“Fallo garrafal y común entre vascoparlantes:” attitudes and salience of grammatical gender agreement in Basque-Spanish contact

2.1 Language attitudes, indexicality and (social) salience

Language attitudes are usually considered to have three components: (i) cognitive components or listeners’ beliefs regarding linguistic features, (ii) affective components or listeners’ feelings regarding linguistic features, and (iii) behavioral components or the manner in which listeners act regarding linguistic features (Díaz-Campos 2014; Garret 2010). Adopting the principles of indexicality, the present paper centers its attention on these three aspects to uncover the social meanings related to the production of N-SGA in BAC Spanish.

Members of a linguistic community associate specific social meanings to particular linguistic features, which is known as indexicality (Silverstein 2003). Indexicality is defined as “a property of speech through which cultural contexts such as social identities (e.g., gender) and social activities (e.g., a gossip session) are constituted by particular stances and acts” (Ochs 1992: 335). Callesano (2023: 3) explains indexicality as the manner in which “linguistic variants point to certain social traits, characteristics, and qualities, such as how variable realizations of syllable-final /s/ in Spanish may index–or perceptually indicate–various levels of social status.”

On the one hand, in direct indexicality, linguistic features contain intrinsic social meanings that are not based on social constructions per se, which is the case of very few words and expressions (Drummond and Schleef 2016; Jaffe 2016; Ochs 1992). One example of direct indexicality might be the use of gendered pronouns and honorifics, although they may also reveal speakers’ ideologies (Jaffe 2016). On the other hand, indirect indexicality refers to when linguistic features convey social meanings created by society (Drummond and Schleef 2016; Ochs 1992). However, these indirect indexicalities are not arbitrary, but rather strongly linked to ideologies (Bucholtz 2009). Therefore, one difference between direct and indirect indexicality is the origin of their social meanings: direct indexicalities arise naturally, while indirect indexicalities relate to the transmission of social meanings that lack a natural origin (Drummond and Schleef 2016; Ochs 1992). Linguistic features may also have more than one social meaning, with each of them activated depending on the context (Eckert 2008). Inasmuch as indexicalities are constructed by speakers, they are very likely to evolve over time (Eckert 2008). Additionally, the meanings are not absolute, but rather part of a continuum: each word or part of speech can convey more than one meaning depending on the context in which they are employed (Eckert 2008).

In terms of N-SGA in BAC Spanish, the social meanings that are indexed by its use can be categorized as a form of indirect indexicality. Nevertheless, its indexical field would not be arbitrary, since there may be a link to production patterns. Indeed, N-SGA production in BAC Spanish has been traditionally linked to the speech of elderly native speakers of Basque with limited access to education (Fernández Ulloa 1997). Therefore, it may be the case that the indexical field of N-SGA in BAC Spanish is connected to its production.

Despite the fact that linguistic attitudes and social meanings can be linked to language production (Díaz-Campos 2014), the level of consciousness with which language is produced is still an area that requires further investigation (Lass 1997; Thomason 2001), probably due to the difficulty related to delimiting the level of awareness. Linguistic awareness is related to the concept of salience, which is considered in this exploration of N-SGA. While varying (and sometimes contradictory) definitions of salience can be found (Boswijk and Coler 2020), one general definition (for example, used in phonology) is that salient linguistic features are those that are more noticeable or prominent to speakers (Rácz 2013), with Siegel (2010: 129) defining salience as “the characteristic of being easily noticeable, prominent or conspicuous.” A linguistic feature can be salient because it was cognitively preactivated (e.g., mentioned recently by someone), or it can be salient because “the stimulus stands out because it is incongruous with a given ground” (Blumental-Dramé et al. 2017: 1).

The concept of social salience is also relevant to the exploration of language variation and change. This concept can be defined as “the relative availability of a form to evoke social meaning” (Levon and Fox 2014: 185). One method with which to explore (social) salience is through the use of experimental approaches: under this methodology, “a linguistic variable is defined as salient to the extent that the presence of that variable in the speech signal contributes to accurate performance by listeners in a perceptual experiment” (Macleod 2015: 84). Therefore, following the aforementioned definitions for (social) salience, the present paper understands the concept of salience as the extent to which speakers of a variety of a language are aware of the presence of a particular linguistic characteristic (Hickey 2000).

