Semantic processing in language production includes different meaning relations determining the selection of lexical representations that best express the intended message. Here, we discuss assumptions of the swinging lexical network proposal (SLN), proposed to account for effects of different semantic relations in a variety of experimental paradigms, with effects ranging from semantic facilitation to interference. The SLN is based on two assumptions. First, conceptual and lexical processing proceed in parallel and may exhibit opposite effects of conceptual priming and lexical competition. Second, the amount of lexical competition is determined by the co-activation of an inter-related lexical cohort and is thus sensitive to the number and strength of active competitors. We discuss behavioural effects across different experimental paradigms and semantic relations in light of the SLN and suggest that by adopting the basic assumptions we can account for a wide range of semantic facilitation and interference effects in language production.