Upload
phunghuong
View
212
Download
0
Embed Size (px)
Citation preview
Department of Numerical Analysis and Computing Science TRITA-NA-P0503 • IPLab-247
Informing the design of page-based writing tool through longitudinal and naturalistic case studies
Ann Fatton and Kerstin Severinson Eklundh
Interaction and Presentation Laboratory (IPLab)
Ann Fatton oand Kerstin Severinson Eklundh Informing the design of page-based writing toolthrough longitudinal and naturalistic case studies Report number: TRITA-NA-P0503, IPLab-247 Publication date: May 2005 E-mail of author: [email protected]
Reports can be ordered from:
Interaction and Presentation Laboratory (IPLab) Numerical Analysis and Computing Science (Nada) Royal Institure of Technology (KTH) S-100 44 STOCKHOLM, Sweden
telephone; + 46 8 790 6280 fax: + 46 8 10 2477 e-mail: [email protected] URL: http://www.nada.kth.se/nada/iplab/
- 1 -
ABSTRACT
This paper describes and evaluates an alternative model for computer-based writing, called the paper model. According to this model, pages are handled as independent
objects that can be freely manipulated and organized by the user. The paper model
attempts to bring to computer-based writing some of the benefits of paper handling when it comes to getting an adequate mental representation and a general overview of the text.
Three studies are reported, conducted as evaluations in the design cycle of an
electronical, page based writing tool based on the paper model. The first study looked at
the adequacy of the paper model to support in-process planning, i.e. planning that goes on
beyond an initial planning phase at the beginning of a writing task. The second and third,
ethnographically oriented case studies focused on real use situations of the tool and long-
term effects of the page-based text presentation.
The results of the studies are used to address a number of design issues in the paper
model. All studies provided valuable results to be used in future implementations of the
model. The results generally support our hypothesis that a page-based text presentation,
as a complement and alternative to a scroll-window presentation, provides writers with an
improved overview of the text, increased support for working with several documents
and a general flexibility that corresponds well to the diversity of writing activities,
writing styles and writing tasks that a computer-based writing environment should
support.
- 2 -
- 3 -
CONTENTS
1. INTRODUCTION
2. PERSPECTIVES ON WRITING AND TECHNOLOGY
2.1. Research perspectives and models of writing
2.2. The influence of computers on writing processes
2.3. Overview problems when (reading and) writing with computers
2.4. A tradition for paginated text presentation
3. USING PAPER AS A DESIGN MODEL FOR COMPUTER-BASED WRITING
3.1. The implementation of the paper model – historical perspective
3.2. Introducing a new metaphor for on-line tools – early design choices and implementation requirements
3.2.1. Writing on fixed-sized pages
3.2.2. Are the text and the page different objects?
4. THE PAPER WRITING ENVIRONMENT
4.1. The current version of Paper
4.2. Design iterations
5. A STUDY OF IN-PROCESS PLANNING IN PAPER
5.1. Goal and research questions
5.2. Study design
5.3 Results
5.4 Discussion
6. TWO LONGITUDINAL, NATURALISTIC CASE STUDIES OF WRITING IN PAPER
6.1. Goal and research questions
6.2. The subjects and their writing tasks
- 4 -
6.3. Settings
6.4. Data collection
6.4.1. Summary
6.5. Findings
6.5.1. The subjects’ general perception of the Paper program
6.5.2. Writing on Paper pages
6.5.3. A limited need for a scrolled text presentation
6.5.4. Optimizing screen space and getting an overview by creating
personal “writing environments”
6.5.5. Working with several documents – a natural strategy when writing in Paper
6.5.6. Improving strategies for handling page splits
7. DISCUSSION AND DESIGN IMPLICATIONS
7.1 Gaining overview of the text with Paper pages
7.2 Working with several documents
7.3 Visual orientation and feeling of control
7.3.1. A feeling of control
7.4 How to improve the page split functionality
7.4.1 Better support for “repairing the damages” after a page split
8. GENERAL CONCLUSIONS
- 5 -
1. INTRODUCTION
Athough the introduction of computer technology on a broad basis during the 1980’s
brought major changes in the way texts were composed, writing remains a difficult
activity. The ease by which the text can be edited and spelling errors corrected in a word
processor has not alleviated problems such as generating ideas, finding an adequate
rhetoric tone, improving the structure of a lengthy text, etc. Moreover, the new
technology introduced specific problems, such as limited screen space and a difficulty to
get a sense of the text written so far. While recent improvements in word processors have
presented the user with additional functionality (e.g. revision tracking, insertion of
pictures and advanced printing options), little has been done about the basic underlying
metaphor that guides the user’s interaction with the text.
In this article, we introduce the paper model for computer-based writing and its
implementation in a page-based writing tool, which has been developed in a number of
iterations as a result of empirical studies. In the Paper program, the text is written and
presented on separate, fix-sized pages, which can be independently manipulated by the
user. We will argue that this way of presenting the text, as an alternative and complement
to a traditional scroll-window presentation, has many advantages, in particular improved
overview of the text, and a general flexibility that corresponds well to the diversity of
activities involved in composing a long document. However, the paper model also
presents a number of design challenges, which we have addressed throughout the study as
a result of the evaluations.
After an introduction to the paper model and the research behind it, three studies
are presented. The aim of the first one was to assess the adequacy of the paper model to
support so-called in-process planning (as opposed to pre-writing planning). We found
evidence that Paper is a flexible and powerful writing tool that facilitates integrated
planning during the whole writing process.
- 6 -
In the second and third studies we focused on real use situations of Paper and
long-term effects of a page-based text presentation. Two subjects used Paper for several
months to produce work-related texts, and their activities were observed with
ethnographic methods. Both case studies provided valuable results and a basis for a
discussion informing the design of future implementations of the paper model.
- 7 -
2. PERSPECTIVES ON WRITING AND TECHNOLOGY
2.1 Research perspectives and models of writing
Writing is a complex and cognitively demanding activity, not easy to study either in
controlled experiments or in real-life situations. Different writing styles, widely varying
writing tasks and diverging writing environments are factors that have to be taken into
account by researchers. In addition, writing is often a long-term activity, stretching over a
considerable period of time. Setting up a controlled experiment might lead to the
simplification of some of the parameters at play, whereas studying real-life writing can be
quite complicated from a practical point of view.
Flower and Hayes’ cognitive model of writing (Flower and Hayes, 1981) had a
great influence both on writing research in general and on the methods used to study
writing. The model described writing as a hierarchically organized mental process.
Planning, transcribing and revising defined the basic activities of the process, between
which the writer switches during composition. Think-aloud methodology was used to
uncover the on-going cognitive processes in empirical studies. In this framework, the
ability to plan was given a central role in the characterization of good writers. Planning
was seen as consisting of three subprocesses: generating ideas, organizing ideas and text,
and setting goals for the writing activity. A neglected aspect in this cognitivist model was
how writers interact with their environment, and how they rely on external sources and
tools in building their texts.
Another factor that influenced research on the writing process during the eighties
was the increasing spread of computer technology. Many studies were done on the effects
of computer use on the writing process (see e.g. van Waes, 1992; Lutz, 1987; Haas,
1989). At the same time, computer technology had an impact on the very way writing
was studied. For instance, computers could be used to log writers, providing an exact
history of the activities undertaken or the commands used during the writing session.
- 8 -
In recent years, a social-cognitive approach to writing has gained ground upon the ”pure”
cognitive theory of writing. Writing is no longer seen only as an individual activity: the
social aspects and the context of the writing situation are increasingly taken into account. Flower (1989) outlined the frame for an interactive theory that strives to explain how
”cognition and context interact, in specific but significant situations”. She proposed three
principles of interaction between the cognition and the context that she hoped would
inform a multi-perspective approach to writing.
“One principle is that cultural and social context can provide direct cues to cognition. The second is that context is also and always mediated by the cognition of the individual writer. And the third is that the bounded purposes that emerge from this process are highly constrained but at the same time meaningful, rhetorical acts.” (Flower, 1989, p.287)
Nystrand (1989) used a slightly different approach in his social-interactive model
of writing as an ongoing discourse between the writer and the reader. According to the
model, every time the readers understand a written text, writers and readers interact.
Brandt (1992) went even further in her argumentation when she advocated tearing down
the limits between the social and the cognitive, instead of considering the interaction
between them. According to Brandt, there is no such thing as an interaction going on
between the social and the cognitive, since it would imply that cognition and context are
two separate things.
Along with new perspectives on models of writing a need emerged to develop and
adapt methods to study writing from a more situated stance. Qualitative methods,
including case studies, interviews and direct observations (Patton, 1987; Yin, 1984) were
increasingly applied. Chin (1994) found that approximately 30% of the studies published
during the past 6 years in two major journals for empirical research on composition
involved subjects being interviewed. However, she points out that not much research has
been done on how the interviews should be conducted, on how the data gathered should
be analyzed, nor on the way the results should be reported. According to Chin, a critical
analysis of qualitative methods such as interviewing is needed within the writing research
community.
Van der Geest (1996) argues that, in order to be able to use case studies as a
reliable method in studying real-life writing situations, a common agreement is needed
- 9 -
about how document production processes should be studied. This would then allow
researchers to build up a large set of similar data, which in turn would facilitate an
analytical generalization based on replication of case studies. To that end, van der Geest
proposes a framework of factors and characteristics of document production that should
be included/used in such studies: e.g. the actors involved in the document production
process, the activities they undertake, and the communication between them.
