Is Lexical Access Mediated by the Syllabic and/or CV
Structure of Words? Exploring Transposed-Letter
Priming Effects
Jerid Francom
University of Arizona
jeridf@u.arizona.edu
1 Observations
Aciocnrdg to a rcaerheesr at Cmargbdie Uiistenvry, it deos not meattr in waht oerdr the leterts in a
wrod are; the olny inmaptort tinhg is taht the frsit and lsat lteter be in the rghit pclae. This observation,
circulated in a popular email circa 2003, underlines the freedom that is afforded to letter position in visual
word recognition. Although it can be argued that other components of the language processor aid in this
type of example, i.e. syntax and semantics, there is clearly a range of freedom for the orthographic system
that is curiously large when considering the fact that the lexical processor must also be precise enough to
distinguish between the positions of ‘s’ and ‘u’ in the words casual and causal. How the lexical access
system is organized such that it can be flexible enough to make the correct lexical retrieval for FIRST when
given frsit and precise enough, when given casual, to recall CASUAL to the exclusion of causal is currently a
growing topic of debate.
The lexical phenomenon pointed out in the Cambridge email is known in the literature as Letter Trans-
position Similarity Effects. One of the most robust characteristics of the phenomenon, and highlighted in
this email, is the observation that a non-word prime in which the inner-letters have been transposed, such as
rghit is much more similar to the target RIGHT than a non-word in which the first and last letter have been
transposed, such as tighr (Perea and Lupker, 2003; Peressotti and Grainger, 1999).
However current research probing these effects suggest that letter-position is even more unconstrained.
Schoonabaert and Grainger (2004) show that letter-deletion also primes consistently, in pairs such as mircle-
MIRACLE, despite the fact that the deletion of the vowel ‘a’ changes word-length and removes lexical input.
In addition, Perea and Lupker (2004) point to evidence suggesting that non-adjacent transpositions show
strong priming effects as well, caniso-CASINO.
This evidence may in fact give the impression that letter-transposition effects show that lexical processing
is not positionally constrained at all. However, research also shows that there are limits to the ability of the
lexical system to freely encode letter position. As pointed out earlier, the word periphery positions appear
to be ‘special’ in visual word recognition. Hence the non-word nositiop does not prime POSITION, (Jordan
et al., 2003). Perea and Lupker (2003) provide evidence showing that transpositions are sensitive to word-
final transpositions, *judeg-JUDGE.1 In other findings, Inhoff et al. (2003) provides evidence that in fact not
only the first letter but first letters are particularly crucial in lexical access, *omuse-MOUSE.
Although evidence suggests that letter position may be highly flexible, even deleted from a stimulus
input and still show priming, the same flexibility is not extended to letter replacement (Perea and Lupker,
2003). Replacement of the letters ‘a’ and ‘o’ with ‘s’ and ‘r’, *bslcrn- BALCON, not only does not induce but
inhibits priming. In addition, letter transposition appears to be limited to some abstract sense of linear order
as findings from experiments conducted by Peressotti and Grainger (1999) show. Despite the orthographic
overlap no priming is found in pairs like *nlcb-BALCON where there is priming in pairs like blcn-BALCON.
2 Theoretical Background
These findings from Transposed-Letter Effects pose difficult questions for position-specific coding
schemes employed in many computational models of visual word recognition. Models such as the Inter-
active Activation Model (IA) (McClelland and Rumelhart, 1981), the Dual-Route Cascaded Model (DRC)
(Coltheart et al., 2001) and the Multiple Read-Out Model (MROM) (Grainger and Jacobs, 1996) use slot-
1An asterisk is used to indicate pairs in which no priming occurs.
89
Proceedings of the 24th NWLC, 3-4 May 2008, Seattle, WA

coding to read specific activation levels for letter position and identity. These models can adequately deal
with the findings that letter periphery position and letter replacement is more rigidly tied to successful access
of a lexical item. However, in light of empirical evidence that indicates that lexical access is also flexible
enough to allow letter-deletion and adjacent and non-adjacent inner-letter transposition that respects some
notion of relative position we are left without a satisfactory account of the lexical access system.
