The Effects of Human Finger and Chinese Character on Chinese Handwriting Performance on Mobile Touch Devices

Abstract

Objective:

The aim of the present study is to investigate Chinese handwriting on mobile touch devices, considering the effects of three characteristics of the human finger (type, length, and width) and three characteristics of Chinese characters (direction of the first stroke, number of strokes, and structure).

Background:

Due to the popularity of touch devices in recent years, finger input for Chinese characters has attracted more attention from both industry and academia. However, previous studies have no systematical consideration on the effects of human finger and Chinese characters on Chinese handwriting performance.

Method:

An experiment was reported in this article to illustrate the effects of the human finger and Chinese characters on the Chinese handwriting performance (i.e., input time, accuracy, number of protruding strokes, mental workload, satisfaction, and physical fatigue).

Results:

The experiment results indicated that all six factors have significant effects on Chinese handwriting performance, especially on the input time, accuracy, and number of protruding strokes.

Conclusion:

Finger type, finger length, finger width, direction of the first stroke, number of strokes, and character structures are significantly influencing Chinese handwriting performance. These factors should be taken into more consideration in future research and the practical design for Chinese handwriting systems.

Keywords

finger input Chinese characters Chinese handwriting mobile touch devices

Introduction

The developments of touch technologies enable people to handwrite with fingers on mobile devices. It is possible to handwrite Chinese characters on smart phones, palmtop computers, tablet computers, and other mobile devices with fingers such as thumb, index finger, and middle finger. Finger-based input for Chinese entry attracts more interest from mobile designers and application developers than stylus-based input. The human thumb is a distinctive finger out of the five human digits and different from the other fingers with respect to dimension, degree of freedom, and movement range. Characteristics of fingers can be analyzed but not improved or developed. For example, people cannot change their fingers’ lengths. Previous studies on Chinese handwriting have focused more on designing an optimal stylus or algorithm for improving the accuracy and efficiency of the Chinese handwriting process. Little research focuses on exploring the influences of the characteristics of the human finger on Chinese handwriting. Therefore, it becomes interesting to understand the process of finger interaction for Chinese input through consideration of finger characteristics such as type and dimension.

Finger-based input for Chinese entry has many advantages. Besides eliminating the necessity of bringing a stylus for mobile touch devices, finger-based input also facilitates the one-hand input position. The two-hand position with thumb or index finger and the one-hand input position with thumb make possible different handwriting situations. It is a common situation that when riding the subway, a user handwrites with the thumb on a smart phone while the other hand grasps the subway handhold. In this situation, users handwrite a few simple characters in a short time. In two-hand situations, users handwrite more and complex characters. The complexity of Chinese characters has a certain influence on the Chinese handwriting process according to the diversity of handwriting situations. Moreover, unlike alphabetical letters, Chinese characters are hieroglyphic, composed of basic radical and directed strokes, which probably lead to the different handwriting performance of thumb and index finger because of their different movement characteristics. Current studies on character characteristics referred mainly to the reading efficiency of Chinese display instead of the input or the use of character segmentation in Chinese character recognition. These studies failed to investigate the fundamental influence of the character characteristics on the Chinese handwriting process.

In the area of human-computer interaction, the input method of Chinese handwriting is a classic topic. It has recently appealed to more and more researchers due to the popularity of mobile touch-sensitive devices. Handwriting input for Chinese input is more natural and similar to Chinese handwriting on paper. Therefore, people who are not familiar with other input methods such as Chinese Pinyin prefer to use handwriting. In this article, the characteristics of the human finger and Chinese characters were both studied. This study conducted an experiment to illustrate the systematic influences of three characteristics of human fingers (type, length, and width) and three characteristics of Chinese characters (direction of the first stroke, number of strokes, and structure) on mobile touch devices.

Literature Review

Handwriting with thumb or index finger was different from handwriting with a stylus. Studies reported the results of Chinese handwriting by using pen and paper (Chan & Lee, 2005; Goonetilleke, Hoffmann, & Luximon, 2009; Wu & Luo, 2006). Thumb and index finger are quite different from each other in respect of size, degree of freedom, and number of joints. Movement of thumb is different from movement of index finger (Häger-Ross & Schieber, 2000; Zhang, Braido, Lee, Hefner, & Redden, 2005). In addition, more characteristics of the human hand had attracted research interest in studies related to touch. Features of the human hand such as hand shape, fingertip, uniqueness of thumb to other fingers, contact area, angle between fingers, and surface were investigated in these studies (Ahsanullah, Mahmood, & Sulaiman, 2010; Epps, Lichman, & Wu, 2006; Holz & Baudisch, 2011; Wang & Ren, 2009). For example, Holz and Baudisch (2010) used fingerprints to improve touch accuracy.

On the other hand, Chinese characters are symbolic characters, so handwriting Chinese has unique characteristics compared with handwriting alphanumeric letters. Most of the Chinese characters are composed of multiple components. The rest of Chinese characters are indecomposable characters. Character structure describes how the components compose a character. Components are composed of strokes. Therefore, Chinese characters are composed of strokes. In previous studies, characteristics of Chinese characters such as direction of strokes, character structure, and number of strokes were widely studied to evaluate the Chinese reading performance (Huang, Patrick Rau, & Liu, 2009; Hwang, Wang, & Her, 1988; Yen, Tsai, Chen, Lin, & Chen, 2011) and handwriting recognition system design (Srihari, Yang, & Ball, 2007; Su, Zhang, Guan, & Huang, 2009; Tang, Tu, Liu, Lee, & Lin, 1998). Moreover, letter trajectory was involved in research on handwriting learning for English letters (Vinter & Chartrel, 2010).

