Cartosemiotics

1. Basics

1.1. The field of study

Cartosemiotics, also called cartographic semiotics, is the semiotic study of cartographic models (or cartographic representation forms), such as maps, globes, relief models, animations, and many others. These models have in common that they represent the space of the earth (or, by extension, of another celestial body or of the sky) by means of a model space. Berlyant (2001-02) has offered some thoughts about a comprehensive semiotics of cartographic representation forms (he calls it geosemiotics). However, semiotic research has mostly been devoted to the map in the traditional sense (as defined in EDC, entry 21.1 (German and Russian sections) and entry 811.1), that is, to the most versatile, most widely used and best studied kind of cartographic models. Its major characteristics are as follows. First, it shows the earth’s surface, at each point, as seen from vertically above and has no dimension free to accommodate elevation, having only two dimensions (the horizontal ones). Second, it has a projection which consistently and uniformly matches points on the map plane with points on the curved globe surface. As a consequence of both circumstances, it is a plan representation. Third, it is reduced according to a stated scale (ratio of reduction), so that features are shown smaller than they are. Since projections involve distortions in the rendering of certain geometric traits, a stated scale is preserved over the map face only to the extent that the underlying projection permits. Fourth, information about the mapped section of earth space is expressed by means of man-made marks. In view of the aforesaid state of research, an overview of cartosemiotics must currently be limited to the semiotics of maps. This temporary thematic limitation also explains why “map semiotics” is often used as a quasi-synonym of “cartosemiotics”.

The aims of cartosemiotic research are intellectual enlightenment as well as practical application. The latter fact is easily understood if one considers that map makers engage in various activities of sign production: they structure the information which is to be conveyed, select and use means of expression, and (ideally) see to it that a map can be read as intended (s. 6.1).

As a distinct area of scholarly inquiry, cartosemiotics arose in the 1960s with about half a dozen studies and has by now reached a fair level of maturity. It is particularly popular in Central and Eastern Europe. Although some cartosemiotic studies have been done by semioticians (such as W. Nöth and D. Schmauks), the bulk of the relevant research has been carried out by cartographers and published in the cartographic literature; consequently, it is still insufficiently known in general semiotics. Many contributions have originated in the Commission on Theoretical Cartography of the International Cartographic Association (ICA) and in international correspondence seminars, published as annual discussion papers (Schlichtmann and Wolod-tschenko 2005). Research done up to the 1990s has been summarized in an international inventory compiled by a working group of the ICA (Schlichtmann, ed., 1999); here also earlier comprehensive reviews are cited. Among relevant entries in reference works, those written by Nöth (2000: 487-490), W.G. Koch (2001) and Tainz and Koch (2002) are the most recent ones. A dictionary of cartosemiotics is in preparation. It may be added that semiotic thinking has also spread into the modern field of geoinformatics, which is related to cartography (Brodersen 2008, especially ch. 9).

In cartosemiotics, there are different conceptual and terminological frameworks or schools of thought, just as there are in general semiotics. This is particularly noticeable in the analysis of map symbolism – also called map language or cartographic language -, that is, the sign-system type that underlies mapping and map use. A synthesis of various approaches is desirable but may only be partially feasible. Rather, we must live with different conceptual orientations and be able to “translate” between different frameworks. On the other hand, all schools of thought must address the same problems. The discussion which follows in sections 2 to 7 is based on the present author’s views of cartosemiotics.

Most research in cartosemiotics has been devoted to map symbolism. Starting with the pioneering works of Aslanikashvili (1967, 1974: 20-52), Bertin (1967, 1983) and Bocharov (1966), several conceptions of map symbolism have been developed. Syntheses or comprehensive discussions of central problems have, in recent times, been presented by Casti (2000), Head (1984 and later), Hussy (1998), Kekelia (2006), Liouty (1984 and later), Neytchev (1997, 2001), Nižnanský (2006), Palek (1986, 1991, incorporating findings of co-researcher Nebeský), Pravda (1990 and later), Schlichtmann (1985 and later), and Żyszkowska (2000). A comparative study of the existing conceptions is a task for the future, although Head (1991: 256-259), Pravda (1993, 1994: 32-40) and Wolodtschenko (1991, 1994a) have taken some steps towards this goal. The characteristics of map symbolism are by now reasonably well known. Still, as may be expected, scholars continue to test and develop their ideas and sometimes find it necessary to modify their conceptions and terminologies.

