m_topn picture
Atlantic Monthly Sidebar

Go to this issue's Table of Contents.

N O V E M B E R  2 0 0 0 

(The online version of this article appears in three parts. Click here to go to part one or part two.)

The Web in My Own Language

WHEN the conversations I have with friends and acquaintances about the future of English veer immediately toward technology -- especially the Internet -- it's understandable. Much has been made of the Internet as an instrument for circulating English around the globe. According to one estimate that has been widely repeated over the past few years, 80 percent of what's available on the Internet is in English. Some observers, however, have recently been warning that this may have been the high-water mark. It's not that English-speakers are logging off -- au contraire -- but that other people are increasingly logging on, to search out or create content in their own languages. As the newsletter that The English Company prepared for the British Council asserted in September of 1998, "Non English speakers are the fastest growing group of new Internet users." The consensus among those who study these things is that Internet traffic in languages other than English will outstrip English-language traffic within the next few years.

There's no reason this should surprise us -- particularly if we recall that there are about 372 million people in the world whose native language is English and about 5,700 million people whose native language is something else. According to the same newsletter, a recent study by Euro Marketing Associates estimated that

nearly 44% of the world's online population now speak a language other than English at home. Although many of these Internet users are bilingual and speak English in the workplace, Euro Marketing suggest that advertisers of non-business products will more easily reach this group by using their home language. Of the 56 million people who speak languages on the Internet other than English, Spanish speakers represent nearly a quarter.

The study also estimated that 13.1 percent of all Internet users speak an Asian language at home -- Japanese, for the most part. A surge in Internet use like the one that began in the United States half a dozen or so years ago is now under way in a number of other populous and relatively well-off places.

Illustration by Christoph NiemannAs has been widely noted, the Internet, besides being a convenient vehicle for reaching mass audiences such as, say, the citizenry of Japan or Argentina, is also well suited to bringing together the members of small groups -- for example, middle-class French-speaking sub-Saharan Africans. Or a group might be those who speak a less common language: the numbers of Dutch-speakers and Finnish-speakers on the Internet are sharply up.

The Internet is capable of helping immigrants everywhere to remain proficient in their first language and also to stay current with what is going on back home. Residents in the Basque communities of Nevada and émigrés from the Côte d'Ivoire, for instance, can browse the periodicals, and even listen to the radio stations, of their homelands -- much as American expatriates anywhere with an Internet connection can check the Web sites for CNN, ABC, MSNBC, and their hometown papers and radio stations.

No matter how much English-language material there is on the Web, then, or even how much more English material there is than material in other languages, it is naive to assume that home computers around the world will, in effect, become the work stations of a vast English language lab. People could use their computers that way -- just as we English-speaking Americans could enlist our computers to help us learn Italian, Korean, or Yoruba. But, the glories of learning for its own sake aside, why would we want to do that? Aren't we delighted to be able to gather information, shop, do business, and be entertained in our own language? Why wouldn't others feel the same way? Consider, too, that many people regard high technology as something very much like a new language. Surely it's enough for a person to try to keep his or her hardware and software more or less up-to-date and running smoothly without simultaneously having to grapple with instructions or content in an actual foreign language.

Studies of global satellite television -- a realm that is several years more mature than the Internet -- also point to the idea that most people like new technology better when it speaks their own language. As Richard Parker wrote in Mixed Signals,

Satellites can deliver programming and advertising instantaneously and simultaneously across the more than two dozen languages spoken in Western Europe, but the viewers -- as repeated market research shows -- want their television delivered in local tongues. Contrary to a history in which both motion pictures and early television broadcasts relied heavily on dubbing of foreign (often U.S.) programming, an affluent and culturally confident Europe now appears to be more linguistically divided than ever before.

Parker distinguishes between the "technologically feasible supply" of foreign programming and the "economically viable demand" for it, warning that we should be careful not to confuse the two. A few years ago, for example, Sweden aired a "reality-based" TV series, Expedition: Robinson (the word expedition has entered Swedish from English), and it quickly became a national obsession. But its success did not inspire American television networks to import the series; rather, they developed new shows, such as Big Brother and Survivor.

English by Accident

AT one point in my conversation with David Graddol, he made a little sketch of something for me on a proof of his article "The Decline of the Native Speaker." The sketch was meant to remind me that technology has begun to blur the distinctions between languages in intriguing ways -- and to suggest how those ways are themselves starting to overlap. Both the Internet and a range of technological applications only distantly related to it, he wanted me to see, are poised to expand what we are able to do with English.

