1. Introduction In Croatia there is a rich history of research of its three dialects, the turbulent processes of language standardization and the status of language varieties. The specific internal variety dynamics is also reflected in language practices of popular music songwriters from the 1960s to the present day, with special notice that there was a tendency of writing lyrics in standard Croatian during the nineties. It was a war and post-war period during which political and linguistic tendencies were designed to strengthen national unity and identity. The aim of the research is to investigate the circular process in which socio-political and linguistic circumstances are reflected in author’s language choices in musical texts, which in turn through rapid media widespread influence language ideologies, national affirmation of the Dalmatian regiolect and supra-regional language norms at the lexical and phraseological level. The theoretical framework of this interdisciplinary research consists of language biographical studies, text linguistics and contact linguistics, dialectology and musical text linguistics. It uses sociolinguistic terminology, and where the research results require, Croatian dialectological terminology is applied. The research is based on the qualitative analysis of narrative interviews with ten most popular authors selected because they write in the Dalmatian regiolect and standard Croatian, in periods of significant change in language policies. The formal-functional structure of the narrative, discourse strategies used by the informants to express positioning and the construction of individual and collective identities are analysed. Furthermore, by analysing their musical texts (lyrics) it is determined whether they write in standard Croatian during the nineties and to what extent do the narrative statements about their linguistic tendencies match their actual authorial language practice. The analysis of their idiolect, recorded in the interview, determined which language variant they speak and if it matches their authorial language. All the data is then combined in order to recreate the picture of certain language ideologies in the last seventy years and to reconstruct language biographies of the most important songwriters in the Dalmatian regiolect area. 2. Theoretical framework and previous research According to Ivo Žanić (2016), the beginning of sociolinguistics in Croatia is considered to be the 1952 article by Slavko Pavešić which presented language practices of Croatian speakers as a specific dialectal situation. In 1997 Dalibor Brozović pointed out that the three dialects essentially define Croatian language and enrich it with a built-in heritage. However, such an affirmative attitude towards dialects was not viewed favourably by language purists in certain years and regional varieties were usually stigmatized and corrected as wrong within the educational system. This influenced the public perception and status of Croatian dialects and other language varieties. In the first half of the 20th century, the Adriatic area was not integrated into the national identity narrative and Katičić (2004: 14) asserted that the Dalmatian regiolect was unfairly neglected as a result of national linguistic standardization which was "performed in the light of inappropriate linguistic ideology". However, the integration of the regiolect into the national cultural and linguistic framework took place in the 1960s mainly through media, especially TV series Malo and Velo Misto, popular music and the Split Festival. A natural absorption occurred of the Mediterranean sociocultural lexicon into the Novoštokavian, standard language norm (Žanić, 2016: 195). Thus, one can say that popular music "actually corrects what the educational system often fails to set in elementary school" (Žanić, 2015a). Based on the Bourdieu's 1992 theoretical model, standard language tends to acquire a hegemonic position and prestige in relation to other varieties through the mediation of schools and through its connection with elite groups, but it seems that this is not the case in Croatia (Žanić, 2015b). The first Croatian research on the status of varieties conducted in 1988 showed that high school students actually attribute prestige to the Split and Zagreb vernaculars, as idioms of the most important urban national centres, and not to the standard language which has a reduced status (Jakovčević, 1988). More recent research (Sujoldžić, 2009; Šimičić and Sujoldžić, 2013) also show a rather low status of the standard, which "is not surprising given the frequent perception of this variant as artificial and formalized" (Sujoldžić, 2009: 101). Based on research of the role of varieties in popular music in Croatia, sociolinguist Ivo Žanić confirms that the three dialects or regional varieties – Kajkavian, South Chakavian and North Chakavian – are recognized in popular music genres. The sociolinguistic role of popular music and festivals in the emancipation of dialects and their speakers was highlighted in 1972 by linguist Ljudevit Jonke who claimed that dialects are vital thanks to music and that they transcend their local area reach. The aforementioned role of the festival was also apostrophized by Stjepko Težak (1997: 21) and Dubravka Sesar who pointed out that through the mediation of popular music and festivals dialects do not disappear but "consolidate and achieve a certain degree of autonomy" (2004: 74). The