Past research has demonstrated a greater “naturalness” of comparative constructions with the unmarked adjective. Not only do subjects typically create sentences of the form “A is longer than B” more frequently than “B is shorter than A,” but such constructions are also more quickly verified. One explanation for this preference is in terms of Linguistic Marking (e.g., Clark, 1969a, b). Because the semantic featural coding hypothesized for unmarked adjectives is simpler than that of marked adjectives, the observed facilitation effects would be expected. However, an equally viable explanation is in terms of the greater positive affect of the unmarked adjective (see Osgood et al., 1957). Since linguistic marking and affect are typically confounded in past research (the unmarked adjective is also the most affectively positive), the present experiment discriminates between these two interpretations by providing also contexts in which linguistic marking and affect are not confounded. Here, analysis of the comparatives constructed by 96 subjects revealed the preference for the unmarked adjective only where it was also the most affectively positive. Where the marked was the more positive, the reverse was observed, as predicted by Osgood (et al., 1957).