A Comparative Quality Assessment Of Machine And Human Simultaneous Interpretation Of Technical Speeches

With the development of technology,machine interpretation and technical interpretation by humans have gained increasing significance.This thesis is a comparative assessment of machine and human simultaneous interpretation to analyze their differences of performance in technical speeches and texts with different frequencies of technical terms.The thesis chose four human interpreters and two machine interpretation apps to interpret two texts by the same speaker about hydrogen fuel cell from Innovation China.The interpretations were recorded,transcribed and then analyzed according to the assessment criteria based on the Theory of Quantitative and Qualitative Assessment put forward by Yang Chengshu.From the quantitative perspective,human interpreters generally outperformed machine in technical speeches.The average scores for humans and machines are 84.63 and 67.19.In terms of accuracy,language usage and coherence,human interpreters had better performance,but they missed more information than machines,especially when the frequency of technical terms is high.As for technical terms,humans were more accurate but not as stable as machines.From the qualitative perspective,humans outperformed machines in professional skills and flexibility.It is suggested that developers should focus on fixing the speech recognition problem while improving machine interpretation based on its strengths in technical interpretation.On the other hand,interpreters should also improve themselves to face the competition.Finally,it is suggested that more samples should be collected,and researchers should build more comprehensive assessment criteria and explore new fields of research.