Vasja VEHOVAR <firstname.lastname@example.org>
Zenel BATAGELJ <email@example.com>
Katja LOZAR <firstname.lastname@example.org>
University of Ljubljana
The Internet is making the world global, with information instantly accessible from any computer linked to the network. A question that arises is, How much of the textual information will remain in local languages? This paper presents a general overview of this issue and also a specific empirical study of a typical small country (Slovenia). It is argued that the language problem significantly affects the process of Internet adoption. Internet penetration has already reached the majority of computer-oriented and well-educated segments of the population. However, the forthcoming segments (and nations) are much less familiar with the English language. We should therefore shed some light on this important and sensitive language issue. First, the issue of Internet growth is addressed. Some standard models (e.g., Rogers) for dissemination of the innovations are applied for two basic indicators of Internet penetration: the number of hosts per habitants and the percentage of population using the Internet. In the next step the attitudes toward English as the Internet language are studied. The results from international surveys (such as GVU) are presented, and some other specific studies are discussed. Specifically, the Slovenian example is presented in more detail. The RIS (Research on Internet in Slovenia) project provides detailed data on this topic. The data are based on three national Slovenian World Wide Web (WWW) surveys (RIS '96, n=1,220; RIS '97, n=3,500; and RIS '98, n=6,500), several national Slovenian household surveys (n=5,000 to 10,000), and a survey of Slovenian companies and institutions (n=4.300) (detailed description: www.ris.org). It is shown that language matters when the extent and structure of the Internet use are in question. As a consequence, the models of the Internet diffusion are split into two parts: one for the English-speaking part of the population and the other for the non-English-speaking part. The spread has been significantly different. Next, the impact of the level of English language knowledge is discussed in detail. It is, of course, trivial to claim that better English performance improves the intensity and the benefits of Internet usage. The extent of this difference is addressed and the attitudes toward Internet use among non-English-speaking segments of the population are discussed in more detail. Finally, on the basis of data from the RIS project, the language structure of WWW homepages in Slovenia is discussed in the light of the above findings. Also addressed is the issue of the top visited sites where the important dilemma arises: Which sites are the most visited in a small country with non-English original language? Are these the local sites, or the global ones, such as Yahoo? As for now, the top domestic site (www.ijs.si/slo/) and the top international page (www.altavista.com) are approximately balanced. Detailed comparisons are given for the frequency and the structure of the visitors for each group of WWW sites. In the conclusion, some possible future developments are outlined.
The World Wide Web radically increases the use of computer-mediated communication. It also dramatically enhances the process of economic globalization. However, the majority (70%) of the content on the Web is in English; 5 percent is in Japanese, 4 percent in German, and only 20 percent in other languages (Nadeau et al., 1998). The obvious question is whether the very nature of the Web discriminates against non-English-speaking cultures.
The obstacles for non-English-speaking populations can be divided into two groups (Schmeiser, 1996). The first is technology: there is no ASCII analogue for the non-Latin characters used by the Chinese, Arabic, Hebrew, and other languages. These problems with character sets are summarized by Bourbonnais and Yergeau (1996), and the relevant software developments are discussed by Leong and colleagues (1998). In this paper, however, we deal with the second obstacle, the dominance of the English language.
Non-English-speaking countries face several language-related problems. First, it is practically impossible to globally promote a Web site in a non-English language. However, creating and maintaining multilingual pages highly increases the costs. In addition, if a non-English-speaking country (or culture) is small or technologically undeveloped (i.e., with a small number of Internet users), it is also difficult to establish an audience large enough to justify the investment needed for elaborated Web applications. Even for the globally known Web sites, such as The LA Times, The New York Times, The Chicago Tribune, Amazon.com, and Yahoo, the costs of development and maintenance are extremely high and, at least within the first years, the online revenues hardly cover these costs (Vehovar, 1999). Except for a few large non-English-speaking nations, small non-English audiences hardly enable the development of Web applications such as bookstores, travel agencies, search engines, portals, and news agencies in their own language. On the other hand, these applications have already been successfully developed in English, and users from countries with a non-English language have already been relying on them extensively.
Spennemann and colleagues even say that "while some languages have a large number of speakers and territorially cohesive populations, such as German, French, Spanish, Russian and Japanese, stand a good chance of survival in the electronic world ... others, such as Marchallese or Pohnpeian (each with less than 100,000 speakers worldwide) do not stand any chance at all" (Spennemann et al., 1996).