2.2 The influence of computers on writing processes
In the early days of word processing technology, many believed that it would make
certain aspects of writing easier compared to writing with pen and paper (see e.g. Daiute,
1983). But as time has gone by, and an increased number of comparative studies have
been conducted, the picture is more complex. Some results are even contradictory. For
instance, a study performed by Lutz (Lutz, 1987) showed that writing technology has an
impact on the way the subjects revised: subjects revise more on-line and focus on
surface-level concerns when writing with a computer. Other studies (e.g. Hill, Wallace
and Haas, 1991), reported no significant difference between the revising processes of
writers working with pen and paper and the same writers revising on a computer. One
possible explanation for these divergent results is given by Hill et al.:
”Given the complexity of the writing process, the sensitive nature of its interaction with the technology of word processing, and the differences in computer technology, these conflicting results are not surprising.” (p. 86)
However, it seems that most studies agreed on the fact that expert writers make more
revisions than novice ones and that these revisions generally are of a higher level than
those done by novices (Hill, Wallace and& Haas, 1991; Hayes et al, 1987; Lutz, 1987;
van Waes, 1992; 2003). This holds for computer supported writing as well as for writing
with pen and paper. Moreover, the texts produced may also be affected by the writing
medium. For instance, Haas found that texts composed at the computer were longer and
of lower quality than the one written by hand (Haas,1989b). On the other hand, some
studies have reported on higher quality for computer produced texts (e.g Owston, Murphy
and Wideman 1992).
- 10 -
When referring to the computer as a writing medium, one has to keep in mind that
the computer and the word processor used in a specific study together define a special
case of a computer setting. Often, in expressions like ”writing with computers”, ”the
computer setting” or ”comparing pen and paper with computer”, the word computer
includes the word processor as well. In the case of the studies mentioned earlier, the word
processor is of the traditional kind, showing the text in a scrolling window. Hence, the
studies done so far have looked at computer-supported writing in the special case where a
word processor using scroll functionality to display the text is used. Of course, some
features in computers are independent of the kind of word processor used: the ease of
changing the text, the relatively limited size of the screen, etc., but the range of the effects
of the special word processor used often remain unknown. Only in a few studies, the
effects of the functionality of the word processor used were mentioned and questions
were raised about their influence on the results of the study. For example, in Haas
(1989b) and in Lutz (1987), the effect of the limited visual access to one’s text was
discussed and were believed to disturb what would be a ”natural” writing process. Haas
wrote that it ”... may actually discourage attending to the whole text”.
When it comes to planning, differences have been found between computer and
pen and paper: there is less total planning and planning is more concerned with lexical
and syntactical issues when subjects use a word processor (Haas, 1989a). Also, less
planning is done before writers actually start writing the text. Haas found that when
writing with pen and paper 42,6% of the total planning was ”initial planning”, while the
percentage for the computer setting was only 28,3 %. Lutz (1987) observed that: ”... more
than half the participants said they had begun without a plan and were willing to just ’see
how it [the text] would go’”.
Moreover, planning activities that take place during a pre-writing phase (before
the actual writing starts) sometimes called initial planning, have been devoted special
attention and several planning tools have been designed to support that cognitively
demanding phase of writing (Kellogg, 1989; Lansman, Smith and Weber, 1993; Smith
and Lansman, 1989). One reason for encouraging writers to plan more at the computer is
that results show that good writers seem to plan more than basic writers (Bereiter and
- 11 -
Scardamalia, 1987; Flower and Hayes, 1980, 1981) and that good writers plan at a higher,
conceptual level (Haas, 1989a)
Unfortunately, such planning tools do not seem to be used to a great extent.
According to Hovmöller (1995), one reason seems to be that they are not integrated with
the remaining writing process. Another reason could be details of the design that give
these programs bad affordances, which may discourage the already cognitively loaded
writer. Yet another reason could simply be that the writer wants to start writing at once
and would rather start directly in the word processor than spend time with some special
planning tool. This seems relevant at least in moderately-sized writing tasks.
2.3 Overview problems when (reading and) writing with
computers
A lack of adequate overview has been reported in many studies of computer-based
writing. The characteristics of the computer and word-processing can make it difficult for
writers to grasp the overall organisation and flow of a text.
During the 1980’s, Haas interviewed writers about their experiences using, and
learning to use, a computer for writing (Haas, 1996). One problem the writers reported
was their difficulty to get a sense of the text. They used different expressions to describe
their experience: “My text is hard to pin down on-line”; “ There is a problem getting a
feel for the piece”, etc. (Haas 1996, p 58). More precisely, sense of the text is described
as:
“[…] the feeling a user may have that he or she has a good grasp of the structural and semantic arrangement of the text – the absolute and relative location of each topic and the amount of space devoted to each.” (Hansen and Haas, 1988; p.1084)
Many possible factors are mentioned by Hansen and Haas (1988) to have a
negative effect on the users’ sense of the text. One of them is the lack of overview due to
the limited screen size, which reduces the context for each part of the text. Another factor
is the scrolling functionality that results in a changing text line at the top of the window.
Yet other factors mentioned are, for example, limited legibility and poor responsiveness
(while scrolling).
- 12 -
A related result was reported in Severinson Eklundh and Sjöholm’s (1991)
longitudinal survey of 70 writers using word processors at the Royal Institute of
Technology. About three quarters of the subjects, in one form or another, expressed that
they had a too restricted view of the text when composing long documents. This resulted
in many print-outs, motivated partly by the need to read the text in order to get an
overview of what had been written so far. Process models of writing emphasize the need
to revise (and thus to read) the material as it is being written. As a part of the writing
process, writers need to read their text both at a local and global level in order to get a
global overview of their text. However, using a computer to write, “many writers found it
difficult to access large parts of their writing or move quickly to a specific place in the
text” (Haas, 1996). Compared to pen and paper, computer based reading and writing has
several drawbacks, that Haas suggests could be alleviated by the use of print-outs:
“Possibly it takes less time to shuffle through several sheets of paper to find a particular paragraph than to move through comparable screens of text. Pen-and -paper writers might also be aided by being able to spread out three, or four, or a dozen sheets of paper in front of them, getting a sense of the text as physical object. Other writers reported that the “intimacy” that they had developed with pen and paper is difficult to achieve on the computer. Still others found it difficult to reorganize their text on-line or to detect errors”. (Haas, 1996; p.55)
A distinction has been made between physical and logical overview (Severinson
Eklundh, 1992), and both are equally needed to provide the user with a global perspective
of the text. Physical overview is the way in which the text is presented to the
writer/reader through its typographical layout and the physical placement of its parts on a
sequence of pages. Logical overview is how the conceptual structure of the text is
presented to the writer or reader. For instance, the outline facility in MSWord may
provide a logical overview of the text to the user.
2.4 A tradition for paginated text presentation
When writing or reading on paper, we take many properties of the medium for granted.
To flip through written pages in order to get a quick look and feeling for the structure and
length of the text is an effective way of getting “oriented” in the text: we do it almost as a
reflex. Moreover, there is evidence for a spatial memory of the text when reading on
paper, i.e. readers recall the location of text on the pages (Rothkopf, 1971). Lovelace and
- 13 -
Southall (1983) showed that within-page location and the content of words from a prose
passage can serve to cue recall of one another when reading on paper. In short, paper has
many good qualities that the human mind has well-developed cognitive processes to use.
One effect of computer technology is that it affects the directness of inscription
(Chandler, 1994). With a word processor, inscription is both indirect and delayed, since
the text is not directly put on paper (as in handwriting), nor is it done ”simultaneously”
(printing is a separate process). Chandler (1994) also compares directness of inscription
to a sense the writers have of the text as physical substance. Many writers experienced
the fluidity of screen text as something disturbing, a problem that can be compared to the
lack of an adequate “sense of the text” described by Hansen and Haas (1988). Writers
seem to have a relationship to the page, as an entity of text and a physical object that has
the affordance to be manipulated in tangible ways. In a study by O’Hara and Sellen
(1997), the superiority of paper handling compared to computer screens when it comes to
reading (for the purpose of writing) was assessed. They came to the following
conclusion:
“But in the support of reading for the purpose of writing, this study has shown that the benefits of paper far outweigh those of on-line tools.” (O’Hara and Sellen, 1997; p.341)
An experimental study by Piolat, Roussey and Thunin (1997) of reading and
revising tasks presented evidence that spatial references offered by page presentation on
the screen help the writer to grasp the overall organization and flow of the text. The
results showed that the user’s overall comprehension of the text is better with a page wise
presentation than a scrolled presentation. What remains to be shown, from a perspective
of designing writing tools, is:
1- How can the page presentation be included in a flexible writing environment?
2- Will such an environment be a successful tool for long term writing tasks?
In the following sections, we will describe our approach to addressing these questions.
- 14 -
3. USING PAPER AS A DESIGN MODEL FOR
COMPUTER BASED WRITING
Looking at some of paper’s properties and how to bring them into a computer based
writing tool, has been the agenda of a research project that started already in 1988 at the
Computer Science Department of the Royal Institute of Technology in Stockholm. The
project attempts to address some of the problems related to computer-based writing
reported in the literature, specifically the lack of overview over lengthy texts experienced
by the users. Over the years, several prototypes have been designed with a common focus
on the paper model, and where each version builds upon user experiences in various
writing tasks and computer environments.
The basic idea is to use paper as a metaphor in the design of an electronical
writing tool, trying to transfer some of the qualities of paper to the computer. Metaphors
and analogies are broadly used in system design, as a powerful way of taking into
account the users’ previous experience and knowledge, while they interact with the
system interface (Norman, 1986; Allwood 1991; Neale and Carroll, 1997; Booth, 1990;
Wozny 1989; Sasse, 1997; Mynatt and Macfarlane, 1987). The use of paper is well
embedded in humans’ working, reading and writing habits, which motivates taking some
of the affordances of paper into consideration when designing new word processing
technology (Sellen and Harper, 2002). If appropriately designed, the new technology will
support the user in achieving her goals with an optimal combination of “means” taken
from both the paper and digital worlds.