2.2 Basque and Spanish in the Basque country

The Basque Country is situated in northern Spain and southern France and is comprised of the BAC (subdivided into the provinces of Bizkaia, Gipuzkoa, and Araba) and Nafarroa in the Spanish territory, and Iparraldea (subdivided into Lapurdi, Nafarroa Beherea, and Zuberoa) in the French territory. The Basque Country can also be divided into the aforementioned three spaces based on the languages that are spoken and their corresponding linguistic policies; following the prohibition of all languages other than Spanish during Francisco Franco’s dictatorship (1939–1975), Basque and Spanish are now recognized as co-official languages in the BAC and in (the northern area of) Nafarroa, as established by the 1978 Spanish Constitution, while Basque is not an official language in France.

This linguistic landscape has a direct effect on the number of speakers of the minoritized language, in this case, Basque. This means that, for instance, the highest number of Basque speakers can be found in the BAC, and for this reason, this project is centered on this particular area: from the 2.1 million inhabitants that comprise it, 41 % consider themselves native Basque speakers, 15 % understand Basque but cannot speak it, and 44 % can neither understand nor speak it (Eusko Jaurlaritza 2020). In terms of language use on the street, the use of Basque in the BAC has increased by 1.8 points from 1989 to 2021 (Altuna Zumeta et al. 2022). With the number of bilinguals increasing in the BAC (Gómez Seibane 2020; Rodríguez-Ordóñez 2021a, 2021b) due to several linguistic policies (e.g., the Act of Normalization of the Basque Language, Law 10/1982, which installed Basque in the educational system), it is believed that the number of monolingual Spanish speakers is decreasing,^[4] mainly among younger speakers (Leivada et al. 2023). In addition, there has been a decrease in monolingual Basque speakers, which are virtually absent in contemporary times.

The use of language in the Basque Country as a whole is still closely related to identity as the use of terms such as euskaldun zaharra and euskaldun berria illustrates. As Gondra (2024: 2) describes, “Basque people call themselves euskaldun, which comes from euskera ‘Basque’ plus the possessive suffix -dun/-tun, meaning ‘person who has (knowledge of) the Basque language.’” The former is the term typically used to refer to Basque native speakers or speakers of local varieties of Basque (not the unified [Batua] variety). The inverse of this term, euskaldun berria is used to refer to the population whose first language is not Basque. These speakers learn Basque at a later age and generally do not learn it through family transmission (Ortega et al. 2015; Rodríguez-Ordóñez 2021a; Urrutia Cárdenas 2006).

It is widely known that Spanish and Basque are two different languages: the former is a Romance language that belongs to the Indo-European family, while the latter’s origin is still unknown. However, they are not isolated from one another. In the BAC and Nafarroa, Basque and Spanish have been in constant contact since the romanization of the Iberian Peninsula and, particularly, after the Middle Ages, with bidirectional linguistic effects (Camus and Seibane 2018; Gómez Seibane 2020; Reguero Ugarte 2024). For instance, Latin and Romance elements such as loanwords are present in present-day Basque (Camus Bergareche 2011). Regarding Spanish, this contact situation has given rise to a particular variety of Spanish that is spoken in these territories: BAC Spanish. This variety of Spanish has been categorized as ciertamente compleja [certainly complex] by previous scholars (Camus 2013), as it is believed to be formed by a continuum of linguistic modalities that are conditioned by diverse geographic, historic, and sociolinguistic factors (Camus and Gómez Seibane 2018).

2.3 Gender in Spanish

Spanish is one of the Romance languages that include GA, which is defined as “a permanent characteristic of nouns”^[5] (Carroll 1989). The use of GA serves to classify nouns into at least two groups (Kramer 2020), as is precisely the case in Spanish: feminine and masculine^[6] (Real Academia Española, RAE 2016). This distinction between masculine and feminine is similar to that found in other Romance languages where it may be based on social gender identity: many feminine nouns refer to females (e.g., madre [mother]), whereas many masculine nouns refer to males (e.g., padre [father]) (Corbett 2013b; Kramer 2020). Nevertheless, both genders include nouns whose gender is not based on social gender identity (Corbett 2013b), but instead is arbitrarily assigned (Kramer 2020). Masculine is categorized as the default or unmarked gender in Spanish and is used to refer to groups that include females and males (Beatty-Martinez and Dussias 2019; Delgado 2018; Goebel-Mahrle and Shin 2020; Harris 1991; Roca 1989, Roca 2005).

