Thinking Stomach

This dataset was generated to test the hypothesis that eating in narrative fiction functions as a multi-dimensional human practice rather than a single-variable phenomenon (e.g., hedonic pleasure, symbolic representation, or individual rational choice). The study hypothesizes that food scenes co-activate six registers of practice—Essendi (being), Cognoscendi (knowing), Agendi (acting), Vivendi (living), Faciendi (making), and Operandi (operating)—and that stable coupling patterns between these registers recur across texts.

Data were collected from a purposive corpus of 27 novels selected for variation in period, genre, setting, and narrative style. The unit of analysis is the Food Narrative Unit (FNU): a bounded segment in which eating, cooking, provisioning, or bodily response to food is narratively foregrounded. FNUs were extracted using an explicit protocol with documented scene boundaries and decision rules. Each FNU was coded for dominant mode, secondary mode, and coupling tags (e.g., COG-ESS, FAC-AG, OP-AG). The deposited materials include coding outputs and audit-oriented logs/metadata to support transparent interpretation.

The data show that eating is consistently represented as a multi-register process. Across the corpus, recurrent couplings indicate that sensory experience is frequently linked to knowledge and self-formation (COG-ESS), craft and preparation often become practical agency (FAC-AG), and operational or infrastructural conditions shape what actions are possible (OP-AG). Additional recurring links (e.g., VIV-FAC, COG-OP) show that routine life and governance structures are embedded in food scenes rather than external to them.

Notable findings include: (1) cross-genre stability of key coupling pathways, (2) repeated movement from bodily sensation to interpretation and decision, and (3) co-presence of ethics, social order, and material constraint within single narrative episodes. These findings support an interpretation of eating as a distributed epistemic-ethical process.

This is a qualitative, theory-guided coded dataset intended for reuse and re-analysis. Coupling frequencies should be interpreted as patterned salience within this corpus, not population-level prevalence. Other researchers can use the FNU protocol and coding schema to replicate on new corpora, compare subsets (period/genre/region), conduct coupling-network analyses, or test alternative theoretical models on the same annotated units. Source material is published literary text; no human-subject personal data are included.