Short abstract:

While the field of digital methods in the humanities and social sciences encompasses a diverse range of approaches and methodologies, this presentation aims to offer an updated and detailed categorisation of the various streams within digital methods.

Long abstract:

The field of digital methods in the humanities and social sciences encompasses a diverse array of approaches and methodologies. Richard Rogers's "manifesto" (2009), which advocates for the principle of "follow the medium," provides a common foundation in this area. Initially associated with computational analytics, "big" datasets, and visualisation software, the first wave of digital methods has faced challenges in the “post-API” era, especially due to significant changes in the relationship between commercial platforms and academic research. However, digital methods in the humanities and social sciences are not limited to computational techniques and automated API-based scraping; they also incorporate critical perspectives and a mixture of various approaches–including qualitative ones, reflecting the dynamic and evolving nature of this interdisciplinary field. Building on earlier works that mapped this domain (Marres, 2012; Snee et al., 2016), this presentation aims to offer an updated and nuanced categorisation of the various streams within digital methods.

Long abstract:

Autoethnography is a mindset as much as a set of techniques that draws on the strength of subjective ways of knowing, long recognized in non-western contexts, to build deep and granular analyses of cultural meaning emerging from one’s own experiences of situations. Autoethnography embraces contingency, that the ‘truth’ of a situation depends on a variety of factors, it highlights reflexivity as a tool for examining one’s own lens for understanding, and it overtly acknowledges that methods shape what is understood.

This approach may seem to contradict computational analytics, whereby data is used in forms already abstracted from experience, among other differences. However, when considering the inductive goals of pattern recognition, autoethnography and largescale data analytics can be aligned, if not combined.

This contribution describes how a research team builds on Markham’s “Guided Autoethnography” approach to conduct an aggregated autoethnography across four countries. Researchers train young adults to conduct autoethnographies of their own digital lived experience, which produces raw data interwoven with ‘thick description’ ethnographic interpretations. Researchers analyze these, generating emergent themes, which in turn are used to reconfigure some of the large pool of aggregated data into new datasets for larger-scale qualitative or computational cross-country analyses.

We discuss merits, challenges, and possibilities of using seemingly disparate epistemological logics in parallel. To retain their independent strengths, they are used in separate stages: autoethnography provides granular level patterns, while computation explores patterns at scale. The question raised is how to ensure the original ethnographer/participants remain connected to possible largescale interpretive discussions.

Long abstract:

For some researchers (Winne, 2022; Lodge et al., 2018), learning analytics developed in technology-enhanced learning systems could be used “to analyze and report learner performance and to highlight the part of learning where self-regulation processes could be improved” (Heikkinen et al., 2023, p. 3061). It is accepted in the scientific community that learning analytics is defined as “the measurement, collection, analysis and reporting of data about learners and their contexts, for purposes of understanding and optimizing learning and the environments in which it occurs” (Siemens & Long, 2011, p. 34). For learners, the challenge is to transform this data into relevant information (Long & Siemens, 2011; Wang, 2018) so that they can track, understand, and evaluate their learning. This raises questions about both what data is relevant to the learner and how the data is presented and visualized through dashboards to make it meaningful and actionable. Addressing these questions also raises methodological challenges. In this panel, we will discuss the use of photo-elicitation (Harper, 2002) as a “multimodal narrative” that includes stories, photos, and traces produced by learners to examine their own self-regulation process. This form of “multimodal narrative” generated through photo-elicitation helps to “make visible” information and analytic practices in support of active learning, and to lead participants into a form of self-reflexivity about these uses of learning analytics.

Long abstract:

The rise of digital technology has opened new avenues for research, particularly by harnessing the extensive power of computational methods to analyze large datasets. However, a purely computational research method raises various issues. Millerand, Myles, and Proulx (2020, note, for instance, that works relying on computational methods tend to adopt a positivist view, asserting that the analysis of massive data would allow for a more objective analysis and representation of reality. On the other hand, these authors also note that the use of traditional and more qualitative methods to study digital practices encounters limitations when faced with large data corpora. They thus suggest the relevance of combining elements from several categories of research methods.

