Upload
nguyenhuong
View
217
Download
1
Embed Size (px)
Citation preview
TO WHAT EXTENT DOES SCREEN READER SOFTWARE SUPPORT THE INFORMATION SEEKING BEHAVIOUR OF
VISUALLY IMPAIRED PEOPLE
A study submitted in partial fulfilment of the requirements for the degree of
Master of Science in Information Systems
at
The University of Sheffield
by
Alisdair Connell
September 2002
ACKNOWLEDGEMENTS
I would like to take this opportunity to thank Alan Thorpes, Steve Hambleton and
Micheline Beaulieu, who have generously given their time and support throughout
this research and without whom it would not have been possible.
Alan Thorpes who has generously given his time to help in finding participants and
providing access to a PC installed with the JAWS screen-reader software at the
Ravenscroft Resource Centre.
Steve Hambleton of the Sheffield Royal Society for the Blind (SRSB) has kindly
helped to find participants, recommend some relevant web sites, and develop the early
stages of the research.
Micheline Beaulieu has provided supervision and invaluable contribution towards the
development of ideas throughout the research project, from development of the
methodology to analysis of the results.
CONTENTS
Page 1. Abstract
Page 3. Introduction
Page 5. Literature Review
-Page 7. Visual Impairment and Assistive Information Technology
-Page 11. Web Design and Standardisation
-Page 13. Beyond Access
-Page 15. Functional Characteristics of The Screen Reader Interface
Page 19. Job Access With Speech (JAWS) Version 3.7
-Page 2. Important Features
-Page 22. D'Amour, and Roy (2002) Critique of JAWS
Page 24. Method
-Page 25. Research Question
-Page 26. Aims
-Page 27. Objectives
-Page 28. General methodological Framework
-Page 29. Procedure
-Page 32. Sample
-Page 33. Participant Profiles
-Page 37. Semi-Structured Interview
-Page 39. Demonstration
Page 41. Findings
-Page 42. Context of Usage
-Page 44. Extent of JAWS Support for Internet Explorer and HTML
-Page 48. Web Design and The Web Environment
-Page 51. Fundamental Limitations of The Screen Reader Interface
-Page 58. Transferable Recommendations
-Page 62 Further Research
-Page 64. References
Page 68 Appendices
-Page 79 Acronyms
-Page 70 Interview Summaries
ABSTRACT
Background:
The screen reader is the technology upon which visually impaired people are reliant to
use the Internet. There may be some problems with the functionality of the interface
it provides.
Research Question:
To what extent does screen reader software support the information seeking behaviour
of visually impaired people?
Method:
Qualitative interpretative analysis of narrative, generated by interviews and
demonstrations with users.
Procedure:
Consisted of three parts, all fused into one discussion:
1. A semi-structured interview addressing; context of use, functional comparison to
alternative technologies, general usability issues and an exploration of specific
problems and positive features of screen reader software.
2. A demonstration of use of the Internet with the JAWS screen reader, exploring
functionality and features of the software.
3. A profile of personal characteristics.
Sample:
10 visually impaired people who use the Internet with a screen reader. Basically
opportunistic, though some purposive selection was used to provide comparison of
different age groups and different levels of familiarity with the technology.
Findings:
- Inaccessible web design is the biggest problem with using the Internet with a
screen reader.
1
- Speech is a functional mode for processing continuous narrative. But there some
limits to the preservation of meaning encapsulated in structural complexity and
graphics
- JAWS does not support of information elements supported by HTML and this
presents a conflict with attempts to standardise web design.
- Speed and fatigue are critical factors in deciding ones preferred interface.
- The screen reader presents problems for older users with confounding sensory
impairments or memory problems.
- An attempt to distinguish between usability and accessibility is advised for
research into assistive technologies and interface design.
- The importance of facilitating scanning is paramount for any technology that
supports use of the Internet.
- Advanced features of JAWS are underused, so training in these features may be
beneficial.
2
INTRODUCTION
The idea for this research evolved from an interest in speech interfaces technology
and a suggestion that usability of speech interfaces technologies for visually impaired
people might be worthy of research.
Following this, a review of relevant literature and discussions with Steve Hambleton
of the Sheffield Royal Society for The Blind (SRSB) and Micheline Beaulieu, helped
to clarify and refine the research question and develop appropriate methodology.
Williamson (2000) highlights the importance and potential of the Internet to remedy
information inequities in information provision and Berry (1999) notes how the
screen reader is the assistive technology through which visually impaired people
access the Internet. Hill (1996) provided a previous dissertation on visually impaired
access to Library and Information Services in Sheffield to critique. Though focussed
more on the quality of information provision by this organisation than the usability
technical interface solutions, the use of qualitative methodology employed helped to
guide the formulation of methodology for this research.
Originally, a comparison between voice navigable interfaces for Web based
applications (voice-web), accessible through the traditional telephone, were to be
compared with the screen reader interface. However, the relevance to visually
impaired people of voice-web applications was considered questionable and the
limited availability of voice –web applications to specific application domains (news,
weather) was not thought conducive to a realistic comparison between the two.
Steve Hambleton discussed the importance of the dominant software tool for using the
PC and accessing the Internet; the screen reader. Berry notes how screen readers
“could not be described as easy to use by the non-technical user”, Moreover Nielsen
(2001) notes how Internet users reliant on a screen reader are less likely to be
successful in their information-seeking task than sighted users. Clearly there are
limitations and problems with the screen reader. What are they?
3
It is a more complete understanding of the limitations and usability that the following
research question aims to address:
To what extent does screen reader software support the information seeking of
visually impaired people?
Inevitable this research question is fuzzy and open. Indeed this can be considered
strength, if the nature of its fuzziness is analysed and understood, since clarification of
a fuzzy, open question is in the interest of a holistic understanding.
The question encompasses the relationship between provision of content and usability
of the interface that provides a medium to present that content. The need to recognise
this was noted by Date 1996 “access to premises is not enough” from Hill.
Furthermore, it must be recognised that the interface for screen reader access to the
Internet is an emergent property of both the software and web design. And so the
interaction between these two factors is particularly important. In an environment of
complex interactions where insights into human behaviour are being examined, a
qualitative method permits a holistic understanding of the situation.
A semi-structured interview was deemed to provide the best framework for
interpretivistic analysis. This allows for the development and capture of narrative in
unforeseen directions, while still providing some structure in the form of questions
that provide starting points from which to explore relevant issues. Inductive analysis
and interpretation of the narrative generated by these interviews was employed to
draw some conclusions and transferable recommendations surrounding the usability
of screen reader software.
4
LITERATURE
REVIEW
5
Visual Impairment and Assistive Information Technology
Visual Impairment Demographics
People with visual impairment make up a significant proportion of the population in
this country. Types of visual impairment vary but in the UK; “a person is registrable
as blind if she can only see the top letter of the eye test chart (used by opticians and
doctors) or less, at a distance of three metres” and “a person is registrable as partially
sighted if she can only read the top letter of the eye test chart at a distance of six
metres” (RNIB). There were 190,322 registered blind people and 157,683 registered
partially sighted people in Great Britain in 1997 but the registration is voluntary and
the RNIB estimate that “almost a million people are blind or partially sighted” in this
country, with at least 42 million blind worldwide.
Any effort to assist people with sight problems must also consider the skewed age
demographics of blindness. “70 per cent of people with impaired vision are over 75”
and “Six in every ten visually impaired people have another serious illness or
disability such as arthritis or deafness. Many have more than one other disability”
(RNIB). These multiple problems limit the extent to which other non-visual modes
(tactile or auditory) can be used to provide alternative means of encoding and
receiving information.
Information Inequality
So how do visually impaired people seek information? To some extent they rely on
other people but information technologies that are accessible and usable by visually
impaired people present an important and useful means through which to find
independence and autonomy. Only 25% of visually impaired people of working age
are employed (RNIB) and Cawthra, L. (1999) finds a specific deficit in provision of
sexual health information for blind adolescents. Evidence (if it were needed) that
such fundamental issues as health and employment are compromised by information
inequality for people with visual impairment. But society’s reliance on visual media
to communicate, dictates that information inequality is not restricted to any particular
domain, permeating every aspects of visually impaired peoples’ lives.
6
Assistive Technology for Presenting Text
The RNIB highlight this inequity by describing the access to literature as a “right” and
raising awareness of the denial of this right through their “right to read week”.
Clearly then, “reading” is defined here, as something beyond a visual experience,
something more abstract. Accordingly, there are a multiple of types of reading
experience that may have important differences between them. The following are the
most widely used:
Enlarged Visual Text
Many visually impaired people are able to read large visual text. This can be
presented in the form of large print, screen magnification software on a PC, or CCTV
devices that display an enlarged image of paper print on the screen.
Braille
Braille is a tactile form of text consisting of characters represented by 6 raised dots. It
is the closest non-visual alternative to visual text, and read (by experienced Braille
readers) at the same speed as sighted readers, but very few blind people can read
Braille (19,000 according to an RNIB estimate). There is also an interface output
device called a ‘refreshable Braille line’, which provides a dynamic Braille output
display for a PC.
Audio
Audio tapes or CD recordings of literature are available. But recording is a laborious
and expensive process and so audio versions of all written communications are not
widely available and are often out of date for dynamic information domains such as
news and weather.
Arkenstone/Kurzweil Readers
Text To Speech technology such as Kurzweil 3000 software can be used in
conjunction with a scanner to convert a physical copy of visual text literature into
speech and has been successfully used in libraries (as noted by Hill, 1996). However,
as Hill points out, actually obtaining visual literature presents significant problems for
visually impaired people.
7
Responsibility
Since the availability of accessible literature can be considered a right, the concept of
a duty to fulfil that right becomes important in this context. The RNIB, on their web
site, discuss how the responsibility for provision of accessible literature to the blind is
indistinct and therefore patchy. Traditionally commercial publishing has focussed on
paper print, it is widely recognised that local authorities fail to provide Braille or
audio copies of literature and often do not offer the complete service that is required.
Moreover the legal protections afforded to owners of copyright are extending to
electronic copies.
“Electronic publishing and digital rights are moving in a direction which will further
exclude blind people”. (Marriot 2002)
The Internet however, offers another important source of information, with some
advantages over paper or audio alternatives; it can have regularly updated content and
there is no need to struggle around a physical repository of literature. Documents can
be retrieved on-line, without sighted assistance.
The Internet and The Screen Reader
The World Wide Web to date has mostly involved HTML files accessed using
graphical web browsers on a PC. The availability of a wealth of electronic text
creates huge potential to provide more equitable access to information for the blind
and visually impaired (since technology can present it in auditory form). However,
the emergence of the Internet into the world of visually impaired people may not be a
panacea for inequities in information access. Williamson et al (2000) notes how
technology offers an alternative to reliance on people. But, “Along with the
enthusiasm about the value of the Internet for people with disabilities, there also
seems to be a lot of hype” (Williamson et al, 2000). Indeed the development of the
Internet has been described as a double-edged sword, since many web sites are not
possible to use with a screen reader. Is the potential threat of the Internet to redefine
the way people find information really such a welcome development? Not only has
the Internet presented new technical capabilities, but arguably a sociological
revolution is associated with recent IT developments. The coercive global inclusion
and communitarian language of the information age and its prophets (Treanor, 1996)
8
is hardly a gentle invitation to an already excluded group. Nevertheless the Internet
offers realistic hope, as well as fear, (Schneidermann, 2000) of potentially vast
improvements to information access.
