Speaking to be Heard over the Roar of AI
LanguageCert, 24 January 2025
When I attended the IATEFL Conference last April, I was struck by the ubiquity of the conversation around large language model, generative AI (AI) and heard my co-panellist brilliantly describe it as the ‘roar of AI’i. This roar is becoming deafening, disorienting, and threatening to significantly disrupt higher education. Disruption in higher education In June, I heard the roar on the radio in a news story about how researchers at the University of Reading submitted under the names of fake students 33 take-home assessments produced by AI. The researchers did not edit or amend the assessments in any way. The AI-generated answers were submitted to five undergraduate modules for the Psychology BSc degree. The markers, who were unaware of the project, identified only one of the answers as being produced by AI and the remaining papers were awarded half a grade boundary higher than answers from real studentsii. In July, I heard the roar from Australia when I read a series of articles about the use of AI in Australian universities. The articles included several unnamed lecturers across disciplines reporting an increase in the use of AI and how software detection methods cannot adequately or fully detect when students use AIiii.
Now in November, the roar can be heard louder than ever with reports of the rapid increase in the inappropriate use of AI at Russell Group universities in the UKiv.
A call for change in assessment methods
Within the roar, I can hear a chorus of voices clamouring that assessment methods at universities and colleges must change. During the COVID-19 pandemic, higher education institutions had to turn to take-home exams or unsupervised coursework, and these assessment systems have remained in place. The emergence of undetectable AI is rendering these methods obsolete and shattering their integrity.
A solution: reviving the viva
Looking to the future, I think one approach to countering AI's unaccountability, anonymity and undetectability is to place greater emphasis on speaking and verbal reasoning skills. To revive the viva.
The viva, or viva voce, is arguably the oldest form of assessment and has been used continuously for centuries. It is the final trial of the PhD rite of passage, but it is not employed across the board at an undergraduate level (although vivas are used for sciences, medicine, law and other disciplines). The value of the viva voce as an antidote to AI is inherent in the phrase’s literal translation - ‘with the living voice’ - from Medieval Latin.
Vivas should complement, not replace, written assessments
I am not suggesting vivas should replace written assessments. Rather, I argue for the more widespread use of vivas across disciplines throughout higher education to complement written assessments. Increasing the use of vivas would place significant demands on institutions and students. It is resource-intensive, time-consuming and administratively challenging. It would require students to develop speaking and verbal reasoning skills necessary to articulate and defend complex ideas under pressure. A skill set often described as oracy.
The importance of oracy in education
Andrew Wilkinson devised the term oracy in the 1960s. His choice of the term oracy was deliberate to position speaking skills as equally important as literacy and numeracy. Originally oracy referred to the goal of speakers being fluent, articulate, and confident in their native language. Over time the concept and teaching of oracy have been applied and introduced to bilingual and additional language educational settings. The term oracy may not be well-known worldwide, but there is an ever-growing awareness of the importance of speaking skills.
The link between poor socio-economic disadvantage and poor oracy
Studies show a strong link between socio-economic disadvantage and poor oracy, which, in turn, negatively impacts individuals' educational and employment outcomes. Oracy is integral to the skills of critical thinking, creativity, collaboration, and communication. Recent research shows that employers see speaking skills as becoming ever more important in the workplace and favour the widespread use of vivas throughout educationv.
Designing a test of real-world speaking skills
With the above in mind, when my colleagues and I began the process of specifying the assessment focus and format of LANGUAGECERT Academic, we looked closely at the present and future needs of students, universities and employers for real-world speaking skills. We identified and defined these skills, and incorporated them into the design of the speaking test, which focuses on communicative language ability and interactional competence.
Human interlocutors and markers
The speaking test is conducted 1:1 with a human interlocutor who facilitates the interaction. The interlocutor asks questions, takes part in the role-playing of real-life scenarios, listens to presentations, and asks follow-up questions. An overall mark is awarded by the interlocutor for task fulfilment and communicative effect. A second human examiner then marks the test recording. Different aspects of speaking performance, including coherence, fluency, the accuracy and range of grammar and vocabulary are all marked according to separate and specific criteria.
Encouraging ownership
In the final part of the speaking test, the test-taker is shown an infographic and asked to speak about the topic of the visual for two minutes. The interlocutor then chooses from a set of follow-up questions to prompt the candidate to expand on their presentation.
This task encourages the test-taker's ownership of the infographic input and their presentation output. Ownership means taking responsibility for their position based on the analysis and evaluation of the information provided, being accountable for the reasoning behind their argument, and justifying their position when prompted by the examiner.
Equipping students for future success
The LANGUAGECERT speaking test is designed to help students develop the speaking and oracy skills they will need to survive, thrive, and flourish at university or college. It is also designed to give students confidence in their own abilities. Students must have the confidence to speak up in the seminar room, to convince in their viva and to overcome the temptation of taking the convenient option of AI.
The complexity of the challenge
The challenges posed by the impact of AI are complex. There are not any overnight or easy solutions. The generative AI genie is out of the bottle and is here to stay. It will take considered and consistent action to stop AI from becoming higher education’s Pandora’s Box.