Lexical decisions to word targets were analysed as follows. Incorrect responses were removed from the reaction time RT analysis 2. There were no reaction times smaller than ms or larger than ms that had to be excluded from the analyses. RTs and error rates are presented in Table 2. Standard errors are shown in parentheses. Models were selected using chi-squared log-likelihood ratio tests with regular maximum likelihood parameter estimation. Trial order was included to control for longitudinal task effects such as fatigue or habituation.
To assess whether the obtained effects were modulated by individual differences in vocabulary, spelling or reading proficiency, the vocabulary, spelling and reading proficiency scores were standardized. A composite measure of reading ReadZ was calculated by averaging the standardized scores of the two reading tests.
All continuous variables were centred i. The model was refitted after excluding data-points whose standardized residuals were larger than 2.
No other effects were significant. Priming effects for targets preceded by compound word, compound-nonword, and non-compound nonword primes relative to the unrelated control condition , as a function of individual differences in reading and spelling proficiency. Positive proficiency scores represent individuals who are better readers than spellers.
Negative proficiency scores represent individuals who are better spellers than readers. Error analyses followed the same logic as the RT analyses. We applied a binomial variance assumption to the trial-level binary data using the function glmer as part of the R-package lme4. The results of Experiment 1 are clear-cut.
Masked Priming: The State of the Art - Google книги
Priming is obtained independently of whether the embedded target is accompanied by a real morpheme e. The priming effect seen in the pimebook-BOOK condition is consistent with previous embedded word priming studies Beyersmann et al. Moreover, greater priming was obtained in the compound word condition relative to the two nonword prime conditions, which was particularly pronounced in individuals who were better readers than spellers we return to this point in the General Discussion.
Embedded words are activated simply by mapping the letters of the input string onto existing whole-word representations in the orthographic lexicon, independently of morphological structure. However, if it is true that the activation of embedded words is based on a clearly non-morphological process, one crucial question to ask is why Fiorentino and Fund-Reznicek failed to observe priming in the window-WIN condition? We hypothesise that the absence of priming in this condition can be explained by lateral inhibition between the lexical representations of window and win.
Of course, there would also be lateral inhibition between opaque compound words e. Here, however, the presence of the second constituent e. The goal of Experiment 2 was to further manipulate the position of the embedded words. Experiment 2 was designed to test this hypothesis.
The aim of Experiment 2 was to compare edge-aligned embedded word priming e. If the activation of embedded words is simply due to lower-level orthographic letter-to-word mappings, we would expect that the amount of priming should be determined by the predicted amount of orthographic overlap between prime and target. The predicted amount of orthographic overlap differs depending on different orthographic coding schemes cf.
Match Calculator software, Version 1. This model thus predicts significantly less priming in the outer condition compared to the edge-aligned and mid conditions.
Forty-seven students from the University of New South Wales, all English native speakers, participated for course credit. We used the exact same non-compound nonwords with word final edge-aligned embeddings e. In addition, we created a mid-embedded condition e.
Some primes were slightly changed to avoid having letters of the target being repeated in the second constituent of the prime. Finally, we used the items of the unrelated control condition of Experiment 1, but the first constituent was replaced with the non-morphemic first constituent of the non-compound prime e. All items are listed in Appendix B. The procedure of the masked priming task was identical to the one used in Experiment 1.
Since Experiment 1 showed no evidence for the influence of vocabulary, reading and morphological awareness on masked non-compound priming, we did not assess individual proficiency measures in Experiment 2. Incorrect responses were removed from the reaction time RT analysis 5. Reaction times smaller than ms or larger than ms were excluded from the analyses 1. RTs and error rates are presented in Table 3. As in Experiment 1, we used linear mixed-effects modelling to perform the main analyses Baayen, ; Baayen et al.
A linear mixed-effects model was created with two fixed effects factors primetype: edge-aligned, mid, outer, unrelated; trial order and two random effects factors random intercepts for subjects and items. All continuous variables were centred.
