Juror perceptions in bilingual interpreted trials

ABSTRACT

The underlying principle of a jury trial is to be judged by one’s peers. Defendants who don’t share the same cultural and linguistic background as the jurors can hardly be considered peers. Interpreters are needed to bridge the language gap when defendants cannot speak the language of the court; but language is not the sole factor that may impinge on jury perceptions of the defendant. The mode of interpreting, the position of the interpreter in the court, and the location of the interpreter have been shown to make a difference. Whether jury perceptions of the interpreter affect their perceptions of the defendant is yet unknown. This paper reports the results of an experimental study of simulated criminal jury trials in Sydney, Australia, that explored this issue.

KEYWORDS:

• Credibility assessments
• interpreting mode
• English competence

1. Introduction

The underlying principle of a contemporary trial by jury is for defendants¹ to be judged by their peers (Horan 2012), thus making it a fairer process than if judged by a judicial officer² who generally comes from a less diverse, elite social background (Millon 1992). When defendants are not from the same cultural and linguistic background as the jurors, they can hardly be considered peers. Research has indeed shown that those who belong to different outgroups are assessed more harshly by jurors, including being judged as less credible (Shuman, Stokes & Martinez 2011; Culhane, Hosch & Daudistel 2014).

When the defendant cannot speak the language of the court, interpreters are needed to bridge the language gap. Removing the language barrier to place the defendant on an equal footing with one who can speak the language of the court is not an easy task. The accuracy of the interpretation is the first aspect to be considered. Achieving accuracy of the content and the manner of the message to produce a similar reaction in the listener as the original, is very complex, and attainable only by the highest qualified interpreters (Shulman 1993; Hale 2007; Goodman-Delahunty, Martschuk, Hale, Doherty & Taibi 2018). But even with the best interpreters, there are factors that may impinge on jury perceptions of the defendant that will be beyond the interpreter’s control (Gallez & Maryns 2014; Stern, Ozolins & Hale 2015; Hale & Napier 2016). The mode of interpreting (simultaneous or consecutive) (Hale, Martschuk, Ozolins & Stern 2017), the position of the interpreter in the court and the location of the interpreter (present or remote) (Hale, Goodman-Delahunty, Martschuk & Lim, 2022) have been shown to make a difference. Whether jury perceptions of the interpreter affect their perceptions of the defendant or witness who does not speak the language of the court is unknown. Traits such as the gender, ethnicity, appearance and even the language competence of the interpreter may impact jury evaluations of the interpreter, and in turn, of the defendant. This paper explores these issues through the presentation of results of an experimental study of 12 live simulated jury trials in Sydney, Australia. Eight of the trials were bilingual (4 English-Mandarin and 4 English-Spanish), with interpreters.

1.1. Jury perceptions of defendants and witnesses in monolingual cases

The credibility of witnesses who testify in legal proceedings is judged on two main dimensions: (a) their believability which is associated with the demeanour and presentation style of the witness (messenger) (O’Barr 1982), and (b) the plausibility and veracity of the content of the evidence (truthfulness of the message). These features are assessed by the jury or judicial officer based on both the witness’s style of speech, manner of delivery and the coherence of the content. More recent empirical studies of witness credibility distinguished two psychometrically derived factors: Communication Style (verbal and nonverbal skills) and Poise (emotional control) (Cramer, DeCoster, Neal & Brodsky 2013). Evidence from a witness whose testimony is rated more credible is usually accorded more weight, and preferred over that of a witness whose evidence is rated lower in credibility (Goodman-Delahunty, Cossins & Martschuk 2016a; Goodman-Delahunty, Martschuk & Cossins 2016b).

Theoretical dual process models of persuasion and decision-making, such as the Elaboration Likelihood Model (Petty & Cacioppo 1986) and the Heuristic Systematic Model (Chaiken & Trope 1999), have been applied successfully to analyse the way juries assess the credibility of lay and expert witnesses in reaching a verdict (Bornstein & Greene 2011; Freckleton, Goodman-Delahunty, Horan & McKimmie 2016; Goodman-Delahunty et al. 2016a; Goodman-Delahunty et al. 2016b). The models anticipate that at times, juries will be influenced by the message or content of the evidence presented at trial, and at other times, by the messenger or witness, independent of the content of the evidence. Two modes or routes to persuasion are posited, sometimes called Type I and Type II, or central and peripheral routes, each of which entails different amounts of thinking or cognitive effort. The central route requires more effortful thinking than the peripheral route. The central route is used when jurors have the time, cognitive capacity and motivation to apply substantial effort to assess the content of the evidence; the peripheral route is used when a juror is distracted from the message, when the evidence is more complex and difficult to understand, such as expert evidence, or when a juror is tired (Freckleton et al. 2016). The dual process model can be applied to the context in which witnesses speak a language other than English and the witness testimony can be difficult to understand either due to a strong accent or to a distraction through the presence of an interpreter who may interrupt the flow of the interaction between the questioning lawyer and the witness.

Another potential influence on laypeople’s evaluations of witnesses is their accent (Lippi-Green 1994; Sobral Fernandez & Prieto Ederra 1994). Research has shown that different accents can elicit different perceptions depending on the prestige of the native language of the speaker. For example, a US study of undergraduate students’ perceptions of non-native English, showed that English-native speakers perceived accented English of speakers of western European countries more favourably than the accented English spoken by eastern Europeans, Asians or Mexicans (Lindemann 2005). Another US study of the effect of accented English on witness credibility disclosed that Lebanese accented testimony was perceived less favourably than German accented testimony. However, Mexican accented testimony produced no significant differences when compared to German accented testimony (Frumkin 2007). Frumkin speculated that the mock-jurors’ familiarity with Mexican accented English made it more acceptable to them. According to the in-group/out-group theory, people favour members of their own group over members of an out-group (Aronson, Wilson, Akert & Sommers 2015), except for the ‘black sheep’, that is, deviant in-group members who are disfavoured more than comparable out-group members (Marques & Yzerbyt 1988). This in-group/out-group effect extends to the perception and evaluation of witnesses, such as culpability and sentencing judgments of defendants of another ethnicity (Mitchell, Haw, Pfeifer & Meissner 2005). The same bias is likely to occur when witnesses speak with different accents. An example of how accent bias affected juror judgements emerged in a simulated civil negligence trial where black American defendants were rated significantly more negligent than similarly situated White and Mexican defendants. The highest monetary compensation was awarded to a White plaintiff when the defendant was a Black American who spoke with a stereotypical Black accent (Cantone, Martinez, Willis-Esqueda & Miller 2019). Accents can identify speakers with a particular group. When they are identified as members of an out-group, it can generate negative reactions. The opposite can be the case when the accent identifies people as part of an in-group (Bresnahan, Ohashi, Nebashi, Liu & Morinaga Shearman 2002).

