L2 phonology learning is constrained by a critical period (Lenneberg, 1967) or a sensitive period (Bornstein, 1989). Flege (1999) proposes that foreign accent in L2 emerges at about the age of five to seven years, and at this age efficiency in perceiving and producing L1 sounds increases. Brown (2000) investigated various English contrasts and found that Japanese and Korean speakers were significantly worse than the native controls in discriminating certain contrasts since L1 does not contain the relevant features. The present study compares a developmental stage of Korean adults and children in L2 sound acquisition and examines the influence of L1 segments on the acquisition of various non-native contrasts through discrimination, identification, and production tests.

The English /s-q /, /l-r/, /d-_/, and /ĵ-z/ contrasts which are not contrastive in Korean were chosen since these pairs may cause a differing degree of difficulty for Korean speakers. The native contrast /p-t/ was also included to check that errors were not caused by the difficulty of the task. The English voiceless fricative /s/ corresponds to Korean phoneme /s/ or /s’/. Since Korean does not have voiced obstruents, and /t/ and /č/ become /d/ and /ĵ/ respectively between vowels, the English voiced obstruents /d/ and /ĵ/ correspond closely to Korean phonemes /t/ and /č/. The interdental fricatives /q /and /_/ are completely absent in Korean, and the possible correspondences are /s’/ and /d/ when we refer to the loanwords. The trill /r/ is an allophone of the Korean phoneme /l/ although the Korean trill /r/ is not exactly same as English /r/. The /z/ sound is almost absent in Korean, and it is possibly mapped onto Korean /č/.

The following are hypotheses: (1) There will be no contrast in which Korean adults will perform significantly better than Korean children, while the children will perform better than adults on some contrasts. (2) While adults will show strong L1 influence, children may show the influence of child language phonology. For example, adults may have most difficulty in perceiving and producing the contrast which is free variations in Korean, /ĵ - z/, and have some difficulty in allophonic variants in complementary distribution, /l — r/, but they may not show any difficulty in /s-q / and /d-_/ contrasts since they may have acquired the completely new sounds /q / and /_/. On the other hand, children may have the most difficulty in acquiring the contrasts which include marked and rare sounds in many other languages, /s-q / or /d-_/. (3) If Korean adults and children show difference between the contrasts, substitution patterns in production will also be different between the two groups. While adults will substitute similar L1 sound for English sounds which are not acquired yet, children will substitute less marked sounds for more marked sounds.

The results show that whereas Korean adults did not perform significantly better than Korean children on any contrast in any tests, the children performed significantly better than the adults on the /ĵ—z/ contrast in discrimination, identification, and production tests. Conversely, Korean children did not show any difficulty in discriminating and identifying the /ĵ—z/ contrast even though they produced these sounds worse than the control sounds. In the discrimination test, the children performed poorly on the /s-q / and /d-_/ contrasts. The fact that the children showed much better performance on these /ĵ — z/ contrasts than the adults tells us that the children are not influence by L1 as much as the adults are. In addition, the adults also discriminated the /d—_/ contrast poorly, and had difficulty producing the /d/ sound accurately. On the other hand, Korean adults performed well in the perception of the /l — r/ contrast, allophonic variants in Korean. It is proposed that Koreans may use the difference of sound length to distinguish between these two sounds. Since in Korean the /l/ sound is very often followed by another /l/ sound, Koreans perceive the /l/ sound as longer than the /r/ sound. However, even though adult learners perceive the /l/ and /r/ sounds accurately, the production data reveal that they still have difficulty in producing those sounds.

The analyses of substitution patterns in production show the difference between the two groups. While the adults show a limited range of substitution patterns, children show a variety of substitution patterns. Half of the adults’ production of /z/ was evaluated as /ĵ/ by native English speakers, and the adults made hypercorrection mistakes by substituting /q / and /_/ for /s/ and /d/ respectively. The most common errors made by the children are the substitutions of /s/ and /d/ for /q / and /_/ respectively.

The results obtained in this study suggest that L2 adult learners are still capable of forming phonetic categories for certain L2 consonants. The errors made by L2 adult learners do not result from their having passed a critical period. However, it is also shown that a certain L2 sound is extremely difficult to master no matter how experienced L2 adult learners are. Even though due to the small number of subjects some statistical difference is on the border line, this pilot study clearly shows the different pattern of acquiring L2 between adults and children. In addition, L1 sound system differently affects each L2 contrast and sound.