Linguistic, psychological and epistemic vulnerability in asylum procedures: An interdisciplinary approach

This article analyzes three video-recorded asylum interviews, their written records and the corresponding decisions by the Finnish Immigration Service. The goal is to identify the causes and consequences of vulnerability in instances that are particularly important when assessing whether the asylum seeker has a well-grounded fear of persecution. A combination of linguistic, psychological and epistemic perspectives on vulnerability shows that these three dimensions are closely intertwined in asylum interviews. Linguistic vulnerability is linked for the most part to interpreting, whereas psychological vulnerability stems from the difficulty in recounting traumatic experiences. Both linguistic and psychological vulnerabilities are central forces that also lead to epistemic vulnerability. Epistemic vulnerability, we claim, gives rise to certain practices within the asylum procedure, which again represents the materialization of the discourses of reporting, truth and credibility.


Introduction: Vulnerability, data and methods
In this article, we mobilize the concept of vulnerability in order to analyze the dynamics of the asylum determination procedure. The concept originates from the Latin verb vulnerare ('to wound'). Hence, vulnerability often has a negative connotation as it refers to the possibility of being either physically or emotionally harmed (Snyman, 2015: 280). Vulnerability is widely discussed in the context of international protection regimes, where it generally refers to a group or individual in need of special support or protection because of age, disability or risk of abuse.
In this article, vulnerability is regarded as a shared condition intertwined with the institutional asylum determination process and situationally affecting the dynamics of asylum interviews rather than representing a group characteristic or an individual quality (also of hierarchies of knowledge and power, and an institutional disposition towards knowledge (see also Heritage, 2012). This disposition affects what kind of information becomes valued as knowledge. In other studies, the disposition has been called an epistemology of ignorance (Bohmer and Shuman, 2007) and a culture of disbelief (Jubany, 2011), where everything that the applicant discloses is suspect and certain elements in the account are given knowledge value while others are omitted (Kynsilehto and Puumala, 2015). Epistemological vulnerability within the asylum procedure stems from the difficulty in verifying subjective accounts. In this article, the analysis of epistemic vulnerability focuses on the production and selection of valuable information throughout the asylum procedure. As an inherent quality of asylum procedures, epistemic vulnerability is manifested through the discourses of truth, reporting and credibility, which denote the ways in which the practice of power has become intertwined with what is regarded as knowledge (Foucault, 1980). These discourses materialize through specific ways of posing questions, giving testimony, framing, claiming and evaluating knowledge.
By combining these three forms of vulnerability and reflecting on their connection to discourses operational throughout the procedure, the article aims at producing a multifaceted picture of the complex institutional, professional and personal constraints in asylum interviews. We argue that understanding these constraints may facilitate the identification of phenomena and practices that are problematic, and help to design better practices for asylum interviews and the asylum determination process as a whole. The cases analyzed in this article are based on three asylum interviews conducted in Finnish by the agents of the Finnish Immigration Service in the agency's premises in 2014 and 2015. The asylum applicants included a Georgian male speaking Georgian, an Iraqi female speaking Arabic and a Russian female speaking Russian. The Georgian male was assisted by a male counsel, and the other two by a female counsel. As the interviews were conducted in Finnish, an interpreter speaking the language used by the applicant as their first language and Finnish as their second language was present. The Arabic interpreter was male, whereas the other two interpreters were female. All interviews were conducted by a female immigration agent; the Georgian applicant was interviewed for 4 hours 44 minutes, the Iraqi applicant for 3 hours 30 minutes and the Russian applicant for 5 hours 10 minutes. The written record was sight-translated orally for verification at the end of the interview except in the case of the Russian applicant, for whom a second meeting was organized to this end due to time constraints.
The interview data were obtained by recording the interviews with two video cameras in order to gain an optimal view of all participants' actions. Selected passages from the tapes were transcribed and translated into English. The data also include the written records, produced by the agent conducting the interview and constituting the basis on which each case is assessed, as well as the decisions made by another agent. Relevant passages from these texts in Finnish were translated into English. Thus, the translations follow the idiosyncratic features of the source texts as closely as possible. These features originate mostly from the interpreters' idiolectal characteristics, such as atypical verb tense usage or constituent order. Research permissions were granted institutionally by the Finnish Immigration Service and individually by all interview participants. In addition, the research had been ethically pre-approved by the University of Tampere ethical committee.
The passages to be analyzed in detail were chosen on the basis of their specific importance when evaluating whether the applicant was suffering from a well-founded fear of being persecuted. Hence, the passages were selected from sections in which the applicants were requested to explain in their own words the reasons why they were applying for international protection.
