What We Talk about When We Talk about Talking

Blog Post Authored by Drs. Maria Otero, Gabriel Armshaw, Alonzo Andrews, and Lee Mason

Dr. Maria Otero, PhD, BCBA,  is a researching clinician in early intervention at Cook Children’s
Hospital. She’s a graduate of the University of North Texas, where she earned her Ph.D. Her
work is driven by a deep interest in verbal behavior, especially in understanding and supporting children who communicate using speech-generating devices (SGDs). She is passionate about the assessment of verbal behavior and improving communication outcomes through research-informed practice.

---

Recent discussions within the literature have repeatedly noted that, in the natural environment, verbal behavior tends to be multiply controlled. This is particularly true of extended and more complex verbal episodes. Yet we largely rely on frameworks that emphasize the functional distinction of verbal operants. Across this four-part series, we aim to explore the antecedent variables contributing to verbal behavior and describe a classification scheme for analyzing their interaction. 

What is the meaning of language?

The question is probably as old as history, and certainly as old as 350 BCE, when Aristotle described language as a means to convey the mind, mediated by societally accepted signs and symbols. Aristotle’s conventions about language were still deeply embedded in the mind for almost two millennia until the great thinker, René Descartes,¹ flipped things around and described language as the very criterion of the mind. “The mind” continued to be intertwined with language even through John Locke’s ideas about the arbitrariness of words that separated the speaker’s private reflections from their experiences of the physical world. 

David Hume continued Locke’s experiential approach to examining language as the association of ideas. Specifically, Hume identified three principles that helped establish new associations: resemblance, contiguity, and causality. Hume’s associations were influential to Edward Thorndike’s law of effect and John B. Watson’s habit formation, both of which ultimately led to B.F. Skinner’s radical behaviorism, which allowed for an analysis of the world within the skin. For Skinner, language was not about the mind, but about the functional control of the environment. 

Though grossly oversimplified, it is critical for us to acknowledge the evolution of the philosophy of language, which is far from complete! To presume a direct lineage without the contributions of other significant influences would be a mistake and directly contradicts the framework presented below.

A Better Way of Talking about Verbal Behavior

Tired of comma counting, Skinner (1959) established a framework for talking about language as Verbal Behavior. A functional analysis of language gives us the means of classifying the contingencies surrounding any bit of verbal behavior by its environmental relations. In Part II of his text, Skinner described different variables that control the speaker’s verbal behavior: Chapter 3 focused on The Mand; Chapter 4 described Verbal Behavior Under the Control of Verbal Stimuli, including echoic, textual, transcription, and intraverbal behavior; and Chapter 5 introduced The Tact. Table 1 outlines Skinner’s codification scheme.

Table 1: Skinner’s 1957 Elementary Verbal Operants
Antecedent	MO	Non-Verbal	Verbal
Behavior	Mand	Tact	Echoic	Textual	Intraverbal
Consequence	Specified Reinforcer	Generalized Conditioned Reinforcer	Generalized Conditioned Reinforcer

Skinner’s verbal operants are functionally independent in the sense that the acquisition of one operant does not necessarily result in the acquisition of another operant with the same response topography. For example, just because I can imitate certain words does not mean I can use them contextually in a conversation. Functional independence, however, does not imply that a verbal response can be controlled by only one source at a time. 

Although Skinner’s initial work has remained consistent, some adjustments have been made to refine its analytic utility. Jack Michael (1982) introduced two new categories to help account for the full range of topographies omitted by Skinner’s terms. Rather than “textual”, he defined codic behavior as a response evoked by a verbal discriminative stimulus that has point-to-point correspondence but no formal similarity. This would be like taking dictation or writing down what someone spells out via sign language. Instead of “echoic” and “transcription,” Michael defined duplic behavior as that in which a response is evoked by a verbal discriminative stimulus that shares both point-to-point correspondence and formal similarity. This includes echoing what you hear, as well as copying text written on a whiteboard onto a notebook. Table 2 shows Michael’s reframing of the verbal operants.

Table 2: Michael’s 1982 Recategorization
Antecedent	MO	Non-Verbal	Verbal
Behavior	Mand	Tact	Codic	Duplic	Intraverbal
Consequence	Specified Reinforcer	Generalized Conditioned Reinforcer	Generalized Conditioned Reinforcer

In that same issue of VB News,² Ernie Vargas (1982) suggested yet another terminological change. He argued the importance of differentiating verbal behavior under the control of non-verbal stimuli from verbal behavior under the control of other verbal behavior. The former he referred to as Extraverbal. For the latter, he proposed a term that Skinner had already used to describe a type of verbal behavior under the control of a verbal discriminative stimulus: Intraverbal. But the same term cannot simultaneously describe the class and a member of the class,³ so he introduced a new term, sequelic, in coordination with Michael’s discussion of codic and duplic control.⁴ Think of your favorite movie sequel, or related words like sequela or sequence. A sequelic response not only comes after the antecedent verbal stimulus, but it shares a relationship with it. Skinner referred to this relation as “contiguous usage,” which we’ll revisit in a later post. Table 3 is a table illustrating Vargas’ refinements. 

Table 3: Vargas’ 1982 Reframing
Antecedent	Extraverbal		Intraverbal
Behavior	Mand	Tact	Codic	Duplic	Sequelic
Consequence	Specified Reinforcer	Generalized Conditioned Reinforcer

As the science of verbal behavior evolves, as all sciences do, behavior analysts have called for a more precise and consistent terminology (Blair & Farros, 2019) with respect to how we talk about verbal behavior. One area that continues to be a sticky wicket is that of multiple control. Although we’ve been conditioned to emphasize the functional independence of the elementary verbal operants, in practice, verbal behavior is rarely isolated to one stimulus event or source of control (Michael et al., 2011; Palmer, 2016). Identifying multiple sources of antecedent control allows us to begin to gain insights into generative language and the emergence of untrained verbal relations. To do this, we demonstrate a new approach to classifying antecedent stimuli that allows us to account for multiple sources of control at once, allowing us to better understand how untrained and novel verbal behavior emerges. 

In applied research and clinical practice, four of Skinner’s verbal operants – mand, tact, intraverbal, and echoic – have received particularly extensive attention as components of more complex verbal behavior.  Researchers and practitioners regularly employ these four verbal operants as the building blocks for developing more complex forms of verbal behavior. Based on this evidence, and using the terms Skinner gave us, we reconceptualize his framework in a way that emphasizes the importance of multiple control. Yes, the verbal operants are functionally independent, but the ubiquity of multiple control means it must be placed at the forefront. Table 4 presents a restructuring of the three-term contingency to emphasize the possibility of co-occurring antecedent variables. Whereas, historically, the operants have been defined by the middle term (e.g., “tacting”), we are suggesting the emphasis be placed on the initial term (e.g., “tact control”) to afford discussions of multiple control across extraverbal and intraverbal sources.

Table 4: Three-term Verbal Contingencies

	Extraverbal		Intraverbal
Antecedent	Mand	Tact	Duplic	Sequelic
Behavior	Verbal Behavior
Consequence	Generalized Conditioned Reinforcer

Palmer (2016) argued that behavior analysts should distinguish between the term intraverbal, as a verbal operant class, and intraverbal control as a means of describing one of many concurrent variables. We take this one step further to say that we should always be referring to controlling relations, whether talking about many or one in isolation. 

A Continuum of Control

Although the above table was designed to illustrate multiple control, something is still missing from the analysis. Emphasizing the functional independence of the verbal operants requires us to treat them as dichotomous variables, which may limit our ability to describe the more ambiguous relations that result from complex interactions. Given the ubiquity of multiple control, a framework for describing the simultaneous sources surrounding any given verbal response is necessary. Skinner referred to supplementary sources as formal and thematic stimulation, though – once again – we’ve made some alterations to Skinner’s original definitions to emphasize multiple control with greater terminological precision. You’ll note that our taxonomy, itself, was also influenced by multiple sources, including Hume’s writings on associations and we untangle Skinner’s concept of “contiguous usage.” Specifically, we describe formal control as having contiguity to emphasize the spatial (i.e., tact) or temporal (i.e., echoic) relationship between verbal response and antecedent stimulus. That is, the antecedent and behavior occur relatively close together. In contrast, thematic control shows discontiguity, with a gap in time (i.e., sequelic) and/or space (i.e., mand) occurring between the antecedent and verbal response. To echo Vargas, our intent is not to redraw any of Skinner’s verbal relations; we have merely adjusted their boundaries to account for properties shared among verbal operants.

The Punnett square below provides a simple layout to display the interaction between primary and supplementary sources of control, allowing each verbal operant to be cross categorized. Its matrix structure reveals nuanced combinations, where each quadrant reflects a unique interaction of antecedent variables. This framework preserves the individual verbal operants for describing pure sources of control, while providing a succinct way to talk about convergent multiple control using specific terms for the interaction of different operants. As verbal episodes become even more complex, overarching terms like primary or supplementary can be used for a more comprehensive account of controlling relations, though at the sacrifice of precision.

