Skip to main content
NCBI home page
The Analysis of Verbal Behavior logoLink to The Analysis of Verbal Behavior
. 2023 Mar 29;39(2):169–189. doi: 10.1007/s40616-023-00182-3

Revisiting Topography-Based and Selection-Based Verbal Behavior

Anna Ingeborg Petursdottir 1,✉, Einar T Ingvarsson 2
PMCID: PMC10697920  PMID: 38075505

Abstract

In topography-based verbal behavior, different antecedent stimuli control different topographies of responding, whereas in selection-based verbal behavior, different antecedent stimuli control the selection of visually distinct stimuli from an array of options. In this article, we point out three variable characteristics of selection-based behavior, highlighted by recent technological developments, that affect its similarity to topography-based behavior: The extent to which stimuli can be constructed from minimal units, the size and composition of the selection array, and the similarity of response-produced stimuli to verbal stimuli that are prevalent in the speaker’s verbal community. Although a distinction between topography-based and selection-based behavior has merit, particular characteristics of a selection-based verbal behavior modality may often be more relevant for researchers and clinicians to consider than its status as selection-based.

Keywords: Augmentative and alternative communication, Matching-to-sample, Selection-based behavior, Topography-based behavior, Verbal behavior

The distinction between topography-based and selection-based verbal behavior was introduced into the literature by Michael (1985). Most verbal behavior, according to this distinction, is topography-based, meaning that different antecedent stimuli control different topographies of responding. In vocal speech, responses under the control of different stimuli (e.g., saying “apple” versus saying “orange”) involve different movements of the lips, tongue and jaw; thus, they differ in topography. The audible response products differ as well (i.e., “apple” sounds different from “orange”); however, it is the differential response topographies rather than the differential response products that characterize the behavior as topography-based. Other common verbal behavior modalities also involve topography-based behavior. Writing is topography-based, as is the typing of a fluent typist who responds to different antecedent stimuli (e.g., those produced by their own covert verbal behavior) with different sequences of small finger movements and presses relative to the home position on a familiar keyboard. Sign languages, as well, involve topography-based behavior, in which the different response topographies consist of motor responses and facial expressions.

Selection-based verbal behavior, by contrast, exists primarily in the context of certain types of augmentative and alternative communication (AAC) systems in which the user communicates by pointing to or otherwise selecting pictures, symbols, or textual stimuli. Instead of controlling different response topographies, different antecedent stimuli control the selection (e.g., via pointing or touching) of visually distinct stimuli from an array of options. For example, a child who uses the Picture Exchange Communication System (PECS; Bondy & Frost, 1994) to mand for preferred items may mand for juice by picking up a picture of juice and handing it to an adult, and for popcorn by picking up a picture of popcorn. The response topography—picking up a picture and handing it to an adult—may be exactly the same in both cases, only the picture that is selected differs. Similarly, speech-generating device (SGD) applications typically involve selection-based behavior, such that the user communicates by selecting (e.g., by touch) symbols or letters on a screen. Purely selection-based verbal behavior may be rare as a form of human communication outside of AAC systems. However, selection of stimuli in match-to-sample (MTS) tasks is often used as a proxy for ordinary verbal behavior in research on derived stimulus relations (e.g., Dougher et al., 2007; Roche & Barnes, 1996) and in research on symbolic, or language-like, behavior in non-humans (e.g., Epstein et al., 1980; Kuroda et al., 2014; Wasserman et al., 2015). In fact, some of the most successful attempts to establish verbal behavior in non-human primates have employed selection-based communication systems (e.g., Savage-Rumbaugh et al., 1978; Segerdahl et al., 2005).

The purpose of this article is to revisit Michael’s (1985) distinction between topography- and selection-based verbal behavior, and point out some variable characteristics of selection-based behavior that may be of relevance to research and practice in behavior analysis. We will begin by reviewing the arguments of Michael (1985) and others for distinguishing between the two, and some of their implications. Next, we will consider some technological developments in both everyday communications and AAC systems that have introduced novel forms of topography- and selection-based verbal responding, sometimes blurring the distinction between the two. Finally, we will discuss how certain characteristics of selection-based behavior, highlighted by said technological developments, may serve to increase or decrease its similarity to topography-based behavior, and potential implications in basic and applied domains.

Distinction Between Topography-Based and Selection-Based Behavior

As we review Michael’s arguments for distinguishing between topography- and selection-based behavior, we will use the example of a topography-based and a selection-based tact evoked by the sight of a cat. The cat is a nonverbal stimulus that controls the emission of a tact response due to a prior history of reinforcement of that response in the cat’s presence (Skinner, 1957). In the topography-based example, the response consists of vocalizing the syllable “cat.” In the selection-based example, it consists of pointing to the textual stimulus CAT from among an array of textual stimuli (i.e., the selection-based tact can be thought of as a type of MTS arrangement in which the sight of the cat is the sample and the textual stimuli are the comparison stimuli; alternatively, the comparisons could be pictures or symbols). For simplicity’s sake, we will assume this response occurs without any autoclitic frames (e.g., “It is a…”); however, such frames could accompany the tact in either modality, assuming the selection-based system includes options for textual or other frame selection. It should be noted that the selection-based tact is not the same as a listener discrimination (a.k.a. receptive identification). In a listener discrimination, the role of the verbal and the nonverbal stimulus is reversed, such that a verbal stimulus (e.g., the spoken word “cat” or the textual stimulus CAT) occasions selection of a nonverbal stimulus (e.g., looking at or pointing to a cat).1

According to Michael (1985), one important difference between the selection-based and the topography-based example is that the topography-based tact response requires only a simple discrimination, such that saying “cat” is more likely to be reinforced in the presence of a cat than in its absence or in the presence of non-feline animals. By contrast, the selection-based tact response requires a conditional discrimination, in which the cat serves as a conditional stimulus that signals increased probability of reinforcement for responding to (i.e., pointing to) another stimulus (i.e., the textual CAT). Whereas the topography-based tact requires only successive discrimination among nonverbal stimuli, the selection-based tact also requires a simultaneous discrimination among all textual stimuli that may be available to select from. Conditional discriminations, of course, are not unique to selection-based verbal behavior; they are ubiquitous in topography-based verbal behavior as well (Axe, 2008). For example, regardless of whether the tact consists of saying “cat” or pointing to CAT, it is more likely to be reinforced in the presence of the verbal conditional stimulus “What is it?” than in the presence of “What color is it?” or “What is it doing?” Michael (1985) argued that nonetheless, selection-based verbal behavior always involves an additional layer of conditionality relative to its topography-based parallel.

Relatedly, Michael (1985) pointed out that selection-based, but not topography-based responding, requires a repertoire of scanning or searching. In order for the sight of the cat to evoke the selection-based tact, the speaker must scan the environment to search for the appropriate textual stimulus to select. By contrast, no such behavior is necessary for the sight of the cat to evoke the vocal response “cat.” In addition to the necessity of a scanning repertoire, the need for scanning each time a verbal response is emitted may introduce a delay between the antecedent stimulus and the response, potentially resulting in loss of control by the antecedent by the time the selection response occurs. Although topography-based verbal behavior does not require the acquisition of a scanning repertoire, it does require learning the topography of each response (Potter & Brown, 1997), whereas in selection-based behavior, the same topography of stimulus selection (e.g., pointing) is applied widely across stimuli.