The model was refitted after excluding data-points with standardized residuals larger than 2. Error analyses revealed a similar pattern as the RT analyses. The results of Experiment 2 confirm the second constituent priming effects found with non-compound nonword primes e. Crucially, the significant priming effects found in this condition in Experiment 2 contrast with the non-significant priming effects seen with primes that share the same number of letters with targets but are either edge-aligned but non-contiguous e.
The first result is evidence that orthographic overlap in itself is not sufficient to generate priming effects, and the second result is evidence that having the target word embedded in the prime is not sufficient to generate priming effects. The complete set of results points to edge-alignment and lexical status of the letters shared by prime and target as two key factors driving the priming effects seen in Experiment 1. The primary goal of our study was to investigate embedded word activation processes operating during the processing of compound words and compound nonwords by using masked priming combined with the lexical decision task.
In Experiment 2, significant priming effects were found in the edge-aligned embedded word condition e.
This replicates the embedded word priming effects previously obtained with derived nonword primes e. Contrary to Beyersmann et al. Importantly, Experiment 2 demonstrates that priming is only obtained for edge-aligned embedded words e. The second key finding is that the magnitude of priming was the same whether the target word was the first or the second constituent of the prime textbook-BOOK vs. This result is compatible with previous studies investigating compound word processing Crepaldi et al. Unlike prefixes and suffixes, embedded words are not subject to positional constraints, and this can explain why the identification of stem morphemes is position independent Crepaldi et al.
Our results also converge with previous evidence for embedded word activation mechanisms in monomorphemic nonwords e. In line with our present findings, Taft et al. Experiment 1 revealed a third key result, namely that greater priming was obtained in the compound word condition relative to the two nonword prime conditions. This suggests that the whole-word representation of the compound word prime contributed to priming effects.
The fact that Fiorentino and Fund-Reznicek found no difference between transparent and opaque compound word primes suggests that the difference between compound word and compound-nonword primes seen in our study is driven by whole-word form representations of the compound word and not by morpho-semantic representations.
However, our observation that priming in the compound word condition was particularly pronounced in those participants who were better readers than spellers Figure 1 , suggests that future research should examine the role of language proficiency in determining the relative size of transparent and opaque compound word priming before ruling out a role for morpho-semantic representations.
Highly relevant with respect to this possibility are the results reported by Andrews and Lo , who found that individuals with higher proficiency in spelling relative to vocabulary showed stronger priming from opaque derived word primes corner-CORN. In other words, those individuals who were relatively poor spellers tend to rely more heavily on whole-word processing, which was also the case in Andrews and Lo Our results concerning the effects of language proficiency are in line with the hypothesis that participants who are better readers than spellers place more emphasis on direct access to whole-word orthographic representations from print compared with morpho-orthographic segmentation processes.
These participants would therefore benefit more from the lexical representation of real compound words textbook , presumably because they would be particularly proficient in rapidly mapping input letter strings onto their corresponding whole-word representations. On the other hand, those participants who were better spellers than readers would be less efficient in mapping a complete letter string onto its whole-word representation and therefore rely to a greater extent on morpho-orthographic segmentation mechanisms. Importantly, the present data provide further support for embedded word activation mechanisms operating at the level of whole-word representations.
More precisely, our results allow us to rule-out an explanation of embedded word priming effects in terms of lower-level letter overlap between the prime and the target.
- Masked Priming: The State of the Art - Google книги.
- Edited by Robert B. Kaplan.
- Advances in space science and technology. Volume 5.
- Serie: Macquarie Monographs in Cognitive Science.
- Introduction to Nonlinear Thermomechanics: Theory and Finite-Element Solutions!
- Wedding Photography Field Guide!
If orthographic overlap, independently of whether the overlapping letters form a word or not, were the factor driving embedded word priming effects, then we should have observed similar priming from the non-contiguous primes e. Of course, it could be argued that having non-contiguous orthographic overlap is not an appropriate comparison, and that a more appropriate comparison might involve contiguous letters in so-called partial primes e.
On the other hand, finding significant priming with such partial primes would force us to re-consider our word-based interpretation of the present constituent priming effects. Taken together, the present findings, in addition to evidence obtained from prior research using different paradigms e.