Verbal Behavior: A Practitioner's Guide to Skinner's Operants, Mand Training, Naming, and Relational Frame Theory

The elementary verbal operants

Skinner's account sorts verbal behavior by function rather than topography: a mand is evoked by a motivating operation and reinforced by the specific item or event requested, a tact is evoked by a nonverbal stimulus and reinforced by generalized social consequences, an echoic reproduces the point-to-point form of a preceding verbal stimulus, and an intraverbal is evoked by another person's verbal behavior without point-to-point correspondence. That functional split is the reason two topographically identical utterances, a child saying "juice" to request a drink versus naming a picture of juice, count as different operants under different contingencies, and it's why programming decisions in this file track function, not just what the learner said. Worth flagging up front: the constituent evidence behind this concept's grade is heavily weighted toward tact, listener, naming, and relational procedures. Mand-specific experimental work is thin in this particular study set, so treat mand-training decisions as informed by the broader applied literature and a functional assessment, not primarily by the citations below.

Tact training and stimulus modality

Tact acquisition research in this set concentrates on how the teaching stimulus is built. Pairing a tactile stimulus with a visual cue during initial teaching let children with autism generalize to tactile-only probes without any additional training step (Ruffo et al., 2025), and the same compound-stimulus logic held for auditory tacts, where pairing a sound with a known picture produced faster mastery and cleaner stimulus control than auditory-only trials, with visuals faded out afterward (Bergmann et al., 2023). Tact generalization across languages is a separate problem from tact acquisition. Instructive feedback delivered during first-language tact teaching produced some bilingual generalization, but not reliably enough to skip a fallback plan of brief rehearsal plus no-no prompting when the second-language tact doesn't transfer on its own (Erhard et al., 2025).

Listener responding and joint control

Listener-side programming has its own entry points, distinct from the speaker-side operants above. For learners who already imitate object-directed actions but don't yet respond reliably to spoken names, object imitation trials can bootstrap the auditory-visual conditional discrimination that listener responding depends on (Thakore et al., 2025). Once basic listener responding is underway, inserting a self-echoic step, having the learner repeat the auditory sample before selecting the corresponding picture, produced faster mastery of listener selection than a traditional conditional-discrimination drill run without it (Zhou et al., 2025), consistent with the joint-control account that repeating the sample helps bridge the delay between hearing the name and scanning the array.

Naming and bidirectional naming

Naming programming targets the point where speaker and listener repertoires for the same stimulus converge without separate training for each direction. Serial multiple exemplar training, teaching one stimulus's listener and speaker responses to criterion before moving to the next, has produced bidirectional naming in preschoolers with autism who didn't already have it (Salomonsen et al., 2024). Naming can also be shaped incidentally rather than through discrete trials: combining direct eye contact, a point, and an animated labeling voice while presenting a new item increased incidental name acquisition for preschoolers with and without autism, a low-cost addition to any natural-environment teaching moment (Hempkin et al., 2025).

Relational framing and equivalence-based instruction

The most complex operants in this set build derived, untaught relations on top of directly taught ones. Equivalence-based instruction, directly teaching a subset of stimulus relations and probing for the rest to emerge without training, has produced new vocational knowledge in a young adult (Belisle et al., 2023) and value-to-action links in autistic teens after only chain-based teaching (Chastain et al., 2026). Matrix training extends the same recombinative logic to a full grid: teaching half the cells of an object-by-preposition matrix produced correct tacts and listener responses on the untaught half without direct teaching (Lee et al., 2025). Relational frame training proper, building deictic frames such as I/you and here/there, has been demonstrated as a systematic teaching sequence in a single autistic child (Chastain et al., 2025), and every one of these relational demonstrations sits in teens or older learners with an established basic repertoire, not in early, minimally verbal learners.

Organize programming around verbal behavior specifically when a functional assessment, a VB-MAPP or ABLLS-R style tool, shows the learner's operants are uneven across function even where topography looks similar, a child who tacts an item spontaneously but can't request it, or responds to a spoken name from an array but can't say it back. A domain-based curriculum that teaches "labels" and "requests" as separate skill lists can miss that unevenness because it never asks which contingency is actually controlling the response. Assessment-driven placement, not a default program template, is what decides where a learner enters.

Within VB programming, lead with listener responding and joint control when the learner already imitates object-directed actions but doesn't yet respond reliably to spoken stimuli. Object imitation is the more available prerequisite in that case, and bootstrapping the auditory-visual conditional discrimination from it gets a listener repertoire moving faster than starting cold on conditional discrimination drills (Thakore et al., 2025). Once basic listener responding is in place, add a self-echoic step before the selection response rather than running listener trials without it, since the added repetition step accelerated mastery in a head-to-head comparison (Zhou et al., 2025).

Reach for relational and equivalence-based methods, matrix training, deictic framing, equivalence-based instruction, only after a learner has an established basic tact and listener repertoire to build on. Every relational demonstration in this evidence base sits in teens or adults with existing verbal repertoires (Chastain et al., 2025; Chastain et al., 2026; Belisle et al., 2023; Lee et al., 2025), not in early or minimally verbal learners building their first operants. Introducing relational training before basic repertoires are solid is reaching past what this evidence supports.

Be precise about what this Strong grade actually covers before leaning on it for a mand-training decision. The constituent studies here are concentrated in tact acquisition, listener responding, naming, and relational training, mand-specific experimental work is largely absent from this set. That doesn't mean mand training is unsupported in the field, it means this particular grade shouldn't be cited as the evidence base for a mand-training choice. Ground mand-priority decisions in a direct functional assessment of the learner's current requesting repertoire and the broader mand literature, not in this concept's citation list.

Start from the assessment data, not a template. Pull the domain-level scores from the learner's most recent VB-MAPP or ABLLS-R and let the uneven cells, strong tacting with weak listener responding, or the reverse, decide which operant gets programming attention this quarter, rather than running every domain at once at a shallow pace.

For tact teaching, build in the compound stimulus from session one rather than adding it later as a fix. Pair a new tactile target with a visual cue during initial teaching and probe tactile-only once the compound is at criterion, most learners generalize to the tactile-only presentation without a separate training phase (Ruffo et al., 2025). The same logic applies to auditory tacts: present the sound alongside a known picture, fade the picture once mastery is stable, and expect faster, cleaner stimulus control than teaching the sound alone (Bergmann et al., 2023). If a tact needs to transfer into a second language, don't assume instructive feedback alone will carry it, build in a brief rehearsal plus no-no prompt cycle as the planned fallback rather than the surprise fix when generalization stalls (Erhard et al., 2025).

For listener and joint-control work, use imitation as your entry ramp when spoken-word control isn't there yet. If the learner already imitates object-directed actions, follow each successful imitation with the spoken object name and a selection opportunity to start transferring control from the modeled action to the word (Thakore et al., 2025). Once listener selection trials are running, add a self-echoic step, have the learner repeat the auditory sample before pointing, rather than running a plain conditional-discrimination format by default (Zhou et al., 2025). And outside discrete trials entirely, when you label something new for a learner in the natural environment, combine direct eye contact, a point, and an animated "this is a ___" rather than a flat label, that combination measurably increases incidental name acquisition and costs nothing extra to run (Hempkin et al., 2025).

For naming and relational work, teach one stimulus's listener and speaker responses to criterion before moving to the next rather than splitting attention across many targets at once, that serial, one-at-a-time structure is what produced bidirectional naming without separate training for each direction (Salomonsen et al., 2024). Once a learner has an established basic repertoire, a 3x3 matrix of objects by prepositions, teaching half the cells directly and probing the untaught half, is an efficient way to generate new tacts and listener responses without teaching every combination by hand (Lee et al., 2025). For older or more sophisticated learners, map the direct A-B and B-C relations for a target domain and probe for the derived A-C and C-A relations before spending session time teaching them directly, equivalence-based instruction has produced new vocational and value-action knowledge this way with only half the relations directly taught (Belisle et al., 2023; Chastain et al., 2026).

Finally, don't wait for a dedicated mand program to build requesting into the day. Short auditory scripts, three or four words on a sticky note attached to materials, faded systematically after a few correct uses, can turn passive discrete-trial wait time into learner-initiated requests without overhauling the DTT structure already in place (Freeman et al., 2024). Treat that as an add-on to existing sessions, not a substitute for a functional mand assessment, since the direct mand-training literature isn't the evidence this particular tactic draws on.