Legend Series #5 - Atsushi Mizumoto (Kansai University)
- haswell247
- May 4
- 3 min read
Chris Cooper came on board with Lost in Citations in 2024 and, to date, has produced 10 interviews, giving him the third-most interviews contributed. Chris is a really switched-on interviewer, knowledgeable and insightful, and his interviews reflect his genuine attention to detail and interest in the topic being discussed. His most-interacted-with episode was the interview with Atsushi Mizumoto from Kansai University about the corpus-linguistic analyzer software CAF. By linking the well-researched area of corpus-based linguistic analysis with the most up-to-date online tools, this interview is a great example of Chris’s value to this project.
As a start to this blog, I have to say that, unlike all the other blogs I have rewritten after re-listening to the audio and rereading the transcript, this was the interview I had to go over the most to get to the finer details of the work being discussed. You see, I am not by nature very good with technology, have a limited background in quantitative analysis, and have an inherent mistrust of systemic linguistics (this goes back to the work I had to do as part of my MA, which still gives me nightmares). That being said, I fully respect the art of what Professor Mizumoto is developing and his ability to explain it in a way that doesn’t sound like witchcraft to people who use almost exclusively pencil and paper to work things out (which, in this story, is me).
Chris begins the interview by noting how many interviewees we have had on the podcast from Kansai University: there have been 16 interviews with people from this institution covering subjects such as university internationalization, metaphor and metonymy, translation work in Japanese hospitals, and, of course, Professor Mizumoto’s CAF analyzer, a link to which can be found here: langtest.jp.
They then talk about how the introduction or acceptance of AI is not always easy for many people already well established in the field; when one has done something one way for so long, why would they want to change?
[11:39]
Chris:
Some of the teachers that I work with, some people are kind of against using AI. They've completely ban it. And then I've been more like in the last couple of years, actually, probably just the last year or a year and a half, because it's really, yeah, it's still really soon, right? Trying to give my students advice how they should use and shouldn't, like you said, kind of but it's still kind of difficult finding the boundaries
...
Atsushi:
Exactly, because that's all based on the teachers' or learners' beliefs, perceptions, or mindsets. Learning, acquisition, learning, language learning, yeah, teachers didn't learn in that way.
And, that is the way you have to address changes in technology: when there is no apparent allegory for the procedure in your experience, why would you embrace it? Atsushi Mizumoto’s job is made harder by the perception that AI doesn’t just make the job more efficient, it replaces the need for a human participant.
Professor Mizumoto then goes on to give specific examples of how the system he is developing helps and supports rather than overtakes and replaces human participation:
[33:33] Open AI API, okay, works in the same way as open AI API. So in any API. And you know, the documentation gives you all the Python code, so I was really surprised, actually to find the how accurate it is. Yeah, and especially for this specific task, accuracy measurement or the detecting errors in writing, because it's really consistent. Because, you know, when it comes to using generative AI, it produces different outputs every time, but right, not for this one … you're familiar with Grammarly, it's like the stimuli.
At this point, generative AI is giving you, the human operator, suggestions for your output and edits, but you are still in control. As someone who uses Grammarly extensively in the final editing stage of my writing, I like the options I receive from the perspective of the reader: am I making my point as clearly as possible, or have I, as I am wont to do, gone down an ungrammatical wormhole of a sentence from which I need rescuing? Case in point…
I’ll not parse out any more of the interview for this blog - I think you will get a much better understanding from listening to Chris and Atsushi go back and forth on the topic and the contents of the tools available at the website, once again available here: langtest.jp.
Go on - have a listen!
Comments