4 Steps to Take Before Transcribing the Data in Your Qualitative Research

THE LAI BLOG

4 Steps to take before transcribing the data in your qualitative research

by Diana Carrillo / Landmark Associates Inc. / Transcription / 23/09/19

One of the biggest logistical challenges in qualitative research is the transcription of recorded video or audio data into a written document that can be coded, organized and analyzed. While on the surface, data transcription may seem like a pretty straight forward process, there are actually several considerations you need to have in mind before you settle on a particular source or method of transcription. These considerations multiply in importance when you are dealing with many hours of recorded data and if you are working in multilingual settings or with international teams.

Simply feeding your audio and video files to an automated platform or even working with a human transcription and translation service has its limitations if you don’t go into the process prepared. The following are four main steps you should take before moving forward with the transcription process in your qualitative research:

Step 1: Decide what kind of transcription you need

The very first thing you need to do is to be clear about the type of transcription you will need for your research. There are basically three main types of transcription to choose from:

• Verbatim

Each and every word is transcribed into text exactly as it sounds. This includes filler sounds, such as “uh” and “ummm” as well as filler words, such as “like,” “so,” and “you know.” Contextual content, such as emotional expression and speaker tone may also be noted.

• Intelligent

In this case the transcribed text excludes most filler sounds, words, and non-contextual pauses and instead renders a “clean” written version of the recorded conversation

• Edited

With edited transcription, the text is not only free of filler speech and pauses, the transcriptionist also edits the spoken sentences so that they make more sense.

Step 2: Determine who will do the transcribing

You have three options when it comes to choosing who will do the transcription of your data:

1. Selecting a designated member of your research team
2. Using an automated transcription platform
3. Outsourcing to a company that employs professional human transcribers

Each option has its own benefits and limitations. The problem with relying on a team member to do the transcribing is the amount of time, energy, and expertise the task requires. Consider, for example, that the industry standard for transcribing one hour’s worth of recorded data is more than four hours, and that is a situation where the audio quality is good, the transcriber is trained, and this person also has the right equipment. Tapping a team member, however, may be the most cost effective option– especially if there is only a few minutes or hours of audio to transcribe.

If you are working on a tight budget and can’t assign someone in-house to take on the transcription, then you can choose an automated transcription platform which relies on speech recognition technology to convert audio to text. While this is definitely the quickest and easiest option, you may experience problems with the resulting transcript’s accuracy since automated systems are currently limited particularly where the audio quality is poor, or there are heavy accents, industry jargon, and numerous people speaking simultaneously.

Outsourcing your transcribing work to a professional human transcription service is the most popular option among academics and qualitative researchers. In this case, with a vetted company, you can expect the highest accuracy and quality. This is extremely important if your recordings involve local dialects, industry-specific terminology, or speakers with heavy accents. It also gives you the most control over the type of transcription needed as well as its format. The biggest downside to using a professional human transcription is that it is the most expensive option.

Step 3: Decide which language(s) to use

When you are dealing with speakers of multiple languages or are working on an international research team, you need to decide which common language to use and after that, how and when to conduct the translation. One transcription and translation method involves generating the transcript in the language in which the interview was conducted without being simplified or condensed in anyway. After the transcript has been completed it is then translated it into another language. This is a good strategy if want the ability to cross-check the original text against the translation.

Step 4: Determine which metadata to include and which transcription structure to use

If you are using human transcription, whether via a team member or an outside professional service, then you should also consider which metadata you want included. The metadata involves things like: the name or ID of the participant(s), the date and location of the recording, and the name of interviewer. Usually, these qualities are placed at the top of the transcript. At the end of the transcript, you may want to include any important observations or notes made by your team members.

You should also consider what the transcription should look like so there is a consistent structure and format throughout. This includes when to start a new line, how pauses of different length are recorded, as well as how sounds, such as coughing, sighing, and laughter are noted.

Once you’ve given consideration to the above 4 steps, you’ll be in the best position to get the transcription you need so you can move your research forward in the easiest and most accurate way.