J Psycholinguist Res (2007) 36:47–63 DOI 10.1007/s10936-006-9032-9 O R I G I NA L PA P E R

Looks aren’t Everything: Pseudohomophones Prime Words but Nonwords do not Laree A. Huntsman

Published online: 13 January 2007 © Springer Science+Business Media, LLC 2007

Abstract After examining literature that deals with phonological and orthographic effects associated with pseudohomophones, the current effort deviates from the norm by using fewer pseudohomophones (20%) and extending the lags between primes and targets ( M = 8). Word and pseudohomophone primes were found to facilitate lexical decision response latencies to word targets. Response latencies to word targets were not influenced by nonword primes, however. The presence of pseudohomophone effects was demonstrated by longer response latencies and higher error rates for pseudohomophones (e.g., DREEM) that were equated in orthography to nonword controls (e.g., DROAM). Despite the frequency effect observed for base words, the pseudohomophones did not exhibit an effect of base word frequency. The results suggest that phonological codes exert an influence on lexical representation but are not frequency sensitive. Keywords Psycholinguistics · Reading · Word recognition · Phonology · Pseudohomophones · Repetition priming

Introduction The relative significance, independence, and speed of operation of orthographic and phonological codes for contacting items in lexical memory has been debated extensively in the literature for over 30 years (see Coltheart, Davelaar, Jonasson, & Besner, 1977; Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001; Lukatela, Lukatela, & Turvey, 1993; Huntsman, 1998). Pseudohomophones, which are nonwords that are homophonic with words (e.g., KNEAL for KNEEL) have played a special role in this debate. In one of the first pseudohomophone experiments, MacKay (1972) asked

L. A. Huntsman (B) Department of Psychology, San Jose State University, San Jose, CA 95192-0120, USA e-mail: [email protected]

48

J Psycholinguist Res (2007) 36:47–63

participants to proofread sentences for spelling errors. In this experiment, two types of spelling errors were used: pseudohomophones (e.g., WERK for WORK) and nonhomophonic nonwords which had been equated for orthographic equality (e.g., WARK for WORK). Results indicated that the nonwords were identified as misspelled words more frequently than were the pseudohomophones. Likewise, participants performing the lexical decision task take more time to correctly reject pseudohomophones (e.g., KEAP for KEEP) than to reject control nonwords such as KEAB (Rubenstein, Lewis, & Rubenstein, 1971; Coltheart et al., 1977; Patterson & Marcel, 1977; Barron, 1978; Cohen & Freeman, 1978; Barry, 1981; McQuade, 1981; Pring, 1981; Mitterer, 1982; Besner & Davelaar, 1983). Martin (1982) matched pseudohomophones to nonwords that were as orthographically similar to words as were the pseudohomophones. For example, the ‘O’ was changed to ‘E’ in WORD resulting in the pseudohomophone WERD. The nonword was created by changing the ‘O’ to ‘E’ in COST resulting in CEST. Using speeded lexical decision trials with the pseudohomophones constituting 25% of the nonwords, Martin found that the response latencies to pseudohomophones were longer than those to pronounceable nonwords which were not matched on visual similarity to words. However, response latencies to the visually similar control nonwords were not significantly longer. Taft (1982) matched pseudohomophones on phonological similarity to their base words but varied orthographic similarity. For example, SKREAM is different from SCREAM by a one-letter change and SKREME is different from SCREAM by a twoletter change. It is known that pseudohomophones are processed more efficiently than nonwords in naming (McCann & Besner, 1987) but less efficiently than nonwords in lexical decision (Besner & Davelaar, 1983; McCann, Besner, & Davelaar, 1988). If the pseudohomophone effect is influenced by the orthographic similarity of the pseudohomophones to words, then the orthographically more-similar one-letter change pseudohomophones would be expected to have shorter response latencies and fewer errors than the orthographically less-similar two-letter change pseudohomophones when the phonological lexical decision task is used. Conversely, when the lexical decision task is employed, the effect would be expected to reverse. Indeed, Taft’s results indicated that one-letter change pseudohomophones (e.g., SKREAM) yielded faster phonological lexical decision (“does this item sound like an English word?”) response latencies and fewer errors than did two-letter change pseudohomophones (e.g., SKREME). In a semantic categorization task, Van Orden, Johnston, & Hale (1987) found that false positive error rates were elevated when participants responded to pseudohomophone foils of category exemplars (e.g., SUTE for the clothing category) than when they responded to nonword foils which had been equated for orthographic similarity (e.g., SURT for the clothing category). In addition, false positive errors occurred equally often for pseudohomophones (e.g., SUTE) and words that were homophonic heterographs (e.g., HARE of the pair HARE and HAIR). Pseudohomophone repetition effects Using a short-term priming procedure where prime and target trials were separated by a mean lag of two intervening trials, Besner, Dennis, & Davelaar (1985) examined priming effects for pseudohomophone and nonword primes that had been equated for orthographic similarity. Pseudohomophone primes exerted a priming effect on lexical

