Los políticos alternan de estilo entre variantes regionales y nacionales como una forma de construir una identidad coherente y atraer a los votantes. En España, investigaciones previas en el sur han mostrado cómo los conservadores tienden a favorecer variantes nacionales normativas, mientras que los socialistas emplean variantes regionales para conectar con las comunidades rurales y la clase trabajadora. Sin embargo, se ha investigado poco el discurso político en español en el norte del país. El presente estudio analiza cómo seis hablantes rurales y seis políticos de Galicia emplean la variación sociofonética, incluyendo grupos consonánticos, /n/ final de palabra, altura vocálica, /d/ intervocálica y /s/ en coda. Los resultados revelan que el discurso de los políticos refleja tendencias previamente observadas: el uso varía según el partido y el contexto social del hablante, siendo la elisión de /d/ intervocálica más frecuente entre políticos que entre hablantes rurales. Esto respalda hipótesis previas sobre la variación estilística en el discurso político del español peninsular. Además, al analizar fenómenos regionalmente relevantes, este estudio establece una base para el discurso político gallego, mostrando cómo las elecciones estilísticas se alinean con patrones más amplios en el español peninsular y demuestran el uso consciente del lenguaje para navegar la identidad y el posicionamiento sociopolítico.

Speakers often style-shift between prestige and non-prestige phonetic variants, but politicians in particular uniquely leverage language to reflect their social positions and appeal to voters. In southern Spain, studies by 1 and 2 reveal that politicians employ regional variants to connect with local communities, indexing regional identity and solidarity through sociophonetic variants. Leftleaning politicians in Spain have been shown to favor regional features, while conservatives instead opt to use normative national variants, reflecting distinct political goals. Similar findings in British and US English (e.g., 3 suggest these tendencies may be widespread in political speech, but further research across the Iberian Peninsula is needed.

The current study examines sociophonetic variation in the northwestern Autonomous Community of Galicia to explore whether the indexical tendencies identified here align with those observed in southern Spain. It analyzes regional features characteristic of Galician Spanish, including consonant cluster elision 4, velarization and reduction of wordfinal /n/ (e.g., Tomé 56, atonic raising of the mid-vowels (e.g., 67, as well as broader tendencies from across northern varieties of Peninsular Spanish, including intervocalic /d/ reduction 89, and coda /s/ retention 8SamperPadilla, 2011). Just as Spanish has been described to have had a major impact on urban prestige varieties of the Galician language (e.g., González 1011, the first four features in the previous list are in turn associated with norms in the Galician language, and as such are perceptually governed by a complicated field of indexical meaning and language ideology.

Galicia's political landscape is notable, in contrast to some other parts of Spain, for its three dominant parties: two national (left and right of center) and a left-leaning regional nationalist party. The Spanish 1 speech of twelve individuals -six rural speakers and six politicians (two from each party) -was analyzed to compare community and political speech. This study investigates correlations between political a�iliation and regional variation, as observed in southern Spain, and considers the influence of social factors such as gender and age on variable usage.

Sociolinguistic style was initially framed as attention paid to speech in early sociolinguistic research 1213, with less attention correlating with greater use of vernacular features. However, recent studies indicate that stylistic inventories are more varied, extending beyond the simple binary of vernacular versus non-vernacular 14. For example, 15 demonstrated how speakers' goals shape stylistic choices, allowing for coherent identity construction, in his analysis of a style-shifting Welsh radio DJ. Social meaning can be attributed to certain stylistic choices, as 16 describes, with salient features permitting interpretation and the conveyance of social meaning in language. Speaker agency also plays an important role in understanding styleshifting 17, as taking a stance shows accommodation toward or away from certain groups and can influence speakers' linguistic choices as a result. There is also an important interaction between the forms that speakers use and the meaning-based social system surrounding language, which 18 terms "bricolage." In this model, linguistic variables have a contextual meaning, connecting speakers with their communities of practice and serving to craft the self alongside other social signals like clothing and self-presentation.

In politics, "bricolage" describes how politicians use symbols like language to align with political parties and voters. As Hernández-Campoy and Cutillas-Espinosa (2010, p. 307) describe in their analysis of a female Murcian president, María Antonia Martínez, "the use of local features -very much associated with the working class world and with progressive ideas -might be a strategic way of building a particular image and projecting her socialist identity… [while] the use of standard features may be associated with conservative ideas and the accent of the bourgeoisie." In an interview with the former politician, the authors (2013, p. 96) find that she consciously developed a linguistic persona to represent herself as Murcian, using high rates of regional features to associate with her working-class background despite her careers both as a politician and a lawyer.

