You are on page 1of 9

Suhrid: A Collaborative Mobile Phone Interface for Low

Literate People
Syed Ishtiaque Ahmed1, Maruf Hasan Zaber2, Mehrab Bin Morshed2, Md.Habibullah Bin Ismail2,
Dan Cosley1, Steven J. Jackson1
1Department of Information Science, Cornell University, Ithaca, NY, USA 14850
2Department of Computer Science and Engineering, BUET, Dhaka, Bangladesh 1000
{sa738, drc44, sjj54}@cornell.edu, {maruf.zaber.09, mehrab.morshed, bahar61119}@gmail.com
ABSTRACT and audio based commands and content. While these efforts have
The design of accessible mobile phone interfaces for low literate shown benefit, they also face important limits, including the need
people usually assumes an individual model of use, and are often for additional computational support from the device, which may
limited by the low technical expertise and/or cognitive ability of not be available in low-resource environments, and the technical
users in marginal communities of developing countries. Drawing and cognitive abilities of users interacting with icons and audio
on previous ICTD scholarship around shared and intermediated commands. As a result, the design of an effective phone interface
use of technology and our own ethnographic field study, we for low-literate people remains an ongoing challenge.
introduce a collaborative model of use in the design of Suhrid, a Our work is built around a shift in perspective about the use of
mobile phone interface that helps low literate users perform ‘personal’ devices like mobile phones from an individual model to
common phone tasks by receiving remote help from higher- a more communal model, in which users figure not as atomized
literacy members of their community. The results of our six week individuals, but as nodes within wider social networks that can be
long deployment of Suhrid among 10 low literate rickshaw pullers drawn on to overcome barriers that literacy poses to technology
in Dhaka, Bangladesh, indicate the potential of collaborative use use. Prior studies in ICTD have shown that technology use in
models to help low-literate people more effectively use mobile developing and low-resource contexts is frequently collective or
phones while strengthening bonds between them and the people in distributed in nature, with the use of technologies like mobile
their community who provide help. phones shared among the user’s family, friends, and other
community members [4, 27]. Intermediate use of technologies is
Categories and Subject Descriptors similarly prevalent; several studies have shown that low literate
H.5.2 [User Interfaces]: Interaction styles people often take help from digitally literate people close to them
for operating their own mobile phones [23, 28].
General Terms
Human Factors At the center of intermediate use lies the practice of ‘help’ by able
members of the community. We see an opportunity to leverage
this practice through a community-sourced model that connects
Keywords low-literate users to higher-literacy remote peers in their
Collaborative Interface; Low Literate User; Gift Economy
immediate network to both accomplish tasks and strengthen social
bonds. In this paper, we present a design intervention that exploits
1. INTRODUCTION the social values and practices of a community of rickshaw pullers
Making technologies accessible to low-literate users is a long- in Dhaka and their intermediate use of technologies to provide
standing challenge for ICTD researchers and practitioners. The low-literate members access to their basic mobile phone
rapid growth of mobile phone penetration in the developing world operations. Our previous ethnography on the same garage
in the last two decades has driven important change across many informed us of the intermediate use of mobile phones among this
aspects of life, from education and health to political participation community. Based on that and an additional focus group study,
and the informal economy. However, low literacy has limited we designed, developed, and deployed Suhrid, a phone
these impacts, especially among poor and marginal populations; application that allows the rickshaw pullers to remotely get help
therefore helping low-literate users reap the benefits of mobile from their garage owner for placing phone calls and saving
devices is an increasingly important question in ICTD research. contacts. A six-week field deployment of Suhrid with 10 rickshaw
Previous work in this area has attempted to overcome this pullers showed that it effectively helped low-literate users make
problem by using non-textual interfaces that incorporate graphic better use of their phones. More generally, we argue that Suhrid
shows the potential of designs that leverage shared and
Permission to make digital or hard copies of all or part of this work intermediate use in contexts where such use is common.
for personal or classroom use is granted without fee provided that
copies are not made or distributed for profit or commercial 2. RELATED WORK
advantage and that copies bear this notice and the full citation on Literacy has long posed important challenges to mobile phone use
the first page. Copyrights for components of this work owned by among poor and marginal populations. In their 2006 paper,
others than the author(s) must be honored. Abstracting with credit Chipchase et al. documented various ways illiterate people used
is permitted. To copy otherwise, or republish, to post on servers or mobile phones [7]. Their study showed that some functions (e.g.,
to redistribute to lists, requires prior specific permission and/or a turning the phone on or off, accepting incoming calls) were easy
fee. Request permissions from Permissions@acm.org. for the users, while other functions (e.g., sending text messages,
ACM DEV 2015, December 01 - 02, 2015, London, United Kingdom
finding contacts from the contact list) were more difficult.