In a writing environment built on the paper metaphor, the writer at the computer
is able to write text on simulated paper pages. When the current page is filled with text, a
new blank page appears. Like sheets on a desktop, the pages can be spread out on the
screen to yield an overview of the text, and the user can rearrange the pages and organize
them to reflect the needs of her current writing task.
To increase the writer’s global perspective of the text is one important goal with the
paper model. This can be achieved by enabling the user to view the whole document at
one time. This includes the possibility to spread out the pages on the screen, making it
possible to see several pages simultaneously, either consecutive pages or pages from
- 15 -
different parts of the text. Similarly, the possibilities of gathering pages and moving them
by direct manipulation help the writer to organize the desk and thereby keep a better
overview of a specific part of the text.
The paper model is also expected to facilitate spatial orientation within the document.
When writing or reading on paper, we make constant use of the spatial arrangement of
the text to remind ourselves of its inherent structure. By presenting text on separate pages
(windows in the system), and making it possible to manipulate and rearrange these pages,
it is possible to have a similar use of physical cues in computer based writing. For
instance, a graphical representation on the screen of the “depth” of a document (number
of pages) may be useful to the user’s general perception of the electronic document.
A related property of the paper model is its strive to display text in as static a
manner as possible. The idea is, by allowing the text to stay stable relative to a given
window, to improve the writers’ spatial memory of the text. There is evidence for spatial
memory and spatial encoding of information when subjects read on paper (Rothkopf,
1971; Lovelace and Southall, 1983). Several studies have shown that subjects remember
the location of text items more accurately than they would by chance and that this helps
them memorize the content of the text. It was assumed that the paper model could help in
introducing some of these benefits of spatial memory in computer supported writing.
3.1 The implementation of the paper model – historical
perspective
The implementation of the paper model into a prototypical writing tool called Paper has
been an important part of the project. Over the last ten years, Paper has evolved from a
simple prototype, to a more elaborate version used in the recent studies (see Fig. 1).
Paper was written in the Smalltalk language1. Important changes made to the program
over the years include adding the possibility to easily switch between different page
presentation types, a speed upgrade, and a more transparent, direct-manipulative
interface.
1 Smalltalk is a graphical, interactive environment for object-oriented programming (see e.g. Lewis,
1995)
- 16 -
Figure 1. The development of the Paper program and the studies during the same period of time
1996 2000
Studies
Versions of Paper
1988 1992 1998/99
First version of the
Paper prototype
Smalltalk-80 on
Tektronix
4405/4406
Object Works 4.0
version, following
the Macintosh
standard.
Easier to use for non
Smalltalk-users.
First formative
evaluation study –
important interface
design issues
uncovered (Fatton,
1992)
Visual Works 2,5
version running on PC
and much faster.
Toolbar and other
directmanipulative
facilities added.
Possibility to switch
between page
presentations
Study to assess the
adequacy of Paper to
support planning during
the entire writing process
(Fatton and Severinson
Eklundh, 1996)
Longitudinal study of
Paper with 2 subjects
writing realistic writing
tasks in Paper
Parallel projects
testing Paper as a
collaborative
writing tool result
in a version called
CoPaper and a
Unix version of
Paper with special
co-writing
functionality
- 17 -
3.2 Introducing a new metaphor for on-line writing tools –
early design choices and implementation requirements
The paper metaphor might seem very concrete and straightforward at first sight, but its
translation into writing tool functionality presents several design problems. The results of
the first formative evaluation study conducted in 1992 (Severinson Eklundh, Fatton and
Romberger, 1996) pointed at some, for the paper model, central questions.
3.2.1. Writing on fixed-sized pages Maybe the most challenging and central design issue for the paper model, is the one of
page overflow. What should happen, in terms of functionality, in the writing tool when:
• the writer reaches the end of the page and needs another page to write on?
• the writer needs to add text in an already full page?
In early phases of the project, different solutions to these problems were envisaged, e.g.
pages that could grow dynamically or the use of “stick-it” notes for insertion of extra text.
However, as mentioned earlier, a goal has been that the program should keep the location
of the text in relation to the page constant as far as possible to support the writer’s spatial
memory (Lovelace and Southall, 1983; Rothkopf, 1971; Piolat et al 1997). The solution
currently chosen to the problem therefore is as follows: whenever a page becomes full, a
new one appears automatically to the right of the full page, and the user goes on writing
without interruption. This preserves the continuity of the writing process, an important
aspect of writing which should not be altered by the paper model.
1 1 2
blank
blank
- 18 -
Figure 2. A page split.
This page split solution has both drawbacks and advantages. The fact that it keeps the text
as static as possible on the page is, of course, a great advantage. One problem, however,
is that the page split generates blank areas that do not have any counterpart in a
traditional word processor (see Fig. 2). Those blanks areas are different from, for
instance, a succession of return characters. Furthermore, too many of those blanks were
shown, in our studies, to cause the text to become fragmented. This, in turn, resulted in an
impoverished local text context around the point of insertion. The fact that a command is
needed in Paper in order to eliminate these blank areas, is an example of how the
solution to the page overflow problem has repercussions on other parts of the program.
Consequently, many different aspects have to be taken under consideration when
designing the functionality to support a page split solution.
3.2.2. Are the text and the page different objects? When using a traditional word processor, the text is the conceptual object in the user’s
mental model on which the commands operate and that is shown in the scrolling window.
Even though page presentation is possible in some parts of the program, like in Print
Preview or Page Layout in MSWord, they are usually only used to display text on, and
they cannot be freely manipulated or moved on the screen.
However, in an implementation of the paper model, two distinct objects with their
own commands to manipulate them coexist: the text and the page. For instance, the user
can execute editing commands on the text and specific page spreading commands on the
pages. This could potentially result in some confusion for the user, especially if it is not
obvious enough whether a command yields a page or the text. For instance, the command
Select All could be referring to a selection of the whole text of the document (useful for
editing) or a selection of all the pages of one document (useful for executing commands
on pages). Through good system design, for instance gathering the different command
types in different menus, misinterpretations by the users can be reduced.
- 19 -
The paper model also raises the question of the relationship of the text to the page.
Since pages are displayed on the screen as objects that can be manipulated by the user,
she/he might think that the pages are entities of text, i.e. that the text “belongs” to the
page. Waern (1989) reflected on the difference between pen and paper and word
processing in that respect:
“…the text “belongs” to a particular position on the paper. It “occupies” this place, which means that in order to change the text it is necessary either to rub it out and replace it by another text or to cut it out together with the paper it occupies. The paper limits the possible movements of the text.” (p. 160)
In the case of the use of a paper metaphor in word processing, one has to consider
the following questions: what should happen when the user decides to “throw away” a
page (with a Delete command, for instance)? Will the text also be lost? Or are the pages
only “windows” to the text, in the same sense that the scrolling window is in traditional
word processors? Functionality should support the design choices made and whatever
disturbing mismatch between the source and target domains of the metaphor should be
handled in an optimal way (Neale and Carroll, 1997; Carroll and Mack, 1985). In the
earliest versions of Paper, only limited functionality supported handling the text as a
whole. The text was divided into pages and editing commands applied only on one page
at a time. During informal tests of the program, this “one page” functionality was found
to cause problems for users. It worked against one of the greatest advantages of word
processing, i.e. the possibility to operate on the whole document with just one command,
e.g. a formatting command. Hence, as a new version was designed, the decision was
taken to give priority to text functionality at the document level rather than the page level.
Another relevant aspect of introducing the paper model into word processing is
that handling pages might have some similarities with handling graphical objects in
painting and drawing applications. If the users have experience from such systems prior
to their contact with the paper model, it might be easy for them to understand such
commands as Send to Back or Bring to Front when it comes to ordering the pages.
4. THE PAPER WRITING ENVIRONMENT
- 20 -
For reasons of space, we only present the latest version of Paper here, the one that has
been used in the two last case studies presented in Chapter 6. The version used in the
study of Chapter 5, is an earlier version referred to in 3.2 The implementation of the
paper model – Historical perspective.
4.1 The current version of Paper
Paper has been developed primarily for research purposes and therefore lacks many of
the functions in commercial word processors. There are only basic functions for
formatting the text. Some standard heading styles and normal text formats are
implemented and the text can be saved with formatting. There are, however, currently no
functions for creating tables, inserting pictures or language checking.
General functionality
The user can choose to display his/her text in different ways: either in a scroll-window
(very much like the one in MSWord or other well-known word processors) or on separate
“sheets of paper” implemented as rectangular windows. The user can also “shrink” the
documents into small window icons with only their titles visible (see Fig. 4.a). When the
icon is “re-opened”, the pages are displayed in their previous positions on the screen.
Figure 3. The menu and the tool bar .
All the pages in one document have the same size, but pages in different
documents can be of different sizes. There are four pre-defined sizes: small, medium and
large and “standard printer size” (during the study presented in chapter 6, this page
format was called A4).
Writing and editing on “sheets of paper”
- 21 -
When the user starts Paper, a blank untitled sheet is available for writing. This page is
then the active page, which is indicated by the color of the title bar. To make a page
active the user just clicks on it. Each page in Paper is labelled with the name of the
document to which the page belongs and a page number showing the relative order of the
pages within the document.
The user then starts to write on the empty page, or opens an older document to work
with. When the page is full, a new page automatically appears to the right of the full
page, enabling the user to see the first and the second pages simultaneously on the screen
and providing him or her with a local context to write the second page.
