Research in Dcvelqmental
Disabilities, Vol. 11.pp. 379-393.1990
Printed in the USA. All rights reserved.
0891-4222190$3.00 + .c0 CopyTight8 1990Pergamon Press plc
Correspondence Training: Review and Current Issues Ruth A. Baer University of Kentucky
This article reviews the literature on correspondence training and discusses issues of theoretical, conceptual, or applied importance. Generalization, maintenance, and application to mentally retarded individuals are addressed. The relationship between correspondence training procedures and the concept of self-control is discussed. Directions for future research are &scribed.
It is generally acknowledged that the relationship between an individual’s verbal and nonverbal behavior is important to society (Israel, 1978; Risley & Hart, 1968; Rogers-Warren & Baer, 1976). The efforts of socializing agents, such as parents and teachers, often are focused on teaching verbal behavior regarding appropriate conduct. The apparent underlying assumption is that such verbal repertoires subsequently will lead to desirable nonverbal behavior. For example, parents and teachers may assume that a child who can verbalize the importance of not hitting people will refrain from hitting, or that a child who promises to look both ways before crossing the street will actually do so on future street crossing occasions. Further, society appears to value correspondence between people’s verbalizations and actions. People who state an intention to do something but later fail to do it, or who falsely describe what they have previously done, are likely to incur the disapproval of others. Thus, the study of the relationship between verbal and nonverbal behavior is important. This review will focus primarily on correspondence training procedures.
Requests for reprints should be sent to Ruth A. Baer, Ph.D., Department of Psychology, Kastle Hall, University of Kentucky, Lexington, KY 405060044. 379
380
R. A. Baer
Usually, these involve teaching subjects to make statements about their own behavior, often in the form of, “I’m going to today.” Correspondence between these statements and the subject’s behavior then is reinforced. Correspondence training procedures have been applied to a broad range of subjects and target behaviors. Several theoretical and applied issues will be discussed. EARLY RESEARCH ON EFFECTS OF VERBALIZATIONS Early research with preschool children showed that limited changes in nonverbal behaviors could be brought about by changing the subjects’ corresponding verbal behaviors. For example, Lovaas (1964) used a procedure in which children engaged in a 15-minute conversation with a mechanical puppet. The puppet provided praise and trinkets each time the subject mentioned the name of a particular food item, such as “celery.” As numerous prompts were provided, the child had many opportunities to say “celery” and receive positive consequences during the session. Immediately following the session, the child’s consumption of celery during a snack period was observed. For some subjects, consumption of celery after puppet sessions was much greater than on days when no puppet session occurred. However, the effects was small and decreased with repeated trials. Sherman (1964) found similar results using play with specific toys as target behaviors. Brodsky (1967) studied the effect of reinforcing changes in verbal behavior in a retarded, institutionalized female adult. During interview sessions, reinforcement was provided contingent on statements about engaging in social behavior, such as, “I like to play with my friends.” Although the frequency of these statements during interview sessions increased, no changes were noted in the subject’s actual social behavior in other settings. EARLY CORRESPONDENCE
TRAINING RESEARCH
More recently, researchers have required subjects to emit verbalizations about their own behavior in the form of statements of intention (often called promises) or reports about their past behavior. Three experimental conditions are typically used in correspondence training studies. During baseline, subjects’ rates of engaging in the target behavior(s) are determined. Subjects may also be asked a question, for example, “What will/did you do (during playtime)?” in order to evoke a verbalization, but no specific verbalizations are trained. During Reinforcement of Verbalization, subjects are taught to emit a spe-
Correspondence Training
381
