This paper examines the extent to which the writing style of texts published on highly usable web-sites provides input that is amenable to Internet-based machine translation (MT) services, thus giving rise to a form of MT-friendly controlled language, in a loosely defined controlled translation environment for web content. Controlled languages are traditionally divided into two groups, human-oriented and machine-oriented, depending on the purpose and applications for which they have been developed. Even though this distinction is justified and useful in a number of respects, it is recognised that these two types of controlled languages often share significant features.The paper investigates the common ground between human-oriented and machine-oriented controlled languages in the context of reader-friendly monolingual on-line material that needs to be translated into a variety of target languages by means of web-based MT services. The paper argues that a usability-oriented writing style for originally monolingual web content not only improves the readability of on-line material for human Internet users, but can also dramatically improve the performance of web-based MT systems. The discussion suggests in conclusion that authoring conventions and guidelines primarily aimed at improving readability for human users in the Internet environment also have a significant impact on the successful application of on-line MT technology, if monolingual web content needs to be disseminated in a variety of languages.
Controlled Language, Web Usability and Machine Translation Services on the Internet
Federico Gaspari
2004-01-01
Abstract
This paper examines the extent to which the writing style of texts published on highly usable web-sites provides input that is amenable to Internet-based machine translation (MT) services, thus giving rise to a form of MT-friendly controlled language, in a loosely defined controlled translation environment for web content. Controlled languages are traditionally divided into two groups, human-oriented and machine-oriented, depending on the purpose and applications for which they have been developed. Even though this distinction is justified and useful in a number of respects, it is recognised that these two types of controlled languages often share significant features.The paper investigates the common ground between human-oriented and machine-oriented controlled languages in the context of reader-friendly monolingual on-line material that needs to be translated into a variety of target languages by means of web-based MT services. The paper argues that a usability-oriented writing style for originally monolingual web content not only improves the readability of on-line material for human Internet users, but can also dramatically improve the performance of web-based MT systems. The discussion suggests in conclusion that authoring conventions and guidelines primarily aimed at improving readability for human users in the Internet environment also have a significant impact on the successful application of on-line MT technology, if monolingual web content needs to be disseminated in a variety of languages.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
