Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 1 Processing Speaker Variability in Repetition and Semantic/Associative Priming Chao-Yang Lee and Yu Zhang Ohio University Author Note Chao-Yang Lee, Division of Communication Sciences and Disorders, Ohio University; Yu Zhang, Division of Communication Sciences and Disorders, Ohio University. We thank Danny R. Moates for assistance in recruiting participants, Lauren Baer and Eleni Gkikas for assistance in administering the experiments, and Juliana Gursky for editorial assistance. Correspondence concerning this article should be addressed to Chao-Yang Lee, Division of Communication Sciences and Disorders, Ohio University, Athens, OH 45701. E-mail: leec1@ohio.edu Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 2 Abstract The effect of speaker variability on accessing the form and meaning of spoken words was evaluated in two short-term paired priming experiments. In the repetition priming experiment, participants listened to repeated or unrelated prime-target pairs, in which the prime and target were produced by the same speaker or different speakers. The results showed that the magnitude of repetition priming was reduced when the prime and target were produced by different speakers, indicating that speaker variability affected access to word forms. In the semantic/associative priming experiment, participants listened to semantically/associatively related or unrelated prime-target pairs, in which the prime and target were produced by the same speaker or different speakers. The results showed that the magnitude of semantic/associative priming was reduced in different-speaker trials, but only for targets produced by the female speaker. There was no evidence that the speaker variability effect varied as a function of the interstimulus intervals used in this study (50 ms and 250 ms). These findings suggest that speaker variability affects spoken word recognition, but primary at a relatively shallow level of processing. Keywords: spoken word recognition, speaker variability, repetition priming, semantic/associative priming Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 3 Processing Speaker Variability in Repetition and Semantic/Associative Priming Introduction During spoken word recognition, a listener extracts information from the acoustic signals generated by speakers and maps that information onto internal lexical representations. Lexical representations are usually assumed to be abstract phonological codes storing only lexically contrastive information. Therefore, the process of recognizing spoken words entails discarding surface acoustic variability to arrive at abstract phonological codes (Johnson & Mullennix, 1997). However, there is emerging evidence that acoustic variability is encoded in lexical representations (e.g., Nygaard, Sommers, & Pisoni, 1994), suggesting that the mapping from acoustic signals onto the lexicon does not necessarily involve reduction of acoustic variability (Pisoni, 1997). Given the many sources of acoustic variability in the acoustic signals, a foundational issue is whether all sources of acoustic variability affect spoken word recognition in the same way. Additionally, because words/morphemes are the smallest meaningful unit of speech, a related issue is whether acoustic variability affects access to the phonological form of spoken words, the meaning of spoken words, or both. Given the time-sensitive nature of processing speech, one could also ask whether various sources of acoustic variability share a similar time course during spoken word recognition. The answers to these questions have important implications for the nature of spoken word recognition. The purpose of this study was to address these issues by examining the effect of speaker variability on accessing the form and meaning of spoken words. To that end, we investigated the effect of speaker variability on repetition priming (Forster & Davis, 1984) and semantic/associative priming (Meyer & Schvaneveldt, 1976). Speaker variability refers to acoustic variability generated as a consequence of the speakers’ idiosyncratic features (Luce & McLennan, 2005). Structural features such as the shape, size and length of the vocal tract and Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 4 behavioral features as reflected by different dialect users are potential sources of speaker variability (Johnson & Mullennix, 1997; Palmeri, Goldinger, & Pisoni, 1993). Acoustic-phonetic research over several decades has shown that the same spoken message can vary significantly across speakers. However, listeners are able to understand different speakers of the same language without much difficulty. How listeners achieve this perceptual constancy despite speaker variability is one of the foundational issues in spoken language comprehension. Speaker specificity in lexical representation It is typically assumed that lexical representations are abstract and contain only lexically contrastive information, implying that surface acoustic variability not directly relevant to lexical identity is discarded during the signal-to-representation mapping process. By this account, speaker-specific information would not be part of the abstract phonological code stored in lexical representations. Consequently, memory for words should not be influenced by variability across speakers. This prediction is supported by Jackson and Morton (1984), who used a long-term priming paradigm to examine word recognition in a “test” phase following a “study” phase. In the study phase, participants listened to words spoken by a female or a male speaker and made semantic judgments about the words. In the test phase, participants listened to the same set of words, presented in noise, which were produced by either the same speaker or different speakers. The results showed no differences in word recognition accuracy between words spoken by the same speaker and different speakers. Schacter and Church (1992) asked listeners in the study phase to judge the pitch of the speakers from words spoken by six speakers. In the test phase, listeners were asked to identify words, presented in noise, which were spoken by the same or different speakers. Although the study phase was intended to focus the listeners’ attention on Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 5 speaker-specific information, word recognition accuracy was similar under both the same- and different-speaker conditions. By contrast, other studies employing the long-term priming paradigm showed that memory for words contains detailed information about speakers and that this information affects word recognition. A speaker effect emerged when Schacter and Church (1992) changed their task in the test phase (from word identification to stem completion) and presented the test stimuli in quiet (instead of in noise). Listeners were asked to propose a multisyllabic word based on auditory input of the first syllable of a word. This “stem completion” was more accurate when the stimuli in the test phase were produced by the same speaker as in the study phase. Church and Schacter (1994) asked listeners to identify words that were low-pass filtered to preserve fundamental frequency (F0) information, but not formant information. The results showed that word identification was more accurate when the stimuli were produced by the same speaker in the study phase. In addition, identification accuracy was also affected by within-speaker F0 variation. Specifically, when the same speaker was used in the study and test phase, word identification was more accurate when the F0 between study and test was the same, suggesting that detailed F0 information is encoded in memory for words. Subsequent studies identified additional factors that contribute to the speaker variability effect in long-term priming. Goldinger (1996) showed