The Myth of an ‘ Invisible Mediator ’ : An Australian Case Study of English-Japanese Police Interpreting

Introduction: Interpreter roles in legal contexts This article explores the roles and management of discourse by police interpreters. The traditional view of the legal interpreter as a ‘conduit’ (Russell 2000: 36-37; Laster & Taylor 1994: 112; Yoshida 2007: 20) suggests that interpreting processes are not relevant as evidence in court. It has also been suggested that people who use interpreters tend to expect them to be invisible (Laster & Taylor 1994; Wadensjö 1998; Roy 2000; Leung & Gibbons 2008). Thus, while police records of interviews are important evidence in court, when interviews are mediated by an interpreter there are few traces of the interpreting process in the evidence that appears in court. Similarly, as has been demonstrated by existing studies (Berk-Seligson 1990; Eades 1996; Hale 2004), court transcripts do not show the ways in which aspects of interpreting processes may have had impacts on the outcomes of the cases.

2001: 13).In other words, interpreters should not let their personal opinions influence their performance.
However, it has been demonstrated that legal interpreters do play roles other than that of 'conduit,' while engaging in various types of discourse management (Angermayer 2005;Berk-Seligson 2002;Leung & Gibbons 2008;Russell 2000Russell , 2002;;Wadensjö 1998).Such role shifts may affect the outcome of interactions between the police investigator and the interviewee (Berk-Seligson 2002;Russell 2000Russell , 2002;;Wadensjö 1998).The present study explores this issue of interpreter roles, through a focus on the problem-solving strategies of Japanese-English interpreters when mediating interviews conducted by the Australian Federal Police.This focus has been chosen because 'role shifts' become salient in the data set when problem solving is required or a problem is anticipated.Specifically, the article addresses four questions.First, can interpreters maintain their invisibility when a communication problem is perceived?Second, what roles do interpreters take in handling communication problems?Third, how and why do interpreters' role shifts occur?And fourth, what are the practical and theoretical implications of the interpreters' handling of problems?

The study
The data used for the study analysed in this essay are Australian Federal Police recordings of four interviews with suspects who were Japanese citizens.Three interviews come from a drug-smuggling case that took place in Melbourne.All the suspects in this case were found guilty and deported to Japan. 1 The remaining interview in the data set is also from a drug case, but it took place in Sydney and the suspect was acquitted after a trial.Three of the interpreters (referred to as Interpreter 1, 2 and 4) were speakers of Japanese as a first language, and one, (Interpreter 3), was a speaker of English as a first language.Interpreter 1 had Professional Interpreter accreditation from the National Accreditation Authority for Translators and Interpreters (NAATI), 2 and the other three interpreters had Paraprofessional Interpreter accreditation in Japanese-English interpreting.It has been claimed that the minimum level of accreditation for 1 One of the suspects is Chika Honda, whose story associated with this case was reported extensively in the media in Japan and Australia (Susuki 2000;Unde 2000;Hyland 2008). 2 According to NAATI, Professional Interpreters 'are capable of interpreting across a wide range of subjects involving dialogues at specialist consultations,' while Paraprofessional Interpreters 'generally undertake the interpretation of non-specialist dialogues' (NAATI n.d.).

Nakane
The Myth of an 'Invisible Mediator' PORTAL, vol. 6, no. 1, January 2009. 3 legal interpreting should be Professional, even though this does not necessarily guarantee minimum quality due to the specialised nature of legal language (Laster & Taylor 1994: 30;Hale 2004: 25-27).Thus, three of the interpreters in the present study could be said to lack sufficient qualifications for legal interpreting.However, the present study is concerned with what actually occurs in the legal process, not in ideal circumstances.Since interpreters with paraprofessional accreditation are given assignments by the police and the court in Australia, it is worthwhile exploring the discourse processes in this data set in order to determine what practical implications they have.

