GÖTEBORG TRANSCRIPTION STANDARD

 

Department of Linguistics

Göteborg University

 

Version 6.3 - DRAFT

 

Joakim Nivre

Jens Allwood

Leif Grönqvist

Elisabeth Ahlsén

Magnus Gunnarsson

Johan Hagman

Staffan Larsson

Sylvana Sofkova

 


Table of Contents

 

1.     Introduction. 3

2.     Basic Format Requirements. 3

2.1        Global Structure. 3

2.2        Special Characters. 4

2.3        File Format 4

3.     Contributions. 4

3.1        Line Format 5

3.2        Words. 5

3.3        Prosody and Pauses. 7

3.4        Overlapping Contributions. 8

3.5        Comment Anchors. 10

3.6        Nonvocal Contributions. 11

4.     Comments. 11

4.1        Comment Lines. 11

4.2        Standard Comments. 12

4.3        Vocal Sounds. 13

4.4        Properties of Speech. 14

4.5        Special Expressions. 16

4.6        Clarifications. 17

4.7        Events and Moods. 19

4.8        Properties of the Recording. 20

4.9        Non-standard Comments. 21

5.     Sectioning. 21

6.     Time Coding. 22

7.     Other Information Lines. 22

8.     Transcription Header. 23

Appendix A -     Transcription Format Overview.. 27

Appendix B -     Standard Comments. 30

Appendix C -     Nonstandard Symbols. 31

 


1.              Introduction

 

This document describes the standard for machine-readable transcriptions of spoken language first used within the research program Semantics and Spoken Language at the Department of Linguistics, Göteborg University.[1] The standard is called the Göteborg Transcription Standard (GTS), and consists of two parts, one language independent part called GTSG (GTS General), and one language dependent part. At the Department of Linguistics in Göteborg this latter part has been what has become known as Modified Standard Orthography (MSO), but GTS does not require that. Within the limits of GTSG, any system can be used, for example the four systems described in 3.2 below. This document describes GTSG; MSO is described in Nivre, Joakim (1999) Modified Standard Orthography (MSO6). Department of Linguistics, Göteborg University.

 

The document is organized as follows. In section 2, we describe the basic format requirements of transcription files. Section 3 is devoted to the transcription of utterances, or contributions, while sections 47 deal with the coding of various types of background information through the use of so-called information lines. In section 8, finally, we give instructions for the composition of the transcription header.

 

 

2.              Basic Format Requirements

2.1             Global Structure

 

Every transcription is divided into two consecutive parts:

 

·      Header

·      Body

 

The header contains background information about the transcription and the recorded activity, such as date of recording, participants, etc. (cf. section 8 below). The body is the transcription proper, consisting of two kinds of elements:

 

·      Contributions

·      Information lines

 

Contributions are the transcribed utterances of participants in the recorded activity (cf. section 3 below). Information lines contain information related to contributions or to other aspects of the recorded activity and can be further subdivided into four kinds (described in sections 47 below):

 

·      Comment lines

·      Section lines

·      Time lines

·      Other information lines

 

 

2.2             Special Characters

 

Every line in a transcription file is either a contribution line (containing the transcribed speech of a participant) or an information line of some kind. The status of a line is indicated by its first character. The following four characters are used for this purpose:

 

·      The dollar sign ($) indicates that the line is a contribution line (cf. section 3). (A useful mnemonic is to think of this symbol as ‘S’ for Speaker.)

·      The at sign (@) indicates that the line is a header line, a comment line, or an other information line (cf. sections 4, 7 and 8).

·      The number sign (#) indicates that the line is a time line (cf. section 6).

·      The paragraph sign (§) indicates that the line is a section line (cf. section 5).

 

The use of these special characters is subject to the following conditions:

 

·      The special characters must always appear in the first position on a new line, separated from the previous material by a line feed or a so-called “hard line break” (obtained normally by pressing the return key). This means that the special characters can only be used in the special function here and never inside a contribution or information line.

·      Only the special characters may appear in the first position on a new line. This means that lines should never be broken, by means of a “hard line break”, inside a contribution or information line.

2.3             File Format

 

All transcriptions should be stored as text files. Such files can be produced by using a plain text editor (such as emacs or vi) or by using a regular word processor (such as Word or WordPerfect) and making sure that the final transcription is saved as plain text.

 

 

3.              Contributions

 

The basic unit of transcription is the contribution. A contribution can be defined as a continuous stretch of communicative activity from one participant, bounded either by inactivity or by communicative activity from another participant. A contribution may be expressed vocally, through bodily gestures, or both. In the following, we will begin by restricting our attention to the vocal aspects of contributions, which is what is primarily captured in the transcription. (Nonvocal aspects of contributions will be discussed in section 3.6 below.)

 

Restricting our attention to the vocal aspects of contributions, we may refine our earlier definition and say that a vocal contribution (or an utterance) is a sequence of words uttered by one participant, bounded either by silence or by the uninterrupted speech of another participant (or by the start/end of the activity).

 

Two things are worth noticing about this definition. First of all, a contribution always belong to a single participant. This means that even if two participants utter a phrase collectively (simultaneously or by completing one half each), this will count as two separate (individual) contributions. Secondly, the condition saying that a contribution may be bounded by silence is not meant to rule out the possibility of pauses within a contribution. In fact, the recommended practice in transcribing is to interpret instances of silence primarily as pauses and to consider a contribution as ended only when (i) another participant talks without overlap, or (ii) it is clear that the activity (or subactivity) is over. The rule of thumb is thus to make contributions as long as possible in the transcription.

 

 

3.1             Line Format

 

Every contribution in a recorded activity should be transcribed as a continuous unit, satisfying the following conditions:

 

·      A contribution is started by a speaker prompt and is ended by any of the four special characters: $, @, #, §.

·      A speaker prompt consists of a dollar sign ($), followed by a speaker initial, a colon and a space, e.g. ‘$A: ’.

·      A speaker initial consists of one uppercase letter (A-Ö), which is used consistently within a transcription to identify a participant as the speaker of the contribution. The special speaker initial X is used for contributions where the speaker can not be identified. If there are more than 28 participants in an activity, so that the whole alphabet of speaker initials is exhausted, speaker initials consisting of two uppercase letters (AA, AB, AC, etc.) can also be used. (All speaker initials used in a transcription must be “declared” in the transcription header; cf. section 8.)

 

To exemplify the format of contributions, consider the following transcription of three contributions, the first and third by speaker G, the second by speaker A:

 

$G: gula sidornas informationservice go morron

$A: ja hejsan ja skulle vilja hyra en bil skulle gärna vilja få reda lite om va man kan göra de

$G: jaa ä: vilken typ av fordon e du intresserad av

 

The contribution by speaker A in this example is too long to fit onto a single line in the printed version of the transcription, and the second line above has been indented to make the example more readable. In producing the actual transcription file, however, no such formatting should ever be used, and special care should be taken not to insert any “hard line breaks” inside contributions (cf. section 2.2). In other words, after a contribution has been started, no line breaks should be made before the end of the contribution, when a new line is started with another speaker initial or one of the other special characters indicating line status (cf. section 2.2 again). Note finally that a contribution must never be interrupted, or split into several parts, by material from another contribution or by an information line.

 

 

3.2             Words

 

When transcribing the words uttered in a contribution, different standards may be used with respect to the level of detail at which features of spoken language are rendered in the transcription. We distinguish four different standards:

 

·      Standard Orthography (SO): Standard orthography is used for all words, which means that no special features of spoken language are rendered in the transcription. However, we depart from standard orthography by not using capitalizion for proper names or abbreviations/acronyms and by not using any punctuation.

