General Discussion

Using singular they when a referent’s gender is unknown or unspecified has become increasingly common and accepted by U.S. English speakers in recent years. Using they/them pronouns is not exclusive to nonbinary people, but it is a particularly common choice for them, since singular they is the most common gender-neutral alternative to he/him and she/her (Cassian, 2021, 2022; Cheung et al., 2020; Parks et al., 2023). The majority of U.S. adults (50–70% across age groups) know at least a little about people using they/them pronouns, and 11–46% personally know someone who uses they/them in 2021, up from 8–32% in 2018. While these proportions are highest in younger generations, they have been increasing across all generations, as well as across political affiliations and religious groups (Minkin & Brown, 2021; Parker et al., 2019). This increase in familiarity with gender-neutral pronouns corresponds to increases in both the visibility and the size of the trans and gender diverse (TGD) community. Estimates based on probability-sampling surveys suggest that 1 in 200 adults in the U.S. consider themselves⁴ transgender (Crissman et al., 2017; Herman et al., 2022; Meerwijk & Sevelius, 2017). This number has been increasing since the mid-2010s (CDC, 2021), especially as population-based surveys are changing their question designs to recognize nonbinary people who fall under the TGD umbrella, but may not also identify as transgender. Among Gen Z adults (ages 18–26 in 2021 and 2022), 2.3% identify as trans and 3.3% as none of these (out of man, woman, transgender), which corresponds to the population of the fifth-largest city in the U.S. (Twenge, 2023).

The backlash to an increasing number of teens and young adults openly expressing TGD identities has been severe. 2023 has seen a record number of bills introduced in U.S. state legislatures targeting healthcare for, civil rights protections for, and education about TGD people (Human Rights Campaign, 2023). Many of these bills include policies preventing people from respecting that someone has changed their pronouns, particularly in educational contexts (e.g., the recently-passed TN S.B. 0466/H.B. 1269, 2023). “Using pronouns” has become a symbol of whether or not you acknowledge and respect TGD people’s existence. Indeed, people describe being misgendered as alienating, devaluing, invalidating, and painful (Cordoba, 2020; Goldberg et al., 2019; Gunn, 2020; Johnson, 2019; Pitcher, 2017; Saltzburg & Davis, 2010; Truszczynski et al., 2020), and people who are misgendered more frequently show higher rates of depression, stress, suicidality, disordered eating, body dissatisfaction, and low self-esteem (Galupo et al., 2020; McLemore, 2015, 2018; Mitchell et al., 2021; The Trevor Project, 2020).

However, this opposition represents a loud minority. Many people are open to learning: across age groups, 40–60% of U.S. adults say they would be somewhat or very comfortable using a gender-neutral pronoun for someone who asked (Minkin & Brown, 2021). Still, learning a new language form can be hard, and many people continue to make errors, despite supportive intentions. This often results in seemingly-counterintuitive errors like “she uses they/them pronouns” (e.g., Example 1). I began by arguing that errors like these occur because learning to use they/them pronouns requires a change in pronoun production processes. Instead of selecting pronouns based on morphosyntactic gender features of a name (e.g., Schmitt et al., 1999) or based on semantic/conceptual gender features of a person (e.g., Antón-Méndez, 2010), speakers may need to retrieve an episodic memory of a person’s stated pronouns or the pronouns they have observed other speakers using in reference to that person. To that end, this dissertation investigated how people learn to associate pronouns with a person, how they use that information in language production and comprehension, and what factors can support learning.

Experiment 1 demonstrated that speakers can learn to associate pronouns with a person, in a context where the use of they/them could not be predicted from the name. While explicit memory for a person’s pronouns strongly predicted accurate production, it did not guarantee it. In Experiment 2, reading a brief PSA about gendered language increased the proportion of speakers who produced singular they at least once and their accuracy when doing so—arguably by motivating participants to pay more attention to information about the character’s pronouns during the introductions, then to attempt to retrieve the information later during the sentence completion task. Experiment 3 then showed that when information about a person’s pronouns is made more salient by directly stating it during introductions and made more available by including it on nametags, speakers can and do use this information to correctly produce singular they. These results all support a model where speakers can use episodic memory for a person’s stated pronouns to select which pronoun to produce, behavior that is not accounted for in current processing models.

