REVISTA DE LLENGUA I DRET JOURNAL OF LANGUAGE AND LAW #79 YOU SPEAK SOME SPANISH? INDICATORS OF INTERPRETERS’ (NON-) PERFORMANCE IN SPANISH CRIMINAL COURTS Carmen Bestué* Mireia Vargas-Urpí** Abstract In this paper we will attempt to establish the reasons for low-quality performance of court interpreters in the context of criminal proceedings in the courts of Barcelona. Drawing on a corpus of authentic recorded hearings, we explore four possible factors that may have an impact on interpreters’ performance: length of the hearing, conversation management problems, interpreters’ experience, and type of offence being tried. Furthermore, we also analyse unjustified non- renditions, which are considered an important indicator of non-performance. The findings are discussed in the context of the linguistic competence of the interpreter, their role in the interaction with all the participants, and the established rules of professional codes of conduct. This paper will draw attention to the need to improve the qualification of interpreters, but also to raise awareness of the role of the interpreter among members of the judiciary in Spain, with a view to protecting the procedural rights of those who do not speak or understand the language in which the proceedings are conducted. Keywords: court interpreting; role of the interpreter; procedural rights; quality of interpreting. PARLES UNA MICA DE CASTELLÀ? INDICADORS DE LA (IN)EFICIÈNCIA DELS INTÈRPRETS A LES SALES PENALS ESPANYOLES Resum En aquest article intentarem determinar els motius que hi ha darrere de la baixa qualitat de la feina dels intèrprets judicials en el context dels processos penals als tribunals de Barcelona. A partir d’un corpus de judicis reals enregistrats, explorem quatre possibles factors que poden afectar l’actuació dels intèrprets: durada del judici, problemes de gestió de les converses, experiència dels intèrprets i tipus de delicte que es jutja. A més, també analitzem faltes de rendiment injustificades, que es consideren un indicador important de la ineficàcia. Les conclusions es comenten en el context de la competència lingüística de l’intèrpret, el seu paper en la interacció amb tots els participants i les normes establertes dels codis de conducta professionals. En aquest article se subratlla la necessitat de millorar la qualificació dels intèrprets, però també de conscienciar els membres del poder judicial espanyol sobre el paper dels intèrprets, amb l’objectiu de protegir els drets processals de les persones que no parlen ni entenen la llengua en què es du a terme el procediment. Paraules clau: interpretació judicial; paper de l’intèrpret; drets processals; qualitat de la interpretació. * Carmen Bestué, associate lecturer, Department of Translation, Interpreting and East Asian Studies, Universitat Autònoma de Barcelona. carmen.bestue@uab.cat. 0000-0001-7945-9348. Mireia Vargas-Urpí, Serra Húnter associate lecturer, Department of Translation, Interpreting and East Asian Studies, Universitat ** Autònoma de Barcelona. mireia.vargas@uab.cat. 0000-0001-6302-581X. Article received: 27.10.2022. Blind reviews: 02.03.2023 and 27.03.2023. Final version accepted: 08.05.2023. Recommended citation: Bestué, Carmen, & Vargas-Urpí, Mireia. (2023). You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts. Revista de Llengua i Dret, Journal of Language and Law, 79, 116-139. https://doi. org/10.58992/rld.i79.2023.3897 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Contents 1 Introduction 2 Previous studies 3 Method 3.1 Description of the TIPp corpus 3.1.1 The annotation of textual and interactional problems 3.2 Revisiting the TIPp project 3.2.1 Factors that can influence interpreters’ performance 3.2.2 The problem of unjustified non-renditions 4 Discussion and conclusions Acknowledgements References Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 117 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts 1 Introduction In 2017, after completing the TIPp project, which will be further described in this article, we realised that, on average, in hearings involving interpreting between Spanish and English, French or Romanian, 54% of the length of the oral interaction stage was not interpreted to users with limited or no language proficiency in the official language of the hearing. Furthermore, of the 46% that was actually interpreted, the low quality of the interpretation – with frequent mistakes by the interpreters in the three language combinations, such as omissions, mistranslations and impartiality problems (for more details, see Arumí & Vargas-Urpí, 2018; Bestué, 2018, 2019; Orozco-Jutorán 2017, 2018, 2019; Vargas-Urpí, 2017, 2018; and Vigier, 2019) – raised doubts about the notion that just having an interpreter present is sufficient to guarantee defendants’ rights. The TIPp project built a corpus of transcriptions of recordings of oral proceedings that took place in Barcelona between January and June 2015, just a few months before the transposition of Directive 2010/UE/64. The results obtained were so alarming that we needed to provide a better understanding of the situations in which they occurred, in order to identify possible factors that might have had an impact on the quality of the analysed interpretations. This article presents a qualitative approach that involves the study and description of real interpreting practices in Spanish courts. As already described elsewhere (see, for instance, Ortega Herráez, 2011; Blasco Mayor & Del Pozo Triviño, 2015), the Spanish Justice Administration does not require professional interpreters to be hired for court interpreting. This has been repeatedly denounced by professional associations such as APTIJ.1 2 Previous studies Studies focusing on interactions have been key to the development of court interpreting research, as they reflect real shortcomings and challenges that court interpreting has to face as a discipline and a profession to ensure that participants’ rights to communication are safeguarded. However, as also mentioned in Orozco-Jutorán (2018), studies focusing on a large corpus of interactions are still rare in court interpreting research. Among the few exceptions are Berk-Seligson (1990/2002), based on 114 hours of interaction in US courtrooms, Angermeyer (2015), based on tape-recordings of 60 arbitration hearings in New York courts, Cheung’s (2017) 100-hour corpus of court interpreting records from Hong Kong, and Gallez & Maryns’ (2014) case study recorded in an attempted murder case tried before a Belgian court. In fact, in Monteoliva-Garcia’s (2016) Annotated bibliography on legal and judiciary interpreting, containing 100 works, only 19 publications explicitly mention the analysis of recordings. Of these, only 7 publications account for more than 15 recordings, namely: Angermeyer (2015), already mentioned; Braun and Taylor’s (2012) 41 recordings of simulations of pre-trial encounters; Russell’s (2002) 20 recordings of police interviews; Pöllabauer’s (2004) 20 audio-recorded asylum hearings; Martínez-Gómez’s (2014) 19 recordings of prison interviews; Hale’s (2004a) 17 audio-recorded hearings, and Pérez-González’s (2006) 16 videos of mock trials. In most studies focusing on interactions, the study of interpreters’ accuracy, e.g., Hale’s studies about the interpretation of discourse markers (1999) or the interpretation of questions (2002), reflects how interpreters’ linguistic choices have an impact on their renditions and, thus, on the role they play (either consciously or unconsciously) in the interpreted event. Other studies have also considered the interaction that takes place in an interpreter-mediated court hearing, i.e., how other participants or contextual factors may impinge on interpreters’ performance, such as Angermeyer’s (2015) discussion of the impact of interpreting modalities (simultaneous vs. consecutive), or Cheung’s (2017) analysis and distinction between self-prompted and other- prompted non-renditions. However, none of these previous studies has focused on external factors such as length of the hearing or type of offence. This study presents the results of the largest court interpreting corpus that has been compiled and analysed in Spain to date and seeks to encompass both perspectives (textual and interactional) to discuss the impact of the interpreter’s role on interpreting quality. 1 URL: https://www.aptij.es/aptij/situacion-de-la-profesion/ Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 118 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts 3 Method This article revisits part of the results obtained in the TIPp project,2 which are mainly based on a quantitative approach. However, this article uses a qualitative approach that may shed light on the possible factors and indicators of the general low quality of interpreter performance in the TIPp’s corpus. This section will first briefly summarise the TIPp project’s methodological approach and will then focus on the method followed for this article. 