Work

Universal Dependencies and Semantics for English and Hebrew Child-directed Speech

Year: 2022

Type: conference-paper

Abstract: While corpora of child speech and child-directed speech (CDS) have enabled major contributions to the study of child language acquisition, semantic annotation for such corpora is still scarce and lacks a uniform standard. We compile two CDS corpora—in English and Hebrew—with syntactic and semantic annotations. We employ a methodology that enforces a cross-linguistically consistent representation, building on recent advances in dependency representation and semantic parsing. Our semi-automatic syntactic annotation follows the Universal Dependencies standard (UD; de Marneffe et al., 2021), adapted to suit the CDS genre. To induce semantic forms, we develop an automatic method for transducing UD structures into sentential logical forms (LFs). The two representations have complementary strengths: UD structures are language-neutral and support direct annotation, whereas LFs are neutral as to the syntax-semantics interface, and transparently encode semantic distinctions.

Source: Edinburgh Research Explorer (University of Edinburgh)

DOI: 10.7275/2fhp-bf70

Authors: Szubert, Ida, Abend, Omri, Schneider, Nathan, Gibbon, Samuel, Goldwater, Sharon, Steedman, Mark

Institutions: University of Edinburgh, Hebrew University of Jerusalem, Georgetown University

Language: English

Topic: Natural Language Processing Techniques

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funders: James S. McDonnell Foundation, European Commission

Award: Understanding synergies in language acquisition through computational modeling