Getting consonantal context for vowels

[carissa abrego]

Hi everyone, 

I'm doing a multi-speaker corpus study on vowels using FAVE-extract, which is working great for us so far. However, I also want to know which consonants immediately precede and follow each vowel. Has anyone looked at something similar using automated methods?  Is there a script out there that can take an aligned text grid as input and give me the preceding and following segments for all vowels (in a transcript, or in a tier)?

 I would hope to be able to add that data as 2 new columns (say "preceding" and "following") for each vowel token.  My corpus is very large, so although I do have the word-context information thanks to FAVE-extract, listing consonants by hand won't do, and I'd rather not reinvent the wheel if someone already has a similar script. Thanks for any leads or tips!

Best,

Carissa

[Josef Fruehwald]

Hi Carissa,

Sorry for the delayed response! FAVE-extract in the most recent FAVE release (v1.2 https://github.com/JoFrhwld/FAVE/releases/tag/v1.2) does exactly this. The following columns are added to the text output:

• pre_seg: The preceding segment on the phone tier
• fol_seg: The following segment on the phone tier
• context: Where in the word the vowel appears (initial, final, internal, coextensive)
• pre_word_trans: full CMU transcription of the preceding word
• word_trans: full CMU transcription of the target word
• fol_word_trans: full CMU transcription of the following word

As well as more human readable versions of the plotnik coding.

-Joe

[carissa abrego]

Yes thank you!  Now it was my turn not to respond--I actually figured this out soon after I asked. Apparently, the last time I downloaded FAVE was two days before the release of 1.2.  Guess that's what I get for not pulling from github.  Thanks!