된소리의 발달

Abstract

The tense obstruents dealt with in this paper are /pˈ, tˈ, kˈ, sˈ/ which developed before the end of the 15th century and made the cosonant system virtually the same as that of present-day Korean. When the Korean alphabet was invented in 1443, /sˈ/ already existed. The most plausible hypothesis for the development of /sˈ/ is that it has developed as initial geminate /ss/ from #sVsV due to syncope of the first vowel. It is supposed that /pˈ, tˈ, kˈ/ have developed as geminates /pp, tt, kk/ from the clusters /sp, st, sk/, respectively, initial ones during the second half of the 15th century, and medial ones towards the end of the 15th century, which dates are estimated in this paper. The most important aim of the paper is to explain the change of the clusters /sp, st, sk/ into /pp, tt, kk/. Let us begin with the medial ones. The syllable-final /s/ changed into /t/ in the early 16th century, so the series of intervocalic changes /sp, st, sk/＞/tp, tt, tk/＞/pp, tt, kk/ appear to be beyond question. The first is the syllable-final change /s/＞/t/ and the second is place assimilation. For example, /əskəj/＞/ətkəj/＞/əkkəj/ ‘shoulder' looks as if perfect. However, the syllable-final change /s/＞/t/ raises a big question, since, as a one-step change, [s]＞[t ̚] is not understandable. So I have set up [st ̚] as an intermediate stage in a previous paper. Thus the syllable-final change in question, I think, is [s]＞[st ̚]＞[t ̚]. The [st ̚] is not a cluster but a complex segment that begins as a fricative and ends as an unreleased stop. Now I set up [st ̚] also for the change of the medial clusters /sp, st, sk/ into /pp, tt, kk/. This change can be represented as (1), and the syllable-final change /s/＞/t/ as (2). (1) a. [sp]＞[st ̚p]＞[spp]＞[pp] b. [st]＞[st ̚t]＞[tt] c. [sk]＞[st ̚k]＞[skk]＞[kk] (2) a. [sp]＞[st ̚p]＞[tp] b. [st]＞[st ̚t]＞[tt] c. [sk]＞[st ̚k]＞[tk] d. [s]＞[st ̚]＞[t ̚] (before a pause) Roughly speaking, (1) represents the case of intra-morphemic clusters and (2) the case of inter-morphemic ones. In (1) the stop part of [st ̚] underwent place assimilation, but in (2) the assimilation was blocked by morpheme boundaries or there was no trigger. It is naturally supposed that (1) and (2) were initiated together but (2) was completed later than (1) because in the case of (2) deletion of [s] was delayed due to the influence of [s] of pre-vocalic allomorphs. Needless to say, this order cannot be compatible with the above-mentioned series of changes /sp, st, sk/＞/tp, tt, tk/＞/pp, tt, kk/. There is supporting evidence. That is, spellings that are in accord with the above supposition are found in documents of the early 16th century, which fact is presented for the first time in this paper. Now we can believe the reality of the complex segment [st ̚] set up by me. Thus there is no reason to think that the initial clusters /sp, st, sk/ have changed into /pp, tt, kk/ in a different way from (1).