cific statement when the experimenter asks the question. Typically, the experimenter prompts the subject to say, “I’m going to (play with crayons),” or “I (played with crayons).” Reinforcement is provided immediately, contingent on the subject’s emission of the verbalization. Observation of the rate of occurrence of the verbalized response (e.g., playing with crayons) continues. This procedure yields mixed results. Occasionally, researchers find that reinforcement of verbal behavior alone leads to consistent changes in corresponding nonverbal behavior (Baer, Williams, Osnes, & Stokes, 1985; Crouch, Rusch, & Karlan, 1984). Most studies, however, have found that reinforcement of verbal behavior alone has minimal effects on corresponding nonverbal behavior (Baer, Osnes, & Stokes, 1983; Baer, Williams, Osnes, & Stokes, 1984; Israel & Brown, 1977; Israel & O’Leary, 1973; Risley & Hart, 1968; Williams & Stokes, 1982). During Reinforcement of Correspondence, the experimenter asks the question of the subject, and the subject emits the trained verbalization. Reinforcement is contingent on correspondence between the verbalization and the subject’s behavior. If the subject states that he/she will play with crayons, for example, then reinforcement is withheld until after playtime, and is presented only if the subject actually plays with crayons. This procedure consistently results in increased correspondence between subjects’ verbalizations and behavior. Risley and Hart (1968) provide an example. During baseline, preschool children were asked what they had done that day, and praise was provided for all verbalizations. During Reinforcement of Verbalization, teachers reinforced any report of having used a particular toy (paint or blocks) during a preceding play period. Although the children’s rate of reporting use of that toy increased, their actual use of it did not. During Reinforcement of Correspondence, only true reports of having used a particular toy were reinforced, and the subjects’ use of the target toy increased. Furthermore, after correspondence had been trained across two toys, reinforcement of the report alone was sufficient to increase the corresponding use of the third toy. Use of the third toy was described as generalized correspondence, because correspondence with this response had not been previously trained, and because it occurrred even though reinforcement was not contingent on it. Similar results were obtained by Israel and Brown (1977). who first asked children to state what toy they intended to play with and then gave them the opportunity to do so. With the first toy, Reinforcement of Verbalization resulted in no increase in use of the promised toy, while Reinforcement of Correspondence resulted in substantial increases. With the second target toy, Reinforcement of Verbalization alone led to increases in both promises to use and actual use of the toy.
R. A. Baer
These and similar early studies (e.g., Jewett & Clark, 1979; RogersWarren & Baer, 1976; Williams & Stokes, 1982) indicated that correspondence between verbal and (non)verbal behavior could be trained using reinforcement. These studies also raised methodological and practical issues. For example, some authors used a say-then-do procedure, while others used a do-then-say procedure. As subsequent research has used the say-do procedure almost exclusively, the do-say procedure will not be discussed further. In addition, generalization and maintenance are important issues. Following training with a few responses, correspondence generalized to new responses in some cases, but not all, for reasons that are unclear. Lastly, many of these early studies targeted toy play behaviors. Although increased rates of play can be an important educational goal for young children, successful application to a greater diversity of behaviors with educational or clinical significance would enhance the practical utility of the procedures. GENERALIZATION Generalization of correspondence is seen when, after training correspondence with a few responses, the subject’s verbal behavior reliably controls his performance of other, untrained responses (Karlan & Rusch, 1982). Generalized correspondence is also known as generalized verbal control (Williams & Stokes, 1982). Generalization of correspondence to untrained behaviors should enhance the utility of corresondence training procedures. A child who shows generalized correspondence will follow through on a wide range of stated intentions, in addition to those that have been trained. If generalized correspondence is assured, the behavior change agent need only evoke from the child a statement of intention to engage in a desired response, and the response is very likely to occur (if it is in the child’s repertoire). Thus, correspondence training is potentially a powerful behavior change technique, if generalized correspondence can be developed. Several studies have documented generalization of correspondence across responses (Guevremont, Osnes, & Stokes, 1986a; Israel & Brown, 1977; Risley & Hart, 1968; Rogers-Warren & Baer, 1976; Williams & Stokes, 1982). After correspondence between verbal and nonverbal behavior was established for one or a few responses, reinforcement of an untrained verbalization was sufficient to alter the rate of corresponding nonverbal behavior, at least for some subjects and some target behaviors. Williams and Stokes (1982) suggested that this generalized verbal control might be conceptualized as generalization through training of sufficient examplars (Stokes & Baer, 1977). In contrast, other studies (Israel & O’Leary, 1973; Karoly & Dirks, 1977)
Correspondence Training
383
