Late-Romantic pieces often feature frequent enharmonic exchanges and adjacency of harmonically distant chords, contributing to a weakening of a single key as the agent for overall tonal coherence. Attempts to abstract from individual composers’ works to style-general concepts are the ‘Theory of Tone Fields’ (TTF) and ‘Neo-Riemannian Theory’ (NRT). While both promote astonishingly similar concepts, their analytical practices diverge. NRT focuses on transformations between pairs or sequences of chords (local dependencies). TTF, on the other hand, tries to reveal the presence of tone fields on different structural levels and insists on the importance of non-local dependencies. A recent model to formalize these in tonal music is the ‘Generative Syntax Model’ (GSM), but instances of extended tonal harmony are out of its scope.
The aim of the present paper is to introduce and integrate the aforementioned theories into a coherent framework to model syntactic relationships in late-Romantic music. The model is capable of expressing non-local dependencies, as well as alternative interpretations of ambiguous cases as distinct derivations of grammatical rules.
Analyzing a number of late piano pieces by Liszt shows that the interaction of tone fields reveals deeper hierarchical structures which are not apparent when only focusing on salient events on the musical surface. This attempt faces several challenges concerning central aspects of tonality in classical and early-Romantic music such as long-distance dependencies (e.g. tonic prolongation) and recursive embeddings (e.g. sequences of fifths). Addressing these challenges will shed light on the question whether these features still prevail in late-Romantic tonality.