‘The Subversion of Dialects’: Changing Attitudes towards Rural Varieties of Galician

Abstract

The gheada and the seseo are the two pronunciations most stigmatised by the top-down standardising tradition of Galician from the mid-19th century. Social stereotypes of peasantry, ignorance, and vulgarity were built on them. Nowadays, those stereotypes are the basis for indexical pointing. These pronunciations were outlawed from schools in the past. Today, despite having been considered standard by The Royal Galician Academy since 1982, they are almost absent from the classrooms, including those of Galician language and literature. This situation is detrimental to the linguistic capital of its users as compared to that of standard speakers. Nonetheless, since the end of the 20th century, there has been a social resignification of the gheada and seseo, symbolically used to express authenticity, ethnolinguistic adherence, and/or socio-political and cultural resistance. Currently, the emergence of vernacular language ideologies (VLIs) counterbalances the weight of standard language ideologies (SLIs) on these phenomena. This article analyses the linguistic attitudes of a sample of young people towards these two dialectal varieties as opposed to the standard pronunciations. It also identifies the indexical associations of contrasting varieties and their evolution over time. For this purpose, the matched-guise technique in combination with semantic differential scales (SDSs) has been applied. The results show that whereas standard pronunciations index social success, dialectal pronunciations index solidarity. However, while the standard indexical values are very stable, a rise in dialectal ratings is observed over fifteen years, which means an improvement of the attitudes towards them. As in other European minority languages, this phenomenon indicates a process of value levelling of the linguistic varieties and the growing weight of the VLIs in late modernity in Galicia.

1. Introduction

The increasing attention towards the standardisation processes of minority languages and the associated ideologies has brought to light the existence of ideological tensions within them. These tensions arise from the conflict between the acceptance of the standard and the desire to preserve their dialectal diversity (Auer 2005; Coupland and Kristiansen 2011; Coupland 2014; Darquennes and Vandenbussche 2015; Lane et al. 2018; Ayres-Bennett and Bellamy 2021). Two general trends have been identified: ideologies of standardisation (SLIs) and ideologies of polynomia (Ó Murchadha 2021).

In communities with SLIs hegemony, revitalisation and planning processes are structured “willy nilly, around the same received notions of language that have led to their oppression” (Woolard 1998, p. 17), although the standard also creates a sense of identity and differentiation from the hegemonic presence of the dominant language. In communities with polynomic ideologies, the notion of prestige is an open concept, and there is no focused variety, but rather there is a tendency to recognise the legitimacy of all of them to perform the functions of the standard (e.g., Jaffe 2021, for Corsican; Kristiansen 2021, for Norwegian; or Ó Murchadha 2021, for Irish). In the first type of communities, the standard acts as a centripetal force that shifts the balance between the norm and variation in favour of reinforcing the norm. In the second type, variation is reinforced to the detriment of the norm (Kristiansen 2021).

As Joseph (1987) argues, in regimes of standardisation, a synecdochic relationship is established between the standard variety and the language. One of its consequences is that other varieties are downgraded to sub-standard status and labelled vulgar, deviant, incorrect, or even ungrammatical, both by normative–prescriptive treatises and by ordinary people. These cultural systems are characterised by the reification, idealisation, and mythologisation of the standard as the ‘best’ variety (Lippi-Green 1997; Milroy and Milroy 1991; Milroy 2001; Kristiansen 2021) and are underpinned by the principle of authority that has traditionally characterised the ‘top-down’ model of language planning (Ayres-Bennett 2021). According to this principle, a group of planning agents are vested with legitimacy to select the ‘best version’ of the language and prescriptively disseminate it ‘downwards’, so that it will be accepted by the community. For this purpose, a legal and institutional corpus is used, which forms what Lippi-Green (1997, p. 64) calls the ‘dominant bloc’ and Silverstein (1996, p. 286) the ‘paraphernalia of standardisation’. SLIs are thus an integral part of socio-political systems and have ideological implications that go beyond the linguistic realm. As Woolard (2021, p. 2) points out, “Language ideologies are not only about language”. In their mediating role between linguistic practices and social structure, SLIs are instrumental in rationalising and naturalising the indexical value of language varieties and their denotative capacity. Indexical signalling such as ‘the best language’ → ‘the best speaker’ → ‘the best human qualities’, presupposes the existence of ‘the worst ones’, which turns SLIs into tools of classification and social exclusion. In regimes with strong SLIs, skill in the use of the standard determines who is legitimised to speak, both in the domain of the public and the published (Foucault [1970] 2002), legitimacies currently challenged thanks to social networks.

A feature of late modernity in European states is the weakening of the standard norm and even the delegitimisation of its ideological foundations (Kristiansen and Coupland 2011; Kristiansen and Grondelaers 2013). This phenomenon is linked to attitudinal changes favourable to vernaculars and their media visibility, especially among young people (Ó Murchadha 2021; Kristiansen 2009, 2021), which generates processes tending towards polynomia, such as demotisation and de-standardisation or vernacularisation (Coupland and Kristiansen 2011; Coupland 2014; Kristiansen 2021; Amorós-Negre 2024). Demotisation involves the rise of stigmatised linguistic forms to prestigious varieties without diminishing the social hegemony of the ideologies of standardisation. In such cases, the referent of ‘the best language’ shifts from one variety to another, giving rise to a new standard (Kristiansen 2021). In contrast, de-standardisation produces an erosion of SLIs and a rejection of the ideologeme ‘the best language’, with the gradual loss of recognition of the superiority of some language variety. Finally, as understood by Coupland (2014, pp. 83–87), vernacularisation includes the changes towards a more positive valuation of vernaculars and the progressive loss of hegemony of SLIs. All these phenomena arise from some mismatch between the stages of norm planning that depend on specialised agents and institutions (Haugen 1966, selection, codification and elaboration) and the phase of social acceptance that depends on the agency of the speech community.1

2. Objectives and Methodology

This article is in line with the growing sociolinguistic interest in the study of levelling processes in the valuation of language varieties in different European contexts (Coupland and Kristiansen 2011) and the organisation of their indexical signalling (Silverstein 2003; Eckert 2008). In the Galician context, it joins other research that addresses attitudes and ideologies towards linguistic variation using different methodologies (González et al. 2003; Beswick 2007; Loureiro-Rodríguez 2008; Iglesias 2013, 2020; O’Rourke 2018; Formoso et al. 2017; Suárez Quintas 2017; Fernández-Rei and Aguete 2023). Our general objective is to analyse quantitatively the linguistic attitudes of a group of young people towards gheada and seseo (two dialectal pronunciations of Galician) and their standard alternatives.

The choice of these pronunciations responds to the need to pay attention to linguistic units which “in the normative tradition, had previously been stigmatised as ‘bad’, ‘incorrect’, ‘colloquial’ or ‘vulgar’, or which have been made ‘invisible’ in printed texts and in metalinguistic sources of the time” (Elspaß 2021, p. 108). More specifically, we seek to answer the following questions: (i) whether there are factorial dimensions underlying attitudinal evaluations and what these are; (ii) whether there are significant differences between dimensions in the evaluation of the three varieties; (iii) whether there is a significant relationship between evaluations and language use; (iv) whether there have been chronological changes in the evaluation of the three varieties and their indexical associations, indicating ongoing processes of vernacularisation.

For analytical purposes, we assume the existence of two sociolinguistic variables: (g-h) and (θ-s), with standard variants (g) and (θ), respectively. These correspond to the phonetic contextual alternatives [g]~[ɣ̞]/[ħ] and [θ]/[s]. We studied attitudes towards three combinations of these variants, which give rise to the dialect with gheada (GH), with gheada and seseo (GHS), and to the standard variety (E = (g)-(θ)). In areas where there is gheada, there is seseo, but not vice versa; therefore, we have not evaluated seseo in isolation.

