Zero-shot Cross-Linguistic Learning of Event Semantics

TL;DR

This paper presents a zero-shot cross-lingual model for predicting lexical aspects of image captions across diverse languages, revealing surprising similarities in event framing despite linguistic differences.

Contribution

It introduces a novel zero-shot learning approach for event semantics prediction across typologically diverse languages without requiring annotated data for each language.

Findings

Lexical aspects can be predicted across languages without annotated data.

Languages show similar event framing patterns despite linguistic differences.

The model generalizes well to unseen languages.

Abstract

Typologically diverse languages offer systems of lexical and grammatical aspect that allow speakers to focus on facets of event structure in ways that comport with the specific communicative setting and discourse constraints they face. In this paper, we look specifically at captions of images across Arabic, Chinese, Farsi, German, Russian, and Turkish and describe a computational model for predicting lexical aspects. Despite the heterogeneity of these languages, and the salient invocation of distinctive linguistic resources across their caption corpora, speakers of these languages show surprising similarities in the ways they frame image content. We leverage this observation for zero-shot cross-lingual learning and show that lexical aspects can be predicted for a given language despite not having observed any annotated data for this language at all.