Source: Twitter.com

Today, 60% of Twitter’s 105 million registered users are based outside of the United States.

And half of all tweets are in a language other than English.

This is a remarkable trend, particularly since Twitter has only been localized into five languages so far.

A few months ago, I set out to better understand how large, multinational companies are using Twitter to reach users around the world.

I studied more than 225 companies across 21 industry verticals (representing 80% of the Interbrand 100). And I interviewed a number of people who manage Twitter feeds in different markets.

This work resulted in the report Twittering in Tongues. This report is a first stab at a phenomenon that is very much in its early days, so it’s hard to draw any sweeping conclusions. But there are some clearly emerging trends, which I discuss. I also highlight a number of Twitter’s inherent international limitations and provide some recommendations for companies considering localized Twitter feeds.

Here are a few findings/recommendations from the report:

• Most companies have yet to launch international Twitter feeds. Only one-third of the 225 companies studied support one or more Twitter feeds outside of their domestic markets. What makes this ratio interesting is that every one of 225 companies studied supports two or more localized web sites. So these are all companies that do business in three or more countries. A number of companies that support more than 20 local web sites still only use Twitter for their domestic markets.
• Sony leads the pack with support for 20 international Twitter feeds, mostly through its Sony Music division. Microsoft, Cisco Systems, and PricewaterhouseCoopers are also out in front with support for 10 or more country specific Twitter feeds. CAVEAT: Counting feeds is a tricky business. Not all corporate feeds are actively managed (which I did not count) and not all local feeds are easy to find.
  
• Brazil rules. Brazil is by far the most popular Twitter market outside of the US. Nearly half of the companies that support one or more international feeds have targeted Brazil. Not surprisingly, Brazilian Portuguese is the second most popular language used on Twitter.
• Local Twitter success depends on local web site promotion. It’s also no surprise that the local feeds with some of the highest numbers of followers also had high visibility on their local web sites. Companies such as Dell and Samsung lead in this respect. Below is a screen shot from Samsung’s Brazil home page; Twitter gets prime real estate.
• Twitter is local by design. Based on my interviews, most of the in-country Twitter feeds have been launched without any central approval process or even awareness. This also applies to local Facebook and YouTube pages. The evolution is local Twitter feeds is similar to the evolution of local web sites in the 1990s. Back then, local offices often created their own sites, with their own designs and platforms. Over the years, the central offices reined in these disparate sites — sometimes going too far and dampening local enthusiasm. The key challenge I see executies facing now is balancing local control with global consistency. While consistency is important, it should not come at the expense of local enthusiasm and innovation. In the end, the success of local Twitter feeds depends on the local offices.

For more information:

• Report: Twittering in Tongues
• Twitter blog post on global growth

This book provides a strong counterpoint to Tom Friedman’s The World is Flat.

While The World is Flat may leave an executive thinking We have to be in Brazil and Russia and China and India yesterday! Pankaj emphasizes a more measured, sober approach to expanding globally. He also makes a good case for looking beyond the BRIC countries.

Pankaj argues that there are very few truly global companies. Most companies are going through a phase of semiglobalization in which “levels of cross-border integration are generally increasing and, in many instances, setting new records, but fall far short of complete integration and will continue to do so for decades.”

Pankaj says that companies should ask themselves if they should even go global to begin with. At a minimum, he recommends that companies apply his “CAGE distance framework.” CAGE refers to the four types of distance that companies must overcome to succeed in a new market: Cultural, Administrative, Geographic, and Economic.

This is a dense book and it feels academic at times. But don’t let that stop you from reading it. It is an important book and could help many executives avoid a lot of headaches as they invest millions and millions in, say, Brazil or Russia or India or China.

Here is a blurb from a New York Times review:

Very few companies are globally global, Mr. Ghemawat observes. Even Toyota became No. 1 in autos by linking operations within the Americas, within Europe and within Asia, rather than across them. Definitions of region can vary — not just continents but trans-Atlantic, Greater China, trans-Indian Ocean, Eurasia — and Mr. Ghemawat examines a variety of regional hub strategies. But the latter, too, is no strategic panacea: regional platforms can grow into regional fiefdoms.

