Whilst completing my Norwegian MOOC, I often interacted with a chatbot. It looked like this.
Pretty straight forward. The chatbot says something, asks a question, and I’m expected to respond. I actually found the inclusion of this chatbot quite useful. It was a nice way to wrap up the lesson and put into use what I’d learned because the chatbot, unsurprisingly, was programmed to try and elicit the structures I had just focused on. It also allowed me to see some natural Norwegian vocabulary that might crop up in a similar situation. Compared to the strange chatbots my classmates and I used to mess around with in our school days, the University of Oslo’s chatbot was quite intelligent. I rarely got any nonsense responses and it kept up with me most of the time. There was an occasion, however, where the conversation became a bit frustrating when I responded in the negative to a question and the chatbot had clearly been programmed to continue with a positive answer. This is an issue highlighted by Chantarotwong (2006) and is related to the chatbot not being able to store your responses in its “memory”, leading to questions that don’t make sense or are irrelevant. In addition, Coniam (2008) points out how chatbots don’t deal well with misspelling, something that is likely to inhibit learners’ usage.
Dale (2016) refers to 2016 as “the year of the chatbot”. If you consider chatbots as programmes like the one I posted above, this might seem a bizarre claim, but consider the so-called “digital assistants” that are built into your handheld devices: Siri, Cortana, S Voice. These are chatbots of a kind; they just work through voice rather than text. Even with all our technological advances, chatbots have not quite hit the mark yet, though. As exemplified in this pronunciation teaching video, Siri can still sometimes get it very wrong – even when a native speaker is using the device!
So what do chatbots get right, and how can we and our students use them?
Shawar & Atwell (2007) suggest one interesting usage, which is to allow students to ask chatbots questions and then access the logs to see what students are struggling with. This might sound a bit long-winded to some teachers, but imagine you have some shy students who don’t like to ask questions, or you’re working on a distance course where you can’t always be available. In those kinds of situations, this might work very well! In the case that the chatbot is not able to provide assistance, at least the teacher can step in and fill in the gaps. My only concern when reading this suggestion was privacy issues, but I imagine if a system like this was implemented, it would be on an educational server with students well aware that the teacher could read their input. I see how some might argue that the learners could just post the question to the teacher, but for some, the relative anonymity might be attractive, as might the freedom to ask a question whenever, wherever, even when they are not in front of their computers or in the classroom.
Coniam (2008) proposes that chatbots deal best with simple, single clauses. In regards to technology like Siri, this may allow low-level learners in particular the opportunity to practice basic phrases and questions in a low-risk environment. As portrayed in the pronunciation video above, you can use voice-activated chatbots like Siri to control your device, doing simple tasks like setting an alarm or playing music. Asking the device to perform an action allows learners to see a response or reaction to their utterance, which is a little bit like how a conversation works!
Finally, there is room to combine the fields of computer science and programming with language learning. Bii (2013) offered a scenario in a Biology class where students create the chatbot themselves, using facts provided by the teacher. In this way they are learning a new skill – coding – as well as creating content that can be used later and that may be used to produce language. I believe this kind of usage could also lend itself quite nicely to activities like WebQuests, where learners could search for the information they need to complete the task via a chatbot, or a digital assistant, like Siri, as chatbots seem to fare better when used as mini-encyclopædia.
How do you feel about chatbots and how might you use them? Do you think we will ever create a chatbot that can offer engaging or fully coherent conversation?