A Corpus-Based Analysis of Canonical Word Order of Japanese Double Object Constructions

Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, pages 2236–2244, Berlin, Germany, August 7-12, 2016. c⃝2016 Association for Computational Linguistics A Corpus-Based Analysis of Canonical Word Order of Japanese Double Object Constructions Ryohei Sasano Manabu Okumura Tokyo Institute of Technology {sasano,oku}@pi.titech.ac.jp Abstract The canonical word order of Japanese double object constructions has attracted considerable attention among linguists and has been a topic of many studies. How- ever, most of these studies require either manual analyses or measurements of hu- man characteristics such as brain activities or reading times for each example. Thus, while these analyses are reliable for the ex- amples they focus on, they cannot be gen- eralized to other examples. On the other hand, the trend of actual usage can be col- lected automatically from a large corpus. Thus, in this paper, we assume that there is a relationship between the canonical word order and the proportion of each word or- der in a large corpus and present a corpus- based analysis of canonical word order of Japanese double object constructions. 1 Introduction Japanese has a much freer word order than En- glish. For example, a Japanese double object con- struction has six possible word orders as follows: (1) a: Ken-ga Aya-ni camera-wo miseta. Ken-NOM Aya-DAT camera-ACC showed b: Ken-ga camera-wo Aya-ni miseta. Ken-NOM camera-ACC Aya-DAT showed c: Aya-ni Ken-ga camera-wo miseta. Aya-DAT Ken-NOM camera-ACC showed d: Aya-ni camera-wo Ken-ga miseta. Aya-DAT camera-ACC Ken-NOM showed e: Camera-wo Ken-ga Aya-ni miseta. camera-ACC Ken-NOM Aya-DAT showed f: Camera-wo Aya-ni Ken-ga miseta. camera-ACC Aya-DAT Ken-NOM showed In these examples, the position of the verb miseta (showed) is fixed but the positions of its nomina- tive (NOM), dative (DAT), and accusative (ACC) ar- guments are scrambled. Note that, although the word orders are different, they have essentially the same meaning “Ken showed a camera to Aya.” In the field of linguistics, each language is as- sumed to have a basic word order that is funda- mental to its sentence structure and in most cases there is a generally accepted theory on the word order for each structure. That is, even if there are several possible word orders for essentially same sentences consisting of the same elements, only one of them is regarded as the canonical word order and the others are considered to be gener- ated by scrambling it. However, in the case of Japanese double object constructions, there are several claims on the canonical argument order. There have been a number of studies on the canonical word order of Japanese double ob- ject constructions ranging from theoretical stud- ies (Hoji, 1985; Miyagawa and Tsujioka, 2004) to empirical ones based on psychological exper- iments (Koizumi and Tamaoka, 2004; Nakamoto et al., 2006; Shigenaga, 2014) and brain science (Koso et al., 2004; Inubushi et al., 2009). How- ever, most of them required either manual analyses or measurements of human characteristics such as brain activities or reading times for each example. Thus, while these analyses are reliable for the ex- …

Content truncated to 500 words