GA in Spanish is syntactically manifested in the noun phrase by establishing agreement between the controller^[7] and other elements such as determiners and adjectives (Balam et al. 2021; Boers et al. 2020; Clegg 2011; Cuza and Pérez-Tattam 2016; Horáková et al. 2014; Montrul and Potowski 2007; Poplack et al. 1982). According to standard Spanish grammar rules, the canonical inflectional form for feminine nouns is -a, which accounts for 96 % of the cases (Teschner and Russell 1984). For the masculine nouns, the canonical inflectional form is -o, which accounts for 99.9 % of nouns with this gender (Teschner and Russell 1984). Although there is no direct relation between nominal suffix and gender assignment in Spanish (Eddington 2002), the canonical inflectional forms are usually reliable gender predictors (Gonzalez et al. 2019; Green 1988; Parafita Couto et al. 2015; Teschner and Russell 1984). The feminine canonical ending and SGA is illustrated in (1), and the canonical ending for masculine and SGA is illustrated in (2).

This broad description of the Spanish GA system is based on standard norms, which do not account for the possibility of producing N-SGA in language contact situations or the gender-inclusive forms such as -x, -e, or -@ as in niñxs, niñes, or niñ@s (Michnowicz et al. 2023; Papadopoulos 2022; Rodríguez Iglesias 2018; Velázquez-Lora 2022).

2.4 Gender in Basque

Different from languages such as Spanish, Basque does not include grammatical gender^[9] (Euskaltzaindia 1991, 2002), although there are words in which the social gender identity can be expressed as in ama [mom] and aita [father] (Gómez Seibane 2008). It is also notable that while Basque does not include grammatical gender in general terms, it does include it in the verbal domain of varieties of Basque that use the familiar second-person singular pronoun hi:^[10] verb agreement morphemes differ based on the gender of the addressee, using -k for male speakers as in duk [have], and -n for female speakers as in dun [have] (Alberdi 1995; Euskaltzaindia 1991, 2002; Gómez Seibane 2008; Hualde 2003; Laka 1996; Pérez-Tattam et al. 2019). Moreover, although categorized as a low-usage phenomenon that is primarily observed in spoken language, scholars have acknowledged that some western Basque varieties exhibit a Romance-style gender marking on some nouns by using -o as in gixajo [poor fellow (masc.)] and -a as in gixaja [poor fellow (fem.)] (Euskaltzaindia 1991, 2002; Gómez Seibane 2008; Laka 1996; Parafita Couto et al. 2015; Trask 2003; Zubiri and Zubiri 2012). There is, however, a need for empirical studies that systematically address this gender production in Basque to delimit its presence and the factors that predict its production in the current spoken/written varieties.

2.5 Non-standard gender agreement in BAC Spanish

N-SGA in BAC Spanish is not a recent phenomenon, with historical records having attested its presence, for instance, in written texts from the 15th and 16th centuries in Bizkaia (Etxague Burgos 2012; Gómez Seibane 2008), or in 18th century Spanish texts in Nafarroa (Etxague Burgos 2012). N-SGA production is not exclusive to BAC Spanish. In fact, it is a common feature in language contact situations, particularly if one of the languages in contact does not have grammatical gender (e.g., Alarcón 2011; Avelino Sierra 2021; Balam et al. 2021). Although the research on N-SGA in BAC Spanish is limited, numerous publications have examined its presence in other language contact situations (e.g., Albirini et al. 2013; Delgado 2018). Indeed, they have identified certain social and linguistic factors that favor this type of GA production in languages in contact (age, language of schooling, level of education, bilingual profile, origin, controller’s gender and animacy, inflectional form and number, frequency or familiarity with the words, and syntactic distance) (e.g., Balam et al. 2021; Goebel-Mahrle and Shin 2020; Husein 2021; Shin et al. 2019; Torres Sánchez 2021; Trawick and Bero 2021).

