Corpus Gesproken Nederlands (COREX)
version 1.4

Manual

This manual was last updated: 25 Apr 2002

Authors: Paul Kilpatrick, Birgit Hellwig

Introduction

The Corpus Gesproken Nederlands (CGN) is a database of recordings and annotations that contains 10 million spoken Dutch words. The Corex program allows you to listen to, view and analyse the corpus. It supports the following features:

  • easy navigation to sub-parts of the corpus, either based on predefined groupings (the sex and age of the speaker, the region in which (s)he grew up, the text type) or based on user-defined groupings (for search purposes or as search results),
  • display of synchronised audio and annotation data,
  • display, search and statistical analysis of annotation data,
  • display and search of metadata descriptions (i.e., information about the kind of data that is contained in the corpus such as information about the speakers).

This manual describes the Corex program commands. It is organised in a sequential manner based on the windows and panels that open up in the program. The following is a list of the Corex windows and panels and their corresponding sections in this manual:

1Corpus Browser Window

2Metadata Search Panel

3Corex Viewer

4Content Search Panel

5Statistics Panel

Notation Conventions

The following notation conventions are used:

  • Menu items, icons and buttons are written in the font MS Sans Serif.
  • Screen displays are written in boldface.
  • Shortcut keys are written in small caps.
  • Information on troubleshooting starts as follows: !

Table of Contents

1Corpus Browser Window......

1.1Metadata Descriptions Tree Panel

1.1.1Navigating through the CGN corpus......

1.1.2Displaying information......

1.1.3Selecting parts of the corpus for purposes of analysis......

1.2Bookmarks Panel

1.3Info/Content Panel

1.4Description Panel

1.5Browser Action Panel

2Metadata Search Panel......

2.1Specify the search options

2.1.1Select the category to be searched......

2.1.2Enter the search term......

2.1.3Add or delete a search query......

2.2Initiate and stop the search

2.3Display the search results

2.4Save the search results

2.5Exit the Metadata Search panel

3Corex Viewer......

3.1View Menu

3.1.1Audio Player......

3.1.2Wave Panel......

3.1.3Print to file......

3.1.4Time Synch......

3.1.5Visible Tracks......

3.2Options Menu

3.2.1Praat Synch......

4Content Search Panel......

4.1Track Selection Menu and Text Field Box

4.2Regular Expression Search

4.3Add Criterion button

4.4Add Query button

4.5Delete Query button

4.6Start/Stop Search button

4.7Save Results As … button

4.8Open Result Set … button

4.9Save Results As Corpus … button

4.10Close button

5Statistics Panel......

Appendix A......

User’s Guide

1Corpus Browser Window

Starting the Corex program opens up the window IMDI-BCBrowser for Corpus Gesproken Nederlands (referred to in this manual as the “Corpus Browser” window).

In the Corpus Browser window you can view and access the directory of the Corpus Gesproken Nederlands (CGN): you can read information about the kind of data that is contained in the corpus, you can access the annotation and audio files, and you can initiate searches and do statistical counts.

The Corpus Browser window contains the following five panels:

1.1Metadata Descriptions Tree panel

1.2Bookmarks panel

1.3Info/Content panel

1.4Description panel

1.5Browser Action panel

1.1Metadata Descriptions Tree Panel

The Metadata Descriptions Tree panel allows you to navigate through the predefined structure of the Corpus Gesproken Nederlands (CGN). It thereby gives you access to the audio, the annotation and the metadata description contained in the corpus. In addition, there are five buttons at the bottom of the panel that allow you to select parts of the CGN corpus for purposes of analysis.

1.1.1Navigating through the CGN corpus

The CGN corpus is displayed when the Corpus Browser window initially opens up. Double-click on the CGN corpus icon to display its corpus nodes; double-click on any corpus node to open it and to display the next level in the hierarchy.

! Note: The Metadata Descriptions Tree distinguishes between open and closed corpus and session nodes. These are represented through different icons:

Most of the program commands do not work when the node is closed. Therefore, if any of the commands do not seem to work, please make sure that the node is open. Double-click on its icon to open it.

! Note: Because of the large amount of data that is loaded, it may take some time until Corex responds to the ‘open’ command.

The CGN corpus branches into five sub-corpora. All five sub-corpora contain the same kind of data, but they group it according to different parameters:

  • CD List: the CGN corpus as it is grouped according to the audio CDs that are supplied with Corex.
  • Ages: the CGN corpus as it is grouped according to the age and sex of the speaker(s).
  • Sexes: the CGN corpus as it is grouped according to the sex and age of the speaker(s).
  • Regions: the CGN corpus as it is grouped according to the native region of the speaker(s) (i.e. his/her place of residence between the ages 4 and 16).
  • Text types: the CGN corpus as it is grouped according to the discourse genre.