Various measurements have been adopted to evaluate handwriting performance. Input time is the first and most frequently used index in the input performance evaluation (Chan & Lee, 2005; Lam, Au, Leung, & Li-Tsang, 2011). Accuracy was employed to describe how many handwritten characters were recognized by the handwriting system (Lam et al., 2011). Besides input time and accuracy, the protruding stroke was involved in research on Chinese handwriting (Ren & Zhou, 2009; Tu & Ren, 2012). Both the length and the number of protruding strokes were considered in these studies. In recent studies, subjective measurements also facilitated the handwriting performance evaluation. Users’ satisfaction, fatigue, and preference were tested with the help of questionnaires or interviews (Chan & Lee, 2005; Tu & Ren, 2012). Users’ satisfaction and preference were usually considered in the evaluation of interface design of Chinese handwriting (Chan & Lee, 2005; Ren & Zhou, 2009; Tu & Ren, 2012; Wang & Ren, 2009). In the field of kinesthesis, movement like handwriting is measured through diverse biofeedback signals. For example, surface electromyography (sEMG) is a common way to evaluate physical fatigue (van den Broek, 2010).

First, the investigation on Chinese handwriting is directly beneficial for the design of a Chinese handwriting system such as interface design or recognition system design. Exploration on the effects of the human finger and Chinese character on Chinese handwriting performance is very important to design issues such as entry size, shape, and location. For example, the optimal entry size was suggested to be 14 mm × 14 mm on PDA (Ren & Zhou, 2009) and 25 × 25 mm for both thumb and index finger (Tu & Ren, 2012). Second, exploration on finger input for Chinese characters is useful to other interdisciplinary researches such as kinesthesis and education. For example, Lam et al. (2011) successfully used Chinese handwriting performance to differentiate typical children and children with dyslexia. Visual and motor training methods were demonstrated to be effective to improve handwriting performance (Poon, Li-Tsang, Weiss, & Rosenblum, 2010; Vinter & Chartrel, 2010).

Most of the current research on Chinese handwriting focused on developing an optimal recognition algorithm (Srihari et al., 2007; Su et al., 2009; Tang et al., 1998). Characteristics of Chinese characters such as stroke direction, number of strokes, and character structure were involved in establishing a better recognition algorithm to improve the Chinese handwriting performance (Jin & Wei, 1998; Li, Jin, Zhu, & Long, 2008; Su et al., 2009). However, the direct influences of these characteristics on Chinese handwriting performance are not clear. For example, Chinese character strokes had four basic directions, which are called horizontal stroke, vertical stroke, left-descending stroke, and right-descending stroke. It is unknown if Chinese handwriting performance is different in the four directions. Moreover, it is unknown that thumb and index finger have any preference on the four directions considering the Chinese handwriting performance.

Significance

Thus, the objective of the present study is to explore the effects of characteristics of the human finger and Chinese character on the Chinese handwriting performance. This present study is mainly different from previous studies in two aspects. First, previous studies rarely consider the effect of finger types, finger length, finger width, stroke direction, number of strokes, and character structure on the Chinese handwriting performance. Second, in human-computer interaction Chinese handwriting performance was usually measured as input time, accuracy, number of protruding strokes, and subjective ratings such as easy-to-write. Few objective measurements are used to evaluate the physical fatigue of Chinese handwriting. Thus, it is interesting to discover if the EMG method is an effective tool to evaluate the handwriting fatigue. Moreover, the results of the present study can benefit the design on Chinese handwriting system and interdisciplinary research.

Methodology

Experiment Design

This study design was a within-group experiment. Six independent variables, finger types, finger length, finger width, stroke direction, number of strokes, and character structure, were included in the experiment. Finger types had two levels, using thumb and using index finger. According to the Chinese standard of hand size, finger length and width were divided into three levels separately (CNIS, 1996). For finger length, three levels were small (thumb: less than or equal to 56 mm, index finger: less than or equal to 84 mm), medium (thumb: 56~62 mm, index finger: 84~92 mm), and large (thumb: more than or equal to 62 mm, index finger: greater than or equal to 92 mm). For finger width, three levels were small (thumb: less than or equal to 18 mm, index finger: less than or equal to 14 mm), medium (thumb: 18~21 mm, index finger: 14~16 mm), and large (thumb: more than or equal to 21 mm, index finger: more than or 16 equal to mm). For stroke direction, four levels were tested, horizontal stroke, vertical stroke, left-descending stroke, and right-descending stroke. For the number of strokes, three levels were tested, simple (less than 8 strokes), medium (8–13 strokes), and hard (more than 13 strokes). For character structure, four levels were tested, left-to-right (e.g., “海” and “树”), up-to-down (e.g., “吴” and “售”), surrounding (e.g., “围” and “凰”), and indecomposable (e.g., “十” and “木”). A factorial design was adopted to study the three factors related to Chinese character. There were 48 Chinese character types in total (4 directions × 3 number of strokes × 4 character structure). Only 39 Chinese character types were used in the experiment as they were in the list of frequently used characters in modern Chinese (Fu, Sun, Zhang, Wei, & Zhu, 1988). A Latin square design was used to arrange the order of character types to avoid the order effect.

Participants

Thirty-nine engineering students were invited to this experiment (20 female, 19 male). Fifteen participants were senior undergraduate students and 24 were graduate students. As for finger use, 15 participants used their index fingers as their dominant finger to touch the mobile phone and 24 used their thumbs. When positioning the mobile device, 21 participants preferred using the one-hand position while 18 used the two-hand position most of the time. Of all the participants, 31 have had experience in using touch devices before. More demographic information is presented in Table 1.

