Modelling music with grammars

James Kippen & Bernard Bel

Modelling music with gram­mars: for­mal lan­guage rep­re­sen­ta­tion in the Bol Processor. In A. Marsden & A. Pople (eds.): Computer Representations and Models in Music, London, Academic Press, 1992, p. 207-238.

Abstract

Improvisation in North Indian tabla drum­ming is sim­i­lar to speech inso­far as it is bound to anun­der­ly­ing sys­tem of rules deter­min­ing cor­rect sequences. The par­al­lel is fur­ther rein­forced by the fact that tabla music may be rep­re­sent­ed with an oral nota­tion sys­tem used for its trans­mis­sion and, occa­sion­al­ly, per­for­mance. Yet the rules are implic­it and avail­able only through the musi­cians’ abil­i­ty to play cor­rect sequences and recog­nise incor­rect ones. A lin­guis­tic mod­el of tabla impro­vi­sa­tion and eval­u­a­tion derived from pat­tern lan­guages and for­mal gram­mars has been imple­ment­ed in the Bol Processor, a soft­ware sys­tem used in inter­ac­tive field­work with expert musi­cians. The paper demon­strates the abil­i­ty of the mod­el to han­dle com­plex struc­tures by tak­ing real exam­ples from the reper­toire. It also ques­tions the rel­e­vance of attempt­ing to mod­el irreg­u­lar­i­ties encoun­tered in actu­al performance.

Excerpts of an AI review of this paper (Academia, June 2025)

Summary of the Work

The man­u­script explores the use of for­mal gram­mar the­o­ries to mod­el impro­vi­sa­tion­al struc­tures in North Indian tabla drum­ming. Drawing an anal­o­gy to lan­guage, the authors dis­cuss how grammar-based rep­re­sen­ta­tions can cap­ture the rhythm, struc­ture, and con­straints of tabla musi­cal sequences, which often rely on mnemon­ic syl­la­bles (“bols”) rather than Western script. By intro­duc­ing and detail­ing the Bol Processor soft­ware, the authors demon­strate how the tool can gen­er­ate, ana­lyze, and com­pare tabla phras­es in ways that blend both deter­min­is­tic and prob­a­bilis­tic approach­es. Throughout the work, real-world exam­ples from per­for­mances and teach­ing tra­di­tions bol­ster the the­o­ret­i­cal explanations.

Strengths

Clear Theoretical Rationale

The authors pro­vide a thor­ough back­ground for why lin­guis­tic concepts—particularly gram­mat­i­cal formalisms—are suit­able for ana­lyz­ing tabla com­po­si­tions. They effec­tive­ly frame the sim­i­lar­i­ties between lan­guage rules and musi­cal rules, espe­cial­ly regard­ing how musi­cians dif­fer­en­ti­ate “cor­rect” from “incor­rect” sequences.

Comprehensive Treatment of Formal Grammars

The dis­cus­sion includes mul­ti­ple types of gram­mars (context-free, context-sensitive, mark­ing of struc­tur­al posi­tions, and pat­tern gram­mars) and shows how each can be adapt­ed or extend­ed to cap­ture nuances of tabla music. The step-by-step expla­na­tions of gram­mar deriva­tions, con­texts, pars­ing, and mem­ber­ship tests are detailed and instructive.

Practical Demonstration via the Bol Processor

By anchor­ing the dis­cus­sion in a work­ing piece of soft­ware, the man­u­script tran­scends the­o­ry and illus­trates prac­ti­cal method­ol­o­gy. Concrete exam­ples of how the Bol Processor is used—both in gen­er­a­tion (modus ponens) and mem­ber­ship checks—highlight the val­ue of the tool for eth­no­mu­si­co­log­i­cal field­work and analysis.

Engagement with Complex Musical Realities

The man­u­script acknowl­edges that musi­cal per­for­mance often devi­ates from neat­ly defined the­o­ret­i­cal mod­els. The authors address how irreg­u­lar­i­ties and “mav­er­ick” vari­a­tions might still be cap­tured or rec­on­ciled with­in the for­mal frame­work, although at the cost of increas­ing­ly com­plex grammars.

Probabilistic Extension

The move from pure­ly deter­min­is­tic gram­mars toward prob­a­bilis­tic gram­mars is a notable con­tri­bu­tion, espe­cial­ly giv­en impro­vi­sa­tion­al con­texts. Assigning weights to pro­duc­tion rules cap­tures more real­is­tic musi­cal sce­nar­ios, enabling the soft­ware to sug­gest typ­i­cal (rather than mere­ly pos­si­ble) musi­cal phrases.

Overall Assessment

The man­u­script pro­vides a robust, method­olog­i­cal­ly sound illus­tra­tion of how for­mal gram­mars can be adapt­ed to mod­el and ana­lyze the com­plex world of North Indian tabla impro­vi­sa­tion. Its detailed explo­ration of gram­mar types and pars­ing approach­es is com­mend­able and is sup­port­ed by real-world musi­cal exam­ples that demon­strate both the poten­tials and the evolv­ing nature of such com­pu­ta­tion­al mod­els. The Bol Processor, as pre­sent­ed, appears to be a valu­able con­tri­bu­tion not only for eth­no­mu­si­col­o­gists and com­pu­ta­tion­al musi­col­o­gists but also for any schol­ar inter­est­ed in struc­tur­al approach­es to musi­cal impro­vi­sa­tion. The incor­po­ra­tion of prob­a­bilis­tic rea­son­ing fur­ther deep­ens the system’s applic­a­bil­i­ty to performance-based con­text and high­lights how for­mal lan­guage mod­els can be refined through inter­ac­tive feed­back with expert practitioners.

Download this paper