his/her mother-tongue does not. This makes intuitive recognition more difficult for a
language which is not the researcher’s mother tongue. However, those cues are easily
learned by practice, and their identification is of crucial scientific importance.
Note that because prosody interacts with syntax (and other domains) in cognitive
processing, there often are “mismatches” between prosodic grouping and traditional
syntactic units (although in narratives, clause-boundaries tend to coincide with
intonational boundaries). This is normal, and should not make researchers uneasy.
There might be intonation units consisting of one word. Do not hesitate to mark those
units.
Note also that there can be mismatches between intonation units and the final translation
into English. This is a practical problem, linked to the on-screen display of our texts. One
solution to that is to index longer stretches of discourse to the translation (i.e. have
translation units that cover more than one IU). For prosodic units larger than the
Intonation Unit as defined in CorpAfroAs, see Izre’el and Mettouchi (draft), pdf on the
CorpAfroAs website.
A good way of learning to identify intonation units is by beginning with folktales or
fables: intonation units will be more stylized and easy to recognize, and this will
familiarize researchers with that type of segmentation. Conversation (which is more
difficult to segment) would then be broached more easily.
3.2. Transcription and segmentation in Praat
Once the recording is finalized in the form of a .wav file corresponding to a narrative or a
conversation (leaving a small blank at the beginning and the end of the file), and once it
has been given a name (LanguageCode_InitialsRersearcher_type_number.wav (type =
CONV(ersation) or NARR(ation) ; number = file’s serial number ), it is ready for annotation.
Example of a file name: KAB_AM_NARR_03.wav is a file in Kabyle, recorded and
transcribed by Amina Mettouchi, that is a narrative, the third one in the Kabyle database.
It is recommended to have those files stored in a special Praat folder, within a larger
corpus folder, which for us was named: CorpAfroAs.
When you open Praat you will find three menus at the top of the ‘Praat Objects’ screen:
New, Read, Write.
Go into Read and select:
Read from file…
Comentários a estes Manuais