MI Diaries: ethical and practical challenges

1.1 The MI Diaries project

MI Diaries is a longitudinal panel study of Michigan English from April 2020 onwards, run by the Sociolinguistics Lab at Michigan State University (MSU). Participation is open to anyone aged 3 or older who is currently living in Michigan. Each week, diarists^[1] receive 8–12 diary prompts, which differ by age group (“kids”, ages 3–12, “teens” ages 13–17, and “adults” ages 18 and over). New questions are developed each week to keep diarists engaged, and are designed to elicit a range of speaking styles (Labov 1984).^[2] Diarists self-record and submit entries via our mobile app. Each week, we select 2–5 anonymized snippets to feature on our website’s ongoing archive. To date, we have received over 700 diary submissions from over 150 participants, adding up to over 9,000 min of audio diary. Data collection is ongoing.

Data collected from this project will enable us to track the retreat from the Northern Cities Shift occurring in Michigan English (see, e.g., Nesbitt et al. 2019; Wagner et al. 2016) and test existing hypotheses about sociolinguistic calibration and incrementation by pre-teen and teenage speakers (Holmes-Elliott 2021; Labov 2001), as well as post-adolescent life span change by adults (Buchstaller 2016; Sankoff and Wagner 2020).

Our focus in this manuscript, however, is on the ethical and practical challenges we faced in the development of this project. Where relevant, we refer the reader to the Supplementary Materials (SM) for more detail. Early in the fall of 2020, the project developed explicit mission, values, and vision statements (SM 1). We include our values in Table 1, as they have been critical in guiding our solutions to the challenges presented in this paper. In what follows, we begin with some technical challenges related to remote audio recording, before moving on to a variety of intertwined practical and ethical concerns.

2.1 Remote audio recordings

Sociolinguistic fieldwork typically involves the long-term immersion of the researcher in the speech community (Eckert 1989; Labov 1963; Rampton 2007). The challenges associated with approximating these conditions via remote means were significant. Practically, we needed to ensure sufficient acoustic quality, but social distancing prevented us from using high-end microphones. Participants also had to trust us enough to speak in an unmonitored style, despite it being impossible for us to build face-to-face rapport.

2.1.1 Acoustic quality

A number of very recent papers have addressed remote audio collection in great detail (e.g., Freeman and De Decker 2021; Hinskens et al. 2021; Leemann 2021; Miley et al. 2021), and we are reassured that our approach of employing the native lossless recording format from smartphones has been found to closely approximate a Zoom H4N recorder, particularly for normalized vowel measurements (De Decker and Nycz 2011). We refer the reader to Sanker et al. (2021) for a detailed breakdown of recording quality across devices for various phonetic features.

Originally, diarists self-recorded on their phones and uploaded recordings to the first author’s non-MSU cloud account. Unfortunately, the cloud system at MSU prohibits “anyone access” uploads, meaning that diarists personally uploading recordings to a cloud server was not a viable long-term solution. Initial conversations with MSU’s IT department turned us toward developing a web app. However, this had to be abandoned after a few weeks of work, when we discovered that long duration recordings (longer than 5 min) required a prohibitively long buffering time.

We ultimately turned to building a mobile app (Figure 1). Having an official app housed on Apple’s and Google’s app stores helped to lend legitimacy, promoting our project value of trustworthiness. But getting MSU to agree took some convincing: we had to demonstrate to our MSU Apple representative that a web app would be insufficient, and provide examples of other mobile research apps^[3] along with a plan for app maintenance and updates. We were able to use MSU’s nonprofit developer profiles, saving us several hundred dollars per year; this saving allowed us to hire a student developer to build the app itself. The app is very simple, but sufficient for our needs.^[4]

Figure 1:

Screenshots showing the app interface for MI Diaries. Diarists can create multiple profiles on a single device, allowing parents to use the same device for multiple children. Diarist recordings are sent to our secure server; diarists may also optionally include images with their submission.

2.2 Recruitment and retention

For a longitudinal study like MI Diaries, participants must be over-recruited (since many will be lost to attrition over time) and, wherever possible, retained. These imperatives have shaped our decisions regarding recruitment techniques. But they have also informed a myriad of efforts to make MI Diaries “sticky” for participants, from easy submission of diaries to the creation of a sense of community (SM 4). To date, the project has recruited nearly 700 diarists, and received audio diaries from just over 150. Of these, roughly half are regular contributors. Although comparisons are hard to draw, this is a good retention rate for a longitudinal sociolinguistic study; compare, for example, 50% for the Montreal French study (Sankoff and Wagner 2020) and 28% for the Springville study (Cukor-Avila and Bailey 2018).

