Mediated phonological–semantic priming in spoken word production: Evidence for cascaded processing from picture–word interference

The cognitive architecture that allows humans to retrieve words from the mental lexicon has been investigated for decades. While there is consensus regarding a two-step architecture involving lexical-conceptual and phonological word-form levels of processing, accounts of how activation spreads between them (e.g., in a serial, cascaded, or interactive fashion) remain contentious. In addition, production models differ with respect to whether selection occurs at lexical or postlexical levels. The purpose of this study was to examine whether mediated phonological–semantic relations (e.g., drip is phonologically related to drill that is semantically related to hammer) influence production in adults as predicted by models implementing cascaded processing and feedback between levels. Two experiments using the picture–word interference (PWI) paradigm were conducted using auditory (Exp. 1) and written (Exp. 2) distractors. We hypothesised that a mediated semantic interference effect would be observable in the former with the involvement of both spoken word production and recognition, and in the latter if lexical representations are shared between written and spoken words in English, as assumed by some production accounts. Furthermore, we hypothesised a mediated semantic interference effect would be inconsistent with a postlexical selection account as the distractors do not constitute a relevant response for the target picture (e.g., drip-HAMMER). We observed mediated semantic interference only from auditory distractors, while observing the standard semantic interference effect from both auditory and written distractors. The current findings represent the first chronometric evidence involving spoken word production and recognition in support of cascaded processing during lexical retrieval in adults and present a significant challenge for the postlexical selection account.