Attribution and Alignment: Effects of Local Context Repetition on Utterance Production and Comprehension in Dialogue
This addresses the problem of improving dialogue systems for human users by aligning model behavior with cognitive patterns, though it is incremental in scope.
The study investigated whether language models produce human-like repetition in dialogue and what processing mechanisms they use for lexical re-use during comprehension, finding that humans prefer and benefit from repetition, which contrasts with typical model penalties.
Language models are often used as the backbone of modern dialogue systems. These models are pre-trained on large amounts of written fluent language. Repetition is typically penalised when evaluating language model generations. However, it is a key component of dialogue. Humans use local and partner specific repetitions; these are preferred by human users and lead to more successful communication in dialogue. In this study, we evaluate (a) whether language models produce human-like levels of repetition in dialogue, and (b) what are the processing mechanisms related to lexical re-use they use during comprehension. We believe that such joint analysis of model production and comprehension behaviour can inform the development of cognitively inspired dialogue generation systems.