HED LANG - A Hierarchical Event Descriptors library extension for annotation of language cognition experiments

Abstract

Experimental design in language cognition research often involves presenting language material while measuring associated behavior and/or neural activity. To make the collected data easily and fully analyzable by both the original data authors and others, it is important to have detailed information about the stimulus presentation events, including the nature and properties of the presented stimuli, using a common vocabulary and syntax. We present HED LANG, a library extension of the Hierarchical Event Descriptors (HED) event annotation schema for time series behavioral and neuroimaging data. HED LANG provides an open source, standardized vocabulary for building detailed, human- and machine-actionable annotations for language cognition datasets. It builds on existing annotation systems in linguistics and is supported by a suite of HED tools for annotating, validating, searching, and characterizing HED-tagged datasets. HED LANG is specific enough to allow event data annotation at the range of levels needed to support many current research paradigms and analyses. Both HED and HED LANG are open to community input and participation, enabling them to evolve with continuing developments in language cognition research.

Fig. 1

Summary overview of the embedding of a few of the LANG schema tags (green) under three top-level Standard schema tags (blue).

Fig. 2

Added language item properties. Summary overview of the language item properties in HED LANG. To examplify the additional tree structure we also show the first level of categories added under Lexical-role.