widespread of the Dalmatian regiolect started in the sixties through media and popular music and gained full national recognition in the former state of Jugoslavija during the eighties, also called the Golden period of Split Festival. A crucial historic overturn happened in 1990 when Jugoslavija started fragmentizing and the Croatian War of Independence changed political and linguistic circumstances. In Croatia the 1990s were largely characterized by linguistic purism and prescriptivism intended to distinguish Croatian from Serbian (Kapović, 2013: 392) and patriotic songs sung in Croatian by means of which politicians and linguists were empowering national identity through Croatian language independence and purity. On the other hand, it was during the late eighties and nineties when a strong urban music scene developed in Split under the influence of American and British pop and rock bands. It is notable that under both circumstances authors of popular music in the area of Dalmatian regiolect started writing lyrics in the standard Croatian variety more often than in the regiolect. Only in the region of Istria did the northern Chakavian dialect become stronger through the "Ča wave" music genre whose authors were influenced by the linguistic policy of deliberate preservation of dialects (Žanić, 2015a). In the 21st century the ideas of nation, language, and identity started being questioned under the influence of neoliberalism, globalization, and economic and political conditions that lead countries into post-nationalism (Heller, 2011). At the turn of the millennium, the three original dialects in Croatia transformed into a trinity combined of standard – Zagreb vernacular –Split vernacular (or Dalmatian regiolect), and the standard's "functional and thematic comprehensiveness was challenged", especially in media and popular music (Žanić, 2010). Various language varieties and dialects started to be heard more often in popular TV shows on national and private TV, so there seems to be more liberty in using one’s idiolect freely in media discourse and thus in music lyrics as well. Today new authors are open to linguistic innovations and use idioms creatively (Žanić, 2015a). However, their idiom is no longer an "authentic dialect" but a hybridized slang dialect (Berruto, 1994) in which the dominant distinguishing features of the Chakavian dialect are eliminated (Vranić and Zubčić, 2015: 53) and anglicisms give it a modern touch and prestige. Choosing a non-standard language variety in lyrics is thus an act of identification which enables authors to escape into linguistic anarchism and individualism (Robertson, 1994). Through their linguistic choices they position themselves in the world and define the groups they consider to be members of and those they want to distinguish themselves from (Novak, 2012: 147–199). However, Peti-Stantić and Langston (2013: 3) conclude that it is usually not possible to completely objectively describe the relations between language communities and language varieties by means of theoretically accepted classifications of languages and dialects, but it is necessary to take into account various sociolinguistic factors. Because we are reconstructing biographies in this research we are dealing with all notions of identities, individual, collective, local, regional, national, and authorial. In contemporary linguistics, researches of the connection between language and identity began with variationist sociolinguistics in the 1970s which presented results statistically. In the last few decades, under the influence of poststructuralist thought, the essentialist notion of an unchanging and permanent identity has been rejected so identities are viewed as dynamic, fluid, changing and often contradictory processes (De Fina et al., 2006; Benwell and Stokoe, 2006) and are investigated in qualitative studies. Kalogjera (2007: 261) pointed out that identity is most clearly expressed in discourse, in communicative interaction, "when a communicatively competent individual chooses expressions from one’s language repertoire by means of which one builds and expresses identity, that is, the identity of belonging to a certain group". So, when describing a speaker's identity and social roles, later approaches such as interactionist sociolinguistics started taking into account the situational context and social and discourse conventions of language use (Bamberg et al., 2007). Identity is thus expressed and built through narrative, which was defined as the oral transmission of personal experiences by Labov and Waletzki in 1967. Labov later set theoretical guidelines for the sociolinguistic analysis of narratives and Schiffrin (1996) also emphasized that the central function of the narrative is the sociolinguistic construction of identity. With the development of sociolinguistics, the narrator as the bearer of cultural and social norms, the functional characteristics of the narrative, the situational context and narrative expression of identity started being studied. In this study we adopt the concept of identity as it is determined in Western, postmodern societies, as a fragmented and plural concept and as a reflection of socio-historical circumstances (Lucius-Hoene and Deppermann, 2004), and determine in which ways and by which discourse means the authors/informants express and build their identities. Because the informants produce lyrics, which are texts intended to be sung and are thus