 

·      Modified Standard Orthography (MSO): Standard orthography is used except in cases where spoken language has variant pronunciations which are not recognized in standard orthograpy. For example, the singular first person pronoun in Swedish is written jag but may be pronounced either ja or jag. Similarly, the Swedish neuter adjective suffix -igt may be pronounced either -it or -igt.  Whenever the use of nonstandard orthography introduces an ambiguity which is not present in standard orthography, the forms are disambiguated, if possible by enclosing missing letters in curly brackets (ja{g} and -i{g}t), otherwise by means of numerical indices (thus, å0 and å1 correspond to standard orthography och and att, respectively).

 

·      Phonematic Transcription (PM): Phonematic transcription means using one symbol for each phoneme (and not distinguishing allophones), given a phonological analysis of the language in question and using a special (machine readable) phonetic symbol system (e.g. SAMPA).

 

·      Phonetic Transcription (PT): Phonetic transcription means taking into account features below the phonological level (allophones, coarticulation, etc.). Different levels of detail are possible, as well as different symbol sets.

 

Normally, a single standard should be used consistently throughout a transcription. (This standard should also be specified in the transcription header; cf. section 8.) The standard mostly used within the research program Semantics and Spoken Language and especially in the so-called core corpus is MSO for Swedish, which is further described in an appendix to this document. In our examples below, we will assume that this is the standard used.

 

When orthographic transcription (either standard or modified) is used, words should be transcribed using only lower case letters and no punctuation marks. Note in particular that numerals should always be transcribed with letters (one, two, three, etc.) and never with digits. Upper case letters may be used to indicate heavy stress (cf. section 3.3) but not for capitalization of e.g. proper names. In addition to the lower and upper case letters, the following special symbols may be used in the transcription of words:

 

·      The digits (09) and the asterisk (*) are used (in MSO) as indices to disambiguate spoken language forms that would otherwise introduce ambiguities that do not exist in standard orthography. (The asterisk is used instead of a numerical index when the transcriber is uncertain about the interpretation.) Note that the index must follow the word without space. Note also that digits may not be used to transcribe utterances of numerals.

·      Blank space ( ) is used to separate words.

·      Apostrophe (') may be used to indicate a glottal stop, e.g. ja'a.

·      Round brackets (()) are used to enclose any word or sequence of words about which the transcriber is uncertain concerning the correct transcription (normally for reasons of limited audibility), e.g. de va (den) de.

·      Three dots enclosed in round brackets ((...)) are used to indicate a passage that cannot be transcribed at all (normally for reasons of limited audibility).

·      Plus (+) is used at the beginning or end of a word to indicate that the word is incomplete. This is used (i) when a word is interrupted, e.g. transkri+, and (ii) when a speaker pauses in the middle of a word, e.g. transkri+ // +bera. Note that the plus must occur immediately before or after the word without space.

 

Here is a short excerpt from a transcription illustrating the use of these symbols:

 

$A: ja de{t} står de{t} förstås inte nånstans om ä:1 domm leve+ om dom levererar bilar vi{d} dörren (dom här) olika (...)

$G: de{t} får du nog ringa å0 höra ja'a

$A: då ska du ha tack fö{r} hjälpen

 

Finally, it is important to note that the transcription of a word must always be continuous, in the sense that no special symbols can be inserted in the middle of a word. This applies to all the special symbols listed above (which are all attached to the beginning or end of a word or occur as separate tokens) as well as to markers for pauses, overlap, and comments (cf. section 3.33.5).

 

For ethical reasons, all information that makes it possible to identify the participants in an activity should be excluded from the transcription. This means that all proper names should be replaced by pseudonyms, and that addresses, phone numbers, and other biographical information should be suppressed, e.g. by replacing them with dummy words such as gatan (the street) for any street name, etc.

 

 

3.3             Prosody and Pauses

 

The representation of prosody in transcriptions is limited to two aspects:

 

·      Uppercase letters are used to indicate emphatic or contrastive stress (but not ordinary word or sentence accent).

·      Colon (:) is used to indicate lengthening of continuant sounds (but not ordinary quantity, e.g. long vowels).

 

Here is a simple example illustrating the use of these symbols:

 

$A: ö:1 va{d} sa: du

$B: ja{g} sa att du skulle komma HIT

 

Note that the lengthening symbol precedes the numerical index at the end of a word.

 

Pauses are indicated by means of the slash symbol (/). One slash (/) signifies a short pause, double slash (//) an intermediate pause, and triple slash (///) a long pause. Here are some simple guidelines for deciding what counts as a short, intermediate and long pause, respectively:

 

·      A short pause (/) has a duration of the same order of magnitude as a word (given the current speech rate).

·      A long pause (///) has a duration of several seconds and is noticeable as a “gap” in the speech flow.

·      When in doubt, mark a pause as intermediate (//).

 

If the transcriber wants to give a more precise characterisation of pause length, a number signifying the duration in seconds can be added immediately after the pause symbol (i.e. /, //, or ///). However, this usage is only recommended for extremely long pauses (on the scale of thirty seconds or longer), e.g. ///34.5. Here is a short transcription illustrating the use of pause markers:

 

$G: ja: e:1 // ska vi se om vi kan hitta nånting som ligger / i de{t} området / de{t} bästa ja{g} kan hitta e0 nog på hyrgatan bilen biluthyrning aktiebolag /// telefonnumret dit e0 tretti noll noll sjuttifem

$A: tretti noll noll sjuttifem bilens biluthyrning

$G: aa

$A: // ja de{t} står de{t} förstås inte nånstans om ä:1 dom levererar bilar vi{d{ dörren dom här olika (...)

 

Note the occurrence of a pause at the beginning of the last contribution. In the current standard, silence can only be marked as a pause in a contribution. This means that noticeable silence between contributions must be marked either as a pause at the end of the preceding contribution or as a pause at the beginning of the succeeding contribution. In some cases, there may be clear indications that the silence belongs to one of the participants rather than to the other (this is the case in the example above). If this is not the case, then the silence should as a rule be marked as a pause at the end of the preceding contribution. (Note also in the example, how street names, company names, and telephone numbers have been altered to prevent identification of the persons or companies concerned.)

 

 

3.4             Overlapping Contributions

 

Contributions that occur sequentially should naturally be transcribed in the order in which they occur. However, sometimes one or more contributions overlap in time, wholly or partly. In this case, the following rules apply:

 

·      Each contribution must still be transcribed as an undivided whole.

·      The overlapping contributions are placed sequentially in the transcription according to their starting point in time.

·      Square brackets ([ ]) with matching numerical index are used to enclose the material in different contributions that overlap in time.

 

Here is a simple example where the beginning of a new contribution overlaps the end of the preceding contribution:

 

$A: // ja de{t} står de{t} förstås inte nånstans om ä:1 dom levererar bilar vi{d} dörren dom här olika företagen [4 utan man får ringa ]4

$G: [4 de{t} får ]4 du nog ringa å0 höra jaa

 

Note that the number used to index the brackets should be unique (within the transcription) for each overlap, and that the number must follow the bracket immediately without any intervening space, while the indexed bracket as a whole should be surrounded by spaces. Note also that the brackets should never be inserted in the middle of words. If an overlap starts or ends in the middle of a word, the whole word should be included in the overlap brackets, and the exact timing can be indicated by means of a comment (cf. section 4).

