Textual Analysis of Conceptual Associations in CEFR B2 Level Texts: A Network-based Semantic Representation Approach
Issued Date
2025-09-01
Resource Type
eISSN
2651088X
Scopus ID
2-s2.0-105029960421
Journal Title
Suranaree Journal of Social Science
Volume
19
Issue
3
Rights Holder(s)
SCOPUS
Bibliographic Citation
Suranaree Journal of Social Science Vol.19 No.3 (2025)
Suggested Citation
Siripol P., Rhee S., Thirakunkovit S., Liang-Itsara A. Textual Analysis of Conceptual Associations in CEFR B2 Level Texts: A Network-based Semantic Representation Approach. Suranaree Journal of Social Science Vol.19 No.3 (2025). doi:10.55766/sjss278569 Retrieved from: https://repository.li.mahidol.ac.th/handle/123456789/115184
Title
Textual Analysis of Conceptual Associations in CEFR B2 Level Texts: A Network-based Semantic Representation Approach
Author(s)
Author's Affiliation
Corresponding Author(s)
Other Contributor(s)
Abstract
Background and Objectives: Lexical cohesion is vital for text comprehension, especially for learners progressing through CEFR levels. While research has focused on logical relations like synonymy and part-whole relationships, conceptual associations remain underexplored. These associations are crucial for cognitive processing and discourse comprehension but appear to be underrepresented in CEFR-B-leveled texts, which may potentially hinder learners’ preparation for C1-level demands. This study examines the patterns and prevalence of conceptual associations in B2 texts, their comparison with logical relations, and the impact of topic complexity on their distribution. Methodology: The study was conducted in two phases, with the initial phase involving the collection of verified B2 level texts. In the second phase, automated analysis via the UCREL Semantic Analysis System (SAS) was used to categorize words into broad conceptual groups, while a manual approach based on Town’s (2021) taxonomy was used to verify their actual association. A semantic network analysis based on Yang and González-Bailón’s (2017) framework was adapted to examine concept clustering. The semantic network was automatically generated and quantified by the numbers of nodes and clusters present in B2 level texts. Main Results: B2 texts showed an uneven use of lexical cohesion, relying more on simpler, explicit logical relationships. In all five texts examined, the use of logical relations (such as parent-child and part-whole relationships) outnumbered the use of other conceptual associations. In the five texts combined, logical relations occurred 92 times, whereas conceptual associations occurred 39 times. While this aids initial clarity, it creates a gap for learners moving to higher proficiency levels, where they need to connect ideas less explicitly via modeled B2 texts. Despite the similar totals of cohesive relationships (logical and conceptual associations) from the approximately 20 relationships in each text, B2 texts vary significantly in their use of conceptual association. Discussions: The dominance of logical relations in B2 texts may limit learners' development of abstract reasoning and inferencing skills, which are critical at higher proficiency levels. While logical relations provide structural clarity, they lack the deeper conceptual connections needed for C1 comprehension. Topic variations also affect conceptual richness, emphasizing the need for intentional text selection. A balanced integration of conceptual associations with logical relations