“Nobody has figured out the optimal way to organize a complex global economy,” he concludes. That is because no single optimal strategy exists. Companies are left to pursue what Mr. Ghemawat labels A.A.A:  — adaptation, aggregation, and arbitrage — or, in straightforward English, multiple variants of individual tailoring.

This map was designed to fit on a cubicle wall and include the major country code top level domains (ccTLDs) that a global Webmaster may encounter.

And although the map includes a whopping 180 ccTLDs, that is not all of them. To include all of them, I would have to develop a much-larger map, which is what I ended up doing.

And here it is, our newest and biggest map, shown below:

This map effectively includes all ccTLDs — 245 in all. Yes, there are more country codes than there are countries. That’s because you don’t have to be a country to have your own country code. The uninhabited Bouvet Island has a code: .bv. Even Antarctica has its own code: .aq.

This map also sizes the country codes based on the population of a given country or territory, which is why China and India feature so prominently. And this map also includes some key statistics regarding country populations and the most popular country codes based on registrations.

To learn more, go to www.bytelevel.com/map/ccTLD.html.

The company says it translates more than a billion words each year into more than 150 languages.

Here are then 10 most-popular languages this year:

1. Canadian French
2. Spanish
3. French
4. German
5. Italian
6. Chinese
7. Japanese
8. Dutch
9. English
10. Russian

�We have observed a strong increase in European language translations � particularly from the Eastern regions as the EU expands,� notes Chris Boorman, chief marketing officer at SDL. �While Canadian French, Spanish, European French and German have held on to the top four spots since 2006, movements further down the list are strong indicators of international business change.�

I’m happy to see SDL sharing this information. It certainly reinforces what I’m seeing on my end. As I wrote two years back, large companies have moved beyond simply translating FIGS languages (French, Italian, German, Spanish) and are now focusing on emerging markets, such as BRIC (Brazil, Russia, India, and China). While FIGS still dominate SDL’s list, the BRIC languages are coming on strong.

SDL also released the following chart detailing the fastest-growing languages overall.

Eastern Europe is certainly getting a lot of attention these days, and not just because the EU requires it across many industries. Deloitte, for example, has over the past 12 months launched Web sites for Serbia, Montenegro, and Bulgaria.

So what languages should we expect to see on the top 10 list next year? According to Chris Boorman:

�Contenders for the future could also include Greece, Nordics and the Baltic states, as there is already a growing demand for translation into these languages,� he concluded. �So we could be in for more surprises in next year�s list.�

Which is what drove me to develop the Map of the World Wide Web.

This map aligns country codes over their respective countries. And it includes a color-coded legend for reference.

So now when I come across, say, .lv, I can know that its Latvia simply by looking at the map.

The map includes 180 of the 250 county codes currently in use. There are actually more country codes than countries, but that’s a whole separate post.

You can order here: www.bytelevel.com/map.

UPDATE: We also offer a larger, more in-depth version of this map at www.bytelevel.com/map/ccTLD.

The article is excerpted from our new Local by Design: Brazil Report.

Here’s what else is included in this issue of Global by Design:

There are choices you must consider when translating for some countries with multiple languages and into some languages that are used in multiple countries.

English is English, Spanish is Spanish, Portuguese is Portuguese, Belgian is Belgian, Swiss is Swiss and Chinese is Chinese — true? Well not quite. For starters, two of the six are not languages (but describe great chocolates!).

This article explains some of the choices you must consider when translating for some countries with multiple languages and into some languages that are used in multiple countries.

I will make some generalizations, but especially if you are taking products into different countries, do seek advice from your local subsidiary or distributor — a consumer-oriented environment is likely to be more sensitive to language variants than business to business.

Most countries have multiple languages and some languages are spoken in many countries. My native United Kingdom counts as native languages English, Welsh, Gaelic (Scottish and Irish), Cornish (in part of the southwest), Manx (on the Isle of Man between England, Scotland and Ireland) and others. However, English is the dominant language understood by just about the entire population and although the Welsh and Scots may disagree, there is little practical need to translate into other UK languages.