 

Several overlaps in one utterance

It frequently happens that several contributions from one participant are completely overlapped by a contribution from another participant. In this case, the overlapped contributions must still be transcribed as separate contributions (and placed after the longer, overlapping contribution). Thus, the following pattern is not permitted:

 

$A: ä+ en personbil räcker gott [2 de{t} e0 bara ja{g} ensam ]2 så de{t} kan va{ra} en ganska liten [3 bil ]3 faktis{k}t

$G: [2 en personbil ]2 [3 mm ]3

 

Instead, it should look as follows:

 

$A: ä+ en personbil räcker gott [2 de{t} e0 bara ja{g} ensam ]2 så de{t} kan va{ra} en ganska liten [3 bil ]3 faktis{k}t

$G: [2 en personbil ]2

$G: [3 mm ]3

 

 

Simultaneous starts

Another case that deserves mention is the case where two speakers in overlap start simultaneously. Unless this occurs at the very beginning of the transcription, the words uttered by the speaker who spoke in isolation before the overlap should always be treated as a continuation of his prior contribution. Thus, the following is not a permissible sequence of contributions:

 

$A: då i så fall vände{r} ja{g} mej nog hellre till firman  vi{d} centralen //

$G: [3 aa ]3

$A: [3 (har) du ]3 telefonnumret dit

 

Nor is the following:

 

$A: då i så fall vände{r} ja{g} mej nog hellre till firman vi{d} centralen //

$A: [3 (har) du ]3 telefonnumret dit

$G: [3 aa ]3

 

Instead, it should be:

 

$A: då i så fall vände{r} ja{g} mej nog hellre till firman  vi{d} centralen // [3 (har) du ]3 telefonnumret dit

$G: [3 aa ]3

 

Note also that this rule must be adhered to even if, in the above example, G’s aa should happen to begin a fraction of a second before A’s continuation, since it is not permitted to insert overlap brackets in the middle of words.

 

 

Nested overlaps

When more than two speakers are involved in overlapped speak, a common mistake is to nest the overlaps in different ways. This is an example of how it should not look:

 

$U: tänker du styr+ styra diskussionen lite grann så de{t} blir [1 ordning på dom för [2 här ]1 e0 ju ]2

$Z: [1 (...) (rutinerna [2 här) ]1 ]2

$A: [2 kan du {i}nte ]2 ta den där stolen den e0 mycke{t} ö{h} mycke{t} mer likadan som (...)

 

Here overlaps 1 and 2 are intermingled in an incorrect way. Instead U's and Z's utterances should be transcribed as two sequential overlap segments:

 

$U: tänker du styr+ styra diskussionen lite grann så de{t}

    blir [1 ordning på dom för ]1 [2 här e0 ju ]2

$Z: [1 (...) (rutinerna ]1 [2 här) ]2

$A: [2 kan du {i}nte ]2 ta den där stolen den e0 mycke{t}

    ö{h} mycke{t} mer likadan som (...)

 

Chines boxes

Another common mistake in the case of more than two overlapping contributions is the ‘Chinese boxes’, where one overlapped segment is contained within another overlapped segment:

 

$I: mhm m: (...) du har tittat förstås [6 på: [7 (sånt här) ]7 ]6

$A: [6 (...) ]6

$Z: [7 {j}a: ja{g} ]7 har lyssnat lite grann

 

This should be transcribed like this:

 

$I: mhm m: (...) du har tittat förstås [6 på: ]6 [7 (sånt här) ]7

$A: [6 (...) ]6 [7 (...) ]7

$Z: [7 {j}a: ja{g} ]7 har lyssnat lite grann

 

 

3.5             Comment Anchors

 

Any words occurring in a transcribed contribution should be words uttered by the speaker in question. This means that other information relevant to a contribution, such as information about bodily gestures performed by the speaker, about vocal events such as laughing and coughing, or about noisy events occurring in the background, etc., cannot be described directly in the transcribed contribution. For this purpose, the transcription standard allows comments to be inserted into the transcription body. These comments must be placed on separate lines (starting with the special symbol @) in order not to be confused with contributions, but they are connected to specific locations in the contributions through the insertion of comment anchors into the contribution.

 

A comment anchor consists of a pair of angle brackets (< >), which is used to enclose the portion of the contribution to which a particular comment relates. (The comment itself is enclosed in another pair of angle brackets on a following line; cf. section 4.) A simple example:

 

$A: < ja > de{t} stämmer men helst vill ja{g} ha / bilar som man kan man kan få framkörda till sin port och lämna där ute på flygplatsen

@ < ingressive >

 

Note that the angle brackets (like overlap brackets) should always be surrounded by spaces. Note also that angle brackets (like overlap brackets) must never be inserted in the middle of a word. If a comment concerns only part of a word, the comment anchor will still have to enclose the entire word.

 

When several comments are linked to the same contribution, it may be necessary to index comment anchors (and their corresponding comments) numerically (in the same way as overlap brackets). This must always be done when comment anchors are nested, as in the following example:

 

$G: den trakten / jaa / <1 just <2 >2 e0 du >1 skulle du vilja ha ett stort företag eller ett lite mindre elle de{t} kanske inte spelar dej någon större roll

@ <1 slow >1

@ <2 hesitation sound >2

 

Note that the numerical index must follow the bracket immediately without any intervening space (as with overlap brackets). This example also illustrates the use of an empty comment anchor, i.e. a pair of angle brackets that do not contain any word (used in this case to represent a hesitation sound that cannot be transcribed orthographically). It is important to note that when the two angle brackets are separated by a space (as in the example above), this means that the commented event has some duration in time. When an empty comment anchor is used to locate a non-extended event, such as the starting point of a longer event, it should be written without any space between the two angle brackets, thus:

 

$A: då e0 de{t} mest passande att <> få nån firma som kanske finns där

@ < event start: someone singing in the background >

 

Requirements on the actual comments (as opposed to the comment anchors) are presented in section 4 below.

 

 

 

 

3.6             Nonvocal Contributions

 

It is a well-known fact that bodily gestures can have the status of independent contributions in spoken communication. Thus, a head nod can often have the same function as an utterance of the word ja (yes) in Swedish. However, if we are working only with audio-recordings, there is no way in which this can be captured in the transcription. And even when we have access to video-recordings, it can sometimes be difficult to cover the nonvocal aspects of communication extensively. This is reflected in the transcription standard, which only allows the verbal aspects (words, basically) to be transcribed directly in the contribution lines, while nonverbal gestures have to be rendered indirectly through the use of comments. This opens up the question whether it is possible to have contributions without words in the transcription, i.e. contributions whose sole purpose is to act as anchoring points for comments describing a nonverbal gesture. Consider the following example:

 

$A: ä:1 ja{g} ska åka söder ä1 ifrån allså från södra delen av göteborg

$G: < >

@ <gesture: nods>

$A: då e0 de{t} mest passande att få nån firma som kanske finns där

 

In the latest version of the transcription standard, we allow this way of transcribing nonverbal contributions as an option. If the option is used, this must be specified in the transcription header (cf. section 8) since it influences the operation of certain programs for automatic analysis. If the option is not used, the foregoing passage has to be transcribed as follows:

 

$A: ä:1 ja{g} ska åka söder ä1 ifrån allså från södra delen av göteborg <//> då e0 de{t} mest passande att få nån firma som kanske finns där

@ <gesture: G nods>

 

 

4.              Comments

 

Comments are used in transcriptions to give information about events in the recorded activity that cannot be represented directly in the transcribed contribution. They may relate to aspects of the participants’ behaviour, such as prosody (other than emphatic stress and lengthening) and nonvocal behavior (gestures, movements, gaze direction, laughter, etc.). They may also be used to give information about extralinguistic events and background noises such as slamming doors, persons leaving or entering the room, etc. A comment is always linked to a specific location in the preceding contribution through a comment anchor (cf. section 3.5).

