Participants and Setting

Four participants from a 1st-grade inclusion classroom participated in this study. The classroom was located in a K–5 elementary school in a public school district located in a suburban, diverse town within proximity to a large metropolitan area. The students from the elementary school were selected because the school district’s elementary students demonstrated a 45% proficiency test score in reading and, in particular, only 52% of the entire school district’s students demonstrated at or above proficiency in reading. Based on racial and economic data of the participant’s school, the demographics consisted of 61% White students, 28% Hispanic students, 4% multicultural students, 4% Asian students, and 2% Black students. Also, 10% of the students within the school qualified for free and reduced-price lunch.

All participants were typically developing 1st-grade learners who read on grade level and read at a rate of 90 words per minute. There were three girls and one boy who were 7 years of age at the start of the study. All participants were selected by the 1st-grade teacher and experimenter because they required additional support with reading comprehension skills during instruction based on teacher assessment The study consisted of two dyad groupings: Participants A and B were in Dyad 1 whereas Participants C and D were in Dyad 2. The classroom consisted of 15 students, one lead teacher, and one teaching assistant. Some features of this classroom included but were not limited to large group instruction, choral responding, and a class-wide behavior management system. All procedures were conducted by the experimenter at a U-shaped table in the corner of the participants’ classroom.

At the onset of the study, all participants demonstrated the ability to read at a 1st-grade level according to the Qualitative Reading Inventory (Leslie & Caldwell, 2010). However, though the participants were reading on grade level, they each required additional support with reading comprehension. All participants were assessed for the skill of observational learning (OL) to determine whether this verbal developmental cusp was in repertoire as this cusp is an indicator of how a student learns (Greer & Ross, 2008; Neu & Greer, 2019). The assessment procedure for OL consisted of two participants set up in a dyad and the experimenter teaching one nontarget peer participant the names of pictures (e.g., five targets and four exemplars of each) while the target participant observed. The target participant observed the nontarget peer participant receive either reinforcement for correct responses or corrections for incorrect responses (Delgado & Greer, 2009; Fryling et al., 2011). After the peer participant mastered the names of the pictures, the target participant was probed for OL by being required to label the images within the set of stimuli at a criterion of 80% accuracy. Results from the OL assessment showed all four participants had OL in their repertoire, which confirmed for the experimenters that the participants could learn in a dyad/group setting where they observed other students receive reinforcement or corrections (Greer & Ross, 2008; Greer & Speckman, 2009).

All participants functioned at listener/speaker, and reader/writer levels of verbal behavior. Functioning at a listener/speaker level means that the learner demonstrates joint stimulus control due to their ability to emit correct listener responses (e.g., selecting items) and correct speaker responses (e.g., labeling items) as a result of unreinforced observing experiences. Functioning at a reader/writer level of verbal behavior means that the participants were able to read words fluently, read-and-do, and emit writing behavior that had an intended effect on a reader. The following prerequisite repertoires were required to be a participant in this study: participants could read-do (i.e., follow written instructions), demonstrated grade 1 reading level, and books were conditioned reinforcers. According to the Qualitative Reading Inventory (QRI) probe and a probe that measured duration of book-engagement, the participants demonstrated the prerequisite repertoires while engaged with books (Gentilini & Greer, 2021; Greer, 2002; Greer & Ross, 2008). Although the participants had these repertoires, the experimenters selected them because they were performing below criterion according to preprobe data across all four of the dependent variables in this study

Materials

The reading comprehension curriculum used was Corrective Reading (CR; Engelmann et al., 1999a, 1999b). Researchers used the CR placement test materials to conduct the prescreenings and determine appropriate placement within the curriculum. All participants were placed on CR placement test A. The experimenters used the Corrective Reading: Thinking Basics-Comprehension A Teacher Presentation Book 1 and the Corrective Reading: Thinking Basics-Comprehension A workbook (Engelmann et al., 1999a, 1999b) to implement the intervention. The experimenters implemented the first five lessons only found within the Corrective Reading: Thinking Basics-Comprehension Level A workbook (Engelmann et al., 1999b). The experimenter selected this level of the curriculum because it matched the participants preperformance level as well as due to the pedagogical compatibility with training for relational responding.

In terms of reading assessment, the experimenters implemented the fifth edition of the QRI and it was used to assess implicit and explicit reading comprehension skills during pre- and postintervention probes (Leslie & Caldwell, 2010). The QRI assessment worksheets consisted of one Level 1 (i.e., 1st-grade level) story and a Level 1 QRI questionnaire packet. The questionnaire packet consisted of comprehension questions and a scoring section. The experimenters used the test materials from the Level 1 stories because Level 1 was equivalent to 1st-grade reading level and all participants were on 1st-grade reading level at the start of the study. The following QRI stories were used, Mouse in the House, The Bear and the Rabbit, and Marva Finds a Friend (Leslie & Caldwell, 2010). Samples of the QRI assessment sheets are shown in Fig. A1 in Appendix A and Fig. B1 in Appendix B.

Fig. 1

Sample of Sets 1–3 from the Metaphor Probe. The participants were required to attend to the teacher vocal antecedent and respond by providing three metaphor responses. The participants were provided with five opportunities per set to demonstrate that extended metaphors were in repertoire

The experimenter constructed metaphor probes that consisted of sentences that were presented vocally and read from a word document. The document listed questions for each set within the metaphor probe. The participants responded in vocal form. The probe materials used for the derived relations from sentences probe consisted of sentences that were presented vocally and read from a document. The document listed questions for each set within the probe. The participants responded in vocal form. The sentences presented on the probe were developed by the researchers. The material used for the derived relations from the letters/numbers probe, which used arbitrary combinations of letters/numbers consisting of alphabetic letters and Arabic numbers, was visually presented to the participants. There were 10 opportunities to respond and they were presented as such: 1) Visual of “X = G,” 2) visual of “G = T, and then 3) the question to test for combinatorial entailment was presented visually, “T = ?” These dependent measures are explained further in the procedures section.

Dependent Variables

There were four dependent variables measured in this experiment and each variable expounded upon reading comprehension from a scientific behavioral approach. In the current study we evaluated derived relations for combinatorial entailment in three ways. First, we probed metaphors that were exclusively based on the learner’s instructional history without explicit rules provided (e.g., “Name three ways a house is like a seashell”). Next, we also evaluated derived relations from sentences (i.e., deductions) that provided explicit rules because that format was similar to the presentation of deductions in Corrective Reading (e.g., “if a chicken is coral it has 3 legs. Bob’s chicken is coral, what else do you know about it?”). Finally, to evaluate the generality of the thinking skill, we probed combinatorial entailment outside of a story-based or learning history context by presenting arbitrary letter and number relations (e.g., X = G, G = T, so T = ___ ).

Overall, the dependent measures were: (1) number of correct metaphor responses; (2) number of correct derived relations from sentences probe; (3) number of correct derived relations from the letters/numbers probe; and (4) number of correct responses to the QRI reading comprehension probe. The first three dependent variables were used to answer the first research question: to what extent does the CR curriculum sequence affect early readers’ metaphors and derived relations as measured by isolated metaphor probes, sentence-based deduction probes, and arbitrary relation probes using letters and numbers? The final dependent variable was used to answer the second research question: to what extent can beginning readers being taught with CR demonstrate advanced reading comprehension skills.

Metaphor Probe

The purpose of the metaphor probe was to test for the presence of combinatorial entailment. There are four different rudiments of metaphors according to Stewart and Barnes-Holmes (2001). The four components of metaphors are: (1) the learner establishing two separate equivalence relations; (2) the learner deriving an equivalence relation between the two relations; (3) the learner discriminating a formal relation via this equivalence–equivalence relation; and (4) a transformation of function takes place based on the formal relation discriminated in the third element. An example of a metaphor provided by Stewart and Barnes-Holmes (2001) is, “struggling with anxiety is [as] struggling in quicksand.” In this example, both anxiety and quicksand involve a form of struggling (psychological or physical); however, the listener contacts the fact that struggling in quicksand leads to drowning based on his previous reinforcement history and subsequently begins to relate that to anxiety. Therefore, the listener derives relations between the two different events and a transfer of function from one stimulus to the other occurs (Stewart & Barnes-Holmes, 2001, Stewart et al., 2001). The structures tested were the commonalities of features or noncommonalities of features between explicit stimuli and often these were arbitrarily applicable (e.g., “name three ways that a house is like a seashell”). See Fig. 1 for an example of the metaphor probe.

Derived Relation from Sentence Probe

The purpose of the derived relation from sentence probe was to test for the presence of combinatorial entailment (e.g., learning relations between two or more stimuli), after being trained relations between two stimuli or events presented in sentence form. In terms of reading comprehension, combinatorial entailment occurs when a teacher teaches her students that poodle (A) is a type of dog (B), which is relation-1, and that dogs (B) bark (C ), which is relation-2, and the teacher assesses for implicit comprehension by asking the students, “Do poodles bark?” The learners who make the appropriate relation that poodles (A) bark (C) is demonstrating the untaught derived relation that A = C. Overton (1990) stated that inferences is the process by with an individual arrives at a conclusion on the basis of other propositions that were presented to them and accepted (i.e., instructional history). The following is an example of one of the derived relation sentences found in the probe with a non-sense rule, “If a monkey is tiny, it is blue; Nick has a tiny monkey, what else do you know about it?” The participant was required to respond by stating the untaught relation which in the case of the mentioned example would be, “the monkey is blue” based on the provided rule. There was a total of five rules presented and three opportunities to respond to each rule. This probe presented two relations then probed for the untaught relation (Howarth et al., 2015). Participants were presented with a novel set of questions during each probe condition (see Fig. 2 below).

Fig. 2

This is a Sample of the Derived Relation from Sentence Probe. Note. The participants were required to attend to the experimenter’s vocal antecedents and emit vocal responses to each relation

Derived Relation from Letters/Numbers Probe

The purpose of the derived relations probe of letters/numbers probe was to determine if the participants could demonstrate combinatorial entailment (RFT) after being trained on two relations (e.g., if G = T and T = 5, then 5 = G). To transfer this to reading comprehension, combinatorial entailment occurs when an educator teaches their students that fuji (A) is a type of apple (B) which is relation-1, and that apples (B) grow on trees ( C), which is relation 2, and then the teacher assesses for implicit comprehension by asking the students, “where do fuji’s grow?” The learners who make the relation that fuji's (A) grow from trees (C) are demonstrating the untaught derived relation that A = C. Here is another example of how derived relations occur during the reading comprehension. A student learns the following relation while reading a passage, cow = milk and kefir = milk, then the untaught relation between kefir and cow can emerge for the learner (see Figs. 3 and 4). This dependent measure builds from the theoretical and scientific groundwork that explains language development; however, we are extending this to further strengthen how reading comprehension occurs after relations are trained. Similar to the transitive property of stimulus equivalence research, combinatorial entailment refers to a network of relations or a combination of two or more stimulus relations. The participants were required to demonstrate that they could learn relations between multiple numbers/letters stimuli, after being taught relations between three contrived letters and/or numbers. There was a total of 10 opportunities to respond in this probe (see Fig. 5, below).

Fig. 3

A visual of derived relations using the following set of stimuli: cow, kefir, and milk. Note. A learner’s ability to respond accurately to the third untaught relation demonstrates transitivity (stimulus equivalence) and combinatorial entailment (RFT)

Fig. 4

A Visual of How the Deduction Learn unit is Formatted in CR Using the Stimuli, Cow, Kefir, and Milk

Fig. 5

Sample of the Derived Relations from Letters/Numbers Probe Presented Using PowerPoint. Note. The participants were presented with one slide at a time and were trained for the first two relations (e.g., mutual entailment) but were probed on the last relation (e.g., combinatorial entailment)

Qualitative Reading Inventory Probe

Numerous reading assessments test for student performance with implicit and explicit comprehension skills, such as the Developmental Reading Assessment (Beaver & Carter, 2006), Fountas and Pinnell Benchmark System (Fountas & Pinnell, 2010), and Dynamic Indicators of Basic Early Literacy Skills (DIBELS) for retell comprehension (Good & Kaminski, 2002); however, in this study the experimenters measured implicit (i.e., combinatorial entailment) and explicit (i.e., mutual entailment) reading comprehension using the Qualitative Reading Inventory (QRI; Leslie & Caldwell, 2010). The QRI is a common standardized reading assessment used in grade level schools from K-12. The purpose of this dependent measure was to examine readers’ textual responses, reading rate of grade level passages, vocabulary, and comprehension responses. The experimenters used the reading comprehension questions from the QRI to assess for mutual entailment (i.e., explicit comprehension) and combinatorial entailment (i.e., implicit comprehension). For example, Level 1 story Mouse in a House, included four explicit comprehension questions and two implicit comprehension questions. The following is an example of one of the explicit questions presented in the Mouse and the House narrative, “Where did the mouse live in the house?” This is an explicit comprehension question because the answer to this question is explicitly stated in the text. This form of question demonstrates mutual entailment due to the fact that it assesses for a bidirectional relationship between two stimuli and the first relation (A = B) relation is directly presented in the text. In this example, the first relation taught was, mouse lives (A) in a wall (B). The reader simply has to demonstrate that he can respond accurately to this explicit question in both directions (i.e., A = B and B = A or mouse = “lives in wall” and “lives in wall” = mouse).

The following is an example of one of the implicit questions presented in the Mouse in the House narrative, “How many floors did the house have?” Responding accurately to this question demonstrates combinatorial entailment because the answer to this question is not explicitly stated in the text and must be inferred given the other information that the text provided (i.e., A = B, B = C, therefore A = C). The assessment provided the participants with a total of six questions following each story; the majority of the stories contained four explicit questions and two implicit questions. See Figs. A1 and B1 in the Appendices for a sample page of the QRI story and the QRI comprehension questions.

Interobserver agreement (IOA) was conducted for all four of the dependent measures across both dyads with 100% agreement. The purpose of IOA is to assess for procedural fidelity (Cooper et al., 2020). Observer agreement is a critical piece in behavior analytic research as it measures the extent to which the procedure was implemented as described (Essig et al., 2022). More specially interscorer agreement (ISA) was conducted for the QRI probes. ISA assesses for procedural fidelity by scoring permanent products (e.g., worksheets, written text, pictures). ISA was conducted for 100% of the QRI comprehension permanent product probes with 100% agreement.

Independent Variable

CR (Engelmann et al., 1999a, 1999b) is the intervention that was implemented in order to measure its effect on the dependent measures. CR has a sequence that trains derived stimulus relations across multiple exemplars within one mode of responding and these sequences allow the process of relating to become abstracted in the appropriate context (Barnes-Holmes et al., 2004). The CR curriculum provides over 30 lessons and within each lesson are multiple exercises targeting different skills related to reading comprehension (Engelmann et al., 1999a, 1999b). Each lesson within the curriculum provides 12–16 exercises and each exercise teaches comprehension skills involving true/false, inferences, deductions, and some/all/none. The experimenters implemented the first five lessons of the CR curriculum because the first five lessons provided ample opportunities for the participants to engage with the curriculum’s sequence that appears to teach derived relations.

The curriculum rotates between listener (e.g., selection responses such as multiple choice) and speaker responses (e.g., emitting a response vocally). One exercise (e.g., deductions) was considered one instructional unit. Some exercises contained multiple tasks (e.g., Task A and Task B) and each task within an exercise was also considered an instructional unit. Data were recorded on each learn unit (Albers & Greer, 1991) within each instructional unit. A learn unit is a fundamental measure of pedagogy and is composed of interlocking operants between the teacher and student, which involves teacher and student interactions (Greer & McDonough, 1999). In brief, the learn unit consists of an attending learner, the presence of a teacher or a target discriminative stimulus for teacher behavior, an unambiguous antecedent, student response to the teacher target discriminative stimulus, and a consequence for student response to the teacher behavior or discriminative stimulus (Albers & Greer, 1991; Greer & McDonough, 1999). Though many learn unit types are not discrete trials, the learn unit shares similar components to discrete trial teaching (DTT), in the sense that DTT involves a discriminative stimulus (i.e., instructional antecedent), a prompt, a response from the learner, and a consequence (Leaf & McEachin, 1999; Smith, 2001). In learn units, a consequence can include either reinforcement (e.g., praise) for correct student responses or correction operations for incorrect student responses. All learn units include correction operations; however, discrete trials may not include correction operations (Greer & McDonough, 1999). For purposes of this study, instructional unit simply refers to the skill (i.e., exercise and tasks) being taught within each lesson. An instructional unit referred to each exercise, which targeted a specific comprehension skill (e.g., True/False) and each task within each exercise was also considered an instructional unit. The criterion was set to 90% accuracy within one lesson.

Design

A concurrent multiple probe design across two dyads was used in this study (Horner & Baer, 1978). In the current study, Dyad 1 included Participants A and B and Dyad 2 included Participants C and D. The sequence of the experiment was as follows: (1) A preexperimental screening to identify the participants’ level of comprehension for the material in the CR curriculum; (2) preintervention probes targeting all four dependent measures for both dyads; (3) CR intervention for Dyad 1; (4) postprobes for Dyad 1; (5) a second preintervention probe for Dyad 2; (6) CR intervention for Dyad 2; (7) postprobes for Dyad 2; and (8) a second postprobe for Dyad 1.

ProcedurePreexperimental Screening

The experiment began with the implementation of the CR (Engelmann et al., 1999a, 1999b) prescreening assessments. The tests were conducted by following the instructions provided in the curriculum. During implementation of the CR Placement Assessment packet, the experimenters recorded the number of correct responses emitted to comprehension questions by the participants. The participants were required to respond in written and vocal form to comprehension questions delivered by the experimenter. The responses emitted during this assessment determined the participants’ placement during the CR intervention. All four participants were placed in Comprehension Level A of the CR intervention.

Preintervention probes Derived Relation from Letters/Numbers Probe

During the preintervention probe, the number of correct and incorrect responses to the derived relations from letters/numbers probe trials was assessed with a visual presentation containing an arbitrary relation made of contrived relations between letters and numbers (e.g., A = 5 and 5 = H). The experimenters presented the visual of the letters and numbers when presenting each relation and asking each question. The experimenters rotated between two forms of antecedents, “what is the same as ____?” and “what does ____ equal?” There was a total of 10 opportunities to respond and each opportunity presented a visual of each relation and the question. The participants were required to emit a vocal response within 5 s following the presentation of the antecedent. When the participants did not emit a response within 5 s, the response was recorded as incorrect by recording a minus (-) and the experimenters proceeded to the next presentation. Correction operations were not provided for incorrect responses during the probe and the correct responses were not consequated with reinforcement during the probe. The experimenters recorded a plus (+) when the participant emitted a correct response. During all probe trials, the experimenters did not consequate any correct or incorrect responses.

Metaphor Probe

During the metaphor probe, the experimenters presented one set of five questions and each question provided three opportunities to respond, which totaled to 15 opportunities to respond per metaphor question. The experimenters presented the participant with metaphor trials that consisted of an antecedent which prompted the participant to describe up to three commonalities or noncommonalities of features between stimuli such as things (e.g., animals, body parts) or events. One example of a metaphor probe trial is, “Name three ways a fin is the same as a tail.” The response was considered correct when the participant responded with one or up to three accurate commonalities or differences. The participants were required to vocally say “both are” or “they have” or “one is _______ and the other is______” in order to accurately communicate the commonality or difference. The content of the participant’s response needed to be accurate and functional. For example, if the participant emitted a vocal response such as, “a house and a seashell are the same because they both are furry” the experimenters marked this as incorrect due to the fact that the participant did not demonstrate an accurate understanding of the commonalities of features between the house and seashell given that neither are furry. However, if the participant emitted a vocal response such as, “a house and a seashell are the same because they both provide shelter,” then the experimenters considered this response as accurate. An incorrect response was recorded if the participant was unable to tact a commonality or difference between the two stimuli or only emitted a response for one of the two presented stimuli (e.g., You can live in a house) rather than both (e.g., You can live in a house and creatures live in a seashell). See Table 2 for an example of a correct and incorrect metaphor response. The experimenters recorded a plus (+) if the participant emitted a correct response and the experimenters recorded a minus (-) if the participant emitted an incorrect response. All participants’ responses were probed using a novel set after each condition.

