An Examination of the Impact of Subjective Cultural Issues on the Usability of a Localized Web Site - The Louvre Museum Web Site
The number of people who access the World Wide Web for entertainment, business, research, and education is growing daily; statistics in February 2000 point to a possible 250 million Web users (Nua, 2000). This represents an increase of 100 million in just over one year.
Horton, Taylor, Ignacio, and Hoft (1996) give the following reasons why people use the World Wide Web:
However, only 5 to 10% of businesses (Bussin, 1998) adapt their Web sites for audiences that do not speak English. Even when a multilingual version of a site is available, it is often inadequate. Web localization currently focuses on linguistic and technical issues, but layout, graphics, color, and organization are rarely adapted to suit the needs of international users. Linguistic translation, however, only scratches the surface of localization: cultural experience may have a greater influence on usersí reactions (Hall, 1990; Arnold, 1998). People from different cultures do have different expectations and needs, and meeting these needs improves the usability of the localized versions of a multilingual site.
Good Design Practices for Web Sites
Some design rules apply to all sites; these include:
1. Avoiding loud colors and backgrounds and "gratuitous use of bleeding edge technology" (Nielsen, 1996).
2.Including solid content.
3.Making the site structure transparent.
Many sources outline good Web design practice; however, each Web site genre (advertising, information, academic, fun, and so on) requires design rules that match audience expectations. For example, a museum site should be aesthetically appealing as well as easy to use.
The Web has enormous scope as a visual medium, and people typically find it difficult to read large chunks of text on screen (Horton, 1994). Therefore, graphics are a crucial factor in Web design. Nonetheless, visuals and colors have cultural implications that designers need to be aware of. For example, symbols that Americans regard as relatively harmless can be offensive to people in other parts of the world. Hand gestures are especially troublesome (Fernandes, 1995; Subbiah, 1992). In many cases though, well-designed visuals can transcend cultural and political boundaries.
A multilingual site must be more than visually appealing: it must be usable for all its target groups. Quite aside from linguistic considerations, the siteís structure must be transparent for users from all culture groups, regardless of their inherent expectations. Like other hypertext media, the World Wide Web defies the traditional approach to documentation organization and structure ó no site with more than one page has a definite beginning, middle, or end. Access devices and content must also be suitable for all members of each target audience.
Culture can be defined as "the shared knowledge and values of any group" (Kostelnick, 1995: p. 182). Many models of culture exist to describe cultural experience and influence. The Iceberg Model (see Figure 1) is one of the most popular.
Just as only 10% of an iceberg is visible above water, only 10% of a groupís cultural characteristics are obvious or explicit. The remaining 90% are comprised of unspoken rules (like business etiquette) and unconscious rules (like non-verbal behavior).
Hall (1989) defines two types of culture: high context and low context. People from high context cultures (including Japanese, Chinese and Arabic people) use context to communicate most information in a message, while people from low context cultures (including native English speakers, German speakers and Scandinavians) expect lots of detail in visual and verbal communication. Other anthropologists look at culture from different perspectives: for example, Hofstede and Trompenaars look at variables such as individualism versus collectivism, masculinity versus femininity and long-term versus short-term.
Practically speaking, how can designers apply knowledge of culture when building international Web sites? Hoft (1995) recommends that designers determine international variables and use these to develop profiles of international communities. A model for developing international Web sites might look at variables like:
1. Color and visuals.
3. User values and attitudes.
4. National formats (like time, date, currency, and character sets).
5. Information expectations.
This next section outlines a study to determine how these international variables affect the usability of the four language versions of the Louvre Museum Official Web site.
The main aim of the study was to discover how culture difference affects how users react to localized versions of the Louvre site. This was possible by determining whether users from different culture groups experienced different problems when using this site (the Louvre Museum Web site). This site is available in four languages: French, Spanish, English, and Japanese.