Table 5: A Cross Categorization of Controlling Relations

		Supplementary Sources
		Formal	Thematic
Primary Sources	Extraverbal	Tact	Mand
	Intraverbal	Duplic	Sequelic

Critically, the sources of control are presented along continua rather than as discrete categories. For example, consider the many speakers who use augmentative and alternative communication. Does the definition of echoic hold up when a speaker uses a speech-generating device? We use the term formal to indicate the relative proximity of controlling relations. The analysis benefits from a framework that further specifies the antecedent variable according to its extraverbal (i.e., spatial) or intraverbal (i.e., temporal) relations, while acknowledging that – in reality – it is typically a combination of both.  

Alternatively, an extraverbal response indicates a non-verbal stimulus that ranges from formal (i.e., contiguous) to thematic (i.e., discontiguous). When presented with the picture below, we’d classify the response, “Chocolate,” under tact control, and the response, “Golden ticket,” as a function of mand control. Though, in real-life scenarios – say, listening to the rhymes of oompa-loompas while touring your local chocolate factory – multiple control is the rule.

Clearly, we have more explaining to do, which is why this is the first of a four-part series of blog posts. Part 2 of our blog series will focus on primary sources of control, where we examine ranges of verbal and non-verbal stimuli. In doing so, we’ll revisit Vargas’ discussion of intraverbal and extraverbal sources of control. Spoiler alert: Vargas also described controlling relations as a continuum!

Part 3 of the series describes supplementary sources of control. Skinner previously talked about supplementary sources in terms of formal and thematic prompts and probes. We retrofit some new definitions onto these categories that allow us to better define the presence and absence of controlling relations. Specifically, the speaker’s verbal response is more or less controlled by the spatial and temporal presence of intraverbal or extraverbal stimuli.

Part 4 of the series brings everything back together to emphasize the importance of multiple, interacting antecedent relations that control the speaker’s verbal response. By this point, we hope to have made the case for describing verbal operants in terms of their sources of control, rather than as discrete instances of behavior. That is, instead of calling the above response, “Chocolate,” a tact, it is more precise to describe it as under tact control (perhaps among others). We’ll also demonstrate the interaction of verbal operants through the use of a multiaxial radar chart and show how the speaker’s verbal repertoire can be quantified using shape descriptors. 

As questions about language become more complex, they require more complex solutions. Skinner’s functional analysis revolutionized the way we think about language as verbal behavior. Scroll through the posts within the Verbal Behavior Matters blog, and you’ll find a vast array of verbal behavior researchers doing remarkable work based on Skinner’s analysis. The premise that language, like all other operant behavior, is subject to the dynamics of reinforcement, punishment, and extinction has led us from a mentalistic foundation to a heuristic science replete with empirically based interventions for imparting generative language.

---

References

Blair, B. J., & Farros, J. N. (2019). The duplic and codic: The importance of a consistent taxonomy of verbal behavior. The Analysis of Verbal Behavior, 35(2), 235-244. https://doi.org/10.1007/s40616-019-00114-0 

Michael, J. (1982). Skinner’s elementary verbal relations: Some new categories. VB News, 1, 1-3. https://doi.org/10.1007/BF03392791 

Michael, J., Palmer, D. C., & Sundberg, M. L. (2011). The multiple control of verbal behavior. The Analysis of Verbal Behavior, 27(1), 3-22. https://doi.org/10.1007/BF03393089 

Palmer, D. C. (2016). On intraverbal control and the definition of the intraverbal. The Analysis of Verbal Behavior, 32(2), 96–106. https://doi.org/10.1007/s40616-016-0061-7 

Skinner, B. F. (1957). Verbal behavior. Appleton-Century-Crofts. https://doi.org/10.1037/11256-000 

Vargas, E. A. (1982). Intraverbal behavior: The codic, duplic, and sequelic subtypes. VB News, 1, 5-7. https://doi.org/10.1007/BF03392792 

Footnotes

1. Perhaps best known for, “I think, therefore I am,” he is also the progenitor of mind-body dualism. ↩︎
2. Between Michael’s and Vargas’ articles, it’s a very worthwhile six whole pages! ↩︎
3. A classic category error. ↩︎
4. Vargas and Michael both identified the same problems with Skinner’s original framework and had each independently developed new terms to help clarify the controlling verbal relations. However, Vargas deferred to Michael’s terminology, which he felt “seemed best.” ↩︎