Do Syntactic Probes Probe Syntax? Experiments with Jabberwocky Probing

06/04/2021
by   Rowan Hall Maudslay, et al.
0

Analysing whether neural language models encode linguistic information has become popular in NLP. One method of doing so, which is frequently cited to support the claim that models like BERT encode syntax, is called probing; probes are small supervised models trained to extract linguistic information from another model's output. If a probe is able to predict a particular structure, it is argued that the model whose output it is trained on must have implicitly learnt to encode it. However, drawing a generalisation about a model's linguistic knowledge about a specific phenomena based on what a probe is able to learn may be problematic: in this work, we show that semantic cues in training data means that syntactic probes do not properly isolate syntax. We generate a new corpus of semantically nonsensical but syntactically well-formed Jabberwocky sentences, which we use to evaluate two probes trained on normal data. We train the probes on several popular language models (BERT, GPT, and RoBERTa), and find that in all settings they perform worse when evaluated on these data, for one probe by an average of 15.4 UUAS points absolute. Although in most cases they still outperform the baselines, their lead is reduced substantially, e.g. by 53 question: what empirical scores constitute knowing syntax?

READ FULL TEXT

page 1

page 2

page 3

page 4

research
06/23/2022

AST-Probe: Recovering abstract syntax trees from hidden representations of pre-trained language models

The objective of pre-trained language models is to learn contextual repr...
research
05/18/2023

Evidence of Meaning in Language Models Trained on Programs

We present evidence that language models can learn meaning despite being...
research
10/01/2017

Mathematical foundations of matrix syntax

Matrix syntax is a formal model of syntactic relations in language. The ...
research
10/24/2022

The Better Your Syntax, the Better Your Semantics? Probing Pretrained Language Models for the English Comparative Correlative

Construction Grammar (CxG) is a paradigm from cognitive linguistics emph...
research
11/11/2022

The Architectural Bottleneck Principle

In this paper, we seek to measure how much information a component in a ...
research
05/06/2021

Bird's Eye: Probing for Linguistic Graph Structures with a Simple Information-Theoretic Approach

NLP has a rich history of representing our prior understanding of langua...
research
02/24/2022

Probing BERT's priors with serial reproduction chains

We can learn as much about language models from what they say as we lear...

Please sign up or login with your details

Forgot password? Click here to reset