Master’s Thesis

Adaptive Rapid Serial Visual Presentation

Gustav Öquist
[email protected]

Language Engineering Programme

Department of Linguistics
Uppsala University
December 2001

Supervisors: Torbjörn Lager, Uppsala University

Mikael Goldstein, Ericsson Research
 i

Abstract

Can readability on small screens be improved by using Rapid Serial Visual

Presentation (RSVP) that adapts the presentation speed to characteristics found in
the text? In this thesis Adaptive RSVP is introduced together with two algorithms for
adaptation. Findings from a usability evaluation, where the ability to read long and
short texts on a mobile device in a realistic setting was assessed, are reported. In a
balanced repeated-measurement experiment employing 16 subjects two variants of
Adaptive RSVP were benchmarked against Fixed RSVP and traditional text
presentation. No significant differences were found for reading speed and
comprehension for long texts but for short texts all RSVP formats increased reading
speed by 33%. For long texts Adaptive RSVP decreased task load compared to Fixed
RSVP and repeated use of RSVP, regardless of type, was also found to decrease task
load. No differences in task load were however found for short texts. Causes,
implications and effects of these findings are discussed together with directions for
further work.
 ii

Acknowledgements

The author would foremost like to thank Mikael Goldstein for inspiration,
supervision and a lot of support during the work with this thesis. Many thanks also to
Peter Ljungstrand and Staffan Björk at the Interactive Institute for valuable
collaboration. The author would also like to express gratitude to all colleagues at
Ericsson Research’s Usability & Interaction Lab in Stockholm and Sony Ericsson
Mobile Communications in Lund. Much credit is also awarded all of those who
participated in the usability evaluation. Last but not least, the author would like to
put across many thanks to family and friends for love and support.
 iii

Table of Contents

Abstract………………………………………………………………………….. i
Acknowledgements……………………………………………………………… iv
Table of Contents ……………………………………………………………….. iii
List of Figures ………………………………………………………………….. v
List of Tables ………………………………………………………………….… v
List of Equations..………………………………………………………………... v

1 Introduction ........................................................................................................1
1.1 The Assignment ............................................................................................2
1.2 Outline of this Thesis ....................................................................................2
2 Background.........................................................................................................3
2.1 The Reading Process.....................................................................................3
2.1.1 The Revelation of Eye Movements .......................................................4
2.1.2 Physiology of the Eye............................................................................5
2.1.3 Eye and Mind.........................................................................................6
2.2 The Concept of Readability ..........................................................................7
2.2.1 Readability Ratings................................................................................8
2.2.2 Readability Measures.............................................................................8
2.3 Screen Reading .............................................................................................9
2.3.1 Large Screen Readability.......................................................................9
2.3.2 Small Screen Readability.....................................................................10
2.4 Text Presentation on Small Screens............................................................10
2.4.1 Traditional Text Presentation...............................................................11
2.4.2 Dynamic Text Presentation..................................................................11
2.5 Rapid Serial Visual Presentation.................................................................12
2.5.1 Previous RSVP Evaluations.................................................................12
2.5.2 The Reader Evaluation.........................................................................13
3 Design ................................................................................................................15
3.1 Fixed RSVP ................................................................................................15
3.2 Adaptive RSVP...........................................................................................16
3.2.1 Content Adaptation..............................................................................17
3.2.2 Context Adaptation..............................................................................18
3.3 Distribution Model......................................................................................19
4 Implementation.................................................................................................20
4.1 Deployment.................................................................................................20
4.1.1 Program structure.................................................................................21
4.2 Graphical User Interface .............................................................................21
4.2.1 Bailando walkthrough..........................................................................21
4.3 RSVP Characteristics..................................................................................24
4.4 Sonification Features...................................................................................24
4.5 Log Functionality........................................................................................24
5 Evaluation .........................................................................................................25
5.1 The Usability Evaluation ............................................................................25
 iv

5.1.1 Method.................................................................................................25
5.1.2 Design ..................................................................................................25
5.1.3 Subjects................................................................................................25
5.1.4 Apparatus.............................................................................................26
5.1.5 Texts ....................................................................................................26
5.1.6 Setting..................................................................................................27
5.1.7 Instructions ..........................................................................................27
5.1.8 Training................................................................................................28
5.1.9 Procedure .............................................................................................28
5.1.10 Inventories ...........................................................................................28
5.1.11 Caveats.................................................................................................28
5.2 Results.........................................................................................................29
5.2.1 Reading Speed .....................................................................................29
5.2.2 Comprehension ....................................................................................30
5.2.3 Task Load ............................................................................................31
5.2.4 Attitude ................................................................................................33
6 Discussion..........................................................................................................35
6.1 Reading Speed ............................................................................................35
6.2 Comprehension ...........................................................................................35
6.3 Task Load....................................................................................................35
6.4 Attitude .......................................................................................................36
6.5 Interface Issues Affecting Readability........................................................36
7 Conclusions .......................................................................................................38
7.1 Future work.................................................................................................38
References ................................................................................................................39
Web resources..........................................................................................................42
Appendix ..................................................................................................................43
A Excerpt from a Bailando log file ....................................................................43
B Instructions .....................................................................................................44
B1 Microsoft Reader:....................................................................................45
B2 Microsoft Internet Explorer.....................................................................46
B3 Bailando ..................................................................................................47
C Inventories ......................................................................................................48
C1 Comprehension........................................................................................48
C2 NASA-TLX (Task Load Index) ..............................................................49
C3 Attittude...................................................................................................50
 v

List of Figures

1. Cross-section of the eye.……………………………………………….. 5

2. The Reader version 1.0 prototype.…………………………………...…13
3. Fixed RSVP speed-exposure plot…………………………………….... 15
4. Content adaptive speed-exposure plot…………………………………. 17
5. Context adaptive speed-exposure plot…………………………………. 18
6. Excerpt from an XML tagged RSVP file……………………………… 19
7. Bailando running in reading mode on a Compaq iPAQ 3630………..... 20
8. Bailando start screen and library menu………………………………... 21
9. Bailando main screen in reading mode and pause mode…………….... 22
10. Bailando text information screen and settings menu………………….. 23
11. Interface of the Microsoft Reader and the Internet Explorer……….…. 26
12. Interior of the usability lab…………………………………………….. 27
13. Task load ratings for long texts………………………………...……….31
14. Task load ratings for short texts………………………………………...32
15. Task load ratings for repeated use of RSVP………………………….. 32
16. Attitude ratings for long texts………………………………………….. 33
17. Attitude ratings for short texts…………………………………………. 34
18. Attitude ratings for repeated use of RSVP……………………….……. 34

List of Tables

1. LIX value interpretations……………………………………………...… 8

2. Variables used by Bailando for RSVP……………..………………….. 24
3. Texts used in the evaluation…………………………………………… 27
4. Reading speeds for long texts………………………………………….. 29
5. Reading speeds for short texts…………………………………………. 29
6. Comprehension scores for long texts………………………………….. 30
7. Comprehension scores for short texts…………………………………. 30

List of Equations

1. LIX value formula………..……………………………………………... 8

2. Fixed RSVP formula.………………………………………………..… 15
3. Content adaptive formula……………………………………………… 17
4. Context adaptive frequency formula…………………………………... 18
5. Context adaptive sentence length formula……………………………...19
 1

1 Introduction

The Mobile Internet has been widely predicted to cause a revolution in

communications over the next few years and when the time is ripe it is likely to
become an integral part of our everyday lives. However, before the revolution can
start there must be devices, services and applications available that people really
want to use. Enabling these new technologies, bringing them to market and making
desirable content available to them is the challenge facing companies involved in the
wireless business today.
Presently the limited screen space on most mobile devices is a bottleneck for
efficient and usable retrieval of information (Ericsson et al. 2001). Since it is the
customers demand for small devices that sets the limit this constraint is also likely to
remain tomorrow. Combined with the expected increase in texts to be read on mobile
devices due to the advent of the mobile Internet this quandary has made the issues
concerning readability on small screens progressively more important.
Early research on screen reading showed that reading speed decreased by 20-30%
when reading on large screens compared to reading on paper (Mills and Weldon
1987). With time, as screen resolution improved and people got more used to them,
readability on large screens became more or less equal to paper (Muter and Mauretto
1991). The evolution in readability on small screens is however not likely to follow
the same pattern. The resolution will surely get better and thus improve the texts
legibility but decreased readability will still be intrinsic to limited screen space
(Duchnicky and Kolers 1983). Readability may however still be increased by
designing interfaces that display the text in a way more suitable for small screens.
Focus has thus shifted from how screens display static texts to how texts can be
dynamically displayed on the screen. Leading and Rapid Serial Visual Presentation
(RSVP) are the two major techniques that has been proposed for dynamic text
presentation (Mills and Weldon 1987). Leading, or the Times Square Format, scrolls
the text on one line horizontally across the screen whereas RSVP presents the text as
chunks of words or characters in rapid succession at a single visual location. Both
formats offer a way of reading texts on a very limited screen space (Mills and
Weldon 1987; Kang and Muter 1989; Joula et al. 1995; Rahman and Muter 1999; de
Bruijn and Spence 2000; McCrickard et al. 2001).
In this thesis a potential enhancement of the RSVP format is described. The text
output is supposed to become more predictive to the reader by letting the
presentation adapt to characteristics found in the text. It is important to explore the
possibilities of dynamic text presentation since improved readability on small
screens also means improved usability of mobile devices.

"There's nothing more exciting than what software can do to help

users and we see that what we've done so far is really just
scratching the surface.“

- Bill Gates, Microsoft, MVP Summit, November 30, 2001.

 2

1.1 The Assignment

The aim with the thesis work was to twofold. Primarily it was to learn whether the
RSVP format could be improved by letting the text presentation speed adapt to
characteristics found in the text. Secondly, it was to see if the RSVP format could
improve readability on mobile devices. In order to reach these objectives a few
subsequent tasks had to be performed.
To start with the adaptive RSVP format had to be developed, it already existed as an
idea but it had to be put into practice. After that adaptation had to be integrated into
an application for RSVP on a handheld device. Finally, the prototype had to be
benchmarked against other presentation formats in a usability evaluation. The work
with the assignment can thus be roughly divided into the following three tasks:

Design Find out how adaptation could and should work on basis
of previous studies and evaluations.
Implement Construct and test an application for reading texts on a
handheld device that incorporates adaptive RSVP
features.
Evaluate Assess the ability to read using RSVP and other text
presentation formats by benchmarking the prototype
against other applications in a usability evaluation.

1.2 Outline of this Thesis

Substantial parts of this thesis were originally written to be included in three articles
reporting progress and results from the work with this thesis (Goldstein et al. 2001;
Goldstein, Öquist and Björk 2001; Öquist and Goldstein 2001). The thesis will keep
the inherent structure of the work done during the assignment, with a few additions.

Background This thesis is all about reading text on small screens and
this part will give an overview of the essentials about
reading and readability.
Design The RSVP and adaptive RSVP formats are formally put
down in this part together with a novel way of
visualizing this form of dynamic text presentations.
Implementation The implementation of the prototype is presented in this
part together with screen shots, design decisions and
technical specifications.
Evaluation In this part the method and design of the usability
evaluation is described together with an analysis of the
obtained data.
Discussion On basis of the results from the evaluation the feasibility
of using adaptive RSVP for reading on small screens is
discussed in this part.
Conclusions This part summarises the thesis and put forward some
insights reached on basis of the experimental results
together with some directions for future work.
 3

2 Background

By reading we examine and grasp the meaning of painted, written or printed

symbols. As far as we know, pictures were drawn to represent animals and other
objects as early as the Stone Age, around 20,000 B.C. The remnants of the oldest
systematic reading and writing systems are dated back to North Babylonia, around
8,000 years ago. Alphabet signs, as we now interpret them, were used in Egypt at
least 7,000 years ago (Huey 1908; Hill 1999). Reading can essentially be seen as a
very intricate form of pattern recognition that has evolved within mankind for
centuries. As with pattern recognition in general, the interpretation is highly
dependant on convention. The development of language and writing is very much a
result of agreeing on what we choose to associate with certain patterns and how we
present them.
The traditional page format for presentation of text has been fashioned over
thousands of years into its current form. Yet, this thesis puts forward a novel and
maybe also unconventional form of text presentation. One question then
immediately arises. Why replace something that works perfectly well? The answer to
this is that new appliances add new constraints. Today there is an unprecedented
need for reading texts on small screens. It is quite obvious that traditional text
presentation in the page format requires far more screen space than the most mobile
devices have to offer. This is the foremost reason for the development of the
dynamic text presentation formats presented in this thesis. If traditional text
presentation does not work properly then it might be possible to improve readability
by trying some new approaches. However, also new presentation formats must
adhere to the principles for reading that has evolved over time. The background part
of this thesis begins in section 2.1 with a primer on some of the most important
issues regarding the reading process.
The first objective of this thesis was to see if adaptation could improve readability. If
we are to say that one text presentation format is better than another, or that one text
is easier to read than another, we need to be able to measure readability. In section
2.2 the concept of readability is defined and a few different measures of readability
are introduced. The second objective was to learn whether the ability to read on
small screens could be improved by using adaptive RSVP. It has already been
implied that there is problem with readability on small screens. Section 2.3 focuses
on how screen reading and limitations in screen space affects readability. In section
2.4 a few different text presentation formats that can be used for small screens are
presented and the background part of this thesis is concluded in section 2.5 by a
description of previous RSVP evaluations and implementations.

