http://www.mpi.nl/ |
||||||||||||||||||||||||||||||||||||||||||||
Database of Dutch diphone perception |
||||||||||||||||||||||||||||||||||||||||||||
Material -- Recording |
||||||||||||||||||||||||||||||||||||||||||||
| background
material  selection
recordinggating
responses download paper preprint |
To facilitate the pronounciation of the chosen diphones
they were placed in a nonsense environment with whom they formed a legal
sequence in Dutch. In some cases this was necessary because the diphone
by itself is not a syllable (CC diphones) or is not phonotactically legal
(e.g., C-short vowel diphones, since short vowels cannot be syllable-final).
For VV diphones where both vowels are either stressed or unstressed, inclusion
of additional syllables made the sequences easier for the speaker to produce
with correct stress. The nonsense environment always included at least
one phoneme after the target diphone, so that the diphone would not be
final to the item. This prevented excessive lengthening of the diphone.
The environments for CV and VC diphones were also varied to prevent
predictability of the diphone category from the preceding environment.
The table below lists the pronunciation environments for the various kinds
of diphones.
In this way 2294 nonwords were created, most of them with two syllables. A phonetically trained female native speaker of standard Dutch seated in a sound-treated recording booth read all items from a list containing phontically transcribed versions of all items in pseudo-random order. Her speech was recorded on DAT tape using high quality equipment. Any items initially mispronounced were re-recorded. The utterances were low-pass filtered with a cut-off frequency of 7 kHz and digitized at a sampling frequency of 16kHz.
|
|||||||||||||||||||||||||||||||||||||||||||