research material in musical discourse, we adopted analytical methods of musical text analysis which determined how musical discourse is used to express identity and how it can be used as a tool for social change (Pavlovova, 2013). Sociolinguist Ivo Žanić set the framework of musical text sociolinguistics in Croatia and expanded the terminological dichotomy of two types of language discourse – spoken and written – into a triadic division by introducing the so-called sung discourse, as part of the broader musical discourse theory. In this paper we took a step forward and integrated notions from identity theories into musical discourse theory and presented a newly designed model of sung discourse with the implementation of concepts of author’s identity (Ahonen, 2007). (...) In approaching the narrative experience, the most commonly used method is the linguistic biographical method. It was mentioned for the first time as a method of contact linguistics in the 1991 manual of German linguists Bechert and Wildgen. It is an adaptation of the biographical method firstly used in sociology and applied in analysis of sociolinguistic and contact linguistic issues (Piškorec, 2007: 457). It provides a differentiated perspective on language acquisition processes and language use enabling a direct insight into how individual experience is connected to social context (Piškorec and Zelić, 2006: 278). In European tradition, language biographies are collected through biographical interviews of informants in order to reconstruct their linguistic biographical trajectories (Nekvapil, 2000; Franceschini, 2003). Narrative data are then often triangulated e.g. combined with socio-historical and socio-political data (Abramac, 2014: 96). In Croatian linguistics some papers deal with linguistic construction of ethnic identity (Starčević, 2014; Abramac, 2014; Ščukanec, 2017; Grgić, 2018), certain aspects of personal identity construction (Bertoša and Novak, 2014; Podboj, 2019) and narrative identity reconstruction in linguistic biographical narratives (Piškorec, 2007). This interdisciplinary research will implement the language biographical method within a theoretical framework based on sociolinguistics, dialectology and musical text linguistics. 3. Research objective and hypotheses. The aim of this work is to recreate language biographies of popular music lyricists from the Dalmatian regiolect area and thus define their linguistic identities. The fractal identity theory was researched by determining primary language influences, in the family, and secondary influences of education, peers, and socio-historic linguistic circumstances, which combined result in the creation of one's personal language repertoires, idiolects, language ideologies, and authorial languages, i.e. one's language identity. The first thesis is the existence of a circular process in which firstly socio-historic, linguistic circumstances and language ideologies are directly reflected in authorial languages, with emphasis on the reflection of standardization policies and national identity of the newly formed state of Croatia as well as the influence of rock’n’roll urban culture in the 1990s. Secondly, due to a massive, widespread reach through rapid media dissemination those authorial language choices in popular lyrics influence national language ideologies, the perception and status of the regiolect, and also influence the standard Croatian language norm. The second thesis is rooted in the postmodern notion that a person’s identity is of a fractal nature. Thus, language identity consists of various language varieties that are part of a person’s language repertoire from which the speaker consciously or unconsciously selects language items depending on the speaking situation, topic, location or interlocutor. The repertoire of each person is like a collage gradually built from birth, under the primary linguistic influences of the family, then under the secondary linguistic influences of the educational system and peers, and finally under the influence of the social and linguistic context of the historic period in which the person lives and works. In comparison to individual identity definitions, it could be formulated that a person's linguistic identity consists of the expressed language identity, i.e. the language varieties actually used from the repertoire, and the declared language identity, i.e. varieties one believes to be using. In order to provide arguments for the thesis a contrastive analysis was carried out. The language analysis of informants’ idiolects revealed their expressed language identity and the analysis of self-characterization in the linguistic-biographical narrative revealed their declared language identity. This gave a clear picture of their fractal and layered linguistic identity. Furthermore, in their lyrics, authors also construct and express complex, fractal identities, specifically, the identity of the singer-songwriter (variant 2) or both the identities of author and vocalist (variant 1) are constructed, expressed and transmitted according to the model of sung discourse designed in this dissertation. The aim is to confirm the designed model by inserting the results from the narrative statements analysis into both variant 1 and 2. Taking into consideration claims from the first and second thesis the question that arises is which authorial language varieties from the personal authors' repertoires are used in song lyrics, in which