Hyuna Kim and Dr. Laura Smith, Department of German and Slavic Languages
Numerous articles have examined the difficulties Korean speakers have learning the English /l/ and /r/ sounds (e.g., Borden et al 1983; Smith, 2001; Han, 2002). Most Korean learners of English hear and produce no distinction between English /l/ and /r/, because no surface contrast exists in Korean between these two liquid sounds (Iverson & Sohn, 1994). The English retroflex /r/ does not exist in Korean; Korean only has a phoneme /l/ with three distinct allophones: an apical flap [ɾ] in the initial position (as in atom in English), a lateral [l] in the coda position, and a geminate [ll] in the intervocalic position. Thus, in Korean, /l/ and /r/ are in complementary distribution and may be viewed as possible variants of one liquid sound. German is another language in which Koreans find it difficult to contrast the two liquid sounds for the same reasons as in English (Song & Lee, 1995; Park, 2011). Like English, German also has a two-way contrast between /l/ and /r/ sounds. However, the actual pronunciation of German /l/ and /r/ is different from English. For instance, the articulation of German /r/ is typically a uvular fricative, produced by several contact of the uvula and the tongue (Hall, 2003).
In this study I examined how the similarities and differences between English and German /l/-/r/ sounds affect Korean speakers’ production and perception of these sounds. Three research questions were raised: 1) Do Korean learners have similar problems with German that they had with English? 2) Do more advanced speakers perceive and produce the German /l-r/ contrasts better than less advanced speakers? 3) When Koreans have difficulties with their German /l-r/ contrast, what sounds are they producing? Does their second language (L2), English, impact their German? To answer to these questions, 27 Korean subjects learning both English and German at the Hankuk University of Foreign Studies (Yongin, South Korea) participated in the study. All subjects but one had learned English as their second language (L3), and German as their third language (L3). The subjects were divided into two groups based on the time they spent in a German-speaking country: Advanced group spent more than 6 months in a German-speaking country, whereas Beginning group less than 6 months. Both groups completed a production and perception task in addition to filling out a short biographical survey.
For the production task, participants were recorded producing the stimuli consisted of 30 pairs of /l/ and /r/ words in German as well as “l” and “r” in an additional to 96 German sentences not explicitly focusing on /l/-/r/, and one short paragraph of “The Sun and the Wind” in both English and German. The stimuli consisted of /l/ and /r/ sounds in initial position (Land vs. Rand), medial (Balsam vs. Barsam), and final position (Ball vs. Barr). The /l-r/ tokens were analyzed by native speaker rating, which included a native speaker of Korean, English, and German who were all fluent in English and German language. Scorings were determined based on a 3-point Likert scale: 0 = not at all the right sound; the right sound but not a good exemplar; 2 = reaching the native sound of German; and 3 = a clear exemplar of /l/ or /r/. For the perception tasks, participants were to complete two tasks, namely Identification task (ID) and Discrimination task (AX). For the ID task, subjects were asked to identify the word they heard from minimal pair choices on the screen, e.g., Land vs. Rand. For the AX task, subjects heard a pair of words and were asked if they heard the same word or different words.
For the production task, a series of Mixed Model Analysis of Variance were run to investigate the differences between sounds, word positions, and level of subject, well as interactions between these variables. According to the result, Korean learners of German produced the English and German /l/-/r/ fairly consistently, in which scores for all sounds in all positions averaged “2” (the speakers almost reached the native sound of German or English). Also, as expected, the /l/ sound was produced better than /r/ in both English and German because the Korean language has the phoneme /l/ but not /r/. Last, Advanced group constantly received higher ratings of production between /l-r/ contrast in English than Beginning group. However, in German, being in an Advanced group did not appear to improve the scores significantly over Beginning group. Although there was no significant difference between advanced and beginning learners for the German /l-r/ productions, a closer phonetic analysis of their productions reveal that advanced learners are more accurate in the choice of phone they produce and rely less on transfer from Korean and English. This corresponds to Williams and Hammarberg’s study (1998), in which L2 has a heavier influence in early stages of L3 learning, especially when L2 (English) factors are more similar to L3 (German), than L1 (Korean) to L3 (German).
For the perception task, results of the analysis indicated that Advanced group better in ID task, but performed 22% worse than Beginning group in the AX task. There was also an interaction between the word pairings and level. The actual words and word position, however, were not significant predictors for discrimination. While the beginners did seemingly better on the questions in which they responded “same” in the AX task, the more advanced learners were better at distinguishing between the different pairs, i.e., the /l-r/ pairs. It may be that they in being better able to hear the contrast, they are overextending it to even when they are hearing the “same pairings”. Koreans learning languages with a 2-way liquid contrast can come to perceive the contrast above chance levels and can produce the contrast fairly reliably with increased experience. Contrary to many studies, more experience with a language can help with pronunciation although not necessarily perception.
Limitations apply to this study. There was a difference between group assignments between the perceptual and production data because not many students had experienced living in a German-speaking country. Moreover, a few subjects who completed the production tasks did not complete the perceptual tasks. Last, the number of English /l-r/ tokens were smaller than the German tokens. For the future research more tokens of English /l/-/r/ should be included to in order to examine L2 and L3 interference in more depth. So far I gave two conference presentations about this study with my mentor Dr. Smith at GLAC (Germanic Linguistics Annual Conference in Bloomington, IN, April 2012) and SLRF (Second Language Research Forum, October 2012 in Pittsburgh, PA). For our forthcoming conference at New Sounds 2013 (Montreal, Canada), we will examine whether Korean leaner’s English proficiency plays a role in perceiving or producing German /l-r/.