Neural language modeling of free word order argument structure

TL;DR

This paper investigates how neural language models understand complex German verb argument structures with flexible word order, revealing both capabilities and limitations in capturing syntactic dependencies and case assignments.

Contribution

It introduces a novel probing methodology for free word order argument structures and compares the performance of Transformers and LSTMs on this task.

Findings

Transformers and LSTMs outperform chance in capturing argument structures

Humans prefer canonical word orders and plausible case assignments

Models show discrepancies, with LSTMs struggling with ungrammatical sentences and Transformers overgeneralizing

Abstract

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.