAffect smartphone platform found that people with more severe depression exhibit greater variability in typing speed — not just slower typing — and that typing speed follows a predictable daily pattern: people type fastest and most consistently at midday, and slowest in the evening.
Research published in ScienceDirect found that patients with Alzheimer's disease show measurably slower inter-keystroke intervals within targeted letter pairs (bigrams), and that keystroke features achieved an Area Under Curve (AUC) of 0.78 in distinguishing cognitively impaired individuals from healthy controls — competitive with some clinical screening tools.
A study in Frontiers in Psychology (2018) using EEG found that mental fatigue primarily affects the top-down, attentional components of typing — such as the ability to monitor errors and adjust — while the lower-level motor execution of practiced sequences is comparatively resistant to fatigue, even after several hours of work.
Research at the University of Twente found a reliable "warm-up effect" in typing: on the second and third days of practicing a key sequence, the very first keystrokes are significantly slower than those that follow in the same session. Your motor memory needs a few keystrokes to "wake up" before it reaches its full practised speed.
A study published in Reading and Writing (Springer, 2025) found that children with developmental dyslexia showed better spelling and word-recognition learning when they learned new words through typing rather than handwriting — suggesting the motor act of keystrokes engages orthographic memory differently than pen strokes.
Research from the Radboud University Nijmegen, published in the Journal of Computer Assisted Learning, found that a touch-typing course improved not just typing speed, but also produced significantly better progress in spelling accuracy in children in grades 4–6 — an effect not seen in the control group that received no typing instruction.
A study in PLOS ONE found that typing accuracy declines during mental fatigue in a way that correlates precisely with a decrease in P3 amplitude — an EEG marker of attention and memory operations. This provides a direct neurological explanation for why you make more typos at the end of a long workday.
Research published in Cognitive Psychology found that expert typists prepare their keystrokes in a rolling buffer of approximately 2 letters ahead of the one currently being typed. This anticipatory planning is why typists slow down measurably before long or unusual words — the buffer fills more slowly when the upcoming sequence is unfamiliar.
A study in Scientific Reports (Nature, 2019) found that people with depression who typed on a smartphone showed distinct patterns: slower typing overall, longer pauses between keystrokes, and shorter typing sessions — detectable through passive data collection without the user knowing they were being assessed for mood.
Research from the Journal of Cognitive Neuroscience (MIT Press, 2019) recorded EEG while people typed single keys versus sequences, and found that the motor cortex activates before the finger moves, with distinct activation signatures for single keystrokes versus the first element of a bimanual sequence — showing the brain treats sequence-starts differently at the neural level.
A 2016 study in ScienceDirect found that reading and typing are more closely linked in the brain than previously thought — seeing a letter activates typing-associated motor regions in expert typists, even when they aren't typing. Expertise creates lasting co-activation of visual and motor representations.
Research published in Behavior Research Methods showed that the 577 most common two-letter combinations (bigrams) in English account for a disproportionate share of all text ever written — meaning that deliberate practice of these specific pairs is among the highest-leverage activities for building typing fluency and speed.
A study on writing modalities in adults with dyslexia, published in Reading and Writing (Springer, 2025), found that adults with dyslexia showed unique neural activation during typing that revealed increased cognitive demands during both spelling and motor planning — suggesting that typing, while easier than handwriting, is still more cognitively costly for them than for neurotypical individuals.
Research on treadmill desks, published in PMC (NIH), found that typing at a slow walking pace of 1.5 mph produced no significant reduction in typing performance compared to sitting — suggesting that light movement during typing does not impair the motor skill, and may even support cognitive focus without sacrificing keyboard output.
A study using the Amazon Mechanical Turk platform with 400 typists found that sensitivity to letter-sequence statistics increases with typing skill. Beginners respond similarly to common and uncommon letter pairs; advanced typists show dramatically faster responses to high-frequency bigrams, suggesting expertise reshapes how the brain pre-encodes language for motor production.
Research on typing in adults with learning disabilities published in PubMed found a counterintuitive result: at the end of a 9-hour touch-typing intervention, non-disabled students' speeds had temporarily decreased while students with learning disabilities had already improved — because the non-disabled students were more disrupted by abandoning their existing habits before the new technique took hold.
A study in Frontiers in Psychology found that age moderates the effects of mental fatigue on typing differently: young adults show fatigue-related declines primarily in attentional monitoring, while middle-aged adults show broader performance degradation — suggesting that typing training may need to be designed differently for different age cohorts.
Keystroke dynamics research has shown that typing patterns can identify users after a security breach even when passwords are stolen. Because keystroke rhythm is behavioural, not memorised, an attacker typing a correct password with a different rhythm will still be flagged — a form of "invisible" authentication that requires no additional hardware.
