In the performance of poetry, some performers occasionally use a certain ‘softened’ voice quality, deviating from that of the context, for emotive expression. We explored how listeners perceive such instances through a combined methodology of eliciting open-ended descriptions and ratings of pre-defined terms on ordinal scales. The collected responses confirm our intuitions that listeners are sensitive to this voice quality, which gives rise to impressions, such asaffection, yearning,andcontemplation. The chosen methodology allowed us, through variations between verbal and numerical responses, to observe the interaction of this voice quality with other factors, such as semantic content and other salient voice qualities. Finally, the responses highlight the distinction between two delivery styles used by performers — an impassive, grave style, typical of poetry recitals, on the one hand, and a more emotional “theatrical” style, on the other hand.