Problem Representations of Femicide/Feminicide Legislation in Latin America

Femicide/feminicide has become an increasing social concern for local communities, international organizations, and national governments. In 2007, Latin American countries began enacting legislation to prevent and punish femicide/feminicide; however, relatively few researchers have assessed the scope and depth of this legislation. Using Carol Bacchi’s (2009) “what’s the problem represented to be” approach, this study analyzes femicide/feminicide across Latin American countries. The goal of this approach is to assess concepts that are taken for granted within policies and uncover what has been silenced through problem representations. Results provide considerations for future legislative development in Latin America and abroad.


Introduction
Femicide/feminicide, the killing of women because they are women, has received increased attention as a global problem requiring state intervention. Beginning in Latin America in 2007, governments have enacted femicide/feminicide legislation, classifying femicide/feminicide as a distinct crime and/or stipulating specific penalties. More than 16 countries in the region have now enacted femicide/feminicide legislation (Sarmiento et al. 2014). However, since enactment, various groups have critiqued the depth and strength of legislation as well as its capacity to prevent and punish femicide/feminicide (Luffy et al. 2015;Mujica and Tuesta 2014;Sopcich 2015). This criticism is not unique to Latin America, as much is still unknown about femicide/feminicide responses globally. However, because Latin America was the first region to undertake femicide/feminicide legislation, analyzing the countries of this region can provide insight for future legislation development. Drawing on Bacchi's (2009) "what's the problem represented to be" (WPR) approach (see Table 1), this study examines how femicide/feminicide is constructed in Latin American legislative responses.
This study aims to contribute to the body of femicide/feminicide literature by assessing how femicide/feminicide legislation has been characterized as a problem across the region. 1 First, we provide some context for femicide/feminicide in Latin America, including relevant social and cultural issues and the development of gender-specific legislation. Next, we outline the WPR approach, its methods and uses, and its particular relevance to gender-based violence legislation. This is followed by the findings of such an analysis, which are organized according to Bacchi's (2009) questions. Determining how femicide/feminicide has been represented and interpreted through national legislation will improve conceptualizations for future legislative development.

Defining Femicide/Feminicide
Femicide is broadly defined as the killing of women because they are women. However, questions remain about what this killing looks like in practice. A major challenge in measuring femicide rates is distinguishing femicides from female homicides. Factors that distinguish the killing of women from the killing of men suggest that the killing of women reinforces societal notions of what it means to be a woman. Societal concepts of womanhood include subordination, weakness, and femininity. Gender-related motivations are often difficult to ascertain, increasing the difficulty of classifying killings as femicide (Sarmiento et al. 2014).
Feminicide, a term adopted by Latin American feminists, describes woman-killing rooted in gendered power structures and produced by patriarchal and social organizations of gender (Fregoso and Bejarano 2010;Lagarde y de los Ríos 2006). Feminicidal violence is fueled by misogynistic actions that deny women physical and mental security (Fregoso and Bejarano 2010). Feminicide is overt violence, but it can also be the product of discriminatory practices including female genital mutilation or unsafe abortion (Toledo 2017). The term is also used to describe femicide-rooted impunity, functioning as a "tool of patriarchal oppression, while also serving as a tool of racism, economic oppression, and colonialism" (Smith 2006: 417). "Feminicide" includes political connotations, drawing attention to the failure of States' to prevent and punish femicide/feminicide, thereby neglecting their international obligations (Sarmiento et al. 2014). Feminicide, as opposed to femicide, is useful for highlighting the causes of femicide and the contexts in which it occurs. However, this broad, conceptual definition is difficult to apply in practice.
Femicide/feminicide is broadly divided into public and private spheres (Sarmiento et al. 2014). Private sphere femicide/feminicide occurs between intimate partners and family members (Meneghel, Ceccon, et al. 2017). Public sphere femicide/feminicide includes killings by strangers, acquaintances, gang members, and public servants (e.g., police). Furthermore, motivations are not unique to each sphere, because femicide/feminicide in either sphere can be motivated by IJCJ&SD 3 www.crimejusticejournal.com 2020 9(2) revenge, hatred, jealousy, or sexual violence. Regardless of sphere, femicides/feminicides are caused by gendered motivations, which are rooted in the patriarchal notions of women as property or objects of men (Sarmiento et al. 2014).

