Testing the efficacy of an Indic script virtual keyboard: Swarachakra Nirav Malsattar
Manjiri Joshi
ABSTRACT
IDC, IIT Bombay
IDC, IIT Bombay
[email protected]
[email protected]
Indian text input is an area which is now being studied by not only language experts but designers and developers. Since a decade there is work in progress for designs that are intended to develop easy and efficient keyboard for different Indian Languages scripts. Swarachakra (for android) is one novel attempt to resolve this problem. In our studies we found that alphabetical keyboard layout performed better than the Inscript layout for Devanagari Script. This case study discusses the evaluation and evolution of Swarachakra as virtual keyboard. It talks about the various degrees of usability testing which has been done to check its efficacy. User group included were students, adults and elder people with literacy level varying from graduate to low literate. The results of the study shows that with minimal training, a user is able to achieve acceptable comfort and speed with the logical design based keyboard.
Nagraj Emmadi IDC, IIT Bombay
[email protected]
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from
[email protected]. IHCI '14, December 07 - 09 2014, New Delhi, India Copyright 2014 ACM 978-1-4503-3218-7/14/12…$15.00 http://dx.doi.org/10.1145/2676702.2677203
160
AUTHOR KEYWORDS Swarachakra; Indic Keyboards; Indian text input system
Virtual
keyboards;
ACM Classification Keywords H.5.m. Information Miscellaneous.
interfaces
and
presentation:
INTRODUCTION
Inscript
India is a multilingual country with 1,635 rationalized languages & dialects [9]. An Indian currency note conveys its value and features 15 major languages. It is suggestive of the importance of each Indian language. However it seems difficult to see similar tendency when we talk of text input in Indian languages. Text input in Indic scripts has been a challenge for many years. If we see literacy level data by Indian Human Development Survey [4], it is clear that literacy rates have risen sharply for all segments of Indian society. As of 2005, 79% of males and 58% of females aged seven and older could read and write a sentence [4]. Due to lack of localized applications or tools for text input there is a digital divide
BACKGROUND In the area of keyboards, different types of layouts exist and they are implemented according to its performance, speed, and complexity. One of them is a logical keyboard layout where keys are arranged in an order that imitates the very basic way of text book learning methods. In logical keyboard, letters are ordered in the same way as one would find in a primary school textbook. Another layout of virtual keyboard is based on frequency. It orders letters by frequency of use so as to allow rapid typing. In the past a lot of research & implementations based on these, have occurred on text input system for Indic scripts [1] [2] [3] [4] [5] [6]. Inscript layout has been accepted as national standard [6].
161
Figure 1: Standard layout of Inscript Keyboard
The standardized inscript layout has the consonants on the right side of the keyboard and vowels on the left side of the keyboard. Consonants and vowels are arranged according to their frequencies. While each dependent vowel and its corresponding independent vowels are plotted on the same key. Disha A logically ordered design which uses touch screens sliding features. Disha use phonetic group of five characters of Devanagari script. First character in the group is placed in center and other four characters are placed around it clockwise starting from the top.
16
30-45
4-7
Disha
17
30-45
4-7
Swarachakra
10
20-28
Graduates
Swarachakra
13
18-27
Graduate
Swarachakra(With Prediction)
(Table 1: Recruited Participates)
Figure 2: Disha Keyboard
Swarachakra One approach towards better Indic language based text input system [7]. It is a logically based text input system. It uses slide gesture to select consonant + vowel combination, which decreases keystroke required to input text in Indian languages. It enables typing a combination of consonants with just two keystrokes. In our study, we tested predictive & non-predictive versions of Swarachakra keyboard. We enlisted different users for each version of Swarachakra.
METHODOLOGY Participants We recruited total 56 for the study. Table 1 elaborates the difference between users in their age and qualification and keyboards on which they were performed the test. No.
Age
Participa
Qualificati
Keyboard
on
nts 16
30-50
4-7
Inscript
yrs
162
Figure 3: Swarachakra Keyboard without Prediction Design Mapping of each keyboard layout inspected thoroughly. Experiment was conducted on Nokia N900 phones, they have resistive touch screen with resolution of 800 x 400 pixels. The entire study was conducted using stylus for text input. For proper result of speed and accuracy they had a freedom two write with any hand. We did the
usability evaluation with Inscript (used as the benchmark), Disha and Swarachakra in two phases. Phase-I On first day we collected user details (Name, Age, Education Qualification, Mobile Usage, Handset used, have they ever typed in Indian language before this on a mobile or a computer) and asked them to rate the difficulty in checking last missed call (1=very easy, 2=easy, 3=average, 4=difficult, 5=very difficult). The actual tasks were rated compared to this rating. To check their reading & writing skills, they were given a paragraph to read aloud first & then write it down. Next, they were introduced to the keyboards and trained to use assigned keyboard with example words (गणपती, अनिकेत, चिन्ह, कार्यशाळा, वऱ्हाड). They were also
asked them to compare difficulty of the task with the missed-call rating on a 1 to 5. Data collection All the typed data was logged into device for future analysis. This method was followed for all 3 different keyboards for different group of users. We analyzed all the data that was recorded and calculated the errors made, typing speeds and difficulty level of the task.
