In the spring of 2002, a pilot study was conducted that aimed to test the protocol and the procedure that had been developed for the prosodic annotation of part of the data in the Spoken Dutch Corpus.

Below an overview is presented of those samples for which in the pilot study a prosodic annotation was made (for part of the sample).  Table 1a gives an overview of the Dutch data, Table 1b of the Flemish data.* For the pilot, the samples were divided over 12 lots (column 1 in the table). In addition, a training corpus (LC) and a test corpus were defined.

* Please note: the components referred to here are the components as they were distinguished in the original design. See also the design of the corpus.

Table 1a: Dutch data (i.e. data originating from the Netherlands)
 

lot	component	# speakers	duration (# sec)	sample(s)
1	1	2	526	fn000303 (LC)
2	1	2	577	fn000346
3	1	2	616	fn000415
4	2	2	890	fn000081 (LC)
5	2	2	1084	fn000140
6	5	2	343	fn000043
7	6	3	600	fn000115
8	6	-	-	-
9	9	-	-	-
10	11	1	300	fn001601, fn001602, fn001603, fn001609, fn001624, fn001629, fn001631, fn001638, fn001639, fn001642, fn001645
11	12	1	214	fn000039 (LC)
12	13	1	510	fn000061 (0...120 als LC)

Note: Lot 10 comprises several samples that should remain together.
 

Table 1b: Flemish data
 

lot	component	# speakers	duration (# sec)	samples
1	1	2	1045	fv400073
2	1	2	457	fv400089 (LC)
3	2	2	1275	fv400109 (LC)
4	2	2	1002	fv400118
5	2	2	1485	fv400165
6	5	2	634	fv600228
7	6	3	231	fv600005
8	9	1	333	fv600388, fv600764
9	10	2	324	fv600091, fv600551                                                        .
10	11	4 3 2	129  69  173	fv600053 fv600121 fv600268
11	12	2	306	fv6000253
12	13	1	839	fv400011 (LC)

Note: Lots 8, 9 and 10 comprise several samples that should remain together.

Each sample was transcribed four times, each time by a different student. The students were working in two different locations in the Netherlands and two locations in Flanders, two students per location. The students were instructed and supervised by local coordinators.

For a discussion of the way this pilot study was set up and the results that were obtained, we refer to Buhmann et al. (2002), here available in .ps and .pdf format.

The results of this pilot study have also been included in this release of the corpus. The files with the prosodic annotations can be identified by the extension .pro and in so far as they result from the pilot can be found in the directories data_NL/ (Dutch data) and data_VL/ (Flemish data). In each of these subdirectories a subdirectory /aangelev/ can be found which contains the files as they were made available to the students.