The Louvre museum is one of the most famous in the world and its official Web site is well known. The site should be of interest to a range of groups: people interested in art and art history; people planning a trip to the museum; people who want to buy goods associated with the museum; students assigned a research task; or casual Web users. This site was most suitable for testing because:
The site is very large and includes extensive information on both temporary and permanent collections, a virtual tour, visitor information, contact information, and online booking and order facilities. While the original site was created in the French language, most of the site is available through English. The Spanish and, in particular, the Japanese versions of the site are more limited. For example, information on, and views of parts of, the collections are not available through Japanese.
The test approach was cooperative evaluation, possibly the most useful form of usability testing (Nielsen, 1993). A co-operative evaluation session involves an evaluator observing a test participant working with a system. Participants verbalize their thoughts and actions as they work through a task sheet. Tasks reflect actions that a typical user might want to do with the system. A short debriefing session follows the test, during which the evaluator tries to elicit specific information and recommendations from participants. If the session is video recorded, evaluators can recap on the sessions. Tapes are also a valuable source of information about subjectsí non-verbal behavior.
Co-operative evaluation, therefore, facilitates the observation of:
Subjects from each language group were recruited from the student population of the University of Limerick. To ensure that the results were valid and specific to the research in question, the subject sample was controlled to the greatest degree possible. All were aged between 20 and 30 years, all had previous experience with the Microsoft Internet Explorer 4.0 browser, and all were university students. None had lived in Ireland for longer than one year. To achieve a balanced study, each group comprised the same number of participants: there were 16 subjects, four from each language group, two male and two female to counterbalance any possible discrepancies associated with gender roles.
Each observation session took place with a single participant in a university laboratory booked for the study to prevent interruptions. Before each participant arrived:
1. The camera was set up and ready to record each session.
2. The Web site was open on the (French language) home page.
3. The browser history was deleted.
Each session took fifty-five minutes to complete. During the first five minutes, the test was explained to the participant who then read through the task sheet (see Appendix A), and had the opportunity to ask questions about the study. Every effort was made to put the participant at ease and to make him or her aware that the purpose of this research was to investigate the Web site, not the abilities of the participant. Each participant began by spending ten minutes exploring the site, and giving general opinions. After this exploratory exercise, they had to return to the site home page and begin working through the task sheet. Participants had to stop once they had spent thirty minutes working through the task sheet, regardless of whether or not they had completed all tasks. Finally a ten-minute debriefing session followed (see Appendix B for the list of debriefing questions).
It was not possible to test more than four users from each language group due to time constraints. In addition, the subjects were not wholly representative of typical users of these language versions, as they had all spent time living in Ireland and were, to some degree, acclimatized to an English-speaking culture. Testing in the various countries may have produced different results. A further constraint was that the language for all task lists and sessions was English. Although they were proficient in the language, for 75% of the participants English is a foreign language. Finally, while the subject groups were controlled with respect to age, profession, gender, and level of experience, it is a fallacy to surmise that any of the subjects can be wholly representative of a particular group. Even in controlled situations, participants display individual preferences and tastes. Cultural background, while influential, does not eclipse the inherent uniqueness of each user.
This section of the paper outlines the results of the co-operative evaluation tests and debriefing sessions. These tests are a rich source of qualitative information. Subjects express their opinions, and demonstrate their likes and dislikes. The think-aloud protocol is less useful for gathering quantifiable data, although we do achieve concrete information on, for example, the number of tasks each subject completes, and the number of errors subjects make (Figures 2 and 3).
While none of the Japanese subjects completed all tasks, two Irish, one French and one Spanish subject completed all 15 tasks successfully.
As the chart in Figure 3 indicates, Japanese subjects experienced most difficulty with tasks; two had an average of more than one error per task. Of the other groups, Irish participants experienced fewest "errors". Errors correspond to the number of superfluous or erroneous steps that users took to complete a task.
During debriefing participants were asked their favorite and least favorite features of the site, features they would change and features they would add. Table 1 illustrates their responses.