2.1 The Reading Process

It is important to acknowledge that the key to reading, and language in general, lies
deeply embedded within in our mind (Taylor and Taylor 1983; Hill 1999). There has
been a lot of research on how we read and researchers have presented several
different models of how the reading process works. Some are very detailed whereas
others are more generalizing (Reichle et al. 2000). Most agree that we recognize
patterns and then mentally process them in some way. Unfortunately, little is known
for sure about the processing part and there are many disputes around these issues.
However, since we do not known much about the processing, a better starting point
for understanding how we read might be in the other end. By observing how the eyes
 4

move while reading we can tell how the recognition part works, if we do that we
might also learn some about how the processing works as well.
The first evidence of what really happens with the eyes while reading was
discovered in the end of 19th century. An overview of the revelation of the reading
process follows next. The structure of it is partially adopted from an excellent review
over influential eye movement studies by Paulson and Goodman (2000).

2.1.1 The Revelation of Eye Movements

In 1879, the French oculist Emile Javal found that the eye does not sweep smoothly
along the line while reading. Instead it proceeds by making short jumps called
saccades. Between the saccades the eye stays put for a brief time in pauses called
fixations. Upon the end of a line the eye moves to the beginning of the next line in a
single movement called a return sweep (reported in Huey 1908 referred to in
Paulson and Goodman 2000). The findings of Javal are what triggered the following
decades of research in eye movements while reading.
In 1891, Landholt, one of Javal´s contemporaries, discovered that “reading of a
foreign language required more pauses, as did also the reading of detached words,
numbers and list of proper names” (reported in Huey 1908:19 referred to in Paulson
and Goodman 2000). Landholt´s findings are probably the first evidence of that the
reading process is not constant, but varying depending on the type of text being read.
The Landholt study was also important since it was the first to imply that eye
movement studies might tell us something about the cognitive processing that takes
place while reading.
In 1900, Dodge reported that the eye does not retrieve any information at all while
moving. The experiment had a very simple design, but nonetheless the proof was
very convincing. Dodge used a cardboard piece with a slit of 4 mm and behind it he
put different coloured cards. The subjects were told to fixate on one side of the slit
and then move to the other side in one unbroken eye movement. Dodge found that
the subjects could not tell what colour that had been exposed or even if there had
been no slit at all (Dodge 1900 referred to in Paulson and Goodman 2000). The
experiment showed that it is during the fixations that the reader actually processes
information.
In 1908, Edmund Burke Huey provided the first physical records of eye movements
while reading. In the records he made a number of interesting findings. He found
that the eye sometimes moves backwards to reread words and phrases in movements
called regressions. He also found that only 20-70% of the words in a line are fixated.
Further, the first fixation on a line was not found to be on the first but rather on the
second or third word (Huey 1908 referred to in Paulson and Goodman 2000). These
findings provided the first evidence of a reading process where the reader chooses
where and when to fixate next. The results reached by Huey might also have been
the first indication of that reading is not just a simple word identification process, but
rather a process where words are processed simultaneously as chunks.
In 1922, Judd and Buswell reported results from the first study where the subject’s
eye movements had been photographed. The detail level was high and the plates
provided accurate records of eye movements and fixation durations. In the data Judd
and Buswell found evidence for that readers read differently under different
circumstances. They also concluded that reading is not simply a matter of bottom-up
word identification but rather a perceptual process that involves interpretations on
the reader's part (Judd and Buswell 1922 referred to in Paulson and Goodman
2000).
 5

From the early findings we know that a reader cannot perceive anything while
moving the eyes and does not fixate on each word. We also suspect that some kind
of processing takes place during the fixations. The perceptual span of each fixation
is thus decisive to how much information that can be processed at a time.
Physiological studies of the eye offer some useful answers on the physical
constraints for the reading process.

2.1.2 Physiology of the Eye

The receptive part of the eye, called the retina, is essentially a panel full of
photosensitive receptors located on the back of the eyeball (Ø ~42 mm) (figure 1).
The retina has two types of receptors, cones and rods. Cones register luminosity and
colours whereas rods register light changes. Rods are much more sensitive to light
but they cannot detect colors and are also slower to respond. Most of the cones are
located in a tiny area at the centre of the retina called the fovea (Ø ~0,2 mm). The
fovea is surrounded by the parafovea (Ø ~3 mm); in this region there are still many
cones, but also an increasing amount of rods. Outside the parafovea there are few
cones and a decreasing amount of rods, therefore vision becomes progressively less
clear in the periphery of the retina (Procter and Procter 1997).

Figure 1. Cross-section of the eye.

When reading a text, the image of the text is inversely reflected upon the retina. The
retina has a 240-degree field of vision but the maximum resolution is restricted to the
fovea. The fixation target must be located in the fovea since a high concentration of
cones is required for accurate recognition. The foveal field of vision is only one or
two degrees wide and this means that only six to eight characters can be in focus at a
time. The parafoveal region further extends the perceptual span to approximately 20
characters but beyond that acuicity is too low for retrieval. The perceptual span is
centered to the right of the fixation point, at least for readers of left-to-right
languages (Just and Carpenter 1980). Readers pick up information from
approximately eight or nine character spaces to the right of a fixation, and four or so
to the left (Rayner and Pollatsek 1989; Robeck and Wallace 1990; Rayner and
Serano 1994; Rayner 1998).
After information is processed in a fixation, peripheral vision is used to determine
the location of the next fixation. A saccade, or a return sweep is executed to move to
the next fixation target, it must not necessarily be a forward movement. Regressions
are essentially backward saccades that are used for clarification of incomplete
retrieval and they appear about 19% of the time (Just and Carpenter 1980). The
length of a saccade is usually between 1-20 characters and they are performed very
 6

quickly ~40 ms. Fixations take a little longer time, ~230 ms for fast readers and
~330 ms for average readers (Robeck and Wallace 1990). The duration of the
fixations has also been found to vary a lot. In some studies it has ranged between
100-500 ms (Rayner 1998) whereas it in others has been found to vary between as
much as 50-1500 ms (Just and Carpenter 1980).
The planning of saccades and the use of regressions for clarification seems to
indicate that there is more to reading than meets the eye. The large differences
observed in saccade lengths and fixation durations appear to reflect an ongoing
process that changes depending on what is being read. The knowledge of what is
know about the physiology of the eyes and their movements while reading seems to
suggest that perception and recognition is highly dependent on cognitive (i.e.
linguistic) processing.

2.1.3 Eye and Mind

Landholt and others were probably quite early to assume a strong bond between the
eye and the mind. Unfortunately, research in the field of eye movements and reading
came to an abrupt halt around the 1930’s. Nothing significantly actually happened
until the late 1970’s. The reason for the renewed interest was probably that
computers then became available. Eye-tracking equipment usually generates huge
amounts of data and without computers it is time consuming to analyse the results.
Another probable cause for the renewed research efforts might have been the
appealing prospect of creating computational models of the reading process. In any
case, a few models of the reading process have at least evolved since then. They can
be roughly divided into the following two categories: the oculumotor and the
processing driven models (Reichle et al. 2000).
The oculumotor models mostly look at the visual properties of the text (i.e. word
lengths) and the physiological limits of the eye (i.e. perceptual span and saccade
lengths) in order to determine the location and duration of fixations (Reichle et al.
2000). Oculumotor modelling has been successfully used to predict eye movements,
but the models can never explain the whole reading process since they ignore the
fact that language evidently has an impact on reading. The processing models on the
other hand assign linguistic processing a very central role. The general assumption
of these models is that the fixation duration is directly related to the cognitive
processing whereas the fixation targets are determined by a combination of
linguistic, orthographic and oculumotor factors (Reichle et al. 2000). Since this
thesis is mostly concerned with fixation durations the processing models had has
most impact on the work, and particularly the work of Just and Carpenter.
In 1980, Just and Carpenter suggested that, “a reader can take in information at a
pace that matches the internal comprehension process” (Just and Carpenter
1980:329). From this starting point they developed the most widely known
processing model of reading. They began by observing actual gaze durations, the
sum of all fixations on a word before moving to the next, made by college students
reading scientific passages of text. Just and Carpenter found large variations in the
duration of individual fixations as well as the duration of fixations on individual
words. They also found that almost each content word was fixated and that fixation
times were longer on words that were infrequent, thematically important or
clarifying the interpretation of previous words. The gaze durations were also found
to be longer at the end of a sentence thus indicating integrative processing.
From these findings Just and Carpenter developed a reading model based on two
assumptions. The first was the immediacy hypothesis, which state that each word is
 7

immediately processed when it is fixated. The second assumption is the eye-mind

hypothesis, which state that the eyes remain fixated on a word as long as it is
processed (Just and Carpenter 1980). Both assumptions have later been criticized
mainly because they don’t account for context and preview effects, i.e. that other
words than the ones fixated can be predicted out of the context, or perceived in the
parafovea, and therefore also affect the processing times (Paulson and Goodman
2000; Reichle et al. 2000). The resulting reading model presented by Just and
Carpenter is very comprehensible but unfortunately it tries to explain the entire
reading process, from fixation to long term-memory (Just and Carpenter 1980; Just
et al. 1982). Although this made the model quite complex it is still disputed as it is
assumed to simplify matters too much (Reichle et al. 2000). However, although the
model might have tried to cover too much of the reading process it still has merits,
simplifying a complex problem is not necessarily negative, and many approaches
used by Just an Carpenter has also been used in this thesis.
There are also other processing models available but these have had a minor impact
on the work presented here. Rayner’s E-Z reader is a model that is similar to the one
presented by Just and Carpenter, but with a narrower scope since it does not try to
account for high-level linguistic processing. It does however account for preview
and context effects (Rayner 1988, Rayner and Serano 1999, Reichle et al. 2000). The
model is unfortunately quite complex and the underlying assumptions are not as
transparent as in the model presented by Just and Carpenter. Another processing
model is the attention-shift model (Reilly 1993). It utilizes two connectionist back-
propagating neural networks, one for word recognition and one for planning
saccades. From a linguistic viewpoint the attention-shift approach seems a little too
simple to be plausible, however the use a learning algorithm is appealing since
individual differences in reading behavior are likely to be quite large.

2.2 The Concept of Readability

It is important to not mistake readability for legibility. Legibility has to do with the
recognition of text items whereas readability has to do with the processing of
continuous and meaningful text (Mills and Weldon 1987). Readability is typically
rreferred to as the ease of “which the meaning of text can be comprehended” (Mills
and Weldon 1987:331). This is of course a very vague definition, but the assessment
of readability is also affected by a multitude of factors. First, there are many
differences between texts, some are very comprehensive and well written whereas
others can be totally unreadable. Second, there are differences between readers;
some are very experienced whereas others cannot read at all. Third, there are
differences between reading situations, reading reference literature before an exam
differs a lot from reading a novel while waiting for the bus. Fourth, there are
differences between the presentation formats, this thesis might be comfortable to
read on paper but is likely to be strenuous to read on a flickering screen with low
resolution. To summarize: There are so many factors that affect readability that it is
impossible to account for them all.
Since readability is hard to enumerate the solution is to use approximate measures
instead, the estimations may not be precise but they are at least likely to point in the
right direction. The readability estimations used in this thesis can be categorized
according to their use as either ratings or measures. Ratings are used to determine
readability of text based on quantitative predictions whereas measures are used to
evaluate readability based on actual reader performance.
 8

2.2.1 Readability Ratings

Readability of text is usually rated by using readability formulas. Most readability

formulas are quite simple and use a combination of word frequencies, word lengths
and sentence lengths as a basis for the results. Although most formulas use purely
quantitative measures they can give an indication of how hard or easy a text is likely
to be to read. There are several readability formulas for English available (see Tekfi
1987 for an overview) but for Swedish there is only one that is widely known. LIX
(Läsbarhetsindex in Swedish) is a quantitative readability formula for developed by
Björnsson (1968). An estimated value of the readability of a text is calculated on
basis of the percentage of long words, seven or more characters, and the average
sentence length (Eq. 1).

LIX = (lwrd/nwrd) * 100 + (nwrd/nsnt) (Eq. 1).

The readability estimate (LIX) is reached by first dividing the number of words with
seven characters or more (lwrd) by the total number of words (nwrd), the percentage
is then added to the average number of words per sentence, i.e. the number of words
(nwrd) divided by the number of sentences (nsnt). The result is a value between
approximately 1 and 100. The following table offers interpretations of what the LIX
values stand for (Table 1):

Table 1. LIX value interpretations.

LIX value Interpretation Type
20 Very easy Children books
30 Easy Novels
40 Medium News papers
50 Difficult Official documents
60 Very difficult Legal documents

Readability formulas like LIX have been criticized for generalizing too much. A
long word is not necessarily harder to read than a short word and short sentence can
also be less readable than a long sentences (Bruce et al. 1981; Redish 1981).
However, readability formulas have also been found to be useful for many
appliances, especially when detail level is not so important (Klare 1984). In the work
with this thesis LIX has been used to determine the readability of the texts used in
the evaluation (see web resources for an address to a web-based lix counter).

2.2.2 Readability Measures

Readability has mostly been evaluated in terms of reading speed and comprehension
(Mills and Weldon 1987). Reading speed is often calculated as words read per
minute (wpm) whereas comprehension is represented as percent of correctly
answered multiple-choice questions, both these measures are objective. The reading
speed results are mostly reliable when comparing results from different evaluations
whereas the comprehension scores are a little unpredictable since they are highly
dependant on the type of questions asked. The product of the reading speed and
comprehension scores has been suggested as a composite measure for reading
efficiency (Jackson and McClelland 1979; Rahman and Muter 1999; Castelhano and
Muter 1999). The measure is used to avoid problems associated with assumed trade-
 9

offs between speed and comprehension (Wickens 1992). However, since the
comprehension scores are likely to be unreliable there is really no point in mixing
them up with reading speed.
Although a high reading efficiency is likely be a good indication of good readability
it is also common to use a few subjective measures. The most widely used measure
of this type is the attitude inventory. It is especially common to use when different
text presentation formats are compared against each other. Essentially attitude
inventories are a set of questions about experience and preference. Unfortunately the
questions often differ between evaluations so it is hard to compare the results,
nonetheless they can be very illuminating for the evaluators. Another subjective
measure used in evaluations, which actually is comparable, is the standardized task
load inventory NASA-TLX (Task Load Index) (Hart and Staveland 1988). The
inventory is composed of six factors denoting different cognitive demands that are
rated by the subjects after completing a task. The NASA-TLX task load inventory
was rewardingly used in the RSVP evaluation preceding the one presented in this
thesis (Sicheritz 2000, Goldstein et al. 2001).

2.3 Screen Reading

Quite a lot of research in readability has focused on comparisons between reading on

screen and reading on paper. It began with studies of readability on the first
generation Cathode Ray Tube (CRT) screens, also commonly referred to as Visual
Display Units (VDUs). In the majority of the early experiments on VDUs readability
was found to be poor compared to paper. The average reading speed for an English
text on paper is between 220-340 wpm (Kump 1999). Reading speed was around 20-
30% slower although comprehension was roughly the same (Muter et al. 1982; Kang
and Muter 1984). These findings are not too surprising since the first screens were
primitive units with poor legibility due to low resolution and mediocre refresh rates.
The designs of the early experiments have also been criticized mainly because the
reading situations were quite unrealistic (Dillon 1992). For example, in the Muter et
al. study the subjects were seated five meters from the display unit while reading and
the text was presented in large white characters on a blue background.

2.3.1 Large Screen Readability

The Achilles' heel of the first-generation large screens seems to have been the low
resolution. Screen technology evolved rapidly in the 80’s and the second-generation
CRT screens offered far better resolution and also color. However, the breakthrough
in readability, and usability in general, came with the introduction of the Graphical
User Interface (GUI). Later studies using computers with GUIs showed that there
was in fact little or no differences between screen and paper, provided that attention
was paid to such factors as screen resolution, refresh rates, anti-aliasing, text
polarity, etc (Gould and Grischkowsky 1984; Osborne and Holton 1988; Muter and
Maurutto 1991; Muter 1996). Although reading speed and comprehension does not
differ much between high-quality screens and paper the users still seem to prefer
reading on paper. This may be partially due to the fact that reading on a large screen
requires the reader to view the text from a distance and in a fatiguing posture
(Schneiderman 1998). An underestimated aspect is also that most readers are more
used to reading on paper, with time however there might be people that prefer
reading on a screen. However, the screen must not necessarily be seen as a successor
to paper but rather as a complement. There are many things that can be done with a
 10

text on a computer that is hard, or impossible, to do on paper. A common situation is

also to browse for documents on the screen and then print the selected document on
paper for reading. In this light the readability on the large screens of today seems
quite satisfactory. Readability on small screens is however another story.

2.3.2 Small Screen Readability

Most mobile devices utilize flat Liquid Crystal Display (LCD) screens. The early
LCD screens were monochrome and offered poor resolution, a little like returning to
a sized down version of the an early VDU unit. However, LCD technology evolved
as well and today LCD's offer a resolution and colour depth that is comparable to
second-generation CRT screens. The problem with readability on small screens is
however not so much the resolution as it is the limitation in the screen space, which
restricts the amount of information that can be presented at a time. This implies a
higher the rate of interaction by the user to view the text. Reading a longer text on a
small screen can thus be frustrating and, to complicate matters further, users of
mobile devices do not always have access to printing facilities. Researchers have
been studying the effect of display size on reading in order to determine how small it
can become before problems occur.
Duchnicky and Kolers (1983) performed an experiment with varying window widths
and heights and found that a height of 20 lines only increased reading speed by a
mere 9% compared to using a height of 4 lines. Smaller window heights than 4 lines
were however found to be significantly less efficient to use. The results also showed
that a window width of 2/3 of a full page increased reading speed by 25% compared
to using 1/3 of a full page. These widths are however much larger than what the
average mobile device has to offer but the findings seem to suggest that a limited
screen width decreases reading speed. Using a higher density of characters per line
was also found to improve readability, using 80 characters compared to using 40
characters increased reading speed by 30%. This is not very surprising since a lower
density implies less information in the perceptual span at a time and therefore also a
lowered efficiency. The texts read in the Duchnicky and Kolers study were however
quite short, ~300 words.
Dillon et al. (1990) investigated how the reading of longer texts, ~3500 words, was
affected by using window heights of 60 and 20 lines. The results showed that neither
reading speed nor comprehension differed. However, the subjects who read using the
smaller window height were found to perform significantly more jumps and also
altered the direction of reading much more often. Unfortunately, there are has been
quite few evaluations on really small displays. The results from the evaluations
reviewed here does however seem to support the assumption that a small screen
space implies a higher interaction when the text is presented in the traditional way.

2.4 Text Presentation on Small Screens

There are a few different ways to present text on small screens. They can be divided
into traditional and dynamic text presentation formats. The major difference
between the formats is that traditional text presentation requires physical interaction
when reading whereas the text proceeds automatically when dynamic text
presentation is used. The text presentation formats presented here are however not
applicable to small screens alone. Sometimes there is a need for squeezing a lot of
information into a small area on large screens as well.
 11

2.4.1 Traditional Text Presentation

The text on a screen is traditionally presented with a spatial layout in the same
manner as on page. However, since a full page cannot be displayed on a small screen
it is divided into smaller parts. The text can then either be presented as smaller pages
that fits the screen or as a long page, i.e. a scroll, continuing outside the screen. In
the page format turn-page keys are used to move between the pages and in the scroll
format a scroll-bar is used to move in the text. Continuous scrolling has been found
to be preferred compared step-by-step scrolling but the page format is still more
popular to use (Mills and Weldon 1987). Both formats require the reader to interact
physically in order to move forward in the text but the increase in interaction may be
acceptable considered that these formats are more familiar to the readers.

2.4.2 Dynamic Text Presentation

Leading and RSVP are the two most common forms of dynamic text presentation.
Both formats requires very little interaction from the reader since the text proceeds
automatically, the obvious advantage of these techniques with respect to limited
screen displays is however their ability to display text in one single line, leaving the
rest of the space for other pieces of information. In Leading, or horizontal scrolling
(also known as the Times Square Format), the text moves from right to left. Sekey
and Tietz (1982) and Granaas et al. (1984) found that Leading was less effective
than traditional text presentation in the page format. However, in these studies the
text moved forward letter-for-letter. Kang and Muter (1989) evaluated leading that
moved forward pixel-for-pixel and found that to be more effective. Chen and Chan
(1990) evaluated leading with self-paced versus experimenter-controlled
presentation speed. The results showed that the self-paced subjects read faster
whereas the experimenter-controlled subjects remembered more. From these results
they deduced that self-paced reading can be counterproductive and that experience is
important when reading moving text.
RSVP originated as a tool for studying reading behavior (Forster 1970) but has lately
received more attention as a presentation technique with a promise of optimizing
reading efficiency (Joula et al. 1982; Masson 1983; Potter 1984; Joula et al 1995;
Muter 1996; Rahman and Muter 1999, Sicheritz 2000; Goldstein et al. 2001). When
RSVP is used the text is successively displayed as small chunks within a small area,
each chunk typically contains one or a few words depending on the width of the text
presentation window. When reading the text in this fashion it proceeds by itself and
that makes the saccadic eye movements and the return sweeps superfluous. It also
means that regressions, or the rereading of words and phrases, may be effectively
prevented (Rahman and Muter 1999).
Comparisons between RSVP and Leading have so far been inconclusive. Kang and
Muter (1989) found no significant differences between the two techniques whereas
Joula et al. (1995) reported that Leading was inferior to RSVP. McCrickard et al
(2001) used RSVP, Leading and ”fading” techniques to support secondary tasks on
peripheral displays and concluded that Leading techniques are better for
comprehension and memorability while RSVP and ”fading” techniques are better for
rapid identification of items. However, when reading is the primary task it seems
more natural to use RSVP. The eye processes information in fixed gazes and a
format that moves the text successively rather than continuously is therefore
assumed to better adhere to the reading process. This is the foremost reason for
choosing RSVP as an alternative to traditional text presentation.
 12

2.5 Rapid Serial Visual Presentation

The term RSVP was first introduced by Forster (1970) as a name for a technique
used for studying text processing and comprehension. Later RSVP was introduced as
a presentation technique for computer screens with the assumption that the reduced
need for eye movements would reduce cognitive load and optimize reading (Joula et
al. 1982; Masson 1983; Potter 1984). However, the term RSVP has come to label a
wide variety of approaches for text presentation where the only common
denomination has been that chunks of text have been successively displayed. The
designs of most RSVP evaluations and implementations have differed so much that
the findings from one evaluation are not necessarily applicable to another. Some
have presented single words at a time whereas others have presented several words.
A few have presented long texts where as most have presented only smaller
paragraphs. The differences in the reading speed have also been quite large, some
have used a very high presentation speed whereas others have let the readers choose
their own.
A short review of the most influential evaluations where RSVP have been used
follows next, although they differ a lot from each other there is something to learn
from all of them. The review is followed by the description of the Reader version 1.0
evaluation, the first RSVP implementation evaluated on a PDA (Sicheritz 2000,
Goldstein et al. 2001). The Reader evaluation is the one preceding the evaluation
presented in this thesis and much has been learned from the experiences of that
evaluation.

2.5.1 Previous RSVP Evaluations

Joula et al. (1982) presented shorter paragraphs of text on a CRT screen, either in the
page format or in the RSVP format with text chunks of 5, 10 or 15 characters. Each
text chunk was exposed for 200-300ms, which is equal to a reading speed of
approximately 300 wpm. The results showed no significant differences in
comprehension between the reading conditions.
Masson (1983) evaluated how the insertion of blank windows at sentence boundaries
affected the RSVP format. Masson experimented with durations of 500 and 1000 ms
and found that performance increased with blank windows regardless of duration.
Cocklin et al. (1984) compared RSVP with the text divided into either idea units or
ad-hoc chunks. The idea unit segmentation was performed by hand and was based on
clause and phrase boundaries as well as linguistic features. Each chunk averaged 13
characters and the reading speed was approximately 300 wpm. The results showed
that the use of idea units increased comprehension a little but not significantly.
Muter et al. (1988) performed experiments with self-paced RSVP and RSVP that
permitted regressions. The results showed that larger regressions yielded slower
reading and regressions back to the beginning of the sentence were found to be more
frequent than regressions two words back. Overall the results indicated that
permitting reader control was feasible but permitting regressions resulted in lower
performance.
Kang and Muter (1989) compared RSVP to word-by-word, letter-by-letter and pixel-
by-pixel Leading. Except for word-by-word, comprehension was as high for Leading
as it was for RSVP. The comprehension scores for pixel-by-pixel leading were also
found to match RSVP at reading speeds ranging from 100 to 300 wpm. The subjects
in the evaluation were also found to express a significantly higher preference for
pixel-by-pixel Leading.
 13

Fine and Peli (1995) evaluated how visually impaired and elderly subjects read using
RSVP and scrolled text. They found that the visually impaired read at a similar
speed using both formats whereas the elderly read faster using RSVP,
Rahman and Muter (1999) compared word-for-word RSVP and sentence-by-
sentence presentation, with or without a completion meter, to traditional text
presentation in the page format. No significant differences were found for
comprehension and reading speed but the subjects liked the inclusion of a
completion meter.
Castelhano and Muter (2001) evaluated the effects of using RSVP with or without
punctuation pauses, variable word durations and a completion meter. They compared
a few RSVP formats to traditional text presentation and sentence-by-sentence
presentation. The results showed that pauses and variations made the RSVP format
significantly more accepted. However, the sentence-by-sentence and traditional page
format remained more popular although RSVP was just as effective.

2.5.2 The Reader Evaluation

Karin Sicheritz implemented the Reader version 1.0 in order to evaluate how it was
to read using RSVP on a PDA compared to using a paper-book (Sicheritz 2000,
Goldstein et al. 2001). The application was implemented on a Casio Cassiopeia E-
105 PDA and offered a graphical user interface (Figure 2).

Figure 2. The Reader version 1.0 prototype, adopted from Sicheritz (2000).