Copyright is held by the owner/author(s). Publication rights licensed to
ACM. Furthermore, understanding and responding to basic information
ACM 978-1-4503-3490-7/15/11$15.00
DOI: http://dx.doi.org/10.1145/2830629.2830640

working closely with communities can A common characteristic across each of the abovementioned lead to new design opportunities. 13] and 1 Houston [11] with mobile phone repairers in Namibia. Our previous work has shown how illiterate rickshaw. Usually the owners of the workers to communicate with microcredit providers [23. contacts from the address book and placing calls remain. and non-use. sustain mobile phone use in many developing countries. Drawing on Mauss.2 Communal Model of Technology Use community practices. . Burrell’s ethnography in Ghana revealed respect (and thus. Kumar studied had 73 rickshaws. and so on networks and rituals of gift-giving and mutual support as a central [32]. Thus. However. the lower tier of the society. Bangladesh depend on literate social peers to ranged from 300 Taka ($3. have reported the practice of helping in pullers on a half. we review key elements of the context people. a small area in Old discovered that low-literate Indian women take help from field Dhaka with many rickshaw garages. However. both cultural contexts. 3. long-term engagement with local and skills. collaborative use model that underpins our design intervention. and farmers [21]. Bellotti et al.2 US cents. For countries due to the lack of material infrastructures and resources audio interfaces. have argued that design strategies was the assumption that a single user would use designs that leverage the altruistic nature of human behavior the device. This spirit is often hidden interfaces that use less text in order to reduce literacy under the layers of official technical support infrastructures in the requirements. remaining battery power. the garage owner the daily income of a rickshaw puller in a day pullers in Dhaka. that spirit is more visible women [9]. Collectively. domestic laborers [20].or whole-day basis. respect. removing text doesn’t necessarily remove usability problems. embarrassment. 28]. mistakes. A number of mobile handset vendors worked to These studies connect in turn to ideas from Mauss’ classic address these problems by providing audio and visual clues for anthropological work The Gift. and the challenges of setting up new ones [10]. which has shown the centrality of battery alerts. A body of ICTD literature has pointed to cognitive challenges stemming from unfamiliarity or out the limitations of designing technologies in developing misrecognition of hierarchical orderings of information [18]. This assumption leads to designs focused on should not only facilitate an extended access to different services communicating with that single user through graphical signals or through a shared economy model. a growing body of ICTD another central challenge of ICTD research. For example. network Bangladesh. [12. meet) the needs of specific contexts [8]. Parikh and Ghosh have We conducted our study in Kamrangirchar. Further. A more detailed review of such provide gifts and thus strengthen bonds. RESEARCH CONTEXT Rangaswamy et al. and Dhaka has a huge number of et al. forming the starting point for the These approaches face a number of practical challenges. commands and their hierarchy. have pointed to the network of actors beyond individual people (almost all male) who earn their living through rickshaw- users supporting the consumption of mobile phone services in pulling. found a similar sort of shared model of Our current study was done with a community of rickshaw pullers technology use in Mumbai slums [27. especially in from that study to help situate the current work. sources have led us to think differently about the site and nature of mobile phone ‘use’. and 2. incoming text messages) was often challenging. revealed only once those supports fall to support low-literate or illiterate populations of Indian village short [26]. and televisions were shared among the members of a community [4]. problems related to operations like finding feature of social life across a wide range of cultural contexts [16]. Besides the difficulty of finding icons that In addition to collaborative use. and trust A common design response to problems of literacy is to develop among the members of a community. such designs depend on the individual user’s memory technologies [2]. Bangladesh whom we studied through a six-month In the case of phones. communities through ethnographic techniques is required to understand the social infrastructure. asking for help in these workers [30] and Indian farmers [25]. these design work can be found in Medhi [17]. Here. hierarchical presentation of icons may fail due practically and ethically. For instance. network connectivity. This sort of long-term engagement points to Individual use is not the only use model. but also often used with the help of others. rickshaw garages buy the rickshaws and rent those to the rickshaw Sambasivan et al.801) to 800 Taka ($10). Work by Jackson et al. 24]. and Uganda has shown the significance and extent of connectivity. how technologies like land-phones. we argue that the essence of gift embedded in 2. based interfaces. we are also committed to make sense to low-literate users across a range of social and collaborative design with specific target communities. cultural values. Such speech interfaces are also involving the existing social infrastructure and community more error-prone than touch or graphical interfaces [25]. the local and global networks of materials and knowledge that These problems often led to confusion. text messages. This communal model extends to the systems of maintenance and repair by which devices and wider infrastructures are sustained in low-income environments. practices may transfer some responsibilities from technologies to the communities. icon-based interfaces have been used developed Western world. the users still need to remember the audio [6].about the phone (e. but also create fields for newer audio clues. Here. 1 Taka is equivalent to approximately 1. the devices are not only shared between ethnography [1]. which has a long work has demonstrated that social use is common in low-resource history of short term research projects that have often failed to contexts. while audio in developing countries through the shared and intermediate interfaces have been developed for low-literate Pakistani health practices around technologies.g. the particular garage that we informal setting for operating mobile phones in India [29]. and this is in support and access basic mobile phone operations [1].. the case of low-literate individuals. computers. in Dhaka. This livelihood is not a wealthy one: after paying rent to India [15]. Without such infrastructure. For icon. Color has also been used contexts may be less of a burden or something to be avoided and with some success to help low-literate users with address books more an integral part of community practice that allows people to [31] and phone contact lists [14].1 Designing for Individual Phone Use technical help produces and reciprocates honor.

