Second, can collocation network analysis provide a new perspective to explain the differences of near-synonyms from a microscopic level. Third, domain network semantic analysis is useful for distinguishing one near-synonym from another at the macroscopic level.
Introduction
Focus of Inquiry
In this study, I will focus on near-synonymous collocations in the maritime industry. Second, can collocational network analysis offer a new perspective to explain differences between near-synonyms from a microscopic level.
Outline of the Thesis
With brief introduction of how people traditionally distinguish between near-synonyms in the dictionary, such as ship vs. It is the semantic domain network analysis that will prove to be an effective way to show the differences between near-synonyms.
Literature Review
A Brief Synopsis
An overview of previous language network analysis papers is also provided in this section to provide explanatory power for descriptions of near-synonyms in Maritime English. Therefore, in section 2.6, the concepts of semantic domains are presented, followed by previous studies on semantic domain analysis.
Maritime English as an English for Specific Purposes (ESP)
- What is ESP?
- Maritime English as ESP
- ESP and Corpus Linguistics
Maritime English is defined as an official language within the international maritime community that contributes to the safety of navigation and the facilitation of maritime transport (Trenker, 2009). To summarize, maritime English refers to the specialized English language used at sea and in port by seafarers.
Synonymy
- Definition of Synonymy
- Synonymy as a Matter of Degree
- Criteria for Synonymy Differentiation
- Near-synonyms in Corpus Linguistics
From the definition and examples given by Cruse (above) it can be seen that he believed that there are three types of synonyms which are absolute synonyms, cognitive synonyms (i.e. propositional synonyms) and near-synonyms (i.e. plesionyms). In this thesis, near-synonyms are defined as "lexical pairs that have very similar cognitive or denotational meanings but may differ in collocational or prosodic behavior" (Xiao & McEnery, 2006, p. 108).
Collocation
- Definition of Collocation
- Collocation in Corpus Linguistics
- Definition of Collocation in Corpus Linguistics
- Collocation vs. Colligation
- Lexical Priming of Collocation in Psychology
Collocation is considered “the co-occurrence of two items (nodal words) within a specified environment (a selected span)” (Sinclair, Jones, & Daley, 2004, p. 10). Firth (1957, p. 183) saw colligation as “the interrelationship of grammatical categories in syntactic structure” and collocation as “actual words in common company.”
Language Network Analysis
- Definition
- Classification
- Basic Concepts
- Previous Studies
Collocation networks can be considered the basic type of network not only because it is common in language, but also because it is based on a stage of meaning expression. Word collocation networks (Ke, 2007), also known as collocation graphs (Choudhury & Mukherjee, 2009; Heyer, Lauter, Quasthoff, Wittig, & Wolff, 2001), are networks of words found in a document or a .
Semantic Domain Analysis
- Concepts of Semantic Domains
- Previous Studies on Semantic Domain Analysis
In addition, Faber, León-Araúz, and Reimerink (2014) drew conceptual networks of cribs showing semantic relationships between words, as shown in Figure 2.10.
Data and Methodology
Maritime English Corpus
- What is a Corpus?
- Characteristics of a Corpus
- Corpus-driven vs. Corpus-based research
- Specialized Corpora for Specialized Discourse
- Maritime English Corpus (MEC)
- Sampling of the MEC
- Size, Balance, and Representativeness
- Multi-word Compounds in the MEC
- Basic Information of the MEC
C11 INTERNATIONAL CONVENTION RELATING TO THE LIMITATION OF THE LIABILITY OF OWNERS OF MARITIME SHIPS, 1957 C12 GENEVA CONVENTION ON TERRITORIAL SEA AND APPENDIX. C34 COMMITTEE MARITIME INTERNATIONAL CONSTITUTION, 1981 C35 UNITED NATIONS CONVENTION ON THE SEA, 1982 C36 UNITED NATIONAL CONVENTION ON CONDITIONS FOR. Neither China nor most of the countries involved in the agreements is a "native English speaking" country;.
This has been particularly relevant in the construction of the MEC, where the texts have been selected to be representative of different text types within the maritime discourse. In this study, I have taken as a starting point decisions about what should be excluded and included in the corpus as "maritime texts" and built the corpus based on the availability and suitability of the maritime texts.
Methodology for Collocates Extraction
For the average word length, due to the compounding process, many words become longer, and thus a larger average word length (4.99) in Compounding MEC compared to the original MEC (4.94). Northcott (2013, p. 215) mentioned that the historical development of languages and the desire for power are the two main factors that create the image of law as something inaccessible, mysterious and frightening. On the other hand, increasing word length will help perpetuate the image of the law as something inaccessible, complex and difficult to understand for the general population, thus allowing the authority to retain power.
However, unless we are specifically looking for grammatical patterns (colligation), we are interested in the patterning of the lexical items in the text that give the text meaning. MI3 is calculated by dividing the observed frequencies of the co-occurring words by the expected frequencies of the co-occurring words therein.
Methodology for Networks Visualization
As for the spans of the windows, I decided to use adjacent collocuts, ie. span windows of one word to the left and one to the right to reduce irrelevant words and show strong semantic relationships between nodes and collocations for the purpose of exploring semantic influences. Therefore, in the range of windows 1–1, items that have a minimum co-occurrence frequency of 2 as a collocation of a given node word and a minimum MI3 score above 3 are considered to be collocation words of the node. Originally designed to discover basic patterns and network structures, it has been used for general research and teaching in social networks in fields as diverse as information science, biology, geography, economics, and political science.