Meanwhile, in their analysis of the female Andalusian president Susana Díaz, Pollock and Wheeler (2022, p. 203) conclude that her reliance on certain regional features reflects "the realities of Díaz's identity as a speaker of a stigmatized dialect (although nonetheless a prestige variety within Andalusia), a 'militant' member of the Socialist Party, and a woman holding a position of major power in Spain." 19 further examined the extent to which individual factors and regional norms a�ect political speech as well, finding that specific goals and motivations also influence speech di�erences. Cruz-Ortiz (2022) analyzes diachronic data from Andalusian politicians over several decades, finding evidence of convergence toward Madrid norms when addressing northern audiences and, on the other hand, cases of individual style-shifting that were important in understanding language use. In all of these situations, socialist politicians used language as a means of showing belonging and of contrasting themselves with the normative tendencies of their opponents and peers.

Researchers have considered the influence of language ideologies and Galician indexicality on political speech, as well as regional attitudes toward Galician. 20 focuses on a speech by the at-the-time president of the region, Alberto Núñez Feijoo, emphasizing a metalinguistic debate regarding the use of Galicia (Castilian Spanish) versus Galiza (the historical form, favored by the left and nationalist parties). In the second half of this study, Amarelo then looks at perceptions of Galician on the gay dating app Grindr, showing how Galician serves as a locus of linguistic conflict in relation to users' decisions to use Spanish or Portuguese orthographic conventions. In a broader perceptual examination of Galician, González González et al. (2003) examine stereotypes surrounding use of the language, including by "neofalantes" who have recently embraced the language in urban areas, and nonurban speakers negatively evaluated as rural and vernacular.

As a result of these divisions in Galician use, 11 finds, Spanish political and discursive norms have been adopted into Galician. Politicians favor the urban variety of Galician, which includes Castilian phonetic influences on vowels, and a tendency to produce word-final nasals as alveolar, rather than velar. Part of the reason for these di�erences may stem from the social contexts in which Galician and Spanish are found. 21 discusses a tendency for Spanish to be associated with prestige and public speech domains historically, while Galician has been relegated to rural contexts. While Galician is a co-o�icial language in the region and is the dominant language in political speech, Castilian Spanish is heavily present in the mass media more broadly, as well as in religion, education, and urban communities. Based on this complex web of indexical meaning associated with the language, Galician presents a particularly charged field of indexicality, which has clear ramifications for its use in political speech.

Style-shifting in political speech is also evident in English-speaking contexts. Starr (2010, 2012) found that U.S. Congress members' pronunciation of Iraq reflected political alignment: Republicans used what they termed the "dismissive" [ae], while Democrats used the "European-influenced" [ɑ], showing divisions along party lines. Meanwhile, 22 showed that Barack and Michelle Obama's speech reflected both linguistic backgrounds and agentive adaptation to their roles, with Michelle displaying more African American English (AAE) features than Barack while conducting interviews as President and First Lady. Finally, in the U.K., Hall-Lew, Friskney, and Scobbie (2017) observed di�erences in CAT vowel production between Scottish National Party and Scottish Labor Party members, linking SNP politicians' more Scottish pronunciations to their secessionist stance. All of these studies suggest that regional sociophonetic variation in political speech reflects not only aspects of the current political situation, but also the social meaning of variants and how they could be applied in unique political contexts.

Galicia, the northwesternmost province in Spain, has a strong bilingual tradition in Spanish and Galician. This study focuses exclusively on sociophonetic variation in Galician Spanish, an area that has received limited sociolinguistic attention. Six phenomena were analyzed: four regional features were selected based on their status in Galician and the regular appearance in descriptions of frequent phonological Galician Spanish phenomena (e.g., Tomé 564-consonant cluster reduction, word-final /n/ velarization and elision, atonic /e/ and /o/ raising-and two features relevant to political discourse in other parts of Peninsular Spanish-intervocalic /d/ reduction and coda /s/ retention. However, as Regueira and Fernández-Rei (2020, p. 331) note, research on Galician Spanish has historically been sparse, with even recent focus being primarily placed on "Galician speakers who occasionally use Spanish," rather than on "Spanish speakers in Galicia." This complicates e�orts to describe trends belonging to the variety.

The first variable, consonant cluster reduction, involves simplifying stops, such as those in acto 'act' [akto] becoming [ato]. 4Rojo ( , p. 1094) describes this as a common phenomenon in many words and with various stops, including /b/ (abstracto → astrato 'abstract'), /p/ (corrupto → corruto 'corrupt'), and /g/ (magnifico → manifico 'magnificent'). Other related phenomena, including interdentalization of the first consonant (i.e., [akto] as [aθto]) or insertion of [j] or [w] in place of [θ], are identified by 4 in Galician Spanish and Martínez-Gil (2022) in Galician.

The second variable is word-final nasal velarization and reduction with vowel nasalization. Tomé Lourido and Evans (2021, p. 53) and 6 both note that velarization in word-final position is a typical Galician feature. While common in several Spanish varieties (e.g., Argentinean and Dominican 23) and described as being part of Galician Spanish 24, this process remains under-researched from a sociolinguistic and phonetic perspective in northwestern Spain.

