Treating the Multimodal Page as a Multilayered Semiotic
3.3 A more complicated example of layout analysis
3.3.2 The layout of the Louvre
The next stage of analysis is to identify the layout units and to organise these within an appropriate layout structure. Here we look for primarily vi-sual groupings that can be motivated perceptually. In the present case, this combines proximity and separation, similarity (e.g., the repeated motif of photograph, bold header and text paragraph), connection (expressed explic-itly by connecting lines), framing (with boxes and colour), relative size and closure.
Visual and spatial prominence let us start naturally with the 3D drawing in perspective of the Louvre. This ‘brings with it’ several closely associated elements: including the text and title upper-left, because of its wrap around behaviour with respect to the left edge of the drawing, and the smaller call-outs surrounding the drawing, because of their connecting lines and prox-imity. No collection of these latter elements can themselves be organised
5The page number and running heads of the two pages are the last units listed (apart from an additional connecting line and the text paragraph associated with unit U046 that were left out on the first run through the numbering process!). There are also some gaps in the numbering—this is because of a contrastive analysis of the layout that we carried out with respect to this page and an earlier version of the same page from the 1993 edition of the guide. The units that occur in both versions have the same numbers. We do not discuss this contrastive analysis here, however.
Figure 3.8 Base units of the Louvre page shown for ease of reference using the page as background. Some of the labels for the connecting lines have been omitted here to improve the readability of the diagram.
into larger configurations because of their mutual lack of proximity (apart from to the drawing) and whitespace framing. The first larger, composite layout element to be formed is therefore that consisting just of the drawing, the overview text and the callouts.
We can distinguish some further detail among the callouts, however, by applying considerations of similarity and difference. On this page there are in fact three classes of callouts. The first class is made up of single bold type text fragments, naming parts of the building: e.g., ‘Pavillon des Sessions’ (U030), ‘Denon Wing’ (U044), ‘Cour Puget’ (U076)—all shown ringed in Figure 3.8 for ease of reference. The second class consists of a combination of a text fragment in bold type within a text paragraph, as in U006 and U007 (on the left, halfway down the page):
The Jardin du Carrousel, now part of the Jardin des Tuileries, was once the grand approach to the Tuileries Palace, which was burned down in 1871 by the Communards.
This is an example where we require an embedded base unit for the bold segment as described above. Other units showing this structure
layout unit base units children layout unit base units children
L1.1.1 composite L1.1.1.1– L1.1.1.3a U037 –
L1.1.1.14 L1.1.1.3b U039 –
L1.1.1.1 U005 –
L1.1.1.4 U044 –
L1.1.1.2 composite L1.1.1.2a
L1.1.1.2b L1.1.1.5 composite L1.1.1.5a
L1.1.1.2a U006 – L1.1.1.5b
L1.1.1.2b U007 – L1.1.1.5a U046 –
L1.1.1.5b U113 –
L1.1.1.3 composite L1.1.1.3a L1.1.1.3b
and so on
Table 3.4 Extract of the correspondence between layout units and base units
are U037+U039, U046+U113, and U099+U100 (moving left to right below the central drawing). These first two classes of callout are not distinguished spatially, however, and so there is little motivation for giving them a distinct status in the layout structure.
This gives a grouping of base units into layout units as begun in Table 3.4.
And, with these layout units defined, we can add in further information con-cerning their realisation. We have already mentioned the 3D drawing of the Louvre that constitutes L1.1.1.1. We also have the bold type small text of the ‘a’ units in the table and the normal type small text of the ‘b’ units that make up all the small callouts. This is how we capture the generalisation that there are two subclasses of callouts being considered here: L1.1.1.2, L1.1.1.3, L1.1.1.5 and so on that contain a header, and L1.1.1.4 and so on that do not. These distinctions are also supported by considering other pages of the book where the first class is sometimes present without the other. The bold elements in these callouts give more weight to the informa-tion being given and also strengthen their role as access elements.