have used similar procedures but have not shown generalization of correspondence to untrained behaviors. Although one could argue that, in these studies, insufficient exemplars.were trained, the number of exemplars required is unclear. Both Israel and O’Leary (1973) and Risley and Hart (1968) used three exemplars (all toy play behaviors), and the latter observed generalization, while the former did not. Williams and Stokes (1982) proposed that failure to obtain generalization of correspondence might be due to the subjects’ discrimination of when the reinforcement contingencies are in effect. That is, the procedures might provide cues that allow the subjects to discriminate whether reinforcement is contingent on correspondence or on the verbalization only. Generalization might be promoted by modifying the procedures to make the contingencies less discriminable. Both Baer et al. (1983) and Baer et al. (1984) accomplished this using delayed reinforcement. Fist, subjects were trained in correspondence with one or a few responses. After subjects were showing consistent correspondence, reinforcement was made contingent only on verbalizations. However, instead of providing reinforcement immediately following the verbalization, the experimenters delayed reinforcement until after the opportunity to engage in the corresponding nonverbal behavior had occurred. Consistently high rates of correspondence were observed during delayed reinforcement conditions, suggesting that the children did not discriminate that reinforcement was contingent on verbal behavior alone, but responded as though reinforcement were contingent on correspondence. MAINTENANCE Two types of maintenance can be examined in correspondence training research. Maintenance of the previously trained behavior is seen when the subject continues to engage in that response even in the absence of both prior statements of intention to do so and subsequent reinforcement for doing so. That is, examination of this type of maintenance is possible when the subject has experienced correspondence training with a particular response, and the experimenters have discontinued both evoking the verbalization from the subject and subsequent reinforcement. Maintenance of the target behavior after statements of intention have been discontinued is rarely investigated. However, studies that report these data reveal that the target response usually does not maintain after the corresponding verbalization is no longer being emitted. Exceptions can be seen in Williams and Stokes (1982), who showed maintenance of two target behaviors in one child, Baer et al. (1984), who showed maintenance in one target behavior in one child, Osnes, Guevremont, and Stokes (1986). who
R.A.Baer
found maintenance in two of three subjects, and Guevremont et al. (1986a), who found maintenance to occur with some target behaviors but not others. Maintenance of correspondence is seen when, after training correspondence with a particular target response, the subject continues to follow through on statements that he or she will engage in that response. In this case, maintenance can be achieved simply by having the subject continue to state that he will engage in the target response, in the absence of programmed reinforcement for doing so. The development of maintenance of correspondence should enhance the practical utility of correspondence training procedures by making it possible to maintain a subject’s behavior in situations in which it may be inconvenient to monitor and reinforce it frequently (Israel, 1978). Maintenance of correspondence has been targeted using the delayed reinforcement procedure described above (Baer et al., 1984). Whitman, Scibak, Butler, Richter, and Johnson (1982) used this technique to program maintenance of correspondence regarding appropriate posture. During the correspondence condition, subjects stated that they would sit correctly during an upcoming class period, and reinforcement was delivered afterwards, contingent on having done so. During the maintenance condition, reinforcement was delivered after the class period, but was contingent only on having emitted the statement to sit correctly beforehand. The delayed reinforcement procedure successfully programmed maintenance of appropriate posture for periods of up to 15 days. Similarly, Baer et al. (1983) demonstrated that the delayed reinforcement procedure resulted in maintenance of correspondence with previously trained responses for periods of up to 17 days. Both Whitman et al. (1982) and Baer et al. (1983) showed considerably longer periods of maintenance than had been seen in most other correspondence training studies. A disadvantage of this delayed reinforcement procedure is that it requires implementation of the entire intervention procedure each day, even during the maintenance and generalization phases. An approach to programming maintenance that permits the extent of the intervention to be reduced over time might prove more cost-effective. Karlan and Rusch (1982) suggested that intermittent application of the correspondence contingencies, in which the opportunity to correspond is present daily while consequences for correspondence are presented less frequently, might successfully program maintenance of correspondence. Baer, Blount, Detrich, and Stokes (1987) investigated this possibility using nutritious snack choices by preschoolers in a daycare center. After correspondence between statements to choose nutritious foods for snack and actual snack choices had been established, an intermittent reinforcement procedure was introduced, in which consequences for correspondence
