nzilbb.formatter.clan (1.4.2)

Serialization for CHAT files produced by CLAN.

NB the current implementation is not exhaustive; it only covers:

Time synchronization codes, including mid-line synchronization.
Overlapping utterances in the same speaker turn are handled as follows:
- If overlap is partial, the start of the second utterance is set to the end of the first.
- If overlap is total, the two utterances are chained together with a non-aligned anchor between them.
Disfluency marking with &+ - e.g. so &+sund Sunday
Non-standard form expansion - e.g. gonna [: going to]
Incomplete word completion - e.g. dinner doin(g) all
Acronym/proper name joining with _ - e.g. no T_V in my room
Retracing - e.g. <some friends and I> [//] uh or and sit [//] sets him
Repetition/stuttered false starts - e.g. the <picnic> [/] picnic or the Saturday [/] in the morning
Errors - e.g. they've <work up a hunger> [* s:r] or they got [* m] to
Pauses - untimed, (e.g. (.), (...)), or timed (e.g. (0.15), (2.), (1:05.15))
%mor line annotations (or %pos line annotations, if present)

Layers to add to fully capture supported tags

(All Type=Text, Alignment=None)

If there's a %mor layer in the transcripts:

(All Type=Text, Alignment=Intervals)

mor: Complete %mor tag(s) for the word token - there can be multiple tags, e.g. contractions and cliticizations.
These tags are then spit into the following parts:
morFusionalSuffix
morGloss
morPOS
morPOSSubcategory
morPrefix
morStem
morSuffix

If there's also a %gra layer in the transcripts:

gra (Type=Text, Alignment=Intervals): Complete %gra tag, one for each %mor annotation. (These tags mark grammatical relations, between a dependent and a head; the dependent is tagged, but currently is not formally linked to its head.)

(All Type=Text, Alignment=Intervals)

cunit: The grammatical unit for each utterance, labelled with utterance terminator
error: Errors - e.g. they've <work up a hunger> [* s:r] or they got [* m] to
linkage: Multiple words in a name joined by _ e.g. Winnie_ther_Pooh
pause: Pauses - untimed, (e.g. (.), (...)), or timed (e.g. (0.15), (2.), (1:05.15))
repetition: Repetition/stuttered false starts - e.g. the <picnic> [/] picnic or the Saturday [/] in the morning
retrace: Retracing - e.g. <some friends and I> [//] uh or and sit [//] sets him

gem (Type=Text, Alignment=Intervals): Parts of the transcript marked for separate analysis.

language: The language the participant speaks.
corpus: A one-word label for the corpus in lowercase.
age: The age of the speaker, using the form years;months.days as in 2;11.17 for 2 years, 11 months, and 17 days.
sex: gender can be used as the attribute name.
group: Any single word label.
SES: Socio-economic status
e.g. WC for working class, UC for upper class, MC for middle class, LI for limited income)
role: The speaker's (standardized) role e.g. Target_Child, Target_Adult, Child, Mother, Father, Participant, Investigator, Adult, Friend, Unidentified, etc.
education: Educational level of the speaker e.g. Elem, HS, UG, Grad, Doc
custom: Any additional information needed for a given project.