PPC Ignore Diacritics

8 views
Skip to first unread message

Nina R Benway

unread,
Oct 21, 2020, 5:15:33 PM10/21/20
to ph...@googlegroups.com

Hello Phon Community,

 

I am trying to run a PPC Analysis using the Ignore Diacritics (only) function, but the diacritic I’d like to ignore is not available in the “Select Diacritics” drop down menu. I can’t seem to specify a diacritic by typing/pasting. Is there a way to specify diacritics that are not available from the drop down menu? The character I’d like to exclude is the ligature 0x0361 in the combining diacritics block.

 

Phon v. 3.14-beta3

 

Thanks,

 

Nina

________________________

Nina R Benway MS CCC-SLP
PhD Candidate

Department of Communication Sciences and Disorders
Syracuse University
NYS Licensed, ASHA Certified Speech-Language Pathologist
NYS Certified Teacher of the Gifted and Talented

nrgslp.github.io

 

 

Greg Hedlund

unread,
Oct 21, 2020, 5:21:23 PM10/21/20
to ph...@googlegroups.com

Hello Nina,

Unfortunately there is no method available right now to ignore the ligature diacritics due to technical reasons. Thanks for bringing this need to our attention we will do our best to accomdate this in the near future. I'll let you know when this feature becomes available.

Cheers,

-Greg

--
You received this message because you are subscribed to the Google Groups "Phon" group.
To unsubscribe from this group and stop receiving emails from it, send an email to phon+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/phon/DM6PR01MB43944A7E11C551E97C65A83AC21C0%40DM6PR01MB4394.prod.exchangelabs.com.

Yvan Rose

unread,
Oct 21, 2020, 7:46:05 PM10/21/20
to ph...@googlegroups.com

Hello Nina,

This diacritic effective 'binds' two phones into a single one, as Phon constructs its phone objects for analysis. So we do not process it at all as other diacritics we have.

 

May I ask why you want to do that? Maybe we can find a roundabout way to attain what you're looking for. (You can also email me directly at yr...@mun.ca if you prefer.)

 

Kind regards,

Yvan

--

Nina R Benway

unread,
Oct 22, 2020, 9:52:25 AM10/22/20
to ph...@googlegroups.com

Hi Greg, Hi Yvan,

 

Thank you for the look under the hood of Phon!

 

We’ve come across cases in our corpus where the target and the actual differ by the ligature. This results in the PPC calculation counting an instance of epenthesis (as it should, algorithmically) despite what the transcriber intended.

 

Right now I think Python/regex are my way forward –other ideas would be appreciated, though.

Yvan Rose

unread,
Oct 22, 2020, 9:57:16 AM10/22/20
to ph...@googlegroups.com

Hi Nina,

Your reply points at discrepancies in IPA transcription standards between your target/actual forms, which will clearly yield analytic artefacts. Once you address this transcription issue (either remove the ligatures or add them everywhere), everything should fall into place nicely.

 

On this note, I would recommend against any post-hoc fix for issues in transcriptions (your primary data source). Instead, I recommend you fix the problematic transcriptions.

 

Does this help?

 

Good luck with the work ahead!

Nina R Benway

unread,
Oct 22, 2020, 11:00:32 AM10/22/20
to ph...@googlegroups.com

Thanks, Yvan. We are on the same page. Also, phonex had a much more elegant solution than regex for quantifying the scope of the issue – the  phonex expression (._.) pulls all the compound phones.

Yvan Rose

unread,
Oct 22, 2020, 12:34:14 PM10/22/20
to ph...@googlegroups.com

Excellent! Let me/us know if we can help in any other way.

Reply all
Reply to author
Forward
0 new messages