Although N-SGA production is a characteristic of BAC Spanish (e.g., Etxebarria Arostegui 2007, 2008; Fernández Ulloa 1997, 2006; Urrutia Cárdenas 2006), there are few studies that have empirically explored its use from a variationist perspective in non-codeswitching BAC Spanish speech (e.g., Basterretxea Santiso 2022, 2024).^[11] Fernández Ulloa (1997) noted that N-SGA in BAC Spanish is commonly associated with elderly Basque native speakers or euskaldun zaharrak with limited access to education. Echaide (1968) argued that N-SGA emerged in BAC Spanish due to Basque not producing grammatical gender, categorizing it as a common phenomenon among speakers with Basque as their primary language and who were over 30 years old. Nevertheless, the association between N-SGA and older, first-language speakers of Basque with limited education has not been explored until recently by empirical studies.

Basterretxea Santiso (2022) challenges the previous common association by suggesting that the production of N-SGA may also be common among young Basque native speakers who have had access to (higher) education. In Basterretxea Santiso (2022), 12 participants engaged in a 45-minute-long spontaneous conversation in Spanish and completed a grammatical choice test. The results showed that N-SGA is also present in the speech of young-adult native Basque speakers (age range: 21–29) who have received formal education (including Basque and Spanish grammar and literature classes). Nevertheless, even though N-SGA was present among these participants, it was infrequent, presumably due to their high level of command of the standard Spanish GA system. Similarly, Basterretxea Santiso (2024) explored the production of N-SGA in BAC Spanish with a diverse population formed by 73 participants to analyze the linguistic factors that favor this production in language contact situations. The study confirmed that N-SGA production was present among all participants (belonging to different generations), and not only among elder Basque native speakers with limited access to education.

Basterretxea Santiso (2022) also studied perceptions towards N-SGA in language contact situations through the use of a written metalinguistic questionnaire. When no specific reference was made to N-SGA in the questionnaire, 10 participants emphasized the importance of “accent” and 6 of them also mentioned the “difficulty” of producing GA. When specifically asked about N-SGA, 10 participants agreed that N-SGA production in their Spanish variety was widespread, and 11 of them shared that they probably produce it in their own speech. Therefore, although the size of the population sample in Basterretxea Santiso (2022) was limited, the study suggests that BAC Spanish speakers are aware of N-SGA production. At the same time, the study did not provide any information regarding the social meanings that may be indexed by its use.

4.1 Participants

A total of 76 participants were recruited from 2022 to 2023 in the three provinces that constitute the BAC: Bizkaia, Gipuzkoa, and Araba. With the aim of obtaining a homogeneous sample, speakers who reside in other areas in which Basque is also spoken (Nafarroa and Iparraldea) were excluded: in Nafarroa, Basque and Spanish are in contact to a lesser extent (depending on the area), and Basque is in contact with French in Iparraldea (Lapurdi, Nafarroa Beherea, and Zuberoa).

The initial group of 76 participants is comprised of speakers from various locations in the BAC, with diverse age, origin, and linguistic and educational backgrounds. These participants were recruited by reaching out to the researcher’s personal contacts, by contacting several associations such as Euskaltegiak (centers for adults learning Basque), and through the snowball sampling method (Milroy and Milroy 1985). Therefore, a randomly filled judgment sample was created with the aim of having diverse participants (Hoffman 2014; Schilling 2013). The final study sample includes 73 participants, as 3 were excluded due to not completing all the study tasks or, similar to Paquet (2018), because they were educated in French instead of Basque or Spanish. In order to conduct statistical analyses following the cross-tabulation guidelines outlined by Meyerhoff et al. (2015), and Silva-Corvalán and Enrique-Arias (2017), the final sample of participants was divided in terms of their first language(s) and age, as is shown in Table 1.

Apart from their first language(s) and age, participants were also stratified in terms of six extralinguistic variables that will be considered for the analysis: Age, Language of Schooling, Highest level of education, Bilingual profile, Origin, and Province of residence. This information was extracted from the background questionnaire that participants completed at the end of their participation and is presented in Table 2.

4.2 Social independent variables

In order to predict the social meanings and attitudes related to N-SGA production in BAC Spanish, the impact of six social variables was explored. A summary of the independent social variables is presented in Table 3. Subsection 4.2.1Sub–Sub4.2.6 provide more details regarding each independent variable.

