November 7, 2018 | Author: Patrick Clark | Category: N/A
Share Embed Donate

Short Description

1 THE CONVERSION OF A LINEAR TEXT TO HYPERTEXT Kerstin Frenckner Kjell Krona Department of Numerical Analysis and Comput...


THE CONVERSION OF A LINEAR TEXT TO HYPERTEXT Kerstin Frenckner Kjell Krona Department of Numerical Analysis and Computing Science Royal Institute of Technology, Stockholm, Sweden

Abstract This report describes the conversion of a rather extensive linear report on legibility of continuous text on computer screens to hypertext. The objectives of the conversion, the resulting hypertext and some difficulties we encountered are discussed. The hypertext is presented on a large screen using several windows and includes a facility for choosing among a number of paths to follow. Each path focuses on some aspect of the material and contains the material dealing with that aspect in the original order of the printed report.

Contents Background.......................................................................... 2 The text................................................................................ 2 Objectives of the hypertext and how we achieved them ....... 3 Paths ............................................................................. 3 Overview........................................................................ 3 Ease of use .................................................................... 4 Use of windows.............................................................. 4 Typographic design ...................................................... 5 The resulting hypertext ....................................................... 5 Some problems on the way ................................................ 14 Coherence of the text.................................................. 14 Updating the text......................................................... 14 Screen size .................................................................. 15 Problems with the implementation.............................. 15 User study.......................................................................... 16 Possible improvements...................................................... 17 Applicability to other texts................................................. 18 Final words......................................................................... 18 References ......................................................................... 19 Kerstin Frenckner, Nada, KTH, 100 44 Stockholm, telephone +46 8 790 8164, electronic mail [email protected] Kjell Krona, Architecture, KTH, 100 44 Stockholm, telephone +46 8 790 8590, electronic mail [email protected],


Frenckner, Krona: Conversion of a linear text to hypertext

Background At IPLab (Interaction and Presentation Laboratory) there is an interest in studying the similarities and differences in text presentation on different media: paper, computer screen and hypertext. One part of this work is the report titled: “Legibility of Continuous Text on Computer Screens — A Guide to the Literature” written by Kerstin Frenckner.

The text The first half of the above mentioned report of about 50000 words is an introduction serving as a general background to research in this interdisciplinary area covering topics as legibility, the reading process, typography, differences between screen and paper and how to measure legibility. These chapters will be of different interest to the reader depending on his previous knowledge and experience. The second half consists mostly of reviews of different articles but it also contains discussions of the results reported in these articles. The articles are sorted primarily according to the reading task performed (reading for understanding, proof-reading etc.) and secondarily according to the kind of computer screen used (alphanumeric or graphic). The report ends with a final discussion on how to increase the legibility of continuous text on a computer screen, a bibliography with some 130 entries, and an index. There are many possible ways of organizing the material of such a report. The material contains an introduction to a diverse field of research with close ties not only to computer science and psychology but also to typography, design, teaching, sociology and medicine. The material also contains a compilation of a vast number of separate experimental reports. So there are essentially two kinds of material: introductory material and reviews of reports. During the work the author tried several ways of organizing the material and finally ended up with what she felt worked the best. When the report was finished it was decided to convert the text into hypertext, since there seemed to be many different ways of reading it.

Frenckner, Krona: Conversion of a linear text to hypertext

Objectives of the hypertext and how we achieved them Paths Many readers are probably just interested in one (or a few) aspect(s) of the experiments, e.g. proof-reading on graphic screens, or the influence of page size, or experiments where eye-movements have been studied. We therefore decided to make different paths through the text. Each such path contains those and only those paragraphs of the original text relevant to the subject of the path. The paths were made by distinguishing 65 different subject areas covered in the report and then manually tagging each paragraph in the report with those subject areas that were deemed relevant to the paragraph. The text was then divided into sections so that the paragraphs in each section were consecutive and tagged identically. Each review (with one exception) was considered a separate section and no section contained more than about one and one half A4-page (8.5"*11") of text. The sections were also distinguished as to whether they contained a specific review or more general background material. Each section was also labeled with one or several headlines according to its place in the original, hierarchic structure of the written report. The sections tagged with a certain subject area correspond to a path, with the sections in the same order as in the printed report. A user can also construct new paths consisting of the union or the intersection of existing subject areas. The subject areas were sorted into five categories, making it easier to find an appropriate path. There is also a path containing all 213 sections; this corresponds closely with the written report in its entirety. There is a table of contents where the reader can choose his path. When a path has been chosen, the table of contents shows only those sections that are in the selected path. We used HyperCard 2.0 on a Macintosh to construct the hypertext. Each section in the text is placed on one HyperCard “card”. Overview One of the main problems of reading from computer screens is the lack of overview. By overview we mean the reader's possibility to grasp the text as a whole, to get the gist of the text. We think overview is promoted by seeing a fairly large portion of the text, by being able to recognize the surroundings of the presented text, and by being able to



