Multimedia and multimodal data

One of the challenges of open research design in ethnomusicology is clarity over the definition of data. Archivist Sean Luyk points out that there is a “notable gap in the literature regarding humanities research data, and very little about ethnomusicology, or music at all for that matter” (Reference Luyk2015:3). This is a problem for ethnomusicologists because our data takes so many forms, extending considerably beyond the audiovisual recordings that tend to be the focus of available literature. Our data include text-based formats that are common across the social sciences and humanities—like questionnaires, interview transcripts, and fieldnotes. But it also includes music notation and transcriptions that may require specialised knowledge, systems, and software to generate, preserve, and access. The ephemera of performances and participatory cultural activities are data, and include promotional materials, sacred or ceremonial items, documentation of protocol, or elements of material culture—like instruments and costumes. Our data include the methods and results of creative processes that exist in a variety of formats, each with distinct challenges for preservation and sharing—and which may be best suited to storage within a physical archive or museum. Our data also comprise sound and video recordings that document wide-ranging sound and movement practices situated in vastly varied contexts.

Recordings serve a variety of purposes, ranging from documentation of performance practices and repertoire to supporting ethnographic observation (or other methodological functions), aiding transmission processes, and surfacing relevant cultural protocols. The range of functions for field recordings is broad, with implications for the nature and re/use of recordings (e.g., quality, perspective, technology, and audience). Landau and Fargion (Reference Landau and Fargion2012) prompt consideration of what is meant by recording “for our own research purposes” in their preface to a special issue of Ethnomusicology Forum. Included articles offer twenty-first-century case studies of how archival recordings offer meaningful tools for community engagement and collaboration, with ethnomusicologists filling roles as “culture brokers” in efforts to repatriate and/or proactively build local archives. The special issue challenges the ambivalence to archives that emerged as a response to our discipline’s colonial history during the twentieth century by demonstrating the decolonising potential of sound collections.

The authors do not, however, clarify what ethnomusicological data are or differentiate uses for recording. As Landau and Fargion point out, “much of the deliberation and debate” related to the methods, technologies, and ethics of collecting and disseminating ethnomusicological data has tended to happen “for and within the audiovisual archiving world” even though “recordings, archives and ethnomusicology have been inextricably linked throughout their histories” (Reference Landau and Fargion2012:128). They suggest that, despite being written by ethnomusicologists who are also archivists, “the existing literature on audiovisual archiving has not reached many ethnomusicologists” (Reference Landau and Fargion2012:128). This siloing of discussion in specialist literature misses something important about the nature of ethnomusicological data: its multimodal nature. Commentary instead focuses on recording for documentation—recording that prioritises the creation of archival objects that are understandable as standalone artefacts—without explicit consideration of how sound recording functions as a research process (cf. Fargion Reference Fargion2009).

Roshanak Kheshti’s (Reference Kheshti2015) theorisation of the “aural traces” of sound recordists, though directed at the production of world music recordings, is useful for thinking through this question of recording for documentation versus research, and what that means in terms of the production and preservation of data. Kheshti points out that recordings capture more than sound; decision-making based on the recordist’s priorities also leaves marks, influencing focus, start and end points, mix, and so on (Reference Kheshti2015:32). Recordings made “for our own research purposes” include traces of research methods and goals. They might, for example, offer low-quality sources that support transcription, document a transmission process rather than sonic or visual practice, or support practitioner reflections on their praxis. Recordings “for research purposes” may document something quite distinct from music or dance, and meaning may be contingent on other types of documentation. Data, in other words, that are truly multimodal, methodological, and ill-suited for publication without substantive contextualisation or links to larger datasets. Focusing the debate about data, preservation, and dissemination within specialist discussions of sound archiving neglects this multimodal nature, leaving us ill-equipped to address challenges inherent in the unique nature of our research.

This lack of clarity has implications beyond our discipline: if we cannot clearly articulate what our data is, how can we expect the people who support us to understand? The multidisciplinary research ethics boards responsible for reviewing our methods and developing institutional policies, for example, may misunderstand the nature of data generated through ethnomusicological research, resulting in frustrations on all sides and overly risk-averse processes that do not necessarily serve the interests of stakeholders (Bishop Reference Bishop2009; cf. Wakaruk et al. Reference Wakaruk, Secker and Morrison2025). Definitional clarity is also a problem for data management professionals, whose systems may be ill-equipped to deal with collections featuring widely varied forms and formats with specific regulatory challenges. Many libraries and archives divide responsibility for their collections by format, meaning that contextualising texts, transcripts, or notation may be separated from audiovisual recordings (e.g., Wittgren Reference Wittgren2025). This challenge is compounded by a reality that not all libraries, archives, and repositories have the infrastructure, knowledge, or mandate to preserve and disseminate all data types. Writing on the proliferation of born-digital heritage objects—inherently multimodal objects that require “the existence of increasingly complex and flexible systems”—Goudarouli et al. point to the increased necessity of “high levels of technical skill from both curator and researcher” to manage “unique preservation and access challenges” (Reference Goudarouli, Sichani and Winters2026:4). In other words, effective management solutions for multimodal data require multidisciplinary collaborations.

