健康与遗传心理学研究室知识库

目的：语言功能异常是孤独症谱系障碍(ASD)的典型症状之一。1957年, 斯金纳从行为学角度提出言语行为，认为人们获得表达和理解语言的能力与学会走路、跳舞等行为的原理相同。2001年，海斯等人在斯金纳的基础上提出关于人类语言和认知的新视角——关系框架理论，澄清了语言的推衍性实质，认为言语行为是将一个刺激与其他刺激建立关系框架的过程。大量的研究结果表明ASD儿童在间接语言尤其是隐喻语言方面存在不足，隐喻作为一种常见的语言现象和语言能力制约着ASD儿童的语言发展和人际沟通，但少有研究致力于教授ASD儿童隐喻语言。关系框架理论提供了一种从行为学角度去探讨隐喻的方式，本研究尝试基于多范例训练范式建立隐喻推理关系反应训练模式并检验其可行性和有效性，同时探讨分析他们在训练过程中的行为表现和变化，从而为今后的实践应用提供理论和实证基础。
方法：本研究分为三个部分。
研究一：初步探索和检验隐喻推理训练的有效性与可行性。采用单一被试设计，将训练分为基线阶段、多范例训练阶段、训练后检验阶段等三个阶段。参与者为北京某培智学校的4名12～16岁ASD儿童，评估指标为每次试验中儿童对隐喻问题的反应正确率。
研究二：孤独症谱系障碍儿童关系反应水平评估。通过微信发布信息招募参与评估的ASD儿童，报名67名，最终参与评估40名，年龄范围2~15岁。采用关系反应测试题、关系前提与能力的训练评估程序等在安静的环境中对儿童进行一对一的施测，同时全程摄像。评估指标为儿童关系反应的正确率。
研究三：进一步完善隐喻推理关系反应训练模式并验证有效性。采用单一被试设计，来自研究二的15名儿童完成了本研究中的训练任务。在对父母进行集中培训后，研究者向父母提供训练材料，由每个儿童的父母在家完成每日训练任务。
结果：研究一：4名儿童顺利完成了所有阶段的评估和训练，且在训练过程中其对隐喻问题的反应正确率有了很大的提升而且还能泛化到未训练过的隐喻上；有2名儿童在训练阶段后期表达出自行创造的隐喻。
研究二：（1）所有儿童在空间关系、比较关系上表现较好，平均反应正确率高于80%，时间关系、对立关系、一致关系次之，而在隐喻推理和层级关系上的反应能力最弱，平均反应正确率低于50%。七种关系类型中，层级关系、隐喻推理、一致关系和对立关系这四个关系类型的反应正确率要弱于4~7岁的正常发育的儿童，且七个关系类型的反应正确率均弱于7岁的正常发育儿童；（2）ASD儿童的关系反应水平在不同严重程度水平之间差异显著，学龄前儿童与学龄儿童的关系反应水平无显著差异；（3）TARPA测试中，3岁以上的ASD儿童几乎都能完成简单辨别任务，大部分儿童都能继续完成基于物理相似性的条件性辨别任务，但在基于非物理相似性的条件性辨别任务上，只有6名儿童全部通过，7名儿童仅未通过刺激功能转化的测试，4名儿童仅未通过联合推衍和刺激功能转化的测试。
研究三：（1）15名完成训练的儿童中有2名儿童在前测的隐喻推理反应正确率就已经达到80%以上的正确率，但在第1次训练时都没有达到80%的正确率，其中一个儿童在训练3次后能达到并维持100%的正确率，另一个儿童经过13次训练后能较稳定地维持80%以上乃至100%的正确率；10名儿童在训练20次左右后就有明显的提升，到训练30~40次时能基本维持80%以上乃至100%的正确率；3名儿童在常规训练中能突破正确率为0%的反应，但进一步的提升需要增加训练次数；（2）采用不同难度水平隐喻故事和不同推衍类型起点的儿童均在隐喻推理和推衍关系反应上有不同程度的提升；（3）15名儿童中，有5名儿童的家长主动报告了儿童在日常生活中自发创造的隐喻；（4）15名儿童的父母平均在第4.67次（标准误为1.98次）时感觉到对训练的操作比较熟悉了，同时，平均在第10.33次（标准误为6.10次）感觉到孩子开始有了进步。
结论：（1）孤独症谱系障碍儿童的推衍关系反应水平较正常发育儿童存在较大的差距，功能程度越低的儿童其关系反应水平越差，且学龄前与学龄期的关系反应水平没有差异，因此干预训练的介入是必要且重要的。（2）孤独症谱系障碍儿童的推衍关系反应水平，尤其是隐喻推理能力能够通过基于关系框架理论的多范例训练范式习得，并且可以泛化到未训练过的推衍关系上。（3）本研究建立的隐喻推理关系反应训练模式具备可行性和有效性，孤独症谱系障碍儿童的父母能够在指导下掌握训练的操作方法并进行个性化的训练。

Objective: Language dysfunction is one of the core symptoms of children with autism spectrum disorders. In 1957, Skinner put forward the concept of verbal behavior. He believed that the underlying process of learning to comprehend and speak language is fundamentally similar to the process of other observable behaviors, such as walking and dancing. In 2001, based on the theoretical foundation of verbal behavior, Hayes et.al proposed a new perspective of human language and cognition, Relational Frame Theory, which explains linguistic generativity in terms of learned contextually controlled relational responding referred to as relational framing. A considerable amount of research has shown that children with autism spectrum disorders (ASD) have difficulty with figurative language, specifically with metaphorical language. As a common language phenomenon and competence, metaphor is a restriction to the language development and interpersonal communication of children with ASD. However, few studies have attempted to teach children with ASD to understand metaphor. Relational framing theory (RFT) came up with a behavioral approach to the topic of metaphorical language and it treats relating, per se as learned behavior. The current study attempt to construct and evaluate the learning model of metaphorical reasoning for children with ASD in China by means of multiple exemplar training for teaching them to establish the relational frames among subjects in a metaphor. Research also explored children’s behavior changes during the training, which is both the theoretical and empirical base of further practice and application.
