A critical discourse study of the representation of immigrant women in a sample of the Spanish press: between victimization and prostitution

This paper analyses the main characteristics of the representation of immigrant women as social actors in a sample from the Spanish press. The data consist of all the news items published on immigrant women in the digital editions of the Spanish newspapers El País , ABC and El Mundo from 1 January 2012 to 1 January 2022. Van Leeuwen’s (2008) social actors’ theory will show how immigrant women are portrayed linguistically in the corpus by examining the main categories observed in the news items analysed: assimilation, role allocation, identification and functionalization. Moreover, corpus linguistics (CL) will be a powerful complementary tool to Critical Discourse Analysis (CDA) because it allows observation of the main patterns and collocations in discourse. The findings indicate that women are not very visible in the corpus and that they are mainly associated with prostitution and low-paid work. Thus, their representation is associated with victimization and vulnerability.


Introduction
Media discourse is important in creating public opinion about social realities, making it an elite discourse because it shows power over people by having an effect on ideology and opinion, at the same time that it is a serious source of information (Bednarek & Caple 2017;Fairclough 1995a;van Dijk 2009). Thus, deconstructing media discourse in general and press discourse in particular allows a deepening of the understanding of the social context that frames the message being transmitted. Specifically, approaching discourse is essential in unveiling social realities, such as the one of immigrant women in the Spanish press. Exploring the discoursive characteristics of immigrant women will shed light on how immigration is represented as a social phenomenon (Bañón 2019).
Press discourse perpetuates the dichotomy 'we'-'they' (van Dijk, 2005), i.e., the majority group, in this case Spanish population, and immigrants. Moreover, othering is highlighted by the general references to immigrants "as a threat to sovereignty" (Holohan 2019: 3), which contributes to perpetuating racist attitudes in people.
The main aims of this paper are: 1) to offer a corpus-based critical discourse analysis of the way immigrant women are referred to in a sample from the Spanish press (2012)(2013)(2014)(2015)(2016)(2017)(2018)(2019)(2020)(2021)(2022) and 2) to observe the main ways they are referred to and portrayed as social actors in the corpus analysed and to observe possible differences in the newspapers under analysis. To do so, some of the most outstanding features of the discourse(s) of immigrant women are examined by considering the categories proposed by van Leeuwen (2008) to analyse social actors in discourse.
Taking into account previous studies (Bringas López 2010; Magalhães & Caetano 2012), the hypothesis of this article is that immi-grant women are not very visible in the Spanish press, they are represented as 'the others' and their portrayal is associated with victimization and prostitution. The main research questions this paper addresses are: in what way are immigrant women portrayed in the Spanish newspapers ABC, El País and El Mundo and, what do we know about them by the way they are portrayed linguistically? Following van Leeuwen's (2008) framework, what are the main categories used to refer to them as social actors?
The perspective offered by CDA is essential for the analysis, since the intention of this research is to explore the relationship between the different linguistic choices used to name and refer to women and the social context framing Spain in a historical period when migrations have been a constant sociological phenomenon. Following the example of Baker et al. (2008), CDA will be combined with CL in order to observe the discourse representation of immigrant women in a sample of the Spanish press over a nine-year period. The notions of keyness and collocation will allow the deconstruction of the lexical patterns associated with immigrant women. Moreover, the categories proposed by van Leeuwen (2008) to refer to social actors in discourse will be considered in the analysis.
After this introduction, the paper is divided in the following way: sections two and three offer the literature review and the theoretical frameworks. Next, I outline the data sources and methodology used for the study. This section is followed by the results and discussion. The article ends with some conclusions based on the analysis.