Femicide/Feminicide in the Latin American Context
Femicide/feminicide is a global problem. However, the rates of its occurrence in Latin America are particularly high, surpassed only by South Africa (United Nations Office on Drugs and Crime 2019). For example, in Honduras, El Salvador, Guatemala, and Colombia, femicide/feminicide rates range from 6 to 14 per 100,000 population and are 8 to 24 times higher than rates in Canada (Alvazzi del Frate 2011; Racovita 2015). 2 Social and political characteristics shape the problem at regional and national levels, accounting for the varying femicide/feminicide rates globally (Castañeda Salgado 2016). In Latin America, the culture of machismo is one characteristic believed to increase femicide/feminicide rates in the region (Prieto-Carrón et al. 2007;Wilson 2014). Machismo is a social construct that describes traditional patriarchal beliefs about the role of women (Kimelblatt 2016). Broadly speaking, machismo is any belief or action that fosters gender inequality and restricts women's autonomy (Luffy et al. 2015). These values demand women's subservience and legitimize the use of violence against them (Wilson 2014). The widespread inequality may provide a partial explanation for why Latin American countries have some of the highest femicide/feminicide rates in the world.
An important predictor setting Latin America apart from many Western countries is states' historic and continued unwillingness to respond. 3 Although Latin American governments have been at the forefront of the development of femicide/feminicide legislation, this legislation has not always translated into practice. State inaction in investigating and prosecuting femicide/feminicide is a condition that fosters violence against women (VAW; Fregoso and Bejarano 2010). Impunity denies Latin American women their right to equal protection under the law, resulting in both an inability to live free from violence and a violation of international human rights obligations (Gherardi 2016).
Governments have also been criticized for failing to publicly denounce femicide/feminicide and disseminate information to the public (Fregoso and Bejarano 2010). Specifically, the public lacks knowledge about the implementation of new laws as well as how they affect victims and perpetrators. Criminal justice officials also lack training on how new laws should be applied. For example, in Guatemala, ambiguity remains on how key legal terms should be interpreted, resulting in gender-based violence not being identified as such (Musalo and Bookey 2014). The general lack of knowledge about legislation has been cited as a reason for the few successful femicide/feminicide prosecutions (Musalo and Bookey 2014). This limitation was the impetus for the development of the Latin American Model Protocol for the Investigation of Gender-Related Killings (Femicide/Feminicide), which provides information on the legislation in various countries and on how police, prosecutors, and other civil servants should approach the investigation of potential femicides/feminicides (Sarmiento et al. 2014).

The "What's the Problem Represented to Be" Approach
Introduced in 1999 and modified in 2009, Carol Bacchi's WPR methodology is used to qualitatively analyze discourse. Bacchi (2009) proposed that, instead of evaluating policies solely on their capacity to solve problems, we need to examine how policies construct problems. Governments draft legislation with intent, carefully choosing what will be included and omitted. Traditional policy analysis tends to focus on measurable outcomes or statistics (Dugan 2003). In the past, legal and social scientists have applied discourse analyses, including statutory interpretation, to discern the meaning of legislation (Sullivan 2016;van Dijk 1997). Traditional policy analysis has therefore focused on how governments address or handle social problems (Goodwin 2012); in contrast, Bacchi's (2009)  conceptualized. Her approach also differs from other policy analyses by considering how policy limits discussion due to the underlying assumption that a particular policy is the best solution (Bacchi 1999). What's the problem represented to be? An approach to policy analysis 1) What's the "problem" … represented to be in a specific policy or policy proposal? 2) What presuppositions or assumptions underpin this representation of the "problem"?
3) How has this representation of the "problem" come about? 4) What is left unproblematic in this problem representation? Where are the silences? Can the "problem" be thought about differently? 5) What effects are produced by this representation of the "problem"? 6) How/where has this representation of the "problem" been produced, disseminated, and defended? How has it been (or could it be) questioned, disrupted, and replaced? Bacchi's (1999;) WPR approach has been used to analyze various social issues including poverty and alcoholism (Bacchi 2015;Bastian and Coveney 2013). 4 Previous WPR research has also compared results cross-nationally, examining Australian and Canadian public health recommendations (Alexander and Coveney 2013) and evaluating controversial drug policies in Australia and Britain (Lancaster et al. 2015). Specific to women's issues, researchers have used this approach to examine gender mainstreaming and domestic and family violence legislation (Bacchi 2009;Murray and Powell 2009). Bacchi (1999) makes connections to legislative development targeting gender-based violence, arguing that developments can help or hinder women depending on how problems are defined. For example, if policies to reduce VAW presume that this problem is defined by the breakdown of social order instead of inequality, then solutions could include allocating funds away from women's shelters and toward police forces (Bacchi 1999). Though Bacchi makes a connection to gender-based violence, the present study is, to our knowledge, the first time the WPR approach has been used to assess legislative responses to femicide/feminicide specifically.

