S1 (FriAM 1-65) - The Psychonomic Society

Posters 4014–4020 Saturday Noon
• PROCESSING LETTER POSITIONS •
(4014)
Transposition Priming Effects Explained With Absolute Letter
Position Coding. WENDY DE MOOR, WOUTER DUYCK, & TOM
VERGUTS, University of Ghent (sponsored by Tom Verguts)—A
basic problem that must be addressed by any word recognition model
is how letter position information within a word is coded. The fact that
we can distinguish between, for instance, GARDEN and GRADEN has led
to the idea of absolute letter position coding—that is, one set of detectors
to code for the first letter, one set for the second letter, and so on
(e.g., McClelland & Rumelhart, 1981). On the other hand, transposition
priming studies have demonstrated priming from, for instance,
GRADEN to GARDEN, leading to word recognition models based on relative
rather than absolute letter position coding (e.g., Peressotti &
Grainger, 1999). In this study, we present a computational model that
demonstrates that transposition priming can also be explained with
absolute position coding if some uncertainty (fuzziness) is introduced
in the coding positions. We also present new empirical data on transposition
priming that support the computational model.
(4015)
The Quiet Clam Is Quite Calm: Transposed-Letter Neighborhood
Effects in Reading. REBECCA L. JOHNSON, Skidmore College—
Response time tasks have found inhibitory neighborhood effects for
word pairs that differ in a transposition of two adjacent letters (e.g.,
clam/calm). Here, two eyetracking experiments were conducted to explore
transposed-letter (TL) neighborhood effects within the context of
normal silent reading. In Experiment 1, sentences contained a target
word that either has a TL neighbor (e.g., angel) or does not (e.g., alien).
In Experiment 2, the context was manipulated to explore whether semantic
constraints attenuate neighborhood effects. Readers took longer
to process words that have a TL neighbor than control words, but only
when either member of the TL pair was likely. Furthermore, this interference
effect occurred late in processing and was not affected by relative
word frequency. These interference effects can be explained either
by the spreading of activation from the target word to its TL neighbor
or by the misidentification of target words for their TL neighbors.
(4016)
The Representation of Letter Position in Orthographic Representation.
SIMON J. FISCHER-BAUM, BRENDA C. RAPP, & MICHAEL
MCCLOSKEY, Johns Hopkins University (sponsored by Brenda C.
Rapp)—The orthographic representations used when reading and
spelling encode not only the letters that are in the word, but also their
positions. How is letter position represented? Is the G in DOG represented
as the third letter, the first letter to the right of a midline, the
last letter, or the coda of an orthographic syllable? We present evidence
from two individuals with acquired dysgraphia (subsequent to
a stroke) who perseverate single letters when spelling. That is, letters
from previous responses intrude into subsequent ones. Perseverated
letters appeared more often than would be expected by chance in the
same position of an error response and the source response. We contrasted
the predictions of different hypotheses regarding the representation
of letter position: syllabically defined position, left, right,
or center alignment and right+left edge alignment. The analyses provide
evidence for both syllabic and right+left edge alignment.
(4017)
Letter Position Encoding and Morphology. MANUEL PEREA, University
of València, JON A. DUÑABEITIA, University of La Laguna,
JOANA ACHA, University of València, & MANUEL CARREIRAS,
University of La Laguna (sponsored by Manuel Perea)—Duñabeitia,
Perea, and Carreiras (2007) found that masked transposed-letter priming
effects tended to vanish when the adjacent letter transposition
crossed a morpheme boundary (walekr–WALKER vs. walibr–WALKER).
They suggested that morphological decomposition operates at an
early stage of visual-word recognition. Recent evidence has also
107
shown that pseudoaffixed words (CORNER) can also be decomposed
early in word processing (Rastle et al., 2004). In the present lexical
decision experiments, we examined the magnitude of the transposedletter
priming effect for suffixed words, pseudosuffixed words, and
nonaffixed words. We did so for adjacent and nonadjacent letter transpositions
across the morpheme boundaries. The results showed that,
for adjacent transpositions, the transposed-letter priming effect was
greater for nonaffixed words than for affixed/pseudoaffixed words.
However, for nonadjacent transpositions, the transposed-letter priming
effect was unaffected by morphology. We examine the implications
of these findings for models of letter position encoding and morphological
processing.
(4018)
Fast and Furious False Recognitions: Jumbled Words and Semantic
Associations. KARA D. DESOUZA, SARA HABER, & WILLIAM P.
WALLACE, University of Nevada, Reno—Jumbled nonwords, in which
the first and last letter of a word remain the same but interior letters
are rearranged, have been shown to be easily readable (Grainger &
Whitney, 2004). This might be due to the nonword being recoded into
the original word (Perea & Lupker, 2003), a process which may suffer
with fast stimulus presentation rates. This experiment investigated
the jumbled word effect in conjunction with false recognitions of critical
items related to semantic associate lists. Four presentation speed
conditions (110 msec, 160 msec, 330 msec, 880 msec) were used to
determine whether false recognitions of jumbled nonwords can be reduced
through short presentation times. False recognition occurred for
neither words nor nonwords at the fastest rate, but did occur for both
at slower speeds. It appears that for nonwords, a significantly longer
presentation time is required than for words, in order for semantically
related false recognitions to occur.
(4019)
Deconstructing the Neighborhood: 1-Gram Effects and Orthographic
Neighborhoods. CHRIS F. WESTBURY, University of Alberta—A
word’s orthographic neighborhood (ON) is defined as the
number of words that differ from it by a single letter. We have examined
the effect on lexical decision RTs of the number and frequency
of words at different positions in a word, using a new frequency dictionary
drawn from a very large new corpus, new nonlinear correlation
tools, and both standard and novel techniques. One finding is that
ON-modulating frequency effects extend much lower in the frequency
range than is usually taken into account. We find strong positional
effects in ON. A few positional effects wash out the effect of
ON itself in linear and nonlinear regression analysis. We demonstrate
that the positional effects can be nonlinear, and much stronger than
linear analysis suggests. More complex and subtle measures of orthographic
typicality than ON may be helpful for understanding how
word typicality impinges on lexical access.
(4020)
The Influence of Sentence Context on Orthographic Neighborhood
Effects in Reading. BRIAN M. FRIEL, Delaware State University—
Two experiments—one involving a lexical decision task (LDT) and
the other employing a self-paced sentence reading task—investigated
the impact of sentence context on orthographic neighborhood effects
in word recognition. In both experiments, the same 24 target words
were used. Half of the targets had many orthographic neighbors (defined
by Coltheart’s N). The LDT results revealed that words with
many neighbors (large N words) yielded faster response times than
words with few neighbors (small N words). In the reading experiment,
the targets in each neighborhood size condition were embedded in
sentences. These sentences differed in terms of the degree to which
context constrained the plausibility of the target’s neighbors. Large N
targets were read faster than small N targets, and sentence constraint
facilitated target recognition. A comparison of reaction times for
items between tasks revealed that the sentence context manipulation
only affected recognition of large N targets.