In this paper we discuss the above statements with some further empirical evidence and a more detailed discussion of future developments. As a case study we use a typical non-English-speaking country -- Slovenia.
With Internet expansion, the question arises about the limits of Internet penetration. At this point, of course, we speak of the Internet as it exists today, at the end of the millennium: awkward, relatively expensive, separated from TV, and with limited links to the entertainment industry. In such a setting one possible limit is simply the percentage of the population that uses personal computers. In developed countries this is often between 30 and 50 percent of the active population. However, in non-English-speaking countries, the lack of knowledge of English presents an additional barrier to a regular (or intensive) use of the Internet.
In this section we first discuss some data on Internet penetration. This data is only illustrative, as we do not consider factors, such as development level of countries, financial and technical resources, history and culture, and type of government. Several authors found that these are all good predictors of Internet penetration (Arnum and Conti, 1998; Elie, 1998; Hargittai, 1998). Nevertheless it is an empirical fact that among the countries with the highest percentage of Internet users (1) we find Scandinavian countries (where English language is common) and English-speaking countries (United States, Canada, Australia, United Kingdom, and New Zealand).
Similarly, we can observe the number of hosts per 100,000 persons. Despite some methodological problems, such as firewalls, doubled Internet protocol (IP) numbers, problems with domains .com, .net, .org, and so on, we can confirm the previous findings. Among the countries with the highest density of Internet hosts we find the same countries as above, with a few additional European countries that are also known for their excellent education in the English language. The top 10 countries (July 1998) are Finland (10,079 hosts per 100,000 inhabitants), Iceland (7,774), Norway (7,224), New Zealand (5,089), Sweden (4,335), Australia (4,205), Denmark (3,656), Canada (3,513), Netherlands (3,346), and Switzerland (2,939)(2) (source: Network Wizards). Again, we cannot prove causality here; however, the data clearly indicate a link between the status of the English language and the corresponding use of the Internet.
Elie (1998) reached similar conclusions by analyzing the relationship between the gross domestic product per capita and the number of hosts in European countries. He showed that the Internet is more widespread in highly developed countries of northern Europe (Finland, Sweden, Norway, Denmark, the Netherlands, and Great Britain) where the English language is commonly used. On the other side, in developed countries of central and southern Europe (France, Italy, Spain, and Greece) -- and also in many others -- with lesser use of English language, Internet penetration is much less (Elie, 1998).
As an example we can compare Ireland with Slovenia or Portugal. These are all small countries with approximately the same level of economic development. However, in Ireland, with English as the primary language, any Web application is automatically available to the global audience. The opposite is also true: All domestic users are automatically global users, which places Ireland in a much better position.
In this section we discuss an example of a country (Slovenia) in which a non-English language stands as a certain barrier to a more intensive use of the Internet. We were also inspired by data from the 9th Graphic, Visualization, & Usage (GVU) Center's User Survey from 1998 in which 42 percent of respondents whose native language is not English claim to know person(s) who could (or would) not learn to use the Internet because of language difficulties.
We will use empirical data on Slovenia provided by the project Research on Internet in Slovenia (RIS), an academic project at the Faculty of Social Sciences, University of Ljubljana. Slovenia, with two million inhabitants (speaking a specific Slovenian language), is located in Central Europe, between Austria, Italy, Hungary, and Croatia. Its Internet penetration is moderate with 1,200 hosts per 100,000 people at the beginning of 1999 (source: RIPE). One-third of the active population (15-65 years) had already accessed the Internet by the end of 1998, and 12 percent of the active population are monthly (9% weekly) users.
In this paper we use data from the following surveys:
In the RIS '98/I telephone survey of the general population (age 15-70), 4 percent of the respondents labeled their knowledge of English as fluent, 16 percent as good, and 22 percent as partial. The remainder of the respondents labeled it as poor (24%) or nonexistent (34%). At this point we can add that Slovenia stands here as perhaps a typical non-English-speaking country in which English is regularly taught in elementary schools (age of 10 years and above).
Chart 1: Percentage of persons speaking English among the general population (age 15-70), among regular Internet users, and among users who responded to the Web survey (3) (source: RIS '98/II telephone survey, RIS '98 Web survey).
