Coding a VIHI file
VIHI files are named in the following format XX_subj#_ageindays.
The first code stands for Typically Developing (TD), Vision-Impaired (VI), or Hearing Impaired (HI).
The second code will stay consistent for all recordings from a participant.
The third code designates an individual recording based on how many days old the participant was when the recording was collected.
So for example, the file HI_424_527 is participant 424 from the sample with hearing impairment, recorded when they were 527 days old.
1. Find your file in SubjectFiles
Find your file in on the Fas-Phyc-PEB-Lab/VIHI/SubjectFiles/LENA. Navigate to the folder of the population your file is from (HI, VI, or TD). Find the folder with the subject # of your file, and choose the subfolder with the age-in-days number of your specific recording file. For example, to find HI_424_527, I would find the appropriate files in:
**NOTE: as of 12/9/2022, use ELAN version 6.2 or earlier. 6.3 and 6.4 are buggy and significantly slow down coding time. This will be updated if new releases resolve lag issues.
3. Annotate the . eaf file, one clip at a time, following ACLEW standards, in the version of your folder in annotations-in-progress
Lab specific differences from ACLEW gold standard tests:
Navigate to each clip by going to Grid view in Annotation mode, then select the "code_num" tier marked with your clip number.
Listen through the whole context segment (2 minutes before the clip and 1 minute after the two-minute segment) before you start coding!
Follow instructions for marking personal information.
It's helpful to have the tutorials up while you code! (specficially, tutorial 8 for in-depth minCHAT rules!)
For EE1 tiers: Segment all electronic media, but do not transcribe it. Enter all annotation values as "0."
Unlike the training files, some clips may be completely silent! They were randomly selected by time frame. Just mark the clip as silent in the coding issues document and move on to the next one.
If an utterance at the beginning of the clip starts before the boundary of the segment on the Code tier, segment the whole utterance. Similarly, if the segment on the Code tier ends mid-utterance, segment the entire utterance.
Slide the task over on the VIHI random sampling coding Asana board as you complete clips 1-5, 6-10, and 11-15, respectively.
4. Take notes in the coding issues document as you go:
Fill out the cast of charcters section to describe each tier in your file: the first name of the speaker (if known), and a brief description of their relationship to the child or role in the recording (e.g. grocery store cashier).
The start and end timestamps of each region.
a description of each clip
anything weird or difficult to code about the file
parts you want to have reviewed
notable behaviors by CHI
notes about the media that's playing
etc. !
5. Add speaker metadata to the ELAN file.
Every tier should have a participant name that aligns with the name of the main transcription tier, e.g. FA1 or CHI or EE1. *** If we get more informative, e.g. using first name or descriptor like mother will mess up the automated minCHAT checker.
6. Export the file as a tab-delimited text file.
Select ok, navigate to your folder, save the file with the same name as the .eaf, and answer "Ok" to the pop-up asking about UTF-8 encoding.
Note: overwrite the old version of the .txt file, no need to change the name if one already exists! We just use these for checking so they should be up to date with the eaf.
7. Submit the .txt file to the automated minchat checker.
Here is the link for the checker.
Here are notes about how to upload and how the checker works.
Open the spreadsheet, track down the annotations it names, and correct the errors that the checker finds in your .eaf file, and then repeat step 6 & 7 as needed. You can ignore warnings about "incorrect tier name" for things like e.g. is_crying-- we made these up for our lab and don't care that they don't follow a three-letter code system.
8. Slide the task to the "ready for superchecker" column on asana.
Last updated