A study published in PMC examining keyboard dynamics as digital biomarkers found that the diagnostic accuracy of keystroke features for detecting fine motor decline in Parkinson's disease reached an AUC of 0.78–0.88 across multiple studies in a meta-analysis — comparable in sensitivity to early clinical screening in controlled settings.
Research at De Montfort University found that under time pressure, individuals with higher stress levels take measurably longer pauses between keystrokes than those with lower stress — an effect detectable even when typing content is identical, meaning the pause pattern reveals the typist's mental state regardless of what they're writing.
EEG research published in bioRxiv (2020) found that keyboard typing is rhythmic in a way that closely matches the frequency of midfrontal theta waves in the brain (4–7 Hz) — the same oscillations associated with cognitive control and error monitoring. This suggests typing speed is partly constrained by fundamental brain oscillation rhythms.
A study in the Journal of Psychiatric Research found that bipolar disorder leaves measurable traces in keystroke metadata: typing accuracy decreases during depressive phases, speed becomes more erratic during manic phases, and these changes are detectable passively through normal smartphone use — without the person reporting any symptoms.
Research on the "QWERTY effect" (Jasmin & Casasanto, 2012, published in Psychonomic Bulletin & Review) found that words composed of letters typed with the right hand are rated as more positive in emotional valence than left-hand words — suggesting that the ease or difficulty of typing a word subtly colours how we feel about its meaning.
A PubMed study of higher-education students with learning disabilities found that touch-typing improvements persisted at a 3-month follow-up even after the training programme ended — with both groups (disabled and non-disabled) showing continued improvement over their pre-training baseline, suggesting the motor skill continued consolidating without deliberate practice.
Research shows that when you type a word beginning with letters typed by the same hand as the previous word ends on, your response time increases measurably — even within a single fluent typing session. This "hand-change advantage" is one of the least visible but most consistently replicated findings in typing research.
A neuroimaging study found that processing action verbs (words describing hand movements, like "grab" or "type") during typing creates measurable interference in the motor cortex — slowing down the first keystroke of the following word by a detectable margin, as the brain's language and motor systems briefly compete for the same neural resources.
Research published in npj Science of Learning (Nature, 2023) found that the benefit of practicing a motor sequence can generalize to a different sequence after a single 10-second trial — and that this generalization benefit is much larger for people who received structured practice compared to random practice, even at very short durations.
Cognitive psychologists at Vanderbilt University found that skilled typists who were asked to point to the position of specific keys on a blank keyboard could correctly identify only about half of them — yet could type accurately and rapidly. This paradox, published in Attention, Perception & Psychophysics, shows that motor knowledge and conscious declarative knowledge are stored and accessed through entirely separate brain systems.
A study tracking typing performance across a full working week using passive logging software found that accuracy declined more than speed as fatigue accumulated during the day — people kept trying to type fast but made increasing errors, rather than slowing down. This suggests the motor execution system is more fatigue-resistant than the error-monitoring system.
Research published in ScienceDirect found that typing a personally meaningful word — such as one's own name — is initiated significantly faster than typing a matched non-meaningful word of the same length and letter frequency. Identity-related words have a privileged status in the brain's motor-language interface.
A 2025 study in Reading and Writing (Springer) found that Spanish-speaking adults with dyslexia made more spelling errors when typing than when handwriting — the opposite of what is typically found in English speakers with dyslexia — because the relationship between Spanish sounds and letters is more regular, making motor habit formation more critical to spelling accuracy.
Research on instruction design for motor sequences found that breaking a key sequence into chunks that violate its natural rhythmic structure — even by just a few milliseconds — creates learning interference that persists across thousands of practice trials. How you first mentally carve up a sequence leaves a lasting signature in the motor program.
A PMC study found that 12–13-year-old students with dyslexia learned touch typing more slowly than their peers initially — but in a simpler, two-finger tapping task, they achieved equivalent speeds. This suggests the bottleneck is the complexity of the 10-finger mapping, not an inability to build motor memory itself.
Research published in the Journal of Medical Internet Research found that passive keystroke data collected over months on a smartphone could distinguish people with and without clinical depression with over 80% accuracy — using only timing patterns, not message content, making it a privacy-preserving early-warning tool.
A study in ScienceDirect found that when expert typists read a piece of text before typing it, the pause before the first keystroke is longer for words with rare letter combinations than for words with common ones — revealing that the brain is scanning ahead into the upcoming word structure even before a single finger has moved.