Analytical framework
The analysis in this article adopts a perspective of interpreters as one of the participants in a communicative event (Wadensjö 1998;Roy 2000;Russell 2000;Angelelli 2004;Nakane 2007;Yoshida 2007) rather than as an 'invisible' mediator.The analysis specifically draws on Goffman's (1981) notion of participation roles played by speakers that was adopted by Wadensjö (1998) in her study of interpreter roles.Goffman (1981: 226) says that a speaker can be an animator, one who has a role of 'the sounding box from which utterances come,' while an author is 'the agent who puts together, composes, or scripts the turns that are uttered.'Additionally, when a speaker takes the role of principal, they are 'the party to whose position, stand, and belief the words attest.'Thus, the 'default' interpreter role would be associated with that of animator, while the editing or modifying of the source text (or the primary speaker's utterance) would make the interpreter an author.Furthermore, at times some interpreters step out of their default 'conduit' or animator role and intervene as a principal, as their own selves, instead of speaking on behalf of the primary speakers.For example, when the interpreter says his/her own name for the record at the beginning of the police interview, they are taking the role of principal.
In identifying problems in interpreter-mediated discourse, the 'organization of repair' (Schegloff et al. 1977: 361) in the tradition of Conversation Analysis (CA) was examined.Recent studies in interpreting have demonstrated the benefit of a CA approach by revealing the active roles of the interpreter in co-constructing discourse with the primary speakers (Wadensjö 1998;Roy 2000;Petite 2005;Nakane 2007).
Furthermore, CA focuses on speakers' orientation to the interaction (Sacks et al. 1974:

Nakane
The Myth of an 'Invisible Mediator' PORTAL, vol. 6, no. 1, January 2009. 4 699-700; Levinson 1983: 286-87).Therefore, an examination of repair sequences can reveal what the participants of police interviews perceived to be interactionally problematic, and how the interpreters handled such problems.
In this article I examine 'repair' in interpreter-mediated police interviews.Wong (2000: 247) defines repair as '[e]fforts to deal with any problems in speaking, hearing or understanding of the talk,' which includes 'confirmation checks, clarification requests, restatements, repetitions, understanding checks.'Repair occurs when the orderliness and coherence of conversation is at risk and, as a result, the participants in conversation deal with 'trouble sources' that may threaten intersubjectivity, or participants' shared assumptions, which enable mutual understanding about what goes on in interaction (Schegloff 1992(Schegloff : 1295-96)-96).Repair can be categorised into 'self repair' and 'other repair,' depending on whether speakers are repairing their own speech or that of other participants.When self-repair is not done immediately by the speaker whose speech turn became the source of problem, it can take the following sequential structure (ten Have 1999: 116-17;Schegloff et al. 1977: 364): This type of repair sequence was also found frequently in the data in the present study.
However, in interpreter-mediated police interviews, the above sequence would have eight parts instead of four because of the interpreting of each turn.Moreover, because of the existence of two possible 'trouble source' turns (the source text and the interpreter's target text), the repair process sometimes becomes complicated.Below, I demonstrate how interpreter roles shift in various problem-solving situations in police interviews.

Interpreter as 'animator' in repair sequence
In the role of animator, the interpreter would speak completely on behalf of the primary speakers.This means that the interpreter would render the next-turn repair initiator (i.e. indication of need for repair by the 'trouble source' speaker), treating it as directed towards the speaker of the 'trouble source' and not towards him/herself.The following extract is an example of repair sequences rendered by the interpreter taking up a role of invisible mediator (for details of transcription conventions, see the appendix): Here, we see an example of a four-part repair sequence starting with the police officer's question.Although the interpreter quickly self-repairs in line 4 for a better translation of 'what did you do,' the suspect appears to find the question confusing, as he initiates a repair in line 6 after a pause.This repair initiator is rendered by the interpreter in line 7, and the police officer repairs the rather vague question (line 1), modifying it to be more specific.Immediately the suspect responds to the interpreted repaired question in line 10, which is rendered as a coherent response in line 11 by the interpreter.The initial question was vague, as evidenced by the police officer's pauses, and the pause the interpreter allowed in line 2 despite the straightforward syntactic and semantic property of the source text.It is possible that the police officer was aware that the question was poorly phrased and made the question more specific despite the next turn repair initiator requesting a repetition rather than a specification.As far as the interpreter role is concerned, the animator role was maintained throughout this sequence.This is the 'default' role expected of legal interpreters who are expected to maintain impartiality and accuracy, as stipulated by the code of ethics.