 

 

4.1             Comment Lines

 

Comments always appear on separate lines in the transcription. The following format requirements apply to comment lines:

 

·      A comment line is started by the special character @ and is ended by any of the four special characters: $, @, #, §.

 

 

 

 

 

·      A comment line contains one or more comments, enclosed in angle brackets (< >), separated by commas, and referring back to a unique comment anchor in the preceding contribution. For example:

$A: {j}a och ä:1 ja{g} vill ju naturlitvis ä:1 hitta nån
     firma som ligge{r} nära mej då < (...) > södra delarna
     av stan
@ < mumbling >

 

When overlapped utterances include comment anchors, the comment line should still be immediately after the utterance line:

 

$A: {j}a och < ä:1 > ja{g} vill ju [22 naturlitvis ]22

@ < event: looks at his watch >

$B: [22 men ska vi inte ]22 fortsätta på torsda{g}

 

The uniqueness condition on comment anchors has three important corollaries:

 

·      A comment line may contain more than one comment only if the comments refer to the same comment anchor. For example:

$G: < dom ligger på teatergatan >
@ < event: music playing on the radio >, < loud >

·      If a contribution contains several (non-nested) comment anchors, the corresponding comments must occur on separate comment lines occurring in the same order as the anchors. For example:

$A: go midda / ä1 de{t} e0 så att jag å0 en go vän < sk+ > tänkte gå ut på en restaurang / och vi skulle vilja gå < > / till nån som säljer < li+ > ä1 som har lite // < > jaa lite kryddigare mat lite // (gärna nåt) exotiskt

@ < cutoff: ska >

@ < clear throat >

@ < cutoff: lite >

@ < click >

 

·      If a contribution contains nested comment anchors, which are therefore indexed numerically, the corresponding comment must be indexed accordingly. For example:

$G: den trakten / jaa / <1 just <2 >2 e0 du >1 skulle du vilja ha ett stort företag eller ett lite mindre elle{r} de{t} kanske inte spelar dej någon större [2 roll ]2

@ <1 slow >1

@ <2 hesitation sound >2

 

·        It has happened that transcribers have broken a single comment into two lines in the following, incorrect way:

 

$A: ja < // >

@ < event: a power drill or some kind or strong electric engine starts, and

@  A looks really surprised and scared. >

 

This is not accepted, but the comment should be kept on a single line.

 

4.2             Standard Comments

 

Some types of comments have been found to be frequently recurring and an attempt to standardize these comments has been made. The standardized comments are at present mainly concerned with sounds or properties of speech, rather than e.g. gestures. The comments can be divided into seven categories:

 

·      Vocal sounds

·      Properties of speech

·      Special expressions

·      Clarifications

·      Events and moods

·      Properties of the recording

·      Non-standard comments

 

It is important to note that standard comments should always be used if possible, which means that the last category, non-standard comments, is relevant only when none of the comments in the preceding six categories is considered appropriate. The rules of usage for the six categories of standard comments are described in section 4.34.8 below.

 

The following notational conventions are used in the description of standard comments:

 

·      Regular font is used for that part of a comment which should be reproduced verbatim, while italics is used to indicate the variable part of a comment. Thus, in the comment < mood: description > the word description is a place-holder which should be replaced by a description of the speaker’s mood. For example:

$A: i+ inte en orange skruv möjli{g}tvis
$B: / < ha{r} ja{g} sagt fel nu > / du ha{r} använt två gula
@ < mood: uncertain >

·      Curly braces ({ }) are used to enclose an optional component of a comment. Thus, the comment < loan language{: word} > may take either of the following two forms: < loan language >, < loan language: word >.

·      Slash (/) is used to separate alternative components of a comment. Thus, the comment < event{ continued/start/stop}: description > may take either of the following four forms: < event continued: description >, < event start: description >, < event stop: description >, < event: description >.

 

 

4.3             Vocal Sounds

 

The comments concerning vocal sounds can be divided into two groups according to whether they are produced by the speaker of the current contribution or by one or more of the other participants. In the former case, the identity of the speaker is left unspecified in the comment, while in the latter case the comment has to include the speaker initial(s) of the person(s) producing the sound (preceded by a colon and separated by commas). These two usages are illustrated in the following example:

 

$A: hur < > e0 läget

@ < laughter >

$B: bra < hur > e0 de{t} själv

@ < laughter: A, C >

 

Note also that when the sound is produced by the current speaker, the comment anchor should always be empty (except for a space indicating the passing of time). (If the sound accompanies the speech, it falls under the category of properties of speech; cf. section 4.4.)

 

The following standard comments belong to the category of speech sounds:

 

@ < hesitation sound{: participant} >
This comment is used only when the hesitation sound cannot be related to any orthographically reproducible form, e.g. ä:.

 

@  < inhalation sound{: participant} >
The speaker or some other participant inhales audibly in order to indicate a turn-taking. Inhalations within contributions need not be commented. Note that the comment < ingressive > (cf. section 4.4) should be used whenever some other sound is uttered while the speaker inhales, and also for the bilabial ingressive sound in Swedish.

 

@  < laughter{: participant} >
The speaker or some other participant laughs.

 

@  < chuckle{: participant} >
The speaker or some other participant chuckles (“småskrattar” in Swedish).

 

@  < giggle{: participant} >
The speaker or some other participant giggles (“fnittrar” in Swedish).

 

@  < sigh{: participant} >
The speaker or some other participant sighs audibly (“suckar” in Swedish).

 

@  < puff{: participant} >
The speaker or some other participan makes an audible puff (“pustar” in Swedish).

 

@  < click{: participant} >
The speaker or some other participant makes a clicking sound (“smackar” in Swedish).

 

@  < clear throat{: participant} >
The speaker or some other participant clears his throat (“harklar sig” in Swedish).

 

@  < cough{: participant} >
The speaker or some other participant coughs.

 

@  < sneeze{: participant} >
The speaker or some other participant sneezes.

 

@  < yawn{: participant} >
The speaker or some other participant yawns.

 

 

4.4             Properties of Speech

 

In contrast to the category of vocal sounds, comments concerning properties of speech always refer to the current speaker. Furthermore, since they are used to describe properties of speech, their anchors normally cannot be empty. The only exception to this rule is the comment < ingressive > (see below). The following standard comments are used to describe properties of speech:

 

@  < ingressive >
The speaker inhales when producing the speech. This comment may also be used to code the bilabial ingressive, primarily used for positive feedback in Swedish. In this case, since the sound has no standard written form, the comment anchor is filled with a slash indicating the passing of time, as in the following example:

$A: ska du me{d} på bio
$B: < / >
@ < ingressive >

 

@  < laughing >
The speaker laughs when producing the speech.

 

@  < chuckling >
The speaker chuckles when producing the speech.

 

@  < giggling >
The speaker giggles when producing the speech.

 

@  < sighing >
The speaker sighs audibly when producing the speech.

 

@  < puffing >
The speaker makes an audible puff when producing the speech.

 

@  < coughing >
The speaker coughs when producing the speech.

 

@  < yawning >
The speaker yawns when producing the speech.

 

@  < high pitch >
The speaker’s pitch (fundamental frequency) is high.

 

@  < low pitch >
The speaker’s pitch (fundamental frequency) is low.

 

@  < quick >
The speaker talks quickly.

 

@  < slow >
The speaker talks slowly.

