Some commentators have suggested posts that pose questions to our readers. I think that the discussion on Peter Henne’s piece, “A Modest Defense of Terrorism Studies,” provides just such an opportunity.
In Remi Brulin’s most recent comment, she asks:
… I am very much interested in better understanding why Peter (and others of course) do believe that the distinction between state and non-state “terrorism” is so important and necessary from an analytical point of view.
For my part, I would tend to think that it could in fact add a lot to our understanding of “terrorism”, of the non-state or state variety. But even if it were not so, even if such difficulties do appear: that is a problem that scholars would deal with at their micro level, at the level of their case studies, of their datasets. I donot see how this can possibly be a reason or argument for defining a whole field of research and expertise.
My flip answer to Brulin is that there’s a significant literature on subjects such as the of targeting civilians, state repression, and mass violence that already engages with “state terrorism.” Some of that literature, I believe, extends its purview to non-state actors. Nevertheless, I think it worthwhile to begin with a premise, disaggregate some issues, and then throw things open to our readership for their opinions.
Let’s begin with a definition: terrorism is a strategy that seeks to instill fear in non-combatants for coercive purposes. This definition faces problems: what is fear? what is a non-combatant? But, for the sake of argument, let’s begin with a definition that does not render all violence in warfare as terrorism, yet is broad enough to include such disparate activities as nuclear deterrence, torture, collective punishment, and blowing up cafes.
So what is at stake — from an analytical and methodological perspectives — in limiting study to non-state actors that engage in terrorism? Will we learn more or less if we include every possible instances of terrorism in our universe of cases, or will we efface causal processes specific to different kinds of actors and contexts?
PS: for additional related arguments, see Phil Arena’s post on the matter.
Notes: First, Morning Linkage regularly runs Monday-Saturday, but only occasionally on Sundays. Second, due to Labor Day and the start of school last week, there will be no podcast this weekend. Podcasts will resume next week.
Daniel H. Nexon is a Professor at Georgetown University, with a joint appointment in the Department of Government and the School of Foreign Service. His academic work focuses on international-relations theory, power politics, empires and hegemony, and international order. He has also written on the relationship between popular culture and world politics.
He has held fellowships at Stanford University's Center for International Security and Cooperation and at the Ohio State University's Mershon Center for International Studies. During 2009-2010 he worked in the U.S. Department of Defense as a Council on Foreign Relations International Affairs Fellow. He was the lead editor of International Studies Quarterly from 2014-2018.
He is the author of The Struggle for Power in Early Modern Europe: Religious Conflict, Dynastic Empires, and International Change (Princeton University Press, 2009), which won the International Security Studies Section (ISSS) Best Book Award for 2010, and co-author of Exit from Hegemony: The Unraveling of the American Global Order (Oxford University Press, 2020). His articles have appeared in a lot of places. He is the founder of the The Duck of Minerva, and also blogs at Lawyers, Guns and Money.
I’ve written quite a bit on this, and I’m just completing a book on it. There are more than analytical issues at stake. However, for what it’s worth I’m in favour of a very deflationary account of terrorism. The problem with most accounts of state terrorism are that just about everything the state does is terrorism.
True but then again most accounts of non-state actor terrorism view just about everything the actor does as terrorism; the Palestinians’ supposed ‘diplomatic terrorism’, throwing stones as terrorism, fighting occupation (Iraq/Afghanistan) as terrorism…
Sorry John, but how does that follow? Fighting occupation isn’t something that can be deemed to be terrorism or not, it depends on how the fighting is pursued, who is fought etc… According to my account, the Brighton bombings weren’t terrorism. Much of the literature confuses ‘terror’ with ‘terrorism’
I’m surprised to hear that, Colin. Could you be more specific about the accounts of state terrorism that use the term in too expansionary a sense? I admit that I haven’t surveyed the “state terrorism” literature as a whole, but for the cases I am familiar with–the military dictatorships in Argentina and Chile–describing the methods used by those regimes for dealing with dissent and holding on to power as “terrorism” certainly seems appropriate. Otherwise, it seems that you end up falling back on a state/non-state distinction, which again begs the question of the analytical and normative reasons for that distinction.