We will first analyze passages related to personal threat, fear of persecution and grounds for international protection in the written decisions. Second, we will examine the corresponding passages in the written interview records in order to show how the decisions produce, assess and use knowledge based on the interview with the asylum seeker. Subsequently, we will analyze excerpts from the transcriptions of all three cases in separate sections, concentrating on the most salient phenomena characterizing each: proactive interpreting and communication of affect, linguistic and psychological alignment between interpreter and asylum seeker, and the possibility of authentic narration and certainty. To conclude, we will link the findings to the discourses of reporting, credibility and truth that are detectable both in the interviews themselves and the written records and decisions into which the interviews are entextualized.

Identifying valid sources of knowledge within the asylum procedure
Our analysis of the decisions and written records begins with an example that concerns a Georgian man who was seeking asylum with his family on the grounds of harassment and misconduct by a former official. The applicant's property had been seized by the authorities, and after he had filed a complaint, an official began to harass and threaten him and his family. According to the decision, the grounds for asylum were insufficient in this case: 1 (1) You have said that you fear that the former police commander will threaten you in your home country. The Immigration Service has considered that you have a possibility to seek protection from the officials in your home country. Your fear is not objectively founded. [. . .] Since you do not suffer from a well-founded fear of persecution, and you are not in a real danger of suffering severe damage or meet the criteria for humanitarian protection, Art. 88-e Aliens Act, regarding subsidiary protection, does not apply in your case.
In the second case, an Arabic-speaking Sunni woman from Iraq was seeking asylum on account of domestic violence spurred by her inter-religious marriage to a Shi'ia. The decision concluded that the applicant was suffering from a well-founded fear of persecution, and her account was accepted as a fact in light of Country of Origin Information: (2) The Immigration Service finds in your favor that although you have previously said that your health deteriorated first due to domestic violence and then as a result of an accident, your account of an accident can be understood to be related to the expectations and roles that are typical in Iraqi society. The Immigration Service accepts your account of the violence targeted at you and the death threat made by your husband. [. . .] You are afraid that your life is in jeopardy in [name of city] due to your mixed marriage. Above, the Immigration Service has accepted your account regarding the threat you faced and the fact that you cannot receive protection from the officials in your home region. Your fear is objectively founded. [. . .] All criteria for international protection are met in your case. The Immigration Service considers that you have a well-founded fear of persecution in your home country, as specified in Art. 87-1 Aliens Act, because of your religion.
In the third case, a Russian applicant stated that she had suffered harassment based on her sexual orientation both at work and in her private life. Her account was accepted on the following grounds: (3) The evidence you have given supports your narration. [. . .] The Immigration Service accepts as a fact that you are a lesbian by sexual orientation and that you have lived in a convivial relationship. [. . .] The Immigration Service has considered, in light of Country of Origin Information, that as a person belonging to a sexual minority, you are in danger of suffering violations of justice in your home country. The Immigration Service considers your fear to be objectively founded. [. . .] All criteria set in the Aliens Act are met in your case. The Immigration Service considers that you have a well-founded fear of persecution in your home country, as specified in Art. 87-1 Aliens Act, due to your sexual orientation, that is, due to belonging to a social group.
These excerpts exemplify the variety of sources of knowledge in asylum procedures. The applicant's verbal account as a key element of the assessment is clearly emphasized. Thus, evidence value is given to the applicants' personal narration, presented in indirect speech (you have said) and narrative reports of their speech acts (based on your account), as well as pieces of evidence produced by the asylum seeker (the evidence you have given). In addition, intertextual links to information originating from official, objective sources (Country of Origin Information and references to the Aliens Act) are used to justify the decisions and render the assessment objective and legitimate.
From an administrative viewpoint, the applicant's role is to give testimony and evidence. The complexity of the evidential assessment, involving not only the applicant's verbal account, becomes clear when analyzing the written records of the interviews. For example, the Georgian applicant's account includes second-hand information in reported speech: (4) In [year] I made a complaint to the prosecutor's office to retrieve my property, and in [year] the same police commander and two of his friends rushed into my home. I was not at home at the time, I was at work, and my wife and two children were at home. I know, as my wife told me, that someone knocked, and my wife opened the door. They requested me, they asked, where is your husband? My wife said she does not know. Then she asked, who are you. My wife had a baby in her hand, [name], in her arms. They came to an argument, they wanted to tear away the child and there was verbal abuse. She was hit a little on the neck and she went indoors. My wife was scared and started to scream for help, the neighbors came and went inside, what is happening here.

