Hearing Community Voices in HCI4D: Establishing Safe Places to Co-Create Counter-Collective Narratives with Women Farmers in Bangladesh

Although listening to community voice is a core value in HCI4D, we have limited methods to capture the community voice of marginalized groups within disadvantaged communities. Working with NGOs and 24 marginalized women farmers in Bangladesh, we promoted psychological safety and empowerment through our configuration of the process. Our stakeholders decided to record and produce a radio-style audio recording that presented their counter-collective narratives for development projects. We reflect on this process using the Benefits of Community Voice framework to document rich insights into community contexts, lived experiences, local knowledge, and building trust and buy-in and through interviews with three NGO workers. We discuss the fundamental need of stakeholders for a safe place to share, the value of letting stakeholders guide method selection, the significance of counter-collective narratives, the benefits of participatory audio to hear community voices for democratizing and sustaining development and design implications of our work for HCI4D.


INTRODUCTION
Global development has helped millions of people in developing countries build more secure, sustainable, and prosperous lives.International aid -from governments, NGOs, and donors -to resourceconstrained countries originated after the Second World War and postcolonialism.Despite billions of dollars in aid spending since then, the practice and impact of international aid are not fully understood, and the Global South still has less socioeconomic development than the rest of the world [52,58,106] with 700 million people still in extreme poverty [15] and 3.1 billion with limited access to safe and nutritious food [142].Most of these people belong to disadvantaged groups that are underserved and marginalized1 particularly in rural areas in developing countries.
Listening to community voices while designing projects, interventions, and technologies has been a central pillar of many development and HCI for development (HCI4D) frameworks [56] and this has led to its becoming a central component of the Sustainable Development Goals (SDGs) of the United Nations [102].Practitioners want to establish relationships with disadvantaged people and to capture their rich narratives and counter-narratives to share their collective voice with physically distant program leads or policymakers.This is particularly challenging with disadvantaged groups within marginalized communities such as women living in socioeconomically disadvantaged patriarchal societies.Although traditional design research approaches, including expert-designed interviews, surveys, focus groups, and participatory work, can collect community views, without proper framing, they can struggle to capture marginalized voices within the community and facilitate their meaningful participation in development [3,56,69,113].
A body of work in HCI4D has focused on capturing the voices of disadvantaged communities to share with decision-makers through participatory media.This is facilitated by the availability of technologies in less developed communities, enabling people to tell and share their own stories [19,43,66,121,151] and making them easy to record.Previous work has utilized audio-based community participation for information gathering and dissemination to aid in a variety of felds, including education [98], health [85,138], agriculture [109], climate [68], entertainment [145] and citizen journalism [100].Despite a rich body of work on HCI4D discussing participatory media, we still need to learn more about systematic and interactive methods to help design efective engagement to bring together marginalized subsets of these groups that are typically unheard as a collective voice for efective community development, a topic that directly interests HCI4D, ICTD, and other disciplines related to development [11,[71][72][73]149].
In this paper, we present the process of working with 24 local women farmers in a rural community in Bangladesh to understand the health and agricultural challenges they face to inform the delivery of NGO programs through co-produced audio.Our methods emphasized their psychological safety and agency throughout the process of capturing community voice and led stakeholders to decide to reconfgure the audio recording to deliver a radio show as a way to document their counter-collective stories.Our fndings highlight the challenges found in the context in which they live, stories of the struggles they face, the skills and knowledge they currently have and want to develop, and their perspectives on NGO's work.Our refections highlight the critical nature of psychological safety to successfully capture sensitive and status-quo-challenging stories from the group, as well as the stakeholders' desire to produce their community voice in their own way and not surrender it to others' contextualization.We discuss the recordings produced from the work with NGO workers to understand their potential.Finally, we refect on how collectively built counter-narratives enhance the stakeholders' sense of safety and build reliability in the community voice data that they generate for decision-making and the implications that their accounts have for HCI4D designers.

RELATED WORK
We begin by examining the most recent work in HCI4D.We then review the literature and debates on giving marginalized groups a voice.Finally, we examine a variety of participatory audio recording methods as a means of involving underprivileged people in the decision-making process for community development in the Global South.

HCI and Global Development
The importance of including the opinions of those underserved and marginalized in decision making is universally acknowledged as the key to achieving lasting social impact [76,140,154].HCI4D researchers have been involved in international development for two decades to create accessible technologies, methods, and innovations for underserved populations.Designers have addressed challenges in health [90,108], nutrition [30,132], education [146], and agriculture [66,109].HCI researchers have used participatory design [23,147], co-design [65,153], and co-creation methods [12,111] to include communities of users as equal partners.Numerous ethnographic studies have examined the ways in which various cultures use technology, as well as the reactions of ethnically marginalized communities to interactive technologies [20,26].Ethnographic studies have tried to close the digital divide [7,8,73], build familiarity with local technologies design practices [16], understand the importance of privacy and security [4][5][6]124], and examine patriarchal cultures to inform the design of safe technology for women [135].Participatory rural appraisal (PRA) is a popular method among NGOs to engage local people and build relationships and analyze to extract information (mainly visual) from local rural people [31].
However, traditional approaches may not be efective in engaging and capturing the experiences and challenges of marginalized groups within disenfranchised communities for a variety of reasons.Irani et al. [74] note that "design research and practice is culturally located and power laden".Bidwell and Hardy notes that text serves as a tangible attribute in modern society, controlling the currency by which people acquire power through social status and historicity, but it can disenfranchise and oppress people from oral cultures [20].Designer positionality may also inadvertently sideline the values, beliefs and conventions of other cultures [21].Largescale data collection, including traditional quantitative surveys, qualitative interviews, focus group discussions, and participatory rural evaluation, are top-down approaches driven by experts that restrict the space for disadvantaged communities to share their experiences through embedded power dynamics [87,89,96,114] leading to unequal participation.Critics claim that this makes these approaches unsuitable for creating environments that incorporate community voice in decisions [38,113,114,123,152].Even participatory approaches designed to address these issues in developing contexts, can fail with marginalized groups.Schelenz and Pawelec [125] criticized HCI4D practices in the Global South citing a "lack of user-centric design of projects" and "insufcient participation/ inclusion of users".Furthermore, practitioners lack knowledge of the needs of the local community and important social, cultural, economic and political factors [11,71,72,143,149].
Other researchers have attempted to address these challenges through varied approaches.Recognizing the inherent challenges of cross-cultural participatory design, Brereton et al. noted crosscultural communication and empathy are needed and signifcant time is required in building cross-cultural trust [25,27].They proposed an approach based on reciprocity working with indigenous Australian people in side-by-side activities to build trust.The local groups involved included many local authority fgures and may not be suitable for specifcally capturing the voices of marginalized groups within indigenous populations, such as the women with whom we aim to work.Soro et al. (2016) [131] propose dialogical investigations as a lens between cultures as a means to gain "inspiration and insight from dialogue" in the design process.Bryan-Kinns et al. (2022) [29] integrate local knowledge with global trends in Qi2He, a co-design framework based on traditional Chinese epistemology, to address the difculties in integrating local knowledge and culture through cross-cultural co-design engagement and post hoc critique of participation.
We see a clear need to create more guidance on the establishment of relationships with marginalized communities, empowering them to share their voices to produce community-generated data for efective decision making [121].This must happen in ways that are understandable and meaningful to those communities [3,69], but voices must also be in a form that can be shared directly with physically distant decision-makers [114].

