An Empirical Comparative Study Of Machine And Human Interpretation From The Perspective Of Error Analysis Theory

With the development of global economy and the frequent communication between countries,interpretation industry has become an increasingly important field.The development and application of artificial intelligence technology have led to the rapid development of machine interpretation.The efficiency of interpretation services has been greatly improved.Machine interpretation has gradually emerged and played an important role in today’s world of crosslingual communication.And the debate about whether machine interpreters can replace human interpreters is endless.Taking Elon Musk’s English Chinese simultaneous interpreting speech at the Space X2022 Starship Press Conference as an example,this thesis makes a comparative study of the three major machine interpretation systems—Iflytek,Baidu AI and i Tour,and six professional simultaneous interpretation interpreters.Under the guidance of James’(2001)error analysis theory,combined with Cai Xiaohong’s(2007)quantitative formula for interpreting quality evaluation and Yang Chengshu’s(2005)quantitative and qualitative evaluation table,error classification and error statistics are conducted by analyzing the interpretation texts of machine and human interpretation from four dimensions of information equivalence,language accuracy,language flow and cultural transformation.This thesis adopts a research method that combines quantitative analysis,and qualitative analysis.The experimental data is statistically analyzed and systematically integrated by using SPSS software.The differences between machine interpretation and human interpretation are compared and the issue of whether machine interpretation can replace human interpretation is explored.The research concludes that,firstly,from quantitative indicators such as mean,variance,and non parametric tests,human interpretation performs better than machine interpretation in the three dimensions of information equivalence,language accuracy,and cultural transformation,while the difference in language flow between the two is not significant.Secondly,from the qualitative perspective,machines can ensure a certain degree of translation integrity under the premise of correct speech recognition.But when there are problems with speech recognition,there will be a large number of inappropriate pauses,causing serious logical errors and losing the overall meaning of the entire discourse.In contrast,human interpreters may make more mistakes in missing numbers and specialized nouns.However,even if there are some omissions,human interpreters can still convey cultural connotations through a variety of strategies such as information integration,free translation,and supplementary information,thereby maintaining the overall semantic integrity and achieving the completeness of information transmission.Speech recognition remains the most important factor affecting the quality of machine interpretation.The ability to filter unnecessary information and the instability of online corpora have led to poor quality of machine interpretation compared with human interpretation.Therefore,machine interpretation can replace some of the physical and mental labor of interpreters in the future,but currently there is still a huge gap from human interpretation and cannot replace it.Human-machine coupling is an inevitable trend.Through empirical research,this thesis discusses the problems and gaps between machine interpretation and human interpretation,analyzes the reasons,finds out the corresponding strategies,and shows the practical application value and significance of the thesis.The research results will help to achieve complementary advantages between human interpreters and machine interpreters,promote interpretation to serve the development of science and technology,and ultimately serve the development and progress of the socialist society and economy with Chinese characteristics.