2.3 Compensation

From the beginning, we felt it was important to compensate participants (Charity Hudley et al. 2020). Compensation is both a participation incentive, and a recognition of diarists’ contributions. This became even more important as the social disparities around COVID became evident early in the summer. While we were fortunate to have access to some internal funds, we could not know how long the pandemic might last, nor how long the funds would survive the dire loss of revenue that universities were suddenly facing. It was imperative that we find the right balance between appropriate participant compensation and sensible management of our funds. An early-abandoned idea was to run a periodic raffle for participants who submitted diary entries, and send a gift card to the single winner each period. Unfortunately, Michigan law put a stop to this; while running one raffle would be fine, running repeated raffles would have required us to apply for a state lottery license.

Our solution was to offer a small gift card ($5) as a thank-you to participants who submitted at least 15 min of audio diary in a given two-week period. This presented several benefits. First, we feel the relatively high hourly rate ($20 per hour) is fair compensation and can help attract diarists with limited time and resources. Second, regular bi-weekly compensation encourages longitudinal participation. Finally, diarists can freely submit longer entries if they wish, without negatively impacting our budget. Feedback from diarists in the early weeks showed that some diarists were intentionally ending their recordings just shy of 15 min, out of a desire to save the project money. This prompted us to incorporate an option during sign-up for diarists to “pay it forward” to another participant instead.

2.4 Funding

Like many rapidly developing research projects, MI Diaries began without directed funding.^[9] Although we invested as much time as possible in applying for external awards, the grant application and announcement cycle was too long for a project that had to immediately compensate participants and personnel. We were fortunate to have some internal funds available. The first author was also awarded a Google Grant ($2,600) to support automated speech recognition (ASR). Our association with a major research university afforded us a financial safety net that is not available to everyone. But other cost-saving measures that we have employed are, we hope, transferable to other institutional environments. We briefly describe some of them here.

2.5 When project values conflict

We often need to navigate a conflict between our top project values of authenticity and trustworthiness. Many diarists share stories of loss and pain. One diarist shared how she broke down in unexpected tears over Halloween, longing for connection with her neighbors. Other diarists have shared stories about topics like infertility, mental health, and the death of family members. In keeping with authenticity, we do not want to paint a rosy view of the pandemic (or real life in general) by only sharing positive “feel-good” diary excerpts in our featured stories. However, we take care in the framing of these stories and prompts so they are less likely to be triggering to participants and listeners.

[associated audio-1-sneller.mp3 with “tears over Halloween” text]

At minimum, this means making sure that featured stories do not expose diarists – however indirectly – to mockery or condemnation. We also anonymize all identifiable words, like family members’ names. However, sometimes our desire to honor diarist entries creates a tension between project values. As an example, one week a diarist described the best gift he had ever given: a photo book for his girlfriend, full of pictures of her recently deceased cat, Rory. The diarist included the name and a picture of the cat. From a confidentiality standpoint, this made the diarist more identifiable, so we considered redacting both. But we ultimately decided to include both the picture and the name, because we felt the diarist had been very intentional about honoring Rory, and that anonymization could have been hurtful to the diarist. For other complicated cases, we adhere to existing protocols where possible. In December, we received a submission about participating in a vaccine trial, which included a photo of the injection. While the diarist’s face was not included, we nevertheless opted not to feature this story out of an abundance of caution regarding US health privacy laws (HIPAA). We have likewise trodden carefully with diary entries about receiving vaccines.

[associated audio-2-sneller.mp3 with “photo book” text]

Press interest in MI Diaries (Araneta 2021; Court and Querolo 2021; Goldberg 2020: inter alia; Herwick 2021; Jenkins 2021) has also presented ethical issues. Reporters, especially radio, TV, and podcast reporters, often ask to broadcast the featured diary excerpts from our website. Initially we resisted, since this went beyond our consent form’s statement that featured stories might be shared with academic audiences. We have since decided that because featured stories are accessible on our website to anyone, providing high-quality versions of those audio files upon request is within our ethical boundaries. But we never give out diarist contact information – despite being regularly asked by reporters if we can connect them to diarists for interviews.

3.1 Geographic accessibility

Since diarists self-record their own diary entries, the geographic barrier to participate is effectively nullified.^[10] Since team meetings also needed to be remote during COVID-related lockdowns, this meant that we were able to expand the team to include lab members who were living in other places, several students from other universities, and youth interns from across the state. We plan to continue running the project meetings remotely even after pandemic restrictions lift.