Voice for the Voiceless: Counter-Narratives in Disadvantaged Communities
Community voice has been widely used to describe a way to address issues with more traditional methods by including various stakeholders from political decision-makers to disadvantaged communities that want their voice heard [67,82].The agency of participants, in terms of their freedom and the opportunity to share their refections, is a critical entry point to sharing an authentic and meaningful voice [39,62,70,78,80].The international development community should be aware of how they portray themselves as the provider of agency, as this language implies that agency is something developed people give to the disadvantaged or marginalized, rather than something the marginalized can take for themselves [50].Freire argues for considerations of agency in sharing and listening to silenced voices to understand perceptions, lived experiences, and expectations [62].However, it is inherently difcult to hear the voice of the marginalized within a community because they have been silenced and face inter-sectional challenges in accessing basic necessities (food, education, healthcare) while being dis-empowered to make decisions that ensure their communities' sustainable development [78,104,114,120,128,134].Gurumurthy observed the critical relationship between "agency", "voice" and "empowerment" in disadvantaged communities [70], highlighting the need to explore more communication-based approaches for community voice.She argues that communities have the ability and accountability to self-express development priorities.Sen's capabilities approach emphasizes enabling human freedoms and comprehending development from the point of view of the people involved [127].Kleine's Choice Framework puts this into practice, focusing on individual development results [88].These approaches prioritize the requirements and concerns of those afected by development initiatives.Furthermore, establishing psychological safety, ensuring privacy, security, and safety are of utmost importance, as are ethical considerations for including marginalized communities in the development process [133].Sen's priorities are refected in the specifc considerations and approaches needed to work with women in patriarchal societies [126].Storytelling has long been considered a powerful approach to bring voices to disadvantaged communities [53,155].Oral storytelling is the most common practice in indigenous cultures that shares and validates personal, mythical, educational, spiritual, and collective experiences through witnessing, remembering, and sharing stories [42,75].In rural Bangladesh, women are restricted in their freedom to make decisions, argue with men, protest, share their desires for their own lives, or even simply voice challenges [104,134,135].These suppressed narratives are forms of counter-storytelling, raising a dissenting voice or challenging the dominant narrative.
Counter-storytelling is one of the most signifcant ways to give voice to the voiceless [45,97].Rooted in critical race theory (CRT), counter-storytelling empowers marginalized communities by exposing dominant narratives and amplifying their perspectives, promoting equity, and exposing marginalized perspectives [45].According to Solórzano and Yosso (2002) [130], counter-stories provide readers with an opportunity to criticize unfair practices and identify opportunities for transformation from the perspectives of the marginalized.DeCuir and Dixson (2004) sought counter stories that exposed and criticized normalized dialogues [44].Delgado (1995a) emphasized the process of analyzing myths, presuppositions, and received wisdom that comprise the common culture around race [46].The "voice" element of CRT, according to Ladson-Billings and Tate (1995), ofers a means of communicating the experiences and realities of the oppressed: "a frst step on the road to justice" (p.58) [92].Counter-narratives ofer alternative perspectives, improving agency, understanding social issues, dismantling oppressive systems, and empowering marginalized individuals by critically examining power dynamics and advocating for change [97].
However, how these stories are formed in terms of rigor and method, validated, evaluated, and represented has been criticised.Chimamanda Ngozi Adichie highlights "The Danger of a Single Story" and stresses the importance of a multiplicity of stories, voices, and perspectives when seeking disadvantaged people's voices [1].Eric McDowell notes "...we must attend to the ways craft itself can create opportunities for constructive and responsible representation" [95].We argue that we should also seek a collective voice in stories and counter-story.By incorporating considerations from participatory research, development studies, design studies, and counter-storytelling approaches, we aim to democratize marginalized groups within excluded communities and amplify their voices.We acknowledge the importance of addressing the critiques raised in this section to ensure that our design methods truly prioritize agency, empathy, dialogue, and diversity to co-design and co-produce their community voice.Ultimately, our study seeks to create an approach in which participants can advocate and authentically express themselves and collectively contribute to community engagement eforts.

Participatory audio and HCI4D
The use of participatory and online media technologies to strengthen communities has increased in recent years.Participatory audio voice initiatives have considerable potential to capture and share community-generated voice in domains from education to journalism.Kumar and Dell (2016) found an increase in the use of multimedia-based interactions in countries in the Global South [47].In many low-and middle-income countries, people in marginalized groups rely on participatory social networks based on audio, text and multimedia, such as WhatsApp, Facebook Messenger, and SMS to communicate [47,139].However, social networks are not always a viable option for underprivileged communities due to the lack of internet and limited literacy.
Participatory audio has been used with populations that have limited access to educational and internet infrastructure to address these issues [55,138,144].The low initial investment and widespread availability of radio have made it a popular choice as a communication medium for underprivileged and rural communities.Radio is a widely accessible technology and a familiar format in these places [61,86].Manyozo (2009) claims that radio is "pervasive, local, extensive, fexible, available, easily understood, personal, portable, swift, and efcient" [93].Voice-based telephony has also been extensively used in international development to empower underprivileged communities.Refugee women in Syria have benefted from telephony platforms that facilitate the exchange of health information in refugee and humanitarian contexts [138].
Accessible audio tools for those living with physical disabilities in the global south include talking audio books for visually impaired people in India [144] and voice-based platforms such as interactive voice response (IVR) to help underprivileged communities access healthcare support, peer support, medication reminders, and health management recommendations [77,85,112].Another example of voice-based innovation inspired by ICTD is India's Avaaj Otalo, which helps rural farmers access and share information on farming techniques and weather forecasts [109].Christnesen et al. developed Krishi Kontho, a system that can send voice messages to farmers' mobile phones and was used to warn farmers in Bangladesh about upcoming storms and other adverse weather events [35].Some ICTD work has been based on creating a community-based voice forum where community members can document and discuss their needs and challenges with policymakers.As a community advocacy tool, CGNet Swara allows citizen journalists to document and report on local issues [100].
Although all of these participatory audio voice initiatives have considerable potential to capture and share community-generated voice, there remain gaps in participatory methods for meaningful community participation in terms of capturing and reviewing stories among disadvantaged communities, sharing collective stories and refecting community voice for decision making.Participatory methods for meaningful community engagement must include ways to share collective stories, capture and review them with communities (specifcally those who are disadvantaged), apply analytical tools to refect on community voice for decision making, and discuss how to use community voice, including as data for decision making [119,121,147].This study aims to start conversations and contribute to these existing gaps through our co-design and co-production of participatory audio work with a disadvantaged women farmer community in rural Bangladesh.

