"What is on your mind?" Automated Scoring of Mindreading in Childhood and Early Adolescence

TL;DR

This paper introduces MIND-CA, a new dataset for automated assessment of mindreading in children and adolescents, demonstrating the potential of NLP methods for this developmental psychology task.

Contribution

The paper presents the first dataset and machine learning approach for automated scoring of mindreading ability in middle childhood and early adolescence.

Findings

Promising results with state-of-the-art NLP models

Successful creation of a large, annotated corpus

Demonstrated applicability of NLP to developmental psychology

Abstract

In this paper we present the first work on the automated scoring of mindreading ability in middle childhood and early adolescence. We create MIND-CA, a new corpus of 11,311 question-answer pairs in English from 1,066 children aged 7 to 14. We perform machine learning experiments and carry out extensive quantitative and qualitative evaluation. We obtain promising results, demonstrating the applicability of state-of-the-art NLP solutions to a new domain and task.