e-mail: p-sundb@uiuc.edu
A semester research paper submitted for:
(click on any blue section of the outline below to go to that section)
1.0 INTRODUCTION -- THE GENERAL ISSUE
1.1 THE PRESENT STUDY
2.0 THE PARTICIPANTS
2.1 WEB-LITERACY AND BROWSER PREFERENCE
2.2 EDUCATIONAL LEVELS
2.3 READING PRACTICES AND PREFERENCES
3.0 THE METHODOLOGY
3.1 ADVANTAGES AND DISADVANTAGES OF AN ON-LINE
QUESTIONNAIRE
4.0 DESCRIPTION OF THE QUESTIONNAIRE
4.1 DESCRIPTION OF THE FIVE SAMPLE WEB SITES
5.0 RESULTS OF THE SURVEY
5.1 CRITIQUES OF THE FIVE SITES
5.2 WEB DESIGN ISSUES ARISING FROM THE
QUESTIONNAIRE
5.2.1 FORMAT ISSUES
5.2.2 NAVIGATION
5.2.3 NON-TEXT MEDIA
6.0 PRACTICAL AND THEORETICAL CONSIDERATIONS
7.0 REFERENCES
At present, there is a concerted effort at cutting-edge US universities to transfer many components of academic courses -- e.g. syllabi, lecture notes, required readings -- from paper to digital form. Computer-assisted Instruction as such is not a new phenomenon, but in the past such instruction was available only on site, either on an individual PC, via a local area network (LAN) in a specific computer lab, or mediated via a central mainframe computer at one university, as was the case with the Plato system in the 1970s (cf. Bruce & Levin's CAI taxonomy 1996 -- on-line). Computer-based courseware is available most universally, however, on the World Wide Web, which is accessed via popular, freely-downloadable Web browsers such as Netscape, Mosaic and Explorer. The main advantage of the World Wide Web (hereafter "the Web") is ease of publication and diffusion; thus, while a certain university course's on-line material is primarily provided for students enrolled in that course, it is theoretically accessible to anyone with a Web browser anywhere in the world.
Here at the University of Illinois at Urbana-Champaign, over 290 courses make at least some material available to students via the Web, courses in such diverse fields as physics, biology, economics, computer engineering, Italian, and religious studies. Various courseware templates are currently being used at U.I.U.C. to create such course sites: Cyberprof, Mallard, and NCSA's Net Learning Place. A listing of some of on-line courses at UIUC may be found at the following Web URLs:
And for a listing of on-line courses nationally:
This massive transfer of academic materials to the Web includes a wide array of academic genres: the course syllabus, on-line class chat forums, lecture notes (e.g. Powerpoint presentation documents), and -- the focus of this project -- academic readings. Many of the courses whose Web sites I looked at keep the heavier reading activities off-line, requiring instead traditional text books or listing required articles available in print form. A few courses, however, are offering such readings in the form of HTML documents, or Web pages. These pages range from long, single-page word-processed documents inserted into HTML format (with only a modicum of editing for on-line presentation) to lavish textual sites which take greater advantage of the Web's multimedia potential, including use of hyperlinks to other sites and use of colorful images, sound and graphics.
The focus of this present study was to investigate the World Wide Web (mediated by various Web browsers) as a reading environment for one subset of all the on-line textual genres: university-level academic readings. It is primarily a heuristic study which attempts to describe such course readings as exist on the Web at present (the latter half of 1996) with an aim to prescribe how Web-mediated readings can be optimized for greater friendliness to and comprehen-sion by student readers. Thus, the current study is both descriptive and (tentatively) prescriptive. To gain objectivity in its descriptions and weight in its prescriptions, this study relies on data collected on-line via a Web questionnaire. The data consist of written reactions by eighteen Web users to five sample course Web sites which span the gamut of academic reading page designs I have encountered on the Web.
Out of the 18 respondents to the questionnaire, eight are graduate students, six of whom are enrolled at the University of Illinois at Urbana-Champaign. Four of the respondents are faculty or staff at a university, and four work full-time in a profession. Two of the 18 respondents, 5 and 10, left no data about their name or location. A slight majority of respondents were male (10), seven were females. Seven of the respondents live outside of Illinois in California, Indiana, Maryland, Michigan, Missouri, and the District of Columbia. One respondent, 10, left no personal information whatsoever, including what her/his gender was (question 1). None of the respondents had taken the classes in which the materials are used. One respondent, however, had been a co-writer (but not Web page designer) of the journal article on teaching teleapprenticeships (Article 5).
Participants were not selected at random; rather, I invited responses to the questionnaire in two ways. The first was by e-mailing its URL (Web address) to listservs of students and faculty in relevant departments and classes -- students in Dr. Levin's Ed Psy 387 course "Computer Uses in Education" and the Division of English as an International Language (DEIL). The second way was by e-mailing specific individuals: friends, former classmates, colleagues, and acquaintances -- people whom I felt would have access to a Web browser and some familiarity with the issues involved in Web design and/or university-level instruction. A number had experience designing Web pages themselves, background in instructional technology, work with traditional layout, and/or writing for publications and other media, experience which made their input more valuable than that of chance Web surfers who might happen on the questionnaire site.