Correspondence Training
385
were presented gradually less frequently over time. Nutritious snack choices maintained during the intermittent reinforcement condition and during a subsequent extinction condition in which the children continued to state daily that they would choose nutritious snacks, but no programmed consequences were provided. This study suggested that intermittent reinforcement of correspondence can be a useful strategy for promoting maintenance. Guevremont, Osnes, and Stokes (1986b) combined an intermittent reinforcement procedure with other procedures designed to make contingencies indiscriminable. On some days, positive consequences were delivered immediately following prompted verbalizations, but not following correspondence. On other days, consequences were delivered following correspondence but not after the prompted verbalization. On some days, no verbalizations were prompted and no consequences were provided. On other days, no consequences were delivered for correspondence, but delayed positive consequences for prompted verbalizations were provided. On still other days, no consequences were provided for either prompted verbalizations or correspondence. Results suggested that this mixed sequence of conditions promoted maintenance of correspondence, both during these conditions and during an extended return to the baseline phase. REMOTE SETTINGS AND TIMES
Several authors have suggested that if verbal control over corresponding nonverbal behavior could be achieved, then desirable behavior in an inconvenient or inaccessible setting, or at a remote time, could be controlled through control of corresponding verbal behavior in a more convenient setting or time. A parent, for example, might be able to increase a child’s rate of compliance at school by prompting and reinforcing at home such statements as, “Today, I’m going to follow my teacher’s instructions.” This possibility was explored in studies incorporating increased time intervals and sometimes a change of setting between the verbalization and the opportunity to engage in the promised response. For example, Baer et al. (1983) taught a preschool child to make statements, at school, that he would perform specific behaviors at home either later that day or early the next morning before returning to school (e.g., choosing fruit for dessert after dinner, putting away clothes after his bath). Thus, several hours passed between the verbalization and the opportunity to engage in the stated response. Further, the verbalization and the promised behavior occurred in two different settings. Similarly, Guevremont et al. (1986a) found that the time interval between the verbalization and the opportunity to engage in the corresponding nonverbal behavior could be systematically increased from 20 minutes to over two
R. A. Baer
386
hours. Generalization ferent setting also was to train correspondence verbalized promise to be achieved.
to a response occurring several hours later in a difobserved. These studies indicate that it is possible with behaviors that are temporally distant from the perform them, and that control across settings can
CORRESPONDENCE
TRAINING
AND SELF-CONTROL
Several authors have suggested that correspondence training procedures are related to the concept of self-control. For example, Kanfer and Karoly (1972) suggested that a statement of intention may serve as a discriminative stimulus for performing the intended response. Correspondence training, then, could be a useful self-control technique if a history of such training would enhance the discriminative function of a verbal self-controlling response (Bijou & Baer, 1978; Israel, 1978). This self-control technique could be used to facilitate generalization across settings and maintenance over time, in that the individual could produce his or her own cues, in those settings and times, for which controlling functions have been developed (Karlan & Rusch, 1982). Recently, some authors have begun to point out a conceptual flaw in the assumptions that underlie much of the correspondence training literature, and that relate correspondence training to the concept of self-control. Correspondence usually is viewed as a set of topographically distinct responses (“doing”) which share a common controlling stimulus (“saying”) (Stokes, Osnes, & Guevremont, 1987). In a successful correspondence training study, the subject engages in the target behavior when he previously stated that he would, but not when he did not make this statement. This leads to the assumption that the child’s verbalization is the stimulus controlling the occurrence of the target response. Because this controlling stimulus is an antecedent cue emitted by the subject, many authors have described correspondence training as a form of self-regulation, or self-control, or as a method of promoting self-regulation (Guevremont et al., 1986a; Israel, 1978; Kanfer & Karoly, 1972; Karlan & Rusch, 1982). The terms “generalized correspondence, ” “generalized verbal control,” and “verbal regulation” are commonly equated (Baer et al., 1984; Guevremont et al. 1986a; Stokes et al., 1987; Williams & Stokes, 1982), reflecting the view that the goal of the procedures is to bring the child’s behavior under the control of the child’s own verbalizations. Matthews, Shimoff, and Catania (1987) questioned the assumption that the child’s verbalization exerts functional control over the child’s subsequent behavior, and that this process could therefore be viewed as an example of self-regulation. They pointed out that the subjects’ responses might