Cartosemiotics overlaps, of course, with other fields of inquiry. It contributes to computer-assisted cartography and the study and development of geographic information systems. Further, it has strong affinities to cognitive science and artificial intelligence, i.e., specialties with which it shares an interest in conceptual modelling.

1.2. General points

Although terms are usually introduced where they are first used, three general points shall, for stylistic convenience, be clarified right at the outset. They relate to the semiotic conception underlying the following arguments, to the status of places, and to the contexts in which signs in maps are embedded.

As for the first point, a sign is taken here as a bifacial entity consisting of a content (meaning) and an expression (sign vehicle, form). (Where required, a content is cited in single quotation marks, an expression between slashes.) Expressions are coupled with contents by descriptive rules which collectively constitute a code. In cartography, the aggregate of sign contents relating to a mapped section of earth space is often called cartographic information (EDC, entry 444.0) or information for short. Referents, finally, are certainly taken into account (e.g., in studying sign production), but they are kept conceptually separate from signs.

Second, mapped information is about places or localities, that is, segments of terrestrial space, or, put differently, about objects which occupy such segments. The terms “object” and “place” will be used interchangeably, depending on which aspect is of interest. Places have attributes or traits, including location in space.

Third, signs in maps function within contexts, where “context” is understood in a general sense (s. 7). The latter relate to the territory (the aggregate of mapped places along with their characteristics), the map author (map maker) and the audience (set of map users). The term “map author” refers to one or more persons who, individually or collectively, decide how the map is to be made. Of prime interest among the users is the subset for whom the map is intended (the intended audience).

2. Major themes of cartosemiotics

How to arrange the subject matter of cartosemiotics? Three basic themes suggest themselves immediately:

(1) Map symbolism (map language), that is, the type of sign systems that are manifested in individual maps (more exactly: in individual map faces) (s. 3).

(2) Processes in which humans handle signs, or sign processes for short (s. 6).

(3) Contexts in which sign systems and sign processes are embedded (s. 7).

Empirical work, however, reveals that this tidy overview is incomplete and that two further themes require attention.

(4) Many signs encountered in the map face are established by the map author and must be introduced for the benefit of the map user. This is done by means of marginal notes, named for their usual position in the margins of the map sheet, outside the map face (s. 4).

(5) Entries in a map face have the basic and criterial function of conveying information about the mapped territory, which, for its part, is one of the contexts mentioned above. But they may also inform about other contexts (and so may material placed outside the map face). Relevant signification phenomena are peripheral to, or superimposed on, those which relate to the world as represented (s. 5).

The basic framework laid out here, along with refinements to be added later, should not be seen as a bank of well-separated pigeonholes but rather as a set of searchlights which illuminate overlapping domains. (In a systematic discussion, such overlaps make for cross-references.) The framework has proven adequate in accommodating current knowledge but remains open to revision and expansion.

3. Map symbolism

Map symbolism is the core subject of the semiotics of maps. It is highly complex, that is, it has components with different characteristics. Notes on this complexity follow; afterwards map symbolism will be discussed under two systematic aspects.

3.1. Complexity

In the description and analysis of map symbolism, one comes across four facets of complexity; these will be referred to repeatedly.

First, the sign inventory of a map comprises two major types of signs. Some of them are pictures, i.e., images of objects (in the first place plan images). Many of their perceptual traits as well as their spatial arrangements are ultimately derived from traits and arrangements of the represented objects. Other signs are what Eco (1976: 227ff.) calls combinational units; they are artifacts, created primarily as instruments of signification and communication. In this respect they rather resemble signs of natural language and can in principle (although certainly not in every detail) be analyzed like the latter. Consequently, we must recognize a picture component and an artifact component of map symbolism. The difference between the two is most obvious where sign units are to be operationally isolated and where their combination to more complex units is studied. The two components reflect different modes of sign production, i.e., invention and replication (Schlichtmann 2006: 24f.).

