A Japanese Bimanual Flick Keyboard for Tablets That Improves

Display Space Efficiency

Yuya Nakamura

and Hiroshi Hosobe

Graduate School of Computer and Information Sciences, Hosei University, Tokyo, Japan

Faculty of Computer and Information Sciences, Hosei University, Tokyo, Japan

Keywords: Text Entry, Touch Screen, Touch Typing.

Abstract: Tablets, as well as smartphones and personal computers, are popular as Internet clients. Tablet users often use

QWERTY software keyboards to enter text. Such a software keyboard usually uses large display space, and

requires its user to largely move their fingers. This paper proposes a Japanese bimanual flick keyboard for

tablets that improves display space efficiency by using 10 character keys. The paper presents an

implementation of the keyboard for an Android tablet, and describes an experiment on its performance

compared with a QWERTY software keyboard. Since the results of a preliminary experiment indicated a

problem with the key layout, the main experiment further introduced an L-shaped layout and a Γ-shaped

layout for comparison. The main experiment examined the keyboard’s input speed, accuracy, and subjective

evaluation, and the results showed trade-offs among these layouts.

1 INTRODUCTION

Tablet users often use QWERTY software keyboards

that occupy large display space. A split keyboard is a

software keyboard that improves display space

efficiency. By splitting a QWERTY keyboard, it

allows its user to put both hands at natural positions.

The improvement of display space efficiency is good

for multitasking.

In Japan, there are about as many users of “flick”

keyboards as QWERTY keyboard users. Flick is a

gesture operation especially used for character input

on touch-screen devices. Flick input usually makes its

user to touch a key with a finger and then slide the

finger upward, downward, to the left, or to the right

for input. Since many flick keyboard users do not use

the roman letter input, it is difficult for them to use a

QWERTY keyboard. Especially, Japanese young

people often find difficulty in efficiently operating

QWERTY keyboards.

This paper proposes a Japanese input software

keyboard that integrates flick input with a split

keyboard. It largely improves display space

efficiency and also shortens finger movement. Its

layout consists of 2×5×2 keys, and its key size is

60×60 pixels. This layout reduces the necessary

display space by 70% in the portrait mode and by 81%

in the landscape mode, compared with a QWERTY

keyboard with the maximum display width. A

preliminary experiment was conducted to explore

input method problems, appropriate key sizes, and

appropriate key layouts and also to estimate user

fatigue. The results of the preliminary experiment

indicated a problem with its 5-row key layout.

Therefore, L-shaped and Γ-shaped keyboard layouts

were further added. The L-shaped and Γ-shaped

layouts reduce the necessary display space by 73% in

the portrait mode and by 83% in the landscape mode.

In addition, the final implementation included the

function of converting Japanese kana characters to

Chinese characters and placed it on the right side of

the keyboard.

To show complementary values of the proposed

keyboard, a main experiment on the comparison of

the 5-row, L-shaped, and Γ-shaped layouts with the

QWERTY key layout was conducted. The main

experiments used two kinds of input tasks, sentence

input and word repetition input. The results indicate

that the error rates of the proposed layouts had

valuable complements in the experiments of both

kinds of input.

170

Nakamura, Y. and Hosobe, H.

A Japanese Bimanual Flick Keyboard for Tablets That Improves Display Space Efﬁciency.

DOI: 10.5220/0008969101700177

In Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2020) - Volume 2: HUCAPP, pages

170-177

ISBN: 978-989-758-402-2; ISSN: 2184-4321

2 RELATED WORK

Various research has been done on keyboards for

tablets. Sax et al. proposed an ergonomic QWERTY

tablet keyboard (Sax, Lau, & Lawrence, 2011). Bi et

al. proposed a bimanual gesture keyboard to reduce

display space and to shorten finger movement (Bi,

Chelba, Ouyang, Partridge, & Zhai, 2012). Hasegawa

et al. studied input of a software keyboard, with a

focus on aging effects and differences between

dominant and non-dominant hands (Hasegawa,

Hasegawa, & Miyao, 2012). Odell studied feedbacks

of software keyboards (Odell, 2015). Takei and

Hosobe proposed a Japanese kana input keyboard that

input 1 character with 2 strokes by using 2×6 keys

(Takei & Hosobe, 2018). Yajima and Hosobe

proposed a Japanese software keyboard for tablets

that reduced user fatigue (Yajima & Hosobe, 2018).