Frenckner, Krona: Conversion of a linear text to hypertext

relate different parts of the text to each other (SeverinsonEklundh, 1991). So we wanted to give the reader some helpful devices for achieving an overview of the text. This means providing assistance for knowing what is in the text, for swiftly moving around in the text, and for knowing the current position in the text. We tried to achieve a good overview by using three separate windows to present the material. The reviews are shown in one window, the general background material in another, and a third window shows the different paths and a table of contents. It is possible to click on a specific section entry in the table of contents. The card containing that section is then shown in the corresponding window. At the top of each card is shown: all levels of headlines that are above this card, the selected path, the number of the card within this path and the total number of cards in this path. There is also a “floating” navigation palette with buttons to show the card before or after the current card in the selected path and to go back to the table of contents. The overview is also promoted by using a large screen. All three windows can then be shown simultaneously and a fairly large amount of text may be visible in each window. Ease of use One of our goals with the hypertext was to make it easy to use even though the material as such is rather large and complex. We hoped that the advantages of the table of contents and the paths would outweigh the navigational difficulties that would be introduced and thus make the text at least as easy to grasp as in the traditional report form. A stack with usage info is provided. It describes the windows, the paths, and how navigation is done. We advise that the usage info is studied before the rest of the material. It can also be accessed from the other cards by clicking a question mark button. The table of contents is always at hand and we feel that the table of contents and the navigation palette provide simple and powerful ways for navigation. If we have succeeded in making the hypertext easy to use is still an unanswered question. We have plans for a user study, but we are also happy to get any comments from other users. Use of windows If a tool of this kind is to be appreciated it has to be not only functional and easy to use, but also has to have a pleasant design. Since the introductory parts and the reviews are quite different in character, style, and content, we decided to present these in windows with differing

Frenckner, Krona: Conversion of a linear text to hypertext appearance. The three main windows were given fixed placements: the table of contents is placed in the lower part of the A3 (21") screen, the concepts and methods and the reviews share the upper part of the screen. These three windows have different placement, contents, and design. The concepts and methods contains most of the general background and some of the discussions. The reviews contains the reviews of the reports and a complete reference. Since our division of the text into cards is based on the contents, there is in some cases more text than fits on the card. Scroll bars are used in such cases. Typographic design In any page design, whether on paper or on a computer screen, there are conflicting demands: the demands for good typography, logical placement, and separation between the different pieces of information conflict with the wish to present as much information as possible on a page. In a text presented on a computer screen, the act of “turning the page” is usually felt to be more troublesome and more conceptually demanding than in a text presented on paper, frequent page turning may easily lead to a loss of overview. A computer screen will thus usually have a higher percentage of its area used for presenting text. The limited resolution of a computer screen calling for fairly large font sizes and the more or less restricted screen area also contribute to this tendency. A more informative design may show less information at a time but put it more clearly in its context. We have tried to achieve as good a compromise as possible in this difficult problem.

The resulting hypertext There are seven HyperCard 2.0 stacks. The Start stack is first opened in the upper left hand corner of the screen. It gives a short presentation and the option of opening four other stacks. See figure 1. The Table of Contents is also opened in the beginning.



