Thesis Christ: How We Mishear Lyrics with Aidan Mokalla
From the “long list of Starbucks lovers” in Taylor Swift’s “Blank Space,” to the “beat of the tangerine” in ABBA’s “Dancing Queen,” it’s more likely than not that you’ve misheard a song lyric or two before. Senior Linguistics and Computer Science major Aidan Mokalla is no different, which brought him to study this phenomenon. Mokalla’s Linguistics thesis aims to explore the ways in which people mishear sounds in music and investigate whether people adapt their expectations of spoken language based on the context.
When people mishear things, the misheard phrases are largely determined by two factors: The first is the similarity of what we hear to what was actually said. As such, people are more likely to mix up similar-sounding words as they interpret the sounds they hear into understandable language.
The second factor that helps to explain this is how people consider how similar what we hear is to something that could logically be said. Importantly, the models we use to determine the possibilities of something being said can change based on context. With this in mind, Mokalla proposes that specific qualities in the context of listening to music can change the models by which people mishear lyrics, as opposed to spoken words. These misheard lyrics, known as mondegreens, make up the main subject of Mokalla’s thesis.
Language as used in singing is different from speaking in a variety of ways. “People when they sing, for example, they pronounce their vowels a bit differently, they pronounce certain consonants differently, like they will have centralized vowels and they’ll change up some stop consonants,” explained Mokalla. As a result, Mokalla theorizes that mondegreens are produced specifically with the different linguistic characteristics of music in mind. “What I hope to find is that the types of things that they mishear are better explained by a model that accounts for people adjusting their expectations to the specific constraints of music,” Mokalla stated.
Mokalla’s interest in mondegreens dates back to childhood, thanks to Sia’s song “Cheap Thrills.” “I was convinced that she was saying ‘We don’t need shalompalomps to have fun tonight,’” recalled Mokalla. “Then I found out years later that it was completely different. I was like, I need to know why this happened.”
Beyond just music, Mokalla sees applications from his thesis research to the question of whether language processing is bottom-up or top-down. In a bottom-up process, new experiences are responsible for our perceptions of information, whereas preconceived expectations shape perceptions in a top-down process. By investigating how the perception of language differs based on context, in this case music, Mokalla hopes to help answer these questions.
Currently, Mokalla is working with his advisor, Linguistics Department Chair Sameer ud Dowla Khan, and the Institutional Review Board (IRB) to get permission to move forward with a survey. In this survey, participants will be tasked with listening to pieces of music and reporting the lyrics they hear. Once approved, the survey will be released online to participants throughout the country.
A double major in Linguistics and Computer Science, also Mokalla’s current mind is his Computer Science thesis. He plans to study deep learning structures known as transformers, looking specifically at the property of induction heads and investigating the ways this property changes when it is given more information. Although the disciplines are different, both theses work with specific models for understanding processes and the ways in which these models change under different conditions. “I think that there is a need across many disciplines and many fields for people that have technical ability in both being critical of models and also using them to explain things that we don’t understand,” said Mokalla.
Much work remains in the thesis process, but Mokalla is excited for the future ahead. “I hope to keep learning about perception and language and music and how it ties together,” he concluded.