Previous research has been conducted on the use of voice commands by different users, encompassing their performance, search pattern strategy, and effectiveness. For instance, Schalkwyk et al. (2010) comprehensively described the history of voice search efforts at Google in their research. They provided an in-depth description of the technology employed at Google and the challenges faced in making voice search a reality. Another study by Yi and Maghoul (2011) investigated the characteristics of search queries submitted from mobile devices using Yahoo! Search for Mobile. They focused on users’ behaviour and found that people may use voice queries when required, which are considerably longer than typed mobile queries. Additional studies have been conducted to understand how users overcome challenges in voice user interfaces (Myers et al., 2018). When users employ voice to search for information, they frequently need to reformulate queries due to voice input errors such as speech recognition errors and improper system interruptions. In a study conducted by (Jiang et al. 2013), it was found that language plays a crucial role in voice search, particularly in relation to user query reformulation behaviours. The study involved user experiments with native English speakers, revealing that these individuals often rephrase their queries using both lexical and phonetic alterations. The research sought to assess the effects of common voice input errors on users’ search advancements and the efficacy of various reformulation strategies in mitigating these errors.

In another study (Guy 2018) analysed voice queries in mobile web searches with a particular focus on the semantic and syntactic characteristics of these queries. His empirical evaluation revealed that the language used in voice queries more closely resembles natural language compared to typed queries. His study suggests that voice search functionality should incorporate voice-based result presentation to facilitate a fully hands-free interaction with users and also emphasise the development of new metrics for evaluating user satisfaction with voice queries. Xing et al. (2020) attempted to grasp users’ performance, query characteristics, query strategy, spoken conversational search, and perception of voice search by reviewing 27 publications. They found that the majority of the studies were primarily conducted with young and well-educated populations. Moreover, (Kiesel et al. 2020) applied voice-based search to investigate argument mining and conversational search. Through an online survey, they found that participants expected to primarily use voice-based argument search at home, either independently or with others. Additionally, they anticipated that such search systems would provide comprehensive information on retrieved arguments, such as the source, supporting evidence, and background knowledge on the mentioned entities or events.

(Crestani and Du 2006) conducted research on the retrieval effectiveness of spoken and written queries. Their results demonstrated that using speech to express one’s information needs enables a more natural expression and encourages the formulation of longer queries. However, despite spoken queries being longer in terms of the number of words, they do not appear to be significantly more effective than written ones. In a recent study, (Jha et al. 2024) conducted an examination of voice-based assistants and their role in voice-based searches for accessing information on the internet. The study observed that the methods through which users retrieve content are influenced by alterations in search engine algorithms. These modifications have implications for the digital marketing approaches employed by websites to attract visitors, establish brand identity, and execute promotional campaigns.

YouTube plays a significant role in entertainment by providing a vast repository of videos catering to diverse interests. It has established a unique role as a repository of popular culture, creating a diachronic archive over time and expanding synchronically in its scope (Arthurs et al., 2018). A key factor in YouTube’s popularity is its accessibility; it is free to use, and while creating an account enhances the experience, it is not necessary to have an account to watch videos. In addition to entertainment, YouTube serves as an online educational platform. For example, research has shown that YouTube is utilised in Malaysia to teach social expression (Raj et al., 2019). Another study by Shoufan and Mohamed (2022) reviewed 647 papers and found that YouTube is a rich, free, easy-to-use, and enjoyable source of learning content for students. Furthermore, YouTube has been identified as an effective pedagogical tool for blending education with entertainment, creating engaging learning experiences beyond traditional classrooms (Hua, 2015). By prioritising entertainment, these channels capture larger audiences and maintain subscriber interest, making learning more accessible and enjoyable.

Moreover, YouTube is increasingly being used as a platform for disseminating health information (Madathil et al., 2015), providing valuable insights into various diseases (Delli et al., 2016; Onder et al., 2021; Diers et al., 2023). During the COVID-19 pandemic, YouTube became a significant source for accessing medical information (D’Souza et al., 2020; Andika et al., 2021). Additionally, YouTube serves as a social media platform that helps farmers improve agricultural practices (Kesherwani et al., 2022), especially through agriculture-related videos that can enhance the skills of young farmers (Paudi et al., 2022). Furthermore, YouTube serves as a valuable source of fishing techniques, enabling fishermen in Indonesia to expand their knowledge base by accessing various videos on different fishing forums (Pratiwi et al., 2023).