3.1 Description of the TIPp corpus The TIPp corpus is composed of the transcriptions of all the hearings where interpreting took place in 10 criminal courts in Barcelona for three language combinations (English-Spanish, French-Spanish and Romanian-Spanish)3 from January to June 2015. It consists of the transcriptions of 55 hearings: 19 (EN-ES), 9 (FR-ES) and 27 (RO-ES). The recordings that make up the corpus were transcribed manually by a team of research technicians who were specially trained for this purpose. To visualise the transcription and annotate it, the EXMARaLDA Partitur Editor tool was used. This is free software where transcriptions can be visualised in musical score format, which facilitates the identification of overlaps and interruptions, among other features of spontaneous interactions. Transcription is verbatim and reproduces the oral mistakes that are typically produced in oral speech. The translations into English of Romanian, Spanish and French are lexical and thus also reproduce the mistakes and ambiguities of the original excerpts in their respective languages. These transcriptions have been provided by the authors for Spanish and French excerpts, and by a research technician for Romanian. 3.1.1 The annotation of textual and interactional problems The annotation of the transcripts in the TIPp project considered two dimensions following Wadensjö’s (1998) distinction between “talk as text” and “talk as activity”: the annotation of textual problems and the annotation of interactional problems. Interactional problems included overlaps, long turns, interruptions, interpreters’ non-renditions and speech style (see Arumí & Vargas-Urpí, 2018). For the purpose of this article, only non- renditions are explained. Annotation of solutions to textual problems and main results Textual solutions were first classified according to a quality scale: adequate, improvable (i.e., that could be improved) or inadequate. Afterwards, the annotation of the types of solutions to textual problems used the following categories, which are further described in Orozco-Jutorán (2017, 2019) and Vigier (2019): • Possible categories for “adequate” solutions: o (EH) Established equivalent o (IM) Making some information implicit o (EX) Making some information explicit • Possible categories for “improvable” solutions: o (CR) Change of register o (NMS) Minor shift in meaning (compared to the source text) 2 TIPp stands for “Traducción e Interpretación en los Procesos Penales” (Translation and interpreting in criminal proceedings). It was funded by the Spanish Ministry of Economy and Competitiveness (ref. FFI2014-55029-R) and developed from January 2015 to December 2017. The PI were Dr Carmen Bestué and Dr Mariana Orozco-Jutorán. URL: https://ddd.uab.cat/pub/ jorconmatpro/2015/tipp/ 3 While Catalan is also an official language in Catalonia, it is rarely spoken in the corpus, as Vargas-Urpí (2020) explains. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 119 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts • Possible categories for “inadequate” solutions: o (O) Omission o (OG) Serious omission o (NT) Not translated o (AD) Addition of information o (ADG) Serious addition of information o (ITER) Inadequate terminology o (FS) Major shift in meaning (substantial distortion of meaning from that of the original message) o (FSG) Serious major shift in meaning o (SS) Incomprehensible (message is not understandable, does not make sense) The difference between “serious” and “less serious” was established considering the possible impact on the communication between the participants in the hearing (see also Orozco-Jutorán, 2017). In the case of omissions, for instance, we annotated the example below as a serious error because the information omitted in the rendition could be relevant to the defendant who is confronted with a plea bargain. Excerpt 1. EN01 Judge: Y la responsabilidad civil de 200 euros que debería de abonar con el otro acusado si nunca saliera condenado. Bueno. Entonces le pregunta si reconoce los hechos y si acepta estas condenas que pide la acusación. And the civil liability of €200, that he should pay with the other defendant if he is convict- ed. Well, ask him if he acknowledges the facts, and he accepts these sentences requested by the prosecution? Interpreter: So, you accept the charges? Defendant: To say that I robbed a wallet, hey, they have to free me? Interpreter: Yeah. Defendant: ….to…ye--…free me…? Interpreter: Yeah. Defendant: So, I have said I robbed the wallet, they have to free me. Interpreter: Sí, sí, sí, sí acepta, y tienen que dejarlo en libertad. Yes, yes, yes, yes, he accepts and you have to free him. Since only 46% of the total length of the hearing is interpreted on average, inadequate solutions have been calculated on the basis of “bilingual hour” or “interpreting time” (i.e., considering only the minutes in which interpreting takes place, as also explained in Orozco-Jutorán, 2017, 2019). Figure 1 shows the average frequency of serious mistakes per bilingual hour in each of the subcorpora. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 120 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Figure 1. Average frequencies of serious mistakes per bilingual hour in each of the subcorpora Source: Own elaboration. The figures presented in the previous graph correspond to a mean frequency of 21.1 serious mistakes per bilingual hour, which is one of the most alarming results of the TIPp project. Less serious errors (i.e., those that may not have a direct impact on the hearing) appear in even higher frequencies, as shown in figure 2. Figure 2. Average frequencies of other inadequate solutions per bilingual hour in each of the subcorpora Source: Own elaboration. Annotation of non-renditions When annotating non-renditions for the TIPp project (see Arumí & Vargas-Urpí, 2018 and Vargas-Urpí, 2019), we made a distinction between justified non-renditions and unjustified non-renditions. On the one hand, by considering some non-renditions as justified, we acknowledge that court interpreters must be able to intervene in the conversation if they need to manage conversation problems to guarantee accuracy in their renditions. On the other hand, unjustified non-renditions are those that imply interpreters taking on another role, as detailed in 3.2.2. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 121 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Unjustified non-renditions are more frequent that justified non-renditions in two language combinations (English-Spanish and Romanian-Spanish), while justified non-renditions are more common than unjustified non-renditions in the French-Spanish combination, as reflected in figure 3. Overall, the Romanian-Spanish subcorpus accounts for the highest frequency of non-renditions. Figure 3. Average justified and unjustified non-renditions per bilingual hour Source: Own elaboration. In order to better relate justified non-renditions to conversation management problems, these were further classified into the following categories: • ‘pause’, when interpreters need to ask for a pause to interpret; • ‘clarification’, when interpreters need to disambiguate a specific aspect in order to interpret accu- rately; • ‘confirmation’, when interpreters ask to verify something they may have misheard; • ‘retrieval’, when interpreters ask for repetition of something they might have partially forgotten, especially in the case of long turns. Unjustified non-renditions were also classified into various categories: • WIA: warning, instructions and advice; when interpreters warn the defendants, or give them advice or instructions; • ‘advocate’, when interpreters provide answers on behalf of the defendants or witnesses; • ‘assistant’, when interpreters add information or questions on behalf of the judiciary staff. 3.2 Revisiting the TIPp project The main objective of this article is to contextualise TIPp’s results by taking into account both quantitative and qualitative data collected during the TIPp project and to triangulate some of the data that had not previously been considered. A spreadsheet was created with the information of each of the hearings in the corpus, including total length, length of the bilingual part (i.e., excluding monolingual parts of the hearing where no interpreting, either whispered or consecutive, was provided to the accused or victim), serious errors per bilingual hour, less serious errors per bilingual hour, justified and unjustified non-renditions per bilingual hour, type of offence and the interpreter’s pseudonym. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 122 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Having all the information about each of the hearings on a single spreadsheet facilitated comparison between all the hearings: it was possible to see which specific hearings had registered fewer mistakes or non-renditions, and which had registered more; and it was then possible to compare these figures with other factors, such as the length of the hearing, the type of offence or the individual interpreter. In specific cases, we also reviewed the transcripts and/or original recordings in order to have a broader context for some of the mistakes or non- renditions identified in the corpus. 3.2.1 Factors that can influence interpreters’ performance The following factors have been considered for the analysis of the corpus: length of the hearing, conversation management problems, interpreters’ experience, and the type of offence being tried. Length of the hearing and interpreters’ performance One factor that is usually believed to affect interpreters’ performance is fatigue (see, for instance, Moser- Mercer, Künzli and Korac, 1998, in a study of conference interpreting); this is why conference interpreting assignments usually involve two interpreters working together in the booth and taking turns to interpret. In court interpreting in Spain, only one interpreter is requested per hearing, regardless of its length. Considering these interpreting circumstances, we might wonder whether the length of the hearing is one of the reasons for decreasing quality in the interpreters’ performance. The TIPp results show that the hearings included in the corpus had an average length of 19.92 minutes, the longest lasting 72 minutes (where only 14 minutes were actually interpreted), and the shortest lasting 2 minutes. Eleven of the hearings lasted more than 30 minutes, although the interpreted parts only exceed 30 minutes in one hearing (FR6).4 Considering the average length of hearings and the small proportion of interpreting in general, we can confirm that fatigue is not the cause of the errors in our corpus, at least not in most of the hearings. As shown in Figure 4, it is not possible to infer a cause-effect relationship between the length of interpreting in the hearing and the average frequency of errors (either serious or less serious), as some of the shortest hearings (e.g., EN03, EN09, EN12, EN16) account for some of the highest average frequencies of errors. Figures are similar in the other two subcorpora. Figure 4. Relationship between the length of interpreting and average frequency of errors per bilingual minute in the English- Spanish subcorpus Source: Own elaboration. For instance, in EN09, the bilingual exchange only lasts one minute. The intervention of the interpreter is confined to the so-called “right of the last word”, a very predictable question formulated by the Judge; however, it is poorly translated in a nonsensical sentence, as we can see in Excerpt 2. Besides his limited competence 4 An ad hoc numeration system has been created to refer to the different hearings that make up each of the subcorpora, using the abbreviation of the interpreting language and a number. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 123 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts in English, the interpreter also shows disregard for professional codes when he adds two non-renditions after the defendant’s first answer. Excerpt 2. EN095 Judge: Pregúntele si quiere decir algo más que no haya dicho con anterioridad. Ask him if he wishes to say anything else he has not previously said. Interpreter: Have you wish to say something else, previous to what you already said? Defendant: Yeah. Interpreter: You wish to say something? Defendant: Yes. Interpreter: Do you wish to speak in Spanish or English? Defendant: In English. Interpreter: Tiene algo que decir. He has something to say. In hearing RO25 (Excerpt 3), the bilingual part also lasts only one minute. Even in such a short bilingual part, a high concentration of errors is found, combining both textual and interactional problems: namely, 4 serious textual errors (1 addition, 2 false meanings, 1 omission), 9 less serious errors and 2 unjustified non-renditions. For instance, the interpreter mistranslates the imprisonment penalty that is accepted by the defendant in the plea bargain, which according to the judge was “18 months”, but it is translated as “1 year and 18 months”. The following excerpt also reflects a tendency to summarise information, as shown in the first rendition, as well as one of the unjustified non-renditions found in this one minute of bilingual exchange. Excerpt 3. RO25 Judge: ¿Usted con esta condena y esta pena así modificada se conforma? Do you agree with this sentence and this penalty thus amended? Interpreter: Dacă sunteţi de acord cu aceasta? And do you agree with this? Defendant: Da, domnule preşedinte, mă conformez cu această situaţie. Yes, Mr. president, I agree with this situation. Interpreter: Sí. Nu vorbeşti cu preşedintele. [In Spanish] Yes. [In Romanian] You don’t talk to the president. Judge: Muy bien. Pues a la vista de la conformidad manifestada (…) Very well. So in view of the agreement expressed (…) Conversation management problems vs. justified non-renditions Conversation management problems (e.g., long turns, overlapping speech, interruptions) may also be regarded as factors that influence interpreters’ performance and that might even provoke some of their justified non- 5 All the examples are extracted from real cases interpreted into Spanish. The English versions presented here are not idiomatic translations but renditions for academic purposes and do not aim to serve as a standard. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 124 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts renditions. Figure 5 reflects the distribution of justified non-renditions in each subcorpus, according to the categories described in section 3. Figure 5. Types of justified non-renditions: average frequency per bilingual hour Source: Own elaboration. The low average frequencies per bilingual hour of certain non-renditions (e.g., pause and retrieval) mean that these types of non-renditions only appear in very few hearings of the corpus, thus resulting in less than one non-rendition per hour in certain language combinations. As shown by the examples presented in this section, the fact that the non-renditions are considered as justified does not mean that these interventions comply with formal protocol. Indeed, according to most professional codes of practice, interpreters should not intervene for clarification or repetition without a formal request to the members of the judiciary (see, for instance, Dueñas et al, 2012: 579). Non-renditions to ask for clarification are the most frequent. The following exchange presents a clear example where we can see a clarification of this kind of justified non-rendition. Excerpt 4. EN02 (1) Defendant: We were in a shop buying water, a 24-hours shop buying water, one (...) = Interpreter: = y estaba en una, en una tienda de 24 horas, comprando agua. And we were in a, in a 24-hour shop, buying water. Defendant: When the police drove by with a bike, two of them. Interpreter: Motorbike? [CLARIFICATION] Defendant: Yeah. Interpreter: Entonces cuando la policía, ah... vino con motos, dos de ellos Then, when the police came ah… came by motorbike, two of them Defendant: So we were talking, I was talking, they came to us, like, they asked to show our hands, both of us, so we show… Interpreter: So they asked you to show them your hands and so on? [CONFIRMATION] Defendant: Yeah. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 125 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Interpreter: Bueno, estuvieron hablando con ellas, le, les pidieron que les mostraran las manos. Well, they were talking to them, they asked her, them to show them their hands. In English, ‘bike’ can be an abbreviation for both ‘bicycle’ and ‘motorbike’. However, in Spanish, there is not an equivalent with both meanings, which is why the interpreter must ask to disambiguate the meaning in order to produce an accurate rendition. The second non-rendition is used to confirm the correct reception of information. While these two non-renditions are justified, the fact that the interpreter does not bring them to the attention of both the members of the judiciary and the defendant may be considered an infringement of the standards recommended in court interpreting, as mentioned, for instance, in EULITA’s6 or NAJIT’s7 codes of ethics. We need to refer to other countries’ codes because there is no mandatory code of ethics or professional behaviour in Spain. Considering the relatively high frequency of conversation management problems,8 non-renditions to ask for a pause were expected. However, pauses account for a very small proportion, as they only occur in 5 of the 55 hearings analysed (3 in the English-Spanish subcorpus, 1 in French-Spanish and 1 in Romanian-Spanish). As Vargas-Urpí (2019) also explains, from the total number of 11 non-renditions to ask for a pause, 10 addressed the defendant and only 1 a member of the judiciary (the defence counsel). The following excerpt reflects a non-rendition where two pauses are requested after a sequence containing various overlapping turns between the prosecutor and the defendant, which may have made some parts inaudible for the interpreter. Excerpt 5. EN11 Prosecutor: Bien... ¿Fue detenido por la policía en ese momento? OK. Were you arrested by the police at that moment? Interpreter: You were arrested by the police that very day? You were caught by the police? Defendant: Yes, eh, because, eh... You want I explain all? Or what? You... okay... Interpreter: ¿Quiere que.. que lo explique todo? ¿O...? Do you want me to… to explain it all? Or…? Defendant: Because a lot of things happened... Prosecutor: [overlapping] Sí, sí… ((inaudible)) Yes, yes… Interpreter: Porque pasaron muchas cosas. Because a lot of things happened. Defendant: Yes. Interpreter: Bueno… Well… 6 According to EULITA’s protocol and demeanour principle: “Should it become necessary for them to assume a primary role in the communication, they must make it clear that they are speaking for themselves, by using for instance the third person (i.e.: “The interpreter needs to seek clarification ...”)”. 7 According to NAJIT’s 5th principle: “When it becomes necessary to assume a primary role in the communication, they [interpreters] must make it clear that they are speaking for themselves”. 8 For instance, in 72% of the hearings, a member of the judiciary’s speech rate is faster than 180 words per minute, (while 150 words per minute is considered the maximum speech rate an interpreter can cope with in court interpreting). There is also an average of 15 overlapping turns per hour, and an average of 24.6 interruptions per hour by the interpreter while interpreting. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 126 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Prosecutor: Bueno… Well… Defendant: ((inaudible)) Interpreter: [overlapping] Hold on... wait, wait, wait... Defendant: (...) Yes, I have to say that. Interpreter: Just a second. Prosecutor: No, no, que explique en relación a esto, que llegó la policía, y le detuvo, y que si… No, no, for him to explain in relation to this, when the police arrived and he was arrested, if he… It must be noted that, even if they may be considered justified because they occur after some overlaps, the pauses requested in Excerpt 5 are not used to ask for a pause to interpret, but to ask for a pause so that the prosecutor can continue questioning. Non-renditions to ask for confirmation might also be caused by conversation management problems: fast speech rate, overlapping turns, interruptions, or the use of English and French as a lingua franca by users with other mother tongues (mainly, African languages), which may require the interpreter to confirm specific details in order to interpret accurately. The following example (Excerpt 6) reflects how an interpreter needs to ask for clarification twice due to the defendant’s non-normative use of English: Excerpt 6. EN02 Defendant: They talked to theirselves, they asked us to = Interpreter: = They talked to…? [VERIFICATION] Defendant: They talked to theirselves, the two policemen = Interpreter: = To both of you? [VERIFICATION] Defendant: No, the two policemen, they talk to theirselves, after = Interpreter: = Ah, ok = Defendant: = ((inaudible)) show them our hands. Interpreter: Eh, estuvieron hablando entre ellos, después de que, de que les enseñaron las manos. They were talking to each other, after they had shown them their hands. The defendant’s use of a non-normative word (*theirselves) makes the interpreter doubt and he tries to confirm the information he has to interpret twice. As also happened with non-renditions to ask for a pause, non-renditions to ask for confirmation more often address defendants or witnesses (22 instances out of 29) than members of the judiciary (7 out of 29). The following extract shows an example of the few non-renditions addressing the members of the judiciary. Excerpt 7. RO06 Prosecutor: Pregúntele si en ese curso, en ese tiempo, ella le entregó 12 objetos cilíndricos que con- tenían hachís con casi ciento, ciento ocho gramos de hachís. Ask him if, in that course [of events], in that time, she handed him 12 cylindrical objects containing hash, with almost a hundred, a hundred and eight grams of hash. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 127 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Interpreter: [in Romanian] Te întreabă dacă tu i-ai înmânat lui ah, 2, deci ca fiind două (…) așa în formă de cilindru care conțineau hașiș. ¿Perdón, cuántos gramos? [in Romanian] He is asking you if you gave eh, two, that’s it two (...) like this in a cylindri- cal form containing hash. [in Spanish] Excuse me, how many grams? Prosecutor: Ciento ocho. A hundred and eight. Interpreter: O sută opt grame. A hundred and eight. While this non-rendition is justified (in case of doubt, it is better to confirm the information to ensure accuracy), it also reflects the interpreter’s lack of specific skills and strategies such as note-taking, since the prosecutor’s turn cannot be regarded as too long or complex. It must be noted that while considered justified if they are used to improve the accuracy of the rendition, we should also discuss whether some of these non-renditions are really necessary for that purpose, as may be the case reflected in Extract 2 (EN09) previously presented, where the defendant’s response is clear (“yeah”) but the interpreter instead of interpreting this answer, insists and reformulates the question (“You wish to say something?”) in an action that could be considered as a deterrent strategy to prevent the defendant’s intervention at this moment of the trial. To contextualise, we must explain that defence attorneys often discourage defendants from using their last words to add more information. In this case, the interpreter seems to be aware of this practice and uses her own voice as a warning to the defendant. Interpreters’ recurrent presence in the corpus vs. interpreting quality Access to the video-recordings of hearings has made it possible to identify all the interpreters participating in the corpus, who have been given pseudonyms as part of the corpus metadata. There are some recurring interpreters in each of the subcorpora. Considering that this corpus is a representative, random sample (see Orozco-Jutorán, 2018: 5), it is noteworthy that Matilde is the interpreter for more than half of the hearings (55.6%) in the French-Spanish subcorpus, while others appear less frequently (Michel, 22.2%; Toni, 11.1% and Benoît, 11.1%). In the Romanian-Spanish subcorpus, 4 female interpreters have been identified: Ana, who interprets for almost half of the hearings (46.2%); Cosmina (26.9%), Gabriela (23.1%) and Rodica (3.8%). This is the only subcorpus where we can confirm that all the interpreters are native Romanian speakers. In the English-Spanish subcorpus there are also some recurring interpreters, such as Alfredo (interpreter for 31.1% of the hearings in this language combination) and Aleix (21.1%). However, there are also four interpreters, Montse, Manolo, Juanito and Jaime, who only interpret in one hearing. The only hearing suspended by the judge due to the poor quality of interpreting was the one interpreted by Montse, who was then substituted by Aleix, in the English-Spanish subcorpus. The fact that recurring interpreters have been identified in each of the subcorpora is interesting for several reasons. First, we cannot infer that the high frequencies of errors and unjustified non-renditions identified in the corpus are due to a lack of experience in court interpreting, since most of the interpreters in the corpus participate in more than one hearing, some of them recurring frequently. The hearings where the most recurring interpreters in our sample (i.e., Matilde, Ana and Alfredo) participate do not reflect a decrease in the number of (either serious or less serious) errors produced or in the number of unjustified non-renditions. For instance, despite being the most frequent interpreter in our corpus, Ana’s non-renditions clearly reflect her lack of awareness of the role of the court interpreter as recommended in deontological codes. In Excerpt 8, Ana’s