Correspondence Training
387
actually be controlled by the experimenter’s prompt, or by the subsequent reinforcement, or both, with the subjects’ verbalization being functionally unnecessary (Catania, Shimoff, & Matthews, 1987). If the subjects* verbalizations can be omitted with no effect on the outcome, then correspondence training may not be a self-control procedure. Until recently, published research did not systematically examine this possibility. Deacon and Konarski (1987) trained one group of mentally retarded adults with typical correspondence training procedures. Manipulations of buttons and levers on an experimental apparatus served as target behaviors. Another group received reinforcement for engaging in the target behaviors, but made no prior verbalizations. Both groups showed increases in the target behaviors over baseline levels, and no differences between the groups were observed. Four of six members in each group apparently developed generalized correspondence. Deacon and Konarski (1987) suggested that the concept of rule-governed behavior could best explain their results. The experimenter’s verbal prompts, verbal feedback, and provision of reinforcement provided sufficient stimuli for the subjects to develop a rule, such as, “To get the reinforcer, I have to do what I say (or what the experimenter says).” If correct, this interpretation implies that the verbalization by the subject is functionally unnecessary to the performance of the target behavior, because the rule can be generated and followed in the absence of verbalizations by the subject. Baer, Detrich, and Weninger (1988) provided additional data. They viewed the typical correspondence training procedure as consisting of four components: the experimenter’s prompt (“What are you going to do today?“), the child’s verbalization (“I’m going to do X.“), the opportunity to engage in the target behavior, and reinforcement. The purpose of their first experiment was to compare a condition in which all four of these components were included to a condition in which the child’s verbalization was omitted but the other three components occurred. These two conditions were systematically compared across several responses in each of three children. The authors reasoned that, if the child’s verbalization plays a functional role, then different results should be seen when the verbalization is included than when it is omitted. No such difference occurred, however. The authors interpreted this as a failure to find evidence of the functional role of the child’s verbalization. The second experiment of this study added another condition, in which no antecedent prompt by the experimenter occurred. Results of this experiment suggested that an antecedent stimulus is important for consistent responding, but that, again, it makes no difference whether the child verbalizes the response to be performed. Weninger and Baer (1990) added further evidence in a study investigat-
388
R.A.Baer
ing the functional role of the child’s verbalization in a correspondence training procedure with a time delay between verbalization and opportunity to respond. Once again, no difference in rate of responding was found between conditions in which the child verbalized the behavior to be performed (worksheet completion) and conditions in which the experimenter provided an antecedent prompt but the child did not verbalize. This line of research casts doubt on the commonly held assumption that correspondence training represents a form of verbal self-regulation, because it suggests that the verbalization actually is not exerting functional control over the target response. As Stokes et al. (1987) stated, “a verbalization and a subsequent nonverbal behavior may covary systematically but both be occasioned by a third variable, for example, an experimenter’s prompt. Here, the relationship is purely correlational and “saying” would not be a necessary component in the sequence, that is, in occasioning doing” (p. 162). Clearly, “verbal control” and “self-regulation” are inaccurate labels for this process if the target response is not functionally related to the emission of the verbalization. However, this research does not prove that self-regulation is not occurring in correspondence training. It simply fails to find evidence that it is occurring. It is possible that two types of regulation are occurring in these studies, and that they are equally effective. That is, during reinforcement of engaging in the target behavior, responding may be controlled by antecedent cues and subsequent reinforcement provided by the experimenter, while during reinforcement of correspondence, self-regulation may be occurring. Other issues remain to be investigated in this research. For example, a systematic investigation of the role of verbalization in generalization of correspondence to untrained behaviors is important. Whether a history of training in which the child emits verbalizations leads to more widespread or more consistent generalization than a history in which the child does not make these verbalizations should be investigated. Another issue warranting investigation is the nature of the child’s verbalization. In most studies to date, the content of the verbalization has been determined by the experimenter. Whether differing results might be seen if the child determined the content of the verbalization is an unresolved question. APPLICATIONS WITH MENTALLY RETARDED ADULTS An important area to which correspondence training procedures have been applied is vocational training in retarded individuals. Rusch and his colleagues have made numerous contributions to this area (e.g., Crouch et
Corresporuhce
Training
389
al., 1984; Gifford, Rusch, Martin, & White, 1984; Rusch, Martin, Lagomarcino, & White, 1987; Rusch, Martin, & White, 1985). These authors note that training the mentally retarded to work as employees in nonsheltered, competitive employment settings has recently emerged as an important topic in the literature. As special training or supervision may not be available for such individuals in these settings, the issue of maintenance of previously learned work skills assumes critical importance. Self-control procedures, viewed by Rusch et al. (1985) as procedures that incorporate a combination of externally and self-produced cues that control responding, seem well suited to address this issue. Correspondence training procedures become important in this context because they involve self-emitted cues that may guide correct performance. Crouch et al. (1984) pointed out that, in order to be effective, worker produced cues must be reliably produced and must act as controlling stimuli. Correspondence training, then, becomes useful because it provides a history of reinforcement for correspondence between verbalizations and subsequent actions. Crouch et al. (1984) applied the correspondence training paradigm to moderately mentally retarded adult employees, in the food service of a university dormitory, who needed to perform specific tasks, such as sweeping, mopping, and setting up a lunch line, more quickly. During the Reinforcement of Verbalization condition, subjects were prompted, a few minutes before they were to begin the task, to state at what time they would begin. About half-way through the task, they were prompted to state at what time they would finish. Praise followed each statement. During the Reinforcement of Correspondence condition, subjects were prompted to make these statements, and they were followed with a neutral response, such as “OK.” Praise was provided contingent on correspondence between the subjects’ statements and their actual work time. Results showed that the Reinforcement of Verbalization condition resulted in significant decreases in work time and increases in prompt completion of tasks. These changes were consistent and stable. The Reinforcement of Correspondence condition was not necessary. These results differ from those of many other studies, which show that Reinforcement of Verbalization is generally not sufficient to promote consistent, lasting changes in responding (e.g., Brodsky, 1967; Guevremont et al., 1986a). Rusch et al. (1987) found similar results. In this study, a moderately retarded adult woman, employed as a lobby cleaner in a fast food restaurant, served as the subject. She was taught to state, in the correct order, each task that she would perform as part of her set-up and clean-up routines. During baseline, she generally completed only about half of the required tasks. During Reinforcement of Verbalization, however, she com-
390
R. A. Baer
pleted between 80 and 100% of the required tasks, in the correct sequence. As this condition was very successful, a Reinforcement of Correspondence condition was not necessary. Again, this is not typical of many other correspondence training studies. The explanation for the discrepancy between the results of these studies and the results of many other studies is not clear. It is possible, as the employees in both studies were likely to lose their jobs if their behavior did not improve, that avoidance of being fired played an important functional role. The verbalizations may have provided the necessary antecedent stimuli to prompt correct responding. In addition, these studies share the conceptual flaw discussed above, in that it was not demonstrated that the verbalizations by the subjects were necessary to promote correct responding. The subjects were always prompted to make these statements. Prompts included instructions to make the statements and modeling of what should be said. It is possible that these extensive prompts provided sufficient antecedent stimuli to cue