|Vol. 20 No. l
|Vol. 20 No. 2
|Vol. 20 No. 3
|Vol. 20 No. 4
|Vol. 20 No. 5
|Vol. 20 No. 6
|Vol. 20 No. 7
|Vol. 20 No. 8
The methods are based upon categories of meaning of kanji strings and a common orthographic characteristics of verbs and adjectives.
My program uses only a small table containing 592 units. In my experiments I have obtained 92.28% correct results.
These methods are simple and of practical use.
Especially, the second type is shown to consist of case particles ``ni'' and ``yori'', for which we discuss their case relations. As a result, the method of processing the conjugation words as case elements will be presented.
In the paper, we classified the type of coordinate structure in accordance with the construction level of the sentences. The types are of morpheme level, terminal level, phrase level, and sentence level. In addition we have refined on them, whether they have a symbol of coordination or not. To eliminate the ambiguity on parsing coordinate structure, the sentences are evaluated to low, if they have problems in either syntactic or semantic process. The major part of the minus point in parsing coordination structure is the degree of morpheme similarity. Finally, we demonstrate our work by using the parser system, JUST, which we developed in respect to the coordinate structure.
I use ``Bunruigoihyou'' ( ``Word List by Semantic Principle'') to classify into the semantic category of words. In this paper, I survey the distribution of polysemic words from the perspective of the categories assigned in the ``Bunruigoihyou''. Most words with a high degree of polysemy are verbs. Among nouns, words belonging to Category ``1.1 Abstract Relation'' and ``1.5 Nature and Natural Phenomena'' have a high degree of polysemy. On the other hand, words belonging to Category ``1.3 Mind and Acts'' have a low degree of polysemy. Words representing the body are highly polysemic from the view-point of more detailed categories.
Also I analyze the meaning structure of polysemic words from thc view- point of semantic categories, assuming the central meaning of a word to be related to the semantic categories assigned in the ``Bunruigoihyou''. Except for words in Category 1.5, derived meaning tends to remain within the original semantic category. On the other hand, polysemic meaning derived from words in Category 1.5 are generally assigned to other categories. (i.e. te, hone, hikari etc. ) The latter derivations are based on a word form, position, function, and essential properties. Category 1.5 derivations are extremely diverse.
The results show the following characteristics of each material as
(1) Speakers' introspection can reflect current tendencies of language change.
(2) Newspaper CD-ROMs, large electronic texts, include a variety of styles and often show interesting examples.
(3) Japanese dictionaries reflect language facts for the last several decades and include old-fashioned descriptions.
This paper describes the analysis and the classification of compound nouns using so-called ``analysis by synthesis'' method, which is on the analogy of speech processing. 34 semantic features are defined to nouns. About every 5 nouns are selected for a feature, and the combination of these nouns make compound nouns. Whether the resulting compound nouns are decided as acceptable or not is examined by hand, and connection rules are made. Abount 70% compound nouns are decided correctly based on these rules.
In Japanese, for example, the frequency of Chinese characters as well as placement of commas are indicators of a writer's characteristic style.
In the present study, the statistical analysis is based on the distribution of particles from 60 diary entries of 6 writers. It is contended that the distribution of particles differ from writer to writer and may therefore be considered one of the features that make up an individual's characteristic writing style, thus providing valuable information for determining or recognizing the authorship of a particular text.