The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/

JUAN-ANDRÉS VILLENA-PONSODA Universidad de Málaga, Spain Fields of research: sociolinguistics, social dialectology, phonological variation, Spanish, Andalusian varieties. THE DILEMMA OF THE RELIABILITY OF GEOLINGUISTIC AND DIALECTOLOGICAL DATA FOR SOCIOLINGUISTIC RESEARCH. THE CASE OF THE ANDALUSIAN EMERGER OF /θ/ Geolingvistikos ir dialektologijos duomenų patikimumo dilema atliekant sociolingvistikos tyrimus. /θ/ skilimo Andalūzijos ispanų kalboje atvejis ANNOTATION Over theoretical and methodological differences between, on the one hand, dialectology and geolinguistics and, on the other hand, sociolinguistics make quite difficult the essential task of using data from the former two disciplines to help the latter one complete present research on the speech communities, particularly when it is about reconstruction of the past of current varieties. The problem is therefore to which extent are geolinguistic and dialect monograph data reliable to retrospectively follow both the history of a given variety and a particular linguistic change. Building on rich results of complex analyses using macrosocial, mesosocial and small-scale variables to explain speech variation, which reveal emergence of a new variety in southern Spain, retrospective comparison with geolinguistic and dialect monograph data discloses ideological drawbacks regarding representation of both the speech community and the linguistic structure, as well as a quite troubling contradiction between geolinguistic and dialect monograph data. KEYWORDS: speech variation, sociolinguistics, dialectology, Spanish, Andalusian varieties. Straipsniai / Articles 9 JUAN-ANDRÉS VILLENA-PONSODA ANOTACIJA A kivaizdūs teoriniai ir metodiniai dialektologijos, geolingvistikos ir sociolingvistikos skirtumai apsunkina labai svarbią užduotį – dviejų pirmųjų mokslų duomenis panaudoti taip, kad jie būtų padedama trečiajam mokslui atliekant kalbinių bendruomenių tyrimą, ypač siekiant atkurti esamą kalbos atmainų istoriją. Taigi, nagrinėjama problema susijusi su klausimu, kiek geolingvistiniai ir monografiniai tarmių duomenys yra patikimi, siekiant retrospektyviai nustatyti tiriamos kalbos atmainos istoriją ir konkretų kalbos pokytį. Remiantis gausiais macrosocialiniais, mezosocialiniais bei miklymens kintamuosius įtraukiančiu sudėtu tyrimu, kuriais siekiama paaiškinti kalbos įvairovę, rezultatas, atskleidžiančiais naujos kalbos atmainos atsiradimą pietų Ispanijoje, atliktas retrospektyvinis palyginimas su geolingvistiniais ir monografiniais tarmių duomenimis parodo ideologinius trūkumus, susijusius su kalbinės bendruomenės ir kalbinės struktūros reprezentavimu, taip pat ir nerimą keliantį prieštaravimą tarp geolingvistinių ir monografinių tarmių duomenų. ESMINIAI ŽODŽIAI: kalbos įvairovė, sociolingvistika, dialektologija, ispanų kalba, Andalūzijos ispanų kalbos atmainos. INTRODUCTION Since data obtained in geolinguistics and dialectology and those from current sociolinguistics have been conceived under very different theories on language and society, reconstruction of the past of a linguistic variety in formation – as well as its relationship with other varieties in the speech community – using geolinguistic and dialect monograph data leads to a dilemma. Although the importance of the findings obtained by traditional dialectologists and geolinguists is evident, it should be emphasised that sociolinguistic data are based on sociological assumptions about sampling and variation that are incompatible with those underlying geolinguistic and dialectological data, particularly the oldest and outdated atlases and monographs (Villena-Ponsoda 2010). The problem is then to which extent geolinguistic and dialect monograph data are reliable to retrospectively follow the history of a given variety and a particular linguistic change.1 1 Data and results discussed in this paper are based on the Spanish Ministerio de Economía y Competitividad DIGYT Research Project on the Sociolinguistic Patterns of Castilian Spanish (ECO-PASOS, FF2015-68171-C5-1 and FEDER funds). I am indebted to Francisco Díaz-Montesinos, who scrutinised the ALEA maps, organised and made comprehensible tables from a huge mass of geolinguistic data, and to Pilar García-Mouton, for her invaluable help and support. Matlde Vida-Castro, Francisco Díaz-Montesinos, Goduno Chela-Flores and Álvaro Molina-García kindly revised previous versions of this paper. I am, however, responsible for its manner and content. 10 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ This paper2 aims at describing a new urban variety in formation which involves a certain number of changes – some of them of a surprisingly fast development – revealing convergence towards the national standard variety and boosted by action of current main-prestige values and attitudes (Vilena-Ponsoda 2008). Apparently, this variety could have originated in Andalusia around the mid-fifties of the 20th century since only few traces of its main features were by then documented by dialectologists and geolinguists. Results of analyses currently carried out show that now we are dealing with a coherent variety, which is perceived as such by its speakers (Vilena, Vida 2017, 2018), and could have emerged as the outcome of the trend of modernisation and urbanisation that took place in southern Spain during the last decades of the 21st century and, above all, after Franco’s rule (1939–1975). Emergence of intermediate varieties between vernacular dialects and the national standard is quite a common process in the European context, and it seems to be related to the fact that urban educated middle-class speakers need means of expression different from both the national standard – too formal or even posh – and vernacular varieties – too rude and socially marked (see Auer 2005: 27–28). Although it is true that an Andalusian spoken regional-standard actually exists – based on the urban dialect of Seville and meant to fulfil the social functions ascribed to the national standard – its influence does not reach any longer urban areas from eastern (Granada, Malaga) or far western (Huelva) Andalusia (see Map 2 below), so that middle-class speakers from these areas trying to flee the vernacular tend to shift to the national standard and adopt its main features (Vilena-Ponsoda 1996, 205). The most interesting trait of this trend of convergence is that it seems to be related to the individual speakers’ profiles, in particular with regard to their motivation to acquire knowledge, their degree of media exposure and awareness of the flow of information in society (i.e. Bourdieu’s notion of the so-called incorporated cultural capital), more than merely to the cultural capital acquired at home or at school, i.e. the institutionalised and the objectified cultural capital (Bourdieu 1977, 1984, 1986). Then, the idea of a proactive speaker who assesses standardisation as a means of social mobility and integration emerges and proves to be of the greatest interest to explain how the new variety has developed so far. 2 The present paper is a revised version of the text of my presentation to the IX Congress of the International Society for Dialectology and Geolinguistics, SIDG (23–27 July 2018, Vilnius, Lithuania). I am grateful to the organisers for their kind invitation and particularly to Jurgita Jaroslavienė from the Research Centre of the Standard Language at the Institute of the Lithuanian Language for her support and understanding. Straipsniai / Articles 11 JUAN-ANDRÉS VILLENA-PONSA Reconstruction of the changes underlying this variety across real time reveals itself as a matter of importance but, at the same time, brings to light the recurrent problem of the reliability