FACULTY OF SOCIAL AND BEHAVIORAL SCIENCES
‘Nothing about us, without us’: the
construction of sex work in the Dutch media
Master thesis submitted by Mehri Zamanbin
Supervisor: dr. D. J. van der Pas Second reader: dr. S.A. Bonjour
Introduction
After a twenty-year discussion, the ban on brothels (“bordeelverbod”) was removed from the Dutch Code of Criminal Procedure in 2000 (Proud and Aidsfonds- Soa Aids Nederland 2018: 10). Although sex work has never been illegal in the Netherlands, until 2000 it was punishable to run a brothel. After the amendment, municipalities had the authority to decide whether sex workers can start a ‘sex company’ in their city. Hence, municipalities determine the maximum number of sex workers in their city and set the minimum age for sex work (eighteen or twenty-one years old). After receiving a permit by their municipality, sex workers can register at the Chamber of Commerce to start a company (Ondernemersplein 2019).
Even though sex work is acknowledged as a legal profession, sex workers face various challenges due to the service they offer. It is relatively laborious for sex workers to get a permit from municipalities, not all banks are willing to open accounts and health insurance companies often deny sex workers based on ethical presumptions or their high-risk profile (Proud and Aidsfonds- Soa Aids Nederland 2018). Furthermore, it is difficult for sex workers to get residency as housing corporations fear that sex workers will work from home (ibid). Thus, Dutch sex workers are denied rights that other workers in the social service industry do have. In addition, in April 2019, more than forty thousand people signed a petition favoring the criminalization of the purchase of sex. The motives behind the movement that started this position was to ‘save’ women in de sex industry and tackle human trafficking (NOS 2019; Parool 2019). This leads to a paradoxical situation: on the one hand, sex work is a legal profession in the Netherlands, whilst on the other hand sex work is associated with human trafficking and immorality.
In order to understand society’s attitude towards certain phenomenon, scholars within social sciences often turn to the media. According to Hunt and Hubbard: “one of the pivotal sources of shared knowledge, prejudices and ideologies in society is the public discourses created and
perpetuated by symbolic elites like journalists” (2015: 26). Consequently, the function of the media is twofold; (1) the media influences society’s values and (2); the media reflects society’s values and believes (Hunt and Hubbard 2015; Gabrietalos and Baker 2008). In addition, newspapers have a financial interest in reflecting the views and attitudes of their readers. Newspapers decide what content will be highlighted or rejected according to their “editorial agenda” (Gabrietalos and Baker 2008: 9). The media is an important actor in perpetuating certain discourses. To answer the question why sex work is treated different than other forms of labor, this thesis will thus assess the role of the Dutch media. This will help identify why sex workers fall victim to several forms of prejudice and exclusion.
This thesis aims to provide insight in the construction of sex work in the Dutch media. Currently, there is only one empirical study, conducted by Hunt and Hubbard (2015), that focusses on the representation of sex work and sex workers in the media. However, Hunt and Hubbard primarily
focus on two South African newspapers and only analyze one specific moment in time. For this thesis, I will investigate how sex work is constructed in the Dutch media looking at multiple newspaper outlets. By studying newspaper articles from 1995 until 2018, I am also able to study how the construction of sex work changed over time. Hence, the main research question in this study is formulated as following: how is sex work constructed by the Dutch media from 1995 until 2018? I break the research questions down in two components:
- what are the discourses discussed in newspaper articles that are related to sex work? - how did the construction of sex work change over time?
Further, developed in the field of computer science, topic modelling is slowly becoming a more popular tool within social sciences. However, most scholars use topic modelling to study framing, themes or issues (Jacob and Tschötschel 2019: 4). Topics modelling is also a useful tool to study discourses (ibid). For example, Törnberg and Törnberg combined critical discourse analysis with topic modelling to study the representation of Muslims in Swedish Internet forums (2016). Thus, besides providing empirical evidence of the construction of sex work by the Dutch media, this thesis contributes to the existing literature by implementing a topic model in order to study discourses. This study proceeds by first discussing discourse theory after which I argue, in line with Cunningham, that studying discourses is useful in order to assess the construction of sex work. Second, I will discuss and review existing literature to identify anti-sex work and pro-sex work discourses. Third, I will describe how I collected the data of this study followed by a description of the method implemented. Thereafter, I will present the empirical results of the topic model and time analysis. Finally, I will conclude with a discussion and some recommendations for future research.
Theoretical background
Discourse(s)
Before I give a definition of discourse, it is important to underline that there are many definitions of discourse (Cheek 2004: 1142). The definition of discourse that I will use is influenced by the work of Michel Foucault and postmodernism. Discourse “provides a set of possible statements about a given area, and organizes and gives structure to the manner in which a particular topic, object, process is to be talked about” (Kress as quoted in Cheek 2004: 1142). It is through discourses that we speak and think a certain way about reality (ibid). There are different possible ways of thinking and speaking about reality, implying that there are always multiple discourses (Cheek 2004: 1143). However, we do not attribute equal authority to every discourse: the knowledge from one discourse can be used to exclude other discourses (ibid). This makes discourses powerful as it determines who can speak with authority and who cannot.
and political interpretations (Torfing 2005: 4-5). Looking at sex work, it is through discourses that we interpret the practice of offering sexual services and give meaning to it. Worldwide, every country adapts its own laws concerning sex work. The legal and political responses vary heavily between countries – in Sweden the demand is criminalized, in Iran sex work is illegal whilst in New Zealand sex work is decriminalized. In each country there are multiple discourses, created by different political and social actors, that influence the legal responses taken against sex work (Cunningham 2016: 48). Hence, it is not the act itself (e.g. offering a sexual service in exchange for money) that gives meaning to sex work, but it is the variety of discourses surrounding it (ibid). Otherwise we would have, worldwide, a uniform response to sex work. In line with Cunningham, I argue that studying discourses is useful in order to explore how meaning is attached to the notion of ‘sex work’ (ibid).
Besides being powerful, discourses are productive as they create subjects and their identities (Hahn and Holzscheiter 2013: 500-501). It is through discourse that we attribute identities to
individuals in society, such as ‘immigrant’ or ‘sex worker’ (ibid). For example, we attribute the ‘victim’ identity to sex workers when we affiliate human trafficking with sex work. By doing so, sex workers are being classified through discursive practices. Sex workers are in this example the political subjects. To sum up, discourses are powerful because (1) they define what is acceptable to be said and thought and (2) they categorize individuals. By categorizing individuals, discourses create identities and their subjectivities (Foucault 1981).
This is where subject agency becomes an important focus within discourse theory. Agency refers to the extent to which political subjects are able/free to use a specific discourse and the extent to which we are all situated within discourse (Bacci 2005: 198). Specifically, the question that some discourse scholars try to answer is; to what extent are political subjects able to use discourse? Are political subjects always situated within discourse? This question is important regarding the position of sex workers and the discourses that are related to sex work. To what extent are sex workers situated within dominant discourse(s) or are they able to respond/move within these discourses?
Foucault is most frequently aligned with viewing political subjects as constituted in discourse (Foucault 1981). According to Foucault, there is a system of exclusion at play when we look at discourses. In order to illustrate this statement, Foucault uses the ‘madman’ as an example. The ‘madman’ can use his voice, but his voice will not be accepted. Within societies, there is a distinction between those who can speak and are heard (and thus have a privileged position), and those who speak but their voices are rejected (Foucault 1981: 53). The experiences of sex workers are often marginalized as anti-sex workers argue that sex workers are victims – either of ‘false consciousness’ or coerced into sex work (McClintock 1993: 7). Sex workers are thus the ‘madman’ in Foucault’s example.