RESULT Two-tailed T-test performed on 10 days written paragraph with different parameters (Speed, Speed with normalized words per minute, Errors, User ratings, User ratings with normalized word per minute).
instructed to “write as you speak” while typing words with conjuncts. The first task was to type given 19 words. These were ordered from easy to hard. These words were chosen so as to cover it cover most of the complex combinations of Devanagari script. User had to speak a word aloud first and then type. (कमल, पवि, अगर, भारत, िदी, रसिका, और, वविर्, अंककत, िााँद, स्टार, प्राण,
अर्य, श्रेर्ािी, शमाय, मस्त, िचिि, मैं, वाऱ्र्ावर). Words shown
one at a time. In the end we asked them to give a difficulty rating to the task compared to the missed-call task. Phase-II From the day 2, we gave them one paragraph every day to type. These paragraphs were chosen from the primary school textbook which covers most complex combinations. These paragraphs were ordered in such a way that first paragraph would be easy and from there till the 10th paragraph complexity of the paragraph increased slowly. After completion of each task, we
163
(Y-axis: Words per Minute Count, X-axis: Paragraph on each day with first day words) Figure 4: Mean Typing Speed of 10 paragraphs for 3 different keyboards and one within prediction.
(Y-axis: number of times, X-axis: Paragraph on each day) Figure 5: Mean Errors in 10 Paragraph using 3 different Keyboards and one within prediction.
Figure 7: Speed and Confidence for First Time Usability Test for 3 Different keyboard and one within prediction.
(Y-axis: User ratings in 1 to 5 scale, X-axis: Paragraph on each day with first day words) Figure 6: Mean user Ratings after each Paragraph for 3 Different keyboard and one within prediction.
Analyzed data and calculated CPM (character per minute) and WPM (word per minute) represent that logically structured keyboard have positive impact while typing in Indic script language.
Figure 8: Errors for First Time Usability Test for 3 Different keyboard and one within prediction.
DISCUSSION By analyzing thoroughly we realized that user who used frequency based keyboard are taking longer time than logical based keyboard. Speed rates of daily session proves that longer the user uses the keyboard the
164
better he/she performs. We believe that this is because of logical design of the keyboards.
In CHI 2004, April 24-29, 2004, ACM Press (2004), 34.
CONCLUSION
[3] Joshi A., Dalvi G., Joshi M., Rashinkar P., and Sarangdhar A. Design and Evaluation of Devnagari Virtual Keyboards for Touch Screen Mobile Phones. MobileHCI 2011, ACM Press (2011), 2-4.
The final results emerge that effect of use of logical structured keyboard for Devanagari Script show improvement in speed, errors, and Likert rates given by user after each session.
[4] Christina L. James and Kelly M. Reischel. Text input for mobile devices: comparing model prediction to actual performance. In CHI '01, ACM Press (2001), 2-4. [5] Lee Butts, and Andy Cockburn. An evaluation of mobile phone text input methods. In AUIC '02, ACM Press (2002), 1-3.
Currently Swarachakra android is available for ten Indian languages Hindi, Marathi, Gujarati, Telugu, Kannada, Malayalam, Oriya, Punjabi, Bengali and Konkani. We are working Swarachakra more Indian languages and also on adding features like prediction, auto-correction.
[6] Sachi Mizobuchi, Mark Chignell, and David Newton. Mobile text entry: relationship between walking speed and text input task difficulty. In MobileHCI '05, ACM Press (2005), 2-5.
ACKNOWLEDGEMENTS
[7] Indian Human Development Survey Report, 2005. http://www.ihds.umd.edu/IHDS_files/06HDinIndia.pdf
The authors would like to thank Prof.Anirudha Joshi for their assistance in conducting this research. In addition we would like to thank anonymous reviewers & others who provided valuable feedback on a previous version of this paper.
REFERENCES [1] Kumar A., Shah H., and Joshi A. Saral: Devanagari Text input system in Mobile phones. In Globicomp September 2009, 1-2. [2] Joshi A., Ganu A., Chand A., Parmar V., and Mathur G. Keylekh: a keyboard for Text entry in Indic Scripts.
165
[8] Standardization of Inscript Keyboard-TDIL Dept. Government of India. http://tdil.mit.gov.in/pdf/keyboard/keyboard%20stand ard.pdf [9] Inscript Keyboard,C-DAC http://en.wikipedia.org/wiki/InScript_keyboard [10] Swarachakra Keyboard http://en.wikipedia.org/wiki/Swarachakra [9] 2011 census general note http://www.censusindia.gov.in/Census_Data_2001/Cen sus_Data_Online/Language/gen_note.html
