Abstract
The present study investigates three relationships with respect to their agreement with the Menzerath-Altmann law on data from Japanese: (1) Sentence length and clause length, (2) clause length and argument length, and (3) sentence length and argument length. The data supports Hypotheses (1) and (3), whereas (2) is not supported. Possible reasons for this result are discussed. Moreover, the influence of sentence structure on the relationships will be investigated with reference to the following aspects: The type of arguments of the individual verb, the depth of the clause, the length of the embedding clause, and the order of arguments in the clause.
Acknowledgements
The study is supported by the Alexander von Humboldt Foundation and the Japan Society for the Promotion of Science (JSPS).
I thank Professor Dr. Naoko Maeda of Gakushuin University for her comments on the definition of the Japanese clause.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 Teupenhayn and Altmann (Citation1984) analysed a relationship between the sentence length in the number of clauses and a clause length.
2 Functions (1), (2) and (3) for hypotheses simply show the relationships are directly proportional or inversely proportional. Function (4) (Altmann, Citation1980) in Section 3 is employed to analyse our data.
3 In Section 4 we investigate these 243 predicates in 240 sentences.
4 A Japanese post position “ga” is a subject marker, and “wa” is a topic marker. However, “wa” also works as a subject marker.
5 There are discussions on the attributive clause and on verbs which have a function to lead a story (Takahashi, Citation1994; Kato, Citation2003; Ooshima, Citation2010).
6 Čech and Uhlířová (Citation2014) listed 13 categories including the time and the place.
7 Maruyama and Ogino (Citation1992), Jin (Citation1996) and Takamatsu (Citation2013) studied a distance among phrases in Japanese. This hypothesis is inspired by these studies.
8 We also investigated a relationship between the clause level and the number of morphemes per clause CL(M), and a relationship between the clause level and the number of morphemes per argument AL(M). However, both show serpentine regression curves, and we cannot obtain a clear tendency for both cases.