Sample for the Present Study
To understand how each country has approached femicide/feminicide, the present authors examined femicide/feminicide legislation implemented in Latin America from 2007 to 2016. 5 Sixteen examples of femicide/feminicide legislation were included from Mexico, Central and South American, and Caribbean countries (see Table 2 for a list of countries). Legislation were included if it defined femicide/feminicide, introduced a punishment, or provided procedures for investigating/prosecuting femicide/feminicide cases. For example, the Dominican Republic added femicide as a crime in its criminal code. Meanwhile, Bolivia enacted legislation that, in addition to defining feminicide, explained how criminal justice actors should investigate and prosecute feminicide. Legislation were located from national websites or databases compiled by international organizations, including the Gender Equality Observatory for Latin America and the Caribbean.
Femicide/feminicide legislation is the product of both regional and national contributions (Kimelblatt 2016). All legislation analyzed were enacted within a nine-year period. Legislation enacted earlier tended to be more restrictive than legislation enacted more recently. Countries have been urged to clearly define femicide/feminicide so that killings can be properly and uniformly identified as such. The inclusivity of recent legislation is thought to be, in part, a lesson learned from countries having already legislated (Sarmiento et al. 2014). Legislative development is important to problem representations because political, social, and cultural atmospheres shape IJCJ&SD 5 www.crimejusticejournal.com 2020 9(2) representations. Femicide/feminicide legislation continues to be amended to reflect an increasing awareness of the causes and indicators of femicide/feminicide.
The analysis began with a thorough read-through of the sample which included legislation from 16 countries. As a method of analysis, the WPR approach, which includes a set of predetermined questions, provides a clear method for examining policies. The first author posed each question, in the order suggested by Bacchi (1999), to each country's legislation before moving on to the next question. This process helped to maintain consistency across countries. Qualitative content analysis often involves multiple readings and going back and forth between stages of analysis (Harding 2013). As such, each country's legislation was examined several times, each time enriching the analytical interpretation.

Question One: What's the "Problem"… Represented to Be in a Specific Policy or Proposal?
The first question is meant to describe the problem and its proposed solutions as constructed within the legislation. Solutions are provisions that aim to change behaviors, actions, or beliefs. The findings in relation to Question One focused on femicide/feminicide definitions, their specificity, and characterizations. This initial question served as a clarification exercise (Bacchi 2009). Analysis examines the effect of representations; therefore, it is critical that representations are identified and explored.
Femicide/feminicide is recognized as a regional and global problem. The international community has encouraged governments to adopt specific mechanisms to prevent, investigate, and eliminate femicide/feminicide, including ending impunity and ensuring the punishment of perpetrators (Puri 2014). Ten countries examined made reference to their international commitments to protect women from violence as a justification for the shift toward a more holistic approach to VAW and defining femicide/feminicide. 6 Two statutes appear particularly influential in the recognition of femicide/feminicide as a global human rights issue: Framing femicide/feminicide as a human rights violation is one method to persuade states to adopt and amend legislation (Weldon 2006). The increasing importance of human rights is, in part, why there has been regional development of femicide/feminicide legislation (ECLAC 2014). A rights-based approach shifts the problem representation by opening the discussion to the international community. An international approach also provides victims a mechanism for holding states accountable when they fail to comply with their international obligation to protect women from violence (Cole and Phillips 2008).

Question Two: What Presuppositions or Assumptions Underpin This Representation of the "Problem"?
Question Two unpacks background knowledge or taken-for-granted assumptions in legislation.
In other words, Question Two asks what meanings must be in place for the problem representation to make sense. Analysis of answers to this question explores the forms of knowledge on which policy depends (Bacchi 2009). For example, in legislation regarding illicit drug use, the presumption that drug users are morally blameworthy is more likely to entail a focus on penalties over therapeutic health services.