More recent models such as the Self-organizing Lexical Acquisition and Recognition (SOLAR) (Davis,
1999) and Sequential Encoding Regulated by Inputs to Oscillations within Letter units (SERIOL) (Whitney,
2001) use spatial encoding techniques in order to avoid the pitfalls of slot-encoding schemes. These models
are particularly adequate for dealing with the data unaccounted for in slot-coding models, namely letter-
transposition and deletion case mentioned here, as letter identity is not calculated in direct correspondence
with position, rather relative position and weighted activation allow for increased letter-position flexibility.
Under this type of system, word transposition is constrained computationally by emphasizing the descending
importance of letter position from left to right including a special slot for word final position.
Spatial models, then, seemingly provide the necessary equipment to both account for the flexibility and
precision noted to be active in visual word recognition. However, these models impose no mediating con-
straint on transpositions that occur among interior letters, from now on inner-letter transposition. Guerrera
(2004) demonstrates in a battery of experiments systematically testing the flexibility of transposed-letter ef-
fects that both the SOLAR and SERIOL models in fact over-generate in target/prime pairs such as *isedawkl,
SIDEWALK in which all letters have been transposed. Spatial coding models predict priming in these cases
where none is found suggesting that there is some form of constraint on the flexibility of letter position cod-
ing. Guerrera suggests a model in which there are in fact two stages in the search; one which checks for
a critical subset of the letters, and another the checks specific letter position from left to right. In this sys-
tem at least two letters in the prime must ultimately match in identity and position for priming effects to be
observed.2
This finding raises the question as to the basis of transposed-letter constraints. In order to address this
question the current study probes the existence of syllabic and/ or CV structural sublexical components in-
volved in lexical access in English as the possible source of transposed-letter effects. Specifically, is lexical
access sensitive to structural notions of syllable or underlying CV pattern?
2.1 Previous Investigation on Syllable and CV Structure
As early as 1976 there has been speculation that syllabic structure has an influence on lexical access.
Taft and Forster (1976) suggest that the first syllable serves as a means to access a lexical entry. In later work
(Taft, 1979) formulated a metric for orthographic syllables coined the BOSS (Basic Orthographic Syllable
Structure) pointing to a unique orthographic syllable component used in mediating lexical access. However,
the most robust evidence for the syllable as strategy in word recognition comes from work on languages
other than English. Carreiras and Perea (2002) found that for Spanish priming is stronger for sequences in
which syllable overlap is maintained pa####, PA.SI.VO , in contrast to priming pairs in which more letters
are overlapped *pas###, PA.SI.VO.
The argument has been made that robust effects pointing to the syllable’s influence on lexical access
are tied to the regularity of syllabic structure in the language in question (Taft and Radeau, 1995). This has
been the main contention against English and the existence of an active sublexical syllabic component. The
observation that the English syllable is more difficult to define compared to the syllable, for example, in
Spanish complicates strategic use of this metric.
Another curious finding, also from Spanish, is that less priming in consonant transposition than vowel
transpositions. Perea and Lupker (2004) produced less priming for non-adjacent consonants transpositions,
such as the pairs *casino, CASINO, than in pairs where non-adjacent vowels were transposed, anamil, ANI-
MAL.
This data appear to support a further layer of sublexical abstraction, namely a CV pattern distinction.3
Furthermore, unpublished evidence from Dutch suggests primes that respect the sequence of vowels and
consonants of the target, such as the pairs bruek, BREUK are more efficient that primes that do not, such
2This is based on target/prime pairs of eight-letter words. The number of necessary position specific matches may vary according to
target/prime word length.
3Independent literature also attests a consonant/ vowels processing distinction (Caramazza et al., 2000; Lee et al., 2002).