Following these reflections, we present a “mixed methodology” that combines computational practices, such as data harvesting and filtering, with more traditional research practices, namely case studies. We start by summarizing the work of Millette et al. (2020), following Marres (2012), who divide four categories of methodologies to study digital practices. We then present the mixed methodology we developed for a study of digital practices in support of the Black Lives Matter movement. Finally, we discuss the advantages and issues that such a mixed methodology may raise for subsequent research works. We argue in this presentation that implementing such a methodology for analysis allows, on one hand, access to the vast quantity of digital data thanks to computational practices, while maintaining the scope of analyses associated with qualitative data analysis.

Long abstract:

How can the metaphor of glitter reveal the productive materiality of the research process? Metaphors carry meanings, highlighting and hiding power dynamics (Cowan and Rault 2022). Glitter, as a metaphor, touches what it isn’t supposed to, sticks to unexpected places, and draws out beautiful and dirty outcomes in weird corners. The metaphor reveals what we mean by speculation as methodology, to connect people in the world in tentative and telling ways, allowing us to conceive of iterative approaches and potential outcomes. As feminist methodology, speculation (Asberg et al., 2015) applies to the conduct of research throughout the process: from the moment of conceiving research questions through to making material changes in society that aim to shift power dynamics. Speculation asserts the importance of examining and enabling diverging material practices (how we do things), political commitments (why we do things the ways we do), and the many specific stances that we each take in relation to the research being conducted (Authors, 2018). This presentation compares two projects. One author examines how a two-year collaboration among six culture policymakers and funders, 12 arts organizations, and 15 academics resulted in three flexible impact assessment tools and approaches. The other author will review a five-year partnership research project, developed from needs in the sex industry to document the role of social media in the recruitment of underage girls. By centering the research participants, the iterative changes in the projects relied on glitter as speculative methodology to ensure the projects were meaningful, responsive and impactful.

Long abstract:

TikTok, a rapidly expanding social media platform, has gained immense popularity for its short video format, particularly among the younger demographic. Often dubbed a ‘meme generator,’ TikTok places greater emphasis on repetitive mimicry rather than individual creativity. Unlike Instagram, where content visibility is influenced by social networks and hashtags, TikTok’s algorithm tailors content primarily based on user interactions within the app. This results in an intriguing experience—an encounter with an algorithm that delivers ‘content without context,’ rather than a cohesive community of like-minded users. Researchers studying TikTok face challenges due to its ever-evolving and unpredictable user base. Unlike platforms such as Instagram, TikTok lacks social anchors like follower networks or hashtags. To address this, we propose a novel methodological approach within the realm of digital ethnography. Our focus centers on digital micro-entrepreneurs in Naples. Termed the “Machine Habitus as Method,” our strategy draws inspiration from Airoldi’s work on the algorithmic connections between humans and artificial social agents. By actively engaging with the platform and providing targeted feedback, we steer the algorithm toward content relevant to our research. This approach not only deepens our understanding of this specific demographic but also serves as a blueprint for investigating other social phenomena within the TikTok ecosystem.

Long abstract:

This abstract offers an innovative approach to working with algorithms in the space between distant and close views of data. For doing so, we will present a project that explored affectively charged social media exchanges about remote schooling in Hungary during the pandemic. Through a cooperation with SentiOneTM, an AI-based social listening tool, we adopted a style of digital research that Blackman (2019) calls ‘transmedial storytelling’: an ethnographic approach for working with digital archives.

Although SentiOneTM offers computational means for data analysis, including commercial sentiment analysis, it was not compatible with our philosophies of affect and emotion. We started with the initial data provided by SentiOneTM‘ but added ‘a human in the loop’ and built an iterative approach into the research design. This included a preliminary reading of media articles related to remote learning during Covid-19 to pinpoint keywords. These were later paired with emotion words to form an algorithm that would crawl Facebook for the desired media articles and social media responses.