Interpreter repair initiation as 'author' in 'off the record' repair
Interpreters are sometimes described as mediators who help their clients to overcome not only linguistic barriers, but cultural barriers as well (Laster & Taylor 1994: 119;Yoshida 2007: 21).The extent to which they function as a 'cultural bridge' is a sensitive issue, as accuracy should not be sacrificed.However, the interpreters are usually the only participants who have access to the two languages used in their assignments, and they can anticipate problems caused by using a direct equivalent in certain situations.
Below is an example of a repair sequence in which the interpreter avoids a possible misunderstanding by quickly repairing 'off the record' to remove the risk: In Japanese, when talking about the number of siblings it is common to hear, for example, literally 'We are three brothers,' including the speaker himself.Thus, through a direct translation between English and Japanese, there is a risk of understanding the number incorrectly.Being aware of this risk, after rendering a literal translation in lines 5-6, the interpreter interrupts the suspect in line 7, questioning whether the 'three brothers' includes himself.Without waiting for the suspect's response to this question, the interpreter modifies the suspect's original reply in line 4 to avoid the risk of the police officer misunderstanding the number of brothers.The subsequent modification of the response to ensure accurate rendition of the original meaning suggests that the interpreter took the role of author, moving from a 'sounding box' towards a 'cultural bridge.'However, this role shift does not go against the code of ethics, as it also recommends to interpreters that they 'shall ask for repetition, rephrasing or explanation' if anything is unclear (NAATI 2001: 14).The repair effectively avoided potential

Nakane
The Myth of an 'Invisible Mediator' PORTAL, vol. 6, no. 1, January 2009. 7 miscommunication that only the interpreter could anticipate and ensured accurate rendition of the primary speaker's message.

Moving between author and principal
Although an 'off the record' repair initiation by the interpreter is necessary at times in order to avoid the pitfalls of seemingly straightforward translation causing cross-cultural misunderstanding, as shown above, there are examples of repairs initiated by the interpreters in their attempts to elicit a coherent or preferred response to the police officer's question.In such cases, they take the role of principal and become a coinvestigator. 3Below is one such example: Ētto juunen kurai.'Uhm about 10 years.' 8 I1: Uh I have been a jeweller for the last 10 years.Extract 3.
In this example, the 'off the record' repair initiation strategy goes beyond an ethically acceptable level.In the unmarked role of animator, the interpreter would render the suspect's utterances in lines 3 and 5 in English.However, judging these utterances as irrelevant to the police officer's question, the interpreter reformulates her initial rendition (line 2) in line 6, as an interruption, latching onto the suspect's utterance in line 5.This reformulated turn is a next turn repair initiator, as she deems the suspect's turn problematic and specifically requests 'the duration.'The suspect repairs in line 7, allowing the interpreter to render a coherent response to the question posed in line 1.
However, the information that S1 dealt with watches (and possibly other things) is lost here although it actually becomes relevant later on in the interview.
This is an example of problematic action by the interpreter, who seeks to avoid a threat to coherent interaction and attempts to elicit answers 'acceptable' to the police officers. 3The interpreter (I1) in this extract is the only interpreter in the data set with a NAATI Professional accreditation.

