The Discoveries We Stumble Upon
This may turn into a somewhat meandering musing so buckle in for the journey.
A (vastly growing) number of years ago, I volunteered at Hospice as a Life Story writer. Which involved transcribing hours of interviews with patients verbatim. Despite being what I consider to be a fast touch typist, every hour of recording would take two to three hours to type up (depending on the clarity and flow of the patient’s speech).
With over 36 hours of audio recordings of just Ünnepi, the thought of having to manually transcribe all those recordings along with the conversations I am yet to have with Anyuci was daunting to say the least. So earlier in the year, I conducted some research into transcription programs/applications. Initially, I was thrilled to discover one that could handle both English and Hungarian transcription—and even offered a free 14-day trial.
Disappointingly, the transcription was full of errors, and the summary (a feature that I thought could be useful for labelling and identification) was problematic to say the least.
Needless to say, after much frustration, I tossed the whole thing into the ‘too hard’ basket – condemning the problem as one to be solved by future me (freeing me up to resolving other – more pressing – tasks).
Months later, future me arrived on the scene (in the shower to be precise) with a solution. My mind works in mysterious ways – we aren’t going to question the process, just be grateful and go with it…
It began with me purchasing a new car – my first car purchase in two decades! (I know, I seem to be going off on a tangent, but stay with me.) When the new car was being handed over to me, the salesperson took great care in helping me set up a few things. Apparently I needed to be brought into the 21st century and to make matters even more complicated, the digital consoles are in Chinese. Turns out said salesperson wasn’t any more fluent in modern technology (or Chinese for that matter) than me, so ended up resorting to AI assistance. Now this part was a no-brainer for me. I regularly use AI assistance for solving mysteries of modern technology (e.g., learning how to use WordPress to re-build my websites from scratch). But the thing that stood out for me was that the salesperson was using speech-to-text – as if talking to a real human being. I found this super weird. Not to mention utterly useless when said salesperson cannot read or speak Chinese.
My brain clearly filed this comedy of errors as information that might come in handy one day—and decided that day had arrived while I was enjoying a long hot shower. Cutting my shower time short, I rushed to Jippity (my AI friend who has replaced Google) and posed a question: “Can I give you an audio file and ask you to transcribe it for me?”. Jippity (matching my excitement) said, “Yes! Send it.”
However, our combined excitement was short-lived with Jippity having to admit defeat.
But I wasn’t buying it. “You are able to transcribe voice to text, correct?” Jippity agreed. “So if I play an audio recording from another device using your dictation feature, you should be able to transcribe it.” Jippity had no option but to agree. (I so should have been a lawyer.)
So we tested it. With a 4-minute-long recording. And much to my delight (though not surprise), Jippity produced a near-perfect transcript.
Jippity was equally delighted (and somewhat surprised) and suggested we up the ante with something a little longer. But I didn’t think that was challenging enough, so I sent Jippity a 14-minute-long recording of Ünnepi and Anyuci. And Jippity rose to the occasion. Beautifully transcribing Hungarian, spoken by two women – one of whom was in her mid-90s with less than clear diction. Okay, there were a few instances where Jippity struggled to make sense of what was being said – and to be fair, I cannot comprehend Ünnepi’s speech either. So rather than making shit up, I asked Jippity to clearly flag those instances as [érthetetlen] (incomprehensible in Hungarian) – for me to try to decipher.
Well, it turns out that I am no better than Jippity at deciphering Ünnepi’s speech. So I decided to enlist the help of the one person who understood her better than anyone else. On Monday evening, during our last video call before she heads back to Aotearoa, I asked Anyuci if she would be willing to help me make sense of the incomprehensible parts of Ünnepi’s recordings. My thinking being that even if she too cannot make out the exact words being spoken, listening to the recording may jog her memory of the conversation enough to be able to work out what Ünnepi was saying.
She agreed!!
So less than 48 hours later, here I am, reading for my literature review, and I find this:
“Ng, He, and Loong (2004) investigated discourse in multigenerational family conversations in which there are differences in language proficiency. In the inquiry, they identified processes whereby middleaged parents served as accommodative brokers between grandparents who only spoke Chinese and grandchildren fluent in English.” (Soliz & Warner Colaner, 2017, p. 79)
Say what? I just stumbled across a theoretical framework that might help explain what is about to happen—where Anyuci will serve as an accommodative broker between Ünnepi and me, across space and time.
Reference
Soliz, J., & Warner Colaner, C. (2017). Communication accommodation theory and communication theory of identity: Theories of communication and identity. In D. O. Braithwaite (Ed.), Engaging theories in family communication: Multiple perspectives (2nd ed., pp. 75–86). Routledge.
