by a muscular contraction (Demski et al. 1973; Bass 1989; Bass and Baker
1997). Such a system puts a clear physiological constraint on the highest
producible fundamental frequency (although some of the fastest-twitching
muscles in the animal kingdom are found in swim bladder muscles; Tavolga
1964). Only a few teleosts produce sound via expulsion of gas from the
swim bladder in a manner analogous with vocalization in tetrapods (e.g.,
minnows through the pneumatic duct, loaches through the anus; Demski
et al. 1973).
The oscillations of the source are never perfectly periodic. Even a nearly
constant fundamental frequency has small perturbations around the mean
frequency, which are called jitter by speech scientists (Lieberman 1961;
Baken 1987), and most vocalizations involve large, nonrandom changes in
fundamental frequency over time. Nonetheless, to a good approximation,
much of normal phonation can be idealized as periodic. In human speech,
such quasiperiodic phonation accounts for the vast majority of voiced
sounds and can be understood using standard linear systems theory and
Fourier analysis. However, in the past decade it has become increasingly
clear that certain types of broadband vocal phenomena are the result of
deterministic chaos in the (nonlinear) dynamics of the vocal source (see
Herzel 1996 for an introduction). In pathological cases in adult human
speech (Herzel and Wendler 1991), in human infant crying (Mende, Herzel,
and Wermke 1990), and in a variety of animal sounds (Wilden et al. 1998),
nonlinearities in the vocal-production mechanism can play an important
role in structuring the acoustic morphology of calls. In these cases, quasiperiodic phonation is replaced by one or more of a variety of irregular or
aperiodic phenomena, including period doubling, biphonation (the presence of two independent frequencies), and deterministic chaos (Fig. 3.2).
Although research into nonlinear phonation is still in its infancy, it appears
likely that such vocalizations play an important role in the communication
systems of many species (see also Tyack and Miller, in press).
The most common example of irregularities in the voice source is provided by calls such as those shown in Figure 3.2, which shows calls from a
normal rhesus macaque. Although the last call is “noisy” in the sense that
it lacks clear periodicity, it has too much spectral structure to be caused
simply by turbulent noise generated by vocal tract constrictions. Instead,
such calls appear to be generated by irregular opening and closing of the
glottis due to strong nonlinearities in the dynamics of the vocal folds; they
are examples of deterministic chaos in glottal dynamics. This hypothesis has
been confirmed in the human voice by direct observation of the vocal folds
with high-speed video endoscopy (Tigges et al. 1997) and is supported in
the case of mammalian vocalization by abundant evidence summarized in
Wilden et al. (1998). Other types of phenomena, such as period-doubling
bifurcations shown in the second call of Figure 3.2, lend further credence
to this hypothesis. Despite the ubiquity of such calls, they have received
little attention in the bioacoustics literature, perhaps because the appro3. Unpacking “Honesty”
77
1997). Such a system puts a clear physiological constraint on the highest
producible fundamental frequency (although some of the fastest-twitching
muscles in the animal kingdom are found in swim bladder muscles; Tavolga
1964). Only a few teleosts produce sound via expulsion of gas from the
swim bladder in a manner analogous with vocalization in tetrapods (e.g.,
minnows through the pneumatic duct, loaches through the anus; Demski
et al. 1973).
The oscillations of the source are never perfectly periodic. Even a nearly
constant fundamental frequency has small perturbations around the mean
frequency, which are called jitter by speech scientists (Lieberman 1961;
Baken 1987), and most vocalizations involve large, nonrandom changes in
fundamental frequency over time. Nonetheless, to a good approximation,
much of normal phonation can be idealized as periodic. In human speech,
such quasiperiodic phonation accounts for the vast majority of voiced
sounds and can be understood using standard linear systems theory and
Fourier analysis. However, in the past decade it has become increasingly
clear that certain types of broadband vocal phenomena are the result of
deterministic chaos in the (nonlinear) dynamics of the vocal source (see
Herzel 1996 for an introduction). In pathological cases in adult human
speech (Herzel and Wendler 1991), in human infant crying (Mende, Herzel,
and Wermke 1990), and in a variety of animal sounds (Wilden et al. 1998),
nonlinearities in the vocal-production mechanism can play an important
role in structuring the acoustic morphology of calls. In these cases, quasiperiodic phonation is replaced by one or more of a variety of irregular or
aperiodic phenomena, including period doubling, biphonation (the presence of two independent frequencies), and deterministic chaos (Fig. 3.2).
Although research into nonlinear phonation is still in its infancy, it appears
likely that such vocalizations play an important role in the communication
systems of many species (see also Tyack and Miller, in press).
The most common example of irregularities in the voice source is provided by calls such as those shown in Figure 3.2, which shows calls from a
normal rhesus macaque. Although the last call is “noisy” in the sense that
it lacks clear periodicity, it has too much spectral structure to be caused
simply by turbulent noise generated by vocal tract constrictions. Instead,
such calls appear to be generated by irregular opening and closing of the
glottis due to strong nonlinearities in the dynamics of the vocal folds; they
are examples of deterministic chaos in glottal dynamics. This hypothesis has
been confirmed in the human voice by direct observation of the vocal folds
with high-speed video endoscopy (Tigges et al. 1997) and is supported in
the case of mammalian vocalization by abundant evidence summarized in
Wilden et al. (1998). Other types of phenomena, such as period-doubling
bifurcations shown in the second call of Figure 3.2, lend further credence
to this hypothesis. Despite the ubiquity of such calls, they have received
little attention in the bioacoustics literature, perhaps because the appro3. Unpacking “Honesty”
77