RNNs as psycholinguistic subjects: Syntactic state and grammatical dependency

By Richard Futrell, Ethan Gotlieb Wilcox, Takashi Morita, Roger Levy. 2018

64 citations recorded by the source corpus.

Recurrent neural networks (RNNs) are the state of the art in sequence modeling for natural language. However, it remains poorly understood what grammatical characteristics of natural language they implicitly learn and represent as a consequence of optimizing the language modeling objective. Here we deploy the methods of controlled psycholinguistic experimentation to shed light on to what extent RNN behavior reflects incremental syntactic state and grammatical dependency representations known to characterize human linguistic behavior. We broadly test two publicly available long short-term memory (LSTM) English sequence models, and learn and test a new Japanese LSTM. We demonstrate that these models represent and maintain incremental syntactic state, but that they do not always generalize in the same way as humans. Furthermore, none of our models learn the appropriate grammatical dependency configurations licensing reflexive pronouns or negative polarity items.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

Neural Network Acceptability Judgments (2018, methodology): “It is also increasingly used in computational linguistics (Linzen et al., 2016; Marvin and Linzen, 2018; Futrell et al., 2018; Wilcox et al., 2018, 2019).”
A Structural Probe for Finding Syntax in Word Representations (2019, background): “Recent work has analyzed model behavior to determine if a model understands hierarchy and other linguistic phenomena (Linzen, 2018; Gulordava et al., 2018; Kuncoro et al., 2018; Linzen and Leonard, 2018; van Schijndel and Linzen, 2018; Tang et al., 2018; Futrell et al., 2018).”
BLiMP: The Benchmark of Linguistic Minimal Pairs for English (2019, neutral): “Anaphora/binding Marvin and Linzen (2018), Futrell et al. (2018), Warstadt et al. (2019b) Subj.”
Structural Supervision Improves Learning of Non-Local Grammatical Dependencies (2019, methodology): “…number of non-local grammatical dependencies, such as subject-verb number agreement and ﬁller-gap licensing (Linzen et al., 2016; Wilcox et al., 2018), although they fail to learn others, such as Negative Polarity Item and anaphoric pronoun licensing (Marvin and Linzen, 2018; Futrell et al., 2018).”
A Systematic Assessment of Syntactic Generalization in Neural Language Models (2020, background): “…capture many syntactic generalizations , ranging from subject–verb agreement Materials and code can be found at https://github. com/cpllab/syntactic-generalization . to long-distance ﬁller–gap dependencies (Linzen et al., 2016; Marvin and Linzen, 2018; Futrell et al., 2018; Wilcox et al., 2019b).”
Studying the Inductive Biases of RNNs with Synthetic Variations of Natural Languages (2019, background): “…large number of studies, some mentioned in the introduction, have used syntactic prediction tasks to examine the generalizations acquired by neural models (see also Bernardy and Lappin 2017; Futrell et al. 2018; Lau et al. 2017; Conneau et al. 2018; Et-tinger et al. 2018; Jumelet and Hupkes 2018).”