of the dialectological and geolinguistic data for sociolinguistic research. The organisation of this paper is as follows. In the first section (The intermediate variety) the new Andalusian urban middle-class variety in formation is described, focusing particularly on its phonology vis-à-vis both the national standard variety and the Andalusian vernacular dialects. The second section (The social meaning of the new variety) analyses sociolinguistic variation and coherence of the intermediate variety, with emphasis on the effect of macrosocial variables on the speakers’ behaviour. In the third section (The proactive speaker. The importance of indirect effects) we deal with the effect of intervening mesosocial and small-scale variables between macrosocial and dependent speech variables – i.e. the speaker’s probability of using the new variety. We show that it is the speaker’s standard orientation – based on his or her capacity of being informed and connected with the outer world beyond the local speech community – what determines his or her decision of using a coherent set of features whose social meaning is related to modernity and standardisation. The fourth section (The dilemma) approaches the central issue of this paper, namely, the limitations of dialectological data of the past to reconstruct the historical development of today’s varieties. To do so, two main points are considered. First, our knowledge of the past speech community, which leads to fundamental methodological differences between, on the one hand, dialectology and geolinguistics and, on the other, sociolinguistics. Second, the problem of the accessible linguistic data, whose observation reveals contradictions between geolinguistic data and those from dialect monographs. Finally, the fifth section (Areoling) briefly accounts for an ongoing project which intends to obtain real data from the time when the variety emerged, based on public or private spoken documents. THE INTERMEDIATE VARIETY The ongoing formation of an intermediate urban middle-class variety between the Spanish national standard (Madrid) and the southern varieties in Andalusia has been understood as the outcome of the trend of leveling and convergence towards the current prestige norm of speech used in Spain by middle-class speakers, whose traditional southern identity has been changing since the mid-fifties and getting close to modern identities of globalisation, 12 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ modernisation and urbanisation (see Villena-Ponsoda 2008; Hernández-Campoy, Villena-Ponsoda 2009; Regan 2017a, b; Villena-Ponsoda, Vida-Castro 2018). This particular trend experienced a sharp rise after the era of Franco (1939–1975), when transition from a mostly rural, traditional and local ideology to modernisation took over (Dietz 2001). The linguistically most interesting point here is that this variety involves a synthesis – to a certain extent, contradictory – of social meanings of (1) modernisation (i.e. standard orientation) and (2) tradition (i.e. traditional southern identity). To do so, a combination is made of national standard features in syllable onset position with southern vernacular features in the coda. This combination is meant to let speakers symbolically flee either the rural or urban working-class background – i.e. in doing so, the intermediate-variety speakers intend to abandon speech features related to social meanings of the ‘traditional’, the ‘oldest’, the ‘rural’, etc. – and, moreover, consciously accept standard features, even though they can entail stopping and even reverting multisecular changes and/or acquiring new phonological contrasts. At the same time, they preserve a great deal of the southern accent – the so-called acento andalú (Andalusian accent), which is particularly identified with unmarked erosive variation processes of the lenition and deletion of codas. Thus, as argued, social meanings of standardisation and modernisation, on the one hand, and those representing loyalty to the southern community values, on the other, can be, to a certain extent, compatible. MAP 1. Mosaic of contemporary Peninsular Spanish dialects and languages (Hernández-Campoy, Villena-Ponsoda 2009: 184). Straipsniai / Articles 13 JUAN-ANDRÉS VILLENA-PONSA On its way towards the national standard variants, the Andalusian urban middle-class variety speakers meet dialect users from southern Castile, Murcia and Extremadura (see Map 1). These dialects share a good deal of features with the Andalusian varieties so that a continuum has traditionally been proposed to represent the geolinguistic variation at stake here (Navarro-Tomás, Espinosa, Rodríguez-Castellano 1933). Therefore, the idea of an intermediate supra-regional variety would apply not only to the sociolinguistic dimension, but also to the geolinguistic space (see Villena-Ponsda, Vida-Castro 2017). Both the linguistic convergence and the underlying new blended identity involve the acquisition, among others, of some standard features, notably splitting of coronal fricative /θ/. This change implies reconstruction of the Middle-Age Castilian contrast – which merged in the early Andalusian dialects since the 13th century – between word sets with apical /s/, (casa ‘house’) and laminal /s/, previously affricate /ts/ (caça ‘hunting’), which has been maintained in the standard variety through contrast between apical /s/ [ˈkaṣa] and (inter) dental /θ/ [ˈkaθa] (see Table 1 below): TABLE 1. Inventory of obstruent consonants in the Andalusian vernacular (left) and in the intermediate variety (right) labial dental palatal velar p t tʃ k tense b d j g lax f θ h fricative labial dental palatal velar p t tʃ k tense b d j g lax f ṣ x fricative tense θ lax 14 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ The splitting of Andalusian /θ/ implies reconstruction of the abovementioned contrast between Middle-Age word sets with /s/ and with /ts/, then /s/, by imitation of the standard. From what we know now, the demerge of /θ/ must have been a hard individual process involving the building-up of an acoustic multi-parametric distance between both fricatives (Regan 2017b; Molina-García 2018). This phonemic contrast is currently used by young educated urban speakers. Scrutiny of data from linguistic atlases, notably the ALPI or Atlas de la Península Ibérica ‘Atlas of the Iberian Peninsula’ (first quarter of the 20th century) and the ALEA or Atlas Lingüístico y Etnográfico de Andalucía ‘Linguistic and Ethnographic Atlas of Andalusia’ (early fifties), as well as the results from dialect monographs describing some urban dialects (notably Alvar 1974, 1993 [1973]), suggests that /θ/ splitting has been a quick change from above. This idea is supported by the fact that very few – if any – traces of its existence have been detected up to the eighties (see Carbonero 1982; Sawoff 1980; Salvador-Salvador 1980; Villena-Ponsoda 2001). The change has spread rapidly over the urban centres of Andalusia, particularly in the eastern part of the area, and affected the lowest status groups only partially (Villena, Ávila-Muñoz 2014; see Map 2). MAP 2. Urban Andalusia A set of other parallel features showing the same trajectory would have likely soared by the same time. These features have been actually analysed so that Straipsniai / Articles 15 JUAN-ANDRÉS VILENA-PONSA researchers are acquainted with the idea of the emergence of an intermediate – relatively coherent – variety between northern standard-close dialects and southern standard-distant dialects (see Vilena-Ponsa, Vida-Castro 2018). In short, it could be said that, in addition to the emerge of /θ/ – which is, of course, the socially most salient change among the set – the intermediate variety adopts either