The dominance of graphic presentation on the web, presents barriers for blind and
visually impaired users in using the content there stored and assistive technologies
have been developed to provide an auditory interface to electronic text. Presently, the
dominant assistive technology for Internet access is the screen-reader.
“The use of a screen reader to access the Web was determined to be paramount by the
blind respondents. They would simply not have any independent access without this
technology.” Berry (1999)
Screen-reader software (such as JAWS, the market leader) reads out text that is shown
on the screen, relying on keyboard input for all navigation and commands. There are
numerous screen reader products available on the market and although only 3 (rather
expensive examples), Window Eyes, Hal and JAWS, dominate the market, there are
some much cheaper alternatives available. The basic function of all screen readers is
to speak text presented on the screen, though most also provide a keyboard echo
(spoken echo of each key pressed). Functionality and commands of the most popular
products are fairly similar and they all support a wide variety of applications
including; email programs, word processors, spreadsheets, web browsers, project
management and research tools,·contact management software, presentation software,
web development tools, software development tools, database management systems
·and sound-editing software. These applications continue to develop and evolve
independently and so screen reader development and compatibility are also very fluid.
Fundamentally screen readers are an ad hoc bolt on application that make available to
the auditory input mode, what is designed first and foremost as a visual interface.
Nonetheless, their success has proven revolutionary in supporting the use of PC’s by
people with visual impairment. A fuller description of the JAWS software package
can be found on page 20.
9
New Developments
Alternative audio-only interfaces (with voice recognition) that are voice navigable and
even available through the traditional telephone are developing in some application
domains (such as news and email). There is also considerable interest in the
development of haptic and multi-modal interfaces, with the principal aim of providing
richer spatial context for the user (e .g. O’ Modhrain and Gillespie, 1998). These
provide a haptic and/or auditory input that relates to spatial representation on the
screen. Haptic interface developments such as a vibrating mouse, or spatial cues from
an auditory mouse, may well improve the navigation of the virtual world for people
with visual impairment, by providing a multidimensional interface without the visual
input mode.
10
Web Design and Standardisation
Much of the literature on making the Internet accessible and usable for blind people
places the onus on web designers. Both the screen-reader and the code of the web-
published document itself make-up the user interface. So usability of screen reader
software for the Internet is inseparable from issues of web site design. The
increasingly harnessed potential for complex graphical web sites causes problems for
screen-readers, which rely on a standardised protocol for the mark-up of text.
Accessibility and usability of web sites using screen-reader software thus requires the
implementation of standardised protocols in web design and the importance of
standardisation has been widely advocated, with awareness-raising campaigns by the
RNIB (and others) and some significant legal developments.
The Legal Situation
The 2002 revision of the Disability Discrimination Act (DDA) explicitly states that
web sites can be considered to deliver services under the Supply of Services and
Goods Act and must therefore not discriminate on the basis of disability. But there is,
as yet, no legal precedent in the UK that defines the extent of implications for web
developers and which types of sites are covered by this obligation (Sloan, 2001). In
the USA in 1999, a civil case was brought against America Online’s (AOL) alleged
breach of the Americans with Disabilities Act, for discriminatory web site design.
Subsequently, an out of court settlement was reached obliging AOL to remedy the
problem, but no legal precedent was set. More compelling, Section 508 of the
Rehabilitation Act (also in the USA) explicitly defines guidelines for accessible web
design, largely drawn from the W3C’s Web Accessibility Initiative. In Australia,
Bruce Lindsay Maguire v Sydney Organising Committee for the Olympic Games
(2000) set a legal precedent by finding contravention of the Australian DDA for
inaccessible web design by the defendant. In this case specific software products,
including JAWS were used to define the standards for accessibility.
Bobby
The Bobby software package produced by Cast is relied upon to a large extent to test
accessible web design. Largely drawn from the W3Cs Web Accessibility Initiative
11
(WAI) guidelines, these standards do not encompass structural complexities, but
ensure that essential visual text or graphics can be spoken (graphics given a text
alternative with the ALT tag).
Cast themselves recognise that Bobby approval is "only one step in the process of
making a site accessible." Web site accessibility is complex and requires much
forward thinking. Standardised protocol can define accessible web design only so far,
but in defining completely usable, meaningful sites, clarity can be a struggle and there
are limits to the extent to which “one size fits all” (Howell, 2002). There are limits to
the extent to which structural features of web design can be standardised. W3C’s
WAI advocates segmentation of web sites into bite-size chunks. But it is important
not to dumb down the meaning there encapsulated and suggestions such as “one idea
per paragraph” (WAI) though to some extent sensical, can seem curiously Orwellian.
12
Beyond Access
Usability of Screen Reader Software
Certainly, huge remediation towards equitable access to the web for visually impaired
people is to some extent realised by screen reader software. But issues of
accessibility extend beyond “access to premises” (Date, 1996), access to an Internet
terminal and even access to assistive technologies. Coyne and Nielsen (2000) note
that that “Technical accessibility is necessary, but not sufficient for usability of a
design”. Technical accessibility may represent the removal of a barrier, but a more
involved analysis of usability is essential. Just because something is technically
possible does not mean every user will succeed. Such usability barriers refer to ease
of use, but can only be understood within a wider context of accessibility barriers that
include factors of cost and availability, but also pervade attitudinal and social factors.
Moreover accessibility can be considered a binary evaluation; access or not. Whereas
the concept of usability suggests that a richer picture of the user experience and its
functionality is of greater value.
“Although there are a number of non-visual access systems for the common GUI, they
could not be described as easy to use by the non-technical user.” (Gill, 1999)
Nielsen (2001) found that visually impaired users using screen-reader software
showed only a 12.5% success rate across various Internet use tasks, while a group
without visual impairment, using conventional graphical browsers had a 78.2%
success rate. Though equal usability was recognised as an unrealistic goal, the study
suggested that some disparity between sighted users and those reliant on the screen
reader can be attributed to “design flaws”, that make this figure unacceptable.
Age and Usability
Zajicek and Morrissey (2001) note how the demographics of the blind population
highlight age as an important factor and specifically advocate designing interfaces for
visually impaired elderly people as a pertinent problem. Machell (1996), for example,
found 66% of those registered as blind or partially sighted were over 75 and that this
was the fastest growing section of the blind population. Thus, although audio-only,
13
and particularly telephone-based interfaces may be more approachable to older users
familiar with older technology, the age demographics of blindness necessitate the
consideration of this extra dimension to interface design. Nielsen and Coyne (2000)
“found that current websites are twice as hard to use for people 65 or over as they are
for younger users”. They go on to suggest this is explained by the relatively young
age of web designers. Indeed it seems likely that websites are made and marketed by
and for the young. This is unfortunate given the potential for this wealth of
electronically represented and readily accessible content to remedy deficiencies in
information access in this very group. However, as the Internet grows and use
diffuses throughout the population, perhaps usability for older groups will improve.
The decline in cognitive processing power and memory associated with age seems
likely to accentuate the difficulty of navigation and comprehension of Internet based
information using Text To Speech (TTS) technologies.
Approachability
Zajicek and Arnold (1999) discuss how users in groups not provided for, often
dissociate themselves from IT and are reluctant to express demand and an interest in
information access. Thus, interface design must incorporate characteristics of
familiarity and approachability, that are potentially valuable for remedying
inequitable access while at the same time supporting functionality to complete the
tasks demanded of it.
14
Functional Characteristics of the Screen Reader Interface
Text
Text is arguably the greatest invention in the history of mankind. “The discovery that
words do not differ from each other holistically but rather by the particular
arrangement of a small inventory [alphabet] of the meaningless units they comprise.”
(Liberman, 1995) gave us phonemes. At some point in history, these phonemes were
given a physical presence and the human narrative world was transformed; the stories
last forever. And since this very first information technology, the distribution and
development of knowledge has been carried out on paper.
Serial and Parallel Processing
Though visual text has always been the most dominant input mode for reading text,
haptic forms, principally Braille, allow people with visual impairment to read. Braille
uses characters defined by 6 raised dots read through the haptic system (by touch).
Bertelson notes how “the most characteristic aspect of the Braille input is that it is
continuous and sequential. This contrasts with visual reading which “on each [eye]
fixation a window if text is made available simultaneously for perceptual analysis”.
From this Bertelson concludes that the sequential characteristics of reading Braille
text make Braille closer than visual text, to the experience of listening to speech.
Certainly, empirical research finds that eye/hand scanning speed is more constant with
Braille than visual reading and on-line adjustments, slowing down to deal with
ambiguous words for example, is possible to a lesser degree with Braille. The point
about how the eye fixes on successive windows of text is particularly interesting,
since windows computer interfaces appear to represent an extension of the natural
interface that can be found at the very “lowest” levels of perceptual processing within
the mind.
Scanning and Conceptual Models
Exposure to graphical text involves simultaneous perceptual processing of words,
whereby the user need only engage in those parts that interest her, compared with the
slow and cumbersome, sequential nature keyboard navigation and speech presentation
involved in screen reader use. The screen-reader interface has been described as one-
15
dimensional, in contrast to the two-dimensional graphical user interface. In
recognition of the difficulties in understanding the organisation of a 2-dimensional
document through speech, Zajicek and Powell (1997) propose a technical solution to
be used in an auditory-only interface; “WebChat”. WebChat organises web content to
enable concurrent processing, by compartmentalising web page contents into a menu
bar. For example, a menu of headings on the page could be created, or the document
could be searched for interesting words. This provides some high level organisation
of the document that helps visually impaired users to build a crucial conceptual model
of the whole page.
In particular, the serial nature of presentation may be restrictive for scanning-type
tasks. The way we selectively attend to elements of a GUI is compromised by text to
speech in comparison to the visual mode where windows of text are selectively
attended by simply moving ones eyes almost unconsciously.
Psychology of Reading
Classical theories of reading advocate a dual route model for cognitive processing.
Either the lexicon is accessed via phonological processing or it is directly recognised
from the grapheme (graphical representation). More contemporary models of the
cognition of reading incorporate the parallel processing characteristics of the brain
into what appears to be a serial task. We read from left to right, but top-down
processing can contribute to the perceptual process of word recognition. That is,
information about the sentence or domain context of an unknown word can push
cognition towards identifying a probable candidate word. McLelland and
Rummelhart’s (1981) interactive activation model, inspired by neural network models
explains this in terms of activation of the higher, semantic representations being able
to constrain the activation of lexical representations to push towards the recognition of
the more likely, appropriate word.