J Psycholinguist Res (2007) 36:47–63

49

decision response latencies over and above any priming due to the orthographic similarity of the nonword primes. Perfetti, Bell, & Delaney (1988) in a backward priming experiment found that identification of a target word (e.g., HEAR) exhibited a processing advantage if it was followed by a pseudohomophone that was orthographically similar to the masking stimulus (e.g., HEER) than if it was followed by a masking stimulus that was only orthographically similar (e.g., HEOR). It has also been found (?) that naming latencies to word targets (e.g., CRUISE) were faster if they had been primed by pseudohomophones (e.g., KROOZE) in a prior lexical decision task. Using both long and short stimulus onset asynchronies, Lukatela & Turvey (1991) found that pseudohomophones (e.g., TAYBLE) and words (e.g., TABLE) facilitated naming latencies for associated word targets (e.g., CHAIR), relative to spelling control nonword primes (e.g., TARBLE). Other studies using pattern masking and eye movement rapid priming paradigms have not found evidence of a phonological relationship between pseudohomophone primes and word targets. Using a repetition priming procedure where primes, that for the most part went undetected by participants, were preceded and followed by a pattern mask (Humphreys, Evett, & Taylor, 1982), it was found that homophone primes produced significant priming effects (e.g., MAID primed MADE), but pseudohomophone primes did not (e.g., BLOO did not prime BLUE). Using an eye movement rapid priming paradigm, Rayner, Sereno, Lesch, & Pollatsek (1995) reported that fixation times to targets (e.g., BEACH) embedded in sentences were facilitated by homophone primes (e.g., BEECH) when compared to visually similar control primes (e.g., BENCH). However, fixation times to targets (e.g., WEED) were not facilitated by pseudohomophone primes (e.g., WEAD) when compared to controls. Are pseudohomophones frequency sensitive? The frequency with which a word occurs in English text is a predictor of response latency in tasks such as lexical decision and naming (e.g., Forster & Chambers, 1973; Frederiksen & Kroll, 1976; Whaley, 1978; Balota & Chumbley, 1984), but no effect of pseudohomophone base word frequency was found by McCann & Besner (1987) who examined naming latencies for pseudohomophones (e.g., TRAX), their corresponding base words (e.g., TRACKS), and nonwords that had been equated for orthographic similarity with the pseudohomophones (e.g., PRAX). While the participants named the pseudohomophones faster relative to the nonword controls, no effect of base word frequency was found on pseudohomophone naming latencies even though a frequency effect was observed on naming latencies of the base words themselves. Furthermore, McCann, Besner, & Davelaar (1988) reported that while the pseudohomophones elicited slower lexical decision responses relative to the nonword controls, no effect of base word frequency was found on pseudohomophone lexical decision response latencies even though a frequency effect was observed on the latencies of the base words themselves. Only when employing a phonological lexical decision task (does this letter string sound like an English word?), were these researchers able to find responses to pseudohomophones to be sensitive to base word frequency. Because frequency effects have been found to be more apparent in the lexical decision task than in the naming task, it is possible that pseudohomophone base word frequency effects may become evident in a study utilizing the lexical decision task. It has been noted that the mean response latency from studies on individual word naming is approximately 500 ms (North, Grant, & Fleming, 1967; Forster &

50

J Psycholinguist Res (2007) 36:47–63

Chambers, 1973; Perlmutter, Sorce, & Myers, 1976; Balota & Chumbley, 1984) and the mean response latency to make a positive lexical decision is approximately 700 ms (Rubenstein et al., 1971; Forster & Chambers, 1973; Meyer, Schvaneveldt, & Ruddy, 1974; Balota & Chumbley, 1984). Perhaps slowing the responses down would give the pseudohomophone base word frequency effect a chance to emerge. Purpose of the current effort In order to assess the role orthographic and phonological processing plays when performing the lexical decision task, pseudohomophones and nonwords were used as primes for their word counterparts. Because many pseudohomophones are spelled similarly to their word counterparts, it is necessary to ensure that pseudohomophones act as primes based on phonological similarity rather than orthographic similarity. One way to accomplish this objective is to yoke each pseudohomophone prime (e.g., DALL) with an orthographic control prime (e.g., DOLP). The pseudohomophone DALL is as orthographically similar to DOLL as is the nonword DOLP. If pseudohomophones prime their word counterparts because of orthographic similarity, then there should be no difference in the priming effect of pseudohomophone primes and nonword spelling control primes. Conversely, if the pseudohomophone primes activate the phonological representations of their word counterparts, then pseudohomophone primes should cause more priming than do nonword spelling control primes. Therefore, the present experiment included the manipulation of pseudohomophone primes (e.g., DALL for the target DOLL) versus spelling control nonword primes (e.g., DOLP for the target DOLL). This manipulation tested for the presence of an effect of the phonological characteristics of pseudohomophone primes while controlling for the fact that these primes are orthographically similar to targets. The idea that both phonological and orthographic influences in word recognition, would predict large priming effects when pseudohomophone primes are spelled similarly to word targets was tested by using pseudohomophone primes that varied in the degree to which they are spelled similarly to their homophone targets. The pseudohomophone prime DALL is very similar in spelling to DOLL. The pseudohomophone prime DAUL is less similar in spelling to DOLL. If an orthographic representation is used directly in word recognition, then the likelihood that priming will be affected by orthographically-similar primes should be a function of the number of orthographic characteristics that they share with a target. Thus, the pseudohomophone DALL is likely to be a better prime for the target DOLL than the pseudohomophone prime DAUL would be. Alternately, if a pseudohomophone is transformed into a phonological representation that is independent of the orthographic structure of the prime, then there should be no difference in the repetition priming effect for pseudohomophone primes such as DALL and DAUL for the target word DOLL. In addition, the experiment will examine whether pseudohomophones will show a word frequency effect. Past research (McCann & Besner, 1987; McCann et al., 1988) found no effect of base word frequency on pseudohomophone naming latencies or lexical decision response latencies. However, these researchers used a stimulus pool in which the pseudohomophones constituted half of the nonword items and primes and targets were separated by only one or two items, if any. McQuade (1981) found that the pseudohomophone effect was apparent when the pseudohomophones constituted a small percentage of the nonwords (13%), but not when they constituted the majority of the nonwords. This led McQuade and other researchers (e.g., Hawkins,