standard or hybridised variants able to avoid vernacular uses (Trudgill 1986; Chambers, Trudgill 2004 [1980]; Siegel 2010 [2006]; Berruto 2005; Almeida 2018; see Table 2). TABLE 2. Phonological changes. Mergers and demergers (adapted from Vilena-Ponsa, Vida-Castro 2018) Varieties Word sets: <θa> <sa> <χa> <tʃa> <da> Standard ˈkaθa ˈkasa ˈkaxa ˈkatʃa ˈkada Intermediate ˈkaθa ˈkasa ˈkaxa~ˈka ˈkatʃa ˈkada~ˈka Vernacular /θ/ = /s/, seseo ˈkasa ˈkasa ˈkaxa~ˈka ˈkatʃa~ˈka ˈkada~ˈka /θ/ = /s/, ceceo ˈkaθa ˈkaθa ˈkaxa~ˈka ˈkaθa~ˈkaθa ˈkada~ˈka /θ/ = /s/ = /χ/ = /ð/, heheo ˈka ˈka ˈka~ˈka ˈka ˈka /θ/ = /s/ = /tʃ/, extended seseo ˈkasa ˈkasa ˈkaθa~ˈka ˈkasa ˈkada~ˈka Gloss hunting house box buttock each As shown in Table 2, on the one hand, the new variety stops vernacular processes, including either frequent and relatively accepted mergers of coronal fricatives (seseo, ceceo) or even their most evolved outcomes (heheo, extended seseo); on the other hand, it adopts phonologically-marked mainstream-prestigious variants (affricate [tʃ], fricative [χ] and approximant [ð], instead of their vernacular unmarked counterparts (fricative [ʃ], aspirated [h] or deleted /χ/ and deleted /ð/). THE SOCIAL MEANING OF THE NEW VARIETY How this new variety has spread out and to what extent we can talk about a coherent variety is a relevant matter of discussion. However, a crucial issue to 16 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ be pointed to is the social profile of its speakers. Results of previous research tend to indicate that it is urban young – mostly female – university-graduate speakers who lead the use of the set of features shaping the intermediate variety, which seems to spread both socially and geographically (Villena-Ponsoda 2001; Moya-Corral, Sosinski 2015; Regan 2017b). Unexpectedly, the contemporary image of the Andalusian speech communities shows growing divergence between two sets of varieties; namely, on the one hand, traditional urban working-class and rural vernacular dialects, which are relatively out of the influence of the current standard-trend of convergence and, on the other hand, the new intermediate middle-class variety (Villena-Ponsoda, Ávila-Muñoz 2014). Features being part of this variety correlate in such a way that their use seems to be affected by coherence restrictions. Variables related to the standard values and attitudes tend to occur in contexts and be used by speakers who are unlikely to promote variables representing vernacular social meanings, and the opposite is also true (see Table 2 above). Then, co-occurrence of the single variables form clusters where we can identify more complex underlying variables or components, able to give us an idea of their common distribution, which is in fact compatible with the notion of a coherent variety (see Guy, Hinskens 2016). Given our purpose of exploring the existence of a central-southern (standard-vernacular) intermediate variety and since our objective is to prove its coherence, a good way of proceeding could be to identify clusters of variables on the basis of their similar variance. Doing so, we would be able to seize which variables are likely to co-occur and the kind of speakers who would use them – i.e. which features shape a particular variety and their social distribution. To find out which features gather according to their respective variance, a factor analysis was carried out3. Based on multiple correlation, two principal components or underlying clusters of variables were found, which are actually able to explain cross-dependence between the individual variables. In other words, these complex underlying variables, components or factors would represent the common basis for the single variables. The overall variation of the data was reduced then to two dimensions or factors that explain 69.5% of the variance (39.1 and 30.4, respectively). A Principal Components Saturation Matrix is shown in Table 3, where the proportion of variance for each variable explained, respectively, by factor 1 and 2 is displayed. Figure 1 below shows the relative position of each single variable in the ideal space delimited by both factors giving us an idea of how they correlate. 3 In the frame of a broader research project, we worked with a sample of 72 speakers of both genders (m = 36, f = 36), stratified by education and age. Fieldwork for this study was carried out during the first decade of the 21st century. See Villena-Ponsoda 2008. Straipsniai / Articles 17 JUAN-ANDRÉS VILLENA-PONSA TABLE 3. Principal Components Analysis. Matrix of rotated components Matrix of rotated components Words Variables Components: F1 F2 <caja> Del_x ,796 ,265 <cad> Del_d ,784 ,158 <casa> Cec eo ,769 −,034 <caza> Seseo −,252 ,913 <caja> Asp_x ,437 ,695 <caja> Fric_x −,673 −,627 <dos, esto> Faithfulness −,647 −,156 Split −,400 −,777 Method of extraction: Principal Components Analysis. Method of rotation: Varimax with Keiser normalization. Rotation converged in 3 iterations. Faithfulness is the variable representing the proportion of code maintenance [dos], [doh] dos ‘two’ against erosive changes affecting every consonant in syllable-final position, either in intervocalic or final position [eto], [eθo] esto ‘this’. See also Table 2. Factor 1 (Standard orientation) opposes standard-oriented variables (in Figure 1, left, negative values) to vernacular variables (right, positive values). The standard variables – which either stop (Fric_x, Faithfulness) or even reverse merges (Split) – are related to social meanings of standardisation, urbanisation and globalisation, while vernacular variables – which likely keep merges (Del_x, Del_d, Ceceo) – convey local and rural values of solidarity. Each set of variables are highly correlated in such a way that their internal correlation – as Figure 1 clearly suggests – supports the idea of two separate varieties related to two different identities. Factor 2 (Regional orientation) seems to represent a different separate dimension, perhaps related to the speaker’s regional orientation, including southern features compatible with those of the spoken standard from urban Seville (Asp_x, Seseo). Given the location of each speaker on both dimensions (standard and regional orientation), based on regression by estimation of individual factor scores, it is possible to spot all of them on the bi-dimensional space defined by F1*F2 and correlate their scores with the speaker variables. Thus, we have here the two fundamental bases for the definition of a particular variety, namely (1) a set of features which co-occur and (2) are socially distributed in a similar way 18 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ (Hudson 1996 [1980]: 22). As we are particularly interested in the use of the variables which are part of the intermediate variety, we will focus on the F1 dimension henceforth. (1) Faithfulness: coda maintenance. Fric_x: fricative [x]. Split: Demerger of /θ/. (2) Seseo: sibilant [s] allophones of merged /θ/. Aspir_x: aspirated [x]. (3) Del_x: deleted /x/. Del_d: deleted /d/. Ceeo: non-sibilant [θ] allophones of merged /θ/. FIGURE 1. Correlation of variables in the standard/regional orientation space As expected, education, gender and age are significant factors constraining the use of the intermediate variety (see Figures 2 and 3). Speakers with non-compulsory educational levels – regardless of their age – favour the use of the new variety (negative values) but even among speakers with low levels of education, the youngest are beginning to accept the standard-like uses (Figure 2). As for the effect of gender (Figure 3), female speakers’ scores have also predominantly negative values on F1, which suggests their preferential use of the standard features. All of