>picture of rachael
Dr. Rachael-Anne Knight 
Senior Lecturer in Phonetics
Department of Language and Communication Science
City University London
Tel: 020 7040 8081
email: knight@city.ac.uk or rachaelanne@cantab.net
 
 
Phonetic transcription research

Background
My interest in phonetic transcription falls out of three related interests.  Of course, as a phonetician I am interested in transcription as one of the tools of the trade, and as a high-level skill, shared by relatively few individuals.  As I also have a background in psycholinguistics, I am interested in how we might unpack and model the transcription process, and what skills underlie it.  Finally, as a teacher I am in interested in the implications for pedagogy: can knowing more about transcription help students to be better transcribers, and how can we facilitate learning transcription?
 
My work on transcription falls under three broad research questions.
1. What are the skills and processes underlying transcription, and how can these be modelled?
2. What factors of the speech signal, and the way it is presented, affect transcription accuracy and reliability?
3. How can innovative teaching methods and technologies enable students to become better transcribers?
 
What are the skills and processes underlying transcription, and how can these be modelled?
As part of my sabbatical (Jan-September 2011) I am trying to work out how the process of transcription can be modelled, and if it can be fitted into existing models of language processing (such as Ellis and Young 1988, or Stackhouse and Wells 1997).  This involves thinking about the skills and knowledge that underlie transcription, and how they relate to other areas of processing (speech perception, symbolic representation etc).
 
Esther Maguire, in her BSc4 honours project (and in Knight and Maguire submitted), investigated if/how short-term memory is related to transcription ability in students.  She found strong positive correlations between measures of short-term and working memory (such as digit span), and studentsí scores on phonetics class tests.  This suggests that we should incorporate short-term memory into our model, and we are currently working out how to do this.
 
What factors of the speech signal, and the way it is presented, affect transcription accuracy and reliability?
In Knight (2010) I investigated how the accuracy of student transcriptions was influenced by the number of times a signal was repeated, and the number of voices in which it was presented.  Accuracy increased significantly between 6 and 10 repetitions, regardless of whether the repetitions were presented in one or two voices.  It was suggested that transcribers may only be able to hold small portions of the signal in their short term memories, and that increased repetitions allow them to attend to more sections of the signal, increasing accuracy.
 
This study was supported by a student survey concerning the issues of repetitions and voices (Knight 2009).  Of 33 second year undergraduate students, more than 75% rated themselves as fairly or very competent in transcription.  In general, students thought that they improve with more repetitions, but appreciated that more than about 10 would be unlikely to improve performance, although they didnít think they would start to make mistakes.  Students were also worried about transcribing voices they hadnít heard before, even though the experimental evidence from the study above showed that students actually perform well with unfamiliar voices.
 
How can innovative teaching methods and technologies enable students to become better transcribers?
In Knight (accepted), I investigated the use students made of exercise podcasts for phonetics.  This study was novel as most podcasts are of lecture material rather than exercise material.  Of course podcasts are ideal for phonetics as they allow students to practice transcription from speech, rather than from orthography. Responses to a student survey showed that students found the exercise podcasts very useful for their learning.  They liked the ability to repeat the recordings many times and felt that there was improvement in their confidence in transcription and in their test scores due to using them.  Students tended to listen to the podcasts on a computer at home, rather than on an mp3 player when on the move.  Many students also listened to the podcasts with family and friends, indicating that there was a social aspect to their learning.  
 
Looking forward
This research is still in the very early stages.  If you would like to offer me data, ideas or suggestions, or collaborate in any other way, please get in touch.
 
Links to the cited articles can be found on my Ďpublicationsípage.
 
Knight, R-A. (2009) 'Feeling confident about transcription (?): A student survey concerning numbers of repetitions and new voices'. Proceedings of the Phonetics Teaching and Learning Conference, 2009, University College London. 
 
Knight, R-A. (2010) 'Transcribing nonsense words:  The effect of numbers of repetitions and voices'. Clinical Linguistics and Phonetics, 24, No. 6: 473Ė484
 
Knight, R-A. (accepted for 2010) Sounds for study: Speech and language therapy studentsí use and perception of exercise podcasts for phonetics (preprint). International Journal of Teaching and Learning in Higher Education, Volume 22, 3.
 
Knight, R-A. and Maguire, E. (submitted). The relationship between phonological short-term and working memory and the phonetic transcription ability of speech-language pathology students.
 
Last updated 10.03.11