METHODOLOGY 3.1 Project overview
Our project was established to understand the health and agricultural challenges of women farmers in Dimla Upazila of Nilphamari District, Rangpur, Bangladesh to support the ongoing eforts of three collaborating NGOs -a local NGO (Pollisree2 ) ofering onthe-ground support and two international NGOs (Oxfam3 and an anonymous and nameless philanthropic network) ofering fnancial support -and the University of the lead author.Our prior research suggested that we could best support these eforts by fnding unheard stories, particularly the local women farmers' counternarratives of the support they received, but we also knew that a safe environment had to be prioritized for them.In this section, we describe our methods and how the community with which we worked came to set the agenda for the work.A note on language around agency: our research is NGO-driven, and though we want to avoid suggesting agency is something we "give" to our stakeholders, we try to accurately refect where an initiative comes from and which elements of the project show our stakeholder's agency rather than our own direction.
Our work focuses on enabling participants to authentically express and advocate for themselves, contributing to community engagement, while addressing critiques from counter-narratives [1,95].We employ a "counter-collective" voice in our approach, where narratives are collaboratively constructed by multiple stakeholders, enhancing story validation, anonymizing personal details, preserving original voices, and protecting participants from feeling targeted.Incorporating participatory research, development, design studies, and counter-storytelling, our objective is to empower marginalized communities in decision-making and amplify their voices.
We start by describing the feld site and stakeholder selection process, followed by feld design activities.We the document how the stakeholders, working in a large group, defned their communication needs, leading to a participatory audio recording phase for a radio show.We also discuss how we sought refections from NGO workers after the recording phase was complete.Finally, we describe our deductive approach to thematic analysis using the Benefts of Community Voice (BoCV) framework [122], focusing on both the data generated and the project process.

Field Site and Recruitment
Our research was carried out at the Dimla Upazila site in northwest Bangladesh close to the Indian border that for the last fve years has been managed by Oxfam Bangladesh.Dimla is located in the Char lands of Bangladesh, a region formed by large river deposits that constantly shift.The region is frequently afected by extreme weather and climate events, and both droughts and foods are common.The local community is heavily dependent on agriculture and river fow but this is challenging due to the particularly sandy soil.Communities in the Char region are politically underrepresented and underserved due to their demographic conditions [35,36].The community with which we worked is involved in development projects that are managed by local NGOs and are primarily funded by international NGOs and UN organizations.Most of these initiatives are focused on empowering women by improving access to information about agriculture, health, and nutrition to improve their general social and economic status and well-being.
We were assisted in recruitment and introductions to our participants by three local project workers who acted as guides and gatekeepers.They helped us fnd 24 women farmers in the Dimla Upazila area who are engaged in agricultural activities and who are residents of households supported by our project.They included older adult women (over 45), young adult women (aged 18-45), and adolescent girls (under 18).We planned our feldwork activities around the convenient locations and schedules of our participants.We ran the workshops as two eight-hour days in a row, with the 24 participants participating at the same time with all the facilitators.This approach was chosen because the participants needed to work in the feld to earn money, and any time missed would lose a days earnings, regardless of whether they worked with us for two or eight hours.As preferred by participants, they were compensated for volunteering their time by being provided foods and drinks (breakfast, lunch, and two snacks) for them and their children.The workshop was mainly facilitated by the frst author, a researcher and practitioner of Bangladeshi descent, with the help of another author and a local Bangladeshi research assistant in March 2019.
Participants worked as farm workers and home farmers, as well as maintaining their households.For those with families, their typical day would start before dawn to prepare food for their family and maintain home crops and livestock.After this, they would usually travel on foot to work on other larger, male-owned farms as day laborers for 8 hours before traveling home to prepare food for their families.Domestic violence was commonplace for them; not having a timely meal prepared could lead to their husband assaulting them.Days of were limited only to large public holidays, such as Eid-ul-Fitor.In the evenings, they would visit with their neighbors and socialize.
Our work with women farmers would reveal that they typically do not have a personal smartphone, despite many having received one from a previous Oxfam intervention.These devices became family phones that they could not keep private.The infrastructure in the region was poor and typical of Char, participants did not have reliable access to electricity, and frequent natural disasters made it unclear whether they would be able to get it in the future.Some went to small 'Recharge and Phone Call' stores to charge mobile devices.The roads were muddy and frequently fooded, and proper sanitation and running water were only accessible in specifc central buildings in the village.

Research ethics.
We obtained the consent of the stakeholders for our workshop and audio production.Our Institutional Research Board granted approval.Because we were aware of the limited literacy of the marginalised community we had in mind, we prepared a visual consent form following the guidelines in [63,120], with written and illustrated representations of our research goals to engage participants in the process and explain considerations that include data sharing, privacy, withdrawal, risk, and benefts of the study.Stakeholders indicated a preference to be acknowledged for their contributions and granted permission for us to use their names, images, and audio in our research, articles, reports, and documentation.

Research activities
We were aware of several challenges that women farmers faced, which NGOs tried to address: The stakeholders considered these issues and chose to focus on health and agriculture.The two-day workshop at Dimla Upazila used participatory audio recording activities.We were interested in the prior knowledge of our stakeholders, current practices and methods, experiences, obstacles, and their approaches to solving problems in their communities.Our objective was to understand the personal and group experiences and stories of our stakeholders to capture a collective voice in the community.Although not part of our initial plan, the stakeholders chose to create a radio show-style audio recording of their communitygenerated voice and share it with the Dimla community and our collaborating organizations, Oxfam and Pollisree.We divided the workshop into two phases, one around brainstorming and storytelling, and the other focused on participatory audio recording, but the second phase process was altered by the participants as we worked.The workshop's frst day was designed for the 24 women, while the second day focused on six individuals (representatives from the frst phase) engaged in the creation of a participatory audio recording.

Phase-1: Designing for Collective Voice
The frst day of the workshop started with a set of structured participant-centered design activities.

Topic familiarization:
On the frst day, participants were provided with our pictorial consent form, outlining the research's goals, issues, withdrawal options, and information sharing, use, and storage.Then, to understand the views of our stakeholders on the two topics they had selected and to start the group discussion, we frst used printed pictures that depicted their daily lives.Stakeholders started discussions with each other to refect on them.

Story sharing:
Stakeholders divided into three groups according to age to refect on, then discuss, their agricultural practices, health challenges, and perceptions of the wider community.The older and adolescent girls' groups each had seven members, while the young adult women's group had ten.Of the seven young girls, four were already married and two had children.All the older adult women were married and had children and grandchildren.All young adult women were married and had at least one child.In facilitating this session, we emphasized counter-collective storytelling, looking for narratives that stakeholders might be reluctant to share outside of the group we had brought together, and encouraged afrmations and elaborations from other group members.We gave them a set of questions on they do and what challenges they face in agriculture and health to stimulate discussion (see Figure 1, left photo).We also gave them note cards and pens.

Counter-collective voices:
All of the participants sat down in a circle for a discussion after their group discussions, where the frst author facilitated the discussion and each group shared its group discussions and stories (see Figure 1, photo on the right).During this activity, the facilitator and research assistant took notes on a fip chart.The group reinforced each other and further validated the idea that the most unheard stories could be voiced in this setting.After the stakeholders had discussed and settled on the issues among themselves, the facilitator surveyed the group to determine what they felt were the most pressing concerns.We made sure that the participants prioritized and agreed narratives as their collective voice.

Phase-2: Co-production of the audio recording
The original plan was to use participatory audio recording to capture short vignettes in the second phase.However, the group wanted to produce something for themselves that was representative of the challenges they faced and that communicated their voice clearly.They did not want the research team to curate or edit what they said, instead looking to be able to do this work themselves, placing the emphasis on the parts of the narrative they felt needed the most attention.The group struck upon the idea of a radio show to narrate their problems, literally capturing their voice in a medium with which they felt familiar while preserving some sense of anonymity.
To do this, instead of all speaking, the large group chose six individuals to represent them and appear in the participatory audio recording the following day.They selected the six stakeholders (two older adults, two young adults, and two adolescents) because they thought they would be the best representatives (see Figure 2, right photo).These six were described as outspoken, four of them being particularly experienced farmers and were seen as leaders or advocates in their typical behavior within the community (see Figure 2, photo left).Two of the group acted as hosts, interviewing the other members, and fnally two additional teenage girls (not included in the six) were selected to perform songs on the radio show but did not narrate their experiences.

