Development of the Pure Mand

According to Mark Sundberg (2005), “A common problem faced by many children with autism is that they are unable to mand for items that are not present.” What Dr. Sundberg is referring to is the challenge of teaching students to emit pure mands. A pure mand is a basic operant defined by Skinner in Verbal Behavior as a “a verbal operant in which the response is reinforced by characteristic consequences and is therefore under the functional control of relevant conditions of deprivation or aversive stimulation”. Pure mands are not under the control of discriminative stimuli such as the item being present or verbal prompts to mand (i.e., “what do you want?”). As noted, in day to day interactions, mands are usually controlled by a variety of factors so that pure mands are probably a relatively rare phenomena. Nonetheless, there are times when people ask for things that are not present. Children with autism may be particularly prone to not developing pure mands.

There has been research that addresses the issue of teaching spontaneous requesting behavior. Most of the studies suggest use of a time delay procedure in fading prompts to teach spontaneity of mands. (Halle, Marshall, & Spradlin, 1979; Halle, Baer, & Spradlin, 1981; Carr & Kologinskym 1983; Charlop, Scheibman, & Thibodeau, 1985; Bennet, Gast, Wolery, & Schuster, 1986; Matson, Sevin, Frideley, & Love, 1990; Ingenmey & Van Houten, 1991; Charlop & Trasowech, 1991; Matson, Sevn, Box & Francis, 1993)

The goal of teaching spontaneous mands is to allow the student to mand without the item being physically present. As we have consistently suggested, teaching the initial mand repertoire will likely be facilitated by teaching in conditions when multiple components occur in the antecedent condition including having the item present. Systematic procedures can now be implemented which allow the fading of discriminative stimuli in the period antecedent to the mand.

The systematic procedures we will review have been outlined in research completed at the Carbone Clinic in Valley Cottage, New York. (Adapted from Sweeney, et al. (2005).

As part of this process instructors will need to select reinforcers to be used in spontaneous mand training that are likely to have a consistently strong motivative operation and are firmly in the student’s mand repertoire when the item is present.

The first step is to conduct a 30 minute probe to determine the current degree to which the mand is under control of motivative operations. The probe session will determine baseline levels of spontaneous manding. If the item is under the control of the motivative operation (learner mands for the item during 30 minute probe session) repeat probe session for 3 consecutive days. If the student mands for the item on all three probe sessions, begin generalization and/or maintenance procedures and start the same probe procedure with a new item.

If the learner does not mand for the item during the probe session, begin sessions that include a time delay between trials to mand with item present. In other words, present the item to the learner, and wait for the learner to mand for the item. Record the student’s mand on your data sheet as prompted mand. Then put the item away and set a timer to begin a 2-minute interval. If the learner mands for the item within the first 15 seconds, deliver the item and record the response as a prompted mand on the data sheet. Again, restart the 2-minute interval. If the learner mands for the item after 15 seconds, deliver the item and record the response as a spontaneous mand. Again put the item away and restart the 2-minute interval. If the learner does not mand for the item after 2 minutes pass, represent the item, wait for the learner to mand for the item, and record the mand as prompted if it occurs. If the student is not motivated for the item, record the trial as no motivation. Restart the timer for another 2-minute interval. Continue this procedure throughout the session.

Prior to each training session a 30 minute probe for spontaneous manding will need to occur. The training procedures would be discontinued for the target item when at the start of each session the first mand for the item occurs within the probe session for three consecutive days in which the student has motivation for the item.

Although the Sweeney, (2006) study focused on a constant duration of time delay interval, it has been suggested based on a case study within the PaTTAN Autism Initiative, Applied Behavior Analysis Supports (Chamberlain, 2008) that use of a gradually increasing time delay interval (from two minutes to longer intervals) may further facilitate acquisition of spontaneous mands.

From Signed Mands to Vocal Mands

The review of these procedures is rather cursory. It is strongly suggested that when implementing procedures to teach vocal responding that skilled professionals familiar with the analysis of verbal behavior and speech and language development be included on the team.

As stated earlier, in the section on selection of an augmentative communication system, when choosing a response form consideration should be given to how the system lends itself to transfer to vocal verbal behavior. Sign language may be the most efficient augmentative system for transference to vocal response forms. This is because of its topographical nature and sign language’s ability to produce functional control across operants; in other words, a sign can serve mand, tact, and intraverbal functions. Several studies have suggested that sign language can promote the development of vocal verbal behavior in some individuals with autism and developmental disabilities (Mirenda & Erickson, 2000; Mirenda, 2003; Tincani, 2004).

However, there is a subset of children with autism for whom sign language may not facilitate vocal production (Mirenda, 2003). In such cases it may be necessary to add other behavioral interventions to increase the development of vocal responding.

In the PaTTAN Autism Initiative, Applied Behavior Analysis Supports, we have been guided by the work of Dr. Vincent Carbone, et al. (2005, 2006) and the work of Tamara Kaspar, SLP/CCC, BCBA in regards to teaching vocal responses to students who are using augmentative communication. The work of Carbone and his collaborators has suggested four approaches to teaching vocal responses. Those approaches include:

  • Stimulus-stimulus pairing: (pair vocal model with delivery of reinforcement)
  • Automatic Reinforcement Procedure
  • Differential reinforcement of vocal responding in the mand frame
  • Echoic training (Kaufman Apraxia procedures)

We will review only one of those procedures here, namely the differential reinforcement of vocal responding in the mand frame.

The differential reinforcement of vocal responding involves a time-delay prompt procedure. An opportunity to mand is presented and the instructor waits for several seconds before delivering a prompt to give the learner the opportunity to emit the vocal response independently. If the student responds in the interval before a prompt is given, differential reinforcement is delivered. Less reinforcement is delivered if prompting is necessary. The basic steps in this process are:

  • Select target vocal response in relation to strong signed mands. The hierarchy of vocal approximations leading to an adult form can be derived from inventories such as the Kaufman Speech Praxis Treatment Kit for Children (Kaufman, 2005). Here is an example of a hierarchy of sounds for shaping the vocal word “block” (from adult form at top, with each successive level of simpler form below):
    • Block
    • Bu walk
    • Bwahk
    • Bahk
    • Bah
  • When sign occurs momentarily withhold reinforcement
  • Say the name of the reinforcer up to three times with a one second pause between each presentation
  • Reinforce immediately delivering the item if a target approximation is produced
  • Reinforce the signed mand (with less reinforcement than if target vocal response occurred) after the third presentation, even if a vocal approximation does not occur.

Here is a treatment integrity checklist related to these procedures:

Chart 16

Procedural Integrity Checklist
Differential Reinforcement of Vocal Responding with Signed Mands
Instructor name: _______________  IOA check? yes no (with:___________)
Date: ________________________  Student: _____________________
  • Did you make sure all mand items are present during the session?  Yes  No
  • Did you have a variety of mand items available?  Yes  No
  • Did you confirm that an MO is in place for teaching items?  Yes  No
  • Did you check to be certain which mand items should be run as differential reinforcement of vocal responding?  Yes  No
  • Did you have the word shells available?  Yes  No
  • Did you intersperse trials for items that are sign only for which we are not running the differential reinforcement procedure?  Yes  No
  • Are you clear on the exact pronunciation of each word shell?  Yes  No
  • Did you hold up the item and get a clear and accurate sign response first?  Yes  No
  • If Adam responds with the target vocal response, did you reinforce immediately with greater quantity of reinforcer (better reinforcement?)  Yes  No
  • If the target vocalization is not said immediately, did you present up to three times before reinforcing (but reinforcement presentation with target response)?  Yes  No
  • Did you record data on the last level of word shell consistently emitted on first trial?  Yes  No
  • Did you remember to keep the sign strong?  Yes  No
  • Did you say the adult form when delivering the reinforcer?  Yes  No

Here is a data sheet for recording the student’s performance on trials during the differential reinforcement of vocal responding with signed mands:

Chart 17


Data Sheet developed at Carbone Clinic

Learner:____________ Date:___________ Instructor:____________ Time:___________
ReinforcerPrompt LevelWhat was said during mandEchoic 1Echoic 2Echoic 3Echoic 4Echoic 5
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      
 V  ITEM MO      



For every mand the instructor will:
  • Record the reinforcer that the learner mands for ("reinforcer" column).
  • Record the prompt level necessary to evoke that response ("prompt level" column).
  • Record what the learner actually says ("what was said during mand" column).
    • If the vocalization during the vocal-mand is clear, deliver the reinforcer.
    • If the vocalization is not clear the instructor will then run up 5 echoic attempts for better articulation, recording exactly what the learner said (or not) for each.
      • If during the 5 echoic trials the learner's articulation is better, the instructor will differentially reinforce the learner depending on the trial. That is, if the learner required all five echoic trials the instructor will reinforce but not as much (magnitude) if the learner had achieved better parity during an earlier trial. That is, the learner will receive more of the reinforcer if less echoic trials were required.


At the end of the session the instructor will graph:
  • Spontaneous versus prompted mands.
    • Calculate the rate per minute for both mands that were prompted versus mands that were spontaneous (solely under the control of the MO) by dividing the frequency for each by the total number of minutes spent manding.
  • Overall prompt level needed to evoke all mands for the day.
    • After the session the instructor will calculate the most frequent prompt level needed for all of the mands for that day and graph this prompt level on the graph.
  • Percentage of vocalizations that was intelligible on first mand attempt.
    • After the session the instructor will calculate the percentage of vocalizations that were intelligible while manding by dividing the number of trials when the learner was clear when manding on his/her first attempt (i.e., those trials where the learner mands for a reinforcer and the vocalization was clear when the learner manded) by the total number of mands for that day.
  • Percentage of vocalizations that improve during the echoic procedure.
    • After the session the instructor will calculate the percentage of vocalizations that improved during the echoic procedure by dividing the number of trials where the learner's vocalization improved during any of the 5 echoic trials by the total number of trials (i.e., one trial is counted for all five, if necessary) when the procedure was implemented for that day.