Amazing ability to read 2 times and more and utilize small screen devices to read large amounts. based on RSVP and long time research.
research conducted with 6th grade students using iPhone.
especially useful for developing countries
2. 2
line user-controlled scroll ('pull' method), while dynamic
refers to push methods where text (paragraph, line, word etc.)
is sequenced to the user. The text presentation formats
presented here are not applicable for small screens alone.
Often a lot of information must be squeezed into a small
window on a large screen.
D. Semantic priming
Semantic Priming refers to the fact that familiar words quickly
activate or “prime” their previously stored semantic
Figure 1: Perceptual span associations: subjects identify a target word (such as
After information is processed in a fixation, peripheral vision DOCTOR) more rapidly when its presentation is preceded by
is used to determine the location of the next fixation. A a relative word (e.g. NURSE) as opposed to an unrelated
saccade, a return sweep, is executed to move to the next “prime” word (e.g. TABLE). Response facilitation of
fixation target, and is not necessarily a forward movement. semantically related words (the priming effect) is believed to
Fixations take about ~230 msec on average for fast readers reflect the organization of concepts in memory: related words
and ~330 msec on average for average readers [80]. Text prime one another because they are stored closely together in
processing studies have shown that readers interpret texts one association network. Below 300 msec, between two inputs,
word, not one sentence, at a time. Semantic Priming refers to priming is automatic, however over 300 msec selective
the fact that familiar words quickly activate or “prime” their processing incurs a processing penalty and un-attentional
previously stored semantic associations. Longer fixations priming degrades [18] [26]. During the reading process, longer
(fixations vary between 100-500 msec) will have an impact on fixations (fixations vary between 100-500 msec) will have an
the priming effectiveness, and consequently comprehension. impact on the priming effectiveness, and consequently
B. Readability comprehension which is improved through priming.
Readability is typically evaluated in terms of reading speed E. RSVP Reading
and comprehension. Reading speed is often calculated as The term RSVP was first introduced by Forster (1970) as a
words read per minute (wpm), whereas comprehension is name for a technique used for studying text processing and
represented as percent of correctly answered multiple-choice comprehension. Later RSVP was introduced as a presentation
questions about the subject matter. The reading speed results technique for computer screens with the assumption that the
are more consistent when comparing results from repeated reduced need for eye movements could reduce cognitive load
performance of the same subject‟s on different texts, whereas and optimize readability ([9] [15] [22]). However, the term
comprehension scores are slightly unpredictable since they are RSVP has come to label a wide variety of approaches for text
highly dependent on the type of questions asked. The product presentation where chunks of text have been displayed
of the reading speed and comprehension scores has been successively. The designs of most RSVP evaluations and
suggested as a composite measure for reading efficiency [2] implementations have differed so much that the findings from
[5]. The measure is used to avoid problems associated with one evaluation are not necessarily applicable to another. RSVP
assumed trade-offs between speed and comprehension. was used for experiments about automatic processing during
rapid reading. The results showed that that readers read and
C. Small Screen Reading
understand RSVP texts presented as fast as 600 WPM – more
Most mobile devices utilize flat Liquid Crystal Display (LCD) than twice the page reading rate [85] .Processing and
screens. Today LCDs offer a good resolution and color depth. comprehension are probably faster during RSVP reading
The problem with readability on small screens is however not because serial presentation eliminates the need for time
so much the resolution, but rather the limitation in the screen consuming strategies readers ordinarily employ when they
space. This limitation restricts the amount of information that decide when and where to move their eye during page reading.
