Section 3.5

HUMAN MORPHOLOGICAL PROCESSING

(very oversimplified)

Are all the morphological forms in the lexicon?

walks, walked, walking, etc.

Turkish (agglutinative language) has 200,000,000,000 possible words

The opposite hypothesis (minimum redundancy) would say only 'walk' is
in the lexicon.

Truth is probably somewhere in-between.

("truth" = good model)

One hypothesis:

- inflectional morphology represented by rules

- some kinds of derivational morphology represented by rules

- other kinds of derivational morphology represented by lexicon

Makes sense when you remember that derivational morphology is only
semi-productive.

How to test: priming experiments

priming = word is recognized faster if you've seen it before (= it is
primed)

one morphological form primes another

but a derivational form doesn't prime another

(Stanners et al, 1979)

Marslen-Wilson: derivational forms can prime their stem if the
meaning is related

cf. govern, government (related meaning)

vs. depart, department (unrelated meaning)