 

@  < loud >
The speaker talks loudly.

 

@  < quiet >
The speaker talks quietly.

 

@  < shouting >
The speaker is shouting when producing the speech.

 

@  < whispering >
The speaker is whispering when producing the speech.

 

@  < mumbling >
The speaker is mumbling when producing the speech.

 

@  < singing >
The speaker is singing when producing the speech.

 

 

4.5             Special Expressions

 

The category of special expressions contains, on the one hand, comments relating word forms in a contribution to their standard orthography or to their actual pronunciation and, on the other hand, comments indicating the special status of word forms as names, abbreviations, etc. The following standard comments belong to this category:

 

@  < SO: expression >
This comment may be used to resolve ambiguity in cases where no index is assigned to an ambiguous word form according to the standard of Modified Standard Orthography (MSO) and where curly braces cannot be used. (The ambiguous word form may later be included and indexed in a new edition of the standard). For example:

$A: ni skulle stått precis < ved > målet
@  < SO: vid >
$B: de{t} gjorde vi också


This comment may also be used for further explanation of a speaker’s contribution, especially if a part of a word or phrase is partially or completely reduced.
For example:

$A: by på landet // < säjer om > de
@  < SO: vad säjer du om >

 

@ < pronunciation: word >
According to a given MSO standard, some words may be written using standard orthography even though they are pronounced in a way that might more accurately be rendered by a non-standard orthography because the pronunciation reflected by the standard orhography is (almost) never used. For example, the Swedish word “projekt” is (almost) without exception pronounced [prUSekt] (i.e. with an unvoiced palato-velar fricative) not [prUjekt] (i.e. with a voiced palatal glide or fricative).[2] Therefore, the standard orthography projekt is used in MSO transcriptions even though the pronunciation is normally [prUSekt]. In case some speaker uses the non-standard pronunciation [prUjekt], this is indicated by a comment in the following way:

$A: ja{g} jobba{de} på ett < projekt >
@  < pronounciation: pro/j/ekt >

Note the use of slashes (
/ /) to delimit the part of the word that is pronounced ”literally”.

 

@  < loan language{: word} >
When words are borrowed from other languages, the orthography can  often be preserved without too much difficulty. For example, the English verb “switch” can be used in Swedish with the inflection paradigm “switcha, switchade, switchat” and the spelling comes naturally. In this case, the simple comment  < loan language > may be used. Sometimes, however, the spelling is not so easily converted. For example, the English verb “trade” might be inflected and spelt “trada, tradade, tradat” but a more satisfactory transcription may be achieved by ”swedifying” the spelling, producing “träjda, träjdade, träjdat” (since “trada” may be misunderstood to be a Swedish word).


If this is done, the comment  < loan language: word > may be used to indicate which word from which language is intended.
Here is an example illustrating the use of this comment:

$A: dom måste få en möjlighet att få sina aktier // ä:1
    < träjdade > (menar) varför < träjdar > man en aktie
    utanför sitt land
@  < loan English: trade>
@  < loan English: trade>

 

@  < name >
The word referred to is a proper name. (If the name is also an acronym, the < acronym > comment should be used instead.) Example:

$A: hon som hette // < inga > hette hon
@  < name >

 

@  < abbreviation >
The word referred to is an abbreviation and is pronounced letter by letter. Example:

$A: ja{g} ska åka till < usa >
@  < abbreviation >

 

@  < acronym >
The word referred to is an acronym, i.e. an abbreviation consisting of initials but pronounced as an ordinary word. (Most acronyms are also names, so the < name > comment need not be used.) Example:

$A: undersökningen gjordes av < sifo >
@  < acronym >

 

@  < letter >
The speech referred to is a letter that is pronounced in its “alphabetic” form. For example, b is pronounced “be”:

$A: ja{g} tvätta{de} den på < b >
@  < letter >

 

 

In case only parts of a word is a letter, that part is given in the comment:

 

       $A: ja{g} tvätta{de} den på < b-hallen >

       @ < letter: b >

 

@  < onomatopoetic >
The speaker imitates some sound, and the “word” used is not part of standard orthography (e.g. “plask” is onomatopoetic but part of standard orthography in Swedish, and need not be commented). Example:

$A: å0 sen kom planet ba{r}a < foom > förbi
@  < onomatopoetic >

 

 

4.6             Clarifications

 

The following standard comments are used for clarifications:

 

@  < cutoff: word >
If a word is cut off in the middle, this comment may be used to indicate which word it (probably) was. For example:

$A: om man tittar på på till exempel < kä+ >
    fusionskraften
@  < cutoff: kärnkraften >

Please note that the symbol
+ should still be used in the transcription to indicate that the word is incomplete.

 

@  < overlap: speech[speech >

@  < overlap: speech]speech >
If an overlap starts or ends in the middle of a word, the standard requires the entire word to be enclosed in the overlap brackets. If the transcriber wants to indicate the exact point within a word where an overlap starts or ends, this comment may be used. For example:

$A: men hur va{r} de{t} med andra < [1 världskriget ]1 >
@  < overlap: världskr[iget >
$B: [1 ja{g} minns ]1 att

 

@  < unclear: {speech}(...){speech} >

@  < unclear: {speech}(speech){speech} >
This comment may be used to specify a clearly audible part of an otherwise inaudible or uncertain word or phrase. Note that the standard requires the entire word or phrase to be marked as inaudible or uncertain in the transcribed contribution. For example:

$A: företaget gjorde vissa < (...) > investeringar
@  < unclear: kapital(...) >

 

@  < incomprehensible >
If some word or phrase is semantically incomprehensible (i.e. cannot be related to any known word or phrase) and yet can be transcribed without problem, this comment may be used. For example:

$A: nänä okej men låt oss nu ponera då att man kan
    < (sebado) > att man me{d} all säkerhet kan bevisa
    sambandet // så spelar de{t} ingen roll
@  < incomprehensible >

 

@  < alternatively: expression1, expression2, ... >
If some word or phrase can be interpreted in several different ways, and the context is not sufficient to decide between them, the most likely word should be written in the contribution and the other possible candidates be given in a comment of this kind. For example:

$C: de{t} e0 allså mönster då som < (han) > återupprepar
@  < alternatively: man >

 

@  < other language{ continued/start/stop}: language  >
If the speaker of a contribution switches to a language different from the main language of the dialogue, this comment is used. For example:

$A: finns det inte mer kaffe då
$B: nä
$A: < could we have a nice cup of tea please >
@  < other language: English >

If the language switch stretches over more than one contribution, the comment may be duplicated and attached to each contribution concerned. In this case, the option continued should be used from the second contribution onwards.

Another way of dealing with long stretches of speech in another language is to put markers at the beginning and end of the passage, using the
start and stop options. Note that the comment anchors should then be empty (with no space), as in the following example:

$C: // va{d} sa du nu <> a nice cup of tea
@  < other language start: English >
$A: yes
$C: good <> // då går vi vidare
@  < other language stop: English >

 

@  < parallel interaction integer  >
If adjacent contributions belong to different dialogues that are going on in parallel, this can be clarified by using numerical indices to refer to one or more of the parallel dialogues. For example:

$A: va{d} e0 klockan
$B: < kommer du >
@  < parallel interaction 1 >
$C: kvart över tre
$D: < ja >
@  < parallel interaction 1 >

If a recorded activity contains parallel interactions, this should also be indicated by a comment in the header.