4.3 Data collection process

4.4 Analyses

A factor analysis was conducted with the results of the modified matched-guise task, following the guidelines outlined by Bandalos and Boehm-Kaufman (2008) to interpret an exploratory factor analysis (e.g., minimum sample size, eigenvalues, etc.). The objective of conducting a factor analysis was to test for potential interdependence among the factors included in the matched-guise task (Loewen and Gonulal 2015). After the factor analysis, mixed-effects linear regressions that included participant as a random effect were conducted to explore the impact of the social independent variables included in the analysis: participants’ Age, Language of schooling, Level of education, Bilingual profile, Origin in terms of rurality, and Province of residence. Interactions between the independent variables and the dependent variable (N-SGA vs SGA) were analyzed with R’s (R Core Team 2022) analysis of variance function. Finally, a thematic analysis was conducted by two raters to analyze the responses to the last two open-ended explicit questions. The inter-rater reliability assessment revealed a 100 % agreement between the two raters for the 47 sentences (out of 73) that they both coded.

5.1 Exaggerated imitation activity

In the 57 exaggerated imitation recordings, only 6 examples of N-SGA were identified, each produced by a different participant. The 6 instances of N-SGA are presented in examples (3) to (8).

Besides N-SGA, all but two participants completing this activity produced at least 3 instances of SGA. Therefore, the results from this experiment suggest that N-SGA is not one of the first characteristics that people think of when asked to exaggeratedly imitate BAC Spanish. However, it is interesting that they only used the masculine -o: this would not be simply a matter of lack of GA, but likely an indicator of their preference for masculine gender (or lack of gender). Other characteristics typically associated with BAC Spanish (e.g., Ciriza 2009; Fernández Ulloa 1997; Klee and Lynch 2009; Oñederra et al. 2014) were more prominent in these recordings: stereotyped Basque tone^[14] (24 participants), use of Basque words such as aitite [grandfather], or baserri [farm] (10 participants), SOV order as in(nosotros) muebles modernos tenemos [latest furniture (we) have] (8 participants), use of swear words (6 participants), or the doubling of adjectives to intensify the meaning as in pasillo largo largo tiene [it has a long long hallway] (5 participants).

5.2 Implicit attitudes

Since it was likely that some of the six adjectives of the semantic differential scale^[15] included in the matched-guise task were related, a factor analysis was conducted. The results suggested that there were two main groups of factors: the factor labeled as Euskaldun (it loaded for the following adjectives: native language, euskaldun zaharra or euskaldun berria, and place in terms of rurality^[16]) and the factor labeled SES (it loaded for the adjectives related to education and resources). There was an extra scale (Agreeableness) that did not load onto any factor, so it was treated as an independent third factor in the analysis. Therefore, in the analyses that follow, the attitudes of N-SGA in BAC Spanish are divided between Euskaldun, SES, and Agreeableness, based on the factor analysis.

Figure 1 presents box plots for the three factors that resulted from the factor analysis by type of GA (N-SGA vs. SGA).

Figure 1:

Mean scores for gender agreement type.

The mean scores of the descriptive statistics in Figure 1 suggest that, in terms of Euskaldun factor, N-SGA in BAC Spanish is more likely to be associated with a Basque native speaker (Euskaldun zaharra) than SGA (N-SGA mean = 3.15, SD = 1.53, N = 73; SGA mean = 3.94, SD = 1.54, N = 73). With regards to the SES factor, mean scores in Figure 1 indicate that N-SGA is associated with lower levels of education and resources when compared to SGA (N-SGA mean = 3.6, SD = 1.12, N = 73; SGA mean = 4.06, SD = 1, N = 73). Finally, the contrast of the perception of the Agreeableness factor appears to be minimum: neither of the guises would be considered as non-agreeable (N-SGA mean = 3.99, SD = 1, N = 73; SGA mean = 4.02, SD = 0.94, N = 73).

The two combined factors and the independent factor (Euskaldun, SES, Agreeableness) were further analyzed through mixed-effects linear regressions that included random effects for participants. This allows us to consider the impact of the sociolinguistic predictors (participants’ Age, Language of schooling, Level of education, Bilingual profile, Origin, and Province of residence) upon the dependent variable (responses to the semantic differential scale divided between SGA vs. N-SGA). When conducting linear regressions, only significant factors (p = 0.05 that was calculated through t values) were kept for the best-fit model.