that word recognition in noise was more accurate when both study and test words were produced by the same speaker. Importantly, the same-speaker advantage in word recognition was observed only when words were encoded at relatively shallow levels of processing (i.e., gender and phoneme classification). By contrast, the speaker effect was attenuated when listeners were asked to encode words at a relatively deep level of processing (i.e., syntactic classification). There is also evidence that the speaker effect is Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 6 contingent on response format. Luce and Lyons (1998) examined the effect of changing speaker on auditory lexical decision (probing implicit memory) and on old/new judgment (probing explicit memory). In the study phase, listeners made lexical decisions on word and nonword stimuli. In the test phase, listeners either performed lexical decisions again or judged whether each stimulus had been presented in the study phase. The results showed that responses were faster to words that were produced by the same speaker in old/new judgment, but not in lexical decision, indicating that speaker variability affected explicit memory, but not implicit memory. Luce and Lyons (1998) proposed that the speaker effect did not emerge in lexical decision because lexical decision requires rapid responses, whereas the speaker effect takes more time to develop because speaker-specific details in lexical representations are not available as early as the more abstract underlying forms. In other words, when processing is fast, the speaker effect is attenuated. As processing unfolds over time, the speaker effect becomes more pronounced. This time-course hypothesis was further tested in McLennan and Luce (2005). In three long-term priming experiments, the authors manipulated ease of word/nonword discrimination (easy vs. difficult) and response format (speeded vs. delayed response). These manipulations required listeners to process words at different speeds (i.e., difficult distinctions and the delayed response format would result in slower processing). As predicted, the speaker effect was found only when discrimination was difficult and in the delayed response format, suggesting that processing speaker-specific information takes time. Mattys and Liss (2008) showed that the speaker effect on long-term priming was more pronounced in stimuli produced by dysarthric speakers than those produced by normal control speakers. Because processing dysarthric speech is more difficult and presumably requires more time than processing non-dysarthric speech, this result is consistent with the idea that processing speaker-specific information requires time. Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 7 Vitevitch and Donoso (2011) showed in a lexical decision task that it was easier for listeners to detect a speaker change when nonword stimuli were word-like than when the stimuli were less word-like. Because the word-like stimuli were more difficult to process, they required more processing time than less word-like stimuli. Therefore, the better detection of speaker change in the word-like stimuli suggests that speaker-specific information requires more time to process. In sum, findings from these long-term priming studies indicate that speaker information is encoded in memory for words. This observation challenges the traditional assumption that lexical representations are highly abstract and devoid of surface acoustic variability. These studies also suggest that processing speaker information takes time. When processing is slowed down due to stimulus difficulty or response format, the effect of speaker variability tends to emerge. Speaker specificity in lexical processing Although the long-term priming studies provided evidence for the richness of lexical representations, it is not clear to what extent such a task can reveal the word recognition process itself (Luce & Lyons, 1998). In particular, the process of spoken word recognition is usually characterized by the activation of multiple word candidates and competition among the candidates (Luce & McLennan, 2005). In addition, lexical activation and competition is a timesensitive, “on-line” process that has been extensively examined with various research paradigms (Grosjean & Frauenfelder, 1996). Few studies, however, have examined the role of speaker variability in these short-term lexical processes. An eye-tracking study by Creel, Aslin, and Tanenhaus (2008) demonstrated the on-line use of speaker information in lexical disambiguation. The stimuli included pairs of words spoken by the same speaker or different speakers. Eyetracking results showed fewer fixations on competitors for words from the different-speaker pairs Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 8 than for words from the same-speaker pairs, indicating that speaker information facilitated disambiguation of lexical competitors. A similar response pattern was found in a second experiment, in which listeners learned to identify visual shapes from novel labels spoken by the same speaker or different speakers. There were more fixations on words with same-speaker cohorts, but fewer fixations on words with same-speaker competitors. These results indicate that listeners are able to use speaker-specific information encoded in lexical representations during lexical activation and competition. Short-term paired priming is a paradigm that has been used extensively to study lexical processing (see Zwitserlood, 1996, for a summary), but has not been applied to the study of speaker variability. In this paradigm, a prime and a target that are related in certain ways (e.g., phonologically) are presented. The accuracy and latency of responses to the target are used to evaluate the effects of the prime-target relationship, thereby revealing the organization of the lexicon and the processes of accessing lexical representations. Of particular relevance to the present study, short-term priming has been used to examine the effects of acoustic variability on spoken word recognition and the time course of the effects. Andruski, Blumstein, and Burton (1994) used short-term paired priming to investigate the effect of subphonetic variability on lexical access. Prime-target pairs that varied in semantic/associative relationship (e.g., kingqueen) were used as stimuli, and participants were asked to make lexical decisions on the targets. The results showed that subtle voice onset time (VOT) differences, which did not change the perception of voicing categories, affected the magnitude of semantic/associative priming, suggesting that detailed acoustic information was not discarded during the recognition process. Rather, subphonetic information affected access to word meaning. In addition, the effect of Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 9 subphonetic variability appeared at a relatively short lag (50 ms), but not at a long lag (250 ms) between prime and target, indicating that the impact of subphonetic variability dissipated quickly. Andruski et al. (1994) demonstrated that acoustic variability that does not alter word identity could still affect lexical processing. The study also provided evidence for the time course of the subphonetic variability effect. Given the multiple sources of acoustic variability in the speech signal, one might ask whether other sources of acoustic variability, such as speaker variability, would affect lexical processing in a similar way. Using the short-term paired priming paradigm to investigate speaker variability could answer the following questions: Does speaker variability affect access to word meaning as subphonetic variability does? Does speaker variability affect access to word form in addition to word meaning? Is the magnitude and time course of the speaker variability effect comparable