- Email: [email protected]
Kanji Lexicality Effect in Partial Repetition Priming: The Relationship between Kanji Word and Kanji Character Processing
Brain and Language 68, 82–88 (1999) Article ID brln.1999.2111, available online at http://www.idealibrary.com on
Kanji Lexicality Effect in Partial Repetition Priming: The Relationship between Kanji Word and Kanji Character Processing Michiko Shimomura Media Center, Osaka City University, Japan Two experiments are described using a partial repetition priming paradigm. The prime type was whether a kanji character is used as a word by itself (word type) or not (nonword type). This property is called kanji lexicality. In Experiment 1, the effect of kanji lexicality was not confirmed, although there was a tendency in the responses for word-type primes to be slower than those for nonword-type primes. In Experiment 2, the kanji lexicality effect depended upon the target familiarity, and no facilitation was obtained for high-familiarity words with word-type primes. The results were interpreted in terms of a modified interactive-activation framework. 1999 Academic Press
Key Words: Kanji words; lexicality; partial repetition priming.
INTRODUCTION
When the prime and the target are orthographically overlapped, a slightly inhibitory effect has been reported for high-frequency targets (Colombo, 1986). This inhibitory effect has been interpreted in an interactive-activation framework (McClelland & Rumelhart, 1981; Taft, 1991, 1994). That is, two mechanisms of facilitatory and inhibitory activation are assumed to be operative at each level. Facilitatory activation spread to nodes at an adjacent level (e.g., from letters to words), and never spread to the same level. Inhibition, on the other hand, occurs between adjacent levels and within the same level in terms of resolution process (Forster & Veres, 1998). The study reported here concerns the activation process underlying lexical access to words written in a Japanese logographic system, namely, kanji I am grateful to Yoshiaki Nakajima, Toshiaki Miura, and Seiki Akai at Osaka University for their useful comments on this research. I also thank the reviewers for their helpful comments on an earlier draft of the manuscript. Requests for reprints should be addressed to Michiko Shimomura, Media Center, Osaka City University, 3-3-138, Sugimoto, Sumiyoshi-ku, Osaka 558-8585, Japan. E-mail: [email protected]. 82 0093-934X/99 $30.00 Copyright 1999 by Academic Press All rights of reproduction in any form reserved.
KANJI LEXICALITY EFFECT
83
words.1 Many studies have been done to clarify the role of orthographic, phonological, and semantic processing in the lexical access of kanji (e.g., Flores d’Arcais, Saito, & Kawakami, 1995; Sakuma, Sasanuma, Tatsumi, & Masaki, 1998; Wydell, Butterworth, & Patterson, 1995). In this paper, the orthographic processing of kanji characters was investigated by focusing on the following property of kanji characters: some kanji characters can appear by themselves as words (word-type) in addition to as constituents of kanji words consisting of some characters; whereas some characters appear only as constituents of words (nonword-type). I call this property kanji lexicality and hypothesize that the relationship between kanji character processing and kanji word processing should vary with kanji lexicality. Nonword-type kanji characters are assumed to be at the adjacent lower level to the kanji-word level in the interactive-activation framework. I hypothesize, on the other hand, that word-type kanji characters have two functions: one as words and the other as constituents. As mentioned above, nodes at the same level should inhibit each other. Thus, if word-type kanji characters and kanji words are at the same level, word-type kanji characters might inhibit the processing of kanji words that include them. I examined this assumption in two partial repetition priming experiments. EXPERIMENT 1
The purpose of Experiment 1 was to compare the partial repetition priming effect with word-type primes and nonword-type primes. In accordance with the earlier mentioned assumption, word-type primes should produce the smaller priming effect. Method Participants. A total of 31 students from Nara National College of Technology participated in the experiment. Materials and design. All words used in this experiment consisted of two kanji characters. The target items were 20 kanji words. For each word, the first or second character was presented as a prime. For half of the words, the first characters were word-type and the second characters were nonword-type. For the rest of the words, the first characters were nonwordtype and the second characters were word-type. Therefore, there were two types of primes concerning kanji lexicality (word-type and nonword-type), which were constructed for each word and appeared in different positions in target words. Neutral primes (symbols) were also used. Sample stimuli are shown in Table 1. Twenty nonwords consisting of two kanji characters were used as foils. Each target item appeared three times in an experiment consisting of 120 trials. In this experiment, there were two conditions for the schedule of stimulus presentation. In one condition, primes were presented for 100 ms and targets appeared after a 300 ms blank display. Therefore, the stimulus onset asynchrony (SOA) was 400 ms in this condition (long condition). In the other condition, primes were displayed for 100 ms and targets appeared
1
‘‘Kanji words’’ hereafter represent words consisting of some kanji characters.
