feature. Each presumed that perception begins with a speech signal, well composed and fit to analyze. This common premise shared by otherwise divergent explanations of perception obliges the models to admit severe and unintended constraints on their applicability. To exist within the limits set by this simplifying assumption, the models apply implicitly to a world in which speech is the only sound; moreover, only a single talker ever speaks at once. Although this designation is easily met in laboratory samples, it is safe to say that it is rare in vivo. Moreover, in their exclusive devotion to the perception of speech the models are tacitly modular (Fodor, 1983), even those that deny it.
Despite the consequences of this dedication of perceptual models to speech and speech alone, there has been a plausible and convenient way to persist in invoking the simplifying assumption. This fundamental premise survives intact if a preliminary process of perceptual organization finds a speech signal, follows its patterned variation amid the effects of other sound sources, and delivers it whole and ready to analyze for linguistic properties. The indifference to the conditions imposed by the common perspective reflects an apparent consensus at the time that the perceptual organization of speech is simple, automatic, and accomplished by generic means. However, despite the rapidly established perceptual coherence of the constituents of a speech signal, the perceptual organization of speech cannot be reduced to the available and well‐established principles of auditory perceptual organization.
Perceptual organization and the gestalt legacy
A generic auditory model of organization
The dominant contemporary account of auditory perceptual organization has been auditory scene analysis (Bregman, 1990). This theory of the resolution of auditory sensation into streams, each issuing from a distinct source, developed empirically in the cognitive era, though its intellectual roots run deep. The gestalt psychologist Wertheimer (1923/1938) established the basic premises of the account in a legendary article, the contents of which are roughly known to all students of introductory psychology. In visible and audible examples, Wertheimer described the coalescence of elementary figures into groups and contours, arguing that sensory experience is organized in patterns, and is not registered as a mere spatter of individual receptor states. By considering a series of hypothetical cases, and without knowing the sensory physiology that would not be described for decades (Mountcastle, 1998), he justified organizing principles of similarity, proximity, closure, symmetry, common fate, continuity, set, and habit. Hindsight suggests that Wertheimer framed the problem astutely, or so it now seems given our contemporary understanding of the functions of the sensory periphery that integrate the action of visual and auditory receptors (Hochberg, 1974).
Setting the indefinitely elastic principle of habit aside, the simple gestalt‐derived criteria of grouping are arguably reducible to two functions: (1) to compose an inventory of sensory elements; and (2) to create contours or groups on the principle that like binds to like. Whether groups occur due to the spectral composition of auditory elements, their common onset or offset, proximity in frequency, symmetry of rate of change in an auditory dimension, harmonic relationship, the interpolation of brief gaps, and so on, each is readily understood as a case in which similarity between a set of auditory sensory elements promotes grouping automatically. A group composed according to these functions forms a sensory contour or perceptual stream. It is a small but necessary extrapolation to assert that an auditory contour consists of elements originating from a single source of sound, and therefore that perceptual organization parses sensory experience into concurrent streams, each issuing from a different sound‐producing event (Bregman & Pinker, 1978).
In a series of ongoing experiments, researchers adopted Wertheimer’s auditory conjectures, and calibrated the resolution of auditory streams by virtue of the historic principles and their derived corollaries. For example, Bregman and Campbell (1971) reported that auditory streams formed when a sequence of 100 ms tones differing in frequency was presented to listeners. According to a procedure that has become standard, the series of brief tones was presented repetitively to listeners, who were asked to report the order of tones in the series. Instead of hearing a sequence of high and low pitches, though, listeners grouped tones into two streams each composed of similar elements, one of high pitch and the other of low pitch (see Figure 1.1). Critically, the perception of the order of elements was veridical within streams, but perception of the intercalation order across the streams was erroneous. In another example, Bregman, Ahad, and Van Loon (2001) reported that a sequence of 65 ms bursts of band‐limited noise were grouped together or split into separate perceptual streams as a function of the similarity in center frequency of the noise bursts. A sizable literature of empirical tests of this kind spans 50 years, and calibrates the sensory conditions of grouping by one or another variant of similarity. A compilation of the literature is offered by Bregman (1990), and the theoretical yield of this research is summarized by Darwin (2008).
Typically, studies of auditory‐perceptual organization have reported that listeners are sensitive to quite subtle properties in the formation of auditory groups. It is useful to consider an exemplary case, for the detailed findings of auditory amalgamation and segregation define the characteristics of the model and ultimately determine its applicability to speech. In a study of concurrent grouping of harmonically related tones by virtue of coincident onset, a variant of similarity in a temporal dimension, Dannenbring and Bregman (1978) reported that synchronized tones were grouped together, but a discrepancy as brief as 35 ms in lead or lag in one component was sufficient to disrupt coherence with other sensory constituents, and to split it into a separate stream. There are many similar cases documenting the exquisite sensitivity of the auditory sensory channel in segregating streams on the basis of slight departures from similarity: in frequency (Bregman & Campbell, 1971), in frequency change (Bregman & Doehring, 1984), in fundamental frequency (Steiger & Bregman, 1982), in common modulation (Bregman et al., 1985), in spectrum (Dannenbring & Bregman, 1976; Warren et al., 1969), due to brief interruptions (Miller & Licklider, 1950), in common onset/offset (Bregman & Pinker, 1978), in frequency continuity (Bregman & Dannenbring, 1973, 1977), and in melody and meter (Jones & Boltz, 1989); these are reviewed by Bregman (1990), Remez et al. (1994), and Remez & Thomas (2013).
Figure 1.1 This sequence of tones presented to listeners by Bregman and Campbell (1971) was reported as two segregated streams, one of high and another of low tones. Critically, the intercalation of the high and low streams (that is, the sequence: high, high, low, low, high, low) was poorly resolved. Source: Based on Bregman & Campbell, 1971.
Gestalt principles of organization applied to speech.
Because explanations of speech perception have depended on an unspecified account of perceptual organization, it has been natural for auditory scene analysis to be a theory of first resort for understanding the perceptual solution to the cocktail party problem (Cherry, 1953; McDermott, 2009), specifically, of attending to a single stream of speech amid other sound sources. However, this premise was largely unsupported by direct evidence. The crucial empirical cases that had formed the model had rarely included natural sources of sound – neither instruments of the orchestra (though see Iverson, 1995), which are well modeled physically (Rossing, 1990), nor ordinary mechanical sources (Gaver, 1993), nor the sounds of speech, with several provocative exceptions. It is instructive to consider some of the cases in which tests of perceptual organization using speech sounds appeared to confirm the applicability to speech of the general auditory account of perceptual organization.
In one case establishing grouping by similarity, a repeating series of syllables of the form CV‐V‐CV‐V was observed to split into distinct streams of like syllables, one of CVs and another of Vs, much as gestalt principles propose