The same is not true for some other countries.

Belgium is a relatively “new” country formed in 1830. Dutch (Flemish) is spoken in the north, and French in the south. Although I have worked for a company that translated Belgian French and Dutch separately from the standard French and Dutch varieties, generally, Belgians understand their neighbors’ languages. Belgian Dutch tends to retain traditional Dutch words whereas the more liberal Netherlands Dutch more readily adopts English words. Belgian French and “standard” French are very similar to each other and standard French (for France) generally serves both.

Switzerland has four languages: German (spoken by the majority), then French, then Italian, and about 1% Romansh. Although the spoken Swiss-German is quite different from that spoken in Germany and Austria, it has no written equivalent and the variants of all languages for Germany, France, and Italy work fine in Switzerland; there is no need to generate separate Swiss variants.

Some languages are common to many countries, especially English, Spanish, Portuguese and French.

To be culturally sensitive, especially at a consumer level, British and American English should be treated as separate locales. However, if you are going to pick just one, American English would be the variant of choice as Brits are more used to American English spelling and terminology and generally accept it (sometimes through grated teeth!), but the same is not true vice-versa. Canada probably has a preference for British English, but is more used to American than their transatlantic counterparts. Other English-speaking countries such as Australia, New Zealand, and South Africa also probably have a preference for British English, but American English will be understood and generally accepted.

European French differs from Canadian French in terms of some terminological and stylistic preferences, however the more technical the subject matter, the fewer the differences. European French also tends to more readily adopt certain English words than Canadian French. If your product or service target markets are Quebec and France, it would be advisable to localize your product into the appropriate language for each market (Canadian French for Quebec and “standard” French for France). The similarity between the two variances of French lends itself to being a “make-from” language: once content has been translated into one or the other versions of French, it is feasible to edit the translation in order to be acceptable in the alternate geography (as opposed to translating from scratch).

Spanish needs additional consideration. Each Spanish-speaking country has its own variant and use of words. So Spain is different from Mexico is different from Peru is different from Argentina is different from … well you get the picture! However, especially at a business level, it is possible to translate into “Latin American Spanish”that will be generally accepted throughout Central and South America. European Spanish is your choice if your target is Spain.

One may also translate into a “Universal Spanish” that is generally acceptable worldwide. However, there will be some compromises the “higher” you go and if you are targeting just one or two countries, and have no plans in the near term to expand elsewhere, you would be generally advised to pick the Spanish that most closely matches your target audience.

Portuguese is a different story! Brazilian and European Portuguese have drifted far enough away from each other that you should translate for one or the other; there is no “Universal Portuguese�.” So if you are targeting Brazil and Europe, you require two separate Portuguese translations.

China has numerous spoken languages and dialects, Mandarin being the most universal and common. However, there are two modern written versions of Chinese: Traditional and Simplified. Traditional Chinese was the written Chinese prior to the formation of the People’s Republic of China (a.k.a. PRC or Mainland China) in 1949. Simplified Chinese was derived from Traditional Chinese by the PRC in order to simplify many of the ideographic characters and promote literacy.

Traditional Chinese is used in the Republic of China on Taiwan. Even though the dialect spoken in Hong Kong and Macau is Cantonese, Traditional Chinese is used for print media, but with a few differences as Cantonese requires a few characters that are not used in Mandarin. Singapore has adopted Simplified Chinese and although Hong Kong has seen more Simplified Chinese since its re-incorporation into PRC in 1997, Traditional Chinese still dominates.

Simplified and Traditional Chinese have many similarities, and those accustomed to Traditional Chinese can read Simplified Chinese more easily than the other way around. However, for the People’s Republic of China, translations should be done into Simplified Chinese. Although a PRC native may be able to read Traditional Chinese, they most likely would not be able to write it as well. Thus, the traditional axiom of using a native translator of the target language still applies.