The people who responded to the questionnaire had no prior contact or ex-perience with the 5 Web sites and generally spent 30-40 minutes on them in total -- although some respondents reported spending extra time to read some of the academic articles fully. Longer-term exposure to the pages and a longer chance to experiment with them may well produce different responses over time: for example, users may well come to loathe a feature that attracted them initially or radically change their opinion of a particular site for or against. Students in courses that are required to use such Web sites over a semester may thus develop very different responses from those of the 18 respondents here. In this, the present study may not be an accurate reflection of user response to authentic courseware used over a longer term
All of the subjects may be presumed to be computer-literate to some extent and familiar enough with Web browsers and the World Wide Web both to locate the questionnaire site and navigate through the questionnaire successfully. The degree of Web literacy (question 2) varies somewhat, however, from experience of "one day to one month" (one respondent), "one to twelve months" experience (one respondent), "one to two years" experience (6 respondents), and "more than 2 years" experience (9 respondents). Again, one respondent left no personal data.
Concerning the Web browser used to view the questionnaire (question 4), the vast majority of respondents (16) used some version of Netscape: Netscape Gold 3.0 (7 respondents), Netscape 2.0n (8 respondents) and Netscape 1.n (1 respondent). As for the other two respondents, one left no information on her/ his browser and the other marked "other" browser, which happened to be OmniWeb 2.0.
Their educational levels (question 3) range from holding a Bachelor's degree (2 respondents) to holding a graduate degree or working towards one (15 re-spondents). One respondent left no information on her/his educational level. While the academic background of these respondents is not representative of the US population as a whole, the present sample of Web users is nonetheless appropriate for the present study since the original intended audience of the selected Web readings is/was university students -- graduate or undergraduate.
As for their practices and preferences reading on the Web (question 5), four of the 18 respondents had taken courses which used academic readings on line. Twelve indicated that they "occasionally" read longer articles/papers on the Web; only one respondent claimed to read such on-line texts "frequently". As for preferences for reading environment, the majority (12) preferred to read for information off line; only one preferred reading for information on the Web; and 2 expressed no strong preference for either reading environment. Not all respondents selected more than one choice -- contrary to the researcher's original intention. Thus, while most responded to questions about reading practices on the Web ("never", "occasionally", "frequently"), not as many responded to preference questions ("off line", "on line", and "no preference"). One respondent left no responses to this question.
Unlike in quantitative studies, the aim of the present study was more heuristic: not to gather "typical" responses of a subset of a certain population to a set group of Web pages , but rather, to gather a wide variety of responses by persons with a variety of expertise in computers generally, in CAI, in publication, and in Web design in particular. It was the variety and quality of the responses in which I was most interested. As Microsoft's consumer research division Web site states, "...we are testing software, not people ...." (see FAQ in References) -- or, in this case: Web pages, not Web users.
Human computer interface and ergonomics should be primary concerns to those who design university courseware on the Web just as such concerns are to developers in the computer industry. A standard method of evaluating this interface is to obtain users' feedback either by having them fill out a survey or recording them -- either on tape or on video -- while they are using a computer .
By way of comparison, the Microsoft Usability Group (http://www.microsoft. com/ usability/tour/study.htm) uses the following methodology for consumer evaluation of Microsoft products. After a subject is given preliminary training in thinking aloud, s/he is videotaped in a laboratory setting interacting with a certain consumer product (software or hardware) which Microsoft is developing. Meanwhile, a Usability Group researcher notes any actions or verbal responses by the subject. Following this, the subject fills out an on-line questionnaire to describe her/his opinions and preferences about the product. The goal is to gain insight into the thought processes and feelings of ordinary users (Microsoft's "audience") in order to aid developers in tailoring the product more closely to them.
In this study, the "products" under development to which subjects are asked to respond are five authentic academic Web sites containing a full-page or longer of normal academic prose. The sites were chosen as representative of the gamut of academic sites used in university courses as required or supplementary readings. The means to gain insight into the thought processes of the target audience in this study is also an on-line questionnaire. It contains both demographic questions about the individual respondent and five text blanks in which to collect the respondents reactions to the five corresponding academic Web pages. Due to the hypertext nature of Web pages and HTML language, the questionnaire is also able to embed examples of the "product", in this case the five Web pages, directly within the questionnaire in the form of hyperlinks.
Disseminating over the Web a questionnaire containing links to relevant sites -- while not as immediate and contextually rich a data-gathering method as video-taped laboratory sessions -- has the advantage of reaching a potentially larger subject pool than is possible in on-site testing at a university. Offering the questionnaire on the Web also allows the researcher to provide questionnaire takers with links to other Web sites, to which subjects can relate directly rather than via a printed description of them, as would be necessary using a traditional questionnaire printed on paper. Data collection is also facilitated, since the respon-ses of respondents who submit their completed questionnaires through an HTML "form" protocol are submitted on-line in a matter of seconds at no cost and compiled automatically in one database document (in this case, "the-file.text").
The main disadvantage is the researcher's loss of control over the filling out and submission processes. Reliability of responses may be compromised since subjects could potentially lie about either (a) their true feelings about the texts -- although this is unlikely given the virtual anonymity of the submission process or (b) the extent of their actual contact with the five sites. However, since taking the 30-40 minutes required to deal thoughtfully with the questionnaire involves some sacrifice of time and effort, it is unlikely that respondents would even bother to undertake to answer the questionnaire unless they were serious.
