This paper describes technology developed to automatically grade students on their English spontaneous spoken language proficiency with common european framework of reference for languages (CEFR) level. Our automated assessment system contains two tasks: elicited imitation and spontaneous speech assessment. Spontaneous speech assessment is a challenging task that requires evaluating various aspects of speech quality, content, and coherence. In this paper, we propose a multimodal and multitask transformer model that leverages both audio and text features to perform three tasks: scoring, coherence modeling, and prompt relevancy scoring. Our model uses a fusion of multiple features and multiple modality attention to capture the interactions between audio and text modalities and learn from different sources of information.

For morphologically rich languages, word embeddings provide less consistent semantic representations due to higher variance in word forms. Moreover, these languages often allow for less constrained word order, which further increases variance. For the highly agglutinative Hungarian, semantic accuracy of word embeddings measured on word analogy tasks drops by 50-75% compared to English. We observed that embeddings learn morphosyntax quite well instead. Therefore, we explore and evaluate several sub-word unit based embedding strategies – character n-grams, lemmatization provided by an NLP-pipeline, and segments obtained in unsupervised learning (morfessor) – to boost semantic consistency in Hungarian word vectors. The effect of changing embedding dimension and context window size have also been considered. Morphological analysis based lemmatization was found to be the best strategy to improve embeddings’ semantic accuracy, whereas adding character n-grams was found consistently counterproductive in this regard.