Political language often has gendered connotations, both in the topics politicians talk about, but also in the words they use. Using a validated dictionary of 700 words rated on the dimension of masculinity and femininity, we analyze approximately 27,000 public statements and interviews of U.S. Senators. We expect that Republicans will use more masculine language than Democrats, and that women will use more feminine language than men. We additionally expect that the effect of partisanship on the use of masculine or feminine language will be larger than the effect of gender of the speaker. Taking advantage of the depth and breadth of data collected on what US Senators tweet, we use a word2vec embedding neural network to identify similar words to those in the dictionary. This expansion gives us a more complete view of the masculine and feminine language that does not rely wholly on ratings from coders.