In linguistics (as opposed to biology or music), transcribing is the systematic representation of speech or audio in a written document. It’s a key part of qualitative research.
Unlike phonetic transcribing, orthographic transcription uses the rules for mapping spoken words onto written forms of the language in question. This mapping is not a straightforward process, because written language is an idealisation made up of a set of distinct and discrete symbols, while speech is continuous and has a potentially unlimited number of components.
In musical terms, a transcript is an arrangement of a composition for an instrument or a voice other than the original. Because the ‘voices’ and rules of speech and writing differ strongly, transcribing is not merely ‘writing out’ or ‘typing out’. It is a subject in its own right because there are different forms and guidelines and because it is so time-consuming.
Transcription was originally done manually and analogue. Nowadays, most transcribing is done on computers. Recordings are usually digital audio or video files, and transcriptions are electronic documents. Two software types can assist the process: one that facilitates manual transcription and automated transcription. Non-automatic transcription hardware (e.g. transcription headphones and a foot pedal) also make transcribing easier.
Because automatic transcription software’s accuracy is sensitive to poor audio quality and because the technology is imperfect, it’s best to check and correct an automatic transcript. The advantage of automatic transcription is that you need only correct instead of starting from scratch.
There are two types of transcribing; depending on the transcribing’s purpose, one or the other is more suitable. Literal, verbatim or narrow transcription seeks to record how something is said through a letter-by-letter transcript. It seeks to capture the details of a conversation, such as which words are stressed, articulated, or spoken more loudly, as well as the points at which the turns-at-talk overlap.
Here, interjections, repetitions, stutters, interrupting words and colloquial language are literally reproduced. Because our speech is mostly grammatically incorrect and spoken sentences are often very long, literal transcripts take longer to read. Verbatim also has the effect of making an interviewee look incoherent or semi-literate. Yet literal transcription is sometimes necessary, such as in psychological research or in transcripts for legal purposes.
If such a high level of detail is not necessary, the second transcription type is considered sufficient. Edited, broad or clean transcribing seeks to render a conversation’s content in a clearly legible form. Here, the transcriber deliberately omits half-sentences, aborted words and interjections, and represents what was spoken as grammatically correct as possible. In an edited transcript, a conversation’s content is perfectly reproduced, while the way something was said is less important.
Hence these recommendations:
“Use punctuation as well as paragraph breaks to facilitate reading.If the interviewee or respondent makes small grammar ‘errors’ such as incorrect verb and subject agreement or incorrect tense, write down the correct form – the rules that apply to written and spoken English differ.Leave out back-channelling indicators (uh huh, mm hmm), false starts (You see it’s…) and word repetitions (the the the or no no no)”.
Edited transcribing is used in interviews that serve as a basis for articles or documentaries, in qualitative research where the focus is on a conversation’s content, in meetings that have to be published, in notes you record just for yourself, and when a transcript is translated for instance from German into English.
You can either transcribe yourself or outsource it. Because third parties are usually unfamiliar with your research, technical terms or product names may be transcribed incorrectly. Further, persons not present during a conversation sometimes don’t understand all the words correctly, especially if the recording’s sound quality is low.
While doing the transcribing yourself takes time, it also has the advantage that, every time you listen to the recording, you are subconsciously also doing much of your analysis and saving time, because you understand what the speakers mean.
To avoid wasting time, consider the following: