Prompt-based Content Scoring for Automated Spoken Language Assessment
Keelan Evanini, Shasha Xie, Klaus Zechner · 2013
This paper investigates the use of prompt-based content features for the automated as-sessment of spontaneous speech in a spoken language proficiency assessment. The results show that single highest performing prompt-based content feature measures the number of unique lexical types that overlap with the listening materials and are not contained in either the reading materials or a sample re-sponse, with a correlation of r = 0.450 with holistic proficiency scores provided by hu-mans. Furthermore, linear regression scor-ing models that combine the proposed prompt-based content features with additional spoken language proficiency features are shown to achieve competitive performance with scoring models using content features based on pre-scored responses. 1