The Behavioral Technology of Discourse Analysis

Leon James

Univ. of Hawaii © 1972


            It is common to elevate language and speech to a universal medium of exchange in all human affairs; .we can hardly find a behavioral setting that does not involve language, either overtly, or privately, in thinking. Given this ubiquitousness of language in behavior it is certain that behaviorists, social engineers, and the helping professions will rely more and more on the technology of discourse. This field is not yet established though a significant beginning was made by B.F. Skinner in the classic Verbal Behavior (1957). Subsequently to its initial interest, which was that language, speech, and thinking can be included in the behavioral tradition, no further significant development in theory was made. The promise of a technology of verbal behavior failed to ignite with success the offices of psychotherapists, social workers, or classrooms. There is actually much talk of the failure of the public educational system to teach adequate skills in reading and writing. The point is that a successful technology of discourse is going to greatly and visibly improve the effectiveness of psychotherapy, counseling, and teaching. What is needed for a successful technology of discourse? This article gives a proposal for cataloguing discourse products in a natural, empirical way rather than through an arbitrary theory. A method of discourse analysis is outlined which shows up the functional relations between discourse-asųproduct and discourseųasųprocess. No previous proposal exists to our knowledge in which this interrelation has been accomplished successfully, from the requirements of the behavioristic tradition.


            Skinner‚s contribution was specifically to prove that discourse production is behaviorally lawful. He outlined a powerful theory that distinguishes three basic behavioral building blocks called „mandsš, „tactsš, and „autoclitics.š Mands are verbal behaviors in response to an internal need and under the control of the individual‚s cumulative history of reinforcement. For example, requests or questions are types of discourse called mands (mand is from demand). Tacts are verbal behaviors that are under the control of external or internal sensory awareness. For example, describing an object or some past event, generates discourse fragments that are tacts (from contact). Autoclitics are verbal behaviors that have an organizing function; for example, grammatical and rhetorical sequences can be selected by the speaker so as to achieve a particular goal. This occurs because listeners are strongly affected by the composition of the discourse they are exposed to. A successful discourse technology, as foreseen by Skinner, would allow the practitioner to know enough about the organization of discourse and its functional properties to be effective in using discourse as a change agent. Clearly, many behaviorists have already incorporated this view in their practice (e.g., Meichenbaum, Tharp, Watson, Heiby, and others). Still, there is clearly a need for advancing the state of our theory of the functional analysis of discourse so as to allow for a more specific application of the behaviorist‚s general idea of mands, tacts, and autoclitics.


            As well, there is a need to rescue the behavioristic point of view on discourse from the low level to which it was assigned by Chomsky‚s devastating attack on it (1960). It will be seen that the present proposal by passes this entire controversy, and merely goes on from where Skinner left off. No doubt there will be many in the cognitive linguistic camp who would relish to retrace the errors of Chomsky‚s arguments in the light of this proposal. We leave the task up to them. Still, Chomsky‚s attack on the behavioristic view of language has cast doubt on the validity and viability of Skinner‚s proposal. It will be seen from the present proposal that the behavioristic view is far from dead, and that indeed, it is the only one right now that offers the potential of an applied discourse technology which society greatly needs.

Behavioral Definition of Discourse.

            Discourse is the visible organization of invisible mental and affective states. By analyzing the elements of discourse and their functional dependence, we gain knowledge on the functional relations in mental and affective states. The control of behavior depends on a knowledge of social reinforcers and their contingent management. Discourse analysis provides information on social reinforcers and contingency schedules. This is because overt discourse is functionally dependent on inner mental and affective states. Thus, the organization of mental and affective states in an individual will determine the visible characteristics of discourse productions by that individual. The converse of this is a technological application: by analyzing the characteristics of the discourse productions of individuals, we can gain information on their inner mental and affective states.


            This idea is generally recognized by all, namely, that language provides clues as to the person‚s thoughts and feelings. Nevertheless, to bring this idea down to the level of technology it needs further behavioristic development. Especially, what is wanted is a functional analysis of discourse content, and the behavioral correlates of the functional units.


            With this in view, we may define discourse as verbal behaviors at three levels of organization. This will be the fundamental premise of our proposal, including of course, the sufficient identification of the three levels. As well, the method of identifying functional units in discourse must be empirical and natural; that is, the units must be in the language of behavioral contingency management. These are technologies of behavior based on reinforcement principles.


            The organization of discourse must parallel the organization of mental and affective states since there is a functional dependence between inner states and outer behaviors. Skinner has argued that inner states are built up and maintained by environmental conditions. This idea is well known in general in that one acts according to one‚s motives and through the means of our habits and skills. In the general sense, therefore, we would expect that discourse organization reflects thoughts and feelings. But in a specific or particular sense, this principle tends to break down in the absence of a behavioristic outlook. For example, Freud argued that slips of the tongue are under the functional control of avoidance affect. Literary criticism and rhetoric are two fields filled with proposals on how to measure the effectiveness of writings. Many people develop an intuitive perception of variations in discourse style; others are effective in the use of metaphor and simile. Still, these personal insights need to be translated into a scientific technology. There is a real problem to overcome in the fact that the content of discourse is so complex that a nonųsystematic approach bogs down in arbitrary cataloguing systems, of which there are many exemplars in the literature. The trouble with these is that not everyone can agree that the arbitrary system is sufficient for their needs. A system of discourse analysis that satisfies the research needs of the linguist does not suit the clinical psychologist or teacher. What is suitable for the language teacher is not for the counselor or on the job training. This proposal is offered as a general solution suitable for all and relying only on known basic laws of behavior.


