Conceptualising Violence in Close Relationships: Discrepancies Between Police Conceptions and the Letter of the Law in Finland

The focus in this Finland-based study is on violence in close relationships—a term that partly overlaps with the more commonly used ‘domestic violence’, ‘family violence’ and ‘intimate partner violence’. We demonstrate how police officers’ conceptualisations of such violence differ from how it is defined in relevant legal documents. The data consists of the Government Bill and legal text on the subject issued as part of a legal reform enacted in 2010, and of a qualitative sample of freelist responses from 79 police officers. We examined both sets of data using theory-driven directed content analysis and deriving from prevailing theoretical frameworks reflecting the family- and gender-based perspectives on violence. The results expose the predominance of a narrow definition of ‘family’ in police understandings of close relationships, but also a notably broad spectrum of conceptualisations of both physical and non-physical forms of violence. In contrast, the legal definition of a close relationship is broader and encompasses multiple types of relationships, whereas forms of violence are more strictly defined. These findings could explain some of the discrepancies between legal policies on violence in close relationships and police responses to it.


Introduction
Violence in close relationships encompasses a vast range of relationships and aggressive behaviour between people. The concept is commonly referred to as family violence (FV), domestic violence (DV) or intimate partner violence (IPV) in the literature, and many researchers study it from the theoretical perspective of violence against women (VAW) or, more broadly, gender-based violence (GBV). These terms are sometimes used in an overlapping manner, or even as synonyms, but they represent recognisable perspectives and different conceptualisations of the relationships between victims and perpetrators, and the types of violence in question (e.g. Bonnet 2015;Lawson 2012). Conceptualisations have the potential to influence legislation. Furthermore, as ideas spread through legislation and professional guidelines, via the media and in public debate to policing practices, they may affect how the police understand and detect violence in relationships. Our aim in this paper is to examine the conceptions used in legal documents in parallel with police definitions of violence in close relationships and to analyse how key aspects of the perspectives and conceptualisations are viewed in Finland. In keeping with the national legislative material analysed, we use the term 'violence in close relationships' when not directly citing the terms used by others.
The tendency in Finland until fairly recently was to consider violence within families and other close relationships, compared with violence between strangers in public places, a private matter that did not require police intervention (Kotanen 2013(Kotanen , 2018. The phenomenon attracted considerable public attention and research interest in the 1990s, which was relatively late by Western standards. The transnational feminist movement has affected policy development in Finland, advocating the defining of DV as a gender issue (Virkki 2017). However, policies and services in Finnish society have been dominated by apparent gender neutrality and the definition of the problem in terms of the gender-neutral concept of FV (Hearn and McKie 2010;Virkki 2017).
A reform of the Finnish Criminal Code in 2011 defined violence in 'close relationships' from the criminal-justice perspective, conveying the message that, even in milder forms conceptualised as petty assault in the Criminal Code, it is unacceptable and therefore a relevant issue of public policy. A critical change was that, for the first time, all forms of assault in close relationships, including petty assaults, became subject to public prosecution. In principle, therefore, violence within families and other close relationships was no longer a private matter if, for instance, the victim was unwilling to pursue a judicial inquiry or the police were reluctant to start a pre-trial investigation. The legislation and law-drafting documents, particularly the Government Bill 78/2010 (hereafter, the Government Bill or the Bill), provide information about the purpose, content, scope and justification of this legal reform.
The police are the frontline responders and pre-trial investigators in the Finnish criminaljustice system, and the aforementioned legal reform enacted a mandatory recording 1 policy in cases of violence in a close relationship. However, police statistics show that there are substantially more FV calls than recorded FV cases . There is reason to suspect a discrepancy in police perceptions of violence in close relationships and its documentation, as it appears that not all cases fulfilling the definitional prerequisites are categorised accordingly (Kotanen and Smolej 2014). It is true that not all the FV tasks the police deal with give reason to report a crime. However, a 2015 study of a major police department in Finland implies that some incidents in which the police detected violence in a close relationship were still left unrecorded ).
There may be several reasons for the discrepancy between the crime records and the number of police FV tasks, such as a lack of evidence. The legal and extra-legal factors associated with recording an offence have been analysed elsewhere , however. Here, we focus on the conceptual aspect and examine how police officers name and define violence in close relationships compared with the conceptions in the Government Bill that subject petty assaults in close relationships to public prosecution. There is no prior empirical research on how Finnish police officers define FV or violence in close relationships. Furthermore, although there have been studies on how police attitudes affect responses to DV (e.g. Belknap 1995;Logan et al. 2006), none thus far show to what extent frontline responders' conceptualisations of this type of violence reflect the relevant legislation.