Much research on flick keyboards has been done

in Japan. Sakurai and Masui proposed a QWERTY

flick keyboard (Sakurai & Masui, 2013). This

keyboard enabled input of Japanese kana characters

and roman letters without mode changes. Fukatsu et

al. proposed an eyes-free Japanese kana input method

called no-look flick (Fukatsu, Shizuki, & Tanaka,

2013). This method enabled flick input for vowels

and consonants in two strokes. Hakoda et al. proposed

a kana input method using two fingers for touch-panel

devices (Hakoda, Fukatsu, Shizuki, & Tanaka, 2013).

This method was also an eyes-free Japanese input

method, but enabled gesture input by two fingers.

Nagasawa investigated, by using questionnaire,

how Japanese university students type on

smartphones and PCs (Nagasawa, 2017). The result

showed that Japanese university students preferred

flick input to a QWERTY input whichever of English

or Japanese letter input was used. Also, the Japanese

Ministry of Education conducted a survey of the

information utilization ability of elementary, junior-

high, and high school students from 2013 to 2016

( [Japanese Ministry of Education, Culuture, Sports,

Science and Technology, 2017]). The results showed

that most students were not able to smoothly enter

text with keyboards.

Research on display space has been done.

Hutchings and Stasko organized a display by creating

a small window for managing and displaying related

information (Hutchings & Stasko, 2004). Hutchings

et al. investigated window management methods for

single-monitor users and multi-monitor users by

creating a tool that tracked window management

events and recorded the window configurations

continuously (Hutchings, Smith, Meyers, Czerwinski,

& Robertson, 2004). This research was done when the

multi-display environment was not popular.

3 PRELIMINARIES

3.1 Japanese Kana Characters

Japanese text is composed of Japanese kana

characters and Chinese characters (called kanji in

Japanese). While a Chinese character typically has a

meaning, a kana character does not; instead, a kana

character is associated with a speech sound. There are

two kinds of kana characters called hiragana and

katakana. Although they are used for different

purposes, they correspond to each other; for each

hiragana character, there is a corresponding katakana

character, and vice versa. In many commonly used

Japanese input methods, kana characters are entered

with keys, and Chinese characters and katakana are

entered with the aid of conversion functions. The

method proposed in this paper is the same at this

point.

There are approximately 50 basic kana characters,

which are further divided into 10 groups that are

ordered, each of which typically consists of 5

characters. The first group is special because its 5

characters indicate 5 vowels that are pronounced “a”,

“i”, “u”, “e”, and “o”. The other 9 groups are

associated with the basic consonants, “k”, “s”, “t”,

“n”, “h”, “m”, “y”, “r”, and “w”. A kana character in

these 9 groups forms the sound that combines a

consonant and a vowel. For example, the 5 characters

of the “k” group are pronounced “ka”, “ki”, “ku”,

“ke”, and “ko”. This grouping of kana characters is

basic knowledge of the Japanese language.

The “k”, “s”, “t”, and “h” groups have variants

called dakuon. Specifically, the dakuon variants of

“k”, “s”, “t”, and “h” are “g”, “z”, “d”, and “b”

respectively. In addition, the “h” group has another

variant called handakuon, which is “p”. Certain

characters have variants that are written in smaller

shapes. Sequences of kana characters can be

expressed with the Roman alphabet by using the

standard Japanese romanization system (ISO, 1989).

This is widely used for computer users to enter

Japanese text with alphabet keyboards such as

QWERTY.

3.2 Japanese Flick Keyboard

Figure 1 is a standard Japanese flick keyboard. The

main character keys are composed of 4×3 key layout.