The third and fourth phenomena-atonic /e/ and /o/ raising-arise from distinctions in Galician mid-vowels (Regueira & Fernández-Rei, 2020, p. 331 [ʊ]. These production di�erences may stem from contact with the Galician language (e.g., 25. De la Fuente-Iglesias and Pérez-Castillejo (2020, p. 316) note that /e/ is lower and /o/ is lower and more backed in Spanish than in Galician, suggesting that bilingual speakers may have greater mobility in their vowel space than monolingual speakers of central Peninsular varieties. 7 examine vowels and intonation contours, finding that while the seven vowel system of Galician is not reproduced in speakers' Galician Spanish, there is a reduction of final vowels that could suggest a maintenance of the covert prestige of Galician.

Intervocalic /d/ reduction, though well-studied from a phonetic perspective in Peninsular Spanish more broadly, has not received attention from a quantitative perspective for Galician Spanish. In Madrid, 8 and 26 identify the most frequent production of intervocalic /d/ as the dental approximate allophone [ð̞ ]. The influence of linguistic factors on this reduction process is considerable: in an examination of the phenomenon in Granada and Málaga, Villena-Ponsoda and Moya-Corral (2016) find that morphology and type frequency, among other linguistic factors, are more important in predicting variation than social factors. Others, such as Molina-Martos and Paredes-García (2014), have found that social and stylistic factors including speech style and topic are also important. While women tend to be more conservative regarding intervocalic /d/ production in Madrid 8, men often favor elision 27). This pattern is presumed to extend to Galicia. (Samper-Padilla, 2011, p. 100). Additionally, socioeconomic class, gender, and age stratify /s/ use: higher-status and younger speakers favor retention, whereas aspiration occurs more among older or working-class speakers (Samper-Padilla, 2011, p. 106). Villena-Ponsoda & Vida-Castro (2020) discuss this as a part of an ongoing change in Andalusian Spanish, blending central and southern Castilian norms as young, urban, middle-class speakers move closer to the national standard while retaining important regional markers that index belonging to an incipient speech community. As with the intervocalic /d/, this study expects that this stratification extends to Galicia, although this will be verified in the analysis.

Overall, these six phenomena were selected as representative of possible contexts for phonetic variation associated with Galician Spanish, in the hopes of capturing variables accessible for style-shifting purposes that politicians would employ in regular speech.

This study contrasts rural speech norms with political speech in Galician Spanish through a variationist lens in order to address three key questions.

First, it examines the linguistic and extralinguistic factors influencing regional variation. Previous research (e.g., 28292 suggests that gender and age shape regional variant use, while political a�iliation plays a crucial role among politicians.

Second, the study compares rural speech with that of politicians. While both groups belong to the same speech community and share general interpretive norms (e.g., 12, politicians are expected to show greater stylistic variability. Particularly those in Galicia from national parties had professional a�iliations that could lead them to produce more normative speech patterns, di�erentiating them from rural speakers, while individual motivations could increase variability in norms of production (e.g., 3019.

Finally, the study situates Galician political speech within a broader conversation on Peninsular Spanish identity work. Research on Murcian, Andalusian, and northern central Peninsular varieties highlights region-specific phenomena, but certain patterns, such as the tendency of female and left-leaning politicians to favor regional variants, associated with rural and working-class speakers, are hypothesized to hold across Spain (e.g., Hernández-Campoy and Cutillas-Espinosa, 2013; 219. Work by VillenaPonsoda, Vida-Castro & Molina-García (2022), for example, has shown how the combination of Andalusian linguistic resources leads to the development of identity, with certain forms taking on social meaning and speakers navigating changes over time to allow them to receive a specific type of evaluation from listeners. The current study investigates whether these broader tendencies extend to Galicia.

To allow for comparison with previous analyses of political speech (e.g., 2, twelve speakers were selected -six representing political speech in Galicia and six drawn from the Corpus Oral y Sonoro del Español Rural (COSER, 'Audible Corpus of Spoken Rural Spanish';31.

Among the politicians, one male and one female speaker were chosen from each of Galicia's three major political parties 32. Selection criteria included holding comparable political positions and conducting public interviews exclusively in Castilian Spanish. The parties represented were the conservative national Partido Popular 'People's Party,' the leftleaning national Partido Socialista Obrero Español 'Spanish Socialist Workers' Party,' and the left-leaning regional Bloque Nacionalista Galego 'Galician Nationalist Bloc.' To control for contextual variation and adopt a speaker-design approach 15, 40 minutes of unscripted interviews with male TV hosts were collected for each politician.

