|
Abstract:
Traditionally, the lexicon and syntax are regarded as
separate domains. Word-internal processes of morphological
combination, particularly for derivational morphology, are largely
ignored in sentence-level research, where whole words form the
input to the parser. This applies even when lexically-specified
information drives processing. As our recent research stresses the
morphological basis of lexical representation, we began to
re-examine the relationship between processes of morphological and
syntactic combination in three self-paced reading experiments.
First, we examine how derivationally complex material is
processed in sentential environments. Experiment 1 investigates
contrasts in morphological complexity (derived/HAPPINESS vs
simple/DELIGHT) co-varied with frequency (target word
predictability was always pretested). There were significant
distinct effects of complexity and frequency on the first
post-target word (T+1), but effects only of frequency at word (T+2)
(effects are often delayed in self-paced reading).
Experiment 2 looks more directly at possible interdependencies
between morphological and syntactic processes using sentences
containing novel or existing morphological combinations under
different degrees of syntactic and pragmatic constraint. Readers
must engage processes of morphological combination with novel
forms, because they have no pre-existing whole-word lexical
representation.
Examples:
Strong constraints:
Mary had cared for her sick mother for a long time.
She watched her with a quiet and
NURSELY(novel)/SAINTLY(established) devotion that was most
amazing;
Weak constraints:
Mary had lived with her mother for a long time.
She watched her with a NURSELY/SAINTLY devotion that was most
amazing.
While there was a strong effect of novelty, degree of constraint
had parallel effects for both novel and existing forms. Both
information from morphemes in novel combinations and from
established words can interact with context during sentence
comprehension. This provides further evidence for a
morphemically-based lexicon, but leaves open whether morphemes or
whole words are the fundamental unit of syntactic processing.
Experiment 3 examined complexity and frequency in inflectional
morphology, which critically involves the syntactic domain. We
contrasted regular (WATCHED) and irregular (SAW) past-tense forms,
co-varied with frequency. Regular past tenses transparently
decompose into stem+affix. Irregulars do not. Separate effects of
complexity and frequency, similar to Experiment 1, were found.
Complexity slows processing in apparently similar ways for both
intra-lexical combination and morphological processes involving
syntax (Experiments 1 and 3). Interactions between either novel
morphological combinations or established words and their
sentential environment are essentially the same (Experiment 2).
These results support the view that processes of intra-lexical
morphological combination and extra-lexical syntactic combination
may be less distinct than standardly assumed.
|
