Comparing mand training and other instructional methods to teach a foreign language

WAI-LING WU, SARAH A. LECHAGO AND LISA A. RETTIG UNIVERSITY OF HOUSTON–CLEAR LAKE

The purpose of the current study was to examine the effects of mand, tact, and native-to-foreign (NFI) and foreign-to-native (FNI) intraverbal training on the acquisition of a foreign language. We used a multiple-baseline design across participants with an embedded adapted alternating treatments design to compare the effects of mand training, tact training, NFI training, and FNI training on the acquisition rate of Chinese words in four typically developing adults. We also examined the emergence of untrained foreign language responses for each training condition. Data for 3 out of the 4 participants suggest that mand training was the most efficient training procedure with respect to acquisition rate. The greatest amount of emergent responding was observed for the mand and tact training conditions. Key words: foreign language, mand training, emergent behavior, intraverbal, verbal behavior

Demonstrating fluency in more than one language is valuable in an increasingly global- ized world (European Union, 2012). Published research suggests multiple benefits to learning a foreign language, including superior communi- cation skills and improved academic perfor- mance (National Education Association, 2007). For example, 50% of all companies globally want applicants to be fluent in a foreign lan- guage (Kassteen, 2014). The American Council on the Teaching of Foreign Languages (n.d.) has compiled multiple research studies to show three major areas of benefits from learning a second language. First, learning a foreign lan- guage can improve academic achievement, including improved performance on standard- ized tests and higher academic performance at the college level. Second, individuals who learn a foreign language exhibit better cognitive per- formance (e.g., problem solving; flexible and effective responding in unexpected circum- stances) compared to their monolingual coun- terparts. Third, those who learn a foreign

language also show more positive attitudes toward other cultures as compared to their monolingual counterparts. Skinner’s (1957) analysis of verbal behavior

involves a functional classification of language. Verbal behavior is categorized based on the antecedent stimuli that influence its emission and the consequences that maintain it. Foreign language teaching based on Skinner’s verbal behavior entails manipulation of environmental variables that influence language. In a tact rela- tion, the response is under the control of the physical properties of the environment and a learner names an object or a picture using the second language. For example, a child who is learning to speak Chinese may see an airplane in the sky and say “fei ji,” which is the Chinese word for airplane. With an intraverbal relation, one may translate the native word into the foreign-language word and vice versa. For example, a teacher in a classroom may ask stu- dents, “How do you say airplane in Chinese?” to which the students would reply, “fei ji.” Conversely, the teacher may ask the students, “What is fei ji in English?” to which the stu- dents would reply, “airplane.” When a learner identifies (points to or physically interacts with) an object or picture when presented with a

Address correspondence to: Sarah Lechago, Ph.D., BCBA-D, University of Houston-Clear Lake, 2300 Bay Area Boulevard, Houston TX, 77058. Email: [email protected]

doi: 10.1002/jaba.564

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2019, 52, 652–666 NUMBER 3 (SUMMER)

© 2019 Society for the Experimental Analysis of Behavior

652

foreign word, this relation would be described as listener behavior. For example, a student would point to a toy airplane when the teacher asks them to find “fei ji.” Often times, when teaching typically developing individuals, when one relation is directly trained (e.g., tact), another relation may emerge in the absence of direct training (e.g., intraverbal; Petursdottir & Haflidadottir, 2009). Emergence of untrained responses leads to the acquisition of more tar- gets with fewer resources and less time required. Skinner’s analysis of verbal behavior and contemporary analyses of generative responding (Miguel, 2016) have much to offer regarding the teaching of a foreign language. There is a limited, but growing, body of

