Go to Semiotic Review of Books Home Page
Go to SRB Highlights
Go to SRB Archives


SRB Archives

This review appeared in Volume 2 (3) of The Semiotic Review of Books.

Reading Urban Landscapes

by O.F.G. Sitwell

The City as Text: the politics of landscape interpretation in the Kandyan Kingdom. By Duncan, James S. Cambridge: Cambridge University Press, 1990, xiv and 229 pp., biblios, illusts., appendix, index. (ISBN O 521 35305 X)

Duncan is methodical. In the second, third, and fourth sentences of the introduction to his book he states his three general goals:

To provide a methodology for interpreting landscapes (...) To illuminate the way in which a landscape, understood as a cultural production, may be integral to both the reproduction and contestation of political power (...) To analyze the relationship between landscape and the pursuit of power in a particular place and time ... (Duncan, 3).

He begins the conclusion of his book by writing:

This book has been devoted to an exploration of the relationship between a landscape and the competing discourses within the discursive field of which it is a constitutive part. While it may be no surprise that a landscape is shaped by discourses, it may be less well understood how the landscape as a signifying system embodying cultural practices can "act back" on the competition between discourses at times thrusting certain ones to the fore and forcing others into the background (Duncan, 181).

Ignoring for the moment the issue of whether the text that comes between these two utterances justifies the claims they make, two things are clear. First, when Duncan begins he avoids the technical language of academic literary criticism. Second, at the end he uses it.

This development may have been the result of a conscious decision because Duncan sees himself as a pioneer. His profession is geography, and it is as a researcher who studies landscapes that he undertook the project that led to this book, but immediately after identifying his three-fold objective he states that his approach "represents a sharp break from the way that American cultural geographers traditionally have studied landacapes" (Duncan, 3). He expands on this point in Chapter 2, "Landscape as a signifying system", where he makes three claims with respect to what follows:

First, it stresses the role that landscape plays in social and cultural processes. Second ... a dialogue is established with researchers in ... the social sciences and the humanities in which the role of objects in social and cultural processes is studied. Third, the more general issues of theory-ladenness and the hermeneutic circle, of the role of commonsense social knowledge in social scientific explanation, and the status of data become important issues in landscape interpretation (Duncan, 13).

It will be apparent that, potentially, two distinct groups of readers could be interested in this book: cultural geographers, and that much larger group of scholars whose outlook, and whose methods of research and exposition have been shaped by developments whose roots lie in either structuralism or hermeneutics. Presumably each party would look for what the other has to offer.

As far as the content of City as Text is concerned, we can be guided by the table of contents. It divides the eight chapters following the introduction in two. The first part, which consists of Chapters 2 - 5, is identified as: "Towards an interpretive frame"; the second (Chapters 6-9) forms: "The politics of landscape interpretation in early-nineteenth-century Kandy".

The function of the second chapter is to make the case that the landscape which a people create can be read as a text. I shall deal with some key points in it later. Chapters 3 to 5 form a continuous flow. They begin with background information about the people of Sri Lanka. We learn, for example, that "it is probable that Lanka was settled by Indo-Aryans from northwestern India ..." and that according to Sinhalese mythology "it was in Lanka that Buddhism would flourish" (Duncan, 25). We are then told a version of Lankan history, and the very use of that name tells us that we are hearing it with Sinhalese rather than Tamil ears. Since the landscape that Duncan is going to interpret is first and foremost that of Kandy, the capital of Lanka after Tamil invaders from the mainland made earlier capitals unsafe for Sinhalese kings, the point of view is justified. The story is told with ever-growing detail until it culminates in events associated with the reign of the last king of Kandy, Sri Vikrama, who ruled from 1798 to 1815. It is these events that form the subject matter of the second part of the text.

In the second part it is the landscape formed by Kandy, both as it was, and, especially, as it was created through the building programme directed by Sri Vikrama, that forms the principal object of Duncan's attention. He considers it, or aspects of it, not once but three times, with each consideration having a chapter devoted to it.