this is in accordance with the idea of an elite change strongly conditioned by social class and particularly by the standard orientation of speakers. The question now is to know why the speaker variables related to the macrosocial dimensions – and particularly social class – are not enough to satisfactorily explain a great deal of the individual behaviour (the proportion of variance explained by extralinguistic factors does not attain 45%). Straipsniai / Articles 19 JUAN-ANDRÉS VILENA-PONSODA F = 22.825 (2), sig. < 0.001, R² = 0.421 FIGURE 2. Education and age differences on the use of the intermediate variety. (source: Vilena, Vida 2018) F1. Gender, p < 0.001 F2. Gender, p < 0.005 FIGURE 3. Effect of gender differences on the use of the intermediate variety 20 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ THE PROACTIVE SPEAKER. THE IMPORTANCE OF INDIRECT EFFECTS One of the most striking facts regarding the intermediate variety is how macrosocial dimensions and mesosocial entities interact to explain the exact nature of the proactive speaker responsible of its spread. As seen from Bourdieu’s theory of cultural capital and social market (Bourdieu 1984, 1986), the incorporated cultural capital (capital incoré) acquired by speakers who nowadays are interested in the flow of the media and the Internet information, acts as an intervening variable multiplying the effect of the objectified and the institutionalised capital (capital objectivé and capital institutionalisé), i.e. respectively, the knowledge acquired at home and the one learned at school, in the broadest sense of the word. The picture so displayed tends to underestimate direct effects of the speaker’s education and that of their parents’ on the dependent linguistic variable while, at the same time, underlines their indirect influence through mesosocial and even small-scale variables. To account for the effect of the speaker variables on the standardisation dimension underlying the use of the intermediate variety, a multivariate model capable of presenting direct and indirect effects on a particular variable would be appropriate. This kind of model allows us to show the effect of one variable which is controlled by that of the other variables, as it is the case with the Structural Equation Model (SEM), also known as Structural Mediator Analysis and Path Analysis. This analysis fits quite well this objective (see Paxton, Hipp, Marquart-Pyatt 2011). SEM is a model based on multiple lineal regression analysis that estimates not only direct effects by certain variables (x, y) on the endogenous or dependent variable (w), but also indirect effects by the same variables or by others across intervening variables (z). So, a diagram composed of different direct (x, w) or indirect (x, z, w) paths between independent (exogenous), intervening and dependent (endogenous) variables is drawn: β (x, w) Stripsnai / Articles 21 JUAN-ANDRÉS VILLENA-PONSODA To represent the diverse paths of influence – indirectly through intervening variables, or directly on the endogenous variables – we hypothesised a path model that focuses on the effect of the speaker’s status (measured by his or her education), controlled by that of their parents’ (mother’s education and father’s occupation). With this, we tried to grasp Bourdieu’s idea (1977, 1984, 1986) of the capital culture institutionalised (education) as determined by the capital objective (i.e. the household context of acquisition). However, even though both types of cultural capital are responsible for a relevant part of the individual’s position in the social, cultural and linguistic market, he or she is also able to obtain and increase a third type of capital, the capital incorporé, which may be seen as the outcome of a personal inversion by the actor (libido sciendi) and affects, among many other things, the individual’s attitudes towards the media and linguistic standardisation in society. FIGURE 4. Structural Equation Modeling on the speech use In addition to this, the speaker’s life and social capital depend not only on the kind of local community (open or closed to external contacts) where he or she lives but also to what extent his or her social network ties are close-knit (Trudgill 1996). However, it has clearly been shown that the effect of the network strength on the speech use is constrained by the speaker’s social status (Villena-Ponsoda 2005). That is the reason why the speaker’s network score was included in our model as an intervening variable. Finally, as seen before (see Figure 2), age – indirectly through education – and gender affect our linguistic variable. In a way, then, as shown below (Figure 4), variables representing the objectified and the institutionalised capital (1) control both the incorporated 22 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ capital and the social capital represented by each individual’s network score (2), and ultimately constrain the speaker’s speech use (3). The analysis consists of four structural equations – i.e. standardised linear regression equations (see Table 4), whose endogenous variables are the following: first, the speaker’s institutionalised capital ‘Education’ (Eq 1), second, the speaker’s incorporated capital ‘Media/Standard orientation’, third, the individual’s network of ties ‘Network’ (Eq 3) and fourth, the linguistic dependent variable ‘Speech use’ (Eq 4). TABLE 4. Direct and indirect effects on the use of the Intermediate Variety Effects on F1 Indirect Direct Total x1 x2 x3 x4 x5 x6 x7 x1*x5 x1*x7 x8 x1 Education – –.210 –.059 –.293 –.562 x2 Mother’s educ. –.121 – –.087 –.058 –.266 x3 Father’s educ. –.106 – –.076 –.051 –.233 x4 Age .063 – .076 .139 x5 Media/Stand. – –.276 –.276 x6 Gender – –.384 –.384 x7 Network – .272 .272 All effects are significant (p < .05). Gender (x6) is a dummy variable, where 0 = male, 1 = female Equations Eq1. Education x1 = x2 + x3 + x4 + e1. R² corr. = .504. Anova (3, 66) F = 24.344, sig. < .001 Eq2. Media/Standarness x5 = x1 + e2. R² corr. = .579. Anova (1, 64) F = 90.231, sig. < .001 Eq3. Network x7 = x1 + e3. R² corr. = .253. Anova (1, 64) F = 23.034, sig. < .001 Eq4. Standarness = x1 + x5 + x6 + x7 + e4. R² = .608. Anova (1, 55) F = 29.020, sig. < .001 Error: √1 − R²; e1 = .70; e2 = .65; e3 = .86; e4 = .63 Explained variance: 1-(e4)² = .60 As shown in Table 4, negative effects on the vernacular speech use – and hence, favouring the use of the intermediate variety – pertain not only to variables representing objectified and institutionalised capital (x1, x2, x3), but also to the variable x5 (incorporated capital). Meanwhile, gender (x6) affects also negatively (i.e. pointing to a female tendency) the linguistic variable. The only positive effects on the speech use here – and hence, favouring the use of the vernacular varieties – are those of age (x4, indirect through x1 and x1*x5) and network scores (x7). Negative and positive effects have to be understood then in the frame of the results of the factor analysis carried out above, where positive values indicate vernacular use on the F1 dimension. Straipsniai / Articles 23 JUAN-ANDRÉS VILLENA-PONSODA The underlying path analysis reveals then an image of complex interactions between variables (see Figure 5 below). On the one hand, the overall negative effect of education (x1) (Beta = −.562) is explained by both its direct influence (Beta = −.293) and indirectly through the intervening variable x5, Media/Standard orientation, Beta = −.210): the higher the speaker’s education and the higher their scores on Media/Standard orientation (Beta = .765), the lower their position on the F1 dimension (Beta = −.276). On the other hand, education acts also as intermediary of the indirect influence of the