1.2. Jury perceptions of defendants and witnesses with interpreters

Some research has pointed to potential biases against witnesses who testify in another language via interpreters, either because of prejudice against their ethnicity (Espinoza, Willis-Esqueda, Toscano & Coons 2015; Minero & Espinoza 2016) or because they may be judged as less intelligent if they need to rely on interpreters (Stephan & Stephan 1986).

For instance, a US study of individual student mock-jurors’ assessments of Spanish and Thai defendants who gave evidence via interpreters, disclosed that non-Hispanic and non-Thai jurors judged Spanish and Thai speaking defendants as more culpable when their testimony was interpreted than when it was presented in English. The study did not describe the characteristics of the interpreters except to say that a ‘courtroom translator is required to speak in neutral tones’ and ‘the inflections of the Spanish speaking defendant were not mirrored in the translation’ (Stephan & Stephan 1986:580). This statement reflects a lack of understanding of the role of competent interpreters and indicates that the interpreting was not accurate, as competent interpreters maintain accuracy of content and manner rather than neutral tones (Gallez & Maryns 2014). Nevertheless, the authors attributed juror assessments to linguistic bias, without any consideration of the effect of inaccuracies in the interpretation. The study showed that in-group membership made a difference, per the in-group/out-group bias: when jurors were Hispanic, their evaluations of Spanish speakers were more favourable. The authors concluded that ‘translation has negative consequences for a defendant who requires a translator if he or she must face jurors from another ethnic group’ (Stephan & Stephan 1986:583). However, the reason for more favourable assessments may have been that the Spanish-speaking jurors could fully understand the testimony in Spanish, while those who did not understand Spanish had to rely on incompetent interpretations. Interestingly, when the judge instructed jurors to ignore the fact that the testimony was translated, the in-group/out-group bias was reduced or eliminated.

A similar study by Hosch et al. (1996, as cited in Mendoza et al., 2000) further revealed that the ethnicity of the jurors impacted assessments of witness credibility. When jurors were from the same ethnic background as the witness, in this case, Mexican American, their assessments of a Mexican witness who gave evidence in Spanish were more favourable when the witness testified in Spanish than in English. However, the authors noted that ratings by jurors of the interpreter were unrelated to the jurors’ assessments of witness credibility.

In a later US study by Mendoza et al. (2000) the speech style of the prosecution witness impacted the conviction rate of the defendant. The more hesitant the prosecution witness, the less credible the jurors perceived the witness to be, which reduced convictions. The opposite was the case when the witness’ speech styles were more powerful. The same witness testified in English in a monolingual condition and in Spanish via a professional court interpreter in a bilingual condition. Again, no association emerged between the individual jurors’ evaluations of the interpreters and their verdicts. However, the authors rightly pointed out that this might be due to the fact that the majority of the jurors spoke Spanish as their dominant language or were balanced bilinguals and were able to understand the witness regardless of the interpretation. They suggested that further research was needed to ascertain the impact of interpreters’ behaviour on jury verdicts.

A more recent US experimental study (Shuman et al. 2011) compared juror evaluations of the same plaintiff in a civil personal injury case in which the language of the plaintiff was varied in three ways. In one condition, mock-jurors were informed that the plaintiff had an Anglo background and spoke English; in a second condition, that the plaintiff had a Hispanic background and spoke accented English, and in the third condition, that the plaintiff with a Hispanic background who spoke Spanish required the services of an interpreter. Mock-jurors drawn from the jury pool in Dallas County, Texas, were randomly assigned to review the same simulated trial video, lasting 10 minutes, in which the testimony of the plaintiff was varied linguistically as described above. The jurors individually awarded a reasonable sum to compensate the plaintiff for his injuries and pain and suffering; they did not deliberate with fellow jurors. Based on previous studies of ethnic bias towards outgroup members, the researchers hypothesised that smaller sums in damages would be awarded to the outgroup plaintiff who testified in Spanish. However, the plaintiff whose evidence was provided via an interpreter received a higher damages award. This was not necessarily a consequence of in-group favouritism from Hispanic jurors, because the majority of the mock jurors were identified as ‘White’. Nevertheless, the authors speculated that the high proportion of women, of college-educated and of Hispanic mock-jurors may have contributed to the positive treatment of the non-English speaking plaintiff, or that this leniency effect among White majority mock-jurors was due to the high prevalence of Hispanic jurors in the Dallas County jury pool. A limitation of this study was the absence of any consideration of the interpreter’s characteristics (i.e. professional background and education) or the accuracy of the interpreting. It also did not specify the mode of interpreting (simultaneous or consecutive), although consecutive interpreting is the default mode of court interpreting in the USA, and the trial with the interpreted evidence took longer than those with monolingual evidence pointing to the likelihood that it was consecutive. Due to these limitations, no inferences can be made regarding the effect of the interpreter on juror evaluations (Shuman et al. 2011). In all the studies described above, neither the quality nor the mode of the interpretation or the characteristics of the interpreters themselves were considered as sources of influence on juror decisions.

An Australian study of the impact of the interpreters’ accented English on mock-jurors’ evaluations of the defendant, showed no negative biases (Hale, Bond & Sutton 2011). The study presented six recordings of interpreted testimony from Arabic, Mandarin and Spanish into English to 224 mock-jurors from Western Sydney, where there is a high migrant population who speak these three languages. Each group of jurors listened to a different recording, which was randomly allocated to jurors who did not speak the language of the witness. The interpretations were scripted to ensure full accuracy. All interpreters had equally high levels of English competence, regardless of accent. Half of the recordings were in non-accented English^3, and the other half were in accented English. When they finished listening to the recording, the mock-jurors were asked to rate the witness on a 5-point Likert type scale on how credible, honest, trustworthy and persuasive they perceived the witness to be. They were also asked to indicate their verdict. The results showed no difference across languages. Interestingly, there were significant differences according to accent. Contrary to previous studies, in this study, those who heard the accented versions were significantly less likely to convict the defendant than those who heard the non-accented versions. Additionally, the accented versions produced more favourable perceptions of the defendant’s credibility, honesty, trustworthiness, and persuasiveness. The authors speculated that the demographic characteristics of the mock-jurors may have explained the results, with 64.2% of bilinguals who possibly spoke accented English themselves. This finding supported the in-group bias theory. Some limitations of the study included the fact that the hearing was brief (10 minutes), presented as an audio recording, and tested a total of four interpreters, one non-accented, and three accented from the three different linguistic backgrounds.

