Incremental Sentence Processing Mechanisms in Autoregressive Transformer Language Models

AbstractAutoregressive transformer language models (LMs) possess strong syntactic abilities, often successfully handling phenomena from agreement to NPI licensing. However, the features they use to incrementally process their linguistic input are not well understood. In this paper, we fill this gap by studying the mechanisms underlying garden path sentence processing in LMs. Specifically, we ask: (1) Do LMs use syntactic features or shallow heuristics to perform incremental sentence processing? (2) Do LMs represent only one potential interpretation, or multiple? and (3) Do LMs reanalyze or repair their initial incorrect representations? To address these questions, we use sparse autoencoders to identify interpretable features that determine which continuation—and thus which reading—of a garden path sentence the LM prefers. We find that while many important features relate to syntactic structure, some reflect syntactically irrelevant heuristics. Moreover, though most active features correspond to one reading of the sentence, some features correspond to the other, suggesting that LMs assign weight to both possibilities. Finally, LMs fail to re-use features to answer follow-up questions.