Privacy and data protection

Much of the tension between open and ethnographic research relates to the privacy of our interlocutors. Ethnographic research depends on rich contextual details that are likely to identify research participants, even when direct identifiers are stripped from the data. Such details are essential to the meaning and significance of the data. Ethnomusicologists take a negotiated approach to identifying research participants, recognising that the co-produced nature of our research comes with ethical obligations to respect interlocutors’ wishes about how they are identified. Some musicians and dancers want to be publicly recognised for their contributions, while others eschew acknowledgement or prefer anonymity (Hield et al. Reference Hield, Wettermark, Hore, Packett, Whitehead, Shields and Wilson2024). The nature of our research area, too, means that separating traditional knowledge from originator(s) poses risks to creators and communities: tracking data provenance is essential to ensuring appropriate access and usage (Zemp Reference Zemp1996; Guy Reference Guy and Jaarsma2002; Draisey-Collishaw Reference Draisey-Collishaw2025). In short, research that focuses on expressive practices often requires identification of practitioners because (1) documenting the practice requires accurate capture of physical likenesses and/or identifying sonic features, and (2) practitioners need to be named in the metadata to protect their intellectual property and/or cultural rights.

Research contexts come with distinct regulatory and structural challenges when it comes to identifying research participants. General Data Protection Regulation in the UK (like the European Union) defines legitimate terms for processing and storing personal data. Under UK GDPR (2016), personal data are “any information relating to an identified or identifiable natural person (‘data subject’),” including “a name, an identification number, location data, an online identifier or to one or more factors specific to the physical, physiological, genetic, mental, economic, cultural or social identity of that natural person” (ICO n.d. Footnote ⁹). This definition covers any information that can identify a person, including images and recorded sounds. While UK GDPR (2016) defines processing personal data for research and archiving as legitimate interests, this regulatory context complicates access to infrastructure for preservation and data sharing, especially when it comes to data that include ethnographic description or creative outputs.Footnote ¹⁰ Digital data repositories that make collections freely available online, for example, will not accept data where individuals are identifiable without their explicit consent. This can be a difficult bar to meet, especially if documentation includes public, informal, and/or participatory performance events.

Archives and repositories that offer “controlled access”—that is, where access is limited to authorised individuals or groups who demonstrate legitimate interest—are better able to preserve and share datasets that include unconsented personal data and data with complex intellectual property and cultural rights, but the availability of suitable options is limited in England and internationally (e.g., Harley et al. Reference Harley, Acord, Earl-Novell, Lawrence and King2010; Hofmann et al. Reference Hofmann, Miksa, Knees, Bakos, Sağlam, Ahmedaja, Yimwadsana, Chan and Rauber2021). A 10 July 2026 search of the Registry of Research Data Repositories (re3data.org n.d.) listed only twenty-four repositories specialising in “musicology” data worldwide, with this number increasing to twenty-nine if “social and cultural anthropology and ethnology” repositories were included. Repositories that have specific geographic and/or cultural remits are often better situated to evaluate what constitutes legitimate access and take risk-managed approaches that are in dialogue with the needs of researchers and the communities they service (e.g., PARADISEC 2026). Research and cultural traditions that fall outside of these specialist remits, however, remain underserviced.

The ethics of privacy are of global relevance and especially complex in the present moment. Widespread adoption of artificial intelligence (AI) is coupled with regulatory uncertainty, resulting in unforeseen risks to individuals and communities represented in research data (Oruc Reference Oruc2025; Brunschweiger Reference Brunschweiger2026). Preserving and disseminating data in 2026 involves privacy considerations that simply were not relevant in 1986, when accessing data involved visiting a physical premises and negotiating usage with an archivist. Even in 2006, when digital and online technologies were changing how data were shared and introducing new threats, there was not the same potential for completely uncontrolled access and unauthorised dissemination by generative AI. And yet, these same technologies hold potential for safeguarding and servicing the needs of our interlocutors. Hemment et al. (Reference Hemment, Kommers, Ahnert, Antoniak, Arbix, Belle, Benford, Brintrup, Bryan-Kinns, Bunz, Caramiaux, Chung, Disley, Du, Duéñez-Guzman, Gius, Medina, Goodlad, Ilany-Tzur, Isaksen, Jirotka, Kennedy, Leslie, Livingstone, Long, Martin, Nalau, Nathan, Noel-Hirst, Ostherr, Prahl, Rana, Ratto, Revell, Rhee, Rutenberg, Seales, Sherman, So, Sobey, Stilgoe, Thain, Ubalijoro, Underwood, Vashishta, Vidmar, Wilkens, Wu, Yuen and Zeilinger2025) point to AI’s potential for responsibly and ethically processing data. Technologies like smart contracts, for example, may support reliable provenance tracking and responsible reuse according to creator/community licensing terms, as well as enabling the confidentiality of data sources. Assessing the risks and accessing the potential of such technologies, however, requires good data governance to start.