In designing a research project there is always a delicate balance between thoroughness in covering the research question and the limits of participants' endurance, goodwill and time. There are obviously many more than 5 types of academic readings commonly found on the Web. However, given the amount of time any thoughtful critique of even one site takes, I felt five critiques to be near the upper limit of what an ordinary (unpaid) respondent would be willing to undertake. As the questionnaire stands, it took the researcher himself nearly 40 minutes to complete the form with any degree of thoroughness -- and I doubt the respondents' zeal for the project could exceed my own!
The questionnaire (www.ed.uiuc.edu/students/p-sundb/ questionnaire.html) is a single Web page HTML document composed of 10 questions: five general demographic and general preference questions and five text blocks for the respondent to critique the five corresponding academic Web pages. Immediately under the title, for legal purposes, the form includes a link to a Web page listing the subject's rights and explaining in brief the purpose of the study (URL: www.ed.uiuc.edu/students/ p-sundb/statement.html).
Out of courtesy to potential respondents, at the beginning of the form I mention the estimated time necessary to complete the form -- 30-40 minutes -- as well as the total number of questions: 10. I also suggest that they scroll to the end first before beginning to fill out the form in order to judge whether they are likely or willing to complete it.
Before the actual questions, the form requests the subject's full name, e-mail address, city and state/country. The five general questions in the first half request the subject's (1) sex, (2) years of experience on the Web, (3) most recent educational level, (4) type of Web browser used to view the questionnaire (e.g. Netscape) and (5) reading practices and preferences on the Web.
Question 5 asks whether the respondent has ever taken courses that required outside readings on the Web, how frequently the respondent reads full-length articles on the Web ("never" to "frequently"), and whether the respondent prefers to read for information on or off the Web.
The second section involves five written Web page critiques. Before actually starting questions 6-10, the respondent is given a brief introduction to the task and instructed to skim the texts "noting positive and negative features" and "any confusion [s/he] had, any strong reactions." It is stressed that skimming, not careful reading is the appropriate mode of reading. The subject is also told to note such features as (1) the general attractiveness of the page, (2) the text and font -- how easy or difficult the size or style makes reading, (3) the formatting -- paragraphing, headers, etc., (4) ease of navigation both within the text and beyond it via hyperlinks, (5) illustrations -- whether they help or hinder the reading task, and (6) helpful extras.
In addition, in case a respondent should still be unsure of what is required after reading the initial instructions for the section, I provide a link to a sample text and my example critique of it. The sample reading text is one whose layout I consider above average for a Web page: "Why we sleep" from the Harvard Mahoney Neuroscience Institute Newsletter, an article containing attractive headers and images along with a text in typical academic prose.
Following the initial introduction and short orientation to Web critiquing, question 6 is presented. A color-highlighted link is offered to each site, followed by a text block in which the respondent is to type in her/his reaction to that site. All pages linked to which are outside the main questionnaire are opened as separate browser windows, allowing the subject to view both the questionnaire and sample Web page at the same time and to type in comments while the object of her/his comments is still visible. Upon completion of each page, the respondent clicks the top left box of the extra window to close it.
After completing all ten questions, the respondent encounters a "submit" button, at which time s/he can click to send off her/his responses to the researcher's database file or else click a second button to clear all fields in the questionnaire (to start over or quit).
The first sample academic Web site (Article 1) is a reading for a physics class, the text of a talk by British physicist S.W. Hawking on the "Origin of the Universe". It was chosen as an example of the bare minimum style of reading site: gray background (if user's browser is set to default), plain text with a minimum of formatting, containing no links or illustrations within the article and depending entirely on scrolling for navigation through the document -- 6 typewritten page lengths. The most basic HTML document, more word-processing document than Web page -- one long, unbroken Web page.
URL: http://www.astro.nwu.edu/lentz/astro/hawking-1.html
The second sample site (Article 2) is a reading for an economics course, an article on welfare reform entitled "Wisconsin Works: Significant Experiment, Troubling Features" from the Center for Law and Social Policy. This site is similar to the first in its minimalism, but makes greater use of HTML features such as internal and external links, horizontal dividers and a greater variety of headers. Esthetically bland, it is still a long, scrollable one-page Web document (10 typewritten page lengths) with plain text on the default gray background, but the author has facilitated navigation by offering internal links directly to various sections of this longer document as an alternative to scrolling. Footnotes are handled as internal hyperlinks: the reader is taken directly down to the relevant footnote, then offered a link immediately back to the context s/he departed from.
URL: http://epn.org/clasp/clwisc.html
The third sample site (Article 3) is a short reading for an economics course. It is actually a Web page created by the professor as one of a set of many issue pages linked to from her course home page. In this case, the issue is "Providing for the Common Good." It differs from the previous two articles in length: roughly the equivalent of a single page of printed text. It presents the issue in summary form, contains links to the professor's lecture notes for that part of the syllabus as well as one external reading (by another author). At the bottom, it contains directional links to the course homepage as well as lecture notes and quizzes. Thus, it is an example of a reading which is more integrated into the body of the course. It also contains a colorful yellow background, colorful icons and dividers, a graph and a photograph.
URL: http://www.cba.uiuc.edu/college/econ/econ214/ issue4.html
The fourth sample site (Article 4) is a suggested (not required) link from a physics course home page. It consists of a series of short, related articles on "Black Holes" and is part of a larger public information site created for NCSA's Multimedia On-line Expo, "Science for the Millennium," winner of various Web design awards. It is lavishly illustrated with professional quality graphics and charts. I chose it as an example of the "high end" of academic Web page design.
