Why Your Typing Isn't Getting Faster — and a Way Out: Over Half Your Time Is Spent in the “Kana Gap”

There is no single cause. In the records, where the time goes and what helps both depend on how fast you already are. The estimates in this article are worked out for the 30–45, 45–60 and 60–80 WPM bands. For what each WPM level means, see Average WPM and Benchmarks.

• Below roughly 30 WPM — the basics stage

  People at or under 20 WPM have a median kana-gap interval of 848 ms, which is very long; many of them may still be converting kana to romaji in their heads (the records can't show what people are thinking). Here, learning romaji and the basics comes first: How to Get Faster at Typing, the romaji input chart, and the Home Position Dojo.

• Around 45–60 WPM — the mid-level plateau (this article's focus)

  The estimate says narrowing the join between kana (the kana gap) pays off more than eliminating mistakes.

• 60 WPM and up — the gap and mistakes both matter

  In the 60–80 WPM band, narrowing the gap takes 65–66 WPM to 76, and zero mistakes takes it to 71.8–71.9 (estimates). The better the typist, the larger the share of mistakes.

When you type Japanese in romaji, the join between two keys is one of four kinds. Take かの (ka-no), typed k, a, n, o. k→a is inside one kana, a→n is the gap between kana, and n→o is inside a kana again. In this article, the “kana gap” means the time from a vowel to the next consonant (the a→n step).

Using Typing Musou's records (June–October 2026), I split the intervals between correctly typed keys into these four kinds.

The gap is only 38.9% of keystrokes but 55.5% of the time. Its average interval is 326 ms, about 2.5 times the interval inside a kana (consonant → vowel, 131 ms on average).

The gap time also includes reading the next word: about 27% of kana gaps fall between two words. What the records can't separate is how much of the rest is picturing the next kana, converting it to romaji, or moving the finger.

So where does the difference between fast and slow typists come from? I split the difference between neighbouring skill bands (8 pairs) by type of join. In all 8 pairs, 49–65% of the difference between bands was the kana gap getting faster.

Between typists at 20 WPM or below and those at 130 WPM or above, the kana-gap interval differs by a factor of about 9. The faster the typist, the less time goes into the joins between kana.

But this is a relationship: fast typists have short gaps. The records alone can't say that shrinking the gap makes you faster.

I worked out “what if this part improved?” for a typical person in each band: split the key intervals by type, shorten the part that improves, and recompute WPM.

For a typist at 45–60 WPM, bringing the gap to the next band's level takes 50 to 58, while zero mistakes gives 54–55. Mistakes are not unimportant, but for a mid-level plateau the gap comes out as the bigger lever.

That doesn't make mistakes meaningless. The better the typist, the larger the share of mistakes; in the 60–80 WPM band, zero mistakes is worth about 6 WPM. One mistake costs about 0.15–0.16 seconds (154 ms and 161 ms from two estimates, comparing within the same person). In Typing Musou a mistake also resets your combo to 0, so it costs damage in battle too.

Also, in Typing Musou battles a wrong key is not accepted and the text doesn't advance. Press the right key and you simply carry on, so there is no deleting and retyping. The Home Position Dojo's stages (except the boss battles and the stages with a speed target) are different: the wrong letter stays in red, and you erase it with BackSpace before retyping. It is practice in noticing a mistake and fixing it with the right finger. For the causes of mistakes and how to reduce them, see how to reduce typing mistakes.

What the records show is that fast typists have short gaps, and that people who keep playing get quicker at reaching the next consonant (see a later section). Which practice actually narrows the gap has not been tested in the records. What follows is the operator's thinking, not the result of a study.

• Read the next kana ahead

  While typing the current kana, read the next kana or the next word with your eyes. Skilled typists have been reported to type with a view of upcoming characters (Salthouse 1984). Typing Musou has a setting, “Preview The Next Prompt”, that shows the next prompt under the current one; it is on by default and appears in battles, Mock Combat, Speed Trial and Accuracy Drill.