However, digital non-participation among marginalised populations, such as people living with disabilities, older individuals, disadvantaged youth, and women, compounds existing disadvantages (Davis & Farmer, 2016). To address this issue, several marginalised individuals with literacy challenges are using voice commands to access information via smartphones.

Several studies have explored various aspects of voice search and video content on YouTube, addressing diverse user needs across different demographics. For example, (Balasuriya et al. 2018) identified barriers and opportunities in voice-activated interfaces, specifically examining how intellectually disabled users in Australia utilise voice commands to search for YouTube content. Similarly, (Kversøy and Kversøy 2018) investigated how individuals with intellectual disabilities in Norway use YouTube as an alternative search tool. Research has also focused on age-related patterns of YouTube usage. (Lee et al. 2022) found that older adults in South Korea actively engage with YouTube, despite concerns about fake news and costs, viewing its content as more unbiased than traditional media. (Hourcade et al. 2015) analysed YouTube use among infants and toddlers, observing that interactive content, particularly educational apps, supports the development of children’s perceptual and motor skills.

YouTube has also been examined regarding the personalisation and dissemination of misinformation. Hussein and Juneja (2020) demonstrated that while demographics such as age, gender, and location have minimal impact on misinformation in search results for new accounts, these factors become influential once a watch history is established. Beyond these studies, other research has addressed the broader implications of algorithmic technology, biases in video recommendations, and factors contributing to video popularity on YouTube (Roth et al., 2020; Bryant, 2020; Foster, 2020; Lutz et al., 2021; Kirdemir et al., 2021; Gran et al., 2021; Brown et al., 2022; Haroon et al., 2022; Vybihal & Desblancs, 2023).

Despite the existing research, a significant gap remains in the literature concerning the experiences of marginalised communities in low- and middle-income countries of the Global South when using YouTube through voice commands. While prior studies primarily focus on user experiences in developed country contexts, there is limited evidence on how marginalised groups, such as small-scale marine fisherfolk in Bangladesh with limited literacy and digital skills, engage with YouTube. Specifically, it is unclear whether these communities can effectively achieve their intended goals when using the platform. Failure or difficulty in accomplishing these objectives may lead to an unjust outcome, reflecting what Beard and Langstef (2018) describe as "unjust use”. To address this gap, the present study adopts an interpretive phenomenological approach to comprehensively examine why individuals from a marginalised community interact with YouTube using voice commands and the subsequent consequences of this engagement.

(Oliver and Foscarini 2014) define information culture as the value attributed to information and attitudes toward it. In line with this concept, the first author investigated various critical information required by fishermen to achieve different purposes in their everyday lives. He collected data on how fishermen accessed this information, focusing on their preferred methods, the reasons for these preferences, their challenges, and their level of trust in different types of information. Ethnographic fieldwork was conducted on the island between December 2023 and January 2024 to collect data for the project. This involved four focus groups, each comprising 8– 10 participants and lasting approximately 1.5 hours, and 10 in-depth interviews with male and female participants. Male participants were fishermen directly involved in fishing, while female participants were family members of fishermen. This approach provided a deeper understanding of the community’s information practices and the socio-cultural dynamics influencing their access to and use of critical information.

The utilisation of an ethnographic approach in social research enables a deeper understanding of behavioural factors among participants, which may not be easily distinguished through other data collection methods. In this study, field data collection took place during the fishing season, which presented challenges in recruiting participants, as most fishermen were either at sea or preparing for upcoming trips. To address this, the first author asked them when they would be available and organised the focus groups and interviews at their preferred time. Given these logistical constraints, non-probability sampling was employed, resulting in a small but targeted sample. Snowball and purposeful sampling were utilised for focus groups and interviews, respectively, to ensure the effective selection of participants.

Embracing the ethnographic approach, the first author who conducted the fieldwork made multiple visits to various villages, including the fish landing station, where fishermen sell their catch. The author speaks the same language as the fishermen, which enabled him to interact directly with them by participating in activities such as drinking tea and sharing snacks. This immersion fostered trust and friendships, allowing the researcher to become a part of their environment. Furthermore, ethnographic fieldwork, strengthened by interviews and focus groups, responds to the realisation that verbal descriptions may not always accurately represent reality and should be complemented by observation. According to (Kawulich 2005), individuals may not always act in accordance with their verbal descriptions of their behaviour. Therefore, the ethnographic approach to data collection serves to illuminate how individuals perceive and articulate their everyday experiences.