J Psycholinguist Res (2007) 36:47–63

51

Reicher, Rogers, & Peterson, 1976; Davelaar, Coltheart, Besner, & Jonasson, 1978) to conclude that processing strategies were induced by including a high percentage of phonologically ambiguous stimuli. Therefore, the present experiment employed lists in which 30% of the trials are comprised of nonhomophonic nonwords and 20% of the trials are comprised of pseudohomophones. With the smaller percentage of pseudohomophones, presented at longer lags, perhaps the pseudohomophone base word frequency effect can be teased out without processing strategies interfering. Method Participants Forty-eight undergraduates from San Jose State University participated for course credit. All participants were right-handed, fluent speakers of English, and all had normal or corrected-to-normal vision. Materials Stimuli consisted of 40 content words of moderate frequency, 40 content words of high frequency, 160 pseudohomophones that were homophonic with the words, and 80 pronounceable and orthographically legal nonwords that were not homophonic with any English word. The first set of pseudohomophones was constructed by changing one letter of the word stimuli while maintaining both orthographic and phonological similarity. The second set of pseudohomophones was constructed by changing two to four letters of the word stimuli which also maintained phonological similarity but lessened orthographic similarity. Pseudohomophones were constructed so that base words representing a wide range of frequencies were included. According to the Kucera & Francis (1967) word count, the resulting set included items ranging in frequency from 1 to 760 occurrences per million. The moderate frequency items have a mean frequency of 9.85 (range = 1–35). The high-frequency items have a mean frequency of 185.40 (36–760). The nonwords were constructed by changing one letter of the word stimuli which maintained orthographic similarity but not phonological similarity. Examples of each of the four types of items are DREAM, GREEN, KEEP, STEAM (words); DREEM, GREAN, KEAP, STEEM (pseudohomophones with one letter change); DREME, GRENE, KEPE, STEME (pseudohomophones with more than one letter change); DROAM, GREEM, KEEB, SPEAM (nonwords with one letter change). The sets of words, pseudohomophones, and nonwords were matched on length (M = 4.70 letters, range = 4–6). In addition, 140 filler items were gathered, consisting of 50 content words similar in familiarity to the experimental words and 20 homophonic, pronounceable, orthographically legal pseudohomophones, and 70 nonhomophonic, pronounceable, orthographically legal nonwords. (See Appendices for a complete list of the stimuli.) Design Four counterbalanced lists containing 300 items consisting of 80 primes, 80 targets, and 140 filler items were constructed. Lists containing 80 yoked quadruplets were constructed such that each word target was preceded by a word prime in the first list, a pseudohomophone prime in the second and third list, and a nonword prime in

52

J Psycholinguist Res (2007) 36:47–63

the fourth list. Each participant saw only one prime-target pair of the quadruplet. By presenting each participant with 20 experimental items of each type, across each set of four participants, all members of each quadruplet appeared equally often. The average number of intervening trials, or lag, between the presentation of the prime and the target was eight items (range = 5–11). The inclusion of filler items insured appropriate lags and discouraged response bias. For each participant: 50% of the trials were comprised of words, 30% were comprised of nonhomophonic nonwords, and 20% were comprised of pseudohomophones. A quarter of the participants were randomly assigned to each list. Lists were constructed by a computer program (Woffindin, 1990) that evenly distributed the prime-target conditions, the number of intervening trials between primes and targets, the string length, word frequency, and the sequence of “word” and “nonword” trials. To summarize the 2 (frequency) ×4 (prime-target condition) experimental design: across lists, each moderate-frequency and high-frequency target word was preceded by its prime in word, pseudohomophone (one orthographic change and more than one orthographic change types), and nonword form. Apparatus Stimulus items were displayed in lowercase letters on a high-resolution computer monitor. Letters were white on a black background. A microcomputer equipped with a World Commerce Inc., Psycholinguistic Testing Station controlled the experiment and recorded responses and response latencies. A comfortable viewing distance was chosen by each participant. The average letter string subtended a visual angle of approximately 2◦ . The response box contained three buttons, one for initiating trials and two for indicating lexical decision responses. Procedure The participants controlled the pacing of trials. To indicate the start of each trial, a fixation asterisk (*) appeared at the center of the monitor. To initiate a trial, the participant used both thumbs to press a button centered on the lower half of the response box causing the disappearance of the asterisk. The stimulus item then appeared in the center of the screen 350 ms later and remained there until the participant made his or her lexical decision response. Responses were indicated by pressing one of two buttons on the upper half of the response box with the appropriate index finger. The left-hand button was used for “nonword” responses and the right-hand button for “word” responses. If the participants made an erroneous lexical decision, feedback was provided by the sound of a “beep.” The participants were asked to make their decisions as quickly as possible without sacrificing accuracy. Each participant completed 24 practice trials before proceeding to the 300 experimental trials. The practice items consisted of words, nonwords, and pseudohomophones not appearing elsewhere in the experiment. The experimental session took less than 30 min for each participant. Scoring of data Data from incorrect trials were excluded from the latency analyses. The occasional extremely long-response latencies, that were more than 2.5 standard deviations greater

J Psycholinguist Res (2007) 36:47–63

53

than the participant’s mean for a particular type of trial, were replaced by the cut-off value of 2.5 standard deviation units plus the participant’s mean for that trial type. Data sets for both subjects and stimulus items were computed from the response latency and error rate data. Two data sets were formed by computing mean response latencies and error rates over stimulus items for each subject (subject analysis data). Two additional data sets were formed by computing mean response latencies and error rates over participants for each stimulus item (item analysis data).

Results and discussion Pseudohomophone effects Mean lexical decision response latencies (ms) and percentage error (PE) for word, pseudohomophone, and nonword primes are presented in Table 1. All descriptive statistics reported were derived from the subject analyses. To determine whether there was a pseudohomophone effect (i.e., slower responses and more errors on pseudohomophones than on nonwords), response latencies and error rates to pseudohomophone one-letter change primes (833 ms and 14.48 PE) and nonword one-letter change primes (797 ms and 6.15 PE) for both subjects and stimulus items were compared. A significant pseudohomophone effect was indicated by longer response latencies (t[47] = 3.61, p < 0.001 by subjects and t[79] = 3.14, p < 0.01 by items) and higher error rates (t[95] = 6.53, p < 0.001 by subjects and t[79] = 4.34, p < 0.001 by items) for the pseudohomophone primes when compared to the nonword primes. To ascertain whether the pseudohomophone effect was influenced by the orthographic similarity of the pseudohomophone primes to their base words (independent of the effect of phonological similarity), the response latencies and error rates to one-letter change pseudohomophone primes (833 ms and 14.48 PE) and two-letter change pseudohomophone primes (782 ms and 5.21 PE) were compared. The significantly longer response latencies (t[95] = 7.71, p < 0.001 by subjects and t[79] = 4.89, p < 0.001 by items) and higher error rates on the one-letter change primes (t[95] = 5.42, p < 0.001 by subjects and t[79] = 4.13, p < 0.001 by items) indicated that pseudohomophone processing was influenced by the orthographic similarity of the pseudohomophones.