If a user flicks a key to the left, upward, to the right,

A Japanese Bimanual Flick Keyboard for Tablets That Improves Display Space Efﬁciency

171

or downward with a thumb, the keyboard inputs a

character corresponding to the direction. The enter

key and the delete key are located on the right side of

the keyboard. When the “123” or “ABC” key is

pressed on the left part of the keyboard, the Japanese

keyboard is replaced with the English letter keyboard

or the number letter keyboard. A conversion space is

located at the top of the keyboard. When a user

touches a word, hiragana characters are converted to

kanji or other characters. If a user touches the upward

arrow, it will show other kanji candidates.

Figure 1: General Japanese flick keyboard (which is

implemented on iPhone XS).

4 PRELIMINARY EXPERIMENT

A preliminary experiment was conducted to explore

appropriate key sizes, key layouts, and input methods

and to estimate user fatigue.

4.1 Method

In this experiment, participants first held a tablet in

the portrait mode, operated it freely with two thumbs

for some time, and were asked about the feeling of its

use. Next, they turned the tablet into the landscape

mode, and did the same things. We recruited 6

participants who all were university students

majoring in computer and information sciences.

We implemented the keyboard on an ASUS

ZenPad 10 tablet (Android OS 7.0, 1920×1200px

screen) as shown in Figure 2. The key sizes are

60×60px, and the keyboard is placed symmetrically

at the lower ends of the display. The “lowercase

convert” key changes the last input character if it can

be converted to a lowercase character or a sonant

mark. When text is entered, it is displayed on the text

field at the top of the display.

Figure 2: Application used in the preliminary experiment.

4.2 Result

No problems were observed in the key sizes and the

input method, but most of the participants found the

5-row key layout inconvenient and felt fatigue

because it required long vertical movement of

thumbs. Also, participants wanted a function for

adjusting the keyboard position.

Therefore, we decided to make a keyboard whose

position would be adjustable, and also to additionally

create an L-shaped layout and a Γ-shaped layout. The

keyboard position adjustment function was

implemented as a slide bar. We also implemented the

function of converting Japanese kana characters to

Chinese characters by using an open-source

dictionary called SKK.

5 PROPOSED METHOD

This paper proposes a Japanese flick keyboard that

reduces display space by splitting the keyboard into

the left and the right. It adopts the key layouts shown

in Figure 3. As a solution to the problems identified

through the preliminary experiment, it introduces an

L-shaped layout (Figure 3-b) and a Γ-shaped layout

(Figure 3-c) in addition to the initial 5-row layout

(Figure 3-a). These 2 new layouts remove the arrows

used to change conversions in the 5-row layout.

These layouts enable the effective use of the

display because the split keyboards give more display

space at the center. Reducing the display space of the

keyboard on a tablet can improve the efficiency of the

user’s multitasking. In particular, in the split screen

mode implemented on Android OS 7.0, display space

is important for the user’s multitasking.

The user holds a tablet with both hands, and

performs flick input with both thumbs. The character

to be entered is displayed above the touched key, and

when a thumb is moved to the left, upward, to the

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

172

right, or downward and is released, the corresponding

character is entered. The bimanual input enables

smooth input without the user’s releasing hands

because the user can keep the inputting posture to

hold the tablet. Another advantage is that it enables

stable input because the user can input while holding

a tablet with both hands.

(a)

(b) (c)

Figure 3: (a) 5-row, (b) L-shaped, and (c) Γ-shaped key

layouts.

The flick input of the proposed keyboard is based

on Table 1. Since hiragana basically has five

characters in one group, it works well with flick input

that uses moving a user’s thumb to 4 directions as

well as keeping it at a neutral position. If a user flicks

a button to the left, upward, to the right, or downward

with a thumb, the keyboard inputs the character

corresponding to the direction. If the user touches the

lowercase convert key, the entered character is

converted to the “Lower 1” character in the table. If

the user touches the lowercase covert key again, the

entered character is converted into the “Lower 2”

character if it exists, and into the original character

otherwise. “Lower 1” is composed of dakuon and

certain characters, and “Lower2” is composed of

handakuon and certain characters.

Table 1: Flick input of the proposed keyboard.

