Stephanie Richter - Academia.edu (original) (raw)
Related Authors
Institut National de Recherche en Informatique et Automatique (INRIA)
Uploads
Papers by Stephanie Richter
Syntactic knowledge involves not only the ability to combine words and phrases, but also the capa... more Syntactic knowledge involves not only the ability to combine words and phrases, but also the capacity to relate different and yet truth-preserving structural variations (e.g. passivization, inversion, topicalization, extraposition, clefting, etc.), as well as the ability to infer that these syntactic variations all adhere to common morphosyntactic rules, like subjectverb agreement. Although there is some evidence that BERT has rich syntactic knowledge, our adversarial approach suggests that it is not deployed in a robust and linguistically appropriate way. English BERT can be tricked to miss even quite simple syntactic generalizations, when compared with GPT-2, underscoring the need for stronger priors and for linguistically controlled experiments in evaluation.
Frequency plays a central role in human cognition, and in language processing in particular. Ther... more Frequency plays a central role in human cognition, and in language processing in particular. There is growing evidence that acceptability judgements are shaped by the statistics of the input. In this paper, we focus on a type of constraint operative in long-distance dependencies (e.g. wh-questions, relative clauses, topicalizations, etc.) which has been claimed to result from verb subcategorization frequency effects. We take a closer look at this hypothesis, and conclude that it does not account for the sentence acceptability contrasts. Rather, the evidence we find suggests that the acceptability of these dependencies hinges on clause-level semantic-pragmatic factors.
Syntactic knowledge involves not only the ability to combine words and phrases, but also the capa... more Syntactic knowledge involves not only the ability to combine words and phrases, but also the capacity to relate different and yet truth-preserving structural variations (e.g. passivization, inversion, topicalization, extraposition, clefting, etc.), as well as the ability to infer that these syntactic variations all adhere to common morphosyntactic rules, like subjectverb agreement. Although there is some evidence that BERT has rich syntactic knowledge, our adversarial approach suggests that it is not deployed in a robust and linguistically appropriate way. English BERT can be tricked to miss even quite simple syntactic generalizations, when compared with GPT-2, underscoring the need for stronger priors and for linguistically controlled experiments in evaluation.
Frequency plays a central role in human cognition, and in language processing in particular. Ther... more Frequency plays a central role in human cognition, and in language processing in particular. There is growing evidence that acceptability judgements are shaped by the statistics of the input. In this paper, we focus on a type of constraint operative in long-distance dependencies (e.g. wh-questions, relative clauses, topicalizations, etc.) which has been claimed to result from verb subcategorization frequency effects. We take a closer look at this hypothesis, and conclude that it does not account for the sentence acceptability contrasts. Rather, the evidence we find suggests that the acceptability of these dependencies hinges on clause-level semantic-pragmatic factors.