Table 1 Mean lexical decision response times (ms), and percentage errors (PE), for word, pseudohomophone one-letter change (PH1), pseudohomophone two-letter change (PH2), and nonword one-letter change (NW1) primes

Moderate frequency Prime Response times PE High frequency Prime Response times PE

Word

PH 1

PH 2

NW 1

SKIRT 778 14.38

SKURT 834 17.71

SCURT 783 5.42

SKART 808 6.46

DREAM 719 3.33

DREEM 831 11.25

DREME 780 5.00

DROAM 786 5.83

54

J Psycholinguist Res (2007) 36:47–63

To determine whether the pseudohomophone effect would persist with two orthographic changes, two-letter change pseudohomophones (782 ms and 5.21 PE) were compared to one-letter change nonwords (797 ms and 6.15 PE). The 15 ms advantage that the pseudohomophones demonstrated over the nonwords was not significant (t[95] = 1.66, p > 0.05 by subjects and t[79] = 1.45, p > 0.05 by items). The 0.94 difference in error rates was also not significant (both subjects and items t < 1). In summary, there was a definite pseudohomophone effect, as evidenced by longer response latencies and higher error rates, when one-letter change pseudohomophones were compared to one-letter change nonwords. However, one-letter change pseudohomophones were significantly slower and more error prone than two-letter change pseudohomophones. Finally, when two-letter change pseudohomophones were compared to one-letter change nonwords, the pseudohomophone effect was not apparent. Repetition priming effects Prime and target examples, mean lexical decision response times (ms), PE, and repetition priming effects are presented in Table 2. All descriptive statistics reported were derived from the subject analyses. Analyses of variance on word targets with frequency (moderate and high) and prime condition (word, pseudohomophone one-letter change, pseudohomophone two-letter change, and nonword one-letter change) as independent variables were performed on response latencies and error rates. The outcome showed that responses to moderate-frequency word targets were slower and less accurate (726 ms and 7.29 PE) than responses to high-frequency word targets (680 ms and 1.88 PE). The 46 ms difference in response latencies was found to be significant (F[1, 329] = 125.18, p < 0.01 by subjects and F[1, 273] = 33.60, p < 0.001 by items) as was the 5.41 difference in error rates (F[1, 329] = 146.41, p < 0.001 by subjects and F[1, 273] = 7.84, p < 0.05 by items). Table 2 Prime and target examples, mean response times (ms), percentage errors (PE), and repetition priming effects for words, pseudohomophone one-letter change (PH1), pseudohomophone two-letter change (PH2), and nonword one-letter change (NW1) Control prime example Moderate frequency Prime example SKIRT Target example Response times 778 Priming effect PE 14.38 Priming Effect High frequency Prime example DREAM Target example Response times 719 Priming effect PE 3.33 Priming effect ∗ Denotes significant priming effect

Target primed by word

Target primed by PH1

Target primed by PH2

Target primed by NW1

SKIRT SKIRT 726 52∗ 7.29 7.09∗

SKURT SKIRT 742 36∗ 13.13 1.25

SCURT SKIRT 766 12 14.58 −0.20

SKART SKIRT 773 5 14.79 −0.41

DREAM DREAM 680 39∗ 1.88 1.45

DREEM DREAM 692 27∗ 3.75 −0.42

DREME DREAM 713 6 2.50 0.83

DROAM DREAM 708 11 3.75 −0.42

J Psycholinguist Res (2007) 36:47–63

55

The analysis also revealed a significant effect of prime condition on response latencies (F[3, 329] = 10.13, p < 0.01 by subjects and F[3, 273] = 7.88, p < 0.01 by items) and error rates to targets (F[3, 329] = 2.80, p < 0.01 by subjects and F[3, 273] = 8.04, p < 0.01 by items) on word targets primed by word (629 ms and 3.54 PE), pseudohomophone one-letter change (664 ms and 5.62 PE), pseudohomophone twoletter change (664 ms and 5.00 PE), and nonword one-letter change (674 ms and 4.59 PE) conditions. The interaction of frequency with prime condition was not significant for the response latencies ( F [3, 329] = 1.40, p > 0.05 by subjects and F[3, 273] = 1.70, p > 0.05 by items) or the error rates (F[3, 329] = 10.02, p > 0.05 by subjects and F[3, 273] = 1.85, p > 0.05 by items). In assessing specific repetition priming effects, the base word prime condition (e.g., the first occurrence of DOLL in the DOLL–DOLL pair) served as a control condition which was individually compared against the base word target condition (e.g., the second occurrence of DOLL in the DOLL–DOLL pair) and the critical target conditions (e.g., DOLL primed with DALL, DAUL, or DOLP respectively). The difference in mean response latency and error rate between the base word prime condition and the base word target condition will be termed the identity priming effect. In order to assess whether the priming effect of the pseudohomophone and nonword primes was different from the identity priming effect individual comparisons among the four target conditions were performed. Thus, the magnitude of facilitation with pseudohomophone and nonword primes was defined relative to the facilitation due to identical repetition of the base word. The results would be taken to indicate “full” repetition priming if the different prime conditions produced results equivalent to identity priming. Conversely, the results would be taken to indicate “partial” repetition priming if the pseudohomophone and nonword prime conditions produced significant priming that is smaller in magnitude than identity priming. Planned comparisons indicated that moderate-frequency word targets were facilitated by word primes and pseudohomophone one-letter change primes, but not by any other type of prime. The identity priming effect (52 ms and 70.09 PE) was statistically significant for both response latencies (t[47] = 6.29, p < 0.001 by subjects and t[39] = 4.80, p < 0.001 by items) and error rates (t[47] = 3.36, p < 0.01 by subjects and t[39] = 3.44, p < 0.01 by items). Response latencies to moderate frequency word targets as a function of pseudohomophone one-letter change primes were significant (t[47] = 2.77, p < 0.05 by subjects and t[39] = 2.67, p < 0.05 by items) but error rates were not (t[47] = 1.42, p > 0.05 by subjects and t[39] = 1.06, p > 0.05 by items), Response latencies to moderate frequency word targets were not affected by pseudohomophone two-letter change (t[47] = 1.21, p > 0.05 by subjects and t[39] = 1.32, p > 0.05 by items), or nonword one-letter change (both subject and item t < 1) prime conditions. Similarly, error rates to moderate frequency word targets did not decrease due to pseudohomophone two-letter change (both subject and item t < 1), or nonword one-letter change (both subject and item t < 1) prime conditions. The difference in mean response latency between the high-frequency base word prime condition (719 ms) and the high-frequency base word target condition (680 ms) yielded an identity priming effect of 39 ms. The difference in mean response latency between the high-frequency base word prime condition (719 ms) and the highfrequency pseudohomophone one-letter change (692 ms), pseudohomophone two-letter change (713 ms), and nonword one-letter change (708 ms) priming conditions yielded partial repetition priming effects of 27, 6, and 11 ms, respectively. These repetition priming effects observed for response latencies to high-frequency

