Pretraining with Artificial Language: Studying Transferable Knowledge in Language Models
This work provides insights into how neural network encoders process human languages and the basis for cross-lingual transferability, though it is incremental in exploring transfer mechanisms.
The study investigated whether structural knowledge from artificial languages transfers to natural language processing, finding that pretraining with a nesting dependency structure yields some transferable knowledge, as shown by improved downstream task performance.
We investigate what kind of structural knowledge learned in neural network encoders is transferable to processing natural language. We design artificial languages with structural properties that mimic natural language, pretrain encoders on the data, and see how much performance the encoder exhibits on downstream tasks in natural language. Our experimental results show that pretraining with an artificial language with a nesting dependency structure provides some knowledge transferable to natural language. A follow-up probing analysis indicates that its success in the transfer is related to the amount of encoded contextual information and what is transferred is the knowledge of position-aware context dependence of language. Our results provide insights into how neural network encoders process human languages and the source of cross-lingual transferability of recent multilingual language models.