To bolster the credibility of the study findings, three key strategies were employed: prolonged engagement, reflexivity, and triangulation (Ahmed, 2024). The author’s one-month stay on the island, coupled with various data collection methods, contributed to the trustworthiness of the data. The rigour and reliability of the findings were further guaranteed by the continuous application of triangulation at various phases of data collection, as outlined by (Johnson et al. 2020). The incorporation of ethnographic fieldwork in conjunction with interpretive phenomenology enabled the documentation of the lived experiences of fishermen concerning their use of YouTube as a source of information.

To conduct data analysis, this project followed the Reflexive Thematic Analysis (RTA) approach proposed by Braun & Clarke (2006). RTA is a theoretically flexible and accessible interpretative method for qualitative data analysis that facilitates the identification and analysis of patterns or themes within a given data set (Braun & Clarke, 2012). This reflexive approach emphasises the active role of researchers in knowledge production, encouraging reflective and thoughtful engagement with both the data and the analytic process. Braun and Clarke identified several theoretical assumptions that should be addressed when conducting RTA, conceptualised as a series of continua: essentialist versus constructionist epistemologies, experiential versus critical orientation to data, inductive versus deductive analyses, and semantic versus latent coding of data (Byrne, 2021). For this research, an experiential orientation was chosen as it aligns with the objective of this research. Adopting an experiential orientation involves understanding that participants’ thoughts, feelings, and experiences reflect their internal personal states, aiming to reflect the experience of a social reality. Additionally, it helps acknowledge the social construction of the research topic.

The first author was responsible for collecting data in the field, analysing it, and disseminating the findings to the other co-authors through regular meetings to ensure their familiarity with the data. The initial codes were established to categorise the types of information accessed by YouTube, including the purpose, implications of information utilisation, proficiency in using YouTube, and justification for relying on the provided information. The data was then analysed to find common underlying patterns from which themes were identified. For example, social norms were identified as an important theme because YouTube used algorithmic technology that went against the societal norms of the fishermen.

Accessing timely weather updates is another crucial information for the fishermen and other residents of the island, as accurate and quickly accessible information can potentially save lives during natural hazards such as cyclones. Fishermen, in particular, frequently check weather forecasts before embarking on expeditions. Failure to get such updates can have serious repercussions. While people can get weather information via traditional sources such as television news or loudspeaker announcements from local governments and non-governmental organisations, these are not the preferred methods because updates from these methods frequently come only after the environmental conditions have already worsened. Smartphones and internet connectivity, particularly through YouTube, offer a quicker and more convenient way for fishermen to receive weather updates, even well in advance of potential cyclones in the Bay of Bengal. As a result, many fishermen consider smartphones and YouTube as their preferred methods of accessing weather information.

Although YouTube is the preferred platform for accessing weather updates, many fishermen lack the necessary skills to effectively navigate smartphone applications. As a result, they rely on the assistance of more proficient peers within their community. For instance, in one village, a fisherman with higher skills in accessing weather information via a smartphone has earned the nickname ‘Abhaoa’ (meaning ‘weather’ in Bengali) due to his role in helping others check weather updates on YouTube. His knowledge stems from prior experience as a volunteer in a flood monitoring project conducted by the Bangladesh Water Development Board, where he learned to use a smartphone for flood monitoring. However, like many others in the community, his illiteracy necessitates the use of voice commands to access weather information.

Fishing is a labour-intensive occupation, and due to the challenging socio-economic conditions in many fishing villages, people often begin fishing from a young age, as by engaging in fishing, they can start earning an income from childhood. On this island, becoming a fisherman is a process of learning by doing, with individuals gaining experience through repeated participation in fishing expeditions. To become a skilled fisherman, one must have prior hands-on experience, as the profession requires direct engagement with the sea, often necessitating extended periods of time away from home.

When at sea, fishermen are constantly occupied with their work, leaving them exhausted upon their return. As a result, they value their leisure time greatly and make every effort to enjoy it. Furthermore, to protect marine species, Bangladesh enforces several fishing bans throughout specific times of year, preventing fishermen from going to sea. During the fishing ban periods, entertainment becomes an important part of their everyday lives, as most fishermen lack alternative employment or skills other than fishing.

