As part of a study into human vocal production and spoken language, a team of linguistic scholars and audio engineers at the University of Southern California has been examining the vocal patterns of a bi-lingual LA beatboxer using an MRI (Magnetic Resonance Imaging) scanner to properly understand this musical artform (see the work of Poolpo Joseph, below), which incorporates percussive sounds made using the mouth, throat, voice and nasal passage and sung or rapped lyrics.
The project team – Michael Proctor, Shrikanth Narayanan, Krishna S. Nayak, Dani Byrd and engineer Erik Bresch – want to understand how each of the elements is produced, to be able describe the elements in phonetic terms where possible and, finally, to examine the range of airstream mechanisms at play.
Their subject is a 27-year-old male professional singer from Los Angeles who had been working as an MC in a rap duo since 1995. He was a native English speaker who is fluent in Spanish and who raps in both English and Spanish.
Michael told Wired:
“We were fascinated by the art of beatboxing, and wanted to learn more about how these artists are able to produce such an astonishing array of sounds. As linguists, we wanted to compare the paralinguistic production of sounds in this type of musical performance with the ways that speakers of different languages produce speech sounds.”
The team made forty recordings using a custom ceramic noise-cancelling microphone systems as the subject lay in the MRI scanner. Each sequence, which lasted between 20 and 40 seconds, was performed at slow (at around 88 beats per minute), medium (95bpm) and fast (104bpm) rates. The data was captured using a realtime MRI protocol developed at USC specifically for studying speech. They also captured video footage of the lips, tongue and other vocal organs at work, which was particularly challenging.
The researchers observed 15 phonetically distinct percussion effects, and sounds that we familiar in other languages, but not in the subject’s – clicks used in Khoekhoe, spoken in parts of South Africa, Botswana and Namibia, and Xhosa from South Africa, and ejective consonants (sometimes known as non-pulmonics) common in Chechen, Quechua (Peru) and the British Columbian Nuxálk language.
The study is ongoing. The current findings are published in the February edition of the Journal of the Acoustical Society of America.