When an insertion for which there is not enough room on the page is made, the
page is split into two pages at the place of the insertion, and writing continues on the first
of the two pages (see Fig. 4.b). The new page, containing the part of the text following
the place of the insertion, is placed to the right of the previous page, and the user can
continue to write the inserted text without interruption. The new page is given a "decimal
number". For instance, if text is added in page 1 and page 2 already exists, the inserted
page is given the number 1.1. In this way, pages are consistently numbered regardless of
changes made, causing page splits.
- 22 -
Fiure 4.a. A Paper screen with three documents: one presented in a scroll window (background), one
presented as medium size Paper pages, and one iconized document to the right. At the top, the menu and
tool bar.
Whenever the user wishes, the text can be compressed by using a command
(Compress text) that deletes all the white areas arising from page splits. When this is
done, pages are renumbered. The user is also free to renumber the pages whenever she/he
wants with the Renumber pages command. The basic text editing operations (Copy, Cut,
Paste and Find) are available in the Edit menu.
Keeping track of the pages on the screen
When there are many pages spread over the desk, overlapping each other, the user will
need support for orientation and navigation among the pages. By using the commands in
the Go menu, the user can flip the pages forward or backwards. The user can also go to a
specific page by writing a page number in the Go to… command dialog box. To “go” to a
- 23 -
certain page means that this page becomes the active one and appears on top of the
others.
The Stack text command can be used to gather the pages of a document into a
neat stack (see Fig. 5). The stack can later on be moved to another position of the screen,
by using the “Move stack” command.
To keep track of the documents opened on the screen, the command Active
document… in the File menu can be used. A list of the documents available is displayed
and the user can choose one of them by clicking on its title.
Figure 4.b. The result of an insertion under the heading 3.4 US LEGISLATION. The page is split into
two: the first one, containing the part of the text before the insertion point as well as the insertion itself, and
the second one, containing the text located below the point of insertion.
- 24 -
Gaining an overview of the text
With the commands in the overview menu, the user can spread out the pages neatly in
several different ways. Pages can be spread out over the whole desk, edge to edge if they
fit completely, or partly overlapping if there is not enough room, overlapping vertically
or horizontally.
To obtain a local context for the active page, the commands View spread (2
pages) and View Context (3 pages) can be used. When using the command Next context,
the pages are “flipped” forward, resulting in the next spread or context being displayed
(the one for the consecutive page in the document).
- 25 -
Figure 5. A Paper screen displaying four documents. One in a scroll window, one blank document with
medium pages, one stacked document with small pages, and one iconized document.
4.2 Design iterations
The version of the Paper program presented above is the result of several iterations, each
one resulting in improvements of the prototype’s functionality. In that process, the first
formative evaluation of Paper in 1992 played an important role. The problems revealed
during the study were mostly related to page splits and the compress text command, to
confusion about different kinds of white areas on the pages (blanks as a result of a page
split or a succession of return characters) and to deleting/closing a page or a whole
document. In order to improve the usability of the Paper prototype, improvements were
made to alleviate these problems. For instance, “hard” page breaks allow the user to
preserve the integrity of certain pages during the execution of a Compress text command.
Moreover, the pages remain in their prior position after the execution of the command.
To avoid the users’ confusion about the nature of white areas on the pages (blanks as
a result of page splits or a succession/sequence of return character), the possibility to
display paragraph marks was implemented.
To increase the users’ perception of the text as a whole, as opposed to text chunks
displayed on separate pages, the possibility to use the arrow keys to move the cursor
back and forth in the text, even across page limits, was implemented.
Menu commands were added that allowed the user to clear a page or delete it. The
closing box of the window was thus disabled. Moreover a command for closing a
whole document (i.e. all the pages of the documents) was added.
5. A STUDY OF IN-PROCESS PLANNING IN PAPER
5.1 Goal and research questions
- 26 -
An important goal with the paper model has been to, by its implementation in a computer
based writing tool, support writers during the demanding process of writing. Planning has
been recognized to be an important activity during this process, both as initial planning
before writers start to write, and as one of the subprocesses that the user switches
between during the whole writing process.
Our assumptions that Paper offers adequate support for initial and in-process
planning (planning that goes on during the whole writing process as opposed to initial
planning) was tested in a case study with four subjects. We expected Paper to be an
appropriate writing environment for both planning and writing on the following grounds:
The planning tool is completely integrated in the word processor, since they are the
same application. This makes the writing environment flexible. For instance, the user
might use small Paper pages to write notes or an outline on. Later on, a ”note” can
become a page, and vice-versa (see also the NoteCards system in Halasz, Moran and
Trigg, 1987). Our observations show that Paper provides objects (e.g. pages) that can
be used differently by different users for different tasks.
Paper offers support for planning during the whole writing process, because of the
improved global perspective due to the page-wise presentation of the text. By
supporting the writer's spatial memory, we expected that it could also make the writer
plan (and revise) at a higher, conceptual level.
Paper is easy to learn and to use: the page is the basic concept in Paper and it is very
simple and concrete.
The study was a formative evaluation in which four subjects participated. The
complexity of the writing activity and the difference in writing strategies explain the
limited number of subjects involved (Fatton and Severinson Eklundh, 1996). The study
was set up to answer questions about how to support planning not only under a pre-
writing phase, but also during the whole writing process. How can the integration of
planning, writing and reviewing be made as smooth as possible? What aspects are
relevant to in-process planning?
- 27 -
5.2 Study design
The four subjects were graduate students in HCI and Computer Science, all used to word
processing, but with varying experience of writing. None of the subjects had any earlier
experience of the Paper program and none had ever used any special planning tool, not
even an outlining tool (this was revealed in the interviews).
The subjects participated in two writing sessions of one hour each, with one
week's interval between the sessions. Before the first session, the Paper system and the
task were introduced to the subjects. The introduction to Paper included examples of how
to work with pages of different sizes. In the task description, the subjects were asked to
write an argumentative text about traffic problems in large cities. The subjects were given
some text material, two articles about city traffic, and told that they would only have
access to them during the first session (to encourage them to plan and take notes during
the first session). The description instructed the subject to use the material to collect ideas
and facts and that they should refer to this material in their text. They were instructed to
write at their own pace, but to have in mind a text of approximately five A4 pages.
After each writing session, a retrospective interview was performed. In the first
interview, the subjects were asked to:
1. describe their usual writing strategies
2. go through (retrospectively) their activities during the session
3. comment on, from their point of view, good and bad qualities of the writing tool.
In the second interview, the questions just included parts 2 and 3.
The sessions and the interviews were videotaped. The experimenter also took
notes during the sessions and answered any questions regarding the Paper program that
would come up. Spontaneous comments from the subjects were written down.
5.3 Results
Using several documents to plan and to write
Three subjects out of four created different documents for different purposes: one or two
note document(s), one outline document, and one document to write their text on, usually
referred to as the “main document” or “the text document” or “the article”. The subjects
- 28 -
named the documents according to their purposes and used them in different ways to
solve the writing task. They often switched between them, looked at the arguments on the
note document and marked them as using them. They spread pages out to get an overview
and created new ones after working a while. Several subjects used different page sizes for
different documents (one subject commented that the size of the page reveals the
importance of the document).
When it comes to manipulating the pages on the screen, we observed a great
variation in the time and effort spent on this activity and in the resulting location of the
pages. One subject placed her two note documents in the left part of the screen, one
below the other, and her main document in the middle of the screen. Another subject
spent a lot of time organizing the pages. She often spread the pages out on the entire
screen, but never seemed to be satisfied with the location of the pages.
The fourth subject had a completely different working strategy: he created only
one all-purpose document for writing down headings, notes and text. He created many
pages (in an “uncontrolled” way) and many blank areas resulting from page splits that did
not try to remove, but instead used to write on. Only much later did the subject try to
organize the text chunks (with cut and paste) and the pages to create a text with a
meaningful organization.
Interview results
All the subjects mentioned the support for overview and simultaneous visual access to
notes and main text as an advantage. One subject (subject 4) compared it with MSWord,
where her outline would have been ”spatially ruined” since she would have written the
running text directly under the headings (she reported, like the other subjects that she had
never used the outline mode in MSWord). In general the subjects thought that Paper
supported their usual way of working and made it easier for them in some aspects. Some
subjects mentioned that the page presentation in Paper offered an alternative to the print-
outs they usually needed in order to have visual access to both points/notes/outline
document and the main text.
The possibility to view different parts of the same document simultaneously, to
check consistency of an argument used in different places (i.e. in the introduction and in
- 29 -
some later part of the text), are other examples of qualities of Paper that were positively
valued by the subjects.
Problems experienced were mostly related to page splits and resulting blanks.
Other disadvantages mentioned in the interviews were the execution of organizing and
compress commands since they changed the location of pages on the screen (one subject
reported that she would then rearrange the pages manually, which was quite time-
consuming). One subject (the one who only wrote in one document) mentioned
orientation problems: he felt he was on ”side pages” all the time and ”far back in the
document”, mostly during the first session. However, the subject developed a better
working strategy in the second session, using the spread out command and trying to get
control over the pages.
5.4 Discussion
The case study showed that the Paper program, based on the metaphor of paper pages,
was perceived as a simple and useful tool for integrated writing and planning. The
possibility to work with several documents (to have them open on the screen at the same
time) was exploited in various ways in the process or planning. The subjects used their
notes document in several ways: all looked at them and read them, one marked, on the
note page, arguments as she was ”using” them, two added new points after beginning to
write the main text. Moreover, the possibility to have different page sizes (and in some
cases even different shapes) for the documents was another way of providing the subjects
with visual cues in organizing their thoughts. This aspect of planning, to have
simultaneous visual access to several documents, is well supported in Paper and could be
improved with extended functionality for handling several documents.
