Current years have met an increased interest in using gender-neutral terms in language to represent individuals with non-binary gender identities, that is, people who do not identify within the male/female binary. This includes not identifying as male or female, identifying as both male and female, or identifying somewhere in between the binary gender system (McNabb, 2017). Non-binary identities have been recognized as a third gender in indigenous cultures for centuries (McNabb, 2017). Some examples comprise the two-spirit Native American identity, the hijras in India, and the muxes in Mexico (see Independent Lens, 2015, for a map of cultures around the world that recognize more than two genders). However, in other cultures the acknowledgment and open discussion of non-binary identities, as well as the use of gender-inclusive language, is more recent. Researchers on the topic have discussed both neutral and innovative gender-fair language strategies to achieve inclusion of all identities (Sczesny et al., 2016). Neutralization proposals comprise epicene or collective nouns or noun phrases that have no grammatical gender marking and are, thus, considered neutral, such as people or humankind in English, and individuo ‘individual’ or el estudiantado ‘the student body’ in Spanish. Innovative proposals include lexical, orthographic, and morphological alternatives. Regarding lexical options, for instance, in Swedish, a new pronoun, hen, has been introduced with a generic singular meaning (Vergoossen et al., 2020). In English, new pronouns, such as ze and xe, have also been proposed (Bradley et al., 2019), although the use of the singular pronoun they seems to be more widely recognized and accepted (Camilliere et al., 2021). Some orthographic options, which have the written form as their central objective, proposed for different languages include the asterisk in German (Körner et al., 2022; Zacharski & Ferstl, 2023) and the underscore in Slovene (Popič & Gorjanc, 2018). In the present study, we focus on a grammatically gendered language, Spanish, and on morphological innovations that have been proposed as gender-fair strategies to achieve language inclusivity. Before we explain the GIMs that have appeared in Spanish, we provide some information on Spanish grammatical gender.

Unlike English, a natural gender language in which gender depends on lexical assignment (e.g., boy for males and girl for females), or morphological distinction (e.g., actor for males and actress for females), Spanish is a grammatically gendered language that marks gender on all nouns, by classifying them into two categories that denote masculine or feminine gender. Generally, words that end with the morpheme –o mark the masculine gender (e.g., libro ‘book’), while words that end with the morpheme –a mark the feminine gender (e.g., silla ‘chair’). The –o = masculine/–a = feminine pattern is consistent 99.9% of the time for nouns ending with –o and 96.3% of the time for nouns ending with –a (Teschner & Russell, 1984). In Spanish, some nouns also end with –e, and 89.4% of these words are masculine. However, two-thirds of the most frequent words that end with –e are feminine (Teschner & Russell, 1984). This indicates that the morpheme –e is not a reliable marker for grammatical gender. Proponents of GIMs in Spanish take advantage of the ambiguous nature of the morpheme –e to suggest its use as one of the GIM options for words that name (i.e., nouns, determiners, pronouns) and describe (i.e., adjectives) people who identify outside of the feminine/masculine binary. In grammatically gendered languages, contrary to what some critics have stated, GIMs do not seek to abolish the grammatical gender categories of other animate and inanimate nouns, but rather pursue both gender neutrality as well as the representation of non-binary individuals in language (Vidal-Ortiz & Martínez, 2018). Based on this double function, GIMs can be considered both neutralization and innovative gender-fair language strategies (Sczesny et al., 2016).

In addition to the proposal of the GIM –e, Spanish speakers have suggested other GIMs, which include –x, –u, and –i. As mentioned above, these morphemes can be attached to nouns (adulte, adultx ‘adult’), pronouns (elle, ellx), articles (le, lx, lu, li ‘the’) and adjectives (alte ‘tall,’ rojx ‘red,’ cansadu ‘tired,’ harti ‘fed up’). As explained by Stetie and Zunino (2024), the GIMs that have been proposed are used both to refer specifically to people who identify as non-binary and as a generic form in plural noun phrases. In Spanish, although traditionally the masculine grammatical gender has been used as the generic plural form (Real Academia Española & Asociación de Academias de la Lengua Española, 2010), proponents of the use of gender-inclusive language argue that this is better achieved with GIMs and, therefore, suggest that the –o be replaced with one of them. According to Stetie and Zunino (2024), the use of these morphological innovations even “work better when they are used in plural form and to replace a generic use than when [they] are used to name or refer to individuals who do not perceive themselves within the traditional sex-gender binarism” (p. 450). This notion is further supported by the findings of a Twitter¹1

Officially known as ‘X’ since July 2023.