Preparing and revisiting collective voice:
In order to refresh the six stakeholders' memories of the topics discussed, the second day started by talking about the key points written on the fip chart.Using the notes from day one as a guide, the six stakeholders decided what questions to ask and what to say in their responses.
The fip chart's notes were placed in front of them to carry out the session and make sure that they covered everything that they decided to share during the audio recording.After rehearsing for a few hours, the group informed us that they were ready to record their radio show.

Capturing participatory audio recording:
We used a general open source recording application suitable for Android-operated mobile phones or tablets on a Samsung Galaxy Tab A, an Android tablet with the built-in audio recording app, along with two separate portable microphones.Our user-friendly tools helped participants share their opinions without the need for expensive equipment, media production expertise, or Internet access, taking into account consumer technology (tablets and mobile phones) that was readily available in the feld.
3.5.3Reviewing, curating, and editing: Stakeholders took about an hour and a half to record their entire session, sometimes pausing to talk, add information if they missed any points, and sing their songs.After recording the session, the facilitators played the audio back and sat with stakeholders to discuss what should be kept and what should be discarded.Pauses and repetitions were removed from the fnal recording.As a result, everyone who participated in the audio recording reviewed and refned the fnal product.The stakeholders named their radio show "Char-er Ful" [Char's Flower].Once the audio recording was complete, we gave it to the participants via fash drive and played it for their large group and community, where the participants' relatives, neighbors, and local NGO workers were all present.

Analytical and refective approach
A deductive thematic analysis was used to analyze the data, which involves reading the data and utilizing an existing theory or structure to identify themes (recurring patterns of meaning) [22,24,110].We used the Benefts of Community Voice (BoCV) framework of Saha et al. (2023) [122], which sets out tools to understand the voice of disadvantaged communities for decision making, but also refects on how well it is captured.Our data for this study include feld observation notes, participant notes on cards, facilitator notes on fip charts, and audio recordings.In addition, we also have the produced radio shows from our stakeholders that represent their counter-narratives about the challenges they face.The frst author translated all raw data into English, and the second Bangladeshi author verifed the translation.We have interpreted the stakeholders' statements to refect their meaning, rather than translating them directly.In terms of NGO workers' refections on the work, the frst author interviewed three NGO workers over phone calls (each call lasted around 30 minutes).While we initially planned to return to them to get feedback shortly after the recording, the onset of COVID caused an interruption in our work.When we fnally returned to the NGO workers, after asking them to listen to the recording again, we captured their refections on the audio show and what they had done since we had left.The author took notes during the interviews and summarized the refections of NGO workers under Section 4.5.
As the authors are experienced in international development, they bring a socio-constructivist perspective [40] on development practice to this analysis and use it to look at the grounded challenges of our participants in the data.The frst author is of Bangladeshi origin, with more than fve years of international development experience and another fve years of experience in HCI4D research.The second Bangladeshi author has 14 years of ICTD experience in Bangladesh.All authors have experience designing participatory media, qualitative research, and working with disadvantaged communities in the Global South.

FINDINGS AND REFLECTIONS
We document the understanding of those involved in the design activities by utilizing the BoCV framework to refect on the entire process, not just the fnal radio show.We have classifed the insights under the four themes proposed by the BoCV framework: understanding the contexts of female farmers, identifying narratives that generate empathy by displaying lived experiences, leveraging local knowledge and abilities, and encouraging trust and buy-in to the development projects [122].We provide excerpts from the participatory audio recording to support our interpretations as evidence from our stakeholders collective voice.

Understanding women farmers' contexts
Contextual understanding in the BoCV framework focuses on external and internal factors that impact the lives of stakeholders.We discuss the economic situations in which our stakeholders fnd themselves, the political norms that impact their community, and the inter-sectional impact of cultural norms.
4.1.1Socio-economic and climate challenges .Through our community engagement, stakeholders shared a variety of contextual information, especially in relation to their current situation and agricultural practices.In Bangladesh, more than 50% of women are involved in agriculture [116] and all our participants participated in diferent types of agriculture, some indoor farming and others outdoor.They highlighted what kind of agricultural activities they do, saying: "We do diferent kinds of farming.For example, rice, corn, vegetable cultivation, fsh farming, livestock including chicken, duck, and cattle farming, and we cultivate and involve ourselves in the household and the feld.We work before and after the harvest collecting food, cleaning, drying, storing it, and all the other associated work. . ." [R3].
Although they are involved in various agricultural activities, they are disadvantaged because they do not receive much beneft from their work [116].They shared the layers of their situational challenges, starting from being in a resource-constrained agricultural setting where they have limited access to economic or educational opportunities.For example, "We are poor people.We depend a lot on farming because we have nothing else to do.None of us are sufciently educated to work in the outside world.Agriculture is the main source of our livelihood, we don't earn much from it... " [R1].
They also highlighted that we needed to understand their geographic and climate disaster issues.Participants highlighted the region-specifc crops that can grow in sandy areas vulnerable to drought: "...those of us who live in the char land produce potatoes, onions, coriander, black cumin, nuts, and various spices.So we usually do all these." [R2] 4.1.2Political contexts related to livelihood.The political environments in which our participants operate were highlighted in their discussions because they felt the government was unconcerned about the 'Charland' and their lives.90% of them are landless or functionally landless [18].Local, national, and international nongovernmental and governmental organizations, including extension workers, work in these areas to promote the development of these communities, particularly through the pursuit of agrarian reform.However, participants draw attention to the difculties they face as a result of their limited access to agricultural resources such as high-quality seeds, fertilizer, insecticides, pesticides, information and other services (such as poultry vaccination and treatment).An audio-recording respondent noted with frustration that "Since we are from rural areas we don't have a good idea of good seeds.We don't get good seeds though we have NGOs who work with the community closely for our agricultural development.Even extension workers (Government agricultural support ofcers) also work for our community.We have to face a lot of problems while farming but we don't get the exact help from them that we need most." [R1] 4.1.3Marriage, social norms and reproductive health.Our participants shared critical health issues that afect them throughout their lives.Their experiences revealed multiple intersecting challenges, linking limited education and economic status with early marriages, prevalent in Bangladesh.In Char, 67% of marriages occur before age 18 [107], higher than South Asia's 51%, placing the country among the top ten for early marriage rates globally [141].Post-marriage, these women often face early pregnancies and childbirth, with the latter posing signifcant health risks during their most vulnerable times.
"In our daily life the most serious problems a woman faces are during pregnancy.Every woman faces that.A new mother faces lots of risks. . .risk of death. . .We mostly prioritize that. . .moreover, after childbirth, the mother and child can both face a lot of problems and the risk of death in the frst 40 days. .