URL: http://www.ncsa.uiuc.edu/Cyberia/NumRel/ BlackHoles.html
The fifth sample site (Article 5) is an optional reading for a course in education, "Teaching Teleapprenticeships: an innovative model for technology integration in teacher training", an article from an on-line academic journal published by the International Society for Technology in Education (ISTE). Like the NCSA example above, it is also professional quality and higher end, albeit more serious in tone and design than NCSA's more popular site. The background is gray, but more textured and artistic. The text is the primary focus, but the article is set up in a "frames" format which allows the table of contents and other para-textual material to be permanently displayed in the left column while the main text changes in the middle section. The contents headers are themselves hyperlinks to the corresponding (Web) pages and sections within the article. Navigation is handled in a consistent and sophisticated manner, for the most part.
URL: http://www.educ.ksu.edu/Projects/JRCE/v28-5/Thurston/article/main.htm
The material in this section is derived from the questionnaire responses submitted on-line by the 18 respondents and stored in the database "the-file.text" in my student folder in the Education College's WWW server. (Click on the link above for a compete transcript of all the responses.)
The first section (5.1) synthesizes the responses to each of the five Web sites, ending by giving each site an overall rating of A (excellent), B (good), C (acceptable) or D (deficient) based on what the researcher interpreted from each respondent's stated overall impression. (In retrospect, for greater reliability and objectivity, I should have asked each respondent to rate each site as a part of the questionnaire itself using the above letter scale.) The second section, 5.2, will discuss general Web design issues which arise from the critiques of the five sample Web sites. These issues are broken down broadly into (a) 5.2.1 format (b) 5.2.2. navigation and (c) 5.2.3 non-text media (graphics and animation). The researcher has interspersed recommendations and implications for academic Web design throughout this section.
The first site , the Hawking article with its "Falling Links" warning at the top (see appendix A for a full transcription of all user critiques), was described by the majority of respondents with words such as "very basic", "unfinished", "unrefined", "neat but utilitarian", "no pizzazz" -- "boring" in general. Some were offended by its plainness ("plain, boring -- nothing like physics!"), others more accepting ("an OKAY document -- nothing that BAD, ... I'd have to really want to read the whole thing to get through it all"), and one respondent even approved of its simplicity, rating all features "good" and concluding "I could tolerate this text very easily, unlike most web pages". A number commented that the length of the text (the equivalent of 6 typewritten pages on one scrollable Web page) was off-putting: "too lengthy -- intimidating." Many objected to the necessity of scrolling through such a long document: "too long to be read from a monitor".
At least five respondents mentioned that they would rather print it out to read and highlight than read it on-line: "I would NEVER read this entire piece on screen. Typical of the type of piece I would print out & highlight," wrote one respondent. Another went so far as to print it out in the midst of filling out the questionnaire.
Several found the formatting unprofessional. "The 'falling links' warning at the beginning is somewhat 'off-putting' -- makes me wonder about the reliability of the page I'm looking at, etc." "I'm always hesitant to spend a lot of time on a page that has the 'under construction' symbol on it. .... Usually, I avoid those pages ..." (Perhaps, the Webmaster for this page meant 'under construction' rather than 'falling links' since there are no links except one at the very end?)
Would they read the article? Some mentioned that they would read the article anyway if the content were interesting enough to them, whereas others were not attracted enough to stay and read unless required: "tempted to go on (unless it's an assignment)", "Boring -- I probably wouldn't stick around to read the whole thing.", "I would have to be a serious academic to want to read this."
Overall grade: C-
Article two, a social critique of Wisconsin's welfare system, was often described as "better than number one", but not by much. General descriptions of it include "not pretty, but functional", "still no flair", "too textbook like". More positively, some respondents commented on it as "well structured" due to more varied formatting (use of headers, boldface, lists, etc.), while more negatively, one of the respondents commented, "page needs overhaul." Opinion was mixed regarding the formatting. A number termed the formatting "appropriate", "easy to get info", "more readable", "an appropriate HTML document making use of HTML features (as opposed to simply a text available via the Web as with #1)".
As with article one, several mentioned that they would prefer to print the article out to read first. One respondent wondered why it was put on line in the first place: "for accessibility?"
While several found the text of the article less "dense" than the first because of the formatting, one respondent complained the article was not really readable due to the "horrendous density of information."
Navigation in Article 2 was generally commented on positively, especially the back-and-forth linking to the footnotes: "I really like the embedded links to the footnotes, and the fact that 'return to text' took me back to the SAME SPOT where I left off ...."
Overall grade: B-
The third site was not a full length article as in the case of the first and second articles; rather, it was a professor's short one-page introduction to an economics topic on her syllabus: "the Common Good." Most questionnaire respondents (15 out of 18) reacted positively to this site. A librarian commented: "Here you're getting closer to good utilization of the capabilities of your program & equipment. It's easy to look at, colorful, broken into small bits of information visually, which is what I expect from most of my experience on the Web." "a great page -- I read lots of this. I think this would be a great course to take!" "I would definitely stay on and read on." They reacted positively to the inclusion of the graph of US military spending and the photograph of soldiers.
Not all were enthusiastic, however. The most frequent complaint was about the bright yellow background of the page: "excruciating on the eyes" was the strongest version. Some respondents also objected to the "magazine"-like format: "inappropriate for assigned readings". Another person, however, found the more colorful magazine format "comfortable". Not all appreciated the Web master's efforts: "'E' for effort but they tried a little too hard to dress this up."