Note:
The applicant is crying, wants to take a break. This narrative, typed on the basis of the interpreter's rendering of the applicant's story, contains several items present in a prototypical narrative (Labov and Waletzky, 1967): an orientation sequence (the first two sentences), complicated action (the bulk of the story) and a resolution marking the end of the aggression (last sentence). The indirect nature of the story is clearly stated (I know, as my wife told me), and the resolution marks the end of the story explicitly (The neighbors came and went inside, asking what's going on here.). Importantly, the physical aggression is played down in the record (she was hit a little on the neck). While the narrative contains a description of the wife's emotional reactions (my wife was scared and started to scream for help), a verbal account of the applicant's own emotions is lacking. Regarding the sources of information, the account is therefore entirely indirect. In addition, the applicant had not produced any documentation about the confiscation of his property, nor about the claim he had made to retrieve it, which had a detrimental effect on the credibility of his story (Bohmer and Shuman, 2007).
While the Georgian applicant started to cry as he was telling his story, the second applicant was already crying as she approached one of the most important points in her story: (5) Note: Emotional reaction: the applicant is crying We were standing in our yard. Our neighborhood has a Shi'ite leader that everyone hates. He has turned in Sunnis who have been killed afterwards. The youth have run up and told that [name], this Shi'ite leader, has been killed in his own shop. I felt so good that I took a deep breath and said that we have gotten rid of him. Immediately after I had said this my husband has hit me in the face and I have taken a fierce fall backwards. My head hit the edge of the door. He said to me, did it make you happy because he is a Shi'a and he has been killed. After I fell against the edge of the door, he has grabbed me by my hair and tried to drag me inside, so that the others would not hear my screaming. He has said that he will kill me in order to get rid of me. I was a little awake when he dragged me inside. I was frightened because he has a gun. I was afraid that he would hit me or kill me. My children have seen that I moved only a little as he was dragging me. I heard my children say to my husband that dad, mom is dead. Then I have passed out and I woke up in the hospital after having been unconscious for three days. There had been so many hard blows that one cervical vertebra had moved and broken. After that, my memory has worsened and I felt that something was going on in my head.
Compared to example (4), this applicant's story is more direct, as it explains the verbal and physical abuse she survived, the feelings she had during the situation, her own verbal reaction prior to the battery and the verbal accounts of the neighbours, her husband and her children, in both direct and indirect speech. In addition, the consequences of the battery are clearly explained, and are concordant with the medical evidence she presented. Overall, the written record depicts a complex narrative including an orientation sequence (the first three sentences), complicated action (the bulk of the story) and an end sequence that can be interpreted as consisting of a resolution and a coda (the last three sentences). In this narrative, the applicant is represented as a passive, helpless object of battery: her only concrete actions are taking a deep breath, verbalizing her relief after the murder of the neighbourhood leader and her screaming. Her body is the object of battery and her mind merely registers her fear and her husband's and children's speech around her (see Barry et al., 2018 on the nature of traumatic memories). Importantly, her inability to move, the emotion she experiences and her registering the speech are embedded in the description of the aggression suffered by her body, which produces a vivid image of a particularly violent and terrifying incident. In addition, compared to the previous example, the concluding sequences are linked to the substance of the narrative and provide information about the tragic consequences.
While the accounts given in examples (4) and (5) are based on the applicants' mentally stored recollections of events, the Russian applicant relies on her written notes while telling her story: (6) Note: The counsel asks what papers the applicant is leafing through.
Sometime before I have worked for an/the 2 LGBT organization. It is an international lesbian and gay organization. This is written by their lawyer, we asked for their help when we left. In this paper, everything is in order.
On tape, one can see that the applicant uses the paper particularly to check the exact dates and chronology of the events. In contrast to the previous two examples, there are neither sidetracks from the main events of the applicant's story, nor accounts of the applicant's emotional state during the interview or during the events that she describes. Instead, she appears to be well aware of the focus and goal of the interview and has prepared her case in advance in order to provide a coherent, accurate and detailed description, not only of what happened but also the places, dates and people involved. The following excerpt from her interview record illustrates the detailed nature of her account: (7) The first time I had problems was in the summer of [year]. My girlfriend, who has arrived in Finland with me, moved to [a city]. On [exact date] we met and wanted to have breakfast at a friend's place. The name of my female friend is [name]. She has lived near my home and lived in an apartment in which there are other people. In the same apartment in which [name] has lived, two other women have lived, I don't know whether they are mother and daughter or two sisters. I know that she has had problems with her neighbors. Subsequently, the applicant gave an account of the abuse her friend had experienced from her housemates, and the abuse that the applicant had experienced on that particular day in the apartment.
