A field study on cultural differences in South African eyewitness reports

This study provides a unique opportunity to study cultural differences in eyewitness memory reports in a real-life, multicultural context. For a previous study (unrelated to questions of culture), we collected 103 interviews conducted by the South African Police Services (SAPS; Vredeveldt et al., 2015). For the present article, we conducted a qualitative analysis of these interviews to answer the research question:

Our sample of police interviews allowed for a comparison of witnesses from three racial groups in South Africa: Black African, Coloured and White. Of course, race and culture are not completely the same thing (for more on definitions of culture, see Jahoda, 2012; Kashima and Gelfand, 2012). In South Africa, diverse cultural identities have been shaped largely by its history. Apartheid created distinct experiences among racial groups: White South Africans were historically privileged, Black South Africans disadvantaged, and those classified as Coloured often in-between (Houston, 2022). However, this is a simplification. Within each group, there were different languages and religions. Indeed, multifold within-group differences are obscured by the classification into Black, White and Coloured. Colonial and White rule thus added cultural complexity to pre-existing practices and norms. Although White rule ended in 1994 and progress towards equality has been made, the effects of earlier stratification and pre-existing cultures still influence social interactions and opportunities. This contributes to a complex society where national identity coexists with diverse cultural heritages. Nonetheless, Black and Coloured South African cultures can be characterised as more collectivist and White South African cultures as more individualist (Eaton and Louw, 2000; Van Dyk and De Kock, 2004). Thus, although the question of “culture” in South Africa is complex, we contend that a comparison between Black, Coloured and White witnesses will allow for meaningful study of some effects of cultural differences on witness testimony.

The findings from cognitive psychology and social psychology seem to converge on what to expect from eyewitness memory reports in the present study. Given that White South Africans are more individualist than Black and Coloured South Africans (Eaton and Louw, 2000), and may experience fewer barriers to disclosure, we expected that White witnesses would provide more detailed and specific descriptions than Black and Coloured witnesses. Given the expected differences in power distance, we also anticipated that Black and Coloured witnesses would have an acquiescent response style (e.g. reflected in deference and a tendency to agree with the interviewer), whereas White witnesses would be more assertive (e.g. reflected in non-compliance, interruptions and overconfidence).

We re-analysed video-recorded eyewitness interviews collected in 2013 (see Vredeveldt et al., 2015). Approval for the original data collection was obtained from the Department of Psychology Research Ethics Committee (University of Cape Town) and the Federal Bureau of Investigation Institutional Review Board (USA). The re-analysis for the current paper was approved by the Ethics Committee of Legal and Criminological Research of the Faculty of Law at VU University Amsterdam, the European Research Council Executive Agency (ERCEA) and the Federal Bureau of Investigation Institutional Review Board (USA).

Within a data collection period of seven months, police interviewers conducted interviews with witnesses as they would have normally, but asked witnesses whether they would be willing to participate in the research. Our inclusion criterion was that the witness had to be at least 18 years old, based on which we excluded four interviews from the original sample. Our final sample consisted of 103 interviews, with 47 male and 56 female witnesses with a mean age of 35.44 years (SD = 15.19), ranging from 19 to 75 years. It included 42 Black witnesses (people of African descent; ethnically mostly isiXhosa), 24 White witnesses (people of European descent; Afrikaans or English), 36 Coloured witnesses, and one “Indian” witness (person descended from the Indian subcontinent of Asia).

Thirteen SAPS officers (6 male, 7 female) conducted the interviews. They were based at five Facial Identification Units in the Western Cape Province (Cape Town, Mitchell’s Plain, Belville, Worcester, George). Their mean age was 35.75 years (SD = 4.63) and they had an average of 5.42 years’ (SD = 3.70) experience with investigative interviewing. Six officers were White, four Coloured and three Black.

SAPS officers from the Facial Identification Units conducted semi-structured interviews with people who had witnessed a crime committed by an unknown perpetrator (mainly victims, some bystanders). The primary goal of the interview was to construct a facial composite. The interviews took place at a police station, in the witness’ home or at the scene of the crime. The interviews concerned serious crimes, involving armed/violent robbery (48%), unarmed/non-violent robbery (28%), (attempted) rape/murder (20%) and other crimes (3%).

