Racial categories and visual descriptions in police emergency calls

Abstract

Taken‐for‐granted presumptions about race are incorporated into the operating logic of many institutions. In emergency police calls, racialization is routinely used to accomplish visual descriptions of persons of interest. This use of racial categories relies on and reproduces inferences about race as visually discernible. These inferences are bound up in colonial histories and ongoing racist violence and discrimination. Our discursive psychology study documents how participants make links between racial category membership and visual appearance and orient to racialization as a delicate but necessary part of an emergency call. We thus show how examining racialization can illustrate the workings of racial inferences even in the absence of overt orientations to prejudice. Understanding how racialization is built into the institutional infrastructure provides an important ‘upstream’ perspective on issues of racialized policing and is a necessary first step in moves towards social justice.

INTRODUCTION

This paper examines the routine ways race is built into institutional settings. We analyse police emergency calls in which racial categories are routinely used to accomplish descriptions of persons. In this context, racialization—commonsense links between racial categories and racial inferences (Shrikant & Sambaraju, 2024)—is integrated into the institutional infrastructure. Call‐takers solicit descriptions from callers, which are recorded in case files and passed to officers so they can identify relevant people when they arrive at the scene. Although participants largely treat this as a legitimate part of the institutional business, our discursive psychology study highlights how these routine sequences rely on racial inferences about physical appearance that are implicated in colonial legacies that continue to manifest in racialized practices of policing.

Discursive psychology research on race and institutional interaction has largely examined how ‘possible ‐isms’ trouble institutional norms of neutrality or broader social norms of antiracism (Whitehead, 2015). In this body of work, the focus is how ‘lay’ participants use race talk and how institutional representatives respond. One of the challenges for institutional representatives is managing competing institutional goals such as neutrality and solidarity (e.g. Tadic, 2024). For example, when callers make racist comments, mediators may depart from the institutionally mandated impartial stance to admonish them (Stokoe, 2015) and radio hosts may sanction callers who use racial ‘stereotypes’ (Whitehead, 2018). These studies highlight the dilemmas that responding to race talk can pose in some institutional contexts.

However, there are other institutional contexts where race talk routinely occurs without participants treating it as problematic. One example is when institutional representatives use racial categories during routine institutional tasks. Many institutions operate bureaucracies that rely on standardized forms as a ‘technology of person production’ (Cahill, 1998) that ‘commonly characterize individuals as aggregations of standardized features or properties—such as race, age and sex—that can distinguish all members of a population’ (Kameo & Whalen, 2015, p. 216). Recording such details in emergency calls is crucial for the organization of assistance. Despite this importance, our study is the first examination of racial categories in emergency calls in Aotearoa New Zealand. We focus on how participants use racial categories and racial inferences to accomplish descriptions of persons in these high‐stakes institutional settings with a novel focus on participants' orientations to the delicacy of such a task.

Racial categories, racialization and racism

Using racial categories in social interaction mobilizes racial inferences about what members of racial categories are like, how they behave, and what they look like. These racial inferences ‘are inseparable from racialization, or the sociopolitical process through which racial categories were developed for purposes of justifying oppression and inequality, and within which race categories have been and continue to be used’ (Shrikant & Sambaraju, 2024, p. 376). In our data, participants use racial categories to accomplish descriptions of physical appearance. In doing so, they mobilize inferences about racial categories members' physical appearance or phenotype. Although descriptions are largely produced as routine parts of the institutional business, inferential links between racial categories and phenotype have a long and troubled history. Racial categories classify people according to supposedly biological differences and are rooted in European imperial‐colonial projects (Shrikant & Sambaraju, 2024). Thus, even when treated as routine and legitimate institutional business, using racial categories to accomplish physical descriptions mobilizes inferences embedded in moral hierarchies.

We take a discursive psychology approach to examine how the routine use of racial categories in person description sequences is tied to broader racial inferences and commonsense knowledge. This approach draws together insights from ethnomethodology, conversation analysis and membership categorization analysis to examine how the psychological world unfolds in social interaction. Although racialization is inherently connected to racism (Shrikant & Sambaraju, 2024), discursive psychology provides a way to examine how participants use such logics without making claims about whether individual speakers are racist or prejudiced (Whitehead & Stokoe, 2015). Indeed, participants orient to many of the question‐answer sequences as unproblematic. Our focus is how such routine sequences rely upon and thus make visible processes of racialization and racial commonsense in the service of relevant institutional tasks. Although at one level, soliciting and recording details about someone's appearance is a routine part of police practice, understanding how these routine practices contribute to racist systems of oppression is an important first step for social justice (Shrikant et al., 2022).

Participants' orientations to past and present sociopolitical structures accountably shape how they build and interpret actions (Tam & Rafaely, 2025). In our context, the fundamental sociopolitical structure is Te Tiriti o Waitangi, which was signed between Māori (the indigenous people of Aotearoa) and the British Crown in 1840. Te Tiriti is foundational to the narrative of a bicultural nation (see Tuffin, 2013) and the myth that colonization in Aotearoa was somehow more humane than other sites of British imperialism (May, 2023). In fact, settler colonialism in Aotearoa is a local instantiation of a global network of white cultural imperialism that opposes indigenous sovereignty (Awatere, 1984).