to those of the subphonetic variability effect? The answers to these questions will contribute to our understanding of the role of acoustic variability in spoken word recognition. The present study In this study, the short-term paired priming paradigm was used to examine the effect of speaker variability on accessing the form and meaning of spoken words. Repetition priming (Experiment 1) and semantic/associative priming (Experiment 2) were used as indices of accessing word form and meaning, respectively. In repetition priming, the processing of a word (target) is facilitated when it is preceded by exactly the same word (prime). The magnitude of repetition priming is usually reduced when the prime and target do not match exactly. If speakerspecific information is part of the information stored in lexical representations, the mismatch in speaker between prime and target should result in attenuated repetition priming. In semantic/associative priming, the processing of a target (e.g., queen) is facilitated when it is Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 10 preceded by a semantically/associatively related prime (e.g., king). If speaker variability affects access to word meaning, a mismatch in speaker between prime and target should result in attenuated semantic/associative priming. In addition, two interstimulus intervals (ISIs) were used to examine the time course of the speaker variability effect. Andruski et al. (1994) found that subphonetic variability affected lexical access at a short lag (50 ms), but not at a long lag (250 ms) between prime and target, indicating the effect was short-lived. On the other hand, long-term priming studies (e.g., Luce and McLennan, 2005) suggest that the speaker variability effect takes time to develop. The use of both 50 ms and the 250 ms ISIs in the present study would allow for a preliminary examination of the time course of the speaker variability effect in short-term priming. Using the same ISIs as in Andruski et al. (1994) would also allow for a direct comparison on the time course between the speaker variability effect and the subphonetic variability effect. The use of these two ISIs was motivated by considerations of auditory memory and backward masking (Andruski et al., 1994). Because auditory details are assumed to fade fast in memory, using a short ISI ensures that a priming effect could be detected. However, too short an ISI may elicit the effect of backward masking (the masking effect of a following stimulus leading to a failure in perceiving the preceding stimulus). According to Massaro (1970), the effect of backward masking from a pure tone can be perceptually neglected when the ISI is at least 250 ms. For white noise, however, it is about 50 ms (Wilson & Carhart, 1971). Therefore, we chose 50 ms and 250 ms as the ISIs in order to explore the time course issue. Finally, examining both repetition and semantic priming may shed light on the issue of depth of processing speaker variability in spoken word recognition. McLennan and Luce (2005) noted that the depth of processing and time course of processing are usually coextensive. Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 11 Therefore, it is difficult to discriminate between an account based on time course and one based on depth of processing (p. 317). The present study attempted to circumvent the issue by manipulating time course (via ISI manipulation) and depth of processing (repetition vs. semantic/associative priming) independently. Processing semantic/associative information presumably involves a deeper level of processing compared to processing phonological information. Based on Goldinger’s (1996) finding that speaker-specific information affected word recognition only when words were encoded at relatively shallow levels, it was expected that the speaker variability effect would be more likely to emerge in repetition priming than in semantic/associative priming. On the other hand, Andruski et al. (1994) showed that subphonetic variability in fact affected access to word meaning, which invites the question of whether speaker variability would have a similar effect as subphonetic variability does. Comparing the effect of speaker variability in these two types of priming would therefore elucidate the depth of processing issue. However, it should be noted that speaker variability is not exactly analogous to subphonetic variability. While a change in VOT may affect word identity, variability across speakers usually does not affect word identity. Viewed this way, speaker variability may not affect access to word meaning as it does to word form. In sum, the goal of the present study was to use the short-term paired priming paradigm to explore how speaker variability affects spoken word recognition. Our major question was whether a speaker difference between the prime and target would result in attenuated repetition and semantic/associative priming. By using two ISIs between prime and target, this study would allow for a preliminary examination of the time course of the speaker variability effect. By investigating both repetition and semantic/associative priming, this study would allow for an examination of the depth of processing issue in processing speaker variability. Compared to Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 12 previous long-term priming studies that addressed the nature of lexical representations, the present study is expected to extend our understanding of how speaker variability is processed during spoken word recognition. Experiment 1: Repetition Priming The purpose of the repetition priming experiment was to examine the effect of speaker variability on accessing the phonological form of spoken words. Prime-target pairs were presented aurally to listeners, who were asked to make lexical decisions on the targets. Each target was paired with four types of real word primes varying in word relation (repetition vs. unrelated) and speaker relation (same speaker vs. different speakers). Table 1 illustrates the design. (Insert Table 1 about here) A word relation effect (i.e., repetition priming) was expected, meaning that responses to targets would be more accurate and/or faster when the targets were preceded by a repetition prime. A speaker relation effect was also expected, meaning that responses to targets would be less accurate and/or slower when the primes and targets were produced by different speakers. The primary hypothesis to be tested in this experiment was that speaker variability would affect the magnitude of repetition priming. We predicted that there would be a word relation speaker relation interaction; that is, the magnitude of the repetition priming would be reduced when the prime and target were produced by different speakers. The time-course hypothesis (i.e., the speaker effect takes time to develop) would also be tested. In particular, we predicted that the word relation speaker relation interaction would be greater at 250 ms ISI than at 50 ms ISI. Method Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 13 Materials. Twenty-six English words were selected as real word targets. Each target was paired with four types of real word primes: (1) a repetition prime produced by the same speaker, (2) a repetition prime produced by a different speaker, (3) an unrelated prime produced by the same speaker, and (4) an unrelated prime produced by a different speaker. The targets and unrelated primes were identical to those used in Andruski et al. (1994, p. 187). In addition, 26 pronounceable nonword items were selected as nonword targets. Each nonword target was paired with the same set of primes as the real word targets such that listeners would not be able to predict the lexical status of the targets. In sum, there were 104 trials with real word targets and 104 trials with nonword targets for a total of 208 