variables representing the context of linguistic acquisition (x2, Beta = −.26 and x3, Beta = −.233) and the speaker’s age (x4, Beta = −.139). Direct effects of these variables on the speaker’s education are positive but indirectly, through education as an intervening variable, they negatively affect F1 or just minimise their original positive influence. Finally, the speaker’s network has not only a direct positive effect on F1 (Beta = .272), but also almost cancels the effect of education (−.059). All effects are significant (p < .05). Gender (x6) is a dummy variable, where 0 = male, 1 = female x1 Speaker’s education x2 Mother’s education x3 Father’s education x4 Age x5 Media/Standard orientation x6 Gender x7 Network strength x8 Speech use FIGURE 5. Diagram of the multi-causal process of the effects of the speaker’s variables on speech use 24 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ Results of the path analysis are then compatible with the idea of the emergence of an intermediate variety between the national standard and the Andalusian vernacular dialects. Speakers who currently use speech features characterizing this variety have low scores on the underlying variable F1 that represents vernacular indexicalities. As easily deduced from results displayed in Table 4 and Figure 5, this new urban variety is spoken mostly by young educated and well-informed standard-oriented female speakers with loose-knit personal networks. THE DILEMMA Bearing in mind such a complex process of sociolinguistic variation, the reconstruction of the past of the intermediate variety seems to be an exciting challenge. However, the central issue now is whether the available data of the past we have access to are complete and satisfactory enough to answer the main questions raised in the course of the research on the present. This is because the data of the past were obtained using the methods of sampling, fieldwork and analysis which are not always compatible with those used to deal with the complexity of contemporary speech community data. In spite of the enormous rigour and generosity distilled from their work, there are profound underlying ideological differences between the results obtained by dialectologists and geolinguists and those achieved in the frame of the present-day sociolinguistic paradigm. On the one hand, basically, in the field of both dialect geography (direct representation of local variation using vocabularies and maps, from Wenker to Gillerón through to the contemporary projects) and dialect monograph (reconstructing the prototypical local dialect, from Ascoli and Wegener through to modern sociolinguistics), dialectologists have carried out an essential and indispensable work. However, their aim was to show that laws of linguistic change are exceptions and that le patois authentique of a local speech community is the purest form of a language, worth preserving from external contamination and studying as a reservoir of original purity (see Villena-Ponsoda 2010: 614–615 et passim). On the other hand, data of the present have to do with a very different outlook, which reflects a dramatic and real change in society. In fact, adopting Durkheim’s (1902 [1893]) conceptual contrast between mechanical and organic solidarity, the transition from traditional to modern industrial communities can be connected with the evolution of methods. Durkheimian mechanical solidarity supporting traditional rural closely knit communities, and reflecting uniformity of the speaker’s états de conscience, was replaced by the contemporary urban sharply stratified loose-knit communities based on organic solidarity and division of labor (Berthele 2004: 732–734). The modern society presents an Straipsniai / Articles 25 JUAN-ANDRÉS VILLENA-PONSA image of heterogeneity which only extralinguistic correlates (occupation, education, age, gender, etc.) could help to organise. Therefore, ideological differences between both paradigms became firstly theoretical and, eventually, methodological. Dialectology and sociolinguistics finished up as separated from each other as the community types they were meant to describe. Within this framework, two issues should be considered when comparing – as it is the case here – contemporary and historical data of such a different type: (1) The speech community. What do we know about the speech community of the past? (2) The actual linguistic data. How was the linguistic structure we know today organised in the past? The speech community dilemma. Data from the Malaga urban speech community regarding the key time-period we are concerned with here are available only from the Atlas Lingüístico y Etnográfico de Andalucía, ALEA (1961–1973). Fieldwork for this atlas was carried out during the 50’s (see Alvar 1955; García-Mouton 1992; Morillo-Velarde 2001) using “multiple and repeated questionnaires” and “optimal subjects” (Alvar 1959: 10, 27–30). Since the idea was – following Gilliéron, Pop, Puscariu and others – to obtain [...] “the snapshot of the speech in a particular moment”, [...] we sought the person able to be the subject of this “linguistic photograph”, avoiding stridency – not the eldest, the most personal or the newest data – just the average, although sometimes it may be necessary to have recourse to the oldest people to pick up a declining phenomenon or a disappearing piece of craftwork. That is the reason why a man between fifty and sixty years old will project very likely the precise image of what we are looking for (Alvar 1959: 29, my translation). To design samples for cities and some of the biggest towns, local areas and the informant’s age, gender and social class were considered as a source of variability, following Jaberg and Jud (1928–1960) procedures for the AIS (Jaberg 1936: 20; see also Jaberg 1959). However, at the end of the day, and in spite of the best intentions, only four subjects were investigated in Malaga city (more than 270 000 inhabitants by then).4 In fact, although this particular atlas contains a good deal of information about the sociolinguistic variation of the time including the so-called “polimorphism” or unconscious variation – the task 4 Three men, one of them defined as a university graduate and the other two as illiterate, and one university graduate woman; no particular details of their age are provided. 26 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ of locating and gathering this information from each map out the six volumes of the atlas is exhausting, and direct data on sociolinguistic differences – taken from parallel questionnaires with different subjects – are not always clearly separated from observations by the researcher based on the informant’s linguistic awareness (see Morillo-Velarde 2001: 15–21 et passim). As a matter of fact, a systematic scrutiny of particular sociolinguistic correlations between speech variants and speaker variables for a certain urban locality within the atlas reveals scarcity of occurrences and, hence, lack of firmness of conclusions. Nonetheless, the main trouble is the existence of two different ways of presenting dialect data. On the one hand, we have the geolinguistic data that are accessible for each locality and each phonological, morphological, lexical or syntactic variable. This data were provided by one particular NORM informant for each point in the map – with supplementary data by a woman for a certain number of localities and variables – and, as mentioned above, with variation at least of gender and education, in the case of cities and the biggest towns (see Alvar 195: 29–30). Of course, this sort of data are direct and undoubtedly affected by the skewness linked to the NORM speaker sampling method. On the other hand, the dialectological