Utilization of Dependency Type per Sentence to Identify Differences among Genres of English Texts

Utilization of Dependency Type per Sentence to Identify Differences among Genres of English Texts Masanori Oya Mejiro University m.oya@mejiro.ac.jp Abstract This study utilized the concept of type per sentence (TPS) with regards to dependency types to identify differences among genres of English texts obtained from an English language corpus. This study also attempted to suggest that TPSs can indicate some of the differences between different genres of texts. TPSs can be used as metrics to indicate how certain genres of texts are more likely to contain certain dependency types compared to others, and the higher TPSs of some dependency types can be shown to be related to higher TPSs of other dependency types. 1. Introduction The basic assumption of Dependency Grammar (henceforth DG) is that we can find a dependency relationship between each word in a sentence and another word in the same sentence, and these dependency relations among words are labelled under certain categories called “types”. The dependency between two words represents their hierarchical relationship, and the “type” of their dependency differentiates it from others. For example, the sentence “The dependency between two words represents their hierarchical relationship” is represented in the following dependency tree. The dependency types are indicated with capital letters: ROOT DOBJ NSUBJ POBJ NUM two words Root represents dependency The between POSS DET relation their hierarchical PREP AMOD Figure 1. The dependency tree for the sentence “The dependency between two words represents their hierarchical relationship.” The sentence, “The type of a dependency differentiates it from others” is represented in the tree below: ROOT NSUBJ PREP DOBJ PREP POBJ POBJ DET from others DET The of dependency a it Root differentiates type Figure 2. The dependency tree for the sentence “The type of a dependency differentiates it from others.” Along with these assumptions, this study assumes that texts of different genres show differences in frequency of use of certain dependency types; this is because each genre’s distinct use of sentence structures can contribute to building differences between various genres, and the differences in frequency of use of dependency types can be partly used as one of the metrics, along with sentence lengths, to indicate the differences among genres. In order to verify this assumption, this study investigates whether the different corpora of different genres actually show different frequencies of use for different dependency types, using type per sentence (henceforth TPS), introduced by Oya (2016), as a metric. The definition of TPS is as follows: “TPS” is the number of examples of a dependency type in a corpus (DT) divided by the number of sentences in the same corpus (S). TPS = DT / S PACLIC 32 523 32nd Pacific Asia Conference on Language, Information and Computation Hong Kong, 1-3 December 2018 Copyright 2018 by the author Intuitively speaking, the TPS of a given dependency type in a given corpus indicates the average occurrence of that dependency type in a sentence in the same corpus. Thus, it is expected that the TPS of the same dependency type …

Content truncated to 500 words