The criminal justice system starkly reflects the institutional racism of the continuing colonial project. Although there are important distinctions between the experiences of Black people in the United States and Māori in Aotearoa New Zealand, the criminal justice system is similarly racialized (see Dawson et al., 2023). Māori are disproportionately subject to surveillance by police (Norris & Tauri, 2024), are seven times more likely than non‐Māori to be subject to police use of force (New Zealand Police, 2019) and are incarcerated at rates grossly exceeding population levels (Aho et al., 2022). These disparities are present‐day manifestations of colonial policies of land dispossession, genocide and state control of Māori (Dawson et al., 2023; Norris et al., 2023). The links between white cultural imperialism and racialized policing is also evident in the poor treatment and discrimination towards other racial minority groups including Pacific (Asafo, 2022), African (Nakhid, 2018) and Asian peoples (see Hendy, 2021).

Delicacies and dilemmas in talking about race

Interactional research has documented how racialization is used and reproduced as part of the largely taken‐for‐granted categorical infrastructure of social life (Shrikant & Sambaraju, 2024; Stokoe et al., 2025). Understanding the social organization of racial categories is important not only for understanding the workings of racist discourse, but for identifying possibilities for anti‐racism (Shrikant, 2022; Whitehead & Lerner, 2009). Analysing racial categories in situ can reveal racial commonsense knowledge, such as which racial category members are responsible for educating others about racism (Shrikant, 2022) and which racial categories are treated as invisible (Whitehead & Lerner, 2009). However, links between racial categories and category features are not pre‐determined but constructed and contested by participants for different ends in social interaction (see Sambaraju, 2021).

Although racial categories can be used as ordinary ways to identify self and other, participants often orient to race talk as delicate. One common orientation to delicacy is to avoid or delay explicit references to race or racism (Durrheim et al., 2015). Other ways to mark delicacy include pre‐emptive accounts and selecting words like ‘nationality’ as euphemisms for race (Wilkinson, 2011). By contrast, referring to race explicitly can be a way to make racism inferentially available. For example, news reports of police shootings that use racial categories construct racism as an explanation for police violence (Shrikant & Sambaraju, 2021).

As in other settler‐colonial societies, understandings of race in Aotearoa are blurred with national identity (Cormack & Robson, 2011). For example, the census category ‘New Zealand European’ conflates ethnicity and national identity, delineating it from other categories like Māori, Samoan, and Chinese (Broman, 2018)¹ and terms like ‘Kiwi’ or ‘New Zealander’ are most often used to refer to white people (Revell et al., 2014). In commonsense ways of talking about race relations, the ethnic majority is unmarked, and colonization is relegated to the past (Nairn & McCreanor, 1990; Wetherell & Potter, 1992). However, categories like Pākehā (the Māori term for descendants of British settlers) or Tauiwi (non‐Māori) can emphasize colonization as an ongoing process (Kirkwood et al., 2005). This sociohistorical context shapes which categories participants use and which category relationships participants treat as taken‐for‐granted (Shrikant & Sambaraju, 2024). The increased scrutiny of policing in the wake of Black Lives Matter has drawn renewed attention to the dynamics of race in institutional interaction. In Aotearoa, there is clear evidence of structural discrimination in the criminal justice system, yet claims of racism are routinely denied (Workman, 2011). Police and media framings construct the issue as unconscious bias (Webb & Deckert, 2023) while activists and critical scholars point to the settler‐colonial legacies of racialized violence (Aho et al., 2022; Dawson et al., 2023). However, in both public and scholarly debates about police reform, interactions between callers and call‐takers are largely overlooked (Gillooly, 2020). We draw attention to the importance of these interactions and show one way race is incorporated into the routine processes through which emergency police calls operate.

Racialization in emergency police calls

Although the organization of emergency call interaction is well understood (see Kevoe‐Feldman, 2019), the ways race is incorporated into the institutional infrastructure has been less examined. Emergency calls are high‐stakes, time–critical interactions organized around the core goal of securing emergency help. Emergency services are a limited resource and unlike other institutional contexts (e.g. customer service), not all requests are granted. Call‐takers are institutional representatives who gatekeep access to the service, assessing the legitimacy, urgency and relevance of callers' requests (Heritage & Clayman, 2010). Call‐takers are accountable to other members of the institution for the details they record (or not) in the casefile (see Whalen & Zimmerman, 1998) which is then animated by dispatchers who assign officers to attend (Clayman & Kevoe‐Feldman, 2023).

In police emergency calls, participants routinely use racial categories during the interrogative series. This is a phase of the call where call‐takers gather information to determine if or how help can be provided (Whalen & Zimmerman, 1987). As well as details about the location of the incident and what has occurred, call‐takers regularly solicit visual descriptions of persons of interest. Racial categories are a key resource for soliciting and producing such descriptions. A small number of studies from the United States have investigated how call‐takers request and record details of persons. Kameo and Whalen (2015) found that call‐takers predominantly used yes/no questions (e.g. is he a white male?) to solicit descriptions that fit the constraints of the system, whereas Garcia (2022) and Hill (2021) documented a broader range of practices including listing racial/ethnic categories, candidate confirmations and open‐ended requests for descriptions. We extend this work with the first analysis of racial categories in Aotearoa New Zealand emergency calls. We additionally examine participants' orientations to delicacy during these routine sequences and show how some callers resist the project of providing a description (and the uses of racial categories pervasively embedded in it).