56

J Psycholinguist Res (2007) 36:47–63

word targets were statistically significant for word (t[47] = 6.45, p < 0.001 by subjects and t[39] = 6.88, p < 0.001 by items) and pseudohomophone one-letter change primes (t[48] = 20.09, p < 0.05 by subjects and t[39] = 2.08, p < 0.05 by items), but not for pseudohomophone two-letter change (t[47] = 1.34, p > 0.05 by subjects and t[39] = 1.77, p > 0.05 by items) or nonword primes (t[47] = 1.09, p > 0.05 by subjects and t[39] = 1.52, p > 0.05 by items). Error rates to high-frequency word targets were not significantly decreased by the identity prime condition (t[48] = 2.01, p > 0.05 by subjects and t[39] = 1.94, p > 0.05 by items), the pseudohomophone one-letter change condition ( t[48] = 1.25, p > 0.05 by subjects and t[39] = 1.11, p > 0.05 by items), the pseudohomophone two-letter change condition ( t[48] = 2.01, p > 0.05 by subjects and t[39] = 1.39, p > 0.05 by items) or the nonword one-letter change condition ( t[48] = 1.31, p > 0.05 by subject and t[39] = 1.06, p > 0.05 by items). Frequency effects Word-frequency effects in the expected direction were indicated by longer response latencies and higher error rates for moderate-frequency word primes (778 ms and 14.38 PE) than for high frequency word primes (719 ms and 3.33 PE). The 59 ms difference was statistically significant (t[47] = 6.62, p < 0.001 by subjects and t[39] = 3.17, p < 0.01 by items). The 11.05 difference in error rates was also significant (t[47] = 5.05, p < .001 by subjects and t[39] = 5.06, p < 0.001 by items). To determine whether pseudohomophone processing was influenced by the frequency of the base word (e.g., were response latencies to DALL influenced by the frequency of DOLL?), the response latencies to pseudohomophones derived from moderate frequency base words (834 ms) were compared to pseudohomophones derived from high frequency base words (831 ms). The 3 ms difference was not significant (both subject and item t < 1). Although the frequency with which a word occurs in English text was found to be a significant predictor of response latency and error rates on the word trials used, no corresponding effect of base word frequency was found on the pseudohomophone trials. Herdman, LeFevre, & Greenham (1994) and Taft & Russell (1992) observed a pseudohomophone frequency effect on participants with slow responses. Therefore additional analyses were performed on the slowest 15 of 48 participants in the present study. The 15 participants identified as the slowest responders had an average lexical decision response latency of 744 ms. Slow-responder responses to pseudohomophones derived from moderate-frequency base words (937 ms) were then compared to slow-responder responses to pseudohomophones derived from high-frequency base words (931 ms). Although in the direction that is typical of frequency effects, the 6 ms difference between pseudohomophones derived from moderate-frequency and high-frequency words was not significant ( t < 1). Therefore, unlike previous studies, the present experiment did not find slow-responders to be sensitive to base-word frequency effects in the lexical decision task.

General discussion The results can be summarized as follows: a phonological activation effect was demonstrated by a processing disadvantage for pseudohomophone one-letter change