We understand linguistic attitudes as ‘the psychological tendencies’ of speakers (Eagly and Chaiken 2005, p. 745), expressed through the attribution of indexical meanings to the three linguistic varieties on fourteen SDS (semantic differential scales). To this end, we applied the matched-guise technique to a sample of 412 secondary school students in two phases separated by 15 years (S-2000 and S-2015) (§ 4.1.). In this method, informants rate the same speaker using contrasting language varieties (‘guises’) (Lambert et al. 1960; Garrett 2010). This is a procedure that allows for psychosocial evidence of registers from the listener’s perspective, since by contrasting them, the listener activates ethnopragmatic (ideological) interpretations of the value of these registers within the linguistic community (Soukup 2013a, 2013b)2.

Data processing was carried out using the SPSS statistical package. Factor analysis was used to obtain the dimensions underlying the scalar scores; a principal components method with VARIMAX rotation was carried out. A repeated measures analysis, Student’s t-test for related samples, and univariate ANOVA for independent samples were also conducted. The level of significance was p ≤ 0.05. The Bonferroni method was used to adjust P values for multiple comparisons.

One of the most important findings is the existence of two factor dimensions (status/solidarity) which correspond to the standard and the dialects, respectively. The value of the standard is not affected by the status/solidarity opposition (positive for both) and remains stable over time. However, dialects score much lower than the standard on status and much higher on solidarity, a dimension in which they improve significantly over time. This trend is more pronounced in the dialect that differs most from the standard (GHS), which even increases in status. This suggests that Galician is undergoing a process of value levelling and re-vernacularising, but with no signs of erosion of the SLIs.

3. The Regime of Galician Standardisation

Galician is one of the co-official languages of Spain. It is spoken in the north-west of the Iberian Peninsula, bordering Asturian to the east and Portuguese to the south, with which it has a high degree of historical kinship.

It is a language of late standardisation, for which there was no official planning policy until the 1980s (Fernández-Salgado and Monteagudo 1995; Regueira 2003, 2005; Alonso-Pintos 2006; Ramallo and Rei-Doval 2015). With the advent of democracy, three fundamental texts were approved in this regard: the Statute of Autonomy (EA, Organic Law 1/1981 of 6 April 1981), the Law of Linguistic Normalisation (LNL, Law 3/1983 of 15 June 1983), and the Orthographic and Morphological Norms of the Galician Language (NOMIG; RAG/ILG 1982). These rules, drawn up by the Real Academia Galega (RAG) and the Instituto da Lingua Galega (ILG), establish a multicompositional standard based on the selection of variants of the three main dialectal blocks of Galician. The glottopolitical importance of these texts is obvious. After being banned for 30 years by the Franco dictatorship, Galician was legally recognised as Galicia’s ‘own language’ and co-official language alongside Spanish (EA, art. 5.1; LNL, art. 4.1). It was also stipulated that the public authorities of Galicia shall promote its use in all areas “of public, cultural and informative life” and shall provide the necessary means “to facilitate its knowledge” (EA, art. 5.3.). As a result of this provision, Galician has been introduced into the regional government and administration, the Galician Broadcasting Corporation (Radio Televisión de Galicia, RTVG), and the education system. The recognition of the right of all Galicians to know and use Galician (LNL art. 1) is materialised by the introduction of the standard variety as an object and vehicle for teaching. As in other political contexts (e.g., Agha 2003; Coupland 2014), school and television have played a key role in the social diffusion of the standard norm as the ‘best version’ of the language and the institutionalisation of a dual standardisation regime in Galicia: that of Spanish and that of Galician.

From the perspective of the top-down language planning outlined above, the SLIs of Galician emerged long before the existence of an official language policy and under the influence of the Spanish hegemonic model, in which the group driving standardisation (the Galician intelligentsia of the mid-19th century) was acculturated. Its emergence coincides with the recovery of Galician for cultural literature and the first attempts at the linguistic codification of the Rexurdimento (Renaissance). The absence of a normative authority created a sense of anarchy and disorder among writers, especially in spelling, which led to the choice of a linguistic model, which some sought in popular oral dialects and others in the language of the bilingual intellectual bourgeoisie or in the medieval literary tradition (González-Seoane 1994; Regueira 2003; Recalde 2008). At the same time, figures emerged who, concerned about the quality of the literary Galician of their contemporaries, played the role of ‘guardians of the language’ (Milroy and Milroy 1991). This concern, now projected onto the uses of ordinary speakers, gave rise to a series of prescriptive texts published under the common denominator of ‘Galician of quality’ (for a critique on this topic, see Rei-Doval 2013; Regueira 2019; Longa 2020).

Language policies bear the ideological imprint of the agents who implement them, since “there is no possibility of scientifically establishing a standard” (Regueira 2019, p. 142). Whatever the more democratic or aristocratic tendency (Joseph 1987) of the proposals for the selection of the standard norm for Galician from the 19th century onwards, they have almost always been marked by anti-ruralist prejudices that proscribed the use of popular phonetic solutions such as gheada and seseo, which were considered vulgarisms or linguistic deformations. In general terms, rural oral Galician was represented in the discourse of the standardising elites as a lacking language (inexpressive, limited, backward, vulgar, neglected…), while the literary standard was associated with dignity, order, culture, urbanity, progress, and the collective identity of Galicia (Recalde 2018, 2021). It can therefore be said that the SLIs regime in Galician reflects, on the one hand, the ideological opposition between tradition and modernity characterised by Gal (2018) and, on the other, reproduces in the minoritised language the sociolinguistic hierarchies imposed by contact with the dominant language. As ideologies of social differentiation (Gal, ibid.), the SLIs of Galician have marked the popular varieties by indexical signalling that lowered their value while increasing the value of the standard in the market of symbolic goods (Recalde 2021).

Postmodern tensions between SLIs and ideologies of the vernacularisation (VLIs)—which can be expressed as varying degrees of resistance to the standard—also exist in the context of Galician (Domínguez 2003; Recalde 2007, 2022; Iglesias 2013, 2020; Zas 2016; Formoso et al. 2017; Regueira 2023). On the one hand, it can be observed how the assumption of SLIs by ordinary speakers leads them to undervalue their dialectal varieties, to which they attribute negative qualities compared to the (‘exemplary’) Galician of television and cultural institutions. On the other hand, there are dissident voices in the public sphere that question the legitimacy of the ‘bearers’ of the standard (be they individuals or institutions) to impose their version of ‘good Galician’. These voices include well-known figures from Galician culture (musicians, performers, poets, oral storytellers, comedians, etc.) but also ordinary speakers who disseminate metalinguistic discourses through social networks in which they claim the value of their dialectal varieties and question the superiority of the standard (Recalde 2022; Regueira 2023). News-peakers also suffer from their own linguistic conflict (O’Rourke and Ramallo 2013). On the one hand, they are seen by some vernacular speakers as the paradigm of good speakers because they learned standard Galician at school. On the other hand, they consider it unfair to be delegitimised by other speakers (both neo and traditional) because of their lack of authenticity. However, they also believe that authentic Galician is the vernacular.

One of the aforementioned dissident voices is the poet, punk musician, and performer Leonardo Fernández-Campos (O Leo), who has published a book of haikus of social criticism and political satire3. In one of these short poems (“Endodighlosia”), he proclaims: “The abuse of the norm serves to continue convincing those at the bottom that they do not know how to speak” (Fernández-Campos 2007, p. 87). This normative dissidence, very common in late-standardised languages (Amorós-Negre and Monteagudo 2024), has been notorious in avant-garde poetry since the 1980s (Nogueira 2022) and, at the end of the millennium, in music (Fernández-Rei 2004; Recalde 2007). Since then, historically discredited popular oral varieties have been used in cultural artefacts, created on the margins of institutions by contesting voices that express in the vernacular not only their identity differentiated (and confronted) to the institutional Galicianness, symbolised by the standard, but also their political opposition to the structures of power in Galicia. The translation of popular speech into the written code leads to the practice of re-spelling (Jaffe 2000), as in the poem quoted above (whose original orthography represents phenomena typical of orality). This is a common strategy in these texts, which makes visible the variation hidden behind the normative uniformity.

