A Corpus-based Language Network Analysis of Near-synonyms in a Specialized Corpus

Abstract

As the international medium of communication for seafarers throughout the

world, the importance of English has long been recognized in the maritime

industry. Many studies have been conducted on Maritime English teaching and

learning, nevertheless, although there are many near-synonyms existing in the

language, few studies have been conducted on near-synonyms used in the maritime industry. The objective of this study is to answer the following three questions. First, what are the differences and similarities between different near-synonyms in English? Second, can collocation network analysis provide a new perspective to explain the distinctions of near-synonyms from a micro-scopic level? Third, is semantic domain network analysis useful to distinguish one near-synonym from the other at the macro-scopic level? In pursuit of these research questions, I first illustrated how the idea of incorporating collocates in corpus linguistics, Maritime English, near-synonyms, semantic domains and language network was studied. Then important concepts such as Maritime English, English for Specific Purposes, corpus linguistics, synonymy, collocation, semantic domains and language network analysis were introduced. Third, I compiled a 2.5 million word specialized Maritime English Corpus and proposed a new method of tagging English multi-word compounds, discussing the comparison of with and without multi-word compounds with regard to tokens, types, STTR and mean word length. Fourth, I examined collocates of five groups of near-synonyms, i.e., ship vs. vessel, maritime vs. marine, ocean vs. sea, safety vs. security, and harbor vs. port, drawing data through WordSmith 6.0, tagging semantic domains in Wmatrix 3.0, and conducting network analyses using NetMiner 4.0. In the final stage, from the results and discussions, I was able to answer the research questions. First, maritime near-synonyms generally show clear preference to specific collocates. Due to the specialty of Maritime English, general definitions are not helpful for the distinction between near-synonyms, therefore a new perspective is needed to view the behaviors of maritime words. Second, as a special visualization method,

collocation network analysis can provide learners with a direct vision of the

relationships between words. Compared with traditional collocation tables, learners

are able to more quickly identify the collocates and find the relationship between

several node words. In addition, it is much easier for learners to find the collocates exclusive to a specific word, thereby helping them to understand the meaning specific to that word. Third, if the collocation network shows learners relationships of words, the semantic domain network is able to offer guidance cognitively: when a person has a specific word, how he can process it in his mind and therefore find the more appropriate synonym to collocate with. Main semantic domain network analysis shows us the exclusive domains to a certain near-synonym, and therefore defines the concepts exclusive to that near-synonym: furthermore, main semantic domain network analysis and sub-semantic domain network analysis together are able to tell us how near-synonyms show preference or tendency for one synonym rather than another, even when they have shared semantic domains. The options in identifying relationships of near-synonyms can be presented through the classic metaphor of "the forest and the trees." Generally speaking, we see only the vein of a tree leaf through the traditional way of sentence-level analysis. We see the full leaf through collocation network analysis. We see the tree, even the whole forest, through semantic domain network analysis.