Current studies on the mapping of discourse participants (e.g., Giorgi 2010, Hill 2007, 2014, Miyagawa 2010, Ross 1970, Sigurðsson 2011, Speas & Tenny 2003) unanimously point out that the [speaker] is crosslinguistically silent (i.e., the feature is not spelled out). This paper presents data where the [speaker] is spelled out in Turkish inverse vocatives. The study argues that this happens when an [affect] feature is mapped to the Speech Act Phrase (SAP) and triggers an imposter operator, which has as side effect a modification in the distribution of features within SAP. This is a marked configuration contrasting with the articulated SAP hierarchy that applies crosslinguistically as the default option (Haegeman & Hill 2013) and supports a silent speaker.