Finally, Michael (1985) noted that in topography-based verbal behavior, there is point-to-point correspondence between the form and the product of the response. In the case of the vocal tact “cat,” the speaker engages in a rapid sequence of distinct behaviors that correspond to the beginning consonant, middle vowel, and ending consonant, producing an auditory stimulus that can be similarly deconstructed by a listener as a particular consonant–vowel–consonant sequence. In the selection-based example, there is no systematic relation between the form of the response and the response product (i.e., the particular stimulus that is selected); various different selection topographies can serve to select the textual stimulus CAT and each of these can also serve to select other stimuli. Relatedly, as later pointed out by Potter and Brown (1997), in topography-based behavior, each response form produces unique feedback to the speaker in the form of kinesthetic stimulation that is absent in selection-based behavior.

Implications for AAC Systems

The overarching point of Michael’s (1985) paper was that topography-based and selection-based verbal behavior were not completely “equivalent forms of the same underlying language processes” (p. 4) given the different discrimination requirements and response–stimulus correspondences revealed by a behavioral analysis. In the previous example, the topography-based tact “cat” and its selection-based counterpart are evoked by the same antecedent stimulus, but the response requirement differs along multiple dimensions. These differences, Michael hypothesized, are likely to be of particular consequence when teaching verbal behavior to individuals with severe language delays due to neurodevelopmental disorders. Specifically, selection-based AAC systems might disadvantage the learner due to the need for conditional discrimination and scanning repertoires, and due to imprecise automatic feedback based on lack of correspondence between response topography and response product. Indeed, difficulties with conditional discrimination acquisition are often observed in individuals with neurodevelopmental disorders (e.g., Kodak et al., 2015; Saunders & Spradlin, 1990). The practical implications of Michael’s distinction for selecting AAC modalities for non-speaking individuals were further discussed by Shafer (1993) and Potter and Brown (1997), along with additional considerations such as the need for external materials (e.g., pictures, communication boards, or electronic devices) in selection-based systems.

A comprehensive review of empirical research comparing topography-based with selection-based verbal behavior is beyond the scope of this article. We will note, however, that with respect to Michael’s (1985) primary concern (i.e., acquisition by non-speaking individuals with neurodevelopmental disorders), the literature is not conclusive. Potter and Brown (1997) reviewed nine published and unpublished studies, mostly translational in nature, that had compared acquisition of topography- and selection-based verbal behavior. Most reported faster acquisition in the topography-based than in the selection-based condition, and this difference was more pronounced when participants were reported to have neurodevelopmental disorders than in studies with typically developing individuals. In addition, some studies (e.g., Sundberg & Sundberg, 1990) reported greater accuracy on selection-based equivalence tests when baseline relations were topography-based than when they were selection-based. More recent studies that have focused on comparing AAC systems in applied settings, have produced more variable findings regarding ease of acquisition and other measures such as generalization, maintenance, and modality preference, with results often in favor or partially in favor of selection-based systems. Some of this research (e.g., Achmadi et al., 2014; Barlow et al., 2013; Couper et al., 2014; Gregory et al., 2009; Tincani, 2004; van der Meer et al., 2012) is difficult to interpret as relevant to Michael’s (1985) concerns due to lack of relevant discrimination requirements during the evaluation (i.e., a successive simple discrimination requirement in the topography-based condition and an added conditional discrimination requirement in the selection-based condition). When such discrimination requirements are not fully present, acquisition data may reflect the ease of acquiring specific topographies (e.g., sign emission vs. pointing) but may not reflect the ease of establishment of the types of discriminative control that is ultimately needed for effective communication (i.e., emitting a different response topography or selecting a different picture or symbol depending on which antecedent stimulus or motivating operation is present). However, even some studies that appeared to include all relevant discrimination requirements have reported faster acquisition of selection-based responding than manual signing (Adkins & Axelrod, 2001; Chambers & Rehfeldt, 2003; see also a translational study by Vignes, 2007). Recent reviews and discussion articles on different forms of AAC have emphasized variability across studies and across participants within the same study, and suggested that different systems may be best suited to different individuals (Bloh, 2016; Carnett et al., 2021; Gevarter et al., 2013). Accordingly, some recent studies have focused on the extent to which potential prerequisite skills may predict ease of acquisition of verbal responses in different topography- or selection-based modalities (LaRue et al., 2016; Valentino et al., 2019).

Implications for Use of MTS Procedures in Research on Symbolic Behavior

Others have considered how Michael’s (1985) distinction between topography- and selection-based verbal behavior may affect the applicability of basic research on stimulus equivalence and other derived stimulus relations to vocal or other common forms of verbal behavior (Hall & Chase, 1991; Polson & Parsons, 2000).

Research on stimulus equivalence is relevant to verbal behavior due to its focus on symbolic behavior (Sidman, 1994). In the basic laboratory paradigm, a training phase establishes overlapping conditional discriminations (baseline relations) among sets of stimuli, followed by testing of additional conditional discriminations among the same stimuli. For example, after acquiring baseline conditional discriminations among A and B stimuli (AB) and among B and C stimuli (BC), a participant may be tested for the defining features of equivalence relations (Sidman & Tailby, 1982), which are reflexivity (AA, BB, and CC relations, often omitted from testing), symmetry (BA, CB), transitivity (AC), and equivalence (CA). These relations are often taught and tested in MTS format and thus involve selection-based behavior. As Hall and Chase (1991) pointed out, a similar arrangement can be employed to train and test simple discriminations involving topography-based responding, which are more characteristic of common human verbal behavior (see also Pérez-González et al., 2008). To illustrate, Hall and Chase (1991) used the example of establishing topography-based intraverbal relations among English (A), Spanish (B), and French (C) words. In an AB trial for “cat,” the experimenter presents the spoken English word “cat” and reinforces saying “gato” (Spanish for cat), and in a BC trial, the experimenter presents the spoken word “gato” and reinforces saying “chat” (French for cat). Thus, the response-produced stimulus in the AB trial (similar to the positive comparison in an MTS task) becomes the antecedent stimulus (similar to the sample stimulus in an MTS task) in the BC trial. The experimenter can then test for emergent relations such as BA, CB, AC, and CA (as well as AA, BB, and CC as echoic relations).

Hall and Chase (1991) noted that this arrangement is not, in fact, completely analogous to the selection-based paradigm. One reason concerns the need to present additional stimuli at test. In a selection-based scenario, testing is accomplished simply by reconfiguring the role of A, B, and C stimuli as samples and comparisons in MTS trials. In the context of Hall and Chase’s (1991) example, if all of the stimuli were textual and all the responses were selection responses, then the BA relation could be tested by presenting the textual stimulus GATO as a sample and English words as comparisons. In the topography-based scenario, the experimenter cannot conduct analogous test trials simply by presenting the spoken word by itself; the experimenter also needs to present additional antecedents or contextual stimuli to restrict the participant’s response to the relation tested (e.g., through instructions like “say gato in English” versus “say gato in French” to distinguish BA from BC trials). These additional antecedents, of course, must already exert control over the emission of responses from different language sets, which may require pre-experimental training (e.g., Belloso-Díaz & Pérez-González, 2015; Carp & Petursdottir, 2012; Pérez-González et al., 2008). When testing selection-based intraverbals, by contrast, the experimenter restricts the participants’ response options in each trial through the presentation of comparison stimuli belonging to a single language set (e.g., CAT is a response option in a CA trial, but CHAT is not). Thus, no additional training or testing stimuli are necessary.