Literature review
The arrival of immigrants in Spain has taken place in the last decades due to the necessity to leave their countries in order to improve their economic and social circumstances (Arango 2002;Rodríguez Breijo 2010). The representation of immigrants in the press has been studied in the last decades by national and international scholars (Bañón 2019;Bañón & Romero 2013;Cruz Moya 2020;Seijas 2014;van Dijk 2015van Dijk , 2018. There are studies that pay attention to the main European discourses and their effects (Borjas 2015;Eberl et al. 2018;Koser Akcapar & Simsek 2018). Generally, investigations coincide on the fact that immigrants are under-represented and shown as delinquents Immigrants in general and immigrant women in particular immigrate to improve their lives and to leave behind the poverty, injustice, war and violence that they suffer in their countries of origin (Benítez Burgos 2015;Catedral 2018;Martínez Lirola 2014a). Women have been present in migrations since recorded history (Dalouh & Soriano Ayala 2020;Martín 2006;Martín & Sabuco 2006). As Pham (2014: 30) notes: The circulation of women has existed since recorded history. Families and groups circulate women as slaves, as brides, as workers, and as prizes. Although some cultures still practice the blatant trade of women as commodities, modern societies circulate women in much subtler form. For women to partake in their own value or worth as a social and economic actor, their sociocultural milieu has to support and encourage such actions.
Despite women being present in migrations, international research shows that they are under-represented in the press (Bouchoucha 2010;Moore et al. 2012;Sadiqi 2010;Zabos 2010). These studies show that the importance of the female migratory movement is not present in the press nor in society due to the invisibility of women in media discourse, which involves minimization of their importance in society. This is even more reinforced because their testimonies are not considered (Dalouh & Soriano Ayala 2020;Milner 2012).
Studies coincide on the representation of immigrant women in the press as 'the others', not very visible, stereotyped and associated only with some topics such as discrimination, prostitution or being victims of illegal associations, which deprives them of the importance they have in the migratory process (Castagnani & Colorado 2009;Martínez Lirola 2014b;Martínez Lirola & Zammit 2017;Gregorio 2009;Mckay et al. 2011;Pérez Grande 2008;Seijas 2014). The discourses of othering and victimization that appear in the press show immigrant women as disempowered, dependent and vulnerable (Gregorio 2009;Martínez-Raposo 2019;Seijas 2014;Zabos 2010), which justifies the necessity to vindicate inclusion and leave behind stereotypes and exclusion (Francos 2020).
The portrayal of immigrant women in the press affects how people perceive them as individual females or as a group (Igartua & Cheng 2009;Igartua et al. 2011;Marling 2010;Martínez-Raposo 2019;Moore et al. 2012). Schrover and Moloney (2013) point out that it is necessary to refer to the differences between migrant men and women. The necessity to make women more visible is also a recurrent topic (Wagner 2019;Pham 2014). Moreover, women from other cultures are stereotyped and marginalized, normally associated with prostitution (García 2021;Gregorio 2009) or with marginal or low paid work (Amrith & Sahraoui The two theoretical frameworks of this article are CDA and CL. On the one hand, CL concentrates on broad discursive patterns through language patterning alignment techniques, frequency, and statistical significance. On the other hand, CDA focuses on clusters, in-depth analysis of concordance lines and whole articles, in Charles et al.'s words (2009: 1): Discourse analysis prioritizes whole texts and their cultural context, identifying patters that extend across sentences and paragraphs. Corpus linguistics tends to use techniques that decontextualize individual texts and focuses on recurrent patternings of small-scale items such as words and phrases.
Language is a form of social behaviour, and it reproduces and creates social structure, i.e., discourse is socially constructed (Halliday 1978). In this sense, language reproduces ideologies, establishes power differences or legitimizes relationships (Fairclough 2002). Consequently, each language choice relates to its meaning and the context surrounding it. In fact, CDA unveils the relationship between social structures and discourse structures in order to uncover inequalities, exploitation, manipulation or power abuse in society (van Dijk 2017).
Discourse is produced by the combination of different choices with implications in meaning that shape people's ideology. Thus, CDA concentrates on how language and power are interrelated because it sees 'language as a social practice' (Fairclough 1995b;Fairclough & Wodak 1997;Fairclough et al. 2011). Consequently, CDA deepens on analysing discourse from a critical perspective, as Baker et al. (2008: 273) point out: We understand CDA to be an academic movement, a way of doing discourse analysis from a critical perspective, which often focuses on theoretical concepts such as power, ideology and domination. We do not view CDA as being a method nor are specific methods solely associated with it. Instead, it adopts any method that is adequate to realize the aims of specific CDA-inspired research.
CDA proposes interdisciplinary research in order to deepen understanding of social phenomena from different angles (Wodak & Meyer 2009). It is not a method but a perspective within discourse studies (van Dijk 2009). Of the different proponents of CDA, the model of discoursive analysis of social actors proposed by van Leeuwen (2008) will be used in this study. This framework is useful to explore how immigrant women are portrayed in the compiled corpus (see section 5).
The main categories proposed by van Leeuwen (2008) to refer to social actors are: 1) role allocation (agent or patient), 2) generalisation and 3) specification (generic or specific reference, i.e., representation as classes or as individuals); 4) assimilation (representation of people in groups), 5) association (groups of social actors) and 6) dissociation (unformed associations); 7) indetermination (unspecified representation of social actors) and 8) differentiation (specification of the different social actors); 9) nomination (reference to the unique identity of social actors) and 10) categorisation (identities and functions shared with others); 11) functionalization and 12) identification (reference to social actors by something they do or what they are); 13) personalization and 14) impersonalisation (representation of social actors as human beings or not); and 15) overdetermination (representation of social actors as participating in more than one social practice). The application of these categories to the corpus will allow observation of the way immigrant women are portrayed in a sample of the Spanish press and how that portrayal is associated with the context that surrounds them when they come to Spain.
The representation of social actors in discourse has been analysed in the last decade. CL is a method of linguistic analysis and description in which corpora are used as the primary data to find "probabilities, trends, patterns, co-occurrences of elements, features or groupings of features" (Teubert & Krishnamurthy 2007: 6). CL is a useful methodology that helps language analysis using computers to observe regularities in the language by analysing large numbers of texts ( Partington et al. 2013). In this sense, CL is very useful for a critical discourse analysis (Baker 2006(Baker , 2009(Baker , 2010a(Baker , 2010bHidalgo Tenorio & Benítez Castro 2020) due to the objectivity of this method to study how language is used, as Gabrielatos and Baker (2008: 7) make clear:

