Using the masked priming paradigm, we examined which phonological unit is used when naming Kanji compounds. Although the phonological unit in the Japanese language has been suggested to be the mora, Experiment 1 found no priming for mora-related Kanji prime-target pairs. In Experiment 2, significant priming was only found when Kanji pairs shared the whole sound of their initial Kanji characters. Nevertheless, when the same Kanji pairs used in Experiment 2 were transcribed into Kana, significant mora priming was observed in Experiment 3. In Experiment 4, matching the syllable structure and pitch-accent of the initial Kanji characters did not lead to mora priming, ruling out potential alternative explanations for the earlier absence of the effect. A significant mora priming effect was observed, however, when the shared initial mora constituted the whole sound of their initial Kanji characters in Experiments 5. Lastly, these results were replicated in Experiment 6. Overall, these results indicate that the phonological unit involved when naming Kanji compounds is not the mora but the whole sound of each Kanji character. We discuss how different phonological units may be involved when processing Kanji and Kana words as well as the implications for theories dealing with language production processes.

In most models of word processing, the degrees of consistency in the mappings between orthographic, phonological, and semantic representations are hypothesized to affect reading time. Following Hino, Miyamura, and Lupker's (2011) examination of the orthographic-phonological (O-P) and orthographic-semantic (O-S) consistency for 1,114 Japanese words (339 katakana and 775 kanji words), in the present research, we initially attempted to measure the phonological-orthographic (P-O) consistency for those same words. In contrast to the O-P and O-S consistencies, which were equivalent for kanji and katakana words, the P-O relationships were much more inconsistent for the kanji words than for the katakana words. The impact of kanji words' P-O consistency was then examined in both visual and auditory word recognition tasks. Although there was no effect of P-O consistency in the standard visual lexical-decision task, significant effects were detected in a lexical-decision task with auditory stimuli, in a perceptual identification task using masked visual stimuli, and in a lexical-decision task with degraded visual stimuli. The implications of these results are discussed in terms of the impact of P-O consistency in auditory and visual word recognition.

The present research involved masked priming lexical decision experiments using, in the crucial condition, masked primes with an orthographic neighbour that was semantically related to the target. Regardless of the lexicality of the prime, a significant priming effect was observed when the relatedness proportion (RP, that is, the proportion of primes and targets that were directly related on the "word" trials) was 2/3 (Experiments 1 and 2). No effect emerged, however, when the RP was 0 (Experiment 3). These results indicate that lexical/semantic activation arises automatically for both the prime and its neighbours. This activated lexical/semantic information appears to be evaluated together with the lexical/semantic information activated by the target, creating a decision bias during the decision-making process, but only when that information often provides a clue as to the nature of the correct decision. Our results, therefore, also provide support for the retrospective account of masked semantic priming.

Two experiments investigated whether Japanese-English bilinguals have integrated phonological stores for their two languages using a masked phonological priming task with Japanese Kanji (logographic) primes and English targets. In both experiments, lexical decisions for English target words were facilitated by phonologically similar Kanji primes. Furthermore, the size of the phonological priming effect was uninfluenced by the participants' English proficiency or target word frequency, which suggests that the priming effect arose from feedback from sublexical phonological representations to lexical orthographic representations. Because of the orthographic and phonological differences between Japanese and English, these findings provide particularly strong support for the Bilingual Interactive Activation (BIA+) model's assumption that representations are integrated across languages.