84
MICHIKO SHIMOMURA
TABLE 1 Sample Stimuli Used in Experiment 1
immediately following the primes. In this case, the SOA was 100 ms (short condition). Sixteen participants were tested in the long condition, and 15 participants were tested in the short condition. Procedure. Each trial consisted of three stimuli. The first was an asterisk displayed for 1 s. Immediately following this was the prime. Finally, the targets were presented until the participants responded. All stimuli were presented at the center of the display. The presentation conditions of primes and targets were as previously mentioned. The experiment was controlled by a Power Macintosh 7600 (Apple Computer). Participants were instructed to make lexical decisions on the targets by pressing the predesignated keys. Items were presented in a pseudorandom sequence, with a different order for each participant.
Results The two presentation conditions were analyzed separately. The grand mean lexical decision times are shown in Table 2. A one-way ANOVA was carried out for the prime type (word-type, nonword-type, and neutral) by subjects (F 1) and by items (F 2). The effect of the prime type was significant by subjects: F 1 (2, 30) ⫽ 33.8, p ⬍ .001 in the long condition, F 1 (2, 28) ⫽ 5.9, p ⬍ .01 in the short condition. The effect was also significant by items: F 2 (2, 38) ⫽ 22.6, p ⬍ .001 in the long condition, F 2 (2, 38) ⫽ 4.0, p ⬍ .05 in the short condition. Newman–Keuls analyses showed that both word-type primes and nonword-type primes significantly facilitated the responses relative to neutral primes in the long condition. Furthermore, nonword-type primes yielded faster responses than word-type primes. In the short condiTABLE 2 Mean Lexical Decision Times (in Milliseconds) for Word Targets Presentation condition Long Short
Prime type Word type
Nonword type
Neutral
522 558
507 545
573 577
Note. The lexical decision times of the different prime positions in the targets (the first and second position) were averaged.
KANJI LEXICALITY EFFECT
85
TABLE 3 Sample Stimuli Used in Experiment 2
tion, nonword-type primes significantly facilitated the responses relative to neutral primes. However, the facilitation of word-type primes was marginally significant by subjects ( p ⫽ .05) and not significant by items in the short condition. The difference between word-type and nonword-type primes was not significant either by subjects or by items in the short condition. To summarize, although there was a tendency in the responses for wordtype primes to be slower than those for nonword-type primes, the difference was not statistically significant. One possible reason is that the word familiarity was not controlled in this experiment. Thus, target familiarity was manipulated in Experiment 2. EXPERIMENT 2
In Experiment 2, both prime lexicality and target familiarity were manipulated. Targets were presented immediately after primes disappeared as in the short condition in Experiment 1. Method Participants. A total of 24 students from Nara National College of Technology participated in the experiment. Materials and design. The target items were 40 two-kanji-character words. For each word, the first character was presented as a prime, and all of these primes were high-frequency characters.2 There were two experimental conditions. The first condition was target familiarity3: half of the words were high familiarity and the rest were low familiarity. The second was prime type, i.e., word-type and nonword-type. For each target, neutral primes (symbols) were presented as baselines. Thus, each word appeared twice in the experiment. A total of 40 nonwords were also used as foils. Sample stimuli are shown in Table 3. Procedure. Each trial consisted of three stimuli. The first was an asterisk displayed for 1 s. Immediately following this, the prime was presented for 133 ms. Finally, a hash mark (#) and the target were presented at the same time until the participants responded. The hash mark 2 Frequency data of kanji characters were found on the WWW site of The National Language Research Institute in Japan when I prepared this experiment. 3 Target familiarity was measured by participants’ rating.