historical periods and with which motivation. The research questions therefore are: are authors’ lyrics a reflection of local and national socio-linguistic changes and language ideologies; with what motivation (personal or social) do authors choose varieties they write in; do the authors use standard Croatian in the nineties; do they use the Dalmatian regiolect prior and after the nineties; do authors write lyrics in their own idiolect or other language varieties and which; what influenced their idiolect (primary influence in childhood and later secondary influences); do they use regiolect archaisms or anglicisms and other loanwords; how do they narratively position standard Croatian and the regiolect; how do they express their narrative identity and do they position themselves locally, regionally or nationally; what was the status of regiolect and standard Croatian language in the public school system and what was/is the status of the regiolect in the media. 4. Methodology and research plan In this research the problem is approached internally, through a qualitative analysis of narratives about the personal experience of authors of popular music lyrics who grew up and worked in periods of crucial changes in terms of national language policy and the status of Dalmatian regiolect. The Croatian language issue is looked at from an internal perspective, primarily in terms of positioning and determining the status of standard Croatian and the Dalmatian regiolect. It is described starting from medieval literature, language policies and changes during the first half of the 20th century, through the second half of the 20th century, the period of the Croatian War of Independence and post-modern aspects of the 21st century. Each historical period brought crucial language policy changes and different attitudes towards the Croatian language and its dialects. Special emphasis is put on the period of the nineties, when Croatia’s identity was forged through language purification, re-standardization and independency from the notion of Serbo-Croatian language. The informants are authors of popular music who were selected based on strictly set parameters regarding the aspects of place, time, relevance, comprehensiveness and number (Plummer, 2001). Hence, they are of Dalmatian regional provenience, members of different generations, authors in different genres (Dalmatian pop, Dalmatian klapa and contemporary hip-hop), nationwide popular with a vast number of popular lyrics. To be more precise in terms of time, informants were selected because they were (1) born around 1945, grew up in post-war Yugoslavia and began acting publicly as authors in the 1960s, at the time of the beginning of regional festivals (Melodije Istria and Kvarner, Split festival) and the popularization of regional varieties; (2) informants born around 1960 who began acting publicly in the mid 1980s and early 1990s; (3) informants born in the early 1970s who began acting publicly as authors in 2000; (4) informants born at the beginning of the 1980s who began acting publicly from the year 2000 and later. In this paper it is researched to what extent authors use the regiolect or standard Croatian language in their texts, in which periods they chose a certain language variety, for what reasons they chose a certain variety, and how conscious or unconscious their language choices are. In order to research their authorial languages a linguistic analysis of complete authors' lyrics was made, collected from the database of the Association for protecting authors' rights (ZAMP). Furthermore, additional data on the respondents was collected from other relevant sources like official biographies on the Croatian Authors Association (HDS) webpage. This approach to the topic is based on the idea that the complex process of the author's public activity at the national level must be studied through the personal experience and self-reflection of the language experience of each informant/author. Such an insider view is the basis of this approach. The methodological framework set for this research is innovative because it includes language biographical studies, qualitative content analysis, but also analytical procedures from text linguistics, contact linguistics, discourse analysis, dialectology and musical text linguistics. The interview method used is that of a semi-structured, narrative interview by means of which informants’ idiolects, i.e. the language of everyday interaction as close as possible to the vernacular, are going to be elicited and recorded. The researcher had open questions and topics prepared in advance so that informants could produce long uninterrupted narratives and so that the course of the interview could be adapted to the unpredictability of answers. The corpus for this research hence consists of (1) recorded and transcribed narratives and idiolects, (2) reconstructed language biographies, (3) texts of popular song lyrics and (4) data from the national Association for protecting authors' rights (ZAMP). The analysis of data was set in seven phases. After transcribing, an extensive discursive macro analysis of the propositional content of the narrative was conducted. It was followed by the analysis of discursive means by which narrative identity is conveyed and constructed in narrative discourse. The third is the linguistic analysis of authors' idiolects recorded in the interview which determined the language variety or varieties the