We can observe a strong relationship between English language and the use of Internet. In the first group (fluent in English), 75 percent used the Internet, but only 1 percent in the last group (non-English speakers) did so. Age and education are the major factors determining both the use of the Internet and knowledge of English: Younger and higher educated persons speak English better and also use the Internet more often. However, the level of familiarity with the English language stands as the strongest predictor of the Internet use in all sociodemographic subgroups.
Of course, the English-speaking population was also the first to use the Internet. Chart 2 shows that almost all experienced users speak English, but only 40 percent of new users speak English well or fluently (source: RIS '98/II telephone survey). There is a certain difference between the telephone and the Web survey but the relation (language versus years on the Internet) is clearly visible in both surveys. We definitely expect that in the future new Internet users will be much less familiar with the English language.
Chart 2: Percentage of Internet users speaking English fluently or well with respect to the year of their first usage (source: RIS '98/II telephone survey, RIS '98 Web survey)
In a small non-English-speaking country, the incoming flow of data through the Web is much larger than the outgoing flow. Arnes, the Slovenian national public Internet service provider (ISP), reports that more than one-third of the network's flow is international. With commercial ISPs the international traffic may reach up to 60 percent (Vehovar, 1999).
Similarly, Slovenian users perform much of their Internet activities on foreign Web sites. The intensive users in the RIS '98 Web survey claim that on average only one-third (31%) of their time spent on the Web was dedicated to Slovenian Web sites. Half of them spent more than 75 percent of their time on foreign sites. Of course, among foreign sites, U.S. Web sites strongly dominate.
There is a remarkable difference between those who speak English and those who do not. The users who labeled their knowledge of English as fluent spent only 25 percent of their time on Slovenian Web sites, while people with less knowledge in English spent more than 35 percent of their time on Slovenian Web sites. At this point we should add that in the self-selected RIS '98 survey on the Web, only 3 percent of users had poor and 9 percent had partial knowledge of English.
The Slovenian example of extensive use of foreign Web sites is not an exception. Even in Sweden, with more than double Internet penetration and a population four times larger than that of Slovenia, the customers, to a large extent, use global services. For example, three U.S. sites are among the top 10 sites most visited by Swedish Internet users (Top Ten Sites, 1998).
The analysis of the most popular Web sites among Slovenian users is also very illustrative. In the RIS '98 Web survey Web site visitation was recorded from answers to an open-ended question about the respondents' most frequently visited sites (recall question). Among the 20 most frequently mentioned sites are 7 foreign Web sites. Foreign sites are marked with a star (*) in table 1. Of course, all of them are in English.
|Web Site||Number of Respondents||% of Respondents|
|Slovenian national Web page (Mat'Kurja)||1103||54.7|
|Slovenian Telephone Directory||361||17.9|
|ARNES-Academic and Research Network of Slovenia||220||10.9|
|National Weather Report (HMZ)||158||7.8|
|Institute Josef Stefan||156||7.7|
|Slovenia Online (Internet Service Provider)||153||7.6|
|Daily newspaper DELO||90||4.5|
|Slovenian virtual library Cobiss||73||3.6|
|Student Internet provider KISS||70||3.5|
|Agency for Financial Transactions||68||3.4|
|Ljubljana Stock Exchange||65||3.2|
|Si21 (News online)||60||3|
The top visited sites were explicitly measured also in the RIS '98/II telephone survey and the results were relatively similar, with Yahoo and AltaVista being far ahead. Again, a clear difference exists among those who speak English and those who do not. Chart 3 shows the percentage of Internet users claiming that they visit these Web pages monthly or more often. We can observe that English-speaking users are more frequent visitors of almost all sites. In addition, we can also observe, of course, a large difference in visits of English pages.
Chart 3: Percentage of those who visit Web pages monthly or more often according to their knowledge of English (source: RIS '98/II telephone survey)
In a small non-English-speaking country, the number of regular (weekly) Internet users is often far below the number needed to pay off the investments in sophisticated Web applications. Commercial Slovenian Web sites with more than 10,000 unique visitors per month are thus relatively rare. That is also true for many shops, newspapers, bookstores, and CD shops. Bookstores are a particularly good example: A dozen bookstores in the domestic language have sophisticated applications (secure server); however, the RIS '98 Web survey showed that in 1998 they sold only a fraction of what Slovenians bought through Amazon.com.