As the labels used for the two parts of the book imply, methodology lies behind the division. In Chapters 4 and 5 Duncan had, in his own words, "only alluded to some of the symbolic significance of some of the landscape element prior to 1800". He then proposes, in Chapter 6:

(To) "thicken" the description by offering a "reading" of ... Kandy, its sacred and profane spaces, buildings, and architectural detail, which I suggest was the king's reading ... I will argue that this landscape is a text, written in the language of concrete, and that it communicated the governing ideas of political and religious life (Duncan, 87).

In the following chapter he offers as a parallel to the text written in brick, plaster, paint, and other building materials (i.e., the language of the concrete), those provided by three civic rituals: "the funeral (of one king) and inauguration (of his successor), the Asala Perahara (an annual ritual that entailed processions that visited all parts of the capital), and the ambassador's audience with the king" (Duncan, 153). His ambition here is to show how "the landscape (as just analyzed) was used in the production of charismatic rule" (Duncan, 119). Finally, in Chapter 8 Duncan offers two alternative readings of the landscape that was Kandy: one was the reading adopted by nobles of the court who became disenchanted with Sri Vikrama and successfully plotted his overthrow; the other was that of the peasants through whose labour the physical structures desired by the king had been brought into material being.

The heart of Duncan's method is provided by intertextuality. Simplifying somewhat, he considers readings of selected texts taken from three different classes of text, and presents the events that culminated in the betrayal of Sri Vikrama to the British as the interplay between the readings, as acted out in the politics of Kandy. It is the interplay to which the concept of intertextuality applies. The three different classes of text are formed by landscape, ritual, and language. Not being an expert on ritual, I shall focus my remarks on the interplay between texts of the first and last types, and still more particularly on the problems involved in interpreting the signs that form the first.

With respect to the verbal texts, we should begin by noting that what Duncan deals with might better be called literary traditions. There are two of these, each of which embodies a model of kingship. On the one hand there are texts of the Asokan tradition, according to which "a king should be mild-mannered, righteous ... protective of Buddhism and responsible for the welfare of his people" (Duncan, 38). The alternative, or Sakran model, which came from Hindu rather than Buddhist sources, presents the king as conqueror; the emphasis is on creativity, very commonly expressed by the enlargement of the kingdom achieved by military conquest, and other activity that would enhance the splendour of the king.

To recapitulate: the landscape of Kandy, especially as modified by the last king, is interpreted as text in Chapter 6. More particularly, Duncan shows that this text conforms to, and reinforces the message of, the Sakran tradition. In Chapter 7 attention is focussed on three sets of ritual, or "narrative" as Duncan also calls them. He justifies their inclusion on the grounds that they took place in, and were adapted to, that particular artifactual setting which is the "text" known as Kandy. In the last substantial chapter Duncan plots the reaction to the king's building programme.

To understand the method Duncan uses in Chapter 6 to interpret the landscape of Kandy, we shall revert to Chapter 2, for it contains a section devoted to "The rhetoric of landscape" (Duncan, 19-22). Here is to be found the heart of Duncan's thesis that geographers have something to offer to scholars in other disciplines. Essentially:

(It is the use of) tropes which allows a landscape to act as a sign system. The first of these tropes is allegory ... Through the use of various conventional forms -- signs, symbols, icons ... powerful people tell morally charged stories about themselves, the social relations within their community, and their relations to a divine order (Duncan, 20).

The second important trope is synecdoche. Duncan draws on Hayden White (1978/:73) to the effect that "synecdoche sanctions the integration of apparently unrelated particulars into a whole" (Duncan, 20).

Without that capacity, buildings would signify little because: "The complexity of the (allegorical) narrative can not be reproduced in toto in the architectural fabric of the city, but through synecdoche it can be effectively alluded to" (Duncan, 21). Finally, "a third trope that comes to mind is that of recurrent narrative structure" (Duncan, 22).

These concepts are brought into play in Chapter 6. It contains two tables of data, both of which have the same title: "The recurrent narrative structure of the landscape". The difference between the tables lies in the allegorical naratives alluded to by the elements of the landscape that form the data presented the body of the two tables. These narratives are different. One is "The world of the Cakravarti"; the other is "The world of the gods". The link between the architectural elements and the allegorical narratives is synecdochal. An example may help.