In short, although rural oral Galician continues to evoke backwardness, ignorance, or rudeness in part of the population, we are witnessing a process of “value levelling” (Ó Murchadha 2021, p. 750) in the form of a recapitalisation of rural culture and its language varieties. As opposed to the standard, these varieties are now reinterpreted and claimed as indices of authenticity and as continuators of “the historical Galician language (...) inherited by generations of speakers” (Regueira 2012, p. 195). The ideologies of the vernacular emerge as a reaction to the relations of domination imposed by the logic of the regimes of standardisation in the Galician socio-political space and are manifested not only in linguistic–performative practices but also through metalinguistic reflexivity or attitudinal reactions that are not overtly conscious, such as those analysed in this research.

4. Gheada, Seseo, and Oral Standard

The gheada consist of pronouncing the ‘g’ in words such as ‘gato’ (‘cat’), typically as an aspirated pharyngeal or fricative velar consonant ([ħ]~[x]). In the prestigious standard variety, this sound is articulated as an occlusive or approximant voiced velar ([g]~[ɣ̞]), depending on the phonetic context. Gheada was once widespread throughout most of Galicia, including all the Atlantic provinces, the western border of Lugo, and half of the provinces of Ourense (Figure 1; green area of gheada). Since the mid-nineteenth century, however, it has declined in cities and in areas of influence. Seseo is the absence of the /θ/:/s/ opposition, leaving only the alveolar unit /s/. It can occur at the end of a word or at the beginning of a syllable, which is the more stigmatised position. Although the implosive seseo is widespread in the western half of Galicia (Figure 2; mauve tones), the plosive position is limited to a narrow strip of territory in the Atlantic area (Figure 2; dark mauve tone), where gheada and seseo are pronounced concurrently.

These pronunciations have had low social prestige, especially the gheada, for over a century. The implosive seseo is prevalent throughout the western half of Galicia, while the plosive position is limited to a narrow strip of territory in the Atlantic area where gheada and seseo coexist. Following the data of IGE (2018), the proportion of Galician speakers residing in areas of gheada and seseo is approximately 90% and 80%, respectively.6 However, the prevalence of these variants cannot be directly inferred from these figures, as they are socially stratified by age and education and also depend on the degree of urbanisation of the area.

The gheada and the seseo are at the focus of the metalinguistic awareness of the sociolinguistic community (Suárez Quintas 2017). Since the nineteenth century, they have been the object of linguistic prejudices and an index of the social characteristics or personality of the speakers. In Labovian terms (Labov [1972] 1983, p. 387), they function as markers of informal style and are, at the same time, stereotypes of rurality and unculture, especially the gheada (Recalde 1994, 1995; Dubert 2006; Loureiro-Rodríguez 2008; Zas 2016; Formoso et al. 2017, etc.).

Gheada is a more marked variant than seseo because its stigmatisation is linked to purist ideologies about its origin and to deficit theories (Recalde 2003, 2022). Until the eighties of the last century, the normative hegemonic discourse considered it an interference of Castilian caused by the diglossic behaviour of the illiterate lower classes, who replaced the phoneme /g/ of their own language with the phoneme /x/ of Castilian (Pensado 1983). Based on testimonies from the eighteenth century (Pensado 1974), its presence in Galician was attributed to the strong pressure of Castilian and the lack of mastery of it by peasants and lower classes in general, who also tended to confuse the phonemes /x/ and /g/. The recurrent discursive association between the illiterate lower classes, diglossia, and learning failure is at the root of the gheada second-order indexical pointing (lack of culture, incivility, rudeness, etc.) (Silverstein 2003)7. These meaning constructions arose in parallel to the metalinguistic discourse of the standard norm in the mid-nineteenth century and spread socially with the expansion of compulsory education in Galicia. All this shows how the SLIs of Galician have contributed to the marking (Bucholtz and Hall 2004) and enclassment (Bourdieu [1982] 1985) of these dialectal variants. In addition, their use has been banned by the education system, which has favoured their social stratification: the higher the level of education, the less the use of these units.

Gheada and seseo are currently accepted by NOMIG as standard alternatives (RAG/ILG 1982), but they have never played this role de facto: they are absent from the linguistic model transmitted by the school, the media, the political–institutional discourse, and the literary canon. In the case of TVG, they have been explicitly excluded from serious programmes, although they may appear in humorous gags. In 2012, the press reported that Galician television had reprimanded a young journalist for using gheada sporadically in a cultural magazine programme. The news was headlined “TVG asks a presenter not to use ‘gheada’. The standard norm returns to the programme ‘A Revista’”. The author of the newspaper article (aligned with the SLIs) described the presenter’s linguistic use as “an unprecedented way of speaking in the channel” (Pato 2012, np [italics ours]). The team of linguistic advisors justified the decision by referring to the image of the channel and the risk to its credibility: “The style must be set by linguistic and corporate lines in relation to the channel’s image. Credibility is at stake and language is one more aspect” (ibid., our italics). As Silverstein (1996) argues, SLIs presuppose a conceptual commitment to the existence of a way of using language denotatively. By establishing an explicit relationship between pronunciation and verisimilitude, speakers with gheada are indexically marked as less reliable than standard speakers. However, the presence of gheada and seseo in TVG is not as unusual as the news reports suggest. They had appeared much earlier in TV series, humour, and entertainment programmes, where they played a performative and enregistered role (Agha 2005; Eckert 2008) of the rural, lower-class, ‘poorly educated’ speaker, a role they still retain today. That is, they are permissible pronunciations if they serve to reinforce culturally embedded sociolinguistic stereotypes and do not challenge SLIs. As Coupland argues when analysing the use of language varieties in the BBC, vernacular styles are not only excluded from mass media: “it has required clear and consistent indexical relations to be matched with relevant vocal performances and performers” (Coupland 2014, p. 90). They can have their place in the broadcasting grid if they respect indexical orthodoxy.

Several authors point to the importance of television in the sociolinguistic changes involving standard varieties in different parts of Europe (e.g., Vaucekauskiene 2012; Coupland 2014; Kristiansen 2021). The TVG, besides banning gheada and seseo, reproduces the prosodic and vowel model of Castilian (Regueira 2019), which is very different from the dialectal Galician of most of its audience (concentrated in rural and semi-rural population centres, e.g., Instituto Galego de Estatística (IGE))8. Today, the broadcasting role of RTVG’s is supported by a Twitter profile with around 300,000 followers (#DígochoEu). In contrast to the rigidity of the linguistic canon of RTVG’s journalistic voices, some research indicates that young speakers of Galician were already in favour of the presence of gheada in the channel’s news at the end of the last millennium (Recalde 1995).

As a counterpoint to the situation described above, since the end of the 20th century, there has been a social resignification of gheada and seseo, used symbolically by the minority language community to express authenticity, ethnolinguistic belonging, and/or socio-political and cultural resistance, especially among young people. In other words, part of the speakers reinterpret the symbolic value of these pronunciations, giving rise to new indexical constructions organised in fluid and dynamic fields of indexicality (Eckert 2008). This change takes place in a context characterised by the essentialist revaluation of rural Galicia, its cultural traditions, and its linguistic expressions in the context of global neo-capitalism. The reorganisation of its fields of indexicality can be observed in different spaces: First, in the metalinguistic discourse of standardising agents since the late 1960s (Álvarez-Blanco and Fernández-Rei 1977; Santamarina 1982; among others). Secondly, in the linguistic attitudes and reflexive awareness of speakers (Beswick 2007; Recalde 2022; Regueira 2023). Finally, in the public use and metalanguage of postmodern cultural products—literature, music, oral narrative, etc.—(Alén 2001; Fernández-Rei 2004; Recalde 2007, 2022; Nogueira 2022).