36
Corpus linguistics methodology allows for a higher degree of objectivity-that is, it enables the researcher to approach the texts (relatively) free from any preconceived notions regarding their linguistic or semantic/pragmatic content. When the starting point is keyword analysis, the analyst is presented with a list of words/clusters which will then be examined in (expanded) concordances for their patterning and contextual use. CL helps to study language patterns in a large number of texts and avoid the criticism of possible subjectivity towards the data (McEnery & Hardie 2012; Pérez Paredes 2021). The basic research methods in CL include key words, collocates, concordances and the generation and study of word lists. It is mainly quantitative and interested in the local context of situation by analysing concordances. Approaching texts in this way allows researchers to be less subject to bias (Baker 2010b;

McEnery & Hardie 2012).
Tognini-Bonelli (2001) established a difference between corpus-based and corpus-driven points of view. The former refers to "a methodology that avails itself of the corpus mainly to expound, test or exemplify theories and descriptions" (Tognini-Bonelli 2001: 65) "in order to validate, refute or refine" them (McEnery & Hardie 2012: 6). The latter "claims instead that the corpus itself should be the sole source of our hypothesis about language" (McEnery & Hardie 2012: 6). In other words, the corpus 'drives' the research in the sense that the analyst observes what is salient to explore in the corpus, and theory is derived from the corpus.

