Material-discursive traces produced by apparatuses of collection, preprocessing, and modeling. Data are relational, situated, and performative: they enact cuts that delineate what becomes an object of inquiry, what counts as evidence, and which learning outcomes are legible in writing pedagogy.
In postqualitative, posthumanist, and new materialist work, data are not considered to be raw facts awaiting interpretation, but effects of specific arrangements of tools, concepts, bodies, policies, and practices. Following Barad, an apparatus is the concrete configuration through which phenomena become determinate; data are the patterned traces this configuration leaves behind. They are material-discursive because they congeal both material processes (sensors, platforms, corpora, interfaces) and discursive selections (categories, rubrics, prompts, ontologies). The traces that result are inseparable from the conditions that made them, and their form performs what can be asked, known, and acted upon.
Building with Haraway, data are situated knowledges. They carry partial perspectives from their sites and histories of production: who designed the instrument, which categories were available, what values and exclusions organized collection, cleaning, and modeling. In generative AI contexts, “data” includes training corpora, fine-tuning and reinforcement feedback, prompts and instructions, interface telemetry and system logs, vector embeddings, and generated outputs that circulate back as future training inputs. Each element participates in composing the “object” under study and the subject positions from which it is knowable. Treating data as performative draws attention to how datasets and dataflows do pedagogical work: they shape what writing becomes, what counts as creativity or authorship, and where accountability is placed in AI-mediated classrooms.
A postqualitative approach treats data generation as world-making rather than extraction. Apparatus design is a primary methodological decision: fieldnotes, version histories, prompts, rubrics, and model outputs are configured to co-produce the phenomenon. Analysis focuses on tracing how agential cuts happen: how categories, metrics, and preprocessing delimit “the data”, and on the ethical accountabilities that follow. Rather than coding toward saturation, inquiry composes with entangled traces, reads them diffractively across theories and materials, and documents how the apparatus reconfigures the questions it can answer. This stance keeps attention on distributed agency and on the politics of classification, measurement, and platform affordances.
In AI-entangled writing instruction, data are the lifeblood of the pedagogical assemblage. Training corpora sediment histories of genre, voice, and citation that shape model tendencies in student-facing tools. Prompts, system messages, and rubrics operate as local apparatuses that enact boundaries around desirable outcomes, stylistic norms, and evaluative criteria. System logs, drafts, revision trails, and chat histories materialize composing processes and mediate how authorship and collaboration are recognized. Embeddings and detectors introduce new forms of legibility that reconfigure trust and accountability. Treating these traces as situated and performative makes visible how instructors’ practices, institutional policies, and platform designs co-produce what writing is and does, how emotions like confidence or anxiety circulate through interfaces, and how creativity is enacted through specific prompt architectures and feedback loops. It invites designing assignments and reflective activities that foreground data provenance, prompt transparency, and the ethics of reuse, while attending to how outputs become new inputs that carry pedagogical consequences over time.