Sociolinguistic authenticity is linked to other kinds of authenticity—ways of being, living and behaving in particular historical and material times and circumstances—and can circulate “via material artifacts (…) embedded under particular sets of social practices” (Johnston 2014, p. 100). Similarly, the stylisation of authenticity can transcend the linguistic system to include other semiotically organised systems of behaviour (Irvine 2001). Since the 1990s, the use of gheada and seseo in lyrics and poetic compositions in Galician has become quite common, as has public discourse that shows a reflexive awareness of their use. Fernández-Campos (O Leo) wrote in a preface to his book of haikus, “This book is pronounced with gheada uvular, palatalised implosive seseo and predorsal explosive seseo” (Fernández-Campos 2007, p. 2). Also, the poet Lupe Gómez, in her book Fisteus era un mundo, says: “I love my gheada. To say ‘jato’ (cat) ‘jaliña’ (chicken). I am proud to say ‘jato’ and ‘jaliña’. I am proud to be from the hamlet’ (Gómez 2001, p. 98). In this case, the gheada is reflected by a non-normative spelling <j>, which doubly transgresses the ideologies of the standard, since the accepted spelling for gheada since 1995 is <gh>. These artists often integrate the stylisation of these phonic variants into performative devices that evoke traditional popular culture (costumes, musical instruments...), but which, when recontextualised and combined with elements of postmodern iconography, acquire new indexical meanings. As Coupland argues, “vernacular performances will be symbolically mediated into new contexts, and into popular consciousness” (Coupland 2014, p. 86). Figure 3 shows an image of O Leo at a concert, dressed in regional costume, with punk boots and the Galician flag in the background. The band’s name ‘O Leo i Arremecaghona’ reproduces a popular oral Galician expression in which the gheada (in bold) is orthographically marked. Johnston also points out that a clear indicator of metapragmatic awareness of the authenticity of language varieties is their commodification. Figure 4 shows the image of a well-known brand of T-shirt with the printed message “o mellor amigo do ghato se non rabuñara” (‘a cat’s good friend if it didn’t scratch’), in which a popular saying is turned on its head. The word gato is also re-spelled as ghato to mark the pronunciation of gheada. As Järlehed (2019, p. 70) argues, the designers of these products “challenge standard language ideologies by their use of rural dialect (gheada) (…) by daring to use ‘imperfect’ and mixed forms of Galician”, and thus achieve an air of “rurban cool” (ibid.).

The integration of gheada and seseo into late modern cultural and marketing products “breaks the conventional link between locality and ‘local language’” (Coupland 2007, p. 122), so that these forms no longer necessarily refer to a place (‘the village home’) but to an identity of choice and a socio-political positioning. These sociolinguistic changes constitute the ideological framework within which the results of the quantitative analysis of this paper must be interpreted.

5. Methodology Description

5.1. Matched-Guise Design and Implementation

Matched-guise is a standard procedure in the speaker assessment paradigm for indirectly measuring ‘hidden’ attitudes towards language varieties. In psychosocial research, it is commonly combined with multidimensional SDSs (Osgood et al. 1957), which use adjectives saturated with evaluative meaning. By contrasting the scalar scores, we access to the indexical meanings of the different styles and their ideological implications regarding the value levelling or re-vernacularisation of Galician.

In this study, three male speakers in their thirties read an argumentative text (in favour of cycling) in each of the varieties analysed. All of them were bilingual with a predominance of Galician. Although several voices were masked, only the voice of a native speaker of the dialect with gheada and seseo was included in the analysis. He also dominated the standard. The voices were interspersed to make them difficult to identify.

The informants listened to six versions of the text, either in colloquial or formal style. In each version, one of the following three combinations of variants could occur: (i) [g]~[ɣ̞]-[θ]→E; (ii) [ħ]-[s]→GHS; (iii) [ħ]-[θ]→GH. As [ħ]-[θ] pronunciations are markers of colloquiality in Galician, only evaluations of the colloquial version were analysed to avoid the ‘parody effect’ (Garrett et al. 2003; Niedzielski and Preston 2000). The test was presented as part of a social research to infer personality traits from language, but the informants did not know which traits to look for.

The raters assigned each speaker a score between 1 and 5 on fourteen bipolar SDSs. Scores 1 and 5 represent the negative and positive poles of the attribute, 3 the neutral value. The choice of scales was based on classic attitudinal studies, which tend to show the existence of two factors: solidarity (underlying scales such as sympathy, kindness, generosity, etc.) and status-success (underlying scales such as intelligence, wealth, culture, etc.). Regional or ethnic varieties of low social prestige tend to score high on solidarity, while standard varieties, dominant languages, or dialects of high prestige score higher on status. To these factors is sometimes added that of dynamism (which groups together scales such as energy, activity, decisiveness, etc.) (cf. Williams et al. 1972; Ryan et al. 1982; Pieras-Guasp 2002; Garrett et al. 2003; Garrett 2010; Kristiansen and Grondelaers 2013; for Galicia, Fernández-Rodríguez 1984; González et al. 2003; Loureiro-Rodríguez et al. 2012, etc.). We ensured that the scales met the criteria of composition and factorial adequacy. Consistent composition requires that the scales relate meaningfully to the concepts being assessed and draw distinctions that are familiar to judges. Adequacy means that they reflect relatively pure measures of a single dimension, for which they should have high saturation on a single factor.

We assume that (i) there is more than one dimension underlying the ratings of the three varieties, (ii) some of these dimensions are related to solidarity/sympathy and social success/status, (iii) dialect varieties will be rated positively on solidarity and negatively on status. We hypothesise that of the fourteen scales, sympathetic, fair, sociable, spontaneous, sincere, and ethical will represent traits associated with solidarity, while intelligent, attractive, educated, active, successful, responsible, self-confident, and classy will represent status. Using FA, the factor components were tested independently for each variety, avoiding the aggregation of varieties into a single correlation matrix. Some of these scales were found to be ambiguous or with low saturation in the factorial explorations (§ 6.3), so they were deleted in the final FA.

5.2. The Sample

The matched-guise was administered to a sample of high school students (N = 412) in two phases fifteen years apart, allowing for a longitudinal comparison of results in real time. The first phase (S-2000, N = 285) was carried out in the academic year 2000–2001, the second in 2015–2016 (S-2015, N = 127). The students were aged between 16 and 18 years, and there was a balance between males and females. Informed consent was obtained before the test was administered, and anonymity and confidentiality of the data were guaranteed. Finally, the students anonymously completed a short sociolinguistic questionnaire with questions about their place of residence, gender, mother tongue(s), usual language(s), and the use of the gheada and seseo (their own and their parents’), which allowed us to observe their reflexive awareness about language use. All the institutes are public, except for the one in Santiago.

The sample was selected in five municipalities in the dialectal area with gheada and seseo. The representation of rural, urban, and semi-urban areas was sought because it is closely related to the use of languages in Galicia and their varieties (standard or dialectal). The GH and GHS varieties are more common in rural or semi-urban areas, while E is more common in urban areas, where Galician is also less widely spoken. According to the IGE, in 2003, 87.6% of the population of municipalities with less than 10,000 inhabitants spoke Galician, compared with 34.6% of those with more than 50,000 inhabitants. In 2018, the use of Galician fell to 81% and 24.9%, respectively, in these two population groups.10 In our sample, the towns of Padrón, Negreira, and Muros belong to the first category and the cities of Vigo and Santiago to the second.

Table 1. Distribution of the sample by localities.

	Population	Economy	S-2000	S-2015	N	%
Muros	8250	Fishing	67	23	90	21.8
Padrón	8350	Small-scale industry	59	27	86	20.8
Negreira	6900	Agrarian	48	26	74	17.9
Santiago	98,700	Tourism Administrative services	62	22	84	20.3
Vigo	293,600	Industry, fishing	49	29	78	18.9
Total			285	127	412	100%

shows the distribution of the sample by location, together with some demographic and socio-economic data (columns 2 and 3).