Of course, the Internet enlarges the opportunities for companies from small countries. The companies are aware of this, particularly of the communication component of the Internet. The RIS '98 telephone survey of Slovenian business companies in 1998 showed that 90 percent of large (more than 250 employees) and 80 percent of medium (50-250 employees) companies had access to Internet, and 42 percent of large companies and 31 percent of medium companies had their own Web site. However, only a few of them offered a type of application or online ordering of their products. To illustrate the language issues we should add that half (53%) of the Web sites were only in Slovenian language, 33 percent were bilingual, and 14 percent were only in English. These numbers are perhaps a good illustration of the dilemmas on the language strategies of commercial Web sites in a non-English country.
In the RIS '98 Web survey 53 percent of all respondents agreed with the statement that important Web pages should be translated into Slovenian language. However, a smaller percentage (20%) agreed with the statement that they would use the Internet more often if there were more contents in Slovenian language (see chart 4). This percentage rises to 67 percent only among those whose knowledge of English is poor. Again, prevalence of English on the Web is an important barrier to the more intensive use of Internet among the non-English-speaking part of the population. The non-English speakers are simply required to master the English language in order to profit from Internet resources to the extent that English speakers can.
Chart 4: Agreement with the statement about the use of Internet if there were more contents in Slovenian language (source: RIS '98 Web survey)
A similar result is found also in the 9th GVU's User Survey from 1998. One-third (36%) of all respondents whose first language is not English agree with the statement that having access to more Web sites that are designed for their language and culture would make them more willing to use the Web. Since they speak English (otherwise they would not complete the GVU's English questionnaire) the percentage is comparable to 9 percent of Slovenian Internet users (RIS '98 Web survey was, of course, in Slovenian) whose English is fluent or 24 percent of those whose English is good.
A slight discrepancy with Slovenian Internet users is in tune with the fact that they are generally enthusiastic about the opportunities the Internet offers to small countries. For example, they disagree with the statement that the Internet would lead to cyber-imperialism of the United States, and they also strongly agree that the Internet would improve the situation of small countries (source: RIS '98 Web survey).
Rogers (1995) gave one of the soundest theories on diffusion of innovations. According to his theory, we could treat the Internet as an interactive innovation; its value increases with every additional user. Rogers developed a general model for the process of adoption of innovation and he also classified the adopters (consumers) into standardized categories, each with very specific social and psychological characteristics: innovators, early adopters, early majority, late majority, and latecomers (Rogers, 1995). The upper line in Chart 5 shows the shares of these categories.
Chart 5: Groups of adopters according to Rogers (Rogers, 1995)
Here, we apply a part of this theory to the process of Internet penetration in Slovenia. Of course, we apply it in a simplified form that serves only as a model for understanding the language issues, and not as a prediction-model for real trends in Internet penetration.
As the rapid growth of Internet penetration in Slovenia has slowed down in 1998, the growth from innovators, early adopters, and early majority has already been exhausted, and the turning point has been reached. In future the number of Internet users will thus grow more slowly.
Let us construct some brief estimates. Thirty-five percent of the active population (15-65 years) (4) in Slovenia are regular users of personal computers (PCs) (source: RIS '98/II telephone survey) and we can roughly treat them as the upper limit for potential Internet adopters (5). This was also confirmed in the answers to a question about the future use of the Internet that made it obvious that all potential Internet users would be recruited almost exclusively (31%) from this very group: Besides the 12 percent of existing (monthly) Internet users, an additional 13 percent have already (6) used the Internet but do not use it (or they don't use it regularly), and 6 percent of current PC-users still intend to use the Internet. However, the majority (61%) of the active population does not plan to use the Internet at all and the largest part of them (43%) even chose the strongest option ("definitely not").
We thus assume that ideally, within a given environment, up to 35 percent of the active population would adopt the Internet. As already mentioned, the upper line in Chart 5 is the pure theoretical distribution of the adoption groups according to Rogers. This is a standardized normal curve with the area below the curve totaling 100 percent. In our example, however, it also denotes the process of adoption if there were no barriers in language. By the end of this process the whole group (i.e., approximately all PC users) would adopt the Internet. The bottom line indicates the same process with an additional language barrier. The construction of the line is based on the above-discussed data on PC usage; data from the past, present, and future (intended) Internet use; and data on the knowledge of the English language. Of course, the area below this curve is less than 100 percent, which simply means that fewer people will adopt the Internet, roughly 40 percent less (in our example this would mean 21% instead of 35%). Clearly, certain parts of the adoption groups are missing; however, almost nothing is missing from innovators, but the large majority is missing from the latecomers.
