the assumption of responsibility for the welfare of the world











2011 December 5


‘Region’ is a term before it is a concept; concepts are eternal and there are an infinite number of them, while terms are limited and ever-changing, both in form, and in the relationship between term and concept (what semioticians refer to as the sign relationship, with term being signifier and concept being signified). If, however, we choose any of several definitions of ‘region’ (that is, any of several concepts that are frequently associated with the term), we can make a sensible analogy with John Agnew’s territorial trap. In fact, understanding the territorial trap can allow us to evaluate the various regional concepts; by using regions of the right definition, we can avoid it.

Though Agnew is discussing states (‘state’ is itself an enormously-problematic term), he intends the state as a kind of region, and his critique applies to the use of the state as a region for other purposes, or sometimes all purposes. Here ‘territory’ and ‘region’ are generally synonymous: a bounded area fixed in location on the surface of the Earth. In this case, the region in question is a state’s territory in one of two ways. First, the state (its government de facto) actually controls the region in question. This is the characterization of empirical statehood, as described by Robert Jackson and Carl Rosberg. Second, the state (its government de jure) has a right to control the region in question, as recognized by some or other political system, usually the Westphalian system and its modern manifestation, the United Nations. This would be Jackson’s and Rosberg’s juridical statehood.

The primary mistake of the territorial trap, in my reading and generalizing to any region, is to believe that the region in question has somehow contained much or all reality (of society, of nature, et cetera). For instance, if the region in question is a juridical state, we would fall into the territorial trap by supposing that the boundaries of the juridical state are likewise the boundaries of a nation, of a society, of a culture, or even of a language or an ecosystem. We would fall into the trap as scholars by using the boundaries of the juridical state for our own explorations. We would fall into the trap, especially, as individuals, when we believe in our minds that there is essentially just one set of regions, that these regions are the juridical states, and that our existence is entirely contained within the juridical state in which we currently reside.

It is the juridical state that is at the heart of the original territorial trap. As has been shown by many (separately, for example, Jackson; Jackson and Rosberg; and Jeffrey Herbst), juridical statehood does not align with any reality of the world besides the diplomatic reality that creates it. In particular, it does not align with empirical statehood, despite the weight given to empirical criteria (the Montevideo Convention, notably) by juridically-oriented scholars like James Crawford. Africa is merely richest in examples of recognized sovereignties that do not control their territories. As Jackson and Herbst separately note, control of the recognized capital city is generally sufficient to be granted the franchise under international diplomacy for the entire recognized territory. (That did not work for the Taliban in Afghanistan, though, or the Union of Islamic Courts in Somalia, so like so much of international diplomacy, there are no reliable principles.) Considering that juridical statehood is purported to be, and is described with the characteristics of, empirical statehood, this misalignment is significant.

The other central misalignment is with the nation. The “nation-state” is a conceptual product of the Westphalian era; whether it has ever existed, whether it has any empirical referents, I quite doubt. Colin Williams notes that aligning nation and state is a primary goal of nationalism; he distinguishes “state nationalism” from “ethnic nationalism” precisely because the state has an interest in confounding the two, in promoting the idea of a nation-state, and in directing any nationalism (here, group pride or chauvinism) towards the state, rather than the ethnic group.

And yet ethnic groups exist distinct from the state, as demonstrated by Peter Osei-Kwame and Peter Taylor in their spatial analysis of Ghanaian politics. Centralizers like Kwame Nkrumah can preach the gospel of “Ghana” as much as they like; that will not keep individuals from knowing who is Ashanti and who is not. In Ghana, as elsewhere, these ethnic identities, which can drive political preferences, are not simply identities, but can be based in solid cultural realities like language (Akan versus Djula, for example). And when people live in specific places (as all but nomads do), these dwellings collectively define regions, and these regions are not as nebulous as would be required to see them summarily dismissed.

