Page 9 - ELG2309 Sep Issue 486

RESEARCH NEWS
Algorithm provides personalised

feedback for trainee teachers

By Gillian Ragsdale appropriate feedback on both feedback; working individually with
An artificial neural network can their diagnoses and justifications. 3) static or 4) adaptive feedback.
be trained to provide effective Since this is potentially very time- Analysis of trainee responses
feedback on trainee teachers’ consuming for teacher trainers, it is showed that the use of adaptive
diagnostic reasoning, according to common to have ‘static’ feedback feedback improved the quality
Michail Sailer and colleagues in in the form of preprepared expert of the trainees’ justifications –
Germany and the UK. diagnoses for comparison. whether working individually
One key skill teachers learn To provide ‘adaptive’ feedback or collaboratively – but not the
during training is to diagnose to individual trainee responses, accuracy of their diagnoses.
what kind of difficulty a student Sailer et al made use of artificial Students working collaboratively
is having and then clearly explain intelligence; in particular, natural actually tended to have poorer expert-time investment could
and justify their diagnosis. language processing (NLP) which diagnostic accuracy, which was pay off in expanded training
One accessible way of practicing can analyse and respond to human somewhat ameliorated by adaptive opportunities with limited further
these skills is to use simulation- language, thus providing real- feedback, especially for more investment. At least we do still
based cases. In this study, the time feedback. The NLP-based complex cases; a curious and need human experts – for now.
simulations were presented using algorithm was trained on relevant unexpected outcome worth further
the computer platform CASUS. diagnostic contexts making use of study. REFERENCE
The simulations provided materials data from previous trainees. Overall, the use of adaptive n Sailer, M., Bauer, E.,
such as transcripts of teacher-parent For the study, 178 pre-service feedback of this kind shows Hofman, R., Kiesewetter, Glas,
conversations, assignments, and teachers were recruited. As well potential for providing immediate, J., Gurevych, I. and Fischer,
descriptions of student behaviours. as comparing ‘static’ vs ‘adaptive’ personalised training. The initial F. (2023) Adaptive feedback
Appropriate diagnoses ranged from feedback, Sailer et al wanted to setting up of the algorithm is still from artificial neural networks
specific learning difficulties, such assess the accuracy of diagnoses quite time-consuming, however, facilitates pre-service teachers’
as dyslexia, to disorders such as made collaboratively vs individually. in this case, 40 expert feedback diagnostic reasoning in simulation-
attention-deficit (ADD/ADHD). Trainees were put into one of four paragraphs were supplied for each based learning, Learning and
In order for the trainees to learn groups: working collaboratively case. For large teacher training Instruction. https://doi.org/10.1016/j.
from the practice cases, they need with 1) static or 2) adaptive programmes, however, this initial learninstruc.2022.101620

LEAVE NOTHING TO CHANCE!

Hone your students’ skills and boost their confidence through our supplementary resources!
-20 %

www.expresspublishing.co.uk

editorial@elgazette.com 9