A corpus-based analysis of Korean segments produced by Chinese learners

Abstract

The increasing demand for learning Korean as a foreign language yields a strong need for a CAPT system that is able to provide automatic tutoring. However, there is limited research on Korean pronunciation produced by non-natives. As a preliminary research towards developing a CAPT system for Chinese learners of Korean, we survey key findings of previous studies. And then, based on corpus analysis, we provide improved descriptions of segmental variation patterns of Korean produced by Chinese learners. The most salient variation is substitutions of liquid sounds: 33.0% of flap sounds were realized as lateral, and 35.0% of lateral sounds were realized as 3 major variation patterns. By quantifying all the variation patterns with statistical data, we resolve disagreements between previous studies, indicate new findings, which are important resources for developing a CAPT system, and lay the groundwork for Korean language learning for various L1 backgrounds.

Linguistic Factors Affecting Evaluation of L2 Korean Speech Proficiency

“…Southeast Asian languages here include Filipino, Cambodian, and Vietnamese, which belong to Austronesian language group. The patterns are organized into five groups (coda deletion and insertion, detensification, aspiration, and vowel epenthesis), which are the salient mispronunciation patterns in non-native Korean speech [20] [21]. Chinese and Japanese L1 speakers show higher coda deletion errors than Southeast Asian learners (Figure 3).…”

Automatic proficiency assessment of Korean speech read aloud by non‐natives using bidirectional LSTM‐based speech recognition

“…Consequently, numerous research results have been reported on automatic proficiency assessment methods for non‐native speech that is read aloud [6–13] and for spontaneous speech [14–17]. However, there has been limited research on proficiency assessment of non‐native Korean speech [18]. Moreover, most research has been focused on the analysis of pronunciation variabilities in non‐native Korean speech.…”

Automatic pronunciation assessment of Korean spoken by L2 learners using best feature set selection

“…Although demand for Korean language education is steadily growing, studies regarding pronunciation assessment of Korean as a foreign language are limited to contrastive analysis between Korean and the native language (L1) of learners [7] [8]. Furthermore, in the viewpoint of CALL/CAPT for the Korean language, except for some quantitative analyses [9][10] [11] [12], there has been a lack of studies regarding automatic pronunciation assessment in a Korean language context. As an early stage of developing a CAPT system for Korean language learners, this paper proposes a method for automatic pronunciation assessment of Korean spoken by L2 learners by selecting the best feature set from a collection of most well-known features in the literature.…”