One of the primary focuses of research on pronoun production has been when speakers choose to use pronouns instead of other referring expressions (Arnold & Zerkle, 2019), but the subsequent choice of which pronoun to use is less well understood. If applied to proper names and pronouns referring to people, production models either assume that speakers select pronouns based on morphosyntactic features of the name, or based on semantic/conceptual features of a person. As an example of the first approach, Schmitt et al.’s (1999) model of lexical access in pronoun production argues that if a lexical concept is activated and sufficiently “in focus” in the discourse, the speaker will produce a pronoun instead of a full noun phrase. In this model, tested with object labels in German, activating the lexical concept for the noun also activates the corresponding grammatical gender node (masculine, feminine, or neuter). If the speaker uses a pronoun, the gender node is selected in order to produce the correct pronoun (Jescheniak & Levelt, 1994; Levelt et al., 1999; Roelofs, 1992). When producing gender-marked pronouns or determiners, competition between different forms can arise from the grammatical gender features of other lexical concepts that are also activated (Schiller & Caramazza, 2003). If each lexical concept can only correctly be associated with a single grammatical gender, models of grammatical gender selection may not need to include competition between multiple grammatical gender features activated by the same lexical concept. The closest analogue may be languages where the singular and plural forms of determiners vary for the same grammatical gender, and one approach argues that the singular form is activated by default and can interfere with activating the plural form (Jescheniak et al., 2014; Schriefers et al., 2002).

In models of grammatical gender processing, it is reasonable to assume that once a speaker decides to produce a pronoun, there is typically only one correct pronoun to use. If extended from pronouns referring to objects to third person gendered pronouns referring to people, Schmitt et al.’s model might predict that the lexical entry for a masculine name would automatically activate the masculine gender node, which is linked to he/him pronouns, and the lexical entry for a feminine name would automatically activate the feminine gender node, which is linked to she/her pronouns. However, this approach would need to be modified to account for how speakers use singular they. One option would be to add a neutral gender node,¹ which is linked to singular they/them pronouns. The lexical entry for a name could activate the neutral gender node, in addition to or instead of the masculine or feminine nodes. This is similar to how Konnelly & Cowper’s morphosyntax model of gender agreement (described in Section 0.3) could function in a processing model. They propose that “innovative” speakers, who only accept specific definite antecedents for singular they that are gender-unspecified, can accommodate a proper name by adding a new lexical entry for the name that does not have a masculine or feminine feature (2020).

This gets us closer to accounting for the use of singular they in the current experiments by allowing a name to activate multiple gender nodes. In Experiments 1 & 2, the masculine names would activate the masculine gender node and possibly also the neutral gender node, and the feminine names would activate the feminine gender node and possibly also the neutral gender node. If the neutral gender node is activated, it is presumably in competition with the masculine or feminine node, with speakers trying to use they instead of defaulting to he or she. However, it is clear that speakers are using more than the lexical information for proper names to select pronouns. None of the current experiments tested a context where names provided enough information to predict pronouns, such as a set of characters where everyone with masculine names used he/him, everyone with feminine names used she/her, and everyone with gender-neutral names used they/them. Instead, Experiments 1 & 2 only used masculine and feminine names, with one third of the characters using they/them and the other two thirds using the expected he/him or she/her. Experiment 3 used all gender-neutral names, where the use of he/him, she/her, or they/them could not be predicted from the name. In both contexts, using singular they required information in addition to the lexical features or knowledge about gender distributions associated with the name. Participants had to associate they/them with the particular character, i.e., this Alex, not all Alex’s in general.

Instead of proposing that pronouns are selected via lexical information associated with the name, pronoun production models can assume that speakers select pronouns based on semantic or conceptual information about the person’s gender (Ackerman, 2019). Here, psycholinguistics research has mostly focused on errors in adult second language learning. For example, English possessive pronouns agree with the possessor (e.g., in Jaime gave the apple to their brother, the pronoun agrees with Jaime), which can result in errors by native speakers of languages where possessive pronouns agree with the possessee (e.g., the pronoun would agree with brother instead of Jaime) (Pozzan & Antón-Méndez, 2017). This is more of an issue about which person the pronoun agrees with, not which gendered form should be selected for the person. Other experiments have looked at errors by native speakers of languages that do not mark gender in third-person singular pronouns, such as Finnish and spoken Mandarin Chinese (Pozzan & Antón-Méndez, 2017; Siewierska, 2013). This results in errors between he/him and she/her, where it is assumed that the speaker has the correct inference about the person’s gender, but not a consistent mapping between gender categories and which pronouns to produce² (Ackerman, 2019), since it is not information their native language requires them to encode (Slobin, 1996).