data, which consist of a set of additional observations on the social meanings related to the speaker’s status, age or gender. These observations are included both as supplementary notes within the maps themselves or by the use of symbols with glosses on the maps’ margins. Based apparently both on these two sources of data and on additional recordings and qualitative observations, dialect monographs were accomplished and then dialect monograph data are also available (Alvar 1993 [1973], 1974). This type of data is influenced then not only by the same ideological and methodological bias as the geolinguistic data, but also by the direct effect of qualitative judgments by the researchers themselves. The dilemma of the linguistic data. Based both on the overall ALEA informants’ profile (men between 50 and 60 years old) and on the information regarding the speakers studied in Malaga city (4 speakers, 1 university-educated woman and 3 men, one of them university-educated and the other two illiterate), it is likely to be ascertained that, in accordance with the apparent-time distribution commented on precedent paragraphs, among these speakers – who would be 85–90 years old by the time when the fieldwork for the Malaga intermediate variety study was carried out – the illiterate male informants would not use the standard-like variants, while the university graduate speakers (man and woman) partially would. (Details on these speakers are shown in Table 5). Stripsniai / Articles 27 JUAN-ANDRÉS VILLENA-PONSODA TABLE 5. ALEA informants in Malaga city (source: adapted from Alvar (1993 [1973]) Informant | 45 | 50 | 583 | 506 Local area | City centre | City centre | Semi-urban | Semi-urban Locality | Malaga | Malaga | Churriana | Iznate Gender | Male | Female | Male | Male Education | University | University | Illiterate | Illiterate The University-educated informants (45 and 50) provided responses only to the phonetic questionnaire, while the illiterate informants (583 and 506) responded to all the items of every part of the whole questionnaire. A “linguistic outline” consisting of four or five pages – which was included in the phonetic part of the questionnaire – as well as a complete page of phonetic transcription from each informant’s speech, were prepared including also recordings and platograms. This material is not available. The image of the speech community based on the data provided by these speakers is, of course, very limited but one would expect that the geolinguistic and dialectological information given was close enough – and not contradictory – to allow the reconstruction of the historical trajectory followed by the changes leading to the intermediate variety. However, both sources of data do not match. Of course, as pointed out above, it would be expected – given the results obtained in the contemporary Malaga study, based on data collected by the early years of the 21st century – that we had to face evidence, on the one hand, of a strongly-marked vernacular variety used by the illiterate speakers and, on the other hand, of a less marked and more standard-like but clearly southern-accented variety used by the university-educated speakers. However, although this is precisely the image that emerges when we consider the dialecto-logical data, this is not so clear when the geolinguistic data are taken into account. When we focus on the most relevant features of the intermediate variety (see Table 2 above), the dialectological information provided by Alvar (1993 [1973] – which is the unique dialect monograph dealing with the urban dialect of Malaga of the first half of the 20th century – confirms the current hypotheses about the set of quick changes enhanced by convergence towards the national standard and supporting the emergence of the new urban intermediate middle-class variety. Although all the vernacular features mentioned by Alvar are related to the illiterate speakers, with the exception of merger of /ʎ/ = /ʝ/ (the so-called yeísmo), which is accepted as normal by university graduates, his description does not suggest that the changes leading to the present-day situation were in progress at the time (Table 6). In fact, the most accepted opinion among the scholars by the mid-fifties was that the Andalusian dialect seemed to be socially homogenous: 28 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ Regarding its social diffusion, the Andalusian variety does not have any limits: it reaches the highest groups in society. And when I call them ‘the highest’, I mean the cultured groups, which are the most likely either to level or to resist diffusion. Pronunciation by a literate Andalusian speaker is clearly similar to that of his or her illiterate fellow citizens. Lexis and procedures of expression may change but pronunciation is the same. In fact, to study the phonetics from Granada, university-graduate speakers were investigated, even university professors (Salvador 1967 [1963]: 64, my translation). On the other hand, an exhaustive scrutiny of the geolinguistic data from the ALEA reveals a rather different image. The information derived from the Atlas and the one provided as a result of qualitative analyses by dialectologists tend to contrast. The geolinguistic image from the ALEA is compatible with a scenario where most of the changes that actually shape the intermediate variety were in progress already by the 50’s, particularly among the urban middle-class speakers. Table 6 displays a comparison between both sources of information. Features as relevant for the characterisation of the intermediate variety as the contrast between /s/ and /θ/ help us understand both positions. Alvar (1993 [1973]) points out that the most frequent pattern of realisation of merged /θ/ among literate speakers is sibilant fricative [s] (seso), while illiterate speakers tend to use non-sibilant fricative [θ] (cece). However, results from scrutiny of the ALEA maps show that variation seems to have started by the time of fieldwork for the atlas, including some traces of the splitting of /θ/ (see Table 7). TABLE 6. Educational differences in the use of the Malaga vernacular features (adapted from Alvar (1993 [1973]) and ALEA) Dialect Monograph data | Geolinguistic data Standard | Illiterate | University | Illiterate | University /s/ /θ/ | poso/pozo | Ceceo [ˈθeθo] | Seseo [ˈseso] | Ceceo/Seseo/Split | Seseo/Split /tʃ/ | pocho [ˈpotʃo] | Fricative [ˈpoθo] | Affricate [ˈpotʃo] | Affricate | Affricate -θd- | dedo | Deletion [ˈdeθo] | Maint. [ˈdeθo] | Variation | Maintenance -θd- | dado | Deletion [ˈdaθo] | Deletion [ˈdaθo] | Deletion | Variation The same can be applied to the rest of features in Table 6. The compared dialectological and geolinguistic results speak for themselves (see also Tables 7–9). The apparent homogeneity of the speech behaviour projected by the dialect monograph analysis is to a certain extent contradicted by the results obtained after a thorough study of the maps, including quantitative account of the responses. Strainsni / Articles 29 JUAN-ANDRÉS VILLENA-PONSODA TABLE 7. Educational differences in the realisation of word sets with <s> and <z, ce, ci> in Malaga (source: adapted from the ALEA) University | Illiterate Male | Female | Male | Male Etim. | [s] | [θ] | Total | [s] | [θ] | Total | [s] | [θ] | Total | [s] | [θ] | Total s | 13 | 1 | 14 | 14 | 0 | 14 | 9 | 258 | 267 | 204 | 38 | 242 92.9 | 7.1 | 100.0 | 0.0 | 3.4 | 96.6 | 84.3 | 15.7 z, ce/ci | 5 | 26 | 31 | 35 | 1 | 36 | 0 | 316 | 316 | 149 | 115 | 264 16.1 | 83.9 | 97.2 | 2.7 | 0.0 | 100.0 | 56.4 | 43.6 Total | 18 | 27 | 45 | 49 | 1 | 50 | 9 | 574 | 583 | 353 | 153 | 506 TABLE 8. Educational differences in the realisation of intervocalic /d/ in Malaga (source: adapted from the