Translating tweets

André Guyon
(Language Update, Volume 10, Number 1, 2013, page 28)

New media are promising markets, but with them also come major challenges. In 2013, the future of technology is in tablets and phones. Whether we like it or not, Twitter is one of those trendy new media people, including celebrities and executives of high-profile organizations, use to communicate. This is creating new demand for translation, and we need to look at how best to translate tweets.

Below are nine characteristics of this new phenomenon:

1. The messages, called tweets, are quite frequent.
2. Their translation cannot wait very long.
3. Tweets are about major events.
4. They have a maximum length of 140 characters.
5. They are often written in textese (similar to chatting and texting).
6. The same person may tweet in more than one language.
7. Communications can be unilingual, bilingual or multilingual.
8. Translation tools are generally not well suited to meet this emerging demand.
9. These texts can be in an infinite number of fields.

The frequency of tweets heralds a certain volume of work, in a variety of fields that are more or less predictable. This kind of translation might therefore be better suited to general translators than specialized translators. However, receiving a series of tweets on a specialized topic would be a good opportunity for general translators to adopt new work habits. I also think that young language professionals would be more attracted to translating tweets. They are more familiar with today’s style of sharing information, which is varied and features short messages.

It goes without saying that we cannot use the regular system of translation requests and purchase orders to invoice and process messages posted on Twitter. That being said, many organizations already provide their clients with packages so that they can send short, frequent content for translation without having to fill out a purchase order every time.

It is a good idea to be available after standard business hours because messages can arrive at any time, day or night.

Finally, the fact that these messages are spontaneous and urgent, like weather warnings, means that they have to be processed quickly.

As I mentioned earlier, some tweets are about major events. It is therefore crucial to find the relevant background documentation quickly. This is what interpreters do, but more or less without any specific tools to assist them. Smart-search software, which would have the same objective as an old National Research Council of Canada project I’ve already mentioned, TerminoWeb,* would quickly assemble a collection of relevant documents.

Specifications

Some components can be greatly improved to better meet the needs related to translating tweets.

The 140-character limit is obviously a challenge for translators. Sometimes, despite their best intentions, translators won’t be able to keep things short and use only 140 characters (because of long official titles, for example).

• A program should indicate, in real time, the number of characters in the tweet and automatically continue the message in another tweet.
  
  
  • The software would notify the translator that the 140-character limit has been exceeded and suggest that the message be posted in two consecutive tweets. At the end of the first tweet, it would automatically insert a symbol meaning “continued in next tweet.”
  • The software would try to find an appropriate place to divide the message. Ideally, it would be after punctuation; if there is no punctuation, it would be after a word, taking into account the “continued” symbol.
  • In the next tweet, the software would insert a symbol meaning “continued from previous tweet.”

Although clients may ask for French-to-English translation, there is no guarantee that they will always write in French (especially in Canada). Also, many Canadian anglophone personalities sometimes like to tweet in French.

• Consequently, given that the speaker’s language may vary, a component to detect language would help assign the message to be translated to the appropriate translator.
• Moreover, it may sometimes be necessary to use a bridging language. For example, one translator may understand Algonquin and French, but not English, and another might understand Spanish and English, but not French.
• Regardless of whether a bridging language is used, a smart queue is needed to store the original messages until they are translated. Why? Simply to avoid having a short “yes” in answer to a long question translated before the question itself.
• When a bridging language is used, it’s best to wait until all the translations are finished, for example, from Spanish to English, then from English to French.
• Unlike for chatting, the translation of tweets can wait a little, but not necessarily very long, especially when two people are exchanging messages.
• Consequently, the smart queue must be able to direct the message to be translated to another translator working in the same language combination if the first translator seems to be away from his or her workstation.
• In the same vein, an instant message function would enable translators to contact colleagues working on a specific thread of conversation for help.

Obviously, all this should be part of an environment to assist translators that would include a memory, a bitext, a machine translation engine, a terminology bank and correction software.

Finally, it is worth decoding textese because every minute counts. In fact, all abbreviations, initialisms and acronyms, which are quite literally the bane of language professionals’ existence, should be decoded. Without even thinking about it, we spend part of our lives learning the initialisms and acronyms that only those working in the same field we do can understand.

As you can see, technology can significantly alleviate our workload when it comes to translating messages on Twitter and other new media, which all require a different approach. Such tools will help language professionals keep clients satisfied and expand their areas of expertise.

Remark

• Back to remark 1* See "Realistic Dreams of a Language Professional," Language Update, Vol. 7, No. 4 (December 2010), p. 17. The article also appears in Favourite Articles, at http://www.btb.termiumplus.gc.ca/tpv2guides/guides/favart/index-eng.html?lang=eng.

© Services publics et Approvisionnement Canada, 2024
TERMIUM Plus^®, la banque de données terminologiques et linguistiques du gouvernement du Canada
Outils d'aide à la rédaction – Favourite Articles
Un produit du Bureau de la traduction