behavior analytic research on teaching foreign languages. Petursdottir, Olafsdottir, and Aradottir (2008) conducted a study with four typically developing children, whose native lan- guage was Icelandic. The experimenters taught two participants to tact objects in Spanish and two other participants to identify stimuli when dictated a word in Spanish (listener training). After training, intraverbal relations were probed for all four participants. One intraverbal assess- ment involved the participants translating a word from Icelandic to Spanish (e.g., saying [Spanish word], when hearing “What is [Icelan- dic word] in Spanish?”), and the other intraverbal probe involved the participants translating Spanish to Icelandic (e.g. saying [Icelandic word] when hearing “What is [Span- ish word] in Icelandic?”). One participant, who received listener training, consistently emitted more foreign-to-native intraverbals (Spanish- Icelandic; FNI) than native-to-foreign intra- verbals (Icelandic-Spanish; NFI). Another par- ticipant, who received tact training, emitted more NFI intraverbals than FNI intraverbals. The two other participants responded similarly using both types of intraverbals. Overall, the results indicated that tact training required more trials to reach mastery as compared to lis- tener training. However, tact training resulted

in a greater emergence of intraverbal responding as compared to listener training. In a subsequent study, Petursdottir and

Haflidadottir (2009) directly compared the four teaching strategies previously examined, namely listener training, tact training, FNI training, and NFI training with two typically developing children as participants. The experimenters employed a multiple-baseline design across par- ticipants with an embedded adapted alternating treatments design. Prior to training a particular relation (e.g., tact), the experimenters also assessed participants’ performance on the three nontraining relations (e.g., listener responses, NFI, and FNI). After directly training one rela- tion, the experimenters conducted posttests for the emergence of the three remaining untrained relations. For example, after completing tact training with Set B, they conducted listener and both types of intraverbal posttests with Set B. The authors also included a control set of target words, for which no training was con- ducted. The results demonstrated that the rate of acquisition was lowest during the NFI train- ing condition. The tact and listener training conditions resulted in the greatest amount of emergent responding (largest number of untrained responses); however, during many of these probe sessions, responding did not reach the 80% mastery criterion. There were a couple of limitations to this study. For one of the par- ticipants, responding during the two intraverbal conditions was not trained to mastery before posttests were conducted. This may be one rea- son why there was a lack of emergence of other relations (tact, listener, and untrained intraverbal) for the sets that were trained using intraverbal relations. Second, only two partici- pants were included in the study and the results were variable, which limited the number of rep- lications. Despite these limitations, this study did much to advance a behavior analytic approach to foreign language training. More recently, Dounavi (2014) compared

the effects of tact training, NFI, and FNI, with

653MAND TRAINING TO TEACH A FOREIGN LANGUAGE

two typically developing adults. Each partici- pant received tact training for two sets of words and intraverbal training for two different sets. Dounavi used pre- and posttests to examine the emergence of untrained responses. Tact train- ing, in which participants emitted the foreign word in the presence of a picture, and NFI training, in which participants emitted the for- eign word after its native equivalent was dic- tated, resulted in emergent responding of the untrained operants to the mastery criterion. For example, if tact training was conducted, emer- gent responding to the mastery criterion for both types of intraverbal responding (NFI and FNI) was observed, and if NFI training was conducted, emergent responding to the mastery criterion for tacts and FNI was observed. For both participants, tact and NFI responding did not emerge to criteria after FNI training. The results of this study indicate that the NFI train- ing was the most efficient method of training in terms of acquisition rate. However, only FNI relations were probed during pretests. Thus, there was no direct comparison between pre- and posttests for NFI for any of the training sets. In the aforementioned studies, tact training,

listener response training, and two types of intraverbal training were compared in terms of efficiency and efficacy. To date, mand training has not been included in any foreign language acquisition study. Mands are verbal relations whose form is under the control of motivating operations (MO), and which typically specify the reinforcer (Skinner, 1957). In 2001, Sundberg and Michael recommended initiating verbal behavior training with individuals with autism by teaching mands due to the multiple benefits it brings to the speaker. Some of the benefits include obtaining desired items, having the speaker’s needs met by a listener, and mak- ing language immediately relevant to the speaker. Additionally, research with individuals with autism has shown that mand training pro- duces a greater emergence of tacts and other

verbal operants, as compared to the emergence of untrained mands after teaching a different verbal operant (Lamarre & Holland, 1985). Based on this recommendation, the present study examined the effects of mand training on the acquisition of foreign language words. The mand training condition of the current study sought to approximate real-life scenarios. We were interested in employing a practical approach to training that considered contextu- ally relevant variables (e.g., asking for food and targeting common items), as current research has been supporting the use of real-life or con- textually relevant scenarios when teaching a for- eign language (Balcikanli, 2017; Ho, Hsieh, Sun, & Chen, 2017). In the current study, we compared the

effects of mand training, tact training, FNI training, and NFI training, as described by Pet- ursdottir and Haflidadottir (2009), on the acquisition of Chinese words in typically devel- oping adults. Additionally, we assessed for the emergence of the three untrained operants across all four training conditions.

METHOD

Participants and Setting Four native English speaking adults partici-

pated in this study. Jesus was a 26-year-old man with a high school diploma, and Eobard and Kip were both 26-year-old men with bach- elor’s degrees. Lily was a 23-year-old woman who held a bachelor’s degree. Participants were recruited from college campuses and the local community. None of the participants had a his- tory of receiving formal instruction in Manda- rin Chinese, nor did they hear Mandarin Chinese on a regular basis. Participants were monolingual, meaning they could communi- cate fluently in only one language. Participants reported that they did not have any language or developmental delays. All experimental sessions were conducted at

a table with two chairs. Jesus’ and Lily’s

WAI-LING WU et al.654

sessions were conducted at the university-based clinic, and Eobard’s and Kip’s sessions were conducted in their homes. Clinic rooms were approximately 5 m by 3 m, and only materials required to complete the session were available. In the participants’ homes, all objects within 1.5 m of the table and not used for sessions were cleared away, and only required materials remained within 1.5 m of the table. Only the participant and the first author were present during sessions. Sessions lasted approximately 5 min each, with 10-12 sessions conducted per day, at least one time per week for 3-5 weeks per participant. All sessions were videotaped for data collection purposes.

Materials The experimenter used five sets of stimuli

for each participant. Each set included three target words (e.g., Xing Pian [chips], La Jiang [salsa], and Cha Zi [fork]). Each set was assigned to an experimental condition (mand, tact, NFI, FNI, and control), for a total of 15 targets per participant. Please refer to Table 1 for the assignments of words to each experimental condition for all four participants. Stimuli were selected based on a questionnaire and an informal interview that were adminis- tered to the participant prior to the experimen- tal sessions. The questionnaire instructed the

participant to list preferred foods. In an effort to control for response effort across the training conditions, each Chinese word contained the same number of syllables within and across stimuli sets for each participant, and the sylla- bles were audibly distinct from each other. Other materials included corresponding items, which are listed in Table 1.

Dependent Variable and Response Measurement A correct response was recorded if the partic-

ipant responded within 5 s of the presentation of a stimulus. An incorrect response was recorded if the participant responded with the incorrect name of an object, or if they did not respond within 5 s of the presentation of a stimulus. A correct mand consisted of the appropriate topography emitted within 5 s of contriving the MO (e.g., participant saying “pu tao” for grape, when grape was withheld during meal times). A correct tact consisted of the appropriate topography emitted within 5 s of the presentation of the object (e.g., participant says “pu tao” within 5 s of a grape presented on the table in front of him and the question, “What is it?”). A correct intraverbal consisted of emission of the corresponding word in the native or foreign language, depending on the condition, within 5 s of the presentation of the

Table 1 Target Words per Condition

Name Mand Tact NFI FNI Control

Jesus Bread (Mian Bao) Scissors (Jian Dao) Stirrer (Jiao Ban) Fork (Cha Zi) Juice (Guo Zhi) PB (Hua Sheng) Spoon (Tiao Gen) Creamer (Nai Jing) Knife (Dao Zi) Straw (Xi Guan) Plate (Pan Zi) Bowl (Xiao Wan) Sugar (Bai Tang) Syrup (Tang Jiang) Wipes (Zhi Jing)

Eobard Cracker (Bing Gan) Oreo (Jia Xing) Chips (Shu Pian) Jelly (Guo Jiang) Tortilla (Da Bing) PB (Hua Sheng) Milk (Niu Nai) Salsa (La Jiang) Bread (Mian Bao) Fork (Cha Zi) Honey (Feng Mi) Cup (Zhi Bei) Spoon (Tang Chi) Knife (Dao Zi) Plate (Xiao Pan)