Second, on both the content and expression sides, map symbolism comprises plan-related and plan-free components. Thus, the cartographic information is divided into plan information (contents pertaining to the two-dimensional earth space as represented) and plan-free information (other contents). Plan information is usually expressed by plan-related expression traits (we leave aside coordinates and other numerical entries), but such traits are also widely used to render items of the plan-free information.

The two aspects mentioned before are conceptually separate but empirically intertwined. This is because in maps – as opposed to aerial photographs – image characteristics of objects are, in most cases, at the same time plan characteristics.

Third, space is alternatively conceptualized in two ways: either as discretized into segments or as a continuous collection of points. A river, a meadow or a county cover segments in discretized space, while values of longitude, elevation or temperature are associated with points in continuous space.

Fourth, the means of expression have various origins. There are plan and side views of objects, and there are emblems and similar entries, writing marks, and marks with abstract traits, including diagrams. Most entries (such as plan images and abstract marks) are created during mapping, others (like writing) are imported from extra-cartographic contexts. Some entries (such as pictures, writing, emblems) come already endowed with meaning or at least with a basic core of meaning; a greater part, however, are abstract in the sense that they are not pre-assigned to contents. Here the expression-content links must be specifically established by the map author.

3.2. First systematic aspect: focus on signs

As noted earlier, map symbolism is to be considered under two systematic aspects. Under the first, the focus is on signs, more exactly: on their components and constitution.

When producing signs, a map author draws on two universes, one made up of concepts, the other of percepts. Both are structured, that is, items or elements within them are organized by relations. (For simplicity, we usually consider items only.) When elements taken from the two universes are coupled, they function, respectively, as contents and expressions of signs. The author of a specific map normally starts by selecting the concepts which are to become sign contents, then selects perceivable entries which are to become expressions and couples these items with the selected concepts. To put it in a simpler but less precise way (as will be done hereafter): the map maker selects contents, then selects expressions and assigns them to contents. The latter two processes are governed by rules.

In view of the preceding overview, three sets of phenomena invite study:

(1) The universe of concepts from which sign contents are selected. It contains all concepts which can function as contents in maps (more exactly: as content types).

(2) The expression material, i.e., the stock of perceivable items from which expressions are selected. It contains all perceivable elements which can function as expressions in maps (more exactly: as expression types).

(3) Codes, i.e., sets of rules by virtue of which expressions are selected and coupled with contents.

3.2.1. The universe of concepts

In principle, all information can be conveyed that relates to places in earth space. To map it, one must be able to render the location of each place by a position in the map face, which is done through the point-to-point matching specified by a map projection. The stock of conveyable concepts is inexhaustible. For example, there are indeterminably many observed and conceivable shapes of lakes. Further, all of them can, in principle, be expressed, including those which have not been encountered before.

The cartographic information can be decomposed in various ways. In the case of maps – which are plan representations – it is convenient to divide it into plan information (or plan-related information) and plan-free information (s. 3.1); for other cartographic models, other divisions may be appropriate. Under either head there are numerous categories. Those of the plan information (relative location, absolute location, plan dimensionality, plan shape, and some others) are, in principle, ubiquitous (expressed in any map) and make up a closed class. The categories of the plan-free information (landforms, water bodies, roads, traffic on roads, etc.) form an open class from which, for each map, the relevant ones are selected.

The aggregate of mapped contents is made up of units (‘lake’, ‘bridge’, ‘30-50 m deep’) and relations (‘beside’, ‘older than’). Relations organize a set of content units into a system. There are many kinds of relations, defined by various criteria, and they form an intricate network. Spatial relations, widely discussed in the literature on cartography and geographic information systems, are one class among others. Some types of conceptual relations leave no trace on the expression side, others are reproduced in the transcription. By way of example, consider a map of grain crops. The contents ‘wheat’, ‘barley’, ‘maize’, apart from having a common core (‘grain’), stand in the relation ‘different from’. This is one of several paradigmatic relations (understood, in the broadest sense, as relations of equivalence), which are prominent in the organization of the contents (Schlichtmann 2004). On their reproduction see s. 3.2.3.

