So, I just acquired a goldmine of data that I can use for linguistic analysis. Sitting in my office are 452 cassette tapes, each containing at least 30 minutes of recorded interviews with an older folks from Heber City, Utah. And that’s about half of the collection: the other half is with a historian in Midway, Utah. So, I’m looking at roughly 400–500 hours of audio. Not sure how I’m going to process it all, but I wanted to kick off the beginning of this long-term project with a blog post describing the history of the tapes, why I’m interested in them, and speculations about the future.
I first heard about the tapes a little over three years ago. In January 2018, the LSA annual meeting was in Salt Lake City. Wanting to take advantage of the trip out there, I applied for and received a grant from the University of Georgia to collect audio in Heber City, aiming for multiple generations within a family to track language change over time. I decided on Heber partly because it was a region of Utah that had never been the subject of acoustic (let alone linguistic) analysis, as far as I know. My parents were living there at the time too, so they could hook me up with some potential contacts.
So on the morning of the first day of my fieldwork,Side note, it’s amazing that I heard about this goldmine literally through the first person I talked to while doing fieldwork? Who knew that there’d be such an amazing collection of audio sitting in someone’s basement nearby? In fact, there could be lots of collections like these, just collecting dust in people’s basements. All it takes is to find the right person! the first thing I did was go to the Heber Valley Visitor’s Center as a way to potentially find some contacts. Literally the first person I talked to told me about a man who had a huge collection of tapes. One person led me to another, and I was talking to an elderly man named Norm Kohler in his nursing home.
Norm was a beloved middle school teacher in Heber City in the 1980s and 1990s. As a history project, he had each of his students get a cassette tape and interview a grandparent. I don’t know what the interview questions were, but I think they mostly concerned life in Heber Valley. He kept all the tapes his students turned in and, over the course of two decades, he ended up with over 1200 interviews! Norm intended to compile them and put together an oral history of the town, but unfortunately was unable to do so. So, just weeks before I met him, he decided it was best to return the tapes to the family members’ of his students and the people they interviewed. So he put an ad in the paper and hundreds of people claimed their tapes and were able to hear their ancestors’ voices, perhaps for the first time.
However, not all the tapes were claimed. I was told a few hundred remained. So, after Norm passed away a few months later, his family held on to them for a while before finally donating them to the Midway Historical Society. (Midway is the town next door to Heber.) Several complications made it difficult for me to get access to the tapes, including outdated contact information on Midway’s website, the society going on an extended hiatus, me living in Georgia, and then covid. But I did my best to reach out to anyone who might know about where the tapes were being stored.
Finally, on Thursday this week, I was contacted by the historian in custody of the tapes. She asked if I was still interested in them, and I most definitely am! So we had a nice chat about what my goals were for them and what the goals were for the Historical Society, and we think there’s mutual interest in getting them digitized and transcribed. So, the next day, yesterday, she happened to be in Provo so she dropped off about half of the tapes—452 of them!—at my office!
So after three years of following tenuous leads, I finally have the tapes!
Why am I so interested?
I am a linguist, so why should I care about these tapes? Well, the obvious reason is that it’s a lot of audio. For my dissertation I analyzed about 40 hours of interviews and that was already a lot of data. This is at least 10 times the amount of audio. In fact, it’s about the size of the Digital Archive of Southern Speech, a subset of the Linguistic Atlas of the Gulf States, that I spent four years in grad school analyzing. So having access to this much audio is absolutely incredible
But it’s not just the amount of audio. There are dozens of oral history projects even in Utah. This particular set is attractive for several reasons:
The nature of the homework assignment ensured good metadata. A few tapes have already been digitized and they all start off introducing the interviewer (the middle-schooler) and the interviewee, with information like the date of interview, their age, and where they grew up.
Because these were all students in the same smallish town in UtahThere were about 4500 people living in Heber in the 1980s and 1990s, which means this sample represents like a significant chunk of the community! the sample will be relatively homogeneous geographically. While it doesn’t ensure that the interviewees (the grandparents) were from Heber or Heber Valley generally, my guess is that a significant number of them are.
Typically, interviews happen with a historian or someone that the interviewer is unfamiliar with. In sociolinguistics, it’s generally accepted that the degree of familiarity with the interviewer can have an influence on a person’s speech. In all cases with these tapes, the interviewer is a teenager and a grandchild of the interviewee. So that lowers the formality of the situation and will likely mean that the interviewees’ speech will be more casual.
Heber Valley has been the focus of very little acoustic research. There may be occasional interviews as parts of the Linguistic Atlas Project or the Dictionary of Regional American English, but no study, as far as I know, has focused on Heber. Instead, most research looks at people from Utah Valley and Salt Lake Valley. This collection of interviews will offer a new spot on the map of Utah dialectology and a nice point of comparison between more urban and more rural areas of the state.
I have virtually no metadata about the interviewees right now, but if their grandchildren were about 14 years old in the 1980s and 1990s, then the speakers in these tapes were born perhaps sometime between 1900 and 1940. There has been some research on the development of Utah English, mostly by David Bowie, but he acknowledges that it was based on public sermons given by upper-class white men. This collection offers a unique look into how other Utahns born around that time talked. And since I have some comparable data from contemporary Heber City residents, I can begin to look at language change in real time.
So there are lots of reasons for why I’m really interested in this collection of tapes. And that’s on top of the oral history the Midway Historical Society wants to create based on them.