appropriate work behavior, and that verbalizations by the subjects were unnecessary. A comparison condition similar to that employed by Baer et al. (1988), in which the prompter explained to the subject what to do, but did not require the subject to say anything, might have yielded similar results. Given the necessity of these extensive prompts each day, the verbalizations by the subject did not appear to decrease the amount of supervision required. Thus, the procedure’s value as a self-control strategy can be questioned. However, the procedure’s value as an effective behavior change strategy is very clear. CONCLUSIONS A reasonable question often asked of correspondence training researchers is, “Why reinforce verbalizations about the desired response rather than simply reinforce the response itself?” The answer is that the ultimate goal of correspondence training procedures is often the development of verbal control. If verbal control is developed, when a behavior change is desired, the change agent need only evoke from the subject a verbalized intention to behave appropriately. Generalization, of course, is the key to verbal control. To achieve verbal control, correspondence must generalize from the few responses with which it is trained, to other responses. Unfortunately, it is not yet clear how generalized correspondence can best be achieved. Although some studies have documented generalized correspondence, the variables that control the occurrence or nonoccurrence of generalization remain unclear. When generalized correspondence does occur, it often requires the continued use of reinforcement, such as the delayed reinforcement of verbalization procedure described above. In
Correspondence Training
391
addition, even when correspondence occurs in the absence of any reinforcment contingent on it, it remains to be demonstrated that the occurrence of the target response is actually controlled by the child’s verbalization per se, rather than by the experimenter’s prompt. Thus, from a theoretical perspective, correspondence training presents unresolved issues. Whether correspondence training should be viewed as a self-control or self-regulation procedure is not yet clear. Future research in correspondence training is likely to help clarify this important question. From an applied perspective, it appears that correspondence training will continue to be a valuable intervention technique for a variety of problems and populations. Its applicability to a wide range of behaviors and individuals has been well documented. Yet, unanswered questions remain in the applied arena. For example, under what conditions are the subjects’ verbalizations necessary or helpful, and under what conditions are they supcrfluous? Perhaps subjects’ verbalizations aid in acquisition of new responses, but are unnecessary for changing the frequency of responses already in the repertoire. Perhaps verbalizations are simply antecedent cues whose source (subject or experimenter) is irrelevant. Additional research is needed to clarify these questions. Perhaps the verbalizations in correspondence training procedures are not necessary, but are still valuable. Interventions that involve verbalizations regarding intentions to perform desirable responses are likely to be appealing to people in general. Behavioral methods are often viewed as manipulative and controlling by lay persons. Using interventions that appear to locate some control within the individual rather than exclusively with external change agents may eventually help to change the image of behavioral technology among the general public. In summary, correspondence training procedures are both practically useful and theoretically interesting. Many questions await additional research. REFERENCES Baer. R. A., Blount, R. L.. Dettich. R., & Stokes, T. E (1987). Using intermittentreinforcement to program maintenance of verbal/nonverbal correspondence. Journal of Applied Behavior Analysis, 20, 179-184. Baer. R. A., Detrich, R., & Weninger, J. M. (1988). On the functional role of the verbalization in correspondence training procedures. Journal of Applied Behavior Analysis, 21.345356. Baer, R. A., Osnes. P. G.. & Stokes, T. F. (1983). Training generalized correspondence between verbal behavior at school and nonverbal behavior at home. Education and Trearmenr of Children, 6.379-388. Baer, R. A., Williams, J. A., Osnes, P. G.. & Stokes. T. F. (1984). Delayed reinforcement as an indiscriminable contingency in verbal/nonverbalcorrespondence training. Journal of Applied Behavior Analysis, 17.429440. Baer, R. A.. Williams, J. A., Osnes. I? G. & Stokes, T. E (1985). Generalized verbal control and correspondence training.Behavior Modrficarion,9.477489.
392
R. A. Baer
Bijou. S. W.. & Baer, D. M. (1978). Behavior analysis of child development. EtnglewoodCliis, NJ: Prentice-Hall, Inc. Brodsky, G. (1%7). The relation between verbal and nonverbal behavior change. Behaviour Research and Therapy, 5,183-19!. Catania, A. C., Shimoff, E., & Matthews, B. A. (1987). Correspondence between definitions and procedures: A reply to Stokes, Osnes. & Guevremont Journal of Applied Behavior Analysis, 20,4ol-m4.