Frenckner, Krona: Conversion of a linear text to hypertext


Figure 1. When the hypertext is first opened this window is shown. It is advisable to start out by reading the usage info in the About this text stack. See figure 2. In this stack, a short general introduction to material and the hypertext presentation is given. The paths, the windows and the methods of navigation are described. The preface and the abstract of the original report are also included. Clicking the Subjects button in the upper right hand corner of the window will get the user to the first card of the About this text stack where there is a list of subjects in the usage info. Any subject may be clicked and this will get the user to the card covering the selected subject. Clicking the left and right arrow on the lower left hand side of the window gets the user to the previous and next card respectively in the About this text stack, whereas clicking the return button gets the user to the card where he was before visiting the About this text stack.

Frenckner, Krona: Conversion of a linear text to hypertext


Figure 2. The About this text window gives usage info.



Frenckner, Krona: Conversion of a linear text to hypertext

Table of contents

Figure 3. The screen may look like this after a while.

Frenckner, Krona: Conversion of a linear text to hypertext Figure 3 shows a picture of the screen when work has progressed for some time. The Table of contents is the active window which, as in all Macintosh applications, is shown by the horizontal lines in the top bar of the window. To the left in this window we see the main categories and the subcategories that can be used to form a path. In this case All pages and All categories is chosen. Above the categories is shown the number of cards in this path (213). To the right, the cards in this path are presented. For each card there is a title. If the content is a review, the authors are shown in a second column and the year in a third column. The division lines between these columns can be moved by dragging the handles. The user may click on a card (title, author, or year) in the table of contents and this will result in a move to the selected card. The question mark button may also be clicked which results in a switch to the appropriate card in the About this text stack. Some of the lines in the table of contents are just headlines. These are shown in italics (Legibility research on paper in figure 3 is such a headline) and correspond to those headlines in the printed version that have another headline directly following. Clicking such a headline will result in a move to the first card under that headline. In the upper left corner of figure 3 is a card in the Concepts and methods stack. In figure 4 the Concepts and methods is the active window. The navigation palette at its left side is used to move to the next or to the previous card in the selected path or to the table of contents. In the top right hand corner of the window you can see the selected path. Some words in the text are italicized. These are references. Clicking a reference results in a transfer to the appropriate card. Returning is done by clicking a return button.



Frenckner, Krona: Conversion of a linear text to hypertext

Concepts and methods

Figure 4. The concepts and methods window is active.

Frenckner, Krona: Conversion of a linear text to hypertext Figure 5 shows the Table of contents when a path has been formed (the same path was chosen in figure 4). When forming a path, a main category must first be chosen by clicking it. Then the corresponding subcategories are shown and a subcategory is chosen. It is also possible to choose the union or intersection of two or more subcategories. In this case the intersection of proofreading and reading for comprehension was chosen. This means that only those cards dealing with both proofreading and reading for comprehension will be in the path and thus shown in the table of contents. You can see that there are only seven such cards.

table of contents med skärning vald

Figure 5. The table of contents when a path other than “all pages” is chosen. There are also some illustrations in the text. These are indicated by the text “Illustration #” and are shown in a separate window by clicking this text. There are two tables in the text. They are called upon in the same way. See figure 6 and figure 7.



Frenckner, Krona: Conversion of a linear text to hypertext

bild med en bild

Figure 6. The left part of the screen when an illustration is shown.

Frenckner, Krona: Conversion of a linear text to hypertext


Figure 7. One of the two big tables is shown. To the right is a review and the table of contents.