Agreed. Interested in an answer also.
I’m inclined to agree with Colin here–I question at times if accounts of “state terrorism” came about due primarily to normative questions and disputes, moreso than issues with explanatory purchase. What I mean is, as Kathryn identifies above, the issue with restricting “terrorism” to non-state actors is that it is easy to then legitimize and delegitimize certain types of violence (delegitimizing non-state actors’ violence versus legitimizing that of the state). To counter this account and normative position if you will, we constantly hear analysts and activists asking, “What about state terrorism? Isn’t the state also culpable in inflicting violence against civilians to instill terror, and for coercive purposes?” Yet, does the term “state terrorism,” (or to be bold) does the term “terrorism” itself, really help us in our causal explanations, as opposed to a term like “political violence” per se? It might if it helps us distinguish between certain types of violence versus others, but this begs the question to me if we can separate out the causes behind terrorism, state or non-state, as opposed to other patterns of violence? Is there a separate explanatory logic here to justify the term even? I’m tempted to say no, that if anything, the term might help in selecting cases/actors for comparison, but we still run into the issue of treating it as if it’s a distinguishable social phenomenon operating according to a separate logic/law/combination of causal mechanisms, when in actuality, this isn’t the case. It seems that the term “terrorism” tells us more about analysts’ normative positions, or even possibly about the identity and constitution of political actors, or even processes of “Othering,” or “securitization” than anything else. At the end of the day, really, it seems to present a much more interesting ethical and normative debate about legitimacy, order, sovereignty, and state authority, I think, than it does in terms of explanatory theorizing that is not as concerned with normative debates, or discussions concerning constitutive effects….
To me an issue that would be at stake from only studying non-state actors in work on terrorism goes back to basics with legitimacy, and debates over who is perceived as legitimately using violence for (legitimate) political ends (I would add “political” to the above definition). The allocation of illegitimate/legitimate boundaries holds significant consequence in how such framing limits and enables different possibilities of response (theoretical and political). This is not to imply that “anything goes” for empirics by moving beyond non-state actors. But, minimizing the state/non-state line would, I think, further open underexplored alternatives to existing research and practice. Perhaps encouraging an identification of violent methods “first”, instead of the violence being (at least partially) defined by how we categorize kinds of actors before even starting analysis? Thanks for the post, a lot to think about!
Dan, I think this is a fair set of questions. I think we should consider the perpetrator of violence as this can inform the reasons for its use. States and nonstate actors can both perform your definition of terrorism (I’d add violence or threats of violence in there, but I think it is implied). This alone is an argument for keeping them together. A good reason to keep them apart is this (obvious?) observation: non state (oppositional) terrorism happens when a weak actor faces a stronger opponent (see Al Qaeda, IRA, PLO, HAMAS, ALF, ELF, and nearly all other groups I can think of off the top of my head), while state terrorism tends to occur when the state actor(s) is/are in a relatively stronger position (see your examples as well as other classically cited cases, such as Stalin in Russia, Hitler in Germany in the 1930s). Studying these phenomena apart may miss this observation (or maybe not) while keeping them together might put scholars on track to try and explain why some groups use it and some states don’t (and vice versa).
Joseph. Couldn’t agree more. And of course, we can do both. At the level of individual case studies, of course sometimes it makes sense to separate, to keep them apart. But at the level of the field, there is absolutely no reason to do so.
Of course, we could shift the covering term to “political violence” and then we would get more observations. Which is a way of suggesting that I think the issue is more complicated than you imply. :-)
‘state terrorism tends to occur when the state actor(s) is/are in a relatively stronger position’
There are cases in which the opposite is true, and the reason for which state terrorism was employed was precisely because of this deficit in state power compared to the power of some rebel force[s]. I think the actions of the Janjaweed in Darfur or possibly the actions of Serbian paramilitary groups might be examples of this. I also think that this supposed asymmetry doesn’t make as much sense in considering terrorism in Lebanon or Iraq, to name a couple examples, as it carried out by multiple, competing, and more symmetrically empowered actors engaged in internecine competition.