Structurally, this narrative is different from examples (4) and (5): most sequences orienting the narrative are embedded in sequences of complicated action instead of being presented at the beginning of the story. Yet resorting to a written document instead of free narration may be problematic, as the asylum agent can take it as a sign of fabrication or the applicant using a 'stock story' instead of narrating a personally experienced event (Kynsilehto and Puumala, 2015). This is the reason why her counsel intervenes, and the agent includes a note of this intervention in the written record (example 6). This intervention shows that there is a contradiction between the requirement of a coherent and detailed narration and the requirement of free narration: it is easier to provide dates, names and places with written notes, whereas these are easily omitted in free narrations (also Bögner et al., 2010).
These examples depict how the conception of knowledge and the process of its valuation become visible through the written records and asylum decisions. The three quite different written narratives already indicate epistemic vulnerability; the practical impossibility of knowing the truth based on verbal accounts and the institutional disposition towards knowledge as objective and neutral. In the following sections, we will focus on salient phenomena related to both linguistic and psychological vulnerability through an analysis of transcribed and translated excerpts from video-recorded interviews, corresponding to the written records and decisions analyzed above. The goal is to illustrate how both linguistic and psychological vulnerability exacerbate epistemic vulnerability and vice versa. The following symbols are used in the transcriptions: pause shorter than 0.2 microseconds (9.0) duration of a longer pause in seconds ( (drinks)

Communicating affect and proactive interpreting
The communication of affect through a proactive interpreter emerges as a key feature of the Georgian applicant's case. The excerpt from the interview transcripts starts towards the end of example (4) from the written records. Unfortunately, the interpreter's voice is audible only when she speaks towards the microphone. Due to the applicant's soft voice, the position of the microphone, poor acoustics and the heavy sound of the agent's typing, which dominates the soundscape in the room, the asylum seeker's speech is not audible either. Therefore, the analysis is based only on the interpreter's and the immigration agent's utterances in Finnish, as well as all participants' non-verbal communication.
While the analysis would have been richer if verbal interaction in Georgian had been audible as well, the analysis of non-verbal interaction and communication in Finnish is sufficient to support our arguments: The written record (see example 4) stated that the applicant wanted to take a break because he was crying (The applicant is crying, wants to take a break.). However, as we can see in the interview transcript, this will was in fact verbalized by the interpreter (wants [.] he [.] go out, line 29). Previously (line 23) the interpreter had verbalized the applicant's crying (he starts [discourse marker] to cry); this line may also contain the translation of the applicant's inaudible words in line 21. The agent had acknowledged his crying with the minimal response particle mm-m? in line 22. It is possible that the interpreter's choice of verbalizing non-verbal communication is prompted by her awareness of the communicative goal of the interaction, namely the production of a written record (c.f. Bolden, 2000;Defrancq and Verfliede, 2018: 230). Thus, the interpreter probably knows from experience that descriptions of the applicants' reactions are often included in the record. Indeed, the agent concentrates on listening to the interpreter's rendition of the asylum seeker's speech and typing the written record: her gaze is directed towards the computer screen, and she glances at the asylum seeker only when he is about to leave the room. In lines 25 and 32, she responds to the interpreter's renditions with the minimal response particles mm-m, mmm-m and joo-o, acknowledging that she has heard what the interpreter has said. As the agent is typing the record, she cannot pay attention to non-verbal communication and take note of the applicant's behaviour and his affective reactions, even though the European Asylum Support Office states in its practical guidelines on personal interviews that these should be taken into account (EASO, 2014). The counsel does not look at the applicant either: her gaze is directed towards the papers in her lap until the applicant stands up. As a result, most verbal and non-verbal interaction takes place between the interpreter and the asylum seeker in a dyadic manner (Pasquandrea, 2012: 133), although the normative form of dialogue interpreting is a triad in which the interpreter functions as a liaison between the two primary participants: the asylum seeker and the government agent (the counsel rarely intervenes, and her direct agency is strongly limited by the law). Hence, there are at least two factors rendering the situation vulnerable in linguistic and interactional terms: the agent does not acknowledge non-verbal communication, and the interpreter not only translates words and actions but provides a modal evaluation of the consequences of the asylum seeker's affective state as well (wants [.] he [.] go out, line 29). At the same time, a comparison between the transcript and the written record indicates that the idiosyncratic and sometimes ungrammatical expressions in the interpreter's Finnish do not appear to have had an adverse effect on the case.