The influence of interpreting accuracy on witness credibility has been investigated through discourse analytical studies of authentic and simulated trials and police interviews. Results have shown that untrained interpreters often fail to reproduce seemingly superfluous non-content features that can be cues to veracity, such as hesitations, fillers, hedges and repetitions (referred to as powerless features) (Berk-Seligson 1990/2002/2017; Hale 2002; Berk-Seligson 2009). Another common inaccuracy is the omission or change of linguistic indicators of reliability, such as vulgar language, pragmatic force or levels of politeness (Jacobsen 2008; Felberg 2016; Taibi & El-Madkouri Maataoui 2016; Hale, Martschuk, Goodman-Delahunty, Taibi & Xu 2020). In one discourse analytical case study (Gallez & Maryns 2014) the interpreter was found to improve the presentational style of the defendant, making him appear more intelligent and socially adapted. The authors argued that jurors were unable to assess the defendant accurately, due to the interpreter’s interference. Quasi-experimental studies using oral recordings of interpreted evidence have shown that when interpreters unwittingly ‘improved’ on the style of the original, by making it more coherent and omitting powerless features, or by adding politeness markers, evaluations of witness credibility were significantly enhanced. The opposite was the case when interpreters added powerless features to the interpretation (Berk-Seligson 1990/2002/2017; Hale 2004/2010).

A large-scale experimental study of 449 Australian mock-jurors’ evaluations of the credibility of a Spanish-speaking defendant yielded no statistically significant differences between jurors who heard the translated evidence via an interpreter and those who heard the evidence directly from the same defendant in accented English (Hale et al. 2017). The trial was scripted and translated by a NAATI⁴ qualified professional translator⁵ to ensure accuracy of content and manner of speech. Professional actors repeated the trial multiple times for each different jury group. The interpreter was also an actor who learned the script and acted her part using two different interpreting modes, consecutive (CI) and simultaneous (SI), for different juror groups. Unlike the US mock-jurors, no mock-jurors in this study spoke Spanish, and they had to rely on the interpretation and their observations of the defendant’s behaviour to assess his credibility. When the jurors were asked how likely they were to convict the defendant, no differences emerged between the SI condition and the monolingual condition. However, jurors who observed the CI condition reported they were less likely to convict the defendant than those who observed the other two conditions. In the SI condition, the jurors heard the interpreter but did not see her, so they were able to concentrate on watching the male defendant while hearing the voice of the interpreter. In the CI condition, the female interpreter stood beside the male defendant and was seen as well as heard. The scripted interpretation was the same in all three conditions. We can speculate that the visual input from the interpreter in the CI condition may have elicited the more favourable response towards the defendant. Physical attractiveness, social status and social respectability have been found to influence jurors’ perceptions of guilt (Salekin et al. 1995; Goodman-Delahunty & Sporer 2010).

Whether the jurors in the study by Hale et al. (2017) judged the defendant based on positive perceptions of the interpreter is unknown. The dual process model described above (Kahneman 2011; Evans & Stanovich 2013) suggests that when testimony is interpreted, it may deflect jurors’ attention from the witness to the interpreter, and this distraction may foster more peripheral route processing, especially in CI, where the presence of the interpreter is generally more intrusive, irrespective of the complexity of the evidence. In CI, the flow of the lawyer-witness interaction is by necessity interrupted by the interpreter, as opposed to SI that is heard while the lawyer or witness speak. With peripheral processing (often referred to as Type 1 or more rapid default processing), the strength of the message content is less important than the association of the messenger with cues that are positively or negatively evaluated, such as the likeability or credentials of a witness and/or possibly the interpreter, rather than the content of the evidence (Type 2 processing that entails distinctive higher order reasoning).

1.3. The present study

To further explore the effect of interpreters on juror evaluations of defendants, we conducted a large experimental study of court interpreting in Mandarin/English and Spanish/English interpreting in a simulated criminal trial in Sydney, Australia. This paper presents the results of part of the study that examined jurors’ perceptions of non-English speaking defendants who gave evidence in Mandarin and Spanish via professional interpreters. Additionally, it presents the results of the jurors’ perceptions of the interpreters. Specifically, the study aimed to answer the following questions: To what extent are jury evaluations of the defendant influenced by:

1. the mode of interpretation?

2. jurors’ perceptions of the interpreter?

3. the interpreting and linguistic performance of the interpreter? and

4. the language of the interpretation?

Based on the dual-process model, Hypothesis 1 predicts that SI will not interfere with the jurors’ evaluations of the defendant, whereas CI will lead to more positive evaluations of the defendant, as found in our previous studies. Similarly, Hypothesis 2 predicts that perceived credibility of the interpreters is associated with perceived credibility of the defendant. Hypothesis 3 predicts that interpreting performance and oral English competence will influence mock-juror perceptions of interpreters and in turn of the defendant. Based on the in-/out-group bias theory, Hypothesis 4 predicts that the language that is typologically closer to the jurors’ language will receive more positive assessments.

2. Method

2.1. Procedure

A call for participants was sent via AUSIT⁶ and NAATI for qualified practising Mandarin and Spanish interpreters. Interpreters were asked to interpret face-to-face in a simulated criminal trial based on drug-possession, in a real courtroom in Sydney, Australia. Fifty-four interpreters were randomly assigned to interpret in the simultaneous or consecutive modes in trials with or without juries. This paper reports results from 12 jury trials, eight of which were interpreted (4: Spanish-English; 4: Mandarin-English), with the remaining four trials in English. In the English language trials, the same two individuals enacted the role of the defendants as in the bilingual trials, comprising monolingual control groups. The trial duration was on average one hour (M = 59.58, SD = 12.94), ranging between 45 and 87 minutes (Monolingual: M = 45 minutes, SD = 0.50; SI: M = 63.5 minutes, SD = 5.45; CI: M = 70 minutes, SD = 12.06).

