What does a monophthongal vowel look like versus a diphthongal vowel in F1/F2 space? Well, I guarantee you, the difference is not as easy to interpret if you only have 2 or 3 measurements per vowel as is the norm. The two graphics I have inserted below (created using R) are what they look like when you take 21-time normalized measurements across the duration of the vowel.
First is a Loess curve created from several hundred BAT tokens extracted from sociolinguistic interviews with 18 African American speakers:
And here is the front lax vowel system (again several hundred tokens for each class) extracted from sociolinguistic interviews with 5 Southern white speakers:
I don’t know about you, but I find this particularly cool! Note that the space between nodes represents rate of spectral change. For African Americans it appears there is a definable steady state but the same isn’t true for Southern white speakers (at least for BET and BAT).
Finally, here are the front lax vowels for both speaker groups (African Americans in green and Southern whites in blue). This is the exact same data presented in the above two graphics except I used a generalized additive model. Monophthongized African American front lax vowels are characterized by parallel F1/F2 movement. Click on the image for a bigger graphic.
This is just a tiny sample of what I’m currently working on; I’m so fascinated by it I couldn’t help but share. Enjoy!
I have a lot of tweaking I plan to do to this methodology (including anchoring the onset/offset to better account for effects of neighboring phonetic environments), but I’m seeing it headed cool places even now.
ETA: This is completely something I’m experimenting with, if I hadn’t made that clear. Comments/suggestions/emails are especially welcome for this reason!
Megan L. Risdal