In map symbolism, just as in natural language, some content units are directly or immediately coupled with the perceivable entries, while others are implied, i.e., conventionally called to mind by virtue of various relations. For example, ‘vineyard’ implies ‘area’ (as an obvious plan feature); ‘wheat’ and ‘barley’ imply ‘grain crop’ (as an included concept). Further, ‘pingo’ implies – or is conventionally associated with – ‘formed under Arctic conditions’ (for a geographer), and ‘glacier’ implies ‘danger’ (for a knowledgeable mountain hiker). Thus, a perceivable entry in the map face may be linked to a bundle of semantic features, immediately expressed or implied. Such bundles can be subjected to a compositional analysis, as proposed by Eco (1976: 105-129) and exemplified for the cartographic information by Du (2003: 1522).

3.2.2. The expression material

The material from which expressions are shaped is usually visual in nature. (In addition, tactile and auditory items are used in special cases but will not be considered further.) It comprises, in principle, all visible items that can be intentionally inscribed in the map face. These include elements which come already endowed with meaning (s. 3.1). The expression material has an autonomous perceptual order, which is frequently put to service in signification.

Putting it most generally, the expression material consists of marks (segmental units), traits (characteristics), and relations which hold between marks and between traits. In cartography, marks are often called symbols, and their creation is known as symbolization. A mark must carry traits in order to be visible. Traits (like shapes or colours) are usually inherent in marks but may also be associated with points in continuous space (like positions in the map face). Traits are values of visual variables (or graphic variables); for example, a specific symbol shape is a value of the variable of shape. Many cartographers have systematized visual variables. The list proposed by Bertin (1983: 96) is the best-known one and is easy to memorize. But it is incomplete and has been extended by other authors. As for relations between traits, paradigmatic ones, which link values of a visual variable (e.g., /red/, /green/, /buff/), most readily come to mind. Further, marks stand in spatial relations (disjunction, overlap, coincidence, and inclusion).

How the expression material is employed depends – or should depend – on the intended map-use or map-reading tasks (Schlichtmann 2003) and practical applications. In this context it bears noting that Bertin’s familiar summary concerning the suitability of visual variables (Bertin 1983: 96) conflates two kinds of tasks: reading for difference, rank order, etc. (which tasks are not inherently spatial) and derivation of spatial distribution phenomena. Summaries developed by some subsequent authors incorporate the required distinction (Geels 1988: 18; MacEachren 1995: 279). Finally, empirical studies have aimed at evaluating the use of graphic material in atlases (Morrison 1984) and at identifying and defining known rules of cartographic visualization from a semiotic point of view (Żyszkowska 2000).

3.2.3. Codes

A code, in a wide understanding of the term, is a set of rules which govern not only the mere coupling of chosen expressions with contents but also the choice of expression material which is suitable to express specific concepts. These two processes, although conceptually distinct, are in practice not easy to keep apart. They are often subsumed under the common head of transcription (of the information). Rules belonging to a code will here be called coding rules or transcription rules. They are, to a large extent, spelled out in the marginal notes, especially in the legend.

Insofar as a code guides the selection of suitable expression material, it reflects intended map-reading operations and anticipated practical applications of the cartographic product. Sometimes a map author does not know or observe all of the relevant rules. For example, he may erroneously render a set of ranked values by entries which are visually different but not ranked, a transcription which makes the resulting map difficult to understand.

A map author may rely on a code which already exists or set up a new one. Further, in studying a specific map, we usually must recognize different codes or code components, distinguished by various criteria.

Coding rules are more or less stringent. At one extreme, plan traits of places (such as absolute location and shape) are rather faithfully reproduced by characteristics of the corresponding map symbols, albeit within the limits set by the map geometry. At the other extreme, items of information (usually plan-free ones) are rendered on the basis of relatively free decisions, so that a mill may be indicated by a red circle or a black rectangle or yet something else. The freedom of transcription is greatest (although not absolute) where the expression material is abstract, i.e., not pre-assigned to meanings (s. 3.1). In this case it is possible to represent indeterminably many and varied mappable phenomena by re-using a limited stock of means of expression, which makes for economy of sign production. On the other hand, it necessitates detailed explanation. Transcription by abstract items is most widely encountered in scholarly map making. Finally, coding rules vary not only in stringency but also in generality. Again taking extreme cases, some pertain to maps in general (global coding), others to individual maps (singular coding).