Frenckner, Krona: Conversion of a linear text to hypertext

Some problems on the way Coherence of the text In a linear text presented on paper the coherence is (or should be) provided by the discourse chosen by the author. We have decided to keep the original text almost unchanged and to make the discourse provided for the original, printed text quite evident to the reader of the hypertext version. He will see the original discourse by choosing the path All pages, which is the default path when opening the hypertext version. By reading these cards (sections) in sequence, he will actually follow the original discourse. If the reader chooses some other way of reading the material, he will be aware that he has done so and will not be surprised by a lack of coherence. Whenever a linear and coherent text is to be divided into different parts (cards) that can be read in any order, either each part has to be self-contained or there will be an obvious risk of lack of coherence. We considered it unrealistic to make each part self-contained, since it would imply duplicating much of the text and placing it in several parts. Although this would probably have helped those readers interested in reading one or a few parts (cards), it would certainly have annoyed those readers interested in several parts (cards) finding the same text in several places. We do, however, believe that we have overcome some the coherency problems. The material is organized in several paths and in each path the material is presented in the order that it is found in the original, printed report. For any reader who reads a path sequentially the number of inferences to material that has not been read should thus be strongly reduced. In the table of contents for each path all headlines that are relevant to the cards in the path are presented. It should thus be easy for the reader to see the context of each card. On each card all headlines that pertain to the card are presented. Updating the text The text covers a topic that is steadily developing and changing. Thus we wanted to provide for easily updating the hypertext if the original text is changed in the future. The hypertext version is therefore generated by a HyperTalk script (HyperTalk is the programming language of HyperCard) that reads the tagged, linear text as a text file and converts it into HyperCard stacks. This generation takes up to one hour of time, but since it will be done so rarely the generation speed will be no problem.

Frenckner, Krona: Conversion of a linear text to hypertext Screen size Originally we wanted to make this hypertext available to everyone with a Macintosh. The largest window was therefore made to fit on a 9" screen. On a small screen the windows were put on top of each other, and on a larger screen they were shown beside each other. But some of us felt that on a small screen, especially the 9'' screen, the text was too crumbled and it was very difficult to grasp the general idea of working with different windows. We didn't want to make the windows too small since it would have hampered the reading, especially for the readers with large screens. But with windows of the same, or almost the same, size as the screen it was difficult to get an overview of the different parts of the text and relate the part of the text that was just being read to the rest of the material. Large screens are also becoming more common. We therefore decided to redesign the text for an A3 (21") screen. A version for small screens is also at hand. Problems with the implementation We decided to use a hypertext environment as a development tool, partly to reduce the development effort to a reasonable level for an experiment, partly because we wanted to evaluate the possibilities of such an environment for a specific problem. From the few hypertext systems available that are both well spread and reasonably robust, we chose to use Apple HyperCard™ (now owned by Claris Corp) since it was (until august -92) distributed free with all new Macintoshes, was well-known to the implementors, and with its flexible scripting language was considered more powerful than other alternatives. There are also several “clones” for HyperCard both on the Macintosh and on the PC (MS-DOS) platform, thereby making a wide distribution of the hypertext possible. Some of the decisions we made were, however, not well suited to the HyperCard environment. In particular, our decision to use several windows, and implement these as separate HyperCard stacks (disk files) resulted in slow response times when changing which window was to be active. This is done frequently, both when using the table of contents to open specific cards, and when following a particular path. The delays may detract from the experience of “browsing” through a book and perhaps confuse the reader by focusing interest on the implementation instead of on the content.



Frenckner, Krona: Conversion of a linear text to hypertext

Another problem is the search mechanism for constructing new paths. In the current implementation this is done with a linear search of the tags on every card, which takes a minute or two on a mid-class Macintosh. One of the reasons behind this is that we wanted every path to have an appropriate table of contents, and much of the time is spent constructing this. In retrospect, a much faster mechanism would probably have been possible to construct. But at the time we felt that in our experimental context the functionality in itself was more important than the actual speed of the system. Of course, the actual speed does make a difference as to the usability of a system.

User study We have plenty of comments from several users in our environment. A simple user study has been performed and a more extensive is planned. Six students attending a one semester course on computer supported text processing on a college level took part. It was during the last week of the course. They had been working on Macintosh computers during the course and had had two lectures on hypertext and one lab using HyperCard. The students were given a short introduction to the text and the structure of the hypertext in a lecture room. Then they were asked to use the hypertext for a while to familiarize with it. After that they were given eleven questions on the text and were asked to answer them and to time themselves. Kerstin was at hand in the room ready to help them. The students were also asked to give comments. Only two of the students got a reasonable feeling for how to use the index and how to navigate in the text. They were the only ones who were able to answer more than a couple of the questions. Answering times varied widely both between these two students (factors up to 1:15) and between the questions (factors up to 1:20). All the students commented that they would rather have read the report in a traditional, paper form. The intention was to give the same students a set of similar questions to answer using the text in a word processor, but due to the poor results with the hypertext this was never done. The poor results with these students may of course be due to the structure of the hypertext being too difficult or inconsistent. The questions may have been too difficult. But it may also be that the students were unfamiliar with the subject, unaccustomed to reading research reports, unfamiliar to reading english, and lacking computer maturity.