Participants were also asked during debriefing to agree or disagree with a number of statements. Their responses are shown in Table 2.
The reasons subjects cited for wishing to use the site again were either to find out more information about the art collections or to view artifacts from the collections. Two subjects who said they would not use the site again remarked that they would prefer a physical visit to the museum than a "virtual" experience.
Some overlapping inevitably occurs when comparing usability between groups. There were some problems that affected all, or a large number of, participants. Likewise, participants across groups admired many features of the site. The problems that were most crucial for most subjects, and the features that most subjects admired, are listed below.
Top Five Usability Problems
1. Navigation tools are very inflexible. Users must depend almost exclusively on the main menu to carry out any task. The navigation bar at the top of each page was meaningless for Spanish, Irish and Japanese users.
2. Menu labels are ambiguous on all language versions, but especially on French and Spanish.
3. Many parts of the site are only available in French. While the English version is almost complete, Spanish and especially Japanese users are forced to work with very limited functions.
4. The organization of the main menu does not correspond to usersí expectations in many respects. For example, many users expected Temporary and Permanent Collections to be grouped together.
5. There are no links to diagrams of the museum on any language version. Subjects came upon them by chance, or by random searches through pages. There is no usable diagram of the museum on any part of the Japanese version.
Top Five Features of the Site
1. Pages are not excessively long, and most users found information with minimal scrolling.
2. Graphics were fast to load. The successful integration of art on this site was noted across all groups.
3. The site has a vast repository of information about artifacts, details of collections, and the history of the museum.
4. The site has a neutral look and feel with no "bleeding edge technology".
5. The site was not offensive politically, culturally, or economically to any of the participants in this study.
The discussion of results focuses on the following international variables:
1. Visuals and Color.
2. Navigation and Access.
3. User Values and Attitudes.
4. Information Expectations.
6. Linguistic Issues.
Visuals and Color
Most visuals on the site are culturally neutral as they depict museum artifacts or photographs of parts of the museum. None of the subjects complained about these graphics. On the contrary, most commented on the utility of featuring examples from each collection.
In general, there are fewer diagrams on the Japanese version of the site. This proved a severe problem; two of the four Japanese participants commented that they were more interested in browsing through pictures than reading text. Figure 4 shows part of the Reception Area under the Pyramid page from the English language site. Figure 5 shows the corresponding Japanese language page, which omits the graphic.
The favorite feature of three of the four Japanese subjects was the Collections part of the site, which has not yet been localized into Japanese. During the experiments it was evident that Japanese subjects only agreed to look at the Collections through English so that they could complete tasks. Under normal circumstances they would not have used this part of the site at all.
Only one (illegible) diagram of the museum is available on the Japanese version of the site (see Appendix C), and those on other versions have French captions. This was most problematic for Japanese users, only one of whom could complete the associated tasks (finding and printing a diagram of the museum).
The colors on the site are neutral ó shades of blue and gray, and white. Three Spanish and three Irish users found the color scheme dull and uninteresting. One Irish user commented that the overall artwork on the site does not reflect the purpose or status of the museum and its associated Web site. However, no user felt that the colors were inappropriate or incongruous, suggesting that the colors were unobtrusive, a positive usability attribute.
Navigation and Access
The primary navigation tool on the site is the main menu, part of a frameset, on the left-hand side of each page. This menu is available on all pages of the site, which is a positive navigation feature. However, because this menu is the sole means of information retrieval, users are restricted in:
1. How much information they can find.
2. How they go about finding information.
3. How they interact with the system.
The reliance of most users on the browserís Back button shows that this menu does not match usersí needs. When a site is created with a diverse cultural audience in mind, and when the site strives for consistency of layout between language versions, as this one does, then developers should consider providing alternative means of information retrieval that support a variety of user techniques. Forty percent of subjects recommended adding either a search facility or an index to the site to facilitate information retrieval.