Foucault’s notion of viewing subjects constituted within discourse makes it seem as if subjects will never be able to produce or resist certain discourses (Miller 1990: 122-123). It is,
instead, important to also explore how political subjects (e.g. the ‘madman’) resists and counters discourses that silence and attribute an identity to him. The political subject in a certain discourse is not merely a subject without any agency, the political subject in fact has possibilities in how they can move within a certain discourse (Miller 1990: 123). Cunningham points out that “if discourse is never fixed and always open to ‘dislocation’ then this means that subject positions, the products of
discourse, are also never fixed and static and can be challenged, resisted and remade” (2016: 48). In relation to sex workers this is important to take into account – their position is marginalized within societies and their experience and voices are often rejected (McLaughlin 1991; McClintock 1993; Levy and Jakobsson 2015; Dziuban and Stevenson 2015). However, as the following paragraphs will illustrate, sex workers are resisting discourses that are producing a certain image of their identity.
What is sex work?
Sex work can, generally, be defined as the exchange of money for sexual services. These services include work provided by erotic dancers, strippers, escort, actors and models in the porn industry, phone sex and so on (Overall 1992: 709). Pro-sex worker movements prefer the word sex worker instead of prostitute, as the latter has a negative connotation. The term sex work was coined by Carol Leigh in the late 1970s (Berg 2013: 693). According to pro-sex worker activists, the term prostitute does not cover the dynamics and complexities of the sex industry. In addition, the term is not neutral. Pheterson (1990: 398) illustrates in her study how researchers contribute to the construction of the category ‘prostitute’. This happens by attaching different meanings to the category. For instance, studies in which prostitution is associated with drugs and abuse, researchers tend to overlook the fact that these issues are not always a part of the category ‘prostitute’. By doing so, researchers contribute to prejudices that are affiliated with the term ‘prostitute’. Pheterson argues that we have to re-think how we use the term ‘prostitute’ and also understand that it is more based upon the ‘bad women’ or ‘whore’ representation than an actual representation of sex workers. Altogether, in this study, the term sex workers will be used when referring to individuals who are offering sexual services.
Anti-sex work discourses
Throughout the years, the anti-sex work movement produced several discourses. Based on the literature on sex workers, we can identify several anti-sex work discourses. We can point out five “major” discourses that are framed as anti-sex work. These discourses can be summarized as
following; (1) sex work as immoral, (2) sex workers as victims, (3) sex work as affiliated with crime, (4) sex work as affiliated with health hazards and (5) sex work as affiliated with human trafficking. These discourses are not fixated in a chronological time order. That is, certain discourses can be more dominant than others and this changes over time. In addition, discourses can also overlap as the following section will illustrate. For instance, the ‘sex workers as victims’ discourse co-occurs with
the ‘sex work as affiliated with crime’ and ‘sex work as affiliated with human trafficking’ discourses. First, the ‘sex work as immoral’ discourse started in the beginning of the 20th century. Sex workers were being studied in order to understand their ‘sick’ personalities and motives to become a sex worker (Vanwesenbeeck 2001: 243). In the Netherlands, around this time, sex workers were perceived as ‘fallen’ or ‘sinful’ women (Outshoorn 2012: 234). The ‘sex work as immoral’ discourse is perpetuated by church organizations and radical feminists. In 1909, the Dutch government
considered sex work as an ‘evil of immorality’ and introduced the Act against Immorality (Post et al. 2018: 9). The legislator of this act argued that sex workers must be approached as victims of immoral behavior, and brothel owners were viewed as the cause of this moral degeneracy in society (ibid). Hence, sex work remained legal but brothels and pimping were banned.
The ‘sex work as immoral’ discourse is reflected in the ‘whore stigma’ that is projected on the identity of sex workers. The argument proposed here, is that by supporting sex workers, one is
supporting men’s indiscriminate access to the sexuality of women (McClintock 1993: 2). The sex worker is selling her honor and her body to a man, for a certain period, in which he ‘buys’ her consent and can do whatever he wants. As such, the ‘whore stigma’ entails that sex work involves no more than a woman ‘selling her body’. In this discourse, sex workers are perceived as being deviant and immoral, as their sexuality is non-reproductive, commercialized and non-monogamous (McClintock 1993; Vanwesenbeeck 2001; Hunt and Hubbard 2015).
Second, the ‘sex workers as victims’ discourse started in the second half of the 20th century. Being victim of childhood trauma and abuse became an explanatory factor of why sex workers entered their field of work (Vanwesenbeeck 2001: 244). Before 1911, brothels were allowed in the Netherlands. Abolitionist feminists and protestants campaigned in the Netherlands against sex work in which sex workers were seen as women who needed saving (Outshoorn 2012: 234). Thus, alongside the ‘sex work as immoral’ discourse, the ‘sex workers as victims’ discourse was perpetuated in order to push the Act against Immorality on the political agenda. This campaign was successful and resulted in the Morality Laws which prohibited brothels. In addition, homosexuality and abortion were also criminalized by the Morality Laws. In practice, authorities did not fully abide by the Morality Laws and as long as the public order was not disturbed by the sex industry, they looked the other way (ibid). In line with the ‘sex workers as victims’ discourse, is the view of radical feminists that sex work is the “absolute embodiment of patriarchal male privilege” (Scoular 2004: 343). Radical feminists have been very vocal about their understanding of sex work as violence against women. Within this discourse, it is argued that sex work reduces women to objects because their bodies are ‘up for sale’. Moreover, the exchange of money for sexual services is inherently exploitative, because there is an unequal power dynamic at play (Scoular 2004: 344). The sex worker becomes merely an object and this objectification is a form of oppression. This discourse also implies that sex work does not represent freedom or choice (Scoular 2004: 345). Radical feminists even drew a comparison between sex work and slavery (Levy and Jakobsson 2015: 594). Remarkable about this discourse is
the fact that it only focusses on cis-women and girls; men or trans sex workers are barely mentioned. This ‘sex workers as victims’ discourse keeps re-occurring until this day. A modern-day example of this discourse is the current Swedish abolitionist law (Levy and Jakobsson 2015). In 1999, Sweden criminalized the purchase of sex (Levy and Jakobsson 2015: 594). Proponents of this discourse argue that sex workers need to be ‘rescued’. As a result, clients of sex workers are criminalized in an attempt to abolish sex work. The case of Sweden is crucial because it illustrates how discourses can be embodied in law and therefore underlines how powerful discourses can be. Third, we identify the ‘sex work as affiliated with crime’ discourse. One of the goals of the 2000 amendment was getting rid of criminal elements of the sex industry. In this amendment, the parliament made a distinction between forced and voluntary sex work (Outshoorn 2012: 235). Forced sex work had to be combatted whilst voluntary sex work should be regulated. Around 2005, there was dismay in the Netherlands about ‘loverboys’. ‘Loverboys’ are young men who force vulnerable girls into the sex industry (Outshoorn 2012: 238). This also gave rise to the discourse that women in the sex industry needed saving (hence, the ‘sex workers as victims’ discourse made its reappearance). At one point, within the discourse on ‘loverboys’, non-native Dutch men were portrayed as the
perpetrators. Antillean men or Dutch-Moroccan men were usually portrayed as ‘loverboys’ who were forcing young white girls into the sex industry (Outshoorn 2012: 238). Hence, pimping became something that became associated with non-native Dutch men. The notion of ‘loverboys’ was
followed by an evaluation of the Wetenschappelijk Onderzoek- en Documentatiecentrum (WODC) in 2005-2006 in which the WODC concluded that pimping was still an issue in the Dutch sex industry (Daalder 2007). In 2007, in the coalition agreement of the new cabinet (Balkenende IV), the Dutch sex industry was framed as a breeding place for crime (Outshoorn 2012: 240).