can be presented at one time thus reading a text on a small Karin Sicheritz implemented an RSVP reader in order to
screen can be frustrating. To complicate matters, users of evaluate how-to read using RSVP on a PDA compared to
mobile devices do not always have access to printing facilities. using paper-based text [5]. The application was implemented
Studies have been done on the effect of display size on reading on a Casio Cassiopeia E-105 PDA and offered a graphical user
in order to determine how small a screen can be before interface. In a repeated-measurement, within-subject
problems occur. Duchnicky and Kolers (1983) performed an experiment, using ten subjects, the RSVP reader was
experiment with varying text window widths and heights and benchmarked against the paper-based text. Different window
found that, readability with window heights that are four lines widths for the RSVP reader, 11 and 25 characters, were
and smaller with width of 1/3 of whole page, were compared to a paper-based text condition. The texts used in
significantly less efficient (highly corresponds to mobile the experiment were the first six chapters from the novel
devices) [4]. “Röda Rummet” (in Swedish by August Strindberg), the
There are several methods to present text on small screens. chapters were between ~2700–6300 words long. The subjects
The methods can be divided into static and dynamic text read the first chapter in the paper-based text and the following
presentation formats. The major difference between the chapters using the RSVP reader prototype. The subjects were
formats is that traditional text presentation refers to a multi- instructed to read as fast as possible. Readability was
3. 3
measured by reading speed, comprehension inventories Each subject was tested with one of the Three experimental
consisting of ten multiple-choice questions, the NASA-TLX conditions (I, II and III) only, 20 subjects for each
(Task Load Index) [82] task load inventory, and an attitude experimental condition, 60 subjects in total. The subjects were
inventory consisting of five questions about difficulty, assigned to one of the three groups at random.
efficiency, comprehension, stimulation and facilitation of the
presentation [82] [5]. Though a text presentation width of 25
characters resulted in the highest reading efficiency, the The task presented to each subject was identical different only
differences were not significant. However, the task load in the reading software used and included:
inventory did reveal significantly higher task load ratings Introduction – An explanation of the procedure and
for all RSVP conditions for all factors but Physical the questionnaires.
demand [82] [5]. Familiarization Text Reading – familiarization text
A later experiment was performed using a Compaq iPAQ was read on the device by each subject, this was done
3630. The initial speed of the text presentation was set to 250 to adjust to the device and reduce initial anxiety.
wpm, but the subjects were allowed to alter the speed at any Test Text Reading – Each subject was asked to read
time. A commercial program was chosen for the traditional identical test text. The WPM (words per minute) rate
text presentation - Microsoft Reader version 1.0. The results was calculated according to the total amount of
showed that the use of RSVP resulted in significantly higher words in the text divided by the end time minus the
task loads when compared to traditional text presentation with start time (as measured by the device).
Microsoft Reader. In addition, regardless of the RSVP Unseen Comprehension – Each subject answered an
condition, in the test the perceived ease, comprehension, ease, unseen comprehension test with 12 multiple choice
comprehension, immersion, and naturalness were rated questions.
significantly lower compared to Microsoft Reader [19] Task Load (derived from NASA-TLX) – Each
F. Gestural Control subject answered a task load questionnaire and a
frustration measuring question as part of it.
Quek et al. (2002) have proposed a framework for classifying
gestures for human-computer interactions, into three „Future Reference‟ Questionnaire – the subject
categories: manipulation, semaphores and gesture-speech completed a set of multiple choice questions on the
approaches [24]. The relevant gesture category for this work is possible use of the reading technique they have
semaphoric: "systems of signaling using flags, lights or arms" experienced in their everyday life and education in
(Brittanica.com). By extension, we define semaphoric gestures the future.
to be any gesturing system that employs a dictionary of hand B. Design
or arm gestures. Semaphoric gestures are linked with the move The null hypotheses were as follows:
towards more ubiquitous computing paradigms and are a
No difference in reading speed across the three
means of reducing distraction to a primary task when
performing secondary task interactions (e.g. locating region on reading methods
satellite image – primary task, while monitoring an error No difference in text comprehension across the three
terminal – secondary task) [17]. reading methods
No difference in task load across the three reading
methods
IV. THE RESEARCH
No difference in frustration level across the three
In Gestural RSVP, gestures manifested by spatial movements reading methods
of the reading device control the speed of the word
presentation and the pause/play control. Research was The hypotheses were tested via the repeated-measurement
conducted with 6th grade students to evaluate whether gestural General Linear Model (GLM). The significance level was set
control will provide a low cognitive load means for controlling to 5%
RSVP reading and thus lower the frustration level. Lowering
the frustration level might enable RSVP to become a popular The data was collected between groups; the hypotheses were
way of reading on small screens and mobile devices. tested based on the following measures:
A. Method Reading Speed – based on the wpm (word per
Sixty subjects were split into three groups: minute) rate in reading the test text.