This is a simple implementation of the spring embedding algorithm of Kamada and Kawai (1989), which is one of the force graph layout algorithms. Computed with force-based graph layout algorithms, this algorithm has the advantage of drawing networks.
Methodology for Semantic Tagging
I then assembled all bidirectional networks connected to the node and used 2D "spring embedding" algorithms to visualize the data. O4 Physical characteristics O4.1 General appearance and physical characteristics O4.2 Assessment of appearance (pretty etc.) O4.3 Color and color patterns. After the completion of analysis by the software, manual trial checking was performed before serious discussions of the marker results were undertaken.
Process of Data Analysis
Collocation Network Analysis of Near-synonyms
- Meaning Differences
- Ship vs. Vessel
- Maritime vs. Marine
- Sea vs. Ocean
- Safety vs. Security
- Port vs. Harbor
- Similarity Degree of Groups of Near-synonyms
- Similarity Degree Based on Number of Shared Collocates
- Similarity Degree Based on MI3 Cosine Similarity
- Collocation Network Analysis
- Ship vs. Vessel
- Maritime vs. Marine
- Sea vs. Ocean
- Safety vs. Security
- Port vs. Harbor
- Advantages and Limitations of Collocation Network Analysis
Ship is defined as "a large vessel for transporting people or goods by sea", originating in the 14th century, from Old English scipian. Maritime is defined as "of the sea, sailing or shipping", originating in the mid-16th century from the Latin maritimus. In the Oxford Advanced Learner's English-Chinese Dictionary (Hornby, 2002, p.), although derived from different words, safety and security are used interchangeably to define each other.
On the contrary, originating from the mid-15th century, from the Latin securitas, security is defined as. Originating in the early 12th century, from Middle English herberwe, haven (haven in BrE) is defined as "a shelter for ships".
Semantic Domain Network Analysis of Near-synonyms
Comparison between Collocation and Semantic Domain Analysis
The total number of collocations, subsemantic fields, main semantic fields of five groups of near synonyms. The total number of collocations, subsemantic domains, main semantic domains that are exclusive to each near-synonym. In this case, it is highly recommended that students refer to sub-semantic areas to generalize the exclusive concept.
An additional interesting observation of the total number of collocates, subsemantic domains and main semantic domains shared by each pair of near-synonyms, as shown in Table 5.3: with the decrease of cosine similarity, the number of shared collocates and semantic domains decreases accordingly , with Total number of collocations, sub-semantic domains, main semantic domains shared by each pair of near-synonyms.
Semantic Domain Network Analysis of Exclusiveness
- Ship vs. Vessel
- Maritime vs. Marine
- Sea vs. Ocean
- Safety vs. Security
- Port vs. Harbor
Therefore, through semantic domain network analysis, we can clearly see which semantic domain is exclusive to the near-synonyms. Through semantic domain network analysis, we can clearly see which semantic domain is exclusive to maritime or marine. Through semantic domain network analysis, we can clearly see which semantic domain is exclusive to the ocean or the sea.
Through semantic domain network analysis, we can clearly see which semantic domain is exclusive to safety or security. Comparing and combining the results from these two semantic domain networks, it appears that Harbor has one G2.2 subcategory (general ethics);
Analysis of Shared Semantic Domains
For the first situation, security tends to merge more often into most shared semantic fields. However, the W3 subsemantic domain tends to correlate with safety with marine at a frequency of 255 compared to 11 for safety. In the third situation, in the main semantic domain I, security only joins financial ones belonging to I1 (Money in general), but security tends to join I2.1 (Business in general) such words.
In the first situation, marine tends to collocation with words in the semantic domains A, B, N, O and X; while seafaring usually collocates with words in semantic domains E, G and P. Table 5.6 shows that both seafaring and seafaring collocate with pollution, environment and environment, but these words more often favor marine (74 >2, 59> 2, 267>4).
Advantages and Limitations of Semantic Domain Network Analysis
Conclusion
Summary
For the third question, semantic domain network analysis proved to be an effective way to show the differences between near synonyms. Semantic domain information discussed in this thesis can help beginners of L1 and L2 understand the specific meanings of near-synonyms. When the learners want to distinguish near-synonyms by words that only coincide with them (i.e. by the exclusivity), main semantic domain network analysis shows us the exclusive domains to a certain near-synonym.
When students want to distinguish near-synonyms through a semantic domain shared by both near-synonyms, it is recommended that both the number of collocations and the frequency of collocations in the main semantic domain be considered. If a word A belongs to a semantic domain shared by both near-synonyms B and C, then in the semantic domain, (i) if.
Limitations and Implications
Mukherjee (Eds.), Dynamics on and of complex networks: Applications Biology, Computer Science, Economics, and the Social Sciences (pp. 145–166). Network Analysis of the Maritime English Corpus with Multiword Compounds: Keyword Networks and Collocation Networks (unpublished doctoral dissertation). In Proceedings of the workshop on Semantic labeling from beyond named entities for NLP tasks in collaboration with 4th.
In Proceedings of the 2nd ACM SIGIR (Association for Computing Machinery Interest Group on Information Retrieval) International Workshop on Query Representation and Understanding (pp. 3-5). Learning collocations: Do number of collocations, node word position, and synonymy affect learning.