Witnesses had to be at least 18 years old and give permission for their interview to be video-recorded before being included in this research project. Witnesses read and signed an informed consent form. Interviewers used a standardised scene report form to conduct the interview in line with how they normally conduct interviews. This consisted of (1) a description of the event through a free-recall question (“Can you tell me what happened?”), potentially followed by follow-up questions about the event, and (2) a description of the perpetrator through questions about the perpetrator’s appearance and characteristics (e.g. race, gender, scars, jewellery), to construct a facial composite. Composites were constructed with the help of booklets containing a set of predetermined facial features, and specialised software. Interviews had an average duration of 35 min, with on average 9% of the interview time spent on describing the event. About half of the witnesses were instructed to close their eyes throughout the interview, as the original study examined the effectiveness of the Eye-Closure Interview (Vredeveldt et al., 2015). The interview ended with several evaluative questions and a debrief.

Prior to re-analysis of the data for cultural differences, we conducted a focus group with experts and a literature review to develop an overview of potential culture-dependent variables and variations in memory reporting (Weiss et al., 2025). This gave us the necessary theoretical background and sensitising concepts for analysing and interpreting the data and highlighted issues to be addressed.

The raw data for re-analysis consisted of video-recordings (including audio) of the eyewitness interviews, which were transcribed and, if conducted in Afrikaans or isiXhosa, translated into English by professional translators. Transcriptions were used to code the data in Atlas.ti.

To analyse the data, we followed Braun and Clarke’s (2006) six steps of thematic analysis, after carefully considering alternative methods of qualitative analysis. Although formally structured as forensic interviews, the officers who gathered information to construct face composites routinely asked open-ended questions, encouraged narrative elaboration and spontaneously followed up with clarifying probes. Therefore, we believe that the actual practice of face-composite interviews resembles semi-structured interviewing closely enough that thematic analysis remains appropriate. We believe that this provided sufficient flexibility for meaningful themes to emerge organically, consistent with thematic analysis principles.

We familiarised ourselves with the data, generated codes, searched for, reviewed, defined and named themes, and wrote up the findings. The six-step approach is not a strictly linear process, and certain steps may happen at the same time, or be repeated. We used investigator triangulation to enhance credibility (Korstjens and Moser, 2018): each interview was coded by the main coder and a second coder from South Africa (one with Xhosa and one with Afrikaans cultural background). A detailed description of how we completed each of the six steps can be found in the OSF preregistration, along with the coding scheme and the standardised scene report form (10.17605/OSF.IO/3U7Z9).

This section describes the six main themes generated to encapsulate differences between witnesses from different cultural backgrounds in our South African sample. Theoretical and practical implications will be addressed in the discussion.

A consistent pattern of cultural differences was noted in the way that witnesses described the events. White witnesses tended to provide more detail in their descriptions than Black and Coloured witnesses. For the 24 White witnesses, 49 instances were coded as having a high level of detail and specificity, whereas for the 42 Black witnesses, 33 instances were coded as such, and for the 36 Coloured witnesses it was 15 instances.

For White witnesses, 46 cases were coded as unsolicited, meaning that the answer was extremely detailed (sometimes superfluously), whereas for Black witnesses, only 21 cases were coded as unsolicited and for Coloured witnesses only 17 instances. Unsolicited detail and specificity was coded separately from a high level of detail and specificity. Interestingly, very long responses were found more often in Black witnesses (25 times) than in Coloured (6) or White (3) witnesses. Thus, higher levels of detail and specificity were not per se related to the length of the response. To illustrate the difference, one highly detailed description by a White witness and one brief description by a Black witness are provided in the Supplementary Materials, Appendix 1.