Nakane
The Myth of an 'Invisible Mediator' PORTAL, vol. 6, no. 1, January 2009. 8 By not rendering the information provided by the suspect and making him focus on the duration of time he worked as a jeweller, this interpreter is violating the code of ethics in terms of accuracy and impartiality.Nevertheless, it should be noted that the interpreter is still, to some extent, engaging in the task of interpreting, as she reformulates the initial question.(Examples in which the interpreter stops interpreting and starts talking as a third party are provided later in this article).The information that was not communicated to the interviewing police officer, however, could have had serious implications for the case.The suspect's business connections with an overseas contact person were probed by the police officer later in the interview.However, when the suspect mentioned a discussion about watches with this contact, the police officer may have drawn a negative inference from the suspect's testimony, as dealings involving watches had not been previously mentioned to him.
The extract below shows a similar case in which the interpreter, faced with a dispreferred response, reformulates in line 5 his first rendition (line 2) without rendering the suspect's initial response (line 4) for the police officer: 1 P3: Why did she meet that person. 2 I3: Naze sonohito ni aimashita ka? 'Why did you see that person?' 3 (0.5) 4 S3: Wakarimasen.
Why did that person um come to meet you?' 6 S3: Annai no hito da tte ittemashita.
'They said he was the guide.' 7 I3: It -it was a tour guide.
In the animator role, the interpreter would have rendered what was said in line 4.
However, the pause in line 3, and the suspect's response 'I don't know' in line 4, may also indicate that the suspect was confused by the question.This is possible because 'aimasu (the past aimashita),' the equivalent of 'meet' in Japanese in line 2, with the 'why' interrogative, implies that the suspect had arranged to meet 'that person.' Noticing the confusion, and possibly being aware that the suspect's response in line 4 may suggest his own incompetence, the interpreter self repairs in line 5 by replacing the term 'au' (to meet) in line 2 with the term 'mukaeru' (to come and meet).This functions as a next turn repair initiator, which successfully (from the interpreter's point of view) elicits a repaired response.However, had the suspect's response 'I don't know' been rendered in English, it may have indicated that, if she had not arranged to meet 'that person,' the police officer's question did not make sense to her.Furthermore, the omission of 'They said' in the rendition in line 7 makes her responsible for the knowledge that 'that person' was a tour guide.In other words, the omission projects her as an insider who knew that the person who met her tour group was a tour guide, rather than a peripheral member of the tour group who only heard from other members of the group that this person was a tour guide.
Thus, although the interpreter does seem to be engaging in interpreting as an author, by avoiding renditions of what he deems irrelevant and of problems associated with interpreting itself, his role leans towards that of principal, as in the previous example.It should be noted that the interpreter in this excerpt has Japanese as his second language, which may explain his correcting himself without rendering the suspect's initial response.In other words, he may have felt insecure about his rendition of the police officer's question into Japanese. 4hese examples of role shifts may occur either because the interpreter feels (inappropriately) responsible for not eliciting preferred or relevant responses from the suspect or because the interpreter is unsure of the quality and accuracy of their first rendition.In the former case (extract 3), at least in Australia, the behaviour of Interpreter 1 breaches the code of ethics because she fails to render the suspect's response (accuracy) and allows her personal judgement of relevance influence her performance (impartiality).In the latter case (extract 4), the interpreter tries to ensure the quality and accuracy of his rendition, while protecting his reputation-that of a competent interpreter-at the cost of denying the police officer access to the information provided by the suspect.Unlike the former example, in the latter, the lack of rendition of the first response 'I don't know' may have favoured the suspect in that a refusal to give information would make the police suspicious (Kurzon 1995: 68).Thus it appears that role shifts may variously favour or disadvantage suspects, depending on the information that is lost or repaired through the interpreters' discourse management.
The interpreter's vulnerability and need to protect their reputation as a competent interpreter (Jacobsen 2008: 154) may also be evoked by requests for clarification by the an example fromSchegloff et al. (1977: 368): B: .hhhWell I'm working through the Amfat Corporation.(Trouble source) A: The who? (Next turn repair initiator) B: Amfah Corporation.T's a holding company.(Self repair) According to Heydon (2005: 99), four-part repair sequences, as shown below, are commonly found in police interviews: