The importance of multiple exemplar instruction in the establishment of novel verbal behavior
Use multiple exemplar instruction—rotating varied examples—to make new, untaught verbal responses appear.
01Research in Context
What this study did
LaFrance et al. (2020) wrote a how-to paper, not an experiment. They explain why you should say 'multiple exemplar instruction' instead of 'multiple exemplar training.'
The authors map the idea onto Skinner’s verbal behavior terms. They show how rotating many examples can create brand-new things a learner has never been taught to say.
What they found
The paper argues that when you vary the teacher, the place, the question wording, and the item set, untaught sentences pop out. The child might say 'white dog runs' even though you only taught 'black cat jumps.'
They call this effect 'novel verbal behavior.' It is the verbal version of stimulus equivalence.
How this fits with other research
Marya et al. (2021) gives a live demo. They used matrix training with kids with autism who used speech-generating devices. After teaching only the diagonal items, the children produced untaught noun-verb tacts. This extends LaFrance’s idea to an SGD population.
Older lab work set the stage. White et al. (1990) showed that training a few sequences created whole ordinal classes. Lazar (1977) got adults to emit untrained sequences after simple matching-to-sample. These studies are the grandparents of the verbal-generativity claim.
Sailor (1971) is an early cousin. Two children learned one plural sound and then used the other plural sound on new words. The result mirrors LaFrance’s point: train a few examples, get new ones free.
Why it matters
Stop chaining endless discrete trials. Pick a set of three to five varied examples that share the same pattern. Rotate them across people, rooms, and materials. Then probe for brand-new responses the next day. If they appear, you just saved hours of direct teaching time.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Pick one target structure (e.g., 'I see ___'). Teach it with five different nouns, two different voices, and two settings. Tomorrow ask for 'I see' plus a never-taught noun and record what happens.
02At a glance
03Original abstract
According to traditional linguistic accounts language, and its generative nature, cannot be taught. From a behavior analytic perspective, language is like any other behavior; it is learned and amenable to change. Based upon Skinner's radical behavioral analysis of verbal behavior, specific procedures have been designed to promote novel verbal relations. However, despite the strength and utility of this approach, using behavioral principles to understand the generativity of language has been challenging. Dependent upon the specific theory (e.g., stimulus equivalence, relational frame theory, bidirectional naming) within the radical behavioral orientation, researchers arrange unique procedures to evaluate the variables responsible for this phenomenon. This paper presents the commonalities and differences of two procedures (i.e., multiple exemplar training, multiple exemplar instruction) with examples of research highlighting the use of both in producing generativity. Further, it describes how multiple exemplar instruction is independent from other procedures leading to this outcome, and concludes by providing recommendations for both research and practice.
Journal of Applied Behavior Analysis, 2020 · doi:10.1002/jaba.611