Table 1:

Descriptive Result of Demographic Information

Item	Average	SD
Age	22.6	1.9
Years of experience of writing Chinese characters	17.8	2.2
Level of Chinese proficiency (grade of Chinese test in the college entrance examination)	119.0	7.9
Length of thumb (cm)	59.90	5.17
Width of thumb (cm)	19.51	1.75
Length of index finger (cm)	89.28	6.71
Width of index finger (cm)	15.13	1.22

Task

In this experiment, there were 169 Chinese characters chosen to input by participants by using their thumbs and index fingers. Table 2 shows some examples of Chinese characters in the experiment. These Chinese characters were selected based on three characteristics of Chinese characters: direction of the first stroke, number of strokes, and traditional structure of Chinese characters. Twenty participants first used their thumbs to hand input the 169 Chinese characters and then used their index fingers to repeat the process. Another 19 participants followed the opposite order, starting with their index finger. As the experiment was a within-group design, 39 participants handwrote the 169 characters twice. There were a total of 13,182 (169 × 39 × 2) trials.

Table 2:

Examples of Chinese Characters in the Experiment

Direction	Type	Structure	Chinese Character	Direction	Type	Structure	Chinese Character
Horizontal	Simple	Left-right	孔	Left-descending	Simple	Left-right	仍
Horizontal	Simple	Up-down	艺	Left-descending	Simple	Up-down	冗
Horizontal	Simple	Surrounding	区	Left-descending	Simple	Surrounding	凤
Horizontal	Simple	Indecomposable	十	Left-descending	Simple	Indecomposable	八
Horizontal	Medium	Left-right	取	Left-descending	Medium	Left-right	股
Horizontal	Medium	Up-down	奈	Left-descending	Medium	Up-down	命
Horizontal	Medium	Surrounding	建	Left-descending	Medium	Surrounding	周
Horizontal	Hard	Left-right	摧	Left-descending	Hard	Left-right	缥
Horizontal	Hard	Up-down	嘉	Left-descending	Hard	Up-down	算
Horizontal	Hard	Surrounding	厮	Left-descending	Hard	Surrounding	魅
Vertical	Simple	Left-right	比	Right-descending	Simple	Left-right	计
Vertical	Simple	Up-down	冗	Right-descending	Simple	Up-down	六
Vertical	Simple	Surrounding	冈	Right-descending	Simple	Surrounding	凶
Vertical	Simple	Indecomposable	卜	Right-descending	Simple	Indecomposable	丫
Vertical	Medium	Left-right	叔	Right-descending	Medium	Left-right	泡
Vertical	Medium	Up-down	咒	Right-descending	Medium	Up-down	单
Vertical	Medium	Surrounding	幽	Right-descending	Medium	Surrounding	疚
Vertical	Hard	Left-right	慢	Right-descending	Hard	Left-right	漆
Vertical	Hard	Up-down	睿	Right-descending	Hard	Up-down	裹
				Right-descending	Hard	Surrounding	遮

Procedure

After an introduction to the experiment, each participant filled out a prequestionnaire, including age, gender, education background, experience of Chinese handwriting, experience of personal computers, and experience of handwriting. The experiment instructor then measured the dimensions of the thumb and the index finger (length, width) of each participant. A handwriting exercise was then completed to make participants familiar with the experiment interface. A surface electromyography sensor was placed on the wrist of participants to record the electromyography signals of their hands. Before each writing task participants were asked to put their arms on the table in a comfortable position and keep still, and meanwhile, sEMG data were recorded for 20 seconds. sEMG data were collected when participants handwrote. After that, participants were asked to complete two handwriting tasks. One was handwriting the 169 Chinese characters using thumb, and the other was using index finger. sEMG data in no movement were recorded again for 20 seconds before the second handwriting task when the participants were at rest. A questionnaire of satisfaction (7-point Likert scale) and a questionnaire of mental workload (adopted from NASA-TLX) were filled out by participants when they completed each handwriting task.

Apparatus

The experiment software was coded with C-Objective on the iOS operation system. An open source OCR engine was used to recognize characters handwritten by participants. The experiment was conducted on iPod 3 with a capacitive touch screen. This device was 3.5 inch, with resolution of 480 × 320. The experiment interface is shown in Figure 1. On the top left there was a hint of the character to write. Input area was centered, which was 25 × 25 mm. There were a rewrite button and a next button on the right. A camera was used to record the handwriting movements of participants.

Figure 1.

Interface of experiment.

Measurements

Input time, accuracy, the number of protruding strokes

Satisfaction, mental workload, physical fatigue, and sEMG were used to evaluate the Chinese handwriting performance. Input time was recorded in two ways, including the input time per character and the input time of the first stroke. The accuracy was recorded as 1 if it was recognized as right and 0 if not. Number of protruding strokes was recorded as the total number of protruding strokes per character. Satisfaction and mental workload were tested in questionnaires after participants completed each handwriting task. Physical fatigue was tested with sEMG signals. The original sEMG data were first transferred to its root mean square (RMS), which was one of the most important indexes for muscle fatigue. Then sEMG was calculated as:

{sEMG}_{relative} = mean of {sEMG}_{handwriting} - mean of {sEMG}_{no movement} .

This was to reduce the influences of individual differences as the sEMG was extremely individually different.