Our focus of interest in this study is on the possible gap between the law and its enforcement in the conceptualisation of violence in close relationships. First, we briefly review the relevant laws and legal reforms to clarify the relevant Finnish legislation. Second, we set out the official instructions given to the police, which are crucial in terms of understanding the broader normative context in which frontline responders carry out their work. Third, we dissect the main elements of the prevailing theoretical perspectives on violence in close relationships and summarise them in two major groupings, i.e. GBV and FV. After presenting our data and methods, we proceed to the results. Finally, we discuss our empirical findings in light of the theoretical perspectives to shed light on how the key elements of violence in close relationships differ in the legislation and in police conceptions, and highlight the potential implications.

The (Changing) Letter of the Law
Finland's Criminal Code does not classify DV or FV as crimes, but it does treat violence in close relationships and between strangers differently. Several amendments to the Criminal Code have changed the former from a private matter to a public concern, and one of the most fundamental changes was the 1994 criminalisation of rape in marriage. Moreover, common assault in private spaces was a complainant offence 2 pre-1995. Since then, it has been subject to public prosecution but with the further option of non-prosecution based on the 'firm and free will of the victim'. The possibility of non-prosecution was strongly criticised as vitiating the original purpose of protecting the victim and ensuring that the perpetrator could not pressurise the already abused victim into dropping the complaint. Section 17, which coded nonprosecution into the law, was removed from the Criminal Code in 2004 (Kotanen 2013(Kotanen , 2018. Although other petty assaults are still complainant offences, those occurring in close relationships have been subject to public prosecution since 2011, as have petty assaults against minors and people at work duties (Ministry of Justice 2015a).
According to the Police Act, '[t]he duty of the police is to secure the rule of law; maintain public order and security; prevent, detect and investigate crimes; and submit cases to prosecutors for consideration of charges' (Ministry of the Interior 2015). The Criminal Code (Ministry of Justice 2015a) defines behaviours that are criminal, punishable and therefore subject to police control. The 2011 amendment includes the legal definition of a close relationship, and this relationship should be recognised whenever behaviour meeting at least the Criminal Code definition of petty assault comes to the attention of the police. A close relationship is defined as that between siblings, spouses and former spouses, children and their parents, and people who live or have lived in a joint household or otherwise are or have been in a corresponding personal relationship with each other. This legal change and definition provide the basis for institutional responses to violence in close relationships because they also impose changes on police practice.
Based on various data and research findings, the Bill views most violence in close relationships as petty assault and hidden crime that do not proceed through the criminal justice system. Under previous legislation, if the police determined the case as petty assault, no criminal investigation would have taken place without the complainant's summary penal order. The reasoning in the Bill is that, to secure the investigation of violence in close relationships that may at first appear mild, the decision about recording the offence should not be at the victim's discretion. Recording the offence and documenting the evidence in such cases should be done immediately by the police in attendance: it is recognised that it is more difficult to collect evidence later in the process, particularly because petty assault does not necessarily produce visible injuries.
Following the 2011 legal reform, in 2015, Finland ratified the Council of Europe Convention on preventing and combating VAW and domestic violence (the Istanbul Convention), thereby giving it the status of law. The Convention covers all forms of VAW and recommends it being applied to all victims of DV, calling for special attention to women and girls as victims of GBV. As part of its implementation in Finland, the Convention was extended to men and boys who experience DV (Ministry of Social Affairs and Health 2017).

The (Stagnant) Definitions in Police Instructions
Police facing potential criminal events have a duty to make a written report at the earliest opportunity. If either a report or direct observation gives the police reason to suspect that a crime has been committed, a pre-trial investigation must be conducted. The person doing the reporting does not need to know what the suspected crime was, as the police will identify it based on definitions in the Criminal Code (Lönnroth and Rantaeskola 2014, p. 77). Therefore, the police need to be able to perceive and name (Felstiner et al. 1980) an incident as violence in a close relationship in order to process the case in the manner required by law. This, in turn, depends on how aware the police officers are of the different definitions of crime and whether their conceptions of violence in close relationships are in line with the law.
Police work is not simply the mechanical implementation of the law, however. There is empirical evidence showing that police officers exercise discretion in their work (e.g. Diemer et al. 2017;Goldsmith 1990;Grant and Rowe 2011), which is often necessary because they have to interpret idiosyncratic events based on generalised legal texts and categories. In the spectrum of different police tasks, intervening in violence in the private sphere has been particularly challenging (e.g. Diemer et al. 2017;Logan et al. 2006).
The instructions covering cooperation between the Emergency Response Centre and the Police of Finland include FV as a task category, instead of violence in close relationships. The definition includes physical violence or noises indicating violence in an apartment, and the parties are not necessarily family members (National Police Board 2014). This categorisation is meaningful for police operational preparation in that the focus is not on the relationship between the people involved but on where the incident takes place, in an apartment or a house as opposed to a public space, for example.
The police should also categorise certain types of crime reporting as FV, defined as follows: Violence within the family, against a family member. It should be understood very broadly in terms of both the parties involved and the type of violence. Physical violence or threats of it constitute family violence when the perpetrator and the victim are in an emotionally close relationship, regardless of where the incident takes place (Ministry of Interior 2005).
