In Depth
The power of Babel
October-November 2009

Audio version of this article.

---

by Norm Goldstein

Blame it on Noah.

Or at least on his family. According to the biblical story, Noah’s descendants—all of whom spoke the same language—settled in Babylon (Babel in the Hebrew) and decided to build a tower that would reach to heaven, an effort to “make us a name.” For this presumption, their language was “confounded,” and they lost the ability to speak intelligibly to one another.

In many ways, the problem persists today, despite linguistic bridges enabling better communication with those speaking different languages. Editors working in English know all too well that there is the confounded problem of confounding alphabets and other systems of writing, many of which do not convert easily to our Latin alphabet.

For centuries, the English language has “borrowed” words—assimilated is probably a better description; most of them are not returned—from a myriad of other languages. Latin, Greek, Hebrew, Arabic, Russian, Sanskrit, French, Spanish, Portuguese, various dialects of indigenous conquered peoples—the list goes on and on, a veritable babble of tongues.

There’s little difficulty in spelling words from languages that share a similar alphabet with English, give or take an accent mark or other diacritic (which are often just left out). Currently, however, English is absorbing more words from languages that do not use the Latin alphabet—Arabic, Chinese, Hebrew, Japanese, and Russian being the most prominent. The process of trying to convert one alphabetic script to another is known as transliteration. In the case of such non-alphabetic writing systems as Chinese and Japanese, the process is called transcription—converting those symbols for units of sound or meaning to an alphabetic script. In the case of English, the script is Latin, and the process is therefore called romanization.

For those editors who need to dive into this sea of scripts on a daily basis, the degree of difficulty varies with the specific language—and the frequency of use.

Familiarity breeds uniformity

Frequency of use and consensus about spelling seem to correlate. We see this happening with words for now-familiar foods, for example. In Turkish, the grilled dish of meat and vegetables served on a skewer is spelled ¸i¸ kebabi, with cedillas under the s’s to mark the pronunciation. The most common of the accepted spellings in English is shish kebab, which is identical to the Armenian form of the term. (Turkey, incidentally, legislated the use of a modified version of the Latin alphabet in 1928, replacing the Arabic-based Ottoman script.)

Even when a word comes directly from Arabic, instead of through a rinse in a romanized alphabet, the spelling can settle down over time. An example is the Arabic term for the Palestinian uprising against Israel, which was spelled intefadah when it began to make news in the 1980s. The Associated Press recently changed its recommended spelling to intifada, bowing to what it considered the more common transliteration.

We don’t have the luxury of familiarity over time, though, with the names, places, and words from Arabic that are rapidly flowing into English because of current events. There is still much debate about the transliterated spellings from an alphabet little known to English speakers. Is it With a hyphen or not? Koran or Quran? Or Qur’an?

In Arabic, Qaida or Qaeda, which means ‘the base’, is a three-syllable word: KAH-i-da. Pro-nunciation is governed by a short second vowel that sounds somewhere between the i in bit and the e in bed. (The Associated Press pronunciation guide in the online version of the Stylebook gives a phonetic spelling of al-Ky’-ee-duh, although its audio pronunciation gives al-Ky’-duh. ) Arabic speakers sometimes elide the word into two syllables, much like KAH-da.

The word, hyphenated, started appearing in the news after terrorist bombings of U.S. embassies in Africa in late 1998, and was at first spelled al-Qaida and al-Qaeda. The AP settled on al-Qaida; The New York Times, on the other hand, chose al Qaeda. The Chicago Manual of Style, in an example of the use of the Arabic definite article, lists “al Qaeda (or al-Qaida) .” The 2009 edition of The American Heritage Dictionary lists the same spellings as CMS; the 2005 edition of Webster’s New World College Dictionary gives the main spelling as Al Qaeda and offers four alternatives: al-Qaeda, al-Qae’da, al-Qaida, and al-Qai’da. Dictionary.com cites all these forms, plus Al-Qaida, al-Qa’ida, al-Quaida, and al-Qa’idah.

While the numerous variants of al-Qaeda listed by Dictionary.com are more confusing than helpful, the variants listed by Webster’s New World point to the fact that most systems of transliteration require diacritics, the marks that appear above, below, or between the letters, which can indicate pronunciation or stand for vowels. A common one in Arabic (and other languages) is the hamza. It indicates a glottal stop in pronunciation and is usually represented in English transliterations by an apostrophe, as in Qur’an. Generally, though, words containing the hamza are transliterated without it, presumably because speakers of English think the apostrophe represents a missing letter rather than a break in pronunciation.

Different languages, different approaches

There is no universally accepted form for transliterating Arabic script, which is used for languages such as Urdu, Uyghur, and Farsi as well as for Arabic and is not uniform in pronunciation. Thus, a babble (Babel?) of spellings—and headaches for editors, whether they’re dealing with breaking news or with longer-lead periodicals and books, and whether the medium is print or electronic. The situation is more or less similar with other languages that don’t use the Latin alphabet:

Chinese Several systems of romanizing Chinese have been developed, the most familiar of which to Westerners was the Wade-Giles system. In 1979, the Chinese officially replaced Wade-Giles with a system called Pinyin, which represents the sounds of Chinese more explicitly. Thus, Peking became Beijing, Chou En-lai became Zhou Enlai, and Mao Tse-tung was spelled Mao Zedong, although most publications stayed with the older spellings for historic figures. Variations on Pinyin exist too, most notably in Taiwan. A comparison chart of six romanization systems can be found at http://pinyin.info/romanization/ compare/hanyu.html.

Hebrew There are also several romanization systems for Hebrew, resulting in variant spellings such as Hanukkah, Hanuka, and Channukah for the same holiday. The differences reflect the varying pronunciations used by Ashkenazi and Sephardi speakers of Hebrew, and also the difficulty English speakers have with the Hebrew sound /ch/. Leo Rosten, in his book The Joys of Yiddish, says it should be pronounced not as in shoe or Chippewa but as the German ch in Ach! or the Scottish ch in loch. (It’s accomplished, he says, by pretending you have a breadcrumb stuck in your throat.)

Languages that use Cyrillic The Cyrillic alphabet is used for Russian, of course, but also for other Slavic languages, including Macedonian, Bulgarian, and Serbian, as well as non-Slavic ones such as Mongolian, Ossetian, and Uzbek. Complicating things even further, not all letters in the Cyrillic alphabet are used in every language that is written with it, and some languages have a history of being written first in one script and then another. There are several systems for transliterating Cyrillic, but no single standard.

The Associated Press Stylebook has some helpful hints for spelling Russian names, with the broad suggestion to use the closest phonetic equivalent in English. Thus, Alexander, not Alexandre. Russian names do not end in -off, so transliterations should end in -ov, as in Romanov. (There’s still a difference of opinion about Dostoevski versus Dostoyevsky.)

Languages of India and Southeast Asia The Bengali alphabet is used to write at least four other major languages, Assamese, Manipuri, Garo, and Mundari. Malayalam, a language spoken in southwest India, parts of Malaysia, Bahrain, Fiji, Qatar, Singapore, and even Israel, is sometimes written in its own syllabic script and sometimes in Arabic. For many languages written in scripts that are difficult to convert to computer use, linguists use a phonetic transliteration script that romanizes the letters, but there are variations between some of the ISO (International Organization for Standardization) standards and other standards that eliminate, or at least reduce, diacritics.

Korean The old McCune-Reischauer system of romanization of the Korean Hangul characters is being replaced with a new system that makes a non-native speaker produce sounds that are closer to the Korean pronunciation and that eliminates diacritics so the system is easier for QWERTY keyboard users to implement. Korea.net, a Web site managed by the Korean Culture and Information Service, provides a helpful introduction to the changes at www.korea.net/Korea/Kor_loca.asp?code=L0602. An example of the differences: in the new romanization system, the word Taej¡n is now spelled Daejeon.

The system of romanization of Japanese is called romaji, a term now used to refer to a few different systems. The most common is a modified version of a system also known as Hepburn romanization (after an American missionary who developed it). Two other main romaji variants have been standardized by the ISO (ISO 3602:1989). These three are also subject to variation, however.

What’s an editor to do?

The American Library Association and the Library of Congress jointly developed a set of standards for the romanization of 56 languages, published in 1997 as the ALA-LC Romanization Tables. A version of the publication that includes the most recently revised tables is available from the Library of Congress Web site at www.loc.gov/catdir/cpso/roman.html.

But the tables are not for the faint of heart, and certainly not for the busy editor searching for a bottom-line spelling preference. The Chicago Manual of Style, which cites the ALA-LC Romanization Tables as its prime resource for alphabetical conversion, takes on the subject with do-not-try-this-at-home warnings:

“Do not attempt to transliterate from a language unfamiliar to you.”

“Grave errors can occur when similar characters are mistaken for each other.”

These warnings come from the introductory paragraphs in the section titled “Languages Usually Transliterated (or Romanized)” in chapter 10 of CMS. But the Manual does have guidelines for specific languages, namely, Arabic, Chinese and Japanese (considered together in the same section), Hebrew, and Russian (Cyrillic), that are more accessible than the ALA-LRC document.

Dictionaries, both printed and online, are at least helpful in providing the most common alternatives, though the first listed word is not necessarily the preferred one. Some dictionaries distinguish between less common variants (introduced by also) and roughly equivalent ones (introduced by or). But other dictionaries don’t make this distinction; the fourth edition of Webster’s New World College Dictionary, for example, uses only or It explains its variant- spelling policy for joint main entries by noting that it “implies that both spellings occur equally, or almost equally, often and that neither one is ‘more correct’ or to be preferred.”

The choice, then, becomes a matter of taste or house style, not “correctness.”

Whose choice? Yours, for a start, or that of the organization you work for. You can surely check out a variety of dictionaries and make an educated pick from among the variants, perhaps choosing the one that seems to be most common.

And, while lexicographers shy from stating their preferences, at least in print, the editors of stylebooks take on the task. Those who work for publications that have the luxury of international correspondents depend greatly on their input. The Associated Press, as a news service with correspondents around the globe, must keep up with the fast-entering names and words from foreign sources. So, The Associated Press Stylebook, especially the online version, is a good source for editors. According to AP International Desk editors, entries for such terms are based primarily on the advice of their editors in foreign bureaus, who in turn rely on their experience, local writers, and a consensus of scholars, local publications, and dictionaries in the language of origin.

The basic approach, then, would be to check a stylebook first, and if a given term is not there, to ask your in-house experts. If you have a foreign desk, that’s the place to go. If editors there have no definitive answer, they’ll go to the correspondents in the field. When all else fails, it’s up to your own publication to make the calls. And once you’ve decided which spellings to follow, be consistent in their use. n

Norm Goldstein is the editor emeritus of The Associated Press Stylebook, and a regular contributor to this newsletter.