The human vocal production system is similar in broad outline to that of other terrestrial vertebrates. All tetrapods (nonfish vertebrates: amphibians, reptiles, birds, and mammals) inherit from a common ancestor three key components: (1) a respiratory system with lungs; (2) a larynx that acts primarily as a quick-closing gate to protect the lungs, and often secondarily to produce sound; and (3) a supralaryngeal vocal tract which filters this sound before emitting it into the environment. Despite this shared plan, a wide variety of interesting modifications of the vocal production system are known. The functioning of the basic tetrapod vocal production system can be understood within the theoretical framework of the myoelastic-aerodynamic and source/filter theories familiar to speech scientists.

The lungs and attendant respiratory musculature provide the air stream powering phonation. In primitive air-breathing vertebrates, the lungs were inflated by rhythmic compression of the oral cavity, or “buccal pumping,” and this system is still used by lungfish and amphibians (Brainerd and Ditelberg, 1993). Inspiration by active expansion of the thorax evolved later, in the ancestor of reptiles, birds, and mammals. This was powered originally by the intercostal muscles (as in lizards or crocodilians) and later (in mammals only) by a muscular diaphragm (Liem, 1985). Phonation is typically powered by passive deflation of the elastic lungs, or in some cases by active compression of the hypaxial musculature. In many frogs, air expired from the lungs during phonation is captured in an elastic air sac, which then deflates, returning the air to the lungs. This allows frogs to produce multiple calls from the same volume of air. The inflated sac also increases the efficiency with which sound is radiated into the environment (Gans, 1973).

The lungs are protected by a larynx in all tetrapods. This structure primitively includes a pair of barlike cartilages that can be separated (for breathing) or pushed together (to seal the airway) (Negus, 1949). Expiration through the partially closed larynx creates a turbulent hiss—perhaps the most primitive vocalization, which virtually all tetrapods can produce. However, more sophisticated vocalizations became possible after the innovation of elastic membranes within the larynx, the vocal cords, which are found in most frogs, vocal reptiles (geckos, crocodilians), and mammals. Although the larynx in these species can support a wide variety of vocalizations, its primary function as a protective gateway appears to have constrained laryngeal anatomy. In birds, a novel phonatory structure called the syrinx evolved at the base of the trachea. Dedicated to vocal production, and freed from the necessity of tracheal protection, the avian syrinx is a remarkably diverse structure underlying the great variety of bird sounds (King, 1989).

Although our knowledge of animal phonation is still limited, phonation in nonhumans appears to follow the principles of the myoelastic-aerodynamic theory of human phonation. The airflow from the lungs sets the vocal folds (or syringeal membranes) into vibration, and the rate of vibration is passively determined by the size and tension of these tissues. Vibration at a particular frequency does not typically require neural activity at that frequency. Thus, relatively normal phonation can be obtained by blowing moist air through an excised larynx, and rodents and bats can produce ultrasonic vocalizations at 40 kHz and higher (Suthers and Fattu, 1973). However, cat purring relies on an active tensing of the vocal fold musculature at the 20–30 Hz fundamental frequency of the purr (Frazer Sissom, Rice, and Peters, 1991). During phonation, the movements of the vocal folds can be periodic and stable (leading to tonal sounds) or highly aperiodic or even chaotic (e.g., in screams); while such aperiodic vocalizations are rare in nonpathological human voices, they can be important in animal vocal repertoires (Fitch, Neubauer, and Herzel, 2002).

Because the length of the vocal folds determines the lowest frequency at which they could vibrate (Titze, 1994), with long folds producing lower frequencies, one might expect that a low fundamental would provide a reliable indication of large body size. However, the size of the larynx is not tightly constrained by body size. Thus, a huge larynx has independently evolved in many mammal species, probably in response to selection for low-pitched voices (Fig. 1A, B). For example, in howler monkeys (genus Alouatta) the larynx and hyoid have grown to fill the space between mandible and sternum, giving these small monkeys remarkably impressive and low-pitched voices (Kelemen and Sade, 1960). The most extreme example of laryngeal hypertrophy is seen in the hammerhead bat Hypsignathus monstrosus, in which the larynx of males expands to fill the entire thoracic cavity, pushing the heart, lungs, and trachea down into the abdomen (Schneider, Kuhn, and Keleman, 1967). A similar though less impressive increase in larynx dimensions is observed in human males and is partially responsible for the voice change at puberty (Titze, 1989).

Figure 1..  