informants speak. The fourth is the lexical analysis of selected authors’ texts which aimed to describe the various authorial languages used in texts and determine to what extent they use archaisms, foreign words, loanwords and hybridized variants of a mixture of varieties. Combined with the data from ZAMP it was determined in which exact years authors used the Croatian standard language, the Dalmatian regiolect or an individual hybrid variety. The fifth analysis is the comparative analysis of idiolects and authorial languages which determined whether authors use, and to which extent or in which circumstances, their idiolect in the authorial language. In the sixth phase all gained results were then triangulated, compared and combined in order to recreate the authors’ language biographies. In the final phase a comparative analysis of all narrative statements according to research topics was made to obtain answers to some of the research questions. All obtained research conclusions were consolidated and the answers to the initial research questions were determined. 5. Results and conclusion This qualitative, interdisciplinary research enriches previous studies in sociolinguistics, dialectology and discourse analysis. It provides an insight into how individual and collective identity is shaped, expressed and mediated through the narrative of personal experience. Results of the discourse macro analysis of their narratives confirmed the research questions as follows. Primary and secondary language influences and overall social, political and linguistic context are reflected in authors’ personal idiolects and in authors’ lyrics. The strongest impact on one's idiolect has the primary influence of parents, or grandparents living in the same household. Some informants have parents speaking different language dialects / varieties and are thus in this paper determined as being linguisitically natural bi-variational speakers and are often fluent speaker of both dialects / varieties. It is significant that the parents of six informants are professors and writers, and were often reading to them in their childhood, which developed their love for literature. Results also illustrate that the usage of dialects in schools was strongly prohibited in the 1950s and 1960s, encouraged in the 1970s and 1980s, and then prohibited again in the 1990s, depending on the socio-historical circumstances and linguistic policies of the time in question. The majority of authors express a strong local and regional identity and some position themselves as representatives of the Dalmatian regiolect. Results of the textual analysis of authors’ lyrics revealed that most authors do not only use their own idiolect, but motivated by various factors they choose different varieties, using them often consciously and purposefully. The authors use mostly (1) the Dalmatian regiolect or the Split vernacular, (2) the standard Croatian language, (3) a personal hybrid authorial language, and less often (4) the Chakavian dialect. The choice of the language variety used in lyrics depends on the purpose of the song, the vocalist who will perform it, the social and linguistic ideologies of the author and the socio-historical context. The main motives for the use of regiolect in popular music are its melodiousness, emotionality and authenticity. The obtained results were compared with the ZAMP data i.e. the years when the lyrics were written. It confirmed the first thesis that authors predominantly used standard Croatian language during the nineties, influenced by socio-historic circumstances. Specifically, data revealed three processes: (1) in the 1990s older authors wrote in standard Croatian language due to political, social and linguistic circumstances of the war and post-war years, in order to join the spirit of national unity and build a post-war collective national identity through popular music, (2) in the 1980s and 1990s young authors from Split wrote in standard Croatian language under the influence of American rock’n’roll culture, in order to be part of the collective rock identity and the newly formed urban Split culture with a deliberate move away from Dalmatian festival music, which was the popular mainstream music at that time, and (3) after year 2000 most authors began writing more often in the Dalmatian regiolect due to the strengthening of Croatian regiolects in the 21st century, personal maturation of some authors, and the Klapa wave phenomenon that arose at the very beginning of the 21st century encouraging the re-popularization of regiolect music and regiolect. This induces the cyclic process in which the more frequent use of a language variety results in its greater widespread, specially when used by some of the most famous authors from the area of Dalmatian regiolect who, through their lyrics, have a great influence on its popularity and prestige nationally and internationally. The prestige of the Dalmatian regiolect was also confirmed because all informants stated that Dalmatia and the Dalmatian regiolect became popular in other Croatian regions and former ex-Yugoslav states thanks to tourism, popular music and the TV series Naše malo misto and Velo Misto by Miljenko Smoje. The research also resulted in a clear picture of today's national and commercial media’s attitude towards the regiolect, which some informants believe has changed, moreover, they stated that a certain marginalization of Split and Dalmatia is currently taking