In the masked priming paradigm, when a word target is primed by a higher frequency neighbor (e.g., blue-BLUR), lexical decision latencies are slower than when the same word is primed by an unrelated word of equivalent frequency (e.g., care-BLUR). This inhibitory neighbor priming effect (e.g., Davis & Lupker, 2006; Segui & Grainger, 1990) is taken as evidence for the lexical competition process that is an important component of localist activation-based models of visual word recognition (Davis, 2003; Grainger & Jacobs, 1996; McClelland & Rumelhart, 1981). The present research looked for evidence of an inhibitory neighbor priming effect using words written in Japanese Kanji, a logographic, nonalphabetic script. In 4 experiments (Experiments 1A, 1B, 3A, and 3B), inhibitory neighbor priming effects were observed for low-frequency targets primed by higher frequency Kanji word neighbors In contrast, there was a significant facilitation effect when targets were primed by Kanji nonword neighbors (Experiments 2 and 3). Significant facilitation was also observed when targets were primed by single constituent Kanji characters (Experiment 4). Taken together, these results suggest that lexical competition plays a role in the recognition of Kanji words, just as it does for words in alphabetic languages. However, in Kanji, and likely in other logographic languages, the effect of lexical competition appears to be counteracted by facilitory morphological priming due to the repetition of a morphological unit in the prime and target (i.e., in Kanji, each character represents a morpheme).

Using the lexical decision task, we compare task performance for three- versus four-character katakana words and two- versus three-character kanji words. While no length effect was observed for katakana words, a significant effect was observed for kanji words, such that lexical decision latencies were longer for three-character kanji words than for two-character kanji words. This effect was observed even when orthographic neighborhood sizes were matched across the two kanji-word groups. Because the kanji words are compound words, but the katakana words are mono-morphemic words, the observation of a length effect only for kanji words may indicate that the effect is due to the difference in the number of morphemes. Assuming morphological decomposition for kanji compound words, the constituent morphemes would have to be recombined to access to whole-word representations. Thus, the length effect could arise within the recombination process because three-character kanji compounds have more complicated morphological structures than two-character kanji compounds. As such, the present results suggest that the nature of reading process is highly dependent on the morphological structure that a word possesses.

In order to examine whether naming performance for Japanese kanji words is determined by the degree of orthographic–phonological consistency at the kanji character level or by the type of readings for the constituent kanji characters (on readings versus kun readings), naming performance was compared for on-reading and kun-reading two-character kanji words with matched orthographic–phonological consistencies in Experiment 1. In Experiment 2, naming performance was compared for kun-reading single-character kanji words with higher and lower on-reading ratios. In Experiment 1, naming performance was comparable for the on-reading and kun-reading words. In Experiment 2, naming responses were faster for kanji characters with lower on-reading ratios than for those with higher on-reading ratios. These results clearly suggested that naming performance for kanji words is modulated by the degree of orthographic–phonological consistency but not by the type of readings possessed by the constituent kanji characters.

Studies using the lexical decision task with English stimuli have demonstrated that homophones are responded to more slowly than nonhomophonic controls. In contrast, several studies using Chinese stimuli have shown that homophones are responded to more rapidly than nonhomophonic controls. In an attempt to better understand the impact of homophony, we investigated homophone effects for Japanese kanji words in a lexical decision task. The results indicated that, whereas a processing disadvantage emerged for homophones when they have only a single homophonic mate (as in the English experiments), a processing advantage occurred for homophones when they have multiple homophonic mates (as in the Chinese experiments). On the basis of these results, we discuss the nature of the processes that may be responsible for producing the processing advantages and disadvantages for homophones. © 2012 American Psychological Association.

According to the orthographic depth hypothesis, automatic phonological activation is not assumed when reading deep orthographies. However, some previous studies have suggested the possibility of automatic phonological activation even for deep orthographies, such as Chinese and Japanese Kanji words. In order to examine whether automatic phonological activation arises when reading Kanji words, this study attempts to examine whether a priming effect can be observed for homophonic Kanji word pairs using lexical decision task with masked primes. Significant homophone priming effects are observed using two different sets of stimulus pairs. Thus, in contrast to the orthographic depth hypothesis, these results clearly indicate that phonological activation automatically arises when reading Kanji compound words.

