Fleeting Moments: Inverted Taxonomies and Outlier Detection applied to Jazz Guitar Vocabulary

Analysis of improvisational language has often been led by observational taxonomies which place emphasis on common vocabulary and familiar structure, often to great effect (Owens. 1996, Givan. 2010). Correspondingly, practice, pedagogy and community are also led by similar taxonomies.

Such categorisation – comprising tendency, familiarity, and trope – help us build systems for codifying, qualifying and classifying the improvisatory lexicon of different idioms and/or performers, so that we might engage with them performatively, pedagogically, developmentally, analytically or otherwise. This kind of organisation often results in the ‘solidification’ or ‘re-reinforcement’ of genre-specific conventions, as well as the rigid cataloguing of attributes ascribed to an individual’s improvisatory ‘voice’. Ultimately, via the tacet prioritisation of predictability, this method encourages a generalisation of an improviser’s approach and the creation of a canonical schema of vocabulary.

What if however, in creating such generalised taxonomies we are overlooking fundamental characteristics – the ‘fleeting moments’ – which, whilst occurring less-frequently, or perhaps less obviously, are in fact those which determine the true colour, depth, nuance and holistic DNA underlying an improvisatory approach?

In data-mining applications, ‘outlier detection’ is considered a vital part of the modelling and analysis process. Ben-Gal asserts that, ‘One of the first steps towards obtaining a coherent analysis is the detection of outlaying observations. Although outliers are often considered as an error or noise, they may carry important information.’ (Ben-Gal, I. 2005; p.131) By adapting outlier detection and applying it to the analysis of improvisatory approaches (more specifically jazz-informed guitar-based improvisation), this paper will eschew the conventional hierarchical approach of defining an improviser by the what is most common, and instead use an inverted taxonomy to explore the significance of any anomalistic, irregular, misplaced, mistaken, trans-idiomatic, anachronistic, and/or forgotten vocabulary; thus developing a broader framework for understanding the improviser’s approach.

By conducting qualitative musical analyses of the improvisatory styles of an array of guitarists working across the jazz landscape – whilst prioritising the detection of outlier variables and considering their impact – this study will provide a fresh perspective on how we understand and interpret improvisatory vocabularies. Drawing from a range of theories and models used across other disciplines, this work aims to augment the conventional schemes applied in jazz and improvisatory analysis and, by expansion, contribute to the wider discourse surrounding multimodal and holistic methodologies of jazz analysis.