| |
Abstract:
Word recognition models differ in their predictions for how
words with pseudo-prefixes, e.g., _deliver_ and _injury_, are
recognized. However, the set of words assumed to consist of pseudo-
prefixes often includes words that linguists argue are complex
(Wurm, 1997). The present study investigates the processing of
words such as _impose_ and demonstrates that they are not a
homogeneous set. Two segment shifting studies (Feldman &
Fowler, 1987) are presented which suggest that words containing a
semantically transparent prefix are processed differently from
words that contain a semantically opaque prefix. These results
suggest that the latter are morphologically complex, e.g.,
consisting of a real prefix, while the former are not.
In the first study, participants were presented with a source
word from one of three conditions: a) items with a semantically
transparent prefix (REFLECT), b) items with a semantically opaque
prefix (REFUSE), and c) words that are clearly morphologically
simplex but have a pseudo-prefix (REMORSE). Words were considered
simplex if the pseudo-stem was semantically opaque and occurred in
no other affixed words (e.g., -fuse is the stem in _refuse_,
_infuse_, _defuse_ while -morse forms no other words besides
_remorse_) and prefix transparency ratings were computed from a
separate rating study. After a brief presentation of the source
word, the (pseudo-)prefix was highlighted simultaneous with the
presentation of a target word (MOVE). Participants were instructed
to shift the highlighted portion of the source word onto the target
word and pronounce a response word (REMOVE).
We hypothesize that source words consisting of a real prefix
will affect the naming latencies to response words (REMOVE)
relative to source words consisting of a pseudo-prefix. Response
words were named significantly slower when the source word
consisted of a semantically transparent prefix relative to source
words with either a pseudo-prefix or a semantically opaque prefix.
Our results suggest that studies that unilaterally treat these
words as morphologically simplex may have a confound in the words
that they investigate.
To support this claim, we present another segment shifting study
that conflates the semantic transparency factor. Bound stem items
(RECEIVE) were compared to free stem items (REACT) and simplex
items (REMORSE). When semantically transparent and opaque prefixes
are combined in a single condition, the bimodal distribution is
averaged. Thus it will appear that bound stem items as a whole are
no different from simplex items. This is what we found. Production
of the target word was fastest when preceded by a free stem item
(REACT) but equally slow when preceded by bound stem items
(RECEIVE) and simplex items (REMORSE). However, a post-hoc analysis
revealed the bimodal distribution. As in Experiment 1, bound stem
items whose prefixes were transparent patterned differently from
bound stem items with opaque prefixes.
Our studies suggest that words that many word recognition
researchers treated as morphologically simplex are, in fact,
complex. The determining factor appears to be whether the initial
string is identifiable as a prefix based on its meaning.
Aronoff, M. 1976. Word formation in Generative Grammar,
Cambridge, MA: MIT Press.
Hockett C. 1954. Two models of grammatical description. Word 10:
210-233.
Feldman, L. B. & Fowler, C. A. 1987. Morphemic segments shift
faster than nonmorphemic controls. Paper presented to the
Psychonomic Society, Seattle, WA.
Feldman, L. B., Frost, R., & Pnini T. 1995. Decomposing Words
Into Their Constituent Morphemes: Evidence From English and Hebrew.
Journal of Experimental Psychology: Learning, Memory, and
Cognition. Vol 21, No 4, 947-960.
Marslen-Wilson, W., Tyler, L., Waksler, R., and Older, L.1994.
Morphology and Meaning in the English Mental Lexicon. Psychological
Review 101 (1) 3-33.
|