Limitations of Speech
In the case of comprehending spoken text using screen-readers, perceptual processes
are not presented simultaneously with long strings of mutually contextually
supportive words, “windows of text”. Thus the parallel processing capacity of the
human mind is not best exploited. Moreover, words must be held in the mind, since
16
regression is a very slow and cumbersome process using these technologies (many
keystrokes rather than a single eye movement). The cognitive processes involved in
using an audio interface and the different temporal characteristics of presentation
present challenges to interface designers that are very different to those of graphic
interface design. Millers (1956) classic psychology study of memory capacity found
that 7 ± 2 units of information (corresponding to mental representations) could be
held in short term memory. And this can be considered relevant to task analyses for
the use of different interface types. With screen-readers, the canvas of mind is limited
to a greater extent by short-term memory. Paper augments our cognitive powers
beyond 7 ± 2 in a way that TTS technology probably cannot. Conversely however,
phonetic speech can be considered a more natural input for language processing in
that it evolved earlier in the history of the human species and is a necessary precursor
to reading in the developmental course of the human lifetime. So cognitive
capabilities for dealing with speech may be better developed and refined. Preliminary
discussions with Steve Hambleton and others suggest that users listening to speech
can deal with text at a faster rate than visual text can be read. But this raises the
question of whether comprehension is sacrificed for speed when using a screen
reader. The speed-accuracy trade-off in performance for interaction with information
systems is a well-established phenomenon. Wickelgren (1977) found that a faster
information system was associated with increased user speed at the expense of
accuracy, with anxiety as an important factor and Underwood and Batt (1996), while
studying speed-reading - an explicitly taught skill – found that “speed is achieved at
the expense of comprehension”.
TTS technology may appear to make text more accessible but the experience and
cognition that accompanies this type of interaction with text is quite different from
that involved in reading to one’s self with visual text or Braille. TTS may well
provide different functionality and meaning to other input modes. So we must
separate content from presentation only with greatest of caution and start from the
assumption that the medium is the message.
17
Navigation, Searching and Scanning
The tasks involved in using the Internet can be understood beyond those involved in
reading continuous narrative. Analysing these types of task may help better predict
the kind of cognitive activity and thus the most usable and desirable features of
interface design. It is the facility to find documents through an auditory interface that
is perhaps so liberating for visually impaired people. Used to struggling around
libraries, shops and shelves and relying on sighted assistance, the Internet provides a
reservoir of information – in one place – that they can use without resorting to sighted
assistance. But the on-line searching processes adds a new dimension to the cognitive
task of reading, since as well as interpreting the meaning of the text read one has to
compare that interpreted meaning with an evolving idea of one’s anomalous state of
knowledge (Belkin, 1980) or information need, in order to select appropriate
documents which are to be read more thoroughly. This involves scanning a large
number of text documents superficially in order to select the most meaningful, the
right ones. The importance of facilitating scanning, is emphasized by Coyne and
Nielsen (2000) and Zajicek and Powell (1997) and so the extent to which the
functionality of the screen reader interface supports scanning can be deemed of the
utmost importance to its usability in supporting use of the Internet.
18
Job Access With Speech (JAWS) Version 3.7
Description
JAWS (Produced by: Freedom Scientific, USA) is the leading product in the screen-
reader market and the most expensive. It uses the PC’s soundcard and its own
integrated speech synthesizer “Eloquence”, unlike other products, which rely on the
Microsoft synthesizer. It also supports the use of a refreshable Braille line. JAWS
has many powerful and complex configurability tools that allow it to be used with a
wide variety of applications. But the commands required to use and navigate the
WWW, using Internet Explorer, are described below.
“Hot-Key” Commands If you want JAWS to ... THEN press ... Toggle PC/Virtual Cursor INSERT + Z Speak list of links INSERT + F7 Speak list of frames INSERT + F9 Speak tool bar items INSERT + F8 Speak address bar INSERT + A Navigate to address bar INSERT + A (pressed twice quickly) Focus to first field CTRL + INSERT + Home Read Current Screen CTRL + INSERT + Down Arrow Read Next Screen CTRL + Page Down Read Prior Screen CTRL + Page Up Move to next link TAB Move to prior link SHIFT + TAB Move to Next Frame CTRL + TAB Move to prior frame SHIFT + CTRL + TAB Speak current link INS + TAB Open a list box ALT + Down Arrow Go Back a page ALT + Left Arrow Go Forward a page ALT + Right Arrow Reformat/linearize page INS + F5 Restore reformatted page to original F5 Speak the Title Bar INS + T Announce active cursor and position ALT + DEL Navigate to first control INS + CTRL + Home Activate forms mode Numpad Slash or Enter
Table Navigation Keystrokes (JFW 3.7 and above)
If you want JAWS to ... THEN press ... Navigate to next cell in table ALT + CTRL + Right Arrow Navigate to previous cell in table ALT + CTRL + Left Arrow Say current cell ALT + CTRL + NumPad 5 Move up one cell ALT + CTRL + Up Arrow Move down one cell Alt + CTRL + Down Arrow Move to first cell in table ALT + CTRL + Home Move to last cell in table ALT + CTRL + End
(Table from Dresser)
19
Important Features
Virtual Cursor
The virtual cursor provides a point of focus on the page, starting automatically at the
top, from which Jaws reads out the text. The automatic movement of the virtual cursor
through the web page, and hence the spoken output, can be stopped at any time with
the CTRL key. Then the user can manually navigate using hotkeys that sequential
manoeuvre the cursor between either characters (←,→), words(CTRL←,CTRL→), or
lines (↓,↑). ALT DEL provides a spoken definition of the position of the cursor (i.e.
line 20, 10 spaces from the left). Pressing enter when the virtual cursor is on a link or
control activates that link or control. When the virtual cursor encounters a graphic it
will speak the ALT text, if there is any. Thus graphics, particularly those that are also
links, must have ALT tagged text to be understood.
Forms
Dealing with web based forms were input is required (such as a text box for a search
engine) requires the virtual cursor to be switched off and Jaws’ forms mode feature to
be activated (Numpad Slash or Enter).
Tables
Jaws can recognise tables (if the virtual cursor is switched on) and will speak the
number of rows and columns in a table upon reaching one. Once inside the table it
provides commands to toggle through the table cells.
Reformatting the Page
This command (INS F5) reformats the web document, removing any images and
organising the text to be presented in a more meaningful linear way. This allows web
documents with more complex organisation structures (particularly columns) to be
presented in a more meaningful order than if with the virtual cursor. That is, columns
are read as independent units rather than jumping from one column to the next as the
virtual cursor moves from left to right of the whole frame.
20
Links List and Frames List Features
Jaws provides powerful Links List (INS F7) and Frames List (INS F9) commands,
that activate and speak out a dialog box containing a list of all the links or frames that
exist on that page. Links or frames can be directly activated from the list for direct
navigation.
21
D'Amour, and Roy (2002) Critique of JAWS
Information Elements not Supported by JAWS
D'Amour, and Roy criticise JAWS 4.0 (the latest version) in the basis that it does not
provide support to all of the elements of textual information supported by HTML.
These criticisms apply equally to JAWS 3.7. They argue that other screen-readers,
such as IBM’s Home Page Reader, , offer a more complete representation of the
HTML document. For example, although JAWS supports the “alt” text equivalent tag
to provide a verbal description of an image, it is does not support the “LONGDESC”
attribute, that defines an alternative longer description. The WAI recommend that
“complex images, like diagrams and graphics, must have a long description via the
"LONGDESC" attribute” (D’Amour and Roy), exemplifying how; the definition of
standards for web design is constrained as well as inspired by an interdependence
with screen reader software.
Other elements not supported by JAWS include:
Frames
- JAWS reads the frame title but not the “LONGDESC” attribute of a frame.
Structural features
- Structure elements such as headings (H1-H6), lists and nested lists are not supported
by JAWS. These may provide a useful overview of the structural layout.
Forms
- JAWS cannot read FIELDSET or LEGENDS elements “which contain precious
information for a user filling out a form.” (D’Amour and Roy).
Tables
- Tables are complex non-visual representations and D’Amour and Roy recognise that
JAWS does present a fairly usable tool for accessing tables. But, JAWS does not
support heading abbreviations.
22
Scope of Support for HTML Information Elements
Category JAWS 4.0 IBM HPR 3.0 N element Structure elements and user control 13% 63% 8 Text equivalents 86% 64% 7 Keyboard access 0% 0% 2 Form information and navigation 67% 100% 6 Table information and navigation 63% 100% 8 Frames information and navigation 50% 25% 2 TOTAL 52% 70% 33 (Table from D’Amour and Roy, 2002)
23
METHOD
24
Research Question
To what extent does screen-reader software support the information-seeking
behaviour of blind and visually impaired people?
25
Aims
The aim will be to explore the functionality of the JAWS 3.7 screen reader interface
in conjunction with Internet Explorer for accessing information on the Internet. The
scope is broad, to permit a holistic understanding of interrelated perspectives on the
functionality of the screen reader interface for using the Internet.
26
Objectives
Establish The Context of Use
An attempt will be made to establish what the participant uses the screen reader, this
may include applications not associated with the Internet. Also what the participation
uses the Internet for and what types of information is the Internet used to access.
Availability and cost are likely to be important factors. All this provides an important
background from which to understand usability issues of the software itself.
Analyse The Usability of JAWS
A detailed discussion of the usability of the features provided by JAWS for using the
Internet will be made. This will include;
- Whether limitations in the information elements supported by JAWS are
problematic for users, particularly when dealing with structurally complex web
design.
- Whether the features that are provided are intuitive and easy to learn
- An exploration of any frustrations that arise may yield understanding of usability
issues and their aetiology. Thus an attempt will be made to analyse and clarify them.
Analyse Any Fundamental Problems With The Screen Reader
There may be some fundamental limitations with the screen reader interface.
Comparison with alternative technologies will help to clarify the role of screen
readers and when and why they are chosen over alternatives. Hence yielding
understanding of the functional characteristics and limitations of the screen reader
interface.
Analyse Related Issues of Web Design
Web design is likely to be a big issue for issues of usability of screen readers with the
Internet. An attempt will be made to understand the relationship between web design,
standardisation and the screen reader.
27
General Methodological Framework
An inductive discourse analysis will be employed, towards gathering together the
many pieces that make up a whole story of screen reader interface design for Internet
access visually impaired people. This will attempt to include an awareness of the rich
context of the cultural values and narrative that surround this issue. The research
question accentuates a need for a holistic understanding, to produce a theory that is
transferable to the real world, outside of the laboratory of scientific enquiry. The aim
should be to create a model that is the “whole cloth” rather than an isolated “building
blocks” (Erlandson et al, 1993) of reality. Although the research question provides a
centre of gravity for theory building, a flexible naturalistic enquiry is best suited to
establishing a more complete narrative context for this model and guiding towards the
most pertinent and relevant lines of enquiry.
Emphasizing the interpretivistic nature of the enquiry, an audit trail, developed
prospectively and retrospectively throughout the study, will explore the development
of ideas in a self-conscious and well contextualised way and will provide some
insights into the transferability and confirmability of the findings. Where such
insights are found that compromise or inspire the transferability of findings, they will
be noted in the results.
Screen readers may be a revolution for inequities in Internet accessibility and usability
experienced by visually impaired people, but they are no panacea. What are their
limitations?
28
Procedure
The concept of a shared emergent design was deemed most appropriate, as this
permits flexibility, sensitive to the ideas and thoughts, expressed through narrative, of
all actors involved in the research and the issue itself. The procedure lasted about an
hour and can be considered to comprise three stages. However, in actuality these
were fused into a continuous progression of ideas, a flowing discussion. The
interview method had to be flexible to the participants involved and so the appropriate
procedure was directed, to some extent, by the participants themselves. In some
cases, participants clearly had things they wanted to talk about and demonstrate and
so the interviewee drove the discussion. In other cases there was greater reliance on
the defined interview structure. All the participants were fairly verbose and the
interview was used, where necessary, to prompt and guide towards interesting and
topical ideas and issues.