Regarding the linguistic characterisation of the sample, Galician is the predominant mother tongue in the secondary schools in the villages, whereas in the urban schools, Castilian is clearly dominant, reaching almost 2/3 in Vigo and 60% in Santiago. These patterns are repeated in the distribution of the usual language: Galician is insignificant in the two urban schools and the majority language in the villages (

Table 2. Use of the languages by high school.

	Muros	Padrón	Negreira	Santiago	Vigo
Only Spanish	11.1	10.5	8.1	27.4	30.8
More Spanish	10.0	23.3	13.5	58.3	65.4
More Galician	32.2	27.9	32.4	9.5	3.8
Only Galician	46.7	38.4	45.9	4.8

). These data are consistent with the latest language censuses of the IGE (2018).11 Table 2 shows the usual language of the samples aggregated by school and

Table 3. Use of the languages by sample year.

	S-2000	S-2015
Only Spanish	14.4	24.4
More Spanish	35.1	30.7
More Galician	22.5	18.9
Only Galician	28.1	26.0

, by year.

As for the use of the three varieties within each municipality, counting only Galician speakers (N = 201), the gheada and seseo decreases as we move from the rural to the urban centre of the cities. The 81% of informants who live in rural villages, 56.2% in towns, and 20.6% in cities say that they speak with gheada (always or sometimes). Seseo is used by 69.4% of rural informants, 43.8% in towns, and 12.7% in cities. It is worth noting that, although the sample belongs to the dialectal area where gheada and seseo coexist, the perception of the use of gheada is higher than that of seseo. On the other hand, women who acknowledge having these dialectal features outnumber men by 7 points (seseo: F = 48.1% vs. M = 41.1%; gheada: F = 57.1% vs. M = 50.3%). This contradicts the female tendency to hide the use of stigmatised varieties which has been found in other sociolinguistic contexts. Also noticeable is the presence of 50 Spanish-speaking students who claim to use gheada, most of whom live in cities (

Table 4. Spanish speakers who state use of GH.

Highschool	N
Muros	8
Vigo	16
Padrón	10
Negreira	3
Santiago	13

). This self-perception is unlikely, but it is significant as an indicator of covert prestige.

As perceptual dialectology studies have shown (Suárez Quintas 2017), gheada is a more perceptible pronunciation than seseo and is more socially stereotyped. According to Niedzielski and Preston (2000, p. 10), theses types of phonetic–phonological units are more accessible to the speaker’s metalinguistic awareness, which may also explain why it is over-declared compared to seseo.

Considering the combination possibilities of the variables (g-h) and (θ-s) (

Table 5. Distribution of varieties by habitat type ^a.

	Urban	Villages	Rural
E	79%	41.9%	18.4%
GH	11.3%	12.1%	13.2%
GHS	9.7%	46%	68.4%
Total	100%	100%	100%

^a Results with aggregated samples.

), the standard variety (E) is dominant in cities, and GHS predominates in rural areas. The use of GH in cities is an effect of the aforementioned perceptual bias of Spanish speakers, and its low presence in towns and villages of the sample is because it does not occur as a vernacular variant isolated from seseo.

Looking at the chronological evolution of the use of the varieties in villages (

Table 6. Distribution of varieties in the village’s high schools ^a.

	S-2000			S-2015
	Muros	Padrón	Negreira	Muros	Padrón	Negreira
GHS	83.6%	78.9%	23.8%	82.6%	46.4%	50%
GH	1.5%	5.3%	38.1%		7.1%	23.1%
E	11.9%	22.8%	38.1%	17.4%	46.4%	26.9%

^a Results with disaggregated samples.

), a drastic increase can be seen in the use of E in Padrón-2015, while in Negreira-2015, the reported use of GHS increases. This may be related to the socio-economic differences between the two centres. Padrón is the most populated, commercial, and industrialised village of the three, while Negreira is an agrarian village. We believe that the increase in the reported use of GHS in Negreira-2015 indicates a reinvestment of the social stigma of the dialectal variety of the area, which acquires identity values, and a rebalancing between SLIs and VLIs. The increase in E in Padrón follows the historical pattern of the use of prestige varieties in the most urbanised villages of Galicia.

6. Results

6.1. Use of (h) and (s) across Generations and Real-Time

Table 7. Evolution of the use of gheada in apparent and real time.

S-2000 (N = 144)				S-2015 (N = 57)
	Students	Fathers	Mothers	Students	Fathers	Mothers
Yes	51.4%	60.4%	56.3%	47.4%	50%	43.9%
Sometimes	33.3%	19.4%	23.6%	40.4%	25%	22.8%
No	15.3%	20.1%	20.1%	12.3%	25%	33.3%
Total	100%	100%	100%	100%	100%	100%

and

Table 8. Evolution of the use of seseo in apparent and real time.

S-2000 (N = 144)				S-2015 (N = 57)
	Students	Fathers	Mothers	Students	Fathers	Mothers
Yes	45.1%	53.5%	49.7%	36.8%	35.7%	28.1%
Sometimes	28.5%	16.7%	19.6%	38.6%	28.6%	29.8%
No	26.4%	29.9%	30.8%	24.6%	35.7%	42.1%
Total	100%	100%	100%	100%	100%	100%

show the results of the Galician-speaking evaluators’ statements on their own and their parents’ use of gheada and seseo, allowing us to observe the rates of reproduction in apparent (intergenerational) and real time (contrasting S-2000 and S-2015).

In apparent time, informants use gheada and seseo more than their fathers. Mothers make the least use of both variants, which may reflect a greater tendency of middle-aged women to avoid stigmatised forms. This contrasts with the self-reported data of young girls, who show a greater use of dialects than boys (§ 5.2.). In real time, among informants in S-2015, there is an increase in gheada and especially in seseo (12 points). However, between S-2000 and S-2015, the use of dialect pronunciations by both parents (especially mothers) drops to 13 points. There is, thus, a strong coherence between the apparent and real time data, suggesting a revaluation of dialectal pronunciations by young people. This may be due to the need to mark generational differences in order to reaffirm their identity in the face of adults or perhaps a reversal of the traditional indexical values of gheada and seseo, which are no longer associated with people of advanced age and little education but rather with adherence to the rural culture of origin.

On the other hand, young people in S-2015 report more variation in the use of these forms with respect to S-2000 and in each sample with respect to their parents. This may indicate a growing metapragmatic awareness of the contextual appropriateness of the sociolinguistic variables (g-h) and (θ-s) as markers of style, reinforced by the fact that the test was carried out in a school context. The exposure of generations of young people to standard Galician in the educational system since the 1980s seems to have consolidated ideas about the functional specialisation of Galician varieties in line with SLIs.

6.2. Intergenerational Transmission of (h) and (s)

In order to observe intergenerational transmission, perceptions of their own use were compared with those of their parents in disaggregated samples (

Table 9. Intergenerational reproduction of gheada.

	Fahter’s use of (h)
Students	S-2000				S-2015
		Yes	Sometimes	No	Yes	Sometimes	No
	Yes	74.7%	21.4%	10.3%	78.6%	7.1%	21.4%
	Sometimes	23.0%	75.0%	24.1%	14.3%	85.7%	50%
	No	2.3%	3.6%	65.5%	7.1%	7.1%	28.6%
	Total	100%	100%	100%	100%	100%	100%
	Χ2 = 101,734 (4 gl), p < 0.01				Χ2 = 28,583 (4 gl), p < 0.01

and

Table 10. Intergenerational reproduction of seseo.