            The three levels of verbal behavior must be the same three levels in which we already distinguish behavior, namely, sensorimotor, cognitive, and affective. Behaviorists use the word „behaviorš to refer to the sensorimotor effects of reinforcement history. Walking, moving, pushing, talking, writing, driving, eating, and so on. In general, these behaviors are public and interpersonal. Each person‚s behavior is a visible discriminant stimulus to others, and so, the behavior of one influences the behavior of another. Skinner referred to thinking, which is private, as verbal behavior under community management. In other words, the discourse content of thinking is under the control of social reinforcers. This relation is evident in general: what we think at any one time is influenced by the environmental requirements.  A characteristic of behavioral settings is that they demand our attention in specific ways; when we drive, when we listen to a lecture, when we compose a letter - these are clear examples of how the content of our thinking varies with the stimulus demands of the setting.


            It is consistent therefore to assume that that all inner states of thinking and affective habits are built up and maintained by social reinforcers as managed in the community throughout its various behavioral settings. Some people might associate our idea with „thought-controlš; but this concept seems to suggest that that the individual is not free to select social reinforcers, when yet there is that freedom. The criticism is fallacious because contingency management relies on the existing social reinforcers, by definition. Individuals remain free to devalue normal reinforcers in a community, and many do, as we witness from all sorts of novel lifestyles. Thus, there is no „thoughtųcontrolš in the control of verbal behavior since the control is managed by means of people‚s own freely chosen values and priorities. The expression control implies, not lack of freedom, but systematic methods of exercising freedom. Behaviorists are dedicated to the maintenance of true freedom through effective techniques of operation. This is the same position as that of legislators: the restriction of alternatives for the sake of the freedom and good of all.


            Social reinforcers are always organized in a community. For example, people can make up lists of valuable things in our society: money, health care, recreation, friendship, pleasant housing, and so on. Whenever these reinforcers are present in the environment they can be utilized for the technology of behavior. The organization of social reinforcers may be also be called community contingency management. Each reinforcer will be present under certain conditions, and the amount available will also vary on specific contingencies. Money is obtained through gifts or services provided; friendship is selective; health care requires membership; and so on. Minimum wage legislation is an example of an attempt to modify the organization of social reinforcers.  Any change in the organization of social reinforcers is accompanied by a change in the behavior of targeted individuals whose behaviors are maintained by those social reinforcers. For instance, an employer may offer merit increases contingent upon performance criteria. The availability of this new source for money creates a new specific channel for social reinforcement and behavioral modification. If the company withdraws the offer there is a concomitant change in the organization of the setting‚s social reinforcers. It is important to realize that the changes in behavior occasioned by a change in social reinforcers occur at three levels of behavior: sensorimotor behavior, cognitive behavior, and affective behavior.


            Let us trace the three levels in connection with the example. The company makes a new rule, crating the availability of a new social reinforcer. Targeted individuals will now be influenced at three levels of behavioral organization. At the sensorimotor level, the influence will be visible in a number of ways: discussion of the new plan, expressions of approval or disapproval, longer work hours, less errors, better quality performances, and so on. These are the results the company is primarily interested in. Nevertheless we know that these sensorimotor behaviors are not under the direct control of the social reinforcers. If they were, then management would be greatly simplified. As managers know, changing a rule or giving a new incentive works with some up to a point. This is no doubt due to conflicting social reinforcers. Still, it is clear that social reinforcers do not operate directly upon sensorimotor performance but through the mediation of cognitive and affective behaviors. The new rule about merit increases tied to specific behavioral performances will operate through the inner states of the individual: mental operations, decision making, contrastive analysis, daydreaming, specific motives, preferences, habits.


            We need therefore a specific proposal for the empirical classification of behavior units in terms of these three levels.  In an important sense, the problem of classifying behavior units and the problem of classifying discourse units is one and the same. This follows from the premise that all discourse units are overt behavioral units, which in turn are under the functional control of social reinforcers. Thus, whatever is being proposed about the organization of discourse must ipso facto apply to the organization of behavior. We shall now proceed to the characterization of the three levels of discourse, and the reader may continually filter every statement through the test of whether it applies equally to behavior laws already known. Any category definition for a discourse unit must have a clear behavioral effect, or else the proposal is no longer behavioristic.


Level C.         Sensorimotor level of Discourse.


            This natural category may be defined as verbal behaviors in the form of sensorimotor habits. These are external, overt, and public. Children, chimpanzees, dolphins, and computers are among the well known examples of successful programs to produce discourse. For example, the earliest discourse fragments produced by little children consist of their attempt to label an object in view or an internal sensation such as hunger or loneliness. The labeling of something produces discourse whose content consists of particular facts about what things are. Describing the memory of an experience, for instance, produces discourse whose content consists of labeling dramatic sequences as they occurred.