Fourth, the ‘sex work as affiliated with health hazards’ discourse. Drug abuse and health hazard are used as a way to illustrate how the sex industry should be abolished. In the Netherlands, around 2005, two members of the Amsterdam City council published a report stating that there is no such thing as voluntary sex work. Included in this report is the statement that, besides considerable drug abuse in the industry, sex workers run the risk of sexual transmitted diseases (Outshoorn 2012: 238). Vanwesenbeeck illustrates in her study (2001: 245) how in the 1980s and 1990s, most studies on sex work were focusing on the relationship between sex work and health hazards such as HIV. Finally, the last discourse this thesis identifies is the ‘sex work as affiliated with human trafficking’ discourse. This discourse also perpetuates the ‘victim’ identity of sex workers (Hahn and Holzscheiter 2013:502). At the end of the 1980s, influenced by anti-sex work feminist movements, international NGO’s and other political actors started taking up this discourse (Hahn and Holzscheiter 2013: 512). In the Netherlands, feminists and women’s policy agencies lobbied to get human
trafficking on the Dutch political agenda (Outshoorn 2012: 236). We see this shift come into effect around 2009, when the Dutch parliament introduced a new bill in order to tackle abuse in the sex industry (Outshoorn 2012: 233; Dziuban and Stevenson 2015: 32). Gradually, Christian Democrats in
parliament started to link trafficking to illegal migration (Outshoorn 2012: 236). This eventually resulted in refusing work permits to individuals from non-EU countries who wanted to work in the sex industry. Even though this was framed as combatting forced sex work, the underlying motive was a fear of ‘foreign’ sex workers entering the country (ibid).
Pro-sex work discourse
Sex workers and their allies started their own movement and organizations in order to counter anti-sex workers discourses. Their slogan ‘nothing about us, without us’ illustrates how sex workers and their allies are demanding that the experiences of sex workers are taken into account. The pro-sex work discourses came into life as a reaction to the anti-sex work discourses. Consequently, we can point out one pro-sex work discourse1, namely the ‘sex work should be recognized as real work’ discourse. There are different arguments within this discourse; (1) sex work is like any other form of labor, (2) sex workers have agency and (3) sex work should be decriminalized. I will discuss these arguments in the following paragraphs.
First, pro-sex workers argue that sex work is like any other form of labor and should thus be treated as other forms of labor (Overall 1992; Berg 2014). The reason Carol Leigh coined the term ‘sex worker’ in the 70s, was to unify workers in the sex industry and to describe the labor they were performing (Berg 2014: 693). In order to understand why sex work is condemned by many feminists and scholars, Overall (1992: 709) raises some questions regarding sex work; why do we find sex work morally worse than other forms of labor, such as cooking or cleaning work? Under capitalism, we all have to sell our labor power in exchange for money. The pro-sex workers movement demands that sex workers are given the right to exchange sexual services on their terms and conditions and not on “terms of the state, the police, pimps, male managers or clients” (McClintock 1993: 4). Within this argument, the focus is on emphasizing that sex work should have the same labor rights as other professions (Dziuban and Stevenson 2015: 5; Hahn and Holzscheiter 2013: 517).
Dutch feminists in the early 1980s moved from a radical anti-sex work narrative towards a pro-sex work narrative. They recognized voluntary sex work as real work (Post et al. 2018: 10) and argued that voluntary sex work should be legalized (Outshoorn 2012: 234). Lifting the ban on brothels in order to improve the position of sex workers was pushed on the political agenda through efforts of the Women’s Movement and their allies in the government. As a result, sex work was not perceived as an issue of morality, but instead became mainly a juridical issue (Post et al. 2018: 10). First, authorities tried to ban brothels on morality grounds, but this shifted into a focus on preventing harm being done to sex workers (Post et al. 2018: 11).
1 Drawing from the definition of discourse used in this study, the pro-sex work discourse contains several arguments arguing
for the recognition of sex work as real work. As such, these arguments are not three separate discourses but are instead three arguments within the same discourse.
Second, pro-sex workers argue that sex workers have agency. By focusing on the motives by which sex workers choose to do their work, sex workers’ organizations try to diversity their identity (Hahn and Holzscheiter 2013: 518). Within this argument it is highlighted that sex workers are capable of making decisions about their work and articulate their needs (Dziuban and Stevenson 2015: 7). By doing so, the pro-sex workers movement is debunking the victim identity that is ascribed to sex workers. Dutch feminists in the 1980s underlined the agency of women in the sex industry and rejected the idea that legal sex work would result in increased trafficking (Outshoorn 2012: 234). This shift in attitude is not unexpected; in the 1960s, the attitude towards sexual freedom was changing in the Netherlands. The focus was based on liberal ideas about individuals’ rights in which sex workers should have the right to offer sexual services for money (Outshoorn 2012: 235). Important here is the argument that the state should not intervene in the sexual behavior of its citizens (Post et al. 2018: 10). Further, McClintock shows internal contradictions in the discourses of the anti-sex workers. On the one hand, sex workers are condescended and silenced as having an inherent lack of agency; through being a victim of ‘false consciousness’ or being forced into sex work (1993: 9). On the other hand, they are condemned for having an excess of agency; because they are catering to the sexual fantasies of men and commodifying their bodies (ibid). Additionally, according to pro-sex workers, the ‘whore’ stigma is problematic because sex workers provide a service after carefully negotiating with their customers. This includes negotiating the time and demanding respect from clients (McClintock 1993: 4). Hence, sex work is not about ‘selling’ your body and losing all forms of autonomy, but it is about offering a sexual service.
Third, pro-sex workers are advocating for decriminalization. Sex workers’ organizations emphasize that criminalizing sex work is problematic because it pushes sex work underground. This also prevents sex workers from calling the police whenever a client is dangerous to them. Thus, criminalization puts more power into the hands of clients, preventing sex workers to empower themselves and combat forced sex work (McClintock 1993: 5). The case of Sweden illustrates that criminalizing clients of sex workers will not decrease the purchase of sex, but instead will force sex work underground (Levy and Jakobsson 2014; 604). Since 2009, there is an increase in international organizations (such as Amnesty International and Human Rights Watch) recognizing sex workers’ rights and call for decriminalization (Dziuban and Stevenson 2015: 17).
Criminalizing sex work puts sex workers in a more vulnerable position (Levy and Jakobsson 2014). Another example is the Dutch system of not granting work permits to non-EU sex workers; their work is pushed underground, making them work in poor conditions. This also makes them a vulnerable group in the Dutch sex industry (Outshoorn 2012: 239). Hence, pro-sex work
organizations are demanding decriminalization and not legalization. Legalization will place sex work under criminal law instead of commercial law (Post et al. 2018: 6). And, as discussed in the paragraph above, sex workers want their work to be recognized as any other form of labor.
In addition, sex workers and their allies criticize the anti-sex workers movement for only referring to cis-women and girls when discussing sex work (Dziuban and Stevenson 2015). This view is problematic because a very large number of sex workers are men or trans. Anti-sex workers also tend to overlook that women are clients of sex workers as well.
Expectations on discourses in the Dutch media
Discourses become dominant when they are repeated; the mass media plays a role in (re)producing these repeated discourses (Hunt and Hubbard 2015: 28). The media does not only reflect society’s values and believes – it also shapes and influences society’s values and believes (ibid). An overview of the a-priori identified discourses are presented in table 1. The time frame of the discourses is also presented; (1) the moment the discourse made its appearance and (2) the moment when the discourse becomes prominent through a certain event.
