When pictures are repeatedly named in the context of semantically related pictures (e.g., “sheep” in the context of “goat” and “horse” – homogeneous context) latencies are longer than when the pictures are repeatedly named in the context of unrelated pictures (e.g., “sheep” in the context of “table” and “hammer” – heterogeneous context). Adaptive models of word production attribute this semantic interference effect in blocked-cyclic naming (BCN) to an incremental learning mechanism which makes semantically related words that were activated but not selected less accessible for future retrieval. Recent results from a picture-word-interference (PWI) study, however, suggested that lexical alternatives (e.g., “bird” for “duck”), might be exempt from this mechanism (Kurtz et al., 2018). We tested whether converging evidence can be obtained in BCN. We embedded pictures named with basic-level category labels (e.g., “bird”) into contexts of pictures of specific exemplars named with subordinate-level labels (e.g., “duck”, “eagle”, “stork”). If lexical alternatives are generally exempt from incremental learning, interference in the homogeneous context should be found for subordinate-level targets but not for basic-level targets. Contrary to this prediction, interference was similar-sized. We discuss implications of these results for theories of incremental learning and the use of the PWI task in this context.