There are few social phenomena as reluctant to be studied under the rigorousness of the scientific method as national identity. The embeddedness within the realm of the doxa, in its twofold Platonian significance of pistis and eikasia (Plato, 1968 [380BC]), of national identity is perhaps better understood when we compare its scientific development with another old, politically loaded, concept like social class, during the last one hundred years of modern social science. Thus, if we observe two of the most widely respected compilations of the state-of-the-art regarding social class and national identity, Wright (2005) and Özkirimli (2010), respectively, we can easily see that both from a theoretical and from a methodological point of view, social class has received a stricter treatment than national identity. This relative underdevelopment on the scientific treatment of national identity, largely discussed by several authors (vid., e.g., Rigger, 2000; McCrone, 1998: 17; Billig, 1995: 51-55; Smith, 1983), cannot be attributed to a lack of time, interest or relevance of the social issues underpinned by national identity. We could mention Ernest Renan’s famous conference Qu’est-ce qu’une nation? delivered at the Sorbonne in 1882 (Renan, 1997 [1882]); the immense amount of texts dedicated to the “national question”, from both the Marxist (Haupt et al., 1974) and the Liberal traditions (Haugaard, 2006); the work of some of the most brilliant scholars of the last decades, such as Ernest Gellner (1983), Eric Hobsbawm (1991) or Benedict Anderson (1991); as well as its undeniable interest, as it played a major role on many of the most destructive conflicts that humanity has suffered throughout its long history (Hall and Maleševic, 2013).

The relevance of national identity in contemporary societies is also undeniable. It was at the core of the 2014 Scottish independence referendum (Mullen, 2014), the 2016 United Kingdom European Union membership referendum (Henderson et al., 2016), and the Catalan movement for independence, initiated in 2012 and currently very active (Serrano, 2013; Lepič, 2017). It is perhaps in this exceptional relevance that we are to find the reason why the study of national identity has been so impervious to the scientific realm of Platonian episteme, as Billig (1995: 22-23) pointed out. However, this relative methodological underdevelopment must be qualified, as there have been some relevant technical achievements that have enabled us to better analyze the questions posed by national identity to the social scientific community; undeniably, the so-called Linz-Moreno Question (LMQ) (Moreno, 1986) is one of these advancements, as it provides a robust methodological tool for the measurement and comparison of people’s national identity feelings.

This article aims to draw upon the path opened by the LMQ, in order to further strengthen the methodological tools at our disposal, which are, as Goldthorpe (2000) insistently states, at the core of any scientific advancement. This task seems particularly urgent today, when national identity is at the core of the issues defining the impact area of political communication (Bouza, 2004) both in Europe (Friend, 2012), and worldwide (Ichijo, 2013).

This article is divided into 5 sections; first, it presents the objectives, hypothesis, and data used, featuring a detailed discussion of the LMQ and of the dual identity approach that usually underpins the interpretations of the LMQ. This presentation is followed by a descriptive analysis of the structure of identity feelings for each AC, using data gathered by the CIS. Based on this descriptive analysis, the article then offers a qualitative analysis of the press, the results of which are aligned with our initial hypotheses. Finally, we present a quantitative analysis of the data based on a multinomial logistic regression using data of CIS’ study 2956, Barómetro Autonómico III. The results of the qualitative and quantitative analyses and their successful alignment with our key hypotheses are discussed in the conclusions.

Agreeing with Merton’s statement that “establishing the phenomenon” is the most important prerequisite preceding any scientific analysis (Merton, 1987: 1), the aim of this article is to improve the quality of the measurement of national identity feelings. In modern social sciences, the LMQ has been established as a widespread technique to measure the levels of national identity in any given society, and it has thus been included in the Eurobarometer since 1992 (Serricchio, 2012). Although there are some alternative measurement techniques —see Luhtanen and Crocker (1992) for a perspective from social psychology—, and it was subjected to criticism about its validity (Ruiz Jiménez, 2007), the LMQ remains dominant, particularly in the fields of Sociology and Political Science —see, inter alia, Rosie and Bond (2008); Billet et al. (2003); Curtice and Heath (2001), and Kiely et al. (2001). We will argue, however, that, although analytically powerful, the LMQ should be treated with methodological caution and reflexivity, in line with Guinjoan and Rodon (2015), who found that the LMQ does not capture appropriately the intensity of state nationalist feelings.