Cause-Related Assumptions
Five countries-Guatemala, El Salvador, Nicaragua, Panama, and Venezuela-acknowledged unequal power relations or gender inequality as contributing to femicide/feminicide. For example, Panama's legislation included a provision that considers a homicide as femicide if a person kills a woman "for any motive based on her condition as a woman or in the context of unequal power relations" (Sarmiento et al. 2014: 150). Inequality was the only root cause of femicide/feminicide found in the legislation analyzed. Eleven countries acknowledged the existence of gender inequality; however, this inequality was not linked specifically as a cause of femicide/feminicide. For example, in its preamble, the legislation of Honduras recognized that existing criminal sanctions do not recognize inequality as the cause of VAW. However, subsequent femicide provisions have not named inequality as a condition of femicide. Identifying inequality as a cause of femicide/feminicide is consistent with decades of social science research contending that power and gender are at the root of VAW (Ellsberg et al. 2015;Morley and Dunstan 2016).
To summarize, countries across Latin America characterized femicide/feminicide differently but tended to include women previously subjected to violence (the types of violence are explored below). The two primary assumptions identified were relationship-related assumptions and IJCJ&SD 8 www.crimejusticejournal.com 2020 9(2) cause-related assumptions. However, most countries did not speculate on the causes of femicide/feminicide.

Question Three: How Has this Representation of the "Problem" Come About?
There are two interconnected objectives of Question Three (Bacchi 2009). The first is to reflect on developments contributing to the identified problem. The analyst should consider political, social, and economic conditions that led to the "problem" designation (Bacchi 2009). This question highlights the conditions that allow a specific problem representation to develop and become dominant (Bacchi 2009). The second objective is to examine opposing problem representations. To achieve this, it is helpful to examine similarities and differences in representations across countries.

Development of Femicide/Feminicide
The development of country-specific femicide/feminicide legislation has been framed by national and regional progress on understanding femicide/feminicide as a social problem. For example, legislation in Chile defined femicide as an intimate partner problem, while more recently enacted legislation, including those in Bolivia and Panama, defined femicide/feminicide under various circumstances (see Appendix 1 for a timeline of legislative enactment). Feminicide provisions in Bolivia included killing because the victim was pregnant or was is in a situation of subordination or dependence with the perpetrator. Another example is apparent in Colombia, which, in 2008, adopted Russell's (2001:13) definition of femicide as "killing a woman because she is a woman." While this definition was consistent with academic literature, it has been criticized more recently for the conceptual difficulty the definition presents in femicide/feminicide classification, particularly for criminal justice employees and other frontline workers (Mujica and Tuesta 2014;Toledo 2017). In 2015, Colombia amended its femicide legislation, enumerating particular circumstances where the killing of a woman is femicide. For example, a killing is also considered femicide when the crime is an act of oppression and domination over a woman's decisions or sexuality.

Similarities and Differences across Countries
Previous research has demonstrated that femicide/feminicide occurs under different circumstances across countries (Alvazzi del Frate 2011; Fregoso and Bejarano 2010). Although femicides/feminicides are rooted in gender inequality (Marcuello-Servós et al. 2016;Russell 2001), the most prevalent femicide/feminicide type differs by country, suggesting that countryspecific conditions may perpetuate certain femicide/feminicide types. 10 Focusing on countryspecific provisions addresses the second objective of Question Three, which acknowledges competing causal factors underlying the problem. Identified themes included (1) kidnapping and human trafficking, (2) sexual violence, (3) gang activity, and (4) aggravating factors. Although these themes often overlap, the degree to which they occur within the context of each country differs significantly, making it important to discuss these themes independently.

Kidnapping and Human Trafficking
Seven countries' legislation included circumstances where the victim is held against her will or another crime is committed before her death. 11 Mexico and Panama included, as characteristics of femicide/feminicide, provisions relating to kidnapping. If the victim is kept "incommunicado" or held against her will before her death, the killing is classified as femicide/feminicide. Bolivian legislation included killings "connected to the crime of human trafficking or smuggling" as feminicides (Sarmiento et al. 2014: 152). Kidnapping and human trafficking in Latin America are common precursors to femicide/feminicide, particularly in the context of drug trafficking, gang activity, and sex trafficking (Fontenla 2010). Similar to femicide/feminicide, human trafficking is another highly gendered international problem (Wilson 2014). Among victims trafficked internationally, up to 80 percent are women, and the largest category of trafficking victims are IJCJ&SD 9 www.crimejusticejournal.com 2020 9(2) young women and children (Hodge and Lietz 2007). Human trafficking does not inevitably lead to femicide/feminicide, and, in fact, it is better for traffickers if the women remain alive, yet trafficking victims can end up as victims of femicide/feminicide, particularly when they try to escape or call for help (Meneghel, Rocha da Rosa, et al. 2013). The exact numbers of trafficked femicide/feminicide victims are unknown due to the difficulty in measuring both human trafficking and femicide/feminicide. However, human trafficking femicide/feminicide has become increasingly recognized by international organizations (Meneghel, Rocha da Rosa, et al. 2013).

