Are nonlinear vocal phenomena as distracting as people think?

What makes baby cries, dog barks and piercing screams so disturbing and difficult to ignore? A common explanation is that their salience is enhanced by vocal roughness and unpredictability caused by irregular phonation. A comprehensive investigation in ten perceptual experiments confirmed that human listeners found nonlinear vocal phenomena (NLP: frequency jumps, amplitude modulation, subharmonics and chaos) distracting and annoying in baby cries and nonverbal vocalisations of adults, including both original recordings and resynthesised versions with manipulated NLP. At least for the tested range of vocalisations, the distraction and annoyance were primarily caused by irregular, rough voice quality during episodes of NLP, and only secondarily by unpredictability and bifurcations between phonatory regimes. In contrast to their clear effects on subjective ratings, NLP had a limited impact on the allocation of attention in dichotic listening tasks, and their presence did not noticeably enhance distraction from the main task in experiments using serial recall and speeded classification. Thus, while irregular phonation typical of distressed baby cries and many animal calls is experienced as unpleasant and subjectively distracting, listeners may be surprisingly adept at blocking or actively avoiding such distractors.