Somewhat strangely you may think, it is easier to translate English into both Simplified and Traditional Chinese than it is translating one into the other. There is a character set issue (Simplified and Traditional Chinese characters do not co-exist in most computer codepages) and translators generally find it easier using English as the source.

So, if your target is PRC (Mainland China) or Singapore, Simplified Chinese is your language of choice. Taiwan, Hong Kong and Macau require Traditional Chinese (with a few differences for Hong Kong and Macau).

I apologize that I have over-generalized to some extent and oversimplified some cultural issues. I am not suggesting that language variants (Belgian French and Dutch) and less common languages (Welsh, Gaelic) are unimportant or insignificant. I have tried to take a practical approach to the localization effort required to take products internationally, and while the Belgian French might prefer their French, they would prefer standard French to English!

About the Author
John Greenwood is Vice President of Business Development for Lombard, Ill.-based InterPro Translation Solutions, which provides translation, software localization, multilingual desktop publishing and project management solutions. He can be reached at (858)486-1848 or jgreenwood@interproinc.com.

Says the Tekrati article, “Four seconds is the maximum length of time an average online shopper will wait for a Web page to load before potentially abandoning a retail site.”

A slow-loading site is second only to high prices in turning away potential shoppers.

A fast-loading Web site is as much a part of a company’s brand as the products it sells.

This issue is even more of an issue for companies that support Web sites in countries that don’t have widespread broadband penetration. Should a company use the same bandwidth-hungry Web site in Brazil, with less than 10% broadband penetration, that it uses within the US?

This is a question every company must ask as it goes global.

A company’s Web localization strategy must take into account the Internet connection speed of the target users. And it also must take into account that Google is successful in many markets outside the US, which means that users around the world have come to expect lightweight, fast-loading Web sites.

For the 2006 Web Globalization Report Card, we “weighed” the home pages of 300 global Web sites. While this does not take into account whether or not that company relies on a partner like Akamai to accelerate Web content delivery, it does shed light on which companies have done the most on their end to keep their Web sites fast-loading. So here are the top 10:

1 Google (www.google.com)
2 Kijiji (www.kijiji.com)
3 Jones, Day, Reavis & Pogue (www.jonesday.com)
4 PayPal www.paypal.com)
5 Wyeth (www.wyeth.com)
6 Manpower (www.manpower.com)
7 AT&T (www.att.com0
8 ST Microelectronics (www.st.com)
9 John Deere (www.deere.com)
10 National (www.national.com)

Google came in at just 13 kilobytes. Most Web sites average around 175 kilobytes. A few of the sites we measured, we won’t name names, came in at more than a megabyte each.

So if you want to provide a fast-loading Web site, keep it under 150 kilobytes to keep it ahead of the pack. These top 10 Web sites all came in under 100 kilobytes.

Today Google also announced the opening of offices in Mexico and Brazil, part of its ongoing effort to expand its presence in Latin America.

Investment bankers use an acronym that is going to have a major impact on the future of the globalization industry: BRIC.

BRIC stands for Brazil, Russia, India, and China, the four countries that pose the greatest opportunities for long-term growth among emerging markets. Of course, the key words are �long term� � these markets are anything but sure bets over the short term.

The localization industry has long used the acronym FIGS, which stands for French, Italian, German, Spanish, the most popular four languages chosen when companies enter Europe. CJK, for Chinese, Japanese, and Korean, is also frequently used when expanding in Asia.

But I see BRIC gaining currency in the localization industry, because where investment bankers see growth so too do the multinationals who hope to capitalize on that growth. And while FIGS and CJK are geographically oriented, BRIC focuses purely on opportunity. This is great news for translators of Tamil, Chinese, Russian, and Portuguese and the vendors who learn to speak �BRIC.�

(NOTE: This essay is from the April issue of Global By Design — on newsstands everwhere!)