The subjects appreciated Paper’s functionality not only for the overview offered
by the simultaneous display of documents on the screen, but also for the page base
presentation of the text. For instance, one subject said that Paper reminded her of real
paper and allowed her to think ”in pages”. When describing her usual writing strategy in
MSWord (this is the word processor she uses), she said that she ”thinks in continuous
texts”. As for the subject that only wrote in one document, he deliberately made some
blank space on the pages and created many pages, inviting him to add text at different
- 30 -
locations, just as his ideas were developing. This subject in fact used the blanks on the
pages in his planning strategy: he needed the blanks to separate ideas or pieces of text at
the beginning of the writing process, and used them later on as visual reminders of places
to write additional text.
The study also shows that Paper is a flexible, non-controlling environment,
without modes and states that add complexity to the writing task. In O’Hara and Sellen’s
study (1997), on-line tools were found to cause problems because they do not support the
seamless integration of note-taking while reading (for writing). In the study, we observed
that the subjects switched many times between different documents and views. The
transition between planning and writing was smooth and easy. This is not the case with
most traditional pre-writing tools, which might explain their limited use by writers
(Hovmöller, 1995).
During the two hours of the writing sessions, the subjects developed quite
different ways of working with Paper. For most writers, working with several distinct
documents seems to be a way to organize their planning activity, but the study also shows
a great variation. Personal variations in writing strategies result in sometimes opposite
demands on the functionality of Paper. For example, one of the subjects asked for a more
”clever” Paper program, regarding the location of re-displayed pages after the executions
of compress text commands. Another subject wished for a more ”manual” way of
handling pages. It is also to be noticed that while the subjects agreed quite well on the
positive aspects of Paper, like overview and visual access to several documents at a time,
they had more different opinions on the problems with Paper due to their different
working strategies. The fact that they all appreciated the overview functionality seems to
show that it is always important to have a global perspective, regardless of what writing
strategy the writer is using.
Although the case study provided us with a lot of data assessing the adequacy of
the use of Paper for integrated planning, it says little about possible long-term strategies
when working with Paper. This is one limitation of the method used. There is a need for
having writers work an extensive period of time in Paper and studying the long-term
effects of the page-wise presentation of text.
- 31 -
6. TWO LONGITUDINAL, NATURALISTIC CASE
STUDIES OF WRITING IN PAPER
When designing a computer supported writing tool, the ultimate question is whether or
not writers can actually use the tool to compose for their daily work purposes. To obtain
an answer to that question requires having writers use the program for a long time in a
realistic setting. At the time when a new, faster version of Paper was released in 1998,
this became an important agenda. We decided to observe subjects, in as natural a context
as possible, writing texts that they had to write as part of their professional or educational
work.
We were also interested in observing whether users would prefer to use a page
presentation compared to a scroll presentation. To that end, the possibility to display
documents either in a scroll window or in separate pages was added to Paper, and the
user could easily switch between the two presentation types. Hence, we were able to
observe the frequency of the use of page presentation, as a result of the subjects’ own
choice between several presentation types.
In the earlier studies (e.g. the planning study), the use of the page handling
functionality differed from one subject to another, and differences in strategies over time
were also observed. This incited us to believe that studying writers’ use of Paper under a
lengthy period of time would reveal longitudinal changes and problems related to
different stages in the writing process. These, in turn, would presumably provide design-
related explanations for either an acceptance or a rejection, by the users, of the paper
model.
Two subjects participated in the longitudinal, naturalistic case studies presented
here. Since we were interested in long-term effects of the page presentation, we observed
the subjects for an extensive period of time, collecting data from several data sources.
Case studies are important to study real life writing processes. As van der Geest (1996)
expresses it:
“studying real-life writing processes poses serious methodological challenges to the researcher. Due to the particulars of the processes studied, such as the “situatedness” of the processes and
- 32 -
the lengthy period of time it often expands over, researchers have to depend on methods such as case studies and ethnographic accounts.” (p. 310)
Our intention to run the study in as naturalistic a setting as possible resulted in an
ethnographically inspired methodology. It was crucial for us that the subjects wrote tasks
that were part of their real work situation. The motivation factor seemed important.
However, this kind of qualitative case study is associated with many problems. The
problems range from practical ones (difficulty of observation, lengthy period of time) to
more theoretical considerations about the method’s validity and the limited generalization
of the findings. However, these problems are remediable, as long as the researcher is
aware of them. For example, a “thick report”, including a detailed description of the
method used as well as of the data collected, improves the quality of the study (Chin,
1994).
At the same time, qualitative studies are also very informative and enable us to
study the real use of and practices related to (new) word processors. In these particular
studies, emphasis is put not so much on the writing process itself, but rather on the ways
in which specific features of a writing tool support the writer during the demanding task
of composing a lengthy text.
6.1 Goal and research questions
The study was aimed at investigating, in an explorative way, if and how the paper model
supports the writer during an extensive, “real life” writing task. Several research
questions guided the study.
• To what extent, in terms of the subject’s subjective satisfaction, does Paper
support writing in a natural setting during an extended period of time?
• What are the impacts of the time factor on possible working strategies in Paper?
• Are scroll and page presentations used in different parts of the writing process? If
this is the case, how is one presentation better suited for a specific task and what
is the writer’s account of that use?
- 33 -
• How can the paper model be implemented in an optimal way regarding previously
observed problems, i.e. page split functionality, page handling and disposition of
pages on the screen?
6.2 The subjects and their writing tasks
Two subjects participated in the study, writing their own, real-life tasks. The second
subject began writing approximately six months after the first case study ended.
Subject A
The first subject (A) was a female, last year student of a 4,5 years program in
Environmental Engineering. A had limited writing experience: she had mostly written
course reports, together with one or two other students. Prior to the study, A had used
MSWord 6.0 on PC. She reported that she did not take advantage of its more advanced
features (e.g. outline mode), rather used it in a ”basic” way.
A wrote her MSc thesis in Paper. A Master’s project represents 20 weeks of full
time work. The report is an important part of the work, and students put a lot of effort in
writing what usually is the largest document they produced until then. This MSc thesis
was in the field of water resource engineering and the final version of the thesis is 70
pages (A4) long.
The case study stretched over an extensive period of time, from September 98
until January 99 with an interruption of one month in October, when A made a visit to a
wastewater treatment plant in Poland.
The study was naturalistic, in the sense that the writing task was a real task that A
had to fulfil in order to get her degree and the setting was as close as possible to the one
that would have been the natural one. Furthermore, the time limits were naturally defined
by the length of the writing task, i.e. A maintained her own time schedule.
Subject B
- 34 -
The subject (B) was a male doctoral student in computer science. He was an experienced
writer, used to writing scientific reports and articles. Since his research area was
collaborative writing, B was familiar with writing research and writing theories in
general. This might have influenced some of his answers and reflections about his use of
Paper.
B wrote an article to be published in a scientific journal. Before the study, he had
already written some text that he brought into Paper to work with. The document
contained findings/results from an empirical study conducted earlier. B also used text
from an abstract written previously for a conference paper.
B wrote for a period of approximately one month. As in the first case study, the
time limits were naturally defined by the length of the writing task.
Since B had a shorter writing period the data collected in the second case study is
less extensive than in the first one (see below).
6.3 Settings
Subject A
Writing mostly took place on a PC with a 17-in. screen, located in a special room at the
Royal Institute of Technology in Stockholm (KTH). The setting and the way the task was
performed were as close as possible to what would define a real situation for this
Master’s thesis work. If A had not had access to the computer offered by the study, she
would have written her thesis on a PC in one of the computer labs that the students have
access to at KTH, and at home.
A could ask questions and come for help at any time. She was also free to write at
any hour that suited her. It turned out that she mostly wrote for an average of four
consecutive hours located around noon. However, she did not write every day of the
week and the writing sessions tended to get longer as the study proceeded.
Since Paper is a prototype, A needed to use a commercial word processor from
time to time, especially at the end of the study to insert tables and format her document in
a way suitable for presentation to the supervisor. The transfer of text between the two
processors was straightforward.
- 35 -
Subject B
The setting was the same as in the first case study. B wrote on a PC with a 19-in. screen2
located in a special room at KTH, in Stockholm. This setting was very close to what
would have been a real situation for B, when writing the article. He wrote in writing
sessions that varied from one to three hours. Sometimes, B wrote for two writing sessions
on the same day for a total of almost 4 hours.
6.4 Data collection
We collected data from different sources: a diary, interviews and spontaneous comments
by the subjects written directly in Paper. However, since B had a shorter writing period
the data collected in the second case study is less comprehensive than in the first one.
Diary The subjects were instructed to fill in a diary form at the end of a writing session.
The form, constructed as a questionnaire with free answers, was divided into four
different sections. The first three questions concerned which writing activities the subject
had been engaged in during the session, which Paper functions had been used, and the
extent to which the subject had worked with several parallel documents. The last question
invited the subject to write open comments about how they felt that the tool worked in
relation to the task. It was viewed as important to see the problems experienced by users
from their own perspective, taking into consideration the context they were working in at
different times.
Interviews Interviews were conducted on a regular basis: one introductory interview,
several intermittent ones, and a final one. The introductory interview aimed at getting to
know the subjects’ background and previous writing experience, as well as their
expectations about Paper and the study. The final interview included retrospective
questions about the task and the use of the tool. All the interviews were open-ended,
though questions were prepared in order to keep focus on matters of interest. Interviews
2 By the time of the second study, a larger screen was available.
- 36 -
were held approximately once a week and lasted about 30 minutes. They took place in
front of the subjects’ computer, and dealt with the following issues:
The writing activities as well as the status of the writing task.