The legal reforms from the 1990s up to 2011 changed the responsibilities of the police in recording offences in cases of violence in close relationships, whereas the aforementioned definitions in police instructions have remained the same. According to police statistics, an increasing number of emergency calls are being categorised as FV . General victimisation surveys covering hidden crime do not indicate an increase in IPV (Danielsson and Näsi 2018) or in violence against children perpetrated by their parents (Fagerlund et al. 2014). It is therefore likely that the increase reported in police statistics implies that people in Finland nowadays are more aware of the problem and are more willing to report it. Simultaneously, the number of recorded FV offences has stabilised at a relatively low level, which suggests that police officers' recording behaviour and documentation have not similarly changed .

Theoretical Frameworks of Violence in Close Relationships
As explained above, in the Finnish context, it is justified to use violence in close relationships as an umbrella concept that is broader than the internationally more commonly used DV or FV (see, however, Frieze 2000; Marcus and Swett 2002;Spitzberg 2009). The chosen terminology in the literature tends to reflect the theoretical framework that guides a specific piece of research on violence (Lawson 2012), thus establishing linkages to these frameworks in national data is relevant regardless of the legislative and linguistic context. According to the relevant literature on the international level, the most striking difference in naming and defining violence in close relationships is between family violence and gender-based violence perspectives. Studies reviewing these theoretical standpoints and their methodological implications could enhance understanding of the theoretical debates and the critical features to look for in empirical data. Our purpose is not to examine or compare theories in terms of their explanatory power, or to look for causes and mechanisms of violence in our empirical data. Both branches of research are decades old, thereby placing their comprehensive review beyond the scope of this paper. We rather aim to utilise existing literature focusing on some of the key points of these theoretical frameworks to identify and describe the similarities and discrepancies in the conceptualisation of violence in close relationships in our empirical data. Table 1 summarises the features identified in the literature to guide the analysis of our empirical data. First, the naming of the phenomenon of interest is a common although not straightforward result of conceptualising and defining it. Research endorsing the perspective of GBV tends to refer to different forms of 'violence against women', 'intimate partner violence', 'domestic abuse' to include coercive control and, before the 1990s, 'wife-beating' and 'wife-battering'. Terms such as 'family violence', 'domestic violence', 'violence against children' and 'child maltreatment' are preferred in FV studies (Ali and Naylor 2013;Bonnet 2015). The use of terminology is not without exception or overlap: some studies with titles including DV, for instance, define it as VAW by male partners and ex-partners (e.g. Hoyle and Sanders 2000;Johnson 2007).
What a phenomenon is called may reflect the unit of analysis. In the tradition of FV studies, violence is interpreted as conflict within the family, hence the family or a relationship between partners could be the unit of analysis (Bonnet 2015;Lawson 2012), whereas in mainstream sociology and criminological research focusing on violence, it could be any person in the general population, male or female, adult or child. General surveys of (crime) victims would appear to have more in common with the FV than the GBV perspective given the status of gender as a background or control variable in the former. The chosen unit of analysis might also reflect the debate concerning the counting of victims or of violent incidents. In the GBV context, criminological research and quantitative surveys of crime victims are criticised for calculating the prevalence of victimisation or the number of violent incidents in the population. According to its critics, mainstream criminology is insensitive to the special characteristics of violence, context and the consequences (e.g. Dobash and Dobash 2004), and fails to recognise and measure the continuum of violence and controlling behaviour by men against their female partners (Bonnet 2015;Walby et al. 2014). On the other hand, it has been suggested that the emphasis on qualitative methods within GBV studies stems from the failure of findings from quantitative surveys to illustrate the gendered nature of IPV (Dutton and Nicholls 2005). The type of relationship is a common marker of a theoretical framework. The focus of DV and IPV studies in the past has been on heterosexual partnerships: as (Murray and Mobley 2009, p. 363) note, because of 'widespread cultural perceptions that intimate partner violence is limited to male perpetrators abusing female victims, discriminatory biases against the LGBT community and same-sex relationships, and reluctance to acknowledge that large numbers of same-sex intimate relationships exist'. The broadening of the gender perspective has widened the lens of GBV such that the experiences of people identifying themselves as LGBT, for instance, are now given special attention (e.g. O'Toole and Schiffman 2007; Wirtz et al. 2020). Including the LGBT population in DV studies entailed moving beyond the heterosexual paradigm of the battered women's movement, which, as Murray and Mobley (2009) argue, limited the availability of terms and theoretical conceptualisations to describe same-sex IPV. The main focus in studies that have more in common with the FV perspective, or seemingly gender-neutral mainstream sociology, has been on family relationships and interaction between family members (Lawson 2012;Bonnet 2015).
