A well-founded system of rules for interview transcription in the context of qualitative research projects. Updated in November 2022 (Dr. Thorsten Dresing, Thorsten Pehl)

The transcription rules we have published so far have been widely used for many years. At the end of 2022, we felt the need to take another critical look at the rules. Even though textbooks contain extremely complex transcription systems, many projects use a rather simple transcription system. In the context of courses or qualification theses and in the first contact with qualitative research, the focus is often more on the semantic content and a simple and reliable applicability of the rules.

We have therefore significantly revised our rules once again and placed great emphasis on clarity and conciseness. Many formulations have been simplified and the explanations that were previously explained using examples have been formulated as rules. This makes the rules easy to learn, easy to communicate and therefore reliable to use.

In the following you will first find the new rules extracted directly (the old/previous rules can be found here on p. 20ff). For those interested in a little more context, here is a basic description of how transcription systems work in general.

—

Rule system for content semantic transcription (2023)

Rule1: Write down every word

– Even if the sentence sounds strange or wrong, write it down as it was spoken.
– Word slurring and dialect are brought closer to written German. If no clear translation is possible, they are retained.
– Incomprehensible words are marked with “(unv.)” or presumed wording “(luck?)”.

Rule 2: Write down short answers, ignore comprehension statements

– Short statements such as “hmm”, “yes”, “exactly” will be ignored if you do not interrupt the flow of speech.
– Short statements such as “hmm”, “yes”, “exactly” are written down if they represent an answer to a question.

Rule 3: Record clear pauses

– Clear pauses in speech of 3 seconds or more are marked by (…).

Rule 4: Note emotions and emphases

– Laughing, sighing or crying are noted in brackets.
– Particularly emphasized words are marked by VERSALS.

Rule 5: Design transcripts uniformly

– Each speech is given its own paragraph, beginning with a person-specific abbreviation with a colon (e.g. I:, B1:, Mr. Müller:).
– Short answers to questions (e.g. “yes” or “um”) are also transcribed in a separate paragraph.

 

 

 

Modules to expand the content-semantic transcription:

The following points are suggestions for possible additions to the rules. These can be selected completely or individually depending on the requirements of the research question.

1. Word and sentence breaks are marked with “/”: “But I thought about Sor/.”
2. Word duplications are always noted.
3. Breaks are marked by ellipsis in brackets depending on their length. Here, “(.)” stands for approximately one second, “(..)” for approximately two seconds, “(…)” for approximately three seconds and “(number)” for more than three seconds.
4. Reception signals and filler sounds of all persons (“hm, ja, aha, ähm” etc.) are transcribed. Exception: Backchanneling by the interviewer while another person is speaking is not transcribed as long as the flow of speech is not interrupted.
5. After the particle “hm”, a description of the stress is recorded in brackets. The following can be used: affirmative, negative, interrogative, e.g. “hm (affirmative)”.
6. Speaker overlaps are marked with “//”. A “//” follows the beginning of the throw-in. The text that is spoken at the same time is then within this “//” and the other person’s interjection is in a separate line and is also marked with “//”.

 

General notes on standardized spelling

Here are some recommendations that are not necessarily part of the transcription rules, but can be helpful in cases of doubt or when transcribing in a team:

1. Several filler sounds in a row are typed without punctuation in between (e.g. “um um um um so here we are …”).
2. The particles “hm” are always spelled “hm” regardless of the stress (not: “hhhhm”, “mhm”, “hmh”).
3. Hesitation sounds are always written “ähm” (not: “äm”, “ehm”, “öhm”).
4. (Measurement) units and symbols are written out in full, e.g. euro, meter, ät, paragraph.
5. Abbreviations are only typed if they are explicitly spoken (“etc.” is only typed when “e te ce” is spoken).
6. If verbatim speech is quoted in the recording, the quote is placed in quotation marks: “And then I said ‘Well, let’s see'”.
7. Word shortenings such as “runtergehen” instead of “heruntergehen” or “mal” instead of “einmal” are written exactly as they are spoken.
8. English terms are treated according to German spelling rules in upper and lower case.
9. Personal pronouns in the second person (du and ihr) are written in lower case, personal pronouns in the polite form (Sie and Ihnen) are written in upper case.
10. Phrases/idioms are also reproduced literally, e.g. “hauen übers Ohr hauen” (instead of “über das Ohr hauen”)

 

---

 

Background: What are transcription systems used for?