90
Francom

as *beruk, BREUK (Martensen, 2006) suggesting that the underlying CV pattern of words is calculated and
employed to select target sets of lexical candidates in lexical access.
Therefore the investigation here, to probe the freedom of transposed-letter effects using structural notions
of syllable and CV structure as a working hypothesis, faces immediate challenges given considerations for
English mentioned. For this reason the stimuli here were specifically chosen to enhance syllabic regularity
by selecting a) the most common and most clearly defined syllabification pattern, which is CVC.CVC and b)
the most constant CV structure, simultaneously. These characteristics are found in words such as NOR.MAL.
This class constitute 25% of six letter combinations, the most dominant pattern for all words with letter length
(4-9) based on the CELEX database for English (Baayen et al., 1993). The belief here is that if syllabic and/
or CV patterns do play a role in English that the more regular patterns in the language should show the most
robust, and therefore most detectable effects.
3 Experiment 1
The purpose of experiment 1 is to test the hypothesis that the syllable is an active component in lexical
access and that syllabic boundaries are employed as structural cues for the search process. If lexical access
does have a structural component that corresponds to syllabic properties the letter-transposition effects should
be sensitive to transpositions of letters that disrupt these structures, ultimately yielding priming asymmetries
between primes where letter-transposition takes place across a syllable boundary and another in which letter-
transposition occurs entirely within the syllable.
3.1 Methods
Participants Nineteen undergraduate students and one graduate student participated in the experiment.
Undergraduates were enrolled in an introductory psychology course at the University of Arizona and received
course credit for their participation. Stimuli Selection and Design Sixty four target words six letters in length
and of mid to low frequency were selected from the CELEX database for English (Baayen et al., 1993). All
were monomorphemic in order to eliminate any other sublexical interference such as morphological structure
in the stimuli (Andrews et al., 2004) and were composed of an underlying CVCCVC ((C)consonant (V)vowel)
pattern for CV consistency across all words and to enhance the regularity of syllabification.
Half of the words were syllabified CV-CCVC such as the word ‘re-gret’ and the other half syllabified
CVC-CVC as in the word ‘nor-mal’.4 The third and fourth letters were transposed in each set of words to
create the non-word primes to avoid left-to-right superiority effects: rerget, REGRET and nomral, NORMAL.
A control set created, controlled for length and frequency in parallel with the target words. All stimuli were
counterbalanced into two lists and randomized for presentation according to the following procedure.
Procedure The experiment was controlled by a Pentium PC, using the Windows DMDX software devel-
oped by J.C. Forster at the University of Arizona (Forster and Forster, 2003). Items were presented as black,
lower-case letters in New Courier font on a white background using a color monitor with a refresh cycle of
10 ms. The forward mask (e.g., ######) and the upper-case target were presented for 500 ms, and the prime
was presented for 40 ms in order to avoid any awareness of the prime.5 Participants were asked to decide
whether the presented word was a English word, and respond by pressing a button, ‘Yes’ or ‘No’. They were
also instructed to respond as quickly as possible without making errors and that this task was not a vocabulary
test. Feedback was given and a practice set of 6 words and 6 non-words was given before testing.
4Feedback from a reviewer points out that the ‘reget’ condition and the ‘normal’ condition also confound stress. I will assume here
that this phonological process is not a factor in orthographic word recognition. This may appear to be somewhat misleading as the topic
under investigation in experiment 1, the syllable, is clearly based upon a phonological entity. Nevertheless, the syllable here is employed
as orthographically analogous to the phonological syllable. On this point, it has been proposed that there is in fact an orthographic
syllable (BOSS) (Taft, 1992) that is not directly congruent to its phonological counterpart. However, I’ve chosen to approach the current
investigation using the phonological syllable as directly correspondent to the ‘orthographic’ syllable as there is no principled reason for
preferring the BOSS calculation.
5In a pilot study, some participants reported being at least partially aware of the prime. Given that priming effects do not exceed 40
ms, this appears to be the maximum gain attainable and therefore, any prime given longer than 40ms is no more effective in producing
priming.
