Guest blog: The death of George Floyd – should we analyse the tape?

Magnus Hamann is a CA researcher with long experience of analysing the kind of police/citizen encounters that are available on YouTube,  especially those that end in violence. In this guest blog, he wrestles with the many dilemmas facing the academic researcher when something is incendiary, and very much in the public eye: How may an interactional researcher approach a case like the police killing of George Floyd? Should they abstain? Take sides?


Magnus Hamann, Loughborough University

Right now (early June 2020), the story of African-American George Floyd’s death at the hands of a US police force circulates the world. The graphic images in those recordings have caused a collective sadness and anger. Emotions that have led to disturbances, to put it mildly, in many US cities. One recording, especially, has gone viral[1].

It shows one of the officers kneeling on his neck until Floyd dies from asphyxiation. That officer was eventually arrested and charged with murder. I’ve studied video clips of police violence before. This time, I started to do so, but stopped. Here’s why.

Emotionally invested

Watching the videos, I ended up feeling what I imagine so many other people who watched the video felt: Anger, shame, sadness and helplessness. Instead of the usual analytical distance, I felt they got to me emotionally. This experience, initially, made me give up watching the recordings, but also made me think about the practice of working on disturbing data at all.

The fact that the data arguably disturbing to watch, should not dismiss it as something we could or should work on. While thinking about this, I realised that there’s lots of work on distressing episodes in life. For example, Shaw (et al. 2016) analyse the conversational strategies employed by doctors who have to tell parent that their new-born child is going to die.

Still, the difference is that in Shaw (et al.)’s case, no-one is going to argue that the parties are in actual conflict. Even less so, a situation where one is physically hurting the other. With a police killing, it is different.

How to see the video?

When I returned to the recordings, I had the idea of trying to see it the same way that Chuck and Candy Goodwin, two of EM/CA’s pioneers, saw the film of the African-American Rodney King being beaten repeatedly and heavily by members of the Los Angeles police department (1997). Here’s part of their comments on the subsequent trial, commenting on how the clip was being discussed in in court:

”By deploying an array of systematic discursive practices, including talk, ethnography, category systems articulated by expert witnesses, and various ways of highlighting the images provided by the tape, the lawyers for the policemen were able to restructure the complex perceptual field visible on the TV screen so that minute body movements of Mr King, rather than the actions of the policemen hitting him, became the focus of the jury’s attention.”

The Goodwins show how an expert witness, defending the police, was able to make the jury see things from the police’s point of view, by introducing an official coding scheme for the jury to see the incident as it played out. In this scheme, what Rodney King did at each moment was codable as a deliberate action – constructing him as an active agent, responsible for the police striking him 47 times with metal clubs. While the case had been described in the media as a very violent beating, the defence manged to get the jury to see that the police-officers had used reasonable and justified force in the situation. So, while their analysis was, to the liberal mind perhaps, revelatory, and could be seen as a form of protest, the analysis itself was not taking sides as it was simply displaying the practices that lead to the emergence of another way to see the recordings in a different way to the public.

Maybe I could follow the Goodwins’ lead, and just be “analytic” and let the chips fall where they may?

Analysis in whose interests?

Previously when I have presented data of police-officers using force that results in a person dying, I have (rightly) been questioned about whether it is even ethical to scrutinize and investigate these situations. I usually argue that we can and should investigate the practices that lead to a person’s death if we do so in a way that is in the interest of the person who died. This does not mean that our analysis should try and clear the person of guilt or try and place a guilt on someone in the interaction. My argument is that if we wish to approach data of other people’s death in an ethical manner, we must do so with an honest curiosity in the practices that led up to the moment where that person loses their life. Our own identities – here, our ethnicities and perhaps our nationalities – should be irrelevant.

As George Floyd’s death is currently a topic of public discussion, I, as a social scientist, stand in a dilemma. I could participate in the public discourse – help shape it or steer it in another direction. The reason that I have chosen not to engage is actually due to the potentially political grounds such an analysis would have to manoeuvre through. While it is both the reason for doing research that we can help inform the public discourse, we also risk that analytical contributions can become political in a way that constitutes an ethical problem:

“The way in which professional coding schemes for constituting control and asymmetry in interaction are used by the police to justify the way that they beat someone alerts us to ethical problems that can arise when, as social scientists we put our professional skills at the service of another profession, and amplify its voice and the power it can enforce over those who become the objects of its scrutiny. “ (C. Goodwin & M H Goodwin, 1997)

So what should we do?

I believe that we should indeed work on recordings such as the one of George Floyd’s death. Not only do they represent disturbing events of human life. They also represent events that shape a society’s collective understanding. These events should be studied. And, I argue, especially within the EM/CA approach. As Chuck and Candy (1997) formulated it back then:

”…, rather than situating such phenomena entirely in a rather general notion of ‘discourse’, it is necessary to investigate in detail the situated practices through which socially relevant talk and vision are accomplished, something made possible by the resources of fields such as conversation analysis.”

I will return to the recording. Writing this blog post has made me realise that I should. But I will not do it now. I think the public interpretation of the recordings is generally the accepted interpretation of what happened. I have not yet seen interpretations/analyses that are being weaponised in order to change the narrative. I fear that introducing an analysis might muddy these waters. At the same time, I fear that by actually thinking like this, I am also politicising my research.

[1] I won’t link to the video here; it can easily be found.


Goodwin, C., & Goodwin, M. H. (1997). Contested vision: The discursive constitution of Rodney King. The construction of professional discourse, 292-316.

Shaw, C., Stokoe, E., Gallagher, K., Aladangady, N., & Marlow, N. (2016). Parental involvement in neonatal critical care decision‐making. Sociology of health & illness38(8), 1217-1242.