Whole Word Morphologizer
Expanding the Word-Based Lexicon: A non-stochastic computational Approach

If we take the enrichment of lexica to be not only the raison d'être of morphology but also the central issue of morphological theory, it seems reasonable to evaluate a theory of morphology, not on the means by which it represents recurring partials or lexical relations, but on its ability to generate new words based on a given lexicon. Within the framework of Whole Word Morphology (cf. Ford and Singh 1991, Ford et al 1997), I designed a small computer program that identifies morphological relations found in a lexicon and creates new words based on these relations.

Under the assumption that the morphology of a language resides exclusively in differences that are exploited in more than one pair of words within its lexicon (c.f. Neuvel and Singh in press), WWM compares every word of a small lexicon (1000 to 5000 labeled phonemic or orthographic forms) and calculates the segmental differences found between them. Some of these differences occur more than once and are translated into bi-directional word-based morphological strategies that can be represented as:

(1) /X/a <––> /X'/b
where:
a. /X/a and /X'/b are words and X and X' are abbreviations of the forms of classes of words belonging to categories a and b (with which specific words belonging to the right category can be unified or onto which they can be mapped).
b. ' represents (all the ) form-related differences between /X/ and /X'/
c. a and b are categories that may be represented as feature-bundles
d. the <––> represents a bi-directional implication ( if X, then X' and if X', then X )
e. X' is a semantic function of X (c.f. Ford et al 1997)

Each word in the lexicon is then mapped onto as many strategies as possible and contrasting new words are added to the lexicon.