91
Proceedings of the 24th NWLC, 3-4 May 2008, Seattle, WA

3.2 Results
Data Modification The mean Reaction Times (RTs) were collected for correct responses and trimmed to
remove outliers. All RT below 300ms and above 1500ms were removed. Overall Test Overall mean latencies
can be seen in Table 1 and 2.
Table 1: Across Syllable Mean RTs
TARGET Across Syllable Priming
NORMAL nomral sketch
(ms) 560.38 589.14 29.13
Table 2: Within Syllable Mean RTs
TARGET Within Syllable Priming
REGRET rerget health
(ms) 574.16 600.10 25.94
The data was analyzed using a 2x2x2 mixed design ANOVA with POSITION(Across Syllable, Within
Syllable) and PRIME TYPE(Related, Unrelated) as within-subjects factors and GROUP(A,B) as the be-
tween subjects factor. In the items test POSITION(Across Syllable, Within Syllable) and GROUP(A,B) were
between subjects and PRIME TYPE(Related, Unrelated) as a within-items factor. The main effects for POSI-
TION were significant by-subjects (F1(1,20)= 11.53, p<.01) but not by-items(F2(1,60)= 1.49, p<1) with no
significant errors in the by-subjects nor by-items tests. However, the main effects for PRIME TYPE were sig-
nificant by-subjects and by-items (F1(1,20)= 18.08, p<.001); (F2(1,60)= 18.84, p<.001) and no significant
errors. The interaction between POSITIONxPRIME TYPE was not significant.
3.3 Discussion
The results here point to significant priming in both Across Syllable and Within Syllable conditions
reaffirming the ability of the lexical access system correlate transposed-letter primes with their correspond-
ing target words. However, the priming is not significantly larger for the Across condition than the Within
condition. The hypothesis that syllable structure mediates the lexical processor cannot be confirmed from the
evidence presented here. In absence of a syllable effect, the evidence points instead to orthographic overlap
as the key dynamic. However, it may be the case that there is a syllable effect but given that the items in this
experiment were specifically chosen to represent the most dominant and clearly defined syllable pattern in
English it is quite likely a negate-able effect.
The question remains, however, whether the consonant/vowel distinction is the relevant strategy for
English instead of the syllable. Experiment 2 aims to investigate this question.
4 Experiment 2
Experiment 2 aims to investigate the hypothesis that the underlying CV pattern cues lexical access.
If primes that share the same underlying CV pattern as their targets prime better than primes in which CV
pattern is dissimilar to their targets, all else being equal, then the CV distinction would appear to be an active
component in early word recognition.
92
Francom

4.1 Methods
Participants Forty undergraduate students participated in the experiment. Undergraduates were enrolled
in an introductory psychology course at the University of Arizona and received course credit for their partici-
pation. Stimuli Selection and Design Eighty target words 6 letters in length and of mid to low frequency were
selected from the CELEX database for English (Baayen et al., 1993). All were monomorphemic and, again
as in Experiment 1, were composed of an underlying CVCCVC ((C)consonant (V)vowel) pattern. From the
eighty target words the following four lists were created. 1) Left Transpositions: the 2nd and 3rd letter po-
sitions were transposed as in wnader, WANDER thus, creating an underlying CCVCVC pattern. 2) Center
Transpositions: the 3rd and 4th positions were transposed as in wadner, WANDER resulting in an underlying
CVCCVC pattern. 3) Right Transpositions: the 4th and 5th letter positions were transposed as in wanedr,
WANDER, underlyingly CVCVCC. 4) Control: a control group of unrelated words was created that was
matched for letter length and word frequency.
Procedure Same as in Experiment 1.
4.2 Results
Data Modification The mean Reaction Times (RTs) were collected for correct responses and trimmed to
remove outliers. All RT below 300ms and above 1500ms were removed. Overall Test Overall mean latencies
can be seen in Table 3 and 4 (by subject and by item).