Generally, however, a majority of the respondents found the page attractive and inviting and even stopped to read it.
Overall grade: A-
The fourth site, a series of pages on black holes, a subset of pages from NCSA's larger site on relativity in physics, inspired the greatest praise: "very professional", "impressive", "this would really work as a good introduction to a subject; this is nearly as good as a good book." An older professor wrote: "Why didn't we have resources like this when I was in school?"
What did they like? The professional-quality layout and graphics were the main points mentioned: "I loved these pages ... I found myself reading more of the text than on any of the other pages because of its layout," "looks a lot like a quality textbook or magazine page." The shortness of the pages and smaller text "bites" were also appreciated: "I didn't feel overwhelmed by the info on the page", "graphics professional and helpful." The respondents were impressed by the greater use of the Web's multimedia potential: "true multimedia approach", "good use of the medium" -- although not everyone. One participant complained, "I want information, not frills." (He also preferred the severe simplicity of texts one and two.)
The more complex navigational devices in Article 4 were frequently commented on, especially the links to related topics interspersed throughout the text. But a number mentioned problems getting lost or finding their way back. The short page-by-page presentation style was seen as an aid to motivation: "encouraged one to surf through", "[multiple links] give a 'what next?' flavor to the subject -- a page-turner, as it were."
Overall grade: A.
The last site, an article from an on-line academic journal, received mixed reactions, mostly positive. The overall appearance was judged on a continuum from "very professional looking", "very stylish", to "drab but professional". It was seen as recognizably "serious" and "academic". The librarian was not very impressed, however: "Another example of a paper transcribed to electronic format with a clickable index or table of contents and capability of moving backwards and forwards. Not imaginative." The sectioning of the article with Netscape's built-in "frames" and other HTML formatting devices such as horizontal bars, and headers was favorably received, as was the menu of links to sections of the articles in the left-most frame.
A very frequent complaint -- not related to reading or layout, but a significant affective feature -- was the slowness of loading the article page by page with its unusual multicolored background and graphics. Frequent complaints about confusing navigational conventions -- links and button icons -- also arose: "could have been more navigationally friendly." The use of small photographs in the text was seen as superfluous: "pictures not really needed, too small, don't come across as professionally as everything else", "photos didn't make much immediate sense". One respondent commented, however, that the photography was properly minimal, keeping the text as the main focus.
Overall grade: B+.
Design preferences are extremely personal. The first reading feature to which I requested respondents to reply was "general attractiveness": the appearance of the page in the browser window. Indeed, the five sample academic sites were purposely chosen for the variety of initial impressions they make on the person accessing them. Some pages were "bare bones," such as the first site's -- essentially, a word-processed document converted into HTML, the simplest form of Web document. Other pages, such as the NCSA black holes site, combined text, hyperlinks and multimedia, .
The questionnaire asked, "Are you tempted to stay and read or surf on?" As with matters of aesthetic taste in general, people's conceptions of "attractiveness" on the Web varied. Some respondents were drawn to color and variety, others to spareness and simplicity. On many of the following design features there were disagreements. Any reader of this paper hoping to legislate one "best academic page design" as a result of this study will find little unambiguous support here. There were some noticeable trends in the respondents' responses, however, and the issues they themselves brought up should offer a useful contribution to the discussion of what constitutes optimal Web page design.
The general appearance of the page in the browser window is the first impression Web readers receive. A few respondents commented on the oddness of Article 2 opening slightly below the top of the page: "page should open at top -- not middle of document." Some objected to different loading times for text and graphics on a page. They felt the whole page should be visible when first loaded. Curiously, no respondent commented on the fact that in Article 3 ("Providing for the Common Good") the page opens with a fourth of it hidden if the browser is set to the default Netscape window size. (I am assuming most respondents had not selected non-default window sizes, although the questionnaire did not ask about this specifically.) One person did mention the problem of "zig-zagging" to view this layout, implying scrolling left and right to see all the information.
The issue of text presentation options was raised by several respondents: reading on-line, reading off-line, text-to-speech options for the visually impaired. Several commented appreciatively on the option of a text-only version of text 5 formatted for printing (for off-line reading) in addition to the version formatted for on-line reading. One woman mentioned that the first text (Hawking) was "clean for printing and for use with text-to-speech" software, suggesting that academic Web designers should be conscious of the preferences and physical disabilities of potential readers -- in a word, "audience". As mentioned previously in 2.3, 12 of the 18 respondents preferred their reading "off-line" and only "occasionally" read for information on-line. "I'd rather have a hard copy of a journal, which is easier on the eyes to read, and a lot more portable."
Aesthetic considerations must therefore be counterbalanced by practical concerns. A plain-Jane Web page, for instance, may offend the aesthetic sensibilities of graphic design connoisseurs, but may be preferred by blind persons. User choice should therefore be foremost in the minds of Web designers. Perhaps an academic reading site should offer at least two versions of itself: one formatted for on-line reading (involving perhaps many pages) and a single-page, stripped-down version suitable for printing -- not all students may have time or the desire to read on-line -- or for conversion to speech.