Two hundred and twenty mock-jurors were recruited via a professional recruitment company through circulation of an advertisement seeking jury-eligible participants for a court study in Sydney. Participants were allocated to one of 12 live simulated trial sessions and asked to complete a pre-trial questionnaire online. In their allocated trial, they were seated with other mock-jurors to observe the proceedings. At the end of the trial, the mock-jurors were randomly assigned to one of two jury groups, each comprising 10–13 jurors. Each group was taken to a separate room to deliberate to a verdict. At the end of the deliberation sessions (up to 90 minutes), each jury submitted its verdict. Finally, each mock-juror completed an individual post-trial questionnaire. This study reports on the jurors’ perceptions of the defendants and the interpreters.

For consistency of delivery, the roles of the judge, expert witness and defendant were performed by professional actors and the roles of the lawyers by law students, all of whom learned the script and were trained to be responsive to the interpreters’ renditions of the speakers’ turns, and to accommodate any necessary changes to the script.

2.2. Design

The study applied a randomised 2 × 3 between-participants design, with the language background of the defendant (Mandarin vs. Spanish-speaking actor) and interpreting mode for the defendant (consecutive vs. simultaneous vs. monolingual mode) as independent variables.

The experiment included 12 trials, each attended by two juries. There were 24 jury deliberations in total. Each deliberation was audio- and video-recorded and then transcribed for analysis.

2.3. Questionnaires

2.3.1. Mock-juror questionnaires

The mock-jurors completed an online pre-trial questionnaire, which included a series of demographic questions. Further, they responded to the Pre-trial Juror Attitude Questionnaire (PJAQ; Lecci & Myers 2008) indicating agreement with statements on a scale from 1 (strongly disagree) to 5 (strongly agree) reflecting their attitudes towards Conviction Proneness, System Confidence, Cynicism Toward the Defense, Racial Bias, Social Justice, and Innate Criminality.

Following the trial, mock juries deliberated until they reached a unanimous or majority verdict, or until they had deliberated for 90 minutes (whichever came first).⁷ After deliberation, individual mock-jurors completed a post-trial questionnaire with a series of questions, namely: whether the defendant was perceived as consistent, reliable, and credible, and was easy to understand. They indicated their agreement with the statements on a scale from 1 (strongly disagree) to 7 (strongly agree).

Using the Observed Witness Efficacy Scale (OWES; Cramer et al. 2013), the mock-jurors assessed the perceived credibility of the defendant and (where applicable) the interpreter, by rating the factors ‘Poise’ (emotional control, anxiety, management, and confidence) and ‘Communication Style’ (verbal and non-verbal behaviours). They indicated their agreement with these statements on a scale from 1 (strongly disagree) to 7 (strongly agree).

Finally, the mock-jurors who participated in the bilingual trials were asked whether they had doubts about the accuracy of the interpreted testimony of the defendant. Again, they indicated their agreement on a scale from 1 (strongly disagree) to 7 (strongly agree).

2.3.2. Interpreter questionnaire

The interpreter questionnaire contained demographic questions about their gender, age, interpreting training and professional experience in interpreting.

2.4. Participants

2.4.1. Jurors

A total of 240 jury-eligible citizens from the greater Sydney area (54.2% female, 45.8% male) participated in the study. Participants were between 18 and 76 years old (M = 53.80, SD = 16.46). Fifty percent of participants had a university degree, 22.9% had a Diploma, 7.1% a trade certificate, and 20.1% completed high school or less. The majority reported they had been called to jury duty before (57.9%) and 17.1% reported they had served as a juror before. The vast majority (90%) reported English as their first language, with most (71.8%) speaking English only and 28.2% reporting another language spoken. Only 2.9% (n = 7) spoke languages relevant to the present study (3 Mandarin, 4 Spanish). Of those who spoke Mandarin, one juror was assigned to a trial with a Mandarin-speaking defendant. Similarly, of those who spoke Spanish, one juror was assigned to a trial with a Spanish-speaking defendant.

2.4.2. Interpreters

Four Mandarin and four Spanish interpreters (5 female, 3 male) between 25 and 74 years of age participated in the eight interpreted trials. All but one interpreter had a university degree, and most were NAATI credentialed⁸ (see Table 1).

Table 1. Demographic information about interpreters.

Trial No.	Interpreting Mode	Gender	Age	A- language	Education in Interpreting	Legal interpreting training	NAATI accreditation	Interpreting experience in years	Court interpreting experience	Other legal settings
Mandarin
5	CI	Male	28	Mandarin	VET diploma	No	Professional	3	Weekly	Monthly
9	CI	Female	20s	Mandarin	Master’s student	Yes	Nil	–	–	–
25	SI	Female	49	Mandarin	Doctoral degree	No	Paraprofessional	9	Quarterly	Monthly
27	SI	Female	32	Mandarin	Master’s degree	Yes	Professional	7	Quarterly	Quarterly
Spanish
17	CI	Male	74	Balanced bilingual	VET Advanced diploma	Yes	Professional	33	Weekly	Monthly
59	CI	Female	72	Italian	BA degree	No	Professional	34	Weekly	Weekly
21	SI	Male	30	English	Master’s degree	No	–	4	Yearly	Quarterly
23	SI	Female	57	Portuguese	Master’s student	Yes	Professional	4	Monthly	Monthly

Notes: Interpreter in Trial 9 assisted by reading the trial transcript, after the scheduled interpreter withdrew from the study. Although interpreter in Trial 23 was a Master’s student, she was a Professional interpreter in Portuguese and Paraprofessional Interpreter in Spanish with many years’ experience. VET = Vocational Education and Training. VET programmes are offered by tertiary institutions that are not part of the Higher Education sector, which cover universities. Degrees are offered by the Higher Education sector.

2.5. Analyses

2.5.1. Analysis of questionnaires

The questionnaires were subject to a series of descriptive and inferential-statistical analyses using the software SPSS Version 27.0 (2020). Specifically, analyses of variance (ANOVA) were conducted to assess the impact of the actors’ ethnic background and the interpreting mode on mock-jurors’ perceptions of the defendant, and if applicable, their perceptions of the interpreter.⁹ Post-hoc analyses were conducted using the Tukey Test. For pairwise comparisons, the effect size Cohen’s d was used, with d = 0.22 considered a small, d = 0.51 a medium and d = 0.83 a large effect size (Cohen, 1988). Correlational analyses assessed associations between (and within) the different measures about the defendant and the interpreters. According to Cohen (1988) r = .11 is considered a small, r = .24 a medium and r = .44 a large effect.

