Íslenskt mál og almenn málfræði - 01.01.1980, Page 17

GEORGE N. CLEMENTS
A Note on Local Ordering
o.
This note is concerned with a recent issue in the theory of rule or-
dering. In his book The Organization of Phonology (1974), S.R. Ander-
son presents a detailed account of what he terms the theory of local
ordering. As it is exemplified and justified in large part with reference
to Icelandic phonology (pp. 141-6, 185-9, 191-5, 200-2), the theory
of local ordering might be of particular interest to scholars concemed
with Icelandic. I will be concerned here with certain of its empirical
predictions that have so far received little attention.
1.
Under the theory of local ordering, rale ordering statements involve
Pairs of rules only. Grammars contain ordering statements of two
sorts. For some rules, it specifies that „Rule A precedes Rule B.“ For
others (the default case), it specifies that „Rule A and Rule B apply
m the natural order“ (interpreted in a first approximation as feeding
and counter-bleeding order). One immediate consequence of this theory
(and one which differentiates it from the linear ordering theory of
standard versions of generative phonology) is that two rules A and B
may apply in the order A,B in one derivation and in the order B,A in
another. (For example, Anderson proposes that Icelandic u-umlaut
must precede syncope in the derivation of jökli from underlying /jak +
ul + e/ but that it must follow it in the derivation of kötlum from
/katil + um/.) A further consequence is that there may be derivational
sequences in which a single rule applies both before and after another
^ule. (For example, Anderson proposes that n-umlaut both precedes
und follows vowel reduction in the derivation of fötnuðum from /fatn-
að + um/.) To the extent that they cannot plausibly be viewed as con-
sequences of cyclic rule application, such „ordering paradoxes" have
been taken as incompatible with the theory of linear ordering, and