Data and methodology
Data consist of all the news items published in the online version of the Spanish newspapers El País, ABC and El Mundo from January 1, 2012 until January 1, 2022, from Monday to Friday. The motivation behind the newspapers' choice is that they are among the most widely read Spanish newspapers. El País is considered a left-wing newspaper, which contrasts with ABC, which is considered right wing. El Mundo has a neoliberal ideology. The former has always leaned towards social-liberalism and progressivism and El Mundo is considered independent and with liberal overtones (Alcaraz-Mármol & Soto-Almela 2022). ABC supports conservative political views and defends the Spanish monarchy (Martínez Lirola 2022a).
I used the link that the Centre of Study and Documentation on Immigration, Racism and Xenophobia, called MUGAK, sends to people who are registered on their web page to collect the news items related to immigrant women in the three newspapers mentioned.
Then the news items were converted to a txt. format and coded according to the following criteria: the initials of the newspaper: ABC for ABC, EP for El País and EM for El Mundo. The initials of each newspaper were followed by the day, month and year in which the news article was published; the last digit was altered if there was more than one example per day (e.g., EP-120318-2), which makes clear that specific day there were two examples related to women in the news item under analysis. Table 1 shows the number of texts analysed distributed per newspaper and the number of tokens. El País offers most of the examples, 42,62% of the texts under study. It is followed by ABC with 36,65% and EL Mundo with 20,71%. Of the news items collected, 70,91% are multimodal, i.e., more than one mode of communication is observed, and 20,09% are monomodal, i.e, there is only one mode of communication, i.e., only written text in the case of the newspapers under analysis. Once the newspapers and the news items were selected, it was decided to use the software Antconc by Anthony (2020) for concordance analysis and cluster lists. 1 Antconc is a very useful program for the type of analysis intended with this article because it allows researchers to view concordances, to calculate lexical and phrasal frequencies, collocations, keywords, etc. This program was used in order to retrieve a list of concordances containing the search word strings woman (mujer) and women (mujeres), and other words such as victims (víctimas), immigrants (inmigrantes) and prostitution (prostitución) (see Table 2). The reasoning behing using these words is to observe up to what extend immigrant women are represented as victims and associated to prostitution in the corpus.
Observing Table 2, it is clear that although ABC has fewer news items than does El País, the frequency of the word víctimas (victims) is higher in ABC (87) than in El País (79). The same thing is observed about the word inmigrantes (immigrants) because the frequency in ABC is 82, which contrasts with 61 in El País. Finally, the frequency of the word prostitución (prostitution) is also higher in ABC (53)  victimization, immigration and prostitution is higher in ABC, which is a clear way of excluding them from the majority group and portraying them as associated with marginalization.

Table 2. Rank and frequency of the keywords
To evaluate the statistical significance between the frequency of certain terms in the news items of El País (O1) with those of ABC and El Mundo (O2,74,869 tokens between the two newspapers the test for statistical significance proposed by log-likelihood 2 (LL) was used (LL of p < 0.01 (critical value = 6.63)). 3 It was created by Paul Rayson from the Computing Department at the University of Lancaster (Log-Likelihood and Effect Size Calculator, 2020) in order to observe how much evidence we have for a difference between two corpora.
LL is obtained by introducing the frequency in corpus 1 (O1) and the frequency in corpus 2 (O2). A LL of more than 5 is considered significant in social sciences; %1 and %2 values show relative frequencies in the texts, + indicates overuse in O1 relative to O2 and -indicates underuse in O1 relative to O2. By doing so, the results presented in Table 3 were obtained. 4 The data presented in Table 3 shows that referring to women as inmigrantes (immigrants), víctimas (victims), inmigrante (immigrant) and the references to prostitución (prostitution) are significant in the second corpus, i.e., in the news items in ABC and El Mundo, with a minus sign before in %DIFF. This contrasts with the references to women as trabajadoras (workers) and refugiada (refugee) that are significant in the first corpus, without a minus sign in %DIFF. Thus, %DIFF is a simple and straightforward effect-size metric for keyness analysis. It indicates the proportion (%) of the difference between the normalised frequencies of a word in two corpora.
Once the terms mujer (woman) and mujeres (women) were chosen to start the research, the collocations and tokens associated with the said words determined the main categories of the discoursive model of social actors proposed by van Leeuwen (2008) chosen in this article. For instance, the fact that there are hardly any examples of immigrant women referred to by their names is the reason the category of nomination is not included. Moreover, due to space limitations, Table 4 shows the specific categories of the ones proposed by the author that will be used in this article (see section 5).

Table 4. Main categories used to refer to immigrant women in the corpus
The fact that this is a corpus-based CDA study means having a mixed methodology. I make use of both quantitative techniques, that is, frequency and keyword lists, complemented by more detailed qualitative textual analysis. The synergy between CDA and CL allows overcoming their limitations: CDA is critizised for being methodologically weak due to its qualitative analysis and the use of small data. In addition, CL is criticised for neglecting the social context of texts by analysing decontextualized examples of language use.

