An Investigation of Onomatopoeias' Senses Classification Based on Modified Verbs

Abstract

One onomatopoeia word has plural senses. A modified verb of an onomatopoeia, which is clue to decide a meaning of onomatopoeia, is investigated in this paper. Modified verbs are extracted from sentences in the "local assembly corpus2010"(LAC2010). The sets of "onomatopoeia and modified verb"(O-MV sets) used with high frequency with asingle sense in the LAC2010 are examined by using two other corpuses. One is a same genre corpus, the "local assemblycorpus 2008-09" (LAC2008-09). The other is the corpus which includes various genres of documents, the BCCWJ. 99.4% of O-MV sets in the LAC2008-09 have a same sense with the LAC2010. A modified verb contributes to decide the sense of onomatopoeia among certain genre of documents. 78.0% of O-MV sets in the BCCWJ have a same sense with the LAC2010. Negative examples in the BCCWJ are mainly caused by (i)errors of dependency analysis, (ii)orthographical variants and (iii)implication of plural senses. Onomatopoeia's sense which is decided with modified verb is not depending on the genre of document. However, there are some biases in occurrence of modified verb among the genre of documents.