Luckily, I’ve had some experience working on a project of this size. For four years at the University of Georgia, I was a part of the team that processed the Digital Archive of Southern Speech, which is a 367-hour subset of the Linguistic Atlas of the Gulf States. So I’ve sat in on transcriber training sessions, seen what kinds of obstacles get in the way of processing, managed thousands of files, and analyzed spreadsheets with a couple million acoustic measurements in them. However, that was only as a graduate student. I’m sure there’s a lot that goes on behind the scenes as a PI that I didn’t see.
Transcribers—Some back-of-the-envelope calculations suggest that I’ll need a sizable grant to get this all processed. Again, I don’t have definite numbers for anything, I know my 452 tapes are a little over half of them, so let’s say there are 700 tapes total. They’re all at least 30 minutes long and I know many went longer, so if I average say 40 minutes per tape, that’s 28,000 minutes or roughly 467 hours. I think the the transcribers for DASS averaged about 13 hours per 50 minutes of audio or so, but this audio is newer and I presume Utah transcribers will be more familiar with Utah speakers I think, so I’ll estimate 10 hours of work per tape. That’s 4670 hours of transcription. At $15 per hour, I’m looking at about $70,000 in student wages. Obviously, I can’t get that much coin internally so it sounds like this is only going to happen with an external grant.
Grad student workers—That’s of course assuming that the only wages I’ll need to pay for are transcribers. This might be getting into “If you give a mouse a cookie” territory, but it would be nice to have some grad students helping out with the project. At UGA, we had at least four and as many as six grad students involved in the project at a time. There was a lot of overlap between our duties, but very roughly speaking, one managed the transcribers, one managed the spot-checks, one managed the acoustic analysis, and one did miscellaneous duties. We were all involved in analysis, and a few others popped in for a semester or two to do additional analysis or perform other duties. To lighten my load, it would be handy to have perhaps three grad students manage the transcribers, check their work, and do the acoustic analysis. I’m fuzzy on what costs are associated with RA-ships at BYU, but I do know it’ll add significantly to the total cost of the project.
Time—How long will transcriptions take? I’ve done transcriptions and they’re soul-sucking work. Even when I was highly motivated to process my own dissertation data, that I collected myself, and under a bit of a time crunch, I could barely put in more than about two hours a day. I surely don’t expect undergraduate transcribers to do more than 10 hours a week. When motivated by money, I’ve seen some at UGA do more, but those students were exceptional. I’ll estimate five hours of work per transcriber per week. So under the assumption of 4670 hours of work total, that’s 934 transcriber-weeks. If a semester is fifteen weeks, that’s 62 transcriber-semesters. If I set a goal of getting all the work done in two years (six semesters if you include summers), it would take ten or eleven transcribers to do it in two years. Of course, these are all very rough estimates, but managing several tens of thousands of dollars and almost a dozen workers for two years is not something I expected to do right away!
Digitizing—Regardless of the cost, number of workers, and time involved, the first step of the process will be digitization. Fortunately, it sounds like the Office of Digital Humanities can take care of that for me! Wow! So my short term goal is to get a batch—maybe 30 or 50 tapes—done first. While they work on digitizing the next batch, I can get started on listening to the first few minutes of the completed tapes and extracting whatever metadata I can from them. Eventually, all the tapes will be digitized and I can have a more concrete idea of how much audio (and consequently, people, time, hours, and money) I’m looking at.
Metadata—After digitizing all of them, my next step will be to finish collecting the metadata. It’ll be nice to have a clear picture of birth years, genders, and birthplaces for all 700 or so people. The most likely scenario is that I won’t get an external grant because they’re extremely competitive, so I’ll have to prioritize which ones to transcribe first. The Historical Society would like to start with some of the prominent members of the community and descendants of the town’s founders. I’d like to find a balance of genders and birth years too, so we’ll probably settle on a subset that satisfies both of our needs. How big? I’m thinking between 35 and 70 (5% to 10% of the tapes). That’s a more reasonably-sized project that I could possibly get funded internally. It could provide me at least a beginning look at the speech community which would help seed an external grant.
Follow-up project?—In case I just need more data to analyze (ha!) wouldn’t it be cool to track down some of the tapes that were given away? Presumably, if an ad in the paper is what it took for the families to get them, then an ad might be a good place to start to find them. We’d digitize the tapes right there for people, give them a copy and return the tape to them of course, but then also add that to the collection for the oral history. I think it would be especially cool to interview those people themselves! That way we can get some contemporary data to compare the tapes to, as well as track change within the family. That’ll have to wait until I get NSF grant number two!
Publications—What’s the end goal? Well, I’ll obviously start cranking out some papers as soon as a reasonable amount of data has been processed. There is a lot going on in Utah English. Many of the stereotyped features are dying out, so these people may provide good acoustic data for what would otherwise be hard to study phonetically today. But there are also lots of other features that I believe are recent innovations, so if they’re infrequent or missing from these speakers, it’ll help establish the timing of when they did develop. Even before I had the tapes, I’ve been thinking a full analysis of this collection deserves a book-length treatment. It likely won’t get done before I’m up for tenure, but maybe it’ll go towards my application for full professor.
The history of the Kohler Tapes is pretty cool, and I’m lucky to be a part of the creation of an oral history of Heber City. It’s so satisfying teaming up with a historical society and finding ways to help the community I’m studying too. Linguistically, they’re interesting to me for lots of reasons, but I think everyone benefits from seeing these tapes get processed. As far as how I’m going to go about processing all of them, I really have no idea what I’m doing so there will be a lot of learning involved. But I’m excited to be involved and to have a clear research trajectory for the next decade or so!
Pillai scores don't change after normalization