Biosemiotics has often been charged with the use of metaphorical language, redundancy with respect to existing biology, weak falsifiability, and lack of specific methods (see, e.g., Barbieri (Biological Theory, 9(2), 239–249, 2014); Deacon (Green Letters: Studies in Ecocriticism, 19(3), 293–311, 2015); Sherman (Chinese Semiotic Studies, 20(2), 231–253, 2024); Vehkavaara (Sign Systems Studies, 30(1), 293–313, 2002). This article examines the epistemological and methodological status of biosemiotics in the light of current philosophy of science and methodological pluralism in life sciences. Rather than asking whether biosemiotics has a single unifying hypothesis or methodology, we want to see whether it can function as a progressive research programme in a broadly Lakatosian sense, while remaining compatible with methodological diversity (Lakatos, 1970; Giere, 2006; Kellert et al, 2006; Chang, 2012; Ruphy, 2016). Our analysis proceeds in four steps. First, we outline evaluation standards appropriate for fields organised around midlevel frameworks that coordinate models, instruments, and standards of evidence rather than for universal ‘methods’. In doing so, we stress model construction, contrasts, and the ability to generate new testable hypotheses, that is, heuristic fertility. Second, we address worries about metaphoricity and conceptual incoherence by tracing how key semiotic concepts (semiosis, code, umwelt, interpretation) can be unified and linked to operational distinctions. Third, we respond to the objections regarding its redundancy and epistemic status by treating biosemiotics as a complementary interpretive layer that functions on top of mechanistic explanations and is useful especially in areas where context-sensitivity, normativity, and agency are central. Fourth, we survey zoosemiotic, ecosemiotic, and biosemiotic studies to identify four recurrent empirical patterns (namely, context-sensitive responses, conventional sign use, norm-guided correction, and remapping under controlled change) related to situations where semiotic and non-semiotic models produce different predictions based on the same data. We conclude that under modest but non-trivial programme-level standards, biosemiotics can be developed as a testable and integrative approach within life sciences and identify domains and methodological practices where its claims remain underspecified and invite further work.
