ABSTRACT
Grammaticalization is commonly assumed to be a process of categorial re-analysis in
which a lexical descriptive element turns into a morpheme with a predominantly
functional role. Frequently this process would seem to convert a lexical head into a
member of the particular functional structure dominating that head, as for example when
verbs with clear descriptive content become re-analyzed as aspectual or modal verbs
occurring in functional heads projected over the VP. In this sense grammaticalization
may be taken to be the result of a combination of movement and re-analysis – movement
of a morpheme from a lexical head position to a higher functional head position and then
eventual re-analysis of the morpheme as being base-generated in the latter functional
head. Such a view of grammaticalization is proposed in Simpson (1998), Roberts &
Roussou (1999) and Wu (2000) and naturally procedes in a simple upwards or ‘vertical’
direction in a tree following the path of movement (e.g. lexical verbs frequently re-
analyze as instantiations of the higher modal-aspectual functional heads projected over
VP). In Simpson & Wu (1998) it is suggested that grammaticalization may also occur in
an essentially ‘horizontal’ direction and that a Chinese nominalizer of type D0 (de) is
currently undergoing re-analysis as a new instantiation of a clausal head (past tense/T0);
such a change does not result from any upwards movement but from the horizontal/
lateral re-analysis of a functional element in the nominal domain as a functional element
in the clausal domain. In this chapter I would like to suggest that this basic type of
horizontal re-analysis argued for in Simpson & Wu (1998) which re-categorizes
nominalizers as clausal functional heads is actually quite widespread as a phenomenon
in Chinese, Japanese and Korean and possibly significant as a general areal feature of
such languages. Due to differences in the surface typological properties of Chinese,
Japanese and Korean it will be shown that the hypothetical re-analysis process is
interestingly revealed by different types of evidence, and that there is also indication of
certain cross-linguistic variation in the way that the nominal elements become re-
analyzed in the clausal functional structure. As a result, the phenomenon is one which
both intriguingly unites Chinese, Japanese and Korean cutting across their typological
differences, and one which also clearly gives rise to certain parametric variation in its
actual realization.