Attuning to affect we selected three ‘data events’ that could be curated into a data-story. They included a variety of digital data including comments, emoji and images. By employing a multimodal analytic that analysed the social actors, visual images, non-linguistic reactions and broader social-economic-political relations, we were able to look at big data patterns but also the micropolitics embodied and embedded in everyday (social media) practices.

Long abstract:

This communication discusses the use of the story completion method (Gravett, 2019; Watson & Lupton, 2022) to explore different applications of AI-based health technologies before their implementation in real-world settings. The contribution of this qualitative method is examined through a research project that gather ‘usage stories’ from people living with Parkinson's disease. To this end, we used story prompts, or story "stems", that depicted fictional patients interacting with three AI-based health technologies. Participants (n=147) were asked to imagine themselves as the protagonist and complete the stories (Lupton, 2021). Participants are required to draw upon their own values, experiences, and perceptions to make sense of the described scenario and create a meaningful narrative (Clarke et al., 2018). The narratives were analyzed using a story mapping technique (Braun et al., 2019). The story maps reveal three key moments in the 'usage trajectories': (1) the initial 'affective forces' (Lupton et al., 2022, Lupton, 2019) that prompt users to act; (2) the 'socio-technical arrangements' (Oudshoorn, 2011; Peine & Moors, 2015) that facilitate or hinder adoption and use; and (3) the anticipated outcomes or future uses. Story maps outline different ‘usage stories’ that provide insight into the conditions that facilitate users’ appropriation and integration of these technologies into their lives. This example serves as a starting point for discussing the strengths and weaknesses of the story completion method for studying the “usage before use” of AI-based health technologies.

Long abstract:

For around ten years, using an experimental approach, we have been developing and testing a method in progress for cartographic analysis of thick corpora, which we have entitled “Cartographic modeling of controversies” (Desfriches Doria, 2022) . This “virtual” method (Marres, 2012) is based on the tradition of cartographic analysis, adapted to the digital context (Millette et al., 2020) at the crossroads of LIS, discourse analysis and sociology.

We will begin by briefly describing this methodology that we apply to media corpora of controversial subjects. We will continue by focusing on the nature of these kinds of polyphonic objects through examples, and the multiplicity of functions that these maps can assume.

Indeed, our cartographies constitute an analogic method of reasoning, which is here applied through digital tools. It is also a method of visual and digital processing of information to show blind spots, convergences, divergences in the positions of the actors in a debate. These maps can also be understood as a tool for reflexivity on the content of the arguments, but also as a tool for mediating the controversy.

Then we will focus on the representation of the embedded forms of enunciation in media discourses, and that the affordances of cartographic tools allow us to show through visual aspects.

Finally we will address the limits of this approach and discuss possible complementarities with more massive data processing and visualization tools.

Long abstract:

The notion of situation has a long history intertwined with ethnomethodology, feminist epistemologies and pragmatist philosophy. Methodological reflections about the study of social relations afforded by digital technologies and platforms has been influenced by the concern of context collapse (Davis and Jurgenson, 2014; Marwick and boyd, 2011; Meyrowitz, 1985) – the disappearance of the spatio-temporal determinants of communication that jeopardizes the ability of actors to make meaning, and of researchers to interpret social phenomena in digital environments. Where the context refers to elements that the researcher must add to the description to explain empirical phenomena, the notion of situation refers to elements that must be taken into account in order to provide a thicker empirical description of the phenomena, emerging from the core of involved actors’ experience and concerns rather than being defined from the outside. It calls for research practices that pay attention to the liveness and indetermination of social media practices , and reflect upon the artificiality of qualitative methods as sociomaterial settings. We are sharing three research projects that designed their own methods to address this challenge, which have in common to elicit participation in the inquiry and to equip their methods with socio-technical devices (telephone, card games, videos etc.). This presentation addresses the challenges of (re)situating the study of social media practices and their role in social life. It does so by investigating offline/online connections and by tinkering with methods and tools for describing the corporeal, spatial and more-than-digital inscription of social media.