Masked translation priming effects were examined for Japanese-English bilinguals using both cognate and noncognate translation equivalents. In Experiment 1, L1 primes and L2 targets were used to examine translation priming effects as a function of target frequency and bilinguals' L2 proficiency. Translation priming effects for cognates were significantly larger than for noncognates, replicating the cognate priming advantage previously reported with different-script bilinguals. In addition, translation priming effects were significantly larger for low- than for high-frequency targets and for less- than for more-proficient bilinguals, whereas the size of the cognate priming advantage was unaffected by either target frequency or L2 proficiency. In Experiment 2, cognate translation priming effects were tested in the L2-L1 direction. There was a significant cognate translation priming effect regardless of L2 proficiency. These results are consistent with the phonological account of the cognate priming advantage, which proposes that cognate translation priming effects are due to the additive effects of phonological and conceptual factors. © 2013 Taylor &amp
Francis.

:There is now considerable evidence that a letter string can activate semantic information appropriate to its orthographic neighbors (e.g., Forster & Hector's, 2002, TURPLE effect). This phenomenon is the focus of the present research. Using Japanese words, we examined whether semantic activation of neighbors is driven directly by orthographic similarity alone or whether there is also a role for phonological similarity. In Experiment 1, using a relatedness judgment task in which a Kanji word-Katakana word pair was presented on each trial, an inhibitory effect was observed when the initial Kanji word was related to an orthographic and phonological neighbor of the Katakana word target but not when the initial Kanji word was related to a phonological but not orthographic neighbor of the Katakana word target. This result suggests that phonology plays little, if any, role in the activation of neighbors' semantics when reading familiar words. In Experiment 2, the targets were transcribed into Hiragana, a script they are typically not written in, requiring readers to engage in phonological coding. In that experiment, inhibitory effects were observed in both conditions. This result indicates that phonologically mediated semantic activation of neighbors will emerge when phonological processing is necessary in order to understand a written word (e.g., when that word is transcribed into an unfamiliar script).

Previous masked phonological priming studies with bilinguals whose languages are written in the same script (e.g., Dutch-French bilinguals) strongly suggest that phonological representations for the two languages are integrated, based on the fact that phonological activation created by reading a word in one language facilitates word identification in the other language. The present research examined whether the same is true for different-script bilinguals (Japanese-English bilinguals). In this study, participants made lexical decisions to English targets (e.g., GUIDE) that were primed by three types of masked Japanese primes: cognate translation equivalents (e.g., (sic), /gaido/, guide), phonologically similar but conceptually unrelated words (e.g., (sic), /saido/, side), and phonologically and conceptually unrelated words (e.g., (sic), /koRru/, call). There were significant priming effects for cognate translation primes (94 ms) and phonologically similar primes (30 ms). Whereas the cognate translation priming effect was modulated by target frequency and L2 proficiency, the phonological priming effect was not. Our results suggest that phonological representations for different languages are integrated even if the languages in question use different scripts. The role of phonological activation in bilingual word recognition is discussed.

It is generally assumed that orthographic-phonological (O-P) consistencies are higher for Japanese kana words than for kanji words and that orthographic-semantic (O-S) consistencies are higher for kanji words than for kana words. In order to examine the validity of these assumptions, we attempted to measure the O-P and O-S consistencies for 339 kana words and 775 kanji words. Orthographic neighbors were first generated for each of these words. In order to measure the O-P consistencies of the words, their neighbors were then classified as phonological friends or enemies, based on whether the characters shared with the original word were pronounced the same in the two words. In order to measure the O-S consistencies, the similarity in meaning of each of the neighbors to the original word was rated on a 7-point scale. Based on the ratings, the neighbors were classified as semantic friends or enemies. The results indicated that both the O-P consistencies for kanji words and the O-S consistencies for kana words were greater than previously assumed and that the two scripts were actually quite similar on both types of consistency measures. The implications for the nature of the reading processes for kana and kanji words are discussed.