3.2 Interdisciplinarity

A big team, in turn, led to more opportunities for growth and connection with faculty, staff, and students in departments outside of Linguistics. This brought valuable nonlinguistic expertise to the core project team, as well as new interdisciplinary collaborations across departments at MSU. A September 2020 presentation of our remote data collection methods led to our app being adapted by a research team in Religious Studies that needed to pivot rapidly from in-person to remote ethnographic methods. We have plans to make our app fully open-source for research projects that may benefit from remote audio recordings. It has also been exciting to support colleagues in academia and in the community who are using the MI Diaries app, audio, transcriptions, and question prompts for a variety of aims across the digital humanities. At the time of writing we are actively encouraging creative uses of MI Diaries data and infrastructure via an upcoming symposium (MSU Digital Humanities Locus symposium, “Spin Offs: The MI Diaries Project as a starting Point”). Echoing Carmichael and colleagues (this collection), we strongly recommend to other linguists with publicly available audio collections that they consider proactively seeking “spin-off” users to achieve broader impact.

3.3 Improving on our foundation

Due to the rapid start of MI Diaries, and our lack of graphic design skills, we did not begin with a polished logo or website. Driven in part by the need to make an app logo easily visible, the student brand team led us in a complete visual overhaul of the project in the fall of 2020, which brought the project’s visual elements more in line with our project values. This resulted in a new website (Figure 2b) and set of logos (Figure 3b and c) that more clearly embody project values. We highlight these visual rebranding efforts here in order to emphasize how for us, starting the project at all – even with low-quality elements – in turn enabled us to draw outside expertise and subsequently improve.

Figure 2:

Updating to a Squarespace website (June 2020) resulted in a more attractive website and enabled a research assistant to update content without access to the first author’s personal website.

Figure 3:

The brand team scrapped our old logo (a), and designed a set of related logos (b–c) to embody the project values trustworthiness, authenticity, and “by Michiganders for Michiganders”.

We have also benefited from hiring undergraduates from MSU’s Experience Architecture major, who helped redesign the website and flyers (see SM 5). We are now developing collaborations (including for-credit internships) with other departments across MSU, including Digital Humanities and Advertising and Public Relations. We continue to look for collaborations that will enable our project to support student learning beyond the linguistic research.

3.4 Giving back to diarists

Finally, we are conscious that diarists are more than just data sources. Monetary compensation in the form of gift cards goes some way to demonstrating our gratitude (Section 2.3). But just as we have thought carefully about providing non-tangible benefits to our student team, so we have done the same for our diarists. We recommend, particularly for researchers working in communities experiencing a crisis, thinking about ways that participation in the project may provide benefits to participants. In June 2021, we sent diarists a survey asking what motivates them to participate. Below we describe a few of the top-ranked benefits.

3.5 Looking to the future

Restrictions on face-to-face human subjects research in early 2020 forced many linguists to adapt their pre-existing studies to remote methods (see, e.g., the papers by Hall-Lew et al. and Nesbitt and Watts in this collection). For MI Diaries, in contrast, the restrictions were a catalyst for creating a brand new project to meet the social moment (see also Abtahian et al. in this collection). We anticipate that the post-pandemic sociolinguistic research landscape will reflect both approaches. On the one hand, there is no substitute in sociolinguistics for the rapport that in-person fieldwork can build, and we eagerly look forward to resuming such fieldwork ourselves. On the other hand, we see a continuing place in the toolbox for both “face-to-face” remote interviews (e.g., via teleconferencing software) and self-recordings.

On balance, launching a new project rather than adapting an existing one has afforded us a lot of freedom to experiment methodologically. We have had to think in creative ways about how to gather vernacular speech data. This has involved adopting new-to-us technologies, establishing a branding and promotion team, collaborating with new partners, and mentoring new kinds of students. The lessons learned can be fruitfully applied to our own and other researchers’ future projects as well.

As a research project that emerged in response to a developing crisis, MI Diaries strongly believes that it is critical that we do more than benefit personally from a time of collective trauma. We continue to actively pursue opportunities to empower and support diarists, both through diary prompts and through community partnerships. We strive to include mentees – including youth – in all parts of the project. We also recognize our own privilege in being employed at an institution like Michigan State University, which has provided support in a number of ways, including funding, IT assistance, and wide promotion of the project. This support has given us a critical foundation to build on, and we hope that our reflections here can contribute useful insights for other linguists who may not have this type of institutional support.

Finally, we reflect that the content of MI Diaries, as a research project designed to document an ongoing disaster, carries its own value beyond sociolinguistic analysis (see, e.g., Carmichael et al. in this collection, for a rich discussion). The inherent interest in documenting an unprecedented historical event has attracted connections with staff, researchers, community members, and media who we might not have otherwise worked with. We are hopeful that through these collaborations, the MI Diaries corpus can be used to benefit the Michigan community in ways that we have not yet thought of. We stress that this kind of interdisciplinary collaboration is a particularly ripe opportunity for (socio)linguists collecting data during a disaster, whose work may be more readily understood by the general public as socially meaningful research.