DATA AND METHOD

The data are drawn from a larger corpus of emergency calls recorded by New Zealand Police in 2020 and categorized by call‐takers as ‘family harm’ (an institutional code that includes family violence). Our university ethics committee and the Police Research Panel stipulated stringent ethical protocols. The second author and a research assistant manually edited sound‐files at police headquarters to replace identifying information (e.g. names, addresses, contact details) with white noise. After checking by a member of Police, de‐identified sound files were released to the research team. We have progressively transcribed calls using Jefferson's (2004) conventions to capture features like overlapping talk, laughter, crying and other aspects of speech delivery (Hepburn & Bolden, 2017).

Unlike other kinds of emergencies, family harm incidents typically involve people who are known to each other. This creates complex epistemic and relational entanglements. Callers may have greater knowledge of the people they are asked to describe, but such knowledge does not necessarily make the task easier. Callers with coexisting institutional and relational category memberships face a dilemma in seeking help from police (Tennent & Weatherall, 2025). Under one category membership (e.g. victim/help‐seeker), calling the police can be evaluated as the right thing to do to get help, but under the other (e.g. family member), it is the wrong thing to get a loved one in trouble with police. These relational dynamics create a unique context within which to examine links between racial categories and visual descriptions.

This paper reports on a subset of 25 calls where call‐takers ask callers to describe the person of interest. In the majority of these cases (n = 18), racialization occurred as a routine part of accomplishing such a description. Our discursive psychology approach combines an analytic focus on the sequential organization of question–answer sequences (Schegloff, 2007) and how membership categories, category features and commonsense knowledge are deployed and negotiated (Stokoe et al., 2025). For each case, we analysed the turn design of call‐takers' requests for descriptions and callers' responses, including how presuppositions, epistemic gradients and action agendas were conveyed, supported or resisted (Hayano, 2013; Heritage & Clayman, 2010). We explored where and how categories were explicitly referenced and how commonsense links between category membership and physical appearance were made inferentially available.

ANALYSIS

We examine the use of racial categories and racial commonsense during question–answer sequences where police call‐takers solicit descriptions of persons of interest. Although these sequences are routine, we highlight participants' orientations to delicacy which attend to the potentially problematic inferences associated with racially categorizing others. The first analytic section shows how both callers and call‐takers orient to race as relevant in accomplishing a description. We analyse how aspects of question design display different orientations to delicacy and how callers align with or resist the presuppositions embedded in such questions. The second analytic section shows how callers resist the relevance of race or the broader project of description by raising and challenging aspects of these presuppositions.

Accomplishing visual descriptions through racialization

Extract 1 demonstrates how a call‐taker produces the request for visual description as both routine and delicate. The extract comes from a call where a distressed caller reports their partner arriving at the property. They have a protection order against him, but police have been unable to find him to serve it, and he has once again left as they started calling. The extract comes two minutes into the call after the call‐taker has already referred the job to dispatch. Our focus is how the call‐taker marks visual description as more delicate than other aspects of details gathering and how the caller responds with sex and race categories.

EXTRACT 1. 98_29.02_111.

The extract begins with the call‐taker initiating the new activity of details gathering about the caller's partner. The first request, for the name (line 1), is answered straightforwardly (line 3)—redacted as part of the ethical requirements for the study. The second request is slightly more complex as the question design suggests the caller may not know the answer (line 6) whereas they answer by providing both age and date of birth (lines 9–10). The confirmation sequence (omitted in the interests of space) displays the participants' orientations to the importance of securing accurate information to record in the case file or check against the police database.

At line 12, the call‐taker receipts the prior information with ‘okay’ and uses a sound‐stretched ‘a::nd’ to design the next request as a continuation of the project (Heritage & Sorjonen, 1994). In contrast to the previous requests, this question is prefaced by an account, ‘just in case our officers need it’ (line 12). Including an account displays the call‐taker's orientation to the question as requiring some explanation (cf. Wilkinson, 2011). The account provides the institutional logic for the request, emphasizing the relevance of the information for attending officers (whose presence the caller is requesting). The call‐taker uses the same modal verb ‘would’ and orientation to contingency (‘be able’) as the initial question (line 1 and line 13) but here uses the plural pronoun ‘we’ to speak on behalf of the institution (Heritage, 2004) rather than as an individual. The request itself is for ‘a description of what he looks like’ (lines 13–14). Thus, the call‐taker displays an orientation that asking for information about visual appearance is more delicate than asking for name or age.