We will follow here the terminology established by Moreno and McEwen (2005) for the description of the different levels as state-wide and sub-state, as we consider this terminology more accurate and value-free than others, such as “regional/national”. The LMQ establishes five categories to complete the sentence “Do you see yourself as…”:

1/ [sub-state identity] only.

2/ More [sub-state identity] than [state-wide identity].

3/ Equally [sub-state identity] and [state-wide identity].

4/ More [state-wide identity] than [sub-state identity].

5/ [state-wide identity] only.

When applied to Spain, the LMQ takes as sub-state identities those associated to each one of the seventeen Autonomous Communities (ACs) in which Spain’s territory is divided. Some ACs have long traditions and historical content, while others were designed ex novo as part of the so-called café para todos (coffee for all) policy, which was at the core of the 1978 Spanish territorial design (Tusell, 1997). In Spain, however, we find different national projects; from the advanced nation-building processes taking place in the Basque Country and Catalonia, to the relatively underdeveloped nation-building processes in other areas like Galicia (Martínez-Herrera, 2002; Linz et al. 1973). These differences among nation-building processes within the Spanish state have been usually explained under the lenses of the neo-institutionalist tradition that considers that political/administrative institutions contribute to the development of nation-building processes (March and Olsen, 1989; Powell and Dimaggio, 1991). For example, the relative underdevelopment of Galician nationalism has been explained as a consequence of historical institutional deficits (Máiz and Losada, 2000: 65). A similar approach is often used to explain differences between the cases of Wales and Scotland; it has thus been argued that Scottish nationalism, much more developed than the Welsh one, is rooted in a solid institutional memory, including differences related to its banking, church, and education systems (Kendrick, 1989; Nairn, 2007), while the comparatively weaker Welsh nation-building process is strongly associated to its linguistic base (Thomas, 1997; Jones and Merriman, 2009).

Our objective is to determine the validity of the LMQ as a proxy variable to measure national identity, regardless of the degree of nationalist consolidation that characterizes the society under examination. Our main hypothesis here is that the LMQ is a good indicator of national identity when applied to societies featuring strong sub-state nation building processes, such as Scotland, but it faces some relevant analytical problems when applied to places where these processes were not so strongly developed, like Galicia.

We find the epistemological foundations of the LMQ in the theory of dual identity (Moreno, 1988; Moreno and Arriba, 1996; Heath and Kellas, 1998). Simply put, dual identity means that in a society where two national projects coexist, people usually display a double sense of identity attachment. As Curtice and Heath put it, “[…] people in England have long tended to think of themselves as both English and British, that is to have a ‘dual identity’” (2001: 157). In this sense, understanding the identity feelings of people living in societies like Scotland or Catalonia as dual, seems an appropriate description of social reality, especially when this assertion is apparently validated by the results of the LMQ when applied in surveys. In fact, in this kind of societies characterized by hosting two coexisting national projects, there are always remarkable proportions of the population that identify themselves simultaneously with their state and their sub-state identities. For instance, as being at the same time Spanish and Catalan or British and Scottish. At first glance, this might seem an obvious descriptive trend, but in this article we will try to problematize the concept of dual identity by going beyond its descriptive surface.

To summarize the basic reason for this problematization of the concept of dual identity, it could be said that single individuals do, in fact, identify themselves with multiple identities on everyday life —woman, teacher, sister, Catalan, etc.— but this fact does not mean that these different identities are all of them on the same layer of social reality. In social sciences, it is well established that the identities assumed by individuals are necessary for the development of social interaction (Goffman, 1959) framed within a network of power relations and capital circulation (Bourdieu, 2012) which are at the core of the formation of human practices (Bourdieu, 1987). This amounts to say that identities can be seen as a resource at the disposal of individuals so that they are able to build up the subjectivity inherent and constitutive of social relations. Such a process, however, is only possible if there is a functional differentiation between the different identities available to each set of social positions. Hence, associations between subjects and identities are not dual but presumably multiple, depending on the synchronic characteristics of social fields, as Moreno himself has argued (Moreno, 2004).