renditions reflect textual errors, such as additions and change of register. Excerpt 8. RO08 Judge: Bien. Dígale que el ministerio fiscal en este acto ha hablado con su letrado y ha rebajado Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 128 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts la condena a la pena de 1 año de prisión. ¿Si se conforma con esta nueva pena? Good. Tell him that the prosecutor in this act has talked to his attorney and he has re- duced the sentence to one year of imprisonment. Does he agree with this new penalty? Interpreter: Ah, că domnul avocat a vorbit cu ah doamna procuroare şi din pedeapsa care îţi cerea de 3 aniîţi face o reducere şi rămâne într-un an ahh. Facem o înţelegere? Eşti de acord? Oh, that Mr. Attorney has spoken to oh, Ms. Prosecutor and for the sentence that they asked for was for 3 years, He will reduce it to 1 year, ehh. Do we make the deal? Do you agree? This example shows an interaction where the judge informs the defendant about the plea bargain reached by the prosecutor and the defence attorney. The interpreter seems to have taken part in the negotiations since she adds in her interpretation extra information that was not included in the judge’s intervention (“for the sentence that they asked for was for 3 years”). In our opinion this addition infringes not only the standard of accuracy but also the standard of impartiality. Considering the role of the interpreter, by adding information the interpreter is stepping into the role of the judge or the defendant’s counsel. Most professional codes of conduct for interpreters state the requirement that the interpreter has to convey everything to the defendant exactly as it is stated in the language of the court without explaining or embellishing in any way. By adding information, the interpreter may think that she is aiding the defendant, but in fact she is overstepping her role, which is to place the defendant on equal footing with a person who understands the language of the court. While, according to Mikkelson (1998), the role of the interpreter should probably be redefined to result in the effective removal of the communication barrier, we believe that this proposal should be limited to the selection of translation techniques better suited to avoid pure linguistic or word-for word translations resulting from the lack of lexical equivalents and considering a listener-centred interpreting process (Stern, 2004:70). The same intervention also presents a change in the level of formality of the register where the question “Does he agree with this new sentence?” is adapted into the informal utterance: “Do we make the deal?” From a semantic perspective, this sentence introduces an ambiguity: either the interpreter is the voice of the judge, or she is involving herself in the decision-making process. In any case, this modulation in the utterance could contribute to facilitating the decision of the defendant and therefore should be avoided. From a pragmatic perspective, changing to first person plural also implies a shift in the way the interpreter aligns with the defendant. By using “we” the interpreter not only aligns with the defendant but may also be adding pressure to accept the bargain as if it were a collective decision. At the closing of the same hearing, Ana again makes an addition, probably motivated by her knowledge of the procedure. Excerpt 9. RO08 Judge: Dígale que se le suspende la ejecución de la condena por un periodo de 2 años. Tell him that the enforcement of the sentence is suspended for 2 years. Interpreter: Deci acest an de zile se suspendează pe doi ani de zile, adică doi ani de zile nu poţi nici un delict. That is to say, this one year is suspended for two years, that is to say that for two years you cannot commit any crime. The previous two excerpts have been extracted from RO8, one of the hearings with the fewest errors detected. Ana does not make serious errors in her renditions, such as false meaning, incomprehensible sentences, omissions, etc., but she shows a clear tendency towards simplification of the language, summarising information in the sentence imposed, lowering the register and relying too much on unnecessary explanation, again reflecting her lack of awareness of the role of the court interpreter and instead adopting a paternalistic relationship towards the defendant. As we do not have access to external sources of information, we have no data about the level of education and training of the interpreters, but, in view of the results, we can infer, as Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 129 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts indicated by Hale (2014b: 321-322), that interpreters seen as advocates and gatekeepers of the interlocutor’s turns are ad hoc or untrained interpreters who do not abide by a professional code of ethics requiring them to be impartial and accurate. Complexity of terminology vs. textual errors The total corpus is composed of abbreviated criminal proceedings corresponding to the Criminal Courts of Barcelona. Abbreviated criminal proceedings are a type of summary trial proceeding created to accelerate the investigation stage. The proceedings are initiated by an investigating judge (juez de instrucción), and the trial stage is conducted by a different court presided over by a single judge, the Criminal Trial Court (Juzgado de lo Penal). Abbreviated proceedings tried in Criminal Courts are reserved for offences with a maximum potential sentence of less than five years in prison. Therefore, more serious offences are not included in our corpus. Figure 6 reflects the nine broad categories that we have created to classify the offences tried in our corpus. Our goal in creating this classification was to establish whether difficulty with expected terminology has an impact on the number of errors and, for this reason, this list was not based on any classification of criminal offences in Spain. Therefore, we have classified the hearings in the following categories: offences against property only (petty theft, theft, burglary, property damage), acts causing harm to a person (assault), offences involving illicit drugs, violence against women, offences against public order, offences for financial or personal gain, dangerous driving, coercion and corporate crimes. This classification is the result of the production of an exhaustive list of domain-specific terms for the purposes of creating a thesaurus (see Bestué, 2021) for court interpreters. In this respect, we considered that the first category (by far the largest one with 31 hearings) may include descriptions of material damage, burglary tools, and damage to property, for instance. As for the second category, acts causing harm to a person, terms related to physical harm, injuries, and other medical terms in general should be prepared in advance by the interpreter. Figure 6. Relationship between type of offence tried and average frequency of errors per hour Source: Own elaboration. Unfortunately, the absence of serious errors does not necessarily indicate accuracy in the textual solutions, as we can see in the following examples from a case on corporate offences where the judiciary members in Spanish and the witness (a French speaker with some knowledge of Spanish) try to overcome the communication problems that were not solved by the interpreter: Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 130 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Excerpt 10. FR06 Witness: Aujourd’hui je n’ai pas d’actions en direct, c’est TARI France. Today I don’t directly have shares, it’s TARI France. Interpreter: Actualmente no tiene acciones directas. Currently, he does not have direct shares. Judge: Porque las tiene TARI Francia que es una filial. Because TARI France, a subsidiary, has them. Witness: Exacto. Correct. Excerpt 11. FR06 Private prosecutor: Con la venia del ilustrísimo señor, eh señor Giraud, ¿Qué cargo ostentaba el señor Pérez en TARI Ibérica en el 2007 y en el 2008? If it pleases the court, ehh Mr. Giraud, what position did Mr. Pérez hold at TARI Ibérica in the years 2007 and 2008? Interpreter: Tu as compris?.. Il vous demande la charge qui occupait, érez… ¿Puede repetir, por favor… el nombre? Did you understand?... He is asking you the charge that érez... held. Could you please repeat, please… the name? Private prosecutor: ¿Qué cargo ostentaba el señor Pérez? What position did Mr. Pérez hold? Interpreter: Pérez, Pérez Pérez, Pérez Private prosecutor: En el 2007 y en el 2008 en TARI Ibérica In 2007 and 2008, at TARI Ibérica Interpreter: En 2007 et 2008 en TARI Ibérica In 2007 and 2008, at TARI Ibérica Witness: Il était administrateur unique de TARI Ibérica. He was the sole director of TARI Ibérica Interpreter: Eh, era administrativo único de TARI Ibérica Eh, he was the sole administrative of TARI Ibérica Private prosecutor: ¿Administrador? Director? Interpreter: Administrateur, administrateur unique, euh, único Director, sole director, ehh, único. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 131 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Private prosecutor: ¿O sea, administrador único? That is to say, sole director? Witness: Sí (In Spanish) Yes Following many other exchanges similar to the examples shown above the private prosecutor (in the Spanish legal system the private prosecutor is an attorney who presses charges against the defendant on behalf of the victim) points out that the interpreter is not translating well and the judge, acknowledging this, decides to address the witness directly and asks him, since the exchange is “highly technical” if the witness can “try to speak” in Spanish (as can be seen in Excerpt 12). Excerpt 12. FR06. Judge: Yo pienso, usted habla un poco de español ¿verdad? I think, you speak some Spanish, right? Interpreter: Sí. Yes. Witness: Yo hablo, hablo. I do speak, I do. Judge: Pues yo pienso que casi, como es muy técnico yo pienso que usted nos lo intenta decir en español. ¿Usted habla un poco español? So I think that maybe, as it is very specialised, I think you should try to tell us in Spanish. Do you speak some Spanish? Interpreter: Si vous comprenez un petit peu l’espagnol? Si tu comprends la question tu peux, m... répondre directement si, si, si y a quelque chose que vous comprenez pas vous pouvez demander mais si vous comprenez … If you understand a little Spanish? If you understand the question you can, m… answer directly if, if if there is something that you do not understand you can ask but if you understand… Witness: Oui, je, je, je comprends pas tout à 100% mais c’est pour ça que vous êtes là mais je comprends suffisamment parfois (.) D’accord, okay. Yes, I, I, I do not understand 100% but this is why you are here but sometimes I understand well enough (.) All right, ok. Prosecutor: Hay cosas que le cuestan, por eso mismo pedimos el traductor… Certain things are difficult for him, that’s why we asked for a translator… Judge: Yo pienso que si el señor se expresa mínimamente en español lo enten ... nos comunicaremos mejor que traduciéndolo en este caso. I think that if the gentleman can express himself basically in Spanish, we will unders… we will communicate better than we would with translation in this case. The judge thus acknowledges that a “broken communication” in Spanish and French would be better than the interpretation provided by the interpreter and, as shown in the following example (Excerpt 13), instead of adjourning the hearing or reprimanding the interpreter, the judge adopts a participant role, trying to correct by means of her Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 132 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts comprehension of French the misunderstandings caused by the interpreter. This attitude of the members of the judiciary seems to indicate that it is normal for an interpreter to be unacquainted with legal terminology in the context of a criminal case and, as is also demonstrated in the following example, shows that when they speak a foreign language to some extent, they monitor, comment on and correct the interpreter’s rendition. Excerpt 13. FR06 Private prosecutor: Cuando usted llegó a la junta el diez de junio del 2008, ¿con quién iba? When you arrived at the meeting, on June 10, 2008, who was with you? Witness: Euh, euh, je suis venu moi en représentation de, les héri--, des héritiers et Ehh, I went on behalf of the, the he--. heirs and Interpreter: Que ha ido él en compañía de los hereditarios. That he went in the company of the hereditary. Judge: No, en representación, no en compañía. No, on behalf of, not in the company of. As we observed in the TIPp corpus, the members of the judiciary only intervene in the interpreting when they detect textual errors (only one of the 55 hearings was adjourned because of bad interpreting, and the reasons set out by the judge were based solely on the linguistic problems detected) instead of the incorrect handling of the interventions and resulting non-renditions by interpreters. In our opinion this indicates a rather narrow understanding of the role of the interpreter as a mere linguistic facilitator and, what is worse, a disregard for other problematic issues. The next section will explore the risks of unjustified non-renditions detected in the TIPp corpus and their impact on the deficient use of interpreting services in Spanish criminal courts. 3.2.2 The problem of unjustified non-renditions In the previous sections we have explored possible factors likely to lead to a reduction in the quality of the interpreting. In this section, we have also included some reflections about the use of the interpreter’s blatant deviance from impartiality in non-renditions that are considered unjustified, thus also reflecting low quality in interpreting standards. Figure 7 shows the frequencies of unjustified non-renditions in the three subcorpora according to the categories presented in section 3.1.1. Figure 7. Frequencies of unjustified non-renditions per subcorpus (average frequency per bilingual hour) Source: Own elaboration. As shown in Figure 7, interpreters of the Romanian-Spanish subcorpus show a higher frequency of non- renditions compared to French-Spanish and English-Spanish interpreters. The numbers speak for themselves, the number of unjustified non-renditions of these interpreters being particularly alarming for the category of “assistant” to the judiciary. As pointed out by Vargas-Urpí (2019), some interpreters use their own knowledge Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 133 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts or experience of the court system to speak on behalf of the members of the judiciary and, on providing extra information, they overstep their role. For example, the following excerpt, interpreted by Ana, shows two examples of unjustified non-renditions, where the interpreter seems to adopt the role of assistant to the members of the judiciary while trying to extract a confirmation from the defendant. Excerpt 14. RO16 Judge: Dígale al acusado que se le pide como autor de un robo con fuerza en tentativa Tell the defendant that he is charged as the perpetrator of an attempted robbery using force. Interpreter: Ah, deci te acuză de un furt cu spargere în grad de tentativă. Ah, in other words, he charges you with attempted robbery using force. Defendant: Nu imi acord deloc. Să îmi iasă creierii pe ochi. I don’t remember anything. That my brains get out of my eyes.9 Interpreter: Ești de acord sau nu ești de acord? Do you agree or not? Defendant: Nu îmi acordo nada, deloc. I don’t remember anything, nothing. Interpreter: Ascultă, tu ești de acord? Listen, do you agree? Defendant: Nu că nu sunt de acord, nu știu deloc... No, I don’t agree, I don’t know anything… Interpreter: Ok, he says he doesn’t remember this. In Excerpt 14, an exchange that should only have had three turns (the interpreter’s rendition of the judge’s question, the defendant’s answer and the interpreter’s rendition) becomes an exchange with six turns due to the interpreter’s unjustified non-renditions. Although the interpreter does not inform the members of the judiciary about the contents of her non-renditions, overall the defendant’s communicative purpose is respected. When revisiting the data, we realised that some of the non-renditions that have been annotated as ‘clarifications’ were not used to disambiguate ambiguous content to interpret accurately, but instead as a means to extract information, as may be seen in the following excerpt. Excerpt 15. RO14 Prosecutor: Pregúntele si cuando vinieron los agentes de policía eh al campamento si él estaba dormi- do, estaba despierto o qué era lo que él estaba haciendo en ese momento. Ask him if when police officers went eh to the camp, if he was asleep, awake or what he was doing at that time. Interpreter: Când a venit poliția ce făceai tu, erai, dormeai, erai... When the police came, what were you doing, were you sleeping… Defendant: Eu eram cu fratele meu în cort, au venit, ne-au legitimat și apoi or găsit cablu, cupru ăsta. 9 The English translation of this expression does not have a real meaning but it is a lexical translation that seeks to reflect the original, and was proposed by a native Romanian translator. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 134 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts I was with my brother in the tent, they asked us to identify ourselves and then they found that cable, that copper. Interpreter: [in Romanian] Dormeai? (.) [in Spanish] Dice estaba a… [in Romanian] Were you sleeping? (.) [in Spanish] He says he was… Defendant: Erau la 3, 4 sute de metri de ei în stup, în boscheți. They were 300 or 400 metres away from them in the shrub, among the bushes. Interpreter: [in Romanian] La cât? (.) [in Spanish] Ah, dice estaba durmiendo cuando llegó la policía yo y mi hermano nos pidió la documentación y después encontró este cable a unos trescientos o cuatrocientos metros en unos matorrales ha dicho. [in Romanian] How many? (.) [in Spanish] Ah, he says he was sleeping when the police arrived, me and my brother, they asked us for our documentation and then they found this cable three or four hundred meters away in some bushes he said. In the previous excerpt, the interpreter is aware that the defendant has not answered the prosecutor’s question, so she insists on the same question (“were you sleeping?”), instead of interpreting the defendant’s digression. Thus, it is clear that not all the ‘clarification’ non-renditions can be considered justified. Furthermore, the final summarised rendition10 at the end of the previous excerpt presents various errors, including an addition of the information the prosecutor had requested (“he says he was sleeping”), which can be considered a serious error because this was not stated by the defendant, even though the interpreter had attempted to extract precisely that information. We must conclude that, in this case, the defendant’s communicative purpose was not respected. 4 Discussion and conclusions In this paper, we have explored possible reasons for the high frequency of errors identified in a corpus of original, authentic hearings of criminal proceedings (the TIPp corpus). First, we can conclude that the length of the hearings does not affect interpreting quality, because long hearings do not present higher error frequencies, while on the other hand, short hearings (some of them less than 3 minutes) do not present lower error frequencies. While justified non-renditions appear less frequently than unjustified non-renditions, among the justified non- renditions the most frequent are those that are for the sake of clarification. These clarifying non-renditions could also be deployed as a means of extracting information and not necessarily for disambiguating a term or question. The non-renditions most closely related to conversation management, such as asking for a pause or retrieving information, are the ones that appear less frequently. Moreover, pauses almost always address the defendant (only in one case did they address the members of the judiciary) and this could reflect a possible bias on the part of the interpreter. In none of the cases were non-renditions communicated to members of the judiciary, who were therefore unable to monitor whether the non-rendition was justified or unjustified. Another factor that we have considered is interpreters’ courtroom experience. The fact that we could not collect information about these interpreters’ professional experience, because we did not conduct the recordings and did not contact them afterwards either, compelled us to consider the recurrence of these interpreters in our corpus. We have observed that some interpreters appear repeatedly in the three linguistic pairs, and since this is a random and generalisable sample, it leads us to assume that these interpreters are accustomed to practising in judiciary contexts. However, we were unable to link more experience with a higher quality of interpreting. In fact, Ana, the interpreter with the greatest number of hearings in our corpus, is the one who made some of the most serious errors and demonstrates a lack of knowledge and awareness of the codes of ethics and standards of practice. 10 Wadensjö (1998) describes summarised renditions as renditions containing the translation of various original turns. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 135 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Even though all the hearings are the same type of proceeding —abbreviated proceedings— and there was a high concentration of the same type of offences, we do not see a reduction in the number of errors. For example, it is quite common to have to interpret in a hearing on theft or robbery, but even in such cases the number of interpreting errors is not less frequent in this category. On the other hand, in hearings that require more specialised terminology, such as those seen in examples from excerpts 10, 11 and 121, what is most surprising is not only that the interpreter shows low proficiency in this specialised terminology, but also that the members of the judiciary accept this lack of preparation as normal. Only in the extreme case of one interpreter who was clearly incapable of interpreting into English (one of 55 hearings) was the case adjourned because of inadequate interpretation. In the remaining hearings, whenever the judges were able to monitor the interpreter’s performance and to detect textual errors, they showed a high level of tolerance towards it and preferred “broken communication” with or without the intermittent intervention of the interpreter rather than adjourning the trial. The fact that this control or monitoring only occurred when textual mistakes were detected, and that this was only the cases with the French-Spanish (one case) and English-Spanish (one case) language combinations, should shed some light on the risks of miscommunication that may occur with more distant languages, as is the case in our corpus with the Spanish-Romanian interpretation, which presents a much higher incidence of interactional and textual errors. For unjustified non-renditions, it must be noted that they often lead to a monolingual exchange between the interpreter and the defendant, which is never interrupted by the judge. We have annotated unjustified non- renditions as WIA (warning, instructions and advice), ‘advocate’, and ‘assistant’, and as such, in some cases they reveal a paternalistic role toward the defendant, as we have seen in the Spanish-Romanian examples. In fact, the Spanish-Romanian subcorpus is by far the combination with the most non-renditions, which leads us to several hypotheses that could explain this phenomenon. First, the knowledge of the environment can lead the interpreter to want to intervene in order to favour the communication and/or the extraction of information from the defendant. Second, the opacity of the language (some members of the judiciary may speak English or French, as we observed in some hearings, but not Romanian) could give the interpreter a sense of greater freedom when intervening with their own voice. Finally, the fact that the interpreter and the defendant or witness are always members of the same cultural context could also influence the relationship that they establish between one another and could even explain the paternalistic attitude identified in some cases. In the English-Spanish and French-Spanish cases, because our transcribers were unable to establish the accent of the interpreters concerned with any certainty, we were unable to identify the cultural background of the interpreter, unlike the Romanian-Spanish cases. Most of the problems we have mentioned involve a certain degree of opacity in the interpreters’ deviances from their role. For example, if the members of the judiciary are not competent in the foreign languages used in court, they cannot control whether or not the interpreter is rendering the message accurately. However, in the case of non-renditions, and especially unjustified non-renditions, there is often a monolingual exchange that is neither communicated in the language of the court nor interrupted by the judge. Therefore, not only can we confirm that the quality of the interpretation is unacceptable in all of the hearings that we have analysed (none of the hearings in our corpus was without substantial errors), but also that the judges did not effectively manage interpreted hearings. On the one hand, monolingual exchanges were accepted without questioning their appropriateness, and on the other, members of the judiciary were very tolerant of the poor quality of language in general, as well as of a lack of knowledge of more specialised legal terminology. In fact, as illustrated in excerpts 10 and 11 from the FR6 hearing, the only interventions of the members of the judiciary were to correct textual errors committed by the interpreter. In the eyes of the members of the judiciary, the interpreter seems to be a low-qualified professional who intervenes as a “tongue”, like in colonial times. They possess some knowledge of a foreign language but are completely unaware of the communication exchanges that should be facilitated by the interpreter or are ill-equipped to