In these instances, fishermen rely on their smartphones for entertainment, with YouTube being a popular choice. They mostly watch local movies, dramas, songs, religious music, and news. Because of their illiteracy, fishermen often utilise voice commands to access content-related entertainment. However, once they are on YouTube, they frequently watch the videos recommended by YouTube and watch content nonstop without actively looking for anything in particular. Accessing YouTube is straightforward for them because it simply requires an internet connection, allowing them to quickly and easily access audio-visual content without the need to register an account. Since they tend to watch videos one after another, their internet data is consumed quickly, requiring frequent recharges, which increases their overall internet usage costs.

Fishermen on the island increasingly rely on smartphones and YouTube for essential information, such as weather updates and entertainment, making these tools integral to their daily routines. This dependency has cultivated habitual usage and made smartphone use more pervasive (Oulasvirta et al., 2012). For many illiterate fishermen, YouTube serves as a crucial resource, enabling them to access information through continuous video content, which compensates for their inability to read. While this reliance underscores the value of YouTube in fulfilling their informational needs, it also results in substantial economic costs to buy mobile internet due to the extensive consumption of video content. (Rumpf et al. 2022) highlighted the significant economic costs associated with problematic internet usage (PUI). In this context, behaviours such as online gambling, viewing adult content, and excessive use of social media are considered forms of PUI that contribute to financial strain.

Fishermen, due to their inability to read, are often unable to make informed decisions about the video content they consume on YouTube. When using voice commands, they cannot understand video captions or titles, forcing them to randomly select videos. This repetitive process of checking one video after another not only increases their data usage but also contributes to the perception that internet costs are unreasonably high. These fishermen act merely as viewers without engaging in any form of interaction on the platform, such as liking, commenting, or sharing videos (Sui et al., 2022). Moreover, they lack knowledge of data-saving techniques, such as adjusting video quality, using data-saving features, or preloading content on Wi-Fi (Whitely, 2024). Although the Internet is seen as an essential infrastructure for audio-video content access on YouTube (Lehr & Sicker, 2017), this increased consumption continues to burden them with high costs.

Fishermen use voice commands in Bengali (the local language), and the system often provides outdated information, especially regarding weather updates. This happens because voice search relies on Google’s audio recognition technologies, which require proper settings to be turned on (YouTube, n.d.). Unfortunately, fishermen are generally unaware of this requirement, as their smartphones were set up with the help of the local mobile technician and without proper guidance or training. Additionally, they often use YouTube without a Google account and remain unaware of the potential benefits, such as improved voice recognition technology based on saved audio data (YouTube, n.d.). This lack of user knowledge limits the effectiveness of YouTube’s voice recognition system and prevents the fishermen from optimising their use of the platform to meet their informational needs.

Search and recommendation algorithms shape content results based on user queries and feedback, retrieving results according to individual user preferences. These algorithms increasingly rely on user profiles to optimise the content they present (Boratto et al., 2020). However, a critical issue with these algorithms is the prevalence of popularity bias, where widely viewed or highly rated content is consistently prioritised, creating a feedback loop that amplifies its visibility over time (Kirdemir et al., 2021). In the context of fishermen seeking entertainment, such as music videos, this bias often results in the prominence of adult content and gambling advertisements. The algorithm misinterprets their repeated consumption of video content as a preference for such material, reinforcing these undesirable recommendations. Consequently, much of the online content fishermen encounter is shaped by algorithms they do not understand, leading to distorted or harmful online experiences (Lewandowsky et al., 2023).

Biases inherent in recommendation systems, stemming from historical data, can exacerbate societal inequalities by influencing everyday information access. Lack of awareness about algorithms can exacerbate the digital divide, leaving vulnerable populations at a greater disadvantage (Gran et al., 2021). When these biases touch on sensitive attributes like religion or gender, they can reinforce discrimination and deepen existing inequalities (Boratto et al., 2019). In the case of young fishermen, persistent exposure to adult content and gambling advertisements not only alters their mindset but also leads to broader societal problems, impacting their families and communities.

Due to widespread illiteracy, many fishermen are unable to read video titles, which leads to confusion and indecision when selecting content. The exponential growth of digital information has exacerbated the problem of information overload, which significantly affects decision-making and productivity (Shahrzadi et al., 2024). For these fishermen, this overload, combined with their limited literacy, makes it challenging to navigate YouTube effectively. Moreover, the algorithm- driven technology behind YouTube fails to consider their literacy, further complicating their ability to access relevant information. As a result, their access to information via YouTube becomes problematic, much like how overconsumption of junk food negatively impacts health, as it does not consider their capacity to engage meaningfully with the platform.