Markovian Versus Rhomboidal Patterning in the Song of Swainson's Thrush

1. This paper examines a claim of NELSON (1973) of rhomboidal patterning in the songs of Swainson's Thrushes reflecting supposed holistic control of the sequences. 2. The subjects were six Swainson's Thrushes recorded in the wild in Quebec. Samples from the six birds ranged from 40 to 346 songs. 3. We used a Ubiquitous Spectrum Analyzer, and Sonagraph to produce frequency-time displays, and the former to produce also amplitude-frequency spectra. 4. The songs consist of 5 to 7 consecutive syllables given in fixed sequence, except for occasional errors. Each bird had 3 to 7 song types, or particular sequences of syllables, also sung in fixed order. 5. Similar syllables may occur in different songs of individual birds. No convincing evidence of "recombination units" of syllable pairs was found. 6. Evidence essential to support the claim of rhomboidal pattern was found wanting. In particular a) the frequencies (pitches) of successive syllables do not fit a pentatonic scale; b) the validity of a keynote is doubted; c) the definition of rhomboidal patterning is so vague that even pseudorandom numbers yield results similar to those claimed for the thrushes. 7. Alternatively the Markovian model of sequences of events appears more suited to the songs of Swainson's Thrushes, based on the evidence of sequences of syllables and songs, and errors in these sequences.