In addition, one participant (I) a computer programmer, was deemed a particularly
useful source of knowledge, but was not available to be interviewed. Instead an open
questionnaire was constructed for him and responded to via email.
1. Interview
Interviews are considered the most appropriate means of capturing a dialogue that can
subjected to interpretivistic analysis. A semi-structured interview design that allows
for the development of narrative in unforeseen directions is deemed most suitable,
since it is the emphasis of the unforeseen that is particularly significant in naturalistic,
qualitative methods. Though some structure in the form of initial questions will be
formulated to provide starting points from which to explore narrative, the extent to
which the interviews will be “a conversation with a purpose” (Dexter, 1970) will be
flexible beyond the research question itself.
The interview questions develop and operationalise the stated objectives. Preliminary
discussions with users and those involved in the SRSB highlighted some related
issues that may provide meaningful context for interpretation and better
understandings of the central question of usability. These are noted on the interview
29
document to facilitate and anticipate responses and guide discussion towards
exploring relevant ideas. However, an understanding of the issues develops
throughout successive interview sessions. Thus the interviewers approach also
evolves.
2. Demonstration
This section aims to explore in a fairly naturalistic setting the information-seeking
task. Deconstructing screen reader functionality will to help analyse how well the
screen-reader provides usable access to the Internet . The participant was encouraged
to carry out a cognitive walk-through commonly performed Internet tasks and web
sites. Speaking out while doing so to explain what he is doing and why, what
problems he is experiencing, what aspects he finds easy and what he has trouble with.
An effort was made to note any discrepancies or incompleteness between the users
understanding of a web page and the sighted interviewers understanding of the GUI.
The demonstration tasks were not prescribed, since this was likely to make users
uncomfortable and give the impression that they themselves were the subject of
scrutiny. Instead they were prompted to demonstrate a particular issue or feature of
JAWS use on the Internet that the interviewer or interviewee brought up. It was
anticipated from preliminary discussions that users would be very much focussed
towards problems of inaccessible web design. And this was exploited to help
generate dialogue about what web design is easiest to use what is difficult and which
features of different web sites are manageable using the various tools that JAWS
provides. The interviewer attempted to guide discussion that may be focussed on web
design towards dialogue from which could be drawn meaningful implications about
the screen-reader tool itself.
An effort was made to demonstrate functionality that reflects a real need and real
potential for empowerment. These considerations were emergent from the interview
and were therefore a natural extension and demonstration of ideas there raised. They
therefore represent valid real world information seeking tasks. Preliminary
discussions reveal that the information blind people need to access is much the same
as the information anyone else needs to access. But, where conventional
communication relies on text or paper print, the blind are excluded.
30
Though every effort was made for the demonstration to progress naturally from the
ideas raised earlier, a document was prepared to inspire insightful discussion and was
referred to by the interviewer, on a very flexible basis.
3. Profile Questions
A profile of the following personal characteristics was taken for each participant:
- Age
- Gender
- Occupation
- Type and history of visual impairment
- Braille reader
- Frequency of use of a PC
- Frequency of use of the Internet:
- Familiarity with computers:
This helped to identify a source of attribution for any important differences in
attitudes, beliefs or behaviour that may be found between people of different age, sex,
occupation… etc. Age and IT familiarity were identified in the literature as likely
sources of difference in experience and characteristics of use and so were included.
Any of these characteristics that did not emerge as part of the interview were saved to
the end, since it was thought that to start the interview with a series of closed
questions set the wrong tone for a natural and flowing narrative.
31
Sample
A basically opportunistic sample of 9 visually impaired volunteer participants
underwent the interview procedure (plus one who responded to an emailed
questionnaire). Most of the participants are registrable as blind, and all have sight
problems severe enough that they must rely on a screen reader to use a PC. Initial
enquiries and literature emphasise that the visually impaired population cannot be
stereotyped and easily defined in terms beyond visual impairment and so we can
expect diversity within the sample to provide meaningful interpretations of what is
usable and why. In particular, there can be valid speculation that age will be an
important dependability factor for building transferable models. To address this
sampling was to some extent purposive, spanning a fairly wide range of ages to
provide some insight into how age differences affect information seeking behaviour
and generate knowledge about how considerations for interface design are
transferable to different age groups. Differences in age then, will be scrutinised for
“heterogeneous patterns” concerning the usability of screen reader software. Other
possible heterogeneous patterns may concern familiarity with IT and gender.
32
Participant Profiles
A
Age: 26
Gender: Female
Occupation: Volunteer for a charity
Type and history of visual impairment: Always had a significant visual
impairment. But sight deteriorated in her teenage years and early twenties to the point
now where she is registrable as blind and has very little sight at all
Braille reader: Yes
Use of computer: Daily
Use of Internet: Daily
Familiarity with computers: Very familiar. She relies heavily on a PC in her
working role
B
Age: 46
Gender: Female
Occupation: Self employed Braille translator
Type and history of visual impairment: She had very little sight until age 2 and is
now completely blind.
Braille reader: Yes
Use of computer: Daily
Use of Internet: Daily
Familiarity with computers: Very familiar. She is reliant on computers to run her
business.
33
C
Age: 30
Gender: male
Occupation: unemployed
Type and history of visual impairment: He is partially sighted and is able to read
magnified text in certain colours.
Braille reader: Yes though not to a high level
Use of computer: Daily
Use of Internet: Almost daily
Familiarity with computers: Very familiar
D
Age: 48
Gender: Male
Occupation: Self employed Braille translator
Type and history of visual impairment: He has always been completely blind
Braille reader: Yes
Use of computer: Daily
Use of Internet: Almost daily
Familiarity with computers: Very familiar. He is reliant on computers to run his
business.
E
Age: 89
Gender: Male
Occupation: Retired
Type and history of visual impairment: sighted for most of his life. But his sight
has deteriorated in recent years and he is now partially sighted.
Braille reader: Yes, though has only recently learnt Braille and not to a high level.
Use of computer: He has only used a computer and the Internet one or two time ever.
But interested in learning how to do so.
Use of Internet: As above
Familiarity with computers: Very unfamiliar
34
F
Age: 46
Gender: Male
Occupation: Full time employment
Type and history of visual impairment: He has always been completely blind.
Braille reader: Yes
Use of computer: Daily
Use of Internet: Daily.
Familiarity with computers: Very familiar
G
Age: 48
Gender: Male
Occupation: Unemployed
Type and history of visual impairment: He has always been completely blind.
Braille reader: Yes
Use of computer: Daily
Use of Internet: Daily
Familiarity with computers: Very familiar
H
Age: 24
Gender: Male
Occupation: Volunteer for a charity
Type and history of visual impairment: He has always been completely blind
Braille reader: Yes
Use of computer: Almost daily
Use of Internet: Rarely, he is not so familiar with using the Internet.
Familiarity with computers: He is familiar with word processing and email and
fairly confident in his all round ability to use a PC
35
I
Age: 38
Gender: Male
Occupation: Full time employment as computer programmer.
Type and history of visual impairment: Partially sighted. He is able to read large
visual text but relies on a screen reader to use a PC.
Braille reader: Yes, though he is not so proficient and not a heavy Braille user.
Use of computer: Daily
Use of Internet: Daily
Familiarity with computers: Very familiar. IT is his profession.
J
Age: 55
Gender: Male
Occupation: Retired
Type and history of visual impairment: He was fully sighted until he lost his sight
at 21 in an accident and is now completely blind.
Braille reader: Yes, though as a second written language, he does not read Braille to
a very high level.
Use of computer: Almost Daily
Use of Internet: Almost Daily
Familiarity with computers: Fairly familiar, he used computers heavily in his
working life and continues to do so recreationally.
36
SEMI-STRUCTURED INTERVIEW
Context of Use
1. How familiar are you with using computers?
Explore frequency of use. When did they start using computers? What sort of assistive
technologies do they use? Can they use a screen magnifier?. Any apparent attitudes to
IT in general?
2. What do you use the Internet for?
Work, academia or recreation, specific task domains. How often do they use the
Internet? Is the Internet an important information source? What is the extent of their
reliance on the Internet?
Functional Comparison to Alternatives
3. What are limitations and benefits of the Internet compared to other
information technologies?
Why do they choose to use the Internet rather than another technology? Factors of
cost and availability of content likely to be important. Explore speed and accuracy
characteristics.
4. Are some technologies better for different purposes?
Specifically, is Braille more engaging than TTS? When do you choose to use Braille
over the TTS or vice versa? Braille better for learning something more thoroughly?
General Usability
5. How easy is it to use the screen reader on the Internet?
Is it intuitive? Is it a pleasant/comfortable interface (speech is natural). Prompt for
positive and negative features of use to explore more thoroughly.
37
6. How easy is it learn to use the screen reader on the Internet?
Any training? How long does it take to learn. Be sensitive to age factors and
familiarity with technology. Learning to touch type may be a considerable barrier
Any frustrations in learning to use? Are you still learning?
Exploration of Specific Problems and Positive Features
7. How well is the screen-reader able to do what you want it to?
Focus on problems. Web design likely to be very important. Problems navigating and
searching. Any tasks that are badly or well supported. Any inadequacies of the screen-
reader software? What is expected of the screen reader?
8. What specific aspects of Internet access using screen-readers do you like?
Speed of access likely to be important. Also ability to retrieve from a reservoir of
documents without sighted assistance. Previously inaccessible content. Probe towards
functionality of software.
9. What specific problems do you find using the internet?
Explore frustrations. Web design is likely to be important. How do they deal with
spatial structure and graphical meaning in web design. Are they able to use the tools
that JAWS provides. Scanning through documents and sequential presentation- does
this present any problems?
10. How could these be remedied?
Desired functionality of screen reader. Any suggested features of the software? Any
suggested improvements for web design.
38
DEMONSTRATION
Could you take me through looking at a few web sites telling me what you are
doing and why, what is going through your mind as you use the Internet?
1. Visit some favourite web sites, or sites you commonly visit.
What is it you remember about familiar web sites? What are your navigational cues?
(links titles). What features in the conceptual model of the interface and of the virtual
world. How is the web/web page represented in your mind? Sequential presentation
2. Visit some new web sites you have not visited before.
Is it difficult to use new web sites? Do you understand it quickly, or does it take a
little while to understand everything on that site? How do you use familiar and
unfamiliar sites differently? Speed?
3. Demonstrate some instances of good web design.
What is good about it? What types of web design does the screen reader handle well?
What page structure is easiest to deal with. Is there a limit to the quantity of
information on a page that makes for a usable interface.
4. Demonstrate some instances of bad web design.
What is bad about it? What features does the screen reader not handle well?
Inaccessible sites and lack of ALT tags likely to be important. Focus on sites the
present usability problems for JAWS caused by structural design considerations.
Deconstruct structural design considerations.
5. Demonstrate any problems with the software/ unachievable tasks.
JAWS’ limited functional support for HTML information elements. For example
longdesc, forms, tables, graphics, headings. Do you ever get lost? What do you do if
you do not know what is happening on the screen? What information on the screen is
not supported by JAWS? What problems does this cause?
39
6. Demonstrate any features of the software you like.
Is it simple to use? Not too complicated? To what extent is the user aware of all the
features of JAWS? Usability of specific features of JAWS.
