TITLE:
Three Types of Episodic Associations for the Semantic/Syntactic/Episodic Model of Language Prospective in Applications to the Statistical Translation
AUTHORS:
Zi-Jian Cai
KEYWORDS:
Language, Semantic/syntactic/episodic linguistic model, Statistical translation, Behavioral classification of living/natural words, Sentential/paragraphic categorization, Frequent word-pairs
JOURNAL NAME:
Open Access Library Journal,
Vol.4 No.8,
August
7,
2017
ABSTRACT: Recently, it was proposed by Cai a new semantic/syntactic/episodic model of language encompassing the sentential meanings, while deriving three corresponding principles from it for machine translation, respectively as first to establish the dictionary of words/phrases, second to translate the grammar, and third to determine the meanings of some words/phrases of multiple meanings by statistical translation. In this article, it is discovered three types of episodic associations for this linguistic model, prospective in applications to statistical translation, as: 1) It is classified the living/natural words and phrases of multiple meanings by behavior, adopting both the zoological/organizational/physical/categorical and affective/behavioral/logic/characteristic/changing characters to classify the nouns and verbs, the affective/behavioral/logic/characteristic/changing/spatial/temporal characters to the adjectives and adverbs, helpful to discern the meanings of them using these episodic associations with others within the sentence. 2) Likewise, it is classified the sentence/paragraph into the category of natural/social subjects like physics, biology, art, economy, etc., which was improved by the Chinese people in television from my original sentential/thematic category. 3) It is suggested to collect the frequent word-pairs during statistical translation, such as “bank money”, “war declaration”, etc., helpful to determine the episodic associations of some prepositions or terminal “which” clauses. It is suggested to use word episodic symbolization to apply them to computer. It is therefore improved the third principle of machine translation as third to determine the meanings of some words/ phrases of multiple meanings by episodic associations with others using the behavioral classification of words, the categorization of sentence/paragraph and the collection of frequent word-pairs.