An approach where speakers select pronouns based on semantic or conceptual information about the person’s gender can better account for they/them pronouns, particularly if you assume that information about a person’s pronouns could be encoded as part of conceptual gender. Traditional approaches treat conceptual gender as primarily automatic inferences (see the mismatching gender stereotype experiments discussed in Section 0.4.1), but one option is to assume episodic information can be added. If someone states that they use they/them pronouns or other speakers are using they/them pronouns for them, this could prompt a revision of the original gender inference that directed to he/him or she/her.³ This model would also need to be modified to account for multiple pronouns being associated with a person, instead of having a one conceptual gender inference to one set of pronouns mapping. Speakers often alternate between they/them and he/him or she/her, like the errors in these experiments or in the Demi Lovato article (Towers, 2021). Alternating between pronouns can also be intentional when referring to people who accept multiple pronouns (e.g., Gurley, 2021; Monáe & Lore, 2022). A one-to-one model also does not account for gender-unspecified specific singular they, particularly when it is used when the speaker does know the referent’s gender, but does not include that information. In this case, the conceptual gender information is not driving pronoun selection, because the speaker is choosing to not specify gender. Overall, the behavior of participants producing singular they in the first three experiments demonstrates that speakers are incorporating additional sources of information and choosing between more options than are accounted for by existing processing models, where pronouns are selected based on the lexical information connected to the name or the gender categorization connected to the person.

Experiment 4 turned from production to comprehension, but with the same underlying question: specific definite singular they does show a processing cost compared to he and she, but it is still unclear how large and how persistent those costs are, and how singular they is fitting into the existing processing mechanisms comprehenders have for he and she. Results from a visual world study showed that, in a context where participants could come to expect certain characters to be referred to with they/them, listeners were able to use singular they to identify the referent, just to a lesser degree than with he and she. This resembles patterns seen in young children (Arnold et al., 2007; Song & Fisher, 2005) and in adult second language learners (Cunnings et al., 2017; Grüter et al., 2012; Speyer & Schleef, 2019). One open question is how these findings will replicate in future years as people continue to gain experience with using they/them pronouns, and if the learning trajectory will be similar to those observed for other pronouns during language development and second language acquisition.

Unlike prior findings for when he or she is ambiguous between two referents (Arnold et al., 2000, 2007; Brown-Schmidt & Toscano, 2017; Falandays et al., 2020) and preliminary findings for singular they using a different task (Arnold et al., 2023), participants in Experiment 4 did not show an order of mention effect for singular they. The pronouns in these stories can refer to either of the two named characters, but is more likely to refer to the character named first than the character named second (Gernsbacher, 1989; Kaiser & Trueswell, 2011). Listeners can use this information to help disambiguate the referent, and thus start looking at the target character—which the visual world paradigm takes to mean that they understand who the sentence is referring to—earlier for characters mentioned first. This pattern was not observed in they trials. The third finding in Experiment 4 was that after hearing he or she, listeners were more likely to be looking at the target character if the competitor character used she/her or he/him than if the competitor character used they/them. The lack of an order of mention effect and the observation that they/them characters behaved differently as competitors than he/him and she/her characters both suggest that while listeners can successfully comprehend they as singular, they may be using different strategies to do so, leaving a number of follow-up research questions.

