Automatic translation in multilingual business meetings

Abstract

Purpose Multilingual meetings continue to be a problem in business communication due to the necessity to translate between different natural language pairs. The paper has developed a new electronic meeting technology that automatically translates comments written in any of 41 languages (e.g. French, Chinese, etc.). The purpose of this paper is to quickly and accurately show on each user's terminal in his or her own native language all comments contributed by the group written in several different tongues. Design/methodology/approach A prototype system, Polyglot II, utilizes instant messaging on Microsoft Windows PCs to exchange comments between client personal computers and a server which in turn, calls the Google Translate API for each translation. In an attempt to measure the accuracy, reverse translations are conducted, e.g. English to French to English, because of the lack of human experts fluent in all of the languages. The final English translations are analyzed for comprehension by 240 college business students. Findings This paper uses reverse translations on 32 of the languages (all that are available at the time of the analysis) with historical transcripts of English text, including grammatical errors. Results show an overall comprehension accuracy of 86 percent for all languages. Italian is the most accurate, and Hindi is the least. Originality/value While other multilingual meeting technologies have been developed, this system provides automatic support for the most languages and is perhaps the most accurate.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 20 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.