There are two broad types of localization issue inherent within design. First, there are ‘hard’ issues such as language and format conventions. Although important, these issues are not my concern here. This article focuses on the ‘soft’ issues that determine the ways in which people in different cultures interact with and respond to websites. As I will show, these underpinning cultural and cognitive dimensions have major implications for international website acceptance.

Through both commercial internationalization projects undertaken by optimum.web limited and associated academic research, I have been exploring the role that culture plays in the successful localization of websites. I will discuss here some of my commercial and research experience, and present my current views on what needs to be achieved in order to better understand the cultural localization of websites.

Culture and Website Acceptability

Researchers working in this area often refer to the concept of ‘cross-cultural usability.’ I prefer to use the term ‘acceptability’ to establish clear differentiation from usability and other associated issues. A website may be considered to be highly usable in all the usual usability senses, but it may not be socially acceptable for various reasons.

At a basic level, usability may be evidenced in a site that is ‘simple to use.’ However, a merely simple site may not be fully acceptable to the target user group. Many other factors, such as aesthetic design and issues relating to branding and trust, determine the way in which users will respond to websites and the extent to which they will adopt them. I define international website acceptability to be: ‘the extent to which the site delivers a user experience that matches local cultural needs and thereby leads to high levels of adoption’. The problem is that people differ across regional, linguistic and country boundaries, and user acceptance of websites is driven principally by their local cultural perspective.

Underpinning Cultural Theories

Cross-cultural acceptability is about making websites an effective means of communication between a global website owner and a local user. Internet-based communication may be a relatively new phenomenon, but in order to fully understand it, we need to start by investigating the much wider research area of intercultural communication.

Although the roots of such study can be traced even further back, the anthropologist, Edward T. Hall, established the original paradigm for intercultural communication. Hall described culture as a selective screen through which we see the world, and believed that basic differences in the way that members of different cultures perceived reality were responsible for miscommunications of the most fundamental kind.

In spite of a wide range of research, it is Geert Hofstede’s dimensions that are most often quoted in relation to cross-cultural acceptability. He conceptualized culture as ‘programming of the mind’ in the sense that certain reactions were more likely in certain cultures than in other ones, based on differences between basic values of the members of different cultures.

Hofstede carried out a study of 116,000 IBM employees distributed through 72 countries using twenty languages in 1968 and 1972. The study was based on rigorous research design and systematic data collection. He proposed that cultures could be defined through three dimensions:

• power distance (PD), the degree of emotional dependence between boss and subordinate,
• individualism/collectivism (IC), integration into cohesive groups versus being expected to look after him/her self, and
• masculinity/femininity (MF), which can be interpreted as toughness versus tenderness.

In addition, for Western cultures another dimension, uncertainty avoidance (UA), represents the extent to which members feel threatened by uncertain or unknown situations. For Eastern cultures, long-term Confucian orientation represents a philosophy of life that is prepared to sacrifice short-term results for long-term gain. Table 1 reproduces a summary of some of the scores that Hofstede provided.

Table 1. Example of Hofstede Scores.

Although Hofstede’s work was not oriented towards website acceptability, his approach is often applied to this field. Most notably, Aaron Marcus provides guidelines for design based upon high and low characteristics of the dimensions. Although there is no lack of theories of culture that may be used when addressing website cultural localization, there is a lack of explicit demonstration that such theories of culture are actually applicable. I will discuss some of my work to date in verifying such theories.

Presenting Cultural Differences to Commercial Clients

A key problem that I have found in implementing these concepts at optimum.web limited in London is the need to provide an accessible means through which the cultural characteristics of website and target users can be discussed with clients. In response to this need, we are developing the concept of a ‘cultural fingerprint’ which can diagrammatically compare the cultural profile of a website (Site Fingerprint) with that of its target cultures (Country or Culture Fingerprint).

Initially, we developed the ‘first-shot’ fingerprint that addresses four of Hofstede’s dimensions and presents Hofstede’s data on a scale from 0 to 10. Taking power distance (PD), for example, Hofstede lists PD scores for 53 countries from a score of 11 for Austria (low PD), to 104 for Malaysia (high PD), with the UK scoring 35. On our 10-point scale, the UK scores 2.58 [10 x (35-11)/(104–11)].