The rural data included four male and two female COSER speakers over the age of 60, residing within 100 miles of A Coruña 3 . These speakers were interviewed by a male Galician interviewer, minimizing variability in the speech context. Their inclusion was intended to test the theory by Hernández-Campoy and Cutillas-Espinosa (2010, p. 307) for southern Spain, which posits that politicians adopt local linguistic phenomena to signal solidarity with rural and working-class communities, particularly in left-leaning contexts. Older rural speakers from the COSER corpus are theorized to embody aspects of Galician identity that politicians may see as advantageous to emulate. Table 1 provides social information, including the name, political party, and position for politicians only, and the age, year of speech, gender, and regional origin of all speakers. Figure 1 maps the location of all speakers, corresponding to their numbering in Table 1.

Based on previous descriptions of Galician and Peninsular Spanish, eight contexts were targeted as either signaling association with the regional variety of Spanish of Galicia or providing a means for contrast with former studies of Madrid and Andalusian Spanish. The six phenomena included the elision of /k p b g/ in consonant clusters, the velarization of /n/ to [ŋ] in word-final position, the lowering of tonic /e/ and tonic /o/, and the raising of atonic /e/ and /o/ 45. Alternately, the two comparative phenomena included in the analysis were elision of intervocalic /d/ and retention of coda /s/, as previously studied in other varieties of Peninsular political speech (Hernández-Campoy & CutillasEspinosa, 2010;30219. The rest of this section focuses on the specific considerations and acoustic analyses used in the consideration of each phenomenon.

First, 4 , in words such as rector 'rector,' corrupto 'corrupt,' and dirección 'direction.' While instances of production, elision and interdentalization were sought out, only the first two were found in the audio data. As represented in Figure 2, cluster retention included the presence of a closure between the two stops, as in (a) prácticas, while elision included no such closure, as in (b) acciones. Next, for the velarization of word-final nasals, described by Tomé Lourido and Evans (2021) as characteristic of Galician Spanish, three productions were identified. These included (a) alveolar /n/, (b) n reduction and vowel nasalization, and (c) velarized [ŋ]. In addition to auditory identification of tokens, classification included inspection of the waveform and identification of F2 in the center 50% of each nasal token as a means of identifying the backness of productions of /n/. The nasal token was identified in Praat as an interval extending from the point where preceding vowel formants reduced in intensity to the end of F0 from the word, and the middle 50% was automatically collected in the Praat script by identifying the first and third quartiles of the segment and averaging F2 within that space. This choice was made following descriptions of average F2 as distinctive for other sonorants (e.g., /l/: Solon, 2015) -however, this acoustic measure did not prove descriptive in quantitatively tracking variation and was removed from consideration after not being able to predictively identify di�erences in nasal productions.

Next, raising in atonic /e/ and /o/ were identified by 4 and Tomé Lourido and Evans (2021) as typical features of Galician Spanish resulting in part from contact with Galician. This process was identified through a comparison of F1 (Hz) in tonic and atonic contexts, such as the di�erence between the tonic café [ka.'fe] 'co�ee' and atonic este ['es.te] 'this.' Tokens with measurable formants in Praat were identified in these contexts using a script in Praat following manual segmentation, and both F1 and F2 were measured in the central 50% of vowels to determine the extent to which raising occurred. To account for di�erences in the articulatory tract of each speaker, the Lobanov (1971) method was used to normalize all vowel productions. This method is used cross-linguistically in languages such as English (e.g., 33, contact varieties of Spanish (e.g., 34, and Peninsular Spanish (e.g., Henriksen, 2017) as a means of reducing variation 4 . The process is discussed in greater detail in Section 3.4.

With respect to intervocalic /d/, retention of [ð̞ ] is most common in northern Spain 835. However, considerable research has associated this phenomenon with political speech in other regions of Spain (Hernández-Campoy & CutillasEspinosa, 2010;36219. Politicians, particularly males, are considerably more likely to elide /d/ in intervocalic positions than members of the general populace. However, as most of this work has been performed in the central and southern regions of Spain, this study examines whether this tendency exists in other regions. While this is a gradient phenomenon, tokens were separated into two broad classification groups based on the degree of occlusion. Finally, the last variable considered in this study, retention of coda /s/, is also tied to northern Peninsular varieties of Spanish. Based on prior descriptions, retention was the expected regional norm in Galicia (e.g., 2635. As Vida-Castro, Villena-Ponsoda, and Molina-Martos (2022) show, social factors and type frequency play a role in variable use, particularly in regions where /s/ reduction is less common, as in the northern regions of the peninsula, since there is more room for, and social meaning associated with, stylistic variation in these contexts. The voiceless sibilant fricative was included in the current study to examine stylistic variation and o�er a comparison to studies of Murcian and Andalusian varieties. As seen in Figure 3, a binary classification of the variable was applied to /s/, distinguishing between (a) retention of the sibilant fricative, with considerable aperiodic noise around 4000-8000 Hz, and (b) cases of weakening or elision where little to no frication remained.