The caller first confirms such a request as acceptable ‘yeah’ before producing two category terms, ‘male māori’ (line 15). These categories can be understood as belonging to two different collections: sex and race/ethnicity. Producing both together displays the caller's understanding of their shared relevance for accomplishing a description. The caller's turn ascribes categories to someone (who she knows) and tacitly claims these categories alone will be sufficient for accomplishing a description of what he looks like which will allow others to identify him. This makes inferentially available the commonsense knowledge that these categories can be ascribed by anyone based on sight alone. Including the sex category violates the normative principle of not telling people things they already know (Sacks, 1995) as the call‐taker's use of gendered pronouns ‘his’ (line 1;6) and ‘he’ (line 14) show this is not in question. However, the sex + race categorization produces a recognizable ‘package’ aligned with the institutional requirements of police details gathering (see Kameo & Whalen, 2015). Although the call‐taker treats the turn as a sufficient answer to the question (line 16), the call‐taker treats it as the first step in a project that includes additional details of appearance (lines 18–19). Nonetheless, both participants orient to inferential links between sex, race and physical appearance as relevant for accomplishing a description.

Extracts 2 and 3 demonstrate different ways racialization is used to accomplish visual descriptions. In both cases, call‐takers shift from open‐ended requests for description to requests for nationality. This word selection thus orients to delicacy by avoiding explicit reference to race (cf. Wilkinson, 2011). In Extract 2, the caller is a neighbour reporting a couple arguing in the street. Despite her limited knowledge of the people involved, she is able to provide visual descriptions that mobilize racial categories and inferences.

EXTRACT 2. 105_1.4_111.

EXTRACT 3. 192_19.4_111.

The call‐taker's first request for ‘their names’ (line 1) is tilted for a negative answer, displaying a presupposition the caller does not know the names of the people arguing, which she confirms (line 3). By contrast, the following question displays a presupposition the caller will be able to ‘describe the female to me’ (line 5). The question extracts one party from the collective (who the caller had identified as the instigator) using a sex category (‘the female’) in an example of the police institutional register (see Kameo & Whalen, 2015). The addition of ‘please’ and ‘for me’ (line 5) present the call‐taker as beneficiary of the request (Clayman & Heritage, 2014), marking the question as delicate in a way that the first question was not. Before the caller can answer, the call‐taker latches an additional turn constructional unit to the turn that specifies ‘nationality’ (line 6). The call‐taker thus uses a non‐visually available category as a euphemism for race in the service of accomplishing a physical description.

The caller's response is delayed (line 7) and departs from the structural preferences of the question in multiple ways, although they align with the call‐taker's use of nationality. The epistemic marker ‘I think’ (line 8) displays an incongruent epistemic stance, suggesting the requested information is not something they know. Yet rather than claim an inability to answer (as they do at line 3), they align with the topic agenda of the question by producing a description. The epistemic hedge ‘I think’ (line 8) inoculates the caller against responsibility for erroneously categorizing others while demonstrating their cooperation with the institutional project. Rather than providing a single category (which would fit the constraints of what nationality is she), they provide two alternative categories, ‘Samo:an? (.) or um:: (0.2) Māori’ (lines 8–9). Using two categories suggests they are linked by a shared category predicate (visual appearance) which illustrates the operation of commonsense racial inferences. Māori and Samoan people share Pacific ancestry and aspects of physical appearance. Despite the caller's displayed uncertainty, the call‐taker treats the answer as sufficient by moving on to the next part of the description: visually discernible age (line 10). This demonstrates that precise or singular category ascription is not necessary for accomplishing a description. The caller's provision of multiple categories displays an inference that neither they nor the officers need to differentiate between these categories to describe or identify the persons of interest.

Extract 3 further illustrates participants' orientations to the commonsense logic that links racial categories and visual appearance. The extract comes three minutes into a call where a distressed caller reports violence between their mother and stepfather. The call‐taker has just asked whether an ambulance is needed, which they decline. Here, the call‐taker shifts to details gathering about the stepfather. Much like Extract 2, the request for a description is followed by a wh‐question about nationality, treating this (along with sex) as a relevant first part of a description. The caller's response—and the subsequent clarification sequence—highlights skin colour as the relevant category feature.

The call‐taker marks the shift to a new activity with the declarative request ‘now I need a description of him’ (line 1). They reveal that sex is one relevant part of such a description but can already categorize the person of interest as ‘male’ (line 1) thanks to the caller's prior description. The wh‐question, ‘what nationality?’ (line 1) treats this as a relevant next detail—showing an institutional orientation to the sex + race categorical package demonstrated in Extract 1 (cf. Kameo & Whalen, 2015).

After a pause (line 2) which is an initial marker of some misalignment (Schegloff, 2007), the caller responds with a sound‐stretched category, ‘m:::āori’ and the additional descriptor ‘white’ (line 3). This displays the caller's understanding that a racial category alone is insufficient to accomplish the description. The additional description ‘white’ illustrates the racial commonsense that without this modifier, the category Māori would not ordinarily be understood in this way. White is not commonly used as a category in Aotearoa and concepts like ‘mixed race’ are understood very differently within Pākehā or Māori worldviews (see Gillon & Webber, 2023). Thus, inferences about skin colour and racial category membership are mobilized to accomplish a description that aligns with the request for a category but treats the referent as visually ambiguous with respect to race.