Graddol uncapped his pen and drew a box in the broad white top margin of the page. "Text to text MT," he wrote in the box, and he said, "Of course you know about machine-translation systems," tapping the box to indicate that it was to represent them. Yes, I did: in fact, The Atlantic published an article about machine translation not long ago (see "Lost in Translation," by Stephen Budiansky, in the December, 1998, issue).

From Atlantic Unbound:

Web Citation: "Speaking in Tongues" (December 1997)
AltaVista is serious about global communication.

As the article explained, there are translation programs -- AltaVista's Babel Fish among them -- available for use free on the Internet. Type some English into the appropriate space on the Babel Fish Web page, or cut and paste it from another source, and choose your "destination" language -- French, Portuguese, Spanish, Italian, or German. Presto! Up will pop a not entirely accurate translation of your chosen text. Or you can do this in reverse, from one of those languages -- or Russian (the English-to-Russian feature is still in the works) -- into English. Some professional translators use machine-translation systems as time-savers, getting the things to hack out rough texts they can then refine.

To the left of his machine-translation box Graddol drew a second box, which he labeled "Speech to text." He tapped it and said, "And you know about the voice-recognition systems that turn spoken words into written words." Yes, those, too. As it happens, I am the proud owner of a Dragon Systems program. Current versions of that and several other voice-recognition programs are reported to render speech into writing with 98 percent accuracy -- not a rate that detail-oriented people are likely to find reassuring (getting two words wrong per hundred can add up), but certainly a rate that allows a user to get a point across.

Speech-to-text systems are now available for a variety of languages. Lernout & Hauspie, an industry leader that recently bought both Dragon Systems and Kurzweil Education Systems, sells products for turning British speech, as well as American, into writing, and also ones for German, Dutch, Spanish, French, Mandarin, and Cantonese.

Graddol drew a third box in the margin to the right, and labeled this box "Text to speech." He said, "And there are also machines that turn written words into spoken words." The Kurzweil reading machine, created to help the blind and visually impaired, and now capable of reading aloud in more than fifty languages, is the most advanced example in use. Simpler machines that turn computer code rather than text into speech are of course commonplace by now. We sometimes hear them when we call 411 and ask for a phone number; we hear them when we're refilling a prescription over the phone and a synthesized voice confirms our prescription number and name; we hear them on airlines' flight-information phone lines. These machines may have a vocabulary as elementary as numbers, the days of the week, and "A.M." and "P.M." But they get the job done, and they hint at how more-complex systems might work.

Now Graddol drew lines from one box to the next. "People are starting to work on connecting all the parts," he said. "Once that happens, a lot of things will be possible."

I could, for example, speak into the microphone that came with my Dragon Systems program and have that program render what I've said in writing; instruct one of the translation programs to turn the text into French; and then use Lernout & Hauspie's French-language speech synthesizer to pronounce the computer's translation. This may strike some as a ponderous process, but surely it would be less complicated than acquiring a creditable French accent the old-fashioned way. Then, too, speech-to-writing and writing-to-speech programs may materialize on the Internet, much as the translation programs have done. In that case I will simply talk into the microphone, miraculous high-tech things will happen somewhere in the ether, and voilà! the computer at the restaurant L'Ami Louis, in Paris, will make my request for a reservation known to the staff, in exquisitely correct spoken or written French, and the maitre d', unwitting, will assign me a good table.

That's the theory, anyway. I have my doubts about how exquisite the actual results will be for quite some time. The interchanging of speech with writing, writing with speech, and English with other languages may, however, yield serviceable results very soon. According to a compilation of funny signs spotted around the world, published by the Far Eastern Economic Review, a Paris dress shop once advertised "Dresses for street walking," and a notice in a hotel elevator in the same city advised, "Please leave your values at the front desk." If we can understand the intention of these signs -- as of course we can -- then surely we will be able to see beyond most of the peculiarities resulting from machines' involvement in language. David Graddol's neat little boxes glossed over myriad difficulties inherent in each step of linguistic interchangeability. But each of these steps is already being accomplished approximately, and implemented not just in experimental settings but in real life.

Even as software developers continue to adapt computers to our linguistic needs and wants, we are -- God help us -- adapting our own language to computers. For example, if I want to see the Amazon.com page about the psycholinguist Steven Pinker's book Words and Rules (1999), it's a complete waste of time to type into the search feature "Words and Rules, by Steven Pinker," correctly capitalized and punctuated. The computer and I will get exactly as much out of the exchange if I type "pinker rules." In effect, in this context "pinker rules" is better English than "Words and Rules, by Steven Pinker."