 

 

4.7             Events and Moods

 

The category of events and moods includes three types of comment:

 

@  < event{ continued/start/stop}: description >
This comment may be used to describe any event that affects the conversation. For example:
$K: < (...) du får vinka (...) >
@  < event: K talks to a dog and wiggles its paw >

If the event stretches over more than one contribution, the comment may be duplicated and attached to each contribution concerned. In this case, the option continued should be used from the second contribution onwards. For example:

$C: // va{d} sa du nu < ett år i gymnasiet >
@  < event: C writes on a paper >
$A: < // a0 >
@  < event continued: C writes on a paper >
$C: < bra // > då går vi vidare
@  < event continued: C writes on a paper >

Another way of dealing with events with long duration is putting markers at the beginning and end of the event, using the
start and stop options. Note that the comment anchors should then be empty (with no space), as in the following example:

$C: // va{d} sa du nu <> ett år i gymnasiet
@  < event start: C writes on a paper >
$A: // a0
$C: bra // <> då går vi vidare
@  < event stop: C writes on a paper >

 

@  < gesture{ continued/start/stop}: {participant} {descr} >
If the speaker gestures (in a wide sense, including all kinds of nonverbal behavior) this comment may be used. If some participant other than the speaker gestures, the participant option should be used. (The continued, start and stop options functions analogously with the same options for the event comment.) The description option is used only if the gesture can be described in a clear and concise manner. Finally, if several gestures occur simultaneously, several comments should be used (separated by commas, see above). Here is an example illustrating the use of this comment:

$A: de{t} gör du < // > va{d} e0 de{t} för sorts saker
     som du tycker e0 finast i naturen < i så fall >
@  < gesture: nods >
@  < gesture: B nods >

 

@  < mood: description >
If the speaker expresses a certain mood, this comment may be used. The description should be as short as possible, e.g. angry, surprised, excited, bored, sad, happy, etc. For example:

$A: // nämen < ska vi inte öppna de{t} då eller >
@  < mood: surprised >

4.8             Properties of the Recording

 

The following comments may be used to describe properties of the recording itself (as opposed to the activity recorded):

 

@  < End of tape{ side side}. Continued on {tape} {side side} >
If the recording is spread over more than one tape or over two sides of one tape, this comment should be used to indicate where the tape (or side) was shifted. For example:

$S: // ah de{t} blir längre bort // ä1 de{t} de{t} blir
     ännu <> längre bort efter den ä1 kanalen där
@  < End of A3602 side A. Continued on side B >

 

@  < damaged: description >
If the tape used in the transcription is damaged so that some speech is inaudible, unclear or uncertain, this may be noted using this comment. For example:

$A: du kan <(...)> eller
@  < damaged: KAV0215 blanked >

 

@  < not transcribed{:  description} >
If, for some reason, a part of the recorded activity is not transcribed, this comment may be used. A short description of the skipped part and/or an explanation as to why this part was not transcribed may also be included. This comment should be used very restrictively. For example:

$S: // just de{t} < >
@  < not transcribed: the name of the informant >

 

 

4.9             Non-standard Comments

 

This type of comment should be used only when no standard comment can be used. The required format is the following:

 

@  < comment: description >

5.              Sectioning

 

The transcription is divided into sections by means of section lines. The following format requirements apply to section lines:

 

·      A section line is started by the special character § and is ended by any of the four special characters: $, @, #, §.

·      The transcription body ends with a special section line: § END.

·      Section lines other than the special END line are optional and simultaneously mark the beginning of a new section and the end of the preceding section, which implies that the whole transcription is divided into a series of continuous sections where every contribution, comment, time line and other information line must belong to a section. The first section line must therefore be placed immediately after the heading. If sectioning is not used at all, the header should instead be followed by a dummy section line: § START.

·      Whatever follows the special character § (and a space) in an ordinary section line (i.e. other than the special START and END lines) is interpreted as the name of the section started by that line.

·      Each section name must be unique within the transcription.

 

The criteria that are normally used for sectioning a transcription are the following:

 

·      Change of activity

·      Change of topic

 

Although these changes may occur simultaneously and it may be hard to determine whether a changing event is of one or the other type, we suggest to first recognize all the activity shifts in the material and base the first sectioning on that. Some typical subactivities (in two different activities) are:

 

§ START

§ Introduction

§ Member presentations

§ Division into groups

§ Coffee break

...

§ END

 

§ START

§ In the kitchen

§ The guests are coming

§ Eating

...

§ END

 

If further sectioning is called for, a closer look is taken at the moments where the topic is changed within subactivities. Typical topic shifts may look as follows. Under Introduction above:

 

§ Short history

§ Last meeting

§ Announcements

 

Under Eating :

 

§ Indian dishes

§ Harry's gravy spot

§ C&Z's wedding

 

Note that the hierarchical nature of such subsectioning is only implicit in the transcription. If there is a need to make the hierarchical structure explicit, this can be expressed in the section names. For example:

 

§ 1.  Introduction

§ 1.1 Short history

§ 1.2 Last meeting

§ 1.3 Announcements

 

Note finally that all section names occurring in a transcription must be listed in the transcription header (cf. section 8).

 

 

6.              Time Coding

 

Time lines can be used anywhere in a transcription to give information about the amount of time elapsed from the start of the recorded activity. Usually, it is a good idea to place the time lines (if they are used at all) in connection with section boundaries. The following format requirements apply to time lines:

 

·      A time line is started by the special character # and is ended by any of the four special characters: $, @, #, §.

·      The time is given in the format HH:MM:SS (Hours:Minutes:Seconds), e.g. 00:12:42.

 

Here is an example of the use of time coding:

 

$C: // va{d} sa du nu // tre år i gymnasiet

$A: // {j}a

$C: bra // då går vi vidare

§ Work life experience

# 00:04:34

$C: va1 har du jobbat me{d} för nåt då

 

7.              Other Information Lines

 

Most information lines are either comment lines (referring to specific positions in the preceding contribution), section lines (indicating the sectional structure of the transcribed activity), or time lines (indicating the amount of time elapsed from the start of the activity). However, it may sometimes be useful to insert background information without connecting it to a specific position in a contribution (either because the information is not tied to any particular point in time in the activity, or because the position of the information line itself is precise enough to indicate the point in time).

 

Such information lines may be used, for example, to give information about a change of transcriber (which may or may not coincide with section borders), which is exemplified below:

 

$A: nej vi kan ju inte bara prata strunt

$B: nej just de{t}

@ Johan Johansson started transcribing here

$A: ja{g} vill att du ska beskriva dom bilder ja{g} visar dej

$B: okej

$A: kan vi ta om de{t} igen

 

The only requirement on these information lines is that they begin with the special character @ (and that they do not split contributions or separate comments from the contributions they refer to).

 

 

8.              Transcription Header

 

The basic requirement on the header format is that every line begins with the at character (@), which identifies the line as an information line. Each line in the header should furthermore begin with a heading, indicating what kind of information the line contains. Below we list the different headings that may occur in a header, specifying for each heading:

 

·      whether it is required (R) or optional (O),

·      whether it can have multiple occurrences in the same header (M),

·      whether it should be automatically generated (A)

·      what kind of information it introduces and in what format.

 

The following headings are permitted in a header:

 

@ Recorded activity ID: (R) The unique identifier of the transcribed acitivity, e.g. A545301 or V343402. See the document File Naming Conventions for a description of this identifier..

 

@ Recorded activity title: (R) The title of the activity (free text).

 

@ Short name: (R) The short name of the activity (maximum 10 characters).

 

@ Recorded activity date: (R) The date of the recording in the format yyyymmdd, e.g. 19921117.