. " [R3]
According to the World Factbook 2020, the maternal mortality rate in Bangladesh is 123 deaths per 100,000 live births, the 4th highest in South Asia (after Afghanistan, Nepal, and Pakistan) [57].
Our participants highlighted the impact of their cultural context on their health.They often kept health issues, particularly those related to sexuality and their bodies, to themselves due to cultural and religious norms viewing these topics as "shameful." Women are discouraged from discussing their menstrual cycles, pregnancies, or related health concerns.A critical issue is the onset of menstruation in young girls, who feel obligated to remain silent about it, unable to even confde in other women.This cultural silence is exemplifed by the difculty young girls face in discussing their challenges with their mothers, who are also constrained by the taboo nature of these topics.
"During this time the mothers are mostly outside of the house.So we can't talk to them properly.And most of the village girls are too shy to talk about these things.They don't even share it with their grandmothers, they follow their own schedule.Because of this, our girls face a lot of problems...our girls also have white discharge.Sometimes they share with us a little bit.They try to fnd a solution but we don't even know why this actually happens.We maintain ourselves and the kids maintain it by themselves.They don't want to share and we don't ask them either..

. " [R2]
In summary,the stakeholders shared their rich situational insights on their important external factors (such as social, economic, political, demographic, and climate factors) and inner factors (such as cultural and religious norms), which are crucial to understanding stakeholders' context.

Creating empathy: rural women farmers' lived experiences
The BoCV framework suggests that hearing community voices creates empathy with stakeholders through the stories and feelings they share.Pragmatically, this helps external parties start to understand their experiences and anticipate their reactions to novel concepts, but it is also important in development work to reinforce the value of the work which can be disheartening.

Lived experiences of agricultural activities .
Our stakeholders face signifcant challenges due to societal discrimination.They expressed frustration with limited market access, particularly affecting rural women, who are culturally discouraged from traveling to markets.Consequently, women often sell their products locally to neighbors, intermediaries, or male farmers, while men typically sell products at the market, sometimes including those produced by women.Their collective voice in the audio recordings highlights these shared frustrations and experiences with this process.
"The markets are a crucial part of our fnancial improvement.We might produce a lot of vegetables, chickens and ducks...But if we can't sell them in the market then there's no point.That's a major problem.So we the women don't have much idea about the market..

. " [R2]
Another experience raised by the stakeholders was how difcult it is to fnd a "fair price" for their goods.They admitted that they were confused about the real prices in the market.As a result, despite having a variety of horticulture, fsh, and poultry productions, they felt like they never received fair compensation for their products.Furthermore, many of the stakeholders shared extrinsic lived experiences about their disappointment in the government's inability to provide them with a standardized price for certain agricultural commodities.They expect the government to set a minimum price for all their goods (including staples, fsh, and poultry) so that they can protect their low-priced sales.Their anger emerged as the following collective voice: "Which crop should we sell or what should be the rate of rice?We don't know these things since the market is very far from us.We don't go there and we don't know the actual price.For instance, for corn, the government hasn't set a rate for it.But the main crop that we work here with is corn.Still, we don't have much idea about the corn market price.We sell them however we like and for random prices.If there was a set rate for it, even the farmers could beneft from it.This is a critical problem." [R1] Stakeholders indicated their agricultural profts are hampered by limited access to information on agricultural services and natural disasters.They highlighted the impact of foods and cyclones on their crops, fsh, and poultry, emphasizing the need for disaster forecasts and protective measures to mitigate fnancial losses.One stakeholder's personal experience underscored how ignorance about natural disasters and market prices can drastically reduce profts, sometimes to a quarter of their potential."...The fsh that get washed of in foods, if we could sell them beforehand then that'd be helpful.But even there we don't know about the market rates.Because if we want to sell fsh before the food the prices would be 100/50 taka [BDT] per kg but after the food, it gets up to 200-300 taka.The price is hiked.If we had any idea about the upcoming disaster notifcation and varied price, then we wouldn't sell the fsh for much less price.

" [R4]
The stakeholders expressed their distress at the lack of knowledge and information about the management of diseases in their agricultural products, leading to signifcant losses in their poultry and fsh production due to unidentifed diseases, and in their horticulture (vegetables and staples) production due to harmful insects.One of the speakers expressed their frustration, saying: "...when the disease starts it eventually kills all the chickens and ducks.So what happens is we have to buy these chickens and ducks every year.This costs us a lot of money instead of some extra earnings for our families." [R3] Their agricultural livelihoods, deeply dependent on climate, are severely afected by extreme weather.Located near the "Tista" river, the community faces erosion of the river banks, leading to sinking of homes and farms and destruction.Daily life is marked by constant tension due to the threat of sudden foods, which can destroy houses, crops, livestock, and fsh, signifcantly afecting their lives and livelihoods.A participant described this experience.
"...we all live near the river Tista. . .The farming that we do, before we can even succeed in farming we have to face the riverside erosion. . .also we face the extreme rain and food. . .Because of the food and river erosion much of a crop, or even all, is ruined...[R1] Another participant reported, "We live in a char area and we do fsh farming.Each year the food comes and washes them away." [R3] 4.2.2Women farmers lived experiences on health.The stakeholders in our study discussed the various real-life experiences their daughters face as adolescents, such as the realization that menstruation is essential for a girl's growth and readiness for motherhood.Most of the women expressed feelings of confusion and inadequacy, as they were unsure how to help their daughters with this process.Furthermore, due to the taboo nature of the subject, girls were hesitant to tell their mothers about their struggles, and mothers were uncomfortable asking them.Furthermore, many mothers had to leave the house during the day to work as farmers, leaving them anxious about their daughters, who were menstruating.
" The women in our study shared their individual stories of menstruation, early marriage, and childbirth, and how these events have impacted their physical and mental health.They expressed concerns about labor due to their limited understanding of the two primary childbirth techniques, natural birth and surgical C-sections, due to a lack of resources.Furthermore, they discussed how the social and cultural contexts in which they live can lead to prenatal anxiety and fears about the delivery process: "The rural people have no idea about the proper way of normal delivery and cesarean delivery [C-section delivery].During childbirth, the most risk is for the baby or the mother.Among rural women, we face this problem a lot." [R1] Stakeholders sadly revealed that despite their participation in food production, they were the last to eat in their households.This lack of access to food resulted in poor health and exacerbated the difculties they faced during pregnancy.The following quote also expresses their actual but unfortunate lived experiences of having food constrained by their socio-cultural contexts."...We eat whatever is left after having meals of our head of the family and children.Sometimes we only take a little as we need to make sure our other family members -husbands, children, and elderly family members such as fathers-in-law have enough food." [R4] In summary, stakeholders candidly shared their real-life stories and subjective feelings, which were shaped by their particular circumstances, including restrictions on economic means, cultural impacts, and learning.

Understanding and leveraging local knowledge and expertise
Our feldwork and the audio recording exercise we conducted revealed the need for a variety of information related to agriculture, health, and nutrition, which encouraged stakeholders to confront their own levels of knowledge and lack of knowledge.This data is useful for decision-makers (program and technology designers) as it shows the types of information that are most important for the development of the community.The stories that were captured not only demonstrate their knowledge and profciency, but also give them a chance to express what kind of knowledge and skills they need to address the difculties they face on a daily basis.