In the present study, we examined the effects of orthographic and phonological neighborhood sizes for Japanese Katakana words using a lexical decision task. Kawakami (2002) reported an inhibitory orthographic neighborhood size effect along with a null phonological neighborhood size effect in his lexical decision tasks. In contrast, Grainger, Muneaux, Farioli, and Ziegler (2005) reported an interaction between orthographic and phonological neighborhood sizes in a lexical decision task. Therefore, we re-examined the effects of orthographic and phonological neighborhood sizes for low-frequency Katakana words in a lexical decision task. Consistent with Grainger et al., we found the interaction between orthographic and phonological neighborhood sizes, indicating that lexical decision performance for Katakana words is modulated by the nature of orthographic-phonological relationships.

:Effects of the number of meanings (NOM) and the relatedness of those meanings (ROM) were examined for Japanese Katakana words using a lexical-decision task. In Experiment 1, only a NOM advantage was observed. In Experiment 2, the same Katakana words produced a ROM advantage when Kanji words and nonwords were added. Because the Kanji nonwords consisted of unrelated characters whereas the Kanji words consisted of related characters, participants may have used the relatedness of activated meanings as a cue in making lexical decisions in this experiment, artificially creating a ROM advantage for Katakana words. Consistent with this explanation, no ROM effect for Katakana words was observed in Experiment 3 when the Kanji nonwords consisted of characters with similar (i.e., related) meanings. These results pose a further challenge to the position that the speed of semantic coding is modulated by ROM for ambiguous words.

According to parallel distributed processing (PDP) models of visual word recognition, the speed of semantic coding is modulated by the nature of the orthographic-to-semantic mappings. Consistent with this idea, an ambiguity disadvantage and a relatedness-of-meaning (ROM) advantage have been reported in some word recognition tasks in which semantic processing is presumed to be required for responding. To further evaluate this idea, we examined ambiguity and ROM effects in lexical decision and semantic categorization tasks. In the lexical decision task, there was an ambiguity advantage but no ROM effect. In the semantic categorization tasks, we used various semantic categories and found a processing time disadvantage for ambiguous words with less related meanings when the decisions were relatively difficult, but observed no effect when the decisions were easier. These results suggest that both the ambiguity disadvantage and the ROM advantage in the semantic categorization tasks are due to decision-making, rather than semantic-coding, processes. The implications of these results for issues concerning the nature of semantic coding and semantic representations are discussed. (c) 2006 Elsevier Inc. All rights reserved.

An ambiguity disadvantage (slower responses for ambiguous words, e.g., bank, than for unambiguous words) has been reported in semantic tasks (L. R. Gottlob, S. D. Goldinger, G. O. Stone, & G. C. Van Orden, 1999; Y. Hino, S. J. Lupker, & P. M. Pexman, 2002; C. D. Piercey & S. Joordens, 2000) and has been attributed to the meaning activation process. The authors tested an alternative explanation: The ambiguity disadvantage arises from the decision-making process in semantic tasks. The authors examined effects of ambiguity on unrelated trials in a relatedness decision task, because these trials are free from response competition created by ambiguous words on related trials. Results showed no ambiguity effect on unrelated trials (Experiments 2, 3c, and 5c) and an ambiguity disadvantage on related trials (Experiments 3a, 3b, 5a, and 5b).

In these experiments, cross-script masked repetition priming and word frequency effects were examined for Japanese words and nonwords as a function of script familiarity and the nature of the task (lexical decision or naming). In the lexical decision task, masked repetition priming effects were observed only for word targets and those effects were larger for targets presented in an orthographically unfamiliar script than for targets presented in an orthographically familiar script. In contrast, in the naming task, masked repetition priming effects were observed for both word and nonword targets and, for word targets, the repetition priming effects were similar regardless of the orthographic familiarity of the targets. In addition, large word frequency effects were observed when the targets were presented in a familiar script, but the effects were diminished or eliminated when the targets were presented in ail unfamiliar script in both tasks. Implications of these results are discussed in terms of the possible loci of the priming effects in the two tasks. (C) 2002 Elsevier Science (USA). All rights reserved.

