Illocutionary Explanation Planning for Source-Faithful Explanations in Retrieval-Augmented Language Models

TL;DR

This paper investigates source-faithful explanations in retrieval-augmented language models for programming education, introducing illocutionary macro-planning to improve adherence to authoritative sources, with mixed success.

Contribution

It proposes illocutionary macro-planning and chain-of-illocution prompting to enhance source adherence in RAG models for programming explanations.

Findings

Non-RAG models have median source adherence of 0%.

Baseline RAG systems show 22-40% adherence.

Chain-of-illocution prompting improves adherence up to 63%.

Abstract

Natural language explanations produced by large language models (LLMs) are often persuasive, but not necessarily scrutable: users cannot easily verify whether the claims in an explanation are supported by evidence. In XAI, this motivates a focus on faithfulness and traceability, i.e., the extent to which an explanation's claims can be grounded in, and traced back to, an explicit source. We study these desiderata in retrieval-augmented generation (RAG) for programming education, where textbooks provide authoritative evidence. We benchmark six LLMs on 90 Stack Overflow questions grounded in three programming textbooks and quantify source faithfulness via source adherence metrics. We find that non Retrieval-Augmented Generation (RAG) models have median source adherence of 0%, while baseline RAG systems still exhibit low median adherence (22-40%, depending on the model). Motivated by Achinstein's illocutionary theory of explanation, we introduce illocutionary macro-planning as a descriptive design principle for source-faithful explanations and instantiate it with chain-of-illocution prompting (CoI), which expands a query into implicit explanatory questions that drive retrieval. Across models, CoI yields statistically significant gains (up to 63%) in source adherence, although absolute adherence remains moderate and the gains are weak or non-significant for some models. A user study with 165 retained participants (220 recruited) indicates that these gains do not harm satisfaction, relevance, or perceived correctness.