Finding syntax in human encephalography with beam search has been selected as one of three “Best Papers (long format)” for the 2018 meeting of the Association for Computational Linguistics!
You can hear John H present the paper in Melbourne July 15-20th. Jon B will be giving a sneak peak tomorrow (Weds June 13th!) at the Lawrence Hall of Science in Berkeley, CA. Or, you can just go and read the paper!
From the abstract:
Recurrent neural network grammars (RNNGs) are generative models of (tree, string) pairs that rely on neural networks to evaluate derivational choices. Parsing with them using beam search yields a variety of incremental complexity metrics such as word surprisal and parser action count. When used as regressors against human electrophysiological responses to naturalistic text, they derive two amplitude effects: an early peak and a P600-like later peak. By contrast, a non-syntactic neural language model yields no reliable effects. Model comparisons attribute the early peak to syntactic composition within the RNNG. This pattern of results recommends the RNNG+beam search combination as a mechanistic model of the syntactic processing that occurs during normal human language comprehension.