Table 3: Condition Mean RTs by-subjects
TARGET Left Center Right
WANDER wnader wadner wanedr
(ms) 539.83 542.68 536.30
priming 22.6 19.75 26.13
Table 4: Condition Mean RTs by-items
TARGET Left Center Right
WANDER wnader wadner wanedr
(ms) 553.48 552.18 545.44
priming 18.46 19.76 26.5
The data was analyzed using a 4x4 ANOVA mixed design with the factor GROUP(A,B,C,D) as a
between-subjects and between-items factor and PRIME TYPE(Left, Center, Right, Control) within-subjects
and within-items factors. The overall test showed main effects for PRIME TYPE by-subjects (F1(3,108)=
7.89, p<.001) but not by-items (F2(3,228)= 2.48, p<.1) with no significant error in either test. Follow-
up pairwise comparisons revealed that the Center condition differed significantly from the Control condition
(F1(1,36)= 10.54, p<.05); (F2(1,76)= 6.95, p<.05) as did the error rates (F1(1,36)= 6.59, p<.05); (F2(1,76)=
4.40, p<.05). Again, the Right condition differed significantly from the Control condition (Right(F1(1,36)=
22.02, p<.001); (F2(1,76)= 6.36,<.05)). The Left condition significantly differed from the Control condition
in the by subjects test Left(F1(1,36)= 15.39, p<.001) but not by items F2(1,76)= 2.20, p>.1. Another striking
result is that neither Left nor Right conditions had significant errors.
Therefore, there was significant priming for Center and Right conditions with the Center condition having
significantly lower error rates than both of the other conditions in comparison with the Control condition. No
significant priming effect was found for Left condition transpositions.
93
Proceedings of the 24th NWLC, 3-4 May 2008, Seattle, WA

4.3 Discussion
These data appear to point away from CV structure as a calculated metric in the lexical access proces-
sor. Center transpositions, the only primes that maintain the underlying CV structure prime less than Right
transpositions which do not maintain CV structure. The results here initially point to a dominant left to right
superiority effect as the Left condition was the only condition that failed to produce consistent priming and
the Right condition showed the strongest priming overall, graphically in Figure 1(a). In line with previous
results (Guerrera, 2004; Davis, 1999) primes with more letter position matches in the left of the word obtain
more robust priming effects.Condition ControlR - noramlC - nomralL - nromal
Mean Erro
r Rates (ms
)
.16
.14
.12
Condition R - noramlC - nomralL - nromal
Mean Prim
ing Effect (
ms)
28262422201816
Page 1
(a) Left to Right Priming
Condition ControlR - noramlC - nomralL - nromal
Mean Erro
r Rates (ms
)
.16
.14
.12
Condition R - noramlC - nomralL - nromal
Mean Prim
ing Effect (
ms)
28262422201816
Page 1
(b) Overall Response Errors
Figure 1: Priming Effects
Thus, if there is a CV pattern effect this effect is lik ly eclipsed by the priming advantage of more posi-
tion specific matches from left to right.
An interesting, but more difficult, result to incorporate is the significant drop in error rates for primes
that shared a common CV structure with the target word seen in Figure 1(b). This result, in effect, says that
participants were less accurate in responding correctly to the target word as a word or not of English when
the prime given did not share the same underlying CV pattern as the target.6
How to interpret these two effects is a difficult question. On first examination, the significant drop in
errors for primes that contained CV sequences that matched their corresponding target forms appears to be
evidence supporting the hypothesis that CV patterns at least recognized in some capacity in lexical access
in English. However, the relevant question here is, what does accuracy in lexical decision tasks mean?; a
question outside the scope of the current analysis.
5 Discussion & Conclusion
Letter-Transposition Effects indicate that the lexical access system permits flexibility in letter position.