86
MICHIKO SHIMOMURA
TABLE 4 Mean Lexical Decision Times and Facilitation (in Milliseconds) for Word Targets Target familiarity High Low
Prime type Word type
Neutral
Facilitation
Nonword type
Neutral
Facilitation
535 605
540 673
5 68
529 597
558 631
29 34
was displayed at the same position of the prime as a masking stimulus, and the target was presented just below the hash mark. Participants were instructed to make lexical decisions on the targets by pressing the predesignated keys.
Results The grand mean lexical decision times are shown in Table 4. The following analyses evaluated the facilitation effect. In the results of a two-way ANOVA, the main effect of target familiarity was significant by subjects: F 1 (1, 23) ⫽ 8.0, p ⬍ .01, but not significant by items. No effect of prime type was obtained either by subjects or by items. The interaction between target familiarity and prime type was marginally significant by subjects, F 1 (1, 23) ⫽ 4.2, p ⫽ .05, and by items, F 2 (1, 9) ⫽ 4.5, p ⫽ .06. For word familiarity, a tendency consistent with some previous studies (e.g., Colombo. 1986) was obtained where low-familiarity words showed slower responses and more facilitation relative to high-familiarity words, even though the difference of facilitation was not statistically significant by items. Of interest is the interaction. The separate analyses of target familiarity showed no significant difference between word-type and nonword-type primes. Moreover, no difference was obtained between high-familiarity and low-familiarity words for the nonword-type prime condition in separate analyses of the prime type. A significant difference was shown only between high-familiarity and low-familiarity words in the word-type prime condition. That is, when the primes were word-type, high-familiarity words were not facilitated at all, although nonword-type primes facilitated targets. These results indicate that the effect of prime lexicality depended upon the target familiarity. DISCUSSION
Two experiments were carried out to evaluate the lexicality effect of kanji characters on kanji word recognition. In Experiment 1, there was a tendency that lexical decisions were faster for nonword-type primes than for wordtype primes, although the difference was not statistically significant when the SOA was short, and facilitation relative to neutral condition was produced regardless of prime type. Thus, evidence of an inhibitory connection between
KANJI LEXICALITY EFFECT
87
word-type characters and words could not be confirmed. In Experiment 2, the material design and the experimental procedure were improved, and the effect of prime lexicality depended upon the target familiarity. In particular, when the primes were word type and the targets were high familiarity, no facilitation was obtained. In an attempt to explain the effect of prime lexicality in the experiment, I suggest a modified version of the interactive-activation framework. The kanji words used here are supposed to be equivalent to English words. The status of kanji characters depends upon their lexicality. Nonword-type characters are at the adjacent level to kanji-word level and activate nodes of kanji words. On the contrary, word-type characters are supposed to have two representations: at the same level as words and at the adjacent level. This is because word-type characters have two functions: as words and as constituents. Kanji character nodes at the adjacent level should spread facilitatory activation to nodes at the kanji-word level, while nodes of kanji words and those of kanji characters at the same level should send inhibitory activation to each other. Thus, the net priming effect with word-type primes may be obtained by subtracting the inhibitory effect from the facilitatory effect. The results in Experiment 2 are interpreted in this interactive-activation framework as follows. Nonword-type primes spread facilitatory activation to kanji words regardless of word familiarity, so the same extent of facilitation was obtained for high-familiarity and low-familiarity targets. On the contrary, word-type primes send both facilitatory and inhibitory activation to kanji words. At the word level, word-type kanji characters and kanji words should inhibit each other, and the processing of