Cartographic practice has given rise to several, partly incompatible principles of transcription, on which the relevant rules are based. The two most common principles are symbol standardization (better: standard coding) and homology of contents and expressions. In the case of standardization, map makers assign the same expression type to a given content type in all or at least many maps. This procedure is practically useful only within specified thematic domains (weather charts, stratigraphic maps in geology). As for homology, imagine that, in a map of grain crops, wheat fields are marked yellow and barley fields pink. Now, /yellow/ and /pink/ are related like the corresponding contents ‘wheat’ and ‘barley’: they have something in common and are different (but not ranked). Thus, the organization of the expressions is homologous (isomorphic, structurally analogous) to that of the contents. Homology is a form of iconicity; it pertains not to units but to relations between units. Here the code regulates the expression-content links not just for individual signs but rather for systems of such. In transcribing the plan information, homology is usually brought about automatically by point-to-point matching; in rendering plan-free contents, one must deliberately create it (this is the most frequent way of extending iconicity to the mapping of the plan-free information). Either way, the autonomous perceptual organization of the expression material is put to service in signification. In the above example, it is the paradigmatic order of the contents which is reproduced by that of the expressions. This transcription principle is prominent in geographical mapping, the chief reason being that it facilitates spontaneous recognition of spatial distributions (Schlichtmann 2003: 56). A second major case of homology is familiar from certain diagram symbols. Here spatial relations between marks (such as overlap and inclusion) are fruitfully employed to reproduce relations between sets of items (e.g., between the set of persons working in a town and the subset of workers commuting into the same town).

3.3. Second systematic aspect: focus on systems

Under the second aspect – or the second searchlight – we consider systems, that is, structured assemblages of items. Again three topics must be dealt with: units, relational order, and combinatorics. In detail:

(1) An analysis presupposes units, which may or may not be distinctly delimited.

(2) Units are linked by relations; this is why we can speak of systems.

(3) Units may combine to form units of higher order (and greater complexity) according to combination or construction patterns.

For convenience, the order by relations will be considered last.

3.3.1. Units

Units of map language are defined by various criteria and are more or less complex. Several researchers have systematized such units. For reports about, and comparisons of, their classifications see Head (1999: 19f.) and Wolodtschenko (1999). One may identify content units, expression units, and sign units (considering content and expression at the same time). Only the latter will be considered here. The present author recognizes signs at three levels. His conception, as developed below, differs in some details from that expounded in earlier publications (e.g., Schlichtmann 1998: 104f.). How units are identified will be noted only in a general way (details in Schlichtmann 2001).

As noted in s. 1.2, a map informs about places. A place is mentally singled out under a focus of interpretative interest (on this concept see W.A. Koch 1971: 302-308). It may be spontaneously recognized as a single item (an island, a railway line, a benchmark) or as an assemblage of items which are considered to belong together (a group of islands, an industrial town surrounded by commuter villages). Different foci may be applied at the same time or in sequence, therefore the identified places may overlap, or one may be included in another. Sign constitution by focus is familiar from map interpretation (sign reception, s. 6.3). A map maker, too, often employs graphic means so as to enjoin map users to take a specific entry or set of entries as indicative of a unit.

Topeme. If a place is shown as a single item, the corresponding sign is a topeme. This is the smallest self-contained entry in a map, hence it can, in principle, occur by itself. Its content is a complex concept of a place along with relevant characteristics of this place, among them its location. Its expression usually contains exactly one locator. This is a visually unitary symbol (i.e., a symbol that can be spontaneously seen as a single mark), the position of which in the map face expresses the absolute location of the mapped place. (A symbol which is not a locator is a complement.) It is the map maker who establishes the unity of the topeme by entering a locator.

Topeme complex. If a place is an assemblage of items which are considered to belong together, a complex sign may be identified which is made up of several topemes (Schlichtmann 2001: 56). Such complexes may overlap or may themselves be grouped in larger assemblages. In a typology of such complex units we would encounter all the descriptive models of spatial organization which are familiar to geographers: distribution area, patterned assemblage of points, network, mosaic, and partitioned area, among others. Fabbri (2000) studies the spontaneous identification of topeme complexes, which he calls chunks, in map interpretation.