I want to emphasize here that the processing cost for singular they is large enough to be statistically significant and relevant for understanding language processing mechanisms, but that this does not necessarily correspond to meaningful differences in processing fluency—people’s subjective experience of how easy it was to understand the story (Alter & Oppenheimer, 2009). In this experiment, visual inspection of the data shows that around 50% of participants were looking at the target character half a second after the onset of he or she, compared to around 35% half a second after the onset of they (Figure 4.7). Similarly, in one reading time study, indefinite gender-unspecified they (e.g., a person, someone) took about 100ms longer than he and she, which the authors note is a smaller, more easily accommodated processing cost than other types of gender and number agreement mismatches (Sanford & Filik, 2007) (see Section 0.4.2). The processing fluency account predicts that language that is more difficult to understand elicits more negative attitudes about the speaker (Dragojevic et al., 2017; Dragojevic, 2020; Dragojevic & Giles, 2016) (see Section 4.1). However, it is an open question how processing costs in psycholinguistics measures correspond to processing fluency—how the time course in a visual world study matches how hard a listener consciously perceives singular they. Like Sanford & Filik I would argue that, when compared to the processing costs seen in other experiments, the processing costs observed here with singular they are within the range that comprehenders can normally accommodate. Instead of singular they being dramatically or permanently more difficult to understand, reference resolution is already complicated and ambiguous, even in contexts that are less socio-politically marked.

From a theoretical perspective, the four experiments in this dissertation demonstrate how singular they pushes language processing models to incorporate additional factors—especially social ones—to account for the full range of people’s language use. From a practical perspective, these experiments show that while learning singular they is hard, it is doable. Multiple strategies can support learning, including providing information about why paying attention to gendered language is important, showing examples of singular they in use, making information about pronouns more salient by including it in introductions, and making information about pronouns more accessible by including it in places like nametags and display names. Learning to use they/them pronouns is worthwhile, because misgendering is a key aspect of how respected and welcome trans and gender diverse people feel in a community, and reducing misgendering is something that individual allies have the opportunity to concretely change.

Ackerman, L. (2019). Syntactic and cognitive issues in investigating gendered coreference. Glossa: A Journal of General Linguistics, 4(1), 117. https://doi.org/10.5334/gjgl.721

Alter, A. L., & Oppenheimer, D. M. (2009). Uniting the tribes of fluency to form a metacognitive nation. Personality and Social Psychology Review, 13(3), 219–235. https://doi.org/10.1177/1088868309341564

Antón-Méndez, I. (2010). Gender bender: Gender errors in L2 pronoun production. Journal of Psycholinguistic Research, 39(2), 119–139. https://doi.org/10.1007/s10936-009-9129-z

Arnold, J. E., Brown-Schmidt, S., & Trueswell, J. (2007). Children’s use of gender and order-of-mention during pronoun comprehension. Language and Cognitive Processes, 22(4), 527–565. https://doi.org/10.1080/01690960600845950

Arnold, J. E., Eisenband, J. G., Brown-Schmidt, S., & Trueswell, J. C. (2000). The rapid use of gender information: Evidence of the time course of pronoun resolution from eyetracking. Cognition, 76(1), B13–B26. https://doi.org/10.1016/S0010-0277(00)00073-1

Arnold, J. E., Payst, N., Kakoki, E., & Wang, L. (2023). Online comprehension of nonbinary ”they” is easier with supporting linguistic context [Poster]. 36th Annual Conference on Human Sentence Processing.

Arnold, J. E., & Zerkle, S. A. (2019). Why do people produce pronouns? Pragmatic selection vs. Rational models. Language, Cognition and Neuroscience, 34(9), 1152–1175. https://doi.org/10.1080/23273798.2019.1636103

Brown-Schmidt, S., & Toscano, J. C. (2017). Gradient acoustic information induces long-lasting referential uncertainty in short discourses. Language, Cognition and Neuroscience, 32(10), 1211–1228. https://doi.org/10.1080/23273798.2017.1325508

Cassian. (2021). Gender census 2021: Worldwide report [Data Set]. https://web.archive.org/web/20220828234326/https://www.gendercensus.com/results/2021-worldwide/

Cassian. (2022). Gender census 2022: Worldwide report [Data Set]. https://web.archive.org/web/20220827231434/https://www.gendercensus.com/results/2022-worldwide/

Centers for Disease Control and Prevention. (2021). 2021 Behavioral risk factor surveillance system survey [Data Set]. https://www.cdc.gov/brfss/annual_data/annual_2021.html

Cheung, A. S., Leemaqz, S. Y., Wong, J. W. P., Chew, D., Ooi, O., Cundill, P., Silberstein, N., Locke, P., Zwickl, S., Grayson, R., Zajac, J. D., & Pang, K. C. (2020). Non-binary and binary gender identity in Australian trans and gender diverse individuals. Archives of Sexual Behavior, 49(7), 2673–2681. https://doi.org/10.1007/s10508-020-01689-9

Cordoba, S. (2020). Exploring non-binary genders: Language and identity (Publication No. 29424421) [Doctoral dissertation, De Montfort University]. ProQuest Dissertations & Theses Global.