The letter x was first introduced into Spanish when Medieval Spanish scholars translated the Arabic word al-shayun, which can mean ‘the thing,’ ‘unknown thing,’ or ‘unknown something’ (Milian, 2017; Salinas & Lozano, 2021; TED, 2012). Medieval Spanish scholars took the sound “ck” from the Greek letter χ (chi) and replaced it with the Latin letter X so that the “sh” in al-shayun could be translated. Since its origins, –x in Spanish has had a history of representing the unknown, therefore providing room for the diverse gender identities in language. Among the Spanish GIMs that have been proposed, –e and –x currently seem to be the most widespread and the most researched (Cabello Pino, 2020; Fábregas, 2022; Michnowicz et al., 2023; Salinas, 2020; Salinas & Lozano, 2017, 2021; Stetie & Zunino, 2022, 2024), while –u, –i, and forms such as the asterisk in Spanish have yet to be examined more carefully. Stetie and Zunino (2024) mention that particularly in Argentina, –e is the most frequently used GIM (p. 447). Salinas and Lozano (2017) study the use of –x specifically in the term Latinx among the LGBTQIA+²2

The acronym stands for lesbian, gay, bisexual, transgender, questioning, intersex, and asexual. The plus sign denotes community members who relate to other forms of gender diverse identities and sexualities.

These innovative GIMs have been labeled direct forms of non-binary Spanish (NBS; López, 2020). Meanwhile, indirect NBS refers to unspecific terms (e.g., estas son mis amistades ‘these are my friends,’ el pueblo puertorriqueño ‘the Puerto Rican people’), what Stetie and Zunino (2024) and other researchers call neutralization strategies to achieve inclusion of different identities in language. These indirect forms are often used unconsciously by Spanish speakers and their gender-neutral value goes unnoticed (López, 2020). In contrast, the use of direct NBS (by employing GIMs), has generated tensions between Spanish grammarians and Feminist scholars, which are rooted in the linguistic sexism debate³3

For a thorough account of the functional grammar standpoint on this debate, see Roca (2005, 2006). For more information on linguistic sexism and the feminist critique of the Spanish language, see Bengoechea (2008, 2015) and Cabeza Pereiro and Rodríguez Barcia (2013).

Related to the debate described above, one group of studies focuses on the perceptions of and attitudes towards gender-inclusive forms. Ivanová and Kyseľová (2022) conducted a qualitative analysis of the answers and comments obtained from an online questionnaire completed by Slovak, Czech, and Polish speakers. The questionnaire focused mainly on comparing (1) the generic masculine form, (2) pair forms (full male and female noun forms or feminine forms), and (3) full male noun forms with female suffixes after a slash or in brackets. The findings displayed arguments reflecting both acceptance and rejection of forms substantiating gender-inclusive language. For instance, some speakers supported the use of the generic masculine as equally representative for both genders, and, hence, neutral, while others validated it as being male-biased and offensive. Speakers also supported the generic masculine more than the other alternatives by referring to its economy, its naturalness, and tradition—“the entrenchment of the generic masculine forms in the language system and the preference to keep the current system unchanged (defending the linguistic status quo)” (p. 413). Moreover, participants tended to associate forms other than the generic masculine with political correctness, feminist ideals, and gender balance (p. 414). To examine the reception of the gender-inclusive underscore in Slovene, Popič and Gorjanc (2018) conducted a corpus analysis of two collections of texts gathered six months apart from different online sources that use this mark, specifically web pages of organizations, members, and supporters of the LGBTQIA+ rights, to see whether its use was prolific, consistent, and extensive. The researchers found that among the LGBTQIA+ community and several other cultural or media outlets, the underscore is very present, it is used consistently, and its use is growing, as exhibited by a significant increase between the two collections of texts in the number of words and tokens including the underscore. With respect to one of the main arguments against the use of the underscore—its complex and difficult use—, the authors state: “the writers who use the underscore use it remarkably consistently as we can hardly find different versions/spellings of words written with an underscore. Thus, one of the main criticisms of the underscore (that it is hard to use) seems to be invalid, at least in our database” (p. 347).

