Sex differences in language asymmetry are age-dependent and small: A large-scale, consonant–vowel dichotic listening study with behavioral and fMRI data

Abstract

Men are often believed to have a functionally more asymmetrical brain organization than women, but the empirical evidence for sex differences in lateralization is unclear to date. Over the years we have collected data from a vast number of participants using the same consonant–vowel dichotic listening task, a reliable marker for language lateralization. One dataset comprised behavioral data from 1782 participants (885 females, 125 non-right-handers), who were divided in four age groups (children <10 yrs, adolescents = 10–15 yrs, younger adults = 16–49 yrs, and older adults >50 yrs). In addition, we had behavioral and functional imaging (fMRI) data from another 104 younger adults (49 females, aged 18–45 yrs), who completed the same dichotic listening task in a 3T scanner. This database allowed us to comprehensively test whether there is a sex difference in functional language lateralization. Across all participants and in both datasets a right ear advantage (REA) emerged, reflecting left-hemispheric language lateralization. Accordingly, the fMRI data revealed a leftward asymmetry in superior temporal lobe language processing areas. In the N = 1782 dataset no main effect of sex but a significant sex by age interaction emerged: the REA increased with age in both sexes but as a result of an earlier onset in females the REA was stronger in female than male adolescents. In turn, male younger adults showed greater asymmetry than female younger adults (accounting for <1% of variance). There were no sex differences in children and older adults. The males in the fMRI dataset (N = 104) also had a greater REA than females (accounting for 4% of variance), but no sex difference emerged in the neuroimaging data. Handedness did not affect these findings. Taken together, our findings suggest that sex differences in language lateralization as assessed with dichotic listening exist, but they are (a) not necessarily reflected in fMRI data, (b) age-dependent and (c) relatively small.