 

@ Anonymized: (R) Indicates whether personal names, etc. have been changed to pseudonyms (Yes) or not (No) (See paragraph 3.2).

 

@  Access: (R) Indicates how restricted access should be to the transcription. One of Restricted, Department, Research and Public, where Restricted means very limited access and Public means ulimited access. Default is Research.

 

@ Activity Type, level 1: (R)  Each transcription is classified according to activity type, following the taxonomy described in ???. This header line specifies the activity type on the top level.

 

@ Activity Type, level 2: (O)  Each transcription is classified according to activity type, following the taxonomy described in ???. This header line specifies the activity type on the second level.

 

@ Activity Type, level 3: (O)  Each transcription is classified according to activity type, following the taxonomy described in ???. This header line specifies the activity type on the third level.

 

@ Aktivitetstyp, nivå 1: (R)  Like Activity Type, level 1, but in Swedish.

 

@ Aktivitetstyp, nivå 2: (O)  Like Activity Type, level 2, but in Swedish.

 

@ Aktivitetstyp, nivå 3: (O)  Like Activity Type, level 3, but in Swedish.

 

@ Activity Artefacts: (O)  Any artefacts important or characteristic for the activity. See t.ex. Allwood 2000[3].

 

@ Activity Medium: (R)  Normally Spoken, face-to-face, but could also have othe rvalues, like Spoken, telephone.

 

@ Duration: (O) The duration of the activity in the format HH:MM:SS
(Hours:Minutes:Seconds), e.g.
00:34:30.

 

@ Participant: (R, M) The speaker initial and ID-number of a participant in the transcribed activity, separated by the equality sign (=), with the speaker’s pseudonym in parentheses after the ID-number, e.g.

A = M52 (Adam). The speaker initial is used in the body of the transcription to tag utterances by the participant in question. The ID-number consists of an M or an F, indicating whether the speaker is Male (M) or Female (F), directly followed by a unique number that has to be provided to the transcriber. If the transcriber has no information about what ID-number to use, the M/F and number should be temporarily replaced by the pseudonym, which should be unique within the transcription, e.g. A = Adam. This pseudonym should later be replaced by an ID-number which is unique within the entire corpus, so that the same ID-number can be used for the same person (and only that person) across different activities. The pseudonym should be used if the speaker is spoken to, or mentioned in the transcription. Note finally that each participant should be specified on a separate line. Thus:

@ Participant: A = M52 (Adam)
@ Participant: B = F36 (Beda)

 

 

 

@ Tape(s): (R) The name of the tape on which the activity is recorded. If the activity stretches over more than one tape, all tape names are given on the same line, separated by commas. For example:

@ Tape(s): A5461, A5462, A5463.

NB: The bracketed plural ending (s) is used regularly to indicate that a heading may be followed by a list of items, separated by commas.

 

@ Recorder(s): (R) The name of the person(s) who did the recording, like Bo Berg.

 

@ Transcription name: (R) The ID of the transcription, like A792701. See the document File Naming Conventions for a detailed description.

 

@ Transcriber(s): (R) The name of the transcriber(s); list field.

 

@ Transcriber ID(s): (R) The ID of the transcriber(s); list field.

 

@ Transcription date(s): (R) The date of the transcription(s); list field.
NB: The number of dates listed in this field should match the number of names listed under
@ Transcriber(s) (i.e. one date per transcriber).

 

@ Transcribed segments: (R) Information about what parts of the recorded activity have been transcribed; either the word All (if all of the activity has been transcribed) or a free text description, e.g. First three minutes, 000-678 on side A of the tape, etc.

 

@ Transcription system: (R) The transcription system used, normally GTS+MSO plus transcription standard version number, thus currently GTS6. NB: If nonvocal contributions are allowed in the transcription (cf. section 3.6), the suffix +NVC should be added to the transcription system, e.g. GTS6+MSO+NVC.

 

@ Checker(s): (R) Name of the person(s) checking the transcription; list field.

 

@ Checking date(s): (R) The date of the checking(s); list field. NB: The number of dates listed in this field should match the number of names listed under @ Checker(s) (i.e. one date per checker).

 

 

 

@ Time coding: (A) Indicates whether the transcription contains time coding (Yes) or not (No) (cf. section 6).

 

@ Section: (A, M) The title of a section of the transcription. Note that the titles of all sections occurring in a transcription should be given on consecutive lines in the header (cf. section 5). For example:

@ Section: Opening
@ Section: Request for car hire firms
@ Section: Closing

 

@ Comment: (O, M) A line beginning with this heading can be used to include into the header any relevant information that is not covered by any of the required headings.

 

Finally, let us consider a sample header, containing all the required headings:

 

@ Recorded activity ID: A545301

@ Recorded activity title: Telephone dialogue, car hire, 2a

@ Short name: TelCar2a

@ Recorded activity date: 19921117

@ Anonymized: Yes

@ Access: Research

@ Activity Type, level 1: Interview

@ Activity Type, level 2: Organized

@ Aktivitetstyp, nivå 1: Intervju

@ Aktivitetstyp, nivå 2: Organiserad

@ Activity Medium: Spoken, telephone

@ Duration: 00:15:12

@ Participant: H = F50 (Harriet)

@ Participant: S = M1003 (Student2)

@ Tape(s): A5453

@ Recorder(s): Tomas Tomasson

@ Transcription name: A5453011

@ Transcriber(s): Lisa Larsson

@ Transcriber ID(s): 100032

@ Transcription date(s): 19930315

@ Transcribed segments: All

@ Transcription system: GTS6+MSO

@ Checker(s): Johan Johansson

@ Checking date(s): 19930417

@ Time coding: No

@ Section: Opening

@ Section: Request for car hire firms

@ Section: Closing

@ Comment: A noise on the telephone line makes it difficult to

hear what S says.

 

 

 


Appendix A -  Transcription Format Overview

 

HEADER

 

@ Recorded activity ID: A545301

@ Recorded activity title: Telephone dialogue, car hire, 2a

@ Short name: TelCar2a

@ Recorded activity date: 19921117

@ Anonymized: Yes

@ Access: Research

@ Activity Type, level 1: Interview

@ Activity Type, level 2: Organized

@ Aktivitetstyp, nivå 1: Intervju

@ Aktivitetstyp, nivå 2: Organiserad

@ Activity Medium: Spoken, telephone

@ Duration: 00:15:12

@ Participant: H = F50 (Harriet)

@ Participant: S = M1003 (Student2)

@ Tape(s): A5453

@ Recorder(s): Tomas Tomasson

@ Transcription name: A5453011

@ Transcriber(s): Lisa Larsson

@ Transcriber ID(s): 100032

@ Transcription date(s): 19930315

@ Transcribed segments: All

@ Transcription system: GTS6+MSO

@ Checker(s): Johan Johansson

@ Checking date(s): 19930417

@ Time coding: No

@ Section: Opening

@ Section: Request for car hire firms

@ Section: Closing

@ Comment: A noise on the telephone line makes it difficult to

hear what S says.

 

BODY

 

Contributions

 

·      A contribution is started by a speaker prompt and is ended by any of the four special characters: $, @, #, §.

·      A speaker prompt consists of a dollar sign ($), followed by a speaker initial, a colon and a space, e.g. ‘$A: ’.

 

Besides lowercase letters, the following symbols may be used in transcribing words in contributions:

 

·      The digits (09) and the asterisk (*) are used (in MSO) as indices to disambiguate spoken language forms that would otherwise introduce ambiguities that do not exist in standard orthography. (The asterisk is used instead of a numerical index when the transcriber is uncertain about the interpretation.) Note that the index must follow the word without space.