Table 2 An example of a correct response and an incorrect response for the metaphor probe

Derived Relation from Sentence Probe. During the derived relations from sentence probe, the experimenters presented the participants with five rules (e.g., If a chicken is coral, it has three legs) per each set. The rules combined nonsense and factual content. The experimenters presented three questions following each rule to probe combinatorial entailment according to RFT (e.g., Bob’s chicken is coral, what else do you know about it?). The participants were presented with a total of five deduction rules and there were three opportunities to respond following each deduction rule. A correct response was recorded if the participant was able to identify the correct relation for each question. If the participant emitted, “I don’t know” or another response aside from the correct relation, the response was recorded as incorrect. The experimenters recorded a plus (+) for correct responses and a minus (-) for incorrect responses and the experimenter did not consequate the trials.

Corrective Reading Curriculum

During the CR (Engelmann et al., 1999a, 1999b) intervention, both dyads were presented with the first five lessons of CR. Each lesson consisted of eight or more exercises and each exercise consisted of four to eight learn units (Albers & Greer, 1991). The learn unit is the smallest component of pedagogy in which it breaks down the exchange that takes place between the teacher and the learner. As mentioned previously, learn unit consists of multiple interlocking contingencies such as a teacher delivered antecedent, a learner’s behavioral response to the antecedent, and a teacher’s response to the learner’s behavior (i.e., reinforcement or a correction operation). Each exercise was counted as an instructional unit because each exercise taught a specific comprehension skill. The CR curriculum requires that the participants master the skill at 90% criterion or above per lesson. If accuracy falls below 90%, participants need to repeat the lesson. The experimenters recorded data on each participants’ responses to learn units (Albers & Greer, 1991; Greer & McDonough, 1999) within each instructional unit. The data collected on each learn unit within each instructional unit allowed the experimenters to analyze whether the participant passed each lesson with a score of 90% or higher. If the student emitted less than 90% correct responses, the student was required to recycle the specific instructional unit to mastery. It should be noted that all students in this study emitted 90% or above during each exercise. If a participant emitted all correct responses during an instructional unit, a plus (+) was recorded. However, when the participant emitted an incorrect response within an instructional unit, the experimenters recorded a minus (-) and the student received a correction operation for that specific response. The correction operation consisted of the experimenters re-presenting the antecedent, modeling the correct answer, and then presenting the antecedent again to then allow the student an opportunity to respond correctly after receiving the model. Praise was withheld for correct responses during all correction operations. Each lesson within CR consisted of one workbook exercise. The workbook exercises required the participants to emit selection responses (e.g., circle the correct letter) when presented with positive and negative exemplars of the answer to the questions. See Table 3 below for a visual display of the total number of instructional units and total number of derived relations learn units within each lesson.

Table 3 This table shows the total number of instructional units (i.e., exercises/tasks) presented within each lesson and the total number of derived relations learn units presented within each lesson of the Corrective Reading intervention Reading Comprehension Assessment

During the QRI probe the experimenters presented each participant with a pencil and one copy of a Level 1 story packet selected from the QRI book. The Level 1 packet consisted of three pages: (1) a written story with corresponding concept-questions; (2) scoring guides for retell responses and accuracy of retell responding; and (3) comprehension questions. The participants were given the direction to independently read the narrative on the first page and answer the six comprehension questions. When it was time for the participants to answer the comprehension questions, the article containing the story was covered so that the participants could not return to it to view the story to answer the questions. The experimenters allowed the participant to view one question at a time by covering the additional questions so that the participant would only answer one question at a time and not be influenced by the other questions. The participants completed the retell portion of the QRI as well; however, for purposes of this study the data for retell was not included. The experimenter scored the comprehension worksheets and a second observer collected data and scored the comprehension worksheets independently for interscorer agreement purposes (ISA).