Intellectual property and cultural rights

Ethnomusicologists have long been aware that recordings made in the field have the potential to attract the interest of individuals who lack commitment to the rights of our interlocutors. Materials get taken and repurposed, used without reference to cultural meaning or protocols, and creators go unacknowledged and uncompensated for their knowledge and art. Scenarios for the exploitation of intellectual property and/or the violation of cultural rights are well-documented (e.g., Zemp Reference Zemp1996; Guy Reference Guy and Jaarsma2002; Pruett Reference Pruett2011), as are the incommensurate nature of cultural rights and existing intellectual property right regimes (e.g., Weintraub and Yung Reference Weintraub and Yung2009; Post Reference Post2017; Reed Reference Reed, Levine and Robinson2019) and unequal impacts of regulation across traditions and geographies (e.g., Stobart Reference Stobart2017; Born and Eisenberg Reference Born and Eisenberg2025). More positively, concurrent with the rise of easy-to-access digital technologies, archiving projects are demonstrating how ownership, control, and access to knowledge might be re/negotiated, either through repatriation of extant collections or through community-engaged digital preservation initiatives (e.g., Landau and Fargion Reference Landau and Fargion2012; Giri Reference Giri2025; Risk and Tulk Reference Risk and Tulk2025). While the issues raised by these initiatives are relevant to open research, reporting tends to be on individual projects with bespoke “archive as output” solutions. Little attention is given to more systemic approaches to data governance and mobilisation, or how risks and rewards might play out in research that does not prioritise archiving as a named project output (cf. Seeger Reference Seeger2023).

Why does taking a more systemic approach to rights management matter? Even under the best of circumstances, rights research is challenging and slow.Footnote ¹¹ Clearing an entire dataset retrospectively is almost impossible without metadata documenting provenance and terms of reuse (Diamond et al. Reference Diamond, Corn, Fjellheim, L’Hirondelle, Maniapoto, Marett, Partridge, Peria, Pirttijärvi and Stålka2017). Indeed, lack of relevant documentation is often a key factor that keeps ethnomusicological data locked away in private collections. That is because sharing data with intellectual property rights involves risks—risks to rights holders, risks to the people and platforms doing the sharing, and risks to (re)users. These legal, financial, and ethical risks are amplified in digital contexts where sharing is, to a degree, uncontrolled (and without the statutory and moral safeguards traditionally provided by brick-and-mortar archives).Footnote ¹² Open data repositories, in consequence, tend to be risk intolerant: only data that exist in the public domain or that have clear agreements in place with all rights holders can be deposited (e.g., UK Data Service 2026a; cf. Hemmasi and Brown Reference Hemmasi and Brown2025; Korn et al. Reference Korn, Kheria and Terras2026).

Risk aversion—and the desire to protect rights holders—might be reasonable, but digging into what this means in practice reveals the impracticalities. UK intellectual property law, for example, grants creators automatic and non-transferable moral rights over their creations. This includes the right of attribution: the right to be named as a creator. Original works also hold copyright(s) that give creators the exclusive right to control the use of their work. While moral rights do not expire, copyright typically lasts for at least fifty years after the death of the creator (Berne Convention 1886), though many countries—like the UK—have extended this to seventy years (UK Intellectual Property Act 2014). Given the multiplicity of rights holders involved in documentation of creative practices (e.g., rights holders for a single audio recording potentially include musicians, composers and lyricists, publishers, and the recordist), it is often impractical to identify (let alone gain consent from) everyone involved. Indeed, the increased duration of copyright over the twentieth and twenty-first centuries, when combined with the non-commercial nature of documentation and high rates of risk aversion, has resulted in what intellectual property rights scholar Naomi Korn terms “a black hole of 20th [and 21st] century content” suitable for digital preservation and dissemination (Korn et al. Reference Korn, Kheria and Terras2026).

