Tuesday, March 8, 2011

Listening Experiment

Over a period of approximately one year, starting in the spring of 2010 and finishing in the winter of 2011, I conducted a listening experiment. I asked 68 Korean university students to listen to five word pairs and circle the word they heard. I read each word pair five times. To ensure that every student heard the word clearly, I then read the list a second time.

The word pairs which I used were light/right, sad/said, suffer/supper, pull/pool and it/eat. Since I read each word pair five times and collected data from 68 students, I had a total of 340 responses for each word pair.

I analyzed the data and compared the results. The word pair which gave my students the least difficulty was light/right. The one which gave them the most difficulty was sad/said. Here I present the results:

338/340, 99%

303/340 89%

292/340, 86%

276/340, 81%

198/340, 58%

The word pair light/right was distinguished correctly in almost every instance. Though the liquids are pronounced differently in Korean, they nevertheless exist. The alveolar flap occurs syllable-initially and intervocalically and the alveolar lateral occurs syllable-finally. They are allophones rather than phonemes. The presence of the two liquids in Korean may explain the percentage of correct responses with this word pair.

However, the word pair sad/said was difficult for many students to distinguish. The vowel of "sad" does not occur in Korean. It is in fact a marked vowel which does not occur in many languages. Also, many English loan words in Korean with the vowel of "sad" are transcribed with the vowel of "said." As a result, many students may perceive the low front vowel as the mid front instead.

The word pair suffer/supper was more easily distinguished than sad/said and it/eat. The percentage of correct responses was 86%. Korean has the voiceless bilabial plosive but does not have the voiceless bilabial fricative. For this reason, Koreans often substitute the plosive for the fricative in speech.

The word pair pull/pool was better distinguished than any word pair other than light/right. The number of correct responses was 89%. Though Korean lacks the high back lax rounded vowel of "pull," most students distinguished the tense and lax vowels correctly. This may indicate that the qualitative difference between the two vowels is perceptually salient for many Koreans.

The word pair it/eat was not as well distinguished as pull/pool. In fact, it was the most difficult word pair to distinguish after sad/said. Though this word pair exhibits the same tense/lax distinction of pull/pool, the results were far from identical. Korean lacks the lax vowels of both "it" and "pull." However, Korean has a high back unrounded vowel which many Korean students may substitute for the lax vowel of "pull." With the vowel of "it," however, no such substitution occurs. It appears that the distinction between pull/pool is more perceptually salient than that between it/eat. It may also be the case that many Koreans are not aware that it/eat have different vowels unless it is made clear to them.

My listening experiment led to a few surprises. I did not expect that the word pair light/right would be distinguished correctly in 99% of instances. This may indicate that an allophonic difference in one's native language is sufficient for successful discrimination of speech sounds in a second language. Another surprise was the difference in correct responses with respect to it/eat and pull/pool. Though both word pairs have the same tense/lax distinction, the percentages were quite different. With pull/pool, the percentage of correct responses was 89% but with eat/it only 81%. Another surprise was with the word pair sad/said. Though Korean lacks the low front vowel of "sad," I expected a higher percentage of correct responses than 58%. This word pair was much harder for students to distinguish than any other. As a result, this may indicate that the low front vowel is very difficult to distinguish for those speakers who lack the vowel in their own native language.

No comments: