Este artigo da Economist sequer menciona a última flor do Lácio... (e o galego?).
Paulo Roberto de Almeida
Languages on the internet
The keenest Wikipedians
WILLIAM GIBSON once said "The future is already here—it's just not
very evenly distributed." I'd include Wikipedia, the wonderful,
sprawling, open-source and free online encyclopaedia, as part of the
future. It also seems to be quite unevenly distributed.
Take a look at the crazy distribution of articles and users among the world's languages. In just five languages does Wikipedia have more than 1m articles: English, German, French, Italian and Dutch. One of these is obviously an outlier: Dutch is spoken by only around 20m people. According to Eurostat, fully 100% of Dutch students are studying English, and anyone can tell you that the Dutch are keen and fluent English-speakers, while Dutch-speaking Flemings usually speak good French. This is strong evidence, with a nod to our earlier posting on the "underwear language", that people are strongly attached to their own languages even when they speak other languages well.
The next order of magnitude carries more surprises. In the box of Wikipedias with more than 100,000 articles fall obvious world languages like Russian, Arabic and Chinese. But this box also includes several languages that do not have their own state, like Spain's Galician, Basque and Catalan—more evidence that people in these regions, even if they can read Spanish, often prefer their own anyway. But even weirder, among stateless languages, are two that have no ethnic nation associated with them at all. It's not entirely surprising that Esperanto, an invented language, has 176,792 articles. Anyone familiar with the Esperanto community (such as Arika Okrent) can tell you how active it is. But Volapük? This much less well-known invented language can claim 119,091 articles in Wikipedia. for what one Village Voice writer claimed were 20 living speakers. Volapük has more articles than Hindi, with its 180m-odd speakers. I can only guess that someone has auto-translated most of those 119,091 articles into Volapük. As for Hindi, this could be a partial exception to the rule that people strongly prefer their own language. Despite a large written tradition in Hindi, it seems likely that many Hindi-speakers read Wikipedia in English.
Languages in the 10,000+ category are similarly ordered to surprise. We now see many dialects and languages with few monoglots, and probably exactly zero people who are literate only in the dialect: how many speakers of Alemannic (13,708 articles) are unable to read standard German? How many Piedmontese readers (who can browse 59,303 articles) cannot read Italian (1,012,838)? Allemanic and Piedmontese rank with languages with tens of millions of speakers, like Javanese (82m speakers, 43,122 articles). As we continue down the list, we see many deeply underserved languages: Xhosa, Nelson Mandela's native language, has 8m speakers but just 146 articles on Wikipedia. At the very bottom of the page, we find hopeful but neglected Herero, for which someone has created a homepage but no articles at all.
Of course the number of articles isn't the only measure we should look at. Volapük is listed as having only 46 active users and zero images. And Wikipedia calculates a "depth" for each language, a measure of how often articles are edited and a discount for how many short "stub" articles there are (and so a rough proxy for quality). Scanning the list, we see that Hebrew, Arabic, Persian and Turkish are much "deeper" than the bigger German or Italian Wikipedias. I'd be fascinated to know exactly why that is, but it must have something to do with how many topics raise controversy in that neighbourhood, even about basic questions of fact.
The "depth" score is calculated by dividing by the total number of articles, so a language with many articles and few edits will have a lower depth than a language with fewer articles with the same number of edits. The deepest Wikipedia? English, with 4.2m articles, is being edited constantly, and emerges as the clear winner. Another manifestation of its unstoppable global march.
Take a look at the crazy distribution of articles and users among the world's languages. In just five languages does Wikipedia have more than 1m articles: English, German, French, Italian and Dutch. One of these is obviously an outlier: Dutch is spoken by only around 20m people. According to Eurostat, fully 100% of Dutch students are studying English, and anyone can tell you that the Dutch are keen and fluent English-speakers, while Dutch-speaking Flemings usually speak good French. This is strong evidence, with a nod to our earlier posting on the "underwear language", that people are strongly attached to their own languages even when they speak other languages well.
The next order of magnitude carries more surprises. In the box of Wikipedias with more than 100,000 articles fall obvious world languages like Russian, Arabic and Chinese. But this box also includes several languages that do not have their own state, like Spain's Galician, Basque and Catalan—more evidence that people in these regions, even if they can read Spanish, often prefer their own anyway. But even weirder, among stateless languages, are two that have no ethnic nation associated with them at all. It's not entirely surprising that Esperanto, an invented language, has 176,792 articles. Anyone familiar with the Esperanto community (such as Arika Okrent) can tell you how active it is. But Volapük? This much less well-known invented language can claim 119,091 articles in Wikipedia. for what one Village Voice writer claimed were 20 living speakers. Volapük has more articles than Hindi, with its 180m-odd speakers. I can only guess that someone has auto-translated most of those 119,091 articles into Volapük. As for Hindi, this could be a partial exception to the rule that people strongly prefer their own language. Despite a large written tradition in Hindi, it seems likely that many Hindi-speakers read Wikipedia in English.
Languages in the 10,000+ category are similarly ordered to surprise. We now see many dialects and languages with few monoglots, and probably exactly zero people who are literate only in the dialect: how many speakers of Alemannic (13,708 articles) are unable to read standard German? How many Piedmontese readers (who can browse 59,303 articles) cannot read Italian (1,012,838)? Allemanic and Piedmontese rank with languages with tens of millions of speakers, like Javanese (82m speakers, 43,122 articles). As we continue down the list, we see many deeply underserved languages: Xhosa, Nelson Mandela's native language, has 8m speakers but just 146 articles on Wikipedia. At the very bottom of the page, we find hopeful but neglected Herero, for which someone has created a homepage but no articles at all.
Of course the number of articles isn't the only measure we should look at. Volapük is listed as having only 46 active users and zero images. And Wikipedia calculates a "depth" for each language, a measure of how often articles are edited and a discount for how many short "stub" articles there are (and so a rough proxy for quality). Scanning the list, we see that Hebrew, Arabic, Persian and Turkish are much "deeper" than the bigger German or Italian Wikipedias. I'd be fascinated to know exactly why that is, but it must have something to do with how many topics raise controversy in that neighbourhood, even about basic questions of fact.
The "depth" score is calculated by dividing by the total number of articles, so a language with many articles and few edits will have a lower depth than a language with fewer articles with the same number of edits. The deepest Wikipedia? English, with 4.2m articles, is being edited constantly, and emerges as the clear winner. Another manifestation of its unstoppable global march.
