Jun 052015

Dr Neil Brown mainly teaches Energy Analysis Techniques, Energy Efficiency, and Mechanical and Electronic Engineering Labs, in the School of Engineering and Sustainable development at DMU.
Traditionally, all feedback in the Energy and Sustainable Development (ESD) Subject Group has been text based due to the use of a specific database for communicating feedback to students. The database was partly developed for the benefit of the Distance Learners in ESD who make up the majority of the cohort.

Neil’s  biggest single marking load is Energy Analysis Techniques, this is a core module to three MSc courses and the assessment comprises of two written components. To provide as much meaningful feedback to students as is possible and to be able to mark efficiently and away from the university whilst offline; he has identified an innovative and efficient way to provide feedback that his students have also embraced.

The approach adopted bypasses the computer keyboard by using speech to text software to simply dictate to the computer. Using this approach it’s possible to generate feedback much more quickly, with less fatigue, and allowing concentration on the subject in hand.  He also uses this technique to generate course notes for Distance Learners and he has found that dictation can be around 5-6x faster than typing.

For marking, the overall process is not sped up massively, but the extra detail possible in feedback means that there are almost zero queries on marks from students, which in itself offers a massive time saving. One recent comment was that a student was ‘blown away’ by the amount of feedback.

For Energy Analysis Techniques, comments on each report are grouped as; general comments, notable good features, and areas for improvement.  Comments could also be placed in the submitted PDF of each assignment. This is done in conjunction with grid marking, where a spreadsheet is used to generate marks based on weighted criteria.  It’s not vital to mark in this way, but grouping comments this way, plus grid marking makes things easier still.

Neil uses Dragon Naturally Speaking 10, which now costs around £30. The basic microphone which comes boxed with the software works reasonably well, but he has found that suppliers of dictation software to GPs etc. offer microphones with much better results – expect to spend around £30-50.
Usually, the dictation is carried out using a basic Dell laptop from 2010, running Windows 7. The Dragon Naturally Speaking CD installs itself in Windows and the software can be configured to run on Linux with some tweaking, and Mac OS. He has also trialled other speech to text solutions such as Google speech recognition and IBM ViaVoice but the Google product proved less reliable on accuracy.  The IBM product worked well but it did require significantly more training.

To dictate, a microphone is plugged into the laptop and the Dragon software is started along with the application (Word, Excel, Open/Libre office, Notepad). Training the software to recognise a specific voice takes around 30 minutes and involves reading some set passages before dictating for real. This ‘training’ can be one-off, although the software does become more accurate with more use by the same person/voice.

A bespoke database had been used in the past, long before Blackboard was used for providing feedback, but now the subject group uses DMU’s Blackboard Learn VLE installation. Blackboard Learn offers the chance to provide audio feedback too, circumventing text altogether. Neil and the ELT Project Officer discussed this and Neil tested this multimedia based audio feedback approach, although after a trial the students stated a preference for text as text is easier to skim read and pick out the salient points. He also felt that the audio files were rather lengthy, handling them became fiddly for a large cohort, so has now reverted to dictation.

This approach to providing rich text based electronic feedback not only benefits students but colleagues who may have a disability could also adopt this technique to speak their feedback.  The software can also control the computer, offering improved functionality for anyone who is differently able.

Neil’s top tips for those who may wish to replicate this practice would be:

  1. Use a good quality microphone – background noise can reduce the accuracy of the software
  2. Set the software to be as accurate as possible and speak clearly
  3. Skim read the output text before releasing to the student as some specialist words or phrases can be misinterpreted
  4. Understand your students – Energy Analysis Techniques students prefer text based feedback but in other subjects it may be more appropriate to provide audio, text, or feedback in other media.

Ian Pettit, Neil Brown