Following previous analyses of political speech (e.g., 2819, dependent variables were treated in a binary fashion, distinguishing "regional" and "normative" variants based on Galician Spanish phonetic norms, as depicted in Table 2. The following variants were treated as regional for the phenomena under consideration: consonant clusters (elision), word-final nasals (velarization and nasalized vowel production), tonic /e/ and /o/ (lowering), atonic /e/ and /o/ (raising), intervocalic /d/ (deletion), coda /s/ (retention).

Independent variables were then selected following categories identified in previous research, as well as those available in the COSER corpus (i.e., those social factors included in Table 1).

The variables selected in this study reflected findings in previous dialectological and sociolinguistic research related to political speech phenomena, permitting a contrast between regional Galician Spanish identity and a normative national one (e.g., 4285. Among the social variables identified in this study were speaker political party, gender, age, audience, and the timestamp when productions were uttered in interviews, which mapped onto a Labovian understanding of style 12. Audience was identified based on the news channel and topic where political speech occurred, with Galician-specific channels being identified as "regional," Madrid-centered channels termed "national," and those related to EU discourse labelled "international."

Eight additional linguistic variables were included, based on expectations of production norms, as shown in the Phenomenon column in Table 2. This included the tonicity of the syllable in which the phenomenon occurred, the duration of the segment, as well as the place of articulation for both the preceding and following context. Three quantitative acoustic measures were included for individual phenomena: intensity range for /d/, normalized F2 for the vowels, and position in word (for /s/). Finally, for the consonant clusters, the coda stop in the cluster was identified (i.e., /k/, /g/, /p/, or /b/).

It is hypothesized that speakers from the COSER corpus will be the most likely to use the variants coded in Table 2 as regional for Galicia, while politicians overall will favor intervocalic /d/ elision more, as 2 identified for Susana Díaz in Andalusia and Cruz-Ortiz (2022) discusses for both Madrid and Andalusia. As the right-leaning PP currently holds political power in Galicia, it is further proposed that speakers on the political left will have a greater tendency to use regional variants in Spanish as a means of distinguishing themselves from the establishment.

Curiously, unlike in Andalusia, a nationalist party also receives considerable support in Galicia -while previous research in Galicia does not o�er strong insight into norms surrounding Galician Spanish by politicians in the BNG to distinguish themselves from members of the larger two national parties in Spanish, their strong support for the Galician language and linguistic decisions within that language are hypothesized to play a role in their Spanish productions as well. As Cruz-Ortiz (2019, 2022) and Pollock (2023)

All data were analyzed using the Praat phonetic software, including a consideration of spectral cues, audible di�erences, and quantitative measurements 37. For each speaker, where possible, 100 tokens representing each of the eight phenomena were collected. In several cases where 100 was not reached, particularly in the COSER data where multiple speakers were interviewed at once and some speakers were further from the microphone or less responsive, as well as for the infrequently produced phenomenon of consonant clusters, all viable tokens in the audio were collected. In the continuous analysis of F1 for vowels, the Lobanov (1971) method was used, and resulting figures were then scaled following the R vowels package formula 38. This takes the normalized values and places them back into a range that is comparable to those expected from raw scores, allowing interpretation while continuing to account for anatomical and acoustic di�erences between vocal tracts.

After descriptive examination of di�erences within the data, the results were subjected to mixed-e�ects logistic and linear regression analyses with random e�ects to account for potential clustering in the data. The Rbrul program for R 39) was used to determine how extralinguistic factors influenced variable use.

In total, 8386 tokens were collected (Table 3). Elision of the initial consonant in clusters beginning with /p k g b/ was frequent, at 73%. Velarization of the word-final nasal was less common at 58%. Intervocalic /d/ elision was the least frequent phenomenon at 49%. Finally, coda /s/ retention was frequent, occurring in 80% of contexts. With relation to the vowel analysis, the overall average variation in F1 between tonic and atonic vowels is well below the perceptible 60 Hz level of di�erence described in previous research on English (i.e., 33. Trends in these phenomena are described in greater detail below, with FN1 signifying the normalized value of F1.