Gustafsson Sendén et al. (2015) examined how attitudes and behavior toward the newly introduced Swedish gender-neutral pronoun hen changed over time. They administered questionnaires from 2012 (the year in which the pronoun hen was introduced) to 2015 in university and community settings in Sweden to obtain data on attitudes and usage of hen, and they considered the influence of factors, such as sexism, political orientation, gender identity, interest in gender issues, and age. They found that time was a significant predictor of attitudes, even when controlling for other predictors. Sexism, political right affiliation, and a strong gender identity were associated with negative attitudes, while younger people, women, students, and those who were interested in gender issues had more positive attitudes towards hen. Similar effects were found for usage of hen, but the overall effects were smaller than those for attitudes. The researchers concluded that attitudes changed faster than behavior, and that although the use of innovative gender-neutral pronouns might be met with resistance at first, attitudes and use can shift over time.

Moving back to Spanish, perceptions of inclusive language were examined by Michnowicz et al. (2023) through an online questionnaire of reported familiarity, opinions, and use of different GIMs among Spanish speakers residing in North Carolina. Overall, the findings show that women report more familiarity with GIMs than men, particularly those in the middle-aged group. Moreover, women have the most positive opinions of the GIMs. Although most participants did not report using any of the innovative morphological forms, those who did reported using –x and –@ more frequently in writing, and –e more often in speech. The authors point out that among the three linguistic contexts considered, a Det-N phrase, a Quant-Det-N phrase, and the word Latin-s, the last was the context in which –x was more frequently used. It is worth noting that this study included one young non-binary participant who was familiar with GIMs, had positive opinions of them, and reported using them, particularly –x. Attitudes towards gender-inclusive language comprise one of the variables considered in the present study.

In addition to studying language attitudes, previous research has examined the way in which different gender-inclusive language strategies are processed and the extent to which they can mitigate biases (Lindqvist et al., 2019; Renström et al., 2024; Stetie & Zunino, 2022, 2024; Vergoossen et al., 2020; Zarwanitzer, 2019). Vergoossen et al. (2020) conducted an eye-tracking study to examine the processing of the innovative gender-inclusive pronoun hen. Participants read pairs of sentences, in which the first sentence included a role noun and the second sentence included a pronoun; both were either gendered (masculine, feminine) or neutral. The authors examined reading time at the noun in the first sentence, at the pronoun in the second sentence, and at the spillover region following the pronoun. A processing cost for the pronoun hen only appeared in the spillover region, where reading time was longer when hen referred to neutral nouns than when it referred to gendered nouns. However, the reading time was similar when hen followed a masculine gendered noun or a feminine gendered noun. Vergoossen et al. (2020) conclude, thus, that as a gender-inclusive pronoun, hen is not difficult to process. The influence of two participant characteristics on the processing of hen were considered, but neither self-reported previous experience with hen nor personal attitudes modulated the processing of hen versus the gendered pronouns.

In a study conducted in English and Swedish, Lindqvist et al. (2019) compared how paired pronouns (e.g., he/she), traditional neutral pronouns (e.g., singular they) and new gender-neutral pronouns (e.g., ze in English and hen in Swedish) influenced biases in recruitment situations. The results from both languages displayed that paired forms and new gender-neutral pronouns eliminated male bias, while traditional gender-neutral words like singular they evoked male biases. The authors conclude that innovative gender-neutral pronouns are optimal for gender inclusivity. In a follow-up study, Renström et al. (2024) tested the extent to which different gender-fair strategies could evoke associations with normative and non-normative gender expressions using the same paradigm from their previous study. In Swedish, paired pronouns evoked gender normative biases, while hen did not. Conversely, the English data showed that both ze and they elicited lower gender normative biases than paired pronouns. The gender normative bias for ze was even lower for participants who were familiar with this new pronoun. In cases where the use of ze and they were explicitly stated as non-binary, their normative bias was eliminated. Therefore, gender-neutral pronouns elicit non-normative gender associations when they are new words that are well known by language users.