The subjects’ working strategy in Paper, as they would describe it in their own
words. This turned out to be a very rich source of data, especially the subjects’ own
wording when talking about Paper and the experience of using it.
What the subjects thought was, at that time, the worst and the best things about
Paper.
Problems, special questions raised since the last time of interview e.g. about the
subjects’ problematic use of a command or the introduction of a change in the
program, etc.
During the interviews notes were taken. Shortly after the interview (usually within
the same day), the notes were transcribed into readable text material (approximately one
typed A4-page per time).
6.4.1. Summary The following data was gathered during the study:
the filled in diary forms. Altogether, A filled in 13 forms and B filled in 8 diary
forms.
the transcript of the interviews. Nine interviews were conducted with A and four with
B. the print-outs of subject A’s spontaneous comments in Paper. A wrote comments on
21 occasions. B did not write any comments directly in Paper.
The qualitative character of the diary data is predominant, and except for
tendencies in use patterns, most of the data is to be taken as A and B’s own account of
their actions and choices.
6.5 Findings
- 37 -
6.5.1. The subjects’ general perception of the Paper program Subject A:
A was positive towards Paper and particularly appreciated working with several
documents. She liked the physical overview mediated by the Paper pages and thought
they offered support for her spatial memory of the text. Presumably, in accordance with
her well-organized way of working, Paper allowed her to structure her work at a high
level of activity. She expressed it in this way: “ What’s nice with Paper, is that you can
work with several things at one time. Paper supports my way of working. […] it supports
the creativity of association of thoughts”.
A used the page presentation a lot. Occasionally, she switched to the scroll
window, mostly for cut and paste operations. A thought that she remembered the text
better in Paper and reported that she felt that she had better control over her work.
Subject B:
In general, B found Paper very useful and satisfying environment to write in, stating that
Paper made writing easier and faster or more effective.
In MSWord (or any scrolling window system), he easily felt a loss of
concentration. The page display in Paper offers ”better readability, better feeling of
achievement”. Good page size also gave less eye fatigue, according to B.
B reported, towards the end of the study, that he did not use print-outs as much as
he used to in MSWord, for instance: ”I found myself not to read so often the hardcopy of
the document during the writing process”. He did use hard copies, but mostly for reading
at home and writing down comments.
In general and because of all these advantages, B reported that Paper could possibly
“speed up the writing process”. Also, B expressed during the interviews that he would
like to use Paper the next time he was going to write a long document.
6.5.2. Writing on Paper pages Many positive comments concerned the way the pages are presented on the screen and
the visual feedback and improved overview they provide. We also found subjective
evidence for the hypothesis that page display in Paper offers better support for the
writer’s spatial memory and an improved overall mental representation of the text. A
- 38 -
reported that ”she remembers the text better in Paper” and that she ”can see what
different pages look like and learn how they look”. She also felt that she had better
control over her work, when working in Paper.
When it comes to page size, A used different page sizes for different documents:
for example, she always made her comments about Paper on medium-sized pages. B
created his own medium size, 470x499 (lines of about 60 characters) that he thought was
optimal. In the interviews, he was very eager to emphasize the importance of the page
size. The size should be “not too small, not too large” according to B, and he gave many
explanations for this: the “perfect” page size made it easier for him to concentrate on the
text, made it easier to read the text (less eye-fatigue), etc. Later on in the study, B opted
for a larger page size. Since the change of page size occurred at the same time as a major
change of strategy (e.g. splitting up the main document into three separate ones and
starting to use icons), one might think that it was because large pages were more
appropriate for B’s new way of working and organizing the screen.
6.5.3. A limited need for a scrolled text presentation One of the research questions that the study was set up to answer, is whether or
not the subjects would use the special page presentation of Paper, and if so, to what
extent and in what situations.
Both A and B did use page presentation to a great extent. According to A’s own
estimation in the final interview, she did so 50% of the time. When asked in the final
interview in what situations she had used the page mode, A answered ”most of the time.
[…] At the beginning, I used scroll quite a lot because I was irritated over the page splits.
Then I got used to it”.
B used almost exclusively the page presentation for displaying documents. He
used the scroll window mode only for cut and paste and printing. In the diary, he reported
his use of scroll with the words: “once, just for curiosity and for printing some pages”,
“once for printing”. In some sessions, he indicated no use of scroll mode at all, even
when he reported many problems with page splits.
A used a scrolled mode more often than B, at least for some time during every
writing session. She also switched between page and scroll modes more frequently.
However, it is difficult to tie the use of the presentation modes to specific writing
- 39 -
activities. It seems that A’s choice of presentation mode was influenced by the following
two factors: an attempt to optimize the use of the limited screen space and an intention to
solve practical tasks such as printing or executing cut and paste operations.
6.5.4. Optimizing screen space and getting an overview by creating personal “writing environments”
After having used Paper for several months, A started to use the expression “write in
stack”, when she referred to the way she used the Paper pages. By this expression, she
meant that she stacked her document and wanted the pages to remain within the stack, as
she was writing. Even new pages, if any, were expected to get into the stack and not to
appear to the right of the full page (see Fig. 6).
Figure 6. The way the new page appears, to the right of the stack, as it works currently in Paper.
b. The way A would like the new page to appear: in the Stack (another way of saying it would be “on top of
the Stack”).
At that point of the study, the stack was no longer just a ”cleaning up” feature: it
was rather used as a self-contained writing environment. Later on, A reported that she
used the View spread page arrangement in exactly the same way: to “write in”, as she
expressed it.
Interestingly, B did something similar when arranging a local context around the
point of work: B used a two pages display (“like a book” he commented), which he tried
to preserve. B used the “View spread (2 pages)” command to display two consecutive
pages. In the first diary form he filled in, he commented on what he appreciated most
with Paper during that particular session: “looking at two pages at the same time” and
“an easy transition between pages”. “When the text was growing, I found myself using
- 40 -
View Context (2 pages) or spreading two pages in other ways. To look at two pages side
by side seems to be the best way to get a context”.
Even though the subjects organized the screen to achieve an effective display of
pages, finding good combinations of page sizes and page types for the different
documents was time consuming and problematic, especially for A. It seems that she did
not feel that she had enough control over the location and the movements of the pages.
She expressed her frustration with the following words: “There are so many combinations
of different page sizes, different dispositions on the screen. I try to arrange a satisfactory
one, and then I start to write and, oops, at the first page split, it’s already ruined, pages
moving all around…”. However, the problems with the pages diminished towards the end
of the study, when A reported that she had found a ”new technique for handling
unwanted page moves: I fix them sooner and I am more preventive in my actions”.
6.5.5. Working with several documents - a natural strategy when writing in Paper Especially for subject A, the support in Paper for working with and handling pages in
several documents at a time on the screen came very well at hand. This is in accordance
with results from the previous study about planning, where the creation and use of several
documents for different purposes was frequent.
- 41 -
Figure 7. A Paper screen typical for A, displaying four documents: one in a scroll window, one in large
Paper pages just beside it, one document with Paper comments in medium pages and one iconized
document.
A created, right from the beginning, several different documents with contents
ranging from facts on different sludge treatment methods to rules and prescriptions for
sludge treatment disposals (see Fig.7). She appreciated having several documents ”at
hand”, which gave her the possibility of making associations and writing at the same
time. After three months of use of Paper, she thought that ”it works very well; I have
been able to get an overview of my text, and also having the document
“questionsandideas” lying just beside on the screen. That’s good, then I don’t forget
anything; instead, I write it down directly”. A thought that the possibility to shrink a
- 42 -
document to a small window to the right of the screen (“iconize” a document) was
important.
In a similar way, B ended up working with several documents to organize his
work and found iconization to be very useful. Even though B started out with one
document, by the time the document had grown to 30 pages, he splitted it in three. He
also created a document called memo.txt. That document was saved and used in later
sessions.
6.5.6. Improving strategies for handling page splits As we mentioned earlier, handling the page overflow problem is probably the most
crucial design issue when implementing the paper model. The amplitude of the disruption
caused by the page splits was related to the kind of activity that the subject was involved
in at the time of the page split. Especially note making and other cognitively demanding
activities were sensitive. For instance, A complained a lot about her outlines and bullet
lists being ruined by the frequent page splits at the beginning of the study and B reported
that page splits were irritating when they occurred at a time he was struggling hard with
the text.
However, both subjects developed working strategies that made page splits less of
a problem. A executed a “Compress text” command almost directly after each page split.
However, the use of “Compress text” became less frequent after a while and by the end
of the study A seemed much more at ease with handling page splits. This may partly have
been due to her learning to insert hard Page Breaks (make it possible to preserve the
breaks between certain pages during the execution of a Compress text command).
B described his working strategy with the following words: “I work until I feel
some kind of chaos (he cannot say exactly what the criteria for chaos are), and then I
execute either a backspace key press (local compress functionality3) or a compress text
operation. Then I have to reorient myself and get back to my two page context”. In the
last interview, B gave an explanation for this need of “reorientation”. Even if the pages
were redisplayed at the same position (we had changed the “Compress text” functionality
3 By locating the cursor at the upper left-hand corner of a page and moving it with the left arrow key to
the previous page, the text is compressed only within these two pages.
- 43 -
of Paper at that point), their position “in depth” can be changed, since the order that
makes them active when redisplaying them might be different from the order in which
they were disposed primarily. That is why B had to “reorganize” after each execution of
“Compress text”. B used a hard page break after each section.