7. Demonstrate use of a search engine.
Do you often use search engines? Do you like using them? Is searching a slow
process? Focus on the scanning task. Any problems trawling through irrelevant
information? Limitations of memory on dealing with a mass of information.
40
FINDINGS
41
Context of Usage
The Importance of Screen Reader Software
It should first be established that the screen reader makes available electronically
published content not otherwise accessible to blind and visually impaired people. The
practical task of getting hold of content is made so much easier by electronic
publishing and screen-readers, that;
“It’s what, as a visually impaired person, I never had. It’s like going into a library or
a bookshop and getting a book on... whatever.” (B)
Textual content is available where it was not before. The alternative of obtaining a
paper Braille translation is logistically a lengthy and expensive process, and
refreshable Braille lines, at several thousand pounds per unit, are not affordable. The
few participants who had used a refreshable Braille line (A and B), did not prefer it
principally, because it was too slow. Many users expressed satisfaction with how the
application of this screen reader software tool to reading web-published documents
was a huge step towards remediation of inequitable access to information for blind
and visually impaired people, particularly in recent years, when “more and more
organisations provide an on-line version of the information” (G) they provide.
Furthermore, the screen reader is the tool that enables visually impaired people to use
a PC. And therefore carry out tasks in and out of the workplace, for which the PC has
become the most widely used and functional tool. In answer to the question; “are
screen-readers pleasant to use?” One respondent answered;
“Screen-readers are an essential part of my daily use of computers and my
productivity has improved greatly since I started with one.” (I)
Understandably then, there is widespread regard for screen-reader software. It is an
important tool, upon which users are very reliant as a source of information that
empower many aspects of their lives.
42
For What Tasks is The Internet Used?
Information seeking was seen as an intrinsically motivated task by itself. And to have
such quantities of information available in one place and without assistance was
considered a resource to be explored and increasingly, relied upon.
“The Internet is the information source” (A)
More specifically, the Internet was used for the following tasks; Email, research, news
and weather (CD or Braille alternatives are commonly out of date) and recreational
interests such as gardening, chess, Big Brother, planning holidays, music and so on.
The Braille translators received their text documents by email.
43
Extent of JAWS Support for Internet Explorer and HTML
Graphical and Structural Complexity
Where the meaning encapsulated in a visual image relies more on structure than on a
continuous narrative, the functionality of the speech interface will of course break
down. But JAWS and HTML provide some powerful if not altogether congruent
features for dealing with this and it is interesting to see how and to what extent, users
interpret these spatial characteristics of web pages through the features that JAWS
provides.
Links List and Frame List provide a Map
The links list and frames list features were explicitly praised by many of the
interviewees. And it can be speculated that they act as a vital higher order map of the
screen. Providing an index - all in one place - that “gives you a good idea of what’s
going on on the screen” (G). Because it is all in one place, a mental map can be
formed of the page, unlike when serially working through all the text (links and non
links) where one item (word or sentence) can be a distance of very many keystrokes
from another.
Headings
This was a problem for some portal sites, which had very many links (sometimes in
excess of 100) and relied upon headings (h1-h6) to provide some vital structural to the
list. This “made no sense” (G) when presented as a continuous list. But, HTML
provides the functionality of various heading levels (H1-H6) and it would seem
inappropriate for JAWS not to support this. Headings offer a form of hierarchy, like
the links list or frames list, that can be very helpful for building a concept of the
organisation of the document. Though it may be a long-winded and hence
unnecessary feature of the audio interface for most users using most sites, this
functionality need only be included as an advanced configurability setting.
44
Tables
All of the participants were able to use and understand tables and were pleased with
the way JAWS was able to navigate and support their use. Commands that speak the
row and column that is currently active were deemed particularly useful.
Complex Graphics
Users described less interest in sites that rely on complex graphics but again JAWS
makes this type of site less usable than it could be by not supporting the
“LONGDESC” tag provided by HTML 4.0.
Forms
Filling in forms was noted by several interviewees, to be very difficult if not
impossible, using JAWS. Screen-readers do not provide the spatial context of the text
box and hence do not adequately provide a label for that text box. It can tell you you
are in a text box, but not which one. Thus making the task impossible without sighted
assistance. B, perhaps the most experienced user, “would normally give up and seek
sighted assistance” when she needed to fill an on-line form. Given the legal
obligations that could unknowingly by imparted by filling a form and the
compromised position of relying on others to input, possibly personal information,
this seems a pertinent functional deficiency of the JAWS software. This is therefore a
failure to support the full functionality of HTML 4.0, which provides FIELDSET and
LEGEND elements that contain this vital information. Doing so could remedy this
inadequacy.
Columns
Since web design is so well defined as the issue among users, problems with columns
were attributed to the web designer rather than the software manufacturer.
In fact JAWS 3.7 has feature to deal with columns; reformatting the page. But only
two of the most experienced users (B and I) knew how to do this. Despite knowing
about this functionality, B demonstrated how she normally made sense of columns
without using it, as did A. This indicates how such structural complexity can be
interpreted without these tools and prompts the suggestion that increasingly complex
tools to deal with structure may in fact present a barrier to understanding spatially and
45
graphically represented meaning that may be too complex and difficult to define for
software to reliably interpret. The reformatting tool to deal with structural complexity
was not demonstrated by any of the participants, and so an analysis of its success
cannot go beyond this speculation.
“Columns, that’s bad web design” (A)
A tendency to blame web designers may inspire a reluctance to demand and explore
these more powerful tools for dealing with structure of web pages. It can be
speculated that, the widespread prevalence of more fundamentally inaccessible web
design may discourage users from tackling sites that are not immediately understood
due to their complex structure but that can be understood using these more complex
software tools
Motivation to Tackle Structure
On coming across difficulties using sites, there is a reluctance to persist and learn
complex tools supplied by the screen reader, to deal with complex structural features
of the web page. Especially when an unknown problem with a site is possibly if not
probably irresolvable. In the context of common accessibility problems (such as
missing ALT tags for vital graphic links) there is little impetus to learn to deal with
difficult sites.
“If I am having real problems I just leave and try a different site” (B)
Ultimately users tend to only use web sites that they find immediately useful and do
not return to those that are not. Possibly, this stagnates users interest in dealing with
structural complexities and hence software developers’ interest in providing such
features. There are, of course, limits to the extent to which structural meaning can be
translated to audio. But it would seem that fundamentally inaccessible web design
may foster an apathy towards exploring those limits.
Cognitive Overload: A Question of Balance
That JAWS does not speak all the text elements available to the HTML coder (such as
LONGDESC) indicates that perhaps the JAWS developers sacrifice a fuller speech
46
description of the visual interface in favour of a manageable quantity of information
so as not too overload the user. Attempts to restrict interface elements to create an
approachable and easy to use interface do not seem unreasonable. But not providing
advanced features to exploit the full capability of HTML 4.0 has limited the
functionality of IE users in this study. There is perhaps a tendency with these Internet
protocols and standards towards functionality that is just enough. Although there is
an attempt made by W3C and legal standardisation to explicitly define protocols that
permit more extensive functionality. HTML provides (structural) mark-up that is just
enough where XML could mark-up content. Likewise, accessible web sites provides
just enough where less tangible usability concerns of designing web page structure
and software features for dealing with structural complexity could augment
functionality.
B, an experienced user complained that “JAWS doesn’t give you everything”.
Though she did not know that unsupported elements (such as FIELDSET and
LEGEND for forms) are available in HTML, she was, dissatisfied with the
incompleteness of the verbal description for structurally and graphically encapsulated
meaning. This experienced user was very focussed on the quantity and completeness
of the content available; more was definitely considered better. At the more
experienced end of the user spectrum, limitations on JAWS support of all the
information elements and functionality of HTML does restrict usability. However,
tools such as the reformatting the page the attempt to deal with structural complexity
were not widely used even by quite experienced users. Perhaps further training in
advanced screen reader use would offer some remedy of this. Or perhaps these tools
really are more trouble than they are worth. Less experienced users, were not
interested in tackling structural complexity. If the web page was not immediately
simple and easy to understand design;
“I would close it down and look for another one” (C)
The screen reader must accommodate this heterogeneity, and the advanced
functionality provided may rarely be used. But arguably these advanced options do
not provide support for HTML functionality that is as extensive as it could be.
47
Web Design and The Web Environment
Issues of inadequate web design for screen-reader use were paramount to all users
concerns and this reflects the large proportion of web sites “something like 30-60% of
sites [that] are innaccessible” (F). Standardisation of web design is a contentious
issue at the moment, especially in the context of recent legal developments that may
oblige suppliers of goods and/or services to make their sites accessible. Defining
accessibility standards will inevitably meet some limits in the extent to which
meaning encapsulated in a visual image will be preserved when translated into an
audio stream of speech.
Usable web design for screen reader users can be divided into two levels, accessibility
and more structural design issues.
Accessibility
The first, more basic level, concerns whether or not text that is graphically presented
can be converted to speech at all. This corresponds to the WAI first requirement of
“perceivability” and in many sites, access to the text is quite literally impossible. This
level can be termed, “accessibility”. Accessibility is a necessary prerequisite, though
not sufficient, for a usable site. Within this term we include a requirement for links,
graphics and text that are visually presented on the screen, to be marked up with a
textually descriptive label using the ALT tag. I and F, both knowledgeable in the area
of web design, pointed this out as the most common and the most easily remedied
reason for inaccessible web design. Most interviewees expressed a preference to have
a description of graphics “It’s nice to know” (A), even if they were not crucial to
understanding and navigating the site. But ALT tags are especially important where a
graphic is also a link. Multimedia and Flash sites were also identified as particularly
problematic. Pop up adverts were a real frustration for many of the participants since
they disrupt intrusively and yet are inaccessible.
Structural Design Considerations
Secondary level issues of web design concern the structural layout of text. Although
text can be read out, the meaning that is encapsulated in the visual image, may not be
48
preserved when presented as a stream of speech. Columns, graphically segmented
pages, large numbers of links, and lengthy portions of narrative, can all lose meaning
when presented as a one dimensional audio stream. Later, the extent to which these
users are able to interpret these complex visual characteristics through the features
JAWS provides, will be discussed. But ultimately, web sites that present information
that is intended to make sense in a rich and complex visual context will always be less
usable for visually impaired users.
Design for Serial Scanning
The inherent lack of organisation in the Internet demands that finding appropriate
information requires significant amount of scanning through text, particularly when
using search engines. But, the difficulties with scanning and searching through a
serially presented and navigated stream of speech are a further amplification of the
frustrating trawl through irrelevant information that sighted users also experience.
And since scanning defines the way people read web sites (Nielsen), design for
scanning can inspire more usable writing for the web. Since speech presentation is
inherently serial, the “Links List” feature found on JAWS is particularly useful to
provide a kind of index for the rest of the site. Interviewees stated how the “Links
List” feature, whereby a list of links on the open page can be presented, was
particularly useful. The links list gives users some immediate access to higher-level
organisation, a kind of index or map (see “JAWS Support for Internet Explorer and
HTML”) of the document and facilitates direct navigational access to a desired
portion of text. Perhaps web designers could exploit this feature to allow visually
impaired users to form a more manageable conceptual pattern of the organisation of
their sites and hence improve their usability. For example, use of internal links that
describe an internally segmented page would provide some meaningful order where
previously there was continuous and time-costly narrative. Limiting the number of
links per page to 7 ± 2 may also be appropriate to create a memorable organisation
that can be more quickly scanned by blind and sighted users alike.
