Sampler Hours- ACLEW
Last updated
Last updated
Being familiar with the CITI training modules is important before working with the data!
Complete the Duke Campus IRB section. Do not complete the Duke Health CITI certification.
When you finish, email the transcript/certificate to Lab Manager.
VPN & drive instructions here: sections 1 and 2, accounts and beginning a shift
ELAN download here (choose the latest version for your operating system, unless you're on an M1/M2 Mac, in which case use version 6.2 for M1): https://archive.mpi.nl/tla/elan/download
You will be transcribing two minutes of a 16-hour long audio file that we recorded to capture a real day in a child's life. You will be using a simplified version of the ACLEW DARCLE annotation scheme, which is a collaboration between baby labs to use a standardized coding format, to make data shareable!
Navigate to Fas-Phyc-PEB-Lab/Seedlings/ACLEW/Training_English. Within the folder, there should be a subfolder named coder_[YourInitials]_sampler. Open it, and double click on the file called TD_419_961_clip10_[YourInitials].eaf.
A file will open in the program called ELAN. The following warning may pop up:
Press "OK." Some things to know about this file:
The file contains the whole 16-hour recording, but you (of course!) don't need to listen to the whole thing. You will listen to five minutes of it: two minutes before the segment you're coding, the two minutes that will transcribe, and one minute after. The minutes at the beginning and the end of the coded segment are there to provide you context and help you figure out what the family is talking about!
To find the segment that you will code in the file (it's NOT right at the beginning):
Go to Options... and make sure "annotation mode" is selected.
Then, in the viewer bar, select Grid (the farthest to the left). In the dropdown, select the "context" tier. One row will pop up in the spreadsheet below the dropdown:
Click on the row (numbered with one), and it will highlight itself in purple and auto-snap you to the right section of the recording in the waveform player below. Press shift+space on your keyboard and listen through the 5 minutes to adjust to the audio "landscape." ctrl+space will pause/unpause the audio.
Essentially, we have created one long annotation to mark out this section of audio. The "context" segment includes 2 minutes before the coding segment and 1 minute after, to help you figure out what is happening at this point in the recording.
Do not start coding at the beginning on the "context" segment-- start at the beginning of the code segment (which starts two minutes later and is visible on the code tier. It already contains segments on speaker tiers).
As you listen, you will notice a bunch of blank annotations on other speaker tiers when you hit the segment you are supposed to code, including one on the "code" tier that marks the boundaries of the two minutes you will transcribe:
Leave them alone for now. Once you have finished your first listen-through, click the following link to find a set of tutorials. You will only read a few of them-- the actual training process for this annotation scheme takes weeks!
*Keep in mind, the file you are coding is slightly different than their examples-- it already has the participants and utterances marked out, but it's helpful to know how those annotations got there.*
When you've got a handle on the tutorials, return to your file and fill in the codes for vcm on the CHI tier, and xds on the FA1 and MA1 tiers (feel free to use the tutorials for reference!). (Don't code for lex and mwu for time's sake).
***If this part is taking forever, skip ahead to transcription when you have about an hour left in your sampler hours.***
In the actual ACLEW DAS annotation scheme, learning the rules for minchat is what takes the longest, so you will not use them for sampler hours.
Instead, just write out what each speaker says during each utterance following conventional spelling/punctuation rules (your best approximation, I'm not here to judge your formal wiriting or anything). To fill in an annotation, double click or press alt/option+m while you have the utterance selected.
Use the following simplified guidelines for special cases (use these instead of trying to decipher tutorial 8):
-unlike they mentioned in the tutorials, for this file, please also try to transcribe what the child says, in addition to the other speakers!
-there should be exactly one and only one period, exclamation point, OR question mark, and it should be at the end of each utterance.
-if you want to mark a special type of speech, like singing or whispering, put the words that are affected by it in <these types of brackets> and then put [sings] in brackets directly after the closing triangle bracket. e.g. "<build me up buttercup> [sings]."
-If a speaker makes a communicative sound that isn't a "word," like a gasp or a laugh for example, just write "&=" followed by the verb in the 3rd person present singular. e.g. &=laughs or &=blowsraspberries.
-if you're not sure how something is spelled (like someone's name or a colloquial term), make your best guess, or go with the most common form according to an online dictionary.
-if you can't tell what someone said, simply write "xxx" for the parts you can't hear! Fill in as many words as you can.
