Consonant transcription

[patrickreidy]

Version 0.1 of the NonWordTranscription.praat script does not include the functionality for transcribing consonants that was agreed upon during the meeting between Mary Beckman, Jamie Byrne, and Patrick Reidy that took place at UMN at the beginning of the month.

At that meeting, it was decided that the transcriber would first be asked to determine the Manner feature value of the consonant among the choices: Stop, Affricate, Fricative, Nasal, Glide, Omitted. If the transcriber selects any value other than 'Omitted', then they are presented with a list of WorldBet phonetic symbols that match the previously selected Manner value, along with the option 'Other', which when selected provides the transcriber with the ability to select the Place and Voicing feature values from drop-down menus.

Before this functionality can be fully implemented the following questions should be addressed:

• For each of the Manner feature values listed above, what WorldBet symbols should be available? Just the ones from the Consonants table on pg. 4 of the WorldBet guide (http://www.ling.ohio-state.edu/~edwards/WorldBet/worldbet.pdf)?
• What transcription data should be entered into the TextGrid? For instances when the transcriber selects a WorldBet symbol, there is the option of entering just the symbol or expanding it out to its Manner, Place, Voicing feature set. Entering the symbol is perhaps easier for a human user to interpret; however, this will mean that there will be transcriptions of two different types: 1) segmental transcriptions; and 2) featural transcriptions (from when the transcriber selects 'Other').
• Should the consonant's segmental accuracy score also be entered into the TextGrid? If so, should this go on a separate tier from the consonant transcription or be concatenated somehow to the consonant transcription?

[tjmahr]

These have all been mostly resolved. The textgrid records the symbol, the feature choices and the score.