The city of Kandy was divided into eastern and western rectangles. The eastern one contained the residence of the king and also the most important religious building in the kingdom. The maps and illustrations that accompany the text show that this rectangle was itself divided in two by a wall that ran north-south across it. Because of the undulating crest that forms its top, this wall has been descriptively named the "wave swell wall". Duncan provides a photograph of it (Figure 10), and also a careful description of the architectural motifs that are literally built into it. In addition to the undulating waves of the crest, there are niches within it, some triangular, some circular; these are spaced in a pattern that is visibly related to the waves. The niches and their shapes carry messages that Duncan interprets as follows:

Each of these shapes and the overall arrangement of these shapes can be seen as a synecdoche for the cosmic Ocean of Milk. In Buddhist iconography the triangle represents leaves in general and the leaf of the sacred bo tree in particular. During the churning of the Ocean of Milk, the leaves of the trees on Mount Mandara were thought to have blended with the waters of the ocean to produce amrita, the elixir of life. Thus I would argue that the triangular niches in the wave wall referred to primordial creativity and fertility associated with the churning of the Ocean of Milk (Duncan, 102).

The location of the wali is also significant as the following passage shows:

According to both the Culavamsa and the Sangaraia Vata, Mount Meru, the central cosmic mountain, arose out of the cosmic ocean. This relationship was expressed spatially in Kandy where the eastern rectangle was bordered on its southern side by the lake (discussed previously, and named by the king: "the Ocean of Milk"). Running along this border was the wave swell wall, symbolizing the waves of the cosmic ocean breaking against the flank of Mount Meru (Duncan, 107).

Rather than describe every synecdochal element separately, Duncan provides a summary of them in two tables. He does more, for the tables themselves are condensations of data that are presented more fully in an appendix that contains four tables, one for each of the tables of Chapter 6, and another pair that support two tables that form part of Chapter 7. A qualification must be entered at this point. There are no references to the appendix in any of the four tables in the two chapters of the text. The converse is also true; there is no explicitly statement in the appendix linking the data it contains to the tables in the body of the book. The link must be made by the readers themselves.

I have no doubt Duncan intended that a link would be made because in the opening section of Chapter 6 he wrote:

These synecdoches (whether iconic, linguistic, or ritualistic) will be discussed in this and the following chapter and a listing of the major ones, showing their location in the landscape, their medium, and the narratives and subnarratives to which they allude is provided in the appendix (Duncan, 89).

Some pages later, Duncan makes the first of a number of references to the two tables of Chapter 6. Table 6.1 which lists landscape elements referring to "The world of the Cakravarti" provides, at a generalized level, the location of 92 such elements; each of these refers to at least one of three subnarratives to be found in the main narrative, though some refer to more than one. Thus the western rectangle itself, and excluding the buildings located within it, contains 10 synecdochal landscape elements; each of them can be read as a single reference to the Cakravarti; 4 of these references are to the first subnarrative, 4 to the second, and 2 to the third. In a similar way, the palace that stands in the eastern rectangle is the location of 22 synecdoches; between them they contain 25 references to the Cakravarti, 10 to the first subnarrative, none to the second, and 15 to the third.

When we turn to Table 1 in the appendix we find that its title is: Location of synecdoches for the narrative "The world of the cakravarti". Column 1 contains the locations; the first 6 entries are "western rectangle". So far, so good, though we might wonder why 6 rather than 10. The first synecdoche is the "rectangular shape" of the western rectangle; the medium of expression is "concrete", and the reference is to the third subnarrative. The remainder of the table contains similar information. All would be well, if it were not for the fact that the numbers do not add up. Problems such as the discrepancy between 6 and 10 occur throughout. The easiest one to demonstrate consists of the references to the second subnarrative. According to Table 6.1 there are a total of 6 -- the 4 of the western rectangle, and 2 others. Table 1 in the appendix shows only the 2 noted above, both located in the western rectangle, one consisting of its division into halves, and the other of its division into 21 blocks (that these may be taken as synecdochal references to the Cakravarti is explained to my Satisfaction by Duncan, 92-93).