Another issue, according to Hall and Chase (1991), is that in the topography-based scenario, relations analogous to symmetry in an MTS task are not in fact symmetric because they do not involve the same response topography. In the above example, AB and BA relations involve the emission of different response topographies (“gato” and “chat”) as opposed to a single selection topography regardless of the stimuli involved. In the case of tacts and corresponding listener discriminations, the task changes entirely: If AB is a topography-based tact (e.g., saying “cat” as a result of seeing a picture of a cat), the BA relation is usually tested in selection-based form (e.g., by asking the participant to point to a picture of a cat).

Related to this point, Polson and Parsons (2000) reported three experiments in which college students were taught intraverbal relations between English and French words. In each experiment, participants were taught to either type (topography-based condition) or select (selection-based condition) English words given their French counterparts (Experiments 1 and 2) or vice versa (Experiment 3). Half of the participants in each condition then received an MTS (i.e., selection-based) test for symmetry or symmetry-like relations, whereas the other half received a topography-based test. Regardless of the teaching condition, performance consistent with symmetry was observed much more readily in the selection-based than in the topography-based test; an effect that was more pronounced in Experiments 1 and 2 than in Experiment 3. Polson and Parsons (2000) interpreted these findings in the context of prior studies suggesting that novel topography-based intraverbals emerge less reliably than do novel stimulus relations in MTS tasks, and cautioned against “extrapolating from studies conducted using the MTS paradigm to real-life exemplars” (p. 123).

New Forms of Topography-Based and Selection-Based Verbal Behavior

Since the publication of Michael (1985), there have been many advances in communication technology. By the early 2000s, text messaging via mobile phones had become a prevalent form of communication between individuals in many parts of the world (Gayomali, 2015). Text messaging on common phone models initially required using a multi-tap text entry technique on a numeric keypad (e.g., pressing “1” three times to type “c” on keypads configured for the Roman alphabet). Many experienced users were able to enter text rapidly with one thumb without looking at the keypad, suggesting they were engaging in topography-based verbal behavior similar to a skilled typist. Other users (e.g., the first author) did not acquire such proficiency and relied on continual scanning of the keyboard to locate needed letters and determine the number of taps needed to select them, similar to a novice typist who types with one finger while scanning a keyboard. Interestingly, although these novice forms of typing and keypad text entry would seem to meet the criteria for selection-based behavior (i.e., a single response topography is used to select letters from an array), they partially share one of the characteristics of topography-based behavior mentioned by Michael (1985): Although the response product does not have point-to-point correspondence with a particular response topography, it does have point-to-point correspondence with the selected letters.

Text entry technology on mobile smartphones has since evolved in ways that further blur the distinction between topography- and selection-based behavior. A common text entry method on current mobile phones involves use of a small touchscreen QWERTY keyboard along with (optional) predictive text technology in which several options for completing a word are present on the screen. The user can tap one of these to select, or ignore them by continuing to type on the keyboard. Use of this technology may combine features of topography- and selection-based behavior. Experienced users entering text (e.g., with two thumbs) likely use different response topographies (i.e., different sequences of directional thumb movements and taps) to construct different verbal stimuli (e.g., different words). However, the flat surface of the keyboard still requires looking and scanning to some extent, and use of the predictive text technology involves clear instances of stimulus selection.

Other technologies used with mobile phones and computers, such as speech-to-text and text-to-synthesized speech, serve to alter the specific forms of correspondence that may exist between a speaker’s response topographies and the response products that stimulate one or more listener. For example, a speaker’s vocalizations may not just produce auditory stimuli but also text stimuli as response products that provide feedback to the speaker, and a remote listener may respond to the text stimuli alone. Further, these technologies make it possible even for clearly selection-based verbal behavior (e.g., selection of pictures, symbols, or whole words on a screen) to produce stimuli (i.e., response products) that are similar or identical to those produced by topography-based behavior.

Parallel to these advances in everyday communications are advances in the development of AAC systems. Gaze-controlled communication systems, developed for individuals whose physical disabilities may prevent vocal or motor-based communication, typically involve stimulus selection from a computer screen, often via eye fixation (selection without fixation is also possible with more advanced technology, e.g., Kristensson & Vertanen, 2012). For some users, the stimuli available to select from may consist of pictures or symbols that represent whole words or morphemes (e.g., Borgestig et al., 2016) as in many traditional selection-based AAC systems. However, for users with writing skills, there are also multiple ways in which eye gaze can be used to construct verbal stimuli from minimal units (i.e., letters of the alphabet). For example, users can select letters from a virtual on-screen keyboard, which may also incorporate predictive features based on the individual’s background and past responding to speed up selection (Elsahar et al., 2019; Ward & MacKay, 2002). Additionally, software has been developed that permits forms of topography-based writing via eye-gaze (e.g., Bee & André, 2008; Wobbrock et al., 2007), although these methods have been found to require more training and be more time consuming than virtual keyboard methods (Porta, 2015). Another approach is switch scanning, which involves pressing a switch when a visual symbol (e.g., a transparent green box) passes over letters or symbols, or holding down a switch and releasing it when the box reaches the relevant letter or symbol. Technology is also emerging that enables users to activate communication systems through patterns of respiration (i.e., breath-activated systems) and brain signals (i.e., brain-computer interface; Elsahar et al., 2019). Regardless of the input method and type of selection unit, the response product can consist of on-screen text or it can be converted to speech-like auditory stimuli. Thus, virtually any form of responding can be converted into response products that are similar to the verbal stimuli that prevail in the larger verbal community.

Speech-generating devices (SGDs), also referred to as voice output communication aids, are forms of AAC that involve the production of a previously recorded or digitized spoken output through portable electronic devices (Gevarter & Zamora, 2018; Lorah et al., 2015; Rispoli et al., 2010). This technology is now fairly common and considered effective as an AAC option (Morin et al., 2018; Muharib & Alzrayer, 2018). The spoken output is clearly characteristic of topography-based verbal behavior, but categorizing the user activation responses is less straight-forward. On the one hand, available interfaces allow users to construct their communication by selecting or touching each letter in sequence, much like in traditional typing. On the other hand, users can select concrete or abstract symbols which correspond with words or other parts of speech. For example, the word “is” might be symbolized by the equal sign ( =), while a cat is symbolized by a picture of a cat. The typed word may also be presented along with the visual symbol or picture. Thus, the most common user interfaces involve some form of letter or symbol selection, but in theory, the user input could involve any of the modalities and formats described in the previous paragraph (e.g., eye-gaze selection; respiration, brain activation).

As discussed above, Michael (1985) suggested that one crucial difference between topography-based and selection-based behavior is point-to-point correspondence between the response form and response product, which is present in the former but not the latter. This distinction is blurred in the case of SGDs systems in which the user constructs the responses out of minimal units such as letters of the alphabet, such that there is point-to-point correspondence between the selected letters and parts of the speech output. Depending on the fluency of typing, the behavior may be selection-based, topography-based or both, such that the additional layer of conditionality noted by Michael (1985) may or may not be present. Additionally, regardless of the input mode, SGDs provide the user with feedback unique to each response in the form of the voice output, which Potter and Brown (1997) suggested as another characteristic feature of topography-based responding. Implications are discussed in the next section.

Variable Characteristics of Selection-Based Behavior

The technological developments reviewed above highlight some variable characteristics of selection-based behavior that may affect how similar or different it is from topography-based behavior (Table 1). First, topography-behavior usually involves fine-grained minimal units (e.g., speech sounds or letters) that can be combined in infinite ways, whereas selection-based behavior may or may not involve similarly fine-grained units (i.e., the minimal unit may be letters, or it may be a larger unit such as a whole word or a symbol representing a word). A related second point is that when selection-based behavior involves larger units such as symbols or pictures, the range of possible responses depends on the number of stimuli that are available to select from—computerized systems usually have the potential to permit speedier selection from a larger number of stimuli compared to non-electronic systems. Third, topography-based behavior typically produces stimuli that are similar to those produced by other speakers in the verbal community, such as vocal speech in a particular language, textual stimuli, or manual signs if used in a community of sign language speakers. This characteristic may be absent in certain forms of selection-based behavior (e.g., picture or symbol selection), but its absence is increasingly being remediated by technology. In this section, we elaborate on these three variable characteristics of selection-based behavior.