As discussed, there is some consensus that word-initial and word-final positions have a privileged status in
this system. Yet there is little agreement on the positional import of inner-letters. The working hypothesis of
the current investigation is that there is a sublexical component to lexical access that is structural in nature. In
other words, accessing the lexicon is mediated by a layer (or layers) in which structural properties of words
are recognized and employed as tools to gain access to a lexical item.
Given evidence from other research, here the investigation centered on two potentially active sublexical
components: the syllabic and CV patterns of words. Although robust priming was found in Experiment 1, no
6In the case of the control items, which elicited the most errors, the mismatch was not only on the CV level but also at the letter
identity level as the controls did not orthographically overlap in any consistent way.
94
Francom

significant differences between prime/target pairs where transpositions appear across the syllable boundary
versus within the syllable were found. This, despite efforts to select stimuli that best represented regular
syllabic patterns to counteract ambiguities particular to English. Experiment 2 also resulted in less-than-
convincing evidence that the CV patterns of English words are employed in lexical access. However, their
may be an effect for CV patterns in English but only if this effect is less than left-to-right superiority effects.
The question now is how to interpret the null findings for syllabic and CV patterns here given their
attested existence in other research. The most likely approach is that statistical regularity in language patterns
can be adopted as strategies for lexical access. In effect, language specific properties are employed by the
lexical access system to serve as cues for lexical access. Results showing syllabic effects come from lan-
guages that are Syllable-Timed languages, in which syllable length and duration are more constant internally,
and across speakers. English on the other hand is a Rhythmically-Timed language and as such allows syllable
boundaries to be manipulated in speech, and may vary even among speakers.
As for the CV distinction, languages such as Spanish and Dutch, discussed here, show a processing
distinction between consonants and vowels at some level. Evidence from Martensen (2006) suggest that in
Dutch this distinction is productive in lexical access to the point that primes that share the underlying CV
pattern with targets show priming effects to the exclusion of those that do not. Notably these languages, that
do show CV processing asymmetries, also share a closer association between phonological and orthographic
forms. English, on the other hand, is not an especially phonemic language.
Table 5: Language Properties and Sublexical Effects:
ST: Syllable-Timed language, PO: Phonemic orthography
Lang ST PO Syll Effect CV Effect
Spanish x x x x
Dutch - x - x
English - - - -
Seen in Table 5, the distribution of syllabic and CV pattern effects appear to come from the very lan-
guages that demonstrate regular syllabic patterns and/ or especially close associations between phonological
and orthographic forms. English may in fact be sufficiently irregular on these dynamics making syllabic and
CV structure a less-than-optimal metric on which to base a viable lexical access strategy. Even more con-
vincing is the finding also from Martensen (2006) that syllabic structure is not active in Dutch; a predicted
result if the timing of a language is intimately linked to the use of syllables in lexical access.
This evidence points to language specific sublexical components in lexical access and not a universal
strategy across languages. The spatial coding models do not need extensive modification to integrate this
hypothesis. The SOLAR model, for example, explicitly incorporates a segmentation-through-recognition
component that could be used to explain cross-linguistic differences in syllable and CV structure recognition.
Reaffirmed here is the inability of slot-coding schemes to adequately deal with input in which prime/target
mismatch in letter-position providing more evidence in favor of spatial coding approaches. Also reaffirmed
are Left to Right Superiority effects, revealing that letter-position is more crucial from left to right. Still
unexplained by these results, however, is existence of a strategy or type of strategy adopted by English to
constrain inner-letter transpositions. Given the findings here, there is mounting evidence that syllabic and CV
patterns are not viable strategies for English.
References
Andrews, S., Miller, B., and Rayner, K. (2004). Eye movements and morphological segmentation of compound words: There is a mouse
in mousetrap. European Journal of Cognitive Psychology, 16(1-2):285–311.
Baayen, R., Piepenbrock, R., and van Rijn, H. (1993). The CELEX lexical database (CD-ROM). Linguistic Data Consortium, University
of Pennsylvania, Philadelphia, PA.
Caramazza, A., Chialant, D., Capasso, R., and Micelli, G. (2000). Separable processing of consonants and vowels. Nature, 403.