The presumed directionality of violence may likewise be embedded in the methodological choices of different researchers. Reflecting the above arguments, studies from the GBV perspective have focused on violence by men against women under the assumption that women resort to violence predominantly as self-defence (Ali and Naylor 2013;Bonnet 2015). Gender-sensitive perspectives in the 2000s and 2010s focused on the concept of intersectionality, in other words on understanding the intersections of age, gender, race, sexuality and class in varying institutional contexts, including interventions in violence (Donovan and Hester 2010;Husso et al. 2017;Subirana-Malaret et al. 2019). Research from either of the above-mentioned perspectives may treat children as mere witnesses or secondary victims when implicitly referring to violence between adults, and therefore does not include violence directed against children as part of the same larger phenomenon. Under the definition of VAW, this is more likely to be connected to the GBV perspective. In terms of directionality, an area of family studies still described as somewhat of a taboo is child-to-parent violence (e.g. Armstrong et al. 2018;Miles and Condry 2016).
An ongoing debate within the two main theoretical streams concerns the types of violent behaviour that should be included. The conflict tactics scale, originally developed by Straus, and its later applications (CTS2) measure conflict in families, listing 'things that might happen when you have differences' (Straus et al. 1996, p. 310): this is different from measuring VAW (Bonnet 2015). Walby et al. (2014) recognise a divergence in the definitions of violence used in mainstream criminology and GBV studies: the former relies on national criminal law, whereas the latter deploys broad definitions used in international policies such as the United Nation's General Assembly's (1993) Declaration on the Elimination of Violence against Women. In line with this debate, conceptualisations of violence focused on criminal offences align more closely with mainstream sociology and criminology, whereas broader definitions referring to coercion and control reflect conceptualisations that are characteristic of the GBV perspective.
As the above review indicates, the key features of the two main theoretical frameworks are closely interwoven such that the naming of violence may indicate the focus or the unit of analysis. This could be connected to the type of relationship, the presumed directionality of the violence and the behaviour inherent in the very concept. The relevance of gender is at the core of the theoretical debate, illustrated in each feature reviewed concerning the two perspectives. In our reading and interpretation of police freelists and legal texts, we identified conceptualisations that refer to or have in common with certain theoretical frameworks that enabled us to analyse the two different types of textual data in parallel.

Data
The Survey and the Freelist Question for the Police We used qualitative data gathered as part of a web-based questionnaire to examine police officers' conceptions of violence in close relationships. The questionnaire was developed for a project called Enhancing Professional Skills and Raising Awareness on Domestic Violence, Violence against Women and Shelter Services (EPRAS), and was distributed during the first data collection phase in December 2017 and January 2018. The research part of the project was conducted jointly by the University of Jyväskylä and the Police University College of Finland. The research plan was approved by the Ethical Committee of the University of Jyväskylä. The purpose of the project, which was coordinated by the National Institute for Health and Welfare, was to develop the first nationwide online training module on violence in close relationships and VAW, targeted at police, social services and healthcare professionals in Finland. The project also included a public awareness campaign to increase awareness of shelter services.
Respondents from two police departments participating in the pilot study were presented with the following statement in a questionnaire during the first phase of the project: 'Please list everything that you consider to be violence in close relationships'. The prompting statement was formulated in line with the established freelisting technique used in cognitive anthropology, described in more detail in the 'Methods' section below. Asking about violence in close relationships, compared with asking about VAW or FV, is likely to produce different types of responses. We chose the term so that we could compare police responses with the concept as defined in the legal text data. We analysed the responses of all 79 police officers who completed the survey and answered the freelist question. The saturation of responses was deemed sufficient for the stated purpose. Table 2 gives the descriptive statistics of the respondents, who differed somewhat in characteristics compared with the personnel in the pilot police departments (PolStat). Around 28% of them were female, compared with 18 in the pilot police departments in 2018. Young police officers aged 24-30 constituted 22% of the sample compared with 11% in the pilot departments. The respective statistics on levels of education were not comparable because of the changes in the educational system and the consequential discrepancies in the classifications. Years of service were similar in the sample and the pilot departments. Of the survey respondents, one-third had not received or could not recall receiving any training related to violence in close relationships as part of their basic training. However, more than 63% had work duties related to such violence daily or weekly, which highlights the need for training.

Legal Text Data
The legal text that we analysed was Government Bill 78/2010, which also includes the final text of the amended Criminal Code. One effect of the Bill was to make petty assault against persons fulfilling their work duties subject to public prosecution, and therefore the original text of the Bill includes sections on workplace victimisation and the reason behind the legal change related to it. These parts were excluded from our analysis because they do not concern the legal definition and conceptualisation of violence in close relationships.

Methods
We used freelisting as our method for collecting data on police perceptions, which is an established technique applied in cognitive anthropology (Borgatti 1998(Borgatti , 1994. The statement 'Please list everything that you consider to be violence in close relationships' was intended to prompt the respondents to disclose culturally salient facts rather than their personal preferences. The freelisting technique is thus based on the assumption that the items listed by respondents belong to a specific cultural domain, in this case 'violence in close relationships', the content and scope of which are shared by members of the social group. The order and frequency of the items listed by the respondents allow the researcher to determine which items are salient across all members of the social group (Quinlan 2005). Freelisting has been used to collect data on illness, medical plants (Quinlan 2005) and romantic love (De Munck et al. 2011), for example.
