An Empirical Study Of Computer-Aided Consecutive Interpretation

Machines have been applied in translation for 85 years,mainly in the form of machine translation(MT)and computer-aided translation(CAT).As new technologies such as voice recognition are introduced,the assistance of machines is both strengthening and diversifying.After IT starts to be applied in interpretation,computer-aided interpretation(CAI)comes into being.CAI refers to interpretation aided by information technologies,including corpus,the Internet and software.CAI can assist interpreters in preparation,recording and searching,and information organization respectively before,during and after interpretation.In the process of interpreting,the computer can rapidly obtain the source text through speech recognition,and perform translation with the aid of corpus,terminology database and cloud resources.The selected target language displayed on the screen can provide reference for the interpreter.For details such as numbers and names that are difficult to remember for human brain,the computer can display them after recognition to reduce the memory burden on interpreters.In addition,powerful terminology database provides accurate translation of proper nouns.Sharing memory tasks with computers enables translators to increase the energy allocated for understanding and expression,thereby improving the quality of interpretation.This paper starts with the application of machines in translation,the history and status quo of CAI,technologies related to CAI and previous research.To determine how can machines provide the best assistance for interpreters,the paper analyzes through empirical research the performance differences between interpreters with and without computer assistance.To conduct this comparison,the paper involves designing a CAI platform that combines named entity recognition,mapping knowledge domain,speech recognition,e-notebook,searching and corpus technologies.The research experiments with 3 news audios,dividing tested interpreters into the CAI group and the non-CAI group.The interpretation results of both groups are then compared in terms of fidelity,delivery and target language qualities.Both quantitative and qualitative analyses are carried out to determine the assistance provided by the CAI platform.The research discovers that for materials relatively fast in speech and intensive in numbers,proper nouns and terminology,CAI can effectively raise the quality of interpretation,especially accuracy,by easing the memory burden of interpreters.For materials otherwise,the influence of CAI is limited,while the quality of interpretation mainly hinders upon the understanding,logic organization and knowledge storage of interpreters.In platform design and the experiment process,the CAI platform used in this paper has many limitations.According to feedback and reflections,the platform can be improved in terms of filtering information,extended functions,corpus and user-defined features.The development of CAI will alter the effort coordination,interpretation process or even the definition of interpretation.In general,CAI can provide solutions to many interpretation problems while changing interpretation itself.