• Repeat the same kind of gap

  Gaps come in kinds (after the vowel U, after the vowel I, vowels in a row, and so on). Grouping one kind and repeating it might make it easier for your hands to learn that shape — that is the operator's idea, not a tested result. Stages 142–144 of the Home Position Dojo are built this way.

• Don't stop at the break between words

  About 27% of kana gaps fall between two words. If you start reading the next word only after finishing the current one, you stall there. The aim is to settle the first kana of the next word while still typing the current one. Stages 146 and 147 of the Home Position Dojo practice this.

Here are three things in the game that fit the ideas above. All are free and need no login. You need a PC, or a tablet with a physical keyboard attached; the game cannot be played on a phone.

• Home Position Dojo, Chapter 15 “Read Ahead — Between the Kana” (Stages 141–150)

  The theme of Chapter 15 of the Home Position Dojo is reading the next kana ahead. Stage 141 introduces the idea; Stages 142 and 143 drill the gap after U and after I; Stage 144 covers vowels in a row, like おう and えい; Stages 146 and 147 are short and long sentences about not stopping between words; Stage 150 is a battle with the Shihan (40 WPM). The dojo opens each stage when you clear the one before it, with no skipping ahead, so reaching Chapter 15 means working through Stage 140 first.

• Mock Combat — a CPU a little faster than you

  Mock Combat is a ladder of CPU matches from Stage 1 to Stage 100. The CPU at Stage n types at 5 + 3 × (n − 1) WPM, so each stage is 3 WPM faster than the last. If you are around 50 WPM, Stage 16 (50 WPM), Stage 17 (53 WPM) and Stage 18 (56 WPM) are the “slightly faster” opponents. Winning opens the next stage, and you can replay any stage you have passed. A match lasts 180 seconds; at time-out the player with more HP wins, and a draw counts as a loss. Whether facing a slightly faster opponent narrows your gap is untested.

• Speed Trial — 60 seconds, same conditions for everyone

  In Speed Trial, you have 60 seconds, with the Samurai at ☆0 and no ultimate skills, to deal as much damage as possible to a sandbag. The result screen shows WPM, accuracy, max combo and mistakes, which makes it a good way to see which band you are in. The prompts are random every time, so a single run's WPM will vary. Look at the best and the middle of several runs.

I followed the same people who played 100 or more Random Battle matches (CPU matches included), comparing their first 10 matches with their last stretch.

Speed went up, but the mistake rate hardly changed. Much of the gain came from shorter intervals between keys, especially the time to reach a consonant (12–20% shorter). The kana gap sits inside that.

Among 5,127 people with 40 or more matches, the median gain between the first 10 and the last 10 matches was +6.1 WPM (+15.8%), and 88.6% improved. Looking at the trend for each of 4,909 people with 30 or more matches, 63.4% clearly improved, 2.0% got worse, and the rest could not be called either way.

These are people who kept playing, so quick improvers may be over-represented. It does not mean that anyone who keeps going will see the same. Mistakes are better treated separately from speed; see how to reduce typing mistakes. For how much and how long to practice, see how much daily typing practice you need.

You can't say that touch typing is required, or that you will inevitably hit a ceiling without it. The 30 participants in the study by Feit et al. (CHI 2016) were chosen to span 34–79 WPM. Speed did not differ significantly between people who had taken a touch-typing course and people who had not, and some people who typed with only one or two fingers per hand exceeded 70 WPM. The same study also reports that those who had not taken a course spent significantly more time looking at the keyboard.

In a large English-language survey (Dhakal et al. 2018, about 168,000 people), people with typing-class experience averaged 54.4 WPM and those without 49.0 WPM — a small difference (effect size 0.27). Participants chose to take part and were mostly English speakers in the US, so this can't simply be applied to romaji input in Japanese.

The records in this article don't distinguish whether players look at their hands. For fixing the habit of looking down, see why you can't touch-type and how to fix it; for which finger covers which key, see the home row position diagram.