Table 1.

Variable characteristics of selection-based verbal behavior

Variable Description Sample research questions
Minimal units The smallest selection units may be at the level of whole words or phrases, or at the level of fine-grained atomic repertoires, such as letters of the alphabet If a system permits selecting from fine-grained minimal units (e.g., letters), does it matter if the user engages in selection-based or topography-based (e.g., fluent typing) behavior?
Size and characteristics of selection arrays When the smallest selection units are word-level or similar, a selection array may include a very large number of options to select from (e.g., some types AAC systems) or a much smaller number of options (e.g., educational software; MTS preparations in research studies) that constrain the user’s selections to particular words or particular types of stimulus relations Are differences between outcomes of topography-based and selection-based training related to the provision of response options in the latter? How does the size and composition of the selection array affect outcomes of selection-based training?
Conventional or non-conventional response-produced stimuli The stimuli that a selection response produces for a listener to respond to may be different from those that prevail in the verbal community (e.g., symbols that are pointed to) or they may be more similar (e.g., synthesized speech or text) Does selection-based production of conventional verbal stimuli affect acquisition or emergence of listener behavior with respect to conventional verbal stimuli?
Open in a new tab

Minimal Units

We previously introduced the notion of point-to-point correspondence between a response topography and its response product (e.g., in vocal speech or fluent, topography-based typing) in the context of Michael’s (1985) distinction between topography-based and selection-based verbal behavior. We also pointed out how some selection-based communication systems may involve a similar level of point-to-point correspondence between fine-grained units of a response and its product (e.g., between selected letters and resulting text or synthesized speech).

Verbal responses can also have point-to-point correspondence with the antecedent stimuli that evoke them, exemplifying what Skinner (1957) termed minimal response repertoires (see also Alessi, 1987). Examples include echoic and imitative repertoires, in which the form and sequence of an individual’s behavior matches (is formally similar) to stimuli produced by another person (Skinner, 1957). Minimal repertoires need not have formal similarity with the antecedent stimuli, however, as in taking dictation, reading text, and following instructions (Alessi, 1987; Palmer, 2012). An important aspect of minimal repertoires is their role in generative responding, in which unique stimulus configurations set the occasion for corresponding behavior that is unique and novel; a process referred to as recombinative generalization (Suchowierska, 2006). Palmer (2012) referred to these repertoires as atomic repertoires, and emphasized their importance for facilitating “… the acquisition of adaptive behavior in a natural context” (p. 67). As an example, a person with an echoic repertoire or a textual repertoire can acquire novel vocal tacts via transfer of stimulus control from echoic or textual responding, respectively, making it unnecessary to shape the vocal response via reinforcement of successive approximations.

To the extent that AAC systems include minimal units (or behavioral atoms, in Palmer ‘s terminology) that can recombine in novel and flexible ways, the distinction between topography-based and selection-based behavior may be less relevant, both from a conceptual and practical perspective. This is especially true when the AAC user has a fluent typing repertoire, the components of which may include robust textual and transcriptive repertoires. As noted above, this may remove the additional layer of conditional stimulus control required in more traditional selection-based systems.

The size of the minimal unit is a relevant consideration for at least two reasons. First, smaller units, or a fine-grained atomic repertoire in Palmer’s (2012) terminology, are likely to be more flexible and have greater capabilities for recombining in multiple novel ways. For example, if the minimal units consist of letters of the alphabet (or phonemes), the units could potentially recombine to form novel words that the AAC user has never heard before. The same would presumably be true of writing systems that use symbols to represent distinct syllables (i.e., syllabaries). If the minimal units are words, communicating a novel word may be impossible. Second, more fine-grained units are likely to have fewer total items, as illustrated by the fact that any alphabet or writing system will have a finite and limited number of letters or symbols, while the number of potential words and sentences in any language is virtually infinite. It follows that smaller, more fine-grained units will generally be easier to display in simple arrays, obviating the need for navigating between multiple screens containing categories of symbols that represent more “coarse” units such as symbols that represent words or phrases. Thus, smaller minimal units have distinct advantages for practical, flexible, and generative communication.

It also might be noted that if a selection-based system involves minimal units, then in the sense described by Hall and Chase (1991), speaker and listener relations are inevitably asymmetrical, as they involve different response topographies. The typed selection-based tact CAT will involve selection of three distinct letters on the keyboard whereas the corresponding selection-based listener response (i.e., responding as a listener to the textual stimulus CAT) will typically entail a single selection response (e.g., picture selection). Again, the involvement of atomic repertoires seems to increase similarity to topography-based behavior, regardless of the extent to which the user’s selection of the minimal units involves different response topographies as in fluent typing. However, many of the individuals that need AACs in order to communicate may lack the necessary prerequisite and component skills in order to acquire fluent, fine-grained atomic repertoires.

Characteristics of Selection Arrays

In the absence of a fine-grained atomic repertoire by the user, the minimal units of a selection-based AAC system are typically at the level of words, such that the user selects from pictures or symbols where each stimulus represents a particular word. Another variable characteristic of selection-based behavior is the size and composition of the selection array in these cases. A beginner learning to use a selection-based system may initially be presented with a very limited number of options to select from (e.g., Frost & Bondy, 2002), similar to MTS preparations in laboratory settings, in which the comparison array often consists of two or three stimuli per trial.2 However, experienced users will typically have a much larger array of stimuli available to select from; for example, a PECS binder or software program containing a large number of different pictures. Further, unlike common practices in MTS research on stimulus equivalence and related phenomena (Green & Saunders, 1998), the array of options available to a user of a selection-based AAC system are not typically restricted to stimuli appropriate to a particular type of relation (e.g., C stimuli when an AC relation is being trained or tested).

Characteristics of selection arrays are potentially important when considering the applicability of research employing MTS preparations to prediction and control over common forms of verbal behavior, such as vocal speech. Concerns about such applicability have previously been discussed in the context of the distinction between topography-based and selection-based verbal behavior (Hall & Chase, 1991; Polson & Parsons, 2000). However, for the reasons described above, selection-based behavior does not necessarily share the characteristics of experimental MTS preparations that are of concern. In the context of the example of teaching intraverbal relations between English, Spanish, and French words (Hall & Chase, 1991), consider a student who engages in purely selection-based typing and has learned to select individual letters to type GATO in the presence of the textual stimulus CAT (AB) and to type CHAT in the presence of GATO (BC). In this case, just as in topography-based behavior, it is not possible to present the stimulus GATO and expect the response CAT (BA) rather than CHAT (BC) in the absence of additional antecedent stimuli. Even if no minimal units are involved, the same is true if the selection array contains a variety of whole words that include both CAT and CHAT. The need for additional stimuli to restrict the student’s response to one language or another is not unique to topography-based behavior; it arises in selection-based behavior as well as soon as the selection array is expanded beyond stimuli appropriate to a particular type of relation.

In addition to the composition of the selection array, its size may be relevant as well. Continuing with the same example, even if only English-language whole-word stimuli were available to select from, it seems possible that GATO might be more likely to evoke selection of CAT if the number of comparison stimuli is limited (e.g., CAT, DOG, DUCK) than in a communication system with a much larger array of response options.