Though I have laboured over the issue of the relationship of the appendix to the tables in the text, I see the problem I have identified as an irritant, not a major flaw in the book as a whole. In fairness to Duncan, I should also emphasize that, as the quotation above shows, he did say that the appendix contains a "listing of the major (synecdoches)". The problem is still worth mentioning, however, because it is symptomatic of one that will trouble those who read the text closely with a view to relating it to the maps shown in Figures 6 and 8, and in relating the information provided on those two maps to each other. As a geographer, and moreover, one who believes that the examination of cultural landscapes has much to offer to the academic community, I regret the fact that (providing we are reading the book at the level of semiology) the most striking flaw revealed by the kind of close reading normally reserved for the examination of a doctoral thesis is of the kind that a geographer above all others should avoid.

The parentheses of the previous sentence imply that I have other reservations about the book, but before I turn to them I want to assess Duncan's achievement in terms of what he saw himself as doing before adding some regrets of mine on what he did not do. In my reading of the book Duncan makes two major claims: first, that he is doing something new in cultural geography; second that this new method will interest scholars in other branches of the social sciences and the humanities.

In the case of the first claim Duncan himself adds some qualifications, identifying other geographers who are currently devoting their energies to the issue of interpreting the cultural landscape as text. While I am very happy to see him do this I feel that he undervalues the work of a number of geographers whose publications predate his own by up to 30 years. It may be that in the case of J.B. Jackson, Duncan feels that "interpretive authority is assumed to result from an unmeditated relation between what is simply 'out there in the landscape' and the informed scholar's stamina for field or archival exercises" (Duncan, 11-12). This raises the issue of the extent to which theory has to be articulated in order to qualify as theory. It was generally accepted that the students of the 1960's were right in massing behind the war-cry that to refuse to take a political stand is to actually take such a stand. In a similar spirit, Karl Popper (1969, 46/47) argues that there is no such behaviour as pure observation; rather, all observation in shaped by theory, no matter how little the observer may be aware of the theory and its presuppositions. As Jackson published more than two dozen papers in the years 1951 - 1970, a selection of which was republished (Zube, 1970), and given the fact that these inspired Meinig to search for the rules desired by those who wish to learn how to read landscapes (Meinig, 1979b in Meinig 1979a), he deserves more than to be merely mentioned, along with D.W. Meinig himself and also W. Zeiinski, in a single sentence. In addition there is the work of Paul Wheatley, to which I shall return below. Even more surprising, given Duncan's emphasis on the need for explicit theorization, is the omission of Glassie (1975). Perhaps Duncan feels that Glassie's narrow focussing of vision on the text provided by the folk housing of Virginia, and consequent lack of obvious concern with intertextuality, justifies the omission (or perhaps it is Glassie's commitment to structural analysis in the manner of Lévi-Strauss).

So far I have limited my list to geographers, but it is far from the case that no scholar from outside that field has ever tried to interpret the meaning contained in the signs of the cultural landscape. This, too, Duncan acknowledges with references to the work of Roland Barthes, Michel De Certeau, and Clifford Geertz (Duncan, 3). But this is all work of the last 25 years. Two generations earlier Emile Male (1910, English translation 1913) sketched by example the method used by Duncan in Chapter 6 and 7 of City as Text; all that is lacking is the technical vocabulary of modern literary criticism.

Given the accuracy of these observations, and noting that Duncan concedes much of the ground in principle, especially with respect to the work on non-geographers, how should we evaluate his first claim? With considerable sympathy in my opinion, for, despite my strictures, to my satisfaction by Duncan, 92-93). those unfamiliar with the methodology of contemporary literary criticism will find in City as Text a clear, as well as a scholarly introduction to a realm that geographers would profit from exploring.

If, then, Duncan's work is to be welcomed in geography, what does it have to offer to scholars in other fields? The answer to this question turns on his claim, quoted above, that in making his in tertextual analysis of a particular landscape, he has something to contribute to "the more general issues of theory-ladenness and the hermeneutic circle, of the role of commonsense social knowledge in social scientific explanation, and the status of data". It is in Chapter 6 that he takes up this theme, writing:

One way to unlock the mythic structure of a landscape, the hermeneutic circle of landscape and myth, is to begin with one element of the landscape and securely anchor it through synecdoche to a set of narratives. Having done this one can then move on to relate that landscape element to other elements of the landscape. These can be explained in turn thorough synecdoches that refer back to the initial narratives. This process entails simultaneous reconstruction of a landscape text and a myth system, with a constant tacking back and forth between elements of each (Duncan, 97).