Examples of unusual vocal adaptations among vertebrates (not to scale). A, Hammerheaded bat, Hypsignathus monstrosus, has a huge larynx (gray) enlarged to fill the thoracic cavity. B, Howler monkeys Alouatta spp. have the largest relative larynx size among primates, which together with the enlarged hyoid fills the space beneath the mandible (larynx and hyoid shown in gray). C, Male red deer Cervus elaphus have a permanently descended larynx, which they lower to the sternum when roaring, resulting in an extremely elongated vocal tract (shown in gray). D, Humans—Homo sapiens—have a descended larynx, resulting in an elongated “two-tube” vocal tract (shown in gray). E, The now extinct duckbilled dinosaur Parasaurolophus had a hugely elongated nasal cavity (shown in gray) that filled the bony crest adorning the skull.

Sounds created by the larynx must pass through the air contained in the pharyngeal, oral, and nasal cavities, collectively termed the supralaryngeal vocal tract or simply vocal tract. Like any column of air, this air has mass and elasticity and vibrates preferentially at certain resonant frequencies. Vocal tract resonances are termed formants (from the Latin formare, to shape): they act as filters to shape the spectrum of the vocal output. Because all tetrapods have a vocal tract, all have formants. Formant frequencies are determined by the length and shape of the vocal tract. Because the vocal tract in mammals rests within the confines of the head, and skull size and body size are tightly linked (Fitch, 2000b), formant frequencies provide a possible indicator of body size not as easily “faked” as the laryngeal cue of fundamental frequency. Large animals have long vocal tracts and low formants. Together with demonstrations of formant perception by nonhuman animals (Sommers et al., 1992; Fitch and Kelley, 2000), this suggests that formants may have provided a cue to size in primitive vertebrates (Fitch, 1997). However, it is possible to break the anatomical link between vocal tract length and body size, and some intriguing morphological adaptations have arisen to elongate the vocal tract (presumably resulting from selection to sound larger; Fig. 1C–E). Elongations of the nasal vocal tract are seen in the long nose of male proboscis monkeys or the impressive nasal crests of hadrosaur dinosaurs (Weishampel, 1981). Vocal tract elongation can also be achieved by lowering the larynx; this is seen in extreme form in the red deer Cervus elaphus, which retract the larynx to the sternum during territorial roaring (Fitch and Reby, 2001). Again, a similar change occurs in human males at puberty: the larynx descends slightly to give men a longer vocal tract and lower formants than same-sized women (Fitch and Giedd, 1999).

Human speech is thus produced by the same conservative vocal production system of lungs, larynx, and vocal tract shared by all tetrapods. However, the evolution of the human speech apparatus involved several important changes. One was the loss of laryngeal air sacs. All great apes posses large balloon-like sacs that open into the larynx directly above the glottis (Negus, 1949; Schön Ybarra, 1995). Parsimony suggests that the common ancestor of apes and humans also had such air sacs, which were subsequently lost in human evolution. However, air sacs are occasionally observed in humans in pathological situations, a laryngocele is a congenital or acquired air sac that is attached to the larynx through the laryngeal ventricle at precisely the same location as in the great apes (Stell and Maran, 1975). Because the function of air sacs in ape vocalizations is not understood, the significance of their loss in human evolution is unknown.

A second change in the vocal production system during human evolution was the descent of the larynx from its normal mammalian position high in the throat to a lower position in the neck (Negus, 1949). In the 1960s, speech scientists realized that this “descended larynx” allows humans to produce a wider variety of formant patterns than would be possible with a high larynx (Lieberman, Klatt, and Wilson, 1969). In particular, the “point vowels” /i, a, u/ seem to be impossible to attain unless the tongue body is bent and able to move freely within the oropharyngeal cavity. Given the existence of these vowels in virtually all languages (Maddieson, 1984), speech typical of modern humans appears to require a descended larynx. Of course, all mammals can produce a diversity of sounds, which could have served a simpler speech system. Also, most mammals appear to lower the larynx during vocalization (Fitch, 2000a), lessening the gap between humans and other animals. Despite these caveats, the descended larynx is clearly an important component of human spoken language (Lieberman, 1984). The existence of nonhuman mammals with a descended larynx raises the possibility that this trait initially arose to exaggerate size in early hominids and was later coopted for use in speech (Fitch and Reby, 2001). Finally, recent fossils suggest that an expansion of the thoracic intervertebral canal occurred during the evolution of Homo some time after the earliest Homo erectus (MacLarnon and Hewitt, 1999). This change may be associated with an increase in breathing control necessary for singing and speech in our own species.

See also vocalization, neural mechanisms of.