Here’s an excerpt:

Whirlpool’s research also revealed that aesthetics were important because washers are considered status symbols. In China, the looks factor is multiplied because “many families keep appliances in the living room.” There’s no place else to put them. In Brazil, Whirlpool jazzed up the control panel with bright yellow buttons and blue lettering. They also carefully selected appliance colors based on by-country preferences . Wash cycles were named on a by-country basis, too (in India, the delicate cycle is called the “sari” cycle, for example). What does one of these machines cost? Just $150 to $200 (about half the average cost in the U.S.). So happy are low-income Brazilians to have these machines that some are said to “treat the washer like a member of the family, referring to her as ‘my little princess’ and my ‘little girl.’”

And here’s a picture of one such machine:

For the full article, go to: http://www.reveries.com/coolnews/2003/december/dec_9.html

The study also notes that “majorities, in most cases strong majorities, in 34 of 44 nations thought the availability of good paying jobs had gotten worse in the last five years. And substantial majorities–82% in France, 67% in the United States, 63% in Mexico–thought the gap between the rich and the poor had worsened.”

What does this all mean? Like all studies, it should be held at arm’s length. After all, a person’s experiences with globalization can vary widely. For intance, it’s not such a bad thing if you save 50% of your stereo equipment, because of increased trade with China, but it’s not such a good thing if you just lost you job to a call center in India.

Globalization is not all good and not all bad, like a lot of forces that have shaped this planet – languages, political movements, technologies. It is a double-edged sword that some countries are more skilled at swinging than others. The U.S., for example, has known how to swing that sword to its advantage for some time, but now other countries are honing their skills — China, India, Russia. It will be most interesting to see how America reacts in the years ahead, as more and more countries start swinging their figurative swords at it.

You can download a copy here. The following is an excerpt from an article that I contributed:

There is no such thing as �rest of world.�

There is good reason why movie executives in Hollywood often ask if a film with �play in Peoria� before releasing it. The United States may be one country, but it is made up of countless cultures and subcultures, based on region, ethnicity, and income. Marketing executives have learned to tailor � (in other words) localize � their promotional efforts to these various groups. Unfortunately, when these same marketing directors take their promotional campaigns to markets outside the U.S., they often do not realize that just as many subtleties exist in other countries, cultures, and regions. Similar challenges exist in Web globalization.

For example, companies often assume that if they translate their site into Spanish that the new site will reach all Latin Americans. Consider Boston Scientific, which built a �Latin American� Web site (the global gateway is shown below). The first problem presented by the Latin American site is that the gateway uses flags for navigation and there is no such thing as a Latin American flag (a good reason to avoid using flags for navigation).

Boston Scientific comes up a flag short for its Latin American site

Yet the flag is a minor detail compared with the Latin American site itself. Naturally, the site is in Spanish, but what flavor of Spanish? There is no such as thing as one Spanish, just as there is no such thing as one English. And where is the Portuguese for Brazilian Web users?

As companies expand outside their native countries, they tend to break up the world into regions: EMEA (Europe, Middle East, Africa), Asia Pacific, the Americas, or worse: ROW (rest of world). Although these groupings do offer a sense of order, they can be dangerous because they lead executives to assume that the people within these regions have more in common than just geography.

Granted, companies have a long, long way to go before they provide Web sites localized for every country, culture, and subculture. But what they can begin doing today is working toward a new way of looking at the world outside their native market � a more personal, less regional, less ROW view of the world.

For example, Toshiba once had a commercial song in China that sang, �Toshiba, Toshiba�� However, it turned out that �to-shi-ba� sounded like �let’s steal it� (tou-chu-ba) in Mandarin Chinese. People really made fun of it.

Fortunately, Toshiba is a Japanese name and its corresponding characters, Dong-Ji, means �the East� and �nobility.� Now Toshiba uses Dong-Ji more and is careful when using the pronunciation of �Toshiba.�

Some brand names travel more easily than others. Here are a few common war stories of brands that didn’t fare so well abroad:

A food company named its giant burrito a BURRADA. Big mistake. The colloquial meaning of that word is “big mistake.”

Ford had a similar problem in Brazil when the PINTO flopped. The company found out that Pinto was Brazilian slang for “tiny male genitals.” Ford pried all the nameplates off and substituted the name Corcel, which means “horse.”

A leading brand of car de-icer in Finland will never make it in America. The brand name: SUPER PISS.