While UK Intellectual Property Law includes a “fair dealing” exception that is relevant for research and preservation,Footnote ¹³ applying this exemption requires a degree of risk tolerance that is lacking in practice. Fair dealing lacks a statutory definition, meaning that the courts assess it on a case-by-case basis that accounts for whether usage affects the market for the original work, as well as whether the amount of the work taken is reasonable and appropriate (Wakaruk et al. Reference Wakaruk, Secker and Morrison2025). Updates in 2014 broadened fair dealing exceptions, especially in the areas of parody, citation, and research, but invoking exceptions in relation to data preservation tends to depend on the availability of controlled access archives (Intellectual Property Office 2014). Small and specialist multimedia archives that more commonly handle these types of data may use a variety of risk management tools (like provenance research and takedown notices) to assess, make available data, and manage potential rights infringements (Korn Reference Korn2021; Korn et al. Reference Korn, Kheria and Terras2026), but such services are unequally available and often underresourced.

Keeping data locked away might be a natural response to avoiding risks, but it also neglects the interests of the people with whom we do research and avoids surfacing the necessity of disciplinary-level debate about what constitutes ethical risk-managed approaches to sharing data with complex rights (Bishop Reference Bishop2009). While informed consent processes (written, aural, and/or continuous), contracts, and recording/image release forms might be norms in our discipline, workflows tend to be designed on a per-project basis and may prioritise mitigating institutional risk over long-term rights management with implications for data preservation and reuse beyond the project. More disciplinary knowledge is needed about the types of metadata tracking that need to be built into data collection or how use of culturally appropriate and internationally recognisable rights and licensing regimes—including Creative Commons, Traditional Knowledge Labels, Nwulite Obodo Open Data Licences (NOODL)—can become a foundation for relationship building with participants and partners (Local Contexts n.d.; Creative Commons 2026; Licensing African Datasets 2026).Footnote ¹⁴

Co-produced knowledge and access

Despite the specific challenges that ethnomusicological research poses to the FAIR Principles, they are an effective starting point for planning research, encouraging attention to consistent data description, use of open storage formats, and consideration of how data might legally and ethically be processed into the future. FAIR also offers standardised language for interdisciplinary communication about research and data management requirements. A fundamental limitation of the Principles, however, is their narrow focus on the characteristics of the data. Indeed, with their origins in the sciences and prioritisation of a digital-first understanding of data, the FAIR Principles do not adequately account for the ethical obligations inherent in co-produced datasets.

Data produced in conjunction with research institutions typically flows through researchers and remains in restricted project archives in compliance with data protection, privacy, and intellectual property regulations. With forethought and planning, researchers can resist this enclosure through contracts and agreements that ensure partners retain control of their data. These efforts, however, are constrained by broader structural challenges (Vince-Myers Reference Vince-Myers2025). Institutional data repositories serve the needs of researchers, with little consideration of other potential users; external partners are thus unable to directly deposit their data, necessitating dependence on researchers with institutional affiliations. Moreover, authentication processes that are in place to assess “legitimate interest” may restrict access to researchers with institutional affiliations. The authority to sustainably preserve, share, and benefit from data remains under the control of research institutions—a significant problem when the goal is more equitable access to knowledge (Ball et al. Reference Ball, Bowring, Hield, Pahl, Vergunst and Graham2019).

The Global Indigenous Data Alliance (GIDA) prompts attention to how histories and power differentials implicate the nature of data sharing. They explain that “greater data sharing alone creates a tension for Indigenous Peoples who are also asserting greater control over the application and use of Indigenous data and Indigenous Knowledge for collective benefit” (GIDA n.d.). As an alternative to FAIR, they propose the CARE Principles (Collective benefit, Authority to control, Responsibility, and Ethics) as a people- and purpose-oriented approach to data governance that both supports open research and prioritises the interests of stakeholder communities. The CARE Principles have been systematised through initiatives like the Traditional Knowledge Labels (Local Contexts n.d.), offering evidence from a variety of global contexts about how open research practices can be compatible with (1) local interests and cultural rights and (2) the data management systems of research, heritage, and memory institutions (Hudson Reference Hudson2024). Notably, applying the CARE Principles depends on the availability of an authoritative body (e.g., band council, council of elders, or appointed community advisory board) that can speak for collective interests (cf. Schulder and Hanke Reference Schulder and Hanke2022). Even in the absence of such a body, core ideas about how communities benefit, who has authority over the data, researcher responsibility, and ethical usage are critical for addressing the shortcomings of the FAIR Principles for ethnomusicological research and data—and for designing the systems that support access.