In order to make our point clearer, we summarize some of the knowledge produced within the academic sub-discipline of nationalist studies, with no intention of presenting an exhaustive review of the huge volume of social scientific literature on the topic, which, as Rogers Brubaker pointed out “has become unsurveyably vast” (Brubaker, 2009: 22); the purpose will rather be to introduce some essential, well-established, arguments that lead us to uphold the aforementioned critique to the dual identity standpoint. Let us start by stating the obvious, i.e., that nations are social constructions in constant discursive reconstruction (Wodak et al., 2009), which involve cultural practices (McCrone et al., 1995), the media (Eder et al., 2002), rites (Abélès, 1990), ceremonies (Balandier, 1994), performances (Vaczi, 2016), fictions and myths (Balibar, 2005), branding image (Dinnie, 2008), and many other everyday subconscious mechanisms (Billig, 1995: 93-127) built around some particular pre-existing diacritics (Barth, 1969) which are articulated around an on-going national project (Armstrong, 1982). It is well-known that language typically plays a central role as a diacritic for sub-state national projects in Spain —see Shabad and Gunther (1982) for the Catalan and Basque cases and Beswick (2007) or Beramendi (2007), for analysis on Galicia— but there is of course a difference between national languages and pre-national forms of talk.

Differences between a national, standardized language and pre-national forms of talk were thoroughly commented by Billig (1995: 13-36), who coined the term “syntax of hegemony” (ibid.: 87), referring to the metonymic process by which a part of a nation’s cultural mosaic claims to represent the whole. Billig uses this concept to analyze the construction of national languages, when a particular dialect becomes the national language in a process linked to power dynamics; in Billig’s words: “The middle class of metropolitan areas typically will make their meanings stick as the official language, relegating other patterns within the national boundaries to ‘dialects’, a term which almost invariably carries a pejorative meaning” (ibid.: 32). In fact, the term dialect has been problematized a number of times, as it usually involves not only a descriptive scientific definition on the natural variation characterizing languages, but also differences regarding the status of each of these varieties —see Haugen (1966) for an in-depth dissertation on the different implications of the term dialect. The use of the term ‘dialect’ in our text refers to its most descriptive non-normative definition.

Differences between pre-national diacritics and standardized national forms do not only apply to languages, but also to whichever ethnic diacritic susceptible of being integrated in the nation-building process in the form of differential facts. This matrix is first developed within the limits of the intellectual field (Hroch, 1985) which, as a subfield dependent on the broader field of political power (Bourdieu, 1971), cannot be independent from the socio-political context; or, as McCrone puts it, “being able to show that there is ethnic homogeneity in a given territory —or, rather, that people living there believe themselves to be homogeneous— is the outcome of political and social processes, not their explanation, their cause” (McCrone, 2001: 23; see also Stavenhagen, 1996). Consequently, whenever nations are in an early stage of construction, i.e., when they are not established, we typically find conflicts among national elites linked to these processes of standardization —see, e.g., Herrero (2011) for an in-depth analysis on the bitter conflict around the orthographic normativization of Galician. These processes usually produce social distinctions in terms of status between well-established national languages and pre-national forms of talk —see, e.g., Álvarez (1993)— which can be seen as manifestations of a general dynamic that produces a distinction between pre-nationalized diacritics and standardized national forms. The point has been developed by Beramendi who, on the Spanish case, concluded:

There is no possible analytical way of discerning the first two identities referred to by Beramendi if we only take into account the five categories of the LMQ, as each of these become a bucket where fundamentally different identities coexist. This impossibility of the LMQ to differentiate regional from national identities might lead to misconceptions; for instance Table 1 uses the examples of Catalonia and Galicia to show how the same responses to the LMQ are associated to different national identifications in strongly and weakly nationalized ACs.