manage them. Court interpreting in Spain has been characterised by its absence of professionalism. In fact, Blasco Mayor (2013) and Blasco Mayor and Del Pozo Triviño (2015) indicate factors such as lack of quality control, training or certification requirements and low fees as the major concerns in this field. Furthermore, as exemplified in excerpt 12, Spanish judges do not seem to grasp even minimally the possible dangers of poor communication. A French witness needs to testify in a case involving a corporate offence. However, the interpreter seems to be incapable of understanding and reproducing the Spanish terminology accurately. To resolve this issue, the Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 136 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts judge relies on the witness’s rudimentary knowledge of Spanish to continue with the oral trial. It is in this case that the judge asks the witness, “I think, you speak some Spanish, right?” Excerpt 12 illustrates how members of the judiciary fail to understand the role of the court interpreter. Likewise, and, perhaps even more disheartening, it reveals a high level of tolerance for low performance on the part of the interpreters hired by Spanish court administrators through intermediary agencies. The way this tolerance of substandard interpreter performance manifests itself reflects a disturbing reality. It does not appear that judges and attorneys are incapable of recognising a poor interpretation, at least as relates to straightforward issues of linguistic incompetence, but rather that they have come to believe that they cannot access better interpretation resources, or they think what they obtain is the standard. It is also clear that members of the judiciary would benefit from specialised training in how to work with interpreters, as well as training in interpreting quality (i.e., what to expect from a qualified, professional interpreter), which would permit them to detect less straightforward issues of inappropriate interpreter protocol and interactions. Ultimately, all these examples illustrate the low level of recognition of the profession of court interpreting in Spain. The most serious consequences of this situation are, as always, for the weaker parties: the defendants or witnesses who do not speak the language of the court. Acknowledgements The English proofreading of the first draft of this article was performed by Jacqueline Minnet and funded by the Department of Translation, Interpreting and East Asian Studies of the Universitat Autònoma de Barcelona. We also thank Dr Anthony Pym, Dr Marta Arumí and Dr Anna Gil-Bardají for their comments on an earlier version of the manuscript. References Angermeyer, Philippe. (2015). Speak English or what? Codeswitching and interpreter use in New York City courts. Oxford University Press. Arumí, Marta & Vargas-Urpí, Mireia. (2018). Annotation of interpreters’ conversation management problems and strategies in a corpus of criminal trials in Spain: The case of non-renditions. Translation and Interpreting Studies, 13(3). Bestué, Carmen. (2018). Aproximación empírica a la labor del intérprete en los tribunales de justicia. In María Jesús Ariza Colmenarejo (Ed.), Derecho de traducción, interpretación e información en el proceso penal (pp. 139-175). Tirant lo Blanch. Bestué, Carmen. (2019). From the trial to the transcription: Listening problems related to thematic knowledge. Some implications for the didactics of court interpreting studies. Fachsprache, Journal of professional and scientific communication, 41(3-4), 159-181. Bestué, Carmen (2021). Tesauro de los procesos penales. Universitat Autònoma de Barcelona. Berk-Seligson, Susan. (1990/2002). The bilingual courtroom: Court interpreters in the judicial process. University of Chicago Press. Blasco Mayor, María Jesús. (2013). Quality of interpreting in criminal proceedings in Spain under European directive 2010/64/EU. Cuadernos de ALDEEU, 25, 165-190. Blasco Mayor, María Jesús, & Del Pozo Triviño, Maribel. (2015). La interpretación judicial en España en un momento de cambio. MonTI, 7, 9-40. Braun, Sabine, & Taylor, Judith. (Eds.). (2012). Videoconference and remote interpreting in criminal proceedings. University of Surrey. Cheung, Andrew K.F. (2017). Non-renditions in court interpreting. A corpus-based study. Babel, 63(2), 174-199. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 137 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Dueñas, Roseann, Vásquez, Victoria F., & Mikkelson, Holly. (2012). Fundamentals of court interpretation: Theory, policy, and practice. 2nd ed. Carolina Academic Press. Gallez, Emmanuelle, & Maryns, Katrijn. (2014). Orality and authenticity in an interpreter-mediated defendant’s examination: A case study from the Belgian assize court. Interpreting, 16(1), 49-80. Hale, Sandra. (1999). Interpreters’ treatment of discourse markers in courtroom questions. Forensic Linguistics, 6, 57-82. Hale, Sandra. (2002). How faithfully do court interpreters render the style of non-English speaking witnesses’ testimonies? A data-based study of Spanish—English bilingual proceedings. Discourse Studies, 4(1), 25-47. Hale, Sandra. (2004a). The discourse of court interpreting: Discourse practices of the law, the witness, and the interpreter. John Benjamins. Hale, Sandra. (2004b). Interpreting culture. Dealing with cross-cultural issues in court interpreting. Perspectives, 22(3), 321-331. Martínez-Gómez, Aída. (2014). Criminals interpreting for criminals: breaking or shaping norms? JoSTrans: The Journal of Specialised Translation, 22, 174-93. Monteoliva-García, Eloísa. (2016). Annotated bibliography on legal and judiciary interpreting. NAJIT. Monteoliva-García, Eloísa. (2018). The last ten years of legal interpreting research (2008-2017). A review of research in the field of legal interpreting. Language and Law / Linguagem e Direito, 5(1), 38-61. Moser-Mercer, Barbara, Künzli, Alexander, & Korac, Marina. (1998). Prolonged turns in interpreting: Effects on quality, physiological and psychological stress (Pilot study). Interpreting, 3(1), 47-64. NCSC-National Center for State Courts. (2011). Guide to translation of legal materials. Consortium for language access in the courts. Orozco-Jutorán, Mariana. (2017). Anotación textual de un corpus multilingüe de interpretación judicial a partir de grabaciones de procesos penales reales. Revista de Llengua i Dret, Journal of Language and Law, 68, 33–56. https://dx.doi.org/10.2436/rld.i68.2017.3034 Orozco-Jutorán, Mariana. (2018). The TIPp project: Developing technological resources based on the exploitation of oral corpora to improve court interpreting. Intralinea, 20. Orozco-Jutorán, Mariana. (2019). A mixed-methods approach in corpus-based interpreting studies. Quality of interpreting in criminal proceedings in Spain. In Lucja Biel, Jan Engberg, Maria Rosario Martín Ruano and Vilelmini Sosoni (eds.) Research methods in legal translation and interpreting (148–165). Routledge. Ortega-Herráez, Juan Miguel. (2011). Interpretar para la justicia. Comares. Pérez González, Luis. (2006). Interpreting strategic recontextualization cues in the courtroom: Corpus-based insights into the pragmatic force of non-restrictive relative clauses. Journal of pragmatics, 38(3), 390-417. Pöllabauer, Sonja. (2004). Interpreting in asylum hearings. Issues of saving face. In Cecilia Wadensjö, Birgitta Englund Dimitrova and Anna Lena Nilsson (Eds.), The Critical Link 4: Professionalisation of interpreting in the community (pp. 39-52). John Benjamins Publishing Company. Russell, Deborah. (2002). Interpreting in legal contexts: consecutive and simultaneous interpretation. Linstok Press. Vargas-Urpí, Mireia. (2017). Court interpreting as a shared responsibility: Judges and lawyers in a corpus of interpreted criminal proceedings. Revista Canaria de Estudios Ingleses, 75, 139-154. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 138 Carmen Bestué, Mireia Vargas-Urpí You speak some Spanish? Indicators of interpreters’ (non-)performance in Spanish criminal courts Vargas-Urpí, Mireia. (2019). When non-renditions are not the exception: A corpus-based study of court interpreting. Babel, 65(4), 478-500. Vargas-Urpí, Mireia (2020). Juicios multilingües en Barcelona desde la perspectiva de la sociología del lenguaje: lenguas dominantes, lenguas minorizadas y lenguas invisibles. Onomázein, 47, 206-224. https://doi.org/10.7764/onomazein.47.08 Vigier, Francisco. (2019). Interpreting in Spanish criminal courts: Preliminary results of the TIPp project’s corpus of real trials. Translation and Translanguaging in Multilingual Contexts, 5(3), 308-320. Wadensjö, Cecilia (1998): Interpreting as interaction. Longman. Revista de Llengua i Dret, Journal of Language and Law, 79, 2023 139