• Whose records, and when

  I tallied them on 2026-10-18, reading the production database only. They are Typing Musou's records (mistake logs from 2026-05-13; key intervals from 2026-06-14, through 2026-10-18), restricted to players who allowed research use. The key-interval analysis covers about 15,000 players whose speed could be determined. Records of pupils using the game in a school class are not included.

• Who is missing

  Plays at 70% accuracy or below, and matches flagged as suspected cheating, are not recorded. The players struggling most are not in the data.

• Not an average

  These are records of people who played Typing Musou — not the average person in Japan, and not figures by age or occupation.

• Relationship, not cause

  “Fast typists have short gaps” is a relationship, not “shrink the gap and you get faster”. The gains of people who kept playing also come from people who kept playing only.

• Estimates

  They recompute WPM after splitting key intervals by type. They are not a promise that real practice gives the same numbers.

• How WPM is counted

  The same as in Typing Musou: correctly typed keys ÷ 5 ÷ minutes, counting romaji keystrokes rather than kana. Mistyped keys are not counted, and the way you spell a kana (si or shi, for example) changes the number of keystrokes.

• Q. Can I measure my kana-gap speed in the game?

  Not directly: the game has no display that shows kana-gap time (as far as the operator has checked). The result screen shows WPM, accuracy, mistakes, weak keys and so on. The practical way to see whether the gap is shrinking is indirect: whether your WPM rises in the same-conditions Speed Trial.

• Q. My Speed Trial WPM is different every time. Which number should I look at?

  Speed Trial prompts are random every run, and word length changes how easy they are, so each run's number moves up and down. Note the best and the middle of several runs, and compare week to week. How to measure is covered in how much daily typing practice you need.

• Q. What should an advanced typist (80+ WPM) fix?

  The estimates here stop at the 60–80 WPM band; I have not made estimates for 80 WPM and above. In the 60–80 band, both narrowing the gap (76 WPM) and zero mistakes (71.8–71.9 WPM) help, and the better the typist the larger the share of mistakes. For people at 80–100 WPM, the median kana-gap interval was 174 ms.

• Q. I'm stuck around 30 WPM. Does this article apply to me?

  The estimate showed the gap paying off more in the 30–45 WPM band as well, but the focus here is mid-level typists at 45–60 WPM. The operator's view is that for people below roughly 30 WPM, firming up romaji and home-position basics is the shorter road. Start with How to Get Faster at Typing and the romaji input chart.

The outside studies cited in this article, and where the in-house figures come from. The URLs are the papers' DOIs, which lead to the publishers' pages (checked October 18, 2026).

1. Feit, A. M., Weir, D., & Oulasvirta, A. (2016). How We Type: Movement Strategies and Performance in Everyday Typing. CHI 2016. — 30 participants (34–79 WPM). No significant speed difference by course; those without a course looked at the keyboard significantly longer
2. Dhakal, V., Feit, A. M., Kristensson, P. O., & Oulasvirta, A. (2018). Observations on Typing from 136 Million Keystrokes. CHI 2018. — 168,960 participants. With a typing course 54.35 WPM, without 49.00 WPM (d = 0.27). Self-selected; 68% from the US
3. Salthouse, T. A. (1984). Effects of age and skill in typing. Journal of Experimental Psychology: General, 113(3), 345–371. — Typing skill was related to, among other things, how many characters ahead the typist needed to see
4. Typing Musou records, tallied by the operator (18 October 2026, read-only access to the production database) — Only players who allow research use. Scope and limits are in the section on how far to trust these numbers

When you stall, first find out which band you are in with Speed Trial. Below roughly 30 WPM, work on the basics; around 45–60 WPM, look not only at mistakes but at the “kana gap” (the join between kana). From 60 WPM up, the gap and mistakes both matter.

All of this is a reading of the records, not a promise. Measure your speed now, then measure again under the same conditions in a few weeks, and you will see your own progress.