The call‐taker's response orients to the need to clarify what white refers to in this context. The request ‘fair skin is he?’ (line 5) seeks confirmation that white is a description of skin colour (rather than, say, an alternative category membership). After some additional trouble in hearing (lines 6–9), the caller confirms this (line 11) in overlap with the call‐taker's reformulation ‘did you say he's white?’ (line 10). Through this confirmation sequence, the participants collaboratively accomplish an understanding that the caller's initial description refers to someone who is not white but may appear to be (see Whitehead & Lerner, 2009). This negotiation reveals how the participants collaboratively distinguish between nationality as visual appearance (synonymous with race) and other possible meanings (e.g. citizenship, identity, etc.).

Extracts 1, 2, 3 demonstrate how racialization is routinely used to accomplish visual descriptions and how racial commonsense knowledge is mobilized. Participants treated race as visually discernible and oriented to links between racial category membership and features of appearance as taken‐for‐granted. Although participants treated the question–answer sequences as legitimate, call takers' use of accounts and word selection displayed orientations to delicacy.

Misalignment in accomplishing visual descriptions

In this section, we analyse cases where callers misalign with call‐takers' requests for visual descriptions. In Extract 4, the caller provides a visual description after some delay but does not treat race as relevant, which the call‐taker treats as noticeably absent. In Extracts 5 and 6, callers resist requests for descriptions by challenging the epistemic gradient of the question. In both cases, call‐takers pursue descriptions across multiple question‐answer sequences.

EXTRACT 4. 42_13.02_111.

EXTRACT 5. 178_17.4_111.

EXTRACT 6. 149_12.4_111 (3:29 – 3:53).

Extract 4 comes from a call where the caller reports a drunk male relative being abusive. After recording his name and age, the call‐taker initiates the request for a description. The caller does not immediately align with this project. We analyse how the call‐taker orients to the eventual description as insufficient by explicitly asking for ethnicity.

Much like Extract 1, the request for a description occurs within the same turn as an account for the request – ‘in case he does leave’ (lines 1–2) and is framed as an open‐ended request, ‘can you give me a description’ (line 1). Unlike Extract 1, the caller does not respond in the next turn, an early indication of dispreference or misalignment (Schegloff, 2007). Instead, after a short silence, the call‐taker provides an additional account, ‘so we can keep an eye out for him’ (line 3). This additional account pursues a response from the caller by highlighting the institutional relevance of a visual description.

The caller's response is delayed by more than a second of silence (line 5) and begins with a drawn out ‘u::m’ (line 6) which displays some trouble in answering. The caller produces two components of a description, ‘short (0.8) half bald’ (line 6). The call‐taker solicits further detail about what he is wearing (line 10), treating the description so far as insufficient. The caller's response is similarly marked by perturbations (line 12). After some follow‐up questions about attire (omitted in the interests of space), the call‐taker issues the request, ‘and sorry what ethnicity is ((name))’ (line 13). Turn‐initial ‘and’ (line 13) marks this request as tied to the ongoing project (Heritage & Sorjonen, 1994) and treats ethnicity as a missing element of the description in progress. Notably, this request occurs in the last available place in the sequence, giving the caller every opportunity to provide a racial category without prompting. Along with the selection of the word ethnicity which participants treat as a more delicate way to refer to race (Wilkinson, 2011), the sequential position of this request demonstrates the call‐taker's orientation to the delicacy (but institutional necessity) of soliciting this information. The apology token ‘sorry’ (line 16)—which is preserved across the repair sequence (line 15) likewise orients to the delicacy of the question.

The caller's response is delayed by over a second of silence (line 17) and begins with a stretched ‘a::h’, which indicates some trouble in speaking. These features treat the question as inapposite or unexpected (Fox & Thompson, 2010). The category ‘New Zealand’ (line 18) is delivered with turn‐final intonation. This response treats the question as a literal request for ethnicity, disattending to the readily available inference that the call‐taker is asking for a visual (i.e. racial) category. Not only do other callers recognize ethnicity as a euphemism for race, the subsequent so‐prefaced turn (line 21) displays the caller's recognition that their initial response was insufficient. The additional description, ‘so he‐he's a European’ (line 21) displays a trouble in speaking which orients to the delicacy in word selection. This turn makes explicit the inferential link that someone categorized as a New Zealand(er) is therefore ‘a European’ (line 21). Although neither turn is type‐fitted to the question, they invoke ‘New Zealand European’, the term used in census. This sequence thus illustrates racial commonsense about what New Zealanders look like, and for whom race is relevant or ‘invisible’ (cf. Awatere, 1984; Whitehead & Lerner, 2009).

A different kind of misalignment between caller and call‐taker is evident in Extract 5. The extract occurs two minutes into a call where a caller reports their grandson has left the property and is arguing with his girlfriend. This is a policeable issue as they mention he has a curfew (which is a common bail condition for people who have been arrested and are awaiting trial) and the call occurs during the nationwide lockdown during the pandemic that restricted movement between households. In the call opening, they referred to him as ‘a moko’, a shortening of the Māori word for grandchild (mokopuna). This person reference term thus makes the category membership Māori relevant to the caller and their moko. When the call‐taker asks for a description (line 1), the caller does not comply.