Kip Chips (Xing Pian) Bagel (Ji Bing) Pretzels (Cui Bing) Crackers (Bing Gan) Toothpick (Ya Qian) Salsa (La Jiang) Plate (Can Pan) Chs Dip (Lu Lao) PB (Hua Sheng) Chs C (Gan Lao) Fork (Cha Zi) Crm Chs (Nai You) Spoon (Tang Chi) Knife (Xiao Dao) Grape (Pu Tao)

Lily Bagel (Ji Bing) Chip (Xing Pian) Apples (Ping Guo) Banana (Xiang Jiao) Cereal (Gu Pian) Crm Chs (Nai You) Salsa (La Jiang) PB (Hua Sheng) Plate (Can Pan) Milk (Niu Nai) Knife (Xiao Dao) Fork (Cha Zi) Spoon (Tang Chi) Bread (Mian Bao) Bowl (Da Wan)

Note. PB = Peanut Butter, Chs = Cheese, C = Cube, Crm = Cream

655MAND TRAINING TO TEACH A FOREIGN LANGUAGE

question (e.g., participant saying, “pu tao” when asked “What is the Chinese name for grape?”). Self-corrections, such as initially responding with the incorrect Chinese word and then immediately following it with the cor- rect Chinese word, were not accepted as correct responses. The experimenter recorded correct and incorrect responses for each trial using pen- cil and paper. For the purposes of this study, tonal differ-

ences in Chinese responses were not included as a criterion in the operational definitions. Different tones were accepted as correct as long as the pronunciation matched the spelling of the word and the number of syllables. Data during all of the experimental conditions were expressed as a percentage of correct responses per session. The number of training trials required to reach the mastery criterion were also examined across training conditions.

Interobserver Agreement and Treatment Integrity A second data collector watched videos of

the sessions and independently collected inter- observer and treatment integrity data on at least 40% of all pretraining probes, posttraining pro- bes, baseline, and training sessions for each par- ticipant. Interobserver agreement and treatment integrity were calculated on a trial-by-trial basis, in which the number of agreements was divided by the number of agreements plus dis- agreements and multiplied by 100 to obtain a percentage. If both observers scored a trial in the same way, it was considered an agreement. If the observers scored a trial differently, it was considered a disagreement. For pretests and baseline, the interobserver agreement was 100% for all four participants. During training, interobserver agreement was 99% (range, 89% to 100%) for Jesus, 100% for Eobard, 100% for Kip, and 100% for Lily. For posttraining probes, the interobserver agreement was 100% for all four participants. Treatment integrity

data were collected on the correct presentation of discriminative stimuli (e.g., “Name the item in Chinese”), the correct delivery of prompts (e.g., echoic prompt “la jiang”), and the correct delivery of consequences per trial per experi- mental condition. Mean treatment integrity across all four participants was 93% (range, 60% to 100%) during pretests, 96% (range, 60% to 100%) during baseline, 98% (range, 80% to 100%) during training, and 94% (range, 55% to 100%) during posttraining pro- bes. Treatment integrity scores were low during three sessions. The experimenter asked ques- tions in the incorrect order in one intraverbal pretest session, removed items in the incorrect order in one mand training session, and lost her place and had to re-present the question in one intraverbal posttest session.

Pre-Experimental Conditions Tact assessment. During the tact assessment,

the experimenter held up an object in front of the participant and asked, “What is it?” The participant was expected to respond with the correct name of the object in English within 5 s of the question. This assessment was con- ducted to ensure that the participant was not learning to tact an object in English for the first time during the tact sessions. The experimenter also wanted to ensure that the participant and the experimenter tacted objects using the same word in English. Each item was presented one at a time and one time, only. The mastery cri- terion was one session with independent responding at 100% accuracy. Echoic assessment. During the echoic assess-

ment, the experimenter said, “Say [Chinese word],” and the participant was expected to repeat the Chinese word. This assessment was conducted to ensure that the participant would be able to imitate the word and that the echoic prompts in later sessions would produce accu- rate responding. The echoic assessment included all the words that were targeted in the

WAI-LING WU et al.656

study. All participants were trained to respond to the mastery criterion, as necessary. The mas- tery criterion was one session with independent responding at 100% accuracy.