2.5.2. Analysis of interpreting performance

Each interpretation was recorded, transcribed and coded for analyses. Interpreter performances were independently assessed by trained assessors who were NAATI certified interpreters in the relevant language and blind to the experimental condition and the interpreters’ backgrounds. The coders/assessors coded each transcript according to a coding guide developed in previous studies and were trained to apply assessment criteria also used in previous studies (Hale, Goodman-Delahunty & Martschuk 2018), giving each interpreter a mark out of 100. Performance was assessed on interpreting accuracy of content and manner, and other skills, such as the use of the correct protocols, management skills and language competence.

In addition to the assessment of interpreting performance, extracts of the interpreters’ oral English performance were assessed by a trained ESL¹⁰ assessor, using a modified tool to assess grammatical accuracy, vocabulary and idiomatic expression, and phonological control along five levels of performance (see Appendix). Phonological control included features of spoken language such as the articulation of individual sounds, intonation, rhythm, word stress, sentence stress and the effect of the speaker’s first language on intelligibility.

3. Results

3.1. Juror pre-trial biases

Juror pretrial biases were assessed using the Pre-trial Juror Attitude Questionnaire (Lecci & Myers 2008), with a possible range from 1 to 5 and a higher score indicating more agreement with each of the factors. Mock-jurors scored moderately (i.e. just below the mid-point of 3) on each of the factors, i.e. System Confidence (M = 2.78, SD = 0.69), Conviction Proneness (M = 2.81, SD = 0.80), Cynicism toward the Defence (M = 2.91, SD = 0.74), Social Justice (M = 2.85, SD = 0.64), Innate Criminality (M = 2.16, SD = 0.80), and Racial Bias (M = 2.50, SD = 0.58). Mock-jurors’ pre-trial biases did not differ between the experimental groups (p > .05).

3.1.1. Perceived credibility of the defendant

Multivariate analyses of variance assessed whether the defendant’s language and the interpreting mode impacted mock-jurors’ perceptions of defendant credibility. Figure 1 shows perceptions of the defendant’s credibility as a function of his language and the interpreting mode. Analyses showed a main effect of the defendant’s language (Spanish vs. Mandarin; F(7, 227) = 8.85, p < .001, Partial η² = .214, Wilk’s Lambda = .786). Mock-jurors perceived the Spanish-speaking defendant as more consistent (d = 0.72, 96% CI [0.46, 0.99], p < .001), reliable (d = 0.72, 96% CI [0.46, 0.98], p < .001), credible (d = 0.68, 96% CI [0.42, 0.94], p < .001), and easier to understand (d = 0.55, 96% CI [0.29, 0.80], p < .001) than the Mandarin-speaking defendant, showing a moderate to large effect. Similarly, they perceived the Poise and Communication Style of the Spanish-speaking defendant (M = 4.03, SD = 0.94) more positively than that of the Mandarin-speaking defendant (M = 3.16, SD = 1.08, d = 0.86, 96% CI [0.59, 1.13], p < .001), showing a large effect size.

Figure 1. Perceived credibility of (a) Mandarin – vs. (b) Spanish-speaking defendant by interpreting mode.

Analyses showed a main effect of interpreting mode for both languages (F(14, 454) = 3.36, p < .001, Partial η² = .094, Wilk’s Lambda = .821). Mock-jurors perceived the Poise and Communication Style of the defendant as more credible when his testimony was interpreted in the CI mode than in the SI mode (Poise: d = 0.47, 96% CI [0.15, 0.78], p = .002; Communication Style: d = 0.51, 96% CI [0.20, 0.83], p = .001), and monolingually in English (Poise: d = 0.81, 96% CI [0.48, 1.13], p < .001; Communication Style: d = 0.72, 96% CI [0.40, 1.05], p < .001). The perceived Poise and Communication Style did not differ between the SI mode and the monolingual trial (Poise: d = 0.33, 96% CI [0.02, 0.64], p = .080; Communication Style: d = 0.22, 96% CI [−0.09, 0.53], p = .316), replicating previous research findings (Hale et al. 2017).

The interaction between the defendant’s language and the interpreting mode was statistically significant (F(14, 454) = 2.46, p = .002, Partial η² = .070, Wilk’s Lambda = .864) for the perceived consistency of the defendant’s testimony. Although the trial scripts were identical in content for all defendant answers (Spanish, Mandarin and English), mock-jurors perceived the Spanish defendant’s evidence as more consistent than that of his Mandarin counterpart when the testimony was interpreted in the SI mode (d = 1.31, 95% CI [0.83, 1.78], p < .001).

3.2. Juror perceptions of the interpreters

Mock-jurors were asked to rate the credibility of each interpreter in terms of Poise and Communication Style using the OWES instrument. Table 2 presents mock-jurors’ credibility assessments of each interpreter, the objectively assessed interpreters’ interpreting performance and oral English competency, together with demographic information about the interpreters. Figure 2 shows the perceived credibility of the interpreters, as measured by the OWES and their oral English language competency as a function of interpreting language and interpreting mode.

Figure 2. Interpreter English language competency and mock-jurors’ perceptions of the credibility of (a) Mandarin and (b) Spanish interpreters.

Table 2. Mock-jurors’ assessment of interpreter credibility (OWES), interpretation of defendant’s testimony and oral English competence of each interpreter.

		Interpreter			Interpreter Credibility				Interpreter Performance
Trial No.	Interpreting Mode	Gender	Age	A- language	Poise	Communication Style	Overall Credibility		Interpreting Performance	Oral English Competence
Mandarin
5	CI	Male	28	Mandarin	5.57 (0.65)	5.61 (0.76)	5.61 (0.64)		76.8 D	8
9	CI	Female	20s	Mandarin	5.41 (0.77)	5.44 (0.82)	5.41 (0.75)		n/a	9
25	SI	Female	49	Mandarin	4.64 (1.04)	4.59 (1.03)	4.74 (1.00)		62.3 P	7
27	SI	Female	32	Mandarin	5.30 (0.68)	5.24 (0.77)	5.32 (0.68)		87.0 HD	9
Spanish
17	CI	Male	74	Balanced bilingual	5.67 (0.85)	5.72 (0.79)	5.69 (0.82)		75.5 D	11
59	CI	Female	72	Italian	5.12 (0.96)	5.26 (0.98)	5.19 (0.88)		54.0 P	10
21	SI	Male	30	English	5.30 (0.85)	5.19 (0.92)	5.32 (0.84)		65.8 C	12
23	SI	Female	57	Portuguese	5.10 (1.14)	5.09 (1.21)	5.15 (1.14)		60.6 P	8

Note: Interpreting performance is the weighted score out of 100. Oral English performance is out of 12. Interpreter in Trial 9 was a student who volunteered to fill in for an interpreter who withdrew, with the assistance of the translated trial transcript. CI = Consecutive interpreting SI = Simultaneous interpreting; HD = High Distinction; D = Distinction; C = Credit; P = Pass.