high-familiarity words might be strongly suppressed by the word-type primes relative to low-familiarity words (e.g., Colombo, 1986; Forster & Veres, 1998, in more detail). In summary, when taken together, the findings suggest that kanji lexicality may affect kanji word processing. The modified interactive-activation framework was proposed to interpret the results, in which word-type kanji characters have two representations, as words and as constituents. Further research is required to evaluate the interactive-activation account mentioned above. REFERENCES Colombo, L. 1986. Activation and inhibition with orthographically similar words. Journal of Experimental Psychology: Human Perception and Performance, 12, 226–234. Flores d’Arcais, G. B., Saito, H., & Kawakami, M. 1995. Phonological and semantic activation in reading kanji characters. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 34–42. Forster, K. I., & Veres, C. 1998. The prime lexicality effect: Form-priming as a function of prime awareness, lexical status, and discrimination difficulty. Journal of Experimental Psychology: Learning, Memory, and Cognition, 24, 498–514. McClelland, J. L., & Rumelhart, D. E. 1981. An interactive activation model of context effects
88
MICHIKO SHIMOMURA
in letter perception: Part 1. An account of basic findings. Psychological Review, 88, 375– 407. Sakuma, N., Sasanuma, S., Tatsumi, I. F., & Masaki, S. 1998. Orthography and phonology in reading Japanese kanji words: Evidence from the semantic decision task with homophones. Memory and Cognition, 26, 75–87. Taft, M. 1991. Reading and the mental lexicon. Hove: Erlbaum. Taft, M. 1994. Interactive-activation as a framework for understanding morphological processing. In D. Sandra and M. Taft (Eds.), Morphological structure, lexical representation and lexical access. Hove: Erlbaum. Wydell, T., Butterworth, B., & Patterson, K. 1995. The inconsistency of consistency effects in reading: The case of Japanese Kanji. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 1155–1168.
Kanji Lexicality Effect in Partial Repetition Priming: The Relationship between Kanji Word and Kanji Character Processing Michiko Shimomura Media Center, Osaka City University, Japan Two experiments are described using a partial repetition priming paradigm. The prime type was whether a kanji character is used as a word by itself (word type) or not (nonword type). This property is called kanji lexicality. In Experiment 1, the effect of kanji lexicality was not confirmed, although there was a tendency in the responses for word-type primes to be slower than those for nonword-type primes. In Experiment 2, the kanji lexicality effect depended upon the target familiarity, and no facilitation was obtained for high-familiarity words with word-type primes. The results were interpreted in terms of a modified interactive-activation framework. 1999 Academic Press
Key Words: Kanji words; lexicality; partial repetition priming.
INTRODUCTION
When the prime and the target are orthographically overlapped, a slightly inhibitory effect has been reported for high-frequency targets (Colombo, 1986). This inhibitory effect has been interpreted in an interactive-activation framework (McClelland & Rumelhart, 1981; Taft, 1991, 1994). That is, two mechanisms of facilitatory and inhibitory activation are assumed to be operative at each level. Facilitatory activation spread to nodes at an adjacent level (e.g., from letters to words), and never spread to the same level. Inhibition, on the other hand, occurs between adjacent levels and within the same level in terms of resolution process (Forster & Veres, 1998). The study reported here concerns the activation process underlying lexical access to words written in a Japanese logographic system, namely, kanji I am grateful to Yoshiaki Nakajima, Toshiaki Miura, and Seiki Akai at Osaka University for their useful comments on this research. I also thank the reviewers for their helpful comments on an earlier draft of the manuscript. Requests for reprints should be addressed to Michiko Shimomura, Media Center, Osaka City University, 3-3-138, Sugimoto, Sumiyoshi-ku, Osaka 558-8585, Japan. E-mail: [email protected]. 82 0093-934X/99 $30.00 Copyright 1999 by Academic Press All rights of reproduction in any form reserved.