At the conceptual level, the above-mentioned characteristics inevitably affect the similarity of particular selection-based arrangements to topography-based behavior. In studies that have compared topography- and selection-based behavior (summarized in an earlier section), it is possible that some differences in acquisition and emergence of derived relations were related to specific characteristics of the selection arrays, or to the fact that selection arrays in general constrain the range of possible responses. To provide an example, a previously mentioned study (Polson & Parsons, 2000) found that participants performed better on a selection-based test than a topography-based test for English–French and French–English intraverbal relations. In the selection-based test condition, four response options were presented in each trial, whereas no response options were provided in the topography-based condition. It is possible that the difference was not due to the topography-based nature of the participants’ typing, but rather the provision of response options. Perhaps participants in the topography-based conditions would have performed just as well if four options had been presented on the screen from which they could have selected one to type. To provide another example, Sundberg and Sundberg (1990) gave a selection-based equivalence test to participants with intellectual disabilities, and found better performance if baseline relations had been trained as topography-based than selection-based relations (see also similar results in comparative studies of emergent relations following topography-based tact instruction and selection-based listener instruction; e.g., Connell & McReynolds, 1981; Cortez et al., 2020, 2022). Would this advantage of topography-based instruction also be found if selection arrays were expanded in selection-based training, or response options provided in topography-based training? In this context, a sizeable literature exists on the differential effects of what is termed recall practice versus recognition practice (Rawson & Zamary, 2019). A training or test trial requiring topography-based responding without response options would be an example of the former, whereas provision of response options would make it an example of the latter, regardless of whether the response is topography- or selection-based.

Production of Conventional Verbal Stimuli

In common topography-based verbal behavior modalities (vocal verbal behavior, writing, typing), the speaker’s unique response-produced stimuli correspond to stimuli to which the same person responds as a listener on other occasions. In AAC systems, this is not necessarily the case to the same extent, because conversation partners don’t necessarily communicate with AAC users through the AAC system, and indeed, often speak to them vocally (Konstantareas, 1987). Therefore, AAC does not negate the utility of a listener repertoire with respect to vocal speech stimuli. In some AAC systems, responding as a listener to vocal speech may require responses to stimuli that differ from the response products of the user’s own verbal behavior (e.g., manual signs or symbol selection).

The extent to which, and the mechanism by which, selection-based behavior produces conventional stimuli (i.e., stimuli that are prevalent in the user’s verbal community) is another variable characteristic of selection-based behavior. As previously mentioned, modern technology permits automatic conversion of selection-based verbal behavior to speech or textual output. In addition, some selection-based systems produce a similar outcome through the mediation of another person. In PECS, for example, the listener (i.e., the recipient of the speaker’s picture communication) is instructed to respond to picture exchanges by tacting the selected stimuli (e.g., say “cookie” when the speaker mands for a cookie by handing over a picture of a cookie; Frost & Bondy, 2002).

Conceptually, there are at least two reasons why a match between verbal response products and vocal speech stimuli may be important. One reason concerns the relationship between speaker and listener behavior in selection-based behavior without voice output. A selection-based tact of a cat, for example, may involve a relation between a nonverbal stimulus (a cat) and a symbol, whereas a corresponding listener discrimination with respect to vocal speech involves a relation between a vocally produced stimulus (“cat”) and the nonverbal stimulus. Teaching one of these relations in isolation cannot be expected to result in emergence of or facilitate acquisition of the other as it often does when the speaker’s response is vocal (e.g., Sprinkle & Miguel, 2012; Su et al., 2019); it is necessary to either teach both directly or establish a third relation involving the vocally produced stimulus and the symbol. When selection-based behavior permits voice output, the addition of this response-produced stimulus would seem to increase the potential symmetry-like properties of a tact and a corresponding listener discrimination, potentially creating some degree of interdependence between the two (i.e., teaching one could bring about emergence or facilitate acquisition of the other). Using a PECS tablet computer application with voice output, Ganz et al. (2015) observed an increase in correct listener discriminations involving vocal stimuli, albeit not to criterion level, after teaching corresponding selection-based tacts to a pre-school age child diagnosed with ASD. However, we are not aware of any other empirical research that has addressed this issue.

Second, young children sometimes begin to spontaneously emit relevant vocalizations as a result of learning to mand using manual signs, PECS, or SGDs (e.g., Charlop-Christy et al., 2002; Tincani, 2004). It is likely that the vocal speech stimuli contingent on these AAC communication responses (e.g., the adult saying “cookie” while handing over a cookie following a successful mand) may be responsible for this effect in one way or another. For example, Petursdottir and Lepper (2015) speculated that the reliable pairing of the vocally produced stimuli with mand reinforcement may result in a conditioned automatic reinforcement effect (Sundberg et al., 1996). Interestingly, the automatic voice output associated with SGDs might be expected to either facilitate or hinder this effect. On the one hand, an SGD eliminates the need to rely on a human mediator to emit the appropriate vocal response, potentially resulting in a more reliable, higher-quality model (Gevarter et al., 2016). On the other hand, from the perspective of the automatic reinforcement hypothesis (Sundberg et al., 1996), the immediacy of the vocal SGD output could have two unintended effects on child vocalizations, given that the listener’s delivery of the reinforcer may be more temporally remote from the vocal stimulus than when the listener is also the source of the vocal stimulus. First, a delay between the two stimuli could hinder conditioning of the vocal stimulus as a reinforcer, whether the delay is intentional or unintentional. Second, the immediacy of the vocal stimulus could reduce motivation to vocalize (which results in hearing the reinforcing stimulus) during an intentionally programmed delay to reinforcer delivery. When teaching use of certain communication modalities (e.g., PECS), delays to reinforcement may be deliberately programmed specifically to promote learner vocalizations (e.g., Cagliani et al., 2017; Carbone et al., 2010; Greenberg et al., 2014; Tincani et al., 2006). The delay typically occurs between the learner’s non-vocal mand on one hand, and the simultaneous delivery of the vocally produced stimulus and the reinforcer on the other. However, with an SGD, the delay would occur between the non-vocal mand and vocally produced stimulus on the one hand, and the delivery of the reinforcer on the other. Having already heard the vocal stimulus (and perhaps being able to reproduce it via the SGD) could potentially affect the learner’s motivation to emit other responses (i.e., vocalizations) to produce the reinforcing auditory stimulus. That said, Gevarter et al. (2016) found that for two of four participants, a delay-to-reinforcement intervention component produced an increase in vocalizations along with SGD manding for two of four participants, without any delay between manding and vocal SGD output. Perhaps the delay is not of concern, but additional research may be needed.

Conclusion

In summary, Michael’s (1985) distinction between topography-based and selection-based behavior may have implications for the selection of communication modalities for individuals with language delays, as well as for the applicability of MTS research on derived stimulus relations to vocal or other topography-based verbal behavior. We have pointed out some variable characteristics of selection-based behavior that may also be of relevance to these two concerns, in part by affecting the degree of similarity between behavior that is primarily selection-based and behavior that is primarily topography-based.