Freelisting is a powerful data collection technique under the right conditions. To begin with, one needs to be aware that individuals may be unable to list all the items from their long-term memory, particularly if the domain is broad or the prompt is vague (Quinlan 2005): 'violence in close relationships', for instance, is not a simple category. However, freelisting is potentially powerful if the respondents come from a socially and culturally coherent group. There is only one national police force in Finland, and police officers constitute a professionally coherent group, thus it is reasonable to assume that they have sufficiently unified conceptions. The organisation and accessibility of knowledge depends on the respondents' experience of and expertise in the cultural domain under research (Ross and Medin 2005). However, the expertise of police officers derives from their daily work and not simply from formal legal texts. Through freelisting, therefore, it should be possible to access knowledge that is crystallised in the long-term memory that police officers have accumulated during their careers (Thompson and Juan 2006). The definition of violence in close relationships comprises two conceptual elements, i.e. relationships that are considered close and the type of behaviour that is considered violent. A close relationship might be explicitly specified ('couple'), or implied by a role ('husband') or a collective noun ('home'). An act could be a generic noun ('assault') or a specific verb ('hit'). We also looked for other features in the data guided by the literature on two major theoretical frameworks concerning violence in close relationships, i.e. FV and GBV. The key features therefore include the naming of the violence, relationships included, directionality and type and forms of violence. Given that the positioning of gender is at the core of the differences between the two theoretical perspectives, we also examined the visibility of gender in both types of text data.
We analysed legal documents and police survey responses using the NVivo programme for coding units, thematically based on the literature review. Because the data were of two different types, we also applied lemmatisation and used the SPAD statistical programme to check the frequency distribution of terms within the legal text and the freelist data. Both data sets were subjected to theory-driven directed content analysis (e.g. Hsieh and Shannon 2005). We focused on key features identified in dominant theoretical perspectives on violence in close relationships (see Table 1) that could indicate alignment with or differentiation from these theoretical frameworks when we compared the legal text and the police responses. LT refers to data extracts from the legal text and PFX to the extracts from the police freelist data, X indicating the list number. Table 3 shows the listing by word frequency for the two types of data set. A cursory glance at this simple quantification reveals the differences in focus of the police responses and the Government Bill. The term 'violence' is the most prominent in the police responses, whereas the Bill predominantly refers to 'assault', 'violence' being the third most common reference. The terms 'physical' and 'emotional'/'psychological' are among the top three in the police responses, whereas no such specifications stand out in the Bill. The term 'child' is the eighth most common in both sets of data, implying that concern about children's involvement in violence is of equal importance. 'Family' is the fourth most commonly used term in the police responses but does not feature in the top 20 for the Bill, in which 'woman' and its derivatives nevertheless rank relatively highly. On the other hand, in the police responses, the term 'woman' did not appear in the top 20.

The Naming of Violence
The naming of violence as 'violence in close relationships' in the Bill was conceptually and terminologically the starting point of this study, and it is therefore natural that the term is also predominant in the legal text. The second most commonly used name is FV, followed by IPV. VAW is also notably common in the Bill, accompanied with a few mentions of GBV, whereas disciplinary violence is slightly less recurrent.
There are only 11 direct mentions of named violence in the police responses. They include, in order of frequency, violence in close relationships, FV and IPV. This is exemplified in the following extract: [PF50] Violence in a close relationship means physical, psychological and social assault, bullying, discriminating, etc., and the parties are in some way in a close relationship with each other, possibly in the sense that one person is dependent on the other.
The explicit mentions of FV in the police responses are notable because the term is consistent with police instructions concerning FV tasks and in the categorisation of certain crime reports as FV. Given that naming is the first step in considering a dispute potentially criminal (Felstiner et al. 1980), it is particularly meaningful that the concept of FV deployed in police instructions seems to challenge the term mentioned in the legislation and, therefore, also in our freelist statement. Of course, naming alone does not specify what is included in the definition of violence in close relationships, and it is better understood with regard to the relationships included and other key factors. The percentage is counted separately for both data sets as a proportion of occurrence from all words included in the analysis

Relationship Types
Intimate partnerships or relationships between spouses, accompanied with a few mentions of former spouses and partners, best describes the focus of the Bill. Although one of its main goals was to define in as detailed a manner as possible the close relationships included in the reformed law, the Bill involves repeated vague descriptions of someone living close by, a near relative and close relationships in general. Parent-child relationships and general references to family relations are present but not as common, and there are references to different types of dependant and submissive relationships in the Bill: [LT]… mutual relationship of trust and dependence adds to the shock and hurt caused by the act (p. 10).
