Metaphors in Pre-Trained Language Models: Probing and Generalization Across Datasets and Languages
This research addresses the problem of understanding metaphorical processing in AI for cognitive and NLP scientists, but it is incremental as it builds on existing probing methods without introducing new paradigms.
The study investigated whether pre-trained language models encode metaphorical knowledge by probing their representations across multiple datasets and languages, finding that such knowledge is present, primarily in middle layers, and is transferable between languages and datasets with consistent annotation.
Human languages are full of metaphorical expressions. Metaphors help people understand the world by connecting new concepts and domains to more familiar ones. Large pre-trained language models (PLMs) are therefore assumed to encode metaphorical knowledge useful for NLP systems. In this paper, we investigate this hypothesis for PLMs, by probing metaphoricity information in their encodings, and by measuring the cross-lingual and cross-dataset generalization of this information. We present studies in multiple metaphor detection datasets and in four languages (i.e., English, Spanish, Russian, and Farsi). Our extensive experiments suggest that contextual representations in PLMs do encode metaphorical knowledge, and mostly in their middle layers. The knowledge is transferable between languages and datasets, especially when the annotation is consistent across training and testing sets. Our findings give helpful insights for both cognitive and NLP scientists.