The articulated wish to have a break, as stated in the written record (wants to take a break, example 4), constructs the applicant as an autonomous subject, in control of the situation (c.f. Foucault, 1982). This representation of the applicant's psychological state is completely different from the emotionally overwhelmed subject that can be observed in the video and in the transcript. In addition to the linguistic perspective, this excerpt is also problematic in terms of psychological vulnerability, namely in relation to building rapport and communicating memories and experiences. In what follows, we will explain and show from the neuroceptive perspective (Porges, 2007) the potential reasons why the applicant has difficulty speaking, starts crying and wants to leave the room.
According to the cognitive and interpersonal neurosciences, coherent memorizing and communicating depends on the successful integration of complex mental processes, resulting in flexible and productive adaptation to the needs of the self and the situation (Siegel, 2001). When integration fails, disintegrated states prevail. This may happen, for example, when interoceptive and/or perceptive information are impossible to integrate due to their ambiguous, novel, fear-provoking, or excessive nature. In this excerpt, the applicant's self-regulation of affect and action aimed at communication fail when describing a threatening situation involving his family. He cannot express himself further and starts crying.
As unregulated mental processes represent the risk of the re-emergence of vivid trauma memories, traumatized people tend to avoid them (Ehlers and Clark, 2000). In addition, the absence of active communication easily triggers the fear of violent or inhuman conduct (e.g. Dalenberg, 2004). In the excerpt, the applicant's disintegration cooccurs with the agent's and the counsel's avoidant interaction. Problematically, this lack of co-regulation by the other main participants leaves the applicant alone with his inner state, and vulnerable to fear-provoking interpretations of their intentions. According to Porges's (2007) theory of neuroception, in such a situation, automatic processes detecting situational threat cues are activated, and the autonomous nervous system becomes aroused and prepares the person for fight or flight instead of co-operation. Indeed, the excerpt reveals the physical reaction of leaving the room.
Defensive states disturb both co-operation and cognitive processes, whereas the creation of a supportive and trustful atmosphere enhances pro-social behaviour and memory activation (Hietanen, 2018;Porges, 2007). Gaze, tone, prosody and posture are central elements when detecting the safety of a situation (Porges, 2007). In this excerpt, while the interpreter encourages the applicant to continue his narration even though he is crying, the agent does not establish verbal or non-verbal contact with him. The asylum applicant glances at the agent while crying (line 21), possibly in order to get reassurance. However, the agent's gaze continues to be directed towards the computer screen. As a result, instead of being reassured, the applicant is left with uncertainty regarding the agent's thoughts and attitude towards him. After failing to make a connection with the officer, the applicant drinks water (line 24), which can be a sign of autonomic nervous system arousal, indicating elevated levels of distress and a subsequent psychobiological shift to a defensive state (Cabrera et al., 2018). Sensory stimulation (drinking) can thus be interpreted as an attempt to gain voluntary control over his inner state, and the activation of traumatic memories and/or dissociation. As his attempt to receive external support for these attempts to stabilize his state fails, the applicant leaves the room, looking at the door. As pointed out earlier, this action can be interpreted as a subtle form of the automatic flight reaction triggered by fear, terror, shame or sorrow, or emerging implicit or explicit traumatic memories.
In addition to linguistic and psychological factors producing epistemic uncertainty over what is 'true', non-verbal communication also emerges as a potential source of epistemic vulnerability in the excerpt. While the guidelines regarding asylum interviews (EASO, 2014) emphasize that attention should also be paid to non-verbal cues, the EASO guidelines for evidence assessment also clearly warn against using demeanour as a credibility indicator (EASO, 2015). The role of non-verbal communication in the decision-making process remains unclear and problematic in this case: while the agent does not pay attention to non-verbal communication during the interview, it is mentioned in the record and framed in a problematic manner in light of the video (Puumala et al., 2018). It is impossible to know how this written information is used in decision-making, particularly in credibility assessment. Using non-verbal cues when assessing the credibility of the applicant's account is highly problematic (Rogers et al., 2015). Yet we argue that taking note of applicants' behaviour during the interview is necessary in order to offer interactional support (Puumala and Kynsilehto, 2016). This would enhance the quality of disclosed information. Our analysis reveals that in assessing whether the applicant has a well-founded fear of persecution, it is important to know whether and how non-verbal expressions of affect are perceived and taken into account in the interaction, as this shapes further interaction and narration. Psychological vulnerability stemming from every participant's actions concerning affect and its non-verbal communication, embedded in the interaction, has a strong epistemic value.

