By Dana P Skopal, PhD
As an acronym, it seems people easily say AI, knowing that the term is artificial intelligence.
Interestingly, AI began in 1956 as an academic discipline, while linguistics, as the scientific study of language, began in the mid-19th century. One link between AI and linguistics is how any language is used, and here, we are looking at communication in English.
Another term is generative artificial intelligence tools (GAITs), which are many different computer programs that you can use such as search engines, grammar checkers and chatbots. Many of us would be using the spelling and grammar checkers available in our word-processing programs. Whether writing a business document or school/university assignment, an issue is how AI tools can be used to generate informative texts but be used appropriately and ensure no misinformation. Further, do GAITs impair critical thinking and can linguistic tools help us see the differences between GAIT generated texts and an individual’s own constructed text – that is, their voice?
As linguists, we often tell students to use their voice and take a stance when writing assignments. Voice also links to using active voice, as a good tool in line with the principles of plain English. Professor Ken Hyland, a British linguist, has created many discourse models based on research of how people use language, and there are patterns that allow us to identify a person’s voice in their writing.
Working with Kevin Jiang, Ken Hyland has analysed argumentative essays compared to ChatGPT generated texts. They found that ChatGPT-generated writing was vaguer and more formulaic than human writing, with differences in text structures, nominal groups and stance (a writer’s voice). For example, the percentage of result/ framing signals was considerably higher in the students’ essays than those by ChatGPT. These linguistic signals, such as ‘due to’ and ‘as a result’, indicate either the reason or consequence of an argument and show how a particular line of thought contributes to the overall argument. In a way, this is a form of forensic linguistics.
Writing to get your message across means you should know your main argument. If you don’t, then no matter how many drafts you write or the use of AI tools, your document may be confusing for your reader. Further, as a writer, you need to focus on why your message is important for the organisation/audience.
Reference: Jiang, F. & Hyland, K. (2024). Does ChatGPT Argue Like Students? Bundles in Argumentative Essays. Applied Linguistics. https://doi.org/10.1093/applin/amae052
Copyright © Opal Affinity Pty Ltd 2024