The third class is more complex and requires composite layout elements combining photographs/graphics, a bold type header and a descriptive para-graph. In contrast to the previous class, the header here is not grammatically integrated into the paragraph. An example is the Pavillon Richelieu callout consisting of U062, U064, U065 connected to the Louvre drawing with the connecting line U063 (righthand page, upper left). We capture the internal structure of these more complex elements with specifications of the form:
layout unit base units children
L1.1.3 composite L1.1.3a L1.1.3b L1.1.3c
L1.1.3a U062 –
L1.1.3b U064 –
L1.1.3c U065 –
There turn out to be three variants of this class also: the form shown in this table, which also applies to the layout elements corresponding to the group {U086, U087, U088} (righthand page, lower left); the form identical to this apart from an additional base unit corresponding to a star icon, as shown in the groups linked to the photographs or drawings: U071, U094, U102, U040; and the isolated case U035 and U036 (lefthand page, upper right).
This latter case is interesting in that it represents one possible weak spot in the layout: it is actually just an unattached picture with caption. Its structure is different to all the other picture+text combinations, it has no connecting line indicating that it is a callout and, when we come to look at its contents, we find that it actually depicts a part of the Louvre that is on the mid-right-hand side of the page rather than anywhere in the vicinity of its own units. There are enough signals present to identify its distinct status but it is questionable whether this would be visible at first glance to a reader. This means that we could find misreadings at this point—leading, for example, to a potential visitor approaching the Louvre from the north (which is following convention in this case also towards the top of the page) and being surprised not to see the part of the building depicted in U035: ‘the East facade’.
The key to its interpretation is only found in the context of the book as a whole. In the introduction there is a description of how the diagrams and complex spreads presented in the book are to be read. And there we find the explanation (itself presented as a callout with reference to a miniature example of a typical page):
The facade of each major sight is shown to help you spot it quickly. [p7]
This works well on many pages of the book precisely because in their case another layout structure is more likely. We can see this clearly from Fig-ure 3.9, where a collection of these alternatives taken from other places in the book are shown.
Most of these pages construct a layout structure in which it is most likely that the photograph of the facade is one of several components combining with the overview text and not with satellitic callouts. The three examples to the left of the figure all have the facade photograph to the right of the overview text: this is the situation illustrated in the book’s own explanation given above. The two examples in the central column show a further variant
Figure 3.9 Example layout positioning of the ‘facade’ information from several pages of the Dorling-Kindersley Eyewitness Travel: Paris guide. Text elements are shown by horizontal lined areas, graphics and photographs by diago-nally lined rectangles; the facade photograph of each page is shown as a dark rectangle and other photographs and graphic material as light rectan-gles.
(not explained in the book’s introduction), where the facade photograph is inset within the overview text. In these cases both photograph and text fall under the title for the text/page and so there is no problem with creating a single layout unit including both the text and the facade photograph.
In our Louvre page, however, shown to the right of the figure, we see the problematic configuration caused by the facade photograph being ‘pushed away’ from the overview text by the intervening central drawing of the Louvre as a whole. This almost explicitly sets up a framing in which the facade photograph is grouped with the other photographic callouts rather than with the overview text. Given that such new groupings can always arise by chance when elements are moved around a page design, it is virtu-ally impossible to guarantee that such unwanted side-effects are ruled out.
We will see several more examples of this in a range of different document types in our analyses in later chapters.6
Even though we suggest here that the layout of the current example en-courages a misinterpretation, strictly speaking we can only really provide
6There is also the question of whether it is even worthwhile ruling them out—for example, if the cost of checking for them is high and their consequences for readers low. Here we are simply drawing attention to methods for revealing them; the separate question of what, if anything, should be done about them, we will leave for later.
Figure 3.10 Layout structure for the Louvre page according to the GeM model
this kind of analysis when we have first provided our explicit account of the intended communicative functions of the layout elements. That is, in this case, we need to characterise the function of the layout element express-ing the facade in order to recognise that it needs a different treatment to the other layout elements giving photographic information about the Lou-vre. Otherwise there are no grounds for rejecting its inclusion within the same layout element. This level of information will be provided by the GeM rhetorical structure layer and so we will return to this and several more extensive examples of potential misinterpretations in our discussion in Chapter 4. For now we simply note that the explicit characterisation of the layout structure we have given makes it difficult to overlook such problematic cases—which is one strong motivation for providing it.
The next layer of layout structure for the Louvre spread as a whole is then to add in the loosely attached ’Visitors’ checklist’, ‘Star features’ and
‘Building the Louvre’ insets, the navigation marks upper right and upper left, and the page number and section information. The complete hierar-chical layout structure for the spread that results is shown graphically in Figure 3.10. Again we omit here some of the finer detail discussed above in order to reduce the complexity of the diagram. We also show in the upper left of the figure a view of visual prominence on the page as suggested by filtering that supports the layout provided.