Time is not organized in an interval scale on the horizontal axis in Figure 5. It is ranked only according to the theoretical adoption groups presented with the upper line. (Here, of course, we always speak of adoption groups exclusively in terms of the upper curve, and not in terms of the bottom line). Thus the upper line also does not necessarily achieve the maximum later, but that can happen, too, as was the case in Slovenia. In general, on a usual interval scale the adoption groups would receive large differences in two lines; the bottom line would be much flatter because the language barrier makes the adoption more limited and also much slower. As the majority of late adopters are not familiar with English, they cannot benefit from the existing content on the Web. They have to wait for enough relevant content in their domestic language to appear on the Web.
We showed that knowledge of the English language is an important factor when discussing Internet penetration, and in the previous section we showed five different aspects of its negative influence on Internet development in a non-English-speaking country. We also showed that Internet use is substantially less spread throughout non-English-speaking countries, and also within non-English-speaking sub-population groups. Obviously, the content is the substance of the Web, and when the content is not present, users have nothing to consume.
It is a simple empirical fact that without knowledge of English, Internet users cannot fully use the possibilities of the Internet. With a non-English language the question often arises whether there is enough content at all to become a user. Family tree research and disease history cases are perhaps the most typical subjects of interest to the older inhabitants who cannot learn a language but are still interested in exploring certain topics. It will perhaps take years for such areas to be offered in languages that are limited to a few million inhabitants.
Of course, the basic language strategy for successful orientation on the Internet is to teach the nation English and, as for now, there is no surrogate for this. As long as potential Internet users are not familiar with English, the full potential of the Web cannot be realized. The non-English-speaking countries must develop two additional strategies for the Web language problem, one for forthcoming young generations who are familiar with English; for them the question of the language on national Web sites is more an issue of national identity as they understand enough English to use the Web. Of course, the issue of national identity is extremely sensitive and complicated but very practical; for example, it touches the question of extent of (expensive) bilinguality of the Web pages. For older generations, who are not only unfamiliar with English but are also unfamiliar with computers, the issue is even more problematic, particularly because these people often comprise the majority of the active population. They are generally willing to learn the Internet (not the English), but only if the sheer volume of the content is sufficient. With this group, the solution is clearly reduced toward straight maximizing of the amount of domestic content on the Web.
At this point we can conclude that more contents are definitely needed in domestic languages. This is also a quest of many users and analysts. This is also the opinion of the Internet users from the 9th GVU's User Survey from 1998. In fact, 79 percent of respondents are convinced that in order to bring additional users to the Web, it will be necessary to provide Web sites in their primary language. However, Web content grows slowly, only in close interaction with the consumers and only when the critical mass of the audience exists. It also grows only (or predominantly) on a commercial basis, which may contradict the requests for more content in non-English languages.
Of course, we are now discussing the Internet as it exists in 1999 -- and this may have no relevance to the Internet in the next millennium, particularly if significant progress is achieved in application-building, bandwidth, and navigation. However, certain aspects of the language barriers will still remain problematic. Hopefully, with time the problem will lose some of its edge as expressed in the statement that "a country (or language) has to be highly present with several million Internet users to create an environment less dependent on that created by English-speaking users" (Hargittai, 1998).
The Internet has therefore (another) paradoxical role. It ensures global communication and gives new opportunities to small nations and cultural and other minorities. They can now easily communicate, present, and express themselves globally. On the other hand, the overwhelming majority of the most attractive content on the Web has already been created, but exclusively for those who accept it in English.
1. Arnum, Eric, and Conti, Sergio (1998): "Internet Deployment World-Wide: The New Superhighway Follows the Old Wires, Rails, and Roads." Paper presented at the INET98 Conference, http://www.isoc.org/inet98/proceedings/5c/5c_5.htm.
2. Bourbonnais, Jean, and Yergeau, François (1996): "Languages on the Internet." Paper presented at the INET96 Conference, http://www.isoc.org/inet96/proceedings/a5/a5_3.htm.
