Replicating and Extending "Because Their Treebanks Leak": Graph Isomorphism, Covariants, and Parser Performance
This work highlights methodological issues in statistical NLP analyses, suggesting that controlled experiments are needed to draw reliable conclusions, which is incremental for parser evaluation research.
The study replicated and extended prior research on graph isomorphism's effect on parser performance, finding that the correlation disappears when controlling for covariants in real data, but a strong correlation exists in controlled experiments.
Søgaard (2020) obtained results suggesting the fraction of trees occurring in the test data isomorphic to trees in the training set accounts for a non-trivial variation in parser performance. Similar to other statistical analyses in NLP, the results were based on evaluating linear regressions. However, the study had methodological issues and was undertaken using a small sample size leading to unreliable results. We present a replication study in which we also bin sentences by length and find that only a small subset of sentences vary in performance with respect to graph isomorphism. Further, the correlation observed between parser performance and graph isomorphism in the wild disappears when controlling for covariants. However, in a controlled experiment, where covariants are kept fixed, we do observe a strong correlation. We suggest that conclusions drawn from statistical analyses like this need to be tempered and that controlled experiments can complement them by more readily teasing factors apart.