2025, Vol.26, No.

Modern English Education 2023 KCI Impact Factor : 1.55

pISSN : 1598-0782 / eISSN : 2586-6141
https://journal.kci.go.kr/meeso

It is well-documented that word frequency significantly impacts English consonant cluster reduction (CCR). This study investigated how the realization of English wordfinal consonant clusters differs based on word frequency in spontaneous American English speech. Overall results showed that high-frequency words exhibited higher rates of CCR than low- frequency words, indicating that word frequency exerts a significant influence on CCR. Grammatical conditioning significantly influenced CCR in high-frequency words, with greater promotion of CCR for monomorphemic forms compared to polymorphemic forms. However, in low-frequency words, grammatical conditioning was not significant, as monomorphemic forms showed similar rates of CCR to polymorphemic forms. Voicing agreement with the preceding consonant also proved significant in high-frequency words: homovoicing promoted CCR more than heterovoicing. The preceding consonant appeared to be an effective factor in CCR for both high- and low-frequency words, while the following consonant affected CCR primarily in high-frequency words. Additionally, the factor of age played a significant role only in low-frequency words, with the older group promoting CCR more strongly than the younger group.

Despite extensive research on readability formulas in language assessment, their relationship with student performance in high-stakes English as a Foreign Language (EFL) testing remains understudied. This study investigated the relationship between text readability and student performance across four item categories in English reading comprehension section of the Korean College Scholastic Ability Test (CSAT). Analysis was conducted on 335 reading passages from CSAT examinations (2017– 2024) using four readability formulas (two traditional and two advanced). Results revealed varying patterns of incorrect response rates across item categories. Main Idea Comprehension items showed the most consistent performance patterns (CV = 0.137) while Discourse Structure Inference items displayed the highest variability (CV = 0.165). Correlation analyses indicated weak relationships between readability measures and incorrect response rates, with contextual inference items showing weak but significant correlations with certain readability indices. Subsequent multiple regression analyses revealed that readability measures explained very small or small amounts of variance in incorrect response rates across item categories (R² = 0.015– 0.091). Findings of this study are discussed in terms of developing differentiated approaches to difficulty calibration tailored to each item category’s distinct linguistic features and cognitive demands.