What is transcribing? Transcription (Latin transcribere “to transcribe”) means the transfer of an audio or video recording into a written form.

Typing is necessary because oral statements, even if they have been recorded, are fleeting and memories of a conversation are often incomplete. The content is recorded in writing. This fixed documentation is the basis of many qualitative analysis methods.

Why worry about the “how”?

Once you start typing up an interview, you will realize that there are many variations and possibilities for writing down what you have heard. It can be quite significant whether someone answers a question with “yes” or only says “yoooah” quietly, indistinctly and elongatedly after a pause of many seconds. Questions arise: Do I make a note of hesitation sounds like “um” or would I rather not? Do I write “hammermal” or would it be better to write “haben wir einmal”? How do I deal with dialect, pauses, speech overlaps and more?

A lot of methodological thought can and has been given to all of these upcoming decisions in the past. And it is certainly exciting to consider how and how detailed certain phenomena should be depicted and how to ensure that as few misinterpretations as possible are made during the analysis.[3] However, in research practice, these questions are usually not reinvented every time, because there are tried and tested transcription rule systems, i.e. collections of rules and transcription symbols, which you can easily fall back on.[4]

Transcription systems help

Simple transcripts mainly contain semantic information. You read a text that is slightly smoothed out in colloquial language. The focus is on good readability, easy learnability of the implementation rule and not too extensive implementation time. A detailed transcription system is necessary when the analysis simply has to provide more information because it is partly interpreted at word level. Then pitch progressions, secondary accents, volume, dialect and speaking speed are discussed, and phonetic transcription is used in some cases. Wonderfully precise in its presentation, unfortunately much more complex to produce. If a simple transcript takes about 5 to 10 times as long as the recording, you will need to spend a good 60 times as much time on such multifaceted detailed presentations. All transcription systems have one goal in common: the transcription is rule-based and therefore uniform and comprehensible.

How does that compare in concrete terms? Here are two excerpts from transcripts of the same conversation that were generated using different rules:

Example of a content-semantic transcript

S2: A particularly good example,
These used to be our neighbors.
(…), married for thirty years, (…)
the last child is finally out of the
House, to study, (…) gone.
went, no, to Berlin.

Example of a GAT transcript [5]

S2: n especially ↑`Good example that
warn times
our ↑`NIGHT neighbors.
(- – -)
um (- – -)
↑`Thirty years ́hMarried, °hh
the last child (.) `Finally off_m
́HAUS,
to the stu ́DER, (-)
́Gone, = ́ne, °h
to ber ́LIN, °h

The semantic-content transcript allows quicker access to the content of the conversation. It dispenses with precise details on pronunciation, making it easier to read. The GAT transcript, on the other hand, gives a better impression of intonation and speech rhythm by reproducing colloquial speech and pitch progressions.

In principle, there are many different systems in addition to these two control systems, some of which can also be used on a modular basis. We have described this in detail in the Handbook of Qualitative Research in Psychology (2020) elaborated on this. In this way, a suitable transcription system can be developed for specific research questions.

 

 

 

—

[1] f4transcript, f4analysis and f4x

[2] For example here: Praxisbuch Interview & Transkription and here: Handbook Qualitative Research

[3] This has been done in detail here: Transcription (Thorsten Dresing, Thorsten Pehl in Handbuch Qualitative Forschung in der Psychologie, 2020)

[4] A good overview of transcription systems is provided by
DITTMAR, Norbert (2004). Transcription. A guide with tasks for
Students, researchers and laypeople. Wiesbaden: VS, Verlag f. Sozialwiss.,
KUCKARTZ, Udo (2010). Introduction to the computer-aided analysis of qualitative data. 3rd ed. Wiesbaden: VS, Verlag f. Sozialwiss. and
DRESING, Thorsten & PEHL Thorsten (2017). Transcriptions of qualitative data. Implications, selection criteria and systems for psychological studies. In Mey, Günter & Mruck, Katja (eds.). Handbook of Qualitative Research in Psychology (S. 723-733). Wiesbaden: VS, Verlag f Sozialwiss.

[5] Excerpt from GAT-Feintranskript, http://www.mediensprache.net/de/medienanalyse/transcription/gat/gat.pdf, p. 35 [accessed: 21.11.22].