ALEA) University | Illiterate Male | Female | Male | Male /0d/ | [d] | ∅ | Total | [d] | ∅ | Total | [d] | ∅ | N | [d] | ∅ | Total 8 | 2 | 10 | 8 | 2 | 10 | 32 | 69 | 101 | 43 | 54 | 97 80.0 | 20.0 | 80.0 | 20.0 | 31.7 | 68.3 | 44.3 | 55.7 /#d/ | 2 | 5 | 7 | 4 | 3 | 7 | 3 | 106 | 109 | 1 | 92 | 93 28.6 | 71.4 | 57.1 | 42.9 | 2.8 | 97.2 | 1.1 | 98.9 10 | 7 | 17 | 12 | 5 | 17 | 35 | 175 | 210 | 44 | 146 | 190 58.8 | 41.2 | 70.6 | 29.4 | 16.7 | 83.3 | 23.2 | 76.8 TABLE 9. Educational differences in the realisation of /tʃ/ in Malaga (source: adapted from the ALEA) Literate | Illiterate Male | Female | Male | Male [tʃ] | 16 | 76.2 | 23 | 100.0 | 136 | 97.8 | 116 | 87.2 [ʃ] | 5 | 2.8 | 0 | 0.0 | 3 | 2.2 | 17 | 12.8 Total | 21 | 23 | 139 | 133 30 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ Before this contrast of results between dialect monographs and geolinguistic work, key questions about the emergence and development of the Andalusian intermediate variety remain unanswered. We really do not know whether most of the phonological variables which currently shape this variety were absent from the speakers’ repertoire by the mid twentieth century and suddenly and unexpectedly arose boosted by certain social and cultural changes or whether they had been developing – maybe since the old times – and, hence, they were already observable in that same period as processes of socially conditioned variation. In the first case, we would be facing an example of a quick change from above, perfectly analysable using data from spoken corpus since the last decade of the 20th century. Cross-sectional apparent-time analyses would then be appropriate and their results would probably match with data from dialect monographs. In the second case, we would need more detailed re-analyses of the geolinguistic data and systematic comparison with current sociolinguistic results. Longitudinal trend studies based on, let’s say, archeological data would then be desirable. Although a discussion about the exact nature of the latter sort of data would be the subject of a whole paper, a brief outline is summarised in the next paragraph. ARCHEOLOGY In addition to the well-known disparity of methods and theoretical assumptions between dialectology/geolinguistics and sociolinguistics – not to mention again the underlying ideological perspectives involved – our retrospective attempt to identify clear traces of living sound changes in dialect monographs and linguistic maps has come across a supplementary source of trouble. Subjective assessments of dialect use meet the NORM speaker bias and make comparison of present-day data with those derived from dialectological and geolinguistic work a difficult task. Contemporary work on these traditional disciplines has completely changed the situation. Outstanding results in both fields will make things easier in the future. Nevertheless, now we have to cope with the Andalusian panorama as it is. Re-analyses of geolinguistic and dialect research data, as well as gaining access to broadcast radio and television recordings and documentary film archives, would allow us to (1) scrutinise in real (though severely biased) time the expansion across society of the currently ongoing changes and (2) explore also the ideological lag affecting this kind of retrospective trajectories. Assessment of selection criteria underlying NORM speakers in geolinguistics and traditional Straipsniai / Articles 31 JUAN-ANDRÉS VILENA-PONSODA dialectology (i.e. the idea of a ‘genuine’ or ‘pure’ dialect) is an interesting way of facing this particular issue. Archeologie is an ongoing research project which intends to have access to supplementary sources of spoken and written data allowing the possibility of a comparison.5 Longitudinal studies on radio archive sources (see Van de Velde, van Hout, Gerritsen 196, 199; Hernández-Campoy 2018), as well as complementary written public and private documents (Hernández-Campoy, Conde-Silvestre 2012), will provide comparative evidence for language change and the history of the formation of the Andalusian intermediate variety. REFERENCES ALEA – Atlas Lingüístico y Etnográfico de Andalucía, 1–6 vols, 1961–1973, ed. by Alvar Manuel, A. Llorente, G. Salvador. Madrid: CSIC. Almeida Manuel 2018: Language hybridism: On the origin of interdialectal forms. – Language Variation. European Perspectives VII, ed. by Juan-Andrés Vilena-Ponsoda, Francisco Díaz Montesinos, Antonio Ávila-Muñoz, Matilde Vida-Castro. Amsterdam: SILV, John Benjamins, forthcoming. ALPI – Atlas Lingüístico de la Península Ibérica, 2016, ed. by García Mouton Pilar (co-ord.), Inés Fernández-Ordóñez, David Heap, Maria Pilar Perea, João Saramao, Xulio Sousa, ALPI-CSIC: http://www.alpi.csic.es (22-10-2018). Alvar Manuel 1955: Las encuestas del “Atlas Lingüístico de Andalucía (diciembre 1953 – marzo 1955)”. Granada: Universidad: https://grupo.us.es/ehandalucia/pdf/lecturas/atlas_linguistico_andalucia.pdf (26-10-2018). Alvar Manuel 1959: El Atlas Lingüístico-Etnográfico de Andalucía. Granada: Universidad. Alvar Manuel 1974: Sevilla, macrocosmos lingüístico. – Estudios Filológicos y Lingüísticos. Homenaje a Ángel Rosenblat, ed. by Quiroga-Torealba Luis, Mario Torreba losi, Pedro Díaz-Seijas. Caracas: Instituto Pedagógico, 13–42. Alvar Manuel 1993 [1973]: Notas de aseo al habla de Málaga. – Puerta Nueva. Revista de Educación, 6–13: https://grupo.us.es/ehandalucia/pdf/lecturas/atlas_linguistico_andalucia.pdf (26-10-2018). 5 In the frame of the above mentioned (on footnote 1) Project on the Sociolinguistic Patterns of Castilian Spanish (ECOPASOS, FF2015-6811-C5-1 and FEDER funds). 32 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ Auer Peter 2005: Europe’s Sociolinguistic Unity, Or, a Typology of European Dialect/Standard Constellations. – Perspectives on Variation, ed. by Nicole Delbecque, Johan van der Auwera and Dirk Geeraerts. Berlin: Mouton de Gruyter, 7–42. Berruto Gaetano 2005: Dialect/standard convergence, mixing, and models of language contact. The case of Italy. – Dialect Change. Convergence and Divergence in European Languages, ed. by Peter Auer, Frans Hinskens, Paul Kerswill, Cambridge: Cambridge University Press, 81–95. Berthele Raphael 2004: Dialektsoziologie – Soziolinguistische Aspekte der Dialektologie. – Sociolinguistics: An International handbook of the Science of Language and Society 1. Second edition, ed. by Ulrich Amon, Norbert Dittmar, Klaus J. Mattheier, Peter Trudgill. Berlin/New York: Mouton de Gruyter, 721–738. Bourdieu Pierre 1977: L’Économie des échanges linguistiques. – Langue Française 34, 17–34. Bourdieu Pierre 1984: Capital et marché linguistiques. – Linguistische Berichte 90, 3–24. Bourdieu Pierre 1986: The forms of capital. – Handbook of theory and research for the sociology of education, ed. by J. Richardson. New York: Greenwood, 241–258. Carbonero Pedro 1982: El habla de Sevilla. Sevilla: Ayuntamiento. Chambers J. K., Peter Trudgill 2004 [1980]: Dialectology. Cambridge: Cambridge University Press. Dietz Gunther 2001: Frontier hybridization or culture clash? Trans-national migrant communities and sub-national identity politics in Andalusia, Spain. – Journal of Ethnic and Migration Studies 30(6), 1087–1112. Durkheim Émile 1902 [1893]: De la division du travail social. Étude sur l’organisation des sociétés supérieures. Second edition. Paris: Alcan. García-Mouton Pilar 1992: El Atlas Lingüístico y Etnográfico de Andalucía. Hombres y mujeres. Campo y ciudad. – Actas del Congreso Internacional de Dialectología, IKER 7, 667–685: http://digital.csic.es/handle/10261/13687?locale=es (22-10-2018). Guy Gregory, Frans Hinskens 2016: Linguistic coherence; systems, repertoires and speech communities. Introduction. – Linguistic coherence: Systems, repertoires