trials. The complete set of stimuli is listed in the Appendix. A female and a male speaker of American English from the same town in central Ohio recorded the stimuli. The recordings were made in a sound-treated booth with a condenser microphone connected through a preamplifier and analog-to-digital converter to a computer. The recordings were digitized with the Brown Lab Interactive Speech System (BLISS) (Mertus, 2000) at 22050 Hz with 14-bit quantization. Each stimulus item was identified from the waveform display and saved as an audio file. Peak amplitude was normalized across all stimulus items. Four stimulus lists were created: (1) Targets in the prime-target pairs were produced by the female speaker with an ISI of 50 ms. (2) Targets in the prime-target pairs were produced by the female speaker with an ISI of 250 ms. (3) Targets in the prime-target pairs were produced by the male speaker with an ISI of 50 ms. (4) Targets in the prime-target pairs were produced by the male speaker with an ISI of 250 ms. Each participant was randomly assigned to receive one of the four lists. The purpose of treating ISI and target voice as between-subject variables was to minimize the repetition of targets during stimulus presentation. In the current design, a target is Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 14 paired with four types of primes, meaning that the participants would hear the same target four times during the presentation. Although assigning participants to all ISI and target voice conditions would have had the advantage of controlling for individual variability, it would also have resulted in even more repetitions of the targets. To minimize potential long-term repetition effects due to the repetition, we opted to assign participants to different ISI and target voice conditions, in addition to randomizing the order of stimulus presentation for individual participants. To assess the acoustic differences between the stimuli produced by the two speakers, F0 and duration of the real word targets were measured. The results are shown in Table 2. For each word, an F0 contour was extracted by the Pitch program in BLISS with a time autocorrelation algorithm. The extracted F0 values were then used to calculate the mean F0 for each word. As expected, the mean F0 was higher for the female speaker than for the male speaker, t(25) = 38.4, p < .001. Word duration was also longer for the female speaker than for the male speaker, t(25) = 6.41, p < .001. (Insert Table 2 about here) Participants. Sixty students (40 females and 20 males) at Ohio University participated in the experiment. All participants were native speakers of American English. The age of the participants ranged from 18 to 26 years (M = 20, SD = 2). All participants were screened for normal hearing, defined as pure-tone, air-conducted thresholds of 20 dB HL at octave frequencies from 1000 Hz to 4000 Hz. Procedure. Participants were tested individually in a sound-treated booth. A Windows XP personal computer equipped with the subject-testing program AVRunner in BLISS was used for stimulus delivery and response acquisition. The participants were screened for normal Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 15 hearing immediately before the experiment. Each participant was then randomly assigned to receive one of the four stimulus lists. The order of stimulus presentation in a list was randomized for each participant. The stimuli were presented through a pair of headphones connected to the computer. The participants were told that they would be listening to pairs of word and nonword items. Their task was to decide whether the second item in a pair was a real word or not a word in the English language. They were told to respond with the index finger of their dominant hand by pressing buttons labeled WORD or NONWORD on a computer keyboard. They were also told to respond as quickly as possible because their responses would be timed. Prior to the actual experiment, 10 practice trials were provided to familiarize the participants with the response format. Each experimental session took approximately 20 minutes. Data analysis. The lexical decision responses acquired by BLISS were analyzed for accuracy and reaction time. Accuracy was defined as the percentage of correct responses to real word targets. Nonword targets were intended as fillers and were not included in the statistical analyses. Reaction time was measured from target onset. Only correct responses were included in the reaction time analysis. Analyses of variance (ANOVAs) were conducted on arcsinetransformed accuracy and log-transformed reaction time with word relation (repetition vs. unrelated) and speaker relation (same vs. different) as within-subject factors, ISI (50 ms vs. 250 ms) as a between-subject factor, and participants (F1) and stimulus items (F2) as random factors. Because the acoustic analysis (Table 2) showed that word duration was significantly longer for the female speaker than for the male speaker, word duration was a potential confounding variable in the reaction time analysis. Therefore, we conducted two sets of reaction time analysis, one based on reaction time measured from target onset and the other based on reaction time measured from target offset. In the target-offset analysis, target duration was Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 16 subtracted from reaction time individually for each participant and item. The two sets of analysis generated essentially the same statistical conclusions with few exceptions. Therefore, we would only report the results from the target-onset analysis and note the exceptions. Results and discussion Table 3 shows the mean accuracy of lexical decision responses. The overall accuracy was 96.5% (SD = 4.3), indicating that the participants were highly accurate in this task. An ANOVA showed a main effect of word relation [F1(1, 58) = 41.66, p < .001, p2 = .42, F2(1, 102) = 27.13, p < .001, p2 = .21]. The main effect of speaker relation was also significant, but only in the analysis by participants [F1(1, 58) = 5.64, p = .021, p2 = .09]. As expected, the main effect of word relation showed that responses to repetition trials were more accurate than those to unrelated trials, indicating repetition priming. Contrary to intuition, the main effect of speaker relation indicated that responses to same-speaker trials were actually less accurate than those to different-speaker trials. However, because the difference was minimal (1%) and the overall accuracy was quite high, the statistical difference is not likely to be of practical significance. Therefore, the focus of our interpretation would be on the reaction time data. (Insert Table 3 about here) Table 4 shows the mean reaction time of lexical decision responses. An ANOVA showed a main effect of word relation [F1(1, 58) = 253.02, p < .001, p2 = .81, F2(1, 102) = 642.07, p < .001, p2 = .86]. The word relation speaker relation interaction [F1(1, 58) = 7.4, p = .009, p2 = .11, F2(1, 102) = 5.95, p = .016, p2 = .06] was also significant. The analysis by items further revealed a main effect of ISI [F2(1, 102) = 4.14, p = .044, p2 = .04]. There were no other effects. As expected, the main effect of word relation showed that responses to the repetition trials were faster than those to the unrelated trials, indicating repetition priming. Importantly, the word Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 17 relation speaker relation interaction indicated that the magnitude of repetition priming differed between the same- and different-speaker trials. As predicted, the degree of facilitation by repetition primes was reduced when the prime and target were produced by different speakers. As for the effect of ISI, although the reduction of priming due to different speakers appeared to be more pronounced at 250 ISI as was predicted, the word relation speaker relation ISI interaction did not reach statistical significance [F1(1, 58) = 3.19, p = .079; F2(1, 102) = 0.85, p = .358]. Finally, the word relation speaker relation interaction did not reach statistical significance in the analysis of reaction time measured from target offset. (Insert Table 4 about here) In sum, the accuracy analysis showed that participants were highly accurate in this task, justifying the use of reaction time as a relevant dependent measure. As expected, both accuracy and reaction time analyses showed robust repetition priming. Contrary to expectation, however, processing same-speaker trials was neither more accurate nor faster than processing differentspeaker trials. The absence of same-speaker advantage could be due to the repetition of targets and the use of only two speakers. These two factors could have resulted in excessive familiarity with the speakers, neutralizing the speaker relation effect. Nonetheless, the primary prediction of the experiment (i.e., repetition would be attenuated when prime and target were produced by different speakers) was supported by the reaction time result. The magnitude of repetition priming was reduced when different speakers produced the prime and target than when the same speaker produced the prime and target. This is to our knowledge the first demonstration of short-term repetition priming being modulated by speaker variability. Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 18 By contrast, the prediction about the time course of speaker variability in repetition priming was not supported by the data. It was predicted that the reduction of repetition priming due to speaker variability would be more pronounced at 250 ms ISI than at 50 ms ISI. Although the trend observed in the data is consistent with the prediction, the difference was not statistically significant. Based on the trend, it is speculated that an effect might emerge at a longer ISI than those examined in this experiment. Experiment 2: Semantic/Associative Priming Results from the repetition priming experiment showed that speaker variability affected access to word form. The magnitude of repetition priming was attenuated when the prime and target were produced by different speakers. The purpose of the present experiment was to examine whether speaker variability would have a similar effect on accessing word meaning. As in the repetition priming experiment, prime-target pairs were presented aurally to listeners, who were asked to make lexical decisions on the targets. The targets used in this experiment were identical to those used in the repetition priming experiment. Each target was paired with four types of real word primes varying in word relation (semantic/associative vs. unrelated) and speaker relation (same speaker vs. different speakers). Table 5 illustrates the design. As in the repetition priming experiment, two ISIs (50 ms and 250 ms) were used to explore the time course of the speaker variability effect. Because semantic/associative relationship presumably involves a deeper level of processing than processing word form, results from the present experiment could be compared to those from the repetition priming experiment to address the depth of processing issue regarding speaker variability. (Insert Table 5 about here) Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 19 A word relation effect (i.e., semantic/associative priming) was expected, meaning that responses to targets would be more accurate and/or faster when the targets were preceded by a semantically/associatively related prime. A speaker relation effect was also expected, meaning that responses to targets would be less accurate and/or slower when the primes and targets were produced by different speakers. However, the absence of the speaker relation effect on repetition priming suggests that this effect might not be obtained in semantic/associative priming. That is, if speaker variability does not affect access to word forms, it is unlikely to affect access to word meaning. The primary hypothesis to be tested in this experiment was that speaker variability would affect the magnitude of semantic/associative priming. We predicted that there would be a word relation speaker relation interaction; that is, the magnitude of semantic/associative priming would be reduced when the prime and target were produced by different speakers. If so, the result would indicate that speaker variability affects not only access to word form, but also access to word meaning. Finally, the time-course hypothesis would be evaluated. If processing speaker variability effect takes time, the word relation speaker relation interaction should be greater at the longer ISI (250 ms). Alternatively, if the time course of processing speaker variability is similar to that of subphonetic variability (Andruski et al., 1994), the word relation speaker relation interaction should be greater at the shorter ISI (50 ms). Method Materials. The targets used in this experiment were identical to those used in the repetition priming experiment. Each target was paired with four types of real word primes: (1) a semantically/associatively related prime produced by the same speaker, (2) a semantically/associatively related prime produced by a different speaker, (3) an unrelated prime produced by the same speaker, and (4) an unrelated prime produced by a different speaker. Both Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 20 the primes and targets were identical to those used in Andruski et al. (1994). The same 26 pronounceable nonword items as in the repetition priming experiment were selected as nonword targets. Each nonword target was paired with the same set of primes as the real word targets. In sum, there were 104 trials with real word targets and 104 trials with nonword targets for a total of 208 trials. The stimuli are listed in the Appendix. The same female and male speakers as in the repetition priming experiment recorded the stimuli. Four stimulus lists were created varying in target voice (female vs. male) and ISI (50 ms vs. 250 ms). Each participant was randomly assigned to receive one of the four lists. Participants. Sixty students (45 females and 15 males) at Ohio University participated in this experiment. All participants were native speakers of American English. The age of the participants ranged from 18 to 25 years (M = 20, SD = 2). All participants were screened for normal hearing as defined earlier in the repetition priming experiment. None of the participants participated in the repetition priming experiment. Procedure. The procedure was identical to that used in the repetition priming experiment. Data analysis. The same data analyses were performed as in the repetition priming experiment, except that the two levels of word relation were now semantically/associatively related and unrelated. In addition, we included an extra factor of target voice (female vs. male) in the statistical analyses. Although no predictions were made regarding target voice, preliminary data analysis showed distinct patterns between responses to the two speakers. Therefore, we decided to report data by target voice as well. In other words, there were two within-subject factors (word and speaker relation) and two between-subject factors (ISI and target voice) in the current experiment. Results and discussion Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 21 Table 6 shows the mean accuracy of lexical decision responses. The overall accuracy was 96.4% (SD = 3.5), indicating that the participants were highly accurate in this task. An ANOVA showed a main effect of word relation [F1(1, 56) = 21.02, p < .001, p2 = .27, F2(1, 100) = 15.73, p < .001, p2 = .14]. As expected, responses to semantically/associatively related trials were more accurate than those to unrelated trials, indicating