Experimental Conditions Pretraining probes. Pretraining probe sessions

were conducted prior to baseline. Each session consisted of nine trials per relation, whereby each of the three targets in a set was presented three times. A total of 16 pretraining probe ses- sions were conducted per participant (three ses- sions for mand training, three sessions for tact training, three sessions for NFI training, three sessions for FNI training condition, and four sessions for the control condition). Each of the three pretraining probe sessions was only con- ducted with the three relations that were not targeted for training for a given set. For exam- ple, if tact training was conducted using Set C words, then pretraining probe sessions were conducted for mand, NFI, and FNI responding for Set C words. Note that pretraining probe sessions were conducted with all four relations (mand, tact, NFI, and FNI) for the control set. At the end of each session, the experimenter thanked the participant for completing the ses- sion and provided a small token of food for participating. During the mand pretraining probe sessions,

objects were arranged so that the participants had to request them in Chinese to complete a particular action (Lechago, Carr, Grow, Love, & Almason, 2010). At the start of the session, the participants were provided with the following instruction, “You must ask for the objects you want in Chinese. If the object is back on my side of the table, then you must ask for the object again in order to access it again.” Partici- pants were instructed not to eat for at least three hours prior to all study sessions, in order to enhance the value of food as a reinforcer and to establish the motivation for the mand train- ing condition. Items in the session were

presented on the table in front of the partici- pant so that the participant could snack on items. The experimenter manipulated the items so that certain materials were needed to gain access to the food. For example, chips and salsa were placed in front of a participant, but the salsa was in a half-filled jar with a narrow open- ing such that a spoon was required to scoop salsa out of the jar. The requested object was delivered to the participant contingent upon a correct response. Prompts were not provided for incorrect responses. A statement of acknowledgement was provided for incorrect or no responses (e.g., “OK”). At the start of the tact pretraining probe

sessions participants were given the following instruction, “You will label in Chinese the objects I present to you.” An object was held at eye-level in front of the participant, and the experimenter asked, “What is it?” There were no programmed consequences for cor- rect or incorrect responses, and prompts were not provided during this condition. A state- ment of acknowledgement was provided for responses (e.g., “OK”). In the NFI pretraining probe session, the

experimenter asked, “What is [English name for object] in Chinese?” In the FNI pretraining probe session, the experimenter asked, “What is [Chinese name for object] in English?” There were no programmed consequences for correct or incorrect responses. A statement of acknowl- edgement was provided for responses (e.g., “OK”). Prompts were not provided dur- ing this session. There were no objects within the participant’s view during the intraverbal sessions. Baseline. Baseline sessions were procedurally

identical to the pretraining probe sessions, except during baseline sessions, only those words in the targeted training condition were considered baseline probes. So, for example, if tact training was conducted using Set C words, then baseline testing for tacting Set C words was conducted. Baseline sessions for each of the

657MAND TRAINING TO TEACH A FOREIGN LANGUAGE

four training conditions (tact, mand, NFI, FNI) were conducted prior to training with each set. Each session consisted of nine trials, whereby the three targets assigned to each training condition were probed three times per session. Baseline probes were randomized across sessions such that targets were not presented in a fixed sequence across sessions. At the end of each condition, the experimenter thanked the participant for completing the session and pro- vided a small token of food, such as a chip with salsa or a bite sized peanut butter and jelly sandwich. Training. All training trials were initiated in

a procedurally identical manner to baseline tri- als, and then correct and incorrect responding produced programmed consequences. Training for a condition was considered complete when the participant emitted independent correct responses for 100% of the trials for two consec- utive sessions. Training for a condition was ter- minated if a participant did not show an increase in independent correct responses for five consecutive sessions. If a condition was mastered before the others were mastered, the experimenter stopped conducting training ses- sions for that condition and continued training in the other conditions until responding met the mastery criterion for all of the conditions. In the mand condition, praise and the

requested object were delivered contingent upon a correct independent response. If the participant emitted an incorrect response or did not respond within 5 s of contriving the MO, the experimenter provided a full echoic prompt every 5 s until the participant provided a cor- rect response. The experimenter would have provided the echoic prompt up to three times before terminating the trial and initiating a new trial. However, all the participants responded correctly after the first prompt. Prompted responses resulted in the delivery of the object without praise. For the tact, NFI, and FNI training condi-