The navigation bar on the top of each page was only meaningful for French participants, probably because of the label on other language versions (Menu, instead of Home Page), but also possibly because of its position on the right instead of the left-hand corner of the page (Nielsen, 1997). Furthermore, though not an issue during these tests, users may choose a language version only from the home page. Many users will not access the site directly from the home page, but may be directed, via a search engine, to a page deep within the site. Given that users during these experiments did not recognize the home page link, this could be a serious issue for international users, who have no way of knowing that other language variants are available.
User Values and Attitudes
Although art is not always neutral, and does not always transcend cultural boundaries, no subject involved in this research showed political, social, or economic sensitivity to any information or presentation technique. Nonetheless, participants from different subject groups did display distinctive characteristics.
1. Japanese participants displayed the most distinctive characteristics of all groups. They were more hesitant at the start of sessions, and more wary of making mistakes than other groups.
2. Japanese subjects were also extremely conscientious, three of the four asking at the start of sessions if they should write the answers.
3. All Japanese participants commented on, and appeared pleased by, the association of Shiseido, a Japanese cosmetic company, with the site. Japanese people traditionally value company loyalty (Hoft, 1995).
4. French and Japanese participants showed a strong sense of being personally responsible for errors. Comments such as "Iím sorry. Iím stupid.", and "That was my fault." were common among both these groups. Schriver (1995) notes that learners are often quick to blame themselves for problems that arise from inappropriate presentation techniques.
5. In terms of non-verbal behavior, the Japanese users were also distinctive from the Europeans involved in the study. They were all manifestly reticent during the opening minutes of sessions, but opened up considerably towards the end. Spanish participants were notably talkative from the outset.
6. French participants were most satisfied with the site, possibly because it is best suited to this audience. However, a sense of national pride, and a greater interest in French culture and heritage may explain this trend.
7. Spanish participants were considerably less focused than others; three of the four became sidetracked on more than one occasion.
8. Irish subjects seemed more at ease with the technology than others. This may reflect the difficulty that other subjects had in "thinking aloud" in a foreign language. However, the growth in the Irish technology sector and the comparatively advanced computing facilities at the University of Limerick may also play a role in increasing confidence among Irish users.
On a well-organized system users should have a clear idea of:
1. How the system works.
2. How the information is organized.
3. How much information is available.
4. How this information can be accessed.
When users make errors or comment that they expected an action to produce a different outcome, it is probable that the organization or layout is incompatible with their own mental model of the system (Ravden and Johnson, 1989). This siteís organization proved to be at odds with user expectations on many occasions. The layout of the main menu confused most users who expected, for example, Permanent and Temporary Collections to be located together on the menu.
The Collections menu divides works into categories of art ó Paintings, Sculpture, Antiques, and so on. Two Spanish participants were uncomfortable with the temporal arrangement of information within these categories. In the Paintings category, for example, works are divided by century. These participants would have liked a facility whereby they could access all watercolor or still life paintings, or all paintings by a particular artist, directly from this menu. The current format assumes a basic knowledge of art which not all users will have.
The paragraph layout of some pages was useful only for French participants, who may expect information to move from the general to the specific (Subbiah, 1992). Alternative formats such as tables might alleviate problems of users from other cultures. In addition, three Japanese users suggested that a single column spread would be more accessible for Japanese users than the current two-column layout.
However, the site proved learnable for all user groups. Despite initial problems with some tasks, participants did not have difficulties performing similar subsequent tasks.
On the Louvre site, graphics were fast to load and many subjects from all groups remarked on this even before debriefing. Arnold (1998) comments that fast-loading pages are the single most important feature for Japanese users. This study suggests that they are even more important for Irish users, with two of the four naming fast-loading pages as their favorite feature of the site. According to Nielsen (1999) pages that are slow to load are still very frustrating for all users.