semantic/associative priming. There were no other effects. As in the repetition priming experiment, the overall accuracy of lexical decision responses was fairly high. Therefore, our interpretation would focus on the reaction time results. (Insert Table 6 about here) Table 7 shows the mean reaction time of lexical decision responses. An ANOVA showed a main effect of word relation [F1(1, 56) = 225.82, p < .001, p2 = .80, F2(1, 100) = 128.98, p < .001, p2 = .56] and a main effect of target voice [F1(1, 56) = 6.95, p = .011, p2 = .11, F2(1, 100) = 11.17, p = .001, p2 = .10]. There were also two interactions: speaker relation target voice [F1(1, 56) = 11.02, p = .002, p2 = .16, F2(1, 100) = 6.18, p = .015, p2 = .06] and word relation speaker relation target voice [F1(1, 56) = 5.80, p = .019, p2 = .09]. As expected, the main effect of word relation showed that responses to semantically/associatively related trials were faster than those to unrelated trials, indicating semantic/associative priming. The main effect of target voice showed that responses to the stimuli produced by the male speaker were faster. However, recall that the acoustic analysis showed that the average duration of stimuli produced by the male speaker was shorter (Table 2), suggesting that the faster response to the male speaker could be due to the shorter stimulus duration. Indeed, the analysis on reaction time measured from target offset did not show an effect of target voice. As for the primary hypothesis that the magnitude of priming would be attenuated when the prime and target were produced by different speakers, we did not find a significant word Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 22 relation speaker relation interaction as was found in the repetition priming experiment. However, the highest order interaction (word relation speaker relation target voice) indicated that the magnitude of semantic/associative priming depended on both speaker relation and target voice. In particular, Table 7 shows that the magnitude of semantic/associative priming was reduced in different-speaker trials, but only when the target was produced by the female speaker. By contrast, when the target was produced by the male speaker, the magnitude of semantic/associative priming was not attenuated in the different-speaker trials. (Insert Table 7 about here) In sum, the accuracy analysis again showed that participants were highly accurate in this task. As expected, both accuracy and reaction time analyses showed robust semantic/associative priming. As in the repetition priming experiment, there was no overall effect of speaker relation. Responses to same- and different-speaker trials were comparable in both accuracy and reaction time. Although this result was somewhat counterintuitive, it could be anticipated from a similar null result in the repetition priming experiment. That is, if speaker variability does not affect access to word form, it is unlikely to affect access to word meaning. The critical result from this experiment was the word relation speaker relation target voice interaction. Unlike repetition priming, semantic priming was modulated not only by speaker relation but also by target voice. The magnitude of semantic/associative priming was attenuated in the different-speaker trials, but only when the target was produced by the female speaker. An alternative way of describing this result is that there was more priming when the female speaker produced the prime than when the male speaker produced the prime. Since the female speaker took longer to produce the primes, listeners might have had more time to process the primes by the time the target arrived, resulting in the more pronounced priming. Compared to Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 23 repetition priming, the absence of a robust word relation speaker relation interaction suggests that as the level of processing becomes deeper (from form to meaning), the impact of speaker variability becomes less prominent. In other words, acoustic details arising from speaker differences may not be as important in accessing word form as in accessing word meaning. Finally, there was no evidence that the reduction in semantic/associative priming varied between the two ISIs. This null result did not allow us to distinguish between the time-course hypothesis (Luce & Lyons, 1998) and the transient nature of processing acoustic variability (Andruski et al., 1994). As noted in the discussion of Experiment 1, using a longer ISI may allow for a further test of the time course of processing speaker variability in priming. General Discussion The research question addressed in this study was whether speaker variability affects spoken word recognition. Specifically, we asked whether speaker variability would affect the magnitude of short-term priming, whether the effect would be found at the level of accessing the form or meaning of spoken words, and what would be the time course of the speaker variability effect. In addition to the well-documented repetition and semantic/associative priming effects, the primary prediction of the study was that the magnitude of priming would be reduced when the prime and target were produced by different speakers. We also predicted that the reduction would be more prominent in repetition priming than semantic/associative priming. In addition, based on the time-course hypothesis (Luce & Lyons, 1998), we expected that the reduction in repetition priming would be more pronounced at the longer ISI (250 ms). Based on Andruski et al. (1994), we expected that the reduction in semantic/associative priming would be more pronounced at the shorter ISI (50 ms). Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 24 Because the accuracy of lexical decision responses in both experiments were near ceiling, our interpretations focused on the reaction time results. As expected, the results showed robust repetition priming (Forster & Davis, 1984) and semantic/associative priming (Meyer & Schvaneveldt, 1976). The average facilitation by a repetitive prime was 169 ms, and the average facilitation by a semantic/associative prime was 73 ms. The magnitude of semantic/associative priming was somewhat smaller than what was found in Andruski et al. (1994), who showed semantic/associative priming of approximately 100 to 150 ms with the same set of stimuli as in the present study (p. 174 & 178). Importantly, the magnitude of facilitation in both repetition and semantic/associative priming varied as a function of speaker relation, indicating that priming was dependent on speaker-specific information. As noted in the introduction, long-term priming studies showed that word recognition/memory was slower and/or less accurate by using different speakers across the study and test phases (Schacter & Church, 1992; Church & Schacter, 1994; Goldinger, 1996; Luce & Lyons, 1998; McLennan & Luce, 2005; Mattys & Liss, 2008; Vitevitch & Donoso, 2011). The reduction in the magnitude of priming is usually interpreted as the prime having activated a different lexical form (Luce & Lyons, 1998), suggesting that lexical representation contains speaker-specific information. In the present study, the reduction in the magnitude of short-term repetition priming in different-speaker trials could similarly suggest that the same word produced by different speakers was processed as different lexical forms. That is, because the prime did not activate the same lexical form, in which speaker information was encoded, there was not as much facilitation as there was when the prime and target were produced by the same speaker. Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 25 Alternatively, the reduction in repetition priming could be due to short-lived processes, but not due to activation of a different lexical form. Compared to the long-term repetition priming studies, listeners in the present study had much less time to actually encode the primes before they had to process and respond to the targets. Therefore, the reduction in repetition priming could reflect listeners’ making online perceptual adjustments to speaker differences, a process commonly referred to as speaker normalization (Pisoni, 1997). Although the present results cannot distinguish these two interpretations, available evidence on the representation (e.g., the long-term repetition priming studies reviewed earlier) and on-line processing (e.g., Creel et al., 2008, and the present study) of spoken words clearly suggests that lexical forms contain speaker-specific information and that this information is used during the online recognition process. The effect of speaker variability, however, was less evident in the processing of word meaning. In the semantic/associative priming experiment, the magnitude of priming was reduced when the prime and target were produced by different speakers. However, this conclusion holds only when the targets were produced by the female speaker, but not the male speaker. A few observations could be made from this result. First, when both target voices were considered together, the reduction of semantic/associative priming in the different-speaker trials was not statistically significant, indicating that the effect of speaker variability on accessing word meaning was not conclusive. In the repetition priming experiment, the average magnitude of facilitation in the same-speaker trials was 26 ms. By contrast, this facilitation disappeared in the semantic/associative priming experiment. The same-speaker trials in fact resulted in an average of 2 ms of inhibition, largely due to the results from the male speaker. Because word meaning presumably involves a deeper level of processing than word form, these results suggest that the Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 26 speaker variability effect occurs primarily at a relatively shallow level of processing (Goldinger, 1996). The lack or inconsistency of facilitation in semantic/associative priming suggests that the effect of speaker-specific information does not persist to accessing word meaning. Second, the absence of reduction by speaker variability in semantic/associative priming contrasts with Andruski et al. (1994), who found that subphonetic variability affected access to word meaning. Andruski et al. (1994) showed that fine acoustic structure (i.e., VOT) that did not change phonetic category (i.e., voicing) could still impact access to word meaning. Although both subphonetic and speaker variability involve changes in fine acoustic structure, this contrast suggests that they are processed differently during spoken word recognition. Alternatively, as noted in the introduction, this contrast could be due to the nature of the two sources of acoustic variability. A change in VOT is more likely to affect word identity than a change of speakers is. Therefore, it may not be too surprising that speaker variability did not affect access to word meaning as much as subphonetic variability does. This conclusion, however, is tentative because a reduction in semantic/associative priming was in fact observed when the targets were produced by the female speaker. That is, the reduction of priming due to speaker variability depended on the specific voice used for the targets. Dialect difference is unlikely to account for the contrast between the two target voices because both speakers grew up in the same town. Intelligibility difference is also unlikely to be the reason because there was no overall effect of target voice in response accuracy. On the other hand, because the two speakers were of different genders, the speaker effects could be a gender effect, but not a speaker effect per se. As a first attempt to employ the short-term paired priming paradigm to investigate the speaker variability effect, our motivation for using speakers of different genders was to maximize the speaker differences such that a speaker effect could be Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 27 detected. Future studies could use speakers of the same gender to control for the contribution of speaker gender. A second possibility for the contrast between the female and male results is that the listeners might have had more time to process the female stimuli, which allowed the speaker effect to surface. As shown in the acoustic analysis, average duration of the female stimuli was longer than that of the male stimuli. Results from the reaction time analysis also showed that responses to female stimuli were slower. It is possible that listeners had more time to process the female stimuli, which allowed speaker-specific information to affect semantic/associative priming. The potential relevance of time, however, was not supported by the ISI analyses. Based on the time-course hypothesis (Luce & Lyons, 1998), it was expected that reduction of repetition priming due to speaker variability would be more pronounced at the longer ISI (250 ms). Based on Andruski et al.’s (1994) finding regarding subphonetic variability, it was expected that reduction of semantic/associative priming due to acoustic variability would be more pronounced at the shorter ISI (50 ms). Neither prediction was supported by the data, i.e., the word relation speaker relation ISI interaction was not statistically significant. Therefore, no conclusion could be drawn from the current study regarding the time course of the speaker variability effect in short-term priming. However, we did observe a trend of more pronounced reduction of repetition priming due to speaker variability at the longer ISI (250 ms), which would be consistent with the time-course hypothesis. We suspect that the ISIs used in the present study were not within the time frame in which speaker variability would exert its effect on spoken word recognition. Future studies could investigate whether a longer ISI will allow the speaker variability effect to emerge. Conclusion Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 28 The present study contributes to the literature on spoken word recognition by examining the processing of speaker variability in two well-established cognitive phenomena, i.e., shortterm repetition and semantic/associative priming. Complementing long-term priming studies that showed the relevance of speaker information in lexical representations, the present study is the first to demonstrate that speaker variability affects the access to word form, as indicated by a reduction in the magnitude of short-term repetition priming. By investigating both repetition and semantic/associative priming, the present study was able to address the depth of processing for speaker variability. With the use of two ISIs, the present study allowed for a preliminary investigation of the time course of the speaker variability effect. With the same set of stimuli as used in Andruski et al. (1994), the present study allowed for a direct comparison between two common sources of acoustic variability. Our findings indicated that speaker variability affects access to word form to a greater extent than it does access to word meaning, suggesting that speaker variability was processed at a relatively shallow level. The time course of processing speaker variability, however, is not conclusive from the current data. Given that only two speakers were used for the stimuli and that gender was a potential confound, future studies could explore the role of speaker variability in priming by using more speakers, controlling for speaker gender, and using ISIs that are longer than those employed in the current study. Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 29 References Andruski, J. E., Blumstein, S. E., & Burton, M. (1994). The effects of subphonetic differences on lexical access. Cognition, 52, 163-187. Church, B. A., & Schacter, D. L. (1994). Perceptual specificity of auditory priming: Implicit memory for voice intonation and fundamental frequency. Journal of Experimental Psychology: Learning, Memory, and Cognition, 20, 521-533. Creel, S. C., Aslin, R. N., & Tanenhaus, M. K. (2008). Heeding the voice of experience: The role of talker variation in lexical access. Cognition, 106, 633-664. Forster, K. I., & Davis, C. (1984). Repetition priming and frequency attenuation in lexical access. Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 680-698. Goldinger, S. D. (1996). Words and voices: Episodic traces in spoken word identification and recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 22, 1166-1183. Grosjean, F., & Frauenfelder, U. H. (1996). A guide to spoken word recognition paradigms: Introduction. Language and Cognitive Processes, 11, 553-558. Jackson, A., & Morton, J. (1984). Facilitation of auditory recognition. Memory & Cognition, 12, 568-574. Johnson, K., & Mullennix, J. W. (1997). Complex representations used in speech processing: Overview of the book. In K. Johnson & J. W. Mullennix (Eds.), Talker variability in speech processing (pp. 1-8). San Diego, CA: Academic Press. Luce, P. A., & Lyons, E. A. (1998). Specificity of memory representations for spoken words. Memory & Cognition, 26, 708–715. Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 30 Luce, P. A., & McLennan, C. T. (2005). Spoken word recognition: The challenge of variation. In D. B. Pisoni & R. E. Remez (Eds.), The handbook of speech perception (pp. 591–609). Malden, MA: Blackwell Publishing. Massaro, D. W. (1970). Preperceptual auditory images. Journal of Experimental Psychology, 85, 411-417. Mattys, S. L., & Liss, J. M. (2008). On building models of spoken-word recognition: When there is as much to learn from natural “oddities” as artificial normality. Perception & Psychophysics, 70, 1235-1242. McLennan, C. T., & Luce, P. A. (2005). Examining the time course of indexical specificity effects in spoken word recognition. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 306-321. Mertus, J. A. (2000). The Brown Lab Interactive Speech System [Computer software]. Providence, RI: Brown University. Retrieved from http://mertus.org/Bliss/. Meyer, D. E., & Schvaneveldt, R. W. (1976). Meaning, memory structure, and mental processes. Science, 192, 27-33. Nygaard, L. C., Sommers, M. S., & Pisoni, D. B. (1994). Speech perception as a talkercontingent process. Psychological Science, 5, 42-46. Palmeri, T. J., Goldinger, S. D., & Pisoni, D. B. (1993). Episodic encoding of voice attributes and recognition memory for spoken words. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 309-328. Pisoni, D. B. (1997). Some thoughts on “normalization” in speech perception. In K. Johnson & J. W. Mullennix (Eds.), Talker variability in speech processing (pp. 9-32). San Diego, CA: Academic Press. Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 31 Schacter, D. L., & Church, B. A. (1992). Auditory priming: Implicit and explicit memory for words and voices. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 915-930. Vitevitch, M. S., & Donoso, A. (2011). Processing indexical information requires time: Evidence from change deafness. The Quarterly Journal of Experimental Psychology, 64, 14841493. Wilson, R. H., & Carhart, R. (1971). Forward and backward masking: Interactions and addivity. Journal of the Acoustical Society of America, 49, 1254-1263. Zwitserlood, P. (1996). Form priming. Language and Cognitive Processes, 11, 589-596. Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING Table 1 An example of the repetition priming experimental setup Prime (speaker) Target (speaker) queen (male) queen (male) queen (female) bell (male) bell (female) Word relation repetition repetition unrelated unrelated Speaker relation same different same different 32 Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 33 Table 2 Mean F0 (in Hz) and duration (in ms) of word targets produced by the speakers Mean F0 Word queen dog pie shawl stick spy bread bottom exam pin short foot wild coach book war forest step friend verse drug hand fruit mine hurt hug M (SD) Male 117 105 112 114 116 105 106 107 112 112 144 123 100 119 114 118 105 117 107 112 112 108 117 105 114 110 113 (8) Duration Female 225 213 220 212 236 209 199 190 200 216 245 240 207 227 243 202 214 237 204 210 197 197 247 199 242 205 217 (18) Male 455 476 383 573 632 688 388 353 584 430 643 436 673 551 345 471 589 635 529 479 486 589 467 475 391 413 505 (102) Female 543 544 559 678 726 775 511 538 802 395 798 580 505 671 450 474 728 785 745 569 599 616 749 685 625 504 621 (118) Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 34 Table 3 Mean accuracy of lexical decision responses (in % with SD, n = 15) in the repetition priming experiment ISI (ms) 50 250 M Same speaker Repetition Unrelated 97 (3) 95 (7) 98 (4) 95 (4) 98 (3) 95 (6) M 96 (6) 96 (4) 96 (5) Different speakers Repetition Unrelated 98 (3) 96 (5) 98 (3) 96 (4) 98 (3) 96 (4) M 97 (4) 97 (3) 97 (4) Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 35 Table 4 Mean reaction time of lexical decision responses (in ms with SD, n = 15) and magnitude of facilitation in the repetition priming experiment ISI (ms) 50 250 M Same speaker Repetition Unrelated 747 (114) 921 (126) 784 (149) 974 (154) 765 (133) 947 (142) Facilitation 174 190 182 Different speakers Repetition Unrelated 764 (122) 929 (110) 802 (139) 948 (124) 783 (131) 939 (116) Facilitation 166 146 156 Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING Table 5 An example of the semantic/associative priming experimental setup Prime (speaker) Target (speaker) king (male) queen (male) king (female) bell (male) bell (female) Word relation semantic/associative semantic/associative unrelated unrelated Speaker relation same different same different 36 Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 37 Table 6 Mean accuracy of lexical decision responses (in % with SD, n = 15) in the semantic/associative priming experiment ISI (ms) 50 250 Target voice Female Male M Female Male M Same speaker Semantic Unrelated 96 (4) 95 (5) 98 (3) 95 (3) 97 (4) 95 (4) 99 (2) 94 (5) 98 (2) 95 (3) 98 (2) 95 (4) M 96 (4) 96 (3) 96 (4) 96 (4) 97 (3) 97 (4) Different speakers Semantic Unrelated 97 (4) 96 (4) 98 (2) 96 (4) 97 (3) 96 (4) 97 (3) 96 (4) 97 (3) 95 (3) 97 (3) 96 (3) M 97 (4) 97 (3) 97 (4) 97 (3) 96 (3) 96 (3) Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 38 Table 7 Mean reaction time of lexical decision responses (in ms with SD, n = 15) and magnitude of facilitation in the semantic/associative priming experiment ISI (ms) 50 250 Target voice Female Male M Female Male M Same speaker Semantic Unrelated Facilitation Different speakers Semantic Unrelated Facilitation 857 (90) 775 (104) 816 (104) 892 (144) 823 (171) 858 (159) 848 (113) 774 (106) 811 (114) 881 (141) 816 (156) 848 (150) 942 (110) 834 (115) 888 (124) 989 (139) 869 (135) 929 (148) 85 59 72 97 46 71 915 (101) 856 (88) 886 (98) 945 (132) 899 (156) 922 (144) 67 82 75 64 83 74 Running head: PROCESSING SPEAKER VARIABILITY IN PRIMING 39 Appendix. List of stimuli used in the experiments Semantically/associatively related primes king cat cake cape cane code toast top test tack tall toe tame team page peace pine pace pal poem pill palm pear pit pain pet Unrelated primes bell fort seed meat sheet joke size nut rag thief choice hill fan rain week hip seal gift yawn fog race job jet sound niece mesh Repetition primes/Word targets queen dog pie shawl stick spy bread bottom exam pin short foot wild coach book war forest step friend verse drug hand fruit mine hurt hug Nonword targets /ɡlɔ/ /θrɪdʒ/ /ɡaɪnd/ /dɛnd/ /ɡraɪs/ /fʌlə˞/ /dost/ /ʌŋk/ /tʃɑrʃ/ /mʊk/ /tɪtʃ/ /ɡɑrt/ /dʒɑʊn/ /lɜ˞d/ /ɡʊʃ/ /ʃɔɪl/ /wɛɡ/ /dʒæft/ /klænt/ /tʃænd/ /kruθ/ /hjul/ /wɪndʒ/ /ʃaʊs/ /ve/ /tɛlt/