We continue to fully agree with Michael (1985) that there are differences between topography-based and selection-based behavior, as well as on the potential importance of the differences he noted, such as the different discrimination requirements and prerequisite repertoires. However, we suggest that because of the many and growing number of varieties of selection-based or partly selection-based verbal behavior, the distinction between topography- and selection-based behavior is sometimes conflated with other variables that may be of equal or greater relevance to the issue at hand. We encourage researchers and clinicians to consider the extent to which the distinction between topography-based and selection-based verbal behavior is relevant in each context. For example, imagine a researcher studying computerized programmed instruction to compare acquisition and maintenance in two conditions: One in which participants (who are fluent typists) type their responses (e.g., “reinforcement”) to material on each screen, and one in which they respond by clicking on one of several textual stimuli displayed on the screen (e.g., select the word “reinforcement” from four options). Conceptualizing this simply as a comparison between topography- and selection-based responding would overlook several variables that could contribute to differences between the two conditions. One difference is that participants in the topography-based condition are constructing responses from minimal units, whereas participants in the selection-based condition are not. If differences are found between the two conditions, would these differences still be found if participants in the latter condition also constructed responses from minimal units (e.g., by clicking on letters on the screen)? Another difference is that participants in the topography-based condition have an unlimited number of response options, whereas participants in the selection-based condition have a limited number of options (e.g., four words). If differences are found between the two conditions, would these differences also be found if response options (i.e., provision of several words from which one can be copied) were also provided in the topography-based condition? More generally, statements about a selection-based communication modality or experimental preparation should be qualified by specifying additional characteristics, as they may not apply to all forms of selection-based behavior.

Further, we encourage researchers to systematically examine how these variable characteristics of selection-based verbal behavior (as well as other variable characteristics, such as stimulus iconicity) affect acquisition and emergence of derived verbal relations for users of AAC systems as well as for other learners. A comprehensive list of possible research topics is beyond the scope of this article, but a few examples can be provided based on the considerations we have raised in previous sections (see Table 1). With respect to minimal units and the characteristics of selection arrays, one avenue for future research might be to examine how various methods of text entry (topography-based, selection-based, or a combination of the two) affect skill acquisition and maintenance of learners who possess relevant minimal repertoires, whether within or outside of the context of AAC systems. To provide a concrete example, computer-assisted learning programs (e.g., Duolingo®) sometimes provide learners with a choice between constructing answers by typing their responses on their keyboard (i.e., topography-based text entry) or selecting letters from an array on the screen via mouse click or touchscreen. Does the method of text entry affect learning outcomes? With respect to selection arrays, researchers might investigate how the provision of response options, the number of response options, and other characteristics of selection arrays affect outcomes of training. For example, does the typical absence of response options in a topography-based skill acquisition program, compared to their typical presence in a selection-based program, translate into different outcomes? With respect to production of conventional verbal stimuli via selection-based methods, some potential avenues for future research might include emergence of listener behavior with respect to vocally produced stimuli for SGD users, and effects of text-to-speech technology in foreign-language learning.

Regardless of the status of a particular communication modality as selection-based (i.e., relying on scanning and conditional discriminations), its use in AAC, instructional programs, and basic experimental preparations requires a full consideration of the range of response production variables and response products that are available with current technologies.

Data Availability

Data sharing was not applicable to this article as no datasets were generated or analyzed during the current study.

Declarations

Compliance with Ethical Standards

This article does not contain data from human participants or animals.

Conflict of Interest

The authors have no conflicts of interest to disclose.

Footnotes

1

Michael (1985) used the term manded stimulus selection to refer to this type of listener behavior when it occurs in response to a speaker’s mand, such as “Point to the cat.”.

2

The use of only two comparisons, however, is not recommended practice due to the possibility of establishing reject instead of select control (Johnson & Sidman, 1993; Sidman, 1987).