The ‘sex work as immoral’ discourse was perpetuated in the beginning of the 20th century. Thereafter, this discourse did not make its reappearance. Instead, sex workers became increasingly portrayed as victims and anti-sex workers argued that sex work should be abolished as it was
affiliated with crime, health hazards and human trafficking. Due to this observation in the literature on sex work, I do not expect to find the ‘sex work as immoral’ discourse to be present in the Dutch media. Consequently, I expect four anti-sex work discourses and one pro-sex work discourse to be present in the Dutch media.
Table 1. Overview of a-priori identified discourses regarding sex work. A-priori identified discourses Time frame
Anti-sex work ‘Sex work as immoral’ • Beginning of 20th century.
‘Sex workers as victims’ • Started in second half of 20th century.
• Is constant over time. Makes reappearance in ‘sex work as affiliated with crime’ and ‘sex work as affiliated with human trafficking’ discourses.
‘Sex work as affiliated with crime’ • Goal of the 2000 amendment was getting rid of criminal elements of the sex industry.
• Around 2005 there was a lot of dismay in the Netherlands about ‘loverboys’.
• In 2007, in the new cabinet pact, sex work was framed as a breeding place for crime.
‘Sex work as affiliated with health hazards’ • In the 1980s and 1990s most published studies on sex work affiliated sex work with health hazards.
• In 2005, members of Amsterdam City council published a report in which they affiliate drug abuse with the sex industry.
‘Sex work as affiliated with human
trafficking’ • In 2009, the Dutch parliament introduced a bill in which human trafficking in the sex industry should be combatted Pro-sex work ‘Sex work should be recognized as real work’ • In the 1960s, attitude towards sexual freedom changed in
the Netherlands. Sex workers should have the right to offer sexual services.
• In the 1980s Dutch feminist started recognizing sex work as real work.
• Since 2009 there is growing support from international organizations recognizing sex workers’ rights.
• This discourse is constant over time.
Data and Methods
Content analysis can be applied to all forms of mass media; newspaper articles, radio fragments, television programming, magazines and so on (Macnamara 2012). Instead of analyzing visual data or radio fragments, this study turns to newspaper articles in order to study the construction of sex work in the Dutch media. Newspaper articles are best suited for the purpose of this study due to their availability and accessibility. In addition, newspaper articles give us the opportunity to analyze a large quantity of text in a systematic way and it enables us to study discourses over time (Jacobs and Tschötschel 2019: 2). The availability and accessibility of the data also creates a great opportunity for other scholars to replicate my findings.
The textual data for this study is extracted from the digital archive LexisNexis. The newspaper articles are manually downloaded from LexisNexis using search strings related to sex work. To prevent the risk of excluding relevant articles (e.g. when articles discuss sex work
indirectly), I included as many relevant words in my search string. The results of the search string are presented in the appendix (table 1). First, I looked up how many articles I could find searching for one word. Second, I manually looked through the articles to see if the articles are discussing sex
work/related to sex workers. Consequently, I left out words such as pooier because I am focusing on the construction of sex work. Pimps are working in the sex industry but are not sex workers
themselves. Finally, I collected the dataset using the following search string: prostitutie, sekswerker, sekswerk, prostituee, prostituee, prostituees, prostitueés, bordeel, hoer, bordeelverbod, seksindustrie, prostitutiesector, prostitutiebeleid, seksbedrijf, seksbranche.
The final dataset consists of 86.832 Dutch newspaper articles related to sex work from January 1995 until December 2018. Looking at the availability of newspaper articles, most Dutch newspaper outlets become available on LexisNexis between 1992 and 1999 (see appendix, table 2). By including the earliest date in which most Dutch newspaper outlets are available on LexisNexis until the end of 2018, I incorporated the biggest time frame possible for this thesis. Further, all the Dutch newspaper outlets that are available on LexisNexis are included in the dataset. This means that, alongside bigger newspapers such as Trouw, smaller newspapers such as Eindhovens Dagblad and Dagblad van het Noorden are included.
In order to study the construction of sex work, I use automated content analysis (ACA). One of the first definitions of ACA was given by Berelson, who described content analysis as a research technique for the “objective, systematic and quantitative description of the manifest content of
communication” (Berelson as quoted by Macnamara 2002: 2). There are several techniques available, but I will use Latent Dirichlet Allocation (LDA). LDA is part of the larger field of probabilistic topic modelling which is a form of unsupervised machine learning. It is a statistical method that
(inductively) captures themes that run through texts (Blei 2012: 77). LDA assumes that each document does not solely consist of one topic but a number of topics, which are distributed over words. Topics are defined as “a distribution over a fixed vocabulary” (Blei 2002: 78).
Going through documents, LDA searches for the co-occurrence of words in order to estimate topics (Grimmer and Stewart 2013: 17). For example, a document about women’s sports will include words such as ‘women’ and ‘soccer’. A document about public transport is less likely to include these words. A topic model operationalizes this logic into an algorithm; figure 1 illustrates this assumption.
Figure 1. Illustrates underlying idea of LDA. Each document is a mixture of (hidden) topics and these topics consists of words.
In this study, discourse is operationalized as topics from the topic model. Jacobs and Tschötschel argue and illustrate in their study how topic modelling are an excellent way of studying discourses. Topic model can help us discover hidden themes and study how they are continuously repeated (2019: 9). Topic modelling are also a very useful tool to discover new content in a corpus (Chang et al. 2009: 1). For my first research question I want to assess which discourses are present in Dutch newspaper articles related to sex work. Following the literature on sex work, I formulated expectations. Even though I expect to find four anti-sex work discourses and one pro-sex work discourse, I also want to assess whether there are more discourses present that are not discussed within the literature on sex work. Thus, an inductive approach is implemented in this thesis instead of searching for pre-defined words.
In order to analyze the large corpus, I use R, which is an open source programming environment, designed for statistical analysis (Welbers et al. 2017: 245). The first step consists of importing the newspaper articles into R. Second, the document-level variables of each article need to be separated. The document-level variables (or metadata) I need for further analysis are the publishing dates of the articles and the newspaper outlets. Third, there are preparatory steps that needs to be taken before we can analyze the text. Commonly, this is referred to as “preprocessing” (Welbers et al. 2017: 250). The preprocessing techniques I implemented are; (1) removing numbers, (2) removing
Documents
(newspaper
articles)
Topics
punctuations and (3) removing stopwords from the corpus.
After preprocessing the text, I created a document term matrix (DTM) which is a format for representing a text corpus (Welbers et al. 2017: 252). I use the quanteda package in order to create the DTM. Following Welbers et al., I only keep the top five percent of the most frequent words that appear in less than ten percent of all documents. Within LDA, there is an important parameter that has to be defined before running a topic model. The number of topics, k, has to be specified upfront. The first step is using the “harmonic means”-measure of the model fit, in which the ideal number of topics can be calculated (Traber et al. 2018: 6). After calculating the harmonic mean, the ideal number of topics for this corpus is 37.
The results of the topic model are presented in table 1. The results show us all discourses that are discussed in newspaper articles related to sex work. All topics are listed with the top seven words that are assigned to the topic. I labelled each topic by creating a name that captures the underlying theme of the words that belong to that topic. Because the labelling of the topics is done manually and thus subjective, other scholars might assign different names to the topics. The topics are also assigned to a category in order to organize the results. Additionally, I calculated the per-topic-per-word
probability called ‘beta’ and the results are presented in the appendix. Thereafter, in order to study how the construction of sex work changes over time, I calculated the document-topic probabilities. Within a topic model, each document consists of several topics. The document-topic probability is the estimation of observing a topic in a newspaper article. After calculating the document-topic
probability, I plotted a time analysis of all topics, presented in the appendix. In the results section, I choose to present and discuss the time analysis of six discourses that I deem important.