KANJI LEXICALITY EFFECT
83
words.1 Many studies have been done to clarify the role of orthographic, phonological, and semantic processing in the lexical access of kanji (e.g., Flores d’Arcais, Saito, & Kawakami, 1995; Sakuma, Sasanuma, Tatsumi, & Masaki, 1998; Wydell, Butterworth, & Patterson, 1995). In this paper, the orthographic processing of kanji characters was investigated by focusing on the following property of kanji characters: some kanji characters can appear by themselves as words (word-type) in addition to as constituents of kanji words consisting of some characters; whereas some characters appear only as constituents of words (nonword-type). I call this property kanji lexicality and hypothesize that the relationship between kanji character processing and kanji word processing should vary with kanji lexicality. Nonword-type kanji characters are assumed to be at the adjacent lower level to the kanji-word level in the interactive-activation framework. I hypothesize, on the other hand, that word-type kanji characters have two functions: one as words and the other as constituents. As mentioned above, nodes at the same level should inhibit each other. Thus, if word-type kanji characters and kanji words are at the same level, word-type kanji characters might inhibit the processing of kanji words that include them. I examined this assumption in two partial repetition priming experiments. EXPERIMENT 1
The purpose of Experiment 1 was to compare the partial repetition priming effect with word-type primes and nonword-type primes. In accordance with the earlier mentioned assumption, word-type primes should produce the smaller priming effect. Method Participants. A total of 31 students from Nara National College of Technology participated in the experiment. Materials and design. All words used in this experiment consisted of two kanji characters. The target items were 20 kanji words. For each word, the first or second character was presented as a prime. For half of the words, the first characters were word-type and the second characters were nonword-type. For the rest of the words, the first characters were nonwordtype and the second characters were word-type. Therefore, there were two types of primes concerning kanji lexicality (word-type and nonword-type), which were constructed for each word and appeared in different positions in target words. Neutral primes (symbols) were also used. Sample stimuli are shown in Table 1. Twenty nonwords consisting of two kanji characters were used as foils. Each target item appeared three times in an experiment consisting of 120 trials. In this experiment, there were two conditions for the schedule of stimulus presentation. In one condition, primes were presented for 100 ms and targets appeared after a 300 ms blank display. Therefore, the stimulus onset asynchrony (SOA) was 400 ms in this condition (long condition). In the other condition, primes were displayed for 100 ms and targets appeared
1
‘‘Kanji words’’ hereafter represent words consisting of some kanji characters.
84
MICHIKO SHIMOMURA
TABLE 1 Sample Stimuli Used in Experiment 1
immediately following the primes. In this case, the SOA was 100 ms (short condition). Sixteen participants were tested in the long condition, and 15 participants were tested in the short condition. Procedure. Each trial consisted of three stimuli. The first was an asterisk displayed for 1 s. Immediately following this was the prime. Finally, the targets were presented until the participants responded. All stimuli were presented at the center of the display. The presentation conditions of primes and targets were as previously mentioned. The experiment was controlled by a Power Macintosh 7600 (Apple Computer). Participants were instructed to make lexical decisions on the targets by pressing the predesignated keys. Items were presented in a pseudorandom sequence, with a different order for each participant.