Making a plan: Institutional repositories and open research

We initially planned to deposit data generated through the SPGs in TUoS’s institutional repository, ORDA.Footnote ¹⁹ Using ORDA, research support services told us, is strongly encouraged in the absence of “a subject-specific repository or data centre” (TUoS 2026b)—something we lacked. As the academic team had already used ORDA to store and share early outputs, continuing with the same platform allowed us to build on our existing knowledge and keep our data in a single location. Though intended for making non-sensitive data publicly available online, we (wrongly) hoped that ORDA’s capacity to restrict and/or embargo access would enable us to navigate issues like the identifiability of our research subjects or the complex intellectual property rights inherent in our data (Draisey-Collishaw Reference Draisey-Collishaw2025). We remained keenly aware that ORDA is not a particularly intuitive or externally discoverable interface, especially for non-academics,Footnote ²⁰ but we planned on using more user-friendly and audience-focused formats to curate the stable links (also known as DOIs—digital object identifiers) generated through the platform. It was not an ideal solution but, faced with a dearth of viable alternatives, we figured that ORDA at least offered a stable, long-term access solution.

Based on guidance from TUoS’s open research and copyright teams about key regulatory challenges, our planning prioritised the FAIR Principles. We contemplated the data we anticipated collecting, how we sought to use it, and the risks associated with reuse. Based on CARE, we also considered who might benefit from access. We developed protocols—and elaborate spreadsheets—for recording critical metadata. Our consent forms flagged intentions to share data beyond the project and offered detailed explanations about what it means to be identifiable. We developed written guidelines for what our partners and their participants could share with us, highlighting preferred formats and file types. We also offered detailed descriptions of how intellectual property rights work in England and encouraged creators to apply Creative Commons licences as a means of protecting their rights and enabling reuse. After several iterations of the various forms, templates, collaboration agreements, and supporting workflows, research support services reviewed our proposed processes and deemed them adequate to our open research goals.

Practical challenges: Folk singing in England

The SPG projects focused on amateur participation in the folk scene. Labels like “singaround,” “song share,” “folk club,” “singing session,” “floor spot,” or “chorus singing” are insider terms describing a variety of communal singing practices that involve people coming together, often in a pub, around the practice of singing (Wettermark Reference Wettermark2025). Participation includes both professionals and amateurs, though status and role within the singing activity have more to do with perceived relationship to “the tradition” than professional standing. While there are clear leaders in the performance, that role may pass to others as people take turns, sometimes singing solo, sometimes in groups, and sometimes joining a soloist on choruses (MacKinnon [1993] Reference MacKinnon1994; Hield and Crossley Reference Hield, Crossley, Crossley, McAndrew and Widdop2014; Dunn [1980] Reference Dunn2015; Gordon et al. Reference Gordon, Hield, Porter, Selous and Wettermark2024).Footnote ²¹ Gatherings typically depend on volunteers for organisation, with those individuals determining the character of the activities they lead through (often unconscious) decision-making about everything from times and places for gathering, to the layout of spaces, flow, how participants join in (Hield Reference Hield, Russell and Ingram2013), behaviour moderation (Hield and Mansfield Reference Hield and Mansfield2019), the nature of acceptable contributions (Mansfield Reference Mansfield2026; Miller Reference Miller2026), and marketing.

The scene is steeped in discourses about the Commons, tradition, volunteerism, and aural transmission of repertoire, often mitigating desires to understand intellectual property law as relevant. Informal participatory music scenes such as this one depend on ideas about goodwill, fair play, and shared heritage to function (MacKinnon [1993] Reference MacKinnon1994; cf. McCann Reference McCann2001)—though these values may be contested when creators perceive their rights to have been violated.Footnote ²² Regular gatherings of singers, musicians, and dancers make recordings that they share, either informally on social media or through archives and commercial recording projects. Session facilitators, too, make recordings, compile songbooks, and share playlists that help participants learn repertoire. And, dancers use recordings they find on YouTube to accompany themselves in the absence of live musicians. As Jones and Cameron point out, “historically many tunes that have become part of the folk repertoire have a traceable author. Likewise, it would take a real purist not to acknowledge that much modern music has been written in a traditional style and assimilated to what many would now call folk music” (Reference Jones and Cameron2005:260). Consent from creators to share their work is often assumed by scene membership, enabling largely informal and non-commercial practices to continue with minimal supporting infrastructure.

Though valuing shared heritage practices might discourage some practitioners from thinking too closely about intellectual property rights, understanding how relevant legal concepts apply is particularly important in the context of digital sharing, where reach and usage have unknown dimensions and consequences. One SPG holder, for example, realised there were additional personal and legal risks when they sought to translate their hardcopy songsheets into a shareable online resource. As they often did not have documentation of song sources—let alone relevant rights clearances—they were unable to assess which materials might violate copyright or licensing terms. Given the significant required labour, they opted to put that element of their project on hold until they could adequately assess the risks.