Results and discussion
Considering the years of analysis, it is noteworthy that the number of texts collected is not high, which implies that immigrant women are not very visible in the Spanish press. Thus, the results of the analysis show how language and ideology are intertwined due to the fact that it is observed a tendency of excluding women from media discourse.
Assimilation (representation of people in groups) or individualization Role allocation (agent or patient) Identification (reference to social actors by what they are) Functionalization (social actors associated with an activity, an occupation or role) 40

Assimilation vs individualization
The first search term used in this study was mujer (woman) and mujeres (women), in order to observe if women were referred to as individuals or in groups (van Leeuwen 2008). The results show that women are referred to as groups in the great majority of the examples because the word in plural (mujeres) appears in 751 examples. In addition, they are mentioned as individuals, i.e., the word in singular ('mujer') appears in in 448 examples (see Table 2). Thus, assimilation (references to people as members of a group) predominates over individualization (van Leeuwen 2008).

Role allocation
Regarding roles, the general tendency is to represent women as patient (goals). A number of examples refer to them as victims (190 times) or victim (102 times). In this sense, van Leeuwen (2008: 33) highlights that: "Activation occurs when social actors are represented as the active, dynamic forces in an activity, passivation when they are represented as 'undergoing' the activity, or as being 'at the receiving end of it'". This is a clear example of appraisement, because referring to women as victims is a way of inviting readers to feel pity for them, following van Leeuwen (2008: 45): "Social actors are appraised when they are referred to in terms which evaluate them as good or bad, loved or hated, admired or pitied". In this sense, one of the main characteristics of the discourse under analysis is the portrayal of women as victims, i.e., as disempowered human beings in different ways. For instance, they can be victims by being sexually exploited or being victims of human trafficking, sexual exploitation, or gender violence, as we can see in these examples: (1) una asociación que ofrece apoyo integral a mujeres víctimas de trata con fines de explotación ( 5 an association that offers comprehensive support to women victims of trafficking for the purpose of exploitation) (EP-240921.txt) (2) las prestaciones que se dan a las mujeres víctimas de violencia de género. (the benefits given to women victims of gender violence.) (ABC-110812.txt)

Identification
Identification is one of the dominant categories in the corpus. Following van Leeuwen ( Mundo), i.e., these words are useful to establish differences between classes of people. Thus, inmigrantes, refugiadas and extranjeras highlights the 'we-they' dichotomy by portraying women separated from the main group; it represents them as others and victims in need of help, as we can see in Table 5. Table 5.

Examples of identification
There are other examples of identification in the corpus. Van Leeuwen (2008: 42) distinguishes three types: a) classification (major categories established by society to differentiate classes of people including gender, age race, religion, sexual orientation, etc.); b) relational classification (social actors are represented by "their personal, kindship, or work relations to each other"; and c) physical identification (references to social actors by their physical characteristics).
Regarding classification, apart from the reference to them as immigrants, refugees and foreigners, women are classified in the ways indicated in Table 6. After the term, the number of times it is used in singular and in plural is offered in brackets.
Of the different terms and examples used to classify women in the corpus, the general references as African, Sub-saharan or Latinamerican are scarce. Similarly, there are some nationalities that are not really represented in the corpus such as Chinese or Dominican. However, the number of references to Moroccan women is significant, being the predominant nationality in the corpus (see Table 6 with 77 examples in singular and 63 in plural). In 16 cases, they are associated with being sea- sonal workers (temporeras) in Spain. Another nationality relevant in the corpus is Nigerian; in this case, the news items analysed show that the association of Nigerian women with prostitution is quite common. (9) embarcación, venía aún en la placenta» Una mujer de origen subsahariano dio a luz este (boat, it was still in the placenta" A woman of sub-Saharan origin gave birth to this) (ABC-210514.txt) Other nationalities: Romanian (8/18), Chinese (9/9), Dominican (12/0).

Examples of relational classification
The low number of examples in this category is complemented by the high number of cases in which women are referred to as victims (see Table 2). In this case, sometimes the reference is general, and in other cases it is specified that they are victims of sexual exploitation, human trafficking, poverty or discrimination.
Regarding physical identification (references to social actors by their physical characteristics), the main examples are associated with being black or white women. Black women are also associated with poverty in several examples, which reinforces the exclusion of women through race (Bringas López 2010; Magalhães & Caetano 2012). This contrasts with the reference to white women that are normally associated with white slavery, as we can see in Table 8.

