The success of the large neural language models on many NLP tasks is exciting. However, we find that these successes sometimes lead to hype in which these models are being described as "understanding" language or capturing "meaning". In this position paper, we argue that a system trained only on form has a priori no way to learn meaning. In keeping with the ACL 2020 theme of "Taking Stock of Where We've Been and Where We're Going", we argue that a clear understanding of the distinction between form and meaning will help guide the field towards better science around natural language understanding.

We present a semantic parser for Abstract Meaning Representations which learns to parse strings into tree representations of the compositional structure of an AMR graph. This allows us to use standard neural techniques for supertagging and dependency tree parsing, constrained by a linguistically principled type system. We present two approximative decoding algorithms, which achieve state-of-the-art accuracy and outperform strong baselines. Related WorkRecently, AMR parsing has generated considerable research activity, due to the availability of large-

Psycholinguistic research shows that key properties of the human sentence processor are incrementality, connectedness (partial structures contain no unattached nodes), and prediction (upcoming syntactic structure is anticipated). There is currently no broad-coverage parsing model with these properties, however. In this article, we present the first broad-coverage probabilistic parser for PLTAG, a variant of TAG that supports all three requirements. We train our parser on a TAG-transformed version of the Penn Treebank and show that it achieves performance comparable to existing TAG parsers that are incremental but not predictive. We also use our PLTAG model to predict human reading times, demonstrating a better fit on the Dundee eyetracking corpus than a standard surprisal model.

Most semantic parsers that map sentences to graph-based meaning representations are handdesigned for specific graphbanks. We present a compositional neural semantic parser which achieves, for the first time, competitive accuracies across a diverse range of graphbanks. Incorporating BERT embeddings and multi-task learning improves the accuracy further, setting new states of the art on DM, PAS, PSD, AMR 2015 and EDS.

We investigate the application of modern planning techniques to domains arising from problems in natural language generation (NLG). In particular, we consider two novel NLGinspired planning problems, the sentence generation domain and the GIVE ("Generating Instructions in Virtual Environment") domain, and investigate the efficiency of FF and SG-PLAN in these domains. We also compare our results against an ad-hoc implementation of GraphPlan in Java. Our results are mixed. While modern planners are able to quickly solve many moderately-sized instances of our problems, the overall planning time is dominated by the grounding step that these planners perform (rather than search). This has a pronounced effect on our domains which require relatively short plans but have large universes. We share our experiences and offer these domains as challenges for the planning community.