International Corpus of English

Using VOICE-Online

4 Output styles

This section contains information on the different output styles which VOICE Online offers for viewing search results. It explains the standard output styles and the options for modifying output styles.

4.1 Standard output styles

Search results can be rendered in three different styles: VOICE style, plain style, and KWIC style. The styles can be selected in the drop-down menu next to the word style at the bottom of the content area.
Screenshot of output styles
Selecting output styles
The style which is selected at the time of submitting a search is automatically used to display the search results. If the style is changed after a search has been submitted (i.e. when search results are already displayed), the search results currently displayed will automatically be re-loaded into the new output style.

4.1.1 VOICE style

As the default setting when logging in to VOICE Online, search results are rendered in VOICE style.
Screenshot of VOICE style
Search results in VOICE style

In VOICE style, each search result contains information in three columns (although the display may slightly vary with different internet browsers, see Browser recommendations). The column on the left locates the search result in the corpus by listing the ID of the corpus text (e.g. EDcon4) and the utterance number in the speech event (e.g. 891). The two pieces of information are separated by a colon and together form a unique identfier for each utterance in the corpus (e.g. EDcon4:891). The next (i.e. middle) column gives the speaker ID in the transcribed speech event followed by a colon, e.g. S1: or S2:. The third and broadest column on the right contains the actual utterance produced by the speaker.

VOICE style displays search results with full VOICE mark-up and includes all tags and mark-up features and contextual information specified in the VOICE Transcription Conventions [2.1] (with the exception of some minor modifications explained in section Differences between VOICE Online and VOICE Transcription Conventions [2.1]). Mark-up features, tags, and contextual information are mostly rendered in different colours to distinguish them from actual text, which is black. For a list of tags and mark-up features see List of tags and mark-up. Individual tags and groups of tags can also be switched off according to personal research interests (see Modifying output styles).

4.1.2 Plain style

The second style that can be selected is plain style, which renders a reduced version of the transcript for easy readability.
Screenshot of plain style
Search results in plain style
Plain style renders the speaker ID in the transcript (e.g. S1:) and the utterance produced by the speaker in plain text without tags and most mark-up features. The only mark-up features rendered in plain style are @-symbols for laughter and the square brackets signalling anonymized items (e.g. [place1]). Overlapping parts of utterances are rendered in blue (but without numbered tags). Please note: Features such as uncertain transcription, emphasis, intonation, and lengthening are NOT represented in plain style.

4.1.3 KWIC style

The third style that can be selected is KWIC (KeyWord In Context) style. In this style, the search results are displayed in a concordance with the search item highlighted at the centre.
Screenshot of KWIC style
Search results in KWIC style
A maximum of 50 characters (including whitespaces) within the same utterance before and after the search item is displayed together with the search item. Words at the beginning or end of the extract may consequently be truncated.

Similar to plain style, KWIC style is devoid of mark-up and tags. In contrast to plain style, KWIC style does not render @-symbols for laughter and does not indicate overlapping speech. KWIC style thus also renders a reduced version of the transcript.

KWIC style is the only style where the list of search results does not render complete utterances but all occurrenes of the search item listed in their immediate context. If an utterance includes two or more occurrences of the search item, each of these occurrences appears in the list produced in KWIC style. One page of results always displays all occurrences of the search item within 25 utterances (i.e. the list can contain more than 25 results if the search item occurs more than once in some utterances).

4.2 Modifying output styles

In addition to the three standard styles, it is possible to customize the display of search results by switching off particular tags or 'hiding' certain pieces of information.

4.2.1 Flexible display of mark-up and tags

Whenever VOICE style is selected, it is possible to customize the display of search results by switching off particular tags or groups of tags or by hiding pieces of information such as the utterance identifier and speaker number. The tags and features can be switched off (and on again) by clicking on the coloured symbols which are displayed in the menu bar at the bottom of the application area whenever VOICE style is selected.
Screenshot of the menu bar
Menu bar for switching off/on mark-up features
The features which can be switched off/on are (from left to right as they appear in the menu bar):
  • utterance identifier (e.g. EDcon496:198)
  • speaker ID (e.g. S1:)
  • overlap tags (e.g. <1> </1>)
  • pauses (e.g. (2))
  • contextual notes (e.g. {parallel conversation starts})
  • speaking modes (e.g. <fast> </fast>, <@> </@>)
  • speaker noises (e.g. <coughs>, @@)
  • tag for spelling out (<spel> </spel>)
  • tag for non-English speech (e.g. <LNfre> </LNfre>)
  • other-continuation (=)
  • tag for unintelligible speech (<un> </un>)
  • tag for onomatopoeic noises (<ono> </ono>)
  • tag for pronunciation variation and coinages (<pvc> </pvc>)
  • information about gaps and non-recorded parts
A short descriptor of the feature that is switched off/on also appears when the cursor of the mouse hovers over the symbol in the menu bar.

4.2.2 Modifying output styles: a note of caution

Please be aware that switching off certain tags or features ‘distorts’ the original transcript to some extent and intentionally hides information that is actually given in the corpus/transcript. This needs to be borne in mind for analysis and should also be indicated if utterances with reduced mark-up are used as examples in publications or studies. Switching off certain features may also lead to seemingly(!) empty utterances if the content of an utterance only consists of the feature which is switched off (e.g. laughter, a speaker noise).