ABSTRACT
As research on the construction of a mental representation of referent gender in speech comprehension is scarce, this study examined whether factors identified in reading comprehension exert similar influence in speech comprehension. Conceptually replicating previous research, a sentence continuation evaluation task was set up in two modalities, as a listening task and as a time-confined reading task (i.e. to correspond to the time constraint when listening). In line with previous findings from self-paced reading paradigms we found gender representations in language comprehension to be grounded in the interaction between textual (grammatical) and background (stereotypical) information. Extending previous research, the effect of stereotypical information was however modulated by presentation modality. In all, although speech and reading comprehension share higher-level processes of comprehension, this study provides first evidence that differences in comprehension might occur due to differences such as orthographic access or attention allocation.
Acknowledgements
We thank Magnus Alm, Håvard Tveit and Ane Torsdottir for their support in stimulus planning and preparation, experiment programming and data collection.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 In contrast to lexically marked role nouns such as catwomen, or chairmen or nouns with definitional gender such as ladies or knights.
2 As in Gabriel and Gygax (Citation2008), we also ran the analyses considering differences, either from the Pilot in Gabriel and Gygax (Citation2008), or from recently published norms (Misersky et al., Citation2015). Since none of these analyses signalled any substantial difference, we kept the original materials.
3 Critical information has been underlined for expository purpose.
4 Originally 73 participants, but data from one left-hander was removed, as the dominant hand was not used for the yes-response.
5 In Gabriel and Gygax (Citation2008) yes/no responses had been analysed with by-item and by-participant ANOVAs, using the share of yes-responses as DV for the yes/no responses. To improve comparability, the data were reanalysed fitting mixed-effect models as described above. Results of their analyses did not change.