Crissman, H. P., Berger, M. B., Graham, L. F., & Dalton, V. K. (2017). Transgender demographics: A household probability sample of US adults, 2014. AJPH Transgender Health, 107(2), 213–215. https://doi.org/10.2105/AJPH.2016.303571

Cunnings, I., Fotiadou, G., & Tsimpli, I. (2017). Anaphora resolution and reanalysis during L2 sentence processing. Studies in Second Language Acquisition, 39(4), 621–652. https://doi.org/10.1017/S0272263116000292

Dragojevic, M. (2020). Extending the fluency principle: Factors that increase listeners’ processing fluency positively bias their language attitudes. Communication Monographs, 87(2), 158–178. https://doi.org/10.1080/03637751.2019.1663543

Dragojevic, M., & Giles, H. (2016). I don’t like you because you’re hard to understand: The role of processing fluency in the language attitudes process. Human Communication Research, 42(3), 396–420. https://doi.org/10.1111/hcre.12079

Dragojevic, M., Giles, H., Beck, A.-C., & Tatum, N. T. (2017). The fluency principle: Why foreign accent strength negatively biases language attitudes. Communication Monographs, 84(3), 385–405. https://doi.org/10.1080/03637751.2017.1322213

Falandays, J. B., Brown-Schmidt, S., & Toscano, J. C. (2020). Long-lasting gradient activation of referents during spoken language processing. Journal of Memory and Language, 112, 104088. https://doi.org/10.1016/j.jml.2020.104088

Friedman, E. J. (2014). Cisgenderism in gender attributions: The ways in which social, cognitive, and individuals factors predict misgendering (Publication No. 3623701) [Doctoral dissertation, The City University of New York]. ProQuest Dissertations & Theses Global.

Galupo, M. P., Pulice-Farrow, L., & Lindley, L. (2020). ”Every time I get gendered male, I feel a pain in my chest”: Understanding the social context for gender dysphoria. Stigma and Health, 5(2), 199–208. https://doi.org/10.1037/sah0000189

Gernsbacher, M. A. (1989). Mechanisms that improve referential access. Cognition, 32(2), 99–156. https://doi.org/10.1016/0010-0277(89)90001-2

Goldberg, A. E., Kuvalanka, K., & Dickey, L. (2019). Transgender graduate students’ experiences in higher education: A mixed-methods exploratory study. Journal of Diversity in Higher Education, 12(1), 38–51. https://doi.org/10.1037/dhe0000074

Grüter, T., Lew-Williams, C., & Fernald, A. (2012). Grammatical gender in L2: A production or a real-time processing problem? Second Language Research, 28(2), 191–215. https://doi.org/10.1177/0267658312437990

Gunn, H. A. (2020). The experience of misgendering among trans and gender diverse people (Publication No. 2447222123) [Doctoral dissertation, University of Massachusetts Boston]. ProQuest Dissertations & Theses Global.

Gurley, A. (2021, August 12). ”The crown” star Emma Corrin got candid about their ”scary” decision to come out as queer. Buzzfeed. https://web.archive.org/web/20210813041253/https://www.buzzfeed.com/alexgurley/emma-corrin-queer-journey%0A

Herman, J. L., Flores, A. R., & O’Neill, K. K. (2022). How many adults and youth identify as transgender in the United States? UCLA School of Law. https://web.archive.org/web/20230430114921/https://williamsinstitute.law.ucla.edu/publications/trans-adults-united-states/

Human Rights Campaign. (2023, February 15). Human Rights Campaign working to defeat 340 anti-LGBTQ+ bills at state level already, 150 of which target transgender people – highest number on record. https://web.archive.org/web/20230401063456/https://www.hrc.org/press-releases/human-rights-campaign-working-to-defeat-340-anti-lgbtq-bills-at-state-level-already-150-of-which-target-transgender-people-highest-number-on-record