Group I – „Standard reading‟: texts presented using a Comprehension – percent of correctly answered
standard text reader for iPod touch (mail reader) success multiple choice domain matter questions.
Group II – „Standard RSVP‟: text presented using Task Load – The measure of the task load was based
standard RSVP with button control as presented in on the answers to the Task Load inventory which was
the relevant research [82] [5] [19]. administered to check mental, temporal demands, as
well as performance level for reading the text.
Group III – „Gestural RSVP‟: text presented using
RSVP and controlled by gestural control. Frustration question – this measure was taken to
check the frustration level during reading the text.
4. 4
The future reference test was used for additional perspective was little difference between the two RSVP methods. The null
and future research; however it was not used for the null hypothesis was not kept, but the direction of the change favors
hypothesis testing. the RSVP methods.
C. Apparatus Reading speed Average Standard
Deviation
Regular 15.25
All experiments were performed on five iPod touch second reading 81.83
generation 8GB devices. They were all running version 2.2.1 RSVP 112.39 39.18
of the iPod/iPhone OS. The prototype software was installed GRSVP 114.00 37.90
on the devices and was used for all RSVP conditions. The
initial speed of the text presentation was always set to 92 wpm
To test the null hypothesis, a differential analysis was
(assuming no punctuation), but the subjects were encouraged
performed between the three groups using the WPM data. A
to set a suitable speed (by adjusting the speed multiple times
significant difference was found between the RSVP groups
until they felt it was right for them) in the training session.
and the regular reading group (f=6.15 p<0.01).
Altering text speed for the Standard RSVP as well as Gestural
RSVP conditions was allowed at any time, also after training B. Comprehension
in the test session. The built in Apple mail reader was used for Comprehension was calculated by the percentage of correctly
the regular text reading. The mail reader allows easy scrolling answered multiple-choice questions in the multiple choice
with hand stroke and zoom in and out. comprehension tests.
The null hypothesis concerning no difference in
comprehension between the conditions when reading texts was
kept. Both RSVP methods showed a somewhat higher level of
comprehension (10%) as compared to the standard reading
method. There was no difference between the two RSVP
methods. However this change is not significant enough to
reject the null hypothesis (F=2.135, P=0.128).
Comprehension Average Standard
Deviation
Regular reading 72% 15%
RSVP 80% 13%
Figure 2: Gestural control implementation
GRSVP 80% 14%
V. RESULTS
All subjects completed the experiment. There were only a few C. Task Load
problems with understanding what had to be done or how to Task load was produced by combining three task load
do it. Although RSVP was a new way of reading for all the questions (mental effort, time pressure, and performance
subjects, no one had any problems using the RSVP program. measure). Each one was assigned a value in the range of 1 to 9
In addition, although iPod Touch gestural control was new for on the Likert scale. The null hypothesis concerning the task
all the subjects, understanding the mechanism and operating it load was kept, although the standard reading method showed a
was easily learned. During the familiarization, some subjects somewhat lower task load with no significance (F=1.048,
pressed the Program button which closed the application. P=0.357). A lower standard deviation for the standard reading
However, after explaining and emphasizing this principal method shows stable results around the average while the
during the familiarization – no such incidents occurred with RSVP methods had higher standard deviations, indicating
the actual test text. higher fluctuations.