The call‐taker latches two turn constructional units, the first which introduces the new activity of securing a description (line 1), and the second which is an open‐ended request for information, ‘what does he look like?’ (lines 1–2). This question conveys an epistemic gradient of the call‐taker as unknowing (K−) and the caller as knowing (K+). After a delay (line 3) which heralds misalignment (Schegloff, 2007), the caller challenges this epistemic gradient by claiming the information should already be recorded on the police database. The call‐taker acknowledges but counters the caller's claim with ‘yeah nah’ (line 5) and provides an account that disaggregates ‘the officers that are on duty’ (line 5) from the wider collective indexed by the caller's reference to ‘you fellas’ (line 4). The call‐taker explains that attending officers may not have prior experience with the person of interest. The upshot of this (marked with ‘so’ at line 7) is a renewed request for a description. The if‐then construction minimizes the imposition ‘if I can just grab’ (line 7) and emphasizes the benefit to the call‐taker ‘that'll be fantastic thanks’ (lines 7–8) as a pursuit (see Clayman & Heritage, 2014).

Despite this second request, the caller does not cooperate (line 9). They instead provide a different reason for why a description is unnecessary: their grandson is ‘out here’ (line 10). This response challenges the presupposition that officers need a description to locate the grandson because he is nearby and thus easily identifiable. The call‐taker orients to the misalignment, using an explicit claim of understanding ‘yeah I understand that’ (line 11) to acknowledge the caller's response while still pursuing the question (cf. Weatherall & Keevallik, 2016). They provide a different account, ‘in case he leaves before we arrive’ (lines 11–12), and a third request, ‘can I just grab a description thanks?’ (line 12). The word ‘grab’ formulates the request as a minimal imposition and routine matter. The call‐taker's pursuit of the description displays their understanding of this information as institutionally relevant.

At line 14, the caller does comply with the project of providing a description, although turn‐initial ‘well’ signals the answer is not straightforward. They provide a description of their grandson's hair colour, which the call‐taker orients to as a first component rather than sufficient (line 15). The description of clothing (line 18) is also minimally compliant. At line 22, the call‐taker uses an and‐prefaced turn to mark one piece of information as noticeably absent from the unfolding description, ‘and what um race or ethnicity is he’. The perturbation ‘um’ and the word selection ‘race or ethnicity’ (line 22) orient to the request as delicate, as does the sequential position which provides ample opportunity for the caller to volunteer such detail before being explicitly asked. The question design presupposes that a single category membership is relevant, treats race and ethnicity as interchangeable and conveys the inference that category membership indexes physical appearance.

The caller responds with an open‐class repair initiator (line 24). This can indicate a trouble in hearing but can also be a challenge (Schegloff, 1997), creating an opportunity for the speaker to reformulate a problematic utterance such as racist talk (Robles, 2015). The call‐taker responds by repeating the turn with added emphasis (and dropping the and‐preface), treating the caller's turn as a trouble in hearing (line 26). The caller then provides two categories rather than one, ‘māori and samoan’ (line 27). Unlike Extract 2, these are not offered as alternatives, but as multiple category memberships. The call‐taker treats this as sufficient and thanks the caller (line 28).

Thus, in Extract 5, a caller (whose own category membership as Māori has been made relevant earlier in the call) initially resists providing any information about their grandson's appearance. As a family member, the caller has relational category rights to knowledge, and as he is still nearby, first‐hand access to details like what he is wearing. (Contrast this with Extract 1 where the caller's ex‐partner has just left the scene.) Thus, the misalignment is not based on inability but unwillingness to answer. The caller provides two accounts for not answering, but these are both rebutted by the call‐taker, and they eventually comply. This extract starkly illustrates the gatekeeping role of the call‐taker. It is challenging to resist the constraints of a question in any situation (Hayano, 2013) but in an institutional environment where the question‐asker has the rights to determine access to help, the stakes are even higher. Although the caller demonstrates unwillingness to answer (quite possibly due to the commonsense understanding that such descriptions entail racialization), her ability to secure the help she seeks is contingent on her cooperation.

By contrast, Extract 6 is a case where the caller does not comply, despite multiple attempts by the call‐taker to secure a description. The extract comes three minutes into a call made during lockdown. The caller has reported an ongoing altercation with their adult son (on home detention at their property) and complained that earlier requests for police to take him were ignored. The caller thus claims a social position of complainant as well as help‐seeker and demonstrates resistance throughout the call to the entire project of information gathering. When the call‐taker solicits a description, the caller responds with increasingly angry affective displays. This results in the only case where the gathering of racialized information is completely forestalled.

The call‐taker begins the request for a description with a wh‐question, ‘what does ((name)) look like?’ (line 1). They then list two relevant parts of such a description already provided: sex and age (cf. Extract 3). Continuing intonation marks these components as parts of a list, which the caller is invited to complete. However, two seconds of silence elapse (line 3) which indicates the caller's response will not be straightforward (Schegloff, 2007). Rather than answer, they issue a counter that challenges the basis of the question.