5.2.1 Euskaldun factor

The best-fit model for the Euskaldun factor (the combination of native language, euskaldun zaharra vs euskaldun berria, and place in terms of rurality) shows that there is a main effect of the interaction between GA and Age, GA and Level of education, GA and Bilingual profile, GA and Province of residence, and GA and Language of schooling. Results from the linear regression are presented in Table 5, and they suggest a small effect size (R² = 0.189), as per Ferguson's (2009) standards.

Table 5:

Summary of final mixed-effects linear regression model for Euskaldun.

	Estimate	SE	t-Value	p-Value
Intercept	−4.046e+00	4.078e−01	9.966	6.69e−16
GA = N-SGA	−9.129e−01	3.066e−01	−2.978	0.002
GA = N-SGA + participant age = retired	−5.856e−01	1.842e−01	−3.180	0.001
GA = N-SGA + participant age = adult	9.047e−01	1.433e−01	6.311	3.13e−10
GA = N-SGA + participant level of education = secondary	5.545e−01	1.758e−01	3.154	0.001
GA = N-SGA + participant level of education = university	8.391e−02	1.412e−01	0.594	0.552
GA = N-SGA + Bilingual profile = L1 Basque & L2 Spanish	2.539e−01	1.263e−01	2.010	0.044
GA = N-SGA + Bilingual profile = L1 Spanish & L2 Basque	−2.586e−01	1.300e−01	−1.989	0.046
GA = N-SGA + language of schooling = Model B	−3.384e−01	1.793e−01	−1.887	0.059
GA = N-SGA + Language of schooling = Model D	−6.454e−01	1.979e−01	−3.261	0.001
Non-Significant Fixed Effects: Origin, province of residence

Figure 2 illustrates the significant interaction between GA and Age for Euskaldun. It shows that young-adults perceive N-SGA as sounding less like a Basque native speaker/more rural when compared to the other age groups. However, young-adults still perceive N-SGA as sounding like a Basque native speaker from rural areas, albeit to a lesser extent than the other two groups. Post-hoc pairwise comparisons indicate a lack of significant differences between the three age groups.

Figure 2:

Significant interaction: GA * age.

Table 5 also shows a significant interaction between GA and Level of education, illustrated in Figure 3: participants with a secondary education degree are less likely than the other groups to perceive N-SGA as resembling Basque native speakers/less rural (although still somewhat similar). Post-hoc pairwise comparisons indicate that there are no significant differences among the three groups, suggesting all of them perceive N-SGA as more Euskaldun than SGA.

Figure 3:

Significant interaction: GA * level of education.

Another significant interaction for Euskaldun, shown in Table 5, was that found between GA and Bilingual profile. This interaction illustrated in Figure 4 suggests that L1 Spanish native speakers perceive SGA as sounding more like a Spanish native speaker when compared to the other two groups. Nevertheless, post-hoc pairwise comparisons reveal no significant differences among the three groups.

Figure 4:

Significant interaction: GA * bilingual profile.

According to Table 5, the last significant interaction for Euskaldun was between GA and Language of schooling. The results illustrated in Figure 5 suggest that N-SGA is consistently associated with being a Basque native speaker, particularly for those who were educated in Basque (Model D), followed by those educated in Basque and Spanish (Model B), and finally, participants educated in Spanish (Model A). In this case, post-hoc pairwise comparisons reveal minimal differences between participants in Model A and Model D (p = 0.05), and between participants in Model B and Model D (p = 0.05).

Figure 5:

Significant interaction: GA & language of schooling.

5.2.2 SES factor

Regarding the SES factor (the combination of education and resources), the best-fit model indicates that GA is a significant predictor. Moreover, significant interactions were found between GA and Age, GA and Level of education, GA and Bilingual profile, and GA and Language of schooling. The results of the linear regression for this model are presented in Table 6, which indicates a moderate effect size (R² = 0.269), as per Ferguson (2009).

Table 6:

Summary of final mixed-effects linear regression model for SES.

