This week, I’m in Ann Arbor, Michigan at the New Ways of Analyzing Variation 53 conference! On Friday I gave a talk on the cord-card merger in Utah, based on the Kohler Tapes data.
Download the slides here
This is especially exciting for me because it’s one of the first times I’ve presented concrete results from my Kohler Tapes. This is a collection of cassette tapes containing audio of eighth graders interviewing older community members, who were often their grandparents. This took place in Heber, Utah, mostly in the 1990s. Norm Kohler was the middle school social studies teacher behind the project, and after his passing in 2019 and some collaboration with his estate and the Wasatch Historical Society, I now have the tapes in my office. Over the past year, a CS PhD student named Alex Lyman has been working on transcribing the tapes using an AI model called CrisperWhisper which keeps everything like filled pauses. We’ve found it does a remarkably good job on this audio. Alex is working on a paper that explains the nitty-gritty of that process in more depth, but if you’re interested in doing something similar, let me know. I then processed the 1200 or so files with MFA and new-fave, and I now have a huge spreadsheet of formant measurements to work with.
For this paper, I focused on the
The gist is that there’s tons of variation, both within and across speakers.
As for results, at the word-level, I found that some words tended to be more
They span much of the full range of the x-axis, and are relavely evenly spread out. However, the fact that there is a wide spread means that some words are more often higher and some words are more often lower. This regularity across speakers makes me think it’s a merger by transfer, though I’m still far from making an ultimate conclusion about that.
I found that previous /w/, as in war or warm tended to have higher vowels, as opposed to Bowie (2008)1, who found more lowering in that environment. This plot shows the result of a linear regression showing that. Values on the right correspond to being more
1 Bowie, David. 2008. Acoustic characteristics of Utah’s CARD-CORD merger. American Speech 83(1). 35–61.
For variation across speakers, I didn’t find any effect for gender and, surprisingly, no effect for age. Bowie (2003)2 finds increased amounts of
2 Bowie, David. 2003. Early development of the card-cord merger in Utah. American Speech 78(1). 31–51.
3 Cook, Stanley Joseph. 1969. Language change and the emergence of an urban dialect in Utah. University of Utah Ph.D. Dissertation.
4 Helquist, Val J. 1970. A study of one phonological variable in urban and rural Utah. Salt Lake City: University of Utah Unpublished Master’s Thesis.
One cool thing I found was a marked decrease in the amount of merger among people born in Salt Lake City compared to other people.
It’s too early for me to say much about this, but it does suggest that Salt Lake led the shift away from the merger and lends support to the idea that the merger is associated with rural places.
I played the carillon!
Besides the research part of the conference, I got to play the Beumont Tower Carillon5, which was super fun!
5 Thanks again, Jessi!