I also wonder if the South African state terrorism of the 80s was not borne out of quite an intense sense of weakness and fear, in which the ‘security state’ looked at what happened to the Rhodesians and thought ‘we are all fighting for our lives’.?
Dan, and all. Earlier today I saw Phil Arena’s post and the great thread there. Had two hours to kill in the train so I jotted down several replies that I just posted there. I am hoping that some of you find them of interest.
As to your “flip” reply to me, Dan, I am not really sure I understand it: no one (at least, not me) claims that there is no “significant literature on subjects such as the of targeting civilians,
state repression, and mass violence that already engages with “state
terrorism.”” In fact if you go back to the discussion on Pete Henne’s piece I specifically mentioned the great works by Campbell and Brenner, by Sluka, or by Corradi, Weiss Fagen and Garreton… Besides, the point is precisely that these works are not produced by “terrorism experts,” that articles on these topics are never published by the two main “terrorism studies journals” but do get published in sociology journals, anthropology journals, poli sci journals, etc… and will most likely get published by CST.
So again, Dan, I don’t see how your reply really deals with any of the substance of any of what I have been trying to say at all. Maybe more precisely: Pete Henne’s piece was in part prompted by my article on FP, where I focused on one case study, El Salvador. And yet most if not all the discussions/ debates that have followed have made no reference at all to the specifics of that article, no reference at all to what it shows about how the two main “terrorism studies” journals have:
– failed to write about this conflict (basically deciding that it was “not about terrorism”)
– been silent about the discourse of Reagan at the time, about the fact that he repeatedly said that US military aid to El Salvador was part of the “fight against terrorism”
– silent about the numerous Congressional debates on who the “real terrorists” were in El Salvador at the time
– silent about the fact that the CIA, or the US embassy, consistently wrote about “right wing terrorism” and “ARENA’s terrorist network” and the ties between the security forces and the “death squad terrorists”.
– all this without ever proposing an explanation for such a silence (and then, a few years later, deciding that US policy in El Salvador, and some of its failings or shortcomings, should be discussed under the headings of “COIN” or “LIC”. Not “terrorism.”)
– all this while publishing countless articles (and several Special Issues) devoted to analyses of “US policies against terrorism”, “the US and whether or not it is efficiently fulfilling its role as leader of the Western world in its fight against terrorism”, etc…
What do experts today say about how the two main journals in their field have dealt with this one conflict? Do they have any problem with it? I was genuinely hoping that this would be the one thing experts on “terrorism” today would be interested to talk about and react to when I wrote that piece…
The short answer is that my post is a prompt for a discussion of the methodological and analytic issues of expanding or contracting the “data set” in studies of terrorism, not an attempt to relitigate your beef with terrorism studies. In that context it is perfectly appropriate to note that a fair amount of extant literature evaluates “stats terrorism” and some of it does not draw a sharp distinction between state and non-state actors.
“stats terrorism” should be “state terrorism,” of course. I strongly deny that this is a Freudian slip. And “beef” sounds more dismissive than it was supposed to (I agree with the basic thrust of the criticism, which deserves the kind of elaboration you appear to be giving it).
Fair enough :) As to the specific question of expanding / contracting data sets and such, I would repeat what I said to Phil Arena: many “orthodox” (using the term as
CST scholars do) terrorism experts like
Rapoport, Crelinsten, Wilkinson, all writing as far back as the late
1970s/early 1980s, clearly defended the idea that state and non state terrorism should absolutely be
studied together because they are often mirror image of each other,
exist in a dynamic relationship one with the other, etc. Or see the work
of Nimer on “terrorism in South Africa”, for example in TPV in 1990 I
believe, where he presents the conflict as a perfect example for how
“terrorism” came to be used by many different actors, how the state used
“terrorism” directly, indirectly, through proxies of various forms, white and black, to
fight the ANC, how of course the ANC came to use “terrorism” also in response to
this “terrorism” by the state or some of its allies, etc…
I think the main problem in using a definition such as yours stems from neopositivist methods.