Functionalization
Finally, the category of functionalization is also noteworthy in the corpus because it highlights the function of individuals, i.e., "[…] social actors are referred to in terms of an activity, in terms of something they do, for instance, an occupation or role" (van Leeuwen 2008: 42). The number of cases of women associated with prostitution is significant in the examples compiled. There are 106 references to prostitution (see Table 2). Examples 25 and 26 illustrate this category: (25) el medio y que informan diariamente sobre 250 mujeres que se dedican a la prostitución en (the medium and that they report daily on 250 women who are engaged in prostitution in) (ABC-200814.txt) (26) sanidad pública, vuelve a confirmar su expulsión. Mujeres que ejercen la prostitución, a menudo "nómadas" (public health, reconfirms his expulsion. Women in prostitution, often "nomads") (EP-040912.txt) The abundant references to prostitution contrast with the few references to women with prostitute (9 times) and prostitutes (30 times). In other words, instead of identifying women with being prostitutes, in most cases we find collocations like the ones presented in the previous examples. The association of women with prostitution contributes to their social exclusion and to stigmatizes them and separate them from the majority group by portraying them as 'people problem' This is another example of "appraisement" (van Leeuwen 2008: 45) because prostitution is evaluated negatively by society as good or bad, loved or hated, admired or pitied.
The number of cases of women associated with being workers is not so high: 9 as worker and 64 as workers. Of these examples, the collocation trabajadoras sexuales (sexual workers), which is associated with the previous examples about prostitution, appears in 10 cases, whereas the collocation trabajadoras del hogar (domestic workers) appears in 19 cases in examples such as the following: (27) persona inmigrada, una mujer, sin papeles y trabajadora del hogar, representa un plus de vulnerabilidad", (an immigrant person, a woman, without papers and a domestic worker, represents a plus of vulnerability",) (EP-300819.txt) (28) e alimentos en dos ocasiones, pero las trabajadoras sexuales tuvieron que ir a recogerlos, hacer (and food on two occasions, but the sex workers had to go and collect it, make) (EP-250820.txt) The four categories of the ones proposed by van Leeuwen (2008) presented in this section are used recurrently for the representation of immigrant women as social actors in the example of the Spanish press analysed. Thus, there is a limitation in this study because there are other categories that need to be explored in future articles.
The previous analysis has also confirmed the hypothesis presented in section 1: the fact that the number of texts collected during ten years is not very high foregrounds that the said women are not very visible in the Spanish press. In this sense, observing how immigrant women are represented in the Spanish press allows a deepening of the understanding of the ideology behind the representation. Thus, the scarce visibility of immigrant women is associated with the lack of recognition of women in society and with the necessity of foregrounding them in order to advance in the construction of more equal societies where women's roles and contributions are clearly portrayed (Moore et al. 2012;Wagner 2019). In addition, the invisibility of immigrant women is associated with the lack of empowerment and positive representation, which affects all aspects and areas of their individual and collective lives (Zabos 2010).
The analysis shows that the invisibility of women is reinforced by not presenting their testimonies in the news items (Dalouh & Soriano Ayala 2020). Moreover, the analysis confirms the general tendency to represent women as 'others', portraying them as victims of sexual exploitation or human trafficking and associated with prostitution in many examples (Gregorio 2009; Martínez-Raposo 2019; Seijas 2014) or with low-paid work (Amrith & Sahraoui 2018). In this sense, the partial representation of immigrant women observed in the corpus and the combination of invisibility, victimization and othering does not favour their integration in the Spanish society.
CL has allowed the observation of the patterns of use of the words immigrant woman/women (mujer/es) in the corpus. In this way, the analysis has proved that immigrant women are not very visible in the Spanish press for different reasons: 1. There are not many texts on women; 2. The main linguistic patterns associated with women are limited and mainly associated with victimization, prostitution and marginal work, as the process of checking the collocates of the word 'woman' has shown.
The collocations observed with women/woman have been essential to specify the references to women in the corpus analysed and the main categories of social actors established by van Leeuwen (2008) observed in the corpus. In this sense, the analysis has shown that women are identified with being immigrants, refugees and foreigners (see Table 5). In addition, they are identified with their nationalities (mainly Nigerian and Moroccan) or with more general words such as African, Latinamerican or sub-Saharan (see Table 6) or with the colour of their 46 skin (see Table 8). The findings of the analysis also show the way in which women are 'functionalized', i.e., categorized by what they do, which is mainly associated with being prostitutes or domestic workers.
Thus, the combination of CL and CDA has been useful for the objectives of this article because I have been able to deepen my understanding of the way immigrant women are linguistically represented in a sample from the Spanish press. The discourse analysed belongs to hegemonic discourses that associate immigrant women with victimization, which maintains the dichotomy of 'we-they' or 'us-them'.
The analysis has shown media constructions of immigrant women in an attempt to reveal what this might suggest about how they are portrayed in the selected newspapers. Thus, the analysis has offered information about the social situation of immigrant women. The choice of van Leeuwen's (2008) framework for the verbal representation of social actors in discourse sheds light on the representation of immigrant women in the sample of the Spanish press mentioned.

