This is a really interesting study. But I wouldn't assume correlation implies causation here.
Just because there are more male or female applicants to job postings that contain specific words does not mean the words themselves are attracting certain types of applicants.
Many of the masculine words are words associated with higher-level management and higher pay jobs. There are less women who make it to this level so naturally there are less women applicants to these types of jobs.
Jobs associated with the feminine words such as "caring" and "serves" tend to be lower-level service-oriented jobs. Women don't apply to these because they are attracted to the words. They apply to them because more women are in the economic position to apply to these jobs over high-status high-skilled jobs because of historical and societal issues. I'm not sure categorizing these words as masculine and feminine is doing anything more than reflecting the existing gender-imbalance in labor although it is interesting to think about how this knowledge could be used constructively.