Dependant relationships are also listed in the police conceptions, as evidenced in the extract quoted in the previous section. This is more of an exception than the rule in the police freelist responses, however. Altogether, the 79 freelists included 144 different items referring to a relationship. As could be deduced from the number of descriptions, a single officer's response might include several mentions of relationships. Twelve of the responses do not include any specific description of a relationship. The comments focus on types and forms of violence, indicating the defining of issues through behaviour rather than interpersonal relationships. The concept of family is predominant in the police descriptions of relationships. Family is mentioned briefly, almost self-evidently, and there are no detailed descriptions or examples of family structures and relationships. Most commonly, the concept of family implies nuclear families, but it is also described in terms of cohabitation and a shared household as the determining factor.
[PF05] Violence occurring within a family, either between spouses or against children, also violence perpetrated by children against their parents [PF42] Intra-familial, in the direct line of ascent and descent [PF43] Physical attack in different forms, and concerning those living in the same household Partnerships and spousal attachment was constituted the second most common type of relationship mentioned by police officers. Almost as recurrent were ambiguous concepts such as close relatives and circles of acquaintances. Former partners were mentioned only twice in the police freelist data, which could indicate that, in general, police officers equate ending an intimate partnership with ending a close relationship. This could potentially lead to the misinterpretation that a petty assault on a former spouse or partner would still be a complainant offence. A few of the definitions also draw parallels between a close relationship and an intimate partnership: [PF67] A close relationship means a stable partnership. It does not have to be formalised, and there need not be the same address in the systems. In an investigation, it should be shown that the partnership is continuous and based on mutual agreement The Directionality of Violence The Bill refers to several points indicating that violence enacted in close relationships, at home, within the family and against women, children and individuals who are otherwise dependent or subordinate to the assailant is particularly objectionable and blameworthy. However, it does not explicitly mention directionality. We coded 46 sections in the Bill that indicate a passive description of violence occurring in a relationship, a partnership or a family.
[LT] The numbers describe violence either between people living in the same apartment or between current or former family members (p. 3) In comparison, violence towards another person, particularly one who is ambiguously termed 'a person close to one' is mentioned in 13 instances. When directionality is clearly stated, it most commonly (14 items) refers to parental violence against their children. This complements findings concerning the naming of violence because it could imply that FV in the Bill generally refers to violence against children. The focus on intervention in this respect is also evident in that children are included in the legal reform concerning public prosecution both as family members (close relationship) and as being particularly vulnerable (of a young age). Violence perpetrated by adult children against their elderly parents is mentioned in the Bill on three occasions.
Police freelist responses resemble the legal text closely in terms of passive expressions of action without an agent or directionality of action. Notably, there are as many as 17 instances police freelist responses that name and describe mere actions as nouns, types and forms of violence without any direction, agent or relationship as the context of these acts.
The freelists are internally diffuse, however, and some include references to violence in the passive form without an agent with specifications of violence targeted at certain people as clarifying examples. Violence perpetrated by parents against their children and violence perpetrated by children against their parents are notably and almost equally represented in the descriptions of directionality in the police conceptions of violence in close relationships.

Types and Forms of Violence
The background of the legal reform and the status quo before are generally described in the Bill with reference to research and statistics featuring the term 'violence' as opposed to directly referring to statistics on assaults, for instance. Nevertheless, the focus in the reform was on assault, and the legal terms 'petty assault', 'assault' and 'aggravated assault' recur frequently in the Bill. The majority of violence in close relationships is described as mild physical violence, which from the perspective of criminal law corresponds to petty assault.
The aim in the legal reform was to incorporate emotional and psychological violence, particularly controlling behaviour and abuse of dependants, into the dynamics of physical violence. The reasoning was that physical and mental abuse in close relationships tends to involve control and subordination that clouds the judgement of victims or frightens them. The perpetrator could demand that the victim drop the charges, for example, and this specific characteristic of violence in close relationships hinders its disclosure and detection, thereby helping to keep it private.
[LT] Some provisions in the Criminal Code prescribe the abuse of confidential and dependent relationships as punishable as either a common offence or an aggravated offence (p. 7).
Sexual violence is mentioned only briefly, usually referring to child abuse. The lack of description of sexual violence in the Bill is understandable given its focus on changing the law on assault, whereas sex crimes are regulated in a separate section of the Criminal Code.
As described in the section on (the lack of) 'The Directionality of Violence', distinctive types and forms of violence dominate the police freelist responses. There are almost as many accounts of emotional or psychological violence as of physical violence and more detailed accounts of the former: one or two terms are used to describe physical acts, most often as 'assaults' or other terms aligned with the legislation.
[PF24] Hurting another family member mentally or physically [PF08] Emotional (e.g. scolding, belittling, subjugation in various forms, exerting pressure, morbid jealousy), physical violence, in which the parties are members of the same family, or close relatives A notable although not dominant feature in the police freelists is that five responses include the consequences of violence as part of describing it as a simple act or in different forms.