Long abstract:

the notion of discoverability is used by both cultural institutions and entrepreneurs to explain how artists and cultural workers must adapt their visibility practices to the digital regime to "meet" thier publics. However, field research seems to show that the social practices of discovery have very little to do with the practices of discoverability encouraged by cultural institutions. This conference aims to clarify, based on research into forms of music consumption on streaming platforms in France and Quebec, the hiatus between marketing and cultural consumption in the light of this concept, by proposing a socio-historical reading of the concept and methods for measuring the effectiveness of cultural policies. Our aim will be to show the limits of the computational approach to analyzing a "natively digital" concept.

Long abstract:

This study proposes a postphenomenological methodology to examine technologically mediated trust, with Airbnb as a case study, offering empirical grounding to the philosophical investigation of trust and distrust in the digital society. Airbnb operates within a complex trust ecosystem, balancing interpersonal trust between hosts and guests concerning personal safety and property security, and its role as service provider fostering trust within its community. The case offers a clear instance of a temporally and spatially distributed process of trust formation, maintenance and potential breaches, from initial booking considerations to post-stay reviews. The research design departs from the postphenomenological focus on materialities, which are identified as different sites of mediation: the platform; the devices to access it; the physical spaces; the human body, which moves in space and time and experiences trust both psychological and physiologically (the “gut feeling”). The research design follows Aagaard’s (2017) two elements of postphenomenological research: an exploration of typical use and a critical comparison of multiple versions of the technology. These are combined with principles of constructivist grounded theory (Charmaz, 2014), constantly moving between theory and data, leading to a theory-laden but open-minded collection of data through three different methods: documentary research, go-along interviews and a diary method. While document analysis, media go-alongs and diaries can help explore the typical use of the platform; elicitation methods (scenarios, vignettes, use of props) during interviews and the study of other hosting platforms can achieve a more critical comparison of multiple versions of the same technology.

Long abstract:

This proposal takes inspiration from the critical work on decolonizing knowledge and the advocacy for open science (Mboa 2020; Piron et al 2018), to explore the potential of Artificial Intelligence (AI) as both a facilitator and a domain for decolonial practices. The proposal argues for the development and application of AI in a manner that democratizes access to scientific tools and data, respects local knowledge systems, and fosters the inclusion of marginalized voices in the research process.

This involves critically examining how AI can contribute to dismantling colonial legacies embedded within current scientific research methodologies and knowledge dissemination practices. It advocates for an approach to AI development that is participatory, ensuring that communities in the Global South can contribute to and benefit from AI innovations (Mboa 2023). This approach not only challenges the dominant paradigms of knowledge production but also opens up new avenues for innovation in qualitative research methodologies. Through this exploration, the proposal aims to contribute to the ongoing dialogue on how AI can be a powerful ally in the journey towards decolonizing knowledge and promoting a more just and inclusive global research community.

Long abstract:

Discussants: Claire Balleys (Université de Genève), Sylvie Grosjean (University of Ottawa), Guillaume Latzko-Toth (Laval University), Mary Elizabeth Luka (University of Toronto), Annette Markham (Utrecht University). This roundtable presents a structured and inclusive discussion on the topic of the open panel. In their efforts to grasp the fast-paced transformations of social, cultural, and scientific practices in the context of an ever more digitally mediated life, social and humanities researchers are encouraged to constantly adapt, extend, and transform their methods. Rooted in qualitative, constructivist, and critical approaches stemming from related epistemologies, these methodological developments contribute to building an extended repertoire of qualitative and mixed methods for the digital age. While they take up the idea of “following the medium” by leveraging digital affordances (Rogers, 2013), these “hands-on” methods are not primarily focused on automated analysis nor the visualisation of large corpora of traces generated in an automated way. On the contrary, these methodological developments highlight the assemblages, hybridisations and bricolages that develop on the margins and in the interstices of the old canons of social science and the new canons of data science. The goal of this roundtable is to explore the unique features, innovative aspects, and diverse range of these emerging methods, discussing their role in advancing digital research and qualitative methodologies.