Learning Stress Cues for Word Segmentation by Korean Learners of English

Abstract

It is known that English native speakers recognize the onset of a word at the stressed syllable (Metrical Segmentation Strategy) but in a large volume of studies, Korean learners of English are not sensitive to stress, and thereby they can't the use stress cue for segmentation effectively. The current study investigated whether Korean learners of English can learn to segment a stream of speech by recognizing the stressed syllable as the onset of a word based on statistical regularity, especially transitional probability (TP). The experiments were conducted for 10 - 13 aged learners of English living in Seoul and divided into two parts: Experiment 1 for learning effects depending on stress patterns and the ways of learning, Experiment 2 for the stress pattern preference of Korean learners of English. In Experiment 1, stimuli were designed with 3 sets of 18 pseudo-words composed of 3 syllables (CVCVCV) and conducted pre- and post- tests. Pre-test was proceeded on TP-only stimuli and the word-spotting test. In the post-test, learning sections were included depending on the stress patterns (word-initial or word-final) and the ways of learning (listening or repeating aloud) and then the participants took the word-spotting test. In Experiment 2, all the participants took the preference test in which they chose what they recognized as a word after listening to 18 chains of a repeated pseudo-word stressed at the initial syllable. As a result of Experiment 1, in the case of being given TP-only cue, Korean learners of English could segment the speech by using TP but this was interfered with when they learned word-initial stressed words by listening. However, when they learned word-initial stressed words by repeating aloud, the accuracy increased and this effect was similar to the case that they learned word-final stressed words by listening. Experiment 2 was investigated whether Korean learners of English showed any preference for stress pattern at word-level. The result showed that approximately 89.7% of the participants tended to recognize a word by word-final stress. Taken together, the current study implies that producing experience can facilitate the accurate perception and segmentation in a new language, which can be explained by Motor Theory. Especially, this result suggests the possibility that segmentation at the word-initially stressed syllable could be learned through producing experience by L2 learners. Furthermore, Korean learners might apply the L1 prosody (L-H tone pattern) to a new language prosodic pattern, which could be interpreted by PAM model or word-final lengthening effects. A number of studies have identified that Korean learners have trouble using the stress cue for segmentation since there is not word level stress in Korean. The results of the current study imply that producing experience is a necessary process not only for speaking but also for perception of the speech. In particular, in order to learn the word level stress cue which does not exist in Korean, the experiences of the direct articulation and the memory of the feature in producing play an important role for the accurate segmentation and perception. This might be a useful way of improving listening skills in the English as a Foreign Language environment such as Korea where listening input is exceedingly limited.