Results

Input Time

The input time per character

Table 3 illustrates the input time per character and ANOVA analysis results. Finger type, finger length and width, number of strokes per character, and structure were found to have significant effects on the input time per character. Finger Type × Finger Length and Finger Type × Finger Width had two-level interactive effects. The input time per character was 2.80 seconds on average when using thumb and 2.42 seconds when using index finger. The input time was 2.68, 2.46, and 2.72 seconds for three finger length levels, small, medium, and large. The input time was 2.66, 2.47, and 2.76 seconds for three finger width levels. The input time was 2.61, 2.59, and 2.71 seconds for characters with 2 to 7 strokes, 8 to 13 strokes, and 14 to 19 strokes, respectively. The characters with 14 to 19 strokes led to a significantly higher input time per character than that for the other two levels. The input time per character was 2.63, 2.64, 2.62, and 2.57 seconds for characters of which the structures were left-to-right, up-to-down, surrounding, and indecomposable, respectively. The indecomposable character had a significantly lower input time per character than the other three structures.

Table 3:

The Input Time (sC⁻¹)

		Descriptive Analysis		ANOVA
		M	SD	F Values	p Values
Finger type	Thumb (T)	2.80	0.75	6.35	.01*
	Index finger (I)	2.42	0.59
Finger length	Small	2.68	1.74	28.44	.00*
	Medium	2.46	1.67
	Large	2.72	1.97
Finger width	Small	2.66	1.76	21.25	.00*
	Medium	2.47	1.71
	Large	2.76	1.94
Finger Type × Finger Length	Small (T)Medium (T)	1.921.82	0.960.86	4.58	.01*
	Large (T)	1.90	0.94
	Small (I)	1.78	0.80
	Medium (I)	1.45	0.59
	Large (I)	1.69	0.84
Finger Type × Finger Width	Small (T)Medium (T)	1.821.79	0.860.93	3.57	.03*
	Large (T)	2.03	0.96
	Small (I)	1.73	0.75
	Medium (I)	1.56	0.72
	Large (I)	1.69	0.85
Direction of the first stroke	HorizontalVertical	2.652.58	1.831.76	0.46	.71
	Left-descending	2.64	1.81
	Right-descending	2.61	1.84
Number of strokes	2~7	2.61	1.79	9.94	.00*
	8~13	2.59	1.78
	14~19	2.71	1.9
Character structure	Left-to-right	2.63	1.85	2.41	.06*
	Up-to-down	2.64	1.81
	Surrounding	2.62	1.79
	Indecomposable	2.57	1.76

p < .10.

Furthermore, the linear regression analysis showed a relationship between the input time per character and number of strokes as shown in Table 4. In this study, the corresponding linear regression equations were Y = 0.0208X + 3.576 and Y = 0.0177X + 3.0445 when using thumb and index finger, respectively, as shown in Figure 2.

Table 4:

Summary of Regression Analysis

	R	R ²		Coefficient	t	p Values
Thumb	.162	.026	Numbers	0.0208	2.100	.037
			Constant	3.576	35.402	.000
Index finger	.183	.034	Numbers	0.018	2.385	.018
			Constant	3.045	0.076	.000

Figure 2.

The regression line for average input time for a character stroke number.

Average input time of the first stroke

Table 5 illustrated the input time of the first stroke and ANOVA analysis results. Finger type, finger length, finger width, Finger Type × Finger Length, Finger Type × Finger Width, and direction of the first stroke were found to have significant effects on the input time of the first stroke. The input time of the first stroke was 0.28 seconds on average when using thumb and 0.25 seconds when using index finger. The input time of the first stroke was 0.26, 0.26, and 0.27 seconds for three finger length levels, small, medium, and large. The input time of the first stroke was 0.26, 0.23, and 0.29 seconds for three finger width levels. The input time of the first stroke was 0.25, 0.26, 0.28, and 0.25 seconds for characters of which the first strokes were horizontal, vertical, left-descending, and right-descending, respectively. The characters of which the first stroke was left-descending had a higher input time of the first stroke than the other three directions.

Table 5:

Input Time of the First Stroke (sC⁻¹)

		Descriptive Analysis		ANOVA
		M	SD	F Values	p Values
Finger type	Thumb (T)	0.28	0.19	48.22	.00*
	Index finger (I)	0.25	0.18
Finger length	Small	0.26	1.81	7.55	.00*
	Medium	0.26	0.17
	Large	0.27	0.19
Finger width	Small	0.26	0.19	35.09	.00*
	Medium	0.23	0.15
	Large	0.29	0.21
Finger Type × Finger Length	Small (T)Medium (T)	0.260.29	0.160.22	7.40	.00*
	Large (T)	0.28	0.19
	Small (I)	0.26	0.18
	Medium (I)	0.22	0.14
	Large (I)	0.26	0.20
Finger Type × Finger Width	Small (T)Medium (T)	0.270.24	0.180.16	6.02	.00*
	Large (T)	0.32	0.22
	Small (I)	0.25	0.19
	Medium (I)	0.22	0.14
	Large (I)	0.27	0.20
Direction of the first stroke	HorizontalVertical	0.250.26	0.200.19	7.31	.00*
	Left-descending	0.28	0.17
	Right-descending	0.25	0.18
Number of strokes	2~7	0.26	0.18	0.35	.71
	8~13	0.26	0.19
	14~19	0.26	0.19
Character structure	Left-to-right	0.26	0.19	0.96	.41
	Up-to-down	0.27	0.20
	Surrounding	0.25	0.17
	Indecomposable	0.26	0.17

p < .10.