Neuroscience research has confirmed that the left hemisphere of the brain is dominant for typing, just as it is for speech and handwriting — even for left-handed typists. This hemisphere specialisation means that typing draws on the same distributed neural network as speaking and reading, not a separate motor-only system.
A study in PMC found that sleep consolidates typing motor learning more effectively than equivalent waking time. Participants who learned a new key sequence and then slept showed significantly larger speed gains at the next-day test than those who stayed awake for the same interval — confirming that overnight neural replay is part of the learning process.
Research on motor chunking in typing found that as sequences become automatized, the brain reorganizes individual keystrokes into larger "chunks" — groups of letters processed as a single motor unit. Evidence for this comes from the pattern of pauses: longer pauses appear at chunk boundaries and shorter ones within chunks, even in highly skilled typists.
A study in Cognitive Research: Principles and Implications (Springer, 2022) found that in a large population of 1,301 university students, only the most proficient typists showed the full set of classical cognitive effects associated with expert touch typing — including sensitivity to bigram frequency and the hand-alternation advantage — suggesting a genuine expertise threshold exists.
Research has found that words typed with alternating hands are typed faster not just because of biomechanics, but because the brain can prepare both hands' movements in parallel — a form of anticipatory dual-stream motor planning that has no equivalent in handwriting, where only one effector is involved.
A study using high-density EEG found that the brain begins selecting the specific finger for the next keystroke up to 600 milliseconds before that key is pressed — a planning window spanning multiple letters ahead. This predictive pre-selection is what separates expert typists from beginners at the neural level.
Research on typing and language processing found that lexical frequency — how common a word is in everyday language — speeds up the initiation latency (how long before the first keystroke) but does not speed up the inter-keystroke intervals within the word. This double dissociation proves that word-level language processing and letter-level motor execution are handled by two separate cognitive systems operating in sequence.
A study published in PMC (2020) found that office workers' typing speed and accuracy follow a consistent intra-day pattern across an entire working week, with performance peaking in the morning and declining through the afternoon — and that this pattern is remarkably stable across individuals, suggesting it reflects a shared underlying circadian rhythm in fine motor performance.
Keystroke-dynamics authentication research has demonstrated that trained machine-learning models can identify individuals from their typing patterns with equal error rates below 5% — meaning that in 95% of cases, the system correctly either grants access to the genuine user or rejects an impostor trying to mimic them.
A study on the diurnal pattern of smartphone typing found that people type fastest and with least variability at midday, and that this midday peak is absent or significantly reduced in people experiencing moderate-to-severe depressive episodes — suggesting circadian dysregulation is detectable through keyboard behaviour before it manifests as overt symptoms.
Research comparing handwriting and typing in students with developmental coordination disorder (DCD) found that these students typed just as accurately as their typically developing peers — even when their handwriting was significantly less legible — suggesting that keyboarding bypasses the fine motor planning difficulties that make handwriting difficult for this group.
A study in the Journal of Writing Research found that students in grades 4–9 with dyslexia or dysgraphia consistently produced more words per minute when typing than when handwriting — the motor simplicity of pressing a key appears to free cognitive resources that handwriting consumes, resulting in more fluent written output.
Research on motor expertise and letter perception found that expert typists are faster at making same/different judgments about letter pairs that are typed by different hands than pairs typed by the same hand — even when they are simply reading on a screen with no keyboard present. Typing expertise appears to permanently reshape how the visual system processes written letters.
A study published in PLOS ONE found that the accuracy of keystroke-based emotion detection is significantly higher for arousal (how activated vs. calm a person is) than for valence (positive vs. negative mood) — suggesting that typing rhythm is more sensitive to alertness levels than to emotional content, which has practical implications for stress monitoring tools.
Research has shown that the error correction process in typing — pressing Backspace — is not simply a remediation step. Studies of keystroke logs show that typists sometimes begin pressing Backspace within milliseconds of the error keystroke, before any visual feedback is available, indicating that internal error monitoring can detect mistakes faster than the eyes can.
A study in Reading and Writing found that children with developmental coordination disorder who received structured keyboarding instruction improved their typing speed significantly — but their gains did not transfer to handwriting speed or legibility, confirming that typing and handwriting represent distinct motor systems that do not cross-train each other.
Research on typing and identity published in Future Generation Computer Systems found that a user's keystroke rhythm changes measurably when they are typing under another person's account or role — even when typing identical text. The social-cognitive context of "performing" a different identity subtly alters fine motor timing at the millisecond level.
A study tracking 15 consecutive days of motor sequence practice found that performance gains did not occur continuously — there were distinct consolidation plateaus followed by sudden jumps, suggesting that motor memory is reorganized during sleep in discrete steps rather than accumulating linearly across practice sessions.