·      Blank space ( ) is used to separate words.

·      Apostrophe (') may be used to indicate a glottal stop, e.g. ja'a.

·      Round brackets (()) are used to enclose any word or sequence of words about which the transcriber is uncertain concerning the correct transcription (normally for reasons of limited audibility), e.g. de va (den) de.

·      Three dots enclosed in round brackets ((...)) are used to indicate a passage that cannot be transcribed at all (normally for reasons of limited audibility).

·      Plus (+) is used at the beginning or end of a word to indicate that the word is incomplete. This is used (i) when a word is interrupted, e.g. transkri+, and (ii) when a speaker pauses in the middle of a word, e.g. transkri+ // +bera. Note that the plus must occur immediately before or after the word without space.

 

The following conventions are used for prosody:

 

·      Uppercase letters are used to indicate emphatic or contrastive stress (but not ordinary word or sentence accent).

·      Colon (:) is used to indicate lengthening of continuant sounds (but not ordinary quantity, e.g. long vowels).

 

Pauses are indicated by means of one, two or three occurrences of the slash symbol (/), corresponding to short, intermediate and long pauses, respectively. The following guidelines may be used:

 

·      A short pause (/) has a duration of the same order of magnitude as a word (given the current speech rate).

·      A long pause (///) has a duration of several seconds and is noticeable as a “gap” in the speech flow.

·      When in doubt, mark a pause as intermediate (//).

 

Overlap is marked in the following way:

 

·      Square brackets ([]) with matching numerical index are used to enclose the material in different contributions that overlap in time.

 

And comments are anchored in contributions as follows:

 

·      Angle brackets (< >) are used to enclose the material in a contribution to which a comment refers.

 

 

Comments  

 

·      A comment line is started by the special character @ and is ended by any of the four special characters: $, @, #, §.

·      A comment line contains one or more comments, enclosed in angle brackets (< >), separated by commas, and referring back to a unique comment anchor in the preceding contribution.

 

 

Section lines    

 

·      A section line is started by the special character § and is ended by any of the four special characters: $, @, #, §.

·      The transcription body begins with a special section line: § START.

·      The transcription body ends with a special section line: § END.

·      Section lines other than the special START and END lines are optional and simultaneously mark the beginning of a new section and the end of the preceding section, which implies that the whole transcription — if sectioning is used at all — must be divided into a series of continuous sections where every contribution, comment, time line and other information line must belong to a section.

·      Whatever follows the special character § (and a space) in an ordinary section line (i.e. other than the special START and END lines) is interpreted as the name of the section started by that line.

·      Each section name must be unique within the transcription.

 

 

Time lines  

 

·      A time line is started by the special character # and is ended by any of the four special characters: $, @, #, §.

·      The time is given in the format HH:MM:SS (Hours:Minutes:Seconds), e.g. 00:12:42.

 

 

Other information lines      

 

·      An other information line is started by the special character @ and is ended by any of the four special characters: $, @, #, §.

 

 

EXAMPLE

 

§  Introduction

#  00:00:00

$P: ha1 de0 lugnat ner sej

$D: de0 här kommer du nog att få ha0 kvar därför
    [1 att (...) ]1

$P: [1 men doktorn ]1 e0 de0 inte inne i benet ja1 tycke de0
    känns så /// och här på [2 sidan ]2

$D: [2 ja0 ]2 böj lite (...) < >

@ < event: examination >

 


Appendix B -   Standard Comments

 

SOUNDS

 

@ < hesitation sound{: participant} >

@ < puff{: participant} >

@ < inhalation sound{: participant} >

@ < click{: participant} >

@ < laughter{: participant} >

@ < clear throat{: participant} >

@ < chuckle{: participant} >

@ < cough{: participant} >

@ < giggle{: participant} >

@ < sneeze{: participant} >

@ < sigh{: participant} >

@ < yawn{: participant} >

 

 

PROPERTIES OF SPEECH

 

@ < ingressive >

@ < low pitch >

@ < laughing >

@ < quick>

@ < chuckling >

@ < slow >

@ < giggling >

@ < loud >

@ < sighing >

@ < quiet >

@ < puffing >

@ < shouting >

@ < coughing >

@ < whispering >

@ < yawning >

@ < mumbling >

@ < high pitch >

@ < singing >

     

 

SPECIAL EXPRESSIONS                              CLARIFICATIONS

 

@ < SO: expression>

@ < cutoff: word >

@ < pronunciation: word >

@ < overlap: speech[speech >

@ < loan language}:  word}>

@ < overlap: speech]speech >

@ < name >

@ < unclear: {speech}(...){speech} >

@ < abbreviation >

@ < unclear: {speech}(speech){speech} >

@ < acronym >

@ < incomprehensible > 

@ < letter >

@ < alternatively: expr1, expr2, ... >

@ < onomatopoetic >

@ < other language{ continued/start/stop}: language >

 

@ < parallel interaction integer >

 

 

ACTIVITIES AND MOODS          

 

@  < event{ continued/start/stop}: description >

@  < gesture{ continued/start/stop} :  {participant} {description} >

@  < mood: description >

 

 

RECORDING

 

@  < End of tape {side side}. Continued on {tape} {side side} >

@  < damaged: tape damage-description>

@  < not transcribed{:  description} >

 

 

NON STANDARD

 

@  < comment: description >


Appendix C -  Nonstandard Symbols

 

In older transcriptions or in those written with other standards, we find variants of the symbols defined above (with more or less the same meaning). We can also encounter symbols denoting other phenomena not discussed here but considered important for other studies. We include these symbols here for reference only. They should never be used in standardized transcriptions.

 

 

VARIANTS

 

speech                       Emphasised, stressed speech

 

:

::                                  Degrees of lengthening

:::

 

? speech ?                 Clear question

 

(   )                              Inaudible speech

(xxxx)                        Inaudible speech with number of syllables marked

 

(.)                                So-called ‘micropause’

 

(5.9)         

/5.9                            Pause with duration measured in seconds

 5.9 

 

+ 

++                        Pause markers, used like /, //, and ///

+++ 

 

OOspeechLL            Overlap

+speech               Beginning of overlap (without end marker)

 

 

OTHER SYMBOLS

 

#speech#                  Phonetic representation

 

*speech*              Quieter speech (but cf. below)

>speech<                  Quicker pace

 

'                                   Incorrect word stress

"                                   Contrastive stress (even inside words)

 

.                                   Falling intonation

,                                   Continuing intonation

?                                  Rising intonation

^                                  Rising intonation

!                                   Animated intonation

 

speech/               Speaker self-interruption, self-correction

spee-                    Cut-off sound (speaker stops inside a word)

speech\               Speaker shift with interruption

 

*speech*              Complete non-target language form (but cf. above)

**speech*                  Source/target language mixed form

 

<< info >>           Gaze direction

 

(( info ))               Comment

 

#                                  Turn shift (occurs on a separate line)

 

 

 

 

 



[1] Activities within this program have been funded by The Swedish Council for Research in the Humanities and Social Sciences (HSFR) through the project Semantics and Spoken Language and by The Bank of Sweden Tercentenary Foundation through the project Spoken Language and Social Activity.

[2] We use the SAMPA machine-readable phonetic alphabet for phonological representations.

[3] Allwood, J. (2000) An Activity Based Approach to Pragmatics. In Bunt, H., & Black, B. (Eds.) Abduction, Belief and Context in Dialogue: Studies in Computational Pragmatics. Amsterdam, John Benjamins, pp. 47-80.