Compare it to my preferred definition, from Neumann and Smith (2005) drawing upon a much older one by T.P. Thornton: ‘the deliberate creation of a sense of fear, usually by the use or threat of use of symbolic acts of physical violence, to influence the political behavior of a given target group.’
This is an extremely broad definition. By not limiting terrorism solely to attacks or actions directed at noncombatant targets, one does risk a situation in which a great deal of warfare begins to look like terrorism. Compiling a dataset using this definition would produce something very broad! On the other hand, it allows us to group strategies and methods together in ways that are more conducive to a good analysis of them.
Take, for example, Marighella’s ‘Urban Guerrilla Warfare’. This is something we should want to call terrorism. The logic behind it is that provocative attacks will lead to government reprisals that, through their brutality, will alienate the people and foment revolution. Is this suddenly not terrorism if the targets are soldiers instead of bureaucrats?
Take another example: Maoist insurgency. Does the Maoist model prescribe attacks against noncombatant targets? It does. But it contextualises those attacks within a comprehensive strategy of insurgency in which more conventional military actions, or at least, ‘guerrilla’ military actions, also feature prominently.
The noncombatant targeting criterion may make a great deal of sense in single case analyses that are sensitive to the historical context of a particular situation or crisis – just as the very broad definition I adduced earlier also allows for nuanced, case-specific analysis – but it will lead to some awfully arbitrary exclusions of relevant actors, events, and methods if used to compile a large-N dataset, and that will definitely have a negative impact upon the validity or usefulness of a wide ranges of analyses.
I don’t see the connection between your objection to the provisional definition I offered and “neopositivism.” My only claim in favor of “non-combatants” was that it renders terrorism a subset of political violence; you’ve incorrectly interpolated an argument that this would somehow make our data sets unmanageable. I’m not trying to be pedantic here, but to make clear that nothing in your discussion really hinges on whether or not one embraces neopositivist methodology.
I’ll clarify. What I mean by pointing to neopositivist methodology is that the potential harm of unreasonable or arbitrary inclusions and exclusions of cases becomes quite great. The goal is presumably to designate a subset of ‘political violence’ that will be amenable to useful neopositivist analyses. But as a result, we can’t go to one case X and say ‘in case X the decision to employ one strategy targeting noncombatants and another targeting combatants makes a noncombatant criterion useful in distinguishing terrorism from another method’, then go to another case Y and say ‘in case Y the choice of targets was tactical and the strategy was the same in general’ and both call them cases of terrorism. On the other hand, if we start with a particularly broad definition of terrorism then tailor it for particular cases or particular analyses in a more narrow and specific sense, we are more likely to avoid the problems of harmful in/exclusions. I don’t think we avoid them entirely, because we still must somehow start with an idea of what terrorism is, but we can employ fuzzier methods for doing so such as starting from ‘family resemblances’.
Nothing in my discussion hinges on neopositivist methodology, but I think that the contrast I make does hold true in a practical sense.
I don’t mean to be obtuse, but I’m having trouble both parsing your alternatives and seeing how they lead to the conclusion, i.e., “that the contrast [you] make does hold true in a practical sense.”
Ach, I’m doing a terrible job of communicating then. I apologise, and I’ll keep trying.
Basically, because ‘terrorism’ is such a fuzzy, contested, and multifacted concept, operationalising it in a way that allows us to build a large multi-case dataset is really hard. One example of why it is hard is that in one case of terrorism, targeting noncombatants might be a crucial aspect of why it is terrorism, while in another case, we might really think terrorism is an appropriate term and yet see no strategic meaning to whether combatants or noncombatants were targeted.
On the other hand, if we start from a family resemblances definition, and accept that terrorism cannot be reasonably operationalised across cases by a single set of necessary and sufficient criteria, we make it hard to compile datasets but easier to study terrorism in a way that is sensitive to context.
One solution to the problem I am trying to describe would be to never compile a ‘terrorism’ dataset and instead only compile datasets of particular terrorist strategies, or particular subtypes of terrorism. You could designate ‘occupation resistance’ terrorism or ‘internecine’ terrorism – these are just off the top of my head – and try to navigate the fuzziness of the term ‘terrorism’ by being more specific with it.