Transcription Protocol Updated 11.2023
##With Spacy processing in mind and minimal variables to be extracted from CLAN
##This workflow developed specifically for bilingual participants

1. Reference CS Data analysis SmartSheet (Spanish, Catalan or English) to determine which participants you will transcribe and have been transcribed via whisper (self-assign).

Open

General Guidelines Regarding Transcription

1. Transcribe into the template by making a copy of the template (make sure to turn off spell check so you can note paraphasias and word fragments without it auto-correcting). Make sure to read the rules on utterance segmentation, paraphasias, fillers, fragments, and unintelligible words.

2. After *PAR: and the text it needs to be a TAB (so it's the larger spacing that needs to be present). Between *PAR: and the text there needs to be a TAB (so it's the larger spacing that needs to be present).Do not leave a space between *PAR: and the text.

3. Only transcribe the participant's utterances

   1.  

      For VISTA script probes, start after the clinician's probe phrase, "Tell me about [topic of script]."

   2. For discourse samples, start after the clinician's instructions to talk about what they see in the picture/talk about a personal event.

4. Transcribe EVERYTHING they say, including fillers such as "uh" or "er" and any word fragments or false starts they make. This can mean listening to the clip a couple of times.

5. If the participant makes a comment about the task, or a tangential comment at the beginning or end of the probe, omit that from the transcription.

   1. If the participant is clearly talking about something other than the topic in the middle of the probe such as "my dog won't stop barking," then omit that from the transcription.

   2. If the participant restates the topic of the probe, (e.g. Clinician: "Tell me about your family." Participant: "My family. I was born in…") omit the restatement (in this example the transcription would start: "I was born in…"). Otherwise, if there are somewhat tangential, brief comments in the middle of the probe, include them.

   3. NOT tangential comments: Comments like “como se llama, no me acuerdo", should be transcribed

   4. Examples of comments to omit at beginning and end:

      1. this is hard.

      2. that's all I remember.

      3. Done.

Checking the transcription with specific CLAN commands

Important Note: If you are unsure about using CLAN, please save a back up of the files that you are working with. The command can delete all previous versions of transcripts!

Use CLAN to check for typos and spelling mistakes
Download CLAN: https://dali.talkbank.org/clan/
Commands:

1. Command Check – helps find errors like the use of a space instead of a TAB. Use code to correct this mistake in all the samples. Helps to check more global issues

   1. Example to correct errors: In CLAN output: "please run "chstring +q +1" command in this file to fix this error"

      1. Run this command for all samples to correct the use space instead of the TAB

2. Command Mor (morphosyntax) - helps detect words that have spelling mistakes, creates list of words to check.

3. Command eval - to extract all linguistic data - check words that aren't in CLAN library

4. code for partial word repetitions: eval@+tPAR:+u+r6(&+]

VIDEO TUTORIAL, how to use CLAN: How to use CLAN to detect typos and spelling mistakes (command CHECK and MOR

3. A "target+adult" error may occasionally appear, as seen in the example below.

   1. This error occurs when the time-point is missing. To fix this error, the time-point must be edited on the existing .cha file to look exactly as such:

Open

Open

This can be done by copying and pasting the timepoint from a file that has been checked by CLAN.  

Make sure to Save the final .cha file from CLAN in the " 4. Taskname_coded_chat" folder.
Remember to update Connected Speech Data analysis.

How to connect to the CLAN mor Library on BOX: 

1.  

   1. Download BOX drive for local synchronization: BOX drive provides a way to synchronize files and directories (in this case the CLAN mor library) between you local computer and the MADR Lab's BOX storage. This means any changes made in the CLAN C directory on your computer will be reflected in your BOX cloud storage, and vice versa. This means that if the lexicon is updated on one computer, it is automatically accessible across computers. 

   2. Directory mapping: It is important that the directory is properly linked to BOX so that files and data can be automatically synchronized and accessed through BOX.

   3. Once you have downloaded BOX drive to your computer, follow this path when updating the mor library ("mor lib") in CLAN: 

      1. BOX → SLHS_Grasso → MADR Connected Speech Project → Bilingual_Transcriptions → CLAN_Libraries → CLAN updated mor library 20240725 → spa

   4. Here is an image of what this looks like on CLAN:

      Open

      

   5. When there is an error in the path, CLAN will output the following: 

      1. If this occurs, refer to the path above to fix the error.

Open

@End of instructions!