Musical Syntax II: Empirical Perspectives

Efforts to develop a formal characterization of musical structure are often framed in syntactic terms, sometimes but not always with direct inspiration from research on language. In Chap. 25, we present syntactic approaches to characterizing musical structure and survey a range of theoretical issues involved in developing formal syntactic theories of sequential structure in music. Such theories are often computational in nature, lending themselves to implementation and our first goal here is to review empirical research on computational modeling of musical structure from a syntactic point of view. We ask about the motivations for implementing a model and assess the range of approaches that have been taken to date. It is important to note that while a computational model may be capable of deriving an optimal structural description of a piece of music, human cognitive processing may not achieve this optimal performance, or may even process syntax in a different way. Therefore we emphasize the difference between developing an optimal model of syntactic processing and developing a model that simulates human syntactic processing. Furthermore, we argue that, while optimal models (e. g., optimal compression or prediction) can be useful as a benchmark or yardstick for assessing human performance, if we wish to understand human cognition then simulating human performance (including aspects that are nonoptimal or even erroneous) should be the priority. Following this principle, we survey research on processing of musical syntax from the perspective of computational modeling, experimental psychology and cognitive neuroscience. There exists a large number of computational models of musical syntax, but we limit ourselves to those that are explicitly cognitively motivated, assessing them in the context of theoretical, psychological and neuroscientific research.