In Spanish, studies have examined the processing of GIMs, but not all of them have taken biases into consideration. Zarwanitzer (2019) studied how –x and –e were processed in Spanish nouns, in contrast to nouns with –o (e.g., amigos, amigxs, amigues, ‘friends’) and how the processing of these three morphemes differed amongst young adults and older adults (37–71 years old). By means of a self-paced reading task, participants read a carrier sentence followed by an experimental sentence that contained a noun ending in one of the three morphemes (e.g., Hoy cumple años María. Invitó a sus amigos/amigxs/amigues a tomar el té ‘Today is María’s birthday. She invited her friends over for tea’). The author found significantly longer total reading time for experimental sentences that included nouns with –x and –e, in comparison to those that included nouns with –o, which had the shortest reading time. This result was consistent across both age groups, leading Zarwanitzer to conclude that reading nouns with GIMs –x and –e requires greater levels of language processing than the masculine generic –o. Stetie and Zunino’s (2022) study tested the generic mental representations of the morphemes –o, –x, and –e in stereotyped role names in Spanish. Participants completed a comprehension task in which they read a sentence that included either high or low male stereotypical subject nouns and, then, answered a multiple-choice question about the reference of the noun phrase in the subject position. The researchers measured total reading time of the sentence and total response time to the question. They also examined participants’ responses to the multiple-choice questions. The authors found that the use of the masculine –o evoked more male than mixed group representations, an effect that was moderated by the stereotypicality of the role names. In other words, nouns with high male stereotypically (e.g., carpinteros ‘carpenters’) were significantly more associated with groups of males than with mixed-gender groups. In contrast, GIMs –x and –e elicited more references to mixed groups. They found no significant reading time differences between sentences with generic –o and those with GIMs –x and –e. Regarding response time, participants took significantly longer to respond to questions when the preceding sentence included –o versus when the preceding sentence included –x and –e, and there were no significant response time differences between the two GIMs. These findings evince that the so-called generic –o morpheme is biased towards male representations in contrast to GIMs, which illustrate mixed groups of men, women, and gender diverse individuals. Moreover, unlike Zarwanitzer (2019), Stetie and Zunino (2022) did not find evidence for a higher processing cost for GIMs –x and –e, when compared to the generic –o morpheme. The present study examines the processing cost of the GIM –x in Spanish pronouns by means of an eye-tracking technique. To our knowledge, it constitutes the first study to use eye tracking to examine gender-inclusive language in Spanish.

The role played by gender stereotypes in language processing, considered by Stetie and Zunino (2022), has been further examined in several eye-tracking and self-paced reading studies. For example, Khan and Daneman (2011) used eye tracking to examine the effect of the –man suffix on excluding female referents (e.g., congressman). Participants read a five-sentence paragraph that included a neutral context in the first three sentences, and a gender stereotyped noun in the fourth sentence that either ended in –man (e.g., spokesman) or –person (e.g., spokesperson) and was followed by a reflexive pronoun (e.g., himself or herself). After the reading task, participants answered a questionnaire that measured the gender stereotypes of the nouns included in the study. Results showed a strong male association for the –man suffix. In other words, participants displayed longer fixation durations when a word with the –man suffix was followed by the reflexive pronoun herself. Similarly, in Reali et al.’s (2015) study, after participants completed the eye-tracking task, they answered questions related to sexism and the roles of men and women in society. The results showed a correlation between participants’ eye movements and their gender stereotype beliefs. The more stereotypically masculine a noun was, the shorter the fixation durations were on the masculine pronoun. In Messer and Kennison’s (2018) study, participants answered the Bem Sex Role Inventory (Bem, 1974), a questionnaire that considers attitudes towards masculinity and femininity. Male participants who obtained higher femininity scores had shorter fixation durations when reading the pronoun he followed by a female metaphor. Coming back to Spanish, Stetie and Zunino (2024) used a self-paced reading task to evaluate the influence of role names with stereotypical male or female bias on the processing of Det-N phrases with the GIM –e. They considered three interest zones—the noun phrase with the GIM –e, the three-word spillover region, and the wrap-up region (final word of the sentence)—, as well as total reading time of the sentence. The authors found evidence of a semantic bias effect only on the first spillover word, in which the male-biased condition was processed faster than the female-biased condition. They do not consider the processing difficulty to be linked to the GIM –e, but rather to the lower frequency of occurrence of female-biased role names with GIMs (p. 461). In sum, these studies suggest that gender stereotypes are activated during reading and the conventional notions of gender play a role during language processing. This is taken into account in the present study.