We then adopted a ‘radar diagram’ approach using the score for PD, IC, MF and UA, as shown in Figure 1 below. We refer to this as a ‘first-shot’ fingerprint, as the diagram suggests that each of the four dimensions are equally important, and indeed that they are actually significant at all. Later on, I will show how we are developing the concept to take into account the actual significance of Hofstede’s cultural dimensions within specific countries/cultures.

Figure 1. UK ‘First-shot’ Country Fingerprint.

Site fingerprints are developed from expert evaluation. In this case, up to five of our consultants rate each of the four dimensions on a scale of 0 to 10. They assess against optimum.web criteria that are based upon those proposed by Aaron Marcus, and the average value is used to determine the site fingerprint. By comparing country and site fingerprint, the usability/acceptability of websites can be matched to the target culture. It is also possible to develop a weighted average fingerprint for all target cultures, based upon a known usage profile, and thereby to evaluate whether a ‘cultureless’ standard has been achieved. We have implemented the approach with a number of international clients in the e-finance and e-commerce sectors, such as Land Rover, Six Continents, Amway and HSBC.

Figure 2. Corporate Investor Websites.

Commercial Case Study: Corporate Websites

In the UK, the Financial Times Shareholder Communications survey recently ranked the websites of all members of the FTSE Eurotop 300 Index from the point of view of the international investor. In order to explore the acceptability of these sites to a global audience, we took a subset of sites roughly equally spread out over the FT’s listing. Through expert evaluation, the site fingerprint of each site was generated. We also produced the cultural fingerprints of four countries/cultures in which UK companies seek significant shareholder investment (the U.S., Germany, Japan and Arab countries) and could thereby compare sites with users. Figure 2 shows the site fingerprint of two of these sites (Hanson and ICI), together with those of two key investor countries (the U.S. and Japan).

There are some very interesting differences. The Hanson site was the best of our sites from the FT survey, but its site fingerprint did not seem to match many of the target cultures. However, it could be an effective mode of communication with people in Japan. The ICI site is almost a mirror image of that of Hanson and is quite suitable for a U.S. audience. The question here is: are these two companies really aiming their communications to totally different client and cultural groups? Probably not.

Verifying Cultural Dimensions: Research in China

Although there is no lack of theoretical underpinning for cross-cultural acceptability, there is a lack in explicit demonstration that such theories are actually transferable to our field. Theories such as Hofstede’s were formed some time ago, and while culture is generally agreed not to change too fast, the models were not developed with the intention of being applied to websites. Additionally, there are views that being a global phenomenon, Internet issues may apply globally, thereby transcending local concerns. I have been interested, therefore, in conducting studies with end users to determine the extent to which cultural factors do actually affect international acceptability.

In order to verify the applicability of cultural dimensions to website acceptability, we are undertaking a range of research projects. On the basis that it is an important developing market, I will share some results from research being undertaken in China. Our aim is to determine how strong each of Hofstede’s dimensions may be to Chinese users. My approach to undertaking user studies in cultural usability is influenced by Taguchi’s work in TQM and adopts the factorial design method first introduced by Fisher in the 1920's. Analysis of variance (ANOVA) is used to determine the importance of different factors, in this case, Hofstede dimensions.

A number of studies in China (Beijing, Dalian and Kunming) have been completed in which Chinese users were asked to undertake a series of tasks on Chinese websites. The sites chosen possessed different levels of the factors of power distance, individualism/collectivism, masculinity/femininity and uncertainty avoidance. After accessing the sites, users were then asked to complete a relatively simple quantitative survey instrument, designed to elicit their overall ‘acceptance’ of the site.

The detailed ANOVA is not presented here but, taken as a whole, the four dimensions were found to be significant in determining user acceptance. Roughly 35% of the entire variability within user acceptance in our studies boiled down to these four issues. As many other factors will also have had an effect (such as other differences in the sites we chose, and differences in the users themselves), it is clear that Hofstede has a role to play in website acceptability. However, there were marked differences in the four dimensions. Table 2 summarizes the results from several research projects, and Figure 3 provides a graphical representation.