Where computers' processing ability and our intelligence will eventually converge is anyone's guess. As we teach ourselves, for instance, to speak in a way that will make our voice-recognition systems as productive as possible, developers are tweaking the new versions of them so that if the system misinterprets a word and we need to revise what it writes, the change will be incorporated into its database and it will never make the same mistake again.

Does this matter to the future of English? It may well. What is English, anyway? Is it the list of words and their meanings that a dictionary provides, together with all the rules about how to combine the words into sentences and paragraphs? Much more is involved than that. English is a system of communication, and highly germane to it is what or who speakers of English care to communicate with, and about what. The more we need to use English to communicate with machines -- or with people whose fluency is limited or whose understanding of English does not coincide with ours -- the more simplified the language will need to be.

And yet technology is expanding English, by requiring us to come up with new words to describe all the possibilities it offers. Throughout the past century, according to Twentieth Century Words (1999), by John Ayto, technological domains -- at first the likes of cars and aviation and radio, and eventually nuclear power, space, computers, and the Internet -- were among the leading "lexical growth-areas." What's new of late isn't only words: we have whole new ways of combining the elements of written language. One ready example is emoticons (such as :> and ;-o), which seem to have firmly established themselves in the realm of e-mail. Is www a word? Does one write the expression dot com or .com or what? And then there's professional jargon. In the course of exchanging ideas, global communities of astrophysicists, cardiologists, chip designers, food scientists, and systems analysts are stuffing the English language full of jargon. As science and technology grow increasingly multifarious and specialized, the jargon necessarily grows increasingly recondite: in the journal Neurology, for example, article titles like "Homogeneous phenotype of the gypsy limb-girdle MD with the g sarcoglycan C283Y mutation" are run-of-the-mill. The range of English continues to expand further and further beyond any single person's ability to understand it all.

One more fact worth keeping in mind is that the relationship between science or technology and English is, essentially, accidental. It is chiefly because the United States has long been in the vanguard of much scientific and technological research, of course, that English is so widely used in these fields. If the United States were for the most part French-speaking, surely French would be the language of science and technology; there is nothing inherent in English to tie it to these fields. And if something as earthshaking as the Internet had been developed in, say, Japan, perhaps English would not now be dominant to the extent that it is. Future technology may well originate elsewhere. In the rapidly advancing field of wireless communications devices, for example, Scandinavia is already the acknowledged leader.

Here an argument is sometimes advanced that American culture furthers innovation, openness to new ideas, and so forth, and that our culture, whether by accident or not, is inseparable from the English language. But this takes us only so far. Even if the vanguards in all scientific and technological fields, everywhere in the world, used English in their work, once the fruits of their labor became known to ordinary people and began to matter to them, people would coin words in their local languages to describe these things. Theoretical physicists at international conferences may speak English among themselves, but most high school and college physics teachers use their native languages in class with their students. The Microsoft engineers who designed the Windows computer-operating system spoke English, and used English in what they created, but in the latest version, Windows Millennium, the words that users see on the screen are available in twenty-eight languages -- and the spell-checker offers a choice of four varieties of English.

IN sum, the globalization of English does not mean that if we who speak only English just sit back and wait, we'll soon be able to exchange ideas with anyone who has anything to say. We can't count on having much more around the world than a very basic ability to communicate. Outside certain professional fields, if English-speaking Americans hope to exchange ideas with people in a nuanced way, we may be well advised to do as people elsewhere are doing: become bilingual. This is easier said than done. If learning a second language were so simple, no doubt many more of us would have picked up Spanish or Chinese by now. It is clear, though, that the young learn languages much more readily than adults. Surely, American children who are exposed to nothing but English would benefit from being taught other languages as well.

At the same time, English is flourishing, and people here and everywhere are eager to learn it to the extent that it is practical for them to do so. It would behoove us to make learning English as easy as possible, for both children and adults, in this country and abroad.

However unwelcome this news may be to some, not even headlong technological advances mean that computers will soon be doing all the hard work of coping with other languages for us. For the foreseeable future computers will be able to do no more than some of the relatively easy work. When it comes to subtle comprehension of our world and the other people in it, we are, as ever, on our own.

(The online version of this article appears in three parts. Click here to go to part one or part two.)


Barbara Wallraff is a senior editor of The Atlantic and the author of the book Word Court (2000), which grew out of her bimonthly Atlantic column of the same name.

Illustrations by Christoph Niemann.

Copyright © 2000 by The Atlantic Monthly Company. All rights reserved.
The Atlantic Monthly; November 2000; What Global Language? - 00.11; Volume 286, No. 5; page 52-66.