Similar challenges existed across the datasets associated with Access Folk’s research, often related to ongoing efforts by both the academic team and SPG holders to square norms of the scene with the requirements of research institutions. For data to be held in institutional data repositories, full clearance of intellectual property rights is non-negotiable. For audiovisual documentation, this means identifying and gaining the explicit consent of all performers, requesting licenses from the people who created the documentation, and clearing third-party copyrights for all songs. Similar clearances are needed for ephemera associated with singing activities, like songsheets. At a minimum, this sort of rights research is labour intensive, requiring detailed record-keeping about performers and extensive archival searches to identify unknown songs, sources, and creation dates. Given the nature of the folk scene and song-sharing activities, not to mention gaps in available records about song sources, this type of rights research and clearance often is not possible.Footnote ²³

The informal nature of participation also poses challenges for privacy, data protection, and consent. Participants drop in and out of activities held in public or semi-public spaces. Sometimes they become session regulars but, just as often, they share a single song, coming and going before anyone can catch their name—or a song title and source. Participants, too, might be well-known professionals or local legends who are identifiable to anyone with a passing knowledge of England’s folk scene, ruling out potential for anonymity. Participating might involve singing a solo, joining in imperceptibly on a rowdy mass chorus, or sitting quietly and observing. The nature of such events makes it equally difficult to identify or anonymise everyone present, let alone gain research institution-approved consents and permissions. This is especially true in a scene that depends on the labour of time-poor volunteers who may not have relevant specialised knowledge or the capacity to tackle the substantial documentation associated with provenance and rights clearances.

The participatory nature of the SPG projects added to the inherent challenges, making our open research goals particularly difficult to manage. SPG holders signed collaboration agreements with Access Folk that specified their roles in research activities, obligations for data sharing, identification in relation to their data, and withdrawal processes, but the academic team often did not have direct relationships with participants. Based on the set of templates and processes that TUoS research services approved, informed consent procedures were adapted on a case-by-case basis in consultation with SPG holders. Alterations accounted for distinctions in whether the partner was an organisation (with its own robust consent, safeguarding, and data management procedures) or an individual; the knowledge and experience of the SPG holder; the types of participants involved; the public or private nature of interventions; and the nature of risks associated with data collection activities. In a few cases, the academic team directly ran consent procedures and data collection activities, but more often the SPG partner took a leading role in assessing and administering the most appropriate procedures for their specific context.

As the academic team were rarely present during project activities, we depended on SPG holders to institute negotiated procedures and gather required documentation. This might include posting a notice about research activities, sharing a consent form, circulating a survey, or facilitating contact with a rights holder. The SPG holders’ varied circumstances meant the results were uneven. Shared data, for example, sometimes lacked necessary details about provenance, consent, and licensing. It was not that SPG holders perceived these details as unimportant, but their primary concern was efficiently enabling participation on terms that suited their participants’ needs. One intervention, for example, that was situated in a multicultural urban centre, targeted young, historically marginalised, and politically active participants for informal drop-in singing sessions in an outdoor setting. Written documentation imposed a problematic level of formality on the event and was greeted with suspicion by some participants, so organisers preferred careful conversations for explaining data usage and ways of opting out.Footnote ²⁴ The SPG holder only shared data once they were satisfied that identifiable participants were happy with its use by the Access Folk project. Though this arguably is the most ethical approach given the circumstances, the lack of documentation trail has proven tricky to manage downstream.

We sought to minimise the administrative burden associated with the research elements of projects by providing SPG holders with templates for recording key details about the data they shared—though as feedback from some SPG holders later revealed, we sometimes needed to go further to streamline research processes. We learned that even when we summarise relevant regulations and ethical considerations, our partners may not take that information on board, either due to capacity or interest. SPG holders generously shared photos and videos that they gathered from participants alongside descriptive accounts of their planning, processes, and impact. But data sometimes were shared in formats (or via platforms) that we could not accept. Some SPG holders, for example, preferred the convenience of WhatsApp to secure file uploads; unfortunately, Meta’s lack of compliance with UK GDPR (2016) meant we were unable to accept data over WhatsApp. Repeatedly asking for the same data to be shared differently tested everyone’s patience! Likewise, we all struggled to understand why it was okay to publicly share recordings and photos over Instagram, Facebook, or on partner websites, but that same material could not be included as research data without considerable extra labour on the part of SPG holders and their participants.

While asking our partners to document consent and licensing agreements was the right thing to do in terms of protecting people’s rights (and fulfilling our institutional requirements), relations with some SPG holders soured if requests were perceived as only relevant to university systems—or, indeed, if conversations about ethics and governance were perceived as suggesting negligence on the part of research partners. But without formal documentation of consent and licensing from all rights holders, our options for processing the data remain limited. And there is the rub. At the end of this project, we will be able to produce academic papers reporting on the results of the project. The supporting datasets—the videos, photos, and creative outputs—that are most valuable to our partners may not be findable, accessible, or reusable by scene stakeholders, given the high burden of labour required of community partners (cf. Luyk Reference Luyk2015:10; Post Reference Post2017; Seeger Reference Seeger1986).