Linguistic and psychological alignment between the asylum seeker and the interpreter
Occasionally, interpreters take on a supportive role during the interviews, and their position changes from being a direct interpreter to a cultural broker (see also Inghilleri, 2005). This can be seen in the second case, where a dyadic interaction between the Iraqi applicant and the interpreter is powerfully foregrounded both verbally and non-verbally: This excerpt suggests that a naturally occurring conversation is taking place between the applicant and the interpreter, as seen for example by the way they complete each other's utterances (lines 12 and 13). This kind of dyadic interaction corresponds to what Hale (2007: 41) has labelled mediated interaction, as opposed to directly interpreted interaction in a triad, which is the norm in dialogue interpreting (Pasquandrea, 2012). In dyadic, mediated interaction, the interpreter easily becomes a cultural broker and takes some of the primary speakers' responsibilities, such as summarizing the content of their speech. In this excerpt, the interpreter's role in requesting clarifications is particularly salient. Thus, in line 2, the interpreter inquires who the aggressor was, whereas in lines 5 and 7 he asks for precision regarding the body part that was affected by the blow. In lines 9 and 12, the interpreter checks whether he has understood the applicant's verbal and non-verbal messages correctly. It is only after six dyadic adjacency pairs that the interpreter provides a translation for the agent (lines 15-20), and the rendition also contains information that was mentioned earlier in the interview -the interpreter therefore parses the information for the agent. While the agent has been gazing at her computer screen throughout this passage, she now glances at the interpreter because the rendition contains a deictic expression (tällä tavalla, 'like this'), accompanied by a gesture (line 17). The interpreter's repair initiations, focusing on the agent of the action (line 2) and its target (lines 5, 7 and 10) could be regarded as demonstrating his orientation towards the communicative goal of the interaction as in example (8), discussed in the previous section. Indeed, in the written record, it is important to specify who hits and where, as this information is crucial when assessing the credibility of the story. At the same time, the interpreter's proactive style, prompted by the lack of communication on the part of the agent, exposes the applicant to linguistic vulnerability and produces a written record containing information that the applicant never disclosed (my husband has hit me in the face; see example 5). Indeed, in line 7, the interpreter asks whether the applicant's husband had hit her face, and in line 8, the applicant says that he pushed her and makes a gesture illustrating separation. However, in line 10, the interpreter states (so he hit you in the face a-). While the applicant does not deny this, she never said that her husband hit her face. In addition, what was a wall in her account (lines 6 and 13) becomes a door in the interpreter's rendering (line 20). Linguistic vulnerability therefore translates into epistemic vulnerability: strictly speaking, the written record is no longer the applicant's personal narrative, even though it is assessed as such. At the same time, the example shows that a personal narrative including all the details may not be possible without a proactive interpreter in this case. Hence, despite a strong orientation towards objective knowledge, the excerpt reveals how the discourse of credibility shapes interaction and actually produces epistemic vulnerability within the procedure.
The agent, who focuses on typing the record, illustrates her orientation towards the goal of the encounter as well. However, it is interesting that the applicant is able to regulate the affect that is clearly present in her description of a traumatizing event. This may be due to a close connection and co-creation of the narration between the interpreter and the applicant: the interpreter requests clarifications (lines 2, 5 and 7) and repeats what the applicant has said, and suggests a verbal account of the applicant's gestures (lines 10 and 12). In addition to this verbal interaction, the co-construction of the narration happens non-verbally: eye contact and body postures enable the monitoring and reciprocation of each other's movements and gazes and the organization of verbal turns. All of these features indicate mutual connectedness (Porges, 2007;Seikkula et al., 2015). The high level of connectedness with the interpreter can serve as a source of co-regulation that enables the applicant to express traumatic memories and emotions even in the absence of a connection with the agent and the counsel.
Linguistic and psychological alignment between the interpreter and the applicant therefore facilitate narration. At the same time, the interpreter's active participation in the construction of the narrative may affect the applicant's personal memory of the event, exposing the vulnerability embedded in the memorization of traumatic events. Indeed, rather than being accurate copies of reality, memories are selectively stored representations of events, encoded as patterns of activation in neural networks. Intra-and intersubjective factors present in each instance of recalling memories and sharing them socially further modify them, and people may even regard their modified and fabricated memories as true (Siegel, 2001). Changing memories and changing narrations thereof constitute a source of epistemic vulnerability in asylum determination processes, as they may have an adverse effect on the assessment of credibility.

The possibility of authentic narration?
