Vocal signals can be regarded as special building blocks of behaviour. In most cases, such signals occur boutwise and in a clear serial organisation. Depending on the composition of signal series, one can distinguish between homotype and heterotype building blocks. In this paper, we present software designed to analyse rules of both the organisation and the dynamics of series of acoustical behaviours.

In the study reported here we investigate aspects of the vocal repertoire of the Arabian Babbler (Turdoides squamiceps). Similar to other birds, its repertoire comprises several highly distinctive and easy recognisable call types, though calls of the same type often show considerable variation (Figure 1). Beneath investigations of the phenology of call type usage in different situations, we particular focus on this variation in calls and call series, and the possible relation between this variation and, for instance, the context or the caller's social status.

Most of the calls concerned are more or less tonal with much emphasis on the fundamental (or sometimes the first harmonic). So far, only few attempts have been made to analyse the acoustic structure of such vocalisations in detail, and, probably due to the difficulties of (objective) measuring, in most studies mainly straightforward parameters such as maximum frequency or call duration were used. Since in the current study we wanted to quantify call parameters in more detail we developed software for that purpose. This program analyses spectrograms with a particular focus on the course of the fundamental.

Figure 1. Spectrogram of call series of an Arabian Babbler: (a) call type 'tzwick', (b) call type 'thrill'. Note especially the variation within series.

At the moment, the program allows to analyse spectrograms containing either a single call or a regular call series. If call series are analysed, it measures each call separately. Measures of call properties mainly focus on the fundamental's shape, and, besides straightforward parameters, include parameters that describe the fundamental's course in more detail or indicate features like its regularity. Since the general shape of signals may vary according to the type of the vocalisation, we vary the set of parameters according to the general structure of vocalisations. Additional parameters describe the energy distribution across time and frequency. Since we mainly analyse field recordings that may contain considerable amounts of background noise we also included measures that indicate the reliability of the analysis. Also an optional export of files containing the amplitude distribution over time, the location of the detected signal, and the detected course of the fundamental allows for a later validation of the values calculated.

We use the program in studies of mobbing vocalisations, long range vocalisations and call ontogeny. First results indicate that the microstructure of vocalisations was clearly related to several parameters of the context. For instance, parameters of mobbing calls were related to the kind of predator and its distance to the caller. Interestingly, a presumed increase in the danger of the situation had similar effects on different call types. The application of the developed software thus proved to be useful for elucidating variation in serial acoustic behaviours and its relation to other variables, e.g. the context.

The study was supported by funds of the GIF and the DFG (To 13/30-1).

© 2000 Noldus Information Technology b.v.