Sexual Violence
Nine countries referenced sexual violence as a condition of femicide/feminicide. 12 Sarmiento et al. 2014: 149). While many countries included sexual violence, only Peru included sexual violence as an aggravating circumstance. The number of countries that recognized sexual violence suggests that this precursor to femicide/feminicide is widespread across the region. This finding is consistent with previous research that has recognized sexual violence as a precursor to femicide/feminicide and as rooted in the notion of women as property or objects to be used by men (Russell and Harmes 2001;Trujillo 2009).

Gang Activity
Related to the theme of sexual violence is gang activity. Gang-related femicides/feminicides can occur in the context of women as drug mules, in sex trafficking, public executions, and to send a political message, or settling rivalries between competing gangs (Racovita 2015). Five countries' legislation included provisions applying to group activity; however, only Nicaragua mentioned gangs explicitly by classifying femicide "as a result of group rituals, of gangs, whether or not using weapons of any kind" (First author's translation, "Ley No. 779 [Law Number 779]" 2012 [Nicaragua]: art. 9). Guatemala, Bolivia, and Panama included provisions classifying femicide/feminicide by more than one person or committed on a group dare. These provisions could also apply to femicides/feminicides committed by family members in the name of "honor." 13 In contrast, El Salvador's legislation labeled group feminicide as an aggravating circumstance rather than a condition of feminicide itself. Instead of the typical punishment of 20 to 35 years for feminicide, group feminicide perpetrators in El Salvador serve 35 to 50 years in prison. This distinction might be due to disproportionate levels of gang violence in El Salvador (Prieto-Carrón et al. 2007;Suarez and Jordan 2007). It is difficult to measure the exact effect of gang activity on crime rates, including femicide/feminicide. Some believe femicide/feminicide rates are largely affected by gang activity (Suarez and Jordan 2007), yet others argue that governments blame a disproportionate number of femicides/feminicides on gang activity because it is easier to blame gangs than to address the structural causes (Godoy-Paiz 2012; Prieto-Carrón et al. 2007).

Aggravating Factors
Lastly, eight countries-Brazil, Colombia, Ecuador, El Salvador, Guatemala, Nicaragua, Peru, and Venezuela-included aggravating circumstances, which increase sentence length. These provisions reflect what legislators perceive as more serious forms of femicide/feminicide. For example, Peru's legislation contained seven aggravating factors increasing victim vulnerability, such as "the victim was a minor," "the victim was pregnant," and "the victim had any type of disability" (First author's translation, "Ley que Modifica el Artículo 107[Law That Modified Article IJCJ&SD 10 www.crimejusticejournal.com 2020 9(2) 107]" 2013 [Peru]: art. 108). In these instances, legislators are prescribing tougher sentences for feminicides they perceive as more callous, suggesting a scale of severity. Many of these aggravating factors are supported by research on VAW and victim vulnerability (Sarmiento et al. 2014). These provisions are consistent with the notion that not all women are equally likely to become femicide/feminicide victims. While VAW and femicide/feminicide are widespread issues that traverse social and economic divides, research has demonstrated that particular groups of women may be more vulnerable to violence, including femicide/feminicide (Weil et al. 2018). For example, the young and the elderly, in addition to those with disabilities, are recognized as being at increased risk of femicide/feminicide (Sarmiento et al. 2014).
In summary, Question Three shows femicide/feminicide legislation has developed nationally and cross-nationally across Latin America. This development has resulted in similarities and differences across countries regarding their problem characterizations. Some Latin American countries experience similar social problems such as gang violence and kidnapping, which contribute to femicide/feminicide (Racovita 2015).

