Do face masks introduce bias in speech technologies? The case of automated scoring of speaking proficiency

The COVID-19 pandemic has led to a dramatic increase in the use of face masks worldwide. Face coverings can affect both acoustic properties of the signal as well as speech patterns and have unintended effects if the person wearing the mask attempts to use speech processing technologies. In this paper we explore the impact of wearing face masks on the automated assessment of English language proficiency. We use a dataset from a large-scale speaking test for which test-takers were required to wear face masks during the test administration, and we compare it to a matched control sample of test-takers who took the same test before the mask requirements were put in place. We find that the two samples differ across a range of acoustic measures and also show a small but significant difference in speech patterns. However, these differences do not lead to differences in human or automated scores of English language proficiency. Several measures of bias showed no differences in scores between the two groups.

Exploring Recurrent, Memory and Attention Based Architectures for Scoring Interactional Aspects of Human-Machine Text Dialog

An important step towards enabling English language learners to improve their conversational speaking proficiency involves automated scoring of multiple aspects of interactional competence and subsequent targeted feedback. This paper builds on previous work in this direction to investigate multiple neural architectures -- recurrent, attention and memory based -- along with feature-engineered models for the automated scoring of interactional and topic development aspects of text dialog data. We conducted experiments on a conversational database of text dialogs from human learners interacting with a cloud-based dialog system, which were triple-scored along multiple dimensions of conversational proficiency. We find that fusion of multiple architectures performs competently on our automated scoring task relative to expert inter-rater agreements, with (i) hand-engineered features passed to a support vector learner and (ii) transformer-based architectures contributing most prominently to the fusion.