The difference was particularly apparent for descriptions of time of day and people’s height. White witnesses were more likely to use specific clock times to describe when events transpired (e.g. “20:10 in the evening”, interview 6, White female), whereas African and Coloured witnesses were more likely to refer to the general time of day (e.g. “one Sunday afternoon”, interview 5, Coloured female). Likewise, when describing the height of perpetrators, White witnesses typically did so in metric or imperial terms (e.g. “6 foot 1”, interview 11, White male), whereas Black and Coloured witnesses tended to describe height comparatively, using themselves or the interviewer as a reference (e.g. “Taller than me, his head was just above me”, interview 14, Black female). White witnesses also sometimes described height comparatively, but Black and Coloured witnesses almost never used metric or imperial terms. For instance, a White female witness in Interview 3 described the height of the perpetrator as: “About 1.78 metres. He was slightly taller than me”. By contrast, a typical example for Black and Coloured witnesses is the response from a Coloured female witness in Interview 55: “Taller than me. Tall”. White witnesses used numbers for height more than twice as often as Black witnesses, while Black and Coloured witnesses compared the height of a person much more often (35 times) than White witnesses (10 times).

Police officers would often look for the perpetrator’s height in metric terms, which could not always be answered by Black and Coloured witnesses, as seen in this exchange:

Moreover, the cultural differences in describing height sometimes resulted in miscommunication. Specifically, on several occasions, when police interviewers followed up a relatively vague answer with a closed question about the perpetrator’s height in metric terms, Black and Coloured witnesses obliged by providing an answer in metric terms that seemed speculative, or that led to guesses by the police officer, as seen in this exchange:

Black witnesses’ preference for relative and qualitative descriptions was also apparent in their descriptions of facial features, complexion and build, among other things. One possible explanation for this could be that the perceptual focus of people from collectivistic cultures is more attuned to others, and thus exists in “relative-to-others” terms (e.g. I am taller than him). In contrast, people from individualistic cultures may draw more on non-social measures, such as the metric or imperial system of measurement.

It is important to note that based on these field data, we cannot know whether the more detailed descriptions were more, less or equally accurate compared to the less detailed descriptions. In fact, one explanation for the higher number of details could be that White witnesses had a more lenient reporting criterion, meaning that they may have felt more comfortable or confident to report details even if they were not sure about their accuracy.

Black witnesses appeared to be less confident in their memory than Coloured and White witnesses. This was often expressed as doubt about certain aspects of what they saw. One Black man (interview 30) said: “I didn’t check nicely if he has marks. But he might be having a mark but I didn’t check because after I just feel very shocked and I start to running”. They also often stated that they could not remember something, such as the Black woman (interview 64) who said: “I can’t remember the actual day, but I was on my way to work”. Many Black witnesses also stated that they felt unsure about something, e.g. “I’m not sure about the eyes” (interview 46, Black male).

White witnesses, on the other hand, more often expressed certainty and confidence in their memory, such as a White woman (interview 15) who stated: “Because I can remember him very well. If I can see him, I can recognise that person”. They used words conveying confidence and certainty, such as “definitely” (e.g., interview 21 and 22, White men) or “I’m sure” (interview 33, White woman). Coloured witnesses expressed less doubt than Black witnesses but also less certainty than White witnesses, thus falling somewhere in-between these two groups. As mentioned before, because we do not know how accurate witnesses’ descriptions were, we cannot conclude whether witnesses’ confidence, or lack thereof, was justified or misplaced.

In addition, White and Coloured witnesses corrected themselves more often than Black witnesses. For example, a White woman (interview 3) said: “And that’s when he said to me, if you make a sound I’ll shoot you, please give me your phone. He didn’t actually say please, he just said, give me your phone”. Thus, the increased certainty expressed by White witnesses was accompanied by an increased tendency to correct their own statements, which will be addressed in the Discussion.