Expectations about genre are another "audience" issue in Web design. University students, with long exposure to many different publication and media styles, approach reading on the Web with certain expectations about genre. To such sophisticated readers, the use of high-quality photographs and colorful design are hallmarks of popular magazines, while a Spartan and text-heavy format is associated with academic publications. When this expectation is challenged, say, by a colorful, yet academic page such as Article 3, reactions differ. Among participants in the Web survey, half found the colorful, graphically rich economics site attractive. Others, however, objected: "eye-catching colors -- but why? It's an assigned reading, not a magazine piece." Another termed it "tabloid". In the case of Article 5, on the other hand, many respondents commented, some grudgingly, that the on-line journal article with its gray background looked very "professional", if "drab". "A serious, academic presentation," one termed it. None objected to its looking like what they expected an academic journal to look like. Some even found it "stylish".
Those designing academic Web pages, therefore, need to be aware that innovative use of formatting -- contrary to a genre's expected conventions -- may not be warmly received by all members of its intended audience. Some will find a magazine format "comfortable". Others may find it "distracting" instead. Even the visually more restrained NCSA site with its gray background was lambasted by one respondent: "Did not like it. I want information, not frills."
Number and length of Web pages was another issue raised by several questionnaire respondents. The Web allows theoretically unlimited page length, navigable by scrolling -- in fact, the computer interface metaphor is of a scroll. The Web also allows an unlimited number of multiply-linked pages, a book metaphor. For printing, obviously the single page scrolled format is most convenient. And indeed, this is how Article 5's "text only" option was designed. For on-line reading, however, respondents were fairly unanimous in their preference for shorter and more pages. One woman wrote that text five was "just the right amount of text per page ... otherwise I feel overwhelmed and reluctant to read on." Another defined "good Web format" as "one screen in length, nice short page a major plus on Web," using article three as an illustration. Several people mentioned a motivational "what next?" aspect to having smaller, more numerous pages, making reading more of an adventure of discovery. For other kinds of reading, however, such as skimming and scanning, one respondent wrote that one long [scrollable] page was "easier for scanning than linked pages." But not many of the readers liked scrolling down through long texts. Several found it "intimidating".
Another Netscape formatting feature, frames, as seen in Article 5, met with approval mostly, although one respondent found them "distracting and seldom useful." Most felt that the restrained use of frames as in 5 -- a narrow "side bar" frame at left to hold the index of links to sections of the article and a wide frame a right for the text -- was well-structured. Several commented, however, that the text in the left frame running over into the right frame (on pages following the title page) was "screwed up" and "distracting". This error could perhaps have been avoided by the Webmaster's viewing the final version in several different browsers.
In general, clear sectioning of the text was seen as an aid to reading. This was commented on especially in Article 2 on welfare and in the journal article (5). "[article 5] is sectioned off nicely." On the Web this is achieved primarily by the use of headers, horizontal lines and frames. They give "diversity to the straight text" and make skimming easier. Text 2 had a particularly rich variety of devices to subdivide the text: "more carefully defined sections [than article 1], and good headers on some paragraphs (and indentations of subtopics) aid in rapid skimming." Since skimming is heavily relied on in academic reading, careful sectioning of the text by the page designer using the devices above can actually increase students' speed in accessing information.
A related issue is that of chunking the text into paragraphs and "white" space -- in other words, paragraph size. Respondents wrote about their dislike of "dense" text with "big blocky paragraphs" and little white space, as seen in the first text. Such text density "increases eye strain and lessens the likelihood that I would read it all on the screen." Instead, many respondents expressed their appreciation for text that is "presented in small chunks", "bite-sized pieces" which don't "overwhelm the reader", although chunking can go too far, as well, requiring "too much back and forth", impeding skimming.
The actual letters (font and font size) are, of course, what make up the text, and have perhaps the most direct effect on how successful the Web is as a reading environment. However, font was not an issue for most respondents: they generally rated the default font as "good" or "OK". (Again, the researcher had provided no way to gather data on the actual print size seen by individual respondents.) Smallness and "crowdedness" of the font size, however, was a concern to a number of readers: one wrote that size 12 font was "impossible to read" and enlarged it to 18 using her Web browser "preferences". Another letter problem mentioned was lack of contrast -- or annoying contrast -- between background and letters. One respondent was concerned that the "small, thin text" in the last article "blends into the gray background -- could create problems on bad monitor," making the text nearly impossible to read.
Implications for academic Web design, therefore, are that, due to the unfortunate lack of standardization among current Web browsers, text size, color and background color may need to be specified in the HTML source document to maintain the desired level of contrast and legibility of the text independent of browser. In addition, professors ought perhaps to orient their students to their browser's options so that the students are aware that they have the power to change the browser environment. For example, seeing-impaired (but not blind) students should know enough about Netscape to be able to enlarge the font size of their on-line academic texts as large as is necessary for legibility.
Page background, although in the "back" subconsciously -- and often consciously, can affect both the legibility of the text in the foreground and the attitude of the reader. Background color was one of the features of the Web environment that was most frequently mentioned in responses to the questionnaire. The choice of color or background is obviously not neutral. A large number of respondents complained about Netscape's default gray background, claiming it was "depressing", "generic", "unpolished", "annoying after a while", and "makes me want to sleep. Some, however, found it made the text legible, a "comfortable contrast" with the black text color. Again, students should be made aware that they have the power to change offensive backgrounds through their browser's menu options. Background color also offered symbolic possibilities: one respondent suggested that an article on black holes ought better to be presented on a black background.