J Psycholinguist Res (2007) 36:47–63

57

primes when compared to the nonword one-letter change primes. However, baseword frequency had no effect on pseudohomophone processing even though word frequency had an effect on the processing of the base words themselves. In addition, the pseudohomophone effect was eliminated when pseudohomophone twoletter change primes were compared to nonword one-letter change primes. Finally, lexical decisions to moderate-frequency and high-frequency word targets were facilitated by word primes and pseudohomophone one-letter change primes only. In the following discussion, the implications of each of the presented results will be considered as they pertain to the issue of phonological and orthographic coding in word recognition. Pseudohomophone effects As aforementioned, the prevalent explanation of the pseudohomophone effect states that a pseudohomophone’s phonological representation activates the lexical representation for the base word, which, then, makes it harder for participants to reject the pseudohomophone. In the present experiment, pseudohomophone effects were demonstrated by longer response latencies and higher error rates for the pseudohomophones that were equally close in orthography to the nonwords. These results demonstrate that phonological representations are utilized in word recognition. However, the pseudohomophone effect failed to materialize when pseudohomophones that were orthographically less similar to their corresponding base words were compared to nonwords that were orthographically more similar to their corresponding base words. The elimination of the pseudohomophone effect provides evidence that orthographic codes have a potent effect on the processing of pseudohomophones. As long as orthographic equality exists between pseudohomophones and nonwords, the pseudohomophone effect will persist. When this equality is not sustained, orthographic codes exert more influence on word recognition than phonological codes do. The results of the present experiment demonstrated that responses to pseudohomophones that were orthographically more similar to their corresponding base word had slower and less accurate lexical decision response latencies than the responses to pseudohomophones that were less similar to their corresponding base words. Pseudohomophone effects from other experiments have demonstrated that when pseudohomophones and nonwords are equated for orthographic similarity, pseudohomophones show a processing advantage in naming (McCann & Besner, 1987) and a processing disadvantage in lexical decision (Coltheart et al., 1977; McQuade, 1981; Besner & Davelaar, 1983; McCann et al., 1988). The results of the present study also demonstrated that when pseudohomophones and nonwords are equated for orthographic similarity, the pseudohomophones show a processing disadvantage in lexical decision. But when the pseudohomophones were orthographically less similar than the nonwords to their base words, the pseudohomophones exhibited a processing advantage for lexical decision response latencies. However, this processing advantage exhibited by the pseudohomophones was not significant. In light of these findings, it cannot be concluded that the pseudohomophone effect is a result of orthographic activation. Instead, the results provide evidence that the pseudohomophone effect is a combined effect of both orthographic and phonological sources of lexical activation.

58

J Psycholinguist Res (2007) 36:47–63

Repetition priming effects The current results demonstrate that lexical decisions to word targets were facilitated by word and pseudohomophone primes but not nonword primes. The magnitude of facilitation for pseudohomophone, and nonword primes was defined relative to the facilitation with an identical repetition of the base word. A partial priming effect was evident in that the pseudohomophone prime conditions produced significant but numerically less-than-full effects. With regard to the word prime condition, a common finding in previous repetition priming research is the frequency attenuation effect. This effect is evidenced by larger repetition priming effects for low-frequency words than those for high- frequency words. To look at it another way, the effect represents a diminishment of the word frequency effect when the words have been repeated (Scarborough, Cortese, & Scarborough, 1977; McCusker, 1979; Kirsner, Milech, & Standen, 1983; Monsell, 1985). Curiously, this classic interactive ef fect of frequency and repetition, although in the direction of an attenuation effect, did not achieve significance in the present experiment. Moderate-frequency word primes and high-frequency word primes had nearly equal facilitative effects on target words (52 and 39 ms, respectively). Therefore, the present study demonstrated a persistent frequency effect that did not diminish when words were repeated. Response latencies to moderate-frequency and high-frequency words differed by 59 ms for first presentation as primes and by 46 ms for second presentation as targets. It is possible that the range of frequencies employed may have contributed to the lack of frequency attenuation. The moderate-frequency words had a mean frequency of 9.85 occurrences per million words of text with a range of 1–35. The high-frequency words had a mean frequency of 185.40 occurrences per million words of text with a range of 36–760. Although the means clearly define the items as moderate- and high-frequency words, the range of frequencies employed is not as clear. In a classic study of frequency and repetition which demonstrated the frequency attenuation effect, Scarborough, Cortese, & Scarborough (1977) employed a median frequency of 3 occurrences per million with a range of 1–9 for low-frequency words and a median frequency of 76 occurrences per million with a range of 31–600 for high-frequency words. With regard to the pseudohomophone data, as mentioned previously, if a pseudohomophone is transformed into a phonological representation that is independent of the orthographic structure of the prime, then there should be no difference in the repetition priming effect for pseudohomophone one-letter change primes such as DALL and pseudohomophone two-letter change primes such as DAUL for the target word DOLL. The results lent partial support to the phonological activation theory because one-letter change pseudohomophone primes facilitated target response latencies. If an orthographic representation is used directly in word recognition, then the likelihood that priming will be affected by orthographically similar primes should be a function of the number of orthographic characteristics that they share with word targets. If this were true, then one-letter change pseudohomophones and one-letter change nonwords would have equally primed the targets. This was not the case with the current effort. In light of these findings, it is evident that phonological sources of lexical activation cannot be ruled out. With regard to the nonword prime data, previous research has reported repetition priming effects to be substantially greater for word targets than for nonword targets (Scarborough et al., 1977; Jordan & Huntsman, 1990). Other researchers have

J Psycholinguist Res (2007) 36:47–63

59

reported repetition inhibition effects when nonwords were used as primes and targets (Monsell, 1987b). Still, other researchers have found essentially no repetition effects, or at best transient facilitatory effects, for nonword targets (e.g., Forbach, Stanners, & Hochaus, 1974; Monsell, 1987a; Bentin & Moscovitch, 1988). However, very little research has been reported on the use of nonwords as primes for word targets. Besner et al. (1985) using a mean lag of two intervening items reported a 19 ms priming effect for graphemically related prime target pairs (e.g., CROTH and CROSS) versus unrelated control prime target pairs (e.g., THARK and CROSS). The present experiment used a mean lag of eight intervening items and found that nonword primes had an 8 ms repetition priming effect on word targets that was not significant. Frequency effects Despite the frequency effect observed for base words, and despite the fact that only 20% of the stimuli were comprised of pseudohomophones, responses to the pseudohomophones were not sensitive to frequency. By classifying their participants as slow and fast responders, Herdman et al. (1994) and Taft & Russell (1992) found that for slow responders, pseudohomophone naming was related to base-word frequency. The present experiment did not find slow responders to be sensitive to base-word frequency effects in the lexical decision task.