The Japanese kana to Chinese character

conversion method is executed by using the

conversion space near the right keyboard. When the

user swipes this space with a thumb, it displays next

Chinese characters, and then the user can tap a word

to confirm the conversion. When no character is

entered, it displays the top 15 words obtained from

the Balanced Corpus of Contemporary Written

Japanese (National Institute for Japanese Language

and Linguistics, 2015). Thus this space enables

inputting frequently used words easily.

6 IMPLEMENTATION

The 5-row layout keyboard (Figure 4-a) has the

height of 300px and the width of 120px, the L-shaped

and Γ-shaped layouts have the height of 240px and

the width of 180px. The 5-row key layout reduces the

display space by 70% in the portrait mode and by 81%

in the landscape mode, compared with the QWERTY

layout keyboard (Figure 4-b) with the maximum

display width. The L-shaped and the Γ-shaped layout

keyboard reduce the display space by 73% in the

portrait mode and by 83% in the landscape mode.

The SKK dictionary of size M was used to convert

Japanese kana characters to Chinese characters. The

conversion spaces are located on the red parts of

Figure 4. The conversion to katakana can be done by

swiping the conversion space to the left. In the 5-row

and the Γ-shaped layout, the conversion space is

located on the left side of the right keyboard. In the

L-shaped layout, the conversion space is overlapped

with the right keyboard. In this case, when a user

touches the conversion space, the processing of the

conversion space has a higher priority. The

recognition of flick input is done in the same way as

usual Japanese flick keyboards.

(a) (b)

Figure 4: (a) The proposed keyboard and (b) the

QWERTY keyboard.

Group Neutral Left Up Right Down Lower 1 Lower 2

ああいうえおぁぃぅぇぉ

かかきくけこがぎぐげご

ささしすせそざじずぜぞ

たたちつてとだぢづでど　　っ　　

ななにぬねの

ははひふへほばびぶべぼぱぴぷぺぽ

ままみむめも

やや（ゆ）よゃ　ゅ　ょ

ららりるれろ

わわをんー＠ゎ　　　　

other 、。！？

A Japanese Bimanual Flick Keyboard for Tablets That Improves Display Space Efﬁciency

173

Figure 4-b is the QWERTY keyboard created for

comparative experiments. This key layout is based on

the “ATOK for ASUS” keyboard, but the keys

unnecessary for the experiment were not included.

7 MAIN EXPERIMENT

The tablet used in the main experiment was the same

as the one in the preliminary experiment. We

recruited 10 participants who were Japanese

university students majoring in computer and

information sciences. The main experiment was

composed of two input experiments and subjective

evaluation. The subjective evaluation included

comments. In the main experiment, participants sat

down on a chair and held a tablet with both hands. If

participants were not able to reach the center of the

keyboard in using the QWERTY, they were allowed

to release their hands. The reasons why we selected

the QWERTY keyboard for comparison were that

there are no popular flick-based tablet keyboards and

that split keyboards are not popular in Japan.

The first experiment was the sentence input

experiment, and the second one was the word

repetition input experiment. For the first and the

second experiment, the sentence input speed (WPM

)

and the word input speed (WPM

) are calculated

respectively as follows:

WPM

+ |E| + 1

× 60

WPM

＝

+ 1

× 60

where T is the length of the string, W is the length of

the word, S is the input time, and E is the number of

entered keys after the conversion. The 1’s are added

to the length of a string or a word to consider the enter

key.

For both sentence input and word repetition input,

error rates, NCER (which defines the not-corrected

error rate) and CER (which defines the corrected error

rate) are calculated as follows:

NCER =

INF

C + IF + INF

CER =

C + IF + INF

where C is the total number of correct words, IF is the

number of incorrect but fixed (backspaced) words,

and INF is the number of incorrect (but not fixed)

words. These equations are based on Bi et al.’s

research (Bi, Chelba, Ouyang, Partridge, & Zhai,

2012). However, we adjust them to Japanese

character input.