Results The two presentation conditions were analyzed separately. The grand mean lexical decision times are shown in Table 2. A one-way ANOVA was carried out for the prime type (word-type, nonword-type, and neutral) by subjects (F 1) and by items (F 2). The effect of the prime type was significant by subjects: F 1 (2, 30) ⫽ 33.8, p ⬍ .001 in the long condition, F 1 (2, 28) ⫽ 5.9, p ⬍ .01 in the short condition. The effect was also significant by items: F 2 (2, 38) ⫽ 22.6, p ⬍ .001 in the long condition, F 2 (2, 38) ⫽ 4.0, p ⬍ .05 in the short condition. Newman–Keuls analyses showed that both word-type primes and nonword-type primes significantly facilitated the responses relative to neutral primes in the long condition. Furthermore, nonword-type primes yielded faster responses than word-type primes. In the short condiTABLE 2 Mean Lexical Decision Times (in Milliseconds) for Word Targets Presentation condition Long Short
Prime type Word type
Nonword type
Neutral
522 558
507 545
573 577
Note. The lexical decision times of the different prime positions in the targets (the first and second position) were averaged.
KANJI LEXICALITY EFFECT
85
TABLE 3 Sample Stimuli Used in Experiment 2
tion, nonword-type primes significantly facilitated the responses relative to neutral primes. However, the facilitation of word-type primes was marginally significant by subjects ( p ⫽ .05) and not significant by items in the short condition. The difference between word-type and nonword-type primes was not significant either by subjects or by items in the short condition. To summarize, although there was a tendency in the responses for wordtype primes to be slower than those for nonword-type primes, the difference was not statistically significant. One possible reason is that the word familiarity was not controlled in this experiment. Thus, target familiarity was manipulated in Experiment 2. EXPERIMENT 2
In Experiment 2, both prime lexicality and target familiarity were manipulated. Targets were presented immediately after primes disappeared as in the short condition in Experiment 1. Method Participants. A total of 24 students from Nara National College of Technology participated in the experiment. Materials and design. The target items were 40 two-kanji-character words. For each word, the first character was presented as a prime, and all of these primes were high-frequency characters.2 There were two experimental conditions. The first condition was target familiarity3: half of the words were high familiarity and the rest were low familiarity. The second was prime type, i.e., word-type and nonword-type. For each target, neutral primes (symbols) were presented as baselines. Thus, each word appeared twice in the experiment. A total of 40 nonwords were also used as foils. Sample stimuli are shown in Table 3. Procedure. Each trial consisted of three stimuli. The first was an asterisk displayed for 1 s. Immediately following this, the prime was presented for 133 ms. Finally, a hash mark (#) and the target were presented at the same time until the participants responded. The hash mark 2 Frequency data of kanji characters were found on the WWW site of The National Language Research Institute in Japan when I prepared this experiment. 3 Target familiarity was measured by participants’ rating.
86
MICHIKO SHIMOMURA
TABLE 4 Mean Lexical Decision Times and Facilitation (in Milliseconds) for Word Targets Target familiarity High Low
Prime type Word type
Neutral
Facilitation
Nonword type
Neutral
Facilitation
535 605
540 673
5 68
529 597
558 631
29 34
was displayed at the same position of the prime as a masking stimulus, and the target was presented just below the hash mark. Participants were instructed to make lexical decisions on the targets by pressing the predesignated keys.