Witnesses sometimes used terms and expressions that were unique to their own culture or, in some cases, subculture. This is illustrated in the following exchange between a Coloured male interviewer and witness about the perpetrator’s earrings and hair:

These terms and phrases seem to be understood by members of the same cultural group, as was the case in interview 63, but are likely to either be misunderstood or not understood when the witness and police interviewer are from different cultures. Furthermore, as these terms are so commonly understood within the culture, witnesses at times appeared not to realise that the terms might not readily be comprehensible to those outside the (sub)culture. In some cases, the interviewers had to check if they correctly understood the meaning of the slang word, for example:

We also observed that especially Coloured and Black females used euphemisms when they reported sexualised violence, for example:

The witness does not use the word “rape”, but instead describes it vaguely as “being busy with me”. Similarly, a Black woman (interview 74) said: “He took me to this place and he took my clothes off and did what he did”.

When interacting with police interviewers, White witnesses were more likely than Black and Coloured witnesses to assert themselves and their views and to direct the interaction. Specifically, they were more likely to interrupt the interviewer, to ask for changes to be made to facial composites, to challenge the accuracy of various aspects of the composites being compiled, or to question certain processes or actions. The following exchange between a White female witness and a Black male interviewer illustrates this trend:

In contrast, save for a few exceptions, Black and Coloured witnesses tended to adopt a more deferential and acquiescent attitude towards the police officer. For example, a Coloured male witness (interview 40) said: “Ja. I think that’s right like that sir. Can you make his hair a little bit […] thinner? Is his hair thinner like that now sir?” Another Coloured male witness (interview 61) said: “The mouth […] 8 or 9. You can decide. I really can’t say between 8 or 9. I’m going to say 8.5 so you can decide which one”.

The acquiescent style can also be seen in the following exchange, in which a Black male witness did not contradict the Coloured male interviewer who wrongly summarised what he had just said, while still reiterating his own point:

When recounting the crimes they experienced, several Black and Coloured witnesses adopted a defensive tone, offering explanations or justifications for why they became victims. This tendency is exemplified in the case of one Coloured male witness (interview 107), who gave a justification for why he was drawing money, which resulted in him being robbed: “Because when we withdraw some money, and there’s someone in front who also draw money, you won’t think anything will happen now. […] W: Because he had money in his hand, he just draw money now”.

Even as the crime is happening, witnesses may feel the need to defend their victimhood. For example, a Black woman (interview 14) who was mugged and assaulted reported: “Everyone thought maybe I know that guy and we are together maybe, maybe we are having a fight. I remembered men asking me: “What’s going on?” and I just said: “I don’t know the guy!””

In contrast to most women, men from all cultural groups also often offered explanations for why they did not resist or fight back. A Coloured man said:

Others gave similar reasons. Although gender differences were noted, we did not observe consistent cultural differences in this tendency to justify not fighting back.

One of the most consistent differences that emerged was that White witnesses were more likely than witnesses from other racial groups to describe Black and Coloured perpetrators in formal terms (potentially to avoid coming across as racist, as will be addressed in the Discussion). Specifically, White eyewitnesses sometimes referred to their attacker as a “lady” or “gentleman”; something which almost never occurred among witnesses from other racial backgrounds. For example, when asked about the race of the perpetrator, a White male witness in interview 48 stated: “My opinion of the guy was a Coloured gentleman with dark skin”. Likewise, in recounting the crime that he had experienced, a White man explained:

Our field sample of 103 South African police interviews provided a unique opportunity to explore how cultural differences observed in memory experiments translate to memory reports from real eyewitnesses in a multicultural setting. Here, we review the six identified themes in light of the South African historical context and concepts such as trust in the police and the role of gender.

Cross-cultural psychologists have long highlighted cultural differences in memory and reporting (see Wang, 2021, for an overview). A consistent finding is that people from individualist cultures provide more detailed and specific memory reports than people from collectivist cultures (De Bruïne et al., 2023; Wang, 2021). We also observed this in our field data: White witnesses – who are more likely to have an individualist orientation – tended to provide more detailed and specific descriptions than Black and Coloured witnesses (Theme 1). In addition, White witnesses were more likely than Black and Coloured witnesses to use numeric values to describe concepts such as time and height. This mirrors observations in the international criminal law literature that people from traditional African cultures are not as accustomed to using numeric values to describe such concepts (e.g. Combs, 2010). Cultural differences in detail and specificity can be explained in light of different communication styles and goals during police interviews (e.g. Beune et al., 2011, 2009; 2010). For example, a contrast between low- and high-context communication styles (Hall, 1976) may result in a mismatch between communication goals of the interviewer (gathering information) and the witness (maintaining social relationships).

