Greg Stump: Paradigm templates, stem patterns and lexical representations
One apparent choice to be made in morphological theory is the choice between (a) defining a
language's morphology with RULES OF EXPONENCE, which deduce the word form realizing a given
cell in a lexeme's paradigm from the stem and morphosyntactic property set associated with that cell,
as in (i); and (b) defining a language's morphology with IMPLICATIVE RULES, which deduce the word
form realizing a given cell in a lexeme's paradigm from the word forms realizing one or more other
cells in that paradigm, as in (ii).
- The cell (Xa, {CASE:locative, NUM:singular}) is realized as Xe.
- If Xena is a lexeme's instrumental singular form, then Xe its locative singular form.
Definitions of types (a) and (b) suggest rather different approaches to lexical representation; in
particular, (a) suggests that a lexeme's lexical representation includes a specification of its stem, while
(b) suggests that its lexical representation consists of the word form(s) realizing one or more cells in
its paradigm.
I argue that (a) and (b) needn't be seen as alternatives: that if the defaults and overrides constituting
a language's system of inflection classes are defined in a suitably general way, then rules of
exponence and implicative rules coexist as theorems of this definition. I propose that general
definitions of this sort take the form of a default inheritance hierarchy whose nodes are inflection
classes, and that each inflection class is itself the pairing of a PARADIGM TEMPLATE with a pattern of
stem alternation. I illustrate with examples from the declensional morphology of Sanskrit.
Toc of the proceedings
Maintained by Stefan Müller
Created: October 16, 2010
Last modified: October 30, 2010
|