In the Reader the user could select three different text presentation window widths
(11, 17 and 25 characters), twenty-seven speed levels ranging from approximately
25 to 1000 wpm (low, 1–25 and fast) and five font sizes (8–16 points). The text
presentation could be started, paused, continued and it was also possible to return to
the previous text window. A page indicator showed the current and total number of
pages. While presenting text the application searched for the nearest white space
given the window width and segmented at that point. The text chunk was then
exposed for a fixed time according to the selected speed. If the number of characters
for a single word exceeded the chosen window width the whole word was presented
anyway. The application also segmented the text when commas or other punctuation
 14

marks were detected. Blank windows of 250 ms were inserted between the sentence
boundaries (Sicheritz 2000).
In a repeated-measurement within-subject experiment employing 10 subjects the
Reader prototype was benchmarked against the paper-book. Two conditions with
different window widths for RSVP, 11 and 25 characters respectively, and a sonified
condition (i.e. where sounds were added to enhance the reading experience) were
compared to a paper-book condition. The texts used in the experiment were the first
six chapters from the novel “Röda Rummet” (in Swedish by August Strindberg), the
chapters were between ~2700-6300 words long. The subjects read the first chapter in
the paper-book and the following chapters using the Reader prototype. The subjects
were instructed to read as fast as possible in all but the sonified condition.
Readability was measured by reading speed, comprehension inventories consisting
of 10 multiple-choice questions, the NASA-TLX (Task Load Index) (Hart and
Staveland 1988) task load inventory and an attitude inventory consisting of five
questions about Difficulty, Efficiency, Comprehension, Stimulation and Facilitation
of the presentation (Sicheritz 2000; Goldstein et al. 2001). The results showed that
neither reading speed nor comprehension differed between the conditions. The text
presentation width of 25 characters resulted in the highest reading efficiency but the
differences were not significant. The task load inventory did however reveal
significantly higher task load ratings for all RSVP conditions for all factors but
Physical demand. Also on the attitude inventory was the book condition significantly
more favored for Difficulty, Comprehension and Stimulation (Sicheritz 2000;
Goldstein et al. 2001).
The results obtained from the Reader evaluation were quite encouraging. At least
they showed that reading using RSVP on a PDA was just as efficient as using a
paper-book, although the subjects may not have agreed. The high task load rating
and the lower attitude ratings for the RSVP format were disappointing. Many issues
may have affected the discrepancy between the subjects’ objective and subjective
results. One problem with the Reader evaluation was that it compared screen reading
to reading in a paper-book, which are two very different things. Although the
resulting findings are interesting it would be fairer to see how RSVP performed
compared to traditional text presentation on an actual PDA. There were also a few
problems with the experimental design. All subjects read the first chapter in the
paper-book but only the two RSVP conditions with different window widths were
properly balanced. The problem is that both the text lengths and difficulty ratings
(LIX) differed a lot between the chapters. Although the realism of reading chapters
consecutively may be lost when balancing the conditions it is much easier to draw
conclusions from the results if experimental control is kept. The texts chosen to be
included in the experiment were also quite hard to read, the results may have looked
different if easier texts were chosen. Another problem was that the subjects had to
halt the experiment each time they made a change in reading speed (it was needed to
keep an experimental record over the selected reading speeds). This may have
caused the subjects to refrain from speed changes or at it least likely to have
disrupted the reading experience. A probable cause for the high task load when using
the Reader prototype might also have been that the exposure time for each text
chunk was fixed. This does not seem to adhere very well to the reading process and
Sicheritz and Goldstein also point that out on basis of the results from the Reader
evaluation (Sicheritz 2000, Goldstein et al. 2001).
However, the last hypothesis is exactly what triggered the work with adaptive RSVP
and the prototype presented in this thesis. This hypothesis also brings us to the end
of the background part. The next part is about the design and that is where the
hypothesis becomes a thesis.
 15

3 Design

Unfortunately the Reader prototype (Sicheritz 2000) could not be improved further
to include adaptive RSVP due to technical reasons. Instead an algorithm for Fixed
RSVP that mimicked the performance of the Reader had to be designed. This section
begins with a description of the fixed RSVP algorithm in section 3.1. In section 3.2
the design of Adaptive RSVP is introduced and explained. In section 3.3 the design
of a distribution and processing model for RSVP is presented together with an
example of a devised document format.

3.1 Fixed RSVP

The speed of the presentation when using RSVP is measured in words per minute
(wpm). The exposure time of each text chunk is calculated on basis of the
presentation speed and on how much that can be displayed in the text presentation
window. Unfortunately there is little or no documentation in previous studies on
exactly how the exposure times have been calculated (see Section 2.5 for an
overview). What is known however is that the exposure times have generally been
fixed. In this evaluation the following formula has been employed for calculating the
fixed text chunk exposure times (Eq. 2):

time0 = fchr/(wavg*wpm/60) (Eq. 2)

The average number of characters that can be displayed (fchr) is divided by the
product of the average word length (wavg) for the current language and the
presentation speed (wpm) divided by 60. The result is a fixed exposure time for each
text chunk measured in seconds (timeo).
When presenting a text with RSVP the relation between reading speed and exposure
time can be visualized in a speed-exposure plot where all the individual text chunks
are represented (Figure 3).

1600
Text chunk exposure time (ms)

1400
1200
1000
800
600
400
200
0
0 100 200 300 400 500 600 700 800
Text chunk reading speed (wpm)

Figure 3. Variations in reading speed for individual text chunks containing 1, 2,

3, 4 or 5 words (the dots) presented at a set speed of 300 wpm (the vertical line)
when using RSVP with fixed exposure times (data derived from the training
text).
 16

The plot is a result of presenting the training text used in the usability evaluation at a
constant speed of 300 wpm using the fixed exposure time formula (Eq. 2). The width
of the text presentation window was 25 characters (fchr=21) and the average word
length was set to 7 characters. The obtained reading speed is a result of dividing a
text chunks exposure time by the number of words it contains (1-5). This explains
why there is a variation in reading speed (100-500 wpm) although the exposure time
is fixed (600 ms). Only the text chunks with three words match the selected reading
speed (the vertical line in Fig. 3). If more words appear in a text chunk speed
increases and if fewer words appear speed decreases. The reading speed is thus
inversely related to the number of words that each text chunk contains. One
explanation to the high task load when using fixed RSVP may be attributable to the
fact that the exposure time do not vary although reading speed actually does.

3.2 Adaptive RSVP

Just and Carpenter found that “there is a large variation in the duration of individual
fixations as well as the total gaze duration on individual words” when reading text
from paper (Just and Carpenter 1980:330). The point with Adaptive RSVP is that it
attempts to mimic the reader’s cognitive text processing pace more adequately by
adjusting each text chunk exposure time in respect to the text appearing in the RSVP
text presentation window. By assuming the eye-mind hypothesis (Just and Carpenter
1980), i.e., that the eye remains fixated on a text chunk as long as it is being
processed, the needed exposure time of a text chunk can be assumed to be
proportional to the predicted gaze duration of that text chunk.
As mentioned earlier, the resulting reading model presented by Just and Carpenter is
comprehensible but also quite complex since it tries to explain the entire reading
process. Although the model has been accused for simplifying matters too much
(Reichle 2000) it is still a too complex to utilize here on an as-is basis. The reason
for keeping the adaptation simple is that the approach with varying presentation
speeds is quite novel and its effect on readability unknown. It is better to begin with
some small changes and observe how they affect the results since if a complex
model is used it is much harder to identify the individual factors that affected the
outcome. The adaptive algorithms presented here are therefore quite simple but they
are still based on the assumptions made by Just and Carpenter.
Since very common, known or short words are usually processed faster than
infrequent, unknown or long words, the text chunk exposure times can be adjusted
accordingly (Just and Carpenter 1980; Just and Carpenter 1982). Further, most new
information tends to be introduced late in sentences and therefore ambiguity and
references tends to be resolved there as well. A shorter sentence is also usually
processed faster than a longer one since it conveys less information (Just and
Carpenter 1980; Just and Carpenter 1982). Thus, processing time differs both within
and between sentences and the text chunk exposure times can therefore be adjusted
according to this as well. On basis of these assumptions two adaptive algorithms
were developed that were supposed to decrease cognitive demand (i.e. task load
ratings). The first algorithm adapts the exposure times to the content of the text
chunks whereas the second also looks to the context in the sentences. Both
algorithms inserts a blank window between each sentence if there is not enough
space to begin on the next sentence in the same window, otherwise a delay is added
to the sentence boundary instead. This approach is assumed to generate a natural
spacing between sentences and cause fewer interruptions while reading.
 17

3.2.1 Content Adaptation

In content adaptive mode the exposure time for each text chunk is based on the
numbers of characters and words that are being exposed for the moment. It is
deliberately a very simple form of adaptation based on some general assumptions
about word distribution. Longer words are assumed to be more infrequent and take
longer time to read than shorter words. A higher number of words are also assumed
to take longer time to read and should thus receive more exposure time. The
following formula is used to calculate the text chunk exposure time for content
adaptation (Eq. 3):

time1 = (nwrd+nchr)/(davg*wpm/60) (Eq. 3)

The formula uses the number of words (nwrd) and the number of characters (nchr) as
a basis for the results. Both arguments are added and divided by the product of the
average word length including delimiters (davg) and the currently set speed in words
per minute (wpm) divided by 60. The result is a variable exposure time (time1)
depending on the content the current text chunk.
The effect of using content adaptation is probably best explained by using a speed-
exposure plot (Figure 4).

1600
Text chunk exposure time (ms)

1400
1200
1000
800
600
400
200
0
0 100 200 300 400 500 600 700 800
Text chunk reading speed (w pm)

Figure 4. Variations in reading speed for individual text chunks containing 1,

2, 3, 4 or 5 words (the dotted lines) presented at a set speed of 300 wpm (the
vertical line) when using content adaptation (data derived from the training
text).

Again the plot is a result of presenting the training text at a constant speed of 300
wpm but this time by using the formula for content adaptation instead (Eq. 3). The
average word length including delimiters was set to 7,8 whereas the other variables
were the same as those used for Fixed RSVP. The speed-exposure plot illustrates the
difference compared to Fixed RSVP.
However, even though the exposure time varies the variation in reading speed is
actually smaller for content adaptation than for Fixed RSVP. When using content
adaptation the exposure time for each text chunk is directly related to the number of
words and characters it contains. Reading using this form of adaptation is supposed
to decrease task load since the relation between what is being exposed and the time
for exposure is more proportional.
 18

3.2.2 Context Adaptation

In context adaptive mode the exposure time for each text chunk is based on the
following: The result of content adaptation, the word frequencies of the words in the
chunk and the position of the chunk in sentence being exposed. To begin with each
word in the chunk is looked up in a lexicon with word frequencies. If the word is
common it receives a weight lower than one and if it is rare or not in the lexicon it
receives a weight higher than one. The following formula is used to calculate how
the exposure time is affected by the word frequencies (Eq. 4):

time2 = time1 * ((wfrq1+…+wfrqnwrd)/nwrd) (Eq. 4)

The formula uses the exposure time for content adaptation (time1) and the word
frequency weights for the words in the chunk (wfrq) as a basis for the result. The
word frequency weights are added and divided by the number of words in the text
chunk (nwrd). The product is then multiplied with the content adaptive exposure
time to get the weighted exposure time (time2).
The next step is to give the chunk less exposure time if it appears in the beginning of
a sentence and more if it appears in the end. The following formula is used to
calculate the text chunk exposure time depending on the position in and the length of
the current sentence (Eq. 5):

time3 = (time2+time2* tanh(swrd/savg))/2 (Eq. 5)

The formula uses the intermediary exposure time reached earlier (time2), the number
of words in the sentence exposed so far (swrd) and the average sentence length
(savg). In order to get a smooth drop-off in speed along the sentence, a mean of the
previously calculated exposure time and its product with the hyperbolic tangent
(tanh) of the division of the number of exposed words and the average sentence
length is calculated. The result is a varying text chunk exposure time (time3). The
effect of using context adaptation is here illustrated by using a speed-exposure plot
again (Figure 5).

1600
Text chunk exposure time (ms)

1400
1200
1000
800
600
400
200
0
0 100 200 300 400 500 600 700 800
Text chunk reading speed (wpm)

Figure 5. Variations in reading speed for individual text chunks containing

1, 2, 3, 4 or 5 words (the dotted lines) presented at a set speed of 300 wpm
(the vertical line) when using context adaptation (data derived from the
training text).
 19

The plot is a result of presenting the training text at 300 wpm when using context
adaptation (Eq. 5). The average sentence length was set to 11,5 words and the word
frequency weights ranged between 0,6-1,2. A lexicon with frequencies for the
10.000 most common words in a corpus of 11,9 million words (Press 97) was used
to assign the weights according to a lognormal distribution. The assumption with
using a lognormal distribution was that the expected gaze duration on each word
would decrease with the logarithm of the word’s frequency. This captures the fact
that there are small differences in frequency among infrequent words whereas the
differences larger for common words (Just and Carpenter 1980).
When Context adaptive RSVP is used the variations in exposure time are larger than
for both Fixed and Content adaptive RSVP. Task load is however still supposed to
decrease since the variations are assumed to better match the cognitive demand
while reading.

3.3 Distribution Model

The discussion around the distribution model originated from the question if the
texts were supposed to be processed on the client or on the server. The reason for
raising this question is that the mobile clients usually are quite thin (i.e. has limited
processing power) and that linguistic processing can be quite demanding. If a server
were used the adaptation could be made much more advanced including parsers for
elaborate segmentation and linguistic processing. The server approach also makes it
possible to update the processing capabilities continuously without having to
reprogram the clients. On the other hand it is also appealing to have independent
clients that can take a text and present it without being connected to a server. There
are merits and pitfalls with both approaches and in the end a combination of both
were chosen. In the distribution model applied here the linguistic processing can be
done on both the server and the client. The server is supposed to perform advanced
linguistic processing whereas the client is supposed to do simple linguistic
processing if no server is available or needed. The client-server approach requires
some form of intermediary document format in order to transfer the meta-
information about the text. That format should ideally also be able to transfer
information about the document structure and pointers to additional resources added
to the text (i.e. sounds, images, bookmarks, annotations etc).
The RSVP formulas developed in this thesis are quite simple and can be considered
as a form of client processing formulas. A document markup schema for RSVP was
still designed in the eXtensible Markup Language (XML) format in order to
facilitate the addition of external resources (i.e. sounds, see section 4.4 for
motivation) and encourage further developments on the server side processing (see
web resources specifications). The following excerpt from an XML tagged file (in
Swedish) will serve as an example of what sort of information a RSVP document
might contain (Figure 6):

<?xml><text title=”Glöd” author=”Anette Kullenberg”><chapter

number="1" title=”Väntan”> <paragraph> <sentence> Klockan
var <sound=”ticktwo.wav”> två på morgonen när Tomás Murillo
sköt ner rånarna. Han hade </sound> gömt sig bakom bardisken
och släckt ner och satt på huk och väntade. <bookmark
id=”1”> Bredvid sig hade han</bookmark> en flaska brandy,
men han drack inget. </sentence> <sentence>Han tände en
Kruger ibland och såg på klockan i tändstickans
sken.</sentence></paragraph> </chapter></text></xml>

Figure 6. Excerpt from an XML tagged RSVP file.