used punctuation symbols to remember the contact associated 4. they would often get recorded with their permission. but they could not read numbers when two or more digits were put together. and later transcribed in Bangla help from their garage owner. or missed call lists. They shared their difficulties in finding a support for the pullers in using their phones [1]. A focus group discussion at the rickshaw garage.1 Focus Group and Design Goals mobile phones that our participants were using. and along with the garage owner. then how the design and phone application that could help the rickshaw pullers to remotely implementation of Suhrid supports them. these rickshaw drivers are a largely low literate population. We showed our own smartphones to them.. while the rest never went to a school. when the rickshaw pullers were usually back from their devices. and although they participation in this round of the study. They said getting help from their social connections for two basic operations they usually took help from their garage owner for these tasks. They also wasted money study them. and speak Bangla find ways to save a new contact on their phone. The discussions were audio these often led to frustration. They were familiar with Bangla and English digits. garage at those times and were interested in discussing joined. He also arranged contact from their contact list. were supplied during our discussions. get that help from the garage owner would help them. The rest agreed. one of the rickshaw pullers suggested that many of in each discussion. We shared that designing on a smartphone would be almost always at least six pullers present in these discussions. led the garage owner to be a primary receiving phone calls. His literacy and technical All of our rickshaw puller participants reported that the two main knowledge. however. These while resting. Sometimes they like the rickshaw pullers. Since most of We conducted four focus group discussions at the garage in the them were not using smartphones. owned several models of smartphones and had discussion among the team and the main themes were extracted. used a variety of other phones. Most of those mobile phones were Java-enabled China-made devices costing between 2. he had basic literacy skills. outlets. Suhrid with a given number (e. The number and identity of participants were not the same However. “# is for Mr. and patterns of their easy for the rickshaw pullers because reading and writing contact mobile phone use. Participants were not paid for their names and numbers requires literacy skills.g. They all expressed .000 Taka ($120). Then they became embarrassed after talking the fit with our condition of low-literacy. work. easier for us. An immediate design challenge was the functionality of the 4. Mitra”). Our previous study revealed that all of the rickshaw pullers had their own mobile phones. and they ended up choosing the We chose this particular rickshaw puller community because of wrong person to call. All the members of our team who worked in the field when they placed phone calls to wrong numbers. mirroring general trends of increased phone use among lower income residents of Dhaka over the last five years [5]. and sometimes they decided not to place a social structure of help giving. However. and so on. and our convenience to reach and phone call to avoid this embarrassment. None of them could read or write a complete sentence in any language. saving a contact. Most of the rickshaw pullers dropped out of primary school. processes often did not work well. All of The rickshaw pullers also reported that they often struggled to them were born and brought up in Bangladesh. ## is for Mr. Basic functions such as making calls and saving contacts were not pullers questions around the length. the to the wrong receiver. the use of phones. received. (The picture is taken and savings of about the 50% of the original prices. there were or relatives. often this was because they had to help support their families [22]. We turn now to the design of Suhrid (Bangla for ‘a good friend’). They said they often tried to for electric power supply. Thus. food and drink tried workarounds described in the design goals section below. sometimes they tried to garage so that the rickshaw pullers could recharge their phones remember the position of a contact in their list. they informed us how they struggled when they were constraints that arose from prior work and our own interactions away from their garage. combined with the social and economic connections tasks that they did with their mobile phones were placing and between him and the pullers. We first discuss the major design goals and However. and phone adapters in the memorize the contact numbers by face. the use of mobile phones is very common among the rickshaw pullers. They also found it difficult to save a an application to support the low-literate rickshaw pullers in number from their dialed. The ($25) to 10. were affiliated with a local university close to the garage. sharing their experiences of using smartphones owned by friends leaving in the middle if they had work to do. and they used a variety of evening. We discussed with the rickshaw briefly gave them an idea about the interface. it was difficult for us to design a single interface. on their mobile phones: 1) placing a phone call to a contact and 2) The garage owner also agreed that he often helped them for these. rickshaw pullers who were present in the them were thinking to buy smartphones soon. The rickshaw shared with proper permission of the people in the picture. Although the garage owner was not and translated into English. purpose. but then later forgot. All the participants opined that a mobile with the garage owner and pullers. Choudhuri. their findings were matched in a group tech-savvy person.000 Taka Figure 1. Two members of our team coded the educated in formal schools. Many of them bought their mobile garage owner (wearing a cap) participated with seven other phones from second-hand markets or mobile repair shops at a rickshaw pullers in that session. and as a discussions independently. pullers reported that they would spend around 20-30 Taka (25-35 The faces are blurred for anonymity) cents) per day using their phone. Still. In practice. the rickshaw pullers are also in a lower educational range in society.On average.

then through Suhrid. The top icon was for the seeker was calling the helper through this application. c) dialed numbers. and an both cases the goal was to minimize the number of elements option to add a new contact to the contact list. The Suhrid would show a list with five entries: a) contact list. If the This way. the rickshaw pullers reach their helpers when they were away Hence. (versus an anonymous crowdsourcing model). place a call or save a contact. d) received calls. For seeking help. The rickshaw pullers chose would be sent from the seeker’s phone to the helper’s phone. the low-literate people could place a phone call to helper chose any of these. missed.3 Functions be minimal. We asked them if smartphones would be affordable rest of the number to increase the pullers’ confidence that the for them. the pairing up could be done by adding a helper on the seeker’s 4. their excitement about having an application for them in a smartphone. right). and some cheap Chinese-branded smartphones in the market. If it the icons during our group discussions. phones that they were using. and e) an option call from his phone after getting a confirmation from the seeker. b) seeker’s application would receive the number and place a phone missed calls. the list of seekers subscribed to this helper. the rickshaw pullers needed the most: placing calls and saving contacts. and argued that the differences would 4. We used comic fonts in that interface to give that 4. The garage owner was also we decided to design an Android phone application for helping interested in helping the rickshaw pullers because he knew them. (Figure 2. The seeker could request left the last two digits visible to facilitate referencing the correct to save a number from their dialed. lists of missed. On the right. (Left) The interface for selecting if they wanted to left. After getting . Based on their recommendation. and missed calls. right). Both our own work and previous studies show that low. Figure 2. the low-literate user needed to open Suhrid on literate people often get confused with too many icons in the their mobile