Minimal sign. A topeme contains what we will provisionally call minimal signs. These are signs the expressions of which cannot be broken down into units which are themselves expressions, i.e., which convey meanings. They are equivalent to the monemes (or morphemes) of natural language. Recall, from s. 3.2.3, the land-use map with a number of differently coloured patches, each patch corresponding to a field and each colour indicating a crop. Then, if /yellow/ stands for ‘wheat’, then ‘wheat’ and /yellow/ jointly form a minimal sign. Further, each point within a patch has a position in the map face which, in turn, stands for the absolute location of the corresponding point in earth space. Each location value and the matching position value jointly make up a minimal sign. Thus, there are minimal signs of two kinds in a topeme: some relate to the mapped place as a whole, others to points within it. Here we are encountering the two space conceptualizations noted in s. 3.1.

When isolating minimal signs, an analyst is reminded of the fact that means of expression have different characteristics which are linked to their origins (s. 3.1). Let a map show, at a number of places, the schematized side view of a factory building which indicates an industrial plant. Each symbol token of this shape has one of several colours, and each colour expresses the production orientation of the industry: ‘iron-processing’, ‘chemical’, etc. The respective minimal signs – which belong to a family of several types – can be isolated by a commutation test, and the minimal sign with the meaning ‘industrial plant’ – which is invariant over all relevant entries – can be isolated by a deletion test, both tests operating within a single map face. If now the iconic factory-building symbol is replaced by a circle or another abstract mark, ‘industrial plant’ no longer has a separate graphic expression. Nevertheless, ‘industrial plant’ is correctly decoded by the map user. To account for this fact we assume that the meaning component ‘industrial plant’ is projected from the legend (one of the marginal notes) onto each of the relevant entries in the map face. The situation described here arises where abstract means of expression are used.

Relation between topeme and minimal sign. The topeme occupies a key position in map symbolism. In creating a topeme, a map maker singles out a piece of space and treats it as internally undifferentiated or, more exactly, disregards any internal spatial segmentation which may be present. Information is provided about the place as a whole; for example, one maps the corporate area of a city as a whole and disregards houses, streets, parks, etc. within it. This way the corresponding segment of space in the map face becomes free for the employment of means of expression which usually convey information about the place as a whole (such as ‘city’ or ‘80,000 inhabitants’) and sometimes about objects contained in this place (such as ‘hospital’). This information is made up of the contents of minimal signs.

3.3.2. Combinatorics

Units may be combined to form more complex ones. The underlying rules and patterns constitute a combinatorics. One may study how either contents or expressions are combined or deal with both components together and consider how complex signs are built up of simpler ones. A combinatorics of the latter sort is a syntax (as understood in linguistics). Following Ratajski (1976), we recognize two kinds of map syntax, a local and a supralocal one (Ratajski’s terminology is different). The local syntax pertains to the combination of minimal signs within topemes (Schlichtmann 1994b), and the underlying construction plans are artifacts of the sign system. The supralocal syntax covers the arrangement of topemes and their integration into larger configurations, and the underlying combination patterns ultimately reflect factual arrangements of objects in space (s. 3.1).

In studying the local syntax, an analyst models, for a given topeme, the content structure and the structure of the expressions as observed in the map face. The two structures consistently correspond in two points only: both structures pertain to the same place, and the position of a locator within the map face stands for the absolute location of the mapped place. Otherwise the scope for correspondence is limited. Furthermore, the way in which expression units are combined depends not only on the underlying content structure, but also, and sometimes in the first place, on visual requirements and on characteristics of the expression material. These concerns must be considered in formulating syntactic rules.

The supralocal syntax is governed by different principles. In conceptual terms, it is based on spatial relations (relations of relative location) but also on non-spatial ones (such as linkages between a town providing jobs and surrounding commuter villages supplying a labour force). The spatial relations ultimately reflect a spatial order of facts and are reproduced on the expression side as relations among the relevant locators.

3.3.3. Order by relations

There are relations between content units and between expression units. Both kinds are involved in organizing signs into systems (as is obvious in the local syntax). They have already been summarily covered in the preceding sections. Finally, we remind of the important phenomenon of homology, which implies that relations between expressions correspond to relations in the content universe.