	Mohter’s use of (s)
	S-2000				S-2015
Students		Yes	Sometimes	No	Yes	Sometimes	No
	Yes	80.3%	19.9%	4.5%	81.3%	17.6%	20.8%
	Sometimes	18.3%	67.9%	20.5%	18.8%	76.5%	20%
	No	1.4%	14.3%	75%	0	5.9%	54.2%
	Total	100%	100%	100%	100%	100%	100%
	Χ2 = 116,476 (4 gl), p < 0.01				Χ2 = 35,944 (4 gl), p < 0.01

). Looking at the diagonals (grey areas), we can see that young people reproduce their parents’ language use to a high degree, both in S-2000 and in S-2015: (h) between 65.5% and 85.7%; (s) between 54.2% and 81.3%. The increase in use in relation to the parents is shown in the upper part of the diagonal and the decrease in use in the lower one. As can be seen, in both samples and for both variants, the percentages of intergenerational recovery are higher than those of abandonment, which shows an explicit adherence to the dialects. Also noteworthy is the percentage of young people who claim to use (h) or (s) without these variants having been inherited from the family: (h), from 10.3% to 21.4%; (s), from 4.5% to 20.8%. This re-vernacularisation is very noticeable in S-2015.

6.3. Language Attitudes

The existence of status and solidarity dimensions cannot be taken for granted without testing them with FA. On the one hand, the indexical meanings of language varieties can vary between socio-cultural contexts or from one historical moment to another within the same context. There are no universal dimensions of connotative meaning, neither in number nor in structure (Fernández-Rodríguez 1984; Garrett 2010). On the other hand, the attributes that make up the factors may differ between varieties. In this sample, the Cronbach’s alpha value is high for both (dis)aggregated varieties (between 0.86 and 0.87), indicating a high degree of internal consistency and reliability of the SDS. However, we do not assume the existence of a single factorial matrix for all varieties, as Cronbach’s alpha can also be high for sets of variables with a multidimensional structure. These are reasons for not adopting a single factorial matrix for the aggregated varieties as representing the underlying dimensions for the disaggregated ones. Therefore, independent analyses were carried out for each variety and sample year.

The FA for the aggregated varieties partially confirmed the two hypothesised factors (with some attributes outside the expected dimension) but added another unanticipated factor that grouped attributes related to ethical sense. Since this triple structure was not preserved in the disaggregated varieties, we forced two dimensions in a confirmatory FA. Providing that some ambiguous or low saturating scales caused factorial instability between varieties, they were eliminated after several checks (these are ethical, attractive, responsible, and active). Thus, we obtained an optimal factor structure, due to its stability across varieties and samples, confirming the factors assumed in § 5.1. The status and solidarity dimensions are confirmed by the FA. They are formed by a coherent set of scales in each of the samples that explains about 60% of the variance in the aggregated varieties and between 53.2% and 60% in the disaggregated ones (

Table 11. FA structure ^a.

Aggregated varieties		Aggregated samples (Explained Var. 59%)		S-2000 (Explained Var. 59.7%)		S-2015 (Explained Var. 57.7%)
		F1 Status (29.7% var.)	F2 Solidarity (29.2% var.)	F1 Status (30% var.)	F2 Solidarity (29.7% var.)	F1 Status (29.7% var.)	F2 Solidarity (27.9% var.)
	educated	0.865		0.873		0.848
	classy	0.825		0.834		0.799
	intelligent	0.785		0.774		0.799
	successful	0.732		0.699		0.797
	spontaneous		0.744		0.736		0.744
	sociable		0.739		0.704		0.783
	friendly		0.726		0.724		0.720
	sincere		0.709		0.725		0.689
	fair		0.607		0.638		0.542
Disaggregated varieties S-2000		E (Explained Var. 60.33%)		GH (Explained Var. 56.55%)		GHS (Explained Var. 58.19%)
		F2 Status (27.0% var.)	F1 Solidarity (33.3% var.)	F2 Status (26.66% var.)	F1 Solidarity (29.89% var.)	F2 Status (28.9% var.)	F1 Solidaridarity (29.2% var.)
	educated	0.824		0.811		0.835
	classy	0.853		0.797		0.803
	intelligent	0.635		0.695		0.780
	successful	0.617		0.680		0.649
	spontaneous		0.750		0.677		0.721
	sociable		0.717		0.735		0.642
	friendly		0.708		0.704		0.693
	sincere		0.726		0.714		0.771
	fair		0.654		0.671		0.662
Disaggregated varieties S-2015		E (Explained Var. 60%)		GH (Explained Var. 53.27%)		GHS (Explained Var. 58.71%)
		F2 Status (30% var.)	F1 Solidarity (30% var.)	F1 Status (27% var.)	F2 Solidarity (26.1% var.)	F1 Status (30.1% var.)	F2 Solidarity (28.6% var.)
	educated	0.815		0.826		0.841
	classy	0.787		0.744		0.754
	intelligent	0.787		0.748		0.803
	successful	0.800		0.715		0.725
	spontaneous		0.727		0.782		0.710
	sociable		0.816		0.731		0.851
	friendly		0.733		0.734		0.801
	sincere		0.771		0.606		0.537
	fair		0.573		0.468		0.502

^a Green and gold colours mark Status and Solidarity components, respectively.

).

From the factorial result, we created six new variables by calculating the mean scores in each dimension and variety: E_status, E_solidarity, GH_status, GH_solidarity, GHS_status, GHS_solidarity. Subsequently, a repeated measure analysis was applied, in which varieties and dimensions were considered intrasubject factors, according to a 3 × 2 structure:

1. E	1.1 Solidarity 1.2 Status
2. GH	2.1 Solidarity 2.2 Status
3. GHS	3.1 Solidarity 3.2 Status

Three within-subject and seven between-subject hypotheses were tested (

Table 12. Tested hypothesis ^a.

Intrasubject factor Variety

H1. The mean of E is higher than the mean of GH and GHS.

Intrasubject factor Dimension

H2. The means of the dimensions Status and Solidarity differ from each other.

Variety * Dimension interaction

H3. The mean scores of GH and GHS are higher than E in the dimension of Solidarity, but lower in the dimension of Status.

Intersubject Factors

Variety * Sample interaction

H4. The configuration of the differences between the means of the standard variety and the other two differs according to the sample year.

Dimension * Sample interaction

H5. The dimension * sample interaction is present

Variety * Dimension * Sample interaction

H6. Interaction between variety and dimension differs between S-2000 and S-2015

Variety * Usual language interaction

H7. The means of the GH and GHS varieties differ between language groups.

Variety * Dimension * Usual language interaction

H8. There are intergroup differences related to the language group, in the sense that the more Galician is spoken, the higher the means in the solidarity dimension for the dialectal varieties.

Variety * Use of the Gheada interaction

H9. Means of GH and GHS varieties differ between those who use the gheada and those who do not.

Dimension * Use of Gheada interaction

H10. The use of gheada results in an increase in the distances between means in the dimensions of status and solidarity.

^a * means interaction between variables.

).

Hypotheses H₁ to H₄ and H₆ are confirmed. There are no significant differences between the groups in terms of language use, except for a small difference in usual language, which is difficult to explain, and therefore, it was excluded from the analysis.

Table 13. Confirmed hypotheses.

Effect	Wilks Lambda	f	df of Hypothesis	df of Error
Variety (H1) *	0.980	4157	2	400
Dimension (H2) **	0.580	290,498	1	401
Variety * Dimension (H3) **	0.726	75,354	2	400
Variety * Sample (H4) **	0.945	11,574	2	400
Variety * Dimension * Sample (H6) **	0.958	8849	2	400
Dimension * usual language (H8) *	0.980	2715	3	401

* p < 0.05; ** p < 0.01.

shows the Wilks Lambda and p-statistics for the confirmed hypotheses.

Paired samples t-tests (H₁–H₄) and univariate ANOVA (H₆) were used to find the subgroups responsible for the significant differences in these six hypotheses. All the tests described are significant at a level p ≤ 0.05.

6.3.1. Results for H₁–H₃

For H₁ (Figure 5), the mean of E (3.55) is higher than that of the dialect varieties (GH: 3.27; GHS: 3.39), among which GHS is the best rated. As for H₂, the value of the solidarity dimension (3.67) is significantly higher than that of status (§ 7.).