The presentation of the results is divided into five sections: Task load Average Standard
Reading Speed, Comprehension, Task Load, Frustration, and Deviation
„Future Reference‟ Questionnaire. Under each section, except Regular reading
the future reference questionnaire, the null hypotheses are (A) 2.60 1.244
tested. RSVP (B) 3.05 1.643
A. Reading speed GRSVP (C) 3.03 1.678
Reading speed was calculated based on the total amount of D. Frustration
words read in the text divided by the end time minus start The null hypothesis concerning no difference in frustration
time, including all interruptions (pauses, regressions, speed between the conditions was rejected since it became
changes, etc.). Based on the calculation: 262 ÷ significant (F=23.407, p≤0.01). Comparisons of pairs revealed
(������������������������������������������������������ ÷ 60) that the use of the standard RSVP method resulted in
The reading was significantly faster for both RSVP methods significantly higher (p≤0.05) frustration compared to using
as compared to the standard reading method. However, there
5. 5
regular reading. Comparison of pairs also revealed that the use In the our view, the RSVP method is primarily a way of
of Gestural RSVP had no significant difference from the facilitating reading on small screens and not a way of
standard reading method and had a significant difference from optimizing reading in itself. However, reading in the near
the standard RSVP method. The frustration level can be future may become more mobile and more prone to small
generalized as: "Regular Reading" ~= GRSVP << RSVP. screens. Children will be more exposed to small screen
devices due to their life style. The obtained reading speed -
Frustration Average Standard 50% faster is quite encouraging in view of future needs.
Deviation B. Comprehension
Regular reading
(A) 1.85 1.137 The result that no significant differences in comprehension
RSVP (B) 4.25 1.552 were found is consistent with findings from previous
GRSVP (C) 1.8 1.152 evaluations ([9] [15] [5] [82] [5]). The lack of differences
shows that the GRSVP method at least does not affect
comprehension in a negative way. Comprehension rating at
the levels reported in the results (72-80%) establishes that the
nature of reading have not been degraded to a high-speed low
comprehension method, such as skimming (skimming in
reading standard texts is faster however reports on average
50% comprehension).
C. Task Load
The task-load parameters (not including frustration) did not
show any significant difference between the standard reading
method, the RSVP method, and the GRSVP method. This is
consistent with previous findings [19]. There was, however, a
non-significant lower task load average for the standard
reading method as compared to the RSVP-based methods.
This may be because the standard reading method is more
familiar to the subjects than the RSVP method. Additional
training and usage with the RSVP method may eliminate this
non-significant difference. The results on the Likert 1–9 scale
(increase indicates task load growth) are on average around 3,
Figure 3: The task load inventory including frustration which indicates a low-mid task load. These results are
encouraging for general readability on small screens and
mobile devices for longer texts SMS.
VI. DISCUSSION
D. Frustration
The results show a significant decrease in frustration level
between the GRSVP and RSVP method and a 50% faster Although research results show a significant speed increase
reading speed for the GRSVP method over the standard and although there are obvious advantages in small screens
reading method. The results also show no significance in (e.g. mobile phones), in our opinion, one of the key issues for
comprehension and task load index. These results are in line the low commercial usage of the RSVP method is due, in large
with the basic assumptions for this research. This discussion is part, to the frustration factor. The frustration factor may be
based on these findings. caused by the perceived lack of control over the reading
process as compared to the standard reading method. The
A. Reading Speed significant change in frustration was reported for small screen
The reading speed for Hebrew readers (the same as English mobile devices in previous research [82] [5] [19]. For the
readers) is 100–200 wpm for learning and 200–300 wpm for frustration factor, the significant change between the RSVP
comprehension. This corresponds to the average reading and GRSVP methods and the similarity between the GRSVP
speeds in English. The lower reading speeds obtained for and standard reading methods are very important. The results
children in this research can be attributed to two combined indicate that the standard reading and GRSVP methods
reasons. The first is that the reading skills of sixth graders are received low average level while the RSVP method reached a
still not as mature as adults. The second is the context of this medium level (and significantly higher). The gestural control
research to the subjects was learning, which implies that the and training are assumed to have been formed as procedural
relevant range is 100–200. The significant differences between learning, such that the executing it has low cognitive load and
using the RSVP method and the standard reading method hence low frustration level.