Results
The 37 discourses discussed in Dutch newspaper articles related to sex work are presented in table 1. The results show us several anti-sex work discourses that are discussed in the theoretical framework: crime, human trafficking, loverboys and aids. In addition, we also see topics addressing the economy, politics, art, countries, history and soccer. There are also topics that consist of general words such as ‘think of2’, ‘think’, or ‘be’ in topic 16. Topics like these are difficult to interpret because they are general. In the following section, I will first interpret the discourses that are discussed and identified in the theoretical framework. Second, I will discuss several ‘new’ discourses that are discovered as a result from the topic model.
2In the results section I will discuss the words translated into English. In table 2 the topics are presented with the original
Table 2. Overview of all topics.
Category Topic Discourse name Top seven words
Geography 1 City arnhem, nijmegen, zwolle, deventer, enschede, dagblad, almelo
2 China turkse, china, chinese, moslims, islam, islamitische, marokkaanse
10 Eindhoven eindhoven, noord, zuid, leeuwarden, west, helmond, friesland
11 Japan japanse, oorlog, japan, leger, militairen, angels, soldaten
12 Europe europese, europa, duitse, duitsland, landen, politieke, politiek
13 America new, amerikaanse, york, and, britse, amerika, to
15 Streetwalking zone tippelzone, utrecht, rotterdam, keileweg, gemeenten, zandpad, arnhem
18 Red light district bewoners, wallen, wijk, overlast, panden, ramen, wonen
20 Tourism hotel, langs, toeristen, plek, water, meter, zee
21 Countries landen, india, afrika, thailand, internationale, organisatie, afrikaanse
24 Spanish spaanse, zwarte, suriname, zuid-afrika, braziliaanse, spanje, eiland
25 Travel auto, eten, langs, rijden, loopt, nacht, half
36 City maastricht, belgische, heerlen, antwerpen, edition, limburg, brussel
Anti-sex
work 4 Human trafficking mensenhandel, slachtoffers, uitbuiting, wet, strafbaar, misstanden, aangfite 6 Crime criminele, drugs, criminelen, criminaliteit, misdaad, georganiseerde, crimineel
7 Loverboys jongeren, loverboys, jongens, school, ouders, opvang, hulp
28 Aids russische, aids, rusland, hiv, moskou, trump, ziektes
30 Refugees asielzoekers, illegale, illegalen, vrouwenhandel, vluchtelingen, illegaal, mensenhandel 37 Suspect verdachte, cel, moord, officier, veroordeeld, rechter, advocaat
Art and
media 14 Films Films, film, regisseur, regie, speelt, vs, acteur 19 Writer roman, schrijver, boeken, schrijver, literatuur, liefde, verhalen
32 Art kunst, museum, tentoonstelling, foto’s, kunstenaar, schilderijen, kunstenaars
34 Theatre muziek, theater, voorstelling, stuk, toneel, publiek, speelt
35 Television programma, rtl, serie, tv, televisie, documentaire, aflevering
Other 3 Weekday zaterdag, utrecht, jan, rotterdam, zondag, vrijdag, groningen
5 History eeuw, geschiedenis, schreef, oorlog, koningin, koning, joodse
8 Religion kerk, god, geloof, christelijk, jezus, kerken, bijbel
9 License vergunning, club, Breda, pand, eigenaar, raad, rechter
16 General vind, denk, nee, ga, nou, bent, dingen
17 President berlusconi, italiaanse, franse, president, premier, frankrijk, krant
22 Economy bedrijf, bedrijven, economie, overheid, betalen, sociale, bank
23 Politics partij, pvda, politiek, cda, vvd, minister, kamer
26 Soccer Voetbal, wedstrijd, sport, club, spelers, ajax, wk
27 Family dochter, zag, wist, ouders, vriend, kind, zoon
29 Sexuality seksuele, internet, seksueel, liefde, porno, seksualiteit, vrouwelijke
31 Weekday Zo, wo, di, ma, path, vr, do
A-priori identified discourses
The ‘sex work as affiliated with human trafficking’ discourse and ‘sex workers as victims’ discourse can be traced back in topic 4, labelled human trafficking. Looking at the words in this topic, we see words represented such as ‘victims’, ‘exploitation’, ‘punishable’ and ‘human trafficking’. Another topic in which we can trace back the ‘sex work as affiliated with human trafficking’ is topic 30, labelled refugees. One of the words in this topic is ‘human trafficking’ and it co-occurs with words such as ‘refugees’, ‘illegals’ and ‘trafficking of women’. As discussed in the theoretical framework, the victim identity is ascribed to sex workers whenever sex work is conflated with human trafficking. Hence, in the results of the topic model, the ‘sex workers as victims’ discourse co-occurs with the ‘sex work as affiliated with human trafficking’ discourse.
The third anti-sex work discourse that can be traced back from the topic model is the ‘sex work as affiliated with crime’ discourse. There are three topics, labelled crime, loverboy and suspect, in which the ‘sex work as affiliated with crime’ discourse can be uncovered. Topic 6, labelled as crime, consists of words such as ‘criminal’, ‘criminality’ and ‘crime’. Topic 7, labelled loverboys, consists of words such as ‘young men’, ‘loverboys’ and ‘shelter’. Topic 37, labelled suspect, consists of words such as ‘suspect’, ‘judge’ and ‘court’. Conflating sex work with these issues put sex work in a context of crime and criminality.
Finally, the last anti-sex work discourse that is uncovered from the topic model, is the ‘sex work as affiliated with health hazards’ discourse. Topic 28, labelled aids, consists of words such as ‘aids’, ‘HIV’, and ‘diseases’. The words also co-occur with words referring to foreign countries. After manually looking through the corpus, a lot of articles discuss sex work and health hazards within the international context. For example, Trouw3 published an article titled ‘prosititutie is opslagplaats van het aidsvirus’. The article discusses the prostitution sector in India and how young girls are forced into sex work. According to this article, more than 64 percent of the women working in the prostitution sector have HIV.
When analyzing the topics, we can see that there are some words that re-occur in some of the topics. If the same word is situated within different topics, there are three possible explanations; (1) this means that the word has different meanings; (2) the word is part of the hegemonic discourse if the meaning stays constant across different topics and (3) when a concept re-occurs in several topics, there is most likely a debate in the corpus over how to signify the term (Jacobs and Tschötschel 2019: 7-9). For example, topic 4 (labelled human trafficking) and topic 30 (labelled refugees) both address human trafficking but there are different words ascribed to these topics. Topic 30 consists of words such as ‘illegals’, ‘refugees’, ‘trafficking women’ whilst topic 4 is composed of words such as ‘exploitation’, ‘punishable’ and ‘wrongdoing’. Topic 4 puts human trafficking in a context of wrongdoing and exploitation whilst in topic 30, human trafficking is placed in a context of illegals
and refugees. Consequently, this indicates that human trafficking is a re-occurring discourse in newspaper articles related to sex work.
Overall, four of the five identified anti-sex work discourses are traced back in the results of the topic model. None of the topics consist of words such as decriminalization, work, human rights, rights or self-determination. Hence, the pro-sex work discourse is not traced back in the results of the topic model. This does not indicate that none of the newspaper articles discusses sex work as a recognized form of labor. For example, Utrechts Niewsblad4 published an article titled: “GroenLinks: betere positie sekswerkers” in which the Dutch political party GroenLinks asked for a better working environment in Utrecht. At that time, many sex workers were working in the illegal sector as the municipality of Utrecht did not give sex workers the opportunity to legally work in their city. Besides this article, there are more newspaper articles discussing the self-determination and rights of sex workers. However, the absence of pro-sex work discourses in the topic model implies that articles implementing a pro-sex work discourse are greatly outnumbered by articles implementing other (anti-sex work) discourses.