 20

4 Implementation

In order to evaluate the RSVP algorithms they had to be incorporated into a

prototype on a mobile device. Bailando is the resulting prototype that has been
developed at the Ericsson Research’s Usability & Interaction Lab by the author.
Bailando is an acronym for: Better Access to Information through Linguistic
Analysis and New Display Organization. Bailando was developed for the Compaq
iPAQ 3630 Pocket PC, a small Personal Digital Assistant (PDA) with a touch
sensitive high-resolution color display (see web resources for product specification).
The iPAQ offers much more screen space than the average mobile device but it still
shares many of the properties typical for smaller devices. The size and weight of the
device was similar to a pocket book of approximately 250 pages (Figure 7).

Figure 7. The Bailando prototype running in reading mode on a

Compaq iPAQ 3630.

The motivation for the choice of deployment is given in section 4.1 together with a
brief overview of the structure of the implemented application. The Bailando
prototype is demonstrated more in detail in section 4.2 where a walkthrough of the
application is presented. In section 4.3 the RSVP characteristics used by Bailando
are presented. Bailando was supposed to be used for experiments with sonification,
although this is a little out of scope for this thesis, the functionality for this is
described in section 4.4. Finally in section 4.5 the log functionality of the prototype
is described.

4.1 Deployment

The Compaq iPAQ is a Windows 32-bit platform; the available programming

languages were primarily Visual Basic and Visual C++. It might have been possible
to use Java as well but there were just too many questions marks around the
performance of Java on the iPAQ to make that choice feasible. It was important that
the timing in the application was reliable since the differences in exposure time
when using adaptive RSVP are quite small. In the end Embedded Visual C++ was
chosen as a programming environment since it offered real-time capabilities (see
web resources for product specification). Using C++ also made the program more
extensible and easier to export the to other platforms.
 21

4.1.1 Program structure

Bailando was implemented as multi-threaded application where the main processes

communicated with each other by using semaphores (i.e. flags). Basically there was
one thread to keep control over the graphical user interface, one thread to parse and
process the text and one thread to play sounds (see section 4.4 for motivation). The
widgets and the threads were implemented as objects whereas functions like
redrawing the screen or calculating a text exposure were implemented as methods.

4.2 Graphical User Interface

It was important that the graphical interface was appealing and yet intuitive to use. It
had to look as a professional application since it was supposed to be compared to
other professional applications for traditional text presentation. Since ample screen
space was available on the iPAQ all the application controls were implemented into
the graphical user interface. This also makes the Bailando software easier to run on
other PDA’s since button assignments differ between devices. In order to get a feel
for how Bailando turned out, a step-by-step walkthrough will be presented here with
design choices and explanations given along the way.

4.2.1 Bailando walkthrough

The first that happens when Bailando is launched is that a start screen is shown
(Figure 8, left). This screen is also showed if a text for some reason or another is not
loaded properly, but then an advisory error message is shown as well. The main
interaction control of the application is the toolbar. It is always located at the bottom
of the screen so that the hand does not obscure the screen while it is used. It always
has the same appearance although unselectable items are greyed out. As can be seen
on the start screen the only item selectable on the toolbar is the Menu.

Figure 8. Bailando start screen (left) and the library menu (right).
 22

When the Menu item is selected a pop-up main menu is shown. From the main menu
it is possible to: exit the application, to get information about version, to change
settings and to retrieve texts from the library. If the library alternative is selected the
library screen is shown. The Bailando application supports files in either the XML
format described earlier or in plain text. A text is selected from the library by
selecting it in the list (The file “Glöd” is selected in Figure 8, right).
Once a text is selected in the library the processing of the file starts. First the file is
checked to see if it can be viewed in the Bailando prototype at all. After that the data
format is determined. If it is an XML file it is parsed and the meta-data is stored as
variables and if it is a plain text file some of the file properties are picked up and
stored (i.e. filename and type). The next step is to analyse the text, the number of
characters, words, long words and sentences are calculated in order to determine the
LIX value.
While processing the text the main reading screen is exposed (Figure 9), but not until
the text is processed completely are all the toolbar items selectable. The complete
toolbar contains the controls to start, pause and resume the presentation. The
presentation can also be paused and resumed by touching anywhere on the text
presentation area of the screen. It is important that it is easy to make pauses since the
user is likely to need to do that fast when interrupted in order not to get lost. This
important feature of the interface is introduced to the user by the following
instruction when the text is ready to be viewed: “tap the screen to begin”.

Figure 9. Bailando main screen in reading mode (left) and pause mode (right).

In Bailando the text is presented at an area located slightly above the half of the
screen. The text is presented at one single line that utilizes 2:3 of the screen width
and the vertical alignment is similar to the text presentation area as a whole. Above
the text presentation area there is a border for aesthetical reasons. Below the text
presentation area there is an information area displaying the text title, the progress
bar and the current speed settings (Figure 9). The progress bar is included in order to
support memory of spatial location while reading, as said earlier a completion meter
has been found to increase the user preference for RSVP in a previous evaluation
(Rahman and Muter 1999).
 23

While reading the toolbar show the Pause button and the coloured part of the
background is green (Figure 9, left). If the user feels he missed or did not understand
a presented text window it is possible to go backwards (<<). It is also possible to
skip text by going forward (>>). These orientation features are included to make it
easier for the user to browse through the text. It is also possible to browse through
the text step-by-step in paused mode. The reading speed can be easily increased (+)
or decreased (-) with the speed control buttons in steps of 10 wpm. When the text
presentation is paused the toolbar shows the Read button and the background is then
also turned red as an extra affordance (Figure 9, right).
When a text is loaded the book item on the toolbar also becomes selectable. The idea
was to add bookmark and table of content functionality to this menu but in the
present version of Bailando this menu only lets the user choose to go to the
beginning of the text, to the point furthest read in the text or to view text
information. The text information screen contains the information selected during
the initial processing of the text (Figure 10, left). Apart from showing the texts title,
type and metrics it also displays a readability rating (based on the LIX value for
Swedish). The suggested reading speed is calculated on basis of what and how fast
the reader has read earlier. Bailando was supposed to keep a record of previous
reading speeds for different text difficulties and match a text according to these data.
However, in the present version it only relates the current reading speed to the LIX
value.

Figure 10. Bailando text information screen (left) and settings menu (right).

The last screen-shot from Bailando is the settings options available from the menu
(Figure 10, right). The only setting variable that could be changed through the GUI
in the present version was the adaptation mode (Flat rate is equivalent to Fixed
RSVP). Ideally it would be better if more variables (i.e. average word length, blank
window exposure time, window width etc) could be changed this easily. However,
adaptation was the only setting needed to be changed during the evaluation and due
to time constraints further improvements of the graphical interface were left out.
 24

4.3 RSVP Characteristics

The window width of the RSVP display was 25 characters wide with the text
presented left justified in a 10-pt. sans-serif typeface. The window width was chosen
on basis of the results from the Reader evaluation where a width of 25 characters
gave the best results (Sicheritz 2000; Goldstein et al 2001). Font-size has been found
to have a minor effect on RSVP (Russel et al. 2001) and 10-pt seemed enough.
Bailando supports all the three forms of RSVP that have been described in this
paper. In all RSVP modes the text chunks that contained punctuation marks received
an addition of 250 ms to their exposure times and all words longer than the display
width were hyphenated. In Fixed RSVP mode a blank screen was inserted for 250
ms between each sentence. In content and Context adaptive RSVP mode a blank
screen was inserted for 250 ms only if necessary, otherwise a delay of 250 ms was
added to the exposure time of the text chunks exposing the sentence boundary. The
following table offers a summary of the variables used by Bailando:

Table 2. Variables used by Bailando for RSVP.

Variable Fixed Content Context
Average number of characters displayed 21 - -
Average word length 7 - -
Average word length incl. delimiters - 7,8 7,8
Average sentence length - - 11,5
Default frequency weight - - 1,2
Blank window exposure time 250 250 250
Sentence boundary delay - 250 250
Punctuation mark delay 250 250 250

4.4 Sonification Features

The Bailando prototype also supports Sonified RSVP. That means that it can play
sounds at the same time as it presents text. This is accomplished by parsing XML
tags in the text. When Bailando finds a start-sound tag (see Figure 6 p.9 for an
example) in the text chunk to be displayed it plays the audio file linked to the tag.
When a stop-sound tag is encountered, all playing sounds are stopped. Thus, exact
synchronization between word and sound is attainable and works at any reasonable
reading speed. The sonified features of Bailando have been evaluated in a usability
evaluation performed by Mandana M-Bayat. The results will be published in a
forthcoming article (Goldstein, Öquist and Björk 2002).

4.5 Log Functionality

In order to get reliable results from the usage of Bailando in the evaluation a few log
functions were built in. Every time Bailando is initiated a new log file is created with
a random name and a timestamp. Bailando logs all the user actions in the interface,
i.e. speed changes, pauses and backward or forward movements. It also keeps track
of the text presentation, what text it has displayed, for how long it was displayed,
which variables that were used and what the individual and average reading speeds
became. The log functions have proved to be very useful, both when debugging and
when evaluating (see appendix A for an excerpt).
 25

5 Evaluation

The aim of the evaluation was to see how traditional text presentation, Fixed RSVP,
Content adaptive RSVP and Context adaptive RSVP affected the ability to read on a
mobile device. This part of the thesis begins with a detailed overview of the usability
evaluation in section 5.1. This is followed by an analysis of the results in section 5.2.

5.1 The Usability Evaluation

It was important that the same device was used for all conditions since the look and
feel of the hardware was likely to bias the assessment. Long and short texts were
also included in the evaluation since the experience of extended and brief reading
was thought to differ.

5.1.1 Method

In order to assess the effects caused by reading long and short texts using the four
presentation formats a repeated-measurement experimental layout was adopted. It
also enabled us to see how repeated use of RSVP could affect the results. This is
interesting since RSVP offers a novel way of reading and may take some time to get
used to. The following null hypotheses were set for reading long texts, short texts
and using RSVP repeatedly:

• No difference in Reading speed

• No difference in Comprehension
• No difference in Task load
• No difference in Attitude

The hypotheses were tested in the SPSS V10.0 software using the repeated-
measurement General Linear Model (GLM). The significance level was set to 5%
and the level of multiple comparisons was Bonferroni adjusted.

5.1.2 Design

A balanced within-subject design was employed. Four experimental conditions were

formed where each subject read one long and one short text using each presentation
format. The combinations of long and short texts were fixed creating four text pairs
Aa, Bb, Cc and Dd. The text pairs were balanced against condition and presentation
order generating sixteen combinations. Each subject was randomly assigned to one
of the sixteen combinations.

5.1.3 Subjects

Sixteen paid subjects participated in the experiment. They were all enrolled with the
criteria that they were fluent in Swedish and had a self-reported interest in reading.
The subjects had a mean age of 25 and half of them were male. All were computer
literate and seven had some previous experience of using a PDA. Nine of the
subjects had corrected vision and two were left-handed (see web resources for a link
to subject registration page).
 26

5.1.4 Apparatus

All experiments were performed on a Compaq iPAQ 3630. Bailando was used for all
RSVP conditions. The initial speed of the text presentation was always set to 250
wpm but the subjects were allowed to alter the speed at any time. Two commercial
programs were chosen for traditional text presentation. The Microsoft Reader
version 1.0 for reading long texts and the Microsoft Internet Explorer for reading
short texts (see web resources for links to product specifications). The reason for
including two different programs was their intended context of use. The MS Reader
is custom-made to present longer texts such as e-books whereas the MS Explorer is
designed to present shorter web-content such as news articles (Figure 11).

Figure 11. The interface for the Microsoft Reader (left) and the Internet
Explorer (right).

There are also some important differences between the programs. The MS Reader
uses page-turn buttons to move to between the pages whereas the MS Explorer uses
both page-turn buttons and a scroll-bar. The MS Reader also utilizes a font-aliasing
technique called ClearType (Hill 1999) in order to increase the legibility of the text
whereas the MS Explorer uses regular fonts.

5.1.5 Texts

Four long fiction texts and four short news articles where chosen to be included in
the experiment. One shorter fiction text was also used as a training text. Texts in
Swedish with different readability ratings were chosen. The readability rating was
measured with LIX (Björnsson 1968). The long fiction texts were of a similar length
whereas the short texts differed some (Table 3).
All texts were formatted and processed to fit the different programs used in the
experiment. In order to read the long texts in the MS Reader they were augmented
with cover graphics and converted to the Microsoft e-book format. The short texts
 27

were converted to the HyperText Markup Language (HTML) in order to be viewed

in the MS Explorer. All texts were also converted to plain text so that they could be
presented with the Bailando program.