Many of the studies reviewed in this Introduction have recognized the importance of considering the influence that extralinguistic factors may have on the use and the processing of gender-inclusive language forms. Overall, these studies have considered sociodemographic characteristics, such as sex, age, educational level, region of origin, gender identity, sexual orientation, and political affiliation, as well as traits relating to the experience that individuals may have with gender-inclusive language (i.e., interest in gender studies). In some, but not all occasions, these extralinguistic factors have influenced not only the attitudes that speakers display towards gender-inclusive language, but also in their language use, particularly their comprehension of these gender-neutral forms. Furthermore, these extralinguistic factors are inevitably sometimes correlated. For example, Michnowicz et al. (2023) removed the region of origin from the data analyses in their study due to excessive collinearity with other factors. Pesce and Etchezahar (2019) found a relationship between gender, attitudes towards gender-inclusive language, and rate of gender-inclusive use, by which women had higher positive evaluations and also reported using gender-inclusive language more frequently than men. Therefore, these extralinguistic factors must be examined carefully. The novelty of these gender-inclusive forms coupled with the varied findings of previous studies encourage the need to continue to examine the effect of extralinguistic factors, as we do in the present study.

As mentioned above, by means of eye tracking, this study examines the processing of Spanish pronouns with the GIM –x. The main research question is: How does the linguistic processing of –x differ (or not) from that of the traditional grammatical gender morphemes –o and –a? On the one hand, we hypothesized that, based on the grammatical innovation or novelty of GIMs and on certain previous studies (Zarwanitzer, 2019), the linguistic processing of –x could be more costly than the linguistic processing required for –o and –a, thus reading pronouns ending with –x would result in longer fixation durations. However, on the other hand, based on results found in other sentence processing studies (Stetie & Zunino, 2024; Vergoossen et al., 2020), it could be that pronouns with –x would not result in any increased processing costs. We are also interested in considering the effect that extralinguistic factors, such as gender identity, sexual orientation, experience with gender studies, and attitudes towards GIMs, may have on the linguistic processing of pronouns with –x. We predicted that certain tendencies would arise with respect to the linguistic processing of –x and the extralinguistic factors. Specifically, we expected participants who identify as non-binary (gender identity) and LGBTQIA+ (sexual orientation), who have experience with gender studies and positive attitudes towards GIMs, to tend to process pronouns with –x more easily than other participants. Although there are varied findings, this prediction is in line with previous studies according to which extralinguistic factors play a role in participants’ perceptions and use of gender-inclusive language (Gustafsson Sendén et al., 2015; Michnowicz et al., 2023; Popič & Gorjanc, 2018, but cf. Stetie & Zunino, 2022; Vergoossen et al., 2020). Previous studies have obtained collinearity or overlap between the extralinguistic factors (e.g., Michnowicz et al., 2023; Pesce & Etchezahar, 2019) and, thus, we expected that some of the factors in our study, such as experience with gender studies and attitudes towards GIMs, would be correlated.

Due to the mixed findings reported in the literature, we hypothesized that processing pronouns with –x would be either less costly than processing pronouns with –o and –a or costlier than processing pronouns with –o and–a. The latter prediction is mainly based on –x’s novelty as a GIM, compared to –o and –a, which form part of the speaker’s lexicon from an early age. Below, we present the results of our early and late processing measures (see Tables 4 and 5), followed by the accuracy data (Table 6). The accuracy data is presented last, as it serves as a way to measure general sentence comprehension.