Jescheniak, J. D., & Levelt, W. J. M. (1994). Word frequency effects in speech production: Retrieval of syntactic information and of phonological form. Journal of Experimental Psychology: Learning, Memory and Cognition, 20(4), 824–843. https://doi.org/10.1037/0278-7393.20.4.824

Jescheniak, J. D., Schriefers, H., & Lemhöfer, K. (2014). Selection of freestanding and bound gender-marking morphemes in speech production: A review. Language, Cognition and Neuroscience, 29(6), 684–694. https://doi.org/10.1080/01690965.2012.654645

Johnson, A. (2019). Read between the lines: Teaching and learning while nonbinary trans. In G. Beemyn (Ed.), Trans people in higher education (pp. 27–34). SUNY Press.

Kaiser, E., & Trueswell, J. C. (2011). Investigating the interpretation of pronouns and demonstratives in Finnish: Going beyond salience. In E. Gibson & N. J. Pearlmutter (Eds.), The processing and acquisition of reference. MIT Press. https://doi.org/10.7551/mitpress/9780262015127.003.0013

Konnelly, L., & Cowper, E. (2020). Gender diversity and morphosyntax: An account of singular ”they”. Glossa: A Journal of General Linguistics, 5(1), 1–15. https://doi.org/http://doi.org/10.5334/gjgl.1000

Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech production. Behavioral and Brain Sciences, 22, 1–75.

McLemore, K. A. (2015). Self and identity experiences with misgendering: Identity misclassification of transgender spectrum individuals. Self and Identity, 14(1), 51–74. https://doi.org/10.1080/15298868.2014.950691

McLemore, K. A. (2018). A minority stress perspective on transgender individuals’ experiences with misgendering. Stigma and Health, 3(1), 53–64. https://doi.org/10.1037/sah0000070

Meerwijk, E. L., & Sevelius, J. M. (2017). Transgender population size in the United States: A meta-regression of population-based probability samples. AJPH Transgender Health, 107(2), 1–8. https://doi.org/10.2105/AJPH.2016.303578

Minkin, R., & Brown, A. (2021). Rising shares of U.S. adults know someone who is transgender or goes by gender-neutral pronouns. Pew Research Center. https://web.archive.org/web/20210731145136/https://www.pewresearch.org/fact-tank/2021/07/27/rising-shares-of-u-s-adults-know-someone-who-is-transgender-or-goes-by-gender-neutral-pronouns/

Mitchell, L., MacArthur, H. J., & Blomquist, K. K. (2021). The effect of misgendering on body dissatisfaction and dietary restraint in transgender individuals: Testing a misgendering-congruence process. International Journal of Eating Disorders, 54(7), 1295–1301. https://doi.org/10.1002/eat.23537

Monáe, J., & Lore, D. (2022). Nevermind. In The memory librarian: And other stories of dirty computer. Harper Voyager.

Parker, K., Graf, N., & Igielnik, R. (2019). Generation Z looks a lot like millennials on key social and political issues. Pew Research Center. https://web.archive.org/web/20210210083104/https://www.pewresearch.org/social-trends/2019/01/17/generation-z-looks-a-lot-like-millennials-on-key-social-and-political-issues/

Parks, C., Guskin, E., & Clement, S. (2023, March 23). Most trans adults say transitioning made them more satisfied with their lives. The Washington Post. https://web.archive.org/web/20230323123428/https://www.washingtonpost.com/dc-md-va/2023/03/23/transgender-adults-transitioning-poll/

Pitcher, E. N. (2017). ”There’s stuff that comes with being an unexpected guest”: Experiences of trans* academics with microaggressions. International Journal of Qualitative Studies in Education, 30(7), 688–703. https://doi.org/10.1080/09518398.2017.1309588

Pozzan, L., & Antón-Méndez, I. (2017). English possessive gender agreement in production and comprehension: Similarities and differences between young monolingual English learners and adult Mandarin-English second language learners. Applied Psycholinguistics, 38(4), 985–1017. https://doi.org/10.1017/S0142716417000017

Roelofs, A. (1992). A spreading-activation theory of lemma retrieval in speaking. Cognition, 42, 107–142. https://doi.org/10.1016/0010-0277(92)90041-F