4. Marginal notes

The marginal notes constitute a separate sign system, auxiliary to that of the map proper. In creating them, cartographers import material from other sign-system types, mainly natural language as relayed through writing, further numbers and, to a lesser extent, diagrams and pictures.

Marginal notes have two functions, which may be intertwined. Some of them explain what entries in the map face mean, that is, they establish the relevant code. These notes include projection statement, scale statement and legend (explanation of map symbols; Lessard 1998). Others provide background information about the mapped territory and the theme(s) under consideration, further about the map author and (sometimes) the intended audience. Among these entries are title and credit note (including statement of sources).

Among the marginal notes, the most versatile one is the legend. Apart from being a symbol dictionary, it presents items of information in some order and has several other functions (Schlichtmann 1997). As for the order, it is most often the paradigmatic one which is of interest. Thus, paradigmatic structures are visually modelled by the ways in which entries are arranged in the space of the legend panel (Wolodtschenko 1994b), for example, in a column, in a two-dimensional array, or in layers which correspond to levels of a hierarchy. This way, the legend can elucidate the relational order of the information where the symbolization in the map face cannot do so. For example, in a geological map the age sequence of sedimentary rocks cannot usually be made sufficiently clear by the underlying colour scheme, but it is easily reproduced by the vertical arrangement of entries in the legend. In other words, the legend is a separate device for conceptual modelling.

5. Peripheral signification phenomena

Peripheral signification phenomena permit to draw inferences about the contexts in which a map originated, other than the context constituted by the territory. These phenomena are encountered both within and outside the map face, but for simplicity only the first situation is considered here. We exemplify for two kinds: style and ideology.

Large-scale topographic maps of the Netherlands show rather wide road symbols, while sheets of the equivalent one-inch series of the British Ordnance Survey display rather narrow ones. These are style traits: they recur regularly and are learnt to be characteristic of certain cartographic works. Style traits may point back to an individual map author, a publishing house (as in the above example), or a journal in which the map under discussion is published. They also may inform about a national mapping tradition, the time period during which a map appeared, and other background facets. Usually, features of style can be convincingly identified.

Further, it is often claimed that maps may reflect beliefs about what world and society are like or should be like. So Harley (1988: 68) refers to pre-19th-century maps of Europe in which settlements of the land-holding elite are shown more prominently than peasant villages and links this observation to ideas about the social order as held by map authors and their clients. Ideology as reflected in maps has recently become a popular research topic. Although results tend to be intuitively plausible, there is still some way to go before the conveyance of ideological meaning components is well understood. For a more detailed discussion see Schlichtmann 2008.

6. Sign processes

Signs are not simply given. They are subject to processes in which humans handle signs, or sign processes for short. These can be grouped into three major classes:

(1) Sign production.

(2) Sign employment in dealing with the world and the participants in cartographic communication (map author and intended audience).

(3) Sign reception, understood as the extraction and derivation of information from signs.

These processes involve signs in both map face and margins, but for simplicity only the former ones will be considered. The classification proposed in this section is tentative, and so is the terminology.

6.1. Sign production

Under sign production (or sign creation) the following four groups of processes are subsumed:

(1) Conceptualization of the world, or creation of a structured set of content types.

(2) Selection, from the expression material, of items which are to serve as expression types.

(3) Assigning expression types to content types according to certain principles.

(4) Production of content and expression tokens in actual mapping.

Schlichtmann (2006) provides a detailed overview of the relevant processes. Note, finally, that marginal notes, where they introduce items of map symbolism, function as instruments of sign production.

Many sign-production processes (but certainly not all) leave the map maker some freedom of manipulating contents and expressions. Such manipulations take place at the level of the minimal sign, and so the said processes are of the greatest practical interest at this level. But freedom in sign production is limited and relative. It is limited by the following requirements: the map maker shall see to it that the intended information is conveyed, and the map user shall be able to retrieve it without avoidable error and excessive effort.

6.2. Sign employment

Once signs (more exactly: sign tokens) have been produced, something is done with them. One uses or employs them in dealing with the world and with other people. Here we find a heterogeneous set of processes. They are tentatively assigned to four classes:

(1) Referring, i.e., applying signs to referents. For example, a sign with the content ‘bridge’ is used to refer to a specific bridge.

