v1.4.0

Scoring changes:

• Restructured prosody scoring, such that transcriber is prompted to assign prosody score after transcribing each segment, then prompted to score a frame prosody score for the frame involving the two transcribed segments. (#38)
• Update to allow for 4-point diphthong offglide scoring. (#38)

Transcription changes

• Transcriber is prompted to include a WorldBet transcription that is reflective of the production if the frame prosody score is transcribed as incorrect (0).
• Transcriber is able to more easily transcribe if the target matches the production through the inclusion of a default prompt. (#28.)
• Cosmetic changes added, such as phonemic slashes (//) and reminders of the target and how it has been transcribed, wherever appropriate.

Development led by @caliban19

v1.3.0

New features

• Transcriber can now extract intervals to a Praat collection.

Transcription changes

• Vowel transcription changed to parallel consonant transcription.
• Transcription and scoring separated for non-canonical substitutions.
• Category of "unclassifiable" added to both C and V transcription and added "noise" to both.
• Added schwa (WorldBet 6) to set of lax vowels.

Development led by @marybeckman

version for first wave of transcription

Design changes

This script is no longer responsible for scoring for C and V productions. (It still is responsible for scoring prosody though). As @patrickreidy puts it, the job of this script instead is to guide the user to choose a symbol or a set of features to describe a production. For example, the script leads the user from the consonant that they hear at the beginning of pwaugub to the description p;Stop,Bilabial,Voiceless;. The actual mapping of numerical scores to target-description pairs will be handled by more appropriate scripting tools. That said, the script still includes numerical scores with the descriptions (e.g., k;Stop,Velar,Voiceless;0), but these scores are only an approximation.

Other design changes

• A third point has been added prosody scoring. The speaker loses a point if the non-word produced does not have the correct number of syllables.
• The script now includes a point-tier and a prompt for recording notes on trials.

Transcription changes

• "Omitted" is an option for vowel productions.
• Most non-English phonemes have been removed from the symbol set.
• The generic lateralized fricatives lh and lhv have been added to symbol set.
• ae has the correct height.

Other changes

• The actual packaged-up script in the last release crashed when it was supposed to transition from CV to VC sequences. Oops.
• Fixed an inconsistency where the second tier was named Target2Pros instead of Target2Seg
• The drop-down menus for feature choices in vowel transcription now default to the feature values of the target sound, instead of defaulting to "front", "high", and "tense" on every trial. This change is mainly meant to prevent users from misremembering the correct lax/tense feature for a vowel.

CV, VC and CC Transcription

Design changes

• Procedures are used whenever possible to contain behavior within meaningful subroutines. For example, @transcribe_segment(...) checks whether the segment is a vowel and calls either @transcribe_consonant(...) or @transcribe_vowel(...) procedure. The consonant transcription procedure in turn calls @transcribe_cons_manner(...), @transcribe_cons_symbol(...), @transcribe_cons_place_voice(...), and @score_consonant(...).
• Procedures are also used to contain or (hide) related variables in a common namespace. For example, @nwr_trans_textgrid(...) defines all the important variables needed to define and describe the transcription textgrid -- for example, its filepath (locally .filepath$ or globally nwr_trans_textgrid.filepath$).
• Transcription steps are managed by an event loop with 7 nodes: transcribe T1, score T1, transcribe T2, score T2, transcribe prosody, score prosody, and save results.
• Start-up wizard now re-uses code from segmentation start-up.
• Functionality is broken up over multiple files.

Transcription changes

• CV trial type has been generalized so VC and CV trials are now supported.
• Prosody scoring now based on insertion and deletion of segments.
• Diphthongs scored for frontness and backness based on the first vowel.
• /l/ and /r/ added as "voiced" "alveolar" "glides"

Other changes

• Minimum version of praat is now enforced. Praat commands that uses Command... [args] notation have been replaced with the recent Command: [args] notation.
• Multistep transcription wizards (consonants) support a back-button.