Summary and conclusion Evidence for phonological activation was demonstrated by a processing disadvantage for pseudohomophones that were equally close in orthography to the nonword controls. This finding suggests that a pseudohomophone’s phonological representation activates the lexical representation for its base word. Consequently, participants experienced greater difficulty in rejecting the pseudohomophone. Two-letter change pseudohomophones were not significantly different from one-letter change nonwords, however. These results provide evidence that the pseudohomophone effect is influenced by both orthographic and phonological sources of lexical activation. Additional evidence for phonological activation was evident because one-letter change pseudohomophones significantly primed word targets, whereas one-letter change nonwords did not. The presence of the pseudohomophone effect coupled with the absence of a corresponding base-word frequency effect for pseudohomophones suggests that phonological activation is not the source of the frequency effect in word recognition because a frequency effect was observed for the base words but not the pseudohomophones. However, considering the significant pseudohomophone effect and the ability of pseudohomophones to prime words, we cannot downplay the role that phonology plays in the word recognition process. After all, looks aren’t everything. Acknowledgments This research was supported in part by a San Jose State University Psychology Department Research Grant. Thank you to Katherine Lemkuil, John Robertson, Joseph Tajnai, Tuan Tran, Ruzica Udovicic, and Michael Weinborn for their assistance with stimuli development, data collection, and data tabulation. Thank you to Derek Besner, Alexander Pollatsek, Kathy Rastle, Marcus Taft, and especially, Susan Lima for helpful comments and criticisms. Special thanks go to Guy Woffindin for his invaluable programming assistance.

60

J Psycholinguist Res (2007) 36:47–63

Appendix Appendix A Moderate frequency word, pseudohomophone, and nonword experimental stimuli Word

PH 1

PH 2

NW 1

bead leaf nail tape bleed cough lease purse steam bruise doll lean fuse teen cease cream nurse scoop steer cruise cope lewd soup wake cheat greed plead skirt terse praise wolf mate cord weed cleat kneel purge spear verse scream

beed leef nayl taipe blead kough leese purce steem bruize dall leen fuze tean ceese creem nerse skoop stear cruize kope leud soop wayke cheet gread pleed skurt turse praize wulf maite kord wead cleet kneal perge speer verce skream

bede lefe nale taip blede coff leece perce steme broose daul lene fewze tene seese kreem nerce skoup stier crooze koap lood supe waik chete grede plede scurt turce prayze wolph mait koard wede kleet gneal perje spere vurce screme

beal li nait tafe bleen coagh leabe pursk speam broise dolp leat fube teeb ceave creaf nulse scook steek cruive cose lawd saup wike chead greud pread skart torse proise wolk mafe corg weef cleak kneek pulg speat vorse scroam

PH1 pseudohomophone one-letter change, PH2 pseudohomophone two-letter change, NW1 nonword one-letter change Appendix B High frequency word, pseudohomophone, and nonword experimental stimuli Word

PH 1

PH 2

NW 1

fail feel keep

fayl feal keap

fale fele kepe

faip feeg keeb

J Psycholinguist Res (2007) 36:47–63

61

Appendix B continued Word

PH 1

PH 2

NW 1

found field guide raise trial choice door gain mean need chief floor learn reach white please fear girl seek work clear green leave serve worth screen feed heat play year dream group prove speed youth stream

fownd feeld gide raize tryal choyce doar gayn meen nead cheaf floar lern reech whyte pleese feer gerl seak wirk cleer grean leeve surve wirth skreen fead heet plae yeer dreem groop pruve spead yooth streem

phound feild gyde rayze tryel choyse dore gane mene nede cheafe flore lerne reche wyte pleeze fere ghurl seke wirc clere grene lieve cerve whirth screne fede hete plaie yeir dreme grewp proov spede yewth streme

foung fielp gaide roise triak cloice doot gais meav neek chiel floob leorn reace whike plerse feag girp serk wolk crear greem leate serne worsh scheen feek hoat plaf yoar droam grouf proke sleed yoush streal

PH1 pseudohomophone one-letter change, PH2 pseudohomophone two-letter change, NW1 nonword one-letter change

References Balota, D. A., & Chumbley, J. I. (1984). Are lexical decisions a good measure of lexical access? The role of word frequency in the neglected decision stage. Journal of Experimental Psychology: Human Perception and Performance, 10, 340–357. Barron, R. W. (1978). Reading skill and phonological coding in lexical access. In M. M. Gruneberg, R. N. Sykes, & P. E. Morris (Eds.), Practical aspects of memory (pp. 203–223). London: Academic Press. Barry, C. (1981). Hemispheric asymmetry in lexical access and phonological encoding. Neuropsychologia, 19, 473–478. Bentin, S., & Moscovitch, M. (1988). The time course of repetition effects for words and unfamiliar faces. Journal of Experimental Psychology: General, 117, 148–160. Besner, D., & Davelaar, E. (1983). Suedohomofoan effects in visual word recognition: Evidence for phonological processing. Canadian Journal of Psychology, 37, 300–305.