phone. left). to add a new contact to the contact list (Figure 3. contact list of one of the seekers. or received calls’ lists. On the pullers. After choosing one of was not the first time. They further argued that the better quality of the 4. It also 4. Touching a helper’s name would If the seeker requested help for placing a call to somebody from place a call from the seeker’s phone to that helper’s phone their contact list or from dialed. we chose a system that paired people who knew each other from them. received. or missed call logs. In Suhrid. The helper would choose that contact and the subscribed to that helper (Figure 3. dialed.3. he could see the corresponding list.2 Placing calls an informal and friendly look. left). The first version of UI for low-literate rickshaw Figure 3. phone numbers between the seekers and the helpers. but hid the or could just tell the helper a phone number to save with a particular name. Based on that the helper would do this task when they were co-present with our findings.1 Seeking help pictures and video on a smartphone would rationalize the We found in our field study that rickshaw pullers would only get additional cost they would bear. Hence. application would send that number to the seeker’s phone. On choosing any seeker. So. we assumed (“seekers”) and one for the help providers (“helpers”). outgoing. and choose the helper from the list. the on the screen and the need for literacy skills. we emphasized the two basic phone operations that a seeker they agreed to pair up with. In the middle. middle). and received calls. the helper could easily find that on his interface and find the On the helper’s side. select whether to place a call or save a interface [18]. our previous study showed that some rickshaw pullers were concerned about the privacy of their contacts [1]. (Right) The list of helpers.3. or just simply type it on the software screen. Three screenshots of the helper side interface.3. then the system only sent changes of these these two icons. help from the people they knew. then the sending a request to place a call and the bottom icon was for list of incoming. the next screen appears with the list of helpers lists since the last time they contacted that helper. If it was the first time interface with only two icons (Figure 2. Suhrid starts with an contact. anybody they wanted. we that could also be done in a similar way. They showed us the costs of some of the Java enabled numbers would be private (Figure 3. However. Since this Suhrid has two interfaces: one for the low-literate users task needed some competency with the mobile phone. Suhrid displayed the list of seekers appropriate name. The helper could find the number in one of the lists.3 Saving contacts reported that the rickshaw pullers would often remember the If the seeker asked help for saving a contact to their contact list contacts by the last three digits of their phone numbers. In the contact list. on the seeker’s side. showing only the last two digits of the phone number to address privacy concerns. and the contact list sending a request to save a contact.2 Interfaces interface and adding a seeker in the helper’s interface.

the contact list to the helper’s phone. To mimic the real-life scenario. The average time they took to place a name into their local copy of the helper’s contact list. Java. Five of them could perform both of the tasks in the first contact’. message down to several text messages for sending. verbally with the participants and that other rickshaw pullers present in the garage did not help them either. their improved. So. with our participants. the participants were allowed to practice the This time. lack of confidence. communication between the mobile phones was done through text After the tasks. However. numbers saved in their phonebook. First. select the helper. Third. Like before each of them attempted the test thrice. Then the satisfactorily in the lab. USABILITY EVALUATION week. apparently they 4. The session Each participant was asked to generate three phone calls to was two hours long. and the average time for saving a contact was would then send a message from the helper’s phone to the 1 minute. First Round Although the rickshaw pullers and the garage owner performed after another. and re-construct one single message. The remaining would be saved in both phones. the application would receive these messages one 5. interactive session. and each of them was paid 800 Taka ($10). This observation matches with Medhi et al. and expressed their satisfaction in using application would parse that message to extract each contact and the application.e. We invited the rickshaw pullers and the garage owner to visit our university in order to introduce them to our system formally and 5.. when we explained the The application was built for Android-based smartphones using task again. For example. two trials. struggling with differentiating the functions of two owner through Suhrid for help in placing a phone call and saving graphical objects. they started using the laboratory so everyone could use one.2. interested to see the laboratory and how we work.2 Field Level User Study. We then asked the garage owner about his experience. the new contact complete both of these tasks after the reminder.’s claim . talk our laboratory. which the software would intercept and use to update the helped them remember. As before. so our team members seeker’s. call test was mainly due to their confusion with the two icons in All 12 rickshaw puller participants were told to ask the garage the interface. they could perform both tasks in the first attempt. we conducted one laboratory study and of the 12 participants who attended our demonstration session. but the test they explained that the low success rate in contrast to the could communicate with them through our application. two field level user studies to understand the usability of this We tried to replicate a similar situation of help-seeking there. Five rickshaw pullers succeeded in all three trials. Two participants failed in all three attempts. the garage owner. This way.2 Contact Saving smartphones.1 Laboratory Study each participant the average income of a whole day. when we assured behind this session. and the system needed to break the that he enjoyed the whole process. We made necessary changes to Suhrid based on our made sure that the garage owner/helper couldn’t communicate findings in these studies. Second. we wanted to understand if that would be make a contact list for the seeker in the helper’s phone. On the helper’s phone. There were three reasons phones by their inexperienced use. Four of them forgot the process.) very quickly.4 Implementation misunderstood our instructions. turning a phone on or off. the contacts’ names and mobile phone numbers one after another. Suhrid call was 40 seconds. we had enough smartphones in our them that no such thing would happen. one for ‘calling’ and the other for ‘saving a a contact. The average time for successful call smartphone to play with and the undergraduate. Participants picked up the basic operations Next. helped them. and praised the software. However. person needed instruction one more time. We application. Three of these four participants could contact list on the seeker’s phone. He said This message was often long. All of them expressed their excitement around it. but succeeded in believed that this session helped us develop a better relationship the second two trials. we paid 5. and receive the response from the help in this session. which the rickshaw pullers considered good enough.the number. Each group was given a reasons for this delay. Both the participants and we considered this Bangla and lasted for three hours. After he could not verbally communicate with the participants.2. who was an generation was eventually reduced to 30-35 seconds in the next expert smart phone user. was taken to another room so that three attempts. the helper could save the contact with the appropriate attempt without any help. We first lectured them about time much longer than the usual time one takes to place a phone the use of a smart phone and then we showed them its basic call to somebody. we asked them about their general impression messages. However. when the seeker’s phone needed to send about Suhrid. We then demonstrated Suhrid. reflected in the context they spend most of their time. The first trial of call generation took on average 45-50 seconds A total of 