The analysis presented in this paper suggest that text analysis is a promising and relevant line of research for science studies. With the indicators for gender stereotyping deployed in this paper, we were able to identify different levels of gender stereotyping in panels which do moderately strong correlate with the levels of gender bias in the outcomes of the decision-making process. In the mediation analysis we found that language use in the review reports - and consequently gender stereotyping – explains a large part of the effect of gender on the panel scores. Language use indicates the (implicit) opinions of panel members and reviewers about what the position of women in science should be, and this translates into the gender differences in the scores male and female applicants receive and in the final outcome of the selection process. If this is correct, the process of selecting panel members may need to become much more sensitive to levels of gender stereotyping. Additionally, we addressed gender differences self-presentation: are there differences between men and women in language use in the grant proposals. This seems somewhat the case, and also here we found that part of the gender differences in the panel scores may be due to these differences in language use. One should be aware that we here only looked at some language differences between men and women, and that other aspects of gendered language use need to be included to derive a reliable picture. Further research will test possible improvements of the stereotyping variables, and also include other linguistic characteristics of the texts. Due to size restrictions, we cannot discuss this more in detail in this short paper. Another extension of the paper would be the use of more extended models, which may combine the text analysis of grant applications and of review reports with other aspects of the evaluation process, such as the composition of the decision-making panels and committees, and the specific procedures in which the (evaluative) texts play a role.