As mentioned above, many participants in the survey objected to the yellow background in text 3, finding it "too bright", "almost overwhelming", "startling", and "excruciating on the eyes". Even those who appreciated the color found that it made it harder for them to focus on the text. One woman claimed that she had to resort to the links on the page to escape the irritation to her eyes. One respondent, however, mentioned that it made the text "easier to read" and "less tiring to my eyes". This leads one to suspect that the actual nature of the yellow background color differed greatly from browser to browser and monitor to monitor, a problem academic Web designers need to keep in mind. Again, a Webmaster ought to view a semi-final version of her/his page in several browsers before posting it to a larger audience.
As for the color of the background in the last article, it contributed in large part, I'm sure, to that page's "professional" and "serious" impression. The textured gray-maroon background gave the page a "classy, high-brow look" to one man and was "easy on the eyes" to a woman. On the liability side, however, several respondents mentioned difficulty in reading the text in the foreground due to lack of contrast. Again, this leads one to suspect that different browsers and monitors are producing different visual effects, a worrying situation if the message is primary. Perhaps until greater standardization on the Web is achieved, designers of academic reading pages, especially, should aim for a more conservative and controllable text/background color format. Legibility, not aesthetics, is primary.
Another issue in designing Web pages for course readings is maintaining academic formatting conventions. Academic genres have peculiar formats and components which may work more or less elegantly on the Web. Over-innovation in dealing with these may not please -- or worse, may confuse -- the reading audience. One respondent was unaware that the last article had a title (see Article 5 in Appendix C) since the page designer had included the title in a unconventional graphic at the top of the page which the reader evidently interpreted as the journal title instead. Seeing only the list of authors' names where he expected the title, the reader complained: "the scholars' names get the headline??"
In general, I believe that academic formats can be adapted quite successfully to the hypertext nature of the Web as evidenced by the many positive responses to the use of hyperlinked footnotes in article 2, the clickable outline of the NCSA readings, and the side index menu in Article 5. These and other navigational issues are dealt with in more detail in the next section.
While navigation through articles on the Web is not technically related to their textual component, navigation is vitally related to ease of accessing texts and reader perceptions of and attitudes toward academic texts. Good navigational interface welcomes readers to a site and guides them through it intuitively and effortlessly; bad navigation, on the other hand, confuses and frustrates readers and may well dissuade them from reading the text altogether. Navigation, at its root, is about control -- user control of the reading environment. If the reader feels out of control of her/his movement through the text, little effective reading can occur.
As mentioned at the end of 5.2.1, traditional academic format comprising an abstract, table of contents, headings, subheadings, footnotes, and references can be adapted quite successfully to the Web. Longer academic texts, especially, require navigational aids to help the reader to move through them sequentially or to skip to certain sections directly -- say, the "conclusions" section of a research article. Whatever the pagination system of the Web article -- scroll or book metaphor -- any section or page should theoretically be accessible by clicking a mouse on that page's hyperlinks. Without internal and external links in the article itself, the reader is reduced to relying on the slow and frustrating scrolling method of navigation. Three of the five articles -- the second, fourth and fifth -- made extensive use of internal links for internal navigation.
In the second article on welfare in Wisconsin, several respondents appreciated "the ability to click to the notes and then back to the text" from any footnote in the text. According to this design, all the footnote information is appended at the end of the one-page document and all the links are internal. This may create problems, however, for the reader who, instead of leaping back and forth between text and footnotes, scrolls through to the end and encounters a long list of footnotes with a "return to text" link attached to each. One reader did this and upon encountering the label "return to text" wondered initially whether there might be another text outside of the one currently displayed since, as far as he was concerned, he had never left the welfare text. Perhaps a better link label would have been "Return to original location in text". Poor navigational labels were responsible in large part for respondents' frustration in navigation.
Article 4 on black holes, while attractive at first glance, seriously misled a number of participants, who, arriving at an intermediate level of a very complex hypertext architecture without realizing where in the structure they were, ended up in unexpected places with no way back except via the "back" buttons built into their browser. One man clicked the "back" icon on the site's glossary page and ended up on an unfamiliar page rather than the one he had just left. On the positive side, a number of respondents liked the fact that the navigational options menu appeared in a consistent location both at the top and at the bottom of each page in the series.
Inconsistent conventions in Article 5 led several readers astray . Unaware that the title page they had opened to was not the home page for the whole site, but rather one article of many in on-line journal, some readers who clicked the "backward" icon on the title page were taken to another article, not another page in the same article. This cause disorientation and confusion. Yet these same icons within the article help navigation through the article. Likewise, the only way to begin the journal article from the title page was to click the index of section links in the left frame to access page 2, yet no explicit instruction was given on how to begin. Once inside the article, the index was no longer available, and the user was forced to rely on the navigation icons.
Thus, levels of navigation should be kept visually separate: links between articles in a chain or larger hierarchy and links within articles. "Back" or "Backwards" labels may be insufficient; instead, labels such as "Return to site home page" or "back to previous article" might prove more intuitive. Navigation conventions should be clearly marked and consistent from the first page through the last.
Other than these navigational inconsistencies, however, respondents were generally favorable to navigation via the system in article 5. One respondent, comparing article 2, with its hyperlinked index found only at the beginning of the article and a "return" at the end, to article 5 with its side index menu, said "better to have side frame with navigation buttons on the side so you don't have to read through entire section". Navigational options should therefore be available at all points in the text, as in a side menu.