Accuracy

Descriptive analysis and ANOVA analysis on the accuracy were shown in Table 6. Finger type, finger length, finger width, Finger Type × Finger Length, Finger Type × Finger Width, direction of the first stroke, number of strokes, and character structure were demonstrated to have significant effects on the accuracy. The accuracy was 68% on average when using thumb and 73% when using index finger. The accuracy was 73%, 67%, and 72% for three finger length levels, small, medium, and large. The accuracy was 70%, 70%, and 72% for three finger width levels. The accuracy was 71%, 72%, 73%, and 67% when inputting the characters of which the first strokes were horizontal, vertical, left-descending, and right-descending, respectively. Inputting the characters with the right-descending first stroke led to significantly lower accuracy than the other three directions. The accuracy of inputting characters with 2 to 7 strokes, 8 to 13 strokes, and 14 to 19 strokes were 74%, 67%, and 71%, respectively. Inputting the characters with 8 to 13 strokes led to significantly lower accuracy than the other two levels. The accuracy of inputting the left-to-right, up-to-down, surrounding, and indecomposable characters were 72%, 67%, 73%, and 73%, respectively. The up-to-down character had significantly lower accuracy than the other three structures.

Table 6:

Accuracy (%)

		Descriptive Analysis		ANOVA
		M	SD	F Values	p Values
Finger type	Thumb (T)	68	12	4.15	.05*
	Index finger (I)	73	11
Finger length	Small	73	44	25.90	.00*
	Medium	67	47
	Large	72	45
Finger width	Small	70	46	8.79	.02*
	Medium	70	46
	Large	72	45
Finger Type × Finger Length	Small (T)Medium (T)	6964	4648	6.63	.00*
	Large (T)	72	45
	Small (I)	76	43
	Medium (I)	73	45
	Large (I)	72	45
Finger Type × Finger Width	Small (T)Medium (T)	6766	4748	4.61	.01*
	Large (T)	73	44
	Small (I)	75	44
	Medium (I)	75	44
	Large (I)	72	45
Direction of the first stroke	HorizontalVertical	7172	4545	3.34	.02*
	Left-descending	73	45
	Right-descending	67	47
Number of strokes	2~7	74	44	11.67	.00*
	8~13	67	47
	14~19	71	46
Character structure	Left-to-right	72	45	11.94	.00*
	Up-to-down	67	47
	Surrounding	73	45
	Indecomposable	73	45

p < .10.

Number of Protruding Strokes

As shown in Table 7, it was found that finger type, finger length, finger width, Finger Type × Finger Length, Finger Type × Finger Width, direction of the first stroke, and number of strokes had significant effects on the number of protruding strokes. The numbers of protruding strokes were 0.17 on average when using thumb and 0.14 when using index finger. The number of protruding strokes was 0.14, 0.16, and 0.16 for three finger length levels, small, medium, and large. The number of protruding strokes was 0.14, 0.15, and 0.17 for three finger width levels. The numbers of protruding strokes were 0.18, 0.14, 0.15, and 0.15 when inputting the characters of which the first strokes were horizontal, vertical, left-descending, and right-descending, respectively. Inputting the characters with the horizontal first stroke had a significantly higher number of protruding strokes than the other three directions. The numbers of protruding strokes of inputting the characters with 2 to 7 strokes, 8 to 13 strokes, and 14 to 19 strokes were 0.15, 0.15, and 0.17, respectively. Inputting the characters with 14 to 19 strokes led to a significantly higher number of protruding strokes than the other two levels. Furthermore, ANOVA analysis was conducted to discover the influence of independent variables on the number of protruding strokes in different borders of the input box. There is no significant result found on number of protruding strokes in top, bottom, left, or right borders.

Table 7:

Number of Protruding Strokes (per character)

		Descriptive Analysis		ANOVA
		M	SD	F Values	p Values
Finger type	Thumb (T)	0.17	0.48	14.68	.00*
	Index finger (I)	0.14	0.42
Finger length	Small	0.14	0.47	3.51	.03*
	Medium	0.16	0.48
	Large	0.16	0.42
Finger width	Small	0.14	0.49	6.48	.00*
	Medium	0.15	0.45
	Large	0.17	0.42
Finger Type × Finger Length	Small (T)Medium (T)	0.240.34	0.630.72	19.94	.00*
	Large (T)	0.44	0.52
	Small (I)	0.22	0.58
	Medium (I)	0.30	0.50
	Large (I)	0.36	0.62
Finger Type × Finger Width	Small (T)Medium (T)	0.250.34	0.690.65	3.55	.03*
	Large (T)	0.41	0.53
	Small (I)	0.26	0.58
	Medium (I)	0.29	0.56
	Large (I)	0.33	0.59
Direction of the first stroke	Horizontal	0.18	0.48	2.77	.04*
	Vertical	0.14	0.43
	Left-descending	0.15	0.44
	Right-descending	0.15	0.45
Number of strokes	2~7	0.15	0.44	2.41	.09*
	8~13	0.15	0.44
	14~19	0.17	0.50
Character structure	Left-to-right	0.16	0.46	0.95	.41
	Up-to-down	0.15	0.44
	Surrounding	0.17	0.47
	Indecomposable	0.15	0.44

p < .10.

Satisfaction

From the satisfaction questionnaires, participants showed their preference on using thumb or index finger. Participants’ satisfaction ratings were 4.52 (SD = 0.64) when using their thumbs and 4.57 (SD = 0.55) when using their index fingers on average. There was no significant influence of finger type or finger size on the satisfaction.

Mental Workload

The mental workload questionnaire results indicated that the mean ratings were 63.88 (SD = 12.26) on average when using thumb and 61.36 (SD = 12.20) when using index finger. Similar to satisfaction results, there was no significant effect of finger type or finger size on the mental workload.

Fatigue

The sEMGrelative was −0.25 on average (SD = 7.76) when using thumb and −7.14 (SD = 31.75) when using index finger. There was no significant effect of finger type or finger size on sEMG.