Even if we consider the neo-institutionalist perspective, according to which weak sub-state identities could have been boosted after more than three decades of autonomic system, the LMQ continues to suffer from the same impossibility to differentiate regional vs. national identities. In addition, the identification with the ACs in Spain is rather stable and shares values and synchronic evolution with the identification with Spain as a whole (vid. Fig. 1).

Our hypothesis, then, is that the LMQ can only be a good indicator of national identity feelings in societies hosting one or two advanced nation-building processes, but not in societies characterized for having low intensity nationalisms. In this second kind of societies, we will find simultaneously the same ethnic diacritics adopting nationalized and un-nationalized versions of themselves, from an emic point of view (Headland et al., 1990). This dynamic can be easily identified in the opposition between normative and vernacular languages —vid. O’Rourke and Ramallo (2011), for a comparison between the cases of Ireland and Galicia; vid. Robert (2009) for a study of the Welsh case—. A paradigmatic example of the trend is well-known for social scientists interested in Galicia, where urban Spanish speaking Galician nationalists share the same category of the LMQ with rural Galician speaking Spanish nationalists (Álvarez-Cáccamo, 1993; De Nieves, 2008). The paradox is only apparent, as the example illustrates how the LMQ puts together radically opposed versions of Galician identity, one national, and the other regional. Although some authors —vid. Pattie et al., 1999: 309 for the Scottish case— have explained that identities do not necessarily correlate with constitutional or political preferences, our point that the LMQ is unable to differentiate regional vs. national versions of the same identity remains, making it problematic for some analyses that should be undertaken with methodological reflexivity.

In order to test our hypothesis, we use data from Spain, due to two reasons. The first reason is that the Spanish case offers a particularly adequate framework for comparison among regions, as the Spanish Centro de Investigaciones Sociológicas (CIS) has been producing empirical data on this matter at least since 1983 (CIS, 2014), enabling comparative analysis throughout methodologically homogeneous quantitative datasets. The second reason is that Spain offers a diverse political scenario, containing a well-established nation-state as well as different nationalities and regions (García Ferrando et al., 1994).

Let us start by analyzing data from CIS’ study Autonomic Barometer (III), no. 2956. This study involved the realization of a survey with 11 290 interviews within Spain, with a sample that allows specific analysis for each one of the seventeen Spanish Autonomous Communities (ACs). The fieldwork for this study took place between 13 September and 09 October 2012, and the data is accessible online (CIS, 2012). Figure 1, vid. infra, presents a scatter plot, showing the relationship between two variables: feeling of attachment to the Spanish identity (Y axis) and feeling of attachment to the AC identity (X axis). We have used the 5-10 scale on the Y axis and the 7-10 on the X axis to better show the slight differences of identification with the AC. It can be argued that only group B.2 stands clearly out of the rest of the distribution, but we believe that using the somehow small differences that make up the other three groups is analytically useful; indeed, we believe that the qualitative and quantitative analyses that we present later on in this article justify this fourfold classification.

Figure 2 represents two main groups of ACs: Group A contains the ACs in the upper part of the chart and group B includes the ACs occupying the lower space of the chart. We have then identified two subsets of ACs within each of the above groups. The first subset, A.1, includes Castile-La Mancha, Valencia, Castile and León and Madrid, all of them featuring a high level of identification with Spain (means ranging from 8 to 9, over 10) together with a relatively low identification with the AC (means ranging from 7 to 7.5, over 10). The second subset, A.2, is made up of ACs that present a high level of identification with Spain (means ranging from 8 to 9, over 10) but also a high level of identification with the AC (means ranging from 8 to 9, over 10), and includes Asturias, Extremadura, Aragon, Andalusia, Cantabria and La Rioja. The second group of ACs, group B, includes the ACs located under the best fit line of the scatter plot, hence characterized for having relatively low feelings of identification with Spain (means under 8, over 10). We have subdivided this group into two subsets, B.1 and B.2; the first one contains relatively low identifications with Spain (means ranging from 7 to 8, over 10) together with high levels of identification with the AC (means ranging from 8 to 9.03, over 10), and it includes the Canary Islands, Galicia and the Balearic Islands.

