0. Introduction

In terms of population, the three main dialects used in Seouth Korea are Seoul, Kyungsang, and Chonnam. Kyungsang is well studied and has been known as a tone langauge. In this project, I focus on Seoul and Chonnam dialect.

It has been an issue of debate for a long time whether vowel length is phonemic or not in Korean. While there are various opinions, it is generally agreed that the length distinction is being lost in modern Seoul, especially among younger generation speakers, but is still maintained in certain conservative dialects of Korean, such as Chonnam. Seoul used to distinguish vowel length until recently, and older Seoul speakers still distinguish a long vowel from a short one. So, the loss of vowel length distinction is seen as an on-going sociolinguistic/historical change that is currently taking place among younger generation Seoul speakers.

When I talk about 'Seoul' in this project, it refers to the one spoken among younger generation Seoul speakers including those in Kyonggi-do province surrounding Seoul. It is not clear how many speakers Seoul has since many of the residents in Seoul migrated from other areas. But it it no doubt the major dialect in Korea. 'Chonnam' refers to the regional dialect spoken in Chonnam (or Cholla-namdo) used by about five million speakers. It uses the same alphabet and vocabulary as Seoul. Although Chunnam share most of the Korean segmental phonological rules with Seoul, it is clearly distinguished from Seoul in terms of prosody as well as some vocabulary. In some sense, 'Chonnam' represents the conservative Seoul dialect as well, as far as the vowel length distinction goes.

In this project, I will show that a careful examination of prosodic phenomena relating stress/accent/tone in Korean challenges the commonly assumed idea (eg. Jun 1994) that Chonnam maintains underlying vowel length distinction while Seoul does not. Instead, I will come up with a new theory where vowel length is argued to be an epiphenomenon of stress, which is determined by lexical tonal specification (i.e. tone=>stress=>pitch/length). So the difference in the aspect of vowel length distinction in Seoul and Chonnam is rather due to the change of the stress rule in Seoul concommitant with the change in underlying tonal specification, rather than the vowel length distinction per se.

The nature of the distinct prosody in these two dialects has been investigated in Jun, S-A (1994), where she argues that the tonal events in Korean do not function to mark prominent syllables. She, therefore, classifies both dialects as pitch accent languages, which differ only in the inventory of the tonal shapes--LHLH for Seoul, and LHL and HHL for Chonnam--with some variation in assignment rules. As for the Tone Bearing Unit, she assumes that it is mora in Chonnam dialect, where vowel length is thought to be distinctive, but syllable in Seoul because it does not have a phonological vowel length distinction.

There have been parallel approach from a metrical perspective regarding the prosodic prominence of Korean. Lee, H-B (1987) has suggested that Korean has an iambic foot structure and came up with a set of stress assignment rules for Korean. While many researches on Korean metrical phonology have based their work on this assumption (eg. Kim, J-K 1997, Lee 1994, etc.), Jun, J-H (1995) hs alternatively argued for an unbounded foot for Korean, where the stress is assigned to the leftmost heavy syllable of a word. In any case, having a foot implies being a strss langauge, whose phonetic realization is most commonly reflected in pitch and duration.

Since Chonnam has both pitch accentual properties (i.e. single H per word) as well as stress (i.e. position of H restricted to the initial two syllables, and correlation of pitch and duration), it has been not easy to determine the category of its prosodic system either way. In this paper, I will show that Chonnam is a stress language, whose stress assignment is based on lexical tonal specification of the syllable, while Seoul is a pitch-accent language due to the gradual loss of lexical tonal specification. The tonal specification comes from Middle Korean which had a clear tonal disintction for sino Korean words borrowed from Middle Chinese. This is obvious from the fact that Hangul, the Korean alphabet system invented in mid 15th century, had diacritics for the tonal specification. In Modern Chonnam, although the diacritics are no longer used, and Chinese characters are not used as often as it used to be, the tonal specification of a syllable in Sino-Korean words can be identified by looking at the allophonic realization of the tones depending on the position of the syllable: If a syllable is lexically specified as H, it is realized as H, and long, when in word initial position. Otherwise, it can be regarded as lexically L. Modern Seoul does not have such allophonic variation any more, and all are realized as short in initial position.