Analyses revealed no differences between mock-jurors’ overall perception of Mandarin vs. Spanish interpreters. In other words, both Mandarin and Spanish interpreters were perceived equally credible (d = 0.12, 95% CI [−0.19, 0.44], p = .508). However, interpreting mode influenced mock-jurors’ perception of interpreters. Participants perceived interpreters’ Poise (M = 5.45, SD = 0.84) and Communication Style (M = 5.50, SD = 0.91) more favourably in the CI than the SI mode (poise: M = 5.07, SD = 0.97, d = 0.42, 95% CI [0.10, 0.74], p = .012); communication style: (M = 5.01, SD = 1.02, d = 0.52, 95% CI [0.20, 0.84], p = .002). This effect held when analyses controlled for interpreters’ oral English competency.

3.3. Juror’ perceptions of interpreting accuracy

The mock-jurors who participated in the bilingual trials were asked to indicate individually whether they had doubts about the accuracy of the interpreted testimony. Overall, mock-jurors indicated that they were not skeptical of the interpreting accuracy, as reflected in a mean score of 2.55 (SD = 1.70) out of 7. This was most likely because the vast majority did not speak Mandarin or Spanish and could not discern how well the testimony was interpreted. However, objectively assessed oral English competence was significantly associated with mock-jurors’ perception of interpreting accuracy, such that the more competent the interpreters’ oral English, the less likely jurors were to doubt the interpreted testimony (r = −.27, p < .001).

An analysis of variance investigated whether there were differences between the experimental groups in mock-jurors’ perceptions of interpreting accuracy. Analyses revealed a main effect of language: Mock-jurors who watched the trial with the Mandarin defendant were more skeptical of the accuracy of the interpreted testimony (M = 2.87, SD = 1.92) than their counterparts who watched the Spanish trial (M = 2.22, SD = 1.39; F(1, 155) = 6.06, p = .015, partial η² = .039). However, this effect disappeared when analyses controlled for interpreters’ oral English language competency (F(1, 155) = 0.50, p = .481, partial η² = .003). In other words, the skepticism of the accuracy of Mandarin interpreters seemed to have been related to the poorer oral proficiency in English of Mandarin vs. Spanish interpreters. The differences between the CI and SI were not significant, that is, mock-jurors perceived interpreting accuracy similarly, regardless of the interpreting mode (F(1, 155) = 3.62, p = .059, partial η² = .023. Incl language competency as covariate: F(1, 155) = 2.14, p = .146, partial η² = .014).

The interaction between the defendant’s language and interpreting mode was significant (F(1, 155) = 4.25, p = .041, partial η² = .027), and remained so when analyses controlled for oral English language competence (F(1, 155) = 4.07, p = .045, partial η² = .026). As shown in Figure 3, mock-jurors doubted the interpreting accuracy more in the Mandarin trial when the defendants’ testimony was interpreted in the SI than the CI mode. By contrast, mock-jurors scarcely doubted the interpreting accuracy in the trial with the Spanish speaking defendant.

Figure 3. Mock-jurors’ doubt about accuracy of the interpreted testimony by language. Error bars are 95% confidence intervals.

3.4. Perceptions of the interpreters and the defendant

Correlational analyses were conducted to ascertain whether mock-jurors’ perceptions of the credibility of interpreters and the defendant were associated with each other, and with the defendant’s factual culpability (likelihood of commission of the crime charged). The correlations are in Table 4. Analyses revealed that the perceived credibility of the interpreters was moderately associated with the perceived credibility of the defendant, measured by the OWES factors, Poise and Communication Style (all correlations between r = .23 and r = .25, p < .01). Elevated perceptions of interpreter credibility were associated with elevated perceptions of defendant credibility. However, this effect held only in the CI mode, and was strongest for the perceptions of the defendant’s Communication Style (correlations between r = .24 and r = .25, p < .01; Table 4), while the association between interpreter and defendant credibility disappeared in the SI mode (correlations between r = .16 and r = .17, p > .05; see Table 4). In other words, the perceived Communication Style of the defendant was associated with the perceived credibility of the interpreters in the CI mode where the interpreters were visible to the mock-jurors. By contrast, in the SI mode where the interpreters were not visible because they were hidden from the view of the jury in an interpreting booth, there was no similar association. This finding was consistent with findings in our previous study (Hale et al. 2017), again demonstrating that the SI mode is less intrusive and achieves similar results to those in the monolingual control condition (Table 5).

Table 3. Correlations of mock-jurors’ perceptions of the credibility of interpreters and the defendant.

		1	2	3	4	5
1	Interpreter Poise	–
2	Interpreter Communication Style	.926**	–
3	Interpreter Overall Credibility	.983**	.963**	–
4	Defendant Poise	.249**	.241**	.241**	–
5	Defendant Communication Style	.235**	.244**	.240**	.950**	–
6	Defendant Overall Credibility	.234**	.243**	.234**	.987**	.976**

Note: *p < .05, **p < .01.

Table 4. Correlations of mock-jurors’ perceptions of the credibility of interpreters and the defendant, by interpreting mode.

		1	2	3	4	5	6
1	Interpreter Poise	–	.925**	.981**	.217	.160	.178
2	Interpreter Communication Style	.922**	–	.964**	.190	.166	.178
3	Interpreter Overall Credibility	.983**	.962**	–	.206	.174	.181
4	Defendant Poise	.220	.215	.217	–	.952**	.987**
5	Defendant Communication Style	.248*	.241*	.246*	.943**	–	.975**
6	Defendant Overall Credibility	.221	.221	.222	.987**	.975**	–

Note: SI is above and CI below the diagonal. *p < .05, **p < .01.

4. Discussion

This paper presented the results of a large-scale study that aimed to ascertain the influence of interpreters on mock-jurors’ perceptions of Mandarin and Spanish speaking defendants. Unlike other studies with mock-jurors drawn from undergraduate students, our mock-jurors were recruited from jury-eligible citizens in the general population, with ages ranging from 18 to 76 years, 90% with English as their first language and almost 60% with experience serving on a real jury. The jurors scored moderately on the pre-trial bias questionnaires, with no differences across the experimental groups. In terms of in/out group membership, we can safely say that the mock-jurors were not from the same groups as the defendants.

When a defendant gives evidence via interpreters, the influence of the interpreter needs to be as minimal as possible, for the fair administration of justice. Our study set out to ascertain the extent to which juror evaluations of the defendant were influenced by the mode of interpretation, jurors’ perceptions of the interpreter, interpreting and linguistic performance of the interpreter and the language of the interpretation.

Our first research question examined the influence of mode of interpreting. We hypothesised, based on previous research, that the simultaneous mode would be the least intrusive, producing the closest results to the monolingual control condition, with the consecutive mode producing more favourable results regarding the perceived credibility of the defendant. This hypothesis was confirmed for Spanish interpreting. Results for the Spanish speaking defendant were consistent with findings in previous research (Hale et al. 2017). Mean credibility assessments in monolingual and SI conditions were not significantly different (3.67 vs 3.95), whereas the credibility assessment in the CI condition was significantly higher (4.50) (see Table 5). Importantly, the monolingual and SI conditions produced results that were not statistically different in terms of credibility assessments of the defendant. In our 2017 study, one female interpreter learned the script to ensure total accuracy of interpreting. In this study with eight different interpreters, four per language, with various degrees of interpreting accuracy, we replicated the finding of more positive credibility assessments of the defendant in the CI condition and corroborated that the SI mode is the less intrusive mode. The results, however, were not uniform for both language combinations. In the case of Mandarin, the defendant’s credibility assessments were higher in both modes of interpreting compared to parallel testimony in English without the assistance of an interpreter (3.07 for SI and 3.66 for CI, 2.75 monolingual; Table 5).

Table 5. Mock jurors’ perceptions of the defendant.

Defendant Language	n	Interpreting Mode	Credibility Assessment M (SD)
Mandarin
	39	CI	3.66 (1.10)
	41	SI	3.07 (1.05)
	37	Monolingual	2.75 (0.89)
Spanish
	38	CI	4.50 (1.00)
	41	SI	3.95 (0.88)
	43	Monolingual	3.67 (0.77)

To address the second research question, we compared mock-jurors’ evaluations of the credibility of the interpreters themselves as well as mock-jurors’ perceptions of the accuracy of interpreting. Overall, the results revealed that the perceived credibility of the interpreters was moderately associated with the perceived credibility of the defendant. When asked how credible the interpreters were, the jurors perceived both language groups equally credible. However, the interpreters’ oral English competence was significantly associated with mock-jurors’ perception of interpreting accuracy. The accuracy of interpreters with higher oral English competence was less likely to be questioned by mock-jurors (r = −.27, p < .001). In addition, mock-jurors who attended the trials with the Mandarin-speaking defendant were more skeptical of the interpreting accuracy than mock-jurors who attended the trials with the Spanish-speaking defendant. Similarly, jurors doubted the Mandarin interpreted testimony more in the SI mode than the CI mode. In Spanish, there was no difference in the perceived accuracy of the interpreting between modes. This result might be due to the Mandarin interpreters’ lower level of proficiency in English, as SI requires a higher level of linguistic competence when the message is delivered under time pressure. Overall, our results confirmed hypothesis 2, which predicted that perceived credibility of the interpreters was associated with perceived credibility of the defendant.

However, the answer to research question 3 was not as clear. Inspection of the interpreters’ backgrounds showed that the Spanish interpreters were overall much more experienced than their Mandarin counterparts, whereas assessments of their interpreting performance showed that Mandarin interpreters produced superior results on average (75% for Mandarin vs 64% for Spanish). This may be due to the fact that the native languages of two of the four Spanish interpreters were Portuguese and Italian, whereas the native language of all Mandarin interpreters was Mandarin. However, the Spanish interpreters were more competent in English, with a mean score of 10.25 compared to 8.25 for their Mandarin counterparts. This result is important, because it points to the fact that jurors base their assessments of the interpretation on the interpreters’ English renditions. They are not equipped to judge competence in the other language nor the accuracy of the interpretation, regardless of what they perceive about interpreting competence. One of the criteria in the English competence assessment was the interpreters’ phonological accuracy, which included detection of a foreign accent. Our results suggested that the interpreters’ accented English was negatively associated with jurors’ evaluations of the defendant. Our third hypothesis was therefore partially confirmed. Independent professional assessments of interpreting accuracy did not produce consistent results. Although the Mandarin interpreters were rated more accurate by independent professional raters, their renditions did not produce equivalent assessments from jurors who heard the interpreted versus the monolingual English version enacted by the Mandarin-speaking defendant. Our hypothesis regarding the interpreters’ perceived English language competence was confirmed.

Regarding the difference language made on juror evaluations of the defendant, overall, the Mandarin-speaking defendant was perceived less favourably by mock-jurors than the Spanish-speaking defendant (credibility ratings of 2.75 vs 3.67). This result may be due to a number of different factors, including the English language competence of the interpreters, as discussed above, and the acting skills of the two individuals who played the same defendant role in different languages. For example, the actor depicting the Spanish-speaking defendant was more experienced than the Mandarin-speaking actor, who tended to exaggerate and overact. The influence that an individual actor in a key role can exert was posited to account for findings in a jury simulation study using a full video trial which failed to replicate effects obtained with the same case-facts in a pencil-and-paper version (Gutek, O'Connor, Melancon, Stockdale, Geer & Done 1999). There may be a third reason, which points to our fourth hypothesis. Based on the in-/out-group bias theory, we hypothesised that the language that was typologically closer to the jurors’ languages would receive more positive assessments. Fewer than one in three jurors reported speaking a language other than English, with only two jurors (0.8%) speaking a language relevant to the trial they attended (Spanish in a trial with a Spanish-speaking defendant or Mandarin in a trial with a Mandarin-speaking defendant). Our analyses revealed no significant differences between the monolingual and bilingual jurors’ evaluations. We therefore speculate, that since Spanish is typologically closer to English, an out-group bias may explain the less favourable evaluations of the Mandarin defendant, thus confirming our fourth hypothesis.