An unexpected theme arising from our data was that Black witnesses tended to express more doubt in their memory while White witnesses tended to express more certainty, with Coloured witnesses falling in-between (Theme 2). Research on memory distrust shows that some people are more sceptical about their memories than others (Nash et al., 2023; van Bergen et al., 2010), but little research to date has been conducted on cross-cultural differences in memory distrust (Zhang et al., 2023). Our findings suggest that this would be a fruitful area for further research. In addition, we found that the witnesses who were more certain in our sample (i.e. White and Coloured witnesses), were also more likely to spontaneously correct their own statements. This suggests a relationship between memory confidence and spontaneous corrections. One potential reason could be that the initial expressions of confidence were misplaced, but because we do not know how accurate witnesses’ descriptions were, we cannot draw conclusions about this. This also warrants more research.

Our analysis revealed that both witnesses and interviewers sometimes used language specific to their (sub)cultural group (Theme 3), such as “skyf” (for cigarette) or “mattress hair” (messy hair after just getting out of bed), which are unlikely to be understood by people from different (sub)cultures. A recent review of the role of culture in eyewitness testimony found that language was the most prominently discussed aspect in the literature (Drenk et al., 2024). This emphasis on language may be because it represents the most tangible manifestation of cultural influences. When testimony involves culturally sensitive topics such as sexual acts, taboos, linguistic codes, and language norms, witnesses may face barriers in providing information important to police investigations. This phenomenon is exemplified by Theme 4’s findings on hlonipha (Thetela, 2002), showing that female victims’ reports were hindered by the common practice of respectful language in certain African cultures. It is important for interviewers to be aware of such cultural and linguistic norms, but awareness alone is not sufficient to address the problem. More research is needed to explore how interviewers can facilitate witnesses to provide the required information while respecting their cultural identity, norms, and boundaries.

Because collectivism is highly correlated with power distance (Basabe and Ros, 2005; Ghosh, 2011), we expected higher power distance in Black and Coloured communities in South Africa than in White communities. This expectation was supported by our field data. White witnesses were more likely to adopt an assertive and directive tone toward the interviewer, whereas Black and Coloured witnesses adopted a more deferential tone (Theme 4). This deference was also seen in terms of acquiescent response style (e.g. Cheung and Rensvold, 2000). Specifically, Black and Coloured witnesses were more likely to agree with the interviewer, even when the interviewer incorrectly summarised their response. Interviewers should be cognisant that acquiescent response styles, particularly when coupled with cross-cultural disparities in the use of numerical values to describe concepts such as height and time, may lead to inaccuracies in witness testimony. This underscores the importance of an open-ended information-gathering approach, as opposed to a more directive approach (cf. Méndez et al., 2021).

Unlike racial groups in other countries that have historically been discriminated against (Kearns et al., 2020; Panditharatne et al., 2021; Weitzer, 2014), Black and Coloured South Africans report greater trust in the police than White South Africans (Carter, 2011; Fry, 2013). This may seem at odds with our finding that Black and Coloured witnesses seemed to feel a greater need to justify or defend their victimhood towards the police interviewer (Theme 5). However, research shows that trust in the police is affected by many factors besides race, such as victimization (which most witnesses in our sample had experienced) and negative previous experiences with the police (Dlamini, 2020; Olutola and Bello, 2016). The observed defensive stance among Black and Coloured witnesses may reflect broader societal issues rather than specific distrust in the police. For example, it could stem from awareness of racist assumptions about crime in society or from generally lower levels of social trust (see Posel, 2022).