Question Four: What is Left Unproblematic in this Problem Representation? Can the "Problem" Be Thought About Differently?
Question Four considers representation limitations. There are many other ways to think about the problem of femicide/feminicide, but policies are limited by how problems are represented (Bacchi 2009: 13). The WPR approach directs close attention to the gaps or silences in legislation, looking for elements unspoken (Tonkiss 2004).

Controversial Language
Legislation in four countries contained controversial language rooted in machismo perspectives. As outlined in Question Three, femicide/feminicide is defined as a killing that occurs under various circumstances. In Guatemala, Nicaragua, Panama, and Venezuela, this included if the victim's body was disrespected "to satisfy sexual instincts" (Sarmiento et al. 2014: 148). At the outset, this construction may appear synonymous with El Salvador's sexual classification, whereby feminicide includes crimes committed against the victim's "sexual liberty." However, the word "instinct" implies a mindset, one biologically or socially predisposed, suggesting that men may be prone to sexually assaulting women. This language could mitigate culpability. In the past, Latin American VAW legislation increased discrimination against women by labeling femicides/feminicides as crimes of passion and giving men reduced sentences for killing their partners (Russell 2008;Sanford 2008). If sexual violence is considered a predisposition, men may also receive reduced sentences for sexually-motivated femicides/feminicides.

Private Sphere Femicide/Feminicide
Femicide/feminicide in Latin America is problematized in several ways. Countries with more limited femicide/feminicide definitions did not call attention to other circumstances under which women are killed. Other limited definitions classified femicide/feminicide as an intimate partnerspecific crime. Historically, governments have been hesitant to interfere in the private sphere, including intimate partner violence (Bacchi 2009). However, Chile, Costa Rica, and Ecuador addressed only private sphere violence, suggesting that femicide/feminicide is solely a private sphere problem. These countries, through legislation, fail to recognize the various relationships under which femicide/feminicide can occur and the power inequalities existing across many relationship types involving women and men. This characterization may result in stranger killings not being classified as femicides/feminicides. As a result, perpetrators may receive lighter sentences because their crimes are not recognized as femicide/feminicide. While not specifically omitted, indigenous women are also overlooked throughout the legislation. 14 For example, only nine countries recognized indigenous women in violence prevention initiatives and IJCJ&SD 11 www.crimejusticejournal.com 2020 9(2) acknowledged that indigenous victims face additional barriers to justice. However, no country specifically recognized indigenous victims in their femicide/feminicide provisions.

Femicide/Feminicide Prevention
An increasing trend among countries with femicide/feminicide legislation is a shift away from focusing solely on punishment. Instead, such countries include efforts focusing on femicide/feminicide prevention. In this context, prevention can be defined as any activity that reduces the probability of femicide/feminicide (Weil et al. 2018). A growing body of literature suggests that there are often missed opportunities to prevent femicide/feminicide. For example, femicide/feminicide victims and perpetrators may access certain services before the killing (Ellsberg et al. 2015;Sarmiento et al. 2014). Thirteen countries have now included femicide/feminicide preventions in their legislation. 15 Such provisions range from restraining orders, public education campaigns for school-aged children, increased services for victims, and training for frontline workers. Countries also vary in the number and depth of these provisions. The movement toward proactive instead of, or in addition to, reactive responses represents a drastically different approach. By attempting to reduce the violence that women experience before death instead of waiting until femicide/feminicide occurs, countries may reduce the number of victims and by extension, the number of femicide/feminicide perpetrators.

Question Five: What Are the Effects Produced by This Representation of the "Problem"?
The purpose of Question Five is to critically evaluate policies and to question the effects of specific representations (Bacchi 2015). We consider how problem representations may restrict what is considered relevant and, subsequently, influence people's understanding of the problem (Bacchi 1999). These effects potentially differ between countries and across regions. However, we focused on more general effects that could occur in one or more countries: gender inequality and impunity.

Gender Inequality
Several countries acknowledged gender inequality as an important underlying cause of VAW and femicide/feminicide, yet few countries have explained how they plan to rectify this social imbalance. Femicide/feminicide is recognized as a persistent manifestation of gender inequality (Vives-Cases et al. 2016), but, if the root of the problem is not addressed, femicide/feminicide legislation will remain reactive instead of proactive. Failing to acknowledge the underlying cause may also allow high femicide/feminicide rates to persist despite increased penalties. Continuing gender inequality is also related to the ongoing impunity associated with femicide/feminicide, as women are undervalued, which discourages public servants from preventing and punishing crimes against them. For example, according to Castañeda Salgado (2016), referring to feminicide in Mexico, problems remain because eradicating gender violence requires structural changes that no Mexican government appears willing to undertake. If, as suggested by many, gender inequality is the primary cause of femicide/feminicide, rates may remain high despite the implementation of legislation denouncing femicide/feminicide (Ellsberg et al. 2015;Morley and Dunstan 2016).

