Good news! The ethics forms for the project are submitted, I know where the eye-tracker is and I’m applying for a small internal grant to get the pilots going. Everything is moving at a pleasantly fast pace and I’m feeling like it’s all going to be possible. (I really can’t recommend highly enough getting together with a bunch of wonderful colleagues for a regular writing group. It really lifts the spirits!)
In short, the first set of experiments I’m getting of the ground are the ones examining the influence of non-grammatical gender on the parser’s behavior. Van Gompel and Liversedge (2003), for instance, show that gender and number are lexical features that are only checked after a long distance dependency is formed. This is what makes the Gender Mismatch Effect (GMME) such a great tool for real-time sentence processing studies — the mismatch can cause the parser to apparently slow down while it revises or repairs its mistake. What I want to know, though, is how flexible this gender-checking mechanism is. As Frazier et al. (2015) notes, in reference to the possible [un]grammaticality of sentences with ‘gender mismatches’:
[I]t is not unimaginable that a woman might be named Steven, merely very unexpected, and likewise in the context of a costume party, the individual picked out by the referring expression the cowgirl could conceivably be male.
Moreover, there are people who do not identify as men or women, but as some non-binary gender (of which there are many). Thus, the assumptions underlying many studies that use gender mismatch to test, e.g., long-distance dependency formation may rely on a gender binary that is not sociologically or biologically valid. To be fair, a large proportion of Western populations only have experience with cisgender people, thus rarely have occasion to process sentences that contain non-binary pronouns. But, as Frazier and colleague point out, this type of real-life gender mismatch can occur anywhere costume parties occur, which is presumably a much larger proportion of the population than those who are familiar with non-binary gender identities.
So, I will be testing whether the GMME is something that is flexible that can be mitigated with exposure to non-binary and ‘nonstandard’ uses, or whether it is something rigid that is acquired at a young age* and is difficult to adapt to new experiences and knowledge. Both cases could have implications for how to positively change societal treatment of people who want to be referred to with non-binary pronouns. Both cases could also inform theories of how the parser assimilates non-linguistic (or paralinguistic) information during real-time sentence processing.
(*I assume that many to most people who are familiar with non-binary gender identities learned about them well after the critical period, based on the broad societal attitude toward transgender and non-binary/gender-queer people in Western cultures.)