Identifying use-related residues from stone artefacts has become increasingly important in determining starchy plant exploitation over time and in different locales. Standard methods for processing residues samples are widely available but there is no clear consensus on suitable methods for attributing unknown starch grains to known plant taxa. We revisit the case study of a flaked stone artefact (K/76/S29B) recovered from Phase 1 (c. 10,000 ka) at Kuk Swamp in Papua New Guinea (PNG). Starch grains from taro (Colocasia esculenta) were identified in the residue extraction, but there were other grains that could not be attributed a plant origin at that time. The new analysis applied robust statistical methods, categorial attributes and expert input. In addition to C. esculenta, kudzu bean (Pueraria montana var. lobata) was identified, representing the earliest use of kudzu bean in the PNG highlands. Importantly, we also determined that starch grains from C. esculenta and Dioscorea esculenta are morphologically indistinguishable. We turned to other attributes of potential contributing plant taxa in determining distinguishing features: habit/growing requirements; the sedimentary context of the archaeological find; and environmental settings. Cultural use of both plants, artefacts and artefact technologies can be critical elements in confident identification outcomes, as exemplified here.