In this thesis, I have investigated a number of phonological phenomena in Korean,
such as [Û]/zero alternation, neutralisation, nasalisation, post-nasal voicing,
tensification and the distribution of [Û] in loanwords, within the framework of GP. It
was noted that the distribution of the vowel [Û], unlike other vowels, is highly
constrained and predictable. This leads to the idea that empty nuclei are postulated
underlyingly when [Û] appears and between consonant clusters. Final empty nuclei
occur in consonant-final words, due to the effect of the Coda Licensing Principle
and the Onset Licensing Principle. Thus, multiple empty nuclei can be present in a
given lexical representation.
I have showed that the phonetic interpretation of empty nuclei in various
contexts provides a adequate account of relevant phonological activities.
Neutralisation is analysed as a licensing constraint on final empty nuclei. Only [p, t,
k, m, n, l] can occur and other segments undergo elemental adjustments. This is
due to the weaker licensing ability of licensed empty nuclei, in comparison with that
of unlicensed ones. With respect to [Û] in morpheme-internal position and in nonanalytic
suffixation, the distribution of [Û] is determined by the interaction of the
ECP (inter-onset government, proper government and the domain-final licensing
parameter), the PP and the CLE. In particular, head-final inter-onset government
together with the notion of government-licensing account for the majority of cases
where [Û] appears. In analytic suffixation, due to the fact that the effect of the SCC
preserves the licensed status of a final empty nucleus in an extended domain, we can
establish an inter-onset governing relationship between a stem-final and a suffixinitial
consonant. Unless the requirements of inter-onset government are met,
various phonological processes are triggered to adjust the segmental shape: postnasal
voicing, tensification and nasalisation. Accordingly, we can treat these
phenomena in a unified way.
Regarding the distribution of [Û] in loanwords, it is pointed out that the loanword
phonology is not completely different from the native phonology. Internal [Û] in
loanwords is distributed in the same way as that of native words and accordingly
analysed in the same way. The presence of final [Û] in loanwords is analysed as a
result of the precedence of the PP over the ECP, since the segmental identity of
segments in the source language is maximally preserved. The loanwords ending in
final [Û] are classified as belonging to the genuine loanword category and those
without as pseudo-native words. The postulation of the two categories provides an
appropriate criterion for active phonological participation of the latter category in
the native phonology.
In comparison with previous approaches, the contributions that I have made are
as follows:
• The reformulation of how internal empty nuclei are licensed, i.e. proper and interonset
government are independently active.
• The revision of the governing hierarchy to provide a more adequate account of
problematic cases in previous GP approaches.
• The proposals of the Nasal Condition and the Condition on NC Clusters for the
asymmetrical distribution of the vowel [Û] between mono-morphemic words and in
suffixation in the former, and that of CN and NC clusters in the latter.
• The refinement of domainhood in GP and the postulation of stem domains and
extended-stem domains to account for the effect of opacity in suffixation, regarding
the preservation of the vowel [Û] in a given stem domain.
• A more appropriate account of the fact that internal [Û] in native and loanwords are
distributed in an identical way.
• The classification of obligatory and optional realisation of final [Û] in loanwords,
which leads to postulate the category of pseudo-native words and genuine loanwords.
