Low contrast in letter-stroke facilitates lexical identification

It is a long-lasting dispute within typography, whether serif or sans serif fonts are the most legible. However, different fonts vary on numerous visual parameters and not just the serifs. In the present experiment, we were interested in investigating whether a given difference in reading performances between serif and sans serif fonts, relates to the serifs or relates to the contrast of the letter stroke.
To test these questions, we developed five new fonts where we isolated the stylistic features of serif and letter stroke contrast, and tested these with a lexical decision (LD) task. To obtain LD thresholds for different font styles, an LD task was embedded within a two-interval, forced-choice (2IFC) task. On a given trial, participants were exposed to one word stimulus after the other whereby one of the words was either a correctly spelled word or a non-word (word with one of its letters swapped with another middle letter). Participants (n = 16) indicated in which of the two intervals the real word was placed. As a baseline condition, LD thresholds, for the font Helvetica, were obtained via the QUEST algorithm. Experiment proper consisted of the method of constant stimuli and the word stimuli's font size, which was tailored for each participant, was pseudorandomly varied. Stoke contrast was manipulated to be low or high and type of serif was manipulated to be sans serif or serif, which yielded a 2 by 2, factorial design. There was a main effect of stroke contrast such that low stroke-contrast elicited lower LD thresholds (M = 63.67) relative to the LD thresholds elicited by high stroke-contrast (M = 66.40), F (1, 15) = 7.12, p = .018. Low LD thresholds indicate that word stimuli may be presented at smaller font sizes to make a correct lexical decision and vice versa.