Methods:There are three experiments.
Study1: Establishing metaphorical reasoning in children with autism spectrum disorders: A pilot study. The design of the experiment was a single subject research design, and there were three main phases including baseline, multiple exemplar training and post-training. Participants included fourchildren, ages 12~16. The percentages of correct responses (i.e., accuracy) of each session was collected as evaluation indicators.
Study2: The evaluation of relational responding in children with autism spectrum disorders. 40 children with ASD from two to fifteen years of age participated with 67 children signed up at first. Each child individually completed an assessment of relational respondingand training and assessment of relational precursors and abilities (TARPA)in a quiet room. Data were collected on the percentage of correct response to each item.
Study3: Training derived metaphorical reasoning in children with autism spectrum disorders: A further test. A single subject research design was conducted in this experiment. 27 children from experiment one participated in the training program with fifteen children completed, so far. Each child was trained individually by their own parents at home after a two-hour centralized training for all parents. Daily training materials of each child were provided online when the previous one was accomplished.
Result:
Study1:Results revealed thatthefour children with ASDachievedin post-training phase and probe session, for accuracy data in all sessions reached above 80%. Furthermore, generalization to untrained metaphors was found and twoof themeven demonstrate the generalization to create their own metaphors.
Study2: (1)As a whole,children with ASD performed better in spatial relation and comparison, with an averagepercentage of correct responsesof more than 80%, and temporalrelation, hierarchical relationand coordinationcame the next, while the averagepercentage of correct responseswas weakest in metaphoricalreasoning and hierarchical relationship, which was less than 50%.Among the seven types of relations, hierarchical, metaphoricalreasoning, coordination, and hierarchical relationwereweakerthan those of normally developing childrenaged 4 to 7 years, and the averagepercentage of correct responsesof alltypes were weakerthan those of normally developingchildren aged 7 years. (2)Results showed thatthere werestatistical significant differences amongthe severity ofASD in children’s relational responding. No statisticalsignificancewas found between preschoolers and school-age children in all types of relations. (3) The results of TARPA showed that children above the age of three could almost accomplished the simple discrimination task, and most of them could continue to finish the non-arbitrary conditional discrimination task. However, for the arbitrary conditional discrimination task, only six participantscould pass the whole test. And there are seven participants only failed the last session (i.e., transfer of function), and fourparticipants only failed the last two sessions (i.e., combinatorial entailment and transfer of function).
Study3: (1) Among the fifteen participants, although two of them got high accuracies beforethe training program, theydid not that wellat the first session. With above three training sessions, one of themgot stable accuracy of 100%. Anotherachieved thatwith about thirteen sessions. In the same time, tenparticipants made significant progresses after approximately twenty training sessions and could reached above 80%even 100% after approximately thirty to forty training sessions. And especially, the other three participants broke through the accuracy of 0% after the first round training, though more training sessions were needed for further advance.(2) Participantswho used different difficulty levels of metaphorical stories or different types of relational responding all achieved success in the promotionof metaphorical reasoning and relational responding.(3) Additionally, five participants’ parents reported created metaphorof their children during the daily life. (4) Besides, according to participants’ parents, they could be familiar with the training operation after approximately five sessions and could see progressof their children after approximately ten session.
Conclusion:The result reveals that (1) compared to normally developing children, the competence of derived relational responding of children with autism spectrum disorderswas severely defective. The more severe the symptoms, the weaker the ability of derived relational responding. Furthermore, there is no measurable development with the growth of children, so applicable intervention is indispensable and crucial; (2) by means of multiple exemplar training for establishing derived metaphorical reasoning inchildren with autism spectrum disorders, they could learn to understand derived relational responding especially the metaphorical reasoning. And generalization to untrained metaphors wasremarkable; (3)the training model of derived metaphorical reasoning in the research is viable and effective. Parents of children with autism spectrum disorderscould master the training methods and conduct personalized training under guidance.