Thus, when Alec Murphy analyzes the linguistically-based administrative territories in Belgium, namely Flanders, Wallonia, and Brussels*, he sees the construction of territorial identities where, prior to the reform, none existed. He is essentially identifying the territorial trap, but I disagree with his analysis, precisely because these territories have some empirical basis. There really is a difference between speaking Dutch and speaking French, and language plays a central role in the individual’s life and the community’s interaction. Murphy might note the imperfections of the boundaries drawn, but he doesn’t seem to dispute the fact that most locations in Flanders or Wallonia can be characterized as primarily Dutch-speaking or primarily French-speaking, even when, as often happens, a location is in Flanders administratively but is primarily French-speaking. So, yes, to reify a cultural Flanders based on its administrative boundaries is a problem when the linguistic boundaries of Flanders are elsewhere; but there still is a cultural Flanders. In fact, Flanders is distinct from the Netherlands in a way that Wallonia is not distinct from France, in that (historically) Flanders is primarily Catholic and the Netherlands is primarily Protestant. If we consider only linguistics, then, Flanders is just the southern part of an undifferentiated Dutch region; but if we consider language and religion simultaneously, it suddenly appears, and not that different in appearance from the “reified” administrative Flanders of Murphy’s concern.

There is some value to standardization, and thus some value to consistent use of one set of regions for continued analysis, across fields and over time. This is an attempt to avoid what is commonly called “the modifiable areal unit problem”. The phrase, though, originated with Stan Openshaw and Peter Taylor, who, while noting the merits of standardization, were much more concerned with discouraging inappropriate standardization. It is, for them, more important to get the right regions for the job, than to use the same regions all the time. At the very least, we need to balance the value of standardization against the value of a truly-fitting set of regions used ad hoc. (That many data are only available for administrative units like states is a different problem, not open to solution by any one scholar facing a data analysis.)

All of this points us to a better — more useful — regional concept. We can avoid the territorial trap by following some of the advice of Peter Taylor, and of Openshaw and Taylor, by identifying regions ad hoc, a different set of regions for each purpose, using only regions that are suited to our purpose, and to allowing for the possibility of overlap or border zones where a location might, for a particular purpose, be considered a part of more than one region (Brussels linguistically, for example). To avoid the territorial trap as experienced by the public at large is a grand educational problem, to which geographers can certainly contribute, but only with the collaboration of educators at all levels, including ordinary citizens (as parents, especially). But so long as geographers remain vulnerable to the trap themselves, to supposing that a set of boundaries drawn for one purpose is suitable to any number of other purposes, the message will not get across.

And we ought to go beyond this advice as well, to recognizing as I have noted elsewhere that regions do not need to be defined by boundaries, but can in fact be defined by central features (where we, in the manner of fuzzy logic, identify a region as the area near the feature, with nearness being context-dependent, rather than precisely-defined); in this case, there is no container. Of course, there is a difference between making suggestions and having them implemented, and between implementation in a small group (such as academic geographers) and in a large group (the general public). The territorial trap is a concern across the board. However, it is only a concern because regionalization has in the past been so conventionalized — this is the prerequisite to Agnew’s argument — and yet there is nothing inherent to the regional concept I originally introduced that makes it conventional, eternal, or all-inclusive.


* This is something of a simplification; there is also a small German-language administration, and certain powers and responsibilities are assigned not to linguistic communities but to geographical units. Flanders the geographical unit has merged with the Dutch-speaking community, while Brussels, properly a geographical unit, is notable as the primary area where both main language communities have jurisdiction.


John Agnew. ‘The territorial trap: the geographical assumptions of international relations theory’, p53-80 in the Review of International Political Economy, v1 n1 (1994 Spring).
James Crawford. ‘The creation of states in international law’. Clarendon (Oxford), 1979.
Robert H. Jackson. ‘Quasi-states: sovereignty, international relations and the Third World’. Cambridge University, 1990.
Robert H. Jackson and Carl G. Rosberg. ‘Why Africa’s weak states persist: the empirical and the juridical in statehood’, p1-24 in World Politics, v35 n1 (1982 October).
Jeffrey Herbst. ‘States and power in Africa’. Princeton University, 2000.
Alexander B. Murphy. ‘Linguistic regionalism and the social construction of space in Belgium’, p49-64 in the International Journal of the Sociology of Language, v104 (1993).
Stan Openshaw and Peter J. Taylor. ‘The modifiable areal unit problem’, ch5 (p60-9) in Neil Wrigley and Robert John Bennett, eds. ‘Quantitative geography: a British view’. Routledge, 1981.
Peter Osei-Kwame and Peter J. Taylor. ‘A politics of failure: the political geography of Ghanaian elections, 1954-1979’, p574-89 in the Annals of the Association of American Geographers, v74 n4 (1984).
Peter J. Taylor. ‘World-systems analysis and regional geography’, p259-65 in Professional Geographer, v40 n3 (1988).
Colin H. Williams. ‘The question of national congruence’, ch9 in Johnston and Taylor, eds.


Home of the Stewardship Project
and O.T. Ford