12 rickshaw pullers and the garage owner came to visit including the time to press the call button. inexperience with touchscreens. we The other five made mistakes in the first trial. and participants into four groups. i.1 Call Generation to see if they had any difficulties in using it in the lab. the next 5. The demonstration session was conducted in helper’s application. number or any unsaved contact number from their call history etc. the application on the They said the reason why some of them had initial difficulties was seeker’s phone would compose a formatted text message putting because both smartphones and the application were new to them. To provide them first-hand experiences we divided our feedback. only three of them were successful in all helper. Three of them hesitated in the which was equivalent to a puller’s earnings for their whole day beginning as they thought they might break the expensive smart and also satisfied the garage owner. After that using Suhrid. their confidence using the touch screen seemed to have use of our application. we asked each of our participants to save a random contact (touching instead of pressing buttons. each consisting of three rickshaw confusion of the “save” icon with the “call” icon were the main pullers and one undergraduate. we conducted a field level user study at their garage with 10 After developing Suhrid. First we briefed them how to use the “touch” action to operate the 5. We also invited four undergraduate students to to the helper over the voice call. they were smartphone. The other three participants performed the calling task twice. According to our observations and their features. To avoid the need for an Internet connection.

and task that In both of the focus group discussions we asked them about their we did in the first round. Those questions included if they had successful in all three attempts on both tasks this time.2 Availability applications such as VizWiz [3] enable people to get help from Suhrid also provided the seekers valuable services and help.” appreciation it gave them for the giving of help between them. a interviews were audio recorded with the participant’s permission call would be generated to the selected helper. coding. and was willing to participate in the The helpers also found Suhrid easy to use. I called the Mahajan and asked him to dial the pay when they would seek help from these freelancers.” freelancer was assigned to be the third and final helper for half of the rickshaw pullers. cognitive load associated with graphics [17. One of interest to participate in our study. The average length of the interviews then tell the helper whether he was trying to call somebody or was 30 minutes and participants were given 200 Taka ($2. It was an rickshaw pullers were clearly informed that they would have to emergency. On selecting the helper. Ten them said. Finally. setup.5) each save a contact. we invited the rickshaw pullers to our through Suhrid (Figure 3. After talking to both the rickshaw pullers and the freelance “I was at the hospital for a checkup of my father and I needed to workers. The helper then would use the appropriate lists as compensation. Second Round helper. See? Here is the icon I had to press to call the helpers. payment.1 Ease The rickshaw puller participants were chosen based on their All seekers were satisfied with the ease of using Suhrid. middle) to find the number. user study at the rickshaw garage in Kamrangirchar two weeks Instead. Another person saved the number. its value and availability. received call and their phone book. . smartphones were given to the pullers and two other phones were “This software is very easy to use. Mahajan. but also for every low literate the ease of using the software. To do this. Upon opening the app. I didn’t find any difficulty given to the garage owner and his brother who would work as the while using it. Participants described if I use this phone. He would often come to the to the garage owner. Here is the list and I can select any of them and see owner at the top followed by the number of his brother. One participant commented. as well as the gradual decline of After the success in the second round of field usability testing. Second. The seeker would and conducted in Bangla. This was both a focus group selecting the number. students sometimes at critical times. I remembered Taka (3 cents) each time they asked the freelancers for help. DEPLOYMENT help from the freelance workers. which lasted six weeks. in the way that some crowdsourcing 6. each of the rickshaw pullers had the number of the garage functionality. We were curious about whether It’s very easy to use. we responded to these problems by participants to collect these data. translation. list. One of them ran out of credit in Suhrid. They also expressed concerns about privacy and reluctance to take 6. we had a separate group discussion with the garage owner after the first round. Here he referred owner’s brother as a second helper. and resolution number from the call log. Eight out of 10 participants were experiences of using Suhrid. so we repeated that test and the processes were the same in this round as before. The the name by which he saved it but couldn’t find it.1 Results “Just one click and some other person does the job from another The interviews and focus group discussions revealed a number of place! I don’t have to do anything!! It will be a great help for me important aspects around the use of Suhrid. Each number for me. it was decided that the rickshaw pullers would pay 2 call a doctor. and his brother at the garage. All participants kept the smartphone as compensation.” people would take help from strangers when their normal helpers weren’t available. rickshaw puller succeeded. we recruited two freelance workers.] garage and help them in operating their mobile phones the same way the garage owner did. Upon university guest room one more time.1. The garage owner said. As one participant said. The lists also had a third number. what their the middle of the process while the other one asked the helper to experiences were around the freelance workers. and if they had any other issues with phonebook. The transcription. we collected three main sources of data. He did and I talked with the doctor. decided Suhrid was ready for a real deployment to understand better how Suhrid would help this rickshaw puller community. We received permission from our In our next round of design. Not only for me. We used the same 10 participants. The overall remarks of the participants were overwhelmingly positive. First. These appear immediately (Figure 2. at the end of six weeks. All of the either making a phone call or saving that number in the contact participants were paid 800 Taka ($10) for this two-hour session. why their usage of help him call to a number that had not actually been saved in his Suhrid declined over time. Suhrid would offer the helper an option for discussion and a formal conclusion of the study. and the time. in which situations they used two failed in in one attempt each. of a local university from outside the rickshaw puller community. and then conducted another round of field level allow the rickshaw pullers to express their experiences freely. right). the list of the missed call. strangers. 5. removing the initial interface for choosing a service on the we went to the garage and interviewed each of our participants seeker’s UI. So. 6.1.about the weakness of low-literate people in handling the To evaluate Suhrid. we collected usage data through a text message Suhrid sent each time a rickshaw puller asked for help that included the seeker. And if the Mahajan is not available I press the second The garage owner and the rickshaw pullers recommended the number. the list of helpers would about their experiences throughout the deployment period. The other any difficulties in using Suhrid. 19]. we their usage of Suhrid over time. in the helpers’ list on “When you first gave me the software I understood the full Suhrid.” [Mahajan means the ‘The big person’. We made the necessary changes in the application according We did not invite the garage owner and his friend this time to to this design. if they got help whenever they needed. In one case the helper mistakenly saved a different Suhrid. 6. and the and illiterate people will be happy to use this application. The second number was his brother’s. …. deployment and able to operate Suhrid.3 Field Level User Study.