First, with respect to consonant cluster elision, members of the PSOE were found to be most likely to retain, while politicians from both the BNG and PP elided considerably more, and speakers from COSER had almost categorical elision (Figure 4). Across all speaker cohorts, elision of /k/ -the most frequent consonant in these clusters -was highest. Following that, for velarization of word-final /n/, Figure 5 represents the divisions between the three allophones. While use of the alveolar [n] remained steady, nasalized vowel production was most frequent among COSER speakers, particularly men, while rates of velarized [ŋ] were highest use among members of the BNG and PP parties, reaching 50%. Production of word-final /e/ in tonic and atonic contexts, depicted in Figure 6, show that di�erences in F1 do not reach a level of perceptibility. However, atonic productions of /e/ do show slight raising across all speakers except for those in the PSOE. In particular, rural female speakers in the COSER data had the greatest di�erence in F1 across contexts, approaching 40 Hz. Next, production of word-final /o/ in tonic and atonic contexts followed the same trends as for /e/ ( Figure 7). PSOE politicians showed higher than expected productions of /e/ in tonic syllables, and female COSER speakers had the greatest distinction across contexts. These di�erences also did not reach 60 Hz in any single subdivision of the speakers. Following that, there was a marked distinction between female speakers in COSER and the other speakers for intervocalic /d/ (Figure 8). The highest rates of elision occur among politicians in the PP and BNG, surpassing rates of elision among rural male speakers and PSOE politicians. Rural female speakers elided in around 20% of cases, while politicians from both the conservative and Galician nationalist parties are near 60%. Finally, with respect to coda /s/, politicians retained in around 90% of cases, while COSER speakers had lower rates of reduction ( Figure 9). Rural men in particular reduced in nearly 40% of cases. Similar to word-final /n/ velarization and intervocalic /d/ reduction, members of the PSOE were closest in their productions to COSER speakers. Given the myriad variables influencing these phenomena, as discussed previously by 19, there are inherent problems in attempting to develop a single model to examine their variation. While some phenomena trend inversely to others with respect to regional identity, in other cases the social meaning inherent in one allophonic pair is completely unrelated to that identified in another. For that reason, six mixed-e�ects regressions were run as a means of comparing trends within the data, with significant factors from these models represented in Table 4. Tendencies within the model are described in the table, while at the bottom, vital statistics from each mixed-e�ect regression are provided to clarify the number of tokens, degrees of freedom, log-likelihood, fixed R 2 and total R 2 values in each model.

For consonant clusters, all speakers except those in the PSOE favored elision, as did shorter segments, clusters beginning with /k/, preceding coronal sounds, and older speakers. Second, for velarization of word-final /n/, when comparing the velarized variant to both elided and alveolar productions, [ŋ] was favored preceding dorsal sounds and pauses, following dorsals, in speech produced by politicians, and in longer segments. Third, for the raising of /e/, using a continuous scaled FN1, the linear regression determined that raising was favored in word-final position and atonic syllables, as well as by younger politicians, older rural speakers, in regional-targeted speech, earlier in interviews, and in conjunction with fronting of normalized F2. Fourth, raising of /o/ tended to occur in final and atonic contexts, in regional and rural speech, among older speakers, in shorter segments, and accompanied with backing of normalized F2. Fifth, for intervocalic /d/, elision favored word-final position before a following word-initial vowel (e.g., ciudad y 'city and'), as well as in shorter segments and with reduced ranges in intensity. Finally, coda /s/ tended to be elided or reduced in final position, when followed by labial and coronal sounds, among rural speakers and for regional and national audiences, and in conjunction with a lower COG (Hz) for the segment.

The first research question examined the linguistic and extralinguistic variables influencing variation in the sociophonetic phenomena. Linguistic variables were grouped into general and phenomenon-specific categories. General variables included word position, following and preceding context, and segment duration. Vowel raising and reduction of intervocalic /d/ and coda /s/ were favored in word-final position. Following context a�ected /n/ velarization and coda /s/ reduction, with dorsal sounds and pauses favoring velarization, while labial and coronal sounds promoted elision. Preceding context also mattered: dorsal sounds encouraged /n/ velarization, and coronal sounds favored consonant cluster elision. Shorter segments correlated with cluster elision, intervocalic /d/ reduction, and /o/ raising, while longer nasal segments favored /n/ velarization.

Five phenomenon-specific variables also conditioned variation. Coda /s/ retention correlated with center of gravity (COG), where a lower average frequency indicated aspiration or elision (where no COG was present). Reduced intensity range (dB) between /d/ and the following vowel predicted /d/ reduction. Normalized scaled F2 and lexical stress were significant for vowels, with raising more frequent in atonic positions and linked to the fronting of /e/ and backing of /o/. Finally, consonant cluster elision was most common with the most frequent stop, /k/.

Four social factors, including political a�iliation, age, audience, timestamp, and their interactions, also shaped variation. Politicians more frequently velarized /n/ and raised /e/ compared to rural COSER speakers, but PSOE politicians uniquely avoided cluster elision. Older speakers across groups favored cluster elision and vowel raising. An interaction between age and political a�iliation influence /e/ raising, with younger politicians and older COSER speakers favoring this pattern. This could be seen as relating to the discussion of Galician indexicality discussed by González 10 and 20, in which young urban speakers associate Galician language proficiency with Galician identity. This vowel production norm may be associated both with the older rural speakers who maintain Galician speech and the younger politicians who aim to reclaim it.