After the input experiment, we investigated

subjective evaluation. It was composed of comfort,

efficiency, etc., with the scales of 1 to 10, and

included comments. These items are based on the

ones used in Bi et al.’s research (Bi, Chelba, Ouyang,

Partridge, & Zhai, 2012) that extended NASA TLX

(Hart & Staveland, 1988).

7.1 Sentence Input Experiment

In the sentence input experiment, the phrase set

created by MacKenzie and Soukoreff (MacKenzie &

Soukoreff, 2003), which was used in Bi et al.’s

research (Bi, Chelba, Ouyang, Partridge, & Zhai,

2012), was translated into Japanese and was used.

The main experiment treated both portrait and

landscape modes of each keyboard layout. Each input

method was carried out for 5 types of sentences.

Before the main experiment, participants practiced

for a few minutes. They started the experiment by

pushing the start button on Figure 4-a, and moved to

the next sentence by pushing the enter key. A target

sentence was displayed on the text field at the top of

the display. The sentences included in the list were

about 10-character sentences that mixed Japanese

normal kana (hiragana) characters and Chinese

characters. The list did not include Japanese katakana

characters, numbers, and English letters. The main

experiment was performed for the QWERTY

keyboard first, the 5-row, L-shaped, and Γ-shaped

keyboards in the portrait mode, and then the three

keyboards in the landscape mode as well.

7.2 Result of the Sentence Input

Experiment

The result of the input speed measurement is shown

in Figure 5. The result of analysing WPM

and error

rates with ANOVA did not show a significant

difference about NCER and CER, but a significant

difference (p < 0.05) was shown for the input speed,

which means that the proposed keyboard was slower

than the QWERTY. In addition, the L-shaped layout

indicated the highest WPM

among the proposed

methods. The average error rate became higher

because of the following reasons: mistaking a tap of

the conversion space, misreading of a target sentence,

and partial conversion for correcting unnecessary

parts with the backspace key.

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

174

(a)

(b)

(c)

Figure 5: (a) Sentence input speed (WPM

), (b) corrected

error rate (CER), and (c) not-corrected error rate (NCER).

7.3 Word Repetition Input Experiment

The word repetition input experiment used the same

layouts as the sentence input experiment. The

experiment measured learnability by exploring

changes of the speed in inputting the same word 6

times successively. The used words consist of 4

Japanese normal kana characters, possibly including

sonant marks.

7.4 Result of the Word Repetition

Input Experiment

The result of word input speeds (WPM

) is shown in.

The QWERTY keyboard indicated the highest

WPM

among all the input methods. There were no

large differences among the proposed methods.

ANOVA showed a significant difference (p < 0.05)

through the first to the sixth, which means that the

proposed method was slower than the QWERTY.

ANOVA did not show a significant difference among

the proposed methods.

7.5 Subjective Evaluation

Figure 7shows the result of the subjective evaluation.

ANOVA showed significant differences (p < 0.05)

in comfort, accuracy, absence of frustration, and

mental demand. Overall preference did not show a

significant difference, and there were no large

differences among the proposed methods.

8 DISCUSSION

8.1 Sentence Input

In the result of the sentence input experiment, the

error rates did not show a significant difference, but

the proposed method had a slower input speed than

the QWERTY, showing a significant difference. One

reason is that the participants needed to more

carefully input because of the different key layouts

from the standard one. Another reason is that our flick

keyboard was split unlike the standard flick keyboard

that is composed of 3×4 keys (including punctuation

marks and the lowercase convert key).

Large differences did not appear among the

proposed methods; in fact, ANOVA did not show a

significant difference among the proposed methods.

A reason for the differences of input speed between

the QWERTY keyboard and the proposed methods is

that all the participants were students majoring in

computer and information sciences and therefore

were good at the QWERTY keyboard. In addition,

many of the participants were more used to the

QWERTY keyboard than flick input.

In all the methods, the landscape mode was faster

than the portrait mode. One reason might be that the

portrait mode has longer distances between the input

field and the keyboard as shown in Figure 4.

