|Table of Contents|
|Table of Contents|
We examine the efficiency of writing modular articles, by comparing the size of the modular versions of A05 and A08 to the original versions.
Table 5.2 gives an indication of the absolute and the relative size of the main sections in the original version and of the main modules in the modular version.5.11
|Table 5.2: An indication of the sizes of the sections in the original versions and the modules in the modular versions of A05 and A08. a) For the main sections in the original versions, as well as for the complete collections of the sections (i.e. the entire articles excluding the meta-information), the absolute size is given in terms of the number of words. The relative size of the sections is given as the percentage of the total number of words in the complete set. In b) the absolute size and the relative size of the main modules, including the module summaries, are given, as well as the sizes of the complete set of all (complex) modules and of the subset of all elementary modules are given as well. In c) the relative size of the modules, compared with the corresponding sections, is given as the percentage of the number of words in the sections. In A08, the first 965 words of section 4.Calculations are considered part of the Methods, and the second 701 as part of the Results. In A05, the appendix corresponds to part of the Interpretation module, as does the first part of the section Discussion. The last 228 words in that section correspond to the module Outcome.|
The complete modular versions of the articles turn out to be longer than the original versions: the modular version of A05 is 113% of the original version and the modular version of A08 is 133%. If we restrict ourselves to the elementary modules, the modular versions are about as long as the original articles: the text in the modular version of A05 is about 100% and A08 108% of that in the original version. With respect to the size of the article, the definition of the modular structure has two opposing consequences: the modularised version can be smaller as well as longer than the original one.
On the one hand, information with a wider range can be represented in mesoscopic or macroscopic modules designed for multiple use and made available by means of links to the reader of a particular modular article. Consequently, the Positioning module in A05 is far smaller than the original section 1. Introduction, which includes a description of the model that in the modular version is given in a mesoscopic Theoretical methods module MESO-m3c-mod. This makes the efficiency of the process of writing the modular article higher than that of writing linear article.
On the other hand, the modularised article can be longer than the original version. In some modules, we had to provide additional information in order to make all required details available and to make the modules self-contained. These modules tend to be longer than the corresponding sections. Both in A05 and in A08, the modules Interpretation and Outcome are far longer than their counterparts in the original version. However, in cases where the same information had to be presented in more than one module for the sake of self-containedness, the workload was increased only when the different modules could not contain literal copies. In the two constituent modules of A05-m4bi, for example, a large part of the text about the experimental results could be recycled.
We have added ten module summaries in the complex modules in A05, and twelve in A08, in order to express the coherence of the information that is distributed over the different constituent modules. The module summaries could not be copied directly from the original version; the linear article generally contains only summaries of the complete text, in its abstract and in the Conclusions section. However, not all module summaries were difficult to write; for example, that of the complex modules Outcome are basically tables of contents of the constituent modules. In addition, writing these module summaries can form an intermediate step in the composition of an adequate abstract of the entire article, which the author should provide at any rate (see [Van der Tol, 1999]).
Creating a modular article, authors not only have to write the text and provide non-textual representations of scientific information in larger amounts than in a linear article. They also must make the structure explicit, more than in the case of a linear article: they must compose cohering elementary modules into different levels of complex modules, to make different types of relations explicit in links between modules and parts of modules, and to visualise the structure of the article in a map of contents.
In the modularisation process, all references in the original versions of the articles were recast in links in the modular versions. The linking in a modular structure is far more elaborate than the standard referencing in the original versions. Whereas the original version of A08 contains 34 references to 26 works, the modular version contains 386 explicit links between (parts of) 29 source modules and 58 targets, representing a total of 940 relations, characterised by 43 different types of labels. The main reason why there are so many links in the modular article is that the internal structure is made explicit systematically: about four fifths of the links express the internal coherence of the modularised articles. In addition, in the modularisation process we have isolated information in mesoscopic and macroscopic modules, which then had to be linked to the modules of the article.
We have found that the task of expressing by hand the structure in links, in the composition of modules and in navigation aids (such as the Map of contents) would increase the author's workload to an unacceptable degree. However, in practice a large part of this additional task lends itself to automated procedures and tools. We shall discuss the requirements for such appropriate authoring tools in section 6.2.2.
Summarising, we conclude that modular articles are feasible. Although we have encountered some difficulties in delineating the boundaries of particular modules and in identifying the relevant scientific discourse relations, writing a modular article is sufficiently straightforward on the whole. Writing a modular article will probably require a greater effort than writing a linear article. However, if the modular article better fulfils the reader's requirements, it will be worthwhile, as the author aims at communicating his work to the reader. In the next section, we therefore discuss the modularised articles from the reader's point of view.