Frenckner, Krona: Conversion of a linear text to hypertext To test the latter factor the same procedure was performed with a computer science student. He grasped the structure of the hypertext, the use of the windows and the navigational facilities reasonably fast. But he still had problems answering the questions. He especially missed a good search facility and an index. Conclusions from this study and from our previous experience with people using this hypertext must be that: people need a good computer maturity to be able to handle and appreciate advanced tools using several windows and navigational facilities and that it takes considerable time to grasp a material as vast and varied as this and thus be able to answer questions where the answer can not be found through the table of contents.

Possible improvements We are well aware of some of the things that people have complained about but we have simply not had time to implement improvements. In a linear text a list of references is usually provided at the end of the text. In the hypertext the full reference is instead given on the card reviewing the article in question. An additional, alphabetical list of references would probably be useful but has not been implemented. The same goes for an index. Both were at hand in the original, printed version of the text. We were uncertain about the functionality of a hypertext list of references and of an index since each reference or index word usually leads to several places in the text. What should clicking a reference or an index word then imply? It would also certainly be useful if the currently selected card was highlighted in the table of contents. It would probably facilitate the orientation if the table of contents was scrolled as the reader moves around in the text so that the active card always would be visible in the table of contents. A mechanism for folding and unfolding the table of contents would be useful. This would give the reader the choice of seeing only headlines and the number of cards subordinated to each headline or the full table of contents including all cards in the path. Some kind of history facility, maybe in combination with the use of bookmarks, might be useful. But it has to be kept in mind that the user interface must be kept fairly easy to handle. The hypertext version presently



Frenckner, Krona: Conversion of a linear text to hypertext

at hand uses several windows for presenting different kind of material and has, by the tools for forming paths, an advanced facility for customizing the material for the user's needs. Thus care must be taken not to make the user interface more complicated.

Applicability to other texts This text has some properties that are quite special but others that are quite common to lengthy non-fictious texts. The division into one part with a general background and one part with reviews is special to this text. But it is quite common that several aspects are covered in a more general text and that a certain reader is only interested in one or two of these aspects. A reader of a text on the history of Sweden written in the traditional, chronological order might for example only be interested in the history of farming. If he then can choose a path called “farming”, he can read only those parts of the text that deal with farming. We thus feel that the ideas of •

tagging a text with subject areas,

dividing the text into sections so that each section contains consecutive paragraphs that are identically tagged,

putting each section on a separate card,

providing the user with a table of contents where he can choose a path,

showing the cards in the selected path in the table of contents, and

allowing the user to get to a card by clicking it in the table of contents

are quite usable in many situations. The HyperTalk code may be used to make a hypertext from other linear texts tagged in a similar way.

Final words We are well aware that our work does by no means represent the final solution to the problem of converting linear text to hypertext. But we feel that the idea of providing tools for forming paths containing material about specific subareas of the general topic is quite useful and should be further developed. The idea of presenting the cards in a hypertext

Frenckner, Krona: Conversion of a linear text to hypertext in a table of contents where any card can be reached by clicking it is by no means new. The usefulness of such a table of contents is probably even better when complemented by the idea of forming paths and of showing only those cards that belong to the selected path in the table of contents. The work has been supported by the project PeopleComputers-Work funded by the Swedish Work Environment Fund and The Swedish National Board for Technical Development. We thank them for their support. We also want to thank Hans Smedshammar who took part in the work of tagging the text and Kerstin Severinson-Eklundh who has inspired us.

References FRENCKNER, K. 1990. Legibility of continuous text on computer screens — a guide to the literature. IPLab-25, by NADA, KTH, 100 44 Stockholm SEVERINSON-EKLUNDH, K. 1991. Problems in achieving a global perspective in computer-based writing. Computers and writing IV, Brighton, 1991 and IPLab-38, by NADA, KTH, 100 44 Stockholm.


View more...


Copyright � 2017 SILO Inc.