place. These notions should be researched furthermore. The second thesis based on the fractal identity theory was confirmed by triangulating results of the idiolects, narratives and lyrics analyses and thus determining informants’ language repertoires. They consist of various language varieties used consciously or unconsciously in different context and with different motivation. The first question is what their idiolect is consisted of. It has been confirmed that the primary language influence in the family largely shapes the idiolect of the speaker, because it is the first that the child acquires. However, close relationships with grandparents are also reflected in the informants’ idiolects, so linguistic features were found of the Split dialect (Stanić, Huljić), Split Chakavian dialect (Rončević), Brač dialect (Belan, Mirić), Šibenik dialect (Juras), Ploče dialect (Antić) and Ogorje dialect (Ninčević). Secondary influences of the educational system were confirmed in narrative statements that during the 1950s and 1960s it was not allowed to use dialect in school, from the beginning of the 1970s and during the 1980s the use was encouraged, and then change occurred in the 1990s when dialects were once again unacceptable in school due to re-standardization and purist language policies. Secondary influences of peers were confirmed in statements of processes of linguistic adaptation and the adoption of the speech of their peers for the purpose of social integration, i.e. Stanić adapting to the local speech of Perušić, Ninčević adopting lexemes of Split speech, and Rončević observing different varieties of speech in Split city districts. Belan, Gibonni and Antić testify to the development of the new Split urban speech of the young population in the 1980s, Rončević confirms the disappearance of the Split Chakavian dialect, Stanić confirms the development of the newly formed Split dialect. This confirmed the changes in the Split speech that occurred throughout the second half of the 20th century. In this paper we define language identity as the sum of all language varieties that the speaker possesses in his language repertoire, created under the primary linguistic influences of the family and the secondary linguistic influences of the educational system and socio-historical circumstances of a certain period. In order to find out how the informants define their own linguistic identity and repertoire, an analysis of the speaker's self-characterization in the linguisticbiographical narrative was conducted. The results confirmed the existence of different varieties in their repertoires which they use in situational code switching, depending on the interlocutor or the place of the conversation. Differently motivated, some code switch when they want to achieve comprehensibility, fit in a social group or when they don't want to emphasize their regional affiliation. Some even refuse to code switch in order to preserve their identity’s integrity. This gave a clear picture of their layered linguistic identity and how using various language varieties in different context enables construction of partial identities. The results confirmed the constructivist approach in which identities are determined as fluid and dynamic processes (Hall, 2006) which change depending on the social context. In addition to self-perceiving themselves as representatives of the Dalmatian area and regiolect, they also self-perceive themselves as members of collective identities related to local areas (Split, Brač, Kaštela), region (Dalmatia), ethnicity (Croat, Vlaj, Bodul, Varošanin, global identity) and music genres (rappers, rockers, rebels). The results also showed that Juras and Ninčević, as members of the older generation, express a strong national identity, while all the authors of the middle and younger generation do not express national identity, but rather an extremely strong local and regional affiliation. To conclude, construction of collective identities is layered because regiolect speakers are simultaneously members of local, regional and national groups. The analysis of idiolects showed that the majority of informants speak language varieties which they specified in the self-characterization process, but with more prominent features of the variety acquired during primary influence, and with frequent unconscious thematic and conversational code switching between different varieties. Obtained results point to the conclusion that the informants’ expressed linguistic identity mostly or partially coincides with that linguistic identity declared in the act of self-characterization. Regarding the notion of author’s identity, the informants describe it realistically with images of olive growers, fishermen and tailors, metaphorically comparing it to birds, heights and flying, as well as something religious, intangible, invisible. The content macro-analysis confirmed the two variants of the model of sung discourse as described for the first time in this paper. The song Mirakul is an example of variant 1 in which the author expresses his own identity in the song which then reaches the ultimate receiver, i.e. the listener. The song Slon is an example of variant 2 in which the identity of the vocalist, in this case Dino Dvornik, is added to the expressed identity of the author. Gibonni confirms that the song Slon would have been a completely different song if it had not been sung by Dino Dvornik but by him as a singer-songwriter.