49
Text Only Sites
Text only sites overcome the problems of reliance on graphic, but some interviewees
were not keen on them since content was often less up to date than the original site.
The thought of “getting a separate version” (F) was thought to imply segregation and
this was not appealing.
50
Fundamental Limitations of The Screen Reader Interface
Comparison with Braille
Since Braille presents the only other form of access to text for blind and visually
impaired people, it provides a useful comparison with which to judge the usability of
the speech interface. But, only a very small proportion of the visually impaired
population are able to read Braille. And since visually impaired adults often grow up
with full or limited sight, Braille is commonly a second and therefore less proficient
written language.
The sample was rather unrepresentative of the visually impaired population, as all of
were able to read Braille to some extent. Speech presentation of electronic text was
preferred over hard Braille copies and this was mostly attributed to quite practical
considerations. Two of the interviewees (B and C) were self-employed as Braille
embossers and therefore had ready access to home Braille translation. But even with
these facilities they favoured TTS technologies for most tasks, because TTS is quicker
and because the sheer physical bulk of Braille text makes it unwieldy
“My house is filled with plastic bags full of the stuff” (B).
B only printed out an electronic document when she needed a portable document or
when she needed to write to that document, such as when filling in forms and C only
printed Braille for meetings, where documents presented through speech would
interfere with participation. For those without a Braille embosser, obtaining Braille is
an expensive and lengthy process to the point where the information is often out of
date by the time it is available.
“It can take up to 10 months to get a Braille translation” (D)
“You can get newspapers on CD but online news sites are updated to the minute.” (J)
In addition to these more practical accessibility issues, two other factors became
evident that make speech more favourable than Braille; speed and fatigue.
51
Speed:
“I can read a document or an email at least as fast as my sighted colleague.” (B)
The increased speed at which information can be understood with speech was
described as especially useful for searching through large amounts of text. Although
there were some differences in the speed at which Braille could be read by the
participants, and possibly a hard Braille copy would be a faster medium for
processing information if it was immediately available, most agreed that speech
allowed them to work through information faster or at least allowed them to sustain
speedy processing of information for longer. Powerful configurability tools on
JAWS, such as setting it to read the first word in every paragraph or first in every
sentence, can improve the speed at which information can be scanned. But, only 2 of
the interviewees (I and J) were aware of these configurability tools.
Fatigue:
“Braille can be tiring” (J)
“I am the world’s greatest Braille enthusiast. But if you’ve been reading with your
hand all day, you sometimes just want to listen.” (A)
It seems that the effects of fatigue and discomfort from prolonged Braille reading,
combined with the longer time required to read Braille documents, provide a strong
motivation for the preference of screen-reader software over Braille. Overall,
auditory presentation was preferred for all types of information seeking task unless
practical considerations made this untenable. It can be presumed that cumulative
fatigue and the maintenance of productivity are of paramount importance to users,
particularly in the competitive workplace where screen readers often play an
important role in supporting visually impaired workers.
52
Is Speech Less Engaging?
Initially, this author suspected that speech may provide an interface that is less
engaging than directly reading visual or tactile text. This is given some empirical
credence by generalising from Wicklegren’s finding of a speed accuracy trade-off and
the greater speed at which screen reader users operate. However, interviewees report
and show no signs of such a speed accuracy trade-off and one can assume that they
adjust the speed to a level that can be fully comprehended. Interestingly, word
processor users who use a screen reader, found that playing back a written document
can help to spot typo’s. One interviewee even mentioned how he “ found that [his]
grammar improved” (C) upon playing back what he had written. This indicates that
in some ways speech presentation permits a very thorough and engaging type of
processing. In retrospect it seems that the mind may well have faster and more
powerful processing for speech, a more natural input mode. As a more natural form
of processing than visual or tactile text, indeed a necessary developmental antecedent,
perhaps speech allows a more complete cognitive representation and more
comprehensive analysis of a given portion of text. Perhaps screen readers can even
help sighted users to become more productive. Indeed similar applications are used to
support people with dyslexia.
Synthesised Voices
This study replicated Hill’s finding that synthesised voices are “not conducive to
literary appreciation.” Incontrovertibly, synthetic speech is less engaging than natural
speech. And the quality of voice synthesis was a relevant issue for all interviewees.
It was described as a “monotone” (D), “unrealistic” (J) and had a tendency for
inappropriate, though phonetically accurate, pronunciation (for example UK is
pronounced “uck”). Although the “softer” (F) more human voices were generally
preferred (such as Microsoft’s’ Mary) it was suggested by (K) that older users
sometimes found the less realistic, more robotic type voices easier due to reduced
hearing frequency range. There were complaints that the lack of “human feeling or
emotion” (J) in synthesised speech made it too easy to “drift off”(J). And perhaps this
explains the general feeling of discontent. Indeed, audio-cassettes (that have a human
reader) were commonly more popular than synthesized speech and were preferred by
one interviewee (H) when he had the choice of voice synthesis or tape for software
53
tutorial information. Certainly then, synthesized speech lacks something of the
engagement commanded upon the listener by a real human voice.
However, users can and do learn to attune to a synthesized voice, and find that they
can increase the speed setting to as they learn. “Once you get used to the way a voice
sounds, you should be able to increase the speed to a level at which it is possible to
read everything relatively fast.” (I). Once familiar with a voice, users reported
reading times with a screen reader to be, at least as fast as visual readers and “around
twice as fast” (F) as one can read Braille. Although the speed at which Braille can be
read seems to vary considerably.
Limitations of The Keyboard
E, an 89 year old man, was interested in learning to use JAWS with the Internet but
expressed some doubts as to his cognitive capabilities.
“I’d like to learn how to do it, but I’m not sure about my memory”
Though this person was particularly active for his age, one can speculate that such
doubts are widespread in older populations and this lack of confidence is likely to be a
self-fulfilling prophecy. To this end an interface that is immediately approachable
may encourage more widespread use by older people, who do after all make up the
vast majority of the visually impaired population. Using the keyboard was rather
uncomfortable for E and so perhaps voice navigable interfaces would provide greater
approachability to older users, who are less likely to have used a keyboard in their
earlier life. Younger users found keyboard navigation approachable and fairly easy to
learn;
“If you just want it to read out chunks of text then all you have to do is press the
correct key” (I).
But it can be concluded that the keyboard is less approachable for the present
generation of older users.
54
Searching With a Screen Reader
The capability to search the Internet with a search engine was considered a very
important feature that offers significant remediation of inequitable information
provision for visually impaired people. However, searching and trawling through
large quantities of information presents greater challenges to those who rely on a
screen reader. Specifically though searching for information is “not really a problem”
(H), it was widely understood that, with a screen reader, this is a time costly process.
“But then, so is everything when you are blind.” (A)
The web, as a particularly “disorganised mass” (A) of documents, requires a lot of
selection to find something relevant.
Where a site is made of multiple HTML documents with the same information at the
top of each page, this was thought to be tiresomely repetitive by (J). All the
participants preferred the Google search engine, citing its simplicity as the reason for
this. Though B also made use of its advanced search capability.
Building Conceptual Models
There is no escaping the notion that a web site is a visual thing. But what kinds of
conceptual models do visually impaired users have of the web site interface. I, who is
able to see the screen but relies mostly on a screen-reader, provides an interesting
suggestion.
“I think if you’re relying on a screen reader more heavily, you need to learn to try and
visualize what’s being read” (I)
He was able to remember important features such as links and was therefore able to
count through the links, without fully hearing them, to get to the one he wished to
activate. B described how she could learn the “pattern” of a web site and D how he
could apply patterns learnt using the windows desktop and file structure to web
navigation using Internet Explorer. Possibly software functionality that reorganizes
structurally complex pages presents a barrier to the building of such a conceptual
model and so simplicity, in screen reader web design offers the optimal chance to
understand “what’s being read” (I).
55
Uncertainty and Unpredictability
One interviewee noted how “JAWS doesn’t give you everything”. Because screen-
readers are fundamentally bolt-on tools that translate into speech what is first and
foremost a visual representation, they provide a narrow focus of what is available to
the sighted user at any point in time. Thus users spoke of compromised “trust” (B) in
the screen-reader. If the screen-reader crashes they do not know about it. Screen
readers do not provide a speech interface immediately on booting up, but must wait
until windows is running (B), so any problems prior to this are irresolvable. Or there
may be web content that is undiscovered by the user .
“There might be something I’m missing I don’t know I am missing” (B)
I (a Supernova v4 user though the criticism applies also to JAWS) notes how; “The
level of reliability isn’t up to the point where I would like to use it to purchase
something online for example. Too frequently it will fail to read out text on pages,
such as form fields.”
The unpredictable download time of the web, was mentioned by most participants and
this adds to the feeling of uncertainty and incompleteness. All this corresponds to a
real problem with the first of the “eight golden rules of interface design” advocated by
Schneidermann (pp74);
“Strive for consistency”.
If the extent of standardised protocols for screen reader software and web site design
reaches a point such that, all the content required to use the site is available with a
screen reader, then the interface can be considered more consistent and hence more
usable. Mindful development certainly offers the potential for a more trustworthy
interface. But, even in the virtual environment, there was widespread
acknowledgement that blind users were operating in a visual world. While the
dominant access technology remains a bolt-on application to what is foremost
designed as visual media, such incompleteness is likely to be problematic.
56
Memory, Effort and Distance
The greater demands made on memory when using this interface over visual or
Braille text were brought up by (E),
“I suppose you rely a lot on memory”
The point was made that dealing with the world without sight presents a far greater
challenge to memory and requires visually impaired people to make more of an effort
to register important features of the world around them that they can’t return to with
an easy eye glance.
“I think this is a general point about being blind. I mean, it’s all right for you. You
don’t have to register everything.”
This has the effect of making the important elements of the world further from
conscious experience, more effort to check. This phenomenon is replicated in the
virtual world, with distance from one end of a page to the other not an eye glance but
many keystrokes. Presentation is more removed from direct control with a screen
reader, and so the user does not have such a complete model of his location and
surroundings in the virtual world.
57
Transferable Recommendations
Transferability
It is necessary to consider the transferability of the findings of this investigation only
after careful examination of any limitations that may be imposed by the research
context. This procedure has the air of experiment and so transferability is
compromised. The interviewer is in the room, the participant is asked to talk through
the task, and the task itself is, to some extent, prescribed. All these things detract
from the naturalism of the situation. Moreover the peculiarities of the interviewer
might shape the responses of the interviewed. I am sighted, and this may make my
analysis less aware of issues surrounding visual impairment. It may also make the
discussions less open and free, though I do not feel this was the case. The participants
know I am into studying a technology related degree, and therefore may express more
positive attitudes towards technology than they otherwise would. It certainly seemed
that all of the participants were keen on computers. Not all visually impaired people
use computers and it seems almost certain that in this way the sample is
unrepresentative. With this in mind we must ensure to design also for those that do
not use a screen reader, but could.