General needs of information on
Agriculture.The stakeholders expressed a clear need for information about place-based agriculture.They live in areas that are prone to disasters, and their cultivation practices, such as soil, land, and water bodies, difer from other parts of Bangladesh.Most of their areas are sandy, which is not suitable for growing many crops and vegetables.The stakeholders thought that, in this time of advanced technology, there must be methods to cultivate various crops and vegetables in their localities.:One stakeholder mentioned, "We have sandy soil here.We don't know which kinds of crops to cultivate in a soil like this.We sow whatever we can in the land but sometimes it grows, and sometimes it doesn't.We don't know which process we can follow to work with sand soil so we can work with less irrigation.If we can get to know more about these then the soil won't go to waste and we can utilize it.

" [R2]
The stakeholders expressed a need for agriculture-related information that was pertinent to their livelihoods.They wanted to know more about disease management, horticulture, fsheries, and livestock.Additionally, they had limited knowledge of how to treat their livestock and when to give them vaccinations, so they also asked for information about these topics.
"We face problems with chicken and duck diseases.In my house the problem we face is that we bring medicine from the market, if we use the medicine sometimes it works and sometimes it doesn't.For example, we sometimes can't vaccinate our chickens or ducks on time.We don't know the perfect timing for vaccination.We want to know all these.Another thing is we don't know the perfect timing for vaccination of our ducks.

" [R4]
As previously discussed, our participants have limited access to market information.This was identifed as a major obstacle that prevents them from obtaining a fair price for their products.Furthermore, they lack the necessary information and services related to climate and natural diseases.They have few reliable sources from which to obtain information about upcoming natural disasters or warnings about foods and droughts.They rely on traditional knowledge and the Bangla calendar to predict when foods may or may not occur, which is not always accurate.They also receive limited information from the local government: "Our needs are never-ending.We have to ask for the knowledge that'd help us in our work feld.Such as knowing what good seeds are, how to handle drought or food, or diferent kinds of training and workshops so we can identify diseases and know more about vaccines.We'd know more about these if we could get some training on it." [R1] 4.3.2Health and well-being related knowledge.Stakeholders cultivate vegetables like pumpkins and gourds in home gardens and open felds, using soil-covered straw roofs but they often cannot aford fruits, as these don't grow well on their land and are costly to buy from markets.However, it is widely understood that they have to eat vegetables to meet their nutritional needs due to the health education they received from the NGOs.Although they cultivate various types of vegetables, they usually sell their produce, eggs, and livestock, including chickens and ducks, to buy staples such as rice and potatoes, which are less nutritious.This trade-of, driven by a lack of specifc nutritional knowledge and fnancial challenges, makes it difcult for them to provide nourishing food for their families and themselves.
"Actually, we cultivate a lot of vegetables in our houses but we still don't know which vegetable contains which nutrition.Sometimes we sell all our fresh vegetables in the market and buy adulterated vegetables to eat... Sometimes we produce gourd, tomatoes, brinjals and vegetables like that.So what we do is we sell all these and buy potatoes from the market...We don't know what nutrients potatoes contain or what the vegetables that we produce contain in them...We also don't know which food to feed our family members or the amount of food to give to the elderly and children...We all are rural people, we can't fulfll our nutrition needs by buying fruits from the market.If we could fulfll all the nutrition needs of the vegetables we grow, that would be better. .

. " [R4]
In summary, our community engagement revealed the information needs of stakeholders in agriculture, health, and nutrition, providing valuable insights for community development.Furthermore, the stories collected highlight the extensive knowledge of agriculture that stakeholders possess, as well as their needs to effectively address daily problems.

Building trust and buy-in
The process of hearing community voice helps to build trust and buy-in to development projects.In our extensive two days of community engagement, the commitment to listening to counter-narratives, and letting the stakeholder community set the agenda developed some level of trust with the research team.The stakeholders agreed that sharing their opinions could build a network connecting them with the government and other local entities, fostering idea exchange, enhanced cooperation, and greater accountability in decisionmaking.They identifed local networks such as schools, health facilities, NGOs, and government ofcials as key communicators to understand their challenges and fnd efective solutions.They believed that addressing access issues, knowledge gaps, and the need for community activities could promote community participation, using local resources such as mobile clinics and health personnel.
"We are living here, and they [decision-makers such as governments and NGOs] can ask us...We are here to share what challenges we face and what we need most...they can ask us before they decide and give us big things like building schools, health complexes, and services like those from government agriculture extension ofcials.

" [R4]
Respondents shared anecdotes highlighting how decision-makers (governments and NGOs) often overlook community voices in identifying and prioritizing needs.They reported that their input was rarely sought or considered before launching new community improvement initiatives.For example, a government-built health complex was placed too far from the community for convenient access to services such as treatment, medicine, and childbirth assistance, since there was no prior consultation with the community.They believe that trained local caregivers could efectively support expectant mothers: "During pregnancy, the pregnant woman faces the most problems... Our health complex is very far from where we live...We are a remote community...If we had skilled birth attendants available in our community, it would be very useful during pregnancy to get the quick help that we need immediately rather than going to that faraway health center.Why did they create a health center too far away, or why do we not have trained birth attendants or caregivers?" [R1] Discussing Government attempts at providing iron tablets for girls' health within the community: "The government provides iron tablets now.Most of the clinics are very far from our area.Because of this reason, our daughters refuse to go there.Most of the time the kids go to school so they can't manage the time.And we don't even push our kids to go there.We don't even know that they need these iron tablets.They [government] can provide iron tablets either in their schools or in our community. .

. " [R2]
The discussions exemplify counter-storytelling, revealing trust issues commonly associated with government and non-government organizations.A repeated theme was the disconnect between higherlevel development authorities (such as governments and NGOs) and local communities.They specifcally listed the agricultural support they need, such as quality seeds and fertilizers, disease control information, livestock medication and vaccinations, climate-adaptive farming methods, and training in modern agricultural techniques and technologies.Stakeholders pointed out that their limited connections with government personnel, like Agriculture extension ofcials, result in inadequate development services, further disadvantaging them: " We'd know more about these if we could get some training on it.We don't have a proper connection and network with the agriculture extension ofcials, hence, we are signifcantly lacking proper services from them such as agricultural and market-related support to change our socio-economic conditions. .

. " [R2]
The stakeholders expressed how they manage daily problems without the support of the authorities, indicating that their concerns are often ignored or overlooked.The lack of assistance suggested to them that community well-being isn't a priority for decisionmakers.They described a history of being under-served and neglected by authorities, leading to a sense of powerlessness and inability to hold decision-makers accountable.This created feelings of helplessness and reinforced their belief that their voices and concerns were perceived as insignifcant due to their disadvantaged status.For example, they described three years of rice cultivation problems, with spoilage and rotting paddies, but received no support from the authorities despite reporting the problems.This situation has resulted in widespread frustration, disillusionment, and mistrust in decision-makers.
"We have been cultivating rice for 3 years.In these 3 years, we have seen rice paddy getting spoiled or rotten.The rice gets rotten even before it can produce the rice.We have been facing this in our village for 3 years but we can't identify the disease and can't get any support to solve this problem from the authorities [governments and NGOs] yet.Then, how do we solve this by ourselves as ignored and neglected communities?"[R1] In summary, our community engagement process served as a means of building trust and buy-in amplifed the often ignored counter-narratives of stakeholders.These narratives can play a key role in understanding community needs and establishing collaborative connections with government and local stakeholders, which can ultimately improve the efectiveness of development initiatives.