	Estimate	SE	t-Value	p-Value
Intercept	4.010e+00	2.751e−01	14.576	<2e−16
GA = N-SGA	−8.565e−01	1.760e−01	−4.865	1.22e−06
GA = N-SGA + participant age = retired	−2.546e−01	1.458e−01	−1.746	0.080
GA = N-SGA + participant age = adult	5.970e−01	1.144e−01	5.219	1.96e−07
GA = N-SGA + participant level of education = secondary	4.112e−01	1.384e−01	2.970	0.003
GA = N-SGA + participant level education = university	2.998e−01	1.103e−01	2.719	0.006
GA = N-SGA + bilingual profile = L1 Basque & L2 Spanish	2.969e−01	1.012e−01	2.936	0.003
GA = N-SGA + bilingual profile = L1 Spanish & L2 Basque	2.495e−01	1.041e−01	2.398	0.016
GA = N-SGA + participant language of schooling = Model B	−3.268e−01	1.424e−01	−2.295	0.021
GA = N-SGA + participant language of schooling = Model D	−4.862e−01	1.580e−01	−3.078	0.002
Non-Significant Fixed Effects: Origin, province

As shown in Table 6, there is a significant interaction between GA and Age, presented in Figure 6: adults and retired individuals perceive N-SGA as lower SES when compared to SGA. In contrast, young adults’ perception is essentially the same for both types of GA productions. Post-hoc pairwise comparison indicated a lack of significant difference among the three age groups, suggesting that all of them perceived N-SGA as associated with lower SES.

Figure 6:

Significant interaction: GA * age.

The results in Table 6 also indicated a main effect of the interaction between GA and participants’ Level of education, illustrated by Figure 7. Mean scores in Figure 7 demonstrate that N-SGA instances consistently carry a lower SES perception than SGA. More specifically, participants with a secondary education perceive N-SGA as indicative of higher SES, when compared to the other groups. However, post-hoc pairwise comparisons indicated no significant differences among the three groups.

Figure 7:

Significant interaction: GA * level of education.

Table 6 revealed another significant interaction for SES factor, presented in Figure 8: the interaction between GA and Bilingual profile. Mean scores illustrated in Figure 8 show that the three groups perceive N-SGA as associated with lower SES. Particularly, native (2L1) bilinguals are the group that is more likely to associate N-SGA with lower SES as compared to the other groups. Nevertheless, post-hoc pairwise comparisons indicated no significant differences in the perceptions of the three bilingual groups.

Figure 8:

Significant interaction: GA * bilingual profile.

The model presented in Table 6 indicated a significant interaction between GA and Language of schooling. Mean scores in Figure 9 suggest that, compared to the participants educated in Basque (Model D), participants educated in Spanish (Model A) and the ones educated in Basque and Spanish (Model B) give both types of GA higher SES rating. Nonetheless, post-hoc pairwise comparison show no significant differences across groups: N-SGA is consistently perceived as lower SES than SGA.

Figure 9:

Significant interaction: GA * language of schooling.

5.2.3 Agreeableness factor

The best-fit model for the Agreeableness factor shows that Province of residence is a significant predictor, and that there is a significant interaction between GA and Age. The results from the linear regression for this model are presented in Table 7, which indicates a moderate effect (R² = 0.209) based on Ferguson (2009).

Table 7:

Summary of final mixed-effects linear regression for agreeableness.

	Estimate	SE	t-Value	p-Value
Intercept	4.5176	0.2345	19.262	<2e−16
Province of residence = Bizkaia	−0.5222	0.2385	−2.190	0.031
Province of residence = Gipuzkoa	−0.5645	0.2188	−2.580	0.012
GA = N-SGA + participant age = retired	0.2486	0.1536	1.618	0.106
GA = N-SGA + participant age = young-adult	0.3544	0.1178	3.007	0.002
Non-Significant Fixed Effects: Language of schooling, Level of education, Bilingual profile, origin

The significant effect reported in Table 7 for Province of residence is illustrated in Figure 10: Araba residents perceive both types of GA production more positively than the other participants, which implies that Araba residents perceived all sentences as more pleasant regardless of GA type. In this case, post-hoc pairwise comparisons revealed a significant difference between Araba and Gipuzkoa residents (p = 0.031).

Figure 10:

Effect of province of residence on agreeableness ratings.