Impunity
Ten countries included impunity or institutional violence, defined as violence committed by public servants or public servants who impede victim's access to justice. 16 For example, institutional violence in El Salvador is described as: Acts or omissions of public servants of any order of government that discriminates against or has the purpose of delaying, hindering or preventing enjoyment and exercise of women's human rights as well as their access to the enjoyment of public policies aimed at preventing, attending, investigating, punishing and eradicating the different types of violence ( Impunity has been identified as a major contributor to femicide/feminicide rates in Latin America (Fregoso and Bejarano 2010;Walsh and Menjívar 2016). As noted above, impunity is one condition in Latin America that distinguishes femicide/feminicide within this region. Impunity provisions recognize previous problems in the criminal justice system, attempting to address the systematic issues leading to ineffective femicide/feminicide prosecutions. In countries with continuing impunity, legislation is unlikely to be effective in increasing protection for women against femicide/feminicide. An example of this is found in Guatemala where, in 2011, more than 20,000 cases were filed under the 2008 Ley contra el Femicidio y Otras Formas de Violencia Contra la Mujer (Law against Femicide and Other Forms of Violence against Women). The over 20,000 cases mentioned included femicide, as well as physical, sexual, and emotional abuse against women (Musalo and Bookey 2014). Despite the number of cases filed, less than three percent of these cases resulted in a judgment (Musalo and Bookey 2014: 106). Widespread biases, largescale inequality, and the normalization of VAW remain primary theories for why impunity and, correspondingly, femicide/feminicide rates, remain high in countries that have implemented femicide/feminicide legislation (Walsh and Menjívar 2016).

Question Six: Where Has This Representation of the "Problem" Been Produced, Disseminated, and Defended? How Has It Been (or Could It Be) Questioned, Disrupted, and Replaced?
Question Six is intended to assess discussions surrounding problem representations. Similar to Question Three, Question Six highlights the existence and possibility of contention, destabilizing taken-for-granted "truths" (Bacchi and Goodwin 2016). As an internationally-recognized problem, various femicide/feminicide reactions/solutions have been discussed, tested, defended, and criticized. Since policy and practice are known to differ, this is a key issue addressed by this question. We intend to provide a snapshot of discussions around femicide/feminicide upon which future research might build.

Definitions
Government definitions of femicide/feminicide differed from those used in academia, where definitions are also debated. "Femicide" is more commonly used, although "feminicide" has gained popularity in Latin America. Furthermore, governments tended to use more narrow definitions than their academic counterparts. This discrepancy demonstrates differing productions of the problem and how authors and their perceptions frame problem representations. This distinction is also a product of the utility in defining the "problem." Although an academic definition may perfectly describe the theoretical underpinnings of the issue, a legislative "problem" definition has more parameters. Unlike academics, governments do not use "feminicide" to reference impunity or to take responsibility for previous failures to punish perpetrators (Fregoso and Bejarano 2010;Sanford 2008). For example, Brazil uses the term "feminicide" but does not acknowledge impunity nor any systemic conditions that foster VAW.