indicate that the RSVP method can improve reading speed on E. Future Use
a mobile device; this is similar to the results reported for the
RSVP method on desktops by Bailey, R.W. and Bailey, L.M. The common findings were that the excitement attributed to
(1999). reading as done in the research, namely with a mobile device
like the iPod touch was high. Fifty percent and above rated the
6. 6
experiment experience (with no significance between groups)
better than reading texts in books or desktops. They also noted
they would read more for fun and for education than they read
today with the new reading methods. This result may be due to
temporary enthusiasm attributed to the iPod touch device used.
On the other hand, the subjects are frequent users of mobile
phones and gadgets and the lack of significant differences
between the groups may be a result of each subject
participates only in one test condition and could not compare.
In any case, this fact may suggest that mobile-device reading
is very attractive to the young, dynamic generation; for young
people the availability of a mobile device and its ease of use
make it very attractive.
This research sets to investigate the ability of Gestural RSVP
to make RSVP reading method widely used for small screen Figure 4: GRSVP reader used in the experiment
device. Changing life-long habit like reading might prove
difficult, the future use questionnaire might outline that school
VIII. REFERENCES
children can adopt it willingly.
VII. CONCLUSIONS [1] Bruijn, O. Spence, R. (2000). Rapid Serial Visual
Presentation: A space-time trade-off in information
presentation. In Proceedings of Advanced Visual Interfaces,
The major drawback of the RSVP method appears to be the AVI2000, 189-192.
high frustration felt by the subjects using it. An increase in
frustration is attributed to the perception that control is lost as [2] Castelhano, M.S. and Muter, P. (2001). Optimizing the
compared to the page-like reading process. Therefore, the reading of electronic text using rapid serial visual presentation.
most important finding in this research is that frustration with Behaviour & Information Technology, 20(4), 237-247.
the RSVP method can be reduced to the level of the standard
reading method by using the Gestural control over RSVP [3] Cocklin, T.G., Ward, N.J., Chen, H.C. and Juola, J.F.
reading. (1984). Factors influencing readability of rapid presented text
The initial training and habit change is likely to be a nuisance segments. Memory & Cognition, 12(5), 431–442.
for the new user, a factor which increases with age due to life
habits. There might not be any reason to use the RSVP method [4] Duchnicky, R.L. and Kolers, P.A. (1983). Readability of
when traditional text presentation can be used efficiently. In text scrolled on visual display terminals as a function of
this research the GRSVP method was found to be just as window size. Human Factors, 25, 683-692.
effective as the standard reading method but GRSVP enabled
a significantly faster reading speed. As part of the modern life [5] Goldstein, M., Sicheritz, K. and Anneroth, M., (2001).
style, society is adopting small screen mobile devices for Reading from a small display using the RSVP technique.
extensive usage. RSVP with the Gestural control can provide a Nordic Radio Symposium, NRS01, Nynäshamn, Sweden.
viable method for enabling reading high volume texts and not
only reading short messages (e.g. SMS). [6] Granaas, M. M., McKay, T. D., Laham, R. D., Hurt, L. D.,
The life-style technology is moving ahead in giant steps. & Juola, J. F. (1984). Reading moving text on a CRT screen.
However, reading failed until fairly recently to use the “new” Human Factors, 26(1), 97-104.
technology effectively; there were a few unsuccessful reading
appliances. Yet, recently, reading devices are attracting [7] Hart, S.G. and Staveland, L.E. (1988). Development of
attention and recent effective ones, such as Amazon™ NASA-TLX (Task Load Index): Results of empirical and
Kindle™, are succeeding commercially. The combination of theoretical research. Human Mental Workload, by P.A.
mobile device proliferation and the ability for reading devices Hancock and N. Meshkati (eds.). Elsevier Science Publishers,
to take advantage of technology (presentation, storage etc.) B.V.: North-Holland.
may become the setting for educating a new generation to read
in a more efficient and economical way. [8] Heritz, 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).