The epistemic stance of the question positions the call‐taker as K− relative to the caller as K+ (Hayano, 2013). Their response challenges that epistemic gradient by suggesting this information is already known to the call‐taker (or at least the institution they represent) as there must be ‘a photo of him’ on file (line 4). When no response is forthcoming (line 5), they upgrade the challenge, reminding the call‐taker their son ‘got charged’ (line 6) and is currently ‘on home d!’ (line 8)—meaning he most certainly has a police record. The caller's rising intonation displays an incredulous affective stance that treats the question as inapposite and ridiculous. The caller begins what is projectable as a request (line 9) but drops out as the caller displays a strong negative stance with the exclamation, ‘I couldn't care a fuck!’ (line 10).

Rather than responding to the caller's challenge, the call‐taker issues the neutral receipt ‘okay’ and pursues a forced‐choice question about height (line 12). This treats the project of description as necessary while constraining possibilities for a relevant answer. The caller indicates continuing misalignment with a well‐prefaced response but does respond with a type‐fitted answer (line 14). The next question is tied to the same project with and‐prefacing (Heritage & Sorjonen, 1994). The call‐taker again constrains the possibilities for answering, here using a list format (cf. Garcia, 2022). The list includes the categories Caucasian, Māori, and Polynesian linked by ‘or’ (line 18), displaying an understanding of them as mutually exclusive. This question design produces the request as a routine institutional information‐gathering, effectively disattending to the caller's affective display.

The caller's response does not select any of the proffered categories, instead rejecting the action agenda of the question. Nonanswer responses are dispreferred and typically delayed or hedged (Hayano, 2013). Here, the caller's non‐answer response is latched to the end of the call‐taker's turn and empathetically rejects the question. Although the caller claims they do not ‘know what he is’ (line 19), the expletive and affective stance make clear they are unwilling rather than unable to answer. Notably, this account for not answering displays an understanding of race as personal property, disavowing access to it, rather than making a category ascription (cf. Extract 2).

This is the only case in the dataset where a caller manages to avoid directly categorizing the person of interest. The call‐taker does not pursue the matter further and soon announces that help is being dispatched (not shown).

DISCUSSION

This study has documented how racialization is incorporated into the institutional logic of emergency police calls. We have shown how call‐takers' requests for visual descriptions routinely include or occasion racial categories. This use of racial categories and racial inferences constitutes racialization (Shrikant & Sambaraju, 2024). Commonsense inferences about what members of racial categories look like and thus how they might be identified are largely treated as a routine and legitimate part of the institutional interaction. Our study thus demonstrates the interactional infrastructure that sustains and makes possible the kinds of racialized policing that have attracted scholarly attention and public critique (e.g. Aho et al., 2022).

Our findings show how racialization is pervasively relevant for accomplishing visual descriptions in emergency police calls, the first analysis of its kind in Aotearoa. Even when call‐takers used open‐ended requests for descriptions, callers treated racial categories as relevant (Extract 1). In other cases, call‐takers shifted immediately from open‐ended requests for descriptions to requests for categorization (Extracts 2 and 3), displaying an understanding that such categories are a relevant first part of a description. If callers provided descriptions without including racial categories, call‐takers pursued them with explicit requests (Extracts 4, 5, 6). The only instance in our dataset where a call‐taker abandoned the project of a visual description (and the embedded racialization) was a repeat caller who rejected requests for information they had previously provided (Extract 6). Thus, in this institutional context, racial categories are treated as a necessary component for accomplishing visual descriptions.

Although participants treated race as relevant for accomplishing descriptions, they oriented to it as delicate. We documented a range of practices that oriented to delicacy, including turn design, word selection, and sequential position. Some call‐takers designed requests to include accounts that specified why the information was required (Ex. 1, Ex. 4). Such accounts marked requests for visual descriptions as more delicate than requests for other kinds of information. Call‐takers also used terms like nationality or ethnicity (Extracts 2, 3, 4) or combined multiple terms (e.g. ‘what race or ethnicity is he?’ in Extract 5). These word choices demonstrated orientations to the delicacy of race by avoiding mentioning it explicitly (see Wilkinson, 2011). In cases where callers did not include racial categories in visual descriptions, call‐takers waited until the last possible moment before asking for them (Extracts 4, 5, 6). This sequential organization allows callers to ‘volunteer’ racial categories without call‐takers having to ask, a further orientation to the delicacy of the question (cf. Garcia, 2022). Participants' orientations to delicacy reveal their understandings that using racial categories for visual descriptions requires careful interactional management.