Femicide/Feminicide Observatories
International organizations and scholars from within and outside the region have evaluated and debated femicide/feminicide "solutions" in Latin America. More recently, these actors have made efforts to help define and reduce femicide/feminicide by implementing observatories. In the past, a significant barrier to identifying and combating femicide/feminicide has been the dearth of available data (Weil 2016). Historically, homicide data were not disaggregated by sex/gender. Although sex-specific data has become more widely produced, data on femicide/feminicide rates and sufficient information to classify femicide/feminicide remain largely unavailable (Weil 2016). This issue is not unique to Latin America and has resulted in a global call, by the United Nations Special Rapporteur on Violence against Women, Its Causes and Consequences, for states to develop national observatories to collect and record femicide/feminicide data (Weil et al. 2018). Observatories are expected to accurately inform and disseminate information on femicide/feminicide. Several countries have included the development of national VAW IJCJ&SD 13 www.crimejusticejournal.com 2020 9(2) observatories in their femicide/feminicide or VAW legislation, outlining the goals and responsibilities of their respective observatories. 17 For example, Peru's legislation reads: The National Observatory on Violence against Women and Group Members Family, in charge of the Ministry of Women and Vulnerable Populations, aims to monitor, collect, produce and systematize data and information by following up on public policies and international commitments assumed by the State in this matter. Its mission is to develop a permanent information system that provides inputs for the design, implementation, and management of public policies aimed at prevention and eradication of violence against women and members of the group family (First author's translation, "Ley Para Prevenir [Law for the Prevention]" 2015 [Peru]: art 43]. Countries without legislated observatories may still have an observatory through other mechanisms. However, observatories stipulated through legislation clearly define the rules and responsibilities of actors, which may increase accountability. The implementation of observatories across Latin America has the potential to increase understanding of femicide/feminicide in local, regional, and international contexts. Other countries and regions have also recognized the importance of femicide/feminicide observatories, with the development of these initiatives in Canada and across Europe (Dawson 2018;Weil et al. 2018).

Implementation
A common taken-for-granted truth is that legislation, once enacted, may not always have its intended impact in practice, and governments have been urged to better implement legislation following enactment. In the time that femicide/feminicide legislation has been in place, countries have been criticized for failing to enforce legislation (Sarmiento et al. 2014). Measures following enactment have been described as ineffectively implemented, lacking sufficient monitoring and review, and thus seldom protecting women from violence (Amnesty International 2006; Fregoso and Bejarano 2010). Reasons for implementation failure range from practicalities, such as budgetary constraints, to a blatant disregard for women's lives (Fregoso and Bejarano 2010;Walsh and Menjívar 2016). Currently, there is a lack of empirical research into legislation implementation and enforcement by justice officials (ECLAC 2014: 69). This dearth is not unique to Latin America, as the absence of this research is a problem recognized globally (Mujica and Tuesta 2014). However, the existing research suggests that implementation remains a problem due to highly ingrained societal and political biases about women's equality and the violence perpetrated against them (Walsh and Menjívar 2016).

Conclusion
In the last 40 years, Latin American countries have made significant strides to increase protections for women (Michau et al. 2015). This progress has been greatest in law reform, recognizing VAW and femicide/feminicide more specifically. Although countries have enacted femicide/feminicide legislation in different ways, the decision to legislate is itself an important step toward increasing the protection of women and girls (ECLAC 2014). While there is little argument that these are positive steps in recognizing women's rights, several legislative limitations remain (ECLAC 2014).
Using Bacchi's (2009) WPR approach, we highlighted problem representations of femicide/feminicide across Latin America. Responses to femicide/feminicide have been shaped by how countries have constructed the problem. Through defining femicide/feminicide using victim-perpetrator relationships and motivations, governments shape responses to femicide: that is, femicide/feminicide constructions depict who is protected and how. These representations suggest who commits femicide/feminicide and under what conditions IJCJ&SD 14 www.crimejusticejournal.com 2020 9(2) femicide/feminicide is most likely to occur. These characterizations may affect how some frontline sectors respond to woman-killing and will impact the sentences perpetrators receive.
Most countries recognized that femicide/feminicide occurs in both private and public spheres and that the circumstances of the killing differ within each sphere. Although this finding may seem obvious, it is a critical step in developing effective preventative responses to femicide/feminicide. The circumstances that differentiate public and private femicide/feminicide provide governments with insight into how to better protect women from femicides/feminicides in either sphere.
Bacchi's (2009) WPR analysis also demonstrates how "problems" that appear identical superficially are affected by different underlying representations: problems are shaped by the unique contexts that permit their occurrence. These differences are evident in the circumstances that countries have prioritized as a condition of classifying femicide/feminicide. Femicide/feminicide in each country is characterized by a series of complex factors including impunity, inequality, and gang activity. Increased attention to these factors will help develop targeted and effective policy responses to femicide/feminicide.
Currently, it is unclear the extent to which each country experiences difficulties in implementation. A comprehensive assessment is needed to understand why femicide/feminicide legislation remains ineffective in Latin America and elsewhere. This information may provide insight into agency failures, funding issues, or social attitudes that may hinder the successful implementation of femicide/feminicide legislation. Information regarding the effectiveness of legislation implementation, in combination with this present study, will suggest future steps that can be taken to protect women from femicide/feminicide in Latin America and abroad.