The notion of feedback activation from semantics to both orthography and phonology has recently been used to explain a number of semantic effects in visual word recognition, including polysemy effects (Hino Lupker, 1996; Pexman & Lupker, 1999) and synonym effects (Pecher, 2001). In the present research, we tested an account based on feedback activation by investigating a new semantic variable: number of features (NOF). Words with high NOF (e.g., LION) should activate richer semantic representations than do words with low NOF (e.g., LIME). As a result, the feedback activation from semantics to orthographic and phonological representations should be greater for high-NOF words, which should produce superior lexical decision task (LDT) and naming task performance. The predicted facilitory NOF effects were observed in both LDT and naming.

In this article, ambiguity and synonymy effects were examined in lexical decision, naming, and semantic categorization tasks. Whereas the typical ambiguity advantage was observed in lexical decision and naming, an ambiguity disadvantage was observed in semantic categorization. In addition, a synonymy effect (slower latencies for words with many synonyms than for words with few synonyms) was observed in lexical decision and naming but not in semantic categorization. These results suggest that (a) an ambiguity disadvantage arises only when a task requires semantic processing, (b) the ambiguity advantage and the synonymy disadvantage in lexical decision and naming are due to semantic feedback, and (c) these effects are determined by the nature of the feedback relationships from semantics to orthography and phonology.

The effects of word frequency and spelling-to-sound regularity were examined using standard naming, standard lexical-decision, go/no-go naming, and go/no-go lexical-decision tasks. In both the standard and go/no-go naming tasks, tasks requiring phonological coding, a significant Frequency X Regularity interaction was observed. That is, the regularity effect was limited to low-frequency words. In the standard and go/no-go lexical-decision tasks, tasks not requiring phonological coding, no Frequency X Regularity interaction was observed. These results indicate not only that the Frequency X Regularity interaction is a product of phonological coding processes but also that these processes are similar in the standard and go/no-go naming tasks. Results are discussed in terms of the dual-route and the parallel distributed processing frameworks.

How should a word's orthographic neighborhood affect perceptual identification and semantic categorization, both of which require a word to be uniquely identified? According to the multiple read-out model (Grainger & Jacobs, 1996), inhibitory neighborhood frequency effects should be observed in these types of tasks, and facilitatory neighborhood size effects should not be. In Experiments 1 and 2 (perceptual identification), these effects were examined as a function of stimulus visibility (i.e., high vs. low visibility) to provide as full a test as possible of the model's predictions. In the high-visibility conditions, words with large neighborhoods were reported less accurately than words with small neighborhoods, but there was no effect of neighborhood frequency (i.e., whether the word had a higher frequency neighbor). In the low-visibility conditions, low-frequency words with large neighborhoods and low-frequency words with higher frequency neighbors showed superior identification performance. In the semantic categorization task (Experiment 3), words with large neighborhoods were responded to more rapidly than words with small neighborhoods, but there was no effect of neighborhood frequency. These results are inconsistent with two of the basic premises of the multiple read-out model-namely, that facilitatory neighborhood size effects are due to a variable response criterion (the Sigma criterion), rather than to lexical selection processes, and that the lexical selection processes themselves produce an inhibitory neighborhood frequency effect (via the M criterion). Instead, the present results, in conjunction with previous findings, suggest that large neighborhoods (and perhaps higher frequency neighbors) do aid lexical selection.

Recent research in visual word recognition suggests that the speed with which a word is identified is influenced by the reader's knowledge of other, orthographically similar words (Andrews, 1997). In serial-search and activation-based models of word recognition, mental representations of these "orthographic neighbours" of a word are explicitly assumed to play a role in the lexical selection process. Thus, it has been possible to determine the specific predictions that these models make about the effects of orthographic neighbours and to test a number of those predictions empirically. In contrast, the role of orthographic neighbours in parallel distributed processing models (e.g., Plaut, McClelland, Seidenberg, & Patterson, 1996; Seidenberg & McClelland, 1989) is less clear. In this paper, several statistical analyses of error scores from these types of models revealed that low frequency words with large neighbourhoods had lower orthographic, phonological, and cross-entropy error scores than low frequency words with small neighbourhoods; and that low frequency words with higher frequency neighbours had lower error scores than low frequency words without higher frequency neighbours. According to these models then, processing should be more rapid for low frequency words with large neighbourhoods and for low frequency words with higher frequency neighbours.