62

J Psycholinguist Res (2007) 36:47–63

Besner, D., Dennis, I., & Davelaar, E. (1985). Reading without phonology? The Quarterly Journal of Experimental Psychology, 37, 477–491. Cohen, G., & Freeman, R. (1978). Individual differences in reading strategies in relation to handedness and cerebral asymmetry. In J. Requin (Ed.), Attention and performance, (Vol. pp. 411–426). Hillsdale, NJ: Lawrence Erlbaum Associates. Coltheart, M., Davelaar, E., Jonasson, J. T, & Besner, D. (1977). Access to the internal lexicon. In S. Dornic (Ed.), Attention and performance (VI, pp. 535–555). New Vol. York: Academic Press. Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J. (2001). DRC: A dual route cascaded model of visual word recognition and reading aloud. Psychological Review, 108, 204–256. Davelaar, E., Coltheart, M., Besner, D., & Jonasson, J. T. (1978). Phonological recoding and lexical access. Memory and Cognition, 6, 391–402. Forbach, G. B., Stanners, R. F., & Hochaus, L. (1974). Repetition and practice effects in a lexical decision task. Memory and Cognition, 2, 337–339. Forster, K. I., & Chambers, S. M. (1973). Lexical access and naming time. Journal of Verbal Learning and Verbal Behavior, 12, 627–635. Frederiksen, J. R., & Kroll, J. F. (1976). Spelling and sound: Approaches to the internal lexicon. Journal of Experimental Psychology: Human Perception and Performance, 2, 363–379. Hawkins, H. L., Reicher, G. M., Rogers, M., & Peterson, L. (1976). Flexible coding in word recognition. Journal of Experimental Psychology: Human Perception and Performance, 2, 380–385. Herdman, C. M., LeFevre, J. A., & Greenham, S. L. (1994). Implicating the lexicon: Base-word frequency effects in pseudohomophone naming. Journal of Experimental Psychology: Human Perception and Performance, 20, 575–590. Humphreys, G. W., Evett, L. J., & Taylor, D. E. (1982). Automatic phonological priming in visual word recognition. Memory and Cognition, 10, 576–590. Huntsman, L. A. (1998). Testing the direct-access model: GOD does not prime DOG. Perception and Psychophysics, 60, 1128–1140. Jordan, K., & Huntsman, L. A. (1990). Image rotation of misoriented letter strings: Effects of orientation cuing and repetition. Perception and Psychophysics, 48, 363–374. Kirsner, K., Milech, D., & Standen, P. (1983). Common and modality-specific processes in the mental lexicon. Memory and Cognition, 11, 621–630. Kucera, H., & Francis, W. N. (1967). Computational analysis of present-day American English . Providence, RI: Brown University Press. Lukatela, G., & Turvey, M. T. (1991). Phonological access of the lexicon: Evidence from associative priming with pseudohomophones. Journal of Experimental Psychology: Human Perception and Performance, 17, 951–966. Lukatela, G., Lukatela, K., & Turvey, M. T. (1993). Further evidence for phonological constraints on visual lexical access: TOWED primes FROG. Perception and Psychophysics, 53, 461–466. MacKay, D. G. (1972). Input testing in the detection of misspellings. American Journal of Psychology, 85, 121–127. Martin, R. C. (1982). The pseudohomophone effect: The role of visual similarity in non-word decisions. Quarterly Journal of Experimental Psychology, 34, 395–409. McCann, R. S., & Besner, D. (1987). Reading pseudohomophones: Implications for models of pronunciation assembly and the locus of word-frequency effects in naming. Journal of Experimental Psychology: Human Perception and Performance, 13, 14–24. McCann, R. S., Besner, D., & Davelaar, E. (1988). Word recognition and identification: Do word-frequency effects reflect lexical access? Journal of Experimental Psychology: Human Perception and Performance, 14, 693–706. McCusker, L. X. (November, 1979). Modality, frequency, and repetition effects in word recognition. Paper presented at the meeting of the Psychonomic Society, Phoenix, AZ. McQuade, D. V. (1981). Variable reliance on phonological information in visual word recognition. Language and Speech, 24, 99–109. Meyer, D. E., Schvaneveldt, R. W., & Ruddy, M. G. (1974). Functions of graphemic and phonemic codes in visual word recognition. Memory and Cognition, 2, 309–321. Mitterer, J. (1982). There are at least two kinds of poor readers: Whole-word poor readers and recoding poor readers. Canadian Journal of Psychology, 36, 445–461. Monsell, S. (1985). Repetition and the lexicon. In A. W. Ellis (Ed.), Progress in the psychology of language (Vol. 2, pp. 147–195). London: Erlbaum. Monsell, S. (1987a). On the relation between lexical input and output pathways for speech. In A. Allport, D. MacKay, W. Prinz, & E. Scheerer (Eds.), Language perception and production: Relationships between listening, speaking, reading and writing (pp. 273–311). London: Academic Press.

J Psycholinguist Res (2007) 36:47–63

63

Monsell, S. (1987b). Non-visual orthographic processing and the orthographic input lexicon. In M. Coltheart (Ed.) Attention and performance (vol. XIII, pp. 299–323). London: Erlbaum. North, J. A., Grant, D. A., & Fleming, R. A. (1967). Choice reaction times to single digits, spelled numbers, right and wrong arithmetic problems and short sentences. Quarterly Journal of Experimental Psychology, 19, 73–77. Patterson, K. E., & Marcel, A. J. (1977). Aphasia, dyslexia and the phonological coding of written words. Quarterly Journal of Experimental Psychology, 29, 307–318. Perfetti, C. A., Bell, L. C., & Delaney, S. (1988). Automatic (prelexical) phonetic activation in silent word reading: Evidence from backward masking. Journal of Memory and Language, 27, 59–70. Perlmutter, J., Sorce, P., & Myers, J. L. (1976). Retrieval processes in recall. Cognitive Psychology, 8, 32–63. Pring, L. (1981). Phonological codes and functional spelling units: Reality and implications. Perception and Psychophysics, 30, 573–578. Rayner, K., Sereno, S. C., Lesch, M. F., & Pollatsek, A. (1995). Phonological codes are automatically activated during reading: Evidence from an eye movement priming paradigm. Psychological Science, 6, 26–32. Rubenstein, H., Lewis, S. S., & Rubenstein, M. A. (1971). Evidence for phonemic recoding in visual word recognition. Journal of Verbal Learning and Verbal Behavior, 10, 645–657. Scarborough, D. L., Cortese, C., & Scarborough, H. S. (1977). Frequency and repetition effects in lexical memory. Journal of Experimental Psychology: Human Perception and Performance, 3, 1–17. Taft, M. (1982). An alternative to grapheme-phoneme conversion rules? Memory and Cognition, 10, 465–474. Taft, M., & Russell, B. (1992). Pseudohomophone naming and the word frequency effect. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 45, 51–71. Van Orden, G. C., Johnston, J. C., & Hale, B. L. (1987). Word identification in reading proceeds from spelling to sound to meaning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14, 371–386. Whaley, C. P. (1978). Word-nonword classification time. Journal of Verbal Learning and Verbal Behavior, 17, 143–154. Woffindin, G. C. (1990). Psycholinguistic stimuli sorting program [Computer Program]. Cupertino, CA: World Commerce Ltd.