The ascription of racial categories in our data reveals taken‐for‐granted presumptions about race and visual appearance that structure our social world. These inferences are built on essentialist understandings of race as biological that underpin racist projects of empire and domination (Shrikant & Sambaraju, 2024). Although such understandings have long since fallen out of favour with both physical and social scientists, they continue to circulate in everyday commonsense reasoning (cf. Wetherell & Potter, 1992) and are embedded into the interactional infrastructure of institutions like the police. One presumption is that race can be seen and thus ascribed by anyone through visual inspection. For example, a caller who is unable to provide names can still ascribe racial category membership and age based on visual access. In this context, participants produce descriptions for others to use, rather than prioritizing descriptions the subject might identify with (cf. Wilkinson, 2011). Even as they do so, participants treat such descriptions as delicate. This dynamic demonstrates how institutional imperatives supersede political or ideological sensitivities—and thus some of their practical limits in countering issues like racism (see Whitehead, 2015).

Our analyses highlight how racial commonsense operates both in straightforward cases and in instances of trouble. For example, providing alternate categorizations (Extract 2) treats members of some racial categories as visually interchangeable. By contrast, listing multiple options to choose between (Extract 6) treats the categories as mutually exclusive. Instances of trouble make links between categories and category features visible. Including a modifying description displays an understanding that the person appears differently to what might be ordinarily assumed (Extract 3). Thus, presumptions which often operate as taken‐for‐granted (i.e. for which racial category members is ‘fair skin’ an unexpected feature; what do ‘New Zealanders’ look like?) rise to the surface of interaction. Previous studies have documented the invisibility of whiteness where membership in the category ‘white’ is rarely made explicit (Whitehead & Lerner, 2009; Wilkinson, 2011). Our study is distinctive for showing how such presumptions operate in a context where racialization is linked to visual appearance.

Call‐takers are accountable to other members of the institution for the details they record on the casefile (Whalen & Zimmerman, 1998), which in this institutional context creates an environment for racialization (Kameo & Whalen, 2015). Unlike other institutional contexts where collecting demographic details is secondary to the core business (Wilkinson, 2011), police call‐takers treat securing a physical description as a necessary part of the call. Some callers questioned the legitimacy of call‐takers' requests by claiming such information should already be available on the police database (Extracts 5 and 6)—a possibility call‐takers may attempt to pre‐empt with accounts (Extracts 1 and 4). Callers' attempts to avoid providing a visual description display tacit concerns with what this information might be used for or how it might colour officers' treatment. Members of marginalized communities know all too well the links between racialization and racism in policing, which is particularly salient when callers are seeking emergency help but may not want their relative to be (further) criminalized (see Tennent & Weatherall, 2025). However, possibilities for resistance are constrained in an interactional environment where call‐takers are gatekeepers of the service and treat a visual description as obligatory.

One limitation of the study comes from the nature of the data. Police provided us with call recordings, but not the written casefiles. Thus, we were unable to examine how call‐takers transform callers' descriptions into the written record (cf. Kameo & Whalen, 2015). An important future direction is to examine the transformation of racial categories across the interactions between callers, call‐takers, dispatchers, attending officers and persons at the scene. Such investigations could document how racialization becomes consequential across the course of an institutional encounter, providing further links between ‘upstream’ practices of racialization in emergency calls and their consequences for how officers identify and interact with persons at the scene (see Gillooly, 2020).

Our findings raise complicated questions about applications. There is some immediate relevance for call‐taker training. Call‐takers do not currently receive any training for how to sensitively obtain details about visual appearance, and the orientations to delicacy we observed illustrate some of the recurrent ways call‐takers manage this interactional challenge. Our findings can provide call‐takers with a new perspective on the challenges of their role and increase their awareness of different ways of requesting this information (e.g. Stokoe, 2014). However, there are structural barriers to broader changes in institutional regimes. Although as scholars we might question whether racial categories ought to be used for visual descriptions, racialization is structurally embedded in the interactional routines of the institution. Despite orientations to delicacy, racialization operates as a key component of visual descriptions that may be required for institutional actors to adequately do their job. It is not possible for call‐takers to simply avoid asking about race—at least as the institution of Police is currently configured.

Our discursive psychology approach provides an interactional perspective to the issue of race and policing. Rather than studying racism directly, discursive psychology allows us to examine racialization as an ‘upstream’ categorical practice. Racialization can and often does lead to racism, but it is important to analytically distinguish routine practices of racialization from their possible interactional consequences (see Shrikant & Sambaraju, 2024). In fact, racialization routinely occurs in our data without participants treating it as problematic or prejudicial. Nonetheless, by attending to the links between racial categories and racial inferences, we have shown how racial logics are structurally embedded into the workings of the institution. The use of commonsense knowledge about racial categories and visual appearance in this institutional context is premised on it being intelligible to any member of the public who may call police. Emergency calls are thus one specific institutional context where this more widespread phenomenon of racialization can be observed. Documenting the routine ways racial commonsense is incorporated into institutional practices is crucial for understanding how racial commonsense organizes institutional interaction and for developing grounded moves towards social justice.

AUTHOR CONTRIBUTIONS

Paige Winthrop: Conceptualization; investigation; writing – original draft. Emma Tennent: Conceptualization; investigation; methodology; writing – review and editing; supervision.

CONFLICT OF INTEREST STATEMENT

None of the authors have a conflict of interest to disclose.

DATA AVAILABILITY STATEMENT

The datasets generated and/or analysed during the current study are not publicly available due to the ethical approvals of the study but are available from the corresponding author upon request.