Audience also a�ected variation: regional audiences favored /e/ raising, while national and international contexts encouraged /o/ raising and /s/ retention. Specifically, international audiences prompted normative northern central Peninsular Spanish /s/ retention. Finally, timestamp influenced /e/ raising, which was more likely earlier in interviews.

The second research question explored di�erences between politicians and rural Galician Spanish speakers. While the selected phenomena are associated with regional Galician Spanish (e.g., 4; Samper-Padilla, 2011), previous research (e.g., 7 has focused on bilingual and Galician-dominant contexts, leaving unclear how these features manifest in everyday Spanish among rural and political speakers. It was hypothesized that rural COSER speakers would favor consonant cluster elision, /n/ velarization, and atonic vowel raising, while politicians would exhibit higher coda /s/ retention (expected at ~90%, based on Madrid norms, as described by 19 and greater intervocalic /d/ reduction, as these phenomena have been found to be more accessible for style-shifting in prior studies of political speech (e.g., 361940.

The data largely supported these hypotheses, as shown in Table 5, which summarizes Figures 4 to 9. Speakers in the COSER corpus, particularly men, showed the highest rates of consonant cluster 5 elision, /n/ reduction with vowel nasalization, /s/ reduction, and vowel raising, with women favoring vowel raising over the other phenomena of consonant elision or reduction. Politicians, in contrast, exhibited lower rates of these regional features, with velarized /n/ more common than nasalized vowel production. Intervocalic /d/ reduction deviated from this pattern: male COSER speakers followed PP and BNG politicians in its frequency, while PSOE speakers showed the lowest rates.

Correlations emerged across political a�iliations. Conservative speakers from the national PP exhibited the fewest regional features, except for higher consonant cluster elision, and had the highest rates of intervocalic /d/ reduction. By contrast, left-leaning PSOE and BNG speakers showed comparable patterns: low rates of consonant cluster and /n/ reduction, reduced coda /s/, and an unexpected rise in normalized F1 for vowels in atonic contexts.

It is important to point out that the rise for mid-vowels did not exceed the perceptibility threshold of 60Hz described for English vowels 33)-a comparable measure for Spanish vowels has not been established-making it hard to determine the extent to which this may be an incipient hypercorrection, moving away from expected Galician Spanish raising norms. While studies of perceptions of vocalic variation in Spanish are not extensive, research on English vowels by 41 and others have examined how dialect change can be tracked through incipient changes in vowel formant measures that nonetheless do not reach the level of perceptibility. Leaving that possible 5 Due to a small number of consonant clusters produced by COSER women, the two genders were combined. phenomenon aside, it is also noteworthy in and of itself that these politicians are not raising vowels in atonic final position, and that instead di�erences in vowel production are negligible. All told, this suggests the presence of distinct linguistic strategies across political a�iliations, warranting further investigation.

Considerable research in Spanish and English highlights the role of speaker agency in crafting linguistic identity, particularly in political speech. From vowel patterns among Scottish Labour and National Party politicians in London 3 to coronal stop deletion in by Barack and Michelle Obama 22, and hyper-vernacular regional usage among Murcian 1 and Sevillian politicians 2, there is ample evidence of intentional linguistic strategies in politics. This study extends these observations to Galician Spanish, which has received little sociophonetic examination, to determine how closely regional use reflects use in other Peninsular varieties. A comparison between intervocalic /d/ and coda /s/ production, in particular, is conducted in Table 6 for research in Murcia, Andalusia, northern central Peninsular Spanish and Galicia.

Comparison with southern Peninsular phenomena, like intervocalic /d/ and coda /s/ reduction, reveals regional distinctions. In Galicia, intervocalic /d/ elision rates were higher among politicians (53.7%) than rural speakers (42.9%), aligning with trends in Madrid and Andalusia, where political speakers favor more elision than their rural counterparts. For coda /s/, retention was notably higher in Galicia, with politicians retaining /s/ at 90.6% compared to 67.5% for rural speakers. This gap contrasts with Andalusia and Murcia, where /s/ reduction is nearly categorical, making it less usable for stylistic variation. These findings suggest a stigma attached to /s/ reduction in Galicia, as political speakers align more closely with Madrid norms than their rural counterparts. In comparison to other Spanish varieties, political a�iliation is similarly influential in describing Galician speech norms. Unlike Andalusia, where Pollock (2023) noted a shift toward rural vernacular features among right-leaning politicians as part of a populist appeal, Galician PP politicians maintained more normative productions, particularly for vowels, nasals, and coda /s/. This divergence reflects Galicia's unique political landscape, and the fact that while parliamentary speech is conducted in the Galician language, as Regueira (2016) and 21 both discuss, Castilian Spanish holds a prestige position in the region, present across mass media, and playing a major role in the political and discursive tendencies of Galician as well. While Andalusia has recently experienced significant political shifts, with the rise of altright Vox and a tendency for more rural voters to support right-wing parties, Galicia's centerright PP has maintained power in five of the last six terms 42. While it has been theorized that shifts in Andalusian speech are the result of recent political changes, and a jockeying on the part of VOX and PP politicians to win support among traditionally PSOE voters 1943, Galicia finds itself in a di�erent political situation. Unlike in Andalusia, where left-leaning political support was strong for decades, right-leaning politicians in Galicia do not need to employ rural vernacular features as they are not in need of the same populist strategy to find success. Additionally, the correlation of vernacular and Galician features with the left gives further rationale for these features to be more strongly favored by members of the PSOE and BNG.

