Developing AI-assisted application for pronunciation and speaking evaluation aids in helping students enhancing communicative competency. Recent advancement of speech technology and natural language processing makes it possible to build data-driven computer assisted language learning system with a limited capability of recognizing oral inputs. This paper overviews the needs and development such corpus-based computer assisted system of English learning and suggests a need for more sophisticated corpus construction for intermediate and advanced learners of English who are in need of enhancing their pronunciation and speaking competency.

The purpose of this study is to examine how hedging expressions in terms of lexicon and syntactic structure were presented in English listening test of Korean CSAT as a pragmatic strategy for Korean learners. To achieve this purpose, two questions were raised: Based on English listening test of Korean CSAT, 1) how do hedging expressions appear in terms of lexical structure? and 2) how do hedging expressions appear in terms of syntactic structure? In order to answer the research questions, the items of listening test from 1994 to 2022 were divided into two groups, the monologue and dialogue to figure out differences of their phonetic properties according to Nunan(1991)’s classification. Also, Crompton(1997) and Kim(2008)’s taxonomy of hedges were used because they defined hedging expressions as a pragmatic strategy. This study was analyzed by the corpus program, Antconc. The results indicated that hedging expressions used in the listening test in Korean CSAT appear depending on given taxonomy of hedges differently. First, the lexical hedging expressions were used more frequently than syntactic ones. In addition, these hedging expressions used in the listening test in Korean CAST appear depending on the type of linguistic data differently. Second, the rate of all syntactical hedging expressions in the monologues was approximately six times higher than in the dialogues. Some syntactical hedging expressions also have different meanings depending on the type of linguistic data, even if they are the same forms. More detailed accounts for analyzing are discussed in the paper.

This study is to construct and compare a high frequency word list of different topis in L1 and L2 children’s English newspapers. The Kids Times and My Apple News are selected for L2 children’s English newspapers, and CBBC is chosen for L1 children’s English newspaper. The chosen children’s English newspapers are constructed into text corpus for the analysis of topic-based frequency list extraction. The topics covers five areas: Special Focus, People, Nature & Animals, Sports, International News. The findings are the frequent words concur in both L1 and L2 English newspapers. They are highly co-related to the national curriculum vocabulary list. Discrepancies between L1 and L2 word list can be utilized for inter-cultural awareness learning opportunities.

This study aims to study some similarities and differences in content, vocabulary and linguistic features of maritime safety news and maritime security news in self-built corpora through the two strategies of keyword extraction: comparison against a representative general English corpus and cross-validation. It is found that the content of the maritime safety news is concerned with the safety of operation process, safe regulation, and the recent topic: Covid-19, which is to report the safety of the ship and related personnel. On the other hand, the content of the maritime security news is focused on pirate attacks, robberies, and other external factors of security. Cross-validation shows that maritime safety news prefers to use modal verbs, while maritime security news prefers to use the passive voice. We also explore which syntactic pattern is preferred through adjacent collocates of ‘safety’ and ‘security’ in both of the news. In addition, we investigate their window span (-5 to +5) collocation through examination of several statistical measures to find out similarities and differences between a pair of near-synonyms ‘safety’ and ‘security’ used in the two corpora.

Metadiscourse refers to those aspects of texts which can lead to the construction of writer and reader interaction. Taken this into account, this study set to investigate how men and women used metadiscourse features in social media in the English language context. Being corpus-based in nature, this study privileged a corpus of 145,213 words gathered from such social media as Facebook, Instagram, and Twitter divided by gender of male and female who wrote in English as the language of communication. For analyzing metadiscourse features, Hyland’s classification (2005) was applied that divides metadiscourse features into two main categories of interactive and interactional. The results indicated that in overall both male and female corpora contained more interactional metadiscourse features than interactive. In addition, female corpus contained more metadiscourse features than male corpus. Added to this, it was found that self-mentions were the most frequent instances of metadiscourse features in both male and female corpora. Moreover, it was revealed that although male corpus had instances of code glosses, the female corpus lacked it. The results showed that there was a statistically significant difference between males and females in using metadiscourse features. The findings of this research can have useful implications in such areas as gender studies, comparative studies, corpus-based studies, and contrastive linguistics.