[PF04] Physical or psychological violence, in other words causing pain and suffering to another, and in this case to someone living close to the perpetrator or to a close relative, which may further increase the suffering because of the close relationship [PF03] Victimisation,grief,fear,depression,dependency,desperation,stagnation,powerlessness,weakness In comparison with the other forms of violence, sexual violence is rarely mentioned in the police conceptualisations of violence in close relationships, which could indicate that, as in the legislation, they set sexual violence apart from other violent crimes.

The Visibility of Gender
The Bill refers to various surveys and other relevant research focusing on VAW. It also refers to studies reporting that the majority of FV recorded by the police was perpetrated by men and directed at women, and that more than half of these cases involved currently or previously married couples or cohabitants. Violence against men, including boys, is also frequently mentioned, but a frequency analysis indicated a strong focus on VAW (24 coded sections in comparison with 13 concerning men and boys). In terms of active agency, violence perpetrated by women and mothers is mentioned almost as rarely as that perpetrated by men and fathers, which emphasises the visibility of gender in victimisation but not in perpetration.
As described in the previous section of this paper concerning the types and forms of violence, the psychological aspects of violence in close relationships are, for the most part, discussed in the Bill in terms of subordinate and dependant positions, power relations and controlling behaviour connected to the use of physical violence. These aspects of violence are described as particularly harmful and reprehensible characteristics that make it more difficult to detect such behaviour, to intervene and to end it. Power dynamics in the use of violence have usually been examined in studies on GBV focusing on men's violence against women. Nevertheless, the Bill purports to connect power relations and the abuse of power in this context to a gender-neutral conceptualisation of violence: [LT] The result is the same in physical violence perpetrated by mothers and fathers against their children (p.3).
[LT] On the individual level, subordination based on the threat of violence can also occur between same-sex partners and in other close relationships, such as that between a minor and a violent parent, or an adult child and an elderly parent. The same may apply to the threat of violence by women towards men. Therefore, there are no penal grounds for treating actions related to an intimate partnership or close relationship between the perpetrator and the victim only as crimes perpetrated by men towards women. The premises of the penal code are also gender neutral in its definition of a perpetrator and a victim in other types of offence (p.10, emphasis added).
The following examples from the legal text could be interpreted as including both the recognition of power relations and the neutralisation of gender: [LT] Common to different situations, regardless of gender, are the victim's vulnerability and at least some sort of dependence on the suspect --Violence in close relationships often involves psychological or physical control and subordination exercised by the other party regardless of gender (p. 11).
[LT] The assessment of petty assaults occurring both in close relationships and in working life have special features that on the individual level similarly concern women and men as both offenders and victims. (p. 22).
Police conceptualisations of violence in close relationships were also predominantly genderneutral. However, in contrast to the explicit gender neutrality in the legal text, the police officers were more implicit, avoiding phrases such as 'regardless of gender'. Giving examples of non-physical violence, they used terms such as 'abuse of power' and 'submission' with no link to gender. There are very few gendered references to family members, such as 'mother' and 'father'. The following is a rare example of a gendered, heteronormative description in the police perceptions: [PF81] Violence between a man and a woman, in marriage, cohabiting and partnership.

Discussion
In this study, we analysed conceptualisations of violence in close relationships in Finland as exemplified in legal texts and police freelist data. The Criminal Code and the Government Bill concerning the legal reform of the right to prosecute for petty assault provided a framework for naming and defining the phenomenon, which was set against police interpretations. The Criminal Code, among other laws, guides and defines the bedrock and boundaries of police work, particularly concerning violence in close relationships as set out in the section mentioned above. Police officers are the first responders to violence and gatekeepers of the criminal justice system, and it is therefore relevant to compare their conceptualisations with those of the law. Finland is the context of this study, but the varying terminology and the debate it raises in terms of defining violence is a broader issue that concerns many societies and languages in which violence in the private setting is recognised. The international political and scientific debate concerning the FV and GBV perspectives is a relevant theoretical background regardless of the national language and legislation.
Our study has some limitations, specifically related to the analysis of two different types of qualitative data in parallel. Legal texts with their negotiated definitions are likely to cover a wide range of situations regardless of their frequency, whereas police officers are likely to produce freelist items in order of familiarity, and the more frequent cases of violence in close relationships are likely to become more familiar in practical police work. Nevertheless, the most common situations police officers face are likely to affect how they perceive the overall definition of such violence. Freelisting has proven useful in studies on culturally coherent groups. The terminology prompting the freelist question on the definition of violence in close relationships was derived from the legal text so that we could analyse these two data sets in parallel with theory-driven directed content analysis. Our application of the freelisting method produced a saturated list of conceptions of relationships and violent acts among the police officers in our sample. It would be useful to examine the same data using other methods and to conduct similar analyses on other freelist data to test for reliability.
