Transcription is the process of turning spoken words into written text. At first glance, it may seem straightforward. However, transcription is quite a skill to master; one which can sometimes take years to perfect. Furthermore, it is not just the skills of the transcribers that affect how the written data is produced. Other variables also alter the way the transcription process is approached, and the content that is created. Mission Translate explains these factors and explores how their effect on the results of a research project can have a wider impact.
The skills required to produce high-quality transcriptions
As well as an exceptionally attentive listening ability and top-notch typing skills, to produce a high-quality transcription, transcribers need an in-depth knowledge of the language spoken, its use of vocabulary, grammar and punctuation. Being able to put words into context, identifying correct homophones and using critical thinking to make sense of the speech if it becomes distorted, or if accents are heavy, can make the difference in the accuracy of a transcript. Thorough knowledge of the subject area, as well as an understanding of the correct and most up-to-date use of its terminology, is also vital for successful transcription work.
Transcription takes patience. A transcriber may need to listen repeatedly to a section of audio to glean the exact words that are spoken. Having a passion for language and a dedication for ‘getting it right’ are essential attributes of a transcriber.
Working with a transcription service that has linguists with these qualities will help ensure the accuracy of your study. Incorrectly transcribing dialogue from an in-depth interview or focus group session may lead to skewed results and invalid conclusions. Time and budgets are wasted, or worse, if the inaccuracies are not picked up, reputations can be damaged and brand trust lost.
Different types of transcription
Depending on the goal of the research, different levels of detail and methods of information presentation can be required in a transcription. Clarifying specific instructions before the transcription commences is vital to ensure the data needed is received.
Intelligent or full verbatim?
Full verbatim means absolutely everything is transcribed, including all ums, uhs, grammatical and vocabulary mistakes, false starts and repetitions. If the purpose of the research is to study the use of language, then this may be appropriate. Otherwise, full verbatim can be tedious to read.
When using intelligent verbatim for transcription, all the information is still transcribed, but the content is effectively tidied up. All the ‘ums’, ‘uhs’ and verbal tics, such as ‘you know’ and ‘like’ are removed. False starts to sentences are taken out, but only if they do not add any meaning to what the speaker is saying. Intelligent verbatim is usually the standard method of transcription, as the accuracy of the dialogue is maintained, yet edited to be in a more readable form.
Another strategy sometimes used to speed up the process of transcription is direct analysis or summaries. In this scenario, a client will direct the transcriber in locating specific information within a dialogue that addresses their questions. For this to be successful, the transcriber should have experience in research methods, and they should always be fully briefed in the requirements of the project. Miss out that step and there is the risk of inaccurate or missed information.
Templates and timestamps
Providing a template for the transcription will help guide the transcriber in the information required. Laying out how the transcription should be ordered, and in which format, makes the post-analysis a lot easier, especially if the planned conversation goes off track. Including an outline of the questions asked will also make the transcription process more efficient.
Using timestamps to indicate at what moment in the audio dialogue occurred is essential when transcribing for subtitles. It can also be used when transcribing for other purposes. Having a timestamp means the point in the audio can easily be located should any clarification be required.
Other considerations for the best-quality results
When planning in transcription as part of a research project, there are a few other things to consider to help successfully meet your objectives.
Transcription is a lengthy process. Depending on the complexity of the topic and the audio quality, one hour of audio can take between three to ten hours to transcribe. Allowing sufficient time for this process will ensure that it is not rushed and errors are not made.
As mentioned above, the quality of the audio can have a significant impact on the time it takes to transcribe and the accuracy of the transcript. Avoid background noise and people talking over one another as much as possible to obtain as much information as possible for your audio.
Transcribers should have an up-date knowledge of the topic they are transcribing. However, providing a list of technical vocabulary, e.g. product-specific brand names and industry abbreviations, will again assist with accuracy and consistency in the transcription.
The variables around transcription should not be underestimated. Factoring these into your project planning will improve the quality of the data created, assisting your analysis and leading to more valid conclusions in your research study. Delivering quality analysis to your end-clients will strengthen relationships, build brand reputation and increase the potential of return business.
By Lorna Paice