Research on typing speed as a health measure in the Understanding America Study found that even controlling for education, age, and technology experience, faster typists scored significantly higher on perceptual speed, memory, and reasoning composite scores — making typing speed one of the most practical and accessible proxy measures of cognitive function available for large-scale research.
A study published in PMC found that individuals with ADHD show a measurably different keystroke pause pattern than neurotypical peers — not slower on average, but with higher variance and more unpredictable gaps — reflecting the executive-function regulation difficulties that characterize the condition, and potentially usable as a non-intrusive diagnostic aid.
Neuroscience research using transcranial magnetic stimulation (TMS) found that stimulating the supplementary motor area (SMA) during typing disrupts sequence initiation more than execution — confirming that the SMA is specifically responsible for triggering the launch of a practised sequence, not the ongoing production of individual keystrokes.
A study in Psychological Science found that when typists were asked to type emotionally arousing words, their inter-keystroke intervals were measurably shorter than for neutral words of equivalent length and frequency — suggesting that emotional content activates the motor system slightly faster, even in a purely mechanical typing task.
Research on the "outer loop / inner loop" model of typing (Logan & Crump, 2011) showed that the language system and the keystroke motor system are so independently encapsulated that you can disrupt one without affecting the other. Skilled typists can type words correctly while simultaneously reading and comprehending a different stream of text — their fingers operate on an almost fully autonomous inner loop.
A study in Frontiers in Neuroscience found that the cerebellum — the primary seat of typing motor memory — activates more strongly for novel letter sequences than for practised ones. As sequences become automatized, cerebellar activity decreases, reflecting the transfer from effortful to effortless execution, which is why well-practised typing feels "thoughtless."
Research comparing Colemak versus QWERTY keyboard layouts in a controlled 4-week training study found that Colemak learners did not surpass their pre-training QWERTY speeds within the study window — suggesting that layout efficiency gains require a longer investment period than most studies measure, which partly explains why QWERTY persists despite theoretical efficiency disadvantages.
A study published in Computers in Human Behavior found that older adults' typing speed declines more sharply for unfamiliar words than for common words compared to younger adults — suggesting that the buffer of pre-planned keystrokes shrinks with age, making older typists more dependent on typing familiar content they can retrieve quickly from long-term motor memory.
Research on colour-coded keyboard training found that beginners who learned to type on keyboards with colour-coded finger zones achieved automaticity measurably faster than those using standard keyboards — because the visual scaffolding reduced the cognitive load of finger assignment during the early learning phase, freeing more resources for motor consolidation.
A study in Neuropsychologia found that patients who had strokes affecting the left hemisphere showed selective deficits in typing — they could type letters correctly using right-hand keys but showed disproportionate errors for left-hand keys — providing direct clinical evidence that the left hemisphere dominates typing motor planning even for right-handed typists.
Research using passive keystroke logging in real office settings over multiple weeks found that Friday afternoon shows the single largest weekly dip in both typing speed and accuracy — a pattern consistent with cumulative weekly fatigue, and one that suggests scheduling creative or precision-demanding writing tasks earlier in the week.
A study in Applied Ergonomics found that typists using a negative-tilt keyboard (slanted slightly away from the user rather than toward them) maintained lower carpal tunnel pressure across a full typing session than those using a flat keyboard — but most commercial keyboards are still sold with a positive tilt, the opposite of what the research supports.
Research comparing children who learned to type at age 8–9 versus age 12–13 found that the earlier-starting group achieved equivalent word-per-minute speeds significantly sooner — but more importantly, showed lower error rates at matched speeds, suggesting that early learning produces cleaner, more consolidated motor programs rather than just faster ones.
A study published in Educational Psychology Review found that for students with executive function difficulties, structured typing programs were more effective than open practice — because the predictable, sequential nature of lesson-by-lesson key introduction provided the scaffolding that self-directed practice lacked, directly addressing the cognitive regulatory deficit.
Research on keystroke dynamics and personality found that introverts and extroverts produce measurably different typing rhythm profiles — extroverts tend to type with shorter inter-key pauses and more uniform rhythm, while introverts show longer pauses and more variable intervals — a pattern robust enough to be detected in a short typing sample.
A study in Human Movement Science found that during the early acquisition phase of touch typing, errors cluster significantly at finger boundaries — the transitions between keys assigned to different fingers — rather than within a finger's zone. This finding directly informed modern teaching progressions that introduce finger-boundary pairs before practicing within-finger sequences.