The third case differs from the previous ones in terms of the interaction between the primary participants. Thus, the role of non-verbal communication is minimal, and while the agent focuses on typing the record in this interview as well, she shows willingness to engage in a real exchange with the applicant. Another peculiar feature of this interview is that the Russian asylum seeker herself shows a clear epistemic orientation towards the communicative goal of the encounter, namely the written record, whereas in the previous examples, this role was played by the interpreter. In this interview, the applicant's goalorientation is illustrated by the length and rhythm of her turns, the interpretation that immediately ensues, and the clear pause during which the officer writes down the account. The applicant waits for a sign to continue before returning to her narration. As shown above in example (7) from the written record, the applicant provides particularly detailed and precise accounts of the several incidents of verbal and physical abuse she experienced. However, her recollection of memories lacks vivid contextual information such as sights, sounds and tones (see Brewin, 2007 on sensory details of traumatic autobiographical memory). In the interview, the applicant has papers in front of her, and on one occasion, her counsel points out that she should explain her story in her own words rather than reading from her notes. Given the fallibility of autobiographical memory, and traumatic memory in particular (Brewin, 2007), a written memory-aid is a sensible choice from the applicant's perspective. The written account may also indicate an effort to objectify oneself when disclosing traumatic memories, a common strategy in traumatized individuals (i.e. observer recollection or semantic recollection; Siegel, 2001). Thus, a focus on semantic information may prevent the activation of implicit memory, including sensations and feelings. The refusal to control the memorizing process renders the applicant vulnerable to dysregulation. At the same time, the counsel's intervention is understandable from an epistemic point of view in light of the discourse of truth, as using written materials undermines the applicant's credibility and -contradictorily -the authenticity of her narration.
From the point of view of knowledge transfer, accurate interpretation and entextualization, the communication pattern is ideal: clearly formulated, relatively short sentences are read from the paper and translated one by one. Hence, the interpreter has time to concentrate on listening and rendering accurate translations, and the agent is able to type almost everything the interpreter says. However, misunderstandings that are due to the interpreter's lack of knowledge of the LGBT subculture emerge as a major source of linguistic vulnerability, also giving rise to epistemic vulnerability as regards the decision-making process. The following excerpt from the transcription directly follows the excerpt from the written record reproduced in example (6) In line 1, the applicant mentions Выход (vykhod, '[coming/going] out'), a Russian LGBT organization. In line 3, the interpreter emits a hesitation sound and is about to produce her rendition, but the applicant continues with what appears to be a lapsus in line 4: while showing the contact information of the organization, she says going/coming out of this organization. As a result, in line 5, the interpreter produces a repair initiation. Instead of repeating the name of the organization, the applicant (line 6) explains that it is an LGBT organization. In line 7, the interpreter makes another repair initiation. Following the applicant's repair (line 8), the interpreter checks once again whether lawyers are involved (line 11). The interpreter's rendition in Finnish (lines 13 and 14) shows that both Vykhod and the acronym LGBT are unfamiliar to her, and she appears to think that LGBT is the acronym for a particular organization. Ultimately, the interpreter's confusion and failure to render the organization's name in Russian leads to the counsel's intervention. The fact that the applicant is using both English and Russian names during the interview further increases the confusion of the interpreter. In the end, both Russian (Vykhod) and English (Coming out) names are included in the written record. Another linguistic and epistemic issue is present in the passage in which the applicant's sexual identity is discussed: In the excerpt, the applicant produces a nuanced reflection of how she would like to identify herself. The agent (turn 1) uses the Finnish terms lesbo and homo, the latter encompassing both female and male homosexuals. Both terms are seemingly neutral and are the preferred terms in the Finnish LGBT community. At the same time, like most words used to describe persons belonging to minorities, these words can be derogatory in certain contexts. The translation of the indexical meanings of such terms is impossible in fact, as a literary translation may change the connotative meaning of the word (Määttä, 2016). In her rendition (line 5), the interpreter uses the terms лесбиянка (lesbiyanka, 'lesbian'), used by the applicant in example (10) and гомосексуалист (gomoseksualist, 'homosexual'). The applicant (line 6) clearly rejects the word lesbiyanka and elaborates that she would perhaps prefer гей (gey, 'gay', lines 6 and 7). Hence, the applicant connects herself to a wider, transnational sexual community and shows awareness of the connotations and different contextual meanings of the terms. In Russian, as in many other languages, the English word gay has replaced negatively loaded, older terms, which triggers another translation problem: while the English loanword gay is widely used in Finnish as well, it is rarely used as a self-identification term. Therefore, it is unclear whether translating the Russian word гей by using the Finnish word gay would be a translation or a quotation from the original Russian speech. Based on the interpreter's general lack of knowledge of the LGBT subculture, it is also possible that she is not aware of the international and Finnish usage of the word gay. This hypothesis is supported by the fact that in the interpreter's speech, the Finnish word homo stands for both the Russian word гомосексуалист and гей. In her rendition, the interpreter (line 10) chooses the Finnish word homo, therefore providing an answer that repeats the words mentioned by the agent in her original question, namely homo and lesbo. Hence, while the agent's question emphasized the applicant's own preferences, these preferences are ultimately disregarded because of several factors: the words mentioned in the agent's question, lexical differences between Finnish and Russian, and the choices made by the interpreter. Subsequently, the word homo is transcribed in the written record as the preferred term used by the applicant. Taking into account the applicant's own preferences would have required a linguistic explanation and accrued awareness of the affective values attached to different lexical items in both the Finnish and the Russian gay subcultures.