The significant interaction found between GA and Age for the Agreeableness factor is presented in Figure 11. Mean scores indicate that adults consider N-SGA to be less agreeable than SGA, while retired participants exhibit a minimal difference between both types of GA production. However, the youngest participants perceive N-SGA as more agreeable than SGA. Nevertheless, post-hoc pairwise comparisons show no significant differences among the three groups.

Figure 11:

Significant interaction: GA * age.

5.3 Explicit attitudes

The answers to the two open-ended explicit questions were thematically analyzed using a bottom-up approach. The following themes were identified: Basque native speaker, Spanish native speaker, intelligence, lack of Spanish grammar knowledge, lack of Spanish use, knowledge of Spanish, nice, not formal, rural, and NA (comments regarding physical aspect or irrelevant comments). Specifically, the sentence that included an instance of N-SGA was associated with speakers who acquired Basque as their first language (not simultaneous bilinguals) 35 times, which stood out as the most repeated comment for N-SGA: el castellano no es su primer idioma [Spanish is not their first language], euskaldun zaharra y hablante de euskera [Basque native speaker and Basque speaker], fallo garrafal y común entre vascoparlantes [tremendous error, common among Basque native speaker], or una persona que seguramente hable mucho en euskera debido al error que ha cometido en la utilización del adjetivo. Diría que es “euskaldun zaharra” porque puede que tenga muy interiorizada la gramática del euskera [a person who probably speaks a lot of Basque due to the error they produced in the use of the adjective. I would say that they are “euskaldun zaharra” because Basque grammar may be very interiorized in them]. Interestingly, one of the participants mentioned that it may be a stereotypical idea linked to Basque native speakers: es la idea que temenos de cómo habla un “euskaldun zaharra” pero es poco real [it is the idea we share about the manner in which a “euskaldun zaharra” speaks but it is not very real].

The next most repeated theme in reference to N-SGA in BAC Spanish is the perceived lack of proficiency in Spanish grammar. This comment was repeated 19 times: una persona que no distingue bien el masculino y el femenino [a person who does not distinguish correctly between masculine and feminine], alguien que confunde el género de las palabras [someone who mixes nouns’ gender], or alguien que no diferencia bien el género de las palabras, posiblemente porque en su lengua maternal no lo haya o no coincide [someone who does not distinguish correctly the words’ gender, possibly because there may not be one in their native language or because they do not coincide]. One participant even showed a complete rejection of this type of production: insoportable, las discordancias de género me matan [insufferable, gender disagreements kill me]. The third most repeated theme in relation to N-SGA was speakers’ lack of Spanish language skills: una persona que en su día a día no utiliza mucho el castellano [a person who in their daily life does not use much Spanish], or una persona que ha leído poco [a person who has read very little]. Finally, only three comments were related to the intelligence of the speakers (e.g., una persona poco culta que no ha podido recibir una educación escolar [a very uncultured person who has not been able to receive a formal education]), and two in relation to individuals from rural areas (e.g., de ámbito rural [from rural areas]).

The most repeated theme in reference to the sentence that included an instance of SGA was a strong understanding of Spanish grammar. This theme was repeated 25 times with comments such as alguien que comprende el género y número de la palabra [someone who understands the gender and number of the word], una persona que domina la lengua [a person who dominates the language], or correcto gramaticalmente [grammatically correct]. The second most repeated idea related to SGA in BAC Spanish was that this person is a Spanish native speaker (18 comments): una persona hablante nativa de castellano y euskaldun berria [a Spanish native speaker and a new learner of Basque], or de lengua maternal castellano [Spanish native speaker]. Finally, 15 sentences were classified as not applicable, and other less common comments were also identified: speakers’ intelligence (4 comments; e.g., una persona culta [a cultured person]), lack of Spanish proficiency (3 comments; e.g., una persona que no sabe expresarse en castellano [a person who does not know to express in Spanish]), people from rural areas (2 comments; e.g., una persona de origen rural hablando de un lugar conocido por el oyente [a person originally from a rural area talking about a known place to the listener]), among others.

Therefore, the explicit attitudes gathered in this task suggest that, on the one hand, N-SGA in BAC Spanish is mainly associated with Basque native speakers with limited knowledge of Spanish grammar, and, on the other, SGA production is explicitly linked to Spanish native speakers with higher levels of Spanish grammar knowledge. Interestingly, the explicit attitudes are consistent across participants regardless of the independent variables considered in this research.