Crouch, K. F’.,Rusch, E R., & Karlan, G. R. (1984). Competitive employment: Utilizing the correspondence training paradigm to enhance productivity. Education and Training of the Mentally Retarded, 19.268-275. Deacon, J. R.. & Konarski, E. A. (1987). Correspondence training: An example of rule-governed behavior? Journal of Applied Behavior Analysis, 20.391400. Gifford, J. L., Rusch. F. R., Martin, J. E., & White, D. M. (1984). Autonomy and adaptability: A proposed technology for maintaining work behavior. In N. Ellis & N. Bray (Eds.), International review of research on mental retardruion (Vol. 12, pp. 284-314). New York: Academic Press. Guevremont, D. C., Osnes, P G., & Stokes, T. F. (1986a). Preparation for effective self-regulation: The development of generalized verbal control. Journal of Applied Behavior Analysis, 19,99-104. Guevremont, D. C., Osnes, P. G., & Stokes, T. F. (1986b). Programming maintenance after correspondence training interventions with children. Journal of Applied Behavior Analysis. 19, 215-219.
Israel, A. (1978). Some thoughts on correspondence between saying and doing. Journal of Applied Behavior Analysis, 11,271-276. Israel, A., & Brown, M. (1977). Correspondence training, prior verbal training, and control of nonverbal behavior via verbal behavior. Journal of Applied Behavior Analysis, 10.333-338. Israel, A., & O’Leary. K. D. (1973). Developing correspondence between children’s words and deeds. Child Development, 44.575-581.
Jewett, J.. & Clark, H. B. (1979). Teaching preschoolers to use appropriate dinnertime conversation: As analysis of generalization from school to home. Behavior Therapy, 10,589-605. Kanfer, F. H., & Karoly, P. (1972). Self-control: A behavioristic excursion into the lion’s den. Behavior Therapy, 3,398416.
Karlan, G. R., & Rusch, F. R. (1982). Correspondence between saying and doing: Some thoughts on defining correspondence and future directions for application. Journal of Applied Behavior Analysis, 15, 151-162. Karoly. P., & Dirks, M. (1977). Developing self-control in preschool children through correspondence training. Behavior Therapy, 8.29-05. Lovaas, 0. I. (1964). Control of food intake in children by reinforcement of relevant verbal opcrants. Journal of Abnormal and Social Psychology, 18.672-678. Mauhews, B. A., Shimoff, E., & Catania, A. C. (1987). Saying and doing: A contingency space analysis. JOU~M~of Applied Behavior Analysis, 20,69-74. Osnes. I? G., Guevremont. D. C., & Stokes, T. F. (1986). If I say I’ll talk more, then I will: Correspondence training to increase peer-directed talk by socially withdrawn children. Behavior Modification, 10.287-299. Risley, T. R., & Hart, B. (1968). Developing correspondence between the nonverbal and verbal behavior of preschool children. Journal of Applied Behavior Analysis, 1,267-281. Rogers-Warren, A., & Baer, D. M. (1976). Correspondence between saying and doing: Teaching children to share and praise. Journal of Applied Behavior Analysis, 9.335-354. Rusch, F. R., Martin, J. E., Lagomarcino, T. R., & White, D. M. (1987). Teaching task sequencing via verbal mediation. Education and Training in Mental Retardakm. 22. 229-235. Rusch, F. R., Martin, J. E., & White, D. M. (1985). Competitive employment: Teaching mentally retarded employess IO maintain their work behavior. Educarion and Training of the Mentally Retarded, 20, 182-189.
Correspondence Training
393
Sherman, J. A. (1964). Modification of nonverbal behavior through reinforcement of related verbal behavior. Child Development. 35.7 17-723. Stokes, T. F... & Baer, D.M. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis. 10.349-367. Stokes, T. F.. Gsnes. R G., & Guevremonf D. C. (1987). Saying and doing: Applied, research. and conceptual issues in verbal regulation. Journal of Applied Behavior Analysis, 20, 161-164. Weninger, J. M., C Baer. R. A. (1990). Correspondence training with time delay: A comparison with reinforcement of compliance. Edncnfion and Treartnenl ~Children, 13,3&%X. Whitman, T. L., Scibak, J. W.. Butler, K. M., Richter, R., & Johnson, M. R. (1982). Improving classroom behavior in mentally retarded children through correspondence training. JOIUM~ of Applied Behavior Analysis, 15.545-564. Williams, J. A., & Stokes, T F. (1982). Some parameters of correspondence training and generalized verbal control. Child and Family Behavior Therapy, 4,l l-3 1.