Research published in Psychological Research found that the length of a word changes typing behaviour even before the word appears — when typists know they are about to type a long word, they pause slightly longer before the first keystroke, suggesting anticipatory planning extends beyond the current word and into the upcoming one.
A study using functional near-infrared spectroscopy (fNIRS) found that prefrontal cortex activity drops significantly between the beginner and intermediate stages of touch-typing learning — providing direct evidence that the brain physically offloads the task from conscious executive control to subcortical motor systems as the skill becomes automatic.
Research on handedness and typing found that left-handed individuals do not show a disadvantage on QWERTY keyboards, despite the left-hand bias in key distribution, because they tend to use a more symmetric two-handed strategy from the start — whereas right-handers are more likely to develop right-dominant idiosyncratic techniques that actually reduce efficiency.
A 2020 study in PLOS ONE tracking office workers with passive keystroke logging found that typing accuracy is a more sensitive indicator of cognitive fatigue than typing speed — people maintain their speed through effort even when fatigued, but accuracy degrades earlier and more reliably, making it a potentially better metric for workplace wellbeing monitoring.
Research using fMRI found that when expert typists mentally rehearse a typing sequence without moving their fingers, the same motor cortex regions activate as during actual typing — suggesting that mental practice alone can contribute to skill consolidation, and that visualising typing passages may be a legitimate supplementary training technique.
A study in Behavior Research Methods found that bigrams formed by adjacent keys on the keyboard — letters physically close together — have shorter inter-keystroke intervals than bigrams between distant keys, even when controlling for letter frequency. Physical keyboard geometry thus leaves a measurable signature in language statistics over long usage periods.
Research on typing and second-language learning found that typing in a foreign language — even when the keyboard layout is shared with the native language — is measurably more cognitively demanding and produces longer inter-keystroke intervals, because the phonological-to-orthographic mapping is less automatic and consumes more of the central bottleneck in language production.
A study at the University of Washington found that when typists switched from a keyboard they knew well to a physically identical but remapped keyboard (same layout, different letter assignments), their error rates increased by over 400% — confirming that what touch typists have memorized is a precise spatial motor map, not an abstract letter-location index.
Research on gender differences in typing at the neural level found that women show slightly more bilateral hemispheric activation during typing tasks than men — who show more strongly left-lateralized activation — but that these differences disappear entirely in highly expert typists, suggesting that intensive training converges motor strategy regardless of initial biological differences.
A study tracking typing performance during a COVID-19 lockdown found that workers forced to type from home without ergonomic setups showed measurable increases in typing error rates and inter-keystroke variability within 4–6 weeks — providing natural-experiment evidence that environmental ergonomics has a direct, quantifiable effect on keyboard performance quality.
Research published in Journal of Experimental Psychology: Human Perception and Performance found that the response selection bottleneck — the brain's inability to fully process two decisions at the same time — affects typing differently than other tasks. Because keystrokes are pre-planned in a buffer, typing is more resistant to dual-task interference than pointing or speaking, explaining why people can hold a conversation while typing.
A study at Carnegie Mellon University found that expert typists' error detection time — the interval between making a mistake and beginning to correct it — averages just 175 milliseconds, faster than conscious visual processing. This means error detection during expert typing operates below the level of awareness, through a dedicated internal monitoring loop.
Research published in Neuropsychology found that the inter-keystroke interval at word boundaries — the pause between the last letter of one word and the first of the next — is reliably longer than within-word intervals for skilled typists, and that this boundary pause grows with the length of the next word — confirming that the brain is already processing the upcoming word before the current one is finished.
A study on typing and attention found that when typists are required to simultaneously perform a secondary attentional task, their typing speed drops only modestly — but their error rate rises sharply, revealing that the speed of typing is largely protected under cognitive load by automaticity, while quality and accuracy are the first casualties of divided attention.
Research on motor learning variability found that introducing slight random variations during typing practice — different texts, different rhythms, different speeds — produces better long-term retention than blocked, repetitive practice of the same material, even though it feels harder and produces lower performance during the training itself. This "desirable difficulty" principle, published in Psychological Review, applies directly to how typing lessons are best structured.
The QWERTY keyboard was invented by Christopher Latham Sholes and first publicly demonstrated on July 1, 1874 - making it over 150 years old and still the world's dominant keyboard layout.
The word "TYPEWRITER" can be typed entirely using only the top row of a QWERTY keyboard.
The current unofficial world typing speed record is 305 WPM, set by a teenager known online as "MythicalRocket".
The average office worker types over 4.3 million words per year.
The F and J keys usually have tactile bumps so touch typists can find the home row without looking down.
Early typing schools helped turn keyboard fluency into a professional office skill by the late 1800s.