Handwriting Area

The trajectories of all 169 handwritten characters of each participant were drawn together, which illustrated the handwriting area of each participant, as shown in Figure 3. Each participant had his or her own handwriting area. There were some similarities among these trajectories. The touched areas were approximately fan-shaped. The two top corners were rarely touched by participants. The size of the untouched area in the top left corner was larger than that in the top right corner.

Figure 3.

The handwriting areas of participants.

Discussion

Input Time

The input time per character

Relationship between average input time and number of strokes was an interesting topic in the Chinese handwriting field. Chan and Lee (2005) investigated that the linear equation of average input time (Y) to number of strokes (X) was Y = 144.4X + 189.9 (ms) for pen use. Previous studies had never shown relationships between average input time and number of strokes for thumb and index finger. The present study indicated that the similar linear equations of using thumb and index finger were Y = 20.8X + 3576.3 (ms) and Y = 17.7X + 3044.5 (ms), respectively. These results also demonstrated that the influence of number of strokes on handwriting speed was reduced while using finger compared to using pen.

Input time for thumb was longer than that for index finger. Thumb was shorter and had one fewer joint and was consequently not as flexible as index finger. Therefore, it is necessary to have special consideration on mobile design for thumb writing. For example, input location for thumb should be different than that for index finger. It was also shown that larger finger (i.e., longer and wider) had longer input time but better accuracy and a smaller number of protruding strokes. The possible reason was when participants with larger fingers spent more time and attention on handwriting, consequently they handwrote more carefully. This was confirmed when reviewing videos recorded in the experiment. Previous studies have never considered finger size in Chinese handwriting. The optimal interface design for finger with different length and width is unknown. Thus, adjustable input box can be a good alternative. Users can adjust the input box properties such as size and location to be suitable to their finger sizes. For character structure, the multiple comparisons in ANOVA showed that input time for character with indecomposable structure was significantly lower than the other three levels. It was because very few Chinese characters were indecomposable when it had more than eight strokes. So all the indecomposable characters tested in this study had less than eight strokes.

One interesting fact was that the multiple comparison in ANOVA analysis showed that input time for character with 2 to 7 strokes was not significantly lower than that for characters with 8 to 13 strokes (p = .45). There were two possible reasons resulting in this. One was that participants tend to handwrite more carefully for characters with lower strokes and handwrite faster for characters with more strokes. In the experiment, participants were required to handwrite as quickly as possible while keeping accuracy. There probably was an expected input time. When the actual input time was lower than expected, they tended to write more carefully. Second, participants had a certain handwriting area no matter which character they wrote, as shown in Figure 3. So when handwriting characters with 2 to 7 strokes, participants handwrite each stroke wider or longer. Due to the difficulty in handwriting characters with too many strokes, it is better to help users avoid handwriting them. For example, the handwriting system should detect which character users are going to write based on the strokes already written.

Average input time of the first stroke

Average input time of the first stroke was useful to test if thumb or index finger had any preference on the direction of the strokes. The experiment results demonstrated that there was a significant main effect of direction of the first stroke on the input time. Input time for horizontal and right-descending strokes was significantly shorter than that for vertical and left-descending strokes. In the handwriting position, the horizontal and right-descending directions were the fingers’ most comfortable moving directions, especially for thumb. Fingers had stronger strength in these two directions, making the movement in these two directions more natural and faster. Thus, it is better to assist users to handwrite the difficult strokes. For example, when users start to handwrite a stroke the handwriting system can recognize which stroke users are going to write and help to finish the rest. However, there was no interactive effect of direction of strokes and finger type on the input time. These results showed that although thumb and index finger were different in number of joints and degrees of freedom, it had similar movement in Chinese handwriting. The similarity can also be supported by the motor constancy theory. Motor constancy theory indicated that movements were equivalent regardless of how many degrees of freedom or joints involved (Raibert, 1977). Motor constancy was validated in handwriting (Phillips, Ogeil, & Best, 2009).

Accuracy

According to the experiment results, all six independent variables had impact on the accuracy. Previous studies either ignored their effects or considered only a few of them. The results of the present study suggested that all these factors should be considered in the future research on Chinese handwriting. For example, finger size should be involved in the interface design. According to a report, the average width of thumb and index finger were 18 mm and 16 mm for male Chinese and 17 mm and 15 mm for female Chinese (CNIS, 1996). As Tu and Ren (2012) suggested that the optimal entry size on mobile phones should be 25 × 25 mm, there is probably not enough space for a male Chinese to handwrite with thumb. Thus, the optimal input size should be larger than 25 × 25 mm. As fingers move toward different directions in the input box, 30 × 30 mm may be the minimum. The effect of finger type on number of protruding strokes indicated that thumb required more space than index finger.

Number of Protruding Strokes

Number of protruding strokes of using thumb is more than using index finger, obviously because thumb is not as flexible as index finger. Previous studies indicate that input time and accuracy of thumb are worse than index finger. Therefore, it is necessary to consider a special design for thumb in the Chinese handwriting system as current research had not done enough effort on this topic. Moreover, number of protruding strokes can help to identify three design factors, size, location, and shape of the entry box. In an optimal entry box, user would hardly handwrite out of the box so the number of protruding strokes could be ignored. In the experiment, number of protruding strokes was less than 0.2 per character. It seems acceptable to have 2 strokes out of the entry box after handwriting 10 characters.

Satisfaction, Mental Workload, and Fatigue

Although significant differences were found on input time, accuracy, and number of protruding strokes between using thumb and index finger, participants showed no preference between thumb and index finger according to satisfaction, mental workload, and sEMG. For satisfaction, participants probably had a demanding time to input with both thumb and index finger. Because the Chinese characters were very familiar to participants, relatively little mental workload was required during handwriting. For sEMG, as there was no time limit in the handwriting tasks, participants could control their muscle in a relatively stable level through the experiment time. So participants could slow down the handwriting process if they felt tired.