A plan undone: The limits of our institutional repository

Despite our carefully planned consent and data intake processes, within a year it became apparent that ORDA was not fit for our purposes. Our first brush with this reality came in September 2024 when we tried to share a short field recording from a workshop run by SPG holders Sarah Hinds and Miriam McDonald (Heartwood Social Farming Centre n.d.). Seeking to improve rural access and mobilise multiple “folk song adjacent” interest groups, the duo offered workshops that explored intersections between landwork, craft, song, and folkloric knowledge at Heartwood Social Farming Centre, just outside of Matlock, Derbyshire (Draisey-Collishaw and Wettermark Reference Draisey-Collishaw and Wettermark2024; cf. Camlin Reference Camlin2024). Sharing this particular field recording via ORDA should have been straightforward: I was the recordist and the video was framed so that Hinds’s image and voice were the only identifiable elements. The workshop participants had, in any case, consented to videorecording, photography, long-term archiving, and reuse—details that I discussed directly with the participants before I made the recording. Trying to share the recording showed up a first major structural barrier to using ORDA to store ethnographic data.

The field recording features Hinds using a call-and-response approach to teach a song chorus while participants harvest willow branches for use in basket weaving. The chorus is from “Willow,” an original song by Rachel Newton and Sechou Keita that was released in 2019 on The Lost Words (Spell Songs Reference Fowlis, Keita, Drever, Newton, Porter and Molyneux2019). The content of the field recording, in other words, includes third-party copyrights that require clearance before sharing via ORDA. In this case, identifying the songwriters was easy, though finding out who owns the publishing rights has proven complicated—as of February 2026, we still do not have a license to share the recording. But we also questioned whether we needed a license. Given the content and quality of the recording, did our usage constitute “fair dealing”? Rather than high-quality audiovisual documentation of a performance and song, this lo-fi field recording—made “for our own research purposes”—features a single line of the song chorus sung repetitively. The recording focuses on the means and context for participant engagement, illustrating details outlined in reporting by SPG holders, testimonials by participants, and my own fieldnotes about how the functional approach to singing supports access for different populations. The song and performance are incidental and the recording is unlikely to affect the market value of the original work.

This example exemplifies the nature of multimodal ethnomusicological data: the recording is only meaningful data when combined with other documentation. While the data infrastructure for ORDA is well-suited to preserve and share multimodal digital data, the risk-intolerant nature of the repository means that content under copyright—even when there is an applicable fair dealing exception—cannot be stored or shared. As Wakaruk et al. explain, “Anyone who works with copyright in higher education knows that the interpretation of related legal provisions is often case-specific and does not lend itself to bright-line legal tests. Fair dealing is also similarly constructed,” often leading to overly conservative assessments of what is possible in the existing regulatory context (Reference Wakaruk, Secker and Morrison2025:12–13). When ambiguities exist about songs or rights holders, the assumption must be that content is still under copyright or usage is unconsented, as institutional repositories are risk intolerant. Indeed, as Wakaruk et al. (Reference Wakaruk, Secker and Morrison2025) reveal, growing rates of “copyright anxiety” in the higher education sector result in changes to research, publication, and teaching outcomes that avoid potential risks associated with copyrighted materials, even where a risk-managed approach might be more appropriate (cf. The National Archives 2021).

While we interpreted copyright infringement risks associated with our field recording as low, we were also pragmatic about our chances of swaying ORDA’s intake policies. We started looking for workarounds. We queried, for example, whether licensing from the Performing Rights Society for MusicFootnote ²⁵ might work as a strategy for ensuring copyright compliance and royalty payment to relevant rights holders (PRS 2025). As researchers, however, we were unable to purchase a streaming licence for our collection. PRS’s suggestion that the repository might opt to licence their entire platform likewise proved impractical, running to a cost of thousands of pounds per year. Even our research team had to concede that this was an unsustainable expense, especially given that (1) we were among the only current users seeking a streaming licence, (2) the small number of streams we expected, and (3) the overall low risk associated with the content.