1. In total. All of them said that they did not need to seeker wondered about the privacy of the contacts saying. Furthermore. So. One participant said. One of them said. He was always my last option for help. first trying to them complained that our software damaged his mobile balance. most of the time that garage together. I was confused. Now they know better how much I care for them. “Now I Second. suggesting the value of minimizing hierarchy in the with him. ask him to save a contact for me. Sometimes it takes time but work gets done. they still often took “Normally when I get out of home to go to Mahajan I forget by help from the garage owner in person when both were in the the time for which reason I went there. “yes”. number or a save a new contact. only five cases. We asked graphical interfaces [18]. It was only regarding help.5 Gradual Decline of Usage Despite the value people drew from using Suhrid. I never asked him how he was wife. may indicate that low- literate users might be better at memorizing the relative positions All of the rickshaw pullers reported that the availability of their of contacts than the interpretation of symbolic icons. doing.3 Reluctance to Take Help from Freelancers Later.” (500 Taka is approx. And how did he save contact numbers in comfortable asking help to somebody I do not know.4 Improving Communal Bonds literate people. then the owner’s brother. “It is great to come numbers to identify contacts had to be respected in our design: . saying. The concentration of help First. One of Seekers almost always got help from their helpers. and the helpers reported that using Suhrid with people they knew strengthened their community feelings.” 6. and the freelancers received 3. You cannot help in designing shared-use interfaces. requests was high. Participants informed us that usage declined because initially they Figure 4. we discovered he had not turned off his mobile phone after The seekers expressed their lack of interest in receiving help from calling.” 7. You can see the added respect in their words these days. which they learned to call themselves after taking help once or twice. He said.1. they reiterated that help should come “I didn’t understand how the garage owner was able to call my from people they knew. both reported that they had enjoyed their task. “I don’t feel wife through my phone. they stopped trying. He seemed to be satisfied with our explanation. so the number of of Suhrid was higher in the first few weeks.1. and that caused a big deduction from his balance. owner either to know the phone number or to place a call to his “Although I took help from him. Daily usage of Suhrid during field deployment. I only called him for help nothing else. they often call the same numbers. Use had to take more help to save the new contacts. our study supports previous findings around the struggle of requests on the garage owner and the brother was not a concern to low-literate users with hierarchical presentation of information on them. his brother received 14. the rest of the participants agreed. development. then declined. Also. they took help from the freelance workers. However. get help from their garage owner. and the time and efforts it needed to help one was not interface. garage owner all the time helped them realize how much care the garage owner had for them. The number of requests sought for placing a call was much higher (41) than for saving a number (22).” 6. Suhrid was mainly useful when the puller and doesn’t help me. In the finished talking to him. I noticed after some time that my balance remaining two cases. he also mentioned that he could two icons on the selection screen. call the freelancers because they had almost always got help from their garage owner. closer. navigate through the list of three helpers. One rickshaw puller said.6 Concerns The seekers also shared several concerns around Suhrid. it was used 63 times over the span of six weeks (Figure 4).” Once he my phone remotely? Was the number saved in his phone too?” said that. The confusion that arose in the first field them if the number of incoming helping requests ever bothered study with the selection interface went away with the removal of them. both the seekers design in the context of developing countries. The seekers shared how Suhrid helped them by making the helper Then they would only take help whenever they had to call a new available in the contexts they needed. $6) 6. Another the freelance workers. In He said. and to make them understand how a well-wisher should be. they could not get help from either of them. The pullers improved because of Suhrid because now they were taking fact that rickshaw pullers used the last three digits of phone help even after leaving the garage. The fact that the rickshaw pullers struggled with even overwhelming to him. But with this software I can easily call him and owner were not in the same place. In “I had 50 taka in my mobile when I called Mahajan but when I three such cases. One We explained to him why it would not be possible for the garage seeker who took help from one of our freelance workers said. and some larger lessons pertinent to problems of In contrast to their experience with freelancers. Contact information can be somebody like this if you do not care for them a lot. Also. and deployment of Suhrid generate a number of immediate lessons for better designing a UI for low- 6.” The garage confidential and the privacy associated with different parts of owner said that the relationship between him and the rickshaw contact information is dependent on the users’ interpretation. but were able to sequentially not respond to couple of calls as he was busy in saying prayers. The garage owner said that he always had his mobile phone that screen. Among the 63 requests.1. is zero. DISCUSSION The design. the garage owner received 46. its use declined over time. our study suggests that privacy management is important realize how caring he (the garage owner) is.