On H₃, the t-test for related samples (

Table 14. Contrasts of means for variety by dimension.

Contrasts	t	fd
GH (solidarity vs. status) **	−23,958	411
GHS (solidarity vs. status) **	−27,481	411
Solidarity (E vs. GH) **	−6213	411
Solidarity (E vs. GHS) **	−6213	411
Status (E vs. GH) **	16,571	411
Status (E vs. GHS) **	15,634	411

** p < 0.001.

) confirms that the means of GH and GHS are significantly lower than those of E in status (orange columns in Figure 5) and higher in solidarity (blue columns). The vernacular pronunciation scores converge with each other and move away from the standard in both dimensions, somewhat more markedly in status (mean difference between E and GH = 0.66, and between E and GHS = 0.72). Moreover, the mean differences between the dimensions are significant only for the dialect varieties, not for the standard, which is rated equally in both dimensions.

6.3.2. Results for H₄ and H₆

Figure 6 shows a significant improvement in the evaluation of dialect varieties in S-2015 (H₄), especially for GHS, but not for E. The differences between the groups are significant in both cases: (i) GHS: f (1 fd) = 19,473, p = 0.000; (ii) GH: f (1 fd) = 6.89, p = 0.009.

Table 15. Contrasts between varieties by sample year.

	Contrasts	t	fd
S-2000	E vs. GH **	7976	284
	E vs. GHS **	5248	284
	GH vs. GHS *	−2093	284
S-2015	E vs. GH **	2837	126
	GH vs. GHS **	−4105	126

* p < 0.05; ** p < 0.01.

shows the significant contrasts for both samples (vertical part of Figure 6 at each corner). In S-2000, the evaluation of the varieties follows the pattern E, GH, and GHS, but in 2015, GHS and E are better evaluated than GH. In S-2000, all contrasts are significant, but in S-2015, the improvement of GHS ratings converges with that of E (GHS = 3.6; E = 3.5).

Finally, the relationship between variety, dimension, and sample (H₆) is confirmed. Although the general pattern for each sample is the same as for the aggregated samples in Figure 5, in S-2015, the higher rating of dialect varieties in relation to the standard in the solidarity dimension (mean close to 4) is more clearly reflected —see right-hand side of Figure 7.

The ANOVA shows a significant relationship between sample year and the scores assigned to GH in the solidarity dimension (f = 8688, p = 0.003 for 1 fd) and to GHS in status and solidarity (status: f = 22,238, p = 0.000 for 1 fd; solidarity: f = 8688, p = 0.003 for 1 fd).

Table 16. Contrasts for variety by sample year.

	Contrast	t	fd
S-2000	GH (Solidarity vs. Status) **	17,797	284
	GHS (Solidarity vs. Status) **	23,632	284
	Solidarity (E vs. GHS) **	3428	284
	Solidarity (GH vs. GHS) **	5311	284
	Status (E vs. GH) **	−14,137	284
	Status (E vs. GHS) **	−15,257	284
	Status (GH vs. GHS) **	−3539	284
S-2015	GH (Solidarity vs. Status) **	−15,933	126
	GHS (Solidarity vs. Status) **	−14,162	126
	Solidarity (E vs. GH) **	−6821	126
	Solidarity (E vs. GHS) **	−6821	126
	Status (E vs. GH) **	8647	126
	Status (E vs. GHS) **	5748	126
	Status (GH vs. GHS) *	−2373	126

* p < 0.05; ** p < 0.01.

shows the significant contrasts for variety by year of the sample shown in Figure 7.

7. Discussion and Conclusions

7.1. Gheada and Seseo as Rural Indexes

The urban/rural opposition is confirmed as the main explanatory factor for the distribution of pronunciations associated with the sociolinguistic variables (g-h) and (θ-s), which function as Labovian indicators or first-order indexicals (Silverstein 2003, p. 194). The use of gheada and seseo is more prevalent in less populated municipalities with an agro-fishing economy. Additionally, young people living in villages within these municipalities tend to use these dialects more frequently. This trend is supported by the evolution of its use over time: it remains or increases in less populated towns such as Muros and Negreira but decreases in favour of the standard variety in more populated or dynamic areas like Padrón. The evolution of the use of the local variety (GHS) is noteworthy in this sense: in Padrón, it decreases by 32.5 points in favour of E; in Negreira, it increases by 26.2 points to the detriment of E and GH. The strong association of GHS with rural areas makes it an index of rurality, although, paradoxically, it is the GH variant that is most often used as a register of types of persons (or “persona style”, cf. Eckert 2008, p. 456) in performative uses of the language in Galicia (probably because it is more marked and widespread).

The results of the study show that speakers of the GHS dialectal area have a greater awareness of the use of gheada compared to seseo. This confirms that gheada is a more prominent and noticeable pronunciation in the linguistic consciousness of speakers. The centrality of gheada is linked to its history of stigmatisation and proscription in the standard language ideology of Galician, as well as to the symbolic violence experienced by its speakers. However, it also comes from its indexical function of authenticity and from the transgression of the normativity regime (even in writing), discussed in § 4. This denotes a sociolinguistic change in which the ideologies of Galician re-vernacularisation are making their way.

7.2. Gheada and Seseo: Between Inheritance and Choice

Group loyalties are encoded through the choice of linguistic variants, both in contextualised practices and in metalinguistic statements about use. By choosing certain variants, individuals align themselves with the group associated with them.

Gheada and seseo are examples of inherited pronunciations but also of choice pronunciations. On the one hand, young people tend to maintain loyalty to these variants if they are used within their family’s Galician. On the other hand, a significant group (§ 6.2) chooses to reintroduce gheada and seseo, even though they are not direct family transmission variants: in S-2015, the recovery percentages doubled for (h) and quintupled for (s). Furthermore, around fifty Castilian speakers, mostly urban, claim to use (h), despite it not being present in their mother tongue. In these cases, they are not following their cultural background as Galician speakers do, but rather an idealised stylisation of ‘coolness’ (cf. Eckert 2008, 2014). This is a discursive choice through which they express their ideological alignment with a sociolinguistic stereotype identified with authenticity.

However, the categorical decrease in the use of gheada and seseo among young people (Table 7 and Table 8) leads us to consider the existence of a metapragmatic awareness of the contextual appropriateness of the variables (g-h) and (θ-s). Thanks to the official standardisation regime since 1982 and the action of the school, they acquire the function of style markers in the (h-s) dialectal areas, whereas (g)-(θ) specialises in marking formality (Iglesias 2020). It is possible that the school context in which the test was administered activated this association.

7.3. Gheada and Seseo: Between SLIs and VLIs

Factor analysis confirmed the existence of the postulated dimensions of status and solidarity, made up of the usual attributes in this type of study (Table 11). Attitudes were positive on both dimensions, although they improved on solidarity, which means that Galician, as the national and minority language of the community, evokes more solidarity than status regardless of variety. This conclusion aligns with other research indicating that young people tend to prioritise values of equality and fraternity over institutional values that lead to success (Garrett et al. 2003). Without considering the dimensions, all varieties were positively valued, but speaker E was rated higher than vernacular speakers, which denotes the social prestige of the standard. Among the dialects, GHS was the best valued, suggesting a tension between accepting the superiority of the standard and the recapitalisation of the more distant vernaculars, that is, between SLIs and VLIs (tension also observed by Regueira 2023).

Considering the interaction between variety and dimension (Figure 5, Table 14), in the aggregated samples, attitudes are more favourable for the standard in status and for the vernaculars in solidarity. That is, the standard speaker is considered more cultured, classy, intelligent, and successful than the dialectal speakers, who are valued as more spontaneous, sociable, friendly, sincere, and fair. These findings are consistent with previous research suggesting that dialects or accents associated with dominant or majority groups, high socio-economic status, or formal or public functions receive high scores on the status dimension, while those associated with minority groups, informal interactions, and family domains tend to receive higher scores on the solidarity dimension. (Knops and van Hout 1988; Luhman 1990; Giles et al. 1995; similarly, Soukup 2013b). On the contrary, they partially disagree with the attitudes found in Kristiansen (2009) among young people from Copenhagen, reflecting a great devaluation of traditional regional accents, totally dissociated from positive identity traits, while lower urban class dialects are highly evaluated.

