基于SpeechRater^？不同分数段学生口语作答常见问题分析
An Analysis of Common Errors on Students’ Speaking Performance in Different Levels Based on SpeechRater^？ Scoring System

随着计算机技术的飞速发展，人工智能技术也在不断突破，计算机的自动评分技术作为一种新的评阅方式逐渐应用于口语考试的评分之中。美国教育考试服务中心ETS自2006年开始将SpeechRater^？用于官方在线练习口语的评分中后，经过多年研究不断优化，将其自动评分系统对外开放，从五大方面12个维度给予学生口语表现反馈。中国大陆学生托福口语平均分常年处于19分左右，并且大多数学生口语分数无法突破瓶颈24分，为了帮助学生找到问题所在，本文对SpeechRater^？不同分数段学生口语作答中词汇和语法上的常见问题进行了初探，希望能为教师提供一种新的视角去了解在教学中该如何使用好SpeechRater^？。
As one of the influential achievements in today’s artificial intelligence arena, the automated speech scoring system has been implemented in the official examinations as an innovative approach. Now, ETS, as the developer of SpeechRater^？, has devoted years of painstaking research to update the system after its initial use in TPO in 2006 and has decided to make the scoring system more acces-sible for the public users. The SpeechRater^？ can provide informative feedback for a given response, and help its users to have a better understanding of their speaking performances. Due to the in-creasingly large number of English learners in China, there is a growing demand for better prac-ticing tools of speaking ability. Besides, it is acknowledged that the average TOEFL iBT speaking score of students in mainland China is round 19, which remains unchanged for ten years. Further-more, most middle-level students whose speaking scores are around 24 find it is harder to reach a higher score than they expected. The goal of this study is to find common errors regarding the usage of vocabulary and grammar in speaking responses of students in different levels, and I hope I can provide a new perspective for teachers to get a sense of how to use SpeechRater^？ as a useful tool in teaching.