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  1. Achmadi, D., Sigafoos, J., van der Meer, L., Sutherland, D., Lancioni, G. E., O‘Reilly, M. F., Hodis, F., Green, V. A., McLay, L., & Marschik, P. B. (2014). Acquisition, preference, and follow-up data on the use of three AAC options by four boys with developmental disability/delay. Journal of Developmental and Physical Disabilities, 26(5), 565–583. 10.1007/s10882-014-9379-z
  2. Adkins T, Axelrod S. Topography- versus selection-based responding: Comparison of mand acquisitions in each modality. The Behavior Analyst Today. 2001;2(3):259–266. doi: 10.1037/h0099941. [DOI] [Google Scholar]
  3. Alessi G. Generative strategies and teaching for generalization. The Analysis of Verbal Behavior. 1987;5(1):15–27. doi: 10.1007/BF03392816. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Axe, J. B. (2008). Conditional discriminations in the intraverbal relation: A review and recommendations for future research. The Analysis of Verbal Behavior, 24(1), 159–174. 10.1007/BF03393064 [DOI] [PMC free article] [PubMed]
  5. Barlow KE, Tiger JH, Slocum SK, Miller SJ. Comparing acquisition of exchange-based and signed mands with children with autism. The Analysis of Verbal Behavior. 2013;29(1):59–69. doi: 10.1007/BF03393124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bee, N., & André, E. (2008). Writing with your eye: A dwell time free writing system adapted to the nature of the human eye gaze. In E. André, L. Dybkjær, W. Minker, H. Neumann, R. Pieraccini, & M. Weber (Eds.), Perception in Multimodal Dialogue Systems. PIT 2008. Lecture Notes in Computer Science, 5078. Springer. 10.1007/978-3-540-69369-7_13
  7. Belloso-Díaz C, Pérez-González LA. Exemplars and categories necessary for the emergence of intraverbals about transitive reasoning in typically developing children. The Psychological Record. 2015;65(3):541–556. doi: 10.1007/s40732-015-0131-6. [DOI] [Google Scholar]
  8. Bloh C. Assessing topography-based versus selection-based verbal behavior: A comparison from the literature. Behavior Analysis: Research and Practice. 2016;16(1):12–18. [Google Scholar]
  9. Bondy AS, Frost LA. The picture exchange communication system. Focus on Autism and Other Developmental Disabilities. 1994;9(3):1–19. doi: 10.1177/108835769400900301. [DOI] [Google Scholar]
  10. Borgestig M, Sandqvist J, Parsons R, Falkmer T, Hemmingsson H. Eye gaze performance for children with severe physical impairments using gaze-based assistive technology—A longitudinal study. Assistive Technology. 2016;28(2):93–102. doi: 10.1080/10400435.2015.1092182. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cagliani RR, Ayres KM, Whiteside E, Ringdahl JE. Picture exchange communication system and delay to reinforcement. Journal of Developmental and Physical Disabilities. 2017;29(6):925–939. doi: 10.1007/s10882-017-9564-y. [DOI] [Google Scholar]
  12. Carbone VJ, Sweeney-Kerwin EJ, Attanasio V, Kasper T. Increasing the vocal responses of children with autism and developmental disabilities using manual sign mand training and prompt delay. Journal of Applied Behavior Analysis. 2010;43(4):705–709. doi: 10.1901/jaba.2010.43-705. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Carnett A, Martin C, Gevarter C. Evaluating augmentative and alternative communication modalities for individuals with developmental disabilities: A brief review of the last 5 years. Current Developmental Disorders Reports. 2021;8(2):133–141. doi: 10.1007/s40474-021-00226-0. [DOI] [Google Scholar]
  14. Carp, C. L., & Petursdottir, A. I. (2012). Effects of two training conditions on the emergence of novel intraverbals: An extension of Pérez-González et al. (2008). The Psychological Record, 62, 187–206. 10.1007/BF03395797
  15. Chambers M, Rehfeldt RA. Assessing the acquisition and generalization of two mand forms with adults with severe developmental disabilities. Research in Developmental Disabilities. 2003;24(4):265–280. doi: 10.1016/S0891-4222(03)00042-8. [DOI] [PubMed] [Google Scholar]
  16. Charlop-Christy MH, Carpenter M, Le L, LeBlanc LA, Kellet K. Using the Picture-Exchange Communication System (PECS) with children with autism: Assessment of PECS acquisition, speech, social-communicative behavior, and problem behavior. Journal of Applied Behavior Analysis. 2002;35(3):213–231. doi: 10.1007/978-1-4899-1947-2_8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Connell PJ, McReynolds LV. An experimental analysis of children’s generalization during lexical learning: Comprehension or production. Applied Psycholinguistics. 1981;2(4):309–332. doi: 10.1017/S0142716400009760. [DOI] [Google Scholar]
  18. Cortez MD, dos Santos L, Quintal AE, Silveira MV, de Rose JC. Learning a foreign language: Effects of tact and listener instruction on the emergence of bidirectional intraverbals. Journal of Applied Behavior Analysis. 2020;53(1):484–492. doi: 10.1002/jaba.559. [DOI] [PubMed] [Google Scholar]
  19. Cortez MD, da Silva LF, Cengher M, Mazzoca RH, Miguel CF. Teaching a small foreign language vocabulary to children using tact and listener instruction with a prompt delay. Journal of Applied Behavior Analysis. 2022;55(1):249–263. doi: 10.1002/jaba.885. [DOI] [PubMed] [Google Scholar]
  20. Couper L, van der Meer L, Schafer MCM, McKenzie E, McLay L, Lancioni GE, Marschik PB, Sigafoos J, Sutherland D. Comparing acquisition of and preference for manual signs, picture exchange, and speech-generating devices in nine children with autism spectrum disorder. Developmental Neurorehabilitation. 2014;17(2):99–109. doi: 10.3109/17518423.2013.870244. [DOI] [PubMed] [Google Scholar]
  21. Dougher MJ, Hamilton DA, Fink BC, Harrington J. Transformation of the discriminative and eliciting functions of generalized relational stimuli. Journal of the Experimental Analysis of Behavior. 2007;88(2):179–197. doi: 10.1901/jeab.2007.45-05. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Elsahar Y, Hu S, Bouazza-Marouf K, Kerr D, Mansor A. Augmentative and alternative communication (AAC) advances: A review of configurations for individuals with a speech disability. Sensors. 2019;19(8):1911. doi: 10.3390/s19081911. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Epstein R, Lanza RP, Skinner BF. Symbolic communication between two pigeons (columbia livia domestica) Science. 1980;207:543–545. doi: 10.1126/science.207.4430.543. [DOI] [PubMed] [Google Scholar]
  24. Frost, L., & Bondy, A. (2002). The picture exchange communication system training manual, 2nd ed. Pyramid Educational Consultants.
  25. Ganz J, Hong E, Goodwyn F, Kite E, Gilliland W. Impact of PECS tablet computer app on receptive identification of pictures given a verbal stimulus. Developmental Neurorehabilitation. 2015;18(2):82–87. doi: 10.3109/17518423.2013.821539. [DOI] [PubMed] [Google Scholar]
  26. Gayomali, C. (2015, January 8). The text message turns 20: A brief history of SMS. The Week. https://theweek.com/articles/469869/text-message-turns-20-brief-history-sms
  27. Gevarter C, Zamora C. Naturalistic speech-generating device interventions for children with complex communication needs: A systematic review of single-subject studies. American Journal of Speech-Language Pathology. 2018;27(3):1073–1090. doi: 10.1044/2018_AJSLP-17-0128. [DOI] [PubMed] [Google Scholar]
  28. Gevarter C, O'Reilly M, Rojeski L, Sammarco N, Lang R, Lancioni GE, Sigafoos J. Comparing communication systems for individuals with developmental disabilities: A review of single-case research studies. Research in Developmental Disabilities. 2013;34(12):4415–4432. doi: 10.1016/j.ridd.2013.09.017. [DOI] [PubMed] [Google Scholar]
  29. Gevarter, C., O‘Reilly, M. F., Kuhn, M., Mills, K., Ferguson, R., & Watkins, L. (2016). Increasing the vocalizations of individuals with autism during intervention with a speech-generating device. Journal of Applied Behavior Analysis, 49(1), 17–33. 10.1002/jaba.270 [DOI] [PubMed]
  30. Green, G., & Saunders, R. R. (1998). Stimulus equivalence. In K. A. Lattal, & M. Perone (Eds.), Handbook of research methods in human operant behavior (pp. 229–262). Springer. 10.1007/978-1-4899-1947-2_8
  31. Greenberg AL, Tomaino ME, Charlop MH. Adapting the picture exchange communication system to elicit vocalizations in children with autism. Journal of Developmental and Physical Disabilities. 2014;26(1):35–51. doi: 10.1007/s10882-013-9344-2. [DOI] [Google Scholar]
  32. Gregory MK, DeLeon IG, Richman DM. The influence of matching and motor-imitation abilities on rapid acquisition of manual signs and exchange-based communication responses. Journal of Applied Behavior Analysis. 2009;42(2):399–404. doi: 10.1901/jaba.2009.42-399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Hall GA, Chase PN. The relationship between stimulus equivalence and verbal behavior. The Analysis of Verbal Behavior. 1991;9(1):107–119. doi: 10.1007/BF03392865. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Johnson C, Sidman M. Conditional discrimination and control by negative stimuli. Journal of the Experimental Analysis of Behavior. 1993;59(2):333–347. doi: 10.1901/jeab.1993.59-333. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Kodak T, Clements A, Paden AR, LeBlanc B, Mintz J, Toussaint KA. Examination of the relation between an assessment of skills and performance on auditory-visual conditional discriminations for children with autism spectrum disorders. Journal of Applied Behavior Analysis. 2015;48(1):1–9. doi: 10.1002/jaba.160. [DOI] [PubMed] [Google Scholar]
