This 5-minute presentation provides an update about a project initially presented at the 2018 Forum. The presentation illustrates the preliminary findings of a qualitative study investigating voters’ evaluation of Australian politicians based on language-related features. The study involved 5 focus groups during which participants (all of whom declared beforehand to hold progressive political beliefs) were asked to listen to short audio clips of lesser-known politicians’ speeches, and to evaluate those politicians accordingly. The findings reveal a number of recurrent voter’s erroneous biases and prejudices (e.g. about politicians’ age and competence) based on language-related features concerning politicians’ prosody and lexicon. Even in an apparently ‘best-case scenario’ (i.e. participants with self-proclaimed progressive views), therefore, language-related bias seems to significantly affect voters’ evaluation of politicians and, therefore, the latter’s electability. This may hinder the fulfilment of such key democratic values as liberty, equality, inclusivity and reciprocity, and pose a critical challenge to the processes of democratic reasoning and political representation.