Meanwhile, finger size showed no impact on satisfaction, mental workload, and sEMG either. Balakrishnan and Yeow (2008) conducted a structured questionnaire interview and investigated the relationship between thumb size and text entry satisfaction with mobile keyboard and discovered that users’ satisfaction or dissatisfaction is affected by thumb size, especially for users with a large thumb size. There were two possible reasons for the conflict. One was that finger size indeed had an impact because it was finger input not keyboard input. The other reason was that research on cultural differences proved that Chinese were more tolerant of dissatisfaction than people from other cultures (Hofstede, Hofstede, & Minkov, 1991; Lowe & Corkindale, 1998). More studies were needed to verify the detailed reason.

Handwriting Area

The results indicated that each participant had his or her unique handwriting area. The shapes and the handwriting areas are similar while the sizes are individually different. The shape of the handwriting area of thumb and index finger was quite similar according to experiment results. The similarity can also be supported by the motor constancy theory. In the present study, it is shown between finger and index finger. The differences on handwriting area possibly resulted from the preference on handwriting position, gesture, or motions. For example, the participant can choose to move only finger or finger and wrist together to input. As previous studies have rarely considered the handwriting area, it is interesting and necessary to make more explorations on the handwriting area in future research. While writer identification is a research direction in Chinese handwriting (He & Tang, 2004; He, You, & Tang, 2008), findings in this study demonstrated that handwriting area can be another writer identifier.

Conclusion

An experiment was conducted and reported in this article to illustrate the effects of the human finger (type, length, and width) and Chinese character (direction of the first stroke, number of strokes, and structure) on the Chinese handwriting performance. All these factors had significant effects on the Chinese handwriting performance. Input time, accuracy, number of protruding strokes, mental workload, satisfaction, and physical fatigue were measured in the experiment. Based on experiment results, relationships between human finger, Chinese character, and handwriting performance were illustrated and discussed. Moreover, the design guidelines on Chinese handwriting on mobile touch devices are summarized in Table 8.

Table 8:

Design Guidelines for Chinese Handwriting on Mobile Touch Devices

Variables	Design Guidelines	To Achieve Better Performance on:
Finger type	More space for thumb input	Input time, accuracy, number of protruding strokes, satisfaction, mental workload, and fatigue
	Different input locations for thumb and index finger
Finger length and finger width	Minimum input size: 30 × 30 mm	Input time, accuracy, number of protruding strokes
	Adjustable input box
Direction of the first stroke	Assisting users to handwrite vertical and left-descending strokes	Input time, accuracy, number of protruding strokes
Number of strokes	Avoiding handwriting characters with too many strokes	Input time, accuracy, number of protruding strokes

Key Points

The properties of human fingers (finger type, finger length, and finger width) significantly influence Chinese handwriting performance.

The characteristics of Chinese characters (direction of the first stroke, number of strokes, and character structures) have significant effects on Chinese handwriting performance.

Footnotes

This study was funded by a National Science Foundation China grant 71188001.

Zhe Chen is a PhD candidate in the Human Factors and Ergonomics Institute in the Department of Industrial Engineering at Tsinghua University in Beijing.

Pei-Luen Patrick Rau is the Director of Human Factors and Ergonomics Institute and professor of Industrial Engineering Department at Tsinghua University in Beijing. He has founded and directed the Human-Computer Interaction (HCI) and Usability Research Center at Tsinghua University. He was a Visiting Scholar at Microsoft Research Asia in China, Visiting Professor at the Aachen University of Technology in Germany, and a Visiting Professor at the Chuo University in Japan. His research themes include human factors engineering, human-computer interaction, cross-cultural design, design for elderly, web usability, mobile interaction, game study, human-robot interaction, and customer experience management.

Cuiling Chen is an experienced research assistant in the Human Factors and Ergonomics Institute in the Department of Industrial Engineering at Tsinghua University in Beijing.

References

Ahsanullah Mahmood

A. K. B.

Sulaiman

(2010, June). Investigation of fingertip blobs on optical multi-touch screen. Paper presented at the Information Technology (ITSim), 2010 International Symposium.

Balakrishnan

Yeow

P. H. P.

(2008). A study of the effect of thumb sizes on mobile phone texting satisfaction. Journal of Usability Studies, 3(3), 118–128.

Chan

A. H. S.

Lee

P. S. K.

(2005). Effects of different task factors on speed and preferences in Chinese handwriting. Ergonomics, 48(1), 38–54.

CNIS. (1996). Hand sizing system—adult. National standard of People’s Republic of China (Vol. GB/T 16252-1996). Beijing, China: Author.

Epps

Lichman

(2006). A study of hand shape use in tabletop gesture interaction. Paper presented at the CHI ’06 Extended Abstracts on Human Factors in Computing Systems, Montreal, Canada.

Sun

Zhang

Wei

Zhu

(1988). List of frequently-used characters in modern Chinese. Retrieved from http://www.zdic.net/zd/cc1.htm.

Goonetilleke

R. S.

Hoffmann

E. R.

Luximon

(2009). Effects of pen design on drawing and writing performance. Applied Ergonomics, 40(2), 292–301.

Häger-Ross

Schieber

M. H.

(2000). Quantifying the independence of human finger movements: Comparisons of digits, hands, and movement frequencies. The Journal of Neuroscience, 20(22), 8542–8550.

Z. Y.

Tang

Y. Y.

(2004, August). Chinese handwriting-based writer identification by texture analysis. Paper presented at the Machine Learning and Cybernetics, 2004.