A solution to these problems is to introduce

customize functions. In other words, it is necessary to

Figure 6: Word input speed (WPM

A Japanese Bimanual Flick Keyboard for Tablets That Improves Display Space Efﬁciency

175

Figure 7: Result of the subjective evaluation. For measures 1-5 and 8, the score of 10 indicates the most positive rating, and

for measures 6 and 7, the score of 10 indicates the most negative rating (based on (Bi, Chelba, Ouyang, Partridge, & Zhai,

2012)).

enable the customization of keyboards for user’s

preference. A solution to the problem of inputting

string that is far from the keyboard is to display the

string near the keyboard or to display a predictive

conversion near the input field which is usual on

personal computers.

8.2 Word Repetition Input

The word repetition input experiment did not show

large differences among the proposed methods in

input speed, but ANOVA showed a significant

difference between them and the QWERTY

keyboard. A reason for the increased misjudgement

in continuous input is that flick movement distance

decreased. Another reason is that tapping the

lowercase convert key failed. The direction was

determined by comparing the vertical movement and

the horizontal movement. However, when the input

direction was upward, a misjudgement occurred

because of the sliding of a thumb to the left or to the

right. This misjudgement differed in participants,

which makes it difficult to solve the problem by using

the same configuration for all users. One solution is

to introduce a “reform” mode, which adjusts the

border of the judgment (e.g., by machine learning)

when misjudgement occurs. The problem of tapping

the lowercase convert key was unique to the flick

keyboard, because the QWERTY keyboard obtained

the input of lowercase characters and sonant marks in

the same ways as other characters. The usual flick

keyboard avoids this problem by using predictive

conversion.

8.3 Subjective Evaluation

In the subjective evaluation, ANOVA showed

significant differences between the QWERTY and

the proposed method in comfort, accuracy, absence of

frustration, and mental demand. One reason for the

lower evaluation is that the time of the participants’

use of the proposed method was short. In this research,

the participants used the proposed method for only a

few minutes of warming up. In addition, they were

fully experienced with the QWERTY keyboard

because they majored in computer and information

sciences. The mental demand did not show a

significant difference among the proposed methods,

but the participants got the most tired when they used

the Γ-shaped layout. Since the experiment of the Γ-

shaped layout was conducted after those of the other

input methods, participants possibly confused the key

layout with the other layouts.

Although ANOVA did not show a significant

difference, the physical demand showed an advantage

of the proposed method, which is a common

advantage of split keyboards that reduce finger

movement.

8.4 Participants’ Comments

Comments showed that participants wanted functions

for changing key layouts and key positions. In

addition, a participant wanted a function for

unimanual flick input, together with the current

bimanual keyboard. The reason for the changing

functions is that the proposed layout differed from the

layout of the commonly used flick keyboard. Another

reason is that participants needed to use keys on only

one side for some words. Participants were satisfied

with the slide bar for moving the keyboard position

upward or downward, but some participants also

wanted to move the keyboard position to the left or to

the right in order to fit it to their hands.

There was a comment about the multitap input

function for selecting kana characters without flick

gestures, which is another common way for inputting

kana characters. Although the multitap input function

was implemented, the experiment excluded it for the

comparison purpose. There were comments about

flick input that moving a finger upward or downward

caused fatigue, and that flicking to the inside direction

was hard to perform in the large display. Also,

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

176

participants felt fatigue because they were not able to

see the keyboard at once in the landscape mode.

There were positive comments of participants.

One is that the proposed keyboard will be useful when

the participant gets used to it. Another is that the

proposed method is more useful than the QWERTY

keyboard when the participant holds the tablet in the

landscape mode.

9 CONCLUSIONS AND FUTURE

WORK

This paper proposed a Japanese bimanual flick

keyboard for tablets that improves display space

efficiency. In the proposed method, the 5-row key

layout reduced the display space by 70% in the

portrait mode and by 81% in the landscape mode,

compared with the QWERTY keyboard with the

maximum display width. The L-shaped and Γ-shaped

layouts reduced the display space by 73% in the

portrait mode and by 83% in the landscape mode.