The effects of word frequency were examined for Japanese Kanji and Katakana words in 6 experiments. The sizes of frequency effects were comparable for Kanji and Katakana words in the standard lexical decision task. In the standard naming task, the frequency effect for Katakana words was significantly smaller than that for Kanji words. These results were consistent with the lexical-selection account of frequency effects offered by dual-route models. Contrary to this account, however, frequency effects were smaller for Katakana words than for Kanji words in go/no-go naming tasks, in which participants were asked to name a stimulus aloud only if it was a word. This Frequency X Script Type interaction was not the result of using a go/no-go task because the interaction disappeared in the go/no-go lexical decision task. These results pose a strong challenge for the lexical-selection account of frequency effects offered by dual-route models.

In these experiments, the effects of polysemy were examined as a function of word frequency for Japanese katakana words, words which have consistent character-to-sound correspondences. In the lexical decision task, an additive relationship was observed between polysemy and frequency (i.e., polysemy effects were identical for high and low frequency katakana words). In the naming task, although no word frequency effect was observed, there was a significant polysemy effect which, as in the lexical decision task, was identical for high and low frequency words. The implications of these results for conclusions about the loci of polysemy and frequency effects in lexical decision and naming tasks are discussed.

The effects of priming by ambiguous, auditorily presented word primes were examined. In related conditions, primes were followed by either associatively related or semantically related but associatively unrelated targets. When the targets were presented at prime offset (Experiment 1), priming effects were observed only for associatively related targets, independent of meaning frequency (i.e., whether the target was related to the dominant or subordinate meaning of the ambiguous prime). When the targets were presented after a 700 ms delay experiment 2), however, priming effects were observed only for targets related to the prime's dominant meaning, regardless of the nature of the prime-target relation. These results raise the strong possibility that previously reported differences in the nature of priming effects that had been ascribed to meaning frequency might actually be due to differences in associative strength. These results are discussed in terms of Fodor's (1983; 1990) "anti-semantic" modularity view.

The effects of polysemy (number of meanings) and word frequency were examined in lexical decision and naming tasks. Polysemy effects were observed in both tasks. In the lexical decision task, high- and low-frequency words produced identical polysemy effects. In the naming task, however, polysemy interacted with frequency, with polysemy effects being limited to low-frequency words. When degraded stimuli were used in both tasks, the interaction appeared not only in naming but also in lexical decision. Because stimulus degradation also produced an effect of spelling-sound regularity in the lexical decision task, the different relationships between polysemy and frequency appear to be due to whether responding was based primarily on orthographic or phonological codes. As such, the effects of polysemy seem to be due to the nature of task-specific processes. An explanation in terms of M. S. Seidenberg and J. L. McClelland's (1989) and D. C. Plaut and J. L. McClelland's (1993) parallel distributed processing models is proposed.

What are the effects of a word's orthographic neighborhood on the word recognition process? Andrews (1989) reported that large neighborhoods facilitate lexical access (the neighborhood size effect). Grainger, O'Regan, Jacobs, & Segui (1989) reported that higher frequency neighbors inhibit lexical access (the ''neighborhood frequency effect''). Because neighborhood size and neighborhood frequency typically covary (words with large neighborhoods will usually possess higher frequency neighbors), these findings would seem to contradict one another. In the present study, 6 experiments on the effects of neighborhood size and neighborhood frequency indicated that, at least for low-frequency words, large neighborhoods do facilitate processing. However, the existence of higher frequency neighbors seems to facilitate rather than inhibit processing. The implications of these findings for serial and parallel models of lexical access are discussed.