Results The grand mean lexical decision times are shown in Table 4. The following analyses evaluated the facilitation effect. In the results of a two-way ANOVA, the main effect of target familiarity was significant by subjects: F 1 (1, 23) ⫽ 8.0, p ⬍ .01, but not significant by items. No effect of prime type was obtained either by subjects or by items. The interaction between target familiarity and prime type was marginally significant by subjects, F 1 (1, 23) ⫽ 4.2, p ⫽ .05, and by items, F 2 (1, 9) ⫽ 4.5, p ⫽ .06. For word familiarity, a tendency consistent with some previous studies (e.g., Colombo. 1986) was obtained where low-familiarity words showed slower responses and more facilitation relative to high-familiarity words, even though the difference of facilitation was not statistically significant by items. Of interest is the interaction. The separate analyses of target familiarity showed no significant difference between word-type and nonword-type primes. Moreover, no difference was obtained between high-familiarity and low-familiarity words for the nonword-type prime condition in separate analyses of the prime type. A significant difference was shown only between high-familiarity and low-familiarity words in the word-type prime condition. That is, when the primes were word-type, high-familiarity words were not facilitated at all, although nonword-type primes facilitated targets. These results indicate that the effect of prime lexicality depended upon the target familiarity. DISCUSSION
Two experiments were carried out to evaluate the lexicality effect of kanji characters on kanji word recognition. In Experiment 1, there was a tendency that lexical decisions were faster for nonword-type primes than for wordtype primes, although the difference was not statistically significant when the SOA was short, and facilitation relative to neutral condition was produced regardless of prime type. Thus, evidence of an inhibitory connection between
KANJI LEXICALITY EFFECT
87
word-type characters and words could not be confirmed. In Experiment 2, the material design and the experimental procedure were improved, and the effect of prime lexicality depended upon the target familiarity. In particular, when the primes were word type and the targets were high familiarity, no facilitation was obtained. In an attempt to explain the effect of prime lexicality in the experiment, I suggest a modified version of the interactive-activation framework. The kanji words used here are supposed to be equivalent to English words. The status of kanji characters depends upon their lexicality. Nonword-type characters are at the adjacent level to kanji-word level and activate nodes of kanji words. On the contrary, word-type characters are supposed to have two representations: at the same level as words and at the adjacent level. This is because word-type characters have two functions: as words and as constituents. Kanji character nodes at the adjacent level should spread facilitatory activation to nodes at the kanji-word level, while nodes of kanji words and those of kanji characters at the same level should send inhibitory activation to each other. Thus, the net priming effect with word-type primes may be obtained by subtracting the inhibitory effect from the facilitatory effect. The results in Experiment 2 are interpreted in this interactive-activation framework as follows. Nonword-type primes spread facilitatory activation to kanji words regardless of word familiarity, so the same extent of facilitation was obtained for high-familiarity and low-familiarity targets. On the contrary, word-type primes send both facilitatory and inhibitory activation to kanji words. At the word level, word-type kanji characters and kanji words should inhibit each other, and the processing of high-familiarity words might be strongly suppressed by the word-type primes relative to low-familiarity words (e.g., Colombo, 1986; Forster & Veres, 1998, in more detail). In summary, when taken together, the findings suggest that kanji lexicality may affect kanji word processing. The modified interactive-activation framework was proposed to interpret the results, in which word-type kanji characters have two representations, as words and as constituents. Further research is required to evaluate the interactive-activation account mentioned above. REFERENCES Colombo, L. 1986. Activation and inhibition with orthographically similar words. Journal of Experimental Psychology: Human Perception and Performance, 12, 226–234. Flores d’Arcais, G. B., Saito, H., & Kawakami, M. 1995. Phonological and semantic activation in reading kanji characters. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 34–42. Forster, K. I., & Veres, C. 1998. The prime lexicality effect: Form-priming as a function of prime awareness, lexical status, and discrimination difficulty. Journal of Experimental Psychology: Learning, Memory, and Cognition, 24, 498–514. McClelland, J. L., & Rumelhart, D. E. 1981. An interactive activation model of context effects
88
MICHIKO SHIMOMURA
in letter perception: Part 1. An account of basic findings. Psychological Review, 88, 375– 407. Sakuma, N., Sasanuma, S., Tatsumi, I. F., & Masaki, S. 1998. Orthography and phonology in reading Japanese kanji words: Evidence from the semantic decision task with homophones. Memory and Cognition, 26, 75–87. Taft, M. 1991. Reading and the mental lexicon. Hove: Erlbaum. Taft, M. 1994. Interactive-activation as a framework for understanding morphological processing. In D. Sandra and M. Taft (Eds.), Morphological structure, lexical representation and lexical access. Hove: Erlbaum. Wydell, T., Butterworth, B., & Patterson, K. 1995. The inconsistency of consistency effects in reading: The case of Japanese Kanji. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 1155–1168.