The main goal of this study was to examine how the linguistic processing of GIM –x differed or not from the linguistic processing of traditional morphemes –o and –a. Based on previous studies with inconsistent results, we predicted that the linguistic processing of –x would result in either longer or shorter reading time, when compared to –o and –a. The results extracted from the reading measures indicate that –x was more costly to process than –o at both early and late processing stages, but that, when compared to –a, –x resulted in longer reading time only for gaze duration and regression path time. However, in terms of the overall comprehension of the sentence, no differences in accuracy were found across all three conditions. Taken together, the results confirm our prediction of longer reading time in comparison to –o and partially confirm our predictions in comparison to –a. Regarding accuracy, we did not have a specific prediction. Below, we first interpret the –o comparison results, followed by the –a comparison results and the accuracy results.

In our study, 70% of our participants listed –x as one of their preferred GIMs. One possible explanation for –x’s effect on both early and late processing could be that, although participants were aware of how the morpheme was used, its use was still novel in comparison to –o, especially considering that these participants were raised using –o as the generic plural morpheme in both formal schooling and everyday informal conversations. They were exposed to the morpheme –x later in life, when they got to college (see Román Irizarry, 2021, for data regarding this sample’s exposure to GIMs). When comparing the difference between –o and –x with previous studies, our findings align with Zarwanitzer (2019), but not Stetie and Zunino (2022, 2024) or with Vergoossen et al. (2020). There could be several reasons for the lack of alignment with the latter studies. First, 75% of the sample from Stetie and Zunino’s (2022) study reported using gender-inclusive forms from occasionally to frequently, while all the participants from Vergoossen et al.’s (2020) study were familiar with gender-inclusive hen. In contrast, we only inquired about which GIMs participants used, but not how frequently they did so. This could also explain why our results align with Zarwanitzer’s (2019), since that study did not report participants' use of, exposure to, or familiarity with gender-inclusive forms. Although it is not clear what Vergoossen et al.’s familiarity with hen means in terms of exposure and frequency of use, the higher sample size in that study (n = 130) leads us to speculate that familiarity with hen was related in some way to degree of use.

One key difference between our study and the previous work on GIMs in Spanish is that we examined the processing of –x in Spanish pronouns with antecedents that constituted grammatically gender-neutral nouns with a neutral bias. The observed differences between Stetie and Zunino’s studies and ours could also be due to the nature of the word where –x was used. While they examined –x in nouns, which belong to linguistically open classes, we examined –x in pronouns, which belong to linguistically closed classes that rarely admit new members (Paterson, 2014). It could have been the case that processing –x in pronouns was more costly because unlike nouns, pronouns engage in a resolution process in which readers must link the pronoun with its antecedent. Although this might be the case, this explanation does not account for why our –x/–o results aligned with Zarwanitzer’s (2019). Establishing comparisons between previous processing studies and ours is also complicated due to the different sentence regions for which reading measures were extracted: while some researchers examine reading time of the entire sentences and others consider certain sentence chunks, we measured reading time on the pronouns. Future studies should explicitly compare how processing –x differs in nouns versus pronouns.