Finally, subset B.2 is made up of ACs featuring low levels of identification with Spain (means ranging from 5 to 7, over 10) combined with high levels of identification with the AC (means ranging from 8 to 8.5, over 10), and it includes Navarre, the Basque Country and Catalonia. At this point, we have to note the peculiarity regarding the case of Navarre, as this AC has been traditionally claimed as part of the Basque Country by Basque nationalists —vid., e.g., Pérez-Agote (1989); Conversi (1997: xvi); for a historical perspective during the Second Spanish Republic, vid. Chueca-Intxusta (1999)—. There were two Basque nationalist coalitions concurring in the Spanish 2011 general election within Navarre, Amaiur and Geroa Bai; they obtained 91 623 votes in the AC of Navarre, concentrating 28.25 % of the vote for political parties within this AC. In the 2015 general elections, EH Bildu together with Geroa Bai, both sharing a Basque nationalist ideology, obtained 65 581 votes, 18.75 % of the vote for political parties within the AC (Ministry of Interior of Spain, 2016).

In line with the existing literature, particularly Herranz de Rafael (1996), we defined group A.1 as containing strong state-wide nationalist ACs, groups A.2 and B.1 as containing weakly nationalized ACs, and group B.2 as containing strong sub-state nationalist ACs. We think that the nature of this cleavage, as a division between strongly nationalized ACs —groups A.1 and B.2— versus weakly nationalized ACs —groups A.2 and B.1—, points towards the existence of two different kinds of national conflict in Spain. ACs within subsets A.1 and B.2 in figure 1 will be characterized by having a belligerent structure of actors reproducing nationalism, whereas ACs within subsets A.2 and B.1 will have a rather lower profile regarding their set of actors reproducing national identity. Following Eder, these dynamics at the core of the reproduction of national identity can be studied from a symbolic interactionist perspective, conformed by a basic matrix of three structural positions: ego, alter, and other (Eder et al., 2002). In order to test this hypothesis, pointing towards differences on the degree of conflict associated to each group of ACs, we will develop a qualitative content analysis of the press. If this hypothesis proves to be right, then the appropriateness of our proposed division for Spanish ACs will gain consistence, enabling subsequent analysis to test our main hypothesis, already explained, according to which the LMQ only provides good measurements of national identity in those contexts where we find highly developed sub-state national projects —ACs within subset B.2— but not in those ACs featuring only one state-wide national project —ACs within subset A.1— nor in those ACs featuring low intensity sub-state national projects —ACs within subsets A.2 and B.1.

It has been consistently proven that the press is a central actor in the reproduction of national identity. Despite some skepticism on this relationship (Schlesinger, 1991a; 1991b; 1993), social scientists have produced advancements in this area of research with remarkable success. For a general overview, vid., e.g., Deutsch (1966); Anderson (1991); Tomlinson (1991); Billig (1995: 93-127); Eder et al. (2002); Roosvall and Salovaara-Moring (2010), and Mihelj (2011). For specific case studies vid., e. g., Brookes (1999) for Britain; Law (2001) for Scotland; Rosie et al. (2004) for the UK; Idoiaga, et al. (2016) for the Basque Country; Yumul and Özkirimli (2000) for Turkey.

We have analyzed the front pages of newspapers read at least by 5 % of the adult newspaper reading population within each AC (vid. Table 2). We have consulted the front pages of these newspapers in two symbolic dates, exemplifying those flag-waving situations referred by Billig as moments of special patriotic meaning (Billig, 1995: 37-39). The first one is 11 September, the national day of Catalonia, and the second day is 12 October, the national day of Spain. The first day, also known as Diada Nacional de Catalunya, commemorates the fall of Barcelona, defeated by the Bourbon Spanish troops during the War of the Spanish Succession, on 11 September, 1714. The second day is the Fiesta Nacional de España, also referred to as Día de la Hispanidad, that commemorates the discovery of America by Christopher Columbus, on 12 October, 1492. It is worth noting that newspapers do not act as mere mirrors of state-building processes, but they are also an agent of nation-building in their respective communities; however, we believe that the analysis shows a clear empirical trend, enough to make it useful for our discussion.