Menu labels, especially on the Spanish version proved problematic. For example, the equivalent of the English label "Visitorís Information" is called "How to Use" (Modo de Empleo). The label on the navigation bar that should lead users to the home page is called "Menu" in English. Only French participants used the home page link, which is called Sommaire (summary or menu) in French.
Japanese users wanted to access all parts of the site in their own language version, and would not have used other parts of the site in normal circumstances, even though, 75% liked the Collections part of the site best. This part is not available through Japanese. Considering that many people will use this site to research a holiday or a hobby, and that in the vast majority of cases they are paying for the privilege of viewing the information, users do not want to wade through foreign-language pages. Quite aside from the high costs involved in spending time online, people want entertainment and leisure activities to be hassle-free and, above all, enjoyable.
Conclusions and Recommendations
An increasing number of Web users come from non-English speaking countries. Therefore, localization of Web sites will almost certainly become, if not essential, more important for international business. However, diverse audience needs exceed the limitations of many sites, including the Louvre site. The main problems with this site are ones that have implications for all multilingual sites:
1. Different groups do have different needs and expectations, even from a medium that has developed in a multicultural environment where the lines that separate ethnic groups are continuing to blur. Therefore, multilingual sites need to provide alternative access and presentation devices to facilitate diverse audience needs.
2. Sites that target a multicultural audience should have an international appeal. While most users enjoyed the site, it was clear that French users were most satisfied with the experience, and were most supportive of the content. Ultimately, this site is more suited to one culture group at the expense of others.
3. Confusing menu labels cause severe problems for users who do not know what to expect when they click on an ambiguous link.
4. Users had difficulty locating diagrams scattered around the site, and many stumbled across them accidentally. Japanese users must use alternative language versions to access diagrams, and indeed other important visuals.
The results of this study, together with previous literature, imply some recommendations, both for the Louvre site and for other multilingual sites of this nature. Primarily, designers should provide:
1. Complete localization of all language versions. Where this is not possible, designers should make clear from the outset which parts of the site are not available in any language version, to reduce user frustration. Likewise, the site should encourage users to use foreign language versions if a feature is not available in their language of choice.
2. Neutral graphics with localized labels, captions, and legends, and access to diagrams from the main menu.
3. Alternative ways to navigate a site, including search tools and a site map.
4. Alternative layout features, such as tables and graphics to suit specific audiences.
5. Link labels that make actions predictable for all users.
6. External links to make the transition between sites more fluid for users.
In addition, usability testing with end-users should be a prerequisite to publishing a multilingual site. One recurring theme of this study is that the usability of a site targeting diverse culture groups depends to a great extent on flexibility. The findings of this study indicate that further research is necessary in the areas of multilingual site structure, user navigation techniques, and reasons why sites appeal to one particular audience above others.
On the positive side, the Louvre site demonstrates that neither technology nor art remain a domain for experts. Both can be used and enjoyed by users who have no expertise in either field. Users do not need a strong background in technology to use the World Wide Web. Nor do they need an extensive knowledge of art to appreciate a site such as this one. Sites like this one bring art to communities that would otherwise have no opportunity to partake of a physical visit to the Louvre Museum. Most importantly, this study emphasizes the ability of pictures (as opposed to words or presentation techniques) to transcend cultural boundaries. Users across all groups admired the samples from the art collections.
Arnold, M.D. (1998). Building a truly World Wide Web: a review of the essentials of international communication., Technical Communication 45(2), 197 ñ 206.
