SMT Discuss has been recently upgraded. Due to increased password security, before logging into the upgraded site, registered users will need to change their DIscuss login password by using the "Forgot?" link found on the "Sign In" dialog.

WesterParse: proto-Schenkerian linear analysis of Westergaardian species counterpoint

This is a follow-up to last year’s announcement about an online tool for composing and evaluating simple species counterpoint in the version defined by Peter Westergaard in his 1975 textbook. The name I have given to the ongoing project is WesterParse. An article on the project is forthcoming in the proceedings of the 13th International Conference on Computer Supported Education (2021).

In addition to checking compliance with voice leading rules, the server-side software (written in Python and based on the music21 toolkit) parses the linear structure of each line in the contrapuntal exercise. The parser in effect looks for Ursatz structures and subsequent composing-out elaborations. The pedagogical tool, however, carries out this proto-Schenkerian analysis behind the scenes and does not reveal the result to the student; that is because I think the ability to analyze the structure of lines in counterpoint is a skill the student has to learn from practice. 

As a kind of proof-of-concept for WesterParse, I have now developed a companion website that allows a user to see the results produced by the parser: https://ada.artsci.wustl.edu/wp_web_project/corpus/

To test the validity of the WesterParse algorithms, I assembled a small corpus of 145 examples, about half of which are taken from Westergaard’s textbook. The user can select an example and then click on a button to parse the syntax. In addition to displaying the brief text report on the parse that users of the pedagogical tool would see, the corpus tool also displays the results in musical notation, in the style of Schenker. In cases where the lines are structurally ambiguous, the corpus tool will present multiple analyses. 

WesterParse produces some minor deviations from Westergaard’s analyses, attributable to minor differences in handling ambiguity. Westergaard, for example, allows rule A1 (S1 in my nomenclature) to attach to a nonfinal pitch, whereas WesterParse does not. On the whole, however, WesterParse reproduces Westergaard’s analyses, where he provided such. 

The corpus site also has links to the online pedagogical tool and the code documentation. And the latter includes a link to a list of Westergaard’s counterpoint rules.

If anyone has feedback on the corpus tool or has an interest in collaborating on further development of WesterParse, please contact me privately. 



Robert Snarrenberg

Associate Professor of Music and Comparative Literature

Washington University in St. Louis

Sign In or Register to comment.