The lowest level in the CGN directory is the session. Each session corresponds to a single piece of data. E.g., the session labelled “fn000001” contains a spontaneous commentary of a middle-aged male speaker from the region Gelders Rivierenland. Furthermore, each session is linked to all relevant corpus nodes in the directory. E.g., the session labelled “fn000001” is linked to the corpus node middle-aged male speakers (Sexes/ male/ middle/ 25-34), the node Gelders Rivierenland (Regions/ The Netherlands/ transitional region/ Gelders Rivierenland) and the node spontaneous commentary (Text types/ monologue/ public/ broadcast/ unscripted/ spontaneous commentary).

Each session node contains the following kind of information:

  • metadata descriptions, i.e., information about the actual audio and annotation data (see sections 1.3 and 1.4),
  • links to a *.wav file (i.e., an audio file),
  • links to a number of annotation files (*.pri, *.skp, *.tag, *.syn, *.bpt, *.lxk).

Double-click on any session node to open a viewer panel (the Corex Viewer), which displays the session’s annotations (see section 3). Each session has a corresponding audio file. In order to use the Corex Audio Player or Wave Panel features, you need the CD that contains the relevant audio file (see sections 3.1.1 and 3.1.2).

! Note: You can only access the Corex Viewer and the Audio Player and Wave Panel through double-clicking on the session icon:

Clicking on any of the annotation files will not start the Corex Viewer. Nor will clicking on the audio file start the Audio Player.

! Note: For some sessions the data files are not available yet. Available and non-available files are represented through different icons:

Some files are not available because they exist in compressed form only. The Corex Viewer and the Audio Player nevertheless work normally (see section 3). And note, too, that a missing annotation file does not have any consequences for the analysis: the Corex program will include this session whenever it performs searches or statistical counts.

1.1.2Displaying information

A left mouse click on any corpus node in the Metadata Descriptions Tree panel will highlight this node; and a right mouse click on any open highlighted icon will display a menu, offering you the following options:

  • Session Count: select this option to count the number of sessions contained under this node. The information is displayed in the Info/Content panel.
  • List Sessions: select this option to view a list of all sessions contained under this node. The information is displayed in the Info/Contentpanel.
  • Remove from basket: select this option to remove the node from the list of nodes to be analysed (see section 1.1.3).
  • Add to basket: select this option to add the node to the list of nodes to be analysed (see section 1.1.3).
  • Clone Node: select this option to open a second Corpus Browser window that contains only the highlighted node.
  • Add to Bookmarks: select this option to add the highlighted node to the Bookmarks panel.
  • Show File Content: select this option to display the XML file of the highlighted node in the Info/Content panel.

In addition, all corpus nodes that were created with the help of either the Save Results as Corpus … button in the Content Search panel (see section 4.9) or the Save results in: button in the Metadata Search panel (see section 2.4) contain the following additional option:

  • Remove: select this option to remove the node.

The options Clone Node, Add to Bookmarks, and Show File Content are also available for session nodes. In addition, session nodes contain two further options:

  • COREX Viewer: select this option to open the Corex Viewer, which displays the session’s annotations and gives you access to the audio data (see section 3). This is the default setting for any session - i.e. a double-click on the session node will automatically open the Corex Viewer.
  • Show LR’s: select this option to display the directory information for all the files associated with the session in the Info/Content panel.
  • Portray: If there is a syntax file available for this session you can start up the Portray syntax viewer tool. You are should have installed the Portray scripts yourself (see the readme.txt file).

The following additional options are available for the annotation files (i.e., *.pri, *.skp, *.tag, *.syn, *.bpt, *.lxk):

  • Show format: select this option to display the file format in the Info/Content panel.
  • Show Text Content: select this option to display the file content in the Info/Content panel.
  • Save File Content: select this option to download the file and save it in a non-compressed format.
  • Show Content-Type: select this option to display the file format in the Description panel.
  • Show Description: select this option to display the file description in the Description panel.

All information is either displayed in the Info/Content panel (see section 1.3) or in the Description panel (see section 1.4). The following illustration is an example of how selected information is displayed in the Info/Content panel:

1.1.3Selecting parts of the corpus for purposes of analysis

The Corex program allows you to do content and metadata searches and statistical counts of the corpus (see sections 2, 4 and 5). By default, the analysis is done throughout the whole corpus. However, it is possible to limit the analysis to a part of the corpus: to one (or several) corpus and/or session nodes. The five buttons at the bottom of the Metadata Descriptions Tree panel are needed for the selection process.

To select a corpus or session node, do the following:

  1. Highlight the node through clicking on it with the left mouse button.
  2. Either click the Add button at the bottom of the Metadata Descriptions Tree panel.