The full specification of these layout units includes information about how they are realised, or physically expressed, on the page. We have al-ready mentioned some of these features above: the titles associated with
texts and photographs are in bold, for example. This information also, as suggested, allows us to draw out similarities and generalisations across lay-out units that are not expressed in terms of the hierarchical laylay-out structure.
Such similarities often capture the fact that distinct layout units are serv-ing a similar rhetorical function, even if they are not clearly members of some overarching layout element. Thus, for example, we in fact can iden-tify two distinct classes among the sibling layout elements accompanying the drawing of the Louvre, and three distinct classes among the siblings accompanying the layout element above this. Since these classes are not expressed spatially, it would not be appropriate to add this generalisation into the layout structure itself.
Now we have to specify the final aspect of the layout organisation by defining the area model for the spread that captures just where the layout units are positioned on the page. In this case, we will proceed in two steps as the layout is relatively complex in this respect. To begin, the layout is essentially centered without polarisation. That is, the elements of the page are spread around the central element given by the drawing of the Louvre.
There is no top-bottom or left-right organisation beyond the minimal struc-turing imposed by the Western default reading path. Picking up again com-ments from the previous chapter (cf. Section 2.2.2), we do not see here any particular additional utility in providing this with a given-new or ideal-real interpretation at this point—indeed, it appears to be access and navigation information that provides the strongest functional motivations in this genre.
It is not the case, however, that the page exhibits no symmetry since the virtual canvas of the spread conforms to a grid in the manner described in Section 2.5.1 of the previous chapter.
This grid is suggested in Figure 3.11; it offers three columns for each page of the spread, organised symmetrically across the central binding and with the outer column slightly narrower than the
Figure 3.11
A grid for the Louvre page
inner two. The fact that this is a true grid, rather than an opportune alignment of the ‘ex-tra information’ insets on this page, can be ascer-tained by examining the other pages of the book—
particularly those spreads which are mostly tex-tual. In these cases there is an explicit 6 col-umn spread, often with framing lines separating the columns. From the book as a whole we can also determine a two column grid superimposed on the three column grid: this is the orienting line shown in the middle of each page in the figure.
The two-column variant is used for those pages
which present an introductory text for an entire section: the paragraphs of these introductions are then set in two columns rather than in three columns, clearly showing their different communicative function. This grid organi-sation is a common feature of this entire series of travel guides. For the current spread, then, we have a straightforward column grid in operation, despite the superficial impression of freedom given by the additional place-ment of eleplace-ments around the Louvre drawing.
Figure 3.12 The complex area model for the Louvre page, combining a geometric shape and a vertical grid
Combining these two contributions give us the complex area model shown graphically in Figure 3.12. This composes the vertical grid with a centralised geometric figure with two ‘rings’ running around it: an ‘inner’
ring and an ‘outer’. This illustrates one of the more advanced grid-like organisations that we suggested in the previous chapter: rather than a rectangular grid defining horizontal and vertical alignment possibilities, we have here the geometric shape approximating to the shape of the Louvre outline defining two concentric bands of alignment possibilities.
These bands provide approximate placement opportunities for selected layout elements which can then be pushed left or right according to the virtual canvas constraints imposed by the vertical grid. Allowing several of these elements, particularly those combining photographs and text, to be slightly wider than the vertical grid’s modules adds even further to the impression of a ‘free-ranging’ layout supportive of the centralised organisation while still ensuring that elements are spread relatively uniformly over the page area.
Figure 3.13 Correspondence between the layout structure and the area model for the Louvre page
The last step is then to bring the layout structure and the area model into correspondence. This is shown graphically in Figure 3.13 in the same manner that we did above for the Gannet page. This captures sufficient spatial detail to keep apart the distinct functional contributions made by the various elements, which in this spread are considerable. In particular, we are able to distinguish elements that share functionality by virtue of their spatial proximity and grouping (layout structure) and those elements that can be identified by virtue of their specific placement within the area model (i.e., central, inner or outer ring).
3.4 Conclusion
In this chapter we have defined and illustrated the ‘lowest’ layers of the GeM model. Specifications here decompose any page-based multimodal document into parts and explicitly state how these parts are organised visu-ally to create a layout structure for the pages of the document. All of the layers of analysis of the GeM model that follow, assume that the base and layout layers are in place as described here.