tions, the experimenter delivered praise

contingent upon a correct response (e.g., “You got it!”). If the participant emitted an incorrect response, or took longer than 5 s to emit the response, the experimenter repeated the ques- tion and provided an echoic prompt every 5 s until the participant responded correctly. The experimenter would have provided the echoic prompt up to three times before terminating the trial and initiating a new trial. However, all the participants responded correctly after the first prompt. Correct prompted responses resulted in the experimenter providing a state- ment of acknowledgement in a neutral tone (e.g., “That’s the Chinese name for grape”). Participants were provided with small amounts of preferred food items at the end of the tact, NFI, and FNI training sessions in order to con- trol for potential reinforcer effects in the mand training condition. No items were present dur- ing NFI and FNI training conditions. Posttraining probes. These sessions were con-

ducted after responding had met the mastery criterion for all training conditions. These ses- sions were procedurally identical to the pre- training probe sessions.

Experimental Design A multiple baseline across participants design

was employed to evaluate the rate of acquisition of responding across the four training condi- tions. An adapted alternating treatments design was embedded into the training phase. The order of the conditions was assigned randomly. Pre- and posttraining probe results were com- pared to examine the effects of each training type on emergent responding.

RESULTS

Figure 1 displays the percentage of correct responses for the targeted words across all four training conditions for all participants. None of the participants responded correctly during baseline. The mand training condition required the fewest number of trials to mastery for Jesus,

WAI-LING WU et al.658

Figure 1. The percentage of correct responses for the trained operants in baseline and training for each participant.

659MAND TRAINING TO TEACH A FOREIGN LANGUAGE

Eobard, and Lily. For Kip, the NFI training condition required the fewest number trials to mastery. Figures 2 and 3 display emergent responses (i.e., untrained relations) across all four training conditions and the control condi- tion during pre- and postsession probes. None of the participants responded correctly during pretraining probes. The mand training condi- tion produced the greatest amount of emergent

responding for Jesus. Mand and tact training produced similar results on emergent relations for Eobard and Lily (with tact condition pro- ducing a slightly greater amount of emergent NFI for Lily). For Kip, the tact training condi- tion produced the greatest amount of emergent responding. During mand training, the number of ses-

sions to reach mastery for Jesus, Eobard, Kip,

Figure 2. Jesus’s and Eobard’s emergent responses in pretraining and posttraining probes.

WAI-LING WU et al.660

and Lily were eight, five, eight, and four, respectively (Figure 1). During the posttraining probes (Figure 2), Jesus responded with 100% accuracy for tact, 89% accuracy for NFI, and 89% accuracy for FNI. Eobard responded with 100% accuracy for tact, NFI, and FNI. Kip (Figure 3) responded with 33% accuracy for tact and NFI, and 56% for FNI. Finally, Lily responded with 100% accuracy for tact and FNI, and 89% for NFI.

During tact training, the number of sessions to reach mastery for Jesus, Eobard, Kip, and Lily was eleven, six, eight, and seven, respec- tively (Figure 1). During the posttraining pro- bes (Figure 2), Jesus responded with 67% accuracy for mand, 22% accuracy for NFI, and 100% accuracy for FNI. Eobard responded with 100% for mand, NFI, and FNI. Kip responded with 67% for mand and NFI, and at 100% for FNI (Figure 3). Lily responded

Figure 3. Kip’s and Lily’s emergent responses in pretraining and posttraining probes.

661MAND TRAINING TO TEACH A FOREIGN LANGUAGE

with 100% for mand, NFI, and FNI (Figure 3). During NFI training, …

Our website has a team of professional writers who can help you write any of your homework. They will write your papers from scratch. We also have a team of editors just to make sure all papers are of HIGH QUALITY & PLAGIARISM FREE. To make an Order you only need to click Ask A Question and we will direct you to our Order Page at WriteDemy. Then fill Our Order Form with all your assignment instructions. Select your deadline and pay for your paper. You will get it few hours before your set deadline.