A significant finding is that the attribution of solidarity/status social meanings to GH-GHS and E is not conditioned by the usual language or dialectal variants of the students. This demonstrates the absence of ethnocentric attitudes among young people, as well as the homogeneity of the linguistic value system of groups of peers. Our results differ from other sociolinguistic contexts, where each group evaluates its variety better in all dimensions when contrasted with non-standard varieties, but only in solidarity when contrasted with standard ones (cf. Giles and Powesland 1975; Woolard and Gahng 1990; Coupland and Bishop 2007, etc.).

In the aggregated samples, the results also confirm that the E speaker is immune to the status/solidarity opposition: attitudinal scores floating around 3.5 in both dimensions, a situation that has not changed in fifteen years. The evaluation of the Galician standard is, therefore, positive and stable both in attributes that facilitate social progress and group integration. This trend does not repeat with the dialectal varieties, significantly better rated in the dimension of solidarity (around a 3.9 average) than in status. The data partially confirm those found by Iglesias et al. (2017), since, unlike dialectal pronunciations, the phonetic standard is an unmarked variety in terms of the indexical associations it evokes, even in the (h)-(s) areas. Contrary to what the above authors found, no linguistic or residential group deviated from this trend.

We think that these results are related to the scalar subordination derived from the Galician sociolinguistic context. One of the varieties of the dominated language (E) dominates over the other two (GH-GHS), but at the same time, it is the vernacular of a dialectal area of Galician. This situation can appear as a script in the cognitive representations of the informants and influence the evaluative pattern observed. Thus, a background framework organised around two oppositions would be at work:

i.

The situation of Galician as a language dominated by Castilian, which favours the Castilian = status/Galician = solidarity dichotomy in collective representations.

ii.

The standard = status/dialects = solidarity opposition within Galician itself, activated after decades of planning and linguistic policies that favoured its prestige and social position.

The status/solidarity opposition is irrelevant in the evaluation of the standard speaker because of the combination of these two schemes. On one hand, evaluators accept SLIs as they are aware that the (g)-(θ) variants are considered the ‘correct’ ones, used in institutional, public, and academic spaces, and necessary for formal communication and social advancement. These factors explain their indexical association with status. However, they also recognise that Galician, represented by the standard, is a shared cultural heritage and a symbol of loyalty and ethnolinguistic belonging, which arouses feelings of solidarity, especially when the standard pronunciations are integrated into colloquial discourse. Lastly, the perception that the (h)-(s) variants are not learned but acquired natively and are exclusively linked to the family register gives them an extra dimension of authenticity and local value, which makes their masks better qualified in terms of solidarity, even for those who do not use them. As Giles and Rakič (2014, p. 17) argue, “language cues (such as accents) can index multiple identities at different levels of abstraction”. The same linguistic index can serve to categorise the speaker as a co-group member or non-member, depending on the relevant reference frame for the evaluator. We do not rule out that Castilian evokes solidarity values as the native or adopted language of Castilian speakers (something we cannot know from this research). We assign it a role of a ‘linguistic over-roofing’ that should, in any case, be empirically verified by other studies. The

Table 17. Relationship between attitudes and indexical value of language varieties in Galicia.

	Symbolic Values	Evaluation
Castilian	Global Institutional Cross-community	+ Status (?) Solidarity
Galician-E	Institutional Ethnolinguistic identity Local identity (vernacular language)	+ Status + Solidarity
GH-GHS	Non-institutional Local identity (vernacular dialect) Authenticity	− Status + Solidarity

summarises the symbolic values of the three varieties.

One of the most relevant conclusions of this work has to do with the changes associated with the indexical values of dialectal varieties between S-2000 and S-2015, which confirms Coupland’s statement (2014, p. 81) that “ideological values for ways of speaking are liable to shift over relatively short periods of time”.

Between 2000 and 2015, dialectal pronunciations underwent a revaluation, which did not affect the standard (Figure 6). This revaluation is consistent with the increase in declared use of gheada and seseo in apparent and real time (Table 9 and Table 10). The revaluation affects both varieties, but it is more intense for GHS. In fact, the overall evaluation of GHS in the 2015 sample even surpasses the standard. In both S-2000 and S-2015, the mask with GH received the lowest rating. Therefore, gheada is ‘penalised’ if not accompanied by seseo. The reason may be that, in colloquial conversation, socially significant phonetic variants do not appear in isolation but together. As the co-occurrence of seseo and gheada is the dialectal pronunciation of the sample collection areas, the GH variant could be interpreted as a less authentic form: neither formal nor fully identified with the vernacular of the area. However, a differencialist attitude towards the standard, which GHS opposes doubly, could also have played a role.

Changes in the sociolinguistic meaning of dialectal varieties were observed. In S-2000, the mask rated highest in solidarity was GHS, followed by GH/E, both with the same score. In the status dimension, the ranking was in the order E, GH, and GHS (the last two were negatively rated). However, in S-2015, changes occurred in both dimensions. On the one hand, there was a levelling of the dialects’ scores in solidarity due to the better rating of GH. On the other hand, there was a reversal in the valuation of their status, with GHS now becoming the best qualified variety among the vernaculars (average above 3). While attitudes towards GHS have improved in both dimensions over the past fifteen years, attitudes towards GH have only improved in terms of solidarity. The use of positive symbolic connotations in these pronunciations suggests significant sociolinguistic changes that point to an ideology of vernacularisation, understood as “a shift towards a more positive valorisation of vernacularity, as well as a weakening or restriction of the standard language ideology” (Coupland 2014, p. 86). The tendency to reward dialectal differentiation relative to the standard is consolidated over time: it begins in 2000 with the superior valuation of the GHS speaker over the GH speaker in attributes that symbolise solidarity and social attractiveness and, in 2015, expands to include attributes associated with social success.

These results are in line with research that shows the upward value of stigmatised linguistic varieties in different sociopolitical contexts (Elías-Olivares 1976; Genesee and Holobow 1989; Woolard and Gahng 1990; Coupland and Bishop 2007; Kristiansen 2009; Ó Murchadha 2021; Grondelaers and Speelman 2013). They are related to ideological shifts that signal “a higher level of rebellion against dominant group values”, as Luhman (1990, p. 344) maintains, or “a glimmer of liberal sentiment”, in the words of Coupland and Bishop (2007, p. 85). Contrary to what happens with Danish in Copenhagen (Kristiansen 2021), in Galicia, it is the rural variants that are revalued thanks to the agency of the youth (also Beswick 2007; Thomas 2007; Formoso et al. 2017, etc.). The historical stigma to which the SLIs subjected gheada and seseo is challenged by their provocative emergence in a new ideological system in which ideals of equality, diversity, loyalty, authenticity, and democratisation are expressed in the form of a symbolic recapitalisation of marginalised pronunciations. The emotional adherence towards “deviant” pronunciations from the norm cannot be explained without taking into account the changes in the fields of indexical relations (Eckert 2008) that the standardisation of Galician produced, secondarily, among linguistic varieties and their speakers.

In Galicia, planning and linguistic policies have led to the emergence of focusing forces, symbolised by (g)-(θ). These are counterbalanced by differentiating forces—symbolised by (h)-(s)—arising from resistance to the linguistic uniformity of SLIs. Attitudes towards E show that there have been no changes in the perception of the standard, which is imbued with its own indexical load as a shared token of Galicianness. However, reinforcing the values of solidarity between two stigmatised dialects, and even the integration of GHS into the standard language, can be seen as a form of ‘value levelling’ (Coupland and Kristiansen 2011, p. 29). This means that the idea of a ‘best language’ is not completely eliminated but attenuated and made more complex.