Categories become infrastructural when they cease to function merely as descriptions and begin to organise what institutions can recognise, store, compare and act upon. Aristotle establishes classification as an operation of distinction; Wittgenstein relocates meaning within situated use; Foucault shows that classifications participate in historically specific regimes of knowledge and authority; Susan Leigh Star demonstrates how standards and categories disappear into infrastructures while continuing to organise access and exclusion. Their intersection reveals a material career of language: a word stabilises as a distinction, enters forms and databases, becomes embedded in procedures and eventually acquires consequences for bodies, spaces, resources and decisions. A planning designation can redirect investment; a diagnostic category can determine eligibility; a metadata field can decide whether a cultural object becomes retrievable; a building regulation can normalise particular assumptions about bodies and behaviour. Classification is therefore one of the smallest mechanisms through which collective reality becomes executable. Transdisciplinary research should consequently follow categories beyond discourse, asking where they enter technical formats, who can modify them, which differences they preserve or erase, and what happens once they operate at institutional scale. Precision matters because concepts such as resilience, community, system or infrastructure lose analytical value when they absorb incompatible mechanisms. The critical task is neither to celebrate fluidity nor to abolish classification, but to construct categories whose limits remain visible, whose consequences can be inspected and whose organisation of reality remains open to revision.