The consistent desire of the respondents, seen in all these navigational problems, was to have freedom of movement to go anywhere at anytime, and clearly marked, intuitive links. One complained about no "return to top of page" button at the bottom of article 1. Another arrived at the last page of article 5 and didn't know how to get back to the first page. Web designers need to be aware that users' navigational needs may not always be linear and that readers often backtrack in reading. Testing the user-friendliness of the navigation system of one's Web pages on volunteer subjects would help remove bugs before the page is posted publicly.
As for the navigational buttons/links themselves, several respondents found the artsy, but abstract buttons in the last article (5) "a little too subtle" and "of minimal use to new user since there was no text explaining their purpose." In addition, the functions of the same set of buttons differed on the "front" page and "interior" pages of the article.
In Article 4, the extensive use of hyperlinks in words within the text was seen both favorably and unfavorably by respondents. In a text-only browser, one man opined, the links add interest (color?). But another felt links within text "suck" and wouldn't bother to check them out. Too many text links annoyed a third respondent, who felt they often link to "irrelevant things" or "pages with practically no info". Most felt the links to be relevant to the topic, however. In general, therefore, links within a text should be used sparingly and only when inclusion of the additional information can be rationalized.
"Relevancy" was another main issue with respect to the illustrations used with academic Web pages. Most respondents found the graph and photograph on the defense spending site (3) "contextual and explanatory", but others found especially the photograph of the soldier "uncompelling," serving "no apparent purpose". Particularly in the journal article (5), several objected to the small photographs as "unprofessional" and "irrelevant". They did not object to the quantity of images in the latter case, however, since the text remained the main focus.
In general, readers appreciated any attempt to break up the text and felt that graphics "add a nice dimension to reading" and make it more "interesting". Page designers need to be aware, however, that many readers can also find graphics "junky" and "too much", a sin against good taste. Especially in the case of academic Web pages, where aesthetic considerations are secondary, all graphical materials should be relevant to the topic and relevant in their location on the page. As images add loading time to each page, they must add to the text and not just ornament it.
The video clips in articles 4 and 5 were appreciated by some, but avoided by others leery of the downloading time. One respondent, however, expressed appreciation for the fact that Article 5 explicitly warned about the 15-minute downloading time of the video clip, a courtesy other Web designers would do well imitate.
As a general issue, loading time for Web pages, special backgrounds and graphics was mentioned frequently. One respondent liked the idea of including movie clips and sound files on Web pages in theory, but has decided to wait until the Net downloading time is faster before actually making use of such options currently on the Web. For academic reading sites, however, the multimedia component should continue to play a secondary role to the text itself.
Among the "already converted," who tout Computer-assisted Instruction with such enthusiasm, it is an article of faith that any ordinary educational practice is done better by computer. Objective educators interested in CAI, however, need to test such an assumption case by case.
No single activity is as central to disseminating and receiving information in academia as reading. Despite the ongoing integration of nearly every contemporary medium -- sound, video, graphics -- in digital form, the bulk of scholarly information should continue to be available primarily in textual form. Thus, reading should continue to play a disproportionately large role in higher education for years to come, even in the Digital Age. At the same time, the sheer quantity of textual information is exploding beyond the individual academic's ability to keep up.
The "facts on the ground" are that a mind-boggling quantity of text is already available on the Internet. At present, a small subset of this consists of readings for university courses. How successfully such material is currently being encoded in digital form for display and how efficiently it is being decoded on screen by academic readers should therefore be of prime concern to educational researchers. Given the geometric growth of academic textual material, it is imperative that an optimal reading medium be found -- and soon. Accurate comparisons between media, however, can only be made when each medium (print and computer) is used at its optimal level. Only then can the "best", most human-friendly medium be determined.
It was the aim of this study to analyze the "state of the art" with respect to on-line academic readings circa 1996 and suggest ways to optimize academic reading on the Web. Granted, Web technology is continually in flux. The many cracks through which inexperienced and careless Web designers can now fall may soon be closed up in future browser systems. The relevance of many of the recommendations made in the sections above may thus be short-lived. Yet the principles of effective presentation of academic text should remain rela-tively stable for much longer as the human perceptual apparatus and cognitive faculties of those who read such text are evolving almost imperceptibly. Media are flexible; the human organism is much less so. Determining the principles of optimal text presentation is thus a goal relatively unaffected by the chaotic turns of technological innovation. It is my hope that the responses of those who participated in the questionnaire will add something of value to this search.
More farsightedly, I feel it is worth beginning to explore whether the computer as a text medium (in a larger sense and not only in the currently dominant form of Worldwide Web browsers) can be developed into an optimal reading environment -- a major historic advance in efficiency and speed over the book/print medium that Western civilization has relied on since the codex was invented in the 1st cen-tury AD. Incorporating (a) what is now known from cognitive psychological research on reading and (b) the flexibility of computer interfaces, can we create a computer reading interface that allows reading to reach the upper limits of human visual processing capability?
Bruce, B. and Levin, J. 1996. "Educational Technology: Media for Inquiry, Communication, Construction, and Expression. World Wide Web HTML document: http://www.ed.uiuc.edu/facstaff/chip/taxonomy/
"Frequently Asked Questions". A Web page by Microsoft Corporation's Usability Group. HTML document: http://www.microsoft.com/usability/faq.htm