Reading Reflection 2026.02.05

Reading Reflection 2026.02.05

Yu Borui

For this week's reading, I mainly focused on New perspectives of L2 acquisition related to human-computer interaction (HCI) written by Meunier. 

According to Song et al., while voice-activated AI assistants are highly effective at processing spoken commands from native speakers, they are far less effective when the commands are given by second-language speakers. The accuracy for a second language learner is approximately 55 percent whereas the one for a native speaker rise to around 98 percent. Despite functional fluency in a second language, subtle but significant pronunciation difficulties may persist, impacting human-computer cognition. However, this also proves that speech AI can also serve as a teaching tool for learning and improving second language pronunciation. This is because that AI assistant will give a synchronal feedback when the learner provides a linguistic practice, whether it is positive or negative, they will know what they need to improve and what they have already achieved. As for my own experience, when I use Siri, most of the time, it will immediately understand what I say. But I do make some mistakes, like unclear pronunciation, grammatical mistakes or incorrect use of words. When Siri receives those mistakes, it will mention the confusion, I will get my feedback and then correct my mistakes right away. I personally regard AI assistant as a useful tool for my English learning, whether for listening practice or spoken English refresh, it's highly effective for me. Yet, this could cause some confusion and troubles for English learners in lower English proficency. When they make mistakes, it might be tough for them to acknowledge the mistakes directly pointed by an AI assistant. This will potentially lead to their avoidance of solving problems and their fear of unpredictability and uncertainty of mistakes. So, the most important thing for them is learning to develop and build the psychological construction for using AI assistant as a tool in their progress of linguistic acquisition.

When mentioning HCI and second language aqcuisition, comparing to traditional textbooks, it is relatively flexible. The AI assistant (tools) can be autimatically adjusted according to the learner's personal learning progress. The changes can be found in various aspects, like the difficulty of vocabulary, the complexity of the sentences, even the relativity of the topics. Another advantage is that, even it is only a collection of digital tools instead of human beings, it means that they are more likely to guide the mutual conversation in a higher level of correctness (the students are able to avoid potential mistakes that they might encounter in human-human conversations), also, it is worth mentioning that, the integration system could still make the learners feel that they are talking to a real person.

To be honest, I think the biggest role of AI voice tools in second language learning is actually to make people dare to speak. I feel this way myself. Sometimes I have sentences in my head in class, but I just don't dare to say them, for fear that my pronunciation is not accurate, that others don't understand, or that I am embarrassed to make mistakes. But talking to AI is different. It won't frown or interrupt you. You can repeat it over and over again, even if you are wrong, there is no psychological burden. Especially the function of speech recognition, if it doesn't understand you, it is actually reminding you where there is something wrong with pronunciation. This kind of feedback is quite direct. But I also think it is just a tool after all, and it is still different from communicating with real people. No matter how the machine simulates the dialogue, there is no real emotion, and there is no feeling of suddenly answering words, interrupting and changing tone. Sometimes if you talk to AI for a long time, the expression will become a bit stereotyped, as if it is just to complete the dialogue task, rather than what you really want to express. So I think it is very suitable for practicing pronunciation and reaction speed, but it can't completely replace real communication.