[9] 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.
7. 7
[10] Juola, J.F., Tiritoglu, A., and Pleunis, J. (1995). Reading [24] Quek, F., McNeill, D., Bryll, R., Duncan, S., Ma, X.-F.,
text presented on a small display. Applied Ergonomics, 26, Kirbas, C., McCullough, K. E., and Ansari, R. 2002.
227-229. Multimodal human discourse: gesture and speech. ACM
Trans. Comput.Hum. Interact. 9, 3, 171–193.
[11] Just, M. A., and Carpenter, P. A. (1980). A theory of
reading: From eye fixations to comprehension. Psychological [25] Rahman, T. and Muter, P. (1999). Designing an interface
Review, 87(4), 329-354. to optimize reading with small display windows. Human
Factors, 1(1), 106-117, Human Factors and Ergonomics
[12] Just, M.A., Carpenter, P.A. and Masson, M.E.J. (1982). Society.
What eye fixations tell us about speed-reading and skimming.
(Eye-lab Technical Report) Carnegie-Mellon University. [26] Ratcliff, R., &McKoon, G.(1981). Automatic and
strategic priming in recognition. Journal of Verbal Learning
[13] Karam, M. and m. c. schraefel. 2005. A study on the use and Verbal Behavior, 20, 204-215
of semaphoric gestures to support secondary task interactions.
In CHI ‟05: CHI ‟05 extended abstracts on Human factors in [27] Rayner, K. and Pollatsek, A. (1989). The psychology of
computing systems. ACM Press, New York, NY, USA, 1961– reading. Englewood Cliffs, New York: Prentice Hall.
1964
[28] Rayner, K. (1998). Eye movements in reading and
[14] Kump, P. (1999). Break-through rapid reading. New information processing: 20 years of research. Psychological
Jersey: Prentice-Hall Press. Bulletin, 124, 372-422.
[15] Masson, MEJ. (1983). Conceptual processing of text [29] Robeck, M.C. and Wallace, R.R. (1990). The Psychology
during skimming and rapid sequential reading. Memory and of Reading: An Interdisciplinary Approach, Second edition,
Cognition, 11, 262-274. Lawrence Erlbaum Associates, Hillsdale: New Jersey.
[16] Muter, P., Kruk, R. S., Buttigieg, M. A., and Kang, T. J. [30] Russel, M., Hull, J. and Wesley, R. (2001). Reading with
(1988). Reader-controlled computerized presentation of text. RSVP on a Small Screen: Does Font Size Matter? Usability
Human Factors, 30, 473-486. News, Winter 2001, Software Usability Research Laboratory,
Wichita State University.
[17] Muter, P. (1996). Interface Design and Optimization of
Reading of Continuous Text. In Cognitive aspects of [31] Seigel A.C (1994) Automatic Processing during Rapid
electronic text processing. H. van Oostendorp and S. de Mul Reading: Understanding Point of View, PhD dissertation.
(Eds.). Norwood, N.J.:Ablex.
[32] Shneiderman, B. (1998). Human-Computer Interaction,
[18] Neely, J. H. (1978) Semantic priming and retrieval from 3rd ed., Addison Wesley Longman, Inc, 412-414.
lexical memory. Journal of Experimental Psychology: General
106, 226-254
[19] Öquist, G (2001) Adaptive Rapid Serial Visual
Presentation, Master‟s Thesis.
[20] Öquist, G. and Goldstein, M. (2002). Towards an
improved readability on mobile devices: Evaluating Adaptive
Rapid Serial Visual Presentation. Full paper submitted to
Advanced Visual Interfaces, AVI2002.
[21] 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.
[22] 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.
[23] 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.