5. Conclusion

Our study results confirmed most of our hypotheses, either fully or in part. Importantly, we found that the simultaneous mode of interpreting is the least intrusive and the one that most closely resembles the monolingual trial condition. Our recommendation is therefore to aim to use the simultaneous interpreting mode whenever possible in trials, to minimise the interpreters’ interference on jurors’ assessments of defendants. Highly trained interpreters and adequate simultaneous interpreting equipment would be required for this goal to be achieved, following the example of the International Court of Justice (Stern 2012). Another important finding was that the interpreters’ English competence was associated with jurors’ perceptions of interpreters and of the defendants for whom they were interpreting, regardless of their interpreting performance. This is a very important finding with implications for interpreter education and training as well as for certification regimes.

Our study had some limitations. The comparison between the two languages was problematic due to differences in the performances and appearance of the actors, which was confounded with language. The sample of interpreters was relatively small, eight in total. Further research with larger samples and more language combinations would be beneficial. Nevertheless, this study corroborated previous research on court and police interpreting which demonstrated the benefits of simultaneous interpreting. It showed that overall, compared to a monolingual trial, although interpreters influence assessments of witness credibility, qualified trained interpreters, with high levels of English competence, using the simultaneous interpreting mode did not produce any significant difference in jurors’ evaluations of credibility or culpability. This is a positive result for bilingual proceedings, when adequately trained and certified interpreters are employed in courts that provide them appropriate working conditions.

Ethics approval

The project was approved by the Human Research Ethics Advisory Panel B: Arts, Humanities and Law, University of New South Wales. Ethics approval number: HC17546.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was supported by the Australian Research Council Discovery Project DP170100634. Natalie Martschuk is a recipient of the Griffith University Research Fellow scheme.

Notes on contributors

Sandra Hale

Sandra Hale is Professor of Interpreting and Translation at the University of New South Wales. She is the sole author of The Discourse of Court Interpreting and Community Interpreting and co-author of Research Methods in Interpreting. She is regularly invited to address the judiciary, tribunal members and other legal professionals on how to work effectively with interpreters. She is a NAATI certified conference interpreter with many years’ experience and was the national president of AUSIT from 2014 to 2017.

Natalie Martschuk

Natalie Martschuk is a Postdoctoral Research Fellow at the Griffith Criminology Institute, Griffith University. Her research interests are procedural justice in legal proceedings, investigative interviewing strategies, legal decision-making and witness reliability and credibility. Furthermore, she has a strong interest in quantitative research methods and meta-analyses.

Jane Goodman-Delahunty

Jane Goodman-Delahunty is a Professor of Legal and Forensic Psychology and a Member of the New South Wales Civil and Administrative Tribunal. An experimental psychologist and lawyer, she conducts empirical legal studies that foster evidence-based decisions to promote social, procedural and distributive justice within organisations and the community. Her recent books include Understanding Sexual Harassment (2021, American Psychological Association) and Juries, Science and Popular Culture in the Age of Terror (2017, Palgrave Macmillan).

Julie Lim

Julie Lim is a Research Officer for a number of externally funded research projects in Interpreting and Translation at the University of New South Wales. Her studies have focused on applied linguistics and Asian studies. She is a TESOL practitioner and a certified English language examiner.

Notes

1 The term defendant is used to mean ‘accused’ or ‘defendant’.

2 We use the term ‘judicial officer’ to refer to judges and magistrates.

3 Non-accented English in this study referred to native Australian English accent.

4 NAATI stands for the National Accreditation Authority for Translators and Interpreters, the Australian credentialing body.

5 We use the term ‘translation’ to refer to written translation and ‘interpreting' to refer to oral translation, as per NAATI nomenclature.

6 AUSIT is the Australian Institute for Interpreters and Translators, the national professional association.

7 Results of the deliberations, including verdict, are reported in a separate paper.

8 The expected level of certification for Mandarin and Spanish is the Certified level, which was called Professional level at the time of the study. Nevertheless, some interpreters with the Certified Provisional Level (formerly Paraprofessional level) also practice as court interpreters. The NAATI nomenclature changed in 2018.

9 Separate analyses showed low interdependency of the data due to the jury group deliberations (all intraclass correlations ≤.05). Therefore, no multilevel analyses were performed.

10 ESL = English as a Second Language.

Appendix: Oral English proficiency rating scale adapted from the CEFR Descriptors (Council of Europe 2020)

Table

	Excellent = 4	Very good = 3	Good = 2	Basic = 1	Unsatisfactory = 0
Grammatical accuracy	Maintains consistent grammatical mastery of complex language structures.	Maintains a very good degree of grammatical accuracy; errors are rare and difficult to spot.	Has a good command of language structures and some complex grammatical forms, although some grammatical errors are made.	Demonstrates reasonable accuracy in familiar contexts; generally good control, though with noticeable mother-tongue influence. Errors occur, but it is clear what speaker is trying to express.	Consistently makes grammatical errors.
Vocabulary and Idiomatic expression	Mastery of a very broad lexical repertoire and idiomatic expressions; shows awareness of connotative levels of meaning.	Has a very good command of common idiomatic expressions; and mostly chooses words adequately	Has a good command of common idiomatic expressions; and chooses words adequately, although some unidiomatic expressions are detected and some incorrect word choices are made.	Some command of idiomatic expressions used appropriately. Sufficient vocabulary for familiar topics. Good control of basic vocabulary but errors still noticeable errors with word form and word choice.	Consistently utters unidiomatic expressions and uses incorrect terms.
Phonological control	Can employ the full range of phonological features in English with a high level of control – including prosodic features such as word and sentence stress, rhythm, and intonation – so that the finer points of their message are clear and precise. Intelligibility and effective conveyance and enhancement of meaning are not affected in any way by features of accent that may be retained from other language(s).	Can employ most of the range of phonological features in English with sufficient control to ensure intelligibility throughout. Can articulate most of the sounds of the English language; some features of accent(s) retained from other language(s) may be noticeable, but they do not affect intelligibility.	Can generally use appropriate intonation, place stress correctly and articulate individual sounds clearly; although accent tends to be influenced by their other language, with some effect on intelligibility.	Pronunciation is generally clear enough to be understood. A strong influence from the other language(s) they speak on stress, rhythm and intonation may affect intelligibility. Pronunciation of familiar words is mostly clear.	Pronunciation, intonation and stress at both utterance and word levels can prevent understanding of the message. Accent is influenced by their other language and affects intelligibility.