7. DISCUSSION AND DESIGN IMPLICATIONS
The methodology used in the studies presented here reflects our design focus and a
concern to study writing from the writer’s own perspective. The data gathered in the
studies provided answers to the following research question: does an interface exist, and
in that case how could it be designed, that could adequately support a page based writing
environment? When it comes to the question of the need for such a page based writing
tool, the results from the longitudinal study were especially helpful. One of the subjects,
towards the end of the study, used exclusively Paper pages to “display as much text as
possible on the screen”. At that particular time, this writer’s need for an alternative text
presentation was substantial and the Paper pages were a satisfying solution, even if the
functionality to support them is not yet optimal.
In a general way, we have found that the complexity of writing activities, the
diversity of the writing styles and writing tasks, and the heterogeneity of the user
community, all speak in favor of a flexible writing tool. There is no doubt that page
presentation will be useful, and maybe superior to scroll presentation, in some situations
and for some writing activities. The goal is then not to replace scroll functionality
completely by paginated text presentation, but rather to enrich the writing environment
with an additional text presentation type. The ultimate question is then: how is the
functionality to support such a page based text presentation best designed?
The studies presented here provided us with many interesting answers to that
question. Moreover, to study writers during a lengthy period of time made it possible to
uncover aspects in the use of Paper that were not our primary concerns when we started
our research. For instance, what seems to be a natural working strategy for several
- 44 -
subjects, is to use several documents for different purposes when writing in Paper. We
also discovered a positive emotional effect of using fixed-sized pages to compose at the
computer – several of the subjects’ comments indicated a satisfaction with the work they
had achieved so far and an impression that the writing environment had helped them in
different ways. Furthermore, interesting aspects of the need for overview when writing
and how it is best achieved were revealed to us.
Below, we discuss some of the results from the studies and their impact on how
the paper model should be implemented to provide the users with a more usable writing
tool.
7.1 Gaining overview of the text with Paper pages
In general, we observed an extended use of the Paper pages and a positive attitude from
the users when it comes to using the pages, especially for gaining overview over the text.
However, during the longitudinal study, switching between different text presentations
was frequent, especially for subject A. Interestingly enough, the reasons mentioned for
the use of the scroll presentation were mostly related to problems with the Paper pages.
Although these problems could probably be remedied in future versions of Paper, there is
no reason in general why the scroll presentation should be completely replaced by Paper
pages. Instead, we think that the possibility of switching between different presentations
for the text adds flexibility to the writing environment. Different writers will certainly use
the different presentations for different purposes, at different times during the writing
process, and some writers may even not switch between different presentations at all.
However, the possibility should be present, with each presentation supported by adequate
functionality and the switching very easy and quick to perform.
Another interesting result is the very limited use of the spread out functions, both
in the planning study and in the longitudinal one. Generally, one could explain the limited
use of spread out commands by the fact that the screen surface is limited and that too
many pages cannot be displayed in a satisfactory way. Another explanation is that some
functions in Paper were built on the erroneous assumption that the user is performing
different activities at different times, and that one activity is well separated from another
one. For instance, the functionality supports the following Paper specific display
- 45 -
activities: cleaning up the screen (e.g. with the Stack page command), and spreading out
the pages to get a good overview (e.g. Spread horizontal). However, A (in the first case
study) was using one of these display modes (the stack) as an environment to write in.
This could mean that the users are constantly building up their mental representation of
the text and that, only very seldom, special actions are devoted to “just” getting an
overview. This suggests that the need for a global perspective is an integrated part of
writing processes, situated in the moment-to-moment text creation and reviewing
contexts, and cannot be confined to a special stage or an isolated function of a writing
tool. Paper offers a writing environment where the writer is able to create her/his own
arrangement of pages that presumably will provide her/him with a good overview and
“sense” of the text.
7.2 Working with several documents
In both studies we were able to observe that subjects appreciated the possibility of
working with several documents in the same task. For instance, in the planning study,
only one subject out of four worked with only one document during the writing sessions.
For the other three subjects, working with pages from several documents appears to be an
appreciated strategy when it comes to planning a writing task. In the longitudinal study,
subject B started to work with only one document. This could be explained by the fact
that he already had a document to start working with in Paper, and did not think of
splitting it up into several ones. However, after writing for a while in Paper, he
discovered the possibility of working with several documents. Moreover, subject A
created many documents during the writing sessions: at most, she had 15-20 documents
to work with!
However, having several Paper documents opened means a great amount of pages
on the screen. The pages in one document have to keep together, in what seems to be
quite a rigid way according to the case studies, or the screen gets too messy (see above).
The icons seem to be a very powerful feature in that respect, since they gather all the
pages in one document very quickly and effectively. In fact, B reported that he stopped
using any page handling commands once he discovered the icons.
- 46 -
To make it even easier to work with several documents in parallel in Paper,
several improvements to the functionality could be envisaged:
the possibility of opening several documents from file directly with the Open
command.
a way of synchronizing flipping pages in different documents
fixed places on the screen for the different documents, maybe a in the form of a
Home command for each document.
changing the relative order of the pages in one document.
exchanging pages between documents
better visual feedback/cues to separate documents, e.g. with different colours. This
could lessen the risk for focusing problems and help the writer keep track of the
different documents.
a command of the type ”Find everywhere”, to search through all open documents on
the screen.
7.3 Visual orientation and feeling of control
As discussed earlier, problems with too many pages on the screen is one of the
consequences of the paper model. During the studies, the subjects reported problems with
controlling the pages’ movements and location. One subject in the planning study
complained very much about the way the pages moved/appeared on the screen. She felt a
lack of control over the page location. Another subject, however, asked for a more
“manual” way of handling the pages. In the longitudinal study, A spent significant effort
into rearranging pages in ways that suited her. B abandoned the View spread command
and started to arrange the pages by hand instead, presumably because he felt it was easier
to display the pages exactly where he wanted. This is also in accordance with results from
the early formative evaluation (Fatton, 1992) where subjects seemed to prefer a direct
manipulative way of handling the pages rather than using the commands created for that
purpose. In a future version of Paper, a better trade-off between the user’s possibility of
moving the pages “by hand” on the screen and the degree of system control should be
implemented. However, it is crucial that it should be clear to the user, at any time, where
- 47 -
a page is going to appear, and where a certain page is. Moreover, a page should remain in
its position unless the user does something deliberately with the page. Also direct visual
cues about the active page and the location of the cursor on that page are needed. During
the study, we made the following improvements to reduce focus related problems.
better indication of the active page and cursor’s location with a new color (light blue)
for the selections, one that is different from the color in the title bar, indicating the
active page.
the selections on the pages not active are removed (all the pages used to have one
selection).
All these visual cues for orientation are very important in Paper and should be
improved in order to make its visual properties even more useful to the user.
7.3.1. A feeling of control It is interesting that the subjects, while experiencing problems with controlling the pages’
location on the screen, commented on Paper providing them with “better control of their
work” or a “feeling of achievement”. One subject (B in the longitudinal study) even
thought that Paper made the writing process easier. Another subject, in the formative
evaluation, reported that filling pages with text and putting them aside made her feel
satisfied with her work. This suggests that their feeling of control corresponded to a
higher level of organization of thoughts and to general writing strategies. We believe that
it might also be related to the fact that the subjects were able to build up a satisfactory
mental representation of the text which is an important part of a successful writing
process.
Another possible explanation is that Paper conveys to the writer a feeling of
writing on “real” pages of paper. The page is a central concept/object in the paper model.
While writing in a traditional word processor can make writers feel they are ”deprived of
the reassuring support of paper” or ”worried about the absence of paper” (Chandler,
1994), any page based writing tool will reduce that feeling by explicitly introducing
pages to be written on by the users. Writing is a cognitively demanding activity, and any
ways in which writing can be made easier for the users are precious.
- 48 -
7.4 How to improve the page split functionality
As expected, one of the most intricate and crucial design issues revealed by both studies
is the one of how to handle the page overflow problem. The page split solution
implemented in the current Paper version was observed to cause many blank areas,
which sometimes resulted in poor local context around the point of work.
One possible solution, or at least improvement, to better preserve the local
context, is to implement a local scroll functionality. The idea is that the insertion would
cause the text to flow over to a new page, preserving the local context and the text
location on remaining pages in the document (see Fig. 8.a and Fig. 8.b).
blank
blank
Figure 8.a. Page split as it works in Paper now.
blank
Fig 8.b. Page overflow handling by “local scroll” functionality.
Beside the advantages described above, we anticipate the local page split solution
to bring the following positive consequences to the program:
• less blank areas
- 49 -
• less new pages, leaving more space on the screen and making it easier for the
subject to keep order among the pages
It is not certain that a local scroll would be the best alternative for all users, in all
writing situations. According to our experiences about the subjects’ change of strategy in
handling page breaks during the studies, it is imaginable that the possibility to choose
between different page split strategies for different writing activities would be the optimal
design solution. Notwithstanding, empirical studies of this solution would be needed,
since the consequences of this alternative solution to the problem of page overflow
cannot be easily predicted.
7.4.1. Better support for “repairing the damages” after a page split An example of secondary functionality needed in Paper because of the occurrence of
page splits is the Compress text command. This functionality was improved during the
longitudinal study to help the subjects “repair” the damages caused by the page splits.
The new command is called a ”local compress”: by placing the cursor at the first position
on the new page and pressing the Backspace key, the text on the new page would ”go
back” onto the first one, restituting the local context and leaving only the last part of the
text that did not fit on the first page on the second page (see Fig. 9.a and Fig. 9.b). Also,
the rest of the pages in the document would not be affected by the ”local compress”
command.
