We introduce the automatic annotation of noun phrases in parsed sentences with tags from a fine-grained semantic animacy hierarchy. This information is of interest within lexical semantics and has potential value as a feature in several NLP tasks. We train a discriminative classifier on an annotated corpus of spoken English, with features capturing each noun phrase's constituent words, its internal structure, and its syntactic relations with other key words in the sentence. Only the first two of these three feature sets have a substantial impact on performance, but the resulting model is able to fairly accurately classify new data from that corpus, and shows promise for binary animacy classification and for use on automatically parsed text.
展开▼