The challenge of depositing our data to ORDA also involved privacy and data protection considerations. Another SPG holder, Rupert Philbrick of Music at the Heart of Teesdale, shared audiovisual documentation of a Wassail event with Access Folk, alongside written reflections and interviews about introducing singing into the activities of an existing youth folk band (ages 11–17).Footnote ²⁶ The recording features Philbrick leading a group of interested young musicians in singing “The Wassail Song” at a community event in Weardale, Co. Durham. Again, sharing this recording seemed straightforward: “The Wassail Song” is in the public domain; Music at the Heart of Teesdale applied a Creative Commons license to the recording itself; and, because the Wassail event was the focus of a short documentary being made by an external funder (see Weardale Together 2025), all participants signed recording releases. The recording in question is an outtake that provides high-quality documentation of Philbrick’s work with young folk musicians. While we hold correspondence from Music at the Heart of Teesdale explaining that all persons identifiable in the recording have consented, we do not directly hold documentation of their consent. This introduces questions about data provenance and whether we have the right to deposit the recording as part of our dataset. Though the documentation relates to Access Folk’s research, it was generated by a non-academic partner. As Philbrick does not have a university affiliation, he cannot directly deposit the recording in ORDA. Philbrick, as well, is conscious of the youthful nature of the participants; while happy for the recording to be part of the project dataset—accessible and reusable for future research—he does not want the recording to be freely downloadable online. His terms of sharing demanded controlled access—a specification that ORDA simply cannot answer (Research Data Services, email to author, 29 May 2025).

Investigating alternatives and weighing priorities

In early summer 2025, we turned to the UK Data Service—the UK’s largest digital repository for social science and humanities research—as a solution for our data sharing challenges. Like ORDA in terms of function, this specialist service offers controlled access to data collections, a characteristic that we hoped would enable a risk-managed approach to the rights and provenance issues inherent in our data. Depositing to this service, however, requires institutional affiliation. Even more problematically given our prioritisation of data sharing with stakeholders in the folk scene, institutional affiliation is also required to use controlled access collections. In the end, asking “open for whom” proved unnecessary. The risks around consent and anonymity inherent in our data, the UK Data Service assessed, would require their SecureLab, a service that “provides controlled access to data that are too detailed, sensitive or confidential to be made available under the standard End User Licence or Special License” (UK Data Service 2026b). Unfortunately, this service is only available for use with quantitative datasets, which ruled us out.

Our discussions with the UK Data Service made one thing very clear: though most of our data take the form of interview transcriptions, written reflections, and fieldnotes, even minimal inclusion of audiovisual formats with complex rights all but requires involvement of a sound archive. More specifically, we needed specialist knowledge of relevant regulatory issues, as well as the capacity to take a more risk-managed approach to our dataset. In late autumn 2025, we turned to the British Library Sound Archive: to our knowledge, the only sound archive available to us in England with a national remit suitable to the scope of our project.

In theory, depositing our data with the British Library was an ideal solution: they handle multimedia data, provide controlled access, take a risk-managed approach to rights management, and, importantly, they are accessible to the public—our stakeholders would be able to access the data. The complicating factor was a cyberattack that targeted The British Library in October 2023. The attack crippled the organisation and, more than two years later, leaves the organisation working at reduced capacity (Ash Reference Ash2024; Keating Reference Keating2024; Stavrou Reference Stavrou2025). In this context, curators are reluctant to accession data from external researchers. Beyond capacity and regardless of their ability to manage multimedia data, one curator informed me that the structure of the institution may not be suitable for the preservation and dissemination of multimodal datasets. The curators at the British Library approach data intake by modality; individual curators are responsible for different types of data (e.g., audiovisual recordings, photos, and scores). Each curator evaluates data items separately according to their section’s collection priorities. Different curators, for example, would assess the lo-fi field recording of Sarah Hinds teaching “Willow” and the accompanying documentation that gives the data meaning. While there is potential to link accessioned data and liaise with relevant curators, it is unlikely that our multimodal dataset would be accepted in its entirety. Initial conversations suggest that depositing our data with the British Library will not be simple given the institution’s current reduced capacity, the complexity of our dataset, and the time limits imposed by the end of Access Folk’s funding.

It is now January 2026, and I have been working on Access Folk’s open research strategy for more than two years. I have not been alone on this journey. I am part of an incredible team of researchers, each offering specialised and complementary knowledge relevant to achieving our research goals. Yet even with the luxury of a well-resourced project and a dedicated multi-person team working in concert with the stated priorities of our institution and funder—and with support from key stakeholders within the folk scene—realising the full extent of our open research objectives remains elusive. Our participatory methods, prioritisation of co-produced research and outputs, and publication of results in open non/academic formats mean we have done a good job on the “science communication, open engagement of societal actors and open dialogue with other knowledge systems” elements of UNESCO’s open research definition (2021). Indeed, after the Access Folk project wraps up at the end of January 2027, our open access outputs will continue to be stably shared on ORDA for anyone with the interest and know-how to locate an institutional repository. But though we’ll have a well-documented dataset that is mostly compliant with the FAIR Principles, that data will nevertheless remain largely inaccessible to stakeholders interested in building on our work.