Figure 9.a. A Local compress operation. After a page split has occurred, the cursor is placed at the
beginning of the new page (position to which the arrow points). By pressing the backspace key, the text is
“moved back” to the page of insertion.
- 50 -
Figure 9.b. The result of the local compress. The local context is restored and the cursor is back at the
working point (position to which the arrow points).
8. GENERAL CONCLUSIONS
The results from the two studies presented here speak in favor of the paper model as an
alternative form of text presentation. They show that a page-based text presentation was
well appreciated by the subjects, even though they had to overcome some problems
related to page splits. Page split was confirmed to be the most intricate design issue when
implementing the model, but we believe that the design of the page split functionality
could be improved. On the other hand, it seems that writers might accommodate with the
page split strategy of the current version of Paper.
We found subjective evidence that the paper model improves the overview of the text and
supports the writers’ spatial memory when implemented in a writing tool. Moreover,
insights about how the subjects organized their work in order to build up an adequate
mental representation of the text, as an integrated part of the writing process, gave
interesting input into the further design of writing tools. The studies also revealed that the
possibility of working with pages from several documents is a strength of the paper
model, even though it was not our primary concern when we started our research work.
- 51 -
Future work with the paper model should include different kinds of studies, both
of improved versions of Paper and also about the ways in which the introduction of the
model in computer based writing might influence writing strategies. Studying the use of
the paper model might also reveal further interesting aspects of the need for overview
when writing at the computer and of different ways of achieving a global perspective of
the text. Finally, to investigate further the emotional aspects of writing in Paper is a
challenging research endeavour that might share further light on the complex relationship
between the writers and their tools.
- 52 -
REFERENCES
Allwood, C. M., (1991). Människa-datorinteraktion, ett psykologiskt perspektiv.
Studentlitteratur, Lund. [in Swedish]
Bereiter, C. and Scardamalia, M. (1987). The psychology of written composition.
Hillsdale, NJ: L. Erlbaum.
Booth, P. (1990). An introduction to human-computer interaction. Lawrence Erlbaum
Associates, East Sussex.
Brandt, D. (1992). The cognitive as the social. Written Communication, 9, 315-355.
Carroll, J. M. and Mack, R.L. (1985). Metaphor, computing systems, and active learning.
Int. J. Man-Machine Studies, 22, 39-57.
Chandler, D. (1994). Who needs suspended inscription? Computers and Composition, 11,
191-201
Chin, E. (1994). Ethnographic interviews and writing research: a critical examination of
the methodology. In P. Smagorinsky (ed.) Speaking About Writing. Sage Series in
Written Communication, Volume 8, pp. 247-272.
Daiute, C. (1983). The computer as stylus and audience. College Composition and
Communication, 34, 134-145.
Fatton, A. (1992). Pappersmodellen för datorstött skrivande: Implementation av
användargränssnitt och överblicksfunktioner i SmallTalk. Technical report TRITA-NA-
P9221, Department of Numerical Analysis and Computing Science, Royal Institute of
Technology (KTH). [in Swedish]
- 53 -
Fatton, A. and Severinson Eklundh, K. (1996). How to support in-process planning in a
computer-based writing environment. European writing conferences, Barcelona,
Universitat de Barcelona.
Flower, L. (1989). Cognition, context, and theory building. College Composition and
Communication, 40, 282-311.
Flower, L.S. and Hayes, J, R. (1980). The cognition of discovery: Defining a rethorical
problem. College Composition and Communication., 31, 21-32.
Flower, L.S. and Hayes, J.R. (1981). A cognitive process theory of writing. College
Composition and Communication, 32, 365-387.
Haas, C. (1989a). How the writing medium shapes the writing process: effects of word
processing on planning. Research in the Teaching of English, 23 (2), 181-207.
Haas, C. (1989b). Does the medium make a difference? Two studies of writing with pen
and paper and with computers. Human-Computer Interaction, 4, 149-169.
Haas, C. (1996). Writing technology: Studies on the materiality of literacy. Lawrence
Erlbaum Associates, New Jersey.
Halasz, F. G., Moran, T. P. and Trigg, R. H. (1987). NoteCards in a nutshell. In
Proceedings from the CHI + GI Conference (Human Factors in Computing Systems and
Graphics Interfaces, pp. 45-52. New York: ACM Press.
Hansen, W. J. and Haas, C. (1988). Reading and writing with computers: a framework for
explaining differences in performance. Communication of the ACM, 31, 1080-1089.
- 54 -
Hayes, J. (1996). A new framework for understanding cognition and affect in writing. In
Levy, C.M. and Ransdell, S., The Science of Writing, pp. 1-27. Lawrence Erlbaum
Associates, New Jersey.
Hill, C. A., Wallace, D. L. and Haas, C. (1991). Revising On-Line: Computer
Technologies and the Revising Process. Computer and Composition, 9, no. 1, 83-109.
Hovmöller, M. (1995). Planeringsverktyg för skrivprocessen. Technical report No.
IPLab-89. Department of Numerical Analysis and Computing Science, Royal Institute of
Technology (KTH). [in Swedish]
Kellogg, R.T. (1989). Idea processors: Computer aids for planning and composing text.
In B.K. Britton and Glynn, S.M. (Eds.), Computer Writing Environments: Theory,
Research, and Design (pp. 57-92). Hillsdale, NJ: L. Erlbaum.
Lansman, M., Smith, J. B. and Weber, I. (1993). Using the Writing Environment to study
writer’s strategies. Computer and Compositions, 10(2), 71-92.
Lewis, S. (1995). The art and science of Smalltalk. Prentice Hall; ISBN: 0133713458.
Lovelace, E. A. and Southall, S. D. (1983). Memory for words in prose and their
locations on the page. Memory and Cognition, 11, 429-434.
Lutz, J.A. (1987). A study of professional and experienced writers revising and editing at
the computer and with pen and paper. Research in the Teaching of English, 21(4), 398-
420.
Mynatt, B. T. and Macfarlane, K. N. (1987). Advanced organizers in computer
instruction manuals: Are they effective? Human-Computer Interaction – INTERACT ´87
(pp. 917-921). Amsterdam: Elsevier Science Publishers.
- 55 -
Neale, D. C. and Carroll, J. M. (1997). The role of metaphors in user interface design. In
Helander, M., Landauer, T. K. and Prabhu, P. (eds.). Handbook of Human-Computer
Interaction. Elsevier Science B.V.
Norman, D. A. (1986). Cognitive engineering. In Norman, D.A and Draper, S. W., User
Centered System Design. Hillsdale, NJ: L. Erlbaum.
Nystrand, M. (1989). A social-interactive model of writing. Written Communication, 6,
66-85.
O’Hara, K., and Sellen A. (1997). A comparison of reading paper and on-line documents.
In Proceedings of CHI ’97, Conference on Human Factors in Computing Systems (pp.
335-342). New York: ACM.
Owston, R.D., Murphy S. and Wideman H.H. (1992). The effects of word processing on
students’ writing quality and revision strategies. Research in the Teaching of English,
26(3), 249-276.
Patton, M. Q. (1987). How to use qualitative methods en evaluation. Thousand Oaks,
California: Sage Publications.
Piolat, A. , Roussey, J.Y. and Thunin O. (1997). Effects of screen presentation on text
reading and revising. Int. J. Human-Computer Studies, 47, 565-589.
Rothkopf, E.Z. (1971). Incidental memory and the recall of information in text. Journal
of Verbal Learning and Verbal behaviour, 10, 608-613.
Sasse, M. A. (1997). Eliciting and Describing User’s Models of Computer Systems.
Unpublished doctoral thesis, School of Computer Science, The University of
Burmingham, B15 2TT, England.
- 56 -
Sellen, A. and Harper H. R. (2002). The myth of the paperless office. The MIT Press,
Cambridge, Massachusetts.
Shneiderman, B. (1997). Designing information-abundant web sites: issues and
recommendations. Int. J. Human-Computer Studies, 47, 5-29.
Severinson Eklundh, K. (1992). Problems in achieving a global perspective of the text in
computer-based writing. Instructional Science, 21, 73-84.
Severinson Eklundh, K., Fatton, A. and Romberger S. (1996). The paper model for
computer-based writing. In Cognitive Aspects of Electronic Text Processing. Herre van
Oostendorp and Sjaak de Mul, Editors. Ablex Publishing Corporation. Norwood, New
Jersey.
Severinson Eklundh, K. and Sjöholm, C. (1991). Writing with a computer. A longitudinal
survey of writers of technical documents. International Journal of Man-Machine Studies,
35, 723-749.
Smith, J. B. and Lansman, M. (1989). A cognitive basis for a computer writing
environment. In B.K. Britton and Glynn, S.M. (Eds.), Computer Writing Environments:
Theory, Research, and Design (pp. 17-56). Hillsdale, NJ: L. Erlbaum. van Waes, L. (1992). The influence of the computer on writing profiles. In H.P. Maat and
M. Steehouder (Eds.), Studies of functional text quality, Utrecht Studies in Language and
Communication 1 (pp. 173-186). Amsterdam: Rodopi Publishers.
Van Waes. L. & Schellens. P.J. (2003). - Writing profiles: the effect of the writing mode
on pausing and revision patterns of experienced writers. Journal of Pragmatics, 35, p.
829-853
- 57 -
van der Geest, T. (1996). Studying ”real life” writing processes: a proposal and an
example. In Levy, C.M. and Ransdell, S. The Science of Writing. Lawrence Erbaum
Associates, New Jersey. Waern, Y. (1989). Cognitive aspects of computer supported tasks. John Wiley and Sons,
New York. Wozny, L. A. (1989). The application of metaphor, analogy, and conceptual models in
computer systems. Interacting with Computers, 1:3, 273-283.