Neural language modeling of free word order argument structure
This work addresses the challenge of syntactic modeling in free word order languages for computational linguistics, though it is incremental as it applies existing methods to a specific linguistic phenomenon.
The paper tackled the problem of modeling verb argument structure in German's free word order subordinate clauses, finding that both Transformers and LSTMs perform substantially better than chance but exhibit unexpected discrepancies, such as LSTMs struggling with ungrammatical sentences and Transformers overgeneralizing to infrequent or implausible cases.
Neural language models trained with a predictive or masked objective have proven successful at capturing short and long distance syntactic dependencies. Here, we focus on verb argument structure in German, which has the interesting property that verb arguments may appear in a relatively free order in subordinate clauses. Therefore, checking that the verb argument structure is correct cannot be done in a strictly sequential fashion, but rather requires to keep track of the arguments' cases irrespective of their orders. We introduce a new probing methodology based on minimal variation sets and show that both Transformers and LSTM achieve a score substantially better than chance on this test. As humans, they also show graded judgments preferring canonical word orders and plausible case assignments. However, we also found unexpected discrepancies in the strength of these effects, the LSTMs having difficulties rejecting ungrammatical sentences containing frequent argument structure types (double nominatives), and the Transformers tending to overgeneralize, accepting some infrequent word orders or implausible sentences that humans barely accept.