Discovered discourses
Besides the anti-sex work discourses, the topic model discovered discourses that are not formulated in the expectations of this study. As already mentioned, one of the benefits of topic modelling is
uncovering new content in a corpus.
A large proportion of the discourses discovered by the topic model, can be categorized as geography. Most of these discourses refer to a city or country. For instance, topic 1 is labelled city and all the words consists of cities in the Netherlands. Since the brothel ban is lifted, municipalities decide how many sex workers can start a company in their city and set the minimum age for sex work. After manually reading through newspaper articles in the corpus, many articles write about the implementation of the amendment, the consequences of lifting the brothel ban and the
successes/downfalls of the new amendment. Another interesting discourse is topic 18, labelled Red Light district. This points out that many Dutch newspaper articles related to sex work write about the Red Light District in Amsterdam. Looking at the words in this topic, we see ‘nuisance’, ‘inhabitants’ and ‘living’. The co-occurrence of the word ‘inhabitants’ and ‘nuisance’ could indicate that the area is a nuisance for the inhabitants.
There are also several discourses referring to foreign countries or countries in general, like topic 13 labelled America and topic 21 labelled countries. An intriguing discourse is topic 11, labelled Japan. The topic also consists of words such as ‘war’, ‘army’ and ‘soldiers’. After manually reading
through the corpus, I discovered that there are many Dutch newspaper articles writing about Japanese comfort women.
Besides many discourses regarding geography, another category is art and media. Surprisingly, there are five topics labelled films, writer, art, theatre and television discovered as a result of the topic model. Going through the corpus and reading some of the articles, there are many newspaper articles discussing a movie, art piece or theatre performance that is related to sex work. For instance, an article in Algemeen Dablad5 titled ‘tales of ordinary madness’ discusses an Italian movie
in which an actress plays a prostitute. Besides that, the article does not discuss sex work or give an opinion about sex workers. Hence, sex workers are also a subject within the art world.
Finally, another interesting notion is the word woman and the words it is listed with. For instance, ‘trafficking women’ is listed with words such as ‘human trafficking’, ‘illegals’ and ‘refugees’. Or in the sexuality discourse (topic 29), the word ‘woman’ occurs together with words such as ‘porn’, ‘sexuality’ and the ‘internet’. Whilst on the other hand, we see the word ‘boy’ being listed with words such as ‘loverboy’, which is discussed in the theoretical framework of this study. This is interesting because it affirms that within discourses surrounding sex work, women are portrayed as being the provider of sexual services and/or victims whilst men are viewed as being the buyer of sex work and/or exploiter of sex workers. Hence, the construction of sex work in the corpus of this study is gendered.
Time analysis
The results of the topic model illustrated six anti-sex work discourses. Figure 2 shows the document-topic probabilities of six anti-sex work discourses from 1995 until 2018. A time analysis of the document-topic probabilities of all 37 discourses are presented in the appendix.
The document-topic probability of the human trafficking discourse illustrates how this discourse is slowly, but surely, increasing throughout the years. The probability of the human
trafficking discourse rapidly increases from 2005 until 2009. In 2015, this discourse is at its peak. We see the opposite trend happening with the probability of the refugees discourses, which also puts sex work in a context of human trafficking. Looking at the development of the refugees discourse, this discourse peaked in 1996 but is overall decreasing since then. Drawing from the theoretical
framework of this thesis, in 2009 the Dutch parliament introduced a bill in which human trafficking in the sex industry should be combatted. This might explain why the probability of the human trafficking discourse peaks around 2009. However, a more detailed analysis of the newspaper articles is needed in order to support this conclusion.
The crime, loverboys and suspect discourses represent the ‘sex work as affiliated with crime’
discourse. We see the probability of the crime discourse peak in 1996. Interestingly, throughout the years, the probability of this discourse is dropping. The crime discourse shortly peaks around 2008 but drops again after 2009. In 2007, sex work was framed as a breeding place for crime in the new cabinet pact. Thus, this peak might be related to the conflation of sex work and crime in the new cabinet pact. In the third plot the probability of the loverboys discourse is presented. The probability of the loverboys discourse is most prominent in 2006 and 2009. After 2009, the probability of the loverboys discourse decreases and remains constant from 2013 until 2018. In the literature on sex work it is discussed that there was moral dismay in the Netherlands in 2005 about ‘loverboys’. It is likely that this public moral dismay explains the peak in 2006.
The last discourse representing the ‘sex work as affiliated with crime’ discourse is the suspect discourse. This discourse is most prominent around 2001, right after the 2000 amendment. As
discussed in the theoretical framework, one of the goals of the amendment was getting rid of criminal elements in the Dutch sex industry. This might explain why right after lifting the brothel ban, the suspect discourse is at its highest peak. After 2001, the probability of this discourse is slowly
decreasing. However, the probability of the suspect discourse is in 2018 still more prominent than the other anti-sex work discourses6.
Finally, the ‘sex work as affiliated with health hazards’ discourse is represented by the aids discourse. The probability of the aids discourse is most prominent in 1995. After 1995, the aids discourse is steadily decreasing with the sole exception of 2002 in which the discourse temporarily peaks again. We see a slight upward trend in the probability of the aids discourse from 2016 until 2018. In the literature on sex work, Vanwesenbeeck argued that in the 1980s and 1990s most studies on sex work affiliated sex work with health hazards. This could be a possible explanation of why the aids discourse is more prominent in 1995. It is important to point out that the aids discourse has, even at its peak, the lowest probability of all the anti-sex work discourses.
In conclusion, in the corpus of newspaper articles related to sex work, the suspect discourse has the highest probability and the aids discourse is the least prominent. The human trafficking discourse, however, is the only discourse that exhibits an upward trend. This could indicate that sex work is becoming increasingly affiliated with human trafficking, while the conflation with crime is slowly decreasing. In both these discourses, sex workers are perceived as victims, meaning that the ‘sex workers as victims’ is also prominent.
6Looking at the appendix (figure 2), compared to all the discourses in the topic model, the suspect discourse has the highest
Figure 2. This graph illustrates the document-topic probability of six anti-sex work discourses from 1995 until 2018.
Conclusion and Discussion
Although sex work is acknowledged as a legal profession in the Netherlands, Dutch sex workers face various challenges due to the service they offer. This thesis studies the construction of sex work by the Dutch media from 1995 until 2018. Based on the literature on sex work, there are five anti-sex work discourses and one pro-sex work discourse identified in the theoretical framework. In order to study the construction of sex work by the Dutch media, Dutch newspaper articles are analyzed using topic modelling. Topic modelling is a useful tool to (inductively) find discourses within a corpus.
Even though there are expectations formulated in this thesis, an inductive method is implemented in order to assess if there are more discourses present in the newspaper articles related to sex work. The results show 37 discourses that are discovered in newspaper articles related to sex work. Drawing from the results, there are six topics that can be categorized as anti-sex worker discourses. The results are partly in line with the expectations formulated in the theoretical framework as the pro-sex work discourse is absent from the topic model. However, all four anti-pro-sex work discourses are, as expected, traced back in the results of the topic model. The ‘sex workers as victims’ discourse co-occurs with the ‘sex work as affiliated with crime’ and ‘sex work as affiliated with human trafficking’ discourses. The ‘sex work as affiliated with health hazards’ is represented in one topic, the aids discourse. The time analysis of the probability of the anti-sex work discourses illustrate that the suspect discourse is most prominent whilst the aids discourse has the lowest probability. However, the probability of the human trafficking discourse is the only discourse that exhibits an upward trend In the theoretical framework it is argued that discourses give meaning to the practice of offering sexual services. Hence, it is not the act itself (e.g. offering a sexual service), but the
discourses surrounding it. Looking at the results of this study, six discourses surrounding sex work in Dutch newspaper articles put sex work in a context of crime, human trafficking, health hazards and victimization. There are also other discourses, such as politics, art and history that are not necessarily anti-sex work. However, the absence of a pro-sex work discourse indicates that articles implementing a pro-sex work discourse are greatly outnumbered. Sex workers are vocal about their rights, as illustrated in the theoretical framework. Nevertheless, if the discourses surrounding sex work does not include their voices, it contributes to the interpretation of sex workers as victims. Looking at the discrimination sex workers face, is important to include their experiences.