My reading of this passage is that it contains, though only by implication, a belief that the a secure anchoring" that would, be achieved, unlock the hermeneutic circle is provided by the cultural landscape, and that it would be so provided by the very concreteness of that landscape. If this is correct, then Chapter 8 subverts the hope. It will be remembered that in this chapter Duncan offers two alternative readings of the landscape that was Kandy; these, moreover, are separate from the reading of the king himself. In his conclusion, Duncan underlines the point "Landscapes never have a single meaning; there always exists the possibility of different readings (Duncan, 182). By itself that is unobjectionable, but it leaves us with the hermeneutic circle unbroken, and the status of data in the social sciences as problematic as ever. Never mind what the landscape "says"; the only thing that matters is the answer to the question: Who has power?

Let me now move on to what might be called a sin of omission, but before dealing with it I want to commend the City as Text for its balance, its clarify of organization and (leaving aside the points made above) its clarify of presentation. Though I am not an authority on south Asian history and its religious traditions, and therefore judge by the appearance of and the professional care invested in, the scholars armament of bibliography and notes, I have no reservation in saying that the book creates an air of authority. Read at the level of the author's explict intent on, it deserves serious attention. There are, however, other levels at which it can be read.

Though Duncan nowhere uses the term semiology, every reader of the SRB will agree that The City as Text belongs to the field. What I suggest is that, in order to justify his claim that, as a geographer reading a landscape, he would be able to help resolve the difficulties associated with the general issues of theory ladenness and the hermeneutic circle, etc., he should have gone beyond semiology into semiotics. In making that assertion I am accepting the distinction made by Deely(1990, 2):

If the essential breakthrough in semiology is to take linguistics as a model and apply linguistic concepts to other phenomena -- texts -- and not just to language itself (...) then the essential breakthrough of semiotics (...) is to see that the phenomena of semiosis requires a model within which linguistic phenomena taken together appear as a subset of a much broader range of sign-activity which cannot even be confined to the cultural side of the line defining our ideas of the natural world (...) What I would like to do here, then, is suggest a way of broadening literary semiotics to include in some sense natural phenomena as well as purely cultural and literary texts.

To make my point let us consider the use of the left hand. Since the work of Hertz (1909 in Needham 1973), the academic community should have been aware of the observation which is so widespread as to be legitimately called panspecific, namely that to offer the left hand in greeting is not done, is a consequence of the fact that, to use language plainer than Hertz dared to use, in societies of the type commonly called "primitive", the left hand was used to wipe traces of faeces from the anus following defecation. As a consequence, the hand smelled unpleasant. That it smelled unpleasant, we, who understand in principle how evolution "instills" traits that aid survival, can also understand that smell is a sign which belongs, in Deely's usage, to semiotics. The associations we have with the English word sinister, as also those patterns of behaviour that correspond to the linguistic usages found among the English, belong to semiology. What began in the realm that is proper to the natural is not merely transferred to the cultural, but survives there long after human ingenuity has removed the need.

Though Duncan does not mention any instances where left and right are used as signs of cultural significance in Kandy, he does mention two other signs that, beyond reasonable doubt, belong to the realm of semiotics, as well as one that probably does, and one that may. However, he deals with each of them in passing; true, sometimes that passing amounts to a paragraph (Duncan, 101, 116, 117, 124, 147, 169), but in total they receive less attention than is paid to the wave-swell wall, and when mentioned tend to be associated with minor features of Hindu and/or Buddhist ritual. The signs in question are height, centrality, rectangularly, and number. Of these, Duncan's seeming indifference to centrality and rectangularly are surprising, because they were considered at length by Wheatley (1971). Though Duncan cites that work in connection with these signs, he seems oblivious to the message of their quasi-universally. If Wheatley's work is not authoritative enough, there is also the explanation of the origin of the meanings that are all but invariably attached to the signs of height and centrality offered by Sitwell (1984). Sitwell was heavily influenced by the attempt, successful in his opinion, of Laughlin and d'Aquifi (1974) to ground the roots of the observations measured by structuralists in the bio-electronic workings of the human brain.