We thank Suhrid performed a secondary but no less important role in the Department of Computer Science and Engineering (CSE) of strengthening and reinforcing local social relations. [16]. was around 70 to 90 paisa. the rickshaw pullers praised the garage owner for helping them in important 9. The garage owner Furthermore. They also demonstrate the kind of long-term commitment to “A rickshaw puller is not as poor as you think. and cost-benefit negotiation in extending the reach of help strengthened the relationship between designing such collaborative interfaces in developing countries. we ground our designs in the values. that may extend the depth. Third. design interventions in longer-term programs of ethnography and . our study demonstrates the importance of anchoring Urban Bangladesh. This particular finding may suggest using the Android platform. ACKNOWLEDGEMENT times through Suhrid. I usually spend places and individuals that may serve as a partial antidote to the 500-600 Taka for my mobile phone every month. them. freelance workers for their support during this study. Thus. A related sensitivity concerned asking for help from and practices of a particular community through our engaged field people outside the community. along with a range of ideas towards solutions. they mentioned how a low literate rickshaw puller community in Dhaka. We involved the members of the community in and empower individual users over time. Morshed. This finding echoes in modest form Mauss’ Fulbright Science and Technology Fellowship and Intel Science classic finding around the importance of mutual aid and gift and Technology Center for Social Computing (ISTC*SC) for giving as a central and indeed constitutive moment of social life supporting this research. in general responsive models of community engagement may limit the participants would need to pay for Suhrid themselves. we deployed likely decline over time as well. and 1-2 Taka is approx. M. or communicating through text messages came future researchers to weigh local sensitivities before advancing out of the users’ values and practice. design options that leverage and extend core [1] Ahmed. we ‘scaling up’ of single technologies to wider social and cultural asked our participants if this expenditure would be heavy for contexts. We about one US cent. Suhrid seeks to leverage and only needed help for the new contacts. S. While deployment of our community sourced mobile phone interface for explaining the gradual decline of the usage. M.” (500-600 Taka is approx. remote help. For communication. In Proc. norms. for the seeker and 30 to 40 paisa for the are now working with them towards technology solutions that helper. And I would be kinds of “research tourism” [8] long identified and criticized in happy to spend 1-2 Taka if I could get my purpose served with ICTD work. beyond its immediately instrumental We thank the garage owner. and the effects in extending effective use among low-literate populations. of privacy. notably Bangladesh University of Engineering and Technology (BUET) relations of respect and trust between the garage owner and for providing us laboratory facilities. This finding overcome literacy-based barriers to mobile phone use within our suggests that such a help system might also eventually educate target population. and Afrose. our study indicates a successes and limitations. Suhrid for 6 weeks and conducted a post-deployment user study through interviews and focus group discussions to understand its Beyond the lessons for immediate design. development and learning the contacts while taking help from Suhrid. Jackson. and the rickshaw drivers.I. we present our arguments supporting the potential expressed his satisfaction to be able to come closer to the of gift-based design and long-term engagement in ICTD works. Likewise. they suggest new opportunities to design multiple. For example. We also thank International rickshaw pullers. and encourage of phone numbers.showing only two digits balanced the needs for communicating community engagement. according to their cost-benefit analysis. 2 to 3 cents) 8.. and opened up Suhrid demonstrates how cost influences design choices. like many other technologies for developing countries. One participant said. impact. or determine and remember saving from their everyday income. some rickshaw pullers joined a co-operative society typical use of Suhrid. Unlike emphasizing scaling up ‘one size between helpers and seekers while respecting seekers’ perceptions fits all’ technologies. The opportunities for designing technologies with their participation. power differences between the designers and users. Ecologies of Use and prominently in the field. features and principles of sociality itself ought to figure more Ismail.B. $6-$7. The credit in the balance that was spent for a example.. our study showed that lessons the use of icons. Building on they could recognize the old contacts from their memory and they our ongoing ethnographic work.B. rickshaw pullers because of Suhrid. While such long-term and locally text messages Suhrid generated during the deployment.. M. S. privacy. including both the call itself and the text where they needed to pay regularly. This helped us address many nuances of design that would preferred to get help from another person in their own community be otherwise difficult to find out. Although we reimbursed participants in the end for all the may help with this goal. decisions like over somebody from outside. and conducted two rounds of help-seeking also suggests that their cost for using Suhrid would field level user studies to refine our design.J. If the goal of ICTD work is to empower and support not only users but also the wider networks and communities of which 10. Our study generates a number of number of bigger concerns for ICTD. CONCLUSION Fourth. and they needed help to messages sent in the background. Thus. This gradual decline in each step of our design process. the rickshaw pullers and the garage owner. Design: Individual and Social Uses of Mobile Phones Within Low-Literate Rickshaw-Puller Communities in Second. They all considered this expenditure justified for the appropriate technologies for the same community—opportunities service.H. crowd-based responses to local use challenges in such contexts. we have described the design. and responsibility of ICTD work. REFERENCES they are a part. we observed that the low-literate rickshaw pullers started In this paper. community. Zaber. They said they could extend existing distributed practices of technology use to recognize the faces of the contact names or numbers. First. 2013.. our deep engagement with the community helped relax the Third. 14:1–14:10. revealing only two digits limits to more generic crowd-sourced solutions. DEV-4. Finally. S. We found our participants work. rationalized The rickshaw pullers often discussed with us other problems of the choices of the Android platform and text-based their life. remote collaboration.