More Functionality and Designing For Heterogeneity
Given some remaining fundamental limitations of the screen reader speech-keyboard
interface, the basic features of JAWS do seem fairly usable and easy to learn, offering
an approachable interface for using the Internet. And it is important to preserve this
simple entry level approachability. But, JAWS does not provide support for
information elements provided by HTML that can be critical, specifically the
“LONGDESC” tag, “FIELDSET” and “LEGENDS” for forms and headings “H1”-
“H6”. (Although this list may not be comprehensive.) Including these features as
advanced options permits augmented accessibility and hence practical functionality,
without compromising usability at a more basic level. In this way the requirements of
heterogeneous groups of users, those who use the advanced functionality and those
who prefer a simpler interface, are provided for.
58
Training
Most of the participants did not know how to use the features for dealing with
structural complexity of web pages or the features for reading only the beginning of
each sentence or paragraph (for faster scanning). We can speculate that users of many
users of screen readers may find further training in these advanced features helpful,
although this apparent ignorance may in fact reflect the limited usefulness of these
features. Those that were able to use the scanning tools (F and J) did find them very
useful and training more users in the advanced and flexible features provided by the
developers of screen reader software may well augment the available functionality
and hence improve the usability of this software for most users.
Problems Fundamental To The Screen Reader
Though they are vital as the dominant assistive technology for the support of visually
impaired PC and Internet users, there remain some fundamental problems with the
screen reader interface.
Speech does not seem to present a less engaging interface as was suspected by the
author and can be used at greater speeds than Braille or visual text readers, indeed it
may even have benefits for sighted users.
The keyboard seems to be fairly approachable for younger users but older users may
well find it uncomfortable and confusing. Learning to use the keyboard can be quite
an effort and may put off those not so confident about their ability to use a PC. Voice
recognition technology might provide a more usable intuitive method of input that is
quicker and easier to learn. Also, as a bolt-on application to a primarily visual
interface, access to spatially represented meaning on the screen is likely to always be
incomplete. However, haptic and audio interfaces (e.g. vibrating or audio mouse)
may yield some exciting developments in providing spatial context for visually
impaired users.
Web Design
In terms of web design it can be recommended that standards for accessible text
where it is necessary, should improve the usability of web sites, although
standardisation for structural complexity will be difficult too define. Designers must
59
therefore consider mindfully these issues to ensure the usability of the sites they
produce, specifically by designing for scanning and not overloading the site with
information.
Accessibility Vs Usability
It is important that developers and researchers acknowledge an awareness of the
distinction between accessibility and usability issues and their aetiology. The
interviews revealed how accessibility is a more familiar term among users than
usability. I can think of 3 possible reasons for this:
1. Accessibility has been defined as the issue.
2. Limitations from the human side of HCI imply human inadequacy and are
therefore uncomfortable to talk about.
3. Information provision for users is so poor that accessibility, the first step, is the
more appropriate terminology.
Nonetheless, we should not assume that technical capability to read a chunk of text is
sufficient to meet users needs. Usability criteria demand that the interface organises
and presents content meaningfully through a pleasant to use, intuitive interface, in a
way that supports the task for which it is used. Whereas accessibility in this context is
the technical capability to read a given portion of text. There can be significant
overlap between these terms but seeking to clarify the distinction yields greater
understanding.
Development and The Interdependence of All Things
The development of screen readers and web design is not a simple matter but really
quite complex, for the interface is an emergent phenomenon of both, and the
technology is evolving rapidly. Defining and imposing standardised protocols for
accessibility or usability is therefore very difficult in such a chaotic situation.
The dependence of accessible web design upon the software that accesses it may well
mean that software with limited functional support of HTML such as JAWS 3.7
defines self-fulfilling standard that stagnate the development of newer more extensive
functionality. Accountability for the user experience is shared and emergent and as
such the demands upon web designers and screen reader developers. WAI defines the
legal standards to which web designers are obliged to meet. But WAI guidelines
demand information elements not supported by JAWS, such as LONGDESC. It is
60
important that some independent body has the power to push development forward so
that neither web design or assistive software hold the other back. User demand is also
very important as a push for new developments. But a lack of knowledge about these
unsupported elements means users will not demand them, leaving development at
greater risk of stagnation.
61
Further Research
Speed-Accuracy Limitations
Further research into the speed of presentation chosen by users and the resulting
performance at different cognitive tasks might yield some insight into the limits of
comprehending fast speech and whether the user chooses appropriate speeds for the
task. It may be that screen readers allow information to be processed faster by blind
and sighted users alike. But a more complete understanding of the speed limits of
cognition would help to define which interface technology is most appropriate for
which task. Further investigation of individual differences in the speed-accuracy
characteristics of screen reader use may also be found.
JAWS Features for Dealing With Structural Complexity The interviews revealed how the features provided by JAWS for dealing with
structurally complex web sites, were somewhat under-used. Given that some of the
participants were very experienced users, this might suggest that such features are of
limited use, especially in a web environment that is continually evolving.
Alternatively users may not be well enough informed of these features. Further
exploration of user satisfaction with these features would help to clarify the extent to
which, structural complexity can be handled by screen readers and the extent to
which, if at all, structure of web design should be standardised. Also exploration of
the level of demand by users for these tools and the extent to which, software
developers believe structural features of web design can be presented as audio might
help to clarify these limits.
Age Differences
A more focussed exploration of differences in the usability of screen reader software
for different age groups might be particularly useful given the skewed age
demographics of the visually impaired towards older age ranges. This study
highlights how older users may find the screen reader less approachable. But this
may be attributable to not having grown up in an age were PC use is prevalent or to
problems with memory. It might be useful to find at what age range any such
differences occur and whether these reflect degradation in cognitive ability, IT
experience or more cultural factors. Tests of novel interfaces, haptic audio or speech
62
recognition, with older users would be particularly useful as it seems that learning to
use a screen reader is more challenging for this group.
Scanning
Some kind of experimental comparison of performance at a scanning task using
screen reader with various different types of web pages and structures might provide
useful insight into what is easiest to scan through with a screen reader since it is this
type of reading that is so prevalent with Internet usage.
63
REFERENCES
D’Amour, J-M. and Roy, C. (2002). How assistive Software Supports Web
Accessibility. Technology And Persons With Disabilities Conference 2002.
Associated Students California State University, Northridge, Inc.
Found at: http://www-cod.csun.edu/conf/2002/proceedings/67.htm
Belkin, N. J. (1980). Anomalous states of knowledge as the basis for information
retrieval. Canadian Journal of Information Science, 5, pp133-143.
Berry, J. (1999). Apart or a part?: Access to the Internet by visually impaired and
blind people, with particular emphasis on assistive enabling technology and user
perceptions. Information Technology and Disabilities, 6 (3), pp1-16.
Found at: http://www.rit.edu/~eas/itd/itdv06n3/article 2.htm
Cawthra (1999). Blind Peoples Specific Deficit: Sexual Health in Adolescents.
Health Libraries Review. 16, pp97-105.
Dresser, S. (no date). JAWS for Windows Navigation Keys. State of Connecticut
Web Site Accessibility Committee.
Found at: http://www.cmac.state.ct.us/access/tutorials/jawsnavigation.htm
Erlandson, D. A., Harris, E. L., Skipper, B. L. and Alled, S. D. (1993). Doing
naturalistic enquiry: A guide to methods. Sage, London. pp 11.
From Erlandson et al:
-Citation pp 85: Dexter, L. A. (1970). Elite and specialized interviewing.
Evanston, Illinouis.
De Gelder, B. and Morais, J. (1995). Speech and Reading. Erlbaum, UK.
From de Gelder, B. and Morais:
-Liberman A, M. (1995). pp20 The relation of speech to reading and writing.
64
-Bertelson, P. (1995). pp91 Language by Touch: The Case of Braille
Reading.
Gill, J. M. (1999). Design Features of Terminals to Improve Accessibility by
Visually Impaired Persons. RNIB Scientific Research Unit.
Found at: http://www.tiresias.org/reports/terminal.htm
Hill, K. (1996). The information needs of the blind and visually impaired in
Sheffield. Librarianship Dissertation, University of Sheffield.
From Hill (1996)
-Date (1996). pp22 The Disability Discrimination Act. STV news. 16, pp5.
Julie, H. (2002). RNIB debate sees web usability experts divide over 'one size fits all'.
From RNIB web site.
Found at: http://www.rnib.org.uk/digital/webuse.htm
McLelland, J. L., and Rumelhart, D. E. (1981). An interactive activation model of
context effects in letter perception: Part 1. An account of basic findings.
Psychological Review. 88, pp375-407.
Machell, J. (1996). Library and information services for visually impaired people:
national guidelines. London : Library Association.
Marriott, J. (2002). The Right to Read; Comment. From RNIB web site.
Found at: http://www.rnib.org.uk/newbeacon/june02/welcome.htm
Miller, G. A. (1956). The Magical Number Seven, Plus or Minus Two: Some Limits
on our Capacity for Processing Information. Psychological Review. 63, pp81-97.
O’ Modhrain, M. S. and Gillespie, B. (1998). The Moose: A Haptic User Interface for
Blind Persons. The Archimedes Project.
Found at: http://www.scope.gmd.de/info/www6/access/acc239.html
65
Nielsen, J. (2001). Beyond Accessibility: Treating Users with Disabilities as People.
From Jakob Nielsen's Alertbox, November 11.
Found at: http://www.useit.com/alertbox/20011111.html
Nielsen, J. and Coyne, K. (2000). Web Usability for Senior Citizens: 46 Design
Guidelines Based on Usability Studies with People Age 65 and Older.
Found at: http://www.nngroup.com/reports/seniors/
Schneidermann, B. (1999). Designing the User Interface, Addison Wesley,
Sloan, M. (2001). Web Accessibility and the DDA. The Journal of Information, Law
and Technology (JILT).
Found at: http://elj.warwick.ac.uk/jilt/01-2/sloan.html
Treanor, P. (1996). Internet as Hyperliberalism.
Found at: http://web.inter.nl.net/users/Paul.Treanor/net.hyperliberal.html
Underwood, G Batt, V. (1996). Reading and Understanding: An introduction to the
psychology of reading. Blackwell, UK.
Wickelgren, W. A. Learning and memory. Englewood Cliffs; London (etc.) :
Prentice-Hall, 1977. (Prentice-Hall series in experimental psychology).
Williamson, K. Schauder, D. and Bow, A. (2000). Information seeking by blind
and sight impaired citizens: An ecological study. Information Research. 5 (4)
Found at: http://information.net/ir/5-4/paper79.html
Zajicek M, Arnold A.G. (1999). The Technology Push and The User Tailored
Information Environment.
Found at:
http://www.brookes.ac.uk/schools/cms/research/speech/publications/74_ERCIM.htm
66
Zajicek M., Morrisey W. (2001). Speech output for older visually impaired adults.
Proc. IHM-HCI'2001 pp503 - 513
Found at:
http://www.brookes.ac.uk/schools/cms/research/speech/publications/93_HCI%7E1.pd
f
Zajicek M., Powell C. (1997). Telephone access to the World Wide Web.
