Transcribing Spoken Data

Why do we Transcribe Spoken Data?

Due to the nature of spoken language, once we’ve heard it, we generally can’t hear it again.

To collect spoken data that we can listen to again, we must record it. This can be done either as an audio recording or as an audio-visual recording (video) where we can then listen to the spoken data as many times as we need.

Although having audio recordings is important when analysing spoken data, it’s not always the most useful way to store data, as it can't be analysed and it can make it difficult to find a specific piece of data quickly.

We transcribe spoken data so that we have a written form of it. This makes it much easier to analyse what has been said and how. Looking at the content of the spoken data (such as topics, words and interruptions) can be useful in areas of linguistics like sociolinguistics where we may need to analyse and compare the language of different speakers.

Language differences can vary among speakers and can be related to social aspects such as age, class, gender, occupation, ethnicity and region.

Another reason why we transcribe spoken data is to look at a person’s accent and pronunciation features. This is done by transcribing data using the International Phonetic Alphabet, which we’ll look at in a bit more detail later. Doing this allows greater and more specific speech analysis in fields such as phonetics and phonology.

Fig. 1 - Transcribing data involves writing it out.

Transcription of data in research

Before transcribing, you first need to collect the data. This is done most often through recording spoken language either as an audio recording or recording as a video – having a video may be useful for looking at things such as NVC within a person's speech.

When recording and transcribing data, certain factors need to be considered. These are ethics and the observer’s paradox.

Ethics

In relation to ethics, we need to think about what is the morally right practice as researchers. As spoken language is produced by an individual and is unique to that individual, you need their permission to record them.

If you don’t ask permission before recording someone, it could be considered a breach of that person’s privacy. Every study that requires spoken data has to first go through ethical considerations and make sure that permission has been asked for where it is needed.

The observer’s paradox

The observer’s paradox is the name given to the problem that arises when trying to record natural spoken language. Most natural speech occurs when the speakers are completely at ease and talking casually amongst themselves.

When recording data though, there is usually an observer (the person recording the data) or at the very least a recording device. Due to ethical considerations, the speakers will also know that they are being recorded. As much as people may try to speak naturally, there is always an element of being a bit on edge when you know you’re being recorded or listened to. This may cause the speaker to either consciously or subconsciously alter how they speak.

How to overcome observer’s paradox

When collecting data, you can make certain allowances for observer’s paradox to overcome it. One thing you could do is ask for permission to record someone’s speech in advance of doing it and then record them when they’re not expecting it. With this method, you’ll have to let them listen to what you recorded before you use it as data to make sure they’re happy with you using it.

Another way to try and sidestep the observer’s paradox is to let people know that you are recording them and then lead the conversation through some casual topics before you get to the conversation you want to record.

By doing this, you’ll allow the speakers to get accustomed to being recorded and settle into speaking more naturally by the time it gets to the data you need. This will hopefully encourage more natural speech.

Transcribing Data

Before you start writing out your data into transcript form, you’ll need to write a sentence or two outlining some basic context. This will need to include:

• Where and when the interaction is taking place

• Who the speakers are

• Any contextual information relevant to your study, for example, the gender of the speakers if you’re looking at language and gender

When writing out a transcript, you’ll first need to listen to your recording and write out what was said. It’s a good idea to listen to the recording a few times to make sure you write what you actually hear and not what you expect to hear.

If something is said that is unusual or of note (this will depend on what you’re looking for), it’s a good idea to annotate this on your transcript and to listen through again to see if it appears anywhere else as well.

Making note of specific speech sounds, such as phonemes can be done by using the International Phonetic Alphabet.

What is the International Phonetic Alphabet?

The International Phonetic Alphabet (IPA) was developed in the 19^th century as an internationally recognised system of phonetic symbols. Each symbol corresponds to one specific speech sound, removing the confusion caused by having multiple sounds represented by the same letters.

You can have a look at all of the different symbols are in the IPA chart here.

^{Fig. 2 - IPA Chart.}

How to use the IPA when Transcribing Spoken Data

Using IPA in transcribing spoken data can make your data much more accurate and can be especially useful if you're looking at accent features such as vowel pronunciation in your spoken data. In A-level English language, you won’t be expected to transcribe whole extracts into IPA, but you will be expected to have a basic understanding of it.

Let's look at an example of how the IPA can be used to show pronunciation features.

In the IPA chart, there are also diacritics and suprasegmentals which are the small marks placed next to, under, or on top of vowel or consonant symbols and give much greater information about the prosodic features of the speech sounds.

The use of suprasegmentals and diacritics can be used to show stress, syllables and the linking of speech so that you can represent in written form exactly how something has been said. When adding diacritics and suprasegmentals into your transcription, you need to use square brackets around the transcribed speech to show that it's narrow transcription.

Transcript example

This transcript is an extract from a recorded conversation between two friends (Polly and Laura) who are planning a trip. You can spot some of the features from the table earlier.

What are we looking at in this example?

• Line 1 is an example of an utterance without any notable speech features.

• In line 2, we can see that Laura took a pause of half a second before she started speaking, and then took another micro-pause later on in her utterance.

• In line 4, Polly pauses for two seconds and then we see an example of simultaneous speech. In this simultaneous speech, Polly on line 4 says "how about girls" while Laura on line 6 says "how about we." As the lines are around those two sections of utterances, these are the only two sections that are spoken simultaneously.

• In lines 7 and 8, we can see an interruption where the double slanted brackets are. Here, Polly interrupts Laura and then carries on speaking.

Citing speech transcriptions

When you first reference the transcript you’re talking about in your work, it’s usually good to cite the year and to give an overview of the general context, saying briefly who the speakers are and where the conversation is taking place (providing it’s relevant to what you’re discussing). From then on, it’s usually fine to reference a line number (as all transcripts should have numbered lines) and also state who is speaking to make it clear for your reader.

Quoting transcriptions

When quoting a short utterance or a word, simply put it in quote marks as you would when quoting a book.

When you are explaining something with the help of the IPA, make sure to put that part in slanted brackets.

When quoting multiple lines, do it as a separate section underneath your paragraph and then do your explanation underneath, making sure to still reference specific line numbers.