This thesis illustrates that topic modelling is a useful tool in order to uncover new content in a corpus and inductively find discourses that run through text. For instance, in newspaper articles related to sex work, this thesis discovered that sex work is also a subject within the art world. In addition, discourses surrounding foreign countries and the Red Light district in Amsterdam are also present in the Dutch media. However, it is important to point out that the results of topic modelling should not be over-interpreted and that a close reading of the newspaper articles is necessary to interpret the results.
Drawing from the results of this study, I also formulate some recommendation for future research. First, it is interesting that within discourses that affiliate sex work with crime and human trafficking, an anti-immigration narrative can be traced. The literature on sex work merely touches upon the fact that non-native Dutch men were pointed out as ‘loverboys’, or how non-EU citizens were not allowed to work in the sex industry. Currently, we have a minimal understanding of how anti-sex work discourses are used in order to push other narratives. For instance, it is important to understand how the sex industry is being used in order to push an anti-immigration/xenophobic agenda in the Netherlands.
Second, the time analysis of this study illustrates how the probability of the human trafficking discourse in Dutch newspaper articles is steadily increasing over time. In the introduction of this thesis it is mentioned that forty thousand people in the Netherlands signed a petition favoring the criminalization of the purchase of sex. The motives behind the movement is ‘saving’ women in the sex industry and combatting human trafficking. Pro-sex workers argue that sex workers have agency and are demanding the same labor rights as other professions. We need more research on human trafficking in the Netherlands in order to disentangle the two. The conflation of sex work and human trafficking is harmful for sex workers who are voluntarily working in the sex industry.
Finally, it is important to point out that this thesis did not give account to differences within the sex workers community. Also referred to as the ‘whorearchy’, the sex workers community also has a system of hierarchy between sex workers. For instance, sex workers who provide full body services or work on the streets are treated differently within their own community. For future research it is important to disentangle the various types of services offered within the sex industry.
Appendix
Table 1. Search string in LexisNexis.
Word Date until present N
Sekswerker 01/01/1996 937 Seks werk 01/01/1991 609 Prostitutie 01/01/1980 58.693 Prostituee 01/01/1980 33.571 Prostituees 01/01/1980 40.883 Prostitué 01/01/1980 1.367 Prostitueés 01/01/1980 40.848 Gigolo 01/01/1980 3.949 Sekswerk 01/01/1991 549 Bordeel 01/01/1980 18.199 Hoer 01/01/1980 20.371 Pooier 01/01/1990 7.931 Bordeelverbod 01/01/1991 2.874 Seksindustrie 01/01/1990 4.981 Prostitutiesector 01/01/1995 294 Prostitutiebeleid 01/01/1990 1.809 Seksbedrijf 01/01/1992 678 Seksbranche 01/03/1993 1.455
Table 2. Overview of newspaper outlets and date from when outlet is available on LexisNexis.
Newspaper outlet N Year
De Telegraaf 4.186 1999 Trouw 6.349 1992 NRC Handelsblad 7.319 1990 Volkskrant 6.621 1995 Algemeen Dagblad 3.307 1991 De Gelderlander 5.489 1995 Brabants Dagblad 4.098 1994 Leeuwarder Courant 5.038 1999 Eindhovens Dagblad 3.270 1994 Dagblad de Limburger 2.287 2007 Dagblad van het Noorden 7.059 1999 Algemeen Nederlands Persbureau (ANP) 8.094 1993
BN/De Stem 3.778 1994 Het Parool 6.619 - AD/Amersfoortse Courant 2.298 - AD/Utrechts Nieuwsblad 3.039 - Noordhollands Dagblad 2.883 - De Twentsche Courant 2.728 - AD/Haagse Courant 2.667 - AD/Rotterdams Dagblad 2.548 -
Figure 1. The per-topic-per-word probability (β).
References
1. Bacci, C. (2005). “Discourse, discourse everywhere; Subject “agency” in feminist discourse methodology”, Nordic Journal of women’s studies, 13(03): 198-209.
2. Berg, H. (2014). “Working for Love, Loving for Work: Discourses of Labor in Feminist Sex-Work Activism”, Feminist Studies, 40(3): 693-721.
3. Blei, D. (2012). “Probabilistic Topic Models”, Communications of the ACM, 55(4): 77-84. 4. Boumans, J. and Trilling, D. (2016). “Taking Stock of the Toolkit”, Digital Journalism, 4(1):
8-23.
5. Chang, J., Gerrish, S., Wang, C., Boyd-Graber, J.L. and Blei, D.M. (2009). “Reading tea leaves: How humans interpret topic models”, In Advances in neural information processing systems (pp. 288-296).
6. Cunningham, S. (2016). “Reinforcing or Challenging Stigma? The Risks and Benefits of ‘Dignity
Talk’in Sex Work Discourse”, International Journal for the Semiotics of Law-Revue internationale de Sémiotique juridique, 29(1): 45-65.
7. Daalder, A.L. (2007). Prostitution in the Netherlands since the lifting of the brothel ban. Boom Juridische uitgevers.
8. Dziuban, A. and Stevenson, L. (2015). “Nothing about us without us!: ten years of sex workers' rights activism and advocacy in Europe”.
9. Foucault, M. (1981) Orders of discourse. In: Young, R. (ed.). Untying the text: A post-structuralist reader. Boston: Routledge & Kegan.
10. Grimmer, J. and Stewart, B.M. (2013). “Text as Data: The Promise and Pitfalls of Automatic Content Analysis Methods for Political Texts”, Political Analysis, 21(3): 1-31.
11. Hahn, K. and Holzscheiter, A. (2013). “The Ambivalence of Advocacy: Representation and Contestation in Global NGO Advocacy for Child Workers and Sex Workers”, Global Society, 27(4): 497-520.
12. Hunt, S. and Hubbard, B. (2015). “The representation of sex workers in South African media: Danger, morals and human rights”, English Language and Linguistics, 46: 25-49.
13. Jacobs, T. and Tschötschel, R. (2019). “Topic models meet discourse analysis: a quantitative tool for a qualitative approach”, International Journal of Social Research Methodology, 1-17.
14. Levy, J. and Jakobsson, P. (2014). “Sweden’s abolitionist discourse and law: Effects on the dynamics of Swedish sex work and the lives of Sweden’s sex workers”, Criminology & Criminal Justice, 14(5): 593-607.
15. McLaughlin, L. (1991). “Discourses of prostitution/discourses of sexuality”, Critical Studies in Mass Communication, 8(3): 249-272.
16. McClintock, A. (1992). “Sex Workers and Sex Work: Introduction”, Duke University Press, 37: 1-10.
17. Macnamara, J. (2002). “Media content analysis: Its uses, benefits and best practice methodology”, Asia Specific Public Relations Journal, 6(1): 1-4.
18. Miller, S. (1990). “Foucault on Discourse and Power”, Theoria: A Journal of Social and Political Theory, 76: 115-125.
