Differences between CGN version 1.0 and CGN version 2.0
The most important changes in relation to version 1.0 are:
Detailed information on fixed and reported bugs can be found in the TST-centrale's online bugtracking system, Bugzilla.
You are invited to use Bugzilla to report new bugs (or use e-mail or your telephone to do so).
Please visit www.tst.inl.nl for more information.
Incorporation of updates 1 and 2
Since the release of CGN 1.0, two updates were made available by the MPI and the TST-centrale.
These updates can be downloaded from www.mpi.nl/COREX.
Both updates have been incorporated into this new release.
New data
- Lexicon 2.0.: A new version of the CGN-lexicon. Version 2.0 is more complete than version 1.0 and contains more validated data.
Drs. R. Piepenbrock is responsible for the update. Please see the lexicon documentation under doc_English for more information.
- Frequency files: nltypealph.frq, vltypealph.frq, nltyperank.frq and vltyperank.frq.
These files contain word frequency information for Dutch or Flemish data.
In the lists a distinction is made between the different components (speech types) in the corpus.
Please see the frequency file documentation under doc_English.
- Annotation files: Version 1.0 contained 13 unannotated Flemish wave files. These files have been annotated and time-aligned
with their annotations in version 2.0. See below for an overview of these files:
| fv701103 |
| fv701104 |
| fv701105 |
| fv701106 |
| fv701111 |
| fv701113 |
| fv701116 |
| fv701122 |
| fv701123 |
| fv701124 |
| fv701155 |
| fv701161 |
| fv701162 |
Bugfixes
Many bugs from version 1.0 have been resolved in this version:
- The annotations contain less bugs. Great care was taken to maintain consistency during bug fixing.
- consistency within annotation files: when bugs were fixed in the one annotation layer, all dependant annotation layers were updated too.
- consistency between annotation files: every found bug was fixed in the entire corpus. Bug fixing was done interactively,
which enabled proper judgment per case.
Result:
- 4505 bugs in the plk layer (and if necessary in the dependant syn layer)
- 190 bugs in the ort layer (and if necessary in the dependant layers)
- The COREX exploration software contains less errors. The bugs in COREX were prioritised and solved,
whenever that was possible, in order of priority.
| 1 |
crash or inproper action |
solved 100% |
(04/04) |
| 2 |
very troublesome behaviour, but no crash |
solved 78% |
(17/23) |
| 3 |
annoying behaviour |
solved 79% |
(11/14) |
| 4 |
less important wishes and bugs |
solved 88% |
(07/08) |
| 5 |
least important wishes and bugs |
solved 43% |
(03/07) |
|
Total |
solved 75% |
(42/56) |
Important COREX improvements are: the help menu in TigerSearch works, texts in the info/content part of the COREX main window no longer appear scrolled down,
the install procedure has been updated and the layout of the windows has been optimised for a screen resolution of 1024 x 768 or higher
(although it may sometimes still be necessary to maximise your windows; some components disappear otherwise).
- Metadata-bugs were solved.
- Some speakers had double speaker numbers in version 1.0. In version 2.0, only one number was kept.
| Double number |
substituted by |
| V90002 |
V40013 |
| V90001 |
V40018 |
| V90030 |
V40032 |
| V90031 |
V40033 |
| V90505 |
V40065 |
| V90540 |
V40066 |
| V40158 |
V40088 |
| V40140 |
V40213 |
| V80113 |
V40194 |
| V90022 |
V80063 |
| V90504 |
V80101 |
| V90506 |
V80103 |
| V80137 |
V80119 |
| V90007 |
V80129 |
| V90541 |
V80134 |
| V90037 |
V80144 |
| V90276 |
V90079 |
- In some cases, the same speakernumber was assigned to two different speakers. That problem has been solved.
| file |
speakernumber |
| fn000256 |
N00382 |
| fn000257 |
N00383 (new) |
| fn007645 |
N03697 |
| fn007645 |
N03698 (new) |
Updated documentation
The documentation in doc_Dutch and doc_Eng has been improved and updated to version 2.0.
The COREX manual has been updated to the new COREX version.
Updated metadata
The metadata of 60 speakers has been updated.
Removal of superfluous files
The annotation dvd of version 1.0 contained a number of superfluous files. These files are no longer present in the new version.