Table 3. Texts used in the evaluation.

Text type Title (in Swedish) Author / Source Words LIX
Training text Glöd Annette Kullenberg, chapter 1. 705 25
Long text A Röda rummet August Strindberg, chapter 3. 4272 37
Long text B Nils Holgersson Selma Lagerlöf, chapter 1 and 2. 4230 27
Long text C Valarnas sång Wally Lamb, chapter 1. 4326 31
Long text D Bara Alice Maggie O´Farrel, chapter 1. 4170 29
Short text a Makedonien Dagens Nyheter 2001-06-03 430 44
Short text b Mellanöstern Dagens Nyheter 2001-05-27 384 54
Short text c Alacala Dagens Nyheter 2001-06-02 628 40
Short text d Lundin Oil Dagens Nyheter 2001-05-18 365 49

5.1.6 Setting

The experiment took place in a dedicated usability lab equipped with audio and
video-recording facilities (Figure 12). While reading the subject was seated in a
comfortable chair in a room separated from the experimenter by a one-way mirror.
Before the experiment started each subject had some time to get acquainted with the
facilities in order to create a relaxed setting.

Figure 12. Interior of the usability lab with the observation room
(left) and the test room (right).

5.1.7 Instructions

Each subject received instructions before the experiment that pointed out that it was
the applications and not the individual performance that were being tested. All were
encouraged to ask questions whenever they wanted and also told that they could
terminate the experiment at any time if they felt uncomfortable. Written instructions
were administered before each session that described the principal features of the
current user interface, what kind of text they were going to read and how long time it
was likely to take. The subjects were particularly instructed to read at a pace as
comfortable to them as possible (see Appendix B for instructions).
 28

5.1.8 Training

To begin with each subject participated in two training sessions. In the first session
the subject read the training text using the Microsoft Reader and in the second
session the subject read the same text again using Bailando in Content adaptive
RSVP mode. The subjects did not train on using the MS Explorer as all had prior
experience of using it on desktop computers. During the training sessions the subject
was encouraged to try out and get used to the user interface. The idea behind reading
the same text twice was to give the subjects an early success experience and making
them more willing to experiment with the interface. After the two training sessions
were completed the subject was introduced to the questions in the inventories and
filled them in.

5.1.9 Procedure

Four experimental conditions were administered and each condition was divided into
two sessions. In the first session the subject read a long text and filled in the
inventories, in the second session the subject read a short text and filled in the
inventories. Between the first and second condition the subject had a 15-minute
break and between the second and third condition the subject had a 45-minute lunch
break. Between the third and fourth condition the subject had a 15-minute break
again. The total participation time for each subject was around five hours.

5.1.10 Inventories

After each experimental session there were four inventories to fill in. The first
inventory was a comprehension test with three-alternative multiple-choice questions.
For long texts there were ten questions and for short texts there were five questions.
The second inventory was the NASA-TLX Task Load Index (Hart and Staveland
1988), which was administered to check Mental, Physical, and Temporal demands,
as well as Performance, Effort and Frustration levels. The third was a short attitude
inventory with eight questions regarding the reading experience in terms of Ease,
Effect, Comprehension, Stimulus, Immersion, Comfort, Naturalness and Focus. The
attitude inventory had a bi-polar scale with a null point and two anchor points
labeled High and Low (see appendix C for inventories). After the last experimental
condition the subjects also participated in a follow up discussion on their experiences
during the experiment. These inventories were selected to be included in the
evaluation since the results would then be comparable to the Reader evaluation
(Sicheritz 2000; Goldstein et al. 2001).

5.1.11 Caveats

Two problems occurred while performing the experiment that may have had a minor
effect on the results. The first two subjects received a comprehension inventory
where one question did not have an answer; this was however noted by the second
subject and corrected after that. It also occurred an error while converting the texts;
this caused the Bailando program to insert a few extra blank windows between some
characters in Content and Context Adaptive mode. This error was corrected after the
first three subjects had completed the experiment, the subjects did not note this error
and it was discovered after an analysis of the log files.
 29

5.2 Results

All subjects completed the experiment and there were few problems with
understanding what to do or how to do it. Although RSVP was a novel way of
reading for all the subjects none had any problems using the Bailando program. A
few subjects did not know when the end of the text was reached when using the MS
Reader and thus got confused, this was however only a minor problem. Those
subjects that were left-handed had some problems when using the MS Explorer since
the hand sometimes obscured the screen while using the scroll-bar.
The presentation of the results is divided into four sections: Reading speed,
Comprehension, Task load and Attitude. Under each section the null hypotheses set
for long texts, short texts and repeated use of RSVP is tested.

5.2.1 Reading Speed

Reading speed was calculated as words read per minute based on the total time it
took for the subjects to read a text including all kind of interruptions (pauses,
regressions, speed changes etc).

5.2.1.1 Long texts

Adaptive RSVP improved reading speed some but the null hypothesis regarding no
difference in reading speed between the conditions when reading long texts was kept
(Table 4).

Table 4. Reading speeds for long texts.

Condition Avg. wpm Std. dev.
MS Reader 242 80,4
Fixed RSVP 249 58,5
Content adaptive 260 51,2
Context adaptive 258 79,5

5.2.1.2 Short texts

The null hypothesis regarding no difference in reading speed between the conditions
when reading short texts was rejected since the main factor for reading speed was
significant (F[3,45]=8.4, p=0.040). Pairwise comparisons revealed that all RSVP
conditions increased reading speed significantly (p≤0.002) compared to using
traditional text presentation with the MS Explorer (Table 5).

Table 5. Reading speeds for short texts.

Condition Avg. wpm Std. dev.
MS Explorer 157 53,2
Fixed RSVP 212 46,5
Content adaptive 213 36,8
Context adaptive 203 43,9
 30

5.2.1.3 Repeated use of RSVP

There were no significant differences in reading speed for either long or short texts
between using RSVP for the first, second or third time compared to using traditional
text presentation. The null hypothesis regarding no difference in reading speed
caused by repeated use of RSVP was kept.

5.2.2 Comprehension

Comprehension was computed as percent of correctly answered multiple-choice

questions. For long texts there were ten questions and for short texts there were five.

5.2.2.1 Long texts

The null hypothesis regarding no difference in comprehension between the

conditions when reading long texts was kept. Content adaptive RSVP showed the
best results but the differences between the conditions were very small (Table 6).

Table 6. Comprehension scores for long texts.

Condition Avg. % correct Std. dev.
MS Reader 73 19,6
Fixed RSVP 75 17,9
Content Adaptive 76 17,5
Context Adaptive 71 21,9

5.2.2.2 Short texts

The null hypothesis regarding no difference in comprehension between the

conditions when reading short texts was kept. MS Explorer gave the best result but
the differences between the conditions were small (Table 7).

Table 7. Comprehension scores for short texts.

Condition Avg. % correct Std. dev.
MS Explorer 70 21,9
Fixed RSVP 66 26,0
Content Adaptive 59 19,9
Context Adaptive 66 18,9

5.2.2.3 Repeated use of RSVP

There were no significant differences in comprehension for either long or short texts
between using RSVP for the first, second or third time compared to using traditional
text presentation. The null hypothesis regarding no difference in comprehension
caused by repeated use of RSVP was thus kept.
 31

5.2.3 Task Load

Task load was calculated as percent of millimeters to the left of the tick mark on a
120-mm scale. The factors were not rated within each other.

5.2.3.1 Long texts

The null hypothesis regarding no difference in task load between the conditions
when reading long texts was rejected as all main factors except Physical demand
became significant (F[3,45]≥5.2, p≤0.014). Pairwise comparisons revealed that the
use of RSVP resulted in significantly higher (p≤0.014) task loads compared to using
traditional text presentation with the MS Reader (Figure 13).

100

90
Task load rating (% of mm on a 120-mm scale)

80

70

60

50 Mental

40 Physical

30 Temporal

20 Performance

10 Effort

0 Frustration
MS Reader Content adaptive
Fixed RSVP Context adaptive

Figure 13. NASA-TLX (Task Load Index) ratings for long texts with
median, 25- and 75-percentile represented. Lower ratings are better.

Content adaptive RSVP decreased task load ratings and the only factor that was
rated significantly higher compared to the MS Reader was Frustration level
(p=0.002). Context adaptive RSVP also decreased task load but in a different way.
The only significantly higher factor compared to the MS Reader was Temporal
demand (p=0.001).

5.2.3.2 Short texts

The null hypothesis regarding no difference in task load between the conditions
when reading short texts was kept. All conditions showed a task load comparable to
using RSVP when reading long texts (Figure 14).
 32

100

90

Task load rating (% of mm on a 120-mm scale)

80

70

60

50 Mental

40 Physical

30 Temporal

20 Performance

10 Effort

0 Frustration
MS Explorer Content adaptive
Fixed RSVP Context adaptive

Figure 14. NASA-TLX Task Load Index ratings for short texts with
median, 25- and 75-percentile represented. Lower ratings are better.

5.2.3.3 Repeated use of RSVP

The null hypothesis regarding no difference in task load caused by repeated use of
RSVP when reading long texts was rejected (F[3,45]≥4.5, p≤0.048). When using
RSVP for the first and second time all main task load factors except Physical
demand became significant compared to the MS Reader. However, the third time
RSVP was used Mental demand and perceived Performance and Effort levels
decreased enough not to become significant compared to the MS Reader (Figure 15).
For short texts the pairwise comparisons between using RSVP repeatedly and using
traditional text presentation did not result in any significant differences.

100

90
Task load rating (% of mm on a 120-mm scale)

80

70

60

50 Mental

40 Physical

30 Temporal

20 Performance

10 Effort

0 Frustration
1st RSVP 3rd RSVP
2nd RSVP MS Reader

Figure 15. NASA-TLX Task Load Index ratings for repeated use of
RSVP when reading long texts with median, the 25- and the 75-
percentile represented. Lower ratings are better.
 33

5.2.4 Attitude

Attitude ratings were calculated as percent of millimeters to the left of the tick mark
on a 100-mm scale.

5.2.4.1 Long texts

The null hypothesis regarding no difference in attitude between the conditions when
reading long texts was rejected. Five of eight main factors: Ease, Comprehension,
Immersion, Comfort and Naturalness became significant (F[3,45]≥8.3 p≤0.049).
Pairwise comparisons revealed that reading with RSVP resulted in lower ratings
compared to using traditional text presentation with the MS Reader (Figure 16).

100

90
Attitude rating (% of mm on a 100-mm scale)

80

70
Ease

60 Effect

50 Comprehension

40 Stimulus

30 Immersion

20 Comfort

10 Naturalness

0 Focus
MS Reader Content adaptive
Fixed RSVP Context adaptive

Figure 16. Attitude ratings for long texts with median, 25- and 75-
percentile represented. Higher ratings are better.

Regardless of RSVP condition perceived Ease, Comprehension, Immersion and

Naturalness was rated significantly lower compared to the MS Reader
(p≤0.049). Perceived Comfort was however only rated significantly lower for
Fixed RSVP and Context adaptive RSVP Compared the MS Reader (p≤0.023).

5.2.4.2 Short texts

The null hypothesis regarding no difference in attitude between the conditions when
reading short texts was partly rejected as the main factor for Ease became significant
(F[3,45]=4.1 p=0.047). Pairwise comparisons showed that the Ease of reading when
using Context adaptive RSVP was rated significantly lower compared to using the
MS Explorer (Figure 17).
 34

100

90

Attitude rating (% of mm on a 100-mm scale)

80

70
Ease

60 Effect

50 Comprehension

40 Stimulus

30 Immersion

20 Comfort

10 Natural

0 Focus
MS Explorer Content adaptive
Fixed RSVP Context adaptive

Figure 17. Attitude ratings for short texts with median, 25- and 75-
percentile represented. Higher ratings are better.

5.2.4.3 Repeated use of RSVP

The null hypothesis regarding no difference in attitude caused by repeated use of

RSVP was rejected (F[3,45]≥7.1 p≤0.047). Using RSVP when reading long texts for
the first and second time resulted in significantly lower ratings for Ease,
Comprehension, Immersion, Comfort and Naturalness compared to the MS Reader
the. However, the third time RSVP was used Comprehension and Comfort ratings
had increased enough to not become significant compared to the MS Reader (Figure
18).
For short texts the pairwise comparisons between using RSVP for the first, second or
third time and using traditional text presentation did not result in any significant
differences.
100

90
Attitude rating (% of mm on a 100-mm scale)

80

70
Ease

60 Effect

50 Comprehension

40 Stimulus

30 Immersion

20 Comfort

10 Naturalness

0 Focus
1:st RSVP 3:rd RSVP
2:nd RSVP MS Reader

Figure 18. Attitude ratings for RSVP when reading long texts with
median, 25- and 75-percentile represented. Higher ratings are better.
 35

6 Discussion

That no significant differences were found within the RSVP formats indicate that the
effects caused by adaptation were quite small. Nevertheless, when the results
obtained for RSVP were compared to those for traditional text presentation
significant differences were found. The discussion will first be based on these
findings and then some interface issues will be addressed. Finally some conclusions
and directions for further work will wrap up the thesis.

6.1 Reading Speed

Since no significant differences in reading speed were found for reading long texts
RSVP appears to be just as fast as traditional text presentation with the MS Reader.
The lower reading speeds obtained for short texts is not very surprising as news
articles are generally harder to read (Björnsson 1968; Tekfi 1987). However, the
significant differences between using RSVP and the MS Explorer is very surprising
since RSVP was found to increase reading speed by 33%. RSVP is usually not much
faster than traditional text presentation but these findings indicate that RSVP can
improve reading speed on a mobile device. That there were no significant
differences in reading speed due to repeated use of RSVP might imply that training
does not have any effect on reading speed. It may also indicate that the subjects
selected their own natural reading speed at all times.
When reading long texts using RSVP the resulting reading speed in this evaluation,
~250 wpm, was considerably slower than what Sicheritz (2000) and Goldstein et al.
(2001) reported from the Reader evaluation, ~320 wpm. The reason for this is
probably that the subjects were now instructed to select a comfortable speed rather
than a highest possible speed. In the author’s view RSVP is a primarily a way of
facilitating reading on small screens and not a way of optimising reading in itself.
The average reading speed for an adult Swedish reader on paper is around 240 wpm
(Björnsson 1968; Kump 1999) and in this light the obtained reading speeds are quite
encouraging.

6.2 Comprehension

That no significant differences in comprehension were found is in line with findings

from previous evaluations where blank screens were used (Joula et al. 1982; Masson
1983; Rahman and Muter 1999; Sicheritz 2000; Goldstein et al. 2001). The lack of
differences here at least shows that adaptation does not affect comprehension in
negative way.

6.3 Task Load

The task load ratings for Fixed RSVP and the MS Reader were close to identical to
those obtained for Fixed RSVP and paper-book in the Reader evaluation (Sicheritz
2000; Goldstein et al. 2001). This is surprising since the subjects now selected a
comfortable reading speed. This may imply that the size of the assumed trade-off
between reading speed and cognitive task load is small for the RSVP format, quite
contrary to the size of the well-established speed-accuracy trade-off (Wickens 1992).
 36

Adaptive RSVP was supposed to decrease task load and it seems to have worked as
expected when reading long texts. Compared to the MS Reader the only factor
significantly higher for Content adaptation was Frustration level. Although the
exposure times were directly related to the content some words were probably not
exposed for a duration that matched the time needed for adequate cognitive
processing. It is however encouraging that even the most straightforward form of
adaptation actually decreased task load for many factors.
In Context adaptive mode the only significant factor compared to the MS Reader
was Temporal demand. A probable cause for this is that the variations in exposure
time were too large. However, the relation between what was exposed and the time
for exposure was probably sound since the Frustration level decreased compared to
Content adaptation. It seems that although the variations were too large they
probably occurred at the right places.
Surprisingly there were no significant differences in task load when reading short
texts. When RSVP was used the task load ratings were almost equal to using the MS
Explorer although the reading speed was 33% higher. This indicates that traditional
text presentation neither guarantees low task load nor high reading speed when
screen size is limited. However, the lack of differences in task load when reading
short texts may also imply that the NASA-TLX Task Load Index (Hart and
Staveland 1988) does not give reliable ratings when the time for completing a task is
small.
The results for repeated use of RSVP seems to support this claim since a significant
decrease in task load was found for long texts but not for short. These findings imply
that training only makes a difference when reading long texts and that seems
contradictory. It seems more reasonable that repeated use should make it less
demanding to use RSVP with time regardless of the text length. The implication of
this is that the task load ratings obtained for the short texts may be unreliable and
that repeated use of RSVP most likely decreases task load.

6.4 Attitude

The obtained attitude ratings in this evaluation reveal a pattern similar to what
Sicheritz (2000) and Goldstein et al. (2001) reported from the Reader evaluation.
The subjects performed objectively very well but rated their performance as lower
than it really was. This is very likely to, at least partially, be a result of unfamiliarity
with the dynamic presentation format. A good indication of this seems to be that
both task load and attitude ratings improved with training.

6.5 Interface Issues Affecting Readability

The MS Reader performed much better than the MS Explorer even though both
applications had interfaces tailored to fit their intended use. The sole purpose of the
MS Reader is to present e-books (Hill 1999) and it seems to have an interface well
customized for extended reading. The use of a scroll-bar may partially explain the
lower reading speeds obtained for the MS Explorer but most likely the MS Reader
just facilitated reading better.
Giving the user full control over the presentation, as the Bailando prototype does,
has previously been shown to be counterproductive in terms of reading speed (Chen
and Chan 1990). The obtained reading speeds in this evaluation were however faster
than what both Rahman and Muter (1999), ~180 wpm, and Castellhano and Muter
 37

(2001), ~95 wpm, reported. In both these evaluations the reader’s control was
restricted in order to increase speed. The results reached here do however indicate
that giving the reader control over the text presentation might not be so
counterproductive after all.
The text presentation window width used by Bailando was chosen on basis of the
results from an earlier evaluation (Sicheritz 2000; Goldstein et al. 2001). Twentyfive
characters might however still have been too much. According to eye movement
research on English texts the eye can extract information in each fixation from
approximately 4 characters to the left and 14 to the right from each fixation point
(Rayner and Pollatsek 1989; Robeck and Wallace 1990; Rayner and Serano 1994;
Rayner 1998). When Bailando was used an average of 21 characters were displayed,
this may imply that the number of characters sometimes exceeded the eye’s
capability. In fact, as Castelhano and Muter (2001) point out, the effort required for
suppressing eye movements may actually cause the increase in cognitive load when
using RSVP. However, without proper eye movement recordings it is impossible to
know if, why and when any unwarranted eye movements occurred.
 38

7 Conclusions

The major drawback of RSVP appears to be the high cognitive demand placed on the
subjects. An increase in task load may actually be inherent to the RSVP format since
it remains high evidently independent of reading speed. Therefore, the most
important findings in this evaluation are that task load actually can be decreased by
using adaptation and through training.
Both adaptive algorithms were found to decrease task load for most factors. This
means that better adaptations may decrease task load even further. That task load
decreased with training seems to support that there is something about the RSVP
format that takes some time to get used to. The high initial demand is however still
likely to be a nuisance for the novel user. If adaptation can further lower initial task
load, then RSVP can be considered as a serious alternative for some appliances.
There is really no reason to use RSVP when traditional text presentation can be used
efficiently. In this evaluation RSVP was found to be just as effective as the MS
Reader but also significantly more demanding. However, when traditional text
presentation becomes ineffective it seems to become more demanding to use as well.
The MS Explorer was found to be just as demanding to use as RSVP but
significantly slower. It is probably when traditional text presentation, for one reason
or another, becomes inefficient that RSVP can constitute a realistic alternative. Both
formats are then likely to be demanding to use but compared to traditional text
presentation RSVP will be more effective.
When screen space is so limited that traditional text presentation becomes
ineffective, like it is on most mobile devices of today, RSVP appears to offer an
improvement in readability. The inherent increase in task load may also be
acceptable if it is compensated by an increase in efficiency.

7.1 Future work

Since even the most straightforward form of adaptation improved task load there is a
lot of interesting work that can be done in order to further improve adaptive RSVP.
Future studies on eye movements may offer some useful answers on the limits and
optimal conditions for the RSVP format. It would also be interesting to see how
RSVP works for different languages, especially for non-Latin languages like
Chinese or Arabic. However, most exciting would be to see an evaluation where
RSVP is compared to traditional text presentation on a mobile device with a smaller
screen than the one used in this evaluation.
 39

References

1. Björnsson, C.H. (1968). Läsbarhet, Stockholm, Liber.

2. Bruce, B., Rubin, A. and Starr, K. (1981). Why readability formulas fail.
IEEE Transactions on Professional Communication, PC-24, 50-52.
3. Bruijn, O. Spence, R. (2000). Rapid Serial Visual Presentation: A space-
time trade-off in information presentation. In Proceedings of Advanced
Visual Interfaces, AVI2000, 189-192.
4. Castelhano, M.S. and Muter, P. (2001). Optimizing the reading of
electronic text using rapid serial visual presentation. Behaviour &
Information Technology, 20(4), 237-247.
5. Chen, H.C. and Chan, K.T. (1990). Reading computer-displayed moving
text with and without self-control over the display rate. Behaviour &
Information Technology, 9(3), 467-477.
6. Cocklin, T.G., Ward, N.J., Chen, H.C. and Juola, J.F. (1984). Factors
influencing readability of rapid presented text segments. Memory &
Cognition, 12(5), 431–442.
7. Dillon, A., Richardson, J. and McKnight, C. (1990). The effect of display
size and text splitting on reading lengthy text from screen. Behaviour and
Information Technology, 9(3), 215-227.
8. Dillon, A. (1992). Reading From Paper Versus Screens: A Critical Review
of the Empirical Literature. Ergonomics, 35(10): 1297-1326.
9. Duchnicky, R.L. and Kolers, P.A. (1983). Readability of text scrolled on
visual display terminals as a function of window size. Human Factors, 25,
683-692.
10. Dodge, R. (1900). Visual perceptions during eye movement. Psychological
Review, VII, 454-465.
11. Ericsson, T., Chincholle, D. and Goldstein, M. (2001). Both the device and
the service influence WAP usability, IHM-HCI2001, Volume II, Usability
in Practice by J. Vanderdonckt, A. Blandford and A. Derycke (Eds.), Short
paper, 10-14 September, Lille, France, 79-85.
12. Fine, E.M. and Peli, E. (1995). Scrolled and rapid serial visual presentation
texts are read at similar rates by the visually impaired. Journal of Optical
Society of America, 12(10), 2286-2292.
13. Forster, K. I. (1970). Visual perception of rapidly presented word
sequences of varying complexity. Percep. Psychophys. 8, 215-221.
14. Goldstein, M., Sicheritz, K. and Anneroth, M., (2001). Reading from a
small display using the RSVP technique. Nordic Radio Symposium,
NRS01, Nynäshamn, Sweden.
15. Goldstein, M., Öqvist, G., Bayat-M, M., Ljungstrand, P. and Björk, S.
(2001). Enhancing the reading experience: Using adaptive and sonified
RSVP for reading on small displays. Proceedings of Workshop on Mobile
Devices at IHM-HCI 2001, Lille, September 2001.
16. Goldstein, M., Öquist, G. and Björk, S. (2001). Immersion does not
Guarantee Excitement: Evaluating Sonified Rapid Serial Visual
Presentation. Full paper submitted to Designing Interactive Systems,
DIS2002.
 40

17. Gould, J.D. and Grischkowsky, N. (1984). Doing the same Work with Hard
Copy and with Cathode-Ray (CRT) computer terminals. Human Factors,
26(3), 323-337.
18. Gould, J.D., Alfaro, L., Finn, R., Haupt, B. and Minuto, A. (1987). Reading
From CRT Displays Can Be as Fast as Reading From Paper. Human
Factors, 29(5), 497-517.
19. Granaas, M. M., McKay, T. D., Laham, R. D., Hurt, L. D., & Juola, J. F.
(1984). Reading moving text on a CRT screen. Human Factors, 26(1), 97-
104.
20. Hart, S.G. and Staveland, L.E. (1988). Development of Nasa-TLX (Task
Load Index): Results of empirical and theoretical research. Human Mental
Workload, by P.A. Hancock and N. Meshkati (eds.). Elsevier Science
Publishers, B.V.: North-Holland.
21. Hill, B. (1999). The Magic of Reading. Available at:
http://slate.msn.com/ebooks/ (December 2001).
22. Huey, E.B. (1968). The psychology and pedagogy of reading. Cambridge,
MA: MIT Press. (Originally published 1908).
23. Jackson, M.D. and McClelland, J.L. (1979). Processing determinants of
reading speed. Journal of experimental psychology, 108, 151–181.
24. Joula, J.F., Ward, N.J. and MacNamara, T. (1982). Visual search and
reading of rapid serial presentations of letter strings, words and text. J.
Exper. Psychol.: General, 111, 208-227.
25. Juola, J.F., Tiritoglu, A., and Pleunis, J. (1995). Reading text presented on a
small display. Applied Ergonomics, 26, 227-229.
26. Judd, C.H. and Buswell, G.T. (1922). Silent reading: A study of the various
types. Chicago, IL: University of Chicago Press.
27. Just, M. A., and Carpenter, P. A. (1980). A theory of reading: From eye
fixations to comprehension. Psychological Review, 87(4), 329-354.
28. Just, M.A., Carpenter, P.A. and Masson, M.E.J. (1982). What eye fixations
tell us about speed-reading and skimming. (Eye-lab Technical Report)
Carnegie-Mellon University.
29. Kang, T.J., and Muter, P. (1989). Reading Dynamically Displayed Text.
Behaviour & Information Technology, 1989, 8(1), 33-42.
30. Klare, G. R. (1984). Readability. In Handbook of Reading Research, P.D.
Pearson (Ed.), Longman Inc, New York, 681-743.
31. Kump, P. (1999). Break-trough rapid reading. New Jersey: Prentice-Hall
Press.
32. Masson, MEJ. (1983). Conceptual processing of text during skimming and
rapid sequential reading. Memory and Cognition, 11, 262-274.
33. McCrickard, D.S., Catrambone R. and Stasko, J.T. (2001). Evaluating
Animation in the Periphery as a Mechanism for Maintaining Awareness. In
proceedings of INTERACT 2001, 148-156.
34. Mills, C.B. and Weldon, L.J. (1987). Reading text from computer screens.
ACM Computing Surveys, 19(4), ACM Press.
35. Muter, P., Kruk, R. S., Buttigieg, M. A., and Kang, T. J. (1988). Reader-
controlled computerized presentation of text. Human Factors, 30, 473-486.
 41

36. Muter, P. and Maurutto, P. (1991). Reading and skimming from computer
screens and books: The paperless office revisited? Behavior & Information
Technology, 10, 257–266.
37. Muter, P. (1996). Interface Design and Optimization of Reading of
Continuous Text. In Cognitive aspects of electronic text processing. H. van
Oostendorp and S. de Mul (Eds.). Norwood, N.J.:Ablex.
38. Osborne, D.J. and Holton, D. (1988). Reading From Screen Vs. Paper:
There Is No Difference. International Journal of Man-Machine Studies, 28,
1-9.
39. Paulson, L.J. and Goodman, K.S. (1999). Influential Studies in Eye-
Movement Research. International Reading Association. Available at:
http://www.readingonline.org (December 2001).
40. Potter, M. C. (1984). Rapid Serial Visual Presentation (RSVP): A method
for studying language processing. In New Methods in Reading
Comprehension Research. D.E., Kieras and M.A., Just (Eds.). Hillsdale,
N.J: Erlbaum.
41. Proctor, R.W. and Proctor, J.D. (1997). Sensation and Perception. In G.
Salvendy (Ed.), Handbook of Human Factors and Ergonomics. Second
Edition, Wiley-Interscience, New York, 53-57.
42. Rahman, T. and Muter, P. (1999). Designing an interface to optimize
reading with small display windows. Human Factors, 1(1), 106-117,
Human Factors and Ergonomics Society.
43. Rayner, K. and Pollatsek, A. (1989). The psychology of reading.
Englewood Cliffs, New York: Prentice Hall.
44. Rayner, K. and Sereno, S.C. (1994). Eye movements in reading:
Psycholinguistic studies. In M.A. Gernsbacher (Ed.), Handbook of
psycholinguistics. San Diego, CA: Academic.
45. Rayner, K. (1998). Eye movements in reading and information processing:
20 years of research. Psychological Bulletin, 124, 372-422.
46. Redish, J. (1981). Understanding the limitations of readability formulas.
IEEE Transactions on Professional Communication, PC-24, 46-48.
47. Reichle, E.D., Rayner, K. and Pollatsek, A. (2000). Comparing the E-Z
Reader Model to Other Models of Eye Movement Control in Reading.
CogPrints electronic archive for papers. Available at:
http://cogprints.soton.ac.uk/ (December 2001).
48. Reilly, R. (1993). A connectionist framework for modeling eye-movement
control in reading. In G. d’Ydewalle & Van Rensbergen (Eds.), Perception
and Cognition: Advances in Eye Movement Research, 191-212.
Amsterdam: North Holland.
49. Robeck, M.C. and Wallace, R.R. (1990). The Psychology of Reading: An
Interdisciplinary Approach, Second edition, Lawrence Erlbaum Associates,
Hillsdale: New Jersey.
50. Russel, M., Hull, J. and Wesley, R. (2001). Reading with RSVP on a Small
Screen: Does Font Size Matter? Usability News, Winter 2001, Software
Usability Research Laboratory, Wichita State University.
51. Shneiderman, B. (1998). Human-Computer Interaction, 3rd ed., Addison
Wesley Longman, Inc, 412-414.
52. Sekey, A. & Tietz, J. (1982). Text display by saccadic scrolling. Visible
Language, 16, 62-76.
 42

53. Sicheritz, K. (2000). Applying the Rapid Serial Presentation Technique to

Personal Digital Assistants, Master’s Thesis, Department of Linguistics,
Uppsala University. Available at: http://stp.ling.uu.se/ (December 2001).
54. Taylor, I. and Taylor, M.M. (1983). The Psychology of Reading. Toronto,
Academic Press.
55. Tekfi, C. (1987). Readability formulas: An Overview, Journal of
Documentation, 43(3) 261-73.
56. Wickens, C. D. (1992). Engineering psychology and human performance,
2nd edition, Chapter 8, Harper Collins Publishers Inc., New York.
57. Öquist, G. and Goldstein, M. (2001). Towards an improved readability on
mobile devices: Evaluating Adaptive Rapid Serial Visual Presentation. Full
paper submitted to Advanced Visual Interfaces, AVI2002.

Web resources

1. A web-based LIX counter developed by the author can be found at:

http://stp.ling.uu.se/~gustav/lixcounter/ (December 2001).
2. The specifications of the eXtensible Markup Language (XML) can be
found at: http://www.w3.org/XML/ (December 2001).
3. Product specifications for the Compaq iPAQ can be found at the following
address: http://www.compaq.com/ipaq (December 2001).
4. Product specifications for Microsoft Embedded Visual C++ can be found
at: http://www.microsoft.com/mobile/developer/ (December 2001).
5. Information about the Microsoft Reader and ClearType technology can be
found at: http://www.microsoft.com/reader/ (December 2001).
6. Information about Microsoft Internet Explorer can be found at:
http://www.microsoft.com/mobile/ (December 2001).
7. The registration page for participation in the usability evaluation can be
found at: http://stp.ling.uu.se/~gustav/readability/ (December 2001).
 43

Appendix

A Excerpt from a Bailando log file

System: Bailando initierad

Settings: setwpm=250 avgword=7.8 avgsent=11.5 adaption=1
Aktion: Filen: Bara Alice öppnad
Fildata: Type=RSVPxml Rating=Easy Sent=310 Word=4172 Long=693 Char=18648 Lix=29 Rwpm=283
Aktion: Adaption satt till 2
Aktion: Läs
Legend: Text Exposure Act wpm Avg wpm Set wpm Adapt. Words Chars
Visar: Det enda Alice kan se av 695 517 517 250 2 6 24
Visar: sin far är sulorna på 824 364 434 250 2 5 21
Visar: hans skor. De är bleknat 1261 237 345 250 2 5 24
Visar: bruna, repade av grus 689 348 345 250 2 4 21
Visar: från de trottoarer där 778 308 339 250 2 4 22
Visar: han har gått. Alice 1016 236 319 250 2 4 19
Visar: brukar få lov att springa 711 421 331 250 2 5 25
Visar: på trottoaren utanför 820 219 317 250 2 3 21
Visar: deras hus för att möta 843 355 322 250 2 5 22
Visar: honom när han kommer hem 970 309 320 250 2 5 24
Visar: från arbetet på 675 266 316 250 2 3 15
Visar: kvällarna. När det är 1126 213 305 250 2 4 21
Visar: sommar springer hon 532 338 307 250 2 3 19
Visar: ibland där i bara 542 442 313 250 2 4 17
Visar: nattlinnet, och dess veck 1102 217 305 250 2 4 25
Aktion: Pause
Visar: fastnar runt hennes knän. 1497 160 289 250 2 4 25
Aktion: Gå bakåt
Visar: hon ibland där i bara 457 656 247 250 2 5 22
Aktion: Gå bakåt
Visar: nattlinnet, och dess veck 1033 232 246 250 2 4 25
Aktion: Gå bakåt
Visar: När det är sommar 346 693 211 250 2 4 18
Aktion: Sänker till 240 wpm
Aktion: Läs
Visar: honom när han kommer hem 572 524 179 240 2 5 25
Visar: från arbetet på 565 318 183 240 2 3 15
Visar: kvällarna. När det är 1073 223 186 240 2 4 21
Visar: sommar springer hon 567 317 190 240 2 3 19
Visar: ibland där i bara 578 415 196 240 2 4 17
Visar: nattlinnet, och dess veck 1176 204 197 240 2 4 25
Visar: fastnar runt hennes knän. 1580 151 193 240 2 4 25
Visar: 250 0 191 240 2
Visar: Men nu är det vinter. 853 351 197 240 2 5 21
 44

B Instructions

Välkommen

Du kommer under dagens försök att få läsa ett flertal texter under olika former och
svara på frågor i anslutning till detta. Det är viktigt att du är medveten om att det inte är
dig vi testar utan en applikation, du behöver därför inte känna någon press att göra
något på ett särskilt sätt. Vi vill helst att du använder applikationerna och läser så som
du skulle ha gjort i en helt vanlig vardagssituation.
Du får själv välja i vilken takt du ska läsa. Välj den hastighet som är så naturlig och
behaglig som möjligt för dig. Du får närsomhelst ändra hastigheten under tiden att du
läser. Du kan närsomhelst avbryta ett försök om du skulle uppleva någon form av
obehag. Om du har några frågor eller stöter på problem är det bara att fråga
närsomhelst under försöken.
Under experimentet kommer du att filmas och interaktionen mellan dig och
applikationen att loggas, detta är för att vi senare ska kunna gå tillbaka och se på i
detaljer hur det fungerat. Ingen annan än Ericsson Usability & Interaction Lab kommer
att ta del av det videoinspelade materialet utan ditt uttryckliga medgivande.
Du kommer att få börja med att öva att läsa en text i två olika applikationer så att du kan
vänja dig vid dessa. Sedan kommer du att få läsa fyra längre texter och fyra kortare
artiklar under olika förutsättningar.

Nedan ser du ungefär hur dagen är upplagd:

Förberedelse / övning 9.00 - 9.30

Första betingelsen 9.30 - 10.30
Fika 10.30 - 10.45
Andra betingelsen 10.45 - 11.45
Lunch 11.45 - 12.30
Tredje betingelsen 12.30 - 13.30
Fika 13.30 - 13.45
Fjärde betingelsen 13.45 - 14.45
Uppföljning 14.45 - 15.00

Börja med att fylla i de formulär som finns i detta häfte och så kan vi börja med
övningarna, vi hoppas att du kommer att få en trevlig dag hos oss!
 45

B1 Microsoft Reader:

Instruktioner

Du kommer nu att få läsa en längre text på ca. 4000 ord vilket tar ungefär 20 minuter.
Efter att du läst texten ska du svara på frågorna på nästföljande sidor.
Kom ihåg att du ska läsa med den hastighet som du finner så naturlig och behaglig som
möjligt.
Nedan ser du hur applikationen ser ut och vilka knappar som finns:

På och avknapp

Visar titel, namnet är Visar textposition

samtidigt menyknapp

Visar text

Byt sidaknapp

Börja inte innan du fått klartecken och får du problem eller undrar över något så är det
bara att fråga.

Lycka till!

Betingelse: B1 Text: T2 Person: P7 Moment: M7

 46

B2 Microsoft Internet Explorer

Instruktioner

Du kommer nu att få läsa en kortare artikel på ca. 500 ord vilket tar ungefär 2 minuter.
Efter att du läst texten ska du svara på frågorna på nästföljande sidor.
Kom ihåg att du ska läsa med den hastighet som du finner så naturlig och behaglig som
möjligt.
Nedan ser du hur applikationen ser ut och vilka knappar som finns:

På och avknapp

Visar titel Visar textposition

Visar text

Gå neråt och
uppåtknapp

Börja inte innan du fått klartecken och får du problem eller undrar över något så är det
bara att fråga.

Lycka till!

Betingelse: B1 Text: A2 Person: P7 Moment: M8

 47

B3 Bailando

Instruktioner

Du kommer nu att få läsa en längre text på ca. 4000 ord vilket tar ungefär 20 minuter.
Efter att du läst texten ska du svara på frågorna på nästföljande sidor.
Kom ihåg att du ska läsa med den hastighet som du finner så naturlig och behaglig som
möjligt.
Nedan ser du hur applikationen ser ut och vilka knappar som finns:

På och avknapp

Visar text

Visar titel
Visar textposition
Visar hastighet

Huvudmeny och Öka (+) och sänk (-)

bokmeny hastigheten

Start och paus

samt fram (>>)
och tillbaka (<<)

Börja inte innan du fått klartecken och får du problem eller undrar över något så är det
bara att fråga.

Lycka till!

Betingelse: B4 Text: T1 Person: P7 Moment: M5

 48

C Inventories

C1 Comprehension

Förståelse

Instruktioner: Välj det svar som du anser är riktigt på var och en av de 5

frågorna nedan.

1. Vad var klockan på morgonen då Thomas Murillo sköt ner rånarna?

- Tre
- Fyra
- Två

2. Hur gammal var Thomas Murillo när han sköt ner rånarna?

- 46 år
- 57 år
- 68 år

3. Hur många inbrott hade han haft tidigare?

- Fem
- Tre
- Två

4. Vad tände han medan han väntade på rånarna?

- En Kuban
- En Silviana
- En Kruger

5. Hur många skott sköt Thomás mot rånarna?

- Två
- Tre
- Fyra

Betingelse: B4 Text: T1 Person: P7 Moment: M5

 49

C2 NASA-TLX (Task Load Index)

Betingelse: B4 Text: T1 Person: P7 Moment: M5

 50

C3 Attittude

Värdering

Instruktioner: Markera för varje skala nedan dina skattningar för den nyss
genomförda uppgiften.

1. Hur upplevde du svårighetsgraden när du läste på detta sätt?

Låg Hög

2. Hur upplevde du effektiviteten när du läste på detta sätt?

Låg Hög

3. Hur upplevde du läsförståelsen när du läste på detta sätt?

Låg Hög

4. Hur upplevde du stimulansen när du läste på detta sätt?

Låg Hög

5. Hur upplevde du inlevelsen när du läste på detta sätt?

Låg Hög

6. Hur upplevde du bekvämligheten när du läste på detta sätt?

Låg Hög

7. Hur upplevde du naturligheten när du läste på detta sätt?

Låg Hög

8. Hur upplevde du koncentrationen när du läste på detta sätt?

Låg Hög

Betingelse: B4 Text: T1 Person: P7 Moment: M5

51