``I want to think like an SLP'': A Design Exploration of AI-Supported Home Practice in Speech Therapy

Research Background and Issues

• Problems Identified by the Authors: The paper highlights the importance of family practices in children's language therapy but notes that parents face numerous challenges in implementing these practices, including environmental differences, lack of guidance, difficulty maintaining children's engagement, emotional stress, and coordinating multiple therapy tasks.
• Why This Issue Is Important: Family practices in language therapy are critical for helping children generalize new skills into daily life. However, parents' lack of capability may limit the effectiveness of therapy, potentially leading to long-term adverse effects on children's communication and social development.
• Research Motivation and Related Work:
  • Family practices have been proven to be a part of language therapy, but existing solutions do not adequately address the specific difficulties parents face in implementation.
  • Artificial intelligence (AI) has shown potential in education and healthcare, but how AI can support parents in implementing language therapy practices remains an unresolved issue.

Solution

• Proposed Methods or Solutions: The authors employed a three-step research design to validate the feasibility and challenges of using AI in family language practices:

  1. Parent Interviews: To understand parents' experiences and needs in language therapy family practices.
  2. Design Concept Development: Based on parents' input, six specific AI-supported solutions were designed and presented.
  3. Professional Evaluation: Feedback from speech-language pathologists (SLPs) was used to assess the potential and possible issues of these AI designs.

• Innovative Aspects of the Solution:

  • Proposed and concretized six AI concepts to assist family language therapy, such as a real-time pronunciation feedback assistant, an organizational tool for integrating therapy tasks, and a personalized practice planning tool.
  • Adopted an interdisciplinary approach, integrating expertise from human-computer interaction (HCI), artificial intelligence, and language therapy.

• Implementation Steps and Key Technologies:

  • Designed AI concepts using technologies such as speech recognition, recommendation algorithms, schedule optimization, and task integration.
  • Presented the design concepts in storyboard format, emphasizing collaborative scenarios between users and AI.
  • Investigated the needs and feedback of target users (parents) and key stakeholders (SLPs).

Research Findings

• Specific Findings:

  1. Parent Needs Analysis: Identified five major challenges parents face in family practices, including environmental differences, children's fatigue, and low engagement.
  2. AI Tool Design and Evaluation: Developed six conceptual AI design solutions covering various support modes (e.g., informational, emotional, and practical support).
  3. Professional Evaluation: SLP feedback highlighted the potential of these designs while pointing out possible concerns (e.g., privacy and data management).

• Advantages Over Existing Solutions:

  • Existing tools focus more on direct interaction with children, whereas this study emphasizes supporting parents' capabilities and reducing their burden.
  • Provides cross-therapy task integration tools and personalized family practice recommendation features, helping parents balance busy daily lives.

• Experimental and Evaluation Results:

  • AI designs were found to enhance parents' confidence, reduce common errors, save time, and promote coordination across different environments.
  • Acceptance of different designs varied depending on factors such as the level of supervision involved and risks of excessive screen time.

• Limitations and Future Directions:

  • Limitations: Did not directly address children's perspectives as users; did not consider parents' familiarity with AI as a variable; the sample was predominantly mothers, with limited feedback from fathers.
  • Future Directions:
    1. Explore AI designs based on joint interactions between children and parents.
    2. Expand the study sample to include parents with varying levels of AI familiarity.
    3. Improve privacy and algorithm fairness to ensure the technology serves diverse family backgrounds.

Conclusion

The paper demonstrates the potential of integrating AI into children's language therapy, offering practical support tool concepts for parents and speech-language pathologists. By combining insights from technology, psychology, and language therapy, the study provides new perspectives on family practices. However, the research also cautions designers to balance supervision with parental autonomy while avoiding additional burdens. Furthermore, when applying AI in real-world settings, the dynamic relationships and ethical considerations among children, parents, and AI need further exploration.