The agent's question is not intended to show empathy or sensitivity towards the applicant's preferences: it is used to prove whether her claim of belonging to a sexual minority has accrued truth value. This interpretation is further supported by the fact that in the decision (example 3) the applicant is consistently referred to as a lesbian and her lesbianism as being the reason for her persecution, although the corresponding Russian term was explicitly rejected by the applicant. The seemingly emphatic question was meant to provide evidence for the assessment of the applicant's sexual identity, which was crucial for determining the credibility of the claim. While membership in a particular social group (in this case a sexual minority) is among the criteria that entitle one to international protection in accordance with the 1951 Geneva Convention, a claim is not enough. In line with the discourses of truth and credibility, the claim needs to be assessed and authenticity determined during the asylum procedure.

Concluding remarks
In this article, we have analyzed examples of three asylum decisions and the written records on which they are based, as well as transcriptions of video recordings of these interviews in order to scrutinize the manifestations of linguistic, psychological and epistemic vulnerability in the asylum determination process.
From the point of view of language and interaction, the interpreter emerges as a key figure that not only translates but also mediates non-verbal communication and cultural messages, and coordinates turn organization (see Wadensjö, 1998) in a situation in which the other participants focus on reading or taking notes. In addition to mediating linguistic vulnerability within the procedure, interpretation constitutes a site of linguistic vulnerability, as exemplified by the additions and changes induced by the interpreter. These, as well as idiosyncratic features in the interpreter's language use, are typically inscribed into the written record forming the basis for decision-making.
Regarding psychological factors, the asylum interview as a whole is characterized by accrued vulnerability. Traumatic events are often stored in a fragmented, sensory mode rather than in a semantic and verbally communicable form (Brewin, 2007). In addition, giving an account of a traumatic event is difficult because of the fear of overwhelming emotions, which results in the suppression of details and difficulty in finding the words to convey the meaning of the event (Herlihy et al., 2012;Sorsoli, 2010). The relational and interactional factors of the situation are extremely important for a successful account of a traumatic event (Siegel, 2001;Sorsoli, 2010), as all psychological processes, including memory, communication and regulation of inner states are interpersonal phenomena (Porges, 2007;Siegel, 2001). As we have shown, the asylum interview does not constitute a safe environment for accounts of traumatic events, and the interpreter is the only person engaging in real communication with the asylum seeker. Problematically, the applicant may interpret the immigration agent's emphasis on neutrality and typing as negative feedback, increasing their cognitive and affective load and making it more difficult to provide a coherent narrative (Herlihy and Turner, 2009). When the mediation of this effect is delegated to the interpreter, the immigration agent cannot access the dynamics of knowledge production.
Yet both linguistic and psychological vulnerability are omitted from the decisions, where the discourses of truth and credibility predominate. This becomes visible through formulations such as accepts as a fact and objectively founded, which recur in the decision text, as well as explicit evaluations of the truthfulness of the applicant's narrative and other elements of evidence. Nonetheless, knowledge within asylum procedures is ontologically vulnerable, as shown in our analysis. The discourse of reporting, namely the overwhelming importance of the written record that orients all interaction during the asylum interview (Määttä, 2015), can be regarded as the main articulation connecting linguistic and psychological vulnerability with epistemic vulnerability. In fact, there is an oscillating movement between the discourses of truth, reporting and credibility that gives rise to certain practices of posing questions and conceptions of what is valuable information during the interviews. Contradictorily, these same practices can produce epistemic vulnerability as regards decision-making, which again undermines the quest for truth and objective knowledge that lies at the heart of the discourses. Denial of the vulnerabilities embedded in the asylum procedure does not make them disappear. There is a political and practical need to understand if -and how -these vulnerabilities may lead to erroneous decisions or distortions in the assessment of the claims. Such thorough and detailed scrutiny of the decision-making process would improve the quality of the asylum process and enhance transparent decision-making by addressing potential discriminatory structures within the procedure.

Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/ or publication of this article: The data were collected as a part of Eeva Puumala's postdoctoral project funded by the Academy of Finland (#266009). Notes 1. Italics are by the authors in all examples. 2. As Finnish does not have articles, it is not possible to know whether the text refers to a particular organization. A similar translation problem appears in line 20 of example 9.