However, in comparative experiments, the proposed

method indicated slower input speed than the

QWERTY keyboard.

Future directions include the implementation of

customization functions, a predictive conversion

function, and a function for setting the multitap input.

Since the result of the experiment suggested it is

necessary to implement functions for setting up

appropriate keyboards for different users. There is

room for improvement in experiments. For example,

the usage time of the proposed method was much

shorter than the QWERTY keyboard. It was a main

cause for the difference of the experimental results.

Also, it is necessary to increase the number of

keyboards to be compared. For example, the above-

mentioned problem might be reduced by using an

input method based on a flick keyboard or other input

methods that are less familiar to the subject than the

QWERTY keyboard.

REFERENCES

Bi, X., Chelba, C., Ouyang, T., Partridge, K., & Zhai, S.

(2012). Bimanual Gesture Keyboard. In Proc. UIST,

pages 137-146.

Fukatsu, Y., Shizuki, B., & Tanaka, J. (2013). No-look

Flick: Singlehanded and eyes-free Japanese text input

system on touch screens of mobile devices. In Proc.

MobileHCI, pages 161– 170.

Hakoda, H., Fukatsu, Y., Shizuki, B., & Tanaka, J. (2013).

An eyesfree kana input method using two fingers for

touch-panel devices. IPSJ SIG Tech. Rep., HCI-

154(6):1–8 In Japanese.

Hart, S., & Staveland, L. (1988). Development of NASA-

TLX (Task Load Index): Result of empirical and

theoretical research. Advances in Psychology. vol.52,

pages 139-183.

Hasegawa, A., Hasegawa, S., & Miyao, M. (2012).

Characteristics of the input on software keyboard of

tablet devices: Aging effects and differences between

the dominant and non-dominant hands for input. J.

Mobile Interact., 2(1):23–28. In Japanese.

Hutchings, D. R., & Stasko, J. (2004). Shrinking Window

Operations for Expanding Display Space. In Proc. AVI,

pages 350-353.

Hutchings, D. R., Smith, G., Meyers, B., Czerwinski, M.,

& Robertson, G. (2004). Display Space Usage and

Window Management Operation Comparisons between

Single Monitor and Multiple Monitor Users. In Proc.

AVI, pages 32-39.

ISO. (1989). Documentation—Romanization of Japanese

(kana script) ISO 3602.

Japanese Ministry of Education, Culuture, Sports, Science

and Technology. (2017). Information utilization ability

survey for high school students.

MacKenzie, S., & Soukoreff, W. (2003). Phrase Sets for

Evaluating Text Entry Techniques. In Proc. CHI, pages

754-755.

Nagasawa, N. (2017). How Japanese university students

type on smartphone and PC. Computer & Education

vol.43 pages 67-72. In Japanese.

National Institute for Japanese Language and Linguistics.

(2015). Balanced Corpus of Contemporary Written

Japanese Short Unit Word list data ver.1.0,.

https://pj.ninjal.ac.jp/corpus_center/bccwj/freq-

list.html.

Odell, D. (2015). On-Screen Keyboard: Does the Presence

of Feedback or Tactile Landmarks Improve Typing

Performance? In Proc. MobileHCI, pages 131-136.

Sakurai, Y., & Masui, T. (2013). A flick-based Japanese

input system for a QWERTY software keyboard. IPSJ

SIG Tech. Rep., HCI-154(5):1–4. In Japanese.

Sax, C., Lau, H., & Lawrence, E. (2011). LiquidKeyboard:

An ergonomic, adaptive QWERTY keyboard for

touchscreens and surfaces. In Proc. ICDS, pages 117–

122. IARIA.

Takei, K., & Hosobe, H. (2018). A 2-by-6 Button Japanese

Software Keyboard for Tablets. In Proc. VISIGRAPP,

pages 147–154.

Yajima, T., & Hosobe, H. (2018). A Japanese Software

Keyboard for Tablets That Reduces User Fatigue. In

Proc. COMPSAC, pages 339–346.

A Japanese Bimanual Flick Keyboard for Tablets That Improves Display Space Efﬁciency

177