and speech communities. Lingua 172–173, 1–9. Hernández-Campoy, Juan-Manuel 2018: Virtual sociolinguistics: From real-time surveying to virtual-time archival sources for tracing change longitudinally. – Language Variation. European Perspectives VII, ed. by Juan-Andrés Villena-Ponsoda, Francisco Díaz Montesinos, Antonio Ávila-Muñoz, Matilde Vida-Castro. Amsterdam: SILV, John-Benjamins, forthcoming. Stripsni / Articles 33 JUAN-ANDRÉS VILLENA-PONSODA Hernández-Campoy Juan-Manuel, Villena-Ponsoda Juan-Andrés 2009: Starness and Non-Starness in Spain: Dialect Attrition and Revitalization of Regional Dialects of Spanish. – International Journal of the Sociology of Language 196/197, 181–214. Hernández-Campoy, Juan-Manuel, Juan-Camilo Conde-Silvestre (eds.) 2012: The Handbook of Historical Sociolinguistics. Malden: Wiley-Blackwell. Hudson Ronald A. 1996 [1980]: Sociolinguistics. Second edition. Cambrigde: Cambridge University Press. Jaberg Karl 1936: Aspects géographiques du langage. Paris: Sociétés de Publications Romanes et Françaises. Jaberg Karl 1959: Geografía lingüística. Ensayo de interpretación del Atlas Lingüístico de Francia. Granada: Universidad. Jaberg Karl, Jakob Jud 1928–1960: Sprach- und Sach atlas Italiens und der Südschweiz, 1–8 vols. Zofingen: Ringler. Molina-García Álvaro 2018: Percepción y distancia acústica. La variación paramétrica individual en la escisión fonemática de /θ/ en el español andaluz. Datos de la ciudad de Málaga. Unpublished manuscript. Universidad de Málaga. Morillo-Velarde Ramón 2001: Sociolingüística en el ALEA: variable generacional y cambio lingüístico. – Estudios de Lingüística de la Universidad de Alicante 15, 13–49. Moya-Corral Juan-Antonio, Marcin Sosiński 2015: La inserción social del cambio. La distinción s/θ en Granada. Análisis en tiempo aparente y en tiempo real. – Lingüística Española Actual 37, 33–72. Navarro-Tomás Tomás, Aurelio M. Espinosa, Lorenzo Rodríguez-Castellano 1933: La frontera del andaluz. – Revista de Filología Española 20, 225–277. Paton Pamela, John R. Hpp, Sandra Marquart-Pyatt 2011: Non recursive models. Endogeneity, reciprocal relationships, and feedback loops. California: SAGE. Regan Brendan 2017a: A study of ceceo variation in Western Andalusia (Huelva). – Studies in Hispanic and Lusophone Linguistics 10(1), 119–160. Regan Brendan 2017b: The Effect of dialect contact and social identity on fricative de- merger. PhD Dissertation. The University of Texas at Austin: https://repositories.lib.utexas.edu/handle/2152/62339 (22-10-2018). Salvador Gregorio 1967 [1963]: La fonética andaluza y su propagación social y geográfica (1963–1964). – Estudios dialectológicos, 61–70. Salvador-Salvador Francisco 1980: Niveles sociolingüísticos de seseo, ceceo y distinción en la ciudad de Granada. – Español Actual 37–38, 25–32. 34 Acta Linguistica Lithuanica LXXIX The Dilemma of the Reliability of Geolinguistic and Dialectological Data for Sociolinguistic Research. The Case of the Andalusian Demerger of /θ/ Sawoff Adolf 1980: A sociolinguistic appraisal of the sibilant pronunciation in the city of Sevilla. – Grazer Linguistische Studien 11–12, 238–262. Siegel Jeff 2010 [2006]: Second dialect acquisition. Cambridge: Cambridge University Press. Trudgill Peter 1986: Dialects in contact. Oxford: Basil Blackwell. Trudgill Peter 1996: Dialect typology: Isolation, social network and linguistic structure. – Towards a Social Science of Language 1, ed. by Gregory Guy, C. Feagin, D. Schiffrin, J. Baugh. Amsterdam/Philadelphia: John Benjamins, 3–22. Van de Velde Hans, Marinel Gerritsen, Roeland van Hout 1996: The devoicing of fricatives in standard Dutch: A real-time study based on radio recordings. – Language Variation and Change 8, 149–175. Van de Velde Hans, Roeland van Hout, Marinel Gerritsen 1997: Watching Dutch change: A real time study of variation and change in standard Dutch pronunciation. – Journal of Sociolinguistics 1/3, 361–391. Villena-Ponsoda Juan-Andrés 1996: Convergence and divergence in a standard-dialect continuum: Networks and individuals in Malaga. – Sociolinguistica 10, 112–137. Villena-Ponsoda Juan-Andrés 2001: La continuidad del cambio lingüístico. Granada: Universidad. Villena-Ponsoda Juan-Andrés 2005: How similar are people who speak alike? An interpretive way of using social network in social dialectology research. – Dialect Change. Convergence and Divergence in European Languages, ed. by Peter Auer, Frans Hinskens, Paul Kerswill. Cambridge: Cambridge University Press, 303–334. Villena-Ponsoda Juan-Andrés 2008: Sociolinguistic patterns of Andalusian Spanish. – International Journal of the Sociology of Language 193–194, 139–160. Villena-Ponsoda Juan-Andrés 2010: Community-based investigations: From traditional dialect grammar to sociolinguistic studies. – Language and Space. An International Handbook of Linguistic Variation 1. Theories and Methods, ed. by Peter Auer, J. F. Schmidt. Berlin/New York: Mouton de Gruyter, 613–631. Villena-Ponsoda Juan-Andrés, Antonio Ávila-Muñoz 2014: Dialect stability and divergence in southern Spain. Social and personal motivations. – Stability and divergence in language contact. Factors and mechanisms, ed. by Kurt Braunmüller, Stefen Höder, Karoline Kühl. Amsterdam: John Benjamins, SLV, 207–238. Villena-Ponsoda Juan-Andrés, Matilde Vida-Castro 2017: Between local and standard varieties: horizontal and vertical convergence and divergence of dialects Straipsniai / Articles 35 JUAN-ANDRÉS VILLENA-PONSODA in Southern Spain. – Language Variation. European Perspectives 6, ed. by Isable Buchstaller, Beat Siebenhaar. Amsterdam: John Benjamins, 125–140. Villena-Ponsoda Juan-Andrés, Matilde Vida-Castro 2018: Variation, identity and indexicality in southern Spain. On the emergence of a new intermediate variety in urban Andalusia, ed. by Stavroula Tsiplakou, Massimo Cerruti, forthcoming. Geolingvistikos ir dialektologijos duomenų patikimumo dilema atliekant sociolingvistikos tyrimus. /θ/ skilimo Andalūzijos ispanų kalboje atvejis SANTRAUKA Kadangi geolingvistikos, dialektologijos ir šiuolaikinės sociolingvistikos teikiami duomenys vertinami remiantis labai skirtingomis kalbos ir visuomenės teorijomis, kyla dilema siekiant geolingvistinių ir monografinių tarmių duomenų pagrindu rekonstruoti besiformuojančios kalbos atmainos istoriją. Sociolingvistiniai duomenys yra pagrįsti sociologinėmis prielaidomis apie imties sudarymą ir nuokrypį, bet jos nesuderinamos su tuo, ką suponuoja geolingvistiniai ir dialektologiniai duomenys (Villena-Ponsoda 2010). Taigi, tiriamoji problema siejasi su klausimu, kiek geolingvistiniai ir monografiniai tarmių duomenys yra patikimi, siekiant retrospektyviai nustatyti tiriamos kalbos atmainos istoriją ir konkretų kalbos pokytį. Dabar vykstantis viduriniosios miestiečių klasės kalbos atmainos, tarpinės tarp nacionalinės standartinės (Madrido) ir pietinių Andalūzijos ispanų kalbos atmainų, formavimasis yra suprantamas kaip kalbos panašėjimo į šiuo metu gyvuojantį prestižinį viduriniosios klasės kalbos standartą pasekmė. Jos tradicinės pietietiškos tapatybės kaitos procesai vyksta nuo šeštojo dešimtmečio vidurio ir artėja prie šiuolaikinių modernizacijos ir urbanizacijos diktuojamų tapatybių (Regan 2017; Villena-Ponsoda, Vida-Castro 2017). Pakartotinė geolingvistikos ir tarmių tyrimų duomenų analizė bei palyginimas su dabartiniais sociolingvistikos teikiamais rezultatais leistų, pirma, kruopščiai ištirti šiuo metu vykstančių pokyčių paplitimą visuomenėje realiuoju laiku, antra, išnagrinėti ideologinį atsilikimą, turintį įtakos šio tipo retrospektyvioms trajektorijoms. Įteikta 2018 m. spalio 30 d. JUAN-ANDRÉS VILLENA-PONSODA Universidad de Málaga Campus de Teatinos, E-29071 Málaga, Spain [email protected] 36 Acta Linguistica Lithuanica LXXIX