3. Clemente, Peter (1998). The State of the Net. The New Frontier. McGraw-Hill.
4. Elie, Michel (1998): "The Internet and Global Development." Paper presented at the INET98 Conference, http://www.isoc.org/inet98/proceedings/5d/5d_3.htm.
5. Hargittai, Eszter (1998): "Holes in the Net: The Internet and International Stratification." Paper presented at the INET98 Conference, http://www.isoc.org/inet98/proceedings/5d/5d_1.htm.
6. Farinelli, Marco (1996): "Little Italy, Wide World." CMC Magazine, 8, http://www.december.com/cmc/mag/1996/aug/farin.html.
7. GVU's 9th User Survey (1998): Graphic, Visualization, and Usability Center. Datasets and questionnaires available at http://www.cc.gatech.edu/gvu/user_surveys/survey-1998-04.
8. Leong, Kok Yong, Liu, Hai, and Wu, Oliver P. (1998): "Web Internalization and Java Keyboard Input Methods." Paper presented at the INET98 Conference, http://www.isoc.org/inet98/proceedings/5f/5f_2.htm.
9. Nadeau, Jocelyn, Lointier, Cécile, Morin, René, and Descôteaux, Marc-André (1998): "Information Highways and the Francophone World: Current Situation and Strategies for the Future." Paper presented at the INET98 Conference, http://www.isoc.org/inet98/proceedings/5f/5f_3.htm.
10. Network Wizards Internet Domain Survey. (1999): www.nw.com/zone/WWW/top.html.
11. NUA: How Many Online. (1998): http://www.nua.ie/surveys/how_many_online/index.html.
12. Remec, Matija and Vehovar, Vasja (1998): "Raziskovanje Interneta." In Vehovar (ed.): Internet v Sloveniji. Desk, Izola.
13. RIPE Internet Statistics. (1999): www.ripe.net/statistics/index.html.
14. Rogers, Everet M. (1995): Diffusion of Innovations. 4th ed. Free Press.
15. Schmeiser, Lisa (1996): "Truly World Wide Communication?" CMC Magazine, 3, http://www.december.com/cmc/mag/1996/mar/callwww.html.
16. Spennemann, Dirk H. R., Birckhead, Jim, Green, David G., and Atkinson, John S. (1996): "The Electronic Colonization of the Pacific." CMC Magazine, 2, http://www.december.com/cmc/mag/1996/feb/spen.html.
17. Top Ten Sites. (1998): RelevantKnowledge Sweden. http://www.relevantknowledge.se.
18. Vehovar, Vasja (1999): "Prospects of Small Countries in the Age of the Internet." In: Ebo (Ed.): Cyberimperialism? Global Relations in the New Electronic Frontier. Greenwood Publishers.
19. Vehovar, Vasja, and Batagelj, Zenel (1998): "Uporabniki Interneta v Sloveniji. Projekt RIS '96-'98." In Vehovar (ed.): Internet v Sloveniji. Desk, Izola.
20. Vehovar, Vasja, Batagelj, Zenel, and Lozar, Katja (1998): "Who are Non-respondents in WEB Surveys." Paper presented at the 9th International Workshop on Household Survey Non-response, Bled, Slovenia, http://www.ris.org/ris98/bled/index.htm.
1. Percentage of users (who used the Internet at least once during the past 3 months) among the adult population in the top 10 countries (source: NUA: How Many Online): Iceland 45%, Finland 35%, Sweden 33%, United States 28%, Canada 26%, Australia 23%, Denmark 22%, Switzerland 16%, United Kingdom 16%, and New Zealand 16%.
2. The data for the United States are not available because of previously mentioned methodological problems with domains .com, .net, .org, etc.
3. Similarly to GVU's user surveys, Internet users who respond to RIS Web surveys are more intensive, more experienced, younger, and higher educated users; there is also a higher percentage of men (Vehovar et al., 1998).
4. During the past six years this percentage has been stable and it increases by 1 percent each year only because of the exchange of generations as 2 percent of young incomers (15 years old) -- a large majority of whom are regular users of PCs -- replace those who passed away (here, PC users were large minority).
5. Of course, this upper limit (35%) also undertakes an (unreal) assumption that there will be no dramatic change in technology (e.g., PC-TV). However, studying one effect (language) we must fix the others.
6. This can also be evidence of the lack of sufficient content: The users tried the Internet, but they perhaps found it not interesting enough.