Our findings need to be considered in the complicated cultural context in which they were collected: post-Apartheid South Africa. Even though Apartheid was abolished over 30 years ago, its effects are still evident in South African society. For example, Black and Coloured South Africans are much more likely than White South Africans to live in poverty and have limited access to education and higher-ranking jobs (McKeever, 2024). Therefore, race is closely intertwined with class in South Africa, and we cannot isolate the relative contributions of race and class in our findings. Furthermore, the lingering effects of Apartheid were evident in the unexpectedly formal language employed by White witnesses when describing perpetrators from different racial groups, even in cases where they had been victims of serious crimes (Theme 6). Due to the recent history of Apartheid, White people (particularly with an Afrikaans background) often feel the need to outwardly distance themselves from racist beliefs (Matthews, 2012; even if some of them privately still hold racist notions; Verwey and Quayle, 2012). Our observation of seemingly exaggerated conversational politeness seems to align with this need.

Although our sample of police interviews provided a unique opportunity to study cultural differences, there were also some drawbacks. Most notably, field studies provide less control over conditions than experimental studies. This resulted in an unequal number of witnesses from different cultural groups (ranging from 24 to 42 witnesses per group), though with a sufficient spread to conduct a meaningful comparison. In addition, the interviews in our sample were conducted with the goal of producing a facial composite construction, which meant that most witnesses had already been interviewed about the crime in a previous interview (which was not recorded) and that the goal of the interview was primarily to gain as much information as possible about the appearance of the perpetrator, rather than about the crime. Even though nearly all witnesses did provide descriptions of the crime, they might have done so in more detail if it had been an interview intended to gather facts about the crime. Nonetheless, the description of perpetrators resulted in rich data that could be used to answer our research questions. Moreover, the field context meant that locations varied, with interviews taking place either at the police station, at the witnesses’ homes, or at the crime scene. This might have had implications, for instance for how comfortable a witness felt, which might, in turn, have impacted their statements.

Despite these limitations, our South African field data provide a unique contribution to the literature, which is currently dominated by experimental studies from Western, Educated, Industrialised, Rich and Democratic (WEIRD) societies (Henrich et al., 2010). We encourage researchers to analyse data from investigative interviews conducted in non-WEIRD countries to identify patterns that occur across cultures, and patterns that are culturally specific.

Of course, witness reports about a crime are not provided in isolation. An important area for further study is the interaction between the interviewer and the witness. Some influences of the interviewer were already apparent from the current analysis, such as the deference and acquiescence towards the interviewer (Theme 4). In future work, we plan to conduct a more in-depth analysis of the interviewer’s contribution and the interviewer–witness interactions in our sample.

While the article mainly focused on differences between cultural groups, given that practitioners and researchers need to be particularly aware of such differences in cross-cultural eyewitness testimony, it is important to note that there are other aspects in which the cultural groups did not differ from each other. For example, we did not find cultural differences in how (often) emotions were expressed or the use of minimization or maximisation when describing what happened to witnesses and the impact this had on them.

Finally, as highlighted by the intersectionality approach (Crenshaw, 1989), different aspects of a person’s identity may intersect to create a unique experience. For example, Black women in South Africa are disadvantaged in ways that go beyond simply being Black or being a woman (Allen, 2018; Ndinda and Ndhlovu, 2022). Furthermore, there is notable diversity in perspectives on gender roles among South African cultural groups. Some researchers note that Black African communities continue to uphold traditional gender norms that tend to ascribe more submissive roles to women and more dominant roles to men (Booysen and Nkomo, 2010; Strebel et al., 2006). Our sample included male and female interviewers and witnesses from different racial groups, but we did not have enough unique intersecting identities to conduct a full intersectional analysis. Further research is needed to understand the complex interactions of witnesses and interviewers of diverse genders and cultural backgrounds.

The authors are grateful to the ERC Starting Grant 2018 for funding awarded to Annelies Vredeveldt as part of the larger project ‘Beyond WEIRD Witnesses: Eyewitness Memory in Cross-Cultural Contexts’ (no. 802080), with the Vrije Universiteit Amsterdam as host institution. They also thank Eva van Rosmalen for all the work she did as a research assistant. Additionally, they thank the members of the expert panel for their valuable input, which they used in the construction of the coding scheme.

The supplementary material for this article can be found online.