This paper reports a production study that seeks to further understanding of U.S. English speakers' use of singular they. This preliminary study explores whether the likelihood of a speaker producing singular they depends on whether the antecedent is realized with a (gender-specific) first name or a last name (e.g., Lisa walked in vs. Jones walked in) and whether different kinds of last names (e.g., Knight, Atkinson, Flowers) pattern differently. Last names make it possible to test use of singular they in a probabilistic situation where gender is signaled by a (violable) social convention, but not explicitly encoded on the noun. What assumptions do people make about the referent of a (proper) noun with a 'socially implied' male bias, and how does this impact the use of singular they? This work used a production task where participants saw a first or last name and wrote a paragraph about the person, based on five gender-neutral bullet points. Participants' texts were analyzed for what pronouns, if any, they produced. The results provide strong evidence that last-name-only style has a strong male bias and show that this bias arises with a variety of last name types, but also show how (many) people's use of singular they differentiates between antecedents with and without lexically specified gender. This work explores the links between these findings and current syntactic analyses of singular they and its featural properties, and furthermore, the results reported in this work provide hints that the male bias may not be equally strong for all last names and may be modulated by the semantic associations of names' components, a finding which is potentially unexpected under many analyses of proper names' semantics. However, further work with more participants and a larger number of names is needed.