What completes the mystery is the note of implicit regret on which Duncan ends. He clearly would like to have his book seen as being relevant to a world larger than that of l9th century Sri Lankan history I believe that if he had taken advantage of the Endings emerging from research in neuropsychology he could have found grounds for asserting that the landscape provides a secure base from which to break into the hermeneutic dzle. He might then have gone on to observe that kings (a role that can be extended to include chief executives in general not only always seem to have the choice of either serving the people over whom they rule, or of pursuing the will-o-the-wisp of military triumph celebrated in a building programme emblematic of pomp and circumstances, but in any long established regime, some "king" will always pursue the latter to the point where he either provokes his" people to revolt or he is defeated in war. And if neuropsychology lies outside Duncan's field of awareness, he could have drawn on Hocart (1970), which he does cite, and argued that any truly species-wide phenomena such as the ones in question must date from very early in human history, and may even be of prehuman origin. Had he done either of these things, his book would have interested the entire readership of SRB.


Deely, John(1990) Basics of Semiotics. Bloomington: Indiana University Press.

Giassie, H. (1975) Folk Housing in Middie Virginia. Knoxville: University of Tennessee Press

Hertz, R. (1990) "La Prominence de la main droite: estude sur la polarite religieuse" Revue philosophique 68: 553-580. Republished in English in R. Needham, ed. (1973) Right and Left: Essays on Dual Symbolic classification Chicago: University of Chicago Press: 3-31

Hocart, A.M. (1970, originally published 1936) Kings and Councillors: An Essay in the Comparative Anatomy of Human Society. Edited and with an Introduction by Rodney Needham. Chicago: University of Chicago Press

Laughlin, Charles D. and E.G. d'Aquili (1974) Biogenetic Structuralism. New York: Columbia University Press

Male, Emile (1913) The Gothic Image: Religious Art in France of the Thirteenth Century. London: J.M. Dent (translated from the 3rd French ed. of 1910; republished by Collins, 1961)

Meinig, D.W., ed., (1979a) The Interpretation of Ordinary Landscapes. New York: Oxford University Press

---(1979b) "Reading the landscape: an appreciation of W.G. Hoskins and J.B. Jackson." In Meinig 1979a, 195-244.

Popper, Karl R. (1969) "Science: conjectures and refutations" Conjectures and Refutations London: Routledge and Kegan Paul, 3rd ed. (1st ed. 1963). Originally published in C.A. Mace, ed. (1957) British Philosophy in Mid-Century

Sitwell, O.F.G. (1981) "Elements of the cultural landscape as figures of speech" Canadian Geographer 25.2: 167-80

Wheatley, Paul (1971) The Pivot of the Four Quarters: A Preliminary Enquiry into the Origins and Character of the Ancient Chinese City. Chicago: Aldine

Zube, Ervin H., ed. (1970) Landscapes: Selected Writings of J.B. Jackson. Amherst, Mass.: University of Massachusetts Press.

O.F.G. Sitwell, B.A. (Oxon) 1958, Ph.D. (Toronto) 1968. Educated narrowly in geography, but in the liberal-arts tradition (unfashionable since the 1960s) that sees geography as the attempt to understand why countries have the character they do. Currently Associate Professor, Department of Geography, University of Alberta, Edmonton. Initial publications relevant to semiotics discussed in Paul Bouissac (1988) Semiotics in Canada II. In T.A. Sebeok and J. UmikerSebeok, eds. The Semiotic Web Berlin: Mouton de Gruyter, 161-162. Recent relevant publications: Sitwell,O.F.G.(1990) "The expression of ideology in the cultural landscape: a statement of general principles illustrated with examples taken from Edmonton." In P.J. Smith and E.L. Jackson, eds. A Worid of Real Places Edmonton; Department of Geography, University of Alberta; University of Alberta Studies in Geography: 175-189. ----- (forthcoming) A human geographer looks at religion Religious Studies and Theology

Go to Semiotic Review of Books Home Page
Go to SRB Highlights
Go to SRB Archives