Creativity through Cantonese Text-setting: Algorithmic Approaches and AI Potentialities
Relying on pitch to differentiate word meanings, tonal languages such as Cantonese exhibit a distinct musicality; however, their musical expressions are often constrained by the pitch structure of the language. Numerous instances exist where songs feature mismatched settings: for example, Cantonese hymns may interpret ‘ lord’ (zyu2) as ‘pig’ (zyu1), and school songs may render ‘principal’ (haau6 zoeng2) as ‘flirty elephant’ (haau4 zoeng6). Therefore, research on the tone-melody interface is critical to ensuring the intelligibility of Cantonese musics. Studies on Cantonese opera and popular music have demonstrated that Cantonese texts are only intelligible in a musical context if the pitches of successive syllables are appropriately distanced. However, there has yet to be a comprehensive mapping of possibilities within contemporary Cantonese compositions to understand how composers navigate text-setting constraints.
This study employs empirical tests and musical analysis to define linguistic constraints as a source of creativity in Hong Kong contemporary music. Perception tests were conducted to identify optimal musical intervals and other relevant conditions that ensure Cantonese text remains intelligible within a melody. Using works by Hong Kong composers, this research illustrates how the restrictions of Cantonese text-setting function as a creative algorithmic procedure in contemporary music, at times utilizing hidden or unsung text as a creative device. Ultimately, this investigation theorizes the textual intelligibility of Cantonese music and develops a framework that facilitates the exploration of musical potentialities in otherwise unconstrained situations.
The findings of this project culminated in the creation of a web application capable of converting Chinese text input into intelligible melodies. The “Cantonese Melody Generator” reduces barriers to Cantonese text-setting and assists composers in producing intelligible music while exploring variations within the same text. Furthermore, the ongoing integration of the generator with large-language-model AI applications has expanded its capabilities in composing Cantonese lyrics, thereby opening new avenues for creative expression within this vibrant musical landscape.
Keywords: Cantonese, prosody, text-music relationship, text-setting constraints, Hong Kong
Biography
Kai-Young Chan holds a PhD in Music Composition from the University of Pennsylvania and has taught at The Chinese University of Hong Kong (CUHK) since 2016. His works, including orchestral, chamber, choral, and mixed media compositions, have been performed over 200 times in more than 20 countries by renowned ensembles like the Pittsburgh Symphony Orchestra and the Hong Kong Philharmonic Orchestra.
Chan has received the CASH Golden Sail Music Award and other honors, representing Hong Kong at international events such as the ISCM World Music Days and the International Rostrum of Composers. His compositions are released by labels like Navona and innova, with scores published by Edition Peters and Universal Edition.
At CUHK, Chan has earned the Faculty Outstanding Teaching Award and secured grants for music theory pedagogical tools. His research explores how Cantonese text-setting constraints inspire creativity and aims to enhance the representation of Cantonese-language contemporary music globally.
The Chinese University of Hong Kong, Hong Kong S.A.R., China – chan.kaiyoung@cuhk.edu.hk