Carreiras, M. and Perea, M. (2002). Masked priming effects with syllabic neighbors in a lexical decision task. Journal of Experimental
Psychology: Human Perception and Performance, 28(5):1228–1242.
95
Proceedings of the 24th NWLC, 3-4 May 2008, Seattle, WA

Coltheart, M., Rastle, K., Perry, C., and Langdon, R. (2001). DRC: A dual route cascaded model of visual word recognition and reading
aloud. Psychological Review, 108(1):204–256.
Davis, C. J. (1999). The self-organizing lexical acquisition and recognition (SOLAR) model of visual word recognition. PhD thesis,
University of New South Whales, Australia.
Forster, K. I. and Forster, J. (2003). DMDX: A windows display program with millisecond accuracy. Behavior Research Methods,
Instruments, & Computers, 35:116–124.
Grainger, J. and Jacobs, A. (1996). Orthographic processing in visual word recognition: A Multiple Read-Out Model. Psychological
Review, 103(3).
Guerrera, C. (2004). Flexibility and constraint in lexical access: explorations in transposed-Letter Priming. PhD thesis, University of
Arizona.
Inhoff, A. W., Radach, R., Eiter, B. M., and Skelly, M. (2003). Exterior letters are not priveleged in early stage of visual word recognition
during reading: Comment on Jordan, Thomas, Patching and Scott-Brown (2003). Journal of Experimental Psychology: Learning,
Memory and Cognition, 29(5):894–899.
Jordan, T. R., Thomas, S. M., Patching, G. R., and Scott-Brown, K. C. (2003). Assessing the importance of letter pairs in initial, exterior
and interior positions in reading. Journal of Experimental Psychology: Learning, Memory and Cognition, 29(5):883–893.
Lee, H., Rayner, K., and Pollatsek, A. (2002). The processing of consonants and vowels in reading: Evidence from the fast priming
paradigm. Psychonomic Bulletin & Review, 9(4):766–772.
Martensen, H. (2006). Determinants and interpretation of phonological priming. Mental Lexicon Conference, 2006.
McClelland, J. and Rumelhart, M. (1981). An interactive activation model of context an interactive activation model of context effects
in letter perception. Psychological Review, 88:375–407.
Perea, M. and Lupker, S. (2003). Does jugde activate COURT? transposed-letter similarity effects in masked associative priming.
Memory & Cognition, 31(6):829–841.
Perea, M. and Lupker, S. (2004). Can CANISO activate CASINO? transposed-letter similarity effects with nonadjacent letter positions.
Journal of Memory and Language, 51:231–246.
Peressotti, F. and Grainger, J. (1999). The role of letter identity and letter position in orthographic priming. Perception & Psychophysics,
61(4):691–706.
Schoonabaert, S. and Grainger, J. (2004). Letter position coding in printed word perception: Effects of repeated and transposed letters.
Language and Cognitive Processes, 19(3):333–367.
Taft, M. (1979). Lexical Access Via an Orthographic Code: The Basic Orthographic Syllabic Structure (BOSS). Journal of Verbal
Learning and Verbal Behavior, 18(1):p21–39.
Taft, M. (1992). The body of the BOSS: subsyllabic units in the lexical processing of polysyllabic words. Journal of experimental
psychology. Human perception and performance, 18(4):1004–1014.
Taft, M. and Forster, K. I. (1976). Lexical Storage and Retrieval of Polymorphemic and Polysyllabic Words. Journal of Verbal Learning
and Verbal Behavior, 15(6):607–620.
Taft, M. and Radeau, M. (1995). The influence of the phonological characteristics of a language on the functional units of reading: A
study in french. Canadian journal of experimental psychology, 49(3):330–348.
Whitney, C. (2001). How the brain encodes the order of letters in a printed word: The SERIOL model and selective literature review.
Psychonomic Bulletin & Review, 8:221–243.
96
Francom