Further, Vox's drop in Galician voter support during the 2023 elections 44 underscores a potential disconnect between rural populist strategies and northern voter priorities. Future research should examine whether this reflects an outright rejection of Vox's political approach or simply a resistance to populist rhetoric in Galicia. These observations suggest that Galician political speech maintains closer ties to regional identity and normative prestige as was more common in the south in previous decades, in contrast with the evolving trends of rural populism in present-day southern Spain. This highlights the importance of localized political realities in shaping linguistic strategies.

This study identifies a system of linguistic variation across rural and political speakers in Galician Spanish, providing evidence to support hypotheses on variation in Peninsular Spanish political speech. Rural speakers favor high rates of Galician-coded variants, while left-leaning PSOE and BNG politicians follow, and conservative PP use more normative variants, aligning with trends in northern central Peninsular Spanish and diverging from phenomena related to the Galician language. This mirrors findings by Hernández-Campoy and Cutillas-Espinosa (2010) about 1990s Murcian political speech, where local features index working-class solidarity and socialist identity, while normative features suggest conservative ideologies. It also reflects ideological tendencies described by González González (2003), 20, and others, whereby Galician is associated with local identity, nationalism, covert prestige, and the left.

These results o�er a baseline for both sociolinguistic trends in Galician Spanish as well as in the political speech of Galicia. Intervocalic /d/ elision is more frequent among politicians than rural speakers (and higher for BNG and PP-a�iliated politicians than for rural male speakers), further suggesting that this trend is a nationwide norm (e.g., 3019. Similarly, /n/ velarization 6 was linked to PP speakers, while /n/ reduction with vowel nasalization was associated with rural speech. COSER speakers and older men favored regional variants more than women and younger speakers, consistent with sociolinguistic patterns described for vernacular and standardized linguistic variants (e.g., 12, as well as with the norms observed in other political contexts (e.g., 12.

Finally, the vocalic phenomena in this study present a deviation from previous descriptions of Galician Spanish. While statistically significant di�erences in vowel height were observed, they fell below the threshold described in English (i.e., 60 Hz: 33 and Spanish (i.e., 60-100Hz: Herrero de Haro, 2017). This raises questions about whether vowel raising in Galician Spanish is su�iciently perceptible to carry social meaning, warranting further research using perceptual methodologies. Additionally, these findings suggest less marked variation in vowel height among Spanish speaking politicians in Galicia than has previously been found among bilingual or Galician-dominant speakers (e.g., 611, emphasizing the need for broader sociolinguistic research on Galician Spanish. It also appears that the phenomenon of /n/ velarization is more complex than previously described in the Galician Spanish literature, with multiple socially stratified allophonic variants. Future studies should distinguish between Spanish speakers in Galicia and Galician speakers using Spanish, as 7 recommend, and include bilingualism as a variable to further parse distinctions.

Alongside a need for more research into Galician Spanish, there is also a pressing need for studies of far-right political speech in Spain. While the current study included the BNG, a heretofore understudied regional party, most other work on Peninsular Spanish political speech has focused on established national parties, such as the PSOE and PP (although see 4543, for a comparison to Vox). The rise of Vox and the ideological shifts in parties like Ciudadanos from the center-left to right in recent years highlight the importance of examining how populist politicians use language to appeal to constituents in novel ways. As seen in Andalusia, co-opting vernacular features typically associated with the left could serve as a strategy being used by the right to attempt to attract rural and working-class votes. Determining whether similar trends exist in Galicia or other northern regions would expand our understanding of political speech norms, and how these are shifting across ideological boundaries.

Overall, this study establishes a baseline for Galician Spanish political speech, showing regionally salient variation and the agentive use of stylistic variants. From a quantitative perspective, both linguistic and social variables predict the use of the six sociophonetic phenomena, revealing consistent political identities shaped by accommodation to or divergence from local norms. Members of some right-leaning parties may employ normative national variants to project worldliness and education, while certain left-leaning politicians make use of regional forms to index solidarity and in-group belonging. However, these tendencies are complicated. These findings show how language serves as a critical tool in navigating identity and sociopolitical positioning within the Galician speech community.