(2) Sign employment relating to the participants in cartographic communication. This includes performing basic speech acts (like making statements and issuing commands), but also rather special acts like signalling a political opinion (as in the use of place-names in areas which are subject to political dispute).

(3) Sign employment in representing a world. Most frequently, a map author aims at conveying factually correct information about the real world. (There may be errors, of course, some of which are inevitable, like distortions engendered by a projection). Other aims are conveying incorrect information about the real world (as in propaganda cartography), modelling a future state (as in planning), and portraying a fictitious territory (e.g., in a novel).

(4) Sign employment in addressing practical problems. Here we are concerned with activities based on maps: navigation (finding one’s way), data collection in research, civil or military planning, legal determination of property boundaries, and others.

6.3. Sign reception

Sign reception is the extraction and derivation of information from signs. Cartographers generally speak of map use. To impose order on the relevant phenomena, two ways appear open, which complement each other.

One way is to follow earlier authors (among others, Ormeling and Anson 1996: 75-79), in distinguishing many elementary map-use tasks and grouping them under three heads: map reading (determining location, identifying, counting, etc.), map analysis (measuring, describing patterns, comparing, etc.), and map interpretation (placing observations onto the background of pre-existing knowledge).

The other approach starts from the fact that a map user draws on several origins or sources of information: entries in the map face, their co-texts, marginal notes, implications, and background knowledge, the latter pertaining not only to the subject matter of a map but also to procedures of data handling and mapping. The objective is to reconstruct how the map user assembles and integrates information items of different origins, reconciles incompatible meanings, and establishes limits to his interpretation. Far from being straightforward, this activity may entail more or less complex reasoning. Such reasoning processes can be elucidated by a variant of compositional analysis.

Let us consider a seemingly trivial example. ‘Airport’ implies ‘area’, ‘proper object’, and ‘typically distinctly delimited’. Now assume that an airport is indicated by a point symbol because of the reduction of the map. This symbol conveys the content ‘point’, which conflicts with the implication ‘area’; so we must remove the conflict by reasoning. We usually do so without being aware of it; nevertheless, we once had to learn it. For more interesting and more complicated cases see Schlichtmann 1991: 274-278.

7. Contexts of sign systems and sign processes

Signs of map symbolism are realized, and sign processes take place, on a background constituted by a set of contextual factors, the exact number of which is undetermined. Ultimately they all relate to the territory and the participants in cartographic communication, that is, map author and audience, as defined initially (s. 1.2). For the present purpose, seven such factors are identified:

(1) The mapped territory (see below).

(2) The map author’s identity. It may be reflected in style characteristics (s. 5).

(3) The map author’s ideas and resulting actions of sign production. Here, too, is the place of ideology (s. 5).

(4) Tools and skills employed in transcribing information. For example, a practiced draftsman can be expected to commit fewer errors than an inexperienced one. Further, river symbols drawn by hand or by a vector-based computer program may look different.

(5) Identity of the intended audience. For example, maps made for young pupils tend to be more simplified than maps produced for university students.

(6) Supposed expectations of the intended audience. They may influence how a politically disputed tract of land is shown and named.

(7) Intended map use(s) (see below).

Some of these factors regularly influence the character of the sign system manifested in a map, others do so only occasionally. We elaborate on two factors which are particularly important: territory and map use.

As for the territory, assume the most common situation, where a segment of the contemporary real world is represented with a view to factual correctness. Characteristics of the territory influence its representation, even though they are accessible through concepts and can be conceptualized in various ways. This is only to be expected. Our conceptions, as embodied in signs, are to a sufficient degree adequate to reality, otherwise they would be useless for orientation in the world. The degree of adequacy is particularly high where map signs are pictures, because here sign traits ultimately stem from the mapped objects. This is why, in spite of possible errors, such a map can safely be treated as a terrain surrogate (as in navigation, military operations, and exercises in interpreting topographic maps).

An important consideration, further, is what people are to do with maps. Good maps are designed so that specific map-use operations and specific practical applications are facilitated (for an overview of the former see Schlichtmann 2003). This includes the consideration of the circumstances, or ambient conditions, of map use, for example, conditions in the cockpit of an aircraft. Consequently, the character of the sign system manifested in a map is normally influenced by the ends which it is intended to serve.