The role of letter features in visual-word recognition: Evidence from a delayed segment technique

•We conducted two lexical decision experiments using a delayed segment technique.•Latency data of Experiment 1: advantage of whole preview over the other conditions.•Latency data of Experiment 2: whole = no terminals < no junctions < no midsegments.•Some components of a word's constituent letters are more critical than others.

Do all visual features in a word's constituent letters have the same importance during lexical access? Here we examined whether some components of a word's letters (midsegments, junctions, terminals) are more important than others. To that end, we conducted two lexical decision experiments using a delayed segment technique with lowercase stimuli. In this technique a partial preview appears for 50 ms and is immediately followed by the target item. In Experiment 1, the partial preview was composed of terminals + junctions, midsegments + junctions, or midsegments + terminals — a whole preview condition was used as a control. Results only revealed an advantage of the whole preview condition over the other three conditions. In Experiment 2, the partial preview was composed of the whole word except for the deletion of midsegments, junctions, or terminals — we again employed a whole preview condition as a control. Results showed the following pattern in the latency data: whole preview = delay of terminals < delay of junctions < delay of midsegments. Thus, some components of a word's constituent letters are more critical for word identification than others. We examine how the present findings help adjust current models of visual word identification or develop new ones.