10.

You

Tang

Y. Y.

(2008). Writer identification of Chinese handwriting documents using hidden Markov tree model. Pattern Recognition, 41(4), 1295–1307. doi:http://dx.doi.org/10.1016/j.patcog.2007.08.017

11.

Hofstede

G. J.

Minkov

(1991). Cultures and organizations. London: McGraw-Hill.

12.

Holz

Baudisch

(2010). The generalized perceived input point model and how to double touch accuracy by extracting fingerprints. Paper presented at the Proceedings of the 28th International Conference on Human Factors in Computing Systems, Atlanta, GA.

13.

Holz

Baudisch

(2011). Understanding touch. Paper presented at the Proceedings of the 2011 Annual Conference on Human Factors in Computing Systems, Vancouver, Canada.

14.

Huang

D.-L.

Patrick Rau

P.-L.

Liu

(2009). Effects of font size, display resolution and task type on reading Chinese fonts from mobile devices. International Journal of Industrial Ergonomics, 39(1), 81–89.

15.

Hwang

S.-L.

Wang

M.-Y.

Her

C.-C.

(1988). An experimental study of Chinese information displays on VDTs. Human Factors: The Journal of the Human Factors and Ergonomics Society, 30(4), 461–471. doi: 10.1177/001872088803000407

16.

Jin

Wei

(1998). Handwritten Chinese character recognition with directional decomposition cellular features. Journal of Circuits, Systems and Computers, 8(4), 517–524. doi:10.1142/S0218126698000316

17.

Lam

S. S. T.

R. K. C.

Leung

H. W.

Li-Tsang

C. W. P.

(2011). Chinese handwriting performance of primary school children with dyslexia. Research in Developmental Disabilities, 32(5), 1745–1756. doi:http://dx.doi.org/10.1016/j.ridd.2011.03.001

18.

Jin

Zhu

Long

(2008). SCUT-COUCH2008: A comprehensive online unconstrained Chinese handwriting dataset. ICFHR2008, 165–170.

19.

Lowe

A. C.-T.

Corkindale

D. R.

(1998). Differences in “cultural values” and their effects on responses to marketing stimuli: A cross-cultural study between Australians and Chinese from the People’s Republic of China. European Journal of Marketing, 32(9/10), 843–867.

20.

Phillips

J. G.

Ogeil

R. P.

Best

(2009). Motor constancy and the upsizing of handwriting. Human Movement Science, 28(5), 578–587. doi:http://dx.doi.org/10.1016/j.humov.2009.07.004

21.

Poon

K. W.

Li-Tsang

C. W. P.

Weiss

T. P. L.

Rosenblum

(2010). The effect of a computerized visual perception and visual-motor integration training program on improving Chinese handwriting of children with handwriting difficulties. Research in Developmental Disabilities, 31(6), 1552–1560. doi:http://dx.doi.org/10.1016/j.ridd.2010.06.001

22.

Raibert

M. H.

(1977). Motor control and learning by the state space model. Retrieved from dspace.mit.edu/bitstream/handle/1721.1/6927/AITR-439.pdf

23.

Ren

Zhou

(2009). The optimal size of handwriting character input boxes on PDAs. International Journal of Human-Computer Interaction, 25(8), 762–784. doi:10.1080/ 10447310903025511

24.

Srihari

S. N.

Yang

Ball

G. R.

(2007). Offline Chinese handwriting recognition: An assessment of current technology. Frontiers of Computer Science in China, 1(2), 137–155. doi: 10.1007/s11704-007-0015-2

25.

T.-H.

Zhang

T.-W.

Guan

D.-J.

Huang

H.-J.

(2009). Off-line recognition of realistic Chinese handwriting using segmentation-free strategy. Pattern Recognition, 42(1), 167–182. doi:http://dx.doi.org/10.1016/j.patcog.2008.05.012

26.

Tang

Y. Y.

L.-T.

Liu

Lee

S.-W.

Lin

W.-W.

(1998). Off-line recognition of Chinese handwriting by multifeature and multilevel classification. Pattern Analysis and Machine Intelligence, IEEE Transactions, 20(5), 556–561. doi:10.1109/34.682186

27.

Ren

(2012). Optimal entry size of handwritten Chinese characters in touch-based mobile phones. International Journal of Human-Computer Interaction, 29(1), 1–12. doi:10.1080/10447318.2012.668130

28.

van den Broek

E. L.

(2010). Beyond biometrics. Procedia Computer Science, 1(1), 2511–2519. doi:http://dx.doi.org/10.1016/j.procs.2010.04.284

29.

Vinter

Chartrel

(2010). Effects of different types of learning on handwriting movements in young children. Learning and Instruction, 20(6), 476–486.

30.

Wang

Ren

(2009). Empirical evaluation for finger input properties in multi-touch interaction. Paper presented at the Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, Boston, MA.

31.

F.-G.

Luo

(2006). Design and evaluation approach for increasing stability and performance of touch pens in screen handwriting tasks. Applied Ergonomics, 37(3), 319–327.

32.

Yen

N.-S.

Tsai

J.-L.

Chen

P.-L.

Lin

H.-Y.

Chen

A. L. P.

(2011). Effects of typographic variables on eye-movement measures in reading Chinese from a screen. Behaviour & Information Technology, 30(6), 797–808. doi:10.1080/0144929x.2010.523900

33.

Zhang

Braido

Lee

S.-W.

Hefner

Redden

(2005). A normative database of thumb circumduction in vivo: Center of rotation and range of motion. Human Factors: The Journal of the Human Factors and Ergonomics Society, 47(3), 550–561. doi:10.1518/001872005774860069