Or click with the right mouse button on the highlighted item, and select Add to basket from the drop-down menu. Note that this last option is only available for corpus nodes, but not for session nodes.

The icon of any selected node will change its colour to grey, e.g.:

Furthermore, once an item is selected, the List button will be highlighted in red colour.

  1. Repeat this process to add other nodes to your selection.

! Note: You can only select a node that is open. Double-click on it to open it.

! Note: It is possible to select several different nodes, e.g. the nodes (a) ‘male’ and (b) ‘The Netherlands’. The analysis will then be done over all sessions contained under the node ‘male’ (regardless of region) and all sessions contained under the node ‘The Netherlands’ (regardless of sex). This option does not allow you to limit the search to, e.g., all male speakers that grew up in the Netherlands. To limit your search in such a way, make use of the metadata search options (see section 2).

! Note: When you limit your search to, e.g., the node ‘male’, it is not necessarily the case that the search tool will only find those items that are actually spoken by a male speaker. The node ‘male’, e.g., also contains sessions with dialogues between male and female speakers. Searching such sessions may thus include data from female speakers.

Click the List button to view a list of all selected nodes, e.g.:

You can remove all selected nodes from this list by clicking the Clear button.

To remove a specific node from the list, do the following:

  1. In the Metadata Descriptions Tree panel, highlight the node through clicking on it with the left mouse button.
  2. Either click the Remove button at the bottom of the panel.

Or click with the right mouse button on the highlighted item, and select Remove from basket from the drop-down menu. Note that this last option is only available for corpus nodes, but not for session nodes.

  1. Repeat this process to remove other nodes from your selection.

You can save the selected list for future uses. Click the Save button. The following message informs you that your list has been saved:

! Note: Once you have saved a selected list, you can only remove it by first clicking the Clear button (to remove all selected nodes) and then the Save button (to save an empty list).

When you are satisfied with your selection, turn to the Browser Action panel in order to initiate searches or statistical counts (see section 1.5).

1.2Bookmarks Panel

The Bookmarks panel of the Corpus Browser window displays shortcuts to various nodes in the corpus. Having a bookmark allows you to immediately access such a node, without being obliged to first navigate through the entire corpus. By default, the following bookmarks are displayed:

  • CGN Root: the whole CGN corpus.
  • CD List: the CGN corpus as it is grouped according to the audio CDs that are supplied with Corex.
  • Text Types: the CGN corpus as it is grouped according to the discourse genre.
  • Geographical Dispersion: the CGN corpus as it is grouped according to the native region of the speaker(s) (i.e. his/her place of residence between the ages 4 and 16).
  • Age Dispersion: the CGN corpus as it is grouped according to the sex and age of the speaker(s).
  • Search Results: the saved results of your searches (see sections 2 and 4).

Double-click on any of these items in the Bookmarkspanel to open the corresponding node in the Metadata Descriptions Tree panel.

In addition to the predefined bookmarks, you can create your own bookmarks. Do the following:

  1. In the Metadata Descriptions Tree panel, highlight the relevant corpus or session node by clicking on it.
  2. Right-click on the highlighted node and select Add to Bookmarks from the menu options.

The following dialogue box appears:

  1. Type in a name for the new bookmark.

The new bookmark is added to the Bookmarks panel.

These bookmarks remain available every time you restart Corex. To remove a bookmark, do the following:

  1. In the Bookmarks panel, click on the bookmark that you want to remove.
  2. Right-click on that bookmark and select Delete Bookmark from the drop-down menu.

The bookmark is deleted without further warning.

! Note: You can only delete those bookmarks that were created by yourself, but never the ones that are predefined by Corex.

1.3Info/Content Panel

The Info/Content panel displays information about corpus nodes, session nodes and files. To read the information, click on the corresponding icon in the Metadata Descriptions Tree panel.

! Note: The information is only displayed when the corpus or session node is open. Double-click on it to open it.

In the case of corpus and session nodes, information about the name of the node, the title, and (in the case of sessions) the recording date is displayed, e.g.:

In the case of a Metadata node, the following kind of information is displayed:

  • Location: information about the place where the data was collected.
  • Keys: information about the size of the files, and about the name of the session CD on which the corresponding audio file is located.
  • Project: information about the project within which the data was collected.
  • Collector: information about the person who collected the data.
  • Content: information about the content: the task, the genre, etc.
  • Participants: information about the participants.
  • Source: information about the source.
  • References: information about additional references relevant to the content of the session.

The Info/Content panel can optionally display additional information such as a count of all the sessions contained under a node, a list of all the sessions contained there, the directory information of file(s), the format of file(s) and the content of the XML files. To view this information, do the following: