Tone sensitivity & the Identification of Consonant Laryngeal Features by KFL learners 15 th AATK Annual Conference Hye-Sook Lee -Presented by Hi-Sun Kim- University of Chicago
Three-way laryngeal contrasts in Korean Challenge for KFL learners Kee-Ho Kim et al. (2006): lenis stops hardest for English learners of Korean to identify and tense stops least hard to identify. Importance of Fundamental Frequency in the three-way distinction S.-A. Jun (2000): a phrase-initial tone depends on the laryngeal type of the initial consonant Aspirated and tense obstruents High tone Plain(lenis) obstruents Low tone 달이요 – 딸이요 – 탈이요 L H H
Three-way laryngeal contrasts in Korean Importance of Fundamental Frequency (F0) in the three-way distinction Silva (2006): phrase-initial plain and aspirated stops are neutralized in terms of VOT and the phrase-initial tone comes into play for the distinction of the two. Mi-Ryoung Kim (2001): F0 plays an important role in the identification of plain stops, whereas other acoustic cues as well as F0 are important for the perception of aspirated stops.
Objectives of this study How does the sensitivity to phrase-initial tone affect the laryngeal feature identification of KFL learners? Heritage vs. non-heritage students First-year vs. second-year students Can an intonation training improve the laryngeal feature identification of KFL learners?
Experiment: Stimuli Recorded two native Korean speakers 4 minimal triplets: p-p’-p h, t-t’-t h, c-c’-c h, k-k’-k h 24 original stimuli (12 words x 2 speakers; no change) 24 modified stimuli (initial tone change) Initial tone modification Aspirated and tense: original High tone F0 lowered 탈이요 : originalmodified 딸이요 : original modified Plain: original Low tone F0 raised 달이요 : original modified
Experiment: Stimuli Initial tone modification F0 was modified in Praat, with all the other acoustic cues unchanged. “ 바라요 ” Original F0 (pitch) modified F0 (pitch)
Experiment: procedures Subjects 17 first-year students: 9 heritage and 8 non-heritage students 3 second-year students 7 native speakers Subjects listened to each token and chose one answer among three choices, plain-tense-aspirated. E.g. stimulus: 달이요 (either original or modified) choices: 1. 달이요 2. 딸이요 3. 탈이요
Experiment: Intonation training The experiment was conducted twice, and all the students had a weekly intonation training in between. Weekly intonation training session 15 minute session every week for 6 weeks Explicit instruction on the tone-consonant correlation Assistant helped students look at a native speaker’s and their own intonation contours and practice the native intonation pattern Pre-test and Post-test Pre-test: experiment before the training Post-test: experiment after the training
Experiment: Matching ratio Each stimulus was repeated 5 times in the test set. One subject has 5 responses to an identical token. Matching ratio (per stimulus and subject) Actual number of matching responses/5 달이요 (original) stimulus 4 달이요 & 1 탈이요 Matching ratio: 4/5 = 80% 달이요 (modified) stimulus 2 달이요 & 3 탈이요 Matching ratio: 2/5 = 40%
Results: Native speakers Mean matching ratios Initial tone change significantly changes the identification of the aspirated and plain series. Initial tone change does not affect the identification of the tense series.
Results: Native speakers Confusion ratios for the modified stimuli When the initial tone reversed, the aspirated series are most confused with the plain series, and vice versa. Response type Stimulus type
Results: First-year heritage students Mean matching ratios (post-test only) Similar patterns as native speakers’: aspirated and plain series are sensitive to the tone change (p<.001), but tense series are not. The effect of the tone change is not as sharp.
Results: First-year heritage students Confusion ratios for the modified stimuli When the initial tone reversed, the aspirated series are most confused with the plain series, and vice versa, similarly to the native speakers’.
Results: First-year non-heritage students Mean matching ratios (post-test only) Matching ratios of the original stimuli are very low. Not much sensitivity to the initial tone change (p>.05).
Results: First-year non-heritage students Confusion ratios for the modified stimuli First-year non-heritage students respond similarly to aspirated and plain stimuli: aspirated responses about 50% of the time and plain responses about 30%. Another evidence for their non-sensitivity to tone.
Results: Second-year non-heritage students Mean matching ratios (post-test only) Matching ratios of the original stimuli are low, but it is clear that the identification of all three laryngeal series is sensitive to the tone change (p<.001).
Results: Second-year non-heritage students Confusion ratios for the modified stimuli Similarly to the native speakers’, if the initial tone reversed, aspirated series are most confused with plain, and vice versa. Unique is that this group shows some tone sensitivity in the tense series as well; they learned not to choose tense responses if they have wrong initial tone.
Results: Effect of intonation training No effect with first-year students Both heritage and non-heritage did not improve the laryngeal identification nor tone sensitivity at the post-test (p>.05). Significant effect with second-year students Correct identification ratios improved significantly at the post-test (p<.005). Tone sensitivity also improved significantly at the post- test (p<.001).
Discussion: Tone sensitivity Native speakers have a clear alternation between aspirated and plain, according to the initial tone. It is clear that tone sensitivity is important for the correct identification of the three-way laryngeal contrasts in Korean. Tone sensitivity varied among the KFL learners, depending on their prior knowledge or exposure to Korean: heritage vs. non-heritage and first-year vs. second year.
Discussion: Effect of intonation training Significant positive effect was found only in the second-year students. Why didn’t the training improve the first-year students’ performance? The training might be too short (only 6 weeks) The training took place in their very first Korean course, so it could be harder for them to acquire both segmental (e.g. VOT) and suprasegmental cues (tone) simultaneously for the laryngeal distinctions:. Still implies that it will be more effective to provide explicit instruction on the consonant-tone correlations.
Conclusions This study confirms the importance of tone in the phrase-initial position when it comes to the laryngeal feature identification. This study suggests that prosody, or intonation, be incorporated in teaching the three-way laryngeal contrasts in the KFL curriculum.
References D. Silva Acoustic evidence for the emergence of tonal contrast in contemporary Korean. Phonology 23, H.-S. H. Kim Processing Strategies and Transfer of Heritage and Non-heritage Learners of Korean. Ph.D. Dissertation. University of Hawai‘i. K.-H. Kim, Y. Park, Y. Chun The Production and Perception of the Korean Stops by English Learners. Speech Sciences, 13 (4), M.R. Kim Consonantal and Vocalic Effects in Korean Stop Identification. Speech Sciences, 8 (1), S.-A. Jun K-ToBI (Korean ToBI ) labelling conventions: Version 3. Speech Sciences, 7,
Appendix: test words PlainTenseAspirated Labial 바라요 빨아요 팔아요 palayop’alayop h alayo Alveolar 달이요 딸이요 탈이요 taliyot’aliyot h aliyo Post-alveolar 자라고 짜라고 차라고 calakoc’alakoc h alako Velar 갈이요 깔이요 칼이요 kaliyok’aliyok h aliyo