In this paper, we investigate the effect of enhancing lexical embeddings in LSTM language models (LM) with syntactic and semantic representations. We evaluate the language models using perplexity, and we evaluate the performance of the models on the task of predicting human sentence acceptability judgments. We train LSTM language models on sentences automatically annotated with universal syntactic dependency roles (Nivre et al., 2016), dependency tree depth features, and universal semantic tags (Abzianidze et al., 2017) to predict sentence acceptability judgments. Our experiments indicate that syntactic depth and tags lower the perplexity compared to a plain LSTM language model, while semantic tags increase the perplexity. Our experiments also show that neither syntactic nor semantic tags improve the performance of LSTM language models on the task of predicting sentence acceptability judgments.
展开▼