We have chosen the national day of Catalonia because the national question in Spain is very much focused on Catalonia since the Catalan elections held on 25 November 2012, which were anticipated by the Catalan government, and presented by CiU —the coalition in government at that time—, as a monographic election around independence. Independence was explicitly defended by Convergència Democràtica de Catalunya, one of the two political parties making up the governing coalition CiU —some leaders of the other party, Unió Democràtica de Catalunya, had shown some reluctance to independence, however (Piñol, 2012)—. Independence was also a central issue on the September 2015 Catalan elections (Rivera and Jaráiz, 2016). The pro-independence movement in Catalonia has grown steadily since the 2012 elections, a dynamic that led the Catalan government to organize an illegal referendum of independence on 1 October, 2017, cancelled by the Spanish government. At the time of finishing this article, the Government of Spain had activated, for the first time in history, Art. 155 of the Constitution, which would lead to the intervention of the Catalan government by the State government, provided that it is also approved by the Senate.

We have analyzed the headlines for these two symbolic days in four front pages for each newspaper, the day after each celebration: 12 September, and 13 October, both in 2013; we have also taken into account the same four dates for 2012, in order to better contrast our data. We have classified the messages of each headline into three different categories: negative, neutral and positive, in order to produce a comparative framework for the analysis of the news coverage of Spain’s identity politics. We have classified the headlines as negative when these are focused on criticizing the opponent —either Catalan or Spanish national projects. Accordingly, we have identified the messages as positive when these are focused on praising the own nation. When there is no evidence or neither negative nor positive messages, we will classified them as neutral.

Table 3, vid. infra, presents the results of our qualitative analysis of the press, and we can see that these are aligned with our hypothesis. The front-pages of strong state-wide ACs, grouped within set A.1, show a primacy of negative messages, 60 %, on the Catalan national day, while this trend is reversed on the Spanish national day, with a majority of positive messages, 64.3 %. We have found a primacy of positive messages on those ACs with strong sub-state nationalisms of group B.2 for both national days, 71.4 % on the national day of Catalonia, 68.6 % on the national day of Spain. In line with our hypothesis, trends are softer for the two sets of weakly nationalized ACs, A.2 and B.1; we found a majority of neutral messages on these two groups on the Catalan national day, 54.5 % in group A.2 and 58.3 % on group B.1. The analysis for the Spanish national day show a majority of positive messages in these two sets of weakly nationalized ACs, although this primacy is clearly bigger in the case of the strongly nationalized ACs —62.5 % of positive messages for group A.2 and 53.8 % for group B.1.

Once we have empirically proved that there are, in fact, remarkable differences among our proposed sets of ACs regarding the degree of national conflict associated to them and, thus, strengthening the usefulness of this classification, we will proceed to present some quantitative empirical evidence based again on data from the CIS’ study 2956, Barómetro Autonómico III. One of the questions in this study directly refers to the political status of Spain:

Q.18 What does Spain mean to you? (ONE ANSWER ONLY)

Options “a” and “b”, are expressions of Spanish nationalism, as we can say that people choosing these options consider Spain as a nation and, most importantly, as their nation. Options “d” and “e”, however, do not explicitly consider Spain as a nation, but simply as a State; moreover, people who choose options “d” or “e” consider that the Spanish State is made up from different sub-state nations. Option “c” is a neutral one, for it only describes the reality of Spain being a State within international law. We consider other response categories for Q18, options “f”, “g”, and “h”, as missing values. As Q18 is included in the questionnaire for every AC, we think that it is a good way to test our hypothesis, i.e., that the LMQ is a good indicator of national identity within those territories which are strongly nationalized, but not when applied to those regions which are weakly nationalized.