The nuanced findings regarding the –x/–a comparisons represent new findings for the literature on processing GIMs in Spanish. The lack of a difference between –x and –a in first fixation time hints at this nuanced relationship, since first fixation time represents the earliest point in which effects can be observed due to properties of a word, such as frequency (Liversedge et al., 1998). In this case both –x and –a are less frequent plural forms that serve as the generic, in contrast to the masculine –o, which is the default generic plural in Spanish. With respect to gaze duration, a measurement linked to the sensitivity of semantic and syntactic anomalies (Rayner et al., 2004), we interpret the longer gaze duration time of –x, when compared to –a, as evidence that, although most of our sample included –x as one of their preferred GIMs, readers have not yet fully integrated –x into their linguistic systems. This notion is further supported by the longer regression path time for –x in contrast to –a, which reflect lexical integration difficulties (Rayner et al., 2004). However, once readers move into total time, the –a morpheme results in the longest reading time. We believe this effect suggests two things. First, although –x was associated with early and late processing costs with gaze duration and regression path time, its processing in later stages is not insurmountable. The shorter total reading time, when compared to –a, suggest that, once readers engage in more conscious processes, they are able to overcome the temporary processing costs associated with –x. This interpretation makes sense particularly for speakers who use –x as a GIM. Second, the longer reading time for –a cannot be explained by an argument of novelty, since –a is part of the grammatical gender system acquired by Spanish speakers. Nevertheless, the use of –a in plural contexts is considered marked by Spanish grammarians (Roca, 2005, 2006). In other words, it can only be used to refer to a group of females. When processing the two traditional grammatical genders in Spanish, an asymmetry that favors the processing of the masculine grammatical gender over the feminine grammatical gender has been found (see Beatty-Martínez & Dussias, 2019, for a review on this topic). It is possible that this reported processing asymmetry emerged in our participants’ later processing stages. Although GIM –x also showed a gender asymmetry when compared to –o in all processing stages, the fact that it did not show a clear asymmetry pattern when compared to –a may be suggestive of its initial stage of integration into the grammatical gender system of this participant sample. Further support for the beginning of this integration process is provided by our accuracy findings, which showed no differences between morphemes –o, –a, and –x in predicting accurate comprehension. To our knowledge, this is a new finding since previous studies have not examined whether morphemes –o, –a, and –x differ in their ability to predict accuracy. This finding suggests that when participants are recalling information to answer comprehension questions, the use of –x in pronouns does not pose an additional cost. In other words, any processing cost of –x that appears when reading sentences (online) no longer seems to influence participants’ behavior (offline) when they answer comprehension questions. This finding can be associated with the lack of a difference found by Stetie and Zunino (2022) in total response time to the question.

The findings have implications for models of sentence processing. The longer reading time on pronouns with GIM –x indicate that readers faced increased cognitive demands when integrating this novel morpheme into sentence structures. These effects are in line with predictive processing models, such as those grounded in surprisal theory (Hale, 2001; Levy, 2008). According to surprisal theory, comprehenders use fine-grained probabilistic knowledge derived from prior linguistic experience to form expectations about the structural interpretation of the input that they have received and that which may be upcoming; these expectations immediately determine processing difficulty and guide interpretation preferences. In our study scenario, processing pronouns with the GIM –x may disrupt participants’ predictions due to its relative novelty and low frequency compared to the canonical grammatical gender forms. This, then, may lead to higher surprisal and processing costs, particularly for readers with certain characteristics (e.g., those who identify with binary gender, heterosexuals, those with mixed attitudes towards GIMs). In a related vein, the longer gaze durations and regression path time suggest that readers engage in reanalysis upon encountering GIM –x. This interpretation aligns with approaches, such as good enough processing (Ferreira et al., 2002, 2009), according to which sentence processing is not always necessarily complete and detailed, but simply good enough for the purpose at hand. Therefore, instead of relying on full syntactic analyses, comprehenders use heuristics to parse sentences; as a result, ambiguities may remain unresolved and underspecified during comprehension, and deep processing will only be carried out if it is required by the task that comprehenders must complete. In the context of sentences including pronouns with GIM –x, reanalysis likely reflects the reader’s need to reconcile a novel grammatical gender marker with the broader syntactic, semantic, and perceived sociopragmatic context.

Additionally, the absence of significant accuracy differences across the three experimental conditions, despite increased processing effort for the GIM –x, highlights an important distinction in certain cognitive load theories: processing effort does not necessarily impair comprehension. This finding contributes to the understanding of cognitive resource allocation in sentence processing. While readers incur greater cognitive effort when processing pronouns with –x, they successfully allocate resources to maintain accuracy in comprehension. This discrepancy between effort and outcome can be understood through the resource allocation model, also known as the capacity theory of attention (Kahneman, 1973), which suggests that attention is a limited cognitive resource that can be allocated to more than one task, but the capability to do so depends on the extent to which the tasks require attention. Therefore, readers dynamically adjust their focus and resource distribution depending on the goals at hand. When encountering new, unexpected, and potentially unfamiliar forms like –x, readers may engage in compensatory strategies, such as slower reading or reanalysis, without compromising overall understanding.