NGO Refections on the Radio Show-Style Audio Recording
The NGO workers explained the impact they observed in the feld.NGO workers observed a notable impact on the stakeholders involved.In the weeks following the creation of the radio show-style recording, NGO workers observed a signifcant impact on stakeholders, with participants expressing surprise and women farmers sharing their stories, a frst for them.
"[The stakeholders] said they have never done something like this before, nobody asked them to listen to their stories by capturing it as a radio show which could be shared with others, even with the policymakers.

" [NGO-1]
Refecting on the workshop days and comparing them to other initiatives, the NGO workers mentioned that they had not seen such excitement among the women farmers during their other project activities.The stakeholders played the recording from their mobile phones for their family members, neighbors, friends, and relatives.They wanted to broadcast their recording somewhere online and on an actual radio station.NGO workers posted their radio-style audio recording in their Facebook group, where many people were members as part of the project activities to share information.NGO workers also advocated for participatory audio recording work by women farmers within their networks, such as with agriculture extension services.
"That recording was just awesome.I had a discussion about this community engagement for recording their issues with our agriculture extension ofcer.He was impressed.He wanted to listen to the recording and asked me to share the recording with him." [NGO-1] When considering the approach, the NGO workers clearly appreciated the psychological safety aspect of the method we had used.They had been in situations where the women farmers had not spoken up because of the presence of a person who had made them uncomfortable many times before.
"Audio recording could be a way to capture sensitive stories that, as NGO workers, we always struggle to document women's real experiences.... Giving them options to capture their experiences through an audio recording, they will be more comfortable sharing their stories."  They also mentioned the power of being able to share the community's voices directly with their own superiors.This was something they tried to do in the past but had always felt they had missed.
"We also try to listen to community needs and challenges to share with our top ofcials, but we can't share the ways that communities can share their stories.Something will always be missing.When community people talk about sharing their stories, it comes from the bottom of their inner hearts, which is as powerful and true as their lived experiences.Nobody can teach them to share those stories they captured in their recordings.It's their own language with their greater true feelings and expressions" [NGO -2] The NGO workers also felt that the voices could be shared with many other people beyond their superiors, seeing the recording as a fexible tool to bring community voice into discussions with agriculture extension services, healthcare facilities, charitable donors, and even policymakers.However, they also saw that the recordings were a tool that would not do anything without someone making a signifcant efort to use them.
"If you don't do anything with these recordings, then there is no point just capturing community voices and doing nothing.We need to take action for them and fnd ways to help them with their collective voices.
Governments and NGOs can listen to their needs and challenges and bring solutions as initiatives for them.Otherwise, their problems will always be there!"[NGO-3]

DISCUSSION
Refecting on the current state of the fve-year initiatives of the NGOs in the area through the lens of the benefts that community voice can ofer, we saw strengths and weaknesses in the implementation of the project.We observed the female farmers buying-in to the NGO's previous projects because they saw that the projects addressed their needs.However, we also saw clear examples of areas where the projects did not respond to the community voice of the women farmers, undermining their trust in the project, as detailed in their counter-collective narratives.We were able to document the shortcomings in how the project understood the context of women farmers, responded to their lived experiences, and met information needs.For example, we saw that farmers had some local skills not accounted for and they lacked certain skills and knowledge, especially market pricing.The long-term impact of the work on the lives of women farmers is hard to evaluate in isolation -the NGOs with which we worked were supportive of the method and observed changes in the women after we left.However, shortly after completing the work, the COVID pandemic prevented further contact, and a variety of other interventions are always being implemented by NGO teams, so it is difcult to attribute any long-term changes to a single cause.
We refect on the implications of our work for community participation in HCI4D, starting by highlighting the importance of the secure environment we built to empower marginalized voices and democratize development eforts.We then examine the importance of counter-and collective narratives as the true voices of marginalized communities, noting their importance in guiding accurate program and policy design.Finally, we discuss the importance of participatory audio creation, particularly in the form of radio shows, as a platform for amplifying marginalized voices while also examining the need to develop more context-specifc innovations and approaches for leveraging community voice as data for decision making.

Empowering Marginalized Voices through Safe Community Engagement
We directly heard from stakeholders that they are a marginalized group within their already marginalized local community.These accounts are consistent with research documenting the discrimination women face in patriarchal communities and the risks to which they are exposed [79,115,134,135].In particular, our participants did not have physical, cultural, or psychological spaces to give voice to their counter-narratives.In our two-day participatory community engagement process, we were able to create a safe and welcoming environment for them to amplify their voices and stories emerged that had not been heard in the previous fve years of development work in the area.We document our stakeholders freely talking about the challenges they faced, demonstrating that we had built a safe place for us to talk [81].This builds on other HCI concepts that suggest safe environments enable marginalized community engagement [32,101,129].We saw the emergence of a psychologically safe environment among our stakeholders as a result of following participatory approaches centered on the community [120] and encouraging a change in the problematic power dynamics rooted in patriarchal norms and cultural and religious values [80,135].
5.1.1Pre-process design to empower through safety.Prior to the workshop, we organized a safe meeting place without men present and designed a visual informed consent sheet that went into great detail to explain the purpose of the project and stressed the agency of the stakeholders.Our visual informed consent form, based on Value Sensitive Design [64], helped us develop a relationship with the community and illustrate the objectives, activities and consent needed [120].Our experience is aligned with other work with lowliteracy groups which describes challenges because they may not understand development or design processes, their goals, or see how participation could be applicable to their daily lives [9,34,84,99,118,136] and speculate this may be due to their cultural values not aligning with western knowledge-based economies [48,84,99,103].

5.1.2
In-process democratization.Without men present in the sessions, participants were encouraged to form their own groups and think about their own intra group power structure.Although many of the women were able to socialize in the evenings, this was an opportunity for them to discuss their challenges knowing that men would not interfere.They arranged themselves into groups based on their age so that they would feel comfortable voicing their issues and deference for older people would not interfere with their groups or constrain them.We found that this encouraged open-ended creative sessions of ideation, storytelling, and counter-storytelling and helped to maintain a healthy balance of democratic power within groups aligning with other critical frameworks in the feld [97,130].Self-selecting groups or preexisting groups used in qualitative research are known to build confdence in some settings [28].
5.1.3Agency over process.An essential facet of participatory HCI is the value we place on the agency of stakeholders [147].Although we set out with a plan to capture multiple independent vignettes from our stakeholders, this was refned in light of their desire to produce a concrete, singular piece of audio in the form of a radio show that directly articulated their needs to development program designers.Stakeholders were empowered to delegate to others to voice their concerns.Using ideas of representative democracy, they chose six women to speak for the group as a whole so that all their voices would be heard in the audio production process without all of them needing to dedicate time to participate in the process.The accountability of the recording on the frst day revisited on the second ensured that less articulate women's points were voiced by a member of their community familiar with the challenge.In fact, when analyzing the data, our fndings aligned with the content of the video showing the alignment of the process and the radio show product.Our study builds on the concepts of agency in HCI4D by focusing on agency over the community participation process [94,147,154].Our study suggests that participatory methods to capture the voice of the community can be used for evidence-based data collection and validation, allowing participants to cross-check and interpret data, thus promoting democratic approaches [117].