19. NOS (2019). Jongeren verzamelen 40.000 handtekeningen voor verbod op betaalde seks. Online:
https://nos.nl/artikel/2279641-jongeren-verzamelen-40-000-handtekeningen-voor-verbod-op-betaalde-seks.html
20. Ondernemersplein KVK (2019). Werken als zelfstandig sekswerker in prostitutie. Online:
https://ondernemersplein.kvk.nl/werken-als-zelfstandig-sekswerker-in-de-prostitutie/
21. Overall, C. (1992). “What’s Wrong with Prostitution? Evaluating Sex Work”, Signs: Journal of Women in Culture and Society, 17(4):705-724.
22. Outshoorn, J. (2012). “Policy Change in Prostitution in the Netherlands: from Legalization to Strict Control, Sexuality Research and Social Policy, 9(3): 233-243.
23. Parool (2019). Petitie: betaalde seks moet straat worden. Online:
https://www.parool.nl/amsterdam/petitie-betaalde-seks-moet-strafbaar-worden~b1ccc788/
24. Pheterson, G. (1990). “The category “prostitute” in scientific inquiry”, Journal of Sex Research, 27(3): 397-407.
25. Post, C., Brouwer, J. and Vols, M. (2018). “Regulation of Prostitution in the Netherlands: Liberal Dream or Growing Repression?”, European Journal on Criminal Policy and Research, 1-20. 26. Proud and Aidsfonds- Soa Aids Nederland (2018). Sekswerk en geweld in Nederland. Online:
https://www.soaaids.nl/sites/default/files/media/Prostitutie/2018-rapport-sekswerk-geweld.pdf
27. Scoular, J. (2004) “The ‘subject’ of prostitution: Interpreting the discursive, symbolic and material position of sex/work in feminist theory”, Feminist Theory, 5(3): 343-355.
28. Torfing, J. (2005). Discourse theory: Achievements, arguments, and challenges. In Discourse theory in European politics (pp. 1-32). Palgrave Macmillan, London.
29. Traber, D., Schoonvelde, M., Schumacher, G., Dahiya, T. and de Vries, E. (2016). Issue Avoidance and Blame Attribution in Leader Speeches during the Economic Crisis in Europe. In 6th Annual Conference of the European Political Science Association, Brussels, 23-25.
30. Vanwesenbeeck, I., (2001). “Another decade of social scientific work on sex work: a review of research 1990–2000”, Annual review of sex research, 12(1): 242-289.
31. Welbers, K., Van Atteveldt, W. and Benoit, K. (2017). “Text analysis in R”, Communication
R-Scripts
First script
##### Author: M. Zamanbin ##### Email: mehri.z91@gmail.com
##### Project: Thesis, studying the construction of sex work in the Dutch media ##### Purpose: Loading data into R + Preprocessing newspaper articles in text ##### Last updated: May 2019
##### R version 1.1.383 (10-04-2019)
require(readtext) library(stringr) library(quanteda)
## Working directory ##
setwd("/Users/mehri/Library/Mobile Documents/com~apple~CloudDocs/Research Master/Thesis/R/All data")
##### Function to reshape data into one line per artICLE, input is data as delivered by LexisNexis in one, big text document
lexisnexisarticles <- function(data) {
data <- as.data.frame(data[data[1]!='',]) # delete empty rows
data$firstline <- str_count(data[,1],"(?i)\\d{1,3} of \\d{1,3} documents") data$noarticle <- str_count(data[,1],"(?i)Download Request: Selected Items") data$beginobs <- data$firstline + data$noarticle
data$fillId <- cumsum(as.numeric(data$beginobs)) data$obsnum <- sequence(rle(data$fillId)$lengths)
obsnum = sequence(rle(data$fillId)$lengths) # Number the observations within one article
datawide <- reshape(data, idvar = "fillId", timevar = "obsnum", direction = "wide") # Reshape to wide format, one observation per article
datawide <- datawide[(str_count(datawide[,2],"(?i)documents"))>0,] # flter out the observations that are not articles, but the search request
# Drop obsolete variables
datawide$fillId <- NULL
for (iin1:(1+(length(colnames(datawide))-1)/4)) {
datawide[,paste("beginobs",i,sep=".")] <- NULL
datawide[,paste("firstline",i,sep=".")] <- NULL
datawide[,paste("noarticle",i,sep=".")] <- NULL
datawide[,paste("articlecount",i,sep=".")] <- NULL
}
# Combine all body of article variables
body <- ''
for (iin5:length(colnames(datawide))) { body <- paste(body, datawide[,i]) body <- str_remove_all(body,"NA NA ") }
datawide[,5:(length(colnames(datawide)))] <- NULL
dataclean <- data.frame(cbind(datawide,body)) # body als column achter de titel plakken
rm(datawide) dataclean
}
# Loading data into R #
files <- list.files(getwd(),pattern = "\\.TXT$")
newspapercorpus <-
data.frame(document=character(),outlet=character(),date=character(),header=character(),body=character(),strin gsAsFactors = F)
for (fileinfiles) { print(file)
data <- readLines(file) data <- as.data.frame(data)
dataclean <- lexisnexisarticles(data)
names(dataclean) <- c("document","outlet","date","header","body") newspapercorpus <- rbind(newspapercorpus,dataclean)
}
save(newspapercorpus, file = "firstsave.rda") View(newspapercorpus)
## Combining header and body, removing some columns ##
load("firstsave.rda")
newspapercorpus$text <- paste(newspapercorpus$header, newspapercorpus$body)
newspapercorpus$document <- NULL
newspapercorpus$header <- NULL
newspapercorpus$body <- NULL
#save(newspapercorpus, file = "secondsave.rda")
newspapercorpus$text <- str_remove(newspapercorpus$text, "LOAD-DATE: ((\\w*\\s*)|(\\W*\\s*))* ((\\w*\\s*)|(\\W*\\s*))*")
newspapercorpus$text <- str_remove(newspapercorpus$text, "BYLINE: ((\\w*\\s*)|(\\W*\\s*))*")
newspapercorpus$text <- str_remove(newspapercorpus$text, "SECTION: ((\\w*\\s*)|(\\W*\\s*))*|((\\w*\\d+))")
newspapercorpus$text <- str_remove(newspapercorpus$text, "LENGTH: \\d+ (words)")
newspapercorpus$text <- str_remove(newspapercorpus$text, ":<U+00A0>")
newspapercorpus$text <- str_remove(newspapercorpus$text, "<c2>")
newspapercorpus$text <- str_remove(newspapercorpus$text, "<c0>") newspapercorpus$text <- str_remove(newspapercorpus$text, "Pg.") newspapercorpus$text <- str_remove(newspapercorpus$text, "ISBN") newspapercorpus$text <- str_remove(newspapercorpus$text, "<a0>") newspapercorpus$text <- str_remove(newspapercorpus$text, "<a8>") newspapercorpus$text <- str_remove(newspapercorpus$text,"DATELINE") newspapercorpus$text <- str_remove(newspapercorpus$text, "Blz.") newspapercorpus$text <- str_remove(newspapercorpus$text, "LENGTH") newspapercorpus$text <- str_remove(newspapercorpus$text, " HIGHLIGHT") newspapercorpus$text <- str_remove(newspapercorpus$text, "<c3><a9><c3><a9>n") newspapercorpus$text <- str_remove(newspapercorpus$text, "<e2><80><9e>") newspapercorpus$text <- str_remove(newspapercorpus$text, "HCST04") print(newspapercorpus[45679,3])
save(newspapercorpus, file = "fourthsave.rda")
Second script
##### Author: M. Zamanbin ##### Email: mehri.z91@gmail.com