A secondary goal of this study was to explore the relationship between extralinguistic factors and the linguistic processing of GIM –x. These have proven to play a role in use of gender-inclusive language in previous studies (e.g., gender identity and interest in gender issues in Gustafsson Sendén et al., 2015; sexual orientation in Popič & Gorjanc, 2018). Despite not being able to properly estimate these effects, we can interpret the emerging trends in the data. We hypothesize that these emerging trends might be the ones responsible for the inverse relationship between the –x/–o and –a/–x contrasts identified by the random slopes of our models. In general, the trend that non-binary and LGBTQIA+ individuals had shorter reading time on –x in comparison to –a across all reading measurements suggests that there might be additional variables not captured by this study that affect speakers’ experience with the GIM –x. One possible explanation is that –x might be more commonly used by non-binary individuals and LGBTQIA+ individuals on a daily basis, and this exposure might influence how they process the GIM –x. There was, after all, a tendency for participants with more experience with gender studies to display shorter reading time on –x; this could also reflect more familiarity with –x through increased exposure. Therefore, future studies should endeavor to accurately capturing an individual’s usage of GIMs in different scenarios (work vs. school vs. peers). It is also important for future studies to continue to consider participants’ attitudes towards gender-inclusive language. In the current study, participants with more positive attitudes or increased acceptance of GIM –x tended towards shorter reading time on –x. Taken together, these emerging trends regarding extralinguistic factors evince that sentence processing is not solely a linguistic phenomenon, but rather that it is deeply intertwined with sociocultural and individual experiences. Consequently, models of sentence processing should integrate as core components the social identities, perceptions, beliefs, and experiences of language users. This has been attempted by frameworks that aim for more flexible parsing, such as the constraint-based lexicalist approach (MacDonald et al., 1994), which proposes that the comprehension process is guided by a series of constraints pertaining to multiple sources of information, including syntactic, semantic, and contextual cues. In our study, the influence of individual differences, such as familiarity with gender studies and socio-cultural attitudes, supports models that emphasize the interplay between linguistic knowledge and extralinguistic factors in real-time processing. On a related, more applied note, it is crucial to develop in future work a validated scale that adequately measures attitudes towards gender-inclusive language. This will enable researchers to better disentangle the complex relationship between language attitudes, gender biases, and gender attitudes, as well as help achieve more consistency among different studies regarding the way attitudes are measured.

The present study contributes empirical data to the body of research focusing on the processing of gender-inclusive language in Spanish. To our knowledge, it is the first to use an eye-tracking technique (most studies have used self-paced reading), allowing us to obtain data that reflects both early and later processing stages. It is also the first to examine the processing of the GIM –x on pronouns (most studies have examined Det-N phrases) and to compare the –x to both traditional morphemes, the masculine –o and the feminine –a (most studies have considered only the generic masculine –o). While our study provides valuable insights into the fine-grained processing of GIM –x, we must acknowledge the following limitations. First, although our sample size is close to the median reported in a recent methodological review of eye-tracking studies (King et al., 2019), it is likely underpowered, as pilot testing to obtain the necessary parameters for a mixed-effects power analysis (Brysbaert & Stevens, 2018) was not feasible in our context. Future studies should aim to recruit a larger sample size to test the trends we identified with our extralinguistic factors. Since this may be challenging, particularly regarding gender identity, as documented in previous studies, we encourage researchers to collaborate to recruit more individuals from underrepresented groups. In conclusion, despite these limitations, our study highlights the importance of examining the processing of GIM –x in contrast to both canonical morphemes. A comprehensive understanding of how Spanish speakers integrate GIMs into their language cannot be put forth unless both traditional grammatical genders are considered. In this case, having analyzed –x not only in relation to –o, but also with respect to –a, provided evidence of what may be considered initial steps towards people’s incorporation of gender-inclusive forms into their linguistic system. Future work in this area should continue to examine how different GIMs are processed alongside traditional forms and explore how extralinguistic factors may modulate their processing.