Empowerment through Safe Technologies.
The theme of safety is carried over to the requirements of technology designed to meet the needs of farmers.Although there are a myriad of ICTs that can relay data from a source to an information seeker, the lack of information in this case is both local (so not held in a single authoritative repository) and deliberately concealed (with women farmers actively prevented from discovering the information they need).This leads to two challenges for designers: frst, how this information can be collected when the majority of its owners do not wish to share it with the women farmers, and second, how to deliver this information in ways that are discreet even when used on shared computing devices.We can envision systems that allow one woman, when hearing this information, to share it anonymously with a group online, and we can imagine that innocuous forums are used to present it.However, online access is limited for most women, so how could this information be discreetly shared via radio broadcast?Other work in this region has shown that this type of information can be shared successfully [36], but we must ask how we can ensure that it will reach women, when men usually control smartphones and radios in families.

Participatory documentary radio shows to hear community voices
Radio is a medium, a set of formats, and a technology familiar to our stakeholders.It is the only mass communication technology to which all our stakeholders have access, so when asked to create participatory audio, the stakeholders wanted to create something in this format.Their choice of a 30-minute documentary show spoke to the desire to control the presentation and framing of their challenges with hosts, stories, commentary, and locally produced songs and echoes desires expressed in other participatory media research [14,19].This format is a marked departure from the simple audio format we envisioned, but yields more interesting and authoritative results.The host's commentary gives more weight to the individual stories, and rather than having the stakeholder community produce multiple messages asking for help, placing them in a powerless position -they take on a journalistic role giving weight to the issues they are documenting, similar to community journalism [100].Instead of each stakeholder asking for help with matters they cannot resolve, they create media that illustrate the problems in their community and urge decision-makers to address them.By uniting their voices in this way, marginalized communities can be empowered, encouraging a sense of responsibility and community ownership.
The radio show format also contributed to the sense of a secure environment due to the mild anonymity provided by audio recording, a signifcant consideration within a patriarchal society and sociocultural structures where women's visibility outside of their homes is constrained.We saw that our stakeholders presented previously unspoken stories about women's health, including taboo topics such as pregnancy, childbirth, and menstruation.Although our work highlights the potential to create a safe environment for voice sharing, we emphasize the need for more research on privacy and information sharing, aligning with broader privacy and security issues in HCI4D [5,124].In particular, concepts such as two-password system access, which allows someone to "reveal" fake content when pressured, could be extremely useful in this context [4][5][6].
Our research contributes to the larger body of work that contends that participatory community radio is an important medium to link the disadvantaged with policymakers [13,33,61,86].Our work specifcally highlights the contrast between producing data points and producing more cohesive narratives, and between being a radio producer as opposed to a guest.The format means decision-makers can easily hear from community members through radio broadcasts, providing them with a wealth of information that can be used to shape future programs and interventions.Designers could take this concept from the design process into system deployment to allow the local community to broadcast radio that shares the types of vital information people in the Char regions.Simple local weather forecasts and storm warning could have a huge impact on the success of farming communities, but radio stations do not focus on this region with their programs.Systems such as RootIO [41] shows that low-cost radio broadcast is possible.

Voicing dissent: counter-collective narratives for decision making
Our study demonstrates the value of creating a safe environment that not only respects stakeholder agency, but also empowers and democratizes marginalized groups, enabling them to share their views with less inhibition.By advocating for the establishment of psychological and cultural safety in community engagement, our work aligns with broader initiatives that promote the creation of "safe spaces" [81] and "third spaces" [101] in HCI and decolonization studies, aimed at empowering the disadvantaged, as well as the principles of "psychological safety" in business studies, which contribute to organizational and economic development [54].
Our process not only has provided a platform for stakeholders to capture multifaceted insights that encompass local contexts, lived experiences, emotions, and indigenous knowledge, but has also played a role in the development of trust-based relationships.This helped glean counter-narratives, revealing suppressed experiences that contrasted starkly with ofcial statements made by decisionmakers.For example, within our participant community, NGO workers and government extension agents frequently reported successful work.However, the narratives of our participants revealed the reality that these services had not produced positive impacts due to lack of community consultation or poor design.Their countercollective narrative painted a consistent picture of the many issues that needed attention.Our work went beyond storytelling; it created a vital space for participants to express their dissent collectively and communicate their experiences as a united front.This is an extension of the old ideas of democratization and empowerment found in participatory design [147].However, where that focused on power dynamics between manager and worker stakeholders in the process, HCI4D work can contend with more severe power imbalances.Our method addresses this by removing pressure from individual stakeholders and builds a counter-collective voice that challenges dominant narratives by not including those dominant voices in the work.
The Benefts of Community Voices (BoCV) framework [122] is a useful tool to understand the benefts granted to underprivileged communities through the collection of in-depth knowledge about their contexts, subjective experiences, traditional knowledge, and the trust-building procedure.These insights can help to understand the needs and priorities of the community in the design, implementation, monitoring, and evaluation of development projects [119,121,123].We suggest that decision-makers actively listen to counter-collective narratives that exemplify empowered community voice [113,114,119,121].This understanding of priorities on the ground, not captured by traditional reports or data, aligns with the growing interest in political design in HCI [17,123,150] that aims to incorporate citizen voices into sustainable development and policy.We envision this to narrow the gap between decision-makers and citizens, facilitate community ownership and accountability over projects, and foster long-lasting relationships and trust building between decision-makers and communities [121].This would then lay the groundwork for other co-design approaches by addressing issues that could be outside the scope of the methods they employ [147].Design frameworks could be selected purely on the basis of specifc cultural values or to meet the need for cross-cultural design [29].
Currently, there are no easy ways for women to build their counter-narratives collectively in ways that can be shared.This could be a design space for HCI4D researchers to explore.How can we safely enable private uncensored communication on shared devices?Enabling communication between people like this can lead to enormous changes and upheaval [60].Previous work in HCI has looked at the various ways in which we can support activism for political change [59,91,148] with various groups that experience marginalization including refugees [2,105,137], migrants [51] and survivors of domestic abuse [37,49].However, while HCI has begun to consider the ways it can improve women's safety [10], including in developing countries [83], the specifc challenges in this context demand more exploration as we need to answer questions about what communication can look like between people that we cannot assume to be literate, how they can keep their communications private with shared devices, and what risks design like this might open them up to.

CONCLUSION
This work seeks to broaden the scope of HCI4D studies that incorporate community feedback in design and development by focusing on marginalized groups.Our work highlights the need for psychological safety to capture counter-narratives from marginalized groups within disadvantaged communities and shows that media co-production can support this.Our research has revealed that creating a psychologically safe and empowering environment motivates marginalized stakeholders to share their stories and experiences and ultimately construct counter-collective narratives together.Furthermore, our study has shown the value that stakeholders place in controlling how their stories are presented to decision-makers.Participatory audio recording, particularly in the form of radio programs, amplifes the voices of marginalized people.It is essential that we remain focused on amplifying community voices to promote programs, policies, research, and development that are inclusive and community-driven.This will lead to more efective, democratized, and sustainable HCI4D outcomes, bringing the advantages of development and technological progress to all, particularly those in disadvantaged groups.

Figure 1 :
Figure 1: The frst image shows stakeholder engagement during topic familiarization and story sharing, and the second image shows fnalizing collective voices in their larger group (permission to use images was granted by participants).

Figure 2 :
Figure 2: The frst image shows four experienced women farmers during their talk, and the second image shows six women representatives of the larger group (permission to use images was granted by participants).