  36. Konstantareas MM. Autistic children exposed to simultaneous communication training: A follow-up. Journal of Autism and Developmental Disorders. 1987;17(1):115–131. doi: 10.1007/BF01487264. [DOI] [PubMed] [Google Scholar]
  37. Kristensson, P.O., & Vertanen, K. (2012). The potential of dwell-free eye-typing for fast assistive gaze communication. In Proceedings of the Symposium on Eye Tracking Research and Applications (ETRA '12) (pp. 241–244). Association for Computing Machinery. 10.1145/2168556.2168605
  38. Kuroda T, Lattal KA, García-Penagos A. An analysis of an autoclitic analogue in pigeons. The Analysis of Verbal Behavior. 2014;30(2):89–99. doi: 10.1007/s40616-014-0019-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. LaRue RH, Pepa L, Delmolino L, Sloman KN, Fiske K, Hansford A, Liebross S, Weiss MJ. A brief assessment for selecting communication modalities for individuals with autism spectrum disorders. Evidence-Based Communication Assessment and Intervention. 2016;10(1):32–43. doi: 10.1080/17489539.2016.1204767. [DOI] [Google Scholar]
  40. Lorah ER, Parnell A, Whitby PS, Hantula D. A systematic review of tablet computers and portable media players as speech generating devices for individuals with autism spectrum disorder. Journal of Autism and Developmental Disorders. 2015;45(12):3792–3804. doi: 10.1007/s10803-014-2314-4. [DOI] [PubMed] [Google Scholar]
  41. Michael J. Two kinds of verbal behavior plus a possible third. The Analysis of Verbal Behavior. 1985;3(1):1–4. doi: 10.1007/BF03392802. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Morin KL, Ganz JB, Gregori EV, Foster MJ, Gerow SL, Genç-Tosun D, Hong ER. A systematic quality review of high-tech AAC interventions as an evidence-based practice. Augmentative and Alternative Communication. 2018;34(2):104–117. doi: 10.1080/07434618.2018.1458900. [DOI] [PubMed] [Google Scholar]
  43. Muharib R, Alzrayer NM. The use of high-tech speech-generating devices as an evidence-based practice for children with autism spectrum disorders: A meta-analysis. Review Journal of Autism and Developmental Disorders. 2018;5(1):43–57. doi: 10.1007/s40489-017-0122-4. [DOI] [Google Scholar]
  44. Palmer DC. The role of atomic repertoires in complex behavior. The Behavior Analyst. 2012;35(1):59–73. doi: 10.1007/BF03392266. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Pérez-González LA, Herszlikowicz K, Williams G. Stimulus relations analysis and the emergence of novel intraverbals. The Psychological Record. 2008;58(1):95–129. doi: 10.1007/BF03395605. [DOI] [Google Scholar]
  46. Petursdottir AI, Lepper TL. Inducing novel vocalizations by conditioning speech sounds as reinforcers. Behavior Analysis in Practice. 2015;8(2):223–232. doi: 10.1007/s40617-015-0088-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Polson DAD, Parsons JA. Selection-based versus topography-based responding: An important distinction for stimulus equivalence? The Analysis of Verbal Behavior. 2000;17(1):105–128. doi: 10.1007/BF03392959. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Porta M. A study on text entry methods based on eye gestures. Journal of Assistive Technologies. 2015;9(1):48–67. doi: 10.1108/JAT-12-2013-0037. [DOI] [Google Scholar]
  49. Potter B, Brown DL. A review of studies examining the nature of selection-based and topography-based behavior. The Analysis of Verbal Behavior. 1997;14(1):85–104. doi: 10.1007/BF03392917. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Rawson KA, Zamary A. Why is free recall practice more effective than recognition practice for enhancing memory? Evaluating the relational processing hypothesis. Journal of Memory and Language. 2019;105:141–152. doi: 10.1016/j.jml.2019.01.002. [DOI] [Google Scholar]
  51. Rispoli MJ, Franco JH, van der Meer L, Lang R, Camargo SPH. The use of speech generating devices in communication interventions for individuals with developmental disabilities: A review of the literature. Developmental Neurorehabilitation. 2010;13(4):276–293. doi: 10.3109/17518421003636794. [DOI] [PubMed] [Google Scholar]
  52. Roche B, Barnes D. Arbitrarily applicable relational responding and sexual categorization: A critical test of the derived difference relation. The Psychological Record. 1996;46(3):451–475. doi: 10.1007/BF03395177. [DOI] [Google Scholar]
  53. Saunders KJ, Spradlin JE. Conditional discrimination in mentally retarded adults: The developmentl of generalized skills. Journal of the Experimental Analysis of Behavior. 1990;54(3):239–250. doi: 10.1901/jeab.1990.54-239. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Savage-Rumbaugh ES, Rumbaugh DM, Boysen S. Symbolic communication between two chimpanzees (pan troglodytes) Science. 1978;201(4356):641–644. doi: 10.1126/science.675251. [DOI] [PubMed] [Google Scholar]
  55. Segerdahl, P., Fields, W., & Savage-Rumbaugh, S. (2005). Kanzi’s primal language: The cultural initiation of primates into language. Palgrave Macmillan.
  56. Shafer E. Teaching topography-based and selection-based verbal behavior to developmentally disabled individuals: Some considerations. The Analysis of Verbal Behavior. 1993;11(1):117–133. doi: 10.1007/BF03392892. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Sidman M. Two choices are not enough. Behavior Analysis. 1987;22(1):11–18. [Google Scholar]
  58. Sidman, M. (1994). Equivalence relations and behavior: A research story. Authors Cooperative.
  59. Sidman, M., & Tailby, W. (1982). Conditional discrimination vs. matching to sample: An expansion of the testing paradigm. Journal of the Experimental Analysis of Behavior, 37(1), 23–44. 10.1901/jeab.1982.37-5 [DOI] [PMC free article] [PubMed]
  60. Skinner, B. F. (1957). Verbal behavior. Appleton-Century-Crofts.
  61. Sprinkle EC, Miguel CF. The effects of listener and speaker training on emergent relations in children with autism. The Analysis of Verbal Behavior. 2012;28(1):111–117. doi: 10.1007/bf03393111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Suchowierska M. Recombinative generalization: Some theoretical and practical remarks. International Journal of Psychology. 2006;41(6):514–522. doi: 10.1080/00207590500492534. [DOI] [Google Scholar]
  63. Su, P. L., Castle, G., & Camarata, S. (2019). Cross-modal generalization of receptive and expressive vocabulary in children with autism spectrum disorder. Autism & Developmental Language Impairments, 4, Article 2396941518824495. 10.1177/2396941518824495
  64. Sundberg CT, Sundberg ML. Comparing topography-based verbal behavior with stimulus selection-based verbal behavior. The Analysis of Verbal Behavior. 1990;8(1):31–41. doi: 10.1007/bf03392845. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Sundberg ML, Michael J, Partington JW, Sundberg CA. The role of automatic reinforcement in early language acquisition. The Analysis of Verbal Behavior. 1996;13(1):21–37. doi: 10.1007/BF03392904. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Tincani M. Comparing the picture exchange communication system and sign language training for children with autism. Focus on Autism and Other Developmental Disabilities. 2004;19(3):152–163. doi: 10.1177/10883576040190030301. [DOI] [Google Scholar]
  67. Tincani M, Crozier S, Alazetta L. The Picture Exchange Communication System: Effects on manding and speech development for school-aged children with autism. Education and Training in Developmental Disabilities. 2006;41(2):177–184. [Google Scholar]
  68. Valentino AL, LeBlanc LA, Veazey SE, Weaver LA, Raetz PB. Using a prerequisite skills assessment to identify optimal modalities for mand training. Behavior Analysis in Practice. 2019;12(1):22–32. doi: 10.1007/s40617-018-0256-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. van der Meer L, Kagohara D, Achmadi D, O’Reilly MF, Lancioni GE, Sutherland D, Sigafoos J. Speech-generating devices versus manual signing for children with developmental disabilities. Research in Developmental Disabilities. 2012;33(5):1658–1669. doi: 10.1016/j.ridd.2012.04.004. [DOI] [PubMed] [Google Scholar]
  70. Vignes T. A comparison of topography-based and selection-based verbal behavior in typically developing children and developmentally disabled persons with autism. The Analysis of Verbal Behavior. 2007;23(1):113–122. doi: 10.1007/BF03393051. [DOI] [PMC free article] [PubMed] [Google Scholar]
  71. Ward DJ, MacKay DJC. Fast hands-free writing by gaze direction. Nature. 2002;418:838. doi: 10.1038/418838a. [DOI] [PubMed] [Google Scholar]
  72. Wasserman EA, Brooks DI, McMurray B. Pigeons acquire multiple categories in parallel via associative learning: A parallel to human word learning? Cognition. 2015;136:99–122. doi: 10.1016/j.cognition.2014.11.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Wobbrock, J. O., Rubinstein, J., Sawyer, M., & Duchowski, A. T. (2007). Not typing but writing: Eye-based text entry using letter-like gestures. The 3rd Conference on Communication by Gaze Interaction, COGAIN 2007: Gaze-based creativity, 61–64.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Data sharing was not applicable to this article as no datasets were generated or analyzed during the current study.

Articles from The Analysis of Verbal Behavior are provided here courtesy of Association for Behavior Analysis International

RESOURCES