Violeta Popadic, online English teacher at Acadsoc.
Want to know more about Acadsoc?
Want to be an online English teacher and write for us?
Apply via acadsoc.ph !
Artificial Intelligence has widely applied in many different territories, i.e. driverless cars, or for score judging in 2020 Olympics. It sounds so incredible to make sports so accurate like constructing a building, but we have to admit it is the quickest and fairest way to tell the performance of each athlete. AI has been discussed in Acadsoc blog previously and it seems many of our teachers have a strong interest in it.
A Japanese company called Fujitsu is developing an AI-based software using the data collected by 3D sensors to analyze gymnastics events. With the big data allocated from 2016 Olympics and other gymnastics races, the computer can automatically simulate the exact position of each gymnast, with a combination of the athlete’s height of jumps, distance, and angle of limbs.[i] Hence, from 2020 Olympics, there can be more and more AI judges in all kinds of races.
Such revolution brings up a possibility: can we evaluate student’s English level using AI technology as well? With real-time information captured by speech sensor and converter, enrolled information of the student and certain algorithm, yes, it is possible. And we can easily find the advantage of such evaluation: faster, more accurate and more natural for the student. Using AI, students don’t have to finish thick and complicated answer sheets and hear a subjective judgment from their teachers before class begins. As for teachers, they will be able to concentrate more on teaching instead of mock exams and change their teaching strategy based on the analytics and suggestions given by AI.
What kind of information should be collected?
This can be a very confusing angle for many English teaching schools in China because most of them will try to collect as much information as possible from their students. However, it will be dangerously helpless. Too much irrelevant information may twist the calculation of AI, letting AI learn useless information from the internet and another database, making the judgment of AI far from accurate one and cause higher risks in students’ privacy leakage.
Actually, it has been proved by Acadsoc Research Institute© and many other English teaching research facilities that several key factors will be enough to judge the real English level of a student.
No matter it is just a conversational sentence given by the teacher or it is a question, there should be an expected response to the sentence in a certain environment. For instance, if a teacher tells his/her student ‘You look so cute today.’ Usually, the expected answer should be ‘Thank you and …’ to show a proper manner of gratuity when receiving a compliment in English. While some Chinese student may respond ‘You look so cute as well’ Or ‘Nahh (Laughing), you must be kidding’ or just too shy to say anything, then we shall add the relationship between the teacher and the student into consideration as well, which can be completed by AI. If they are strangers before who firstly met in a classroom, as required, the student shall use more official tones at first. Please note, the ability to give the expected answer does not require the answer to be perfect because a different student may use different expressions due to the difference of their cultural and linguistic background. Some Chinese students may accidentally say ‘谢谢 (Thank you)’ in Chinese, however, AI will not just give a 0 score for that. At least the student knows when to use gratitude tone when hearing English.
This can be difficult but very important for many ELLs (English Language Learners), it requires listening skills to comprehend the conversation or at least several keywords in it. Especially when hearing a long story, non-native speakers can struggle to understand the whole meaning word for word or it may take time for them to convert English back to their mother language. However, with a smart strategy by capturing key information in a story, they should be able to make a suitable response. Please note, not just response in words, but the facial movements, eye movements, and tones.
For instance, the teacher says ‘Chinese Spring Festival is just unreal for me!’ If the student realizes that ‘unreal’ here means cool and shocked, he/she may say something more about Chinese Spring Festival as they know, i.e. their personal experience during Spring Festival with a happy emotion shown in their tone and face. Or if the student is confused, he/she should show a thinking face and say ‘unreal? /Pardon? /Sorry, I don’t get what you mean…’ Or they can just express their happiness that someone has a common feeling as they do… A teacher may be hard to tell by what extent the student understands your words because sometimes the student can use some general response and emotion to pretend that they get what you mean.
However, with the support of rich data (language, facial, tone, eye focus, even heartbeats) collected and analyzed from AI, it is easy to tell if the student really gets what you mean or he/she is pretending to catch up with you. As we believe, the key reason people may get fooled or misled is lack of sufficient information. And as mentioned above, it may require more advanced technology and sensors to engage alike the ones applied in lie detectors maybe.
As discussed by ETS[ii], there is a diversity of English usage among students. Hence, the preference in English expression can be a key factor to take for AI as well. There is no standardized answer to open questions in English; however, AI may tell which part of English you are better at by analyzing the preference of your word, phrase and sentence structure preference. For instance, if someone keeps using oral expressions, short and easy phrase and simple sentence structure with many grammatical mistakes, this may indicate the student might be just all right for daily English speaking, but facing some trouble when using official and academic English. On the contrary, as a result of English learning of Chinese students from books, they usually use too much written English and uncommon expression in conversation. Well, there is no compulsory regulation that what kind of English you should use every day if they are all grammatically correct.
But, it shall be taken by AI as a disadvantage in English Speaking because usually, people don’t have that much time to carefully think about a speech like what they are doing when reading a written sentence. There is no 100% accurate evaluations that AI can make just based on the preference of an English speaker. Such information must be taken into account along with other necessary information. But there is no doubt that, if a student can fluently change their word usage and sentence structure on different occasions, he/she must have a very close level to native English speakers.
We put this as the less important and lowest prioritized factor for AI simply because many native speakers will make grammar mistakes as well, or, there is no so-called correct English in different parts of the world. Even in different states in the US, English grammar can be different. Hence, AI will judge the grammar based not on dictionaries but how exactly native English speakers use it every day in social media, articles, interviews and any other public release.
Nevertheless, there are still some basic criteria that an English sentence shall follow, e.g. 1 verb at most is allowed in 1 single sentence, all verbs should be in the same tense if things happen concurrently, there must be a subject in a sentence (unless in some oral expressions), etc. The key and core of AI judgment here is to make sure a student should use the grammar as most native speakers will use, but not necessarily the grammar listed in a dictionary. With the aid of rich information collected from the internet and a real-time update of English usage everywhere, AI can make more accurate judgments than human beings. Actually, many teachers have started to use some automatic grammar checker so far.
Welcome to join our discussion on how AI will change online English classroom in the future. If you have any tips and ideas, please join us via acadsoc.ph!