Bussin, R.. (1998). News you can use ó emerging trends: multilingual Web sites., Clear Point Consultants Inc consulted February 14, 2000. http://www.clearpnt.com/news-candidates/bussin.htm
Fernandes, T. (1995). Global interface design. Boston: Academic Press
Hall, E.T. (1989). Beyond culture. New York: Anchor Books
Hall, E.T. (1990) Understanding cultural differences: Germans, French and Americans. Yarmouth, Maine: Intercultural Press
Hoft, N. (1995). International technical communication: how to export information about high technology. New York: Wiley
Horton, W. (1994). Designing and writing online documentation: hypermedia for self-supporting products. New York: Wiley
Horton, W., L. Taylor, A. Ignacio, and N. Hoft. (1996). The web page design cookbook. New York: Wiley
Kostelnick, C. (1995). Cultural adaptation and information design: two contrasting views., IEEE Transactions on Professional Communication 38(4), 182 ñ 195
Louvre Museum Official Web Site. (2000). Consulted February 14, 2000. http://www.louvre.fr/
Nielsen, J. (1993). Usability engineering. Boston: Academic Press
Nielsen, J. (1996). Top ten mistakes in Web design., The Alertbox: Current Issues in Web Usability Consulted February 14, 2000. http://www.useit.com/alertbox/9605.html
Nielsen, J. (1997). The difference between Web design and GUI design., The Alertbox: Current Issues in Web Usability consulted February 14, 2000. http://www.useit.com/alertbox/9705a.html
Nielsen, J. (1999). The top ten new mistakes of Web design., The Alertbox: Current Issues in Web Usability consulted February 14, 2000. http://www.useit.com/alertbox/990530.html
Nua Online Relationship Management. (1998). Internet surveys. Consulted February 14, 2000. http://www.nua.ie/surveys
Ravden, S. and G. Johnson. (1989). Evaluating usability of human-computer interfaces: a practical method. Chichester: Ellis Horwood
Schriver, K. (1997). Dynamics in document design. New York: Wiley
Subbiah, M. (1992). Adding a new dimension to the teaching of audience analysis: cultural awareness., IEEE Transactions on Professional Communication 35(2), 196 ñ 217
Appendix A: Task Sheet
You are taking a weekend trip to Paris in May and wish to spend Saturday afternoon (May 15) in the Louvre museum. You are using the Louvre Official Web Site to research your visit. Work through the following list of tasks.
1. Select your language of choice from the list.
2. You will be using the metro to get to the Louvre. Which metro station serves the museum?
3. Find the opening hours for the museum on Saturdays.
4.You would like to have dinner after your visit. Find out the opening hours of the cafes and restaurants in the museum.
5. You want a list of publications about the Louvre palace and museum. Can you find such a list?
6. Find out which collections are guaranteed to be open on the day of your visit.
7. Find the email address of the Practical Information department.
8. You would like a map or diagram of the museum in advance to help you find your way around when you get there. Find a map or diagram and print it.
9. You want a ticket that covers temporary exhibitions and permanent collections. How much will this cost?
10. You are especially interested in the works of 16th-century Italian painters. Find the menu of the 16th-century Italian paintings collection.
11. You are also interested in the works of Jean Baptiste Perroneau, a late 18th-century French painter. Print the information available on his Portrait of Madame de Sorquianville.
12. Find out if the Essence of Architecture exhibition (a temporary exhibition) coincides with your visit, and where it is on.
13. Can you find the location of the Essence of Architecture exhibition on the diagram?
14. As well as paintings, you like Egyptian antiques, and have heard that the Louvre has an extensive collection. Browse through the Coptic Egypt collection.
15. Finally, you see a link to the museum online shop. Look through the selection of posters and prints that you can order.
Appendix B: Debriefing Questions
What did you like most about this Web site?
What did you dislike most?
What do you think most needs to be changed?
What additional features would you like to see on the site?
Did you find the pages
Slow Moderate Fast
Would you use this Web site again? Why/Why not?
Agree or disagree with the following statements:
"It was very easy to use this Web site."
"Using this Web site was a frustrating experience."
"This site allows me to do most of the things I think I would need."
"This site is very pleasant to work on."
Comment on the organization of material on this Web site.
Did you think the information on the site was arranged logically?
Did you have any difficulty finding specific information? Give an example.
Comment on the graphical elements on this Web site.
Comment on the color scheme.
Did you think the site had a
French Spanish/Japanese/Irish International
Appendix C: Louvre Map on All Language Versions