A Measure of Proficiency or Short‐Term Memory? Validation of an Elicited Imitation Test for SLA Research

Abstract

Elicited imitation requires listeners to listen and repeat sentences as accurately as possible. In second language acquisition (SLA) research it has been used for a variety of purposes. Recently, versions of the same elicited imitation test (EIT) have been created in 6 languages with the purpose of measuring second language proficiency (Ortega et al., ; Tracy–Ventura et al., 2014; Wu & Ortega, ). The validity of these EITs has been tested, and results are promising. However, questions remain regarding the extent to which EIT performance is mediated by learners’ memory span. The current study validates a new Korean EIT following the Ortega et al. () design and, as part of that, investigates the potential role of phonological short‐term memory in test performance. Korean as a second language learners (N = 66) took the following tests over 2 days: the Korean EIT, the listening section of the standardized Test of Proficiency in Korean (TOPIK), a speaking test, and a forward digit span test in their first language. Results indicated significant positive correlations between EIT scores and the various proficiency measures, but a weak and nonsignificant correlation between the EIT and forward digit span scores. Together, these results provide support for this EIT as a valid and reliable proficiency measure for use in SLA research.

The Roles of Working Memory and Oral Language Abilities in Elicited Imitation Performance

“…Similar results were found in Kim et al. (), which developed and validated a Korean EIT following Ortega et al.’s () design and explored the role of PSTM in EIT performance. A total of 66 L2 learners across three course levels (beginner, intermediate, and advanced) took the Korean EIT, listening and speaking tests, and a PSTM test.…”

“…Similar findings have been corroborated by SLA studies such as Kim et al. () and Okura and Lonsdale (), which indicated that EITs highly correlated with other measures of L2 proficiency but not WM scores. Okura and Lonsdale () administered a 60‐item English EIT and a 10‐item WM test to L2 learners ( n = 40) across six course levels (Levels 0–5).…”

“…Therefore, Kim et al. () concluded that their Korean EIT does not favor reliance on rote repetition, corroborating the results of Okura and Lonsdale ().…”

“…It should be noted that neither Kim et al. () nor Okura and Lonsdale () explored the potential mediating role of language proficiency in the relationship between WM and EIT performance. It is relevant, too, to acknowledge that Serafini and Sanz () included as one of their assessment tasks an elicited oral imitation task.…”

“…In particular, Ortega et al. () developed and validated parallel versions of the same test in four different languages (English, German, Japanese, and Spanish), and three more parallel versions have been created and validated since then (Chinese: Wu & Ortega, ; French: Tracy–Ventura et al., ; Korean: Kim, Tracy–Ventura, & Jung, ) to promote generalizability and comparability of test scores across more languages and studies. Previous studies have all reported high external and internal validity for these parallel versions, suggesting that EITs may be a reliable tool to measure global (oral) proficiency, or basic language cognition in Hulstijn's term (Hulstijn, , , referring to the language ability that is restricted to the processing of oral language), for research purposes.…”