Temporal structure of spoken-word recognition in Croatian in light of the cohort theory.

This article addresses two issues related to spoken-word recognition: the relationship between the amount of acoustic material and the degree of recognizability at the lexical and phonemic levels and the influence of segmentation ambiguities on the speed and success of the recognition process. The analyses were done on Croatian language materials using the gating paradigm. The results indicate that the degree of recognizability is directly proportional to signal duration and inversely proportional to the complexity of segmentation. A combination of bottom-up and top-down processing is involved in successful word recognition.