International Ergonomics Association 13th Triennial Congress (IEA'97), Tampere,
Finland
Found at: http://www.brookes.ac.uk/schools/cms/research/speech/pubs.htm
Internet Resources
Cast
Found at: http://www.cast.org
Freedom Scientific
Found at: http://www.freedomscientific.com
Jacob Nielsen’s Website
Found at: http://www.useit.com
RNIB
Found at: http://www.rnib.org.uk
Web Accessibility Initiative
http://www.W3.org/WAI
67
APPENDICES
68
ACRONYMS
ADA Americans with Disabilities Act
AOL America Online
CCTV Closed Circuit Television
CD Compact Disc
DDA Disability Discrimination Act
GUI Graphical User interface
HTML HyperText Mark-Up Language
JAWS Job Access With Speech
PC Personal Computer
RNIB Royal National Institute for the Blind
TTS Text To Speech
W3C World Wide Web Consortium
WAI Web Accessibility Initiative (W3C)
69
INTERVIEW SUMMARIES
A
A was able to use a PC with JAWS software installed and the internet only through
her employment and provision by the Access to Work scheme and does not have this
technology at home due to the expense. She considered the Internet a crucial source of
information upon which she was reliant;
“The Internet is the information source”
The internet had allowed her to find out information in a more independent way;
“The alternative is phoning people up”
She used the Internet for academic research, as well as more recreational domains
such as news and weather, and information about the Big Brother reality TV show
and gardening. A was an experienced Braille reader and agreed that Braille has some
functionality that is not provided by a screen reader. But felt that the screen reader
interface was able to support the information seeking tasks for which she used it.
Though she had used a refreshable Braille line, she generally preferred to use the
screen reader interface where it was available, for the following reason;
“I am the world’s greatest Braille enthusiast. But if you’ve been reading with your
hand all day, you sometimes just want to listen.”
“Accessible” web design was identified as a major problem. Use of the BBC web site
was demonstrated and structural problems such as columns and the large number of
links (50-100) were discussed.
“Columns, that’s bad web design
70
A was aware that the configurability tools available to reformat the page but did not
know how to use them. The site could still be used and made sense of without this.
Moreover, it was noted that one had first to make sense enough of the page to realise
that columns were there before reformatting. A appreciated a text description of
graphics;
“It’s nice to know”
Learnt much of her IT skills while at university. It was then that her sight detoriated
and she became increasingly reliant and technology such as screen readers. However
she was wary of expecting too much from developments in technology;
“Technology’s great. But it’s not going to give you your sight back”
Trawling through information when searching was considered somewhat time costly;
“But then so is everything when are blind”
Although A did not know how to use features that read only the beginning of each
sentence or paragraph.
71
B
B relies heavily on her PC with her own copy of JAWS software and her Braille
embosser (printer), to run and operate her business. Although she mainly uses the
Internet to pursue other more recreational interests such as news, weather and music.
Generally she used the Internet to find information “you can’t get anywhere else”.
B had had some training and was therefore aware of some of the more complex
feature for dealing with structure such as reformatting. But had not really explored
their functionality and tended to rely on the basic features
Web design problems demonstrated included problems with completely inaccessible
buttons and animation on Flash sites and text links links that say only “click here”.
This loses any sense of meaning when spoken as part of a links list and without the
context of words around the “click here” link. Familiar web sites could be used much
quicker and this was attributed to a memory of the “pattern” of the site, particularly
important features such as links.
B had the capability to translate literature into Braille when she so required. But only
did so when she needed a portable copy. The physical size of Braille text made it a
less usable medium;
“My house is filled with plastic bags full of the stuff.”
She considered speech to be “very natural” and screen readers to provide a functional
interface for the most information seeking tasks and found reading Braille to be
slower;
“[with a screen reader] I can read a document or an email at least as fast as my sighted
colleague.”
The facility to search a reservoir of information without sighted assistance. Was
considered one of the most positive features of the Internet;
72
“It’s what, as a visually impaired person, I never had. It’s like going into a library or
a bookshop and getting a book on... whatever.”
The Links List feature was discussed and demonstrated to be a very useful tool for
navigating and quickly grasping the contents of the page.
B was aware of the structures for dealing with structural complexity such as
reformatting the page but had not explored how to use them.
Several specific problems with the functionality of JAWS were discussed. A
particular problem was identified with the “Forms Mode” feature in JAWS.
Specifically that; “it can tell you you are in a text box, but not which one”. This was a
real frustration and B complained that she would have to seek sighted assistance in
filling in on-line forms. Another problem demonstrated concerned streaming media.
JAWS needs to be shut down during streaming media and this was thought to be due
to problems in sharing the soundcard. While JAWS was not running she had no
access to any problems that may have arisen, and this sense of incompleteness was a
more general frustration for B.
“JAWS doesn’t give you everything”
“The biggest frustration is that you have to wait until Windows is up and running
before you have speech. If it’s running a scandisk or something I’ve no way of
knowing about it”
73
C
C had recently taken a number of IT training courses and was quite familiar with the
several screen reader products, including HAL and Window Eyes and has a PC with a
copy of Window Eyes at home. He is partially sighted and is able to read magnified
text in certain colours. But is largely reliant on a screen reader as a back-up to use a
PC. C suggested that provision of technology can be very problematic because it can
be difficult to know what is available and “where to look” to find the most appropriate
assistive technology.
He is also able to read Braille, but not that confidently, perhaps explaining an overall
preference for auditory presentation.
Problem identified with pop-up adverts, that are disruptive and often innaccessible
with a screen-reader. C preffered text only sites wherever thgey were available and if
having any problems with a site, would “close it down and look for another one” .
C had the least visual impairment of the sample and so was less reliant on a screen
reader. When scanning through documents, screen readers were too slow and C
found that he was able to do this faster by sight. This highlights the slow pace of
scanning through documents with a screen reader. He also “ found that [his] grammar
improved” when he was able to read through documents he had word processed with
a screen reader. This indicates how perhaps screen readers allow a more complete
comprehension of continuous narrative.
74
D
Like B, A relies heavily on his PC, Braille embosser and screen reader software (in
this case Window Bridge produced by Syntha-Voice Computers Inc) to run and
operate his business. He has always been completely blind and is able to read Braille
proficiently.
The main problems with Braille were considered its size and (though not for this user)
the time to get hold of it.
“It can take up to 10 months to get a Braille translation.”
But advantages of Braille over screen readers were also given. C used Braille print
outs when audio presentation interferes with other concurrent demands on the user
(such as in meetings) and noted that;
“[With Braille] you can flick backwards and forwards. But [with a screen reader]
you’ve got to hear it fully.”
He uses the Internet to find information on news, weather and other recreational
domains although perhaps not as frequently as other participants. Though,
interestingly, he did not use any of the functionality that JAWS provides for dealing
with web page structure. Instead preferring to copy web published documents into a
word processor and read them through the word processor.
C was reluctant to persist with web sites that he found problematic;
“If there are problems with that site then I’ll exit that site.”
C suggested that using the Internet Explorer involved skills transferable from learning
how to use the windows desktop.
“ [It is similar to] moving through filenames, directories, menus”
75
He did not like the synthesised voice on the basis that they were a “monotone” and
also noted two key frustrations with using the Internet. Searching through.
information that looks relevant, but is deceptively irrelevant and inconsistent
download times when he has no way of knowing if something has gone wrong.
76
E
E is a very active 89 year old man who is beginning to learn to use a PC, principally
with the intention of accessing information available on the internet. However, he
expressed concerns about his memory and thought this may limit his ability do so,
specifically expressing concern about whether his memory was good enough.
“I’d like to learn how to do it, but I’m not sure about my memory.”
E was able to use the arrow keys to move navigate within a page but had very little
experience of using a keyboard and was evidently experiencing problems in learning
to use one. In addition, problems with hearing made the speed difficult to
comprehend, although he was able to understand the synthesised voice if the volume
was turned up.
NB/ This interview was rather shorter than the others in order to fit in with E’s
transport arrangements.
77
F
F relies extensively on a PC and screen reader software (HAL) to carry out his
working role and is very interested in IT and the Internet, particularly issues of
accessible web design. He suggested that “something like 30-60% of sites are
inaccessible.” and went on to clarify to types of problem with web site design. More
basic accessibility problems, where essential text cannot be spoken by the screen
reader and problems with dealing with structural complexity. Text only version sites
were discussed but were considered too often out of date and the idea of “getting a
separate version” was not appealing.
He preferred to read documents with a screen reader over Braille because he can do so
“around twice as fast”. The issue of speed was considered important because it
permitted the maintenance of a higher productivity for longer.
The problem of unengaging voices was discussed and F said that the“softer” more
human voices were generally preferred. But that older users sometimes found the less
realistic, more robotic type voices easier due to reduced hearing frequency range.
F used the complex functionality for dealing with structurally complex web pages,
although this was slightly different with HAL. But he felt that this did not greatly
improve the usability of most web sites, especially considering the prevalence of
fundamentally inaccessible web design.
78
G
G has a PC with a copy of Jaws at home. He has begun to use the internet recently,
for job-hunting and to find information on many other aspects of his life. He finds he
is increasingly able to use the Internet to find information without assistance as
“more and more organisations provide an on-line version of the information”. The
immediacy of available content was a big advantage over requesting Braille copies of
information provided by organisations.
An example of good web design was demonstrated. The page had successive bite size
chunks of narrative each relating to a link. This was easy to read through with a
screen reader and made immediate sense.
A problem was encountered on a portal page, where a very long list of links was
divided into meaningful headings using the H1-H6 tags. This was immediately
evident to the sighted interviewer but “made no sense” to the screen reader user.
G particularly noted the usefulness of the Links List and the Frames List;
“It gives you a good idea of what’s going on on the screen.”
The problem with filling in forms was also demonstrated.
G was not aware of the tools for dealing with structural complexity (except tables)
and, like most of the participants, left a site if he was having problems with it. Nor
did he use the functionality to read only the beginning of sentences and paragraphs for
speedy scanning.
79
H
H has a lot of experience using a PC with Jaws and relied on using Word and Outlook
Express applications to complete in his education, although he is less familiar with
using the Internet and Internet Explorer.
The dislike of speech synthesisers was highlighted by a preference to receive
computer tutorials in recorded audio form. These were thought more pleasant to use
than the electronic text tutorials that are read out by the synthesised voice.
The capability to navigate and understand tables was demonstrated and functionality
for dealing with this kind of structural complexity appeared to be particularly
successful. Though H had little experience of using the Internet he was able to carry
out functions fairly successfully with knowing only the basic commands and was able
to discover by trial and error new commands. Perhaps this reflects how younger users
are able to learn more quickly and easily.
80
J
J has experience of using a PC with a screen reader (JAWS) through his working life
and now uses the Internet within more recreational domains such as news and
weather, what’s on listings and chess. One of the major advantages of the Internet
being that up to date content is available without sighted assistance.
“You can get newspapers on CD but online news sites are updated to the minute.”
Though he can read Braille, J notes how “Braille can be tiring” and reading it involves
physical discomfort. It is therefore not pleasant to use. This is particularly the case
for him, since Braille is his second written language and he will therefore never be as
proficient as if he had learnt it as a child.
The synthesised voice was criticised on the basis that unlike a human voice it has no
“human feeling or emotion”. For this reason it is often all to easy to “drift off” when
using a screen reader. This was considered an important issue J talked of “effort” that
could be better spent comprehending the content wasted on deciphering the words
being spoken.
When scanning through documents, this user was able to use the functionality that
sets the screen reader to the beginning of each paragraph or the beginning of each
sentence and found this made the scanning task much faster. He had discovered this
functionality without any specific training and claims to enjoy learning new skills.
81