Saltzburg, S., & Davis, T. S. (2010). Co-authoring gender-queer youth identities: Discursive tellings and retellings. Journal of Ethnic And Cultural Diversity in Social Work, 19(2), 87–108. https://doi.org/10.1080/15313200903124028

Sanford, A. J., & Filik, R. (2007). ”They” as a gender-unspecified singular pronoun: Eye tracking reveals a processing cost. Quarterly Journal of Experimental Psychology, 60(2), 171–178. https://doi.org/10.1080/17470210600973390

S.B. 0466/H.B. 1269. (2023). 113th Tennessee General Assembly. http://web.archive.org/web/20230317012106/https://wapp.capitol.tn.gov/apps/BillInfo/Default.aspx?BillNumber=SB0466

Schiller, N. O., & Caramazza, A. (2003). Grammatical feature selection in noun phrase production: Evidence from German and Dutch. Journal of Memory and Language, 48(1), 169–194. https://doi.org/10.1016/S0749-596X(02)00508-9

Schmitt, B. M., Meyer, A. S., & Levelt, W. J. M. (1999). Lexical access in the production of pronouns. Cognition, 69(3), 313–335. https://doi.org/10.1016/S0010-0277(98)00073-0

Schriefers, H., Jescheniak, J. D., & Hantsch, A. (2002). Determiner selection in noun phrase production. Journal of Experimental Psychology: Learning, Memory, and Cognition, 28(5), 941–950. https://doi.org/10.1037/0278-7393.28.5.941

Siewierska, A. (2013). Gender distinctions in independent personal pronouns. In M. S. Dryer & M. Haspelmath (Eds.), The world atlas of language structures online. Max Planck Institute for Evolutionary Anthropology. https://web.archive.org/web/20230309235830/https://wals.info/chapter/44

Slobin, D. I. (1996). From ”thought and language” to ”thinking for speaking”. In J. J. Gumperz & S. C. Levinson (Eds.), Rethinking linguistic relativity (pp. 70–96). Cambridge University Press.

Song, H., & Fisher, C. (2005). Who’s ”she”? Discourse prominence influences preschoolers’ comprehension of pronouns. Journal of Memory and Language, 52(1), 29–57. https://doi.org/10.1016/j.jml.2004.06.012

Speyer, L. G., & Schleef, E. (2019). Processing ”gender-neutral” pronouns: A self-paced reading study of learners of English. Applied Linguistics, 40(5), 793–815. https://doi.org/10.1093/applin/amy022

The Trevor Project. (2020). 2020 National survey on LGBTQ youth mental health. https://web.archive.org/web/20210104100949/https://www.thetrevorproject.org/survey-2020/

Towers, A. (2021, July 2). Demi Lovato thanks Lizzo for correcting paparazzo who misgendered her: ”I love you”. Entertainment Weekly. https://web.archive.org/web/20210702144731/https://ew.com/music/demi-lovato-thanks-lizzo-correcting-paparazzo-who-misgendered-her/

Truszczynski, N., Singh, A. A., & Hansen, N. (2020). The discrimination experiences and coping responses of non-binary and trans people. Journal of Homosexuality, 69(4), 741–755. https://doi.org/10.1080/00918369.2020.1855028

Twenge, J. M. (2023, May 1). How Gen Z changed its views on gender. Time. https://web.archive.org/web/20230501110039/https://time.com/6275663/generation-z-gender-identity/

---

1. A related question is if the model would need separate nodes for gender-unspecified and gender-specified singular they. But since the experiments here only elicit gender-specified they, I focus on that form.

2. Accounts of grammatical gender errors assume the opposite. Grammatical gender marking errors remain common even in highly-proficient learners, particularly if they are native speakers of a language without grammatical gender (e.g., English speakers learning Spanish) (Antón-Méndez, 2010; Grüter et al., 2012). These errors occur when a speaker selects the wrong grammatical gender feature for a word, then produces the pronoun corresponding to that wrong grammatical gender category.

3. Here, I can only claim that information about gendered language forms is added to knowledge about a person, which could but does not necessarily correspond to changing an underlying gender categorization. See Friedman (2014) for a theoretical model and behavioral data about how people revise their original automatic categorizations of a person’s gender.