We have built a multinomial logistic regression model with Q18 as the dependent variable (DV), and the five categories of the LMQ as independent variables (IVs), controlling by sex and type of occupation. The model allows us to measure the predicted probabilities that respondents of the LMQ have to choose the different categories of Q18 (figures 3 to 7).

The results of the multinomial regression prove that the relationship between national identity and the response to the LMQ is structurally different for each of the four groups of Autonomous Communities. The results are also aligned with our hypothesis, as the category “only AC, not Spanish” of the LMQ is a better predictor of national identity —category “my country” in Q18— for the groups A.1 and B.2 of ACs. Moreover, the results of the regression show consistently lower predicted probabilities for “equally Spanish than AC” in the B.2 group for all Q18 categories, suggesting that this response to the LMQ has contradictory associations to national identity in those ACs with highly developed conflicting national projects.

The scientific study of national identity is among the most urgent tasks for contemporary social sciences. The relevance of identity politics and nationalism in Europe and worldwide has only increased during the recent years. Some of the most relevant recent political events are defined to a great extent by nationalism, including the Brexit (Henderson, et al. 2016), Trump’s presidency and his “America First” politics (Trump, 2015), the Catalan pro-independence process (Crameri, 2014), the 2014 Scottish Independence Referendum (Mullen, 2014), and the internal conflicts of Ukraine (Kuzio, 2016), to name but a few. State-wide national identity holds a prominent role as well, as it underpinned some of the processes of the current European crisis, as Habermas has pointed out (Habermas, 2012), and was shown by the increased electoral support of political options deeply rooted on aggressive state-wide nationalism, like France’s Front National, or the UK Independence Party. The work presented in this article offers a technical improvement which we believe will be useful for the advancement of the scientific study of national identity, something that we consider an urgent task for contemporary social sciences.

Thus, we have provided proof that the most usual measure of national identity feelings through surveys, the Linz-Moreno question, needs to be analyzed with methodological reflexivity. Using data from Spain, we have developed both qualitative and quantitative analyses in order to test if the LMQ provides a right measure of national identity independently of the degree of development of the nation-building processes present in each society. After classifying Spain’s Autonomous Communities into four categories in accordance to the identity feelings displayed by their populations, we saw that the LMQ presents structural differences when considered as a proxy of national identity feelings. In particular, the weakly nationalized ACs are usually considered a good example of what is usually referred to as “dual identity”, but we have shown that identity overlapping is only possible in these regions due to a substantive difference affecting both identities; while one is purely national, the other is regional.

We have presented a critique of the dual identity theory, which is the epistemological foundation of the LMQ; this critique is based on the fact that in societies where there is a clear primacy of a statewide national project, sub-state identity (or identities) is either totally or partially —in case that a sub-state national project developed— subsumed within the dominant state-wide national project, in the form of its regional expressions, producing what we may call a dialectalization of identity, through its regionalization. This regionalization process, based on the creation of symbolic dependencies —affecting those ethnic diacritics (Barth, 1969) through which national or pre-national sub-state identities are socially constructed— initiates a dialectical process in which this regionalization became the thesis, while the antithesis is embodied in diverse social movements that can, to some degree, be identified under the label of pro-independence; in fact we can easily cite many examples of these kind of movements in places where there is sufficient cultural diversity to make the situation too symbolically violent for the said dialectical process not to start, and these would include the defense and promotion of language, natural and historical heritage, music, literature, etc. (Hroch, 1985). Thus, what makes compatible both identities in the dual identity framework is precisely that the two identities are different in nature.

In conclusion, if nationalism should be treated as an ideology, as Billig insists (Billig, 1993) maybe we could start by measuring it as such, making use of a similar scale to the one that we usually employ to measure the left-right ideology, where one of the extremes represents the complete identification with the state-wide national project, and the other one the complete identification with the sub-state national project. We believe that the question “On a scale from 1 to 10, being 1 total identification with [sub-state identity] and 10 total identification with [state-wide identity], where would you place yourself?” will thus constitute a better indicator of national identity, especially for weakly nationalized regions.