2015. 2012. M. 2007. 2 System for Low-Literate Farmers.. [14] Joshi. CSCW’15. ACM. KrishiPustak: A Social Networking Journal of Computer Mediated Communication. Jugaad. (2010). semi-literate users. Kanitkar. 5.. Present and Future. Available at: 2013. Text-free user [3] Bigham. ICTD’07. 2813–2822. [4] Burrell. and Sheikh.. O’Neill. and Toyama. 2006. 13. and Parikh.J. A. K.. J. [23] Parikh. E. 15. K. Ali. Behaviour & Information Novice Users: Past. Ferreira. JP.S. Avaaj Otalo: a field study of an interactive voice [10] Heimerl.L. 2863–2872. Prasad. 1670–1681.K.. M. 51-65. S. Assessing Working Conditions and Livelihoods. N. and Karim. N. Sustainable. 2006. Parikh. Lakhmanan. K. Lancaster University. In Proc. [25] Patel. 739–748. Grinter. S.S.. Toyama. 2013. 2. In Proc. N. M. 32–39.info/ on hierarchical user interface navigation. ICTD’06.. N. Intermediated technology use in developing Human computer interaction with mobile devices and communities. 1. [8] Dearden. 1–9. 824–825. 1. 18. J. and International Development. VM. In Proc of Mobile Phone. and technology for [28] Rangaswamy. and Brewer. and Rifat. community. ACM. R. Some evidence for the impact of limited education http://www. Pompe. 3. music. Computing within Limits and ICTD. 333–342. materiality. S. 873–882.. J. 15. A. 2975– computers. Hasan. 2009. 2015. A. [32] Global Accessibility Reporting Initiative (GARI): Mobile [18] Medhi. 2984. 2007. 2. forum for small farmers in rural India. 2014.. Rangoli: a visual phonebook for low-literate users. [15] Kumar. 68–82. R. 2003. K. User Interface Design for Low-Literate and learning and retention. A. [5] Chen. T.. 1999. Parikh.. In Proc. The gift: Forms and functions of low-literate users. I. CHI’13. Learning.. A. and and the circulation of mobile phones in rural Uganda.. repair. T. 230–250. and sustainability among mobile phone store ecology in a Mumbai slum community: Hybrid repairers in Dhaka. M. Toyama. and Krieshok. 2008. and Nair. CHI’13. considerations for a financial management system for rural.. IG Cunnison. and Here Pheri: towards UbiComp for a global iConference’11. Design the Physically Handicapped.J. T. In Proc. In University. T. 11. G.. Evaluating Shared Access: social equality [21] Medhi-Thies. Things International Development. Memon. I. A. K. 905–914. you can’t read or write? Interactions. In Proc.. N. Rickshaw Cycle Drivers in Dhaka: Fast Start for Mobile Financial Services.. Welankar. Chittamuru. K. services. E. T. Cutrell. Optimal support for peer-to-peer service exchange: rethinking the audio-visual representations for illiterate users of timebanking metaphor. T..gari. Trends in Human-Computer Interaction. Mirza. 16–17. education initiatives in rural Namibia.[2] Bellotti. CHI’10. D. 2015. and Parikh. J. J. 6 (2006). DIS’14. I. In ACM SIGCAPH Computers and [9] Ghosh. 2. Proc. A. 2015. Bangladesh. 2014. E. 2006. 1-72. 2010. 1954. and Sambasivan. Handset/Device Accessibility Report. 107–116. T. Information Technologies & [26] Poole. 553–564. 217–223. CHI'14. end user application program: an empirical study of [17] Medhi. W. L. In Proc. 2012. Ahmed. PhD Thesis. Design ICTD. A Longitudinal Study of Local. 2003. P. N. Sagar. Erasmus [6] Chen. K. See no evil?: ethics in an interventionist [24] Parikh. Edwards. I. A. S. Cutting Chai. Cutrell. Small. and Nardi. [20] Medhi. 2010. . S. In Proc. Ali. exchange in archaic societies. Chetty. Foundations and Technology. 2583–2592. Understanding and [7] Chipchase..R.E. Morgan. In Proc. 1-17. 72–82. Mobiles. In Proc.. N. K. to visual questions.. IEEE Pervasive Computing. Jain.. Information Technologies and [13] Jackson. P.. WWW’07. Towards community-centered [19] Medhi. Gupta. In Proc. 8.. World Bank. S. Y. ACM. In Proc CHI EA’03.L. and Rasmussen. A. Proceedings of the 10th International conference on 2010. and Ghosh. In Proc. In [29] Sambasivan. R.. 2010. London: [31] Wiedenbeck. [12] Jackson. and Chavan. and Chavan. The use of icons and labels in an Cohen and West. 1-19. N. E. 2. The mobile phone innovation. Information Technologies & International studies for a financial management system for micro-credit Development 9. and Toyama. M. N.I. CHI’09.S. 2013. Tongia. 2010. CHI’10. B. LIMITS’15.. 2014. and Rosenfeld. 6. I. VizWiz: nearly real-time answers interfaces for illiterate and semi-literate users. Inventive Infrastructure: An Exploration and motivations for informal technical support.. and Cutrell.S. S. T. K. In Proc. Personal and Ubiquitous Computing. Computer help at home: methods [11] Houston. R. [27] Rangaswamy. Scale Cellular Networks. fall apart: maintenance. Fatma. HealthLine: Speech-based access to health information by [16] Mauss.. et al. groups in rural India. bKash Bangladesh : A [22] Meheri. How do you manage your contacts if designing for intermediated information tasks in India. G.. 733–742. UIST'10. K. Ghosh. Dave. 2014. S. 15–22. [30] Sherwani. 2011. E. networks for enterprise.. et al.