Table 2. Hofstede Research: Summary for China.

A number of detailed findings emerge from this study. First, by far the most significant dimension is that of power distance. The preferred level was, as predicted by generic cultural models, that of high PD. Contrary to expectations, individualism/collectivism is not important to Chinese users, and global expectations of the web would seem to predominate. Uncertainty avoidance is also not important, but perhaps we should expect this from Hofstede, who suggested that long-term orientation is more significant. Chinese users were found to prefer masculine sites, an interesting result, as there is little current evidence in this area. Power distance is very important, and masculinity/femininity less so, but international website designers would be ill-advised to ignore either of these two issues.

Figure 3. Enhancing the Fingerprint.

In order to take into account the actual significance of cultural dimensions, we have now developed an enhanced fingerprint that uses ‘circle sectors,’ as opposed to radar diagrams, as the mode of representation. By so doing, we are able to introduce a second dimension to the diagrammatic representation. The one-dimensional ‘first shot’ fingerprint only addresses the magnitude of each issue (e.g., 2.58 for UK power distance), whereas the two-dimensional ‘enhanced’ fingerprint can address both magnitude and importance.

Figure 4. China Enhanced Cultural Fingerprint.

In the enhanced fingerprint, the sector angle (θ), represents the importance of the factor within the culture/country. The sector angle will be the same for both site and culture fingerprint. The radius of the sector (R) is determined by the magnitude of the factor in the culture - either from Hofstede scores for the country fingerprint or by expert evaluation for the site fingerprint. The enhanced culture fingerprint for China is presented in Figure 4 above. Here, the sector angle for each factor is obtained from the relative percentage importance as given in Table 2. Power distance was by far the most important factor, so it has a much larger sector angle. The radius of each sector in the enhanced fingerprint is obtained from the best available score for that factor/dimension in the target culture/country.

It should be noted that Hofstede’s initial work, being based on IBM employees, did not include China. Our results here include a mix of more recent scores for China and a weighted average of scores for Hong Kong, Taiwan and Singapore. The equivalent Hofstede scores adopted were, therefore, PD=80 (7.4 on scale 1-10), IC=15 (1.1), UA=35 (2.6), MF=50 (5.0).

Discussion

The question we are beginning to answer is the extent to which generic cultural issues really affect website acceptability, and how this information can be easily communicated. There is a lively debate in the cultural localization community between those who attempt to build upon cultural theories and those who claim that their relevance is at best ‘not proven’.

One problem is the lack of equivalence between country and culture, and the fact that many countries have sub-cultures. As someone who is managing a European Union-funded project promoting usability in India, I am very aware that applying data at the country level hides very considerable internal differences. Another issue is the global nature of the web itself, and the extent to which the web is a truly global phenomenon, operating irrespective of culture.

From the studies in China discussed here, and from other work, I believe that we do have evidence that Hofstede has a significant role to play. However, considerable caution needs to be adopted when applying generic models of culture to the world of the web. Certainly, not all factors are equally important, and website owners and developers are advised to concentrate on those factors that are significant. The problem is that we are only beginning to understand which ones are important in different countries/cultures.

Much more research needs to be undertaken in this area. Collectively, we need to undertake further empirical studies in other countries/cultures and to investigate factors proposed by researchers other than Hofstede. Following this, it should be possible to develop specific localized design guidelines. I would welcome suggestions for collaborative projects.

Acknowledgments

I would like to thank and acknowledge the work of Yu Chang in collecting and analyzing the Chinese cultural fingerprint data, and optimum.web consultants in providing expert assessment data for the websites discussed.

Dr. Andy Smith is Technical Director of optimum-web ltd., which provides website usability analysis in the UK and within an international context. He is also Director of the U.K. Centre for Software Internationalization. In addition to his commercial and academic roles, Andy is an Executive Committee Member of the British HCI Group and Co-convener of the Indo European Systems Usability Partnership, which aims to promote usability in India. You can reach Andy at andy@optimum-web.co.uk.