Although the legal text and police conceptions analysed in this paper do not straightforwardly represent either of the two theoretical perspectives summarised here as FV and GBV, the key features of both facilitate the analysis of 'close relationships' and 'violence', specifically in relation to the meanings of gender in phenomena involving violence. Other features guiding our analysis included the naming and directionality of violence. In total, we identified more nuances in both the legal text and the police responses that resemble the FV perspective on violence. This is an understandable result given its prominent position and the seemingly gender-neutral vocabulary in Finnish society (Hearn and McKie 2010;Virkki 2017). In this sense, there may be more similarities in than differences between the law and law enforcement in conceptualising violence in close relationships. However, even discrepancies that may first seem rather subtle are notable from the perspective of criminal procedures in recognising and dealing with a complex issue that is largely a hidden crime.
Given the focus on family in the police conceptualisations, the effect of the FV terminology used in official police instructions and information systems seems evident in the naming of violence. This emphasises the weight of organisational instructions in comparison with legislation as normative control. Both data predominantly describe passive acts and behaviour among certain people or in a relationship without an active agent or direction of the violence. Directionality is most evident in the Bill in sections emphasising the vulnerable status of children. Children are depicted as victims in two ways: parent-child relationships are listed as close, and all petty assaults against minors became subject to public prosecution in the same Government Bill and legal reform as did petty assault in close relationships. The special attention given to children in the legal reform, regardless of their gender, could be seen as conforming to the FV perspective.
Our findings imply that police officers' definitions of violence are not limited to physical forms, and even include aspects of coercive control typically considered within GBV perspectives. This is notable given earlier findings on the weight given to physical injury in police responses to interpersonal violence (e.g. Armstrong et al. 2018;Grant and Rowe 2011). However, what police officers perceive as violence in close relationships may differ from what they consider criminal. According to the Criminal Code, violence in close relationships is explicitly connected to assault crimes and does not cover acts that would be categorised as other than physically harmful, such as defamation and illegal threats. Petty assault does not necessarily entail direct physical violence, but the Bill recognises emotional and psychological violence only in connection to physical violence, as abuse of dependence and power relations.
The focus on family relationships and the emphasis on children rather than solely on intimate partnerships indicate that both the police conceptions and the legal text are, in this sense, closer to FV than GBV perspective. The legal text broadens the scope of a close relationship beyond family relations, whereas police conceptions are notably focused on the nuclear family. There are also more distinct differences. For example, former spouses and partners are mentioned so rarely in the police responses that one might assume that the police perceive the ending of marriage or cohabiting as an end to a close relationship, whereas former spouses and former partners are explicitly mentioned in the definition of a close relationship in the legal text. This calls for more detailed examination of whether marriage and cohabiting affect the likelihood of recording assault crimes in Finland.
Gender has been the focus of studies concerning violence in close relationships from the GBV and VAW perspectives. FV research has placed less emphasis on gender and power relations as causes of violence, focusing more strongly on other social disadvantages and conflicts in the family. According to our data, legislation aims at both gender neutrality and inclusivity, and gives several indicators of the endorsed theoretical perspectives. The legal text cites research on VAW and highlights power dynamics in close relationships on the one hand, but on the other hand it projects an explicitly gender-neutral conceptualisation of violence that clearly distances it from the GBV perspective. Power relations, the abuse of power and dependant relationships in general are of concern in the Bill, regardless of gender. This resonates with prior findings on conceptualisations of gender neutrality in Finnish society (Hearn and McKie 2010;Virkki 2017) and is in line with the national implementation of the Istanbul Convention (Ministry of Social Affairs and Health 2017), in which male victims of DV are also emphasised. The police conceptions represent more implicit gender neutrality, whereby gender and gendered terms are not visible.

Conclusions
The subtle discrepancies in how police officers and the legal text conceptualise violence in close relationships may be indicative of how such cases are treated by the criminal justice system. The legal reform aimed at increasing the number of cases reported to the police, but the increase has been mainly in FV call-out tasks rather than in the number of recorded crimes. Our results indicate that police conceptions are not entirely consistent with the ideas presented in the Bill, and the key feature may be the definition of a close relationship. Researchers analysing crime-record data on repeated assaults and the categorisation of FV crimes by the police in Finland following the legal reform have found that, in terms of law-drafting and the aims in changing the right to prosecute, the main problems regarding police actions include the illogical use of the FV crime-record categorisation, and the lack of instructions on how the code relates to the legal change and the concept of a 'close relationship' (Kotanen and Smolej 2014).
We conclude from our results, which are in line with those of the aforementioned authors, that legal changes and policies aimed at affecting frontline responders' work should be accompanied with a careful assessment of their implications related to field-specific guidelines, training, resources and other prerequisites for successful change in everyday work practices. Legal reforms derive partly and justifiably from research, statistics and other sources of information, and in turn, the chosen theoretical perspectives behind these sources of information define the phenomenon that the law aims to control. Given that the aim of the legal reform in Finland was to increase the detection of violence in close relationships and to consolidate official social control of this type of crime, differences in the conceptualisations should be taken into consideration. These discrepancies are potentially meaningful in any assessment of the implementation of the law, the effects of the legal reform on the detection of crime and the proceeding of cases in the criminal justice system.