Conclusions
This article has been an example of the use of CL for a critical discourse analysis of newspaper articles, which allows observation of the way immigrant women are portrayed in a sample from the Spanish press. The texts analysed show that immigrant women are portrayed as victims, prostitutes and are associated with low-paid work. The use of CL in the analysis has shown that it is a complementary tool to CDA when examining women's representation in the Spanish press. Thus, CL has tracked the distribution of lexical items and collocates in the corpus, and CDA has tracked the relationship between the discourse patterns analysed and the social situation being portrayed. The social perspective of CDA has been useful to see how each linguistic choice has a social purpose and how discourse can reproduce discrimination, marginalization, or power inequality.
This article has shed light on the opaque situation of immigrant women in Spain and has tried to unveil the way in which they are portrayed in discourse in order to vindicate their rights, the importance of the work they do and the way they contribute to the Spanish economy. In addition, the study has contributed to deconstructing the ideology of the texts analysed and the binary opposition of 'we-they', which does not favour the integration of immigrant women in Spain nor make visible their contribution to the country of arrival.
The limitations of the present study are the following: on the one hand, as already mentioned, this article has concentrated on the analysis of only four categories of the ones proposed by van Leeuwen (2008), i.e., assimilation, role allocation, identification and functionalization. Thus, the analysis of other categories such as indetermination, differentiation or overdetermination in future studies will shed light on the way immigrant women are represented in the Spanish press. On the other hand, using three Spanish newspapers is also a limitation. In this sense, other articles should incorporate more Spanish newspapers in the analysis to observe if the results are similar to the ones obtained in this paper. In fact, it would also be possible to incorporate some international newspapers and prepare a contrastive study in order to explore if immigrant women are associated with victimization and prostitution in other countries. In addition, the number of key words used in the study are also limited due to space constraints. For that reason, in future studies, other key terms could be used such as empoderadas (empowered), estudiantes (students) or word that refer to specific jobs. The use of the said key words would be useful to observe up to what extend immigrant women are also portrayed in a positive way.
A corpus-based discourse analysis using van Leeuwen's (2008) categories of social actors has been used to observe how immigrant women are referred to in El País, El Mundo and ABC. It shows how journalists construe immigrant women and their roles in the period under study. The analysis confirms that there is a narrow view of immigrant women due to their partial representation, which does not contribute to seeing them as active citizens. In fact, this partial representation deprives women of power and of social justice. Consequently, it is necessary to rethink the depiction of immigrant women in the press and to rethink the role of women in the migratory movements so that the media can advance in a better representation of immigrant women's realities and contribute, in some way, to vindicate more equal societies.
María Martínez Lirola Departamento de Filología Inglesa/Department of Modern Languages Universidad de Alicante/University of South Africa (UNISA) maria.lirola@ua.es ORCID:orcid.org/0000-0002-6427-425X Recepción: 20/10/2022; Aceptación: 03/01/2023 Notes 1 In a frequency list, AntConc ranks all the words a